DROUGHT-ASSOCIATED TREE MORTALITY: GLOBAL PATTERNS AND

DROUGHT-ASSOCIATED TREE MORTALITY: GLOBAL PATTERNS AND
DROUGHT-ASSOCIATED TREE MORTALITY: GLOBAL PATTERNS AND
INSIGHTS FROM TREE-RING STUDIES IN THE SOUTHWESTERN U.S.A.
by
Alison Kelly Macalady
____________________________
A Dissertation Submitted to the Faculty of the
SCHOOL OF GEOGRAPHY AND DEVELOPMENT
In Partial Fulfillment of the Requirements
For the Degree of
DOCTOR OF PHILOSOPHY
In the Graduate College
THE UNIVERSITY OF ARIZONA
2015
2
THE UNIVERSITY OF ARIZONA
GRADUATE COLLEGE
As members of the Dissertation Committee, we certify that we have read the dissertation
prepared by Alison Kelly Macalady, titled Drought- Associated Tree Mortality: Global
Patterns and Insights from Tree-Ring Studies in the Southwestern U.S.A., and
recommend that it be accepted as fulfilling the dissertation requirement for the Degree of
Doctor of Philosophy.
_______________________________________________________________________
Date: 05/05/2015
Thomas W. Swetnam
_______________________________________________________________________
Date: 05/05/2015
Julio L. Betancourt
_______________________________________________________________________
Date: 05/05/2015
Connie Woodhouse
_______________________________________________________________________
Date: 05/05/2015
David D. Breshears
_______________________________________________________________________
Date: 05/05/2015
Donald Falk
Final approval and acceptance of this dissertation is contingent upon the candidate’s
submission of the final copies of the dissertation to the Graduate College.
I hereby certify that I have read this dissertation prepared under my direction and
recommend that it be accepted as fulfilling the dissertation requirement.
________________________________________________ Date: 05/05/2015
Dissertation Director: Thomas W. Swetnam
________________________________________________ Date: 05/05/2015
Dissertation Director: Connie Woodhouse
3
STATEMENT BY AUTHOR
This dissertation has been submitted in partial fulfillment of the requirements for
an advanced degree at the University of Arizona and is deposited in the University
Library to be made available to borrowers under rules of the Library.
%ULHITXRWDWLRQVIURPWKLVGLVVHUWDWLRQDUHDOORZDEOHZLWKRXWVSHFLDOSHUPLVVLRQ
SURYLGHGWKDWDQDFFXUDWHDFNQRZOHGJHPHQWRIWKHVRXUFHLVPDGH5HTXHVWVIRUSHUPLVVLRQ
IRUH[WHQGHGTXRWDWLRQIURPRUUHSURGXFWLRQRIWKLVPDQXVFULSWLQZKROHRULQSDUWPD\EH
JUDQWHGE\WKHKHDGRIWKHPDMRUGHSDUWPHQWRUWKH'HDQRIWKH*UDGXDWH&ROOHJHZKHQLQ
KLVRUKHUMXGJHPHQWWKHSURSRVHGXVHRIWKHPDWHULDOLVLQWKHLQWHUHVWVRIVFKRODUVKLS,Q
DOORWKHULQVWDQFHVKRZHYHUSHUPLVVLRQPXVWEHREWDLQHGIURPWKHDXWKRU
SIGNED: Alison Kelly Macalady
4
ACKNOWLEDGEMENTS
Over the last 7 years I have been blessed with teachers, mentors, friends and
family who provided me with an exceptional platform for exploration, research, and
professional development. My advisors Julio Betancourt, Tom Swetnam and Connie
Woodhouse were generous with material and moral support, and gave me freedom to
pursue my own scientific and professional interests. I am thankful for their patience with
my winding road to a finished dissertation. I am also grateful for mentorship and
collaborative opportunities offered to me by Craig Allen, Dave Breshears, Harald
Bugmann, Nathan English, Don Falk, Nate McDowell, and Park Williams.
My time at the University of Arizona was enriched by interactions with a great
cohort of graduate student and post-doc colleagues. Friendship and collaborations with
Henry Adams, Alex Arizpe, Toby Ault, Erica Bigio, Adam Csank, Rebecca Franklin,
Dan Griffin, Chris Guiterman, Christine Hallman, Matthias Kläy, Troy Knight, Ellis
Margolis, Steph McAfee, Kiyomi Morino, Greg Pederson, David Tecklin, Tyson
Swetnam, and Erika Wise were particularly important.
Tucson friends Margaret Adcock, Jenny Ault, Caryn Fraser, Julia Guiterman,
Alison Meadow, Hampton Uzzelle and Bridgid Uzzelle helped me maintain sanity during
bumpy stretches. What would I have done without our excellent dinners, hikes, runs, bike
rides, coffees and long conversations about life, work and parenting? Lucy Bassett,
Valerie Craig and Caroline Simmonds helped me keep my eyes on the prize while I
finished writing in Washington, DC.
Time spent in the field or away at conferences and workshops would not have
been possible without help from my extended family. Liza Ketchum and John Straus,
Casey Murrow, Meredith Wade and Jenna Wade-Murrow, Don and Esther Macalady, and
Seth Macalady and Bonnie Treffnen all helped either in the field or the nursery. Thank
you, you are amazing! Finally, I doubt this dissertation could have come to fruition
without the baby-swaddling, bottle-warming, coffee-brewing, home-repairing, truckpacking, knot-tying, tire-changing, fire-starting, chain-sawing skills (not to mention the
encouragement and moral support) of my dear husband, Derek Murrow.
I also wish to acknowledge sources of financial support that enabled my graduate
studies, my dissertation, and related collaborative research. I was supported at various
times by a Science Foundation Arizona Graduate Fellowship, National Science
Foundation GK-12 STEM Education Fellowship, U.S. Department of Energy, Graduate
Research Environmental Fellowship, University of Arizona Marshall Foundation
Dissertation Fellowship, and a grant from the Los Alamos National Laboratory, Institute
of Geophysics and Planetary Physics.
5
DEDICATION
To the biggest helpers of all, Camille and Willa, my desert girls. Now you can call me a
tree doctor.
And to Derek, for your love, support and patience.
6
TABLE OF CONTENTS
ABSTRACT........................................................................................................................9 CHAPTER 1: INTRODUCTION ..................................................................................... 12 Statement of problem........................................................................................... 12 Background .......................................................................................................... 21 Regional context............................................................................................ 21 The role of dendrochronology in studies of tree mortality ........................... 23 Explanation of the research approach .................................................................. 26 CHAPTER 2: PRESENT STUDY ................................................................................... 29 A global synthesis of drought and heat-associated tree mortality ....................... 30 Tree-ring growth records in piñon pine and predisposition to mortality
during severe droughts ......................................................................................... 31 Pre-formed tree defenses and the risk of mortality during drought ..................... 35 Climatic sensitivity of vertical resin duct formation in piñon pine ..................... 37 REFERENCES ................................................................................................................. 40 APPENDIX A: A GLOBAL OVERVIEW OF DROUGHT AND HEATINDUCED TREE MORTALITY REVEALS EMERGING CLIMATE CHANGE RISKS
FOR FORESTS.................................................................................................................55 Abstract ................................................................................................................56 Introduction ......................................................................................................... 57 Methods............................................................................................................... 57
Results ................................................................................................................. 58
Discussion ........................................................................................................... 64
Conclusions ......................................................................................................... 67
Acknowledgements ............................................................................................. 67
Appendix A ......................................................................................................... 67
References ........................................................................................................... 75
APPENDIX B: GROWTH-MORTALITY RELATIONSHIPS IN PIÑON PINE (PINUS
EDULIS) DURING SEVERE DROUGHTS OF THE PAST CENTURY: SHIFTING
PROCESSES IN SPACE AND TIME ............................................................................ 81 Abstract ............................................................................................................... 82 Introduction ......................................................................................................... 82 7
TABLE OF CONTENTS - continued
Materials and Methods.......................................................................................... 83
Results ................................................................................................................... 86
Discussion ............................................................................................................. 93
Conclusions ........................................................................................................... 94
Supporting Information Descriptions ................................................................... 95
Acknowledgements ............................................................................................... 96
Author Contributions ............................................................................................ 96
References ............................................................................................................. 96
Supporting Information..........................................................................................99
APPENDIX C: MORTALITY RISK OF AN ARIDLANDS CONIFER DURING
SEVERE DROUGHT DEPENDS ON RADIAL GROWTH AND INVESTMENT INTO
DEFENSE ........................................................................................................................ 113 Abstract ............................................................................................................... 113 Introduction ......................................................................................................... 114 Methods............................................................................................................... 119
Study Sites .................................................................................................... 119 Sampling of live-dead tree pairs .................................................................. 120 Dendroecological methods .......................................................................... 121 Statistical analyses of growth and duct variables........................................ 122 Models of mortality risk ............................................................................... 123 Results ................................................................................................................. 126
Resin duct and growth characteristics across space and time .................... 126 Relationships between growt and duct attributes ........................................ 127 Models of mortality risk ............................................................................... 129 Discussion ........................................................................................................... 130
Defense anatomy is strongly associated with drought-mortality risk in P.
edulis .......................................................................................................... 130 Spatial and temporal variability in the importance of growth and resin duct
attributes ...................................................................................................... 131 The importance of resin duct size for mortality risk .................................... 133 Relationships between growth, defense and mortality - are tradeoffs
evident? ........................................................................................................ 134 Conclusion and Implications............................................................................... 137
Acknowledgements ............................................................................................. 139
References ........................................................................................................... 139
Tables .................................................................................................................. 148
Figure Legends .................................................................................................... 151
Figures................................................................................................................. 153
Supplementary Materials .................................................................................... 158
8
TABLE OF CONTENTS - continued
APPENDIX D: CLIMATE AND TREE INFLUENCES ON XYLEM RESIN DUCTS IN
PIÑON PINE ................................................................................................................... 159 Abstract ............................................................................................................... 159 Introduction ......................................................................................................... 160 Methods............................................................................................................... 163
Study Sites .................................................................................................... 163 Field and laboratory methods...................................................................... 165 Statistical analyses ....................................................................................... 166 Climate response of ring widths and resin duct attributes .......................... 168 Results ................................................................................................................. 169
Age and size-related trends in resin duct attributes .................................... 169 Statistical properties of tree-ring width and resin duct chronologies ......... 169 Correlations among growth and resin duct mean-value chronologies........ 170 Relationships between climate and tree-ring chronologies ......................... 171 Discussion ........................................................................................................... 174
Cambial age and tree size-reated changes in resin defenses ...................... 174 Statistical properties of RD chronologies .................................................... 175 Climatic controls on radial growth, resin duct frequency and size ............. 177 Implications for dendroclimatology............................................................. 181 Implications for tree defenses under climate change .................................. 182 Conclusions ......................................................................................................... 183
Acknowledgements ............................................................................................. 184
Citations .............................................................................................................. 184
Tables .................................................................................................................. 192
Figure Captions ................................................................................................... 196
Figures................................................................................................................. 199
Supplementary Material ...................................................................................... 206
9
ABSTRACT
Forests play an important role in the earth system, regulating climate, maintaining
biodiversity, and provisioning human communities with water, food and fuel. Interactions
between climate and forest dynamics are not well constrained, and high uncertainty
characterizes projections of global warming impacts on forests and associated ecosystem
services. Recently observed tree mortality and forest die-off forewarn an acceleration of
forest change with rising temperature and increased drought. However, the processes
leading to tree death during drought are poorly understood, limiting our ability to
anticipate future forest dynamics.
The objective of this dissertation was to improve understanding of droughtassociated tree mortality through literature synthesis and tree-ring studies on trees that
survived and died during droughts in the southwestern USA. Specifically, this
dissertation 1) documented global tree mortality patterns and identified emerging trends
and research gaps; 2) quantified relationships between growth, climate, competition and
mortality of piñon pine during droughts in New Mexico; 3) investigated tree defense
anatomy as a potentially key element in piñon avoidance of death; and, 4) characterized
the climate sensitivity of piñon resin ducts in order to gain insight into potential trends in
tree defenses with climate variability and change.
There has been an increase in studies reporting tree mortality linked to drought,
heat, and the associated activity of insects and pathogens. Cases span the forested
continents and occurred in water, light and temperature-limited forests. We hypothesized
that increased tree mortality may be an emerging global phenomenon related to rising
10
temperatures and drought (Appendix A). Recent radial growth was 53% higher on
average in piñon that survived versus died during two episodes of drought-associated
mortality, and statistical models of mortality risk based on average growth, growth
variability, and abrupt growth changes correctly classified the status of ~70% of trees.
Climate responses and competitive interactions partly explained growth differences
between dying and surviving trees, with muted response to wet/cool conditions and
enhanced sensitivity to competition from congeners linked to growth patterns associated
with death. Discrimination and validation of models of mortality risk varied widely
across sites and drought events, indicating shifting growth-mortality relationships and
differences in mortality processes across space and time (Appendix B). Pre-formed
defense anatomy is strongly associated with piñon survivorship over a range of sites and
stand conditions. Models of mortality risk that account for both growth and resin duct
attributes had ≈1019 more support than models that contained only growth. The greatest
improvement in classification was among trees from the 2000s drought, suggesting an
enhanced role for tree defense allocation and/or bark beetle activity during recent warm
versus historic cool drought. Accounting for defense characteristics and growth-defense
allocation is likely to be important for improving representation of drought-associated
mortality (Appendix C). Piñon resin duct chronologies contain climate responses that are
coherent and distinct from those of radial growth. Growth responds positively and
strongly to previous fall and current winter precipitation, and negatively to late spring and
early summer temperature. A relatively equal positive resin duct response to winter
precipitation and positive response to mid-to-late summer drought suggests that changes
11
in climate will affect tree defense anatomy in complex ways, with the outcome
determined by seasonal changes in precipitation and temperature (Appendix D).
12
CHAPTER 1: INTRODUCTION
Statement of problem Anticipating the fate of forests under climate change is a major challenge in the
ecological, earth and social sciences. Forests cover approximately 30% of the earth’s land
surface, represent ~ 45% of terrestrial carbon stores, provide habitat for ~2/3 of the
world’s species, and provision human communities with water, food, fiber and cultural
and spiritual benefits (FAO 2014). In terms of climate moderation, forests have absorbed
approximately one third of anthropogenic carbon emissions in recent decades (Bonan
2008; Pan et al. 2011), while exerting strong control on surface conditions via alteration
of surface energy budgets (Zaitchik et al. 2006; Rotenberg & Yakir 2010; Teuling et al.
2010; Wit et al. 2014) and water cycling (Brown et al. 2005; Chapin et al. 2008).
Many of these diverse functions of forests are likely to be altered by climate
change (Millennium Ecosystem Assessment 2005; Bonan 2008; Diffenbaugh & Field
2013). Vegetation modeling in support of the International Panel on Climate Change’s
fourth (AR4) and fifth (AR5) assessment reports, for example, indicates that increases in
temperature may dampen the ameliorating role of forests within the climate system
(Friedlingstein et al. 2006, 2013; Sitch et al. 2008; Arora et al. 2013). Beyond these
generalities, however, major gaps remain in our understating of how climate change will
affect, and be affected by, forest processes. On the global scale, these gaps manifest as
large uncertainties in projections of coupled climate-vegetation models. In the AR4, these
uncertainties represented close to 40% of the total uncertainty in global temperature
projections for the 21st century (Huntingford et al. 2009).
13
Forest response to extreme events such as heat waves and droughts are a large and
potentially underestimated source of the uncertainty in assessments of climate changeforest feedbacks (Reichstein et al. 2013; Bahn et al. 2014). Global average temperature
has risen by ~0.85°C since 1850, and is projected to further rise between 0.3°C and 4.8°C
by 2100 (IPCC 2013). Superimposed on linear temperature change is an increasing
likelihood for heat waves, heat extremes and drought (Seneviratne et al. 2012). Poleward
expansion of Hadley circulation cells is likely to reduce precipitation in the subtropics
while increasing precipitation in higher latitudes (Seidel et al. 2007; Seager, Naik &
Vecchi 2010; IPCC 2013). Warming-driven changes in potential evapotranspiration and
vapor pressure deficit (VPD), meanwhile, may enhance effective aridity even in some
areas where precipitation increases, increasing the likelihood of drought in the 21st
century across many parts of the globe (Seneviratne et al. 2012; Dai 2013; Cook et al.
2014).
Tree mortality is one of the most important processes controlling long-term
dynamics in forest ecosystems. Reports of increased tree mortality due to drought, heat
and the associated activity of forest insects and pathogens (FIPs) raise the specter of
accelerated vegetation change with changes in climate. Increases in tree mortality with
drought and heat have been observed as upward trends in background mortality rates, as
was recently documented in the western US (van Mantgem et al. 2009), boreal Canada
(Peng et al. 2011), and southern Europe (Carnicer et al. 2011). Drought and heat can also
be associated with episodic canopy collapse or die-off, as in aspen forests of the Rocky
Mountains and western Canada (Worrall et al. 2008; Michaelian et al. 2011), among
14
Eucalyptus trees in Australia (Fensham, Fairfax & Ward 2009; Matusick et al. 2013), in
Austrocedrus and Nothofagus forests of temperate South America (Villalba & Veblen
1998; Suarez, Ghermandi & Kitzberger 2004), and among conifers across western North
America (Raffa et al. 2008; Bentz et al. 2009). Reports such as these now span the
forested continents, and have been escalating, suggesting that tree mortality associated
with drought and warmth must be accounted for in order to improve projections of forest
change under global warming (Allen et al. 2010).
Because of forests’ important role in regulating ecological communities and the
earth system, changes in tree mortality rates are likely to have important ecological and
biophysical consequences (Adams et al. 2010; Allen et al. 2010). Climate-associated tree
mortality may be of particular concern because it can be widespread and often occurs
rapidly (Breshears & Allen 2002; Allen et al. 2010; Breshears, López-Hoffman &
Graumlich 2011). For example drought and bark beetle-associated tree mortality across
millions of hectares of forests in western North America has altered both regional and
local fluxes of nutrients such as nitrogen and phosphorous (Edburg et al. 2012), carbon
(Huang et al. 2010; Hicke et al. 2012), water (Guardiola-Claramonte 2011; Biederman et
al. 2014) and energy (Maness, Kushner & Fung 2013; Huang & Anderegg 2014).
Perhaps most dramatically, this event has changed the status of Canada’s forests from a
carbon sink to a net source (Kurz et al. 2008) while increasing regional summer surface
temperature by around 1°C in British Colombia (Maness et al. 2013) and affecting the
carbon cycle in western USA at least as much as wildfire over the past decade (Hicke et
al. 2013).
15
Despite its importance, tree mortality is one of the least well understood aspects
of forest dynamics (Franklin, Shugart & Harmon 1987; Keane et al. 2001; McDowell et
al. 2008; Bugmann 2013). Lack of understanding, along with emerging evidence about
the sensitivity of tree mortality to drought and heat, has engendered a large increase
(350% over the last decade) in studies on the topic (McDowell et al. 2013b). Much of this
work has been experimental in nature and focused on testing hypotheses about key tree
physiological mechanisms and thresholds linking climate to tree death. The current
framework for such work focuses on two key mechanisms: hydraulic failure and carbon
starvation (McDowell et al. 2008). Hydraulic failure can occur when severe drought
causes water loss from transpiration to exceed available water supply from roots, leading
to critically high tension in the xylem water column, air embolism, loss of conductivity
and cellular desiccation (Tyree & Sperry 1989; McDowell et al. 2008). Carbon starvation
is hypothesized to occur as a consequence of trees closing stomata in order to avoid water
loss and cavitation, which limits photosynthesis and can lead to a negative tree carbon
balance. Prolonged drought may lead to death when trees deplete reserves that are
necessary to meet vital functions such as respiration or defense against biotic attacks
(McDowell et al. 2008, 2011; McDowell 2011). Subsequent advances in this framework
have emphasized the interplay between hydraulic failure and carbon limitation, and
suggest that mortality ensues following a cascade of failures in trees’ interconnected
carbon and water flux systems (Sala, Piper & Hoch 2010; McDowell 2011; McDowell et
al. 2011; Anderegg, Berry & Field 2012a). Biotic agents such as bark beetles may also
promote mortality via a combination of the two mechanisms, as trees must expend carbon
16
reserves to fight off attacks, but may ultimately succumb to mortality from loss of
conductivity associated with unchecked phloem-feeding of beetle larvae or the xylemblocking activity of fugal associates introduced by bark beetle attack (Anderegg et al.
2015).
Recent evidence supporting a role for either hydraulic failure or carbon limitation
suggests that embolism and progressive loss of conductivity (e.g. hydraulic failure) often
occur before death, but reductions in carbohydrate levels may or may not occur. This
suggests that carbon limitation may or may not play a role in drought-associated
mortality. For example, both hydraulic conductivity and non-structural carbohydrate
(NSC) concentrations declined before death during experimental drought in piñon pine
(Plaut et al. 2012; Adams et al. 2013a; Dickman et al. 2014), and NSCs have also been
observed to decline before death in Scots pine (Galiano, Martínez-Vilalta & Lloret 2011;
Poyatos et al. 2013). However hydraulic failure occurred without NSC depletion before
death in aspen (Anderegg et al. 2012b). It is not clear whether these and other contrasting
results reflect species traits (McDowell et al. 2008; Mitchell et al. 2013), or other factors,
but the discrepancies underscore how the physiological mechanisms of droughtassociated tree mortality are still a topic of active investigation and debate.
Even if instantaneous physiological mechanisms and thresholds of mortality are
identified within controlled, short-term experiments, tree mortality is notoriously
complex, and the particular reasons that trees die across landscapes are often difficult to
diagnose (Franklin et al. 1987). Multiple processes operating on different spatial and
17
temporal scales can interact to precipitate or ameliorate the seemingly sudden mortality
of long-lived trees (Franklin et al. 1987; Waring 1987; Manion 1991).
For example, interactions between the effects of climate stress on tree carbon uptake and
hydraulics, tree defenses, and climatic influences on FIPs are a common but poorly
understood aspect of widespread die-off (Raffa et al. 2008). Mortality among many
conifer species in the western US has been associated with drought and heat but also with
attack by bark beetles and the fungus pathogens they introduce to tree vascular tissue
(Raffa et al. 2008; Gaylord et al. 2013). The interacting influences of climate on tree
hydraulic functioning (McDowell et al. 2008), tree defenses (Kane & Kolb 2010;
Ferrenberg, Kane & Mitton 2014), and insect and fungus lifecycles (Bale et al. 2002; Six
& Bentz 2007; Bentz et al. 2010) may all be important for understanding patterns of tree
mortality. However, due in large part to lack of information, such interactions are not
included in ecosystem models, though recent progress has been made in developing
necessary conceptual frameworks (McDowell et al. 2013b; Dietze & Matthes 2014;
Oliva, Stenlid & Martínez‐Vilalta 2014; Anderegg et al. 2015).
A variety of site, stand and tree factors further interact with drought and insects
to affect the likelihood of mortality. Soil and atmospheric moisture gradients can dictate
mortality patterns in a relatively simple way, as was apparently the case with Juniperus
monosperma in Arizona. Here, mortality was concentrated at lower elevations, on sunny
southwestern aspects, and on more drought-prone soils (Bowker et al. 2012). However
the influence of soils, topography and forest condition played a much more variable role
in the mortality of co-occurring piñon pine: A recent review has revealed that piñon
18
mortality during the 2000s drought did not align predictably with landscape moisture
gradients, and the influence of stand attributes such as density and basal area were highly
variable (Meddens et al. 2015). This points to poorly understood interactions between
landscape moisture gradients, competition for resources between trees, and the influence
of insects and other biotic agents. Resolving the impacts of site, stand and tree conditions
on mortality is important for uncovering dominant mortality processes, and will be
essential for developing forest management interventions that promote the resilience of
forests to future increases in drought and heat (Fettig et al. 2007; Grant, Tague & Allen
2013).
Finally, the longevity of trees means that understanding long-term predisposing
factors and acclimatization processes across a range of trees and sites will be important
for fully assessing vulnerability (Lloret et al. 2012; Anderegg et al. 2012a). For example,
previous climate conditions may influence mortality risk in a variety of ways. Multiple
droughts may lead to reduced resiliency via compromises to tree hydraulic infrastructure
(Anderegg et al. 2013; Plaut et al. 2013), photosynthetic capacity and carbon reserves
(Galiano et al. 2011), with mortality depending on recovery in intervening drought-free
periods. However abundant resources and subsequently high tree growth rates early in
life may also increase later susceptibility. For example, subalpine conifers in Colorado
that exhibited rapid early growth had reduced longevities (Bigler & Veblen 2009).
Processes of acclimatization may also work to reduce the sensitivity of systems to
multiple droughts (Lloret et al. 2012), as was the case in a long-term drought experiment
of Mediterranean oak and pine. Here, high mortality rates in the first years of the
19
experiment stand in sharp contrast to low mortality and relatively high growth rates
among survivors as the experiment progressed (Barbeta, Ogaya & Peñuelas 2013).
Multiple information gaps surrounding drought-associated tree mortality
compound the problem of developing adequate representation of mortality in models of
forest dynamics and vegetation change. Many dynamic vegetation models do not include
mechanistic representations of plant mortality (or establishment) (McDowell et al. 2011).
Mortality is instead based on climate envelopes, is set to a fixed annual rate based on tree
size/age or productivity, or is not modeled explicitly at all (McDowell et al. 2011).
Another common formulation in dynamic vegetation, ecosystem demography and socalled ‘gap’ models is to represent mortality using relatively simple algorithms that relate
recent tree growth or carbon balance to the risk of death (Bugmann 2001; Keane et al.
2001; McDowell et al. 2011). Extensive recent research has been undertaken to improve
the physiological understanding of drought-associated tree mortality, in part because it
has been assumed that these simpler representations must be replaced by more
mechanistically detailed approaches. However, performance of all model types has been
uneven, and even newer models with relatively sophisticated representations of plant
water and carbon relations have been challenged to reproduce drought-associated tree
mortality and other drought-related forest trends (Sitch et al. 2008; Fisher et al. 2010;
Galbraith et al. 2010; Manusch et al. 2012; McDowell et al. 2013a; Xu et al. 2013;
Powell et al. 2013). Along with intensive data requirements for parameterizing
mechanistically-detailed models, these results highlight the importance of developing and
testing mortality representations of all types, both for use in shorter-term projections and
20
for testing hypotheses (McDowell 2011; Seidl et al. 2011; Anderegg et al. 2012a; Adams
et al. 2013b; Bugmann 2013).
In summary, recognition of both the importance of and the lack of information
about drought and heat-associated tree mortality has engendered a rapid acceleration in
research on the topic in the last decade (McDowell et al. 2013b). Although significant
progress has been made, many uncertainties and questions remain (Zeppel, Anderegg &
Adams 2013; McDowell et al. 2013b). Key information regarding plant physiological
mechanisms, long and short-term predisposing and acclimatizing factors, and interactions
between tree physiology, environmental stress, and forest insects and pathogens, is
lacking. Filling these information gaps will be essential for developing better projections
of climate-associated forest changes, and will aid in the development of management
responses to enhance forest resilience to climate change.
21
Background
Regional context
Three of the four appendices of this dissertation describe observational studies set
in piñon-juniper woodlands of New Mexico. New Mexico and the broader southwestern
United States represent an ideal natural laboratory for investigating climate-associated
forest decline and tree mortality. Recent climate trends are in line with projections for 21st
century climate, which include an overall drying (Seager et al. 2007; Cayan et al. 2010;
Seager & Vecchi 2010) superimposed on sharply rising temperature and vapor pressure
deficits (Diffenbaugh & Ashfaq 2010; IPCC 2013; Williams et al. 2013). Recent drought
conditions have been associated with increases in tree mortality among a number of
species in the region’s forests (Mueller et al. 2005; Gitlin et al. 2006; Worrall et al. 2008;
Anderegg et al. 2012b; Williams et al. 2013; Twidwell et al. 2014). Drought-stressed
trees and high temperature have also contributed to historically unprecedented bark beetle
outbreaks (Raffa et al. 2008; Bentz et al. 2009; Williams et al. 2013; Hart et al. 2014),
consistent with the interacting role of drought, heat and FIPs in documented cases of
climate-associated mortality globally (Allen et al. 2010).
In the region’s piñon-juniper woodlands, millions of hectares experienced
elevated mortality due to drought, heat and insects between 1996 and 2004, with 40-97%
mortality of piñon trees at many sites (Breshears et al. 2005; Mueller et al. 2005; Floyd et
al. 2009). Mortality of co-occurring juniper species was substantially lower, highlighting
how contrasting tree life history strategies and/or susceptibility to insects and pathogens,
may influence survival (Mueller et al. 2005; Koepke, Kolb & Adams 2010).
22
Due to the widespread nature of piñon mortality under recent, warm, ‘globalchange type drought’ (Breshears et al. 2005), and the contrasting response of cooccurring juniper, piñon-juniper woodlands have become a model system in which to
study mortality processes and impacts. On the spectrum of tree drought strategies, piñon
pine is relatively isohydric, e.g. it exhibits tight stomatal control when water stressed.
This may minimize the threat of desiccation during drought by conserving water, but the
strategy simultaneously results in limited photosynthetic uptake (McDowell et al. 2008;
Breshears et al. 2009). Juniper, in contrast, is relatively anisohydric, maintaining
photosynthesis and transpiration during drought even at the expense of increased
cavitation and the risk of desiccation (McDowell et al. 2008; Breshears et al. 2009).
Current hypotheses about the mechanisms of mortality during drought thus suggest that
piñon may be more vulnerable to carbon starvation during drought as its carbon reserves
are depleted, and that hydraulic failure may be more common in juniper. Experimental
research has both validated and challenged this view. For example, elevated temperature
was linked to increased respiration rates and reductions in time-to-mortality under
drought conditions in a piñon pine experiment, suggesting that tree carbon budgets play a
key role in mortality (Adams et al. 2009). However, results from precipitation
manipulation experiments suggest that piñon is also vulnerable to hydraulic failure (Plaut
et al. 2012; Sevanto et al. 2014), and that insects may play a more important role in piñon
versus juniper death (Gaylord et al. 2013). Thus it remains unclear clear how differential
vulnerability to carbon starvation, hydraulic failure or populations of insects (such as
bark beetles) may have caused discrepancies in mortality rates across the landscape
23
(Meddens et al. 2015). These distinctions are important because correctly identifying the
drivers and thresholds of drought-associated tree mortality in natural forests is likely to
be essential for improving modeling of vegetation dynamics under climate change.
Despite the dramatic nature of recent die-off, piñon pine has also experienced
episodic mortality in the past. Relatively recent historic drought in the 1940s and 1950s
triggered widespread mortality among piñon and other species (Potter 1957; Betancourt
et al. 1993; Allen & Breshears 1998; Swetnam & Betancourt 1998). Although both recent
2000s and historic 1950s droughts register by some metrics as among the most severe in
the last millennium (Williams et al. 2013), the 1950s drought featured greater
precipitation deficits for most seasons across much of the southwest, whereas the 2000s
drought was warmer and characterized by anomalously high vapor pressure deficit
(Breshears et al. 2005; Weiss, Castro & Overpeck 2009; Williams et al. 2013). In many
locations, piñon that died during the 1950s drought remain well preserved on the
landscape, providing an opportunity to study and compare recent tree mortality with past
mortality episodes associated with cooler droughts.
The role of dendrochronology in studies of tree mortality
Dendrochronology is an important tool in the study of tree mortality. For
example, assigning dates to the outside rings of dead trees has allowed for investigations
of the association between mortality rates and climate or other ecological events (Henry
& Swan 1974; Swetnam & Betancourt 1998; Villalba & Veblen 1998; Bigler et al. 2006).
As records of tree growth, tree rings reflect at least to some degree variability in trees’
24
carbon economy, and thus can reflect the physiological processes hypothesized to
underlie tree death (Waring & Pitman 1985; Franklin et al. 1987; Waring 1987;
Dobbertin 2005; McDowell et al. 2008; Fritts 2012). For example, lower radial growth,
along with a variety of other tree growth changes, have been associated with mortality,
consistent with carbon limitation as an important mortality mechanism (Franklin et al.
1987; Manion 1991; Pedersen 1998; Ogle, Whitham & Cobb 2000; Das et al. 2007;
McDowell, Allen & Marshall 2010). In the context of drought, direct environmental
controls on tree cambial activity and short-term growth reductions have been also been
widely documented, suggesting the possibility that low growth among trees that die
during drought reflects increased climate stress more than carbon limitation per se (Hsiao
& Acevedo 1974; Körner 2003; Vaganov, Anchukaitis & Evans 2011; Hoch 2015).
However, chronic constraints on tree gas exchange and photosynthesis– e.g., on the
ability of trees to exchange CO2 and water with the atmosphere –reduce tree carbon
uptake such that carbon available for growth is also likely to become limited (Ericsson,
Larsson & Tenow 1980; Waring & Pitman 1985; Kozlowski & Pallardy 1997; Dobbertin
2005; Bréda et al. 2006; Bansal & Germino 2008; McDowell 2011; Wiley & Helliker
2012; Hoch 2015). Either way, changes in tree growth are directly tied to trees’ water and
carbon economies, and long-term declines in particular are likely to reflect vulnerability
to carbon starvation during lethal droughts (McDowell 2011).
Tree ring-widths or basal area increments sampled from trees that die and survive
may be particularly useful for isolating short versus longer-term physiological processes
associated with tree death, for identifying environmental drivers associated with mortality
25
symptoms, and for testing and calibrating growth-based mortality algorithms for use in
models of forest dynamics. Compared to other records such as forest inventory data, treering growth records are relatively easy to obtain and contain life-long, annually resolved
information that can be sampled retrospectively in trees that die and survive, allowing for
isolation of processes important to mortality risk. Furthermore, tree-ring growth records
from dying and surviving trees can be related readily to climate time series, competitive
indices, and site factors in order to understand how environmental factors may initiate or
compound differences in physiology between living and dying trees (Bigler & Bugmann
2003; Bigler et al. 2007; McDowell et al. 2010; Linares, Camarero & Carreira 2010;
Hereş, Martínez-Vilalta & López 2012). In the context of mortality prediction, tree-ring
growth records have proven to be useful indicators of mortality risk (Wyckoff & Clark
2002), and have been used to formally test and calibrate growth-related mortality
algorithms for use in models of forest dynamics (Wyckoff & Clark 2000; Bigler &
Bugmann 2003, 2004a; b; Bigler et al. 2004; Wunder et al. 2006; Das et al. 2007).
Though they are still quite rare, tree ring chemical and anatomical records hold
great promise for clarifying additional aspects of tree physiology that lead to mortality.
The role of tree carbon investment into defense and tree internal trade-offs between
carbon allocation to growth versus defense may be investigated through study of xylem
anatomical defenses, such as resin ducts in conifers (Kane & Kolb 2010; Arbellay et al.
2014; Ferrenberg et al. 2014). Further advances may be possible by measuring
differences in tree carbon storage capacity – e.g. ray parenchyma (Olano et al. 2013), or
wood trachid and vessel characteristics related to hydraulic conductance (Levanič, Cater
26
& McDowell 2011; Hereş et al. 2014). Records of tree ring stable isotopes, particularly
carbon, may reveal the role of leaf-level photosynthetic capacity and/or stomatal
conductance (McDowell et al. 2010; Levanič et al. 2011).
It may not be possible to diagnose completely the causes of tree death using treering records, and field dendrochronological approaches cannot provide the level of
physiological detail and control of treatment effects that characterize experiments.
However tree-ring based study of tree mortality can reveal aspects of underlying
physiology and long-term predisposing factors across a much wider range of tree ages,
sizes and landscapes. Thus tree-ring field studies provide an important complement to
controlled experiments. Tree-ring studies may be particularly valuable in the context of
drought-associated tree mortality: most tree-ring studies of mortality have focused on
sporadic, individual-tree mortality, rather than on widespread mortality associated with
severe drought. Indeed differences in growth between trees that die and survive during
drought have not been widely assessed (but see Ogle et al. 2000; Suarez et al. 2004;
Bigler et al. 2006; McDowell et al. 2010), and the utility of relatively simple growthrelated metrics as predictors of drought-associated mortality has not been evaluated
thoroughly (McDowell et al. 2013a; Xu et al. 2013). Particularly lacking are observations
of mortality across a wide range of site and climatic conditions, which is essential for
understanding both variability in mortality processes and thresholds, and for determining
the processes and factors that are most robustly related to mortality risk.
Explanation of the research approach
27
A primary goal of this dissertation has been to develop tree-ring based records of
piñon pine mortality from woodlands in New Mexico and to use these records to address
gaps in our understanding of tree mortality during drought. Specifically, this dissertation
addresses the following questions:
•
How sensitive are mortality rates to drought and heat (Appendix A)?
•
What are tree physiological mechanisms that underlie tree mortality during
drought (Appendices B, C)?
•
How important is tree allocation to pre-formed defense against insects in
determining mortality risk during drought (Appendix C)?
•
How variable are mortality processes across space and time? (Appendices B, C)?
•
What is the climate sensitivity of processes that are important to droughtmortality risk (Appendices B,D)?
•
What are the advantages and limitations of current representations of tree
mortality in models of forest dynamics (Appendices A, B)?
I addressed these questions largely through the creation and analysis of a unique tree-ring
dataset containing trees that died and survived at four sites and two severe droughts – the
2000s and the 1950s droughts -- in piñon-juniper woodlands in New Mexico. Data
representing two droughts with distinct temperature and precipitation profiles afforded a
novel opportunity to compare processes that characterized mortality across a range of
climate and site conditions. First, I investigated the role of short and long-term growth as
factors associated with piñon mortality. Differences between mortality processes across
28
space and time were analyzed in order to understand how mortality during recent warm
drought may have been different from historic drought, and to test whether growthmortality relationships in piñon are stable enough to support their use in models of forest
dynamics. The role of climate variability and competition in driving growth differences
between live and dead trees helped clarify the role of stand structure and climate
sensitivity in predisposing trees to die. Second, a record of xylem resin ducts was created
in order to ascertain the role of pre-formed tree defenses and tree allocation to defense
versus growth in precipitating widespread piñon mortality. Last, I assessed the climate
sensitivity of resin duct and growth records in living trees from the dataset, in order to
understand the role of climate versus intrinsic factors in determining growth and defense
levels. This information provides a foundation for understanding how climate variability
and change may alter tree defenses, as well as revealing the potential of certain resin duct
attributes for use as climate proxies. Taken together, these analyses provide new
perspectives on the role of long-term declines versus short-term vulnerabilities,
competition, and defense in precipitating widespread mortality among different
populations of a widespread conifer.
29
CHAPTER 2: PRESENT STUDY
This dissertation is presented as a set of four research papers published or
intended for publication in peer-reviewed journals. These papers are presented as
appendices to the dissertation, and fully contain the methods, results and conclusions of
the dissertation research. All articles include co-authors, and this chapter is intended to
detail my contributions and summarize the most important aspects of each study.
Appendix A was published in the journal Forest Ecology and Management, as a
co-authored article with C.D. Allen and 18 others. I am not the study’s corresponding
author, but my ideas and efforts were central its development and publication.
Specifically, I contributed: 1) criteria for and collection of studies to be included in the
synthesis; 2) organization and summarization of included studies, including coordination
with co-authors to reach consensus on the summary table design and contents; 3) coleadership on the development of the paper’s aim, scope, organization and main
conclusions; 4) significant contribution to analysis and writing throughout the paper,
including leadership in analysis and writing of key sections.
In the remaining Appendices, I am the corresponding author. With the help of coauthors, I lead the development of the research design, executed the analyses, and wrote
the manuscripts. Appendix B was published in the journal PLoS ONE. Appendix C is
formatted for publication in the journal Oecologia. Approximately 50% of the
measurements upon which this manuscript is based were made by co-author Matthias
Kläy as part of his master’s research (which I supervised), and his thesis represented a
30
pilot for the study. Appendix D is formatted for publication in the journal
Dendrochronologia. Brief descriptions of each manuscript follow in the sections below.
A global synthesis of drought and heat-associated tree mortality
Forests play an important role in the earth system, regulating climate, maintaining
biodiversity, and provisioning human communities with water, food and fuel
(Millennium Ecosystem Assessment 2005; Bonan 2008). Forest degradation due to land
use change is widely documented, but the likely impacts of climate change on forests are
not well understood. Recently observed tree mortality associated with drought and high
temperature suggests the possibility that forests may already be responding to climate
change in some areas. We used literature review to further evaluate the idea that forests
may respond to warming and drying with increases in tree mortality. We developed
criteria to identify studies that show increases in tree mortality that are clearly tied to
drought and heat. These criteria allowed for the inclusion of studies that involved biotic
agents such as insects and pathogens, but excluded studies that involved fire, wind-throw,
or ice damage. We also measured trends in scientific publishing, reviewed current
understanding of drought-associated tree mortality at the tree physiological level, and
identified research gaps that currently limit our ability to assess and project mortality
trends.
More than 150 references document 88 cases of tree mortality tied to drought and
heat since 1970. Cases are concentrated in Europe, North America and Australia, but
examples span the forested continents and occurred in water, light and temperaturelimited forests. These cases are not necessarily attributable to climate change, and current
31
observations of forest mortality are insufficient to determine if global trends are
emerging. However, our observation that climate-associated tree mortality is happening
not only in semi-arid environments but also in temperate and rainforests suggests that
increases in temperature may be a common underlying driver (Adams et al. 2009; van
Mantgem et al. 2009). Taken together with data showing an increase in the number of
studies addressing drought and heat-associated tree mortality since 1984, we hypothesize
that forests will become increasingly vulnerable to tree mortality and die-off in the
coming decades.
Our ability to make more robust characterizations of current and future tree
mortality trends is limited by a number of data gaps and scientific uncertainties. First,
there is a clear need for global monitoring of forest mortality patterns and trends, and for
more detailed climate data to relate to data from forests. Improved understanding of the
physiological mechanisms and predisposing factors of tree mortality will be essential for
improving conceptual and process models of tree mortality under climate change.
Understanding interactions between climate, tree defenses and population dynamics of
the forests insects and pathogens is of particular concern. Finally, the implications of
climate-associated tree mortality for ecological and earth systems are largely unexplored,
and such assessments – particular in carbon-rich tropical and boreal forests - will be
particularly important for understanding likely feedbacks between terrestrial ecosystems
and climate change.
Tree-ring growth records in piñon pine and predisposition to mortality during
severe droughts
32
Leading theories of tree death suggest that mortality risk hinges at least in part on
trees’ overall carbon budget (Waring 1987; Manion 1991; Bréda et al. 2006; McDowell
et al. 2008, 2013a). Radial stem growth reflects the integrated effects of past resource
availability and environmental stress on tree carbon economy (Kozlowski & Pallardy
1997; Dobbertin 2005; Fritts 2012), and evidence of carbon limitation as an important
mechanism of mortality is found in studies that link mortality risk to low radial growth
(Waring 1987; Pedersen 1998; Ogle et al. 2000; Wyckoff & Clark 2002; Bigler et al.
2004; Das et al. 2007). Many forest and ecosystem models reflect these observations and
represent tree mortality due to climate, succession, and ‘weak pathogens’ using
algorithms that relate recent average growth rates or carbon balance to the risk of death
(Bugmann 2001; Keane et al. 2001; McDowell et al. 2011). However, universal
relationships such as those incorporated into many models are not necessarily expected
(Keane et al. 2001), and species or guild-specific growth-mortality relationships that may
better reflect mortality risk have not been widely defined. Furthermore, changes in
growth-mortality relationships with environment and ontogeny are largely unexplored,
though they may ultimately limit the applicability of empirical growth-mortality
relationships in models of long-term forest dynamics (Keane et al. 2001; Wunder 2007).
Drought-and-heat associated tree mortality has been increasingly documented in
many forests and may be an emerging phenomenon associated with climate change
(Allen et al. 2010). Yet, understanding of the physiological processes underlying
drought-associated mortality is particularly lacking (McDowell et al. 2008, 2013b;
McDowell & Sevanto 2010; Sala et al. 2010). Specifically, it is unclear whether carbon
33
limitation plays a role in the drought-associated mortality of many species (McDowell &
Sevanto 2010; Sala et al. 2010), and further, whether representations of tree mortality
based on growth history are appropriate for modeling mortality risk associated with
drought (Xu et al. 2013).
We evaluated tree-ring growth records from 263 New Mexico piñon pine trees
that died and survived two severe droughts in order to elucidate physiological and
ecological processes underlying tree death, and to assess whether growth history can be
used to predict the risk of death. The main objectives were to: 1) assess whether growth
histories of piñon pine trees that lived and died are consistent with a role for tree carbon
limitation as a mortality mechanism; 2) test whether growth-mortality relationships can
be defined that consistently reflect the risk of mortality in this system; 3) to elucidate
underlying ecological processes that predispose trees to drought-mortality by evaluating
how competition and response to climate contributed to growth differences between live
and dead trees. Pairs of trees represented mortality at four sites along a latitudinal and
mortality-severity gradient. Two of these sites included trees that died and survived
recent warm drought (2000s), and cooler, drier historical drought (1950s).
Averaged across all sites and droughts, recent radial growth was higher in piñon
that survived versus died (53% higher, p<0.05). Logistic regression models of mortality
risk correctly classified the status ~70% of trees, and included predictor variables
reflecting the influence of low average growth rate, increased year-to-year growth
variability, and the number of abrupt growth changes over years to decades. These results
are consistent with the hypothesis that carbon limitation plays an important role in
34
mortality of piñon during drought. Despite adequate discrimination overall, however,
validation of growth-mortality regressions varied across sites and drought events, with
higher growth-mortality thresholds and/or mortality risk largely decoupled from growth
history at some sites representing the 2000s drought. Weak or non-existent growth
differences between live and dead trees at these sites suggest that mortality during recent
drought resulted from a mix of long-term, chronic factors combined with either hydraulic
failure and/or sudden metabolic decline due to extreme drought, or heavy pressure from
bark beetles that overwhelmed even more vigorous trees. Shifting growth-mortality
relationships across space and time further point to variable underlying mortality factors
and thresholds, and highlight the challenges associated with calibrating growth-based
mortality algorithms for use in models of long-term vegetation dynamics.
At sites where growth history predicted tree status well, differences in growth
tended to be chronic and span multiple decades, with growth histories tending to diverge
after previous severe droughts. Differences in tree response to climate and competition
partly explained divergent growth responses, with muted response to wet/cool conditions
and an enhanced negative response to an index of conspecific competition characterizing
growth among dying trees. These results suggest that in addition chronic tree drought
stress per se, the ability to maximize resource acquisition and growth during years with
abundant water supply may be an important aspect of piñon survival during drought.
Results further highlight that studying the recovery of surviving trees after drought may
be important for furthering our understanding of drought-mortality risk.
35
Pre-formed tree defenses and the risk of mortality during drought
Our understanding of carbon and hydraulic constraints within trees that die due to
hot and dry conditions is fast improving. Symptoms of dying trees may include limited
carbon uptake (Breshears et al. 2009; McDowell et al. 2010), increased carbon
metabolism (e.g., respiration) (Adams et al. 2009; Zhao et al. 2013), hydraulic failure
(Martínez-Vilalta, Piñol & Beven 2002; Anderegg et al. 2012b), and interactions between
these processes (Plaut et al. 2012; Hartmann et al. 2013; McDowell et al. 2013a; Sevanto
et al. 2014). However the allocation of available carbon e.g., to growth, defense, storage,
and reproduction, has received relatively little attention, and the importance of tree
carbon allocation strategies for tree survival during drought have not been thoroughly
investigated (Seidl et al. 2011; McDowell et al. 2013a).
Tree allocation to defense leading up to drought and insect attack may be
particularly important in the context of recent conifer mortality in the southwestern USA.
Here, drought and insects have interacted to produce mortality in a number of species, but
as in many cases globally, it is unclear the degree to which environmental stress, tree
allocation patterns, and insect dynamics influenced mortality risk. Theories of plant
defense suggest that allocation to defense can compromise other tree functions and
influence fitness under different conditions (Herms & Mattson 1992). Using Pinus edulis
as a case study, we investigated how tree allocation to radial growth and resin ducts – a
critical component of trees’ defense anatomy - related to mortality risk during severe
droughts associated with widespread tree mortality and bark beetles. Using tree-ring
records of growth and vertical resin ducts from trees that died and survived two severe
36
droughts (1950s, 2000s) at four sites in New Mexico, USA, we addressed four questions:
1) are xylem resin duct and growth attributes significantly different in piñon that survived
vs. died? 2) Is the characterization of drought-related mortality risk improved by
considering resin duct attributes in addition to radial stem growth alone? 3) How variable
is the relationship between resin duct defenses, growth and tree survivorship across sites
and drought events? 4) Is there evidence for tradeoffs between allocation to growth and
defense in the tree-ring record of growth and vertical resin ducts?
We found that resin duct number, size, and the ratio of resin duct to xylem area
were significantly higher (p<0.05) in trees that survived, with ~58% larger resin ducts in
surviving trees. Differences in resin duct size were both larger and more consistent than
growth differences across sites and drought events. Logistic regression models that
considered both growth and defense anatomy best captured mortality risk, with good
discrimination at all sites and overall correct classification rates exceeding 80%. Resin
ducts were much more important to mortality in trees that died during the 2000s versus
1950s drought, which may reflect amplified bark beetle pressure and/or the increased
importance of defense allocation versus overall tree carbon balance during recent
drought. Growth and defense anatomy were positively correlated overall, but variations
among individual trees and across years suggest the existence of growth-defense
tradeoffs particularly at higher levels of growth. We thus suggest that accounting for
defense traits along with tree carbon allocation priorities will be important for improving
predictions of drought-associated mortality, at least among the genus Pinus.
37
Climatic sensitivity of vertical resin duct formation in piñon pine
Resin ducts are an important component of conifer’s system of defense against
insects, pathogens and mechanical injury. We establish in Appendix C that low levels of
investment into xylem resin ducts relates strongly to the risk of Pinus edulis mortality
during drought and bark beetle attack. This pattern has now also been documented in P.
ponderosa, P. contorta and P. flexilus (Kane & Kolb 2010; Ferrenberg et al. 2014).
Although apparently critical for survival during drought, tree internal and climatic
controls on the formation of anatomical defenses are not well understood.
Dendroclimatological approaches to understanding the sensitivity of resin ducts to
climate have been carried out in a few species, with a response to summer climate found
for Picea abies and Pinus sylvestris (Wimmer & Grabner 1997; Rigling et al. 2003), but
no response found for Pinus nigra (Levanič 1999) or P. leicodermis (Todaro et al. 2007).
To our knowledge, no dendroclimatological investigation of anatomical defenses has
been carried out for any conifer species in western North America. Thus knowledge is
limited about how conifer defense against insects may vary with climate and tree
ontogeny.
We used a dendroclimatological approach to investigate how attributes of vertical
xylem resin ducts in piñon pine are related to tree intrinsic versus climatological
variables. Resin duct chronologies spanning two distinct ~30 year intervals during the
period from 1925-2010, and reflecting resin duct number, size, density and relative area
(% of xylem area containing resin duct tissue) were developed from living trees at four
sites in New Mexico. We documented changes in resin duct attributes with tree size and
38
age, as well as statistical properties and relationships between standardized ring-width
and resin-duct chronologies. We used correlation, partial correlation, and multiple
regressions to explore the sensitivity of resin duct attributes and radial growth to monthly
and seasonal climate.
Resin duct attributes changed variably with tree age and size, with a tendency for
resin duct number to decrease precipitously with cambial age, and for most resin duct
attributes to increase slightly with tree diameter. Resin duct chronologies exhibited higher
levels of intrinsic variability and lower inter-series correlations compared to ring width
chronologies, and resin duct size chronologies were particularly noisy. This reflects the
greater dominance of tree internal versus climate control on defense anatomy, and may be
related to greater genetic control on resin ducts versus growth (Rosner & Hannrup 2004).
Despite low signal strength (especially when compared to ring width), resin duct
chronologies contain a coherent response to climate that is distinct from the response of
radial growth. Similar to the response of conifer ring-widths through out New Mexico
and Arizona (Williams et al. 2013), our piñon ring-width chronologies responded
positively to previous fall, winter and current spring precipitation, and negatively to late
spring and early summer temperature. In contrast, resin duct number, density, size and
relative area exhibited positive relationships with previous year winter and early spring
precipitation. Resin duct number, density, and relative area further exhibited a positive
response to mid-late summer precipitation and drought. These responses persisted even
after the statistical influence of ring width on resin duct formation was removed,
indicating resin duct sensitivity to climate above and beyond the influence of climate on
39
radial growth. In multiple step-wise regressions, seasonal climate variables explained a
greater portion of the variability in ring width chronologies (R2=0.584) than in resin duct
chronologies (R2 values range from 0.331 – 0.519), but all regressions were highly
significant.
These findings have implications for understanding tree defenses and
vulnerability to drought and bark beetle outbreaks with climate variability and change.
Recent southwestern drought may be an early symptom of climate change (Seager et al.
2007; Williams et al. 2013; Garfin et al. 2014), and the region is likely to become even
hotter and drier as the 21st century progresses (Cayan et al. 2010; Garfin et al. 2014).
Anticipated increases in temperature during late summer, along with any increases in
summer drought, could enhance some aspects tree defenses in P. edulis. However, two
factors are likely to negate any benefit: 1) copious winter precipitation – which is very
likely to decline under climate change - appears to be equally important for resin duct
formation as summer drought; and, 2) the attribute of xylem defense anatomy that
appears most important for survivorship in P. edulis – resin duct size (Macalady et al. in
preparation) - is least well-controlled by climate, with the relatively weak response
characterized by a positive relationship with winter and spring precipitation.
Our findings also indicate that the development of additional resin duct
chronologies may help improve climate reconstructions. Resin duct chronologies contain
a distinct climate response especially in summer and early autumn, suggesting that they
have the potential to bolster reconstructions especially of summer climate in the lowland
areas where piñon pine is widely distributed.
40
REFERENCES
Adams, H.D., Germino, M.J., Breshears, D.D., Barron-Gafford, G.A., GuardiolaClaramonte, M., Zou, C.B. & Huxman, T.E. (2013a) Nonstructural leaf
carbohydrate dynamics of Pinus edulis during drought-induced tree mortality
reveal role for carbon metabolism in mortality mechanism. New Phytologist, 197,
1142–1151.
Adams, H.D., Guardiola-Claramonte, M., Barron-Gafford, G.A., Villegas, J.C.,
Breshears, D.D., Zou, C.B., Troch, P.A. & Huxman, T.E. (2009) Temperature
sensitivity of drought-induced tree mortality portends increased regional die-off
under global-change-type drought. Proceedings of the National Academy of
Sciences, 106, 7063–7066.
Adams, H.D., Macalady, A.K., Breshears, D.D., Allen, C.D., Stephenson, N.L., Saleska,
S.R., Huxman, T.E. & McDowell, N.G. (2010) Climate-Induced Tree Mortality:
Earth System Consequences. Eos, Transactions, American Geophysical Union,
91, 153–154.
Adams, H.D., Williams, A.P., Xu, C., Rauscher, S.A., Jiang, X. & McDowell, N.G.
(2013b) Empirical and process-based approaches to climate-induced forest
mortality models. Frontiers in Plant Science, 4.
Allen, C.D. & Breshears, D.D. (1998) Drought-induced shift of a forest–woodland
ecotone: rapid landscape response to climate variation. Proceedings of the
National Academy of Sciences, 95, 14839–14842.
Allen, C.D., Macalady, A.K., Chenchouni, H., Bachelet, D., McDowell, N., Vennetier,
M., Kitzberger, T., Rigling, A., Breshears, D.D., Hogg, E.H. (Ted), Gonzalez, P.,
Fensham, R., Zhang, Z., Castro, J., Demidova, N., Lim, J.-H., Allard, G.,
Running, S.W., Semerci, A. & Cobb, N. (2010) A global overview of drought and
heat-induced tree mortality reveals emerging climate change risks for forests.
Forest Ecology and Management, 259, 660–684.
Anderegg, W.R.L., Berry, J.A. & Field, C.B. (2012a) Linking definitions, mechanisms,
and modeling of drought-induced tree death. Trends in Plant Science, 17, 693–
700.
Anderegg, W.R., Berry, J.A., Smith, D.D., Sperry, J.S., Anderegg, L.D. & Field, C.B.
(2012b) The roles of hydraulic and carbon stress in a widespread climate-induced
forest die-off. Proceedings of the National Academy of Sciences, 109, 233–237.
Anderegg, W.R.L., Hicke, J.A., Fisher, R.A., Allen, C.D., Aukema, J., Bentz, B., Hood,
S., Lichstein, J.W., Macalady, A.K., McDowell, N., Pan, Y., Raffa, K., Sala, A.,
41
Shaw, J.D., Stephenson, N.L., Tague, C. & Zeppel, M. (2015) Tree mortality
from drought, insects, and their interactions in a changing climate. New
Phytologist.
Anderegg, W.R.L., Plavcová, L., Anderegg, L.D.L., Hacke, U.G., Berry, J.A. & Field,
C.B. (2013) Drought’s legacy: multiyear hydraulic deterioration underlies
widespread aspen forest die-off and portends increased future risk. Global
Change Biology, 19, 1188–1196.
Arbellay, E., Stoffel, M., Sutherland, E.K., Smith, K.T. & Falk, D.A. (2014) Resin duct
size and density as ecophysiological traits in fire scars of Pseudotsuga menziesii
and Larix occidentalis. Annals of Botany, 114, 973–980.
Arora, V.K., Boer, G.J., Friedlingstein, P., Eby, M., Jones, C.D., Christian, J.R., Bonan,
G., Bopp, L., Brovkin, V., Cadule, P., Hajima, T., Ilyina, T., Lindsay, K.,
Tjiputra, J.F. & Wu, T. (2013) Carbon–Concentration and Carbon–Climate
Feedbacks in CMIP5 Earth System Models. Journal of Climate, 26, 5289–5314.
Bahn, M., Reichstein, M., Dukes, J.S., Smith, M.D. & McDowell, N.G. (2014) Climate–
biosphere interactions in a more extreme world. New Phytologist, 202, 356–359.
Bale, J.S., Masters, G.J., Hodkinson, I.D., Awmack, C., Bezemer, T.M., Brown, V.K.,
Butterfield, J., Buse, A., Coulson, J.C., Farrar, J., Good, J.E.G., Harrington, R.,
Hartley, S., Jones, T.H., Lindroth, R.L., Press, M.C., Symrnioudis, I., Watt, A.D.
& Whittaker, J.B. (2002) Herbivory in global climate change research: direct
effects of rising temperature on insect herbivores. Global Change Biology, 8, 1–
16.
Bansal, S. & Germino, M.J. (2008) Carbon balance of conifer seedlings at timberline:
relative changes in uptake, storage, and utilization. Oecologia, 158, 217–227.
Barbeta, A., Ogaya, R. & Peñuelas, J. (2013) Dampening effects of long-term
experimental drought on growth and mortality rates of a Holm oak forest. Global
Change Biology, 19, 3133–3144.
Bentz, B., Logan, J., MacMahon, J., Allen, C.D., Ayres, M., Berg, E., Carroll, A.,
Hansen, M., Hicke, J. & Joyce, L. (2009) Bark beetle outbreaks in western North
America: causes and consequences. p. 42. University of Utah Press, Salt Lake
City, UT.
Bentz, B.J., Régnière, J., Fettig, C.J., Hansen, E.M., Hayes, J.L., Hicke, J.A., Kelsey,
R.G., Negrón, J.F. & Seybold, S.J. (2010) Climate change and bark beetles of the
western United States and Canada: direct and indirect effects. BioScience, 60,
602–613.
42
Betancourt, J.L., Pierson, E.A., Rylander, K.A., Fairchild-Parks, J.A. & Dean, J.S. (1993)
Influence of history and climate on New Mexico piñon-juniper woodlands.
General Technical Report RM-236. pp. 42–62. US Department of Agriculture,
Forest Service, Rocky Mountain Forest and Range Experiment Station, Fort
Collins, CO.
Biederman, J.A., Harpold, A.A., Gochis, D.J., Ewers, B.E., Reed, D.E., Papuga, S.A. &
Brooks, P.D. (2014) Increased evaporation following widespread tree mortality
limits streamflow response. Water Resources Research, 50, 5395–5409.
Bigler, C., Bräker, O.U., Bugmann, H., Dobbertin, M. & Rigling, A. (2006) Drought as
an inciting mortality factor in Scots pine stands of the Valais, Switzerland.
Ecosystems, 9, 330–343.
Bigler, C. & Bugmann, H. (2003) Growth-dependent tree mortality models based on tree
rings. Canadian Journal of Forest Research, 33, 210–221.
Bigler, C. & Bugmann, H. (2004a) Assessing the performance of theoretical and
empirical tree mortality models using tree-ring series of Norway spruce.
Ecological Modelling, 174, 225–239.
Bigler, C. & Bugmann, H. (2004b) Predicting the time of tree death using
dendrochronological data. Ecological Applications, 14, 902–914.
Bigler, C., Gavin, D.G., Gunning, C. & Veblen, T.T. (2007) Drought induces lagged tree
mortality in a subalpine forest in the Rocky Mountains. Oikos, 116, 1983–1994.
Bigler, C., Gričar, J., Bugmann, H. & Čufar, K. (2004) Growth patterns as indicators of
impending tree death in silver fir. Forest Ecology and Management, 199, 183–
190.
Bigler, C. & Veblen, T.T. (2009) Increased early growth rates decrease longevities of
conifers in subalpine forests. Oikos, 118, 1130–1138.
Bonan, G.B. (2008) Forests and Climate Change: Forcings, Feedbacks, and the Climate
Benefits of Forests. Science, 320, 1444–1449.
Bowker, M.A., Muñoz, A., Martinez, T. & Lau, M.K. (2012) Rare drought-induced
mortality of juniper is enhanced by edaphic stressors and influenced by stand
density. Journal of Arid Environments, 76, 9–16.
Bréda, N., Huc, R., Granier, A. & Dreyer, E. (2006) Temperate forest trees and stands
under severe drought: a review of ecophysiological responses, adaptation
processes and long-term consequences. Annals of Forest Science, 63, 625–644.
43
Breshears, D.D. & Allen, C.D. (2002) The importance of rapid, disturbance-induced
losses in carbon management and sequestration. Global Ecology and
Biogeography, 11, 1–5.
Breshears, D.D., Cobb, N.S., Rich, P.M., Price, K.P., Allen, C.D., Balice, R.G., Romme,
W.H., Kastens, J.H., Floyd, M.L., Belnap, J. & others. (2005) Regional vegetation
die-off in response to global-change-type drought. Proceedings of the National
Academy of Sciences, 102, 15144–15148.
Breshears, D.D., López-Hoffman, L. & Graumlich, L.J. (2011) When ecosystem services
crash: preparing for big, fast, patchy climate change. Ambio, 40, 256–263.
Breshears, D.D., Myers, O.B., Meyer, C.W., Barnes, F.J., Zou, C.B., Allen, C.D.,
McDowell, N.G. & Pockman, W.T. (2009) Tree die-off in response to global
change-type drought: mortality insights from a decade of plant water potential
measurements. Frontiers in Ecology and the Environment, 7, 185–189.
Brown, A.E., Zhang, L., McMahon, T.A., Western, A.W. & Vertessy, R.A. (2005) A
review of paired catchment studies for determining changes in water yield
resulting from alterations in vegetation. Journal of Hydrology, 310, 28–61.
Bugmann, H. (2001) A review of forest gap models. Climatic Change, 51, 259–305.
Bugmann, H. (2013) Forests in a greenhouse atmosphere: predicting the unpredictable?
Forests and Global Change Ecological Reviews (eds D.A. Coomes, D.F.R.P.
Burslem & W.D. Simonson), pp. 359–380. Cambridge University Press,
Cambridge, UK.
Carnicer, J., Coll, M., Ninyerola, M., Pons, X., Sánchez, G. & Peñuelas, J. (2011)
Widespread crown condition decline, food web disruption, and amplified tree
mortality with increased climate change-type drought. Proceedings of the
National Academy of Sciences, 108, 1474–1478.
Cayan, D.R., Das, T., Pierce, D.W., Barnett, T.P., Tyree, M. & Gershunov, A. (2010)
Future dryness in the southwest US and the hydrology of the early 21st century
drought. Proceedings of the National Academy of Sciences, 107, 21271–21276.
Chapin, F.S., Randerson, J.T., McGuire, A.D., Foley, J.A. & Field, C.B. (2008) Changing
feedbacks in the climate-biosphere system. Frontiers in Ecology and the
Environment, 6, 313–320.
Cook, B.I., Smerdon, J.E., Seager, R. & Coats, S. (2014) Global warming and 21st
century drying. Climate Dynamics, 43, 2607–2627.
Dai, A. (2013) Increasing drought under global warming in observations and models.
Nature Climate Change, 3, 52–58.
44
Das, A.J., Battles, J.J., Stephenson, N.L. & Van Mantgem, P.J. (2007) The relationship
between tree growth patterns and likelihood of mortality: a study of two tree
species in the Sierra Nevada. Canadian Journal of Forest Research, 37, 580–597.
Dickman, L.T., Mcdowell, N.G., Sevanto, S., Pangle, R.E. & Pockman, W.T. (2014)
Carbohydrate dynamics and mortality in a piñon-juniper woodland under three
future precipitation scenarios. Plant, Cell & Environment, 38, 729–739.
Dietze, M.C. & Matthes, J.H. (2014) A general ecophysiological framework for
modelling the impact of pests and pathogens on forest ecosystems. Ecology
Letters, 17, 1418–1426.
Diffenbaugh, N.S. & Ashfaq, M. (2010) Intensification of hot extremes in the United
States. Geophysical Research Letters, 37, 15701.
Diffenbaugh, N.S. & Field, C.B. (2013) Changes in Ecologically Critical Terrestrial
Climate Conditions. Science, 341, 486–492.
Dobbertin, M. (2005) Tree growth as indicator of tree vitality and of tree reaction to
environmental stress: a review. European Journal of Forest Research, 124, 319–
333.
Edburg, S.L., Hicke, J.A., Brooks, P.D., Pendall, E.G., Ewers, B.E., Norton, U., Gochis,
D., Gutmann, E.D. & Meddens, A.J. (2012) Cascading impacts of bark beetlecaused tree mortality on coupled biogeophysical and biogeochemical processes.
Frontiers in Ecology and the Environment, 10, 416–424.
Ericsson, A., Larsson, S. & Tenow, O. (1980) Effects of early and late season defoliation
on growth and carbohydrate dynamics in Scots pine. Journal of applied Ecology,
17, 747–769.
FAO. (2014) State of the World’s Forests 2014. Food and Agricultural Organization of
the United Nations, Rome, Italy.
Fensham, R.J., Fairfax, R.J. & Ward, D.P. (2009) Drought-induced tree death in savanna.
Global Change Biology, 15, 380–387.
Ferrenberg, S., Kane, J.M. & Mitton, J.B. (2014) Resin duct characteristics associated
with tree resistance to bark beetles across lodgepole and limber pines. Oecologia,
174, 1283–1292.
Fettig, C.J., Klepzig, K.D., Billings, R.F., Munson, A.S., Nebeker, T.E., Negrón, J.F. &
Nowak, J.T. (2007) The effectiveness of vegetation management practices for
prevention and control of bark beetle infestations in coniferous forests of the
western and southern United States. Forest Ecology and Management, 238, 24–
53.
45
Fisher, R., McDowell, N., Purves, D., Moorcroft, P., Sitch, S., Cox, P., Huntingford, C.,
Meir, P. & Ian Woodward, F. (2010) Assessing uncertainties in a secondgeneration dynamic vegetation model caused by ecological scale limitations. New
Phytologist, 187, 666–681.
Floyd, M.L., Clifford, M., Cobb, N.S., Hanna, D., Delph, R., Ford, P. & Turner, D.
(2009) Relationship of stand characteristics to drought-induced mortality in three
Southwestern piñon-juniper woodlands. Ecological Applications, 19, 1223–1230.
Franklin, J.F., Shugart, H.H. & Harmon, M.E. (1987) Tree Death as an Ecological
Process. BioScience, 37, 550–556.
Friedlingstein, P., Cox, P., Betts, R., Bopp, L., Bloh, W. von, Brovkin, V., Cadule, P.,
Doney, S., Eby, M., Fung, I., Bala, G., John, J., Jones, C., Joos, F., Kato, T.,
Kawamiya, M., Knorr, W., Lindsay, K., Matthews, H.D., Raddatz, T., Rayner, P.,
Reick, C., Roeckner, E., Schnitzler, K.-G., Schnur, R., Strassmann, K., Weaver,
A.J., Yoshikawa, C. & Zeng, N. (2006) Climate–Carbon Cycle Feedback
Analysis: Results from the C4MIP Model Intercomparison. Journal of Climate,
19, 3337–3353.
Friedlingstein, P., Meinshausen, M., Arora, V.K., Jones, C.D., Anav, A., Liddicoat, S.K.
& Knutti, R. (2013) Uncertainties in CMIP5 Climate Projections due to Carbon
Cycle Feedbacks. Journal of Climate, 27, 511–526.
Fritts, H.C. (2012) Tree Rings and Climate. Elsevier.
Galbraith, D., Levy, P.E., Sitch, S., Huntingford, C., Cox, P., Williams, M. & Meir, P.
(2010) Multiple mechanisms of Amazonian forest biomass losses in three
dynamic global vegetation models under climate change. New Phytologist, 187,
647–665.
Galiano, L., Martínez-Vilalta, J. & Lloret, F. (2011) Carbon reserves and canopy
defoliation determine the recovery of Scots pine 4 yr after a drought episode. New
Phytologist, 190, 750–759.
Garfin, G., Blanco, H., Comrie, A., Piechota, T. & Waskom, R. (2014) Ch. 20:
Southwest. Climate Change Impacts in the United States: The Third National
Climate Assessment pp. 462–486. U.S. Global Change Research Program.
Gaylord, M.L., Kolb, T.E., Pockman, W.T., Plaut, J.A., Yepez, E.A., Macalady, A.K.,
Pangle, R.E. & McDowell, N.G. (2013) Drought predisposes piñon-juniper
woodlands to insect attacks and mortality. New Phytologist, 198, 567–578.
Gitlin, A.R., Sthultz, C.M., Bowker, M.A., Stumpf, S., Paxton, K.L., Kennedy, K.,
Munoz, A., Bailey, J.K. & Whitham, T.G. (2006) Mortality gradients within and
46
among dominant plant populations as barometers of ecosystem change during
extreme drought. Conservation Biology, 20, 1477–1486.
Grant, G.E., Tague, C.L. & Allen, C.D. (2013) Watering the forest for the trees: an
emerging priority for managing water in forest landscapes. Frontiers in Ecology
and the Environment, 11, 314–321.
Guardiola-Claramonte, M. (2011) Decreased streamflow in semi-arid basins following
drought-induced tree die-off: A counter-intuitive and indirect climate impact on
hydrology. Journal of Hydrology, 406, 225–233.
Hartmann, H., Ziegler, W., Kolle, O. & Trumbore, S. (2013) Thirst beats hunger –
declining hydration during drought prevents carbon starvation in Norway spruce
saplings. New Phytologist, 200, 340–349.
Hart, S.J., Veblen, T.T., Eisenhart, K.S., Jarvis, D. & Kulakowski, D. (2014) Drought
induces spruce beetle (Dendroctonus rufipennis) outbreaks across northwestern
Colorado. Ecology, 95, 930–939.
Henry, J.D. & Swan, J.M.A. (1974) Reconstructing forest history from live and dead
plant material--an approach to the study of forest succession in southwest New
Hampshire. Ecology, 55, 772–783.
Hereş, A.-M., Camarero, J.J., López, B.C. & Martínez-Vilalta, J. (2014) Declining
hydraulic performances and low carbon investments in tree rings predate Scots
pine drought-induced mortality. Trees, 28, 1737–1750.
Hereş, A.-M., Martínez-Vilalta, J. & López, B.C. (2012) Growth patterns in relation to
drought-induced mortality at two Scots pine (Pinus sylvestris L.) sites in NE
Iberian Peninsula. Trees, 26, 621–630.
Herms, D.A. & Mattson, W.J. (1992) The Dilemma of Plants: To Grow or Defend. The
Quarterly Review of Biology, 67, 283–335.
Hicke, J.A., Allen, C.D., Desai, A.R., Dietze, M.C., Hall, R.J., (Ted) Hogg, E.H.,
Kashian, D.M., Moore, D., Raffa, K.F., Sturrock, R.N. & Vogelmann, J. (2012)
Effects of biotic disturbances on forest carbon cycling in the United States and
Canada. Global Change Biology, 18, 7–34.
Hicke, J.A., Meddens, A.J., Allen, C.D. & Kolden, C.A. (2013) Carbon stocks of trees
killed by bark beetles and wildfire in the western United States. Environmental
Research Letters, 8, 035032.
Hoch, G. (2015) Carbon Reserves as Indicators for Carbon Limitation in Trees. Progress
in Botany pp. 321–346. Springer.
47
Hsiao, T.C. & Acevedo, E. (1974) Plant responses to water deficits, water-use efficiency,
and drought resistance. Agricultural Meteorology, 14, 59–84.
Huang, C. & Anderegg, W.R. (2014) Vegetation, land surface brightness, and
temperature dynamics after aspen forest die‐off. Journal of Geophysical
Research: Biogeosciences, 119, 1297–1308.
Huang, C., Asner, G.P., Barger, N.N., Neff, J.C. & Floyd, M.L. (2010) Regional
aboveground live carbon losses due to drought-induced tree dieback in piñon–
juniper ecosystems. Remote Sensing of Environment, 114, 1471–1479.
Huntingford, C., Lowe, J.A., Booth, B.B.B., Jones, C.D., Harris, G.R., Gohar, L.K. &
Meir, P. (2009) Contributions of carbon cycle uncertainty to future climate
projection spread. Tellus B, 61, 355–360.
IPCC (2013) Climate Change 2013: The Physical Science Basis. Working Group I
Contribution to the IPCC Fifth Assessment Report (AR5) (eds T.F. Stocker, D.
Qin, G.-K. Plattner, M. Tignor, S.K. Allen, J. Boschung, A. Nauels, Y. Xia, V.
Bex, & P.M. Midgley), Cambridge Univ Press, Cambridge, UK and New York,
NY, USA.
Kane, J.M. & Kolb, T.E. (2010) Importance of resin ducts in reducing ponderosa pine
mortality from bark beetle attack. Oecologia, 164, 601–609.
Keane, R.E., Austin, M., Field, C., Huth, A., Lexer, M.J., Peters, D., Solomon, A. &
Wyckoff, P. (2001) Tree mortality in gap models: application to climate change.
Climatic Change, 51, 509–540.
Koepke, D.F., Kolb, T.E. & Adams, H.D. (2010) Variation in woody plant mortality and
dieback from severe drought among soils, plant groups, and species within a
northern Arizona ecotone. Oecologia, 163, 1079–1090.
Körner, C. (2003) Carbon limitation in trees. Journal of ecology, 91, 4–17.
Kozlowski, T.T. & Pallardy, S.G. (1997) Growth Control in Woody Plants. Elsevier.
Kurz, W.A., Stinson, G., Rampley, G.J., Dymond, C.C. & Neilson, E.T. (2008) Risk of
natural disturbances makes future contribution of Canada’s forests to the global
carbon cycle highly uncerain. Proceedings of the National Academy of Sciences of
the United States of America, 105, 1551–1555.
Levanič, T. (1999) Vertical resin ducts in wood of black pine (Pinus nigra Arnold) as a
possible dendroecological variable. Phyton [Austria], 39, 123–127.
48
Levanič, T., Cater, M. & McDowell, N.G. (2011) Associations between growth, wood
anatomy, carbon isotope discrimination and mortality in a Quercus robur forest.
Tree physiology, 31, 298–308.
Linares, J.C., Camarero, J.J. & Carreira, J.A. (2010) Competition modulates the
adaptation capacity of forests to climatic stress: insights from recent growth
decline and death in relict stands of the Mediterranean fir Abies pinsapo. Journal
of Ecology, 98, 592–603.
Lloret, F., Escudero, A., Iriondo, J.M., Martínez‐Vilalta, J. & Valladares, F. (2012)
Extreme climatic events and vegetation: the role of stabilizing processes. Global
Change Biology, 18, 797–805.
Macalady, A.K., Kläy, M., Bugman, H., Gaylord, M.L., Allen, C.D., Swetnam, T.W. &
McDowell, N.G. (in preparation) Mortality risk of an aridlands conifer during
severe drought depends on radial growth and investment into defense.
Maness, H., Kushner, P.J. & Fung, I. (2013) Summertime climate response to mountain
pine beetle disturbance in British Columbia. Nature Geoscience, 6, 65–70.
Manion, P.D. (1991) Tree Disease Concepts, 2nd ed. Prentice Hall.
van Mantgem, P.J., Stephenson, N.L., Byrne, J.C., Daniels, L.D., Franklin, J.F., Fulé,
P.Z., Harmon, M.E., Larson, A.J., Smith, J.M., Taylor, A.H. & Veblen, T.T.
(2009) Widespread Increase of Tree Mortality Rates in the Western United States.
Science, 323, 521–524.
Manusch, C., Bugmann, H., Heiri, C. & Wolf, A. (2012) Tree mortality in dynamic
vegetation models – A key feature for accurately simulating forest properties.
Ecological Modelling, 243, 101–111.
Martínez-Vilalta, J., Piñol, J. & Beven, K. (2002) A hydraulic model to predict droughtinduced mortality in woody plants: an application to climate change in the
Mediterranean. Ecological Modelling, 155, 127–147.
Matusick, G., Ruthrof, K.X., Brouwers, N.C., Dell, B. & Hardy, G.S.J. (2013) Sudden
forest canopy collapse corresponding with extreme drought and heat in a
mediterranean-type eucalypt forest in southwestern Australia. European Journal
of Forest Research, 132, 497–510.
McDowell, N.G. (2011) Mechanisms linking drought, hydraulics, carbon metabolism,
and vegetation mortality. Plant Physiology, 155, 1051–1059.
McDowell, N., Allen, C.D. & Marshall, L. (2010) Growth, carbon-isotope
discrimination, and drought-associated mortality across a Pinus ponderosa
elevational transect. Global Change Biology, 16, 399–415.
49
McDowell, N.G., Beerling, D.J., Breshears, D.D., Fisher, R.A., Raffa, K.F. & Stitt, M.
(2011) The interdependence of mechanisms underlying climate-driven vegetation
mortality. Trends in Ecology & Evolution, 26, 523–532.
McDowell, N.G., Fisher, R.A., Xu, C., Domec, J.C., Hölttä, T., Mackay, D.S., Sperry,
J.S., Boutz, A., Dickman, L. & Gehres, N. (2013a) Evaluating theories of
drought-induced vegetation mortality using a multimodel–experiment framework.
New Phytologist, 200, 304–321.
McDowell, N., Pockman, W.T., Allen, C.D., Breshears, D.D., Cobb, N., Kolb, T., Plaut,
J., Sperry, J., West, A., Williams, D.G. & others. (2008) Mechanisms of plant
survival and mortality during drought: why do some plants survive while others
succumb to drought? New Phytologist, 178, 719–739.
McDowell, N.G., Ryan, M.G., Zeppel, M.J.B. & Tissue, D.T. (2013b) Feature:
Improving our knowledge of drought-induced forest mortality through
experiments, observations, and modeling. New Phytologist, 200, 289–293.
McDowell, N.G. & Sevanto, S. (2010) The mechanisms of carbon starvation: how, when,
or does it even occur at all? New Phytologist, 186, 264–266.
Meddens, A.J., Hicke, J.A., Macalady, A.K., Buotte, P.C., Cowles, T.R. & Allen, C.D.
(2015) Patterns and causes of observed piñon pine mortality in the southwestern
United States. New Phytologist, 206, 91–97.
Michaelian, M., Hogg, E.H., Hall, R.J. & Arsenault, E. (2011) Massive mortality of
aspen following severe drought along the southern edge of the Canadian boreal
forest. Global Change Biology, 17, 2084–2094.
Millennium Ecosystem Assessment. (2005) Ecosystems and Human Well-Being:
Synthesis. Island Press, Washington, DC.
Mitchell, P.J., O’Grady, A.P., Tissue, D.T., White, D.A., Ottenschlaeger, M.L. &
Pinkard, E.A. (2013) Drought response strategies define the relative contributions
of hydraulic dysfunction and carbohydrate depletion during tree mortality. New
Phytologist, 197, 862–872.
Mueller, R.C., Scudder, C.M., Porter, M.E., Trotter, T., R., Gehring, C.A. &
Whitham, T.G. (2005) Differential tree mortality in response to severe drought:
evidence for long‐term vegetation shifts. Journal of Ecology, 93, 1085–1093.
Ogle, K., Whitham, T.G. & Cobb, N.S. (2000) Tree-ring variation in pinyon predicts
likelihood of death following severe drought. Ecology, 81, 3237–3243.
50
Olano, J.M., Arzac, A., García-Cervigón, A.I., von Arx, G. & Rozas, V. (2013) New star
on the stage: amount of ray parenchyma in tree rings shows a link to climate. New
Phytologist, 198, 486–495.
Oliva, J., Stenlid, J. & Martínez‐Vilalta, J. (2014) The effect of fungal pathogens on the
water and carbon economy of trees: implications for drought‐induced mortality.
New Phytologist, 203, 1028–1035.
Pan, Y., Birdsey, R.A., Fang, J., Houghton, R., Kauppi, P.E., Kurz, W.A., Phillips, O.L.,
Shvidenko, A., Lewis, S.L., Canadell, J.G., Ciais, P., Jackson, R.B., Pacala, S.W.,
McGuire, A.D., Piao, S., Rautiainen, A., Sitch, S. & Hayes, D. (2011) A Large
and Persistent Carbon Sink in the World’s Forests. Science, 333, 988–993.
Pedersen, B.S. (1998) The role of stress in the mortality of midwestern oaks as indicated
by growth prior to death. Ecology, 79, 79–93.
Peng, C., Ma, Z., Lei, X., Zhu, Q., Chen, H., Wang, W., Liu, S., Li, W., Fang, X. &
Zhou, X. (2011) A drought-induced pervasive increase in tree mortality across
Canada’s boreal forests. Nature Climate Change, 1, 467–471.
Plaut, J.A., Wadsworth, W.D., Pangle, R., Yepez, E.A., McDowell, N.G. & Pockman,
W.T. (2013) Reduced transpiration response to precipitation pulses precedes
mortality in a pinon–juniper woodland subject to prolonged drought. New
Phytologist, 200, 375–387.
Plaut, J.A., Yepez, E.A., Hill, J., Pangle, R., Sperry, J.S., Pockman, W.T. & McDowell,
N.G. (2012) Hydraulic limits preceding mortality in a piñon–juniper woodland
under experimental drought. Plant, Cell & Environment, 35, 1601–1617.
Potter, L.D. (1957) Phytosociological study of San Augustin Plains, New Mexico.
Ecological Monographs, 114–136.
Powell, T.L., Galbraith, D.R., Christoffersen, B.O., Harper, A., Imbuzeiro, H.M.A.,
Rowland, L., Almeida, S., Brando, P.M., da Costa, A.C.L., Costa, M.H., Levine,
N.M., Malhi, Y., Saleska, S.R., Sotta, E., Williams, M., Meir, P. & Moorcroft,
P.R. (2013) Confronting model predictions of carbon fluxes with measurements
of Amazon forests subjected to experimental drought. New Phytologist, 200, 350–
365.
Poyatos, R., Aguadé, D., Galiano, L., Mencuccini, M. & Martínez-Vilalta, J. (2013)
Drought-induced defoliation and long periods of near-zero gas exchange play a
key role in accentuating metabolic decline of Scots pine. New Phytologist, 200,
388–401.
51
Raffa, K.F., Aukema, B.H., Bentz, B.J., Carroll, A.L., Hicke, J.A., Turner, M.G. &
Romme, W.H. (2008) Cross-scale Drivers of Natural Disturbances Prone to
Anthropogenic Amplification: The Dynamics of Bark Beetle Eruptions.
BioScience, 58, 501–517.
Reichstein, M., Bahn, M., Ciais, P., Frank, D., Mahecha, M.D., Seneviratne, S.I.,
Zscheischler, J., Beer, C., Buchmann, N., Frank, D.C., Papale, D., Rammig, A.,
Smith, P., Thonicke, K., van der Velde, M., Vicca, S., Walz, A. & Wattenbach,
M. (2013) Climate extremes and the carbon cycle. Nature, 500, 287–295.
Rigling, A., Brühlhart, H., Bräker, O.U., Forster, T. & Schweingruber, F.H. (2003)
Effects of irrigation on diameter growth and vertical resin duct production in
Pinus sylvestris L. on dry sites in the central Alps, Switzerland. Forest Ecology
and Management, 175, 285–296.
Rosner, S. & Hannrup, B. (2004) Resin canal traits relevant for constitutive resistance of
Norway spruce against bark beetles: environmental and genetic variability. Forest
Ecology and Management, 200, 77–87.
Rotenberg, E. & Yakir, D. (2010) Contribution of semi-arid forests to the climate system.
Science, 327, 451–454.
Sala, A., Piper, F. & Hoch, G. (2010) Physiological mechanisms of drought-induced tree
mortality are far from being resolved. New Phytologist, 186, 274–281.
Seager, R., Naik, N. & Vecchi, G.A. (2010) Thermodynamic and Dynamic Mechanisms
for Large-Scale Changes in the Hydrological Cycle in Response to Global
Warming. Journal of Climate, 23, 4651–4668.
Seager, R., Ting, M., Held, I., Kushnir, Y., Lu, J., Vecchi, G., Huang, H.-P., Harnik, N.,
Leetmaa, A., Lau, N.-C., Li, C., Velez, J. & Naik, N. (2007) Model Projections of
an Imminent Transition to a More Arid Climate in Southwestern North America.
Science, 316, 1181–1184.
Seager, R. & Vecchi, G.A. (2010) Greenhouse warming and the 21st century
hydroclimate of southwestern North America. Proceedings of the National
Academy of Sciences, 107, 21277–21282.
Seidel, D.J., Fu, Q., Randel, W.J. & Reichler, T.J. (2007) Widening of the tropical belt in
a changing climate. Nature Geoscience, 1, 21–24.
Seidl, R., Fernandes, P.M., Fonseca, T.F., Gillet, F., Jönsson, A.M., Merganičová, K.,
Netherer, S., Arpaci, A., Bontemps, J.-D., Bugmann, H., González-Olabarria,
J.R., Lasch, P., Meredieu, C., Moreira, F., Schelhaas, M.-J. & Mohren, F. (2011)
Modelling natural disturbances in forest ecosystems: a review. Ecological
Modelling, 222, 903–924.
52
Seneviratne, S.I., Nicholls, N., Easterling, D., Goodess, C.M., Kanae, S., Kossin, J., Luo,
Y., Marengo, J., McInnes, K. & Rahimi, M. (2012) Changes in climate extremes
and their impacts on the natural physical environment. Managing the risks of
extreme events and disasters to advance climate change adaptation. A Special
Report of Working Groups I and II of the Intergovernmental Panel on Climate
Change (IPCC) (eds. C.B. Field, V. Barros, T.F. Stocker, D. Quin, D. Dokken,
K.L. Ebi, M.D. Mastrandrea, K.J. Mach, G.-K. Plattner, S.K. Allen, M. Tignor &
P.M. Midgley), pp. 109–230. Cambridge University Press, Cambridge, UK and
New York, NY, USA.
Sevanto, S., McDowell, N.G., Dickman, L.T., Pangle, R. & Pockman, W.T. (2014) How
do trees die? A test of the hydraulic failure and carbon starvation hypotheses.
Plant, Cell & Environment, 37, 153–161.
Sitch, S., Huntingford, C., Gedney, N., Levy, P.E., Lomas, M., Piao, S.L., Betts, R.,
Ciais, P., Cox, P., Friedlingstein, P., Jones, C.D., Prentice, I.C. & Woodward, F.I.
(2008) Evaluation of the terrestrial carbon cycle, future plant geography and
climate‐carbon cycle feedbacks using five Dynamic Global Vegetation Models
(DGVMs). Global Change Biology, 14, 2015–2039.
Six, D.L. & Bentz, B.J. (2007) Temperature Determines Symbiont Abundance in a
Multipartite Bark Beetle-fungus Ectosymbiosis. Microbial Ecology, 54, 112–118.
Suarez, M.L., Ghermandi, L. & Kitzberger, T. (2004) Factors predisposing episodic
drought‐induced tree mortality in Nothofagus – site, climatic sensitivity and
growth trends. Journal of Ecology, 92, 954–966.
Swetnam, T.W. & Betancourt, J.L. (1998) Mesoscale disturbance and ecological response
to decadal climatic variability in the American Southwest. Journal of Climate, 11,
3128–3147.
Teuling, A.J., Seneviratne, S.I., Stöckli, R., Reichstein, M., Moors, E., Ciais, P.,
Luyssaert, S., van den Hurk, B., Ammann, C., Bernhofer, C., Dellwik, E.,
Gianelle, D., Gielen, B., Grünwald, T., Klumpp, K., Montagnani, L., Moureaux,
C., Sottocornola, M. & Wohlfahrt, G. (2010) Contrasting response of European
forest and grassland energy exchange to heatwaves. Nature Geoscience, 3, 722–
727.
Todaro, L., Andreu, L., Alessandro, C.M. D’, Gutiérrez, E., Cherubini, P. & Saracino, A.
(2007) Response of Pinus leucodermis to climate and anthropogenic activity in
the National Park of Pollino (Basilicata, Southern Italy). Biological Conservation,
137, 507–519.
Twidwell, D., Wonkka, C.L., Taylor, C.A., Zou, C.B., Twidwell, J.J. & Rogers, W.E.
(2014) Drought‐induced woody plant mortality in an encroached semi‐arid
53
savanna depends on topoedaphic factors and land management. Applied
Vegetation Science, 17, 42–52.
Tyree, M.T. & Sperry, J.S. (1989) Vulnerability of Xylem to Cavitation and Embolism.
Annual Review of Plant Physiology and Plant Molecular Biology, 40, 19–36.
Vaganov, E.A., Anchukaitis, K.J. & Evans, M.N. (2011) How Well Understood Are the
Processes that Create Dendroclimatic Records? A Mechanistic Model of the
Climatic Control on Conifer Tree-Ring Growth Dynamics. Dendroclimatology
(eds M.K. Hughes, T.W. Swetnam & H.F. Diaz), pp. 37–75. Springer
Netherlands.
Villalba, R. & Veblen, T.T. (1998) Influences of large-scale climatic variability on
episodic tree mortality in northern Patagonia. Ecology, 79, 2624–2640.
Waring, R. (1987) Characteristics of trees predisposed to die. BioScience, 37, 569–574.
Waring, R. & Pitman, G. (1985) Modifying lodgepole pine stands to change
susceptibility to mountain pine beetle attack. Ecology, 889–897.
Weiss, J.L., Castro, C.L. & Overpeck, J.T. (2009) Distinguishing Pronounced Droughts
in the Southwestern United States: Seasonality and Effects of Warmer
Temperatures. Journal of Climate, 22, 5918–5932.
Wiley, E. & Helliker, B. (2012) A re‐evaluation of carbon storage in trees lends greater
support for carbon limitation to growth. New Phytologist, 195, 285–289.
Williams, A.P., Allen, C.D., Macalady, A.K., Griffin, D., Woodhouse, C.A., Meko,
D.M., Swetnam, T.W., Rauscher, S.A., Seager, R., Grissino-Mayer, H.D., Dean,
J.S., Cook, E.R., Gangodagamage, C., Cai, M. & McDowell, N.G. (2013)
Temperature as a potent driver of regional forest drought stress and tree mortality.
Nature Climate Change, 3, 292–297.
Wimmer, R. & Grabner, M. (1997) Effects of climate on vertical resin duct density and
radial growth of Norway spruce [Picea abies (L.) Karst.]. Trees, 11, 271–276.
Wit, H.A., Bryn, A., Hofgaard, A., Karstensen, J., Kvalevåg, M.M. & Peters, G.P. (2014)
Climate warming feedback from mountain birch forest expansion: reduced albedo
dominates carbon uptake. Global Change Biology, 20, 2344–2355.
Worrall, J.J., Egeland, L., Eager, T., Mask, R.A., Johnson, E.W., Kemp, P.A. &
Shepperd, W.D. (2008) Rapid mortality of Populus tremuloides in southwestern
Colorado, USA. Forest Ecology and Management, 255, 686–696.
54
Wunder, J. (2007) Conceptual Advancement and Ecological Applications of Tree
Mortality Models Based on Tree-Ring and Forest Inventory Data. Ph.D.
dissertation, ETH Zurich, Zurich, Switzerland.
Wunder, J., Bigler, C., Reineking, B., Fahse, L. & Bugmann, H. (2006) Optimisation of
tree mortality models based on growth patterns. Ecological Modelling, 197, 196–
206.
Wyckoff, P.H. & Clark, J.S. (2000) Predicting tree mortality from diameter growth: a
comparison of maximum likelihood and Bayesian approaches. Canadian Journal
of Forest Research, 30, 156–167.
Wyckoff, P.H. & Clark, J.S. (2002) The relationship between growth and mortality for
seven co-occurring tree species in the southern Appalachian Mountains. Journal
of Ecology, 90, 604–615.
Xu, C., McDowell, N.G., Sevanto, S. & Fisher, R.A. (2013) Our limited ability to predict
vegetation dynamics under water stress. New Phytologist, 200, 298–300.
Zaitchik, B.F., Macalady, A.K., Bonneau, L.R. & Smith, R.B. (2006) Europe’s 2003 heat
wave: a satellite view of impacts and land–atmosphere feedbacks. International
Journal of Climatology, 26, 743–769.
Zeppel, M.J., Anderegg, W.R. & Adams, H.D. (2013) Forest mortality due to drought:
latest insights, evidence and unresolved questions on physiological pathways and
consequences of tree death. New Phytologist, 197, 372–374.
Zhao, J., Hartmann, H., Trumbore, S., Ziegler, W. & Zhang, Y. (2013) High temperature
causes negative whole-plant carbon balance under mild drought. New Phytologist,
200, 330–339.
55
APPENDIX A: A GLOBAL OVERVIEW OF DROUGHT AND HEAT-INDUCED
TREE MORTALITY REVEALS EMERGING CLIMATE CHANGE RISKS FOR
FORESTS
This paper was published in the journal Forest Ecology and Management.
Reproduced with written permission from Elsevier © 2010 and the authors
56
Forest Ecology and Management 259 (2010) 660–684
Contents lists available at ScienceDirect
Forest Ecology and Management
journal homepage: www.elsevier.com/locate/foreco
A global overview of drought and heat-induced tree mortality reveals
emerging climate change risks for forests
Craig D. Allen a,*, Alison K. Macalady b, Haroun Chenchouni c, Dominique Bachelet d, Nate McDowell e,
Michel Vennetier f, Thomas Kitzberger g, Andreas Rigling h, David D. Breshears i, E.H. (Ted) Hogg j,
Patrick Gonzalez k, Rod Fensham l, Zhen Zhang m, Jorge Castro n, Natalia Demidova o,
Jong-Hwan Lim p, Gillian Allard q, Steven W. Running r, Akkin Semerci s, Neil Cobb t
a
U.S. Geological Survey, Fort Collins Science Center, Jemez Mountains Field Station, Los Alamos, NM 87544, USA
School of Geography and Development and Laboratory of Tree-Ring Research, University of Arizona, Tucson, AZ 85721, USA
Department of Biology, University of Batna, 05000 Batna, Algeria
d
Department of Biological and Ecological Engineering, Oregon State University, Corvalllis, OR 97330, USA
e
Earth and Environmental Sciences, MS J495, Los Alamos National Laboratory, Los Alamos, NM 87544,USA
f
CEMAGREF, ECCOREV FR 3098, Aix-Marseille University, Aix-en-Provence, France
g
Laboratorio Ecotono, INIBIOMA-CONICET and Univ. Nacional del Comahue, Quintral 1250, 8400 Bariloche, Argentina
h
Swiss Federal Institute for Forest, Snow and Landscape Research WSL, Zurcherstr. 111, CH-8903 Birmensdorf, Switzerland
i
School of Natural Resources and the Environment, and Department of Ecology and Evolutionary Biology, University of Arizona, Tucson, AZ 85721, USA
j
Northern Forestry Centre, Canadian Forest Service, 5320-122 Street, Edmonton, Alberta T6H 3S5, Canada
k
Center for Forestry, University of California, Berkeley, CA 94720, USA
l
Queensland Herbarium, Environmental Protection Agency, Mt Coot-tha Road, Toowong, Queensland 4066, Australia
m
Research Institute of Forest Ecology, Environment and Protection, Chinese Academy of Forestry,
Key Laboratory of Forest Protection of State Forestry Administration, Beijing 100091, China
n
Grupo de Ecologı´a Terrestre, Departamento de Ecologı´a, Universidad de Granada, Granada E-18071, Spain
o
Northern Research Institute of Forestry, Nikitov St., 13, Arkhangelsk 163062, Russian Federation
p
Division of Forest Ecology, Department of Forest Conservation, Korea Forest Research Institute #57, Hoegi-ro, Dongdaemun-gu, Seoul 130-712, Republic of Korea
q
Forestry Department, Food and Agriculture Organization (FAO), Viale delle Terme di Caracalla, 00100 Rome, Italy
r
Numerical Terradynamics Simulation Group, University of Montana, Missoula, MT 59812, USA
s
Central Anatolia Forestry Research Institute, P.K. 24, 06501 Bahcelievler-Ankara, Turkey
t
Department of Biological Sciences and Merriam Powell Center for Environmental Research, Northern Arizona University, Flagstaff, AZ 86011, USA
b
c
A R T I C L E I N F O
A B S T R A C T
Article history:
Received 4 March 2009
Received in revised form 13 August 2009
Accepted 1 September 2009
Greenhouse gas emissions have significantly altered global climate, and will continue to do so in the
future. Increases in the frequency, duration, and/or severity of drought and heat stress associated with
climate change could fundamentally alter the composition, structure, and biogeography of forests in
many regions. Of particular concern are potential increases in tree mortality associated with climateinduced physiological stress and interactions with other climate-mediated processes such as insect
outbreaks and wildfire. Despite this risk, existing projections of tree mortality are based on models that
lack functionally realistic mortality mechanisms, and there has been no attempt to track observations of
climate-driven tree mortality globally. Here we present the first global assessment of recent tree
mortality attributed to drought and heat stress. Although episodic mortality occurs in the absence of
climate change, studies compiled here suggest that at least some of the world’s forested ecosystems
already may be responding to climate change and raise concern that forests may become increasingly
vulnerable to higher background tree mortality rates and die-off in response to future warming and
drought, even in environments that are not normally considered water-limited. This further suggests
risks to ecosystem services, including the loss of sequestered forest carbon and associated atmospheric
feedbacks. Our review also identifies key information gaps and scientific uncertainties that currently
hinder our ability to predict tree mortality in response to climate change and emphasizes the need for a
globally coordinated observation system. Overall, our review reveals the potential for amplified tree
mortality due to drought and heat in forests worldwide.
Published by Elsevier B.V.
Keywords:
Climate change
Drought effects
Forest die-off
Forest mortality
Global patterns
Tree mortality
* Corresponding author. Tel.: +1 505 672 3861x541; fax: +1 505 672 9607.
E-mail address: [email protected] (C.D. Allen).
0378-1127/$ – see front matter . Published by Elsevier B.V.
doi:10.1016/j.foreco.2009.09.001
57
C.D. Allen et al. / Forest Ecology and Management 259 (2010) 660–684
1. Introduction
Forested ecosystems are being rapidly and directly transformed
by the land uses of our expanding human populations and
economies. Currently less evident are the impacts of ongoing
climate change on the world’s forests. Increasing emissions of
greenhouse gases are now widely acknowledged by the scientific
community as a major cause of recent increases in global mean
temperature (about 0.5 8C since 1970) and changes in the world’s
hydrological cycle (IPCC, 2007a), including a widening of the
Earth’s tropical belt (Seidel et al., 2008; Lu et al., 2009). Even under
conservative scenarios, future climate changes are likely to include
further increases in mean temperature (about 2–4 8C globally)
with significant drying in some regions (Christensen et al., 2007;
Seager et al., 2007), as well as increases in frequency and severity of
extreme droughts, hot extremes, and heat waves (IPCC, 2007a;
Sterl et al., 2008).
Understanding and predicting the consequences of these
climatic changes on ecosystems is emerging as one of the grand
challenges for global change scientists, and forecasting the impacts
on forests is of particular importance (Boisvenue and Running,
2006; Bonan, 2008). Forests, here broadly defined to include
woodlands and savannas, cover 30% of the world’s land surface
(FAO, 2006). Around the globe societies rely on forests for essential
services such as timber and watershed protection, and less tangible
but equally important recreational, aesthetic, and spiritual
benefits. The effects of climate change on forests include both
positive (e.g. increases in forest vigor and growth from CO2
fertilization, increased water use efficiency, and longer growing
seasons) and negative effects (e.g. reduced growth and increases in
stress and mortality due to the combined impacts of climate
change and climate-driven changes in the dynamics of forest
insects and pathogens) (Ayres and Lombardero, 2000; Bachelet
et al., 2003; Lucht et al., 2006; Scholze et al., 2006; Lloyd and Bunn,
2007). Furthermore, forests are subject to many other human
influences such as increased ground-level ozone and deposition
(Fowler et al., 1999; Karnosky et al., 2005; Ollinger et al., 2008).
Considerable uncertainty remains in modeling how these and
other relevant processes will affect the risk of future tree die-off
events, referred to hereafter as ‘forest mortality’, under a changing
climate (Loehle and LeBlanc, 1996; Hanson and Weltzin, 2000;
Bugmann et al., 2001). Although a range of responses can and
should be expected, recent cases of increased tree mortality and
die-offs triggered by drought and/or high temperatures raise the
possibility that amplified forest mortality may already be occurring
in some locations in response to global climate change. Examples of
recent die-offs are particularly well documented for southern parts
of Europe (Peñuelas et al., 2001; Breda et al., 2006; Bigler et al., 2006)
and for temperate and boreal forests of western North America,
where background mortality rates have increased rapidly in recent
decades (van Mantgem et al., 2009) and widespread death of many
tree species in multiple forest types has affected well over 10 million
ha since 1997 (Raffa et al., 2008). The common implicated causal
factor in these examples is elevated temperatures and/or water
stress, raising the possibility that the world’s forests are increasingly
responding to ongoing warming and drying.
This paper provides an overview of recent tree mortality due to
climatic water stress and warm temperatures in forests around the
globe. We identify 88 well-documented episodes of increased
mortality due to drought and heat and summarize recent literature
on forest mortality and decline. From this review we examine the
possibility of emerging mortality risks due to increasing temperatures and drought. Climate as a driver of tree mortality is also
reviewed, summarizing our scientific understanding of mortality
processes as context for assessing possible relationships between
changing climate and forest conditions. Note that while climatic
661
events can damage forests in many ways ranging from ice storms
to tornadoes and hurricanes, our emphasis is on climate-induced
physiological stress driven by drought and warm temperatures.
The ecological effects of increased mortality in forests and the
associated consequences for human society remain largely
unassessed. We conclude by outlining key information gaps and
scientific uncertainties that currently limit our ability to determine
trends in forest mortality and predict future climate-induced forest
die-off. Addressing these gaps would provide improved information to support policy decisions and forest management worldwide.
2. Methods
This paper emerged in part from collaborations and presentations developed in special sessions on climate-related forest
mortality at two international meetings: the 2007 annual meeting
of the Ecological Society of America in San Jose, California (Allen
and Breshears, 2007), and the 2008 international conference
entitled ‘‘Conference on Adaptation of Forests and Forest Management to Changing Climate with Emphasis on Forest Health’’ in
Umeå, Sweden (Allen, 2009). In addition to citing contributions
from these sessions, we conducted a systematic search for
published accounts of climate-induced tree mortality since 1970
using the ISI Web of Science and Google Scholar. We used different
combinations of the key words ‘‘tree,’’ ‘‘forest,’’ ‘‘mortality,’’ ‘‘dieoff,’’ ‘‘dieback,’’ ‘‘decline,’’ and ‘‘drought’’ in the searches. We also
consulted regional forestry experts to find examples recorded in
government documents and other sources outside the scientific
literature.
From the extensive set of documents uncovered during these
searches, we used two specific criteria to determine whether
the reference was appropriate for this review. Criteria for
inclusion were that the study included: (1) an estimate of area
affected or amount of adult tree mortality at the stand or
population level, based on ground measurements, aerial photography, or remote sensing, and (2) documentation of a strong
correspondence between increases in mortality and increased
water stress or high temperatures. We included examples where
biotic agents were involved in the mortality, but excluded
examples of fire-driven death. Studies of forest decline or partial
canopy dieback without significant increases in mortality
were also excluded, as were studies that documented only
seedling mortality. To simplify presentation, we standardized
study descriptors and combined references that describe
impacts of the same event on the same tree species but used
slightly different methods or were conducted at different spatial
scales.
To estimate trends in the literature related to climate-induced
forest mortality, we searched the ISI Web of Science using the topic
words ‘‘forest AND mortality AND drought’’ over the available
interval from 1985 to 2009. We then controlled for increases in the
general scientific literature related to forests by standardizing the
number of target articles by the number of citations uncovered by a
search using only the topic word ‘‘forest.’’
For each mortality event (listed as rows in Appendix Tables A1–
A6) we tested the association between the forest type affected by
mortality and the categorized duration of the mortality-triggering
drought (seasonal event vs. multi-year drought) with a Chi-square
analysis, comparing number of observed triggering droughts (by
drought and forest type) versus expected number of triggering
droughts. Forest types were grouped into four major biome types
considering similar water limitations: (1) savannas, (2) conifer
forests and Mediterranean woodlands, (3) temperate evergreen
and deciduous forests, and (4) evergreen broadleaved tropical
forests.
58
662
C.D. Allen et al. / Forest Ecology and Management 259 (2010) 660–684
Fig. 1. White dots indicate documented localities with forest mortality related to climatic stress from drought and high temperatures. Background map shows potential
environmental limits to vegetation net primary production (Boisvenue and Running, 2006). Only the general areas documented in the tables are shown—many additional
localities are mapped more precisely on the continental-scale maps. Drought and heat-driven forest mortality often is documented in relatively dry regions (red/orange/
pink), but also occurs outside these regions.
3. Results
3.1. Examples of recent climate-induced forest mortality
More than 150 references that document 88 examples of forest
mortality met our criteria of events that were driven by climatic
water/heat stress since 1970. The examples range from modest but
significant local increases in background tree mortality rates to
climate-driven episodes of regional-scale forest die-off. We found
examples from each of the wooded continents that collectively
span diverse forest types and climatic zones (Figs. 1–8 and
Tables A1–A6). Despite our collective efforts to secure references
from non-English language sources, this review is clearly more
comprehensive for North America, Europe, and Australia, and
obviously incomplete particularly for some regions, including
mainland Asia and Russia.
Our searches also reveal that published reports of climaterelated forest mortality in the scientific literature have increased
markedly in recent decades. For example, a search of the ISI Web of
Science (23 July 2009) using the topic words ‘‘forest AND mortality
AND drought’’ showed 546 references for the period 1985 through
2009, with a steep increase in articles published since 2003 (Fig. 9),
even when standardized for general increases in the forest-related
scientific literature. The years of elevated mortality documented in
the references that met our criteria also show a clear increase in
mortality events with a jump in 1998 and marked accumulation of
events in the 2000s, particularly the years 2003–2004. Although
these trends could be coincidental or a reflection of greater
scientific interest in the topic of tree mortality, recent increases in
reported events also mirror warming global temperatures.
3.1.1. Continental-scale summaries
3.1.1.1. Africa. Increased tree mortality linked to drought and heat
in Africa (Fig. 2; Table A1) includes examples from tropical moist
forest in Uganda (Lwanga, 2003), mountain acacia (Brachystegia
glaucescens) in Zimbabwe (Tafangenyasha, 2001), mesic savanna
trees in South Africa’s Kruger National Park (Viljoen, 1995), and
centuries-old Aloe dichotoma in Namibia (Foden et al., 2007). In the
Sahel, long-term decreases in precipitation linked to anthopogenic
climate change (Biasutti and Giannini, 2006) have caused a die-off
of mesic tree species in parts of Senegal (Gonzalez, 2001),
especially following the severe drought of 1968–1973 (Poupon,
1980). Recent extreme drought in North Africa (Touchan et al.,
2008) is linked to severe mortality of Atlas cedar (Cedrus atlantica)
from Morocco to Algeria (El Abidine, 2003; Bentouati, 2008; Box 1,
see also Fig. 3).
3.1.1.2. Asia. Reports of forest mortality in Asia (Fig. 4; Table A2)
include death triggered by severe El Niño droughts in 1982/1983
and 1997/1998 in the tropical moist forests of both Malaysian and
Indonesian Borneo (Leighton and Wirawan, 1986; Woods, 1989;
Nakagawa et al., 2000; van Nieuwstadt and Sheil, 2005). Severe
droughts are also associated with increased mortality among many
tree species from tropical dry forests in northwest and southwest
India (Khan et al., 1994), Abies koreana in South Korea (Lim et al.,
2008), Juniperus procera from Saudia Arabia (Fisher, 1997), and
pine and fir species in central Turkey (Semerci et al., 2008). Recent
droughts have triggered mortality of Pinus tabulaeformia across 0.5
million ha in east-central China (Wang et al., 2007), and across
extensive areas of Pinus yunnanensis in southwest China (Li, 2003).
The Russian Federal Forest Agency has mapped zones of forest
health risk (‘‘threat’’) across the Russian Federation, showing 338
million ha as ‘‘low threat’’, 260 million ha as ‘‘medium’’ threat, and
76 million ha of ‘‘high’’ threat, predominantly in southerly portions
of the country (Kobelkov, 2008), where forest health problems due
to drought appear to be concentrated (Ermolenko, 2008).
3.1.1.3. Australasia. In the sub-humid environments of northeast
Australia (Fig. 5; Table A3), multi-year droughts have repeatedly
triggered widespread Eucalyptus and Corymbia mortality (Fensham
and Holman, 1999; Rice et al., 2004; Fensham and Fairfax, 2007),
and have also caused tree death in Acacia woodlands (Fensham and
Fairfax, 2005). There is also documentation of drought-induced
mortality in temperate Nothofagus forests in New Zealand
(Hosking and Hutcheson, 1988).
3.1.1.4. Europe. In Europe (Fig. 6; Table A4), forest mortality due to
dry and warm conditions in the 1990s and 2000s arcs across the
Mediterranean regions, including increased death among many
59
C.D. Allen et al. / Forest Ecology and Management 259 (2010) 660–684
663
Fig. 2. Satellite map of Africa, with documented drought-induced mortality areas indicated with numbers, tied to Table A1 references. Upper photo: Cedrus atlantica die-off in
Belezma National Park, Algeria; 2007, by Haroun Chenchouni. Lower photo: quiver tree (Aloe dichotoma) mortality in Tirasberg Mountains, Namibia; 2005, by Wendy Foden.
woody species in Spain (Peñuelas et al., 2001; Martinez-Vilalta and
Piñol, 2002), increased mortality of oak, fir, spruce, beech, and pine
species in France after the extreme heat wave and drought during
the summer of 2003 (Breda et al., 2006; Landmann et al., 2006;
Vennetier et al., 2007), and increases in mortality of Pinus sylvestris
near the species’ southern range limits in Switzerland and Italy
(Dobbertin and Rigling, 2006; Bigler et al., 2006; Vertui and
Tagliaferro, 1998). A severe drought in 2000 killed many Abies
cephalonica in mainland Greece (Tsopelas et al., 2004) and Pinus
halapensis sub. brutia—the most drought tolerant of the Mediterranean pines—in eastern Greece (Körner et al., 2005). Farther north,
summer drought paired with biotic stressors has been linked to
Fig. 3. Map of northern Algeria climate zones and mortality distribution of Cedrus atlantica. ‘‘Box 1—Atlas Cedar Die-off in Algeria’’ serves as the full caption.
60
664
C.D. Allen et al. / Forest Ecology and Management 259 (2010) 660–684
Box 1. Atlas Cedar Die-off in Algeria
Atlas cedar (Cedrus atlantica) occurs in northern Algeria, distributed in scattered montane populations near the limits of its
bioclimatic tolerance between the Sahara Desert and the Mediterranean Sea (Fig. 3). Since the onset of severe drought from
1999 to 2002 cedar forests have undergone mass mortality,
affecting all age classes (Bentouati, 2008). While all Algerian
cedar forests are affected, the magnitude of mortality varies
along a steep moisture gradient (Fig. 3), with die-off greatest
(up to 100%) in the drier mountains nearest the Sahara, dropping to much lower mortality levels in the moister coastal
mountains (Chenchouni et al., 2008). Prolonged soil moisture
deficits lead to decline and progressive death of cedar trees
over a period of 1–3 years; a variety of insects and fungi have
continued to kill weakened cedar trees since the drought eased
after 2002 (Chenchouni et al., 2008). The Cedrus mortality
began as small patches on drier aspects in the arid near-Sahara
mountains, eventually coalescing into large patches affecting
all ages on all exposures. In contrast, only small patches of old
trees on dry aspects have died in more mesic regions near the
coast. This recent drought also triggered substantial mortality
in other Algerian tree species, including Pinus halapensis,
Quercus ilex, Quercus suber, and Juniperus thurifera. Dendrochronological reconstructions of drought in Algeria show that
this early 2000s dry period was the most severe drought since
at least the middle of the 15th century (Touchan et al., 2008),
consistent with climate change projections for a trend of
increasing aridity in this region (Seager et al., 2007).
mortality of Quercus robur in Poland (Siwecki and Ufnalksi, 1998),
Picea abies in southeast Norway (Solberg, 2004), and with a severe
die-off of Picea obovata in northwest Russia (Kauhanen et al., 2008;
Ogibin and Demidova, in press).
3.1.1.5. North America. Climate-induced tree mortality and forest
die-off is relatively well documented for North America (Fig. 7;
Table A5). Drought and warmth across western North America in
the last decade have led to extensive insect outbreaks and
mortality in many forest types throughout the region, affecting
20 million ha and many tree species since 1997 from Alaska to
Mexico (Raffa et al., 2008; Bentz et al., 2009). Examples of forest
die-off range from >1 million ha of multiple spruce species in
Alaska (Berg et al., 2006) and >10 million ha of Pinus contorta in
British Columbia (Kurz et al., 2008a), to drought-induced
Populus tremuloides mortality across a million hectares in
Saskatchewan and Alberta (Hogg et al., 2008). In the southwestern U.S., die-off of Pinus edulis on over a million hectares
was specifically linked to ‘‘global-change-type drought’’ (Breshears et al., 2005). In the eastern portion of the continent,
declines and increased mortality among oaks, particularly in the
red oak family, have been reported from Missouri (Voelker et al.,
2008) to South Carolina (Clinton et al., 1993) in relation to
multi-year and seasonal droughts in the 1980s–2000s. Drought
during the 1980s followed by an unusual spring thaw in eastern
North America also contributed to decline and mortality of
maples in Quebec (Hendershot and Jones, 1989). In addition,
recent increases in background rates of tree mortality across the
Fig. 4. Satellite map of Asia, with documented drought-induced mortality localities indicated with numbers, tied to Table A2 references. Lower R Photo: Dead Abies koreana,
Mount Halla, South Korea; 2008, by Jong-Hwan Lim. Upper R photo: Pinus tabulaeformis mortality in Shanxi Province, China; 2001, by Yugang Wang. Center photo: Dying Pinus
yunnanensis in Yunnan Province, China; 2005, by Youqing Luo. Upper L photo; Abies cilicicia mortality in the Bozkir-Konya region, Anatolia, Turkey; 2002, by Orphan Celik.
Lower L photo: Dying Pinus nigra near Kastamonu, Anatolia, Turkey; 2008, by Akkin Semerci.
61
C.D. Allen et al. / Forest Ecology and Management 259 (2010) 660–684
665
Fig. 5. Satellite map of Australasia, with documented drought-induced mortality areas indicated with numbers, tied to Table A3 references. R photo: Die-off of mulga, Acacia
aneura, the dominant tree across large areas of semi-arid Australia; 2007, by Rod Fensham. L photo: Eucalyptus xanthoclada mortality in Queensland, northeastern Australia;
1996, by Rod Fensham.
western U.S. have been attributed to elevated temperatures (van
Mantgem et al., 2009).
3.1.1.6. South and Central America. In Latin America (Fig. 8;
Table A6), ENSO-related seasonal droughts have amplified background tree mortality rates in tropical forests of Costa Rica
(Chazdon et al., 2005), Panama (Condit et al., 1995), northwest
Brazil (Williamson et al., 2000), and southeast Brazil (Rolim et al.,
2005), and caused extensive mortality of Nothofagus dombeyi in
Patagonian South America (Suarez et al., 2004). A hot and severe
drought across the Amazon basin in 2005, linked to anomalously
warm sea surface temperatures in the North Atlantic, has also
recently been tied to regionally extensive increases in tree
mortality rates and subsequent aboveground biomass loss,
indicating vulnerability of Amazonian forests to moisture stress
(Phillips et al., 2009) (Fig. 9).
3.1.2. Spatial and temporal patterns of mortality
Climate-induced mortality events in this review include
examples that span a broad gradient of woody ecosystems, from
monsoonal savannas with mean precipitation <400 mm/year, to
subalpine conifer forests with a Mediterranean climate, to tropical
Fig. 6. Satellite map of Europe, with documented drought-induced mortality areas indicated with numbers, tied to Table A4 references. R photo: Pinus sylvestris mortality,
Valais, Switzerland; 1999, by Beat Wermelinger. L photo: Pinus sylvestris die-off, Sierra de los Filabres, Spain; 2006, by Rafael Navarro-Cerrillo.
62
666
C.D. Allen et al. / Forest Ecology and Management 259 (2010) 660–684
Fig. 7. Satellite map of North America, with documented drought-induced mortality localities indicated with numbers, tied to Table A5 references. Top photo: Aerial view
showing severe mortality of aspen (Populus tremuloides) in the parkland zone of Alberta, Canada; 2004, by Michael Michaelian. Lower photo: Pinus ponderosa die-off, Jemez
Mountains, New Mexico, USA; 2006, by Craig D. Allen.
Fig. 8. Satellite map of South and Central America, with documented drought-induced mortality localities indicated with numbers, tied to Table A6 references. Photo:
Nothofagus dombeyi mortality at Rı́o Manso Inferior, northern Patagonia, Argentina; 2004, by Thomas Kitzberger.
63
C.D. Allen et al. / Forest Ecology and Management 259 (2010) 660–684
667
Fig. 9. ISI Web of Science search of the trend in published reports of climate-related forest mortality in the scientific literature, for the years 1985–2009. Plotted bars show the
percent of references using the topic words ‘‘forest AND mortality AND drought’’, relative to all ‘‘forest’’ references. Line represents the linear regression model fitted to the
data (R2 = 0.61; F = 35.73; p < 0.001).
rainforests with mean precipitation >3000 mm/year. These cases
reveal a complex set of mortality patterns in response to drought
and heat stress, ranging from modest and short-lived local
increases in background mortality rates to episodes of acute,
regional-scale forest die-off, which often (but not always) involve
biotic agents like insect outbreaks. At broad spatial scales, droughtrelated forest mortality has been reported near species geographic
or elevational range margins where climatic factors (particularly
water stress) are often presumed to be limiting (Allen and
Breshears, 1998; Foden et al., 2007; Jump et al., 2009; Fig. 1;
Fig. 3 and linked Box 1). Spatially extensive die-offs are commonly
associated with prolonged water deficits, such as in savanna and
temperate conifer forest vegetation types during multi-year
droughts (Fensham et al., 2009; Fig. 10). Notably, however,
drought-induced mortality is not restricted to forests typically
thought to be water-limited, as highlighted by events in tropical
rainforests of Borneo where stand-level mortality reached as high
as 26% after the severe El Niño in 1997/1998 (van Nieuwstadt and
Sheil, 2005), or the Amazon basin in 2005 (Phillips et al., 2009).
Mortality in ever-wet and seasonally dry tropical rainforests
appears to be relatively diffuse and incited most often by short but
extreme seasonal droughts (Fig. 10). In temperate forests, short
(seasonal) droughts may be more likely to induce dieback of
broadleaved (deciduous angiosperm) trees (Fig. 10) than conifer
(evergreen needleleaf) trees because of their increased vulnerability to xylem cavitation (Maherali et al., 2004).
Patterns of tree death are often quite patchy at finer spatial
scales across the synoptic region where drought occurs. Although
mortality is sometimes greatest in locally dry landscape positions
(Oberhuber, 2001; Dobbertin et al., 2005; Worrall et al., 2008),
ecosite variability (soils, elevation, aspect, slope, topographic
position) may interact with density-dependent processes such as
insect outbreaks, competition, or facilitation to produce complex
spatial patterns of mortality at the stand and forest scale (Fensham
and Holman, 1999; Lloret et al., 2004). Greater mortality can occur,
for example, on more favorable sites within the middle of
geographic and landscape distributions where higher tree density
drives increased competition for water or elevated insect activity
(Guarin and Taylor, 2005; Greenwood and Weisberg, 2008;
Fensham et al., 2009; Horner et al., 2009; Klos et al., 2009).
Fig. 10. Differences between observed and expected frequencies of reported forest mortality cases listed in Tables A1–A6, sorted by duration of associated drought events
(seasonal vs. multi-year), with forests grouped into four major biomes. Mortality discriminated by forest type is dependent on drought duration, with more drought-adapted
forest types showing mortality during long droughts and less drought-adapted forest types showing more mortality cases during short-term seasonal droughts. Pearson Chisquare = 23.46, df = 3, p = 0.000012.
64
668
C.D. Allen et al. / Forest Ecology and Management 259 (2010) 660–684
However, high severity drought can drive extensive forest
mortality independent of tree density (Floyd et al., 2009). Higher
mortality rates can also occur on favorable sites where trees do not
invest in adequate root systems or where they otherwise become
hydraulically overextended (Ogle et al., 2000; Fensham and
Fairfax, 2007; Nepstad et al., 2007).
Spatial patterns of mortality at the stand and forest scale are
also heavily influenced by life-history traits and tolerances of
individual species within forests, with drought commonly
triggering differential mortality rates between co-occurring tree
species (Suarez et al., 2004; Gitlin et al., 2006; Fensham and Fairfax,
2007; Newbery and Lingenfelder, 2009; Phillips et al., 2009). Larger
and/or older trees often appear more prone to drought-induced
mortality (Mueller et al., 2005; Nepstad et al., 2007; Floyd et al.,
2009), although this relationship is species-dependent, and in
cases where stands are undergoing intense self-thinning, smaller
sub-dominant trees and saplings are often more affected (Kloeppel
et al., 2003; Elliott and Swank, 1994; Hanson and Weltzin, 2000).
Temporal patterns of drought-related tree mortality also can be
difficult to interpret due to lagged responses in some species, in
which mortality has been shown to occur years or even decades after
drought stress (Pedersen, 1998, 1999; Bigler et al., 2007).
Furthermore, the long-lived nature of trees and their ability to shift
allocation of resources and change their hydraulic architecture
throughout their lives can result in non-linear responses to drought
stress in both space and time. Different sequences of climate events
may also affect the risk of mortality (Miao et al., 2009).
4. Discussion
4.1. Climate-induced forest mortality—are new trends emerging?
The diverse instances of mortality reported here clearly
illustrate that drought and heat can impact trees in many forest
types. However awareness of, and interest in, climate-induced
forest mortality and dieback is not new (Auclair, 1993; Ciesla and
Donaubauer, 1994). Past die-offs have been extensively documented. Historic examples include: widespread death of Eucalyptus, Acacia, and Callitris species in the early 1900s triggered by
the worst drought of the instrumental record in northeastern
Australia (Fensham and Holman, 1999); Nothofagus mortality
during 1914–1915 in New Zealand (Grant, 1984); Picea meyeri
mortality during the 1920s in northern China (Liang et al., 2003);
extensive tree mortality in the southern Appalachian Mountains
and the Great Plains during the dust-bowl droughts of the 1920s–
1930s (Hursh and Haasis, 1931; Albertson and Weaver, 1945);
Pinus sylvestris death during 1940–1955 in Switzerland (Dobbertin
et al., 2007); oak mortality in many European countries following
severe droughts episodes in 1892–1897, 1910–1917, 1922–1927,
1946–1949, 1955–1961 (Delatour, 1983); extensive tree mortality
of Austrocedrus chilensis during El Niño droughts in the 1910s,
1942–1943, and the 1950s in Argentina (Villalba and Veblen,
1998); and die-off of multiple pine species during the 1950s
drought in the southwestern USA (Swetnam and Betancourt, 1998;
Allen and Breshears, 1998). Furthermore, the overwrought
perception of unprecedented forest decline and impending death
due to air pollution in central Europe (where it was referred to as
‘Waldsterben’) and eastern North America that received much
attention in the 1980s provides a cautionary example of
exaggerated claims of widespread forest health risk in the absence
of adequate evidence (Skelly and Innes, 1994).
So are recent occurrences of die-off simply well-documented
examples of a natural phenomenon linked to climate variability, or is
global climate change driving increases in forest mortality? We
recognize that the available data on climate-induced forest mortality
have many limitations: our examples represent a compilation of
idiosyncratic case studies with uneven geographic coverage. The
studies differed greatly in their goals, methods, and definitions of
mortality, and inconsistently report mortality rates, spatial scale and
patterns of mortality, and severity parameters of climate stress. The
recent increase in forest mortality reports that we document could
merely be an artifact of more scientific attention on climate change,
perhaps in concert with a few high profile cases of climate-related
forest die-off. These limitations, and the lack of any systematic global
monitoring program, currently constrain our ability to determine if
global changes in forest mortality are emerging.
Even though our review is insufficient to make unequivocal
causal attributions, our data are consistent with the possibility that
climate change is contributing to an increase in reported mortality.
Documentation of climate-related forest mortality in association
with recent warming and droughts is rising rapidly (Fig. 9), and in
some of these cases the droughts have been the most severe of the
last few centuries. Furthermore, recent research indicates that
warmer temperatures alone can increase forest water stress
independent of precipitation amount (Barber et al., 2000). In
addition, new experimental results show that warmer temperatures
can greatly accelerate drought-induced mortality (Adams et al.,
2009, and associated correspondence). If the recent increase in
mortality reports is indeed driven in part by global climate change,
far greater chronic forest stress and mortality risk should be
expected in coming decades due to the large increases in mean
temperature and significant long-term regional drying projected in
some places by 2100, in addition to projected increases in the
frequency of extreme events such as severe droughts, hot extremes,
and heat waves (IPCC, 2007a; Jentsch et al., 2007; Sterl et al., 2008).
4.2. Climate and plant physiological interactions that drive
forest mortality
Understanding complex spatial and temporal patterns of
climate-induced tree death and forest die-off requires knowledge
of the physiological drivers of tree mortality. The fundamental
mechanisms underlying tree survival and mortality during
drought remain poorly understood despite decades of research
within the fields of forestry, pathology, entomology, and ecology
(Waring, 1987; Manion, 1991; Mueller-Dombois, 1986,1988;
Breda et al., 2006; Ogaya and Penuelas, 2007; McDowell et al.,
2008). Part of the challenge is that tree mortality commonly
involves multiple, interacting factors, ranging from particular
sequences of climate stress and stand life histories to insect pests
and diseases (Franklin et al., 1987; Miao et al., 2009). Based on the
decline spiral model (Manion, 1991; Manion and Lachance, 1992),
drought can operate as a trigger (‘‘inciting factor’’) that may
ultimately lead to mortality in trees that are already under stress
(by ‘‘predisposing factors’’ such as old age, poor site conditions and
air pollution) and succumb to subsequent stem and root damage
by biotic agents (‘‘contributing factors’’ such as wood-boring
insects and fungal pathogens). McDowell et al. (2008) build upon
Manion’s framework to postulate three mutually non-exclusive
mechanisms by which drought could lead to broad-scale forest
mortality: (1) extreme drought and heat kill trees through
cavitation of water columns within the xylem (Rennenberg
et al., 2006; Zweifel and Zeugin, 2008); (2) protracted water
stress drives plant carbon deficits and metabolic limitations that
lead to carbon starvation and reduced ability to defend against
attack by biotic agents such as insects or fungi (McDowell et al.,
2008; Breshears et al., 2009; Adams et al., 2009); and (3) extended
warmth during droughts can drive increased population abundance in these biotic agents, allowing them to overwhelm their
already stressed tree hosts (Desprez-Loustau et al., 2006; Raffa
et al., 2008; Wermelinger et al., 2008). Although these hypotheses
have growing support, our physiological knowledge remains
65
C.D. Allen et al. / Forest Ecology and Management 259 (2010) 660–684
inadequate for confidently predicting patterns of regional die-off,
as well as variation in survival for trees within the same stand.
The degree to which trees regulate water loss during drought may
explain patterns of carbohydrate (and resin) production and
subsequent susceptibility to drought or biotic attack (McDowell
et al., 2008; Zweifel et al., 2009). A continuum of stomatal responses
to drought exist from drought avoidance (isohydry), in which
stomata close at a threshold water potential to minimize further
transpiration, to drought tolerance (anisohydry), in which stomatal
closure is less severe and transpiration continues at relatively high
rates (McDowell et al., 2008). The isohydric response protects xylem
from cavitation through avoidance of severe low water potentials,
but can cause eventual carbon starvation as stomatal closure shuts
down photosynthesis while respiration costs continue to deplete
carbon stores. The anisohydric response can allow continued carbon
gain through maintaining open stomata but at greater risk of
cavitation, which might kill trees directly or could increase the
likelihood of future carbon deficits. Plants that typify each response
have associated traits consistent with their mode of stomatal
regulation, such as deep rooting access to more reliable soil water
and cavitation-resistant xylem for drought-tolerant species.
In addition to hydraulic failure and carbon starvation, a third
physiological mechanism predisposing plants to mortality may
exist—cellular metabolism limitation. This hypothesis suggests that
low tissue water potentials during drought may constrain cell
metabolism (Würth et al., 2005; Ryan et al., 2006; Sala and Hoch,
2009), thereby preventing the production and translocation of
carbohydrates, resins, and other secondary metabolites necessary for
plant defense against biotic attack. The common observation that
trees which succumb to insect attacks have weak resin flow and are
unable to pitch out attacking insects is consistent with constraints on
photosynthetic carbon uptake, cellular carbon metabolism, and/or
tree water relations. A likely sequence for many isohydric species that
is consistent with Manion’s cascade (Manion, 1991) is that climatestressed trees starve for carbon, perhaps due to poor edaphic position
combined with drought, which causes poor resin flow and an inability
to defend against insect attack, which subsequently allows fungi that
are symbiotic with the beetles to colonize and occlude the sapwood,
causing transpiration to cease, drying of the canopy, and eventual
mortality (McDowell et al., 2008, 2009).
The observation that climate-induced tree mortality is happening
not only in semi-arid regions but also in mesic forests suggests that
the global rise in temperature may be a common driver (van
Mantgem et al., 2009; Adams et al., 2009). The mechanisms by which
rising temperature in the absence of severe precipitation deficits may
result in increased tree mortality include impacts on both host
physiology and biotic agents. Increasing temperature raises the
vapor pressure deficit and evaporation to the atmosphere. This
results in increased water loss through transpiration and either
stomatal closure in the case of isohydric species, or decreased margin
of safety from hydraulic failure in the case of anisohydric species.
Rising temperatures may impact the carbon storage of trees in a
particularly negative way because the rate of carbohydrate
consumption required to maintain cellular metabolism (respiration)
is strongly linked to temperature (Amthor, 2000). The first
experiment under controlled climate to isolate the effect of
temperature on drought-induced tree mortality, conducted on Pinus
edulis, indicates a high degree of sensitivity to elevated temperature
and indirectly implicates carbon starvation (Adams et al., 2009).
Warmer temperatures may also be important where cold
winters are usual, in that abnormally warm winter temperatures
maintain significant physiological activity after the growth
season, with tree respiration costs wasting stored carbohydrates
(Damesin, 2003). Even though CO2 uptake can occur during mild
winters and partially compensate for carbon loss during summer
droughts (Holst et al., 2008), the annual C balance often remains in
669
deficit under these conditions. Therefore under climatic warming
scenarios, drought-avoiding tree species may move closer to
carbon starvation, and drought-tolerant species may come closer
to hydraulic failure (McDowell et al., 2008).
Presumably, surviving individuals after a severe climate event
would have some degree of genetic drought resistance that would
be inherited by the next generation (Gutschick and BassiriRad,
2003; Parmesan, 2006; Millar et al., 2007a). But the adaptation of a
tree species to a markedly different local climate, with only one or a
few generations per century, may be too slow to successfully
respond to the rapid present rate of climate change.
Warming temperatures also have direct effects on insect
population dynamics—in particular, outbreaks of some aggressive
bark beetle species are closely tied to temperature (Logan et al.,
2003; Berg et al., 2006; Hicke et al., 2006; Rouault et al., 2006).
Higher temperatures can accelerate insect development and
reproduction, increasing infestation pressure directly (e.g., Wermelinger and Seifert, 1999; Bale et al., 2002; Caldeira et al., 2002;
Gan, 2004), while at the same time heat-induced drought stress
may reduce tree vigor and increase host susceptibility to insect
attack (Mattson and Haack, 1987; Rouault et al., 2006). Warming
temperatures and drought-stressed trees also may foster increased
mortality from non-insect pathogens, particularly fungi (Ayres and
Lombardero, 2000; Desprez-Loustau et al., 2006; Garrett et al.,
2006). However, fungal responses to climatic factors are complex
and uncertain because of interactions with tree host susceptibility
and insect vectors, and some fungi-tree relationships are difficult
to assess because important belowground interactions between
fungi and tree roots are not well studied.
4.3. Consequences of broad-scale forest mortality
Due to the increasingly tight coupling of human and environmental systems, the consequences of broad-scale forest mortality
are important to contemplate. Trees grow relatively slowly but can
die quickly: a 200-year-old tree may be killed by severe drought
within a few months to a few years. Therefore, mortality of adult
trees can result in ecosystem changes far more rapidly than a
gradual transition driven by tree regeneration and growth (Fig. 11).
If forests are forced to adjust abruptly to new climate conditions
through forest die-off, many pervasive and persistent ecological
and social effects will result. Major changes in understory species
may occur (Rich et al., 2008), as well as the possible development
of novel ecosystems due to new combinations of native and
invasive exotic trees that, depending on the climatic tolerances of
seedlings, eventually repopulate the overstory (Walther et al.,
2005; Millar et al., 2007b; Suarez and Kitzberger, 2008).
Abiotic ecosystem impacts may include changes in solar energy
fluxes reaching ground level and reflecting back to the atmosphere,
with potentially large feedbacks to regional climate in some areas
(Bonan, 2008; Chapin et al., 2008), as well as alterations in hydrology
and ecosystem water budgets due to increases in evaporation and
reductions in transpiration (e.g., Huxman et al., 2005), and changes
in groundwater recharge. Potential effects of extensive forest
mortality on water resource availability could have large effects
on human societies (Millennium Ecosystem Assessment, 2005).
In addition, broad-scale forest mortality could change local,
regional, and global carbon budgets (Breshears and Allen, 2002;
Jones et al., 2009). Forests store considerably more carbon than the
atmosphere, and forest die-off could redistribute within-ecosystem
carbon pools and release pulses of carbon back to the atmosphere. A
recent modeling study simulated this type of transformation in
managed forests of Canada, where climate-related increases in fire
and insect disturbance are forecast to turn these forests into a net
carbon source (Kurz et al., 2008b). Meanwhile, climate-related
increases in the spatial extent of mass tree mortality by insects,
66
670
C.D. Allen et al. / Forest Ecology and Management 259 (2010) 660–684
Fig. 11. Abrupt reductions in forest biomass (or ecosystem carbon) can result from
drought-induced forest die-off and occur more rapidly than the relatively slow
countervailing biomass increments from tree natality and growth. Trajectories of
change vary with ecosystem, as do minimum biomass and carbon values, and are
not to scale in this conceptual figure.
notably mountain pine beetle, have recently transformed some
forests of interior British Columbia (Canada) from a net carbon sink
into a net carbon source (Kurz et al., 2008a). Similarly, it is possible
that ‘‘widespread forest collapse via drought’’ could transform the
world’s tropical forests from a net carbon sink into a large net source
during this century (Lewis, 2006, p. 195; cf. Phillips et al., 2009; Jones
et al., 2009). Land-use impacts such as anthropogenic fires and forest
fragmentation, interacting with climate-induced forest stress, are
likely to amplify these effects in some regions, including the Amazon
Basin (Nepstad et al., 2008). Overall, climate-induced forest
mortality and related disturbances will increase global carbon flux
rates at least temporarily, potentially undermining the capacity of
the world’s forests to act as carbon sinks in the coming centuries.
Past forest management may have exacerbated recent mortality in some regions. In portions of western North America, over a
century of fire suppression has fostered the buildup of unusually
high tree densities. Trees in these unnaturally dense forests can
have decreased vigor, which can increase their vulnerability to
multiple mortality factors (Savage, 1997). Extensive reforestations
with pine plantations in regions such as China and the
Mediterranean Basin (e.g., 3.5 million ha reforested with conifers
since 1940 in Spain alone; J. Castro—from agency statistical
sources) may be particularly vulnerable, especially because some
of these plantations are on marginal sites given the excessive
densities and unknown genetic provenances of the trees.
In summary, given the potential risks of climate-induced forest
die-off, forest managers need to develop adaptation strategies to
improve the resistance and resilience of forests to projected
increases in climate stress (Seppala et al., 2009). Options might
include thinning stands to reduce competition, selection of
appropriate genotypes (e.g., improved drought resistance), and
even translocation of species to match expected climate changes
(e.g. Millar et al., 2007b; Joyce et al., 2008; Richardson et al., 2009).
of adequate data on forest health status globally (FAO, 2006,
2007). Existing permanent sample plot networks can detect large
scale events or a generalized background mortality increase, but
are not designed to detect and assess patchy mortality, even at
rather high rates, as is common when forest landscapes are
heterogeneous and in most of the cases of biotic agent outbreaks.
Reliable, long-term, global-scale forest health monitoring, likely
combining remote-sensing and ground-based measurements in
a methodologically coordinated and consistent manner, is
needed to accurately determine the status and trends of forest
stress and mortality on planet Earth. Regional and global maps of
actual patterns of climate-induced tree mortality are also vitally
important for the development and validation of models for
predicting forest die-off in response to climate change.
(2) Understanding the mechanisms by which climate change may
affect forests requires quantitative knowledge of the physiological
thresholds of individual tree mortality under chronic or acute
water stress (Fig. 12). With the exception of information for a
few tree species (McDowell et al., 2008; Zweifel et al., 2009),
there is surprisingly little species-specific knowledge on
regulation of xylem water potentials; therefore, placing various
species on the continuum of isohydry–anisohydry is difficult,
and predicting how diverse species differentially experience
carbon starvation or hydraulic failure is currently impossible.
Similarly, there is almost no knowledge on the patterns or
mechanisms of carbohydrate storage in response to drought
and heat. The potential effects of other components of changing
atmospheric chemistry (e.g., elevated levels of nitrogen
deposition and ground-level ozone) on the sensitivity of trees
to drought remain inadequately known (Grulke et al., 2009).
Research is also needed on how tree phenologies will respond
to climate warming, because increasing winter temperatures
may contribute to depletion of carbohydrate reserves relevant
to carbon starvation thresholds. In addition, better knowledge
is needed on within-species genetic variability and selection of
trees related to drought and heat stress.
(3) More accurate global vegetation maps are needed as essential
inputs to calibrate and validate dynamic global vegetation models.
The extent of forest mortality can only be documented or
modeled if there is precise information on the locations and
extent of pre-die-off forests.
4.4. Key information gaps and scientific uncertainties
The conclusions that can be drawn about recent trends in tree
mortality and the predictions that can be made about future
climate-induced forest die-off are limited by a number of key
information gaps and scientific uncertainties.
(1) Accurate documentation of global forest mortality patterns and
trends requires the establishment of a worldwide monitoring
program. Despite many national and regional forest-monitoring
efforts (e.g., the European Union’s intensive forest health
monitoring EU/ICP-Forests Level II network), there is an absence
Fig. 12. Conceptual diagram, showing range of variability of ‘‘Current Climate’’
parameters for precipitation and temperature, or alternatively for drought duration
and intensity, with only a small portion of the climate ‘‘space’’ currently exceeding a
species-specific tree mortality threshold. ‘‘Future Climate’’ shows increases in extreme
drought and temperature events associated with projected global climate change,
indicating heightened risks of drought-induced die-off for current tree populations.
67
C.D. Allen et al. / Forest Ecology and Management 259 (2010) 660–684
(4) Spatially explicit documentation of environmental conditions in
areas of forest die-off is necessary to link mortality to causal
climate drivers, including precipitation, temperature, and vapor
pressure deficit. Given the difficulties in measuring precipitation and the absence of reliable soil datasets at adequate
resolutions for continental-scale studies, a robust water
availability index, possibly derived from remote sensing, is
needed to help modelers simulate water stress in trees. In order
to disentangle moisture deficit from temperature effects on
tree mortality, more research is also needed to relate spatial
gradients of mortality to variation in temperature. This
research might utilize historical and dendrochronological
records across spatial and temporal gradients where variations
in rainfall deficit and temperature increase are expressed.
(5) Mechanistic understanding of climate-induced tree mortality
requires improved knowledge of belowground processes and soil
moisture conditions (e.g. Brunner et al., 2009). Models often
include detailed algorithms describing aboveground physiological processes but treat belowground processes as a ‘‘black box’’.
Understanding of the impacts of increasing atmospheric CO2,
nitrogen deposition, ground-level ozone, and drought on root
dynamics, productivity, exudation fluxes, and mycorrhizal
interactions would particularly improve belowground modeling.
(6) The direct effects of climate on the population dynamics of almost all
forest insect pests and other biotic disturbance agents remain poorly
understood but are important to modeling climate-induced forest
mortality (Wermelinger and Seiffert, 1999; Logan et al., 2003;
Desprez-Loustau et al., 2006; Breda et al., 2006; Bentz et al., 2009).
Generalization through synthesis of current knowledge on the
dynamics of damaging biotic agents and tree response to attacks
could improve existing mortality functions in forest models.
(7) Feedbacks between physiological stress (and tree mortality) driven
by climate and other forest disturbance processes (e.g., insect
outbreaks, fire) are poorly understood (Allen, 2007). These major
disturbance processes may increasingly drive the mortality
dynamics of forests in a rapidly changing climate, necessitating
improved modeling of their cumulative and collective effects
(Nepstad et al., 2008).
Current models of vegetation response to climate change share
weaknesses associated with the knowledge gaps identified here,
including individual tree-based process models (Keane et al., 2001),
species-specific empirical models (climate envelope models, e.g.,
Hamann and Wang, 2005; Thuiller et al., 2008), climate envelope
threshold models linked to plant functional types in dynamic global
vegetation models (Scholze et al., 2006), and earth system models (Ciais
et al., 2005; Huntingford et al., 2008). The significant uncertainties
associated with modeling tree mortality are reflected in ongoing
debates about the magnitude of die-off risk to Amazon rainforests and
boreal forests from climate change this century, the potential for dieoffs in forests more generally (Loehle and LeBlanc, 1996; Phillips et al.,
2008; Soja et al., 2007), and the degree to which forests worldwide are
likely to become a net carbon source or sink (e.g., Kurz et al., 2008b).
5. Conclusions
This overview illustrates the complex impacts of drought and
heat stress on patterns of tree mortality, and hints at the myriad
ways in which changes in drought and/or heat severity, duration,
and frequency may lead to gradually increasing background tree
mortality rates and even rapid die-off events. Many recent
examples of drought and heat-related tree mortality from around
the world suggest that no forest type or climate zone is
invulnerable to anthropogenic climate change, even in environments not normally considered water-limited. Current observations of forest mortality are insufficient to determine if worldwide
trends are emerging in part due to the lack of a reliable, consistent,
671
global monitoring system. Although the effects of climate change
cannot be isolated in these studies and clearly episodic forest tree
mortality occurs in the absence of climate change, the globally
extensive studies identified here are consistent with projections of
increased forest mortality and suggest that some forested
ecosystems may already be shifting in response to climate.
There are major scientific uncertainties in our understanding of
climate-induced tree mortality, particularly regarding the mechanisms that drive mortality, including physiological thresholds of tree
death and interactions with biotic agents. Recent advances in the
understanding of tree mortality mechanisms suggest that forests
could be particularly sensitive to increases in temperature in
addition to drought alone, especially in cases where carbon
starvation rather than hydraulic failure is the primary mechanism
of tree mortality. However, we currently lack the ability to predict
mortality and die-off of tree species and forest types based on
specific combinations of climatic events and their interactions with
biotic stressors and place-specific site conditions. The potential for
broad-scale climate-induced tree mortality can be considered a nonlinear ‘‘tipping element’’ in the Earth’s climate system (Lenton et al.,
2008), because forest die-offs from drought can emerge abruptly at a
regional scale when climate exceeds species-specific physiological
thresholds, or if climate triggers associated irruptions of insect pests
in weakened forests. Such cross-scale mortality processes in forests
remain poorly understood.
Collectively, these uncertainties currently prevent reliable
determination of actual mortality trends in forests worldwide,
and also hinder model projections of future forest mortality in
response to climate change. As one consequence, the potential for
climate change to trigger widespread forest die-off may be underrepresented in important assessments to date, notably including
the latest major IPCC report (2007b). If extensive climate-induced
tree mortality occurs, then substantial negative ecological and
societal consequences can be expected. Determining the potential
for broad-scale, climate-induced tree mortality is therefore a key
research priority for ecologists and global change scientists, and is
essential for informing and supporting policy decisions and forest
management practices.
Acknowledgements
We thank Rebecca Oertel, Andrew Goumas, Ángeles G. Mayor,
Russell Fairfax, and Megan Eberhardt Frank for literature review
assistance; Jennifer Shoemaker for graphics support; and Julio
Betancourt, Adrian Das, Dan Fagre, Brian Jacobs, Francisco Lloret,
Cynthia Melcher, Catherine Parks, Tom Veblen, and Connie
Woodhouse, and two anonymous reviewers for comments on this
paper. Support was provided by the U.S. Geological Survey,
Biological Resources Discipline, Global Change Program (CDA); the
National Science Foundation and Science Foundation Arizona
(AKM); US DOE NICCR DE-FC02-06ER64159 and Biosphere 2Philecology (DDB); and Chinese Special Research Program for
Public-Welfare Forestry 2007BAC03A02 and 200804001 (ZZ). This
work is a contribution of the Western Mountain Initiative, a USGS
global change research project.
Appendix A
These appendix tables (Tables A1–A6) accompany the continental-scale maps and associated text descriptions, and are the core
compilation of documented examples of drought and heat-induced
tree mortality. Organized by continent and year of mortality event,
concisely listing key information for each documented example,
including an identification number allowing easy visual linkage to the
continental-scale map locations.
2000–2008
Zimbabwe
(Southeast)
Senegal
South Africa
(Northern
Province)
South Africa
(Northern
Province)
Uganda
(Western)
Namibia,
South Africa
Algeria
Morocco
3
4
5
6
7
8
9
10
Med. montane conifer
(300–600)
Med. conifer (348–356)
Savanna (100–200)
Tropical Rainforest (1492)
Savanna (240–500)
Woodland, deciduous
broadleaf (500–600)
Savanna, deciduous
broadleaf woodland
(240–560)
Savanna
Savanna (366)
a
Cedrus atlantica
Cedrus atlantica
Aloe dichotoma
Uvariopsis spp., Celtis spp.
C. mopane, Combretum
apiculatum, Grewia spp.,
Ximenia americana
Dichrostachys cinerea,
Pterocarpus angolensis,
Strychnos madagascariensis,
Terminalia sericea, C. mopane,
many others
Anacardium occidentale,
Cordyla pinnata, Ficus ingens,
many others
Brachystegia glaucescens;
other savanna species
Colophospermum mopane
Acacia senegal, Guiera
senegalensis
Dominant tree taxa
Arid edge of geographic
range
Arid edge of geographic
range
Arid edge of geographic
range
Not reported
Patchy within range
Patchy within range
Arid edges of geographic
range
Not reported
Patchy within range
Middle–lower edges of
elevational range;
arid edge of geographic
range
Spatial concentration
of mortality within
geographic or
elevational range
Multi-year drought
Multi-year drought
Multi-year drought,
high temperatures
Seasonal drought
Multi-year drought
Multi-year drought
Multi-year drought
Multi-year droughts
Multi-year drought
Multi-year drought
Climate anomaly
linked to mortality
10–40
40–80
2–71
19
7
1–78
(speciesdependent)
23
Not reported
13–87
(basal area)
50
Stand/
populationlevel
mortality
(%)b
Subregional
Subregional
Subregional
Not reported
Not reported
Not reported
Regional
Subregional;
500,000 ha
affected
Not reported
Regional
Scale of
impact/area
affected
Not reported
Insects
None
Not reported
None
None
None
Elephants,
scale insects
Not reported
None
Biotic agents
associated with
mortality?c
El Abidine (2003);
Adil (2008)
Bentouati (2008);
Bentouati and
Bariteau (2006);
Chenchouni
et al. (2008)
Foden et al. (2007)
Lwanga (2003)
O’Connor (1999)
Viljoen (1995)
Gonzalez (2001)
Tafangenyasha
(2001, 1998, 1997)
MacGregor and
O’Connor (2002)
Poupon (1980)
Reference(s)d
b
Mediterranean forest types are abbreviated as Med. in this column. Annual precipitation is in mm/yr in parentheses if reported.
Severity of mortality is reported at the stand or population level as percentage of dead trees (depending on study design), unless otherwise noted in the entry. Other common units are annual mortality rate during drought (%/
year), percent dead basal area, and dead wood volume in meters3.
c
If biotic agents are thought to have played a primary role in tree mortality, this is noted in bold type. If biotic agents were involved in mortality but their role was not evaluated or is secondary to climate, the agents are simply
listed.
d
Citations from which reported mortality data is derived are written in bold type. Other citations provide corroborating or secondary evidence. If there are multiple citations without no bold type, reported data reflects numbers
compiled from all citations.
2002–2008
1999
1982–1997
1991–1993
1945–1993
1970–1982,
1991–1992
1988–1992
Savanna (300)
Forest type/mean precip.
672
a
1904–2002
South Africa
(Northern
Province)
2
1972–1973
Senegal
1
Year(s) of
mortality
Location
ID
Table A1
Documented cases of drought and/or heat-induced forest mortality from Africa, 1970–present. ID numbers refer to locations mapped in Fig. 2.
68
C.D. Allen et al. / Forest Ecology and Management 259 (2010) 660–684
Malaysia
(Borneo)
India (Gujarat)
Russia (Far
East)
Saudi Arabia
and Oman
Indonesia
(Sumatra)
Indonesia and
Malaysia
(Borneo)
Indonesia
(Borneo)
Malaysia
(Borneo)
2
3
4
5
6
7
8
9
2002–2007
2003–2008
2005–2008
12 Turkey
(Central
Anatolia)
13 South Korea
14 Russia
Boreal and temperate
Temperate montane
mixed (1400–2000)
Picea spp., Pinus spp.
Abies koreana
Qercus spp., Juniper
spp., Pinus nigra,
P. sylvestris,
Abies cilicicia
Pinus yunnanensis
Subtropical coniferous
plantation
Temperate conifer and
mixed (400–600)
Pinus tabulaeformia
Dipterocarpaceae,
Euphorbiaceae,
Burseraceae,
Myristicaceae
Anacardiacea,
Dipterocarpaceae,
Sapotaceae, Rutaceae
Dipterocarpus spp.,
Lauraceae
Not reported
Juniperus procera,
J. excelsa
Picea jezoensis, Abies
nephrolepis
Acacia senegal,
Holarrhena
antidysenterica,
Helicteres isora,
Terminalia crenulata,
others
Dipterocarpus spp.,
Shorea spp.
Calophyllurn spp.,
Syzygium spp.
Dominant tree taxa
Temperate coniferous
plantation
Tropical rainforest
(2700)
Tropical lowland
swamp (2800)
Tropical rainforest
(2100–3000)
Tropical rainforest
Woodland (559)
Montane mixed conifer
Tropical dry deciduous
Tropical rainforest
(2000)
Montane tropical
rainforest
Forest type/mean
precip.a
Southern portions of
Russian forest zones
Not reported
Southern edge of
geographic range
for P. sylvestris
Not reported
Not reported
Not reported
Not reported
Not reported
Not reported
Lower edges of
elevational range
Mountain slopes
and plateaus,
variable aspects
Not reported
Not reported
Upper-mid
elevational range
Spatial concentration
of mortality within
geographic or
elevational range
Drought
Warm winters/
springs, possibly
drought
Drought
Seasonal
drought
Seasonal
drought
Seasonal
drought
Seasonal
drought
Seasonal
drought
Seasonal
drought
Possibly
drought
Drought
Seasonal or
single-year
drought
Seasonal
drought
Seasonal
drought
Climate
anomaly
linked to
mortality
Not reported
20–50
Not reported
Varied in
different
plantations
30
4.3–6.4
4.2–6.1
0.6–26.3
9.8
>400,000 ha
across the nation
Landscape
Not reported
Landscape; 26,700–
113,000 ha affected
Subregional;
500,000 ha
affected
Not reported
Not reported
Not reported
Not reported
Landscape–
subregional
165,000 ha
affected
14 M m3 timber
lost
30 (J. excelsa)
141,000 ha
affected
Not reported
Not reported
Scale of
impact/
area affected
37–82 (speciesdependent)
12–28
50–100
Stand/
populationlevel mortality
(%)b
Not reported
Not reported
Insects
Bark beetles
(Tomicus
yunnanensis,
T. minor)
Bark beetles
(Dendroctonus
valens)
Not reported
Not reported
Not reported
Not reported
None
Fungi
Ungulates
(Cervus unicolor)
Not reported
Not reported
Biotic agents
associated
with
mortality?c
Ermolenko (2008)
Lim et al. (2008);
Woo et al. (2007)
Semerci et al. (2008)
Li (2003)
Wang et al. (2007)
Nakagawa et al. (2000);
Lingenfelder and
Newbery (2009)
Nishimua et al. (2007)
van Nieuwstadt and Sheil (2005);
Potts (2003); Aiba and Kitayama
(2002); Slik (2004)
Kinnaird and O’Brien, 1998
Fisher and Gardner (1995); Fisher
(1997); Gardner and Fisher (1996)
Man’ko and Gladkova (2001)
Khan et al. (1994)
Woods (1989); Becker et al.
(1998); Leighton and
Wirawan (1986)
Werner (1988)
Reference(s)d
C.D. Allen et al. / Forest Ecology and Management 259 (2010) 660–684
Footnotes as in Table A1.
1986–1988;
1998–2000;
2003–2005
1998–2001
1997–1998
1997–1998
1997–1998
1997–1998
Early 1990s
1987–1988
1987
1982–1983
1976–1980
Year(s) of
mortality
11 China
(Yunnan)
China (Shanxi,
Hebei, Henan)
Sri Lanka
1
10
Location
ID
Table A2
Documented cases of drought and/or heat-induced forest mortality from Asia, 1970-present. ID numbers refer to locations mapped in Fig. 4.
69
673
New Zealand
(West Coast)
New Zealand
(Hawkes Bay)
Australia
(Queensland)
Australia
(Queensland)
Australia
(Queensland)
Australia
(Queensland)
1
2
3
4
5
6
2005
2004
1990–2002
1992–1996
1984–1987
1978–1980
Year(s) of
mortality
Acacia spp.
Eucalyptus spp.,
Corymbia spp.
Eucalyptus spp.,
Corymbia spp.
Eucalyptus spp.,
Corymbia spp.
Nothofagus solandri
Nothofagus fusca
Dominant tree
taxa
Widespread
Patchy within
ranges
Patchy within
ranges
Patchy within
ranges
Not reported
Not reported
Spatial concentration
of mortality within
geographic or
elevational range
Multi-year drought
Multi-year drought
Multi-year drought
Multi-year drought
Spring droughts
Spring droughts
Climate
anomaly linked
to mortality
Not recorded
15.0 (basal area;
unpublished data)
78 stand level;
17.7 across
region
29 (basal area)
24–52
75
Stand/
populationlevel
mortality (%)b
Switzerland
(Valais)
Europe
(Western,
Central)
France
Poland
Greece
Italy (South
Tyrol)
2
3
4
5
6
Location
1
ID
1992
1987–1989
1979–1987
1980–1985
1970–1985
1960–1976
Year(s) of
mortality
Temperate
mixed conifer
(650)
Mediterranean
mixed conifer
(1622)
Temperate
broadleaf
(500–550)
Temperate
broadleaf
(650–850)
Temperate
conifer and
broadleaf
(600–1500)
Temperate
conifer (572)
Forest type/
mean precip.a
Pinus sylvestris
Abies alba Mill.
A. cephalonica
Loud.
Quercus robur
Quercus spp,.
mainly Q. robur
Abies spp., Picea
spp., Pinus spp.,
Fagus sylvatica
Pinus sylvestris
Dominant tree
taxa
Lower/southern edges
of ranges
Middle of elevation
ranges
Not reported
Patchy across ranges
Lower edges of
elevation range
Lower/southern edges
of ranges
Spatial concentration
of mortality within
geographic or
elevational range
Multi-year
drought
Multi-year
drought
Seasonal
drought
Seasonal or
single-year
drought
Repeated
droughts
Multi-year
drought
Climate
anomaly linked
to mortality
Not reported
Landscape–
subregional
Landscape–
subregional
Landscape
111,000 m3
timber lost
1.8/yr in
drought
years
Subregional;
patchy across
500,000 ha
Regional;
patchy across
<1 M ha
Landscape–
subregional
Scale of impact/
area affected
600 ha
affected
Not reported
Regional;
5.5 M ha
affected
Regional;
5.5 M ha
affected
Not reported
Landscape;
5000 ha
affected
None
None
None
None
Leafminer (Neomycta
pulicaris); Fungi
(Nodulisporium spp.)
Various insects
Bark beetles and other
insects
Moths (Tortrix viridiana);
pathogens
(Ophiostoma spp.)
Fungi; bark beetles
(Agriles, Scolytus)
Bark beetles (Scolytus,
Ips, Pityogenes, Tomicus,
Dendrochtonus,
Pytiokteines); Fungi
Not reported
Fensham and Fairfax
(2005)
Fensham and Fairfax
(2007)
Fensham et al.
(2003, 2009)
Fensham and Holman
(1999); Fensham (1998);
Rice et al. (2004)
Hosking and Hutcheson
(1988)
Hosking and Kershaw
(1985)
Reference(s)d
Minerbi (1993)
Markalas (1992); Kailidis
and Markalas (1990)
Siwecki and Ufnalksi
(1998)
Nageleisen (1994);
Nageleisen et al. (1991);
Delatour (1983)
Schutt and Cowling (1985)
Kienast et al. (1981)
Reference(s)d
Beech scale (Inglisia fagi);
Fungi (Hypocrella duplex);
Wood borer (Platypus spp.,
Psepholax spp.)
Biotic agents associated
with mortality?c
Biotic agents associated
with mortality?c
Scale of impact/
area affected
10–50
1–20
5–100
Stand/
populationlevel
mortality (%)b
Table A4
Documented cases of drought and/or heat-induced forest mortality from Europe, 1970–present. ID numbers refer to locations mapped in Fig. 6.
Tropical savanna
(500–850)
Tropical savanna
(500–850)
Tropical savanna
(500–850)
Tropical savanna
(480–2600)
Montane
broadleaf
Montane
broadleaf
Forest type/
mean precip.a
674
Footnotes are as given in Table A1.
Location
ID
Table A3
Documented cases of drought and/or heat-induced forest mortality from Australasia, 1970–present. ID numbers refer to locations mapped in Fig. 5.
70
C.D. Allen et al. / Forest Ecology and Management 259 (2010) 660–684
France
(Ardennes,
Vosges)
Norway
Greece
(Samos)
Austria (Tyrol)
Greece
(South,
Central)
Switzerland
Switzerland
(Valais)
Germany
(BadenWürttemberg)
Spain
Russia
(Northwest)
11
12
13
14
15
16
17
18
19
20
Italy (Aosta)
9
Spain
(Northeast,
Central, South)
Austria (Tyrol)
8
10
Austria (Lower
Austria)
7
2004–2006
2004–2006
2003–2006
1973–1976,
1987–1993,
1996–2000,
2000–2004
2003
2000–2002
2001
2000
1992–2000
1998
1994, 1998
1985–1998
1991–1997
1990–1996
Boreal conifer
Temperate conifer
plantations
Temperate
broadleaf
Temperate
mixed conifer
and broadleaf
(500–600)
Temperate
conifer and
broadleaf
Mediterranean
mixed conifer
(700–1100)
Picea obovata
Pinus sylvestris,
Pinus nigra
Fagus sylvatica
Pinus sylvestris
Picea abies
Abies
cephalonica
Pinus sylvestris
Pinus brutia
Mediterranean
mixed conifer
(700–800)
Temperate
mixed conifer
(710)
Picea abies
Fagus sylvatica
Quercus spp.,
Pinus spp.,
Juniperus spp.
Pinus sylvestris
Pinus sylvestris
Pinus sylvestris,
Pinus nigra
Temperate
conifer
Montane mixed
conifer and
broadleaf
(800–1200)
Mediterranean
mixed conifer
and broadleaf
(537–605)
Temperate mixed
conifer and
broadleaf (550)
Temperate mixed
conifer (840)
Temperate mixed
conifer (650)
Patchy
Not reported
Not reported
Lower/southern
edges of ranges
Not reported
Not reported
Lower edge of
elevational range
Lower edge of
elevational range
Patchy across ranges
Middle of ranges
Patchy within
elevational range;
southern edge of
geographic range
(P. sylvestris)
Lower/southern
edges of ranges
Lower edge of
elevational range
Lower edge of
elevational range
Drought, high
temperatures
Multi-year
drought
Drought, high
temperatures
Seasonal and
multi-year
droughts, high
temperatures
Drought, high
temperatures
Multi-year
drought
Seasonal
droughts
Multi-year
drought
Multi-year
summer
droughts, high
summer
temperatures
Deep frost
after an
abnomally hot
period
Multi-year
drought,
recurrent
summer
droughts
Multi-year
drought
Seasonal
droughts
Seasonal
droughts
Patchy across
13,404 ha
1.9 M ha
affected
208 M m3
timber lost
Landscape–
subregional
98,000 m3
timber lost
Not reported
Landscape–
subregional
Landscape–
subregional
2.0 M m3
timber lost
7–59
Landscape
Landscape–
subregional
Not reported
Landscape–
subregional
Subregional;
patchy across
200,000 ha
Landscape–
subregional
Landscape–
subregional
Landscape
Stand–landscape
5–10/yr in
drought years
vs. 0.17–0.50/
yr in nondrought years
Not reported
Not reported
2–6.6
5–30
0.0–19.4
(speciesdependent)
Not reported
10.0–70.0
27.6–49.2
Bark beetles
(Ips typographus), fungi
Not reported
Bark, ambrosia beetles
(Taphrorychus bicolor,
Trypodendron domesticum);
wood borer
Primary role, bark
beetles (Phaenops
cyanea, Ips acuminatus);
nematodes; mistletoe
Bark beetles
(Ips typographus)
Primary role, bark
beetles (Phaenops
knoteki, Pityokteines
spinidens) mistletoe
Not reported
Not reported
Bark beetles
(Polygraphus
poligraphus)
None
Not reported
Fungi (Armillaria spp.);
wood borers
Various insects
Various insects
Krotov (2007); Tsvetkov
and Tsvetkov (2007); Chuprov
(2007); Shtrakhov (2008);
Kauhanen et al. (2008)
Navarro-Cerrillo et al.
(2007)
Petercord (2008)
Wermelinger et al. (2008);
Dobbertin et al. (2007);
Bigler et al. (2006);
Dobbertin and Rigling
(2006); Rigling et al. (2006);
Dobbertin et al. (2005);
Rigling and Cherubini (1999)
Forster et al. (2008)
Tsopelas et al. (2004);
Raftoyannis et al. (2008)
Oberhuber (2001)
Körner et al. (2005);
Sarris et al. (2007)
Solberg (2004)
French Forest Health
Department (1998–1999)
Peñuelas et al. (2001);
Lloret and Siscart (1995);
Lloret et al. (2004);
Martinez-Vilalta and
Piñol (2002)
Vertui and Tagliaferro
(1998)
Cech and Perny (2000)
Cech and Tomiczek (1996)
71
C.D. Allen et al. / Forest Ecology and Management 259 (2010) 660–684
675
Switzerland
(Grisons)
France
(Provence,
Southern Alps)
France
France
(Eastern
Pyrénées)
France
(Provence,
Maures
Mountains)
21
22
23
24
25
Mediterranean
broadleaf
Temperate
mixed conifer
(800–1000)
Temperate
mixed conifer
and broadleaf
(650–1100)
Mediterranean
conifer
(750–950)
Temperate mixed
conifer (750)
Forest type/
mean precip.a
Quercus suber
Abies alba
Quercus spp.
Fagus sylvatica,
Abies spp.,
Picea abies,
Pinus spp.
Pinus sylvestris
Pinus sylvestris
Dominant tree
taxa
Northern edge to
middle of geographic
range
Lower edge to middle
of ranges
Lower and middle of
elevational range
Lower/southern edges
of ranges
Lower edge of
elevational range
Spatial concentration
of mortality within
geographic or
elevational range
Multi-year
drought
Recurrent
drought, high
temperatures
Spring and
summer
drought,
scorching heat
Multi-year
drought, high
temperatures
Drought, high
temperatures
Climate
anomaly linked
to mortality
10–70
10–30
1–3/yr
20–80
6.3–16.0
Stand/
populationlevel
mortality (%)b
Subregional;
patchy across
120,000 ha
Subregional;
patchy across
150,000 ha
Regional
Subregional;
patchy across
100,000 ha
Landscape–
subregional
Scale of impact/
area affected
1979–1986
1984–1989
3 USA (Midwest)
4 USA (North Carolina)
Temperate
deciduous
(1270–1520)
Temperate
deciduous
Temperate
deciduous
1984
2 USA (Midwest)
Forest type/
mean precip.a
Late 1970s– Upland
1980s
temperate
mixed
Year(s) of
mortality
1 USA (Southeast,
Northeast, Midwest)
ID Location
Acer saccharum,
Fagus grandifolia,
Tilia americana,
Aesculus flava
Betula spp.
Acer spp.
Quercus spp.,
Cayra spp.
Dominant tree
taxa
Not reported
Not reported
Not reported
Not reported
Multi-year drought
Multi-year drought
Drought
Multi-year droughts;
high temperatures
preceded by severe
winters
Spatial
Climate anomaly
concentration
linked to mortality
of mortality
within
geographic or
elevational range
1.0–3.25/yr. in
drought years
Not reported
Not reported
16.6 in stands
across Southeast;
1.2–6.3 in Missouri
Stand/
populationlevel
mortality (%)b
Not reported
Landscape–
subregional
Landscape–
subregional
Regional
Not reported
leafminers;
wood borers;
birch
skeletonizers
Wood borers
(Agrilus spp.)
Wood borers
(Agrilus
bilineatus);
fungi; insect
defoliators
Biotic agents
associated with
mortality?c
Insects (Platypus spp.,
Coroebus spp.)
Bark beetles
(Ips, Pissodes)
Bark beetles; fungi
Bark beetles
(Tomicus, Ips, Pissodes)
Not reported
Biotic agents associated
with mortality?c
Scale of
impact/area
affected
Table A5
Documented cases of drought and/or heat-induced forest mortality from North America, 1970–present. ID numbers refer to locations mapped in Fig. 7.
2006–2008
2003–2008
2003–2008
2003–2008
2003–2007
Year(s) of
mortality
Olano and Palmer (2003)
Millers et al. (1989)
Millers et al. (1989)
Stringer et al. (1989); Starkey
and Oak (1989); Starkey et al.
(1989); Clinton et al. (1993);
Millers et al. (1989); Tainter
et al. (1983); Law and Gott
(1987); Kessler (1989); Jenkins
and Pallardy (1995)
Reference(s)d
Vennetier et al. (2008)
French Forest Health
Department (2003–2008)
Breda et al. (2006);
Landmann et al. (2006);
Rouault et al. (2006);
French Forest Health
Department (2003–2008)
Vennetier et al. (2007);
Thabeet et al. (2009)
Schilli et al. (in press)
Reference(s)d
676
Footnotes as in Table A1.
Location
ID
Table A4 (Continued )
72
C.D. Allen et al. / Forest Ecology and Management 259 (2010) 660–684
1986–1992
1986–1992
1986–1992
1985–1995
1996
1990–1997
1990–2002
1983–2004
1989–2004
8 USA (California)
9 USA (California)
10 USA (California)
11 USA (California)
12 USA (Arizona)
13 Canada (Alberta)
14 USA (Midwest,
Southeast)
15 USA (California)
16 USA and Canada
(Alaska, Yukon)
Not reported
Patchy within
ranges
Not reported
Not reported
Pinus spp.,
Abies spp.
Quercus spp.
Populus tremuloides
Pinus edulis,
Juniperus
monosperma
Pinus flexilis
Multi-year drought,
high temperatures
Multi-year drought
Multi-year drought
Patchy within
ranges
Lower edges
of elevational
range
Patchy within
ranges
Patchy within
ranges
63% increase
in annual
mortality rate
15–50 basal
area reduction
18–47
2.3–25.9
50–75
Not reported
13 (basal area)
23.3–69.2
4–15
10–15
18.2
Stand/
populationlevel
mortality (%)b
Drought, high summer Not reported
temperatures
Drought, high
temperatures
Multi-year drought
Drought preceding
warm winter and
spring
Patchy within
Single-year drought
elevational range
Lower edges
of elevational
range
Pinus spp., Abies spp. Drier edge of
local range;
lower edges
of elevational
ranges
Not reported
Multi-year drought,
high spring and
summer temperatures
Multi-year drought
Drought, high
temperatures
preceded by winter
thaw
Multi-year drought
Spatial
Climate anomaly
concentration
linked to mortality
of mortality
within
geographic or
elevational range
Pinus ponderosa,
Not reported
Calocedrus decurrens,
Abies concolor
Pinus jeffreyi,
Abies concolor
Acer saccharum
Quercus ellipsoidalis,
Q. macrocarpa
Dominant tree
taxa
Coastal rainforest, Picea spp.
boreal (485)
Montane mixed
conifer
(1100–1400)
Upland
temperate
mixed
Boreal forest,
prairie ecotone
(450)
Woodland
(370)
Montane
mixed conifer
Montane
mixed conifer
Montane
mixed conifer
Montane
mixed conifer
(945)
Montane
mixed conifer
(600–800)
1985–early
1990s
7 USA and Mexico
(California and Baja
California)
Savanna (726)
Forest type/
mean precip.a
Temperate
deciduous
(900–1200)
1987–1989
Year(s) of
mortality
6 Eastern North America 1980s
5 USA (Minnesota)
ID Location
Table A5 (Continued )
Subregional;
>1.2 M ha
Landscape–
subregional
Regional;
1.8 M ha
affected
Subregional;
patchy across
1 M ha
Landscape–
subregional
Stand–
landscape
Landscape–
subregional
Landscape–
subregional;
56,000 ha
affected
Landscape
Landscape–
subregional
Subregional;
patchy across
>1 M ha
Not reported
Scale of
impact/area
affected
Primary role,
bark beetle
(Dendroctonus
rufipennis)
Insects,
pathogens
Wood borers
(Enaphalodes
rufulus, Agrilus
spp.); fungi
Insect defoliator
(Malacosoma
disstria)
Not reported
Mistletoe
(Arceuthobium)
bark beetles
(Dendroctonus
ponderosae)
Primary role,
bark beetles
(Dendroctonus
spp.); engraver
beetles (Scolytus
spp.)
Engraver
beetles (Scolytus
spp.)
Bark beetles
(Dendroctonus
spp.)
Bark beetles
(Dendroctonus
spp.)
Insect defoliator
(Malacosoma
disstria)
Not reported
Biotic agents
associated with
mortality?c
Berg et al. (2006)
van Mantgem and
Stephenson (2007)
Starkey et al. (2004);
Oak et al. (2004); Voelker
et al. (2008); Heitzman
et al. (2004); Lawrence
et al. (2002)
Hogg et al. (2002)
Mueller et al. (2005);
Ogle et al. (2000); Trotter (2004)
Millar et al. (2007a)
Ferrell et al. (1994);
Ferrell (1996)
Macomber and Woodcock
(1994)
Guarin and Taylor (2005)
Savage (1997)
Hendershot and Jones (1989);
Payette et al. (1996); Auclair
et al. (1996); Roy et al. (2004);
Robitaille et al. (1982)
Faber-langendoen and
Tester (1993)
Reference(s)d
73
C.D. Allen et al. / Forest Ecology and Management 259 (2010) 660–684
677
2001–2004
2002–2004
2000–2006
2005–2006
1955–2007
1997–2007
2004–2007
1998–2001, Not reported
2005–2008
Long-term
1880–2008
19 USA (Arizona)
20 Canada
(Saskatchewan
and Alberta)
21 Canada (British
Columbia)
22 USA (Colorado)
23 USA
(Western States)
24 Western North
America
25 USA (Minnesota)
26 USA (California)
27 Canada and USA
(Alaska,
British Columbia)
Not reported
Populus
tremuloides,
Fraxinus spp.
Pinus spp., Picea
spp., Abies spp.,
Pseudotsuga
menziesii
Many species
Populus tremuloides
Pinus contorta
Populus
tremuloides
Pinus ponderosa
Pinus edulis, Pinus
monophylla,
Juniperus
monosperma,
Juniperus
scopulorum
Temperate coastal Chamaecyparis
rainforest
nootkatensis
(1300–4000)
Boreal and
temperate mixed
(480–900)
Coniferous
All western
forest types
Montane mixed
(380–1100)
Montane mixed
conifer
(250–1000)
Boreal forest,
prairie ecotone
(360–460)
Coniferous (180)
Woodland
(200–450)
Pinus ponderosa,
Pinus edulis,
Juniperus
monosperma,
Populus spp.
Dominant tree
taxa
Multi-year drought
Middle
Not reported
Lower edges
and middle of
ranges
Not reported
Not reported
None
70% of basal area lost Subregional;
200,000 ha
affected
Bentz et al. (2009)
van Mantgem et al. (2009)
Worrall et al. (2008)
Kurz et al. (2008a)
Beier et al. (2008); Hennon
and Shaw (1997); Hennon
et al. (2005)
Garrett et al. (2006)
Insect defoliators Minnesota Dept. Nat.
Resources (2007)
Primary role,
bark and
engraver beetles
(Dendroctonus,
Ips, Dryocoetes,
Scolytus spp.)
Not reported
Warmer winters
and springs
Not reported
Regional;
60.7 M ha
affected
Regional
Primary role,
pathogen
(Phytophthora
ranorum)
Not reported
Not reported
3.9-fold increase
in annual mortality
rate
Landscape–
Wood borers;
subregional;
cytospora
58,374 ha affected canker; bark
beetles
Regional –
Primary role,
continental;
bark beetle
13 M ha affected (Dendroctonus
ponderosae)
>435 M m3
(timber lost)
Insect defoliators Hogg et al. (2008)
Drought preceded
423,000 dead tress
Landscape–
by wet, warm episodes in northern California subregional
Drought
Drought, high
temperatures
High temperatures
Breshears et al. (2005);
Shaw et al. (2005); Swaty
et al. (2004); Mueller et al.
(2005); Allen (2007);
Greenwood and Weisberg
(2008)
Gitlin et al. (2006);
Burkett et al. (2005)
Reference(s)d
Primary role,
Negron et al. (2009)
bark and
engraver
beetles (Ips spp.)
Primary role,
bark beetles
(Ips confusus);
twig beetles;
pitch moths;
root fungus;
mistletoe
Not reported
Biotic agents
associated with
mortality?c
Subcontinental;
patchy across
10 M ha
Landscape–
subregional
Regional;
1.2 M ha
affected
Landscape–
subregional
Scale of
impact/area
affected
3.6/yr vs. 1.6/yr
in non-drought yrs.
7 –21
6 region-wide,
0–90 stand-level
for Pinus spp.;
4.5 stand-level for
J. monosperma
3.3–41.4 (species
dependant)
Stand/
populationlevel
mortality (%)b
Patchy but
Multi-year drought,
32 (stand scale);
concentrated at high spring and
5.62 (landscape
lower edges of
summer temperatures. scale)
elevational range
Drought, high spring
Middle of
geographic range and summer
temperatures
Southern edge
of geographic
range
Lower edges of
Multi-year drought,
elevational range high temperatures
Patchy within
Multi-year drought,
geographic and
high spring and
elevational range summer
temperatures
Patchy within
Multi-year drought
elevational range
Spatial
Climate anomaly
concentration
linked to mortality
of mortality
within
geographic or
elevational range
678
Footnotes as in Table A1.
2000–2004
18 Southwest, USA
(New Mexico,
Arizona, Colorado,
Utah, Nevada)
Woodland,
conifer
(250–750)
2000–2004
17 USA (Southwest)
Forest type/
mean precip.a
Year(s) of
mortality
ID Location
Table A5 (Continued )
74
C.D. Allen et al. / Forest Ecology and Management 259 (2010) 660–684
75
Phillips et al. (2009)
Not reported
Subcontinental
1.2–1.6 Pg C lost
Single year drought
Not reported
Not reported
Tropical rainforest
2005
Amazon Basin
6
Footnotes as in Table A1.
Suarez et al. (2004);
Suarez and Kitzberger
(2008); Bran et al. (2001)
Wood borers;
woodpeckers
Landscape–
subregional
11–57
Seasonal drought,
high temperatures
Arid edge of
geographic range;
lower elevations
Nothofagus
dombeyi
Temperate
steppe – Montane
broadleaf
(500–1500)
Argentina
(western
Neuquén,
Rio Negro)
5
1998–1999
Chazdon et al. (2005)
Not reported
Not reported
3.1/yr vs. 1.6/yr in
non-drought years
Seasonal drought
Not reported
Not reported
Tropical rainforest
(3962)
Costa Rica
(Heredia)
4
1998
Williamson et al.
(2000); Laurance
et al. (2001)
Not reported
Not reported
1.9/yr vs. 1.21–1.23/yr
in non-drought years
Seasonal drought
Not reported
Not reported
Tropical rainforest
(2000)
Brazil
(Amazonas)
3
1997
Rolim et al. (2005)
Not reported
Not reported
4.5/yr vs. 1.4/yr in
non-drought years
Seasonal drought
Not reported
Not reported
Tropical rainforest
(1200)
Brazil
(Espı́rito
Santo)
2
1986–1989,
1997–1999
Condit et al. (1995);
Leigh et al. (1990)
Not reported
Not reported
2.75/yr vs. 1.98/yr in
non-drought years
Seasonal drought,
high temperatures
Not reported
205 different
species
Tropical rainforest
(2600)
Panama
(Panama)
1
1982–1985
Scale of impact/
area affected (ha)
Stand/population-level
mortality (%)b
Climate anomaly
linked to mortality
Spatial
concentration
of mortality
within geographic
or elevational range
Dominant
tree taxa
Forest type/
mean precip.a
Location
Year(s) of
reported
mortality
679
References
ID
Table A6
Documented cases of drought and/or heat-induced forest mortality from South and Central America, 1970–present. ID numbers refer to locations mapped in Fig. 8.
Biotic agents
associated with
mortality?c
Reference(s)d
C.D. Allen et al. / Forest Ecology and Management 259 (2010) 660–684
Adams, H.D., Guardiola-Claramonte, M., Barron-Gafford, G.A., Villegas, J.C., Breshears, D.D., Zou, C.B., Troch, P.A., Huxman, T.E., 2009. Temperature sensitivity
of drought-induced tree mortality: implications for regional die-off under
global-change-type drought. Proceedings of the National Academy of Sciences
of the United States of America 106, 7063–7066.
Adil, S., 2008. Climate change and forest in Morocco: case of the decay of the cedar in
the Atlas Mountains. In: Poster Presentation At: International Conference
‘‘Adaptation of Forests and Forest Management to Changing Climate with
Emphasis on Forest Health: A Review of Science, Policies, and Practices’’, Umeå,
Sweden: FAO/IUFRO, 25–28 August 2008.
Aiba, S.I., Kitayama, K., 2002. Effects of the 1997–98 El Nino drought on rain forests
of Mount Kinabalu, Borneo. Journal of Tropical Ecology 18, 215–230.
Albertson, F.W., Weaver, J.E., 1945. Injury and death or recovery of trees in prairie
climate. Ecological Monographs 15, 393–433.
Allen, C.D., 2007. Interactions across spatial scales among forest dieback, fire, and
erosion in northern New Mexico landscapes. Ecosystems 10, 797–808.
Allen, C.D., 2009. Climate-induced forest dieback: an escalating global
phenomenon? Unasylva 231/232 (60), 43–49.
Allen, C.D., Breshears, D.D., 1998. Drought-induced shift of a forest-woodland
ecotone: rapid landscape response to climate variation. Proceedings of the
National Academy of Sciences of the United States of America 95, 14839–
14842.
Allen, C.D., Breshears, D.D., 2007.In: Meetings: Organized Oral Session on ‘‘ClimateInduced Forest Dieback as an Emergent Global Phenomenon: Patterns, Mechanisms, and Projections’’. Annual Meeting of Ecological Society of America EOS
88(47), San Jose, California, 7 August 2007, p. 504.
Amthor, J.S., 2000. The McCree-de-Wit-Penning de Vries-Thornley respiration
paradigms: 30 years later. Annals of Botany 86, 1–20.
Auclair, A.N.D., 1993. Extreme climatic fluctuations as a cause of forest dieback in
the Pacific Rim. Water Air and Soil Pollution 66 (3–4), 207–229.
Auclair, A.N., Lill, D., Revenga, J.T.C., 1996. The role of climate variability and global
warming in the dieback of Northern Hardwoods. Water, Air, and Soil Pollution
91, 163–186.
Ayres, M.P., Lombardero, M.J., 2000. Assessing the consequences of global change
for forest disturbances for herbivores and pathogens. The Total Science of the
Environment 262, 263–286.
Bachelet, D., Neilson, R.P., Hickler, T., Drapek, R.J., Lenihan, J.M., Sykes, M.T., Smith,
B., Sitch, S., Thonicke, K., 2003. Simulating past and future dynamics of natural
ecosystems in the United States. Global Biogeochemistry Cycles 17, 1045
doi:10.1029/2001GB001508.
Bale, J.S., Masters, G.J., Hodkinson, I.D., Awmack, C., Bezemer, T.M., Brown, V.K.,
Butterfield, J., Buse, A., Coulson, J.C., Farrar, J., Good, J.E., Harrington, R., Hartley,
S., Jones, T.H., Lindroth, R.L., Press, M.C., Symrnioudis, I., Watt, A.D., Whittaker,
J.B., 2002. Herbivory in a global climate change research: direct effects of rising
temperature on insect herbivores. Global Change Biology 88, 1–16.
Barber, V.A., Juday, G.P., Finney, B.P., 2000. Reduced growth of Alaskan white spruce
in the twentieth century from temperature-induced drought stress. Nature 405,
668–673.
Becker, P., Lye, O.C., Goh, F., 1998. Selective drought mortality of dipterocarp trees:
no correlation with timber group distributions in Borneo. Biotropica 30, 666–
671.
Beier, C.M., Sink, S.E., Hennon, P.E., D’Amore, D.V., Juday, G.P., 2008. Twentiethcentury warming and the dendroclimatology of declining yellow-cedar forests
in southeastern Alaska. Canadian Journal of Forest Research 38, 1319–1334.
Bentouati, A., 2008. La situation du cèdre de l’Atlas en Algérie. Forêt Méditerranéenne 29, 203–209.
Bentouati, A., Bariteau, M., 2006. Réflexions sur le dépérissement du cèdre de l’Atlas
des Aurès (Algérie). Forêt Méditerranéenne 27, 317–322.
Bentz, B.J., Allen, C.D., Ayres, M., Berg, E., Carroll, A., Hansen, M., Hicke, J., Joyce, L.,
Logan, J., MacFarlane, W., MacMahon, J., Munson, S., Negron, J., Paine, T., Powell,
J., Raffa, K., Régnière, J., Reid, M., Romme, W., Seybold, S., Six, D., Tomback, D.,
Vandygriff, J., Veblen, T., White, M., Witcosky, J., Wood, D., 2009. In: Bentz, B.J.
(Ed.), Bark Beetle Outbreaks in Western North America: Causes and Consequences. Univ. of Utah Press, , ISBN: 978-0-87480965-7p. 42.
Berg, E.E., Henry, J.D., Fastie, C.L., De Volder, A.D., Matsuoka, S.M., 2006. Spruce
beetle outbreaks on the Kenai Peninsula, Alaska, and Kluane National Park and
Reserve, Yukon Territory: relationship to summer temperatures and regional
differences in disturbance regimes. Forest Ecology and Management 227, 219–
232.
Biasutti, M., Giannini, A., 2006. Robust Sahel drying in response to late 20th century
forcings. Geophysics Research Letters 33, L11706 doi:10.1029/2006GL026067.
Bigler, C., Braker, O.U., Bugmann, H., Dobbertin, M., Rigling, A., 2006. Drought as an
inciting mortality factor in Scots pine stands of the Valais, Switzerland. Ecosystems 9, 330–343.
Bigler, C., Gavin, D.G., Gunning, C., Veblen, T.T., 2007. Drought induces lagged tree
mortality in a subalpine forest in the Rocky Mountains. Oikos 116, 1983–1994.
Boisvenue, C., Running, S.W., 2006. Impacts of climate change on natural forest
productivity—evidence since the middle of the 20th century. Global Change
Biology 12, 1–21.
Bonan, G.B., 2008. Forests and climate change: forcings, feedbacks, and the climate
benefits of forests. Science 320, 1444–1449.
Bran, D., Pérez, A., Ghermandi, L., Barrios Lamunière, S.D., 2001. Evaluación de
poblaciones de coihue (Nothofagus dombeyi) del Parque Nacional Nahuel Huapi,
afectadas por la sequı́a 98/99, a escala de paisaje (1:250.000).
76
680
C.D. Allen et al. / Forest Ecology and Management 259 (2010) 660–684
Breda, N., Huc, R., Granier, A., Dreyer, E., 2006. Temperate forest trees and stands
under severe drought: a review of ecophysiological responses, adaptation
processes and long-term consequences. Annals of Forest Science 63, 625–644.
Breshears, D.D., Allen, C.D., 2002. The importance of rapid, disturbance-induced
losses in carbon management and sequestration. Global Ecology and Biogeography Letters 11, 1–15.
Breshears, D.D., Cobb, N.S., Rich, P.M., Price, K.P., Allen, C.D., Balice, R.G., Romme,
W.H., Kastens, J.H., Floyd, M.L., Belnap, J., Anderson, J.J., Myers, O.B., Meyer, C.W.,
2005. Regional vegetation die-off in response to global-change-type drought.
Proceedings of the National Academy of Sciences of the United States of America
102 (42), 15144–15148.
Breshears, D.D., Myers, O.B., Meyer, C.W., Barnes, F.J., Zou, C.B., Allen, C.D., McDowell, N.G., Pockman, W.T., 2009. Tree die-off in response to global-change-type
drought: mortality insights from a decade of plant water potential measurements. Frontiers in Ecology and Environment 7, 185–189.
Brunner, I., Graf-Pannatier, E., Frey, B., Rigling, A., Landolt, W., Dobbertin, M., 2009.
Morphological and physiological responses of Scots pine fine roots to water
supply in a climatic dry area in Switzerland. Tree Physiology 29, 542–550.
Bugmann, H.K.M., Wullschleger, S.D., Price, D.T., Ogle, K., Clark, D.F., Solomon, A.M.,
2001. Comparing the performance of forest gap models in North America.
Climatic Change 51, 349–388.
Burkett, V.R., Wilcox, D.A., Stottlemyer, R., Barrow, W., Fagre, D., Baron, J., Price, J.,
Nielsen, J.L., Allen, C.D., Peterson, D.L., Ruggerone, G., Doyle, T., 2005. Nonlinear
dynamics in ecosystem response to climatic change: case studies and policy
implications. Ecological Complexity 2, 357–394.
Caldeira, M.C., Fernandéz, V., Tomé, J., Pereira, J.S., 2002. Positive effect of drought on
longicorn borer larval survival and growth on eucalyptus trunks. Annals of
Forest Science 59, 99–106.
Cech, T., Perny, L.B., 2000. Kiefernsterben in Tirol. Forstschutz-aktuell 22, 12–15.
Cech, T., Tomiczek, C., 1996. Zum Kiefernsterben in Niederösterreich. Forstschutzaktuell 17/18, 12–13.
Chapin III, F.S., Randerson, A.D., McGuire, A.D., Foley, J.A., Field, C.B., 2008. Changing
feedbacks in the climate-biosphere system. Frontiers in Ecology and the Environment 6, 313–320.
Chazdon, R.L., Brenes, A.R., Alvarado, B.V., 2005. Effects of climate and stand age on
annual tree dynamics in tropical second-growth rain forests. Ecology 86, 1808–
1815.
Chenchouni, H., Abdelkrim, S.B., Athmane, B., 2008. The deterioration of the Atlas
Cedar (Cedrus atlantica) in Algeria. Oral presentation at: International Conference ‘‘Adaptation of Forests and Forest Management to Changing Climate with
Emphasis on Forest Health: A Review of Science, Policies, and Practices’’, Umeå,
Sweden: FAO/IUFRO, 25–28 August 2008.
Christensen, J.H., Hewitson, B., Busuioc, A., Chen, A., Gao, X., Held, I., Jones, R., Kolli,
R.K., Kwon, W.-T., Laprise, R., Magaña Rueda, V., Mearns, L., Menéndez, C.G.,
Räisänen, J., Rinke, A., Sarr, A., Whetton, P., 2007. Regional climate projections.
In: Solomon, S., et al. (Eds.), Climate Change 2007: The Physical Science Basis.
Contributions of Working Group I to the Fourth Assessment Report of the
Intergovernmental Panel on Climate Change. Cambridge University Press,
Cambride, United Kingdom/New York, NY.
Chuprov, N.P., 2007. The problem of dying spruce stands in forests of the Russian
European North. In: Dying Spruce Forests of Arkhangelsk Region, Problems and
Means of their Solution, Department of Forest Complex of Arkhangelsk Region,
Arkhangelsk, Russian Federation, pp. 66–71.
Ciais, P., Reichstein, M., Viovy, N., Granier, A., Ogee, J., Allard, V., Aubinet, M.,
Buchmann, N., Bernhofer, C., Carrara, A., Chevallier, F., De Noblet, N., Friend,
A.D., Friedlingstein, P., Grunwald, T., Heinesch, B., Keronen, P., Knohl, A.,
Krinner, G., Loustau, D., Manca, G., Matteucci, G., Miglietta, F., Ourcival, J.M.,
Papale, D., Pilegaard, K., Rambal, S., Seufert, G., Soussana, J.F., Sanz, M.J.,
Schulze, E.D., Vesala, T., Valentini, R., 2005. Europe-wide reduction in primary
productivity caused by the heat and drought in 2003. Nature 437 (7058), 529–
533.
Ciesla, W.M., Donaubauer, M.E., 1994. Decline and dieback of trees and forests: a
global overview. FAO Forestry Paper 120, 90.
Clinton, B.D., Boring, L.R., Swank, W.T., 1993. Canopy gap characteristics and
drought influences in oak forests of the Coweeta Basin. Ecology 74, 1551–1558.
Condit, R., Hubbell, S.P., Foster, R.B., 1995. Mortality-rates of 205 neotropical tree
and shrub species and the impact of a severe drought. Ecological Monographs
65, 419–439.
Damesin, C., 2003. Respiration and photosynthesis characteristics of current-year
stems of Fagus sylvatica: from the seasonal pattern to an annual balance. New
Phytologist 158, 465–475.
Delatour, C., 1983. Les dépérissements de chênes en Europe (Oak die-back in
Europe). Revue forestière française 35 (4), 265–282.
Desprez-Loustau, M.-L., Marcais, B., Nageleisen, L.-M., Piou, D., Vannini, A., 2006.
Interactive effects of drought and pathogens in forest trees. Annals of Forensic
Science 63, 597–612.
Dobbertin, M., Rigling, A., 2006. Pine mistletoe (Viscum album ssp. austriacum)
contributes to Scots pine (Pinus sylvestris) mortality in the Rhone valley of
Switzerland. Forest Pathology 36, 309–322.
Dobbertin, M., Mayer, P., Wohlgemuth, T., Feldmeyer-Christe, E., Graf, U., Zimmermann, N.E., Rigling, A., 2005. The decline of Pinus sylvestris L. forests in the swiss
Rhone Valley—a result of drought stress? Phyton-Annales Rei Botanicae 45,
153–156.
Dobbertin, M., Wermelinger, B., Bigler, C., Buergi, M., Carron, M., Forster, B., Gimmi,
U., Rigling, A., 2007. Linking increasing drought stress to Scots pine mortality
and bark beetle infestations. The Scientific World Journal 7, 231–239.
El Abidine, A.Z., 2003. Forest decline in Morocco: causes and control strategy.
Science et changements planétaires/Sécheresse 14, 209–218.
Elliott, K.J., Swank, W.T., 1994. Impacts of drought on tree mortality and basal area
growth in a mixed hardwood forest of the Coweeta Basin. Journal of Vegetation
Science 5, 229–236.
Ermolenko, A., 2008. Climate change and mass-scale forest dieback: regional,
national and international aspects. Oral presentation at: International Conference ‘‘Adaptation of Forests and Forest Management to Changing Climate with
Emphasis on Forest Health: A Review of Science, Policies, and Practices’’, Umeå,
Sweden: FAO/IUFRO, 25–28 August 2008.
Faber-langendoen, D., Tester, J.R., 1993. Oak mortality in sand savannas following
drought in East-Central Minnesota. Bulletin of the Torrey Botanical Club 120,
248–256.
FAO, 2006. Global forest resources assessment 2005—progress towards sustainable
forest management. FAO Forestry Paper No. 147. Rome.
FAO, 2007. Forest monitoring and assessment for climate change reporting: partnerships, capacity building and delivery. Holmgren, P., L.-G. Marklund, M. Saket,
M.L. Wilkie. FAO Forest Resources Assessment Working Paper No. 142. Rome.
Fensham, R.J., 1998. The influence of cattle grazing on tree mortality after drought in
savanna woodland in north Queensland. Australian Journal of Ecology 23, 405–
407.
Fensham, R.J., Fairfax, R.J., 2007. Drought-related tree death of savanna eucalypts:
species susceptibility, soil conditions and root architecture. Journal of Vegetation Science 18, 71–80.
Fensham, R.J., Fairfax, R.J., 2005. Preliminary assessment of gidgee (Acacia cambagei)
woodland thickening in the Longreach district, Queensland. The Rangeland
Journal 27, 159–168.
Fensham, R.J., Holman, J.E., 1999. Temporal and spatial patterns in drought-related
tree dieback in Australian savanna. Journal of Applied Ecology 36, 1035–1050.
Fensham, R.J., Fairfax, R.J., Butler, D.W., Bowman, D.M.J.S., 2003. Effects of fire and
drought in a tropical eucalypt savanna colonized by rain forest. Journal of
Biogeography 30, 1405–1414.
Fensham, R.J., Fairfax, R.J., Ward, D.P., 2009. Drought-induced tree death in savanna.
Global Change Biology 15, 380–387.
Ferrell, G.T., 1996. The influence of insect pests and pathogens on Sierra Forests. In:
Sierra Nevada Ecosystem Project: Final Report to Congress, vol. II, Assessments
and Scientific Basis for Management Options. Univ. of California, Davis, Water
Resources Center Report No. 37, pp. 1177–1192.
Ferrell, G.T., Otrosina, W.J., Demars, C.J., 1994. Predicting susceptibility of white fir
during a drought-associated outbreak of the fir engraver, Scolytus ventralis, in
California. Can. J. For. Res. 24, 302–305.
Fisher, M., 1997. Decline in the juniper woodlands of Raydah reserve in southwestern Saudi Arabia: a response to climate changes? Global Ecology and
Biogeography Letters 6, 379–386.
Fisher, M., Gardner, A.S., 1995. The status and ecology of a Juniperus-Excelsa subsp.
Polycarpos woodland in the northern mountains of Oman. Vegetatio 119, 33–51.
Floyd, M.L., Clifford, M., Cobb, N.S., Hanna, D., Delph, R., Ford, P., Turner, D., 2009.
Relationship of stand characteristics to drought-induced mortality in three
Southwestern piñon–juniper woodlands. Ecological Applications 19 (5), 1223–
1230.
Foden, W., Midgley, G.F., Hughes, G., Bond, W.J., Thuiller, W., Hoffman, M.T., Kaleme,
P., Underhill, L.G., Rebelo, A., Hannah, L., 2007. A changing climate is eroding the
geographical range of the Namib Desert tree Aloe through population declines
and dispersal lags. Diversity and Distributions 13, 645–653.
Forster, B., Meier, F., Braendli, U.-B., 2008. Deutlicher Rückgang der Fichten im
Mittelland. Vorratsabbau - auch durch Sturm und Käfer. Wald Holz 89 (3), 52–
54.
Fowler, D., Cape, J.N., Coyle, M., Flechard, C., Kuylenstierna, J., Hicks, K., Derwent, D.,
Johnson, C., Stevenson, D., 1999. The global exposure of forests to air pollutants.
Water, Air, and Soil Pollution 116, 5–32.
Franklin, J.F., Shugart, H.H., Harmon, M.E., 1987. Tree death as an ecological process.
Bioscience 27, 259–288.
French Forest Health Department (Département Santé des Forêts), 1998–1999,
2003–2008 Annual Reports.
Gan, J.B., 2004. Risk and damage of southern pine beetle outbreaks under global
climate change. Forest Ecology and Management 191, 61–71.
Gardner, A.S., Fisher, M., 1996. The distribution and status of the montane juniper
woodlands of Oman. Journal of Biogeography 23, 791–803.
Garrett, K.A., Dendy, S.P., Frank, E.E., Rouse, M.N., Travers, S.E., 2006. Climate change
effects on plant disease: genomes to ecosystems. Annual Review of Phytopathology 44, 489–509.
Gitlin, A.R., Sthultz, C.M., Bowker, M.A., Stumpf, S., Paxton, K.L., Kennedy, K., Munoz,
A., Bailey, J.K., Whitham, T.G., 2006. Mortality gradients within and among
dominant plant populations as barometers of ecosystem change during
extreme drought. Conservation Biology 20, 1477–1486.
Gonzalez, P., 2001. Desertification and a shift of forest species in the West African
Sahel. Climate Research 17, 217–228.
Grant, P.J., 1984. Drought effect on high-altitude forests, Ruahine Range, North
Island, New Zealand. New Zealand Journal of Botany 22 (1), 15–27.
Greenwood, D.L., Weisberg, P.J., 2008. Density-dependent tree mortality in pinyon–
juniper woodlands. Forest Ecology and Management 255, 2129–2137.
Grulke, N.E., Paine, T., Minnich, R., Chavez, D., Riggan, P., Dunn, A., 2009. Air
pollution increases forest susceptibility to wildfire. In: Bytnerowicz, A., Arbaugh, M., Riebau, A., Andersen, C. (Eds.), Wildland Fires and Air Pollution.
Developments in Environmental Science, vol. 8. Elsevier Publishers, The Hague,
Netherlands, pp. 365–403.
77
C.D. Allen et al. / Forest Ecology and Management 259 (2010) 660–684
Guarin, A., Taylor, A.H., 2005. Drought triggered tree mortality in mixed conifer
forests in Yosemite National Park, California, USA. Forest Ecology and Management 218, 229–244.
Gutschick, V.P., BassiriRad, H., 2003. Extreme events as shaping physiology, ecology,
and evolution of plants: toward a unified definition and evaluation of their
consequences. New Phytologist 160, 21–42.
Hamann, A., Wang, T., 2005. Models of climatic normals for genecology and climate
change studies in British Columbia. Agricultural and Forest Meteorology 128,
211–221.
Hanson, P.J., Weltzin, J.F., 2000. Drought disturbance from climate change: response
of United States forests. Science of the Total Environment 262, 205–220.
Heitzman, E., Muzika, R.M., Kabrick, J., Guldin, J.M., 2004. Assessment of oak decline
in Missouri, Arkansas, and Oklahoma. In: Yaussy, D.A., et al. (Eds.), Proceedings
of the 14th Central Hardwood Forest Conference, Wooster, OH, 16–19 March
2004. Gen. Tech. Rep. NE-316. USDA Forest Service, Northeastern Research
Station, Newtown Square, PA pp. 510.
Hendershot, W.H., Jones, A.R.C., 1989. Maple decline in Quebec: a discussion of
possible causes and the use of fertilizers to limit damage. Forestry Chronicle 65
(4), 280–287.
Hennon, P.E., Shaw, C.G., 1997. The enigma of yellow-cedar decline—What is killing
these long-lived, defensive trees? Journal of Forestry 95, 4–10.
Hennon, P.E., D’Amore, D.V., Zeglen, S., Grainger, M., 2005. Yellow-cedar decline in
the North Coast Forest District of British Columbia. In: USDA Forest Service
Research Note PNW-RN-549, Pacific Northwest Research Station, 16 pp.
Hicke, J.A., Logan, J.A., Powell, J., Ojima, D.S., 2006. Changing temperatures influence
suitability for modeled mountain pine beetle (Dendroctonus ponderosae) outbreaks in the western United States. Journal of Geophysical Research 111,
G02019 doi:10.1029/2005JG000101.
Hogg, E.H., Brandt, J.P., Kochtubajda, B., 2002. Growth and dieback of Aspen forests in
northwestern Alberta, Canada, in relation to climate and insects. Canadian Journal
of Forest Research-Revue Canadienne De Recherche Forestiere 32, 823–832.
Hogg, E.H., Brandt, J.P., Michaellian, M., 2008. Impacts of a regional drought on the
productivity, dieback, and biomass of western Canadian aspen forests. Canadian
Journal of Forest Research-Revue Canadienne De Recherche Forestiere 38,
1373–1384.
Holst, J., Barnard, R., Brandes, E., Buchmann, N., Gessler, A., Jaeger, L., 2008. Impacts
of summer water limitation on the carbon balance of a Scots pine forest in the
southern upper Rhine plain. Agricultural and Forest Meteorology 148, 1815–
1826.
Horner, G.J., Baker, P.J., MacNally, R., Cunningham, S.C., Thomson, J.R., Hamilton, F.,
2009. Mortality of developing floodplain forests subjected to a drying climate
and water extraction. Global Change Biology 15, 2176–2186.
Hosking, G.P., Hutcheson, J.A., 1988. Mountain beech (Nothofagus-Solandri var.
Cliffortioides) decline in the Kaweka Range, North Island, New Zealand. New
Zealand Journal of Botany 26, 393–400.
Hosking, G.P., Kershaw, D.J., 1985. Red Beech death in the Maruia Valley South
Island, New Zealand. New Zealand Journal of Botany 23, 201–211.
Huntingford, C., Fisher, R.A., Mercado, L., Booth, B.B.B., Sitch, S., Harris, P.P., Cox, P.M.,
Jones, C.D., Betts, R.A., Malhi, Y., Harris, G., Collins, M., Moorcroft, P., 2008.
Towards quantifying uncertainty in predictions of Amazon ‘‘Die-back’’. Philosophical Transactions of the Royal Society B-Biological Sciences 363 (1498),
1857–1864.
Hursh, C.R., Haasis, F.W., 1931. Effects of 1925 summer drought on Southern
Appalachian hardwoods. Ecology 12, 380–386.
Huxman, T.E., Wilcox, B.P., Breshears, D.D., Scott, R., Snyder, K., Small, E.A., Hultine,
K., Pockman, W., Jackson, R.B., 2005. Woody plant encroachment and the water
cycle: an ecohydrological framework. Ecology 86, 308–319.
IPCC, 2007a. Climate change 2007: the physical science basis. In: Solomon, S., Qin,
D., Manning, M., Chen, Z., Marquis, M., Averyt, K.B., Tignor, M., Miller, H.L. (Eds.),
Contribution of Working Group I to the Fourth Assessment. Report of the
Intergovernmental Panel on Climate Change. Cambridge University Press, Cambridge, United Kingdom/New York, NY, USA, 996 pp.
IPCC, 2007b. Climate change 2007: impacts, adaptation and vulnerability. Parry,
M.L., Canziani, O.F., Palutikof, J.P., van der Linden, P.J., Hanson, C.E. (eds.),
Contribution of Working Group II to the Fourth Assessment Report of the
Intergovernmental Panel on Climate Change. Cambridge University Press, Cambridge, UK, 976 pp.
Jenkins, M.A., Pallardy, S.G., 1995. The influence of drought on red oak group species
growth and mortality in the Missouri Ozarks. Canadian Journal of Forest
Research-Revue Canadienne De Recherche Forestiere 25, 1119–1127.
Jentsch, A., Kreyling, J., Beierkuhnlein, C., 2007. A new generation of climate change
experiments: events, not trends. Frontiers in Ecology and the Environment 5,
365–374.
Jones, C., Lowe, J., Liddicoat, S., Betts, R., 2009. Committed terrestrial ecosystem
changes due to climate change. Nature GeoScience 2, 484–487.
Joyce, L.A., Blate, G.M., Littell, J.S., McNulty, S.G., Millar, C.I., Moser, S.C., Neilson, R.P.,
O’Halloran, K., Peterson, D.L., 2008. Chapter 3, National forests. In: Preliminary
Review of Adaptation Options for Climate-Sensitive Ecosystems and Resources.
A Report by the U.S. Climate Change Science Program and the Subcommittee on
Global Change Research. U.S. Environmental Protection Agency, Washington,
DC, pp. 3-1–3-127.
Jump, A.S., Mátyás, C., Peñuelas, J. 2009. The altitude-for-latitude disparity in the
range retractions of woody species. Trends in Ecology and Evolution,
doi:10:1016/j.tree.2009.06.007.
Kailidis, D.L., Markalas, S., 1990. Dryness and the most destructive secondary bark
beetle epidemic on fir in Greece. Ecotopia 8, 38–41.
681
Karnosky, D.F., Pregitzer, K.S., Zak, D.R., Kubiske, M.E., Hendrey, G.R., Weinstein, D.,
Nosal, M., Percy, K.E., 2005. Scaling ozone responses of forest trees to the
ecosystem level in a changing climate. Plant, Cell and Environment 28, 965–
981.
Kauhanen, H., Wallenius, T., Kuuluvainen, T., Aakala, T., Mikkola, K., 2008. Extensive
mortality of spruce forests in Arkhangelsk Region: satellite image analysis. In:
Poster Presentation At: International Conference ‘‘Adaptation of Forests and
Forest Management to Changing Climate with Emphasis on Forest Health: A
Review of Science, Policies, and Practices’’, Umeå, Sweden, FAO/IUFRO, 25–28
August 2008.
Keane, R.E., Austin, M., Field, C., Huth, A., Lexer, M.J., Peters, D., Solomon, A., Wyckoff,
P., 2001. Tree mortality in gap models: application to climate change. Climatic
Change 51, 509–540.
Kessler Jr., K.J., 1989. Some perspectives on oak decline in the 80’s. In: Rink, G.,
Budelsky, C.A. (Eds.), Proceedings of the Seventh Central Hardwood Conference.
Gen. Tech. Rep. NC-132, U.S. Department of Agriculture, Forest Service, North
Central Research Station, St. Paul, MN, pp. 25–29.
Khan, J.A., Rodgers, W.A., Johnsingh, A.J.T., Mathur, P.K., 1994. Tree and shrub
mortality and debarking by Sambar Cervus-Unicolor (Kerr) in Gir After a drought
in Gujarat, India. Biological Conservation 68, 149–154.
Kienast, F., Flühler, H., Schweingruber, F.H., 1981. Jahrringanalysen an Föhren (Pinus
sylvestris L.) aus immissionsgefährdeten Beständen des Mittelwallis (Saxon,
Schweiz). Waldschäden im Walliser Rhonetal (Schweiz). Mitteilungen Eidg.
Anstalt für das forstliche Versuchswesen 57, 415–432.
Kinnaird, M.F., O’Brien, T.G., 1998. Ecological effects of wildfire on lowland rainforest in Sumatra. Conservation Biology 12, 954–956.
Kobelkov, M., 2008. National program on monitoring of large-area decline of boreal
and temperate forests and minimization of its consequences with purpose of
integration with the international plans of actions in connection with climate
change. Oral presentation at: International Conference ‘‘Adaptation of Forests
and Forest Management to Changing Climate with Emphasis on Forest Health: A
Review of Science, Policies, and Practices’’, FAO/IUFRO, Umeå, Sweden, 25–28
August 2008.
Kloeppel, B.D., Clinton, B.D., Vose, J.M., Cooper, A.R., 2003. Drought impacts on tree
growth and mortality of southern appalachian forests (pp. 43–55). In: Greenland, D., Goodin, D.G., Smith, R.C. (Eds.), Climate Variability and Ecosystem
Respnse at Long-Term Ecological Research Sites. Oxford, Univ. Press, NY, p. 459
pp.
Klos, R.J., Wang, G.G., Bauerle, W.L., Rieck, J.R., 2009. Drought impact on forest
growth and mortality in the southeast USA: an analysis using Forest Health and
Monitoring data. Ecological Applications 19, 699–708.
Körner, C., Sarris, D., Christodoulakis, D., 2005. Long-term increase in climatic
dryness in the East-Mediterranean as evidenced for the island of Samos.
Regional Environmental Change Journal 5, 27–36.
Krotov, N.S., 2007. On problems of spruce forest mortality in the Arkhangelsk
Region. In: Dying Spruce Forests of Arkhangelsk Region. Problems and Means
of their Solution, Department of Forest Complex of Arkhangelsk Region,
Arkhangelsk, Russian Federation, pp. 6–11.
Kurz, W.A., Dymond, C.C., Stinson, G., Rampley, G.J., Neilson, E.T., Carroll, A.L., Ebata,
T., Safranyik, L., 2008a. Mountain pine beetle and forest carbon feedback to
climate change. Nature 452, 987–990.
Kurz, W.A., Stinson, G., Rampley, G.J., Dymond, C.C., Neilson, E.T., 2008b. Risk of
natural disturbances makes future contribution of Canada’s forests to the global
carbon cycle highly uncertain. Proceedings of the National Academy of Sciences
of the United States of America 105, 1551–1555 doi:10.1073/pnas.0708133105.
Landmann, G., Dreyer, E. (Eds.), 2006. Impacts of drought and heat on forest.
Synthesis of available knowledge, with emphasis on the 2003 event in Europe.
Annals of Forest Science 3 (6) 567–652.
Law, J.R., Gott, J.D., 1987. Oak mortality in the Missouri Ozarks. In: Hay, R.L., et, al.
(Eds.), Proceedings of the Sixth Central Hardwood Forest Conference, The
University of Tennessee, Knoxville, TN, pp. 427–436.
Laurance, W.F., Williamson, G.B., Delamônica, P., Oliveira, A., Lovejoy, T.E., Gascon,
C., Pohl, L., 2001. Effects of a strong drought on Amazonian forest fragments and
edges. Journal of Tropical Ecology 17, 771–785.
Lawrence, R., Moltzan, B., Moser, K., 2002. Oak decline and the future of Missouri’s
forests. Missouri Conservationist 63, 11–18.
Leigh, E.G.J., Windsor, D.M., Rand, A.S., Foster, R.B., 1990. The impact of the El Niño
drought of 1982–83 on a Panamanian semideciduous forest. In: Glynn, P.W.
(Ed.), Global Ecological Consequences of the 1982–83 El Nino-Southern Oscillation. Elsevier, Amsterdam, pp. 473–486.
Leighton, M., Wirawan, N., 1986. Catastrophic drought and fire in Borneo tropical
rain forest associated with the 1982–1983 El Nino southern oscillation event.
In: Prance, G.T. (Ed.), Tropical Rain Forests and the World Atmosphere. Westview Press, Boulder, CO, USA, pp. 75–102.
Lenton, T.M., Held, H., Kriegler, E., Hall, J.W., Lucht, W., Rahmstorf, W., Schellnhuber,
H.J., 2008. Tipping elements in the Earth’s climate system. Proceedings of the
National Academy of Sciences of the United States of America 105, 1786–1793.
Lewis, S.L., 2006. Tropical forests and the changing earth system. Philosophical
Transactions of the Royal Society B-Biological Sciences 361, 195–210
doi:10.1098/rstb.2005.1711.
Li, L.S., 2003. Pine shoot beetle. In: Zhang, X., Luo, Y. (Eds.), Major Forest Disease and
Insect Pests in China. Beijing Forestry Publishing House, Beijing, China, pp. 217–
226.
Liang, E.Y., Shao, X.M., Kong, Z.C., Lin, J.X., 2003. The extreme drought in the 1920s
and its effect on tree growth deduced from tree ring analysis: a case study in
North China. Annals of Forest Science 60 (2), 145–152.
78
682
C.D. Allen et al. / Forest Ecology and Management 259 (2010) 660–684
Lim, J.H., Chun, J.H., Woo, S.Y., Kim, Y.K., 2008. Increased declines of Korean fir forest
caused by climate change in Mountain Halla, Korea. In: Oral Presentation At:
International Conference ‘‘Adaptation of Forests and Forest Management to
Changing Climate with Emphasis on Forest Health: A Review of Science,
Policies, and Practices’’, Umeå, Sweden, FAO/IUFRO, 25–28 August 2008.
Lingenfelder, M., Newbery, D.M., 2009. On the detection of dynamic responses in a
drought-perturbed tropical rainforest in Borneo. Plant Ecology 201, 267–290.
Lloret, F., Siscart, D., 1995. Los efector demograficos de la sequia en poblaciones de
encina. Cuadernos de la Sociedad Espanola de Ciencias Forestales 2, 77–81.
Lloret, F., Siscart, D., Dalmases, C., 2004. Canopy recovery after drought dieback in
holm-oak Mediterranean forests of Catalonia (NE Spain). Global Change Biology
10, 2092–2099.
Lloyd, A.H., Bunn, A.G., 2007. Responses of the circumpolar boreal forest to 20th
century climate variability. Environmental Research Letters 2, 045013,
doi:10:1088/1748-9326/2/4/045013.
Loehle, C., LeBlanc, D., 1996. Model-based assessments of climate change effects on
forests: a critical review. Ecological Modelling 90, 1–31.
Logan, J., Regniere, J., Powell, J.A., 2003. Assessing the impacts of global warming on
forest pest dynamics. Frontiers in Ecology and the Environment 1, 130–137.
Lu, J., Deser, C., Reichler, T., 2009. Cause of the widening of the tropical belt since
1958. Geophysical Research Letters 36, L03803.
Lucht, W., Schaphoff, S., Erbrecht, T., Heyder, U., Cramer, W., 2006. Terrestrial
vegetation redistribution and carbon balance under climate change. Carbon
Balance and Management 1, 6 doi:10.1186/1750-0680-1-6.
Lwanga, J.S., 2003. Localized tree mortality following the drought of 1999 at Ngogo,
Kibale National Park, Uganda. African Journal of Ecology 41, 194–196.
MacGregor, S.D., O’Connor, T.G., 2002. Patch dieback of Colophospermum mopane in
a dysfunctional semi-arid African savanna. Austral Ecology 27, 385–395.
Macomber, S.A., Woodcock, C.E., 1994. Mapping and monitoring conifer mortality
using remote sensing in the Lake Tahoe Basin. Remote Sensing of Environment
50, 255–266.
Maherali, H., Pockman, W.T., Jackson, R.B., 2004. Adaptive variation in the vulnerability of woody plants to xylem cavitation. Ecology 85, 2184–2199.
Man’ko, U.I., Gladkova, G.A., 2001. Spruce Mortality in the Light of the Global
Decline of Dark Coniferous Forests. Russian Academy of Sciences, Far East
Branch, Vladivostok, 228 pp.
Manion, P.D., 1991. Tree Disease Concepts, 2nd ed. Prentice-Hall Inc., Upper Saddle
River, NJ, 416 pp.
Manion, P.D., Lachance, D., 1992. Forest Decline Concepts. APS Press, St. Paul, MN,
249 pp.
Markalas, S., 1992. Site and stand factors related to mortality-rate in a fir forest after
a combined incidence of drought and insect attack. Forest Ecology and Management 47, 367–374.
Martinez-Vilalta, J., Piñol, J., 2002. Drought-induced mortality and hydraulic architecture in pine populations of the NE Iberian Peninsula. Forest Ecology and
Management 161, 247–256.
Mattson, W.J., Haack, R.A., 1987. The role of drought in outbreaks of plant-eating
insects. BioScience 37, 110–118.
McDowell, N., Pockman, W.T., Allen, C.D., Breshears, D.D., Cobb, N., Kolb, T., Sperry, J.,
West, A., Williams, D., Yepez, E.A., 2008. Mechanisms of plant survival and
mortality during drought: why do some plants survive while others succumb to
drought? Tansley review. New Phytologist 178, 719–739.
McDowell, N., Allen, C.D., Marshall, L., 2009. Growth, carbon isotope discrimination,
and climate-induced mortality across a Pinus ponderosa elevation transect.
Global Change Biology, doi:10.1111/j.1365-2486.2009.01994.x.
Miao, S.L., Zou, C.B., Breshears, D.D., 2009. Vegetation responses to extreme hydrological events: sequence matters. American Naturalist 173, 113–118.
Millar, C.I., Westfall, R.D., Delany, D.L., 2007a. Response of high-elevation limber
pine (Pinus flexilis) to multiyear droughts and 20th-century warming, Sierra
Nevada, California, USA. Canadian Journal of Forest Research-Revue Canadienne
De Recherche Forestiere 37, 2508–2520.
Millar, C.I., Stephenson, N.L., Stephens, S.L., 2007b. Climate change and forests of the
future: managing in the face of uncertainty. Ecological Applications 17, 2145–
2151.
Millennium Ecosystem Assessment, 2005. Ecosystems and Human Well-being:
Synthesis. Island Press, Washington, DC, 137 pp.
Millers, I., Shriner, D.S., Rizzo, D. 1989. History of hardwood decline in the Eastern
United States. Gen. Tech. Rep. NE-126. U.S. Department of Agriculture, Forest
Service, Northeastern Forest Experiment Station, Broomall, PA, pp. 75.
Minerbi, S., 1993. Wie gesund sind unsere Wälder? 10. Bericht über den Zustand der
Wälder in Südtirol. Agrar- und Forstbericht, Autonome Provinz Bozen, Assessorate für Land-und Forstwirtschaft, pp 40.
Minnesota Department of Natural Resources. 2007. Federal conditions report 2007.
24 pp. http://fhm.fs.fed.us/fhh/fhh_07/mn_fhh_07.pdf
Mueller, R.C., Scudder, C.M., Porter, M.E., Trotter, R.T., Gehring, C.A., Whitham, T.G.,
2005. Differential tree mortality in response to severe drought: evidence for
long-term vegetation shifts. Journal of Ecology 93, 1085–1093.
Mueller-Dombois, D., 1986. Perspectives for an etiology of stand-level dieback.
Annual Review of Ecology and Systematics 17, 221–243.
Mueller-Dombois, D., 1988. Towards a unifying theory for stand-level dieback.
GeoJournal 17, 249–251.
Nageleisen, L.-M., Hartmann, G., Landmann, G., 1991. Dépérissements d’essences
feuillues en Europe Occidentale: cas particulier des chênes rouvre et pédonculé
Revue Forestière Française n8 hors série, vol. 2, pp. 301–306.
Nageleisen, L.-M., 1994. Dépérissement actuel des chênes. Revue Forestière
Française 46 (5), 504–511.
Nakagawa, M., Tanaka, K., Nakashizuka, T., Ohkubo, T., Kato, T., Maeda, T., Sato, K.,
Miguchi, H., Nagamasu, H., Ogino, K., Teo, S., Hamid, A.A., Seng, L.H., 2000.
Impact of severe drought associated with the 1997–1998 El Nino in a tropical
forest in Sarawak. Journal of Tropical Ecology 16, 355–367.
Navarro-Cerrillo, R., Varo, M.A., Lanjeri, S., Hernández-Clemente, R., 2007. Cartografı́a de defoliación en los pinares de pino silvestre (Pinus sylvestris L.) y pino
salgareño (Pinus nigra Arnold.) en la Sierra de los Filabres. Ecosistemas. 2007/3.
url: http://www.revistaecosistemas.net/articulo.asp?Id=495.
Negron, J.F., McMillin, J.D., Anhold, J.A., Coulson, D., 2009. Bark beetle-caused
mortality in a drought-affected ponderosa pine landscape in Arizona, USA.
Forest Ecology and Management 257, 1353–1362.
Nepstad, D.C., Tohver, I.M., Ray, D., Moutinho, P., Cardinot, G., 2007. Mortality of
large trees and lianas following experimental drought in an amazon forest.
Ecology 88, 2259–2269.
Nepstad, D.C., Stickler, C.M., Soares-Filho, B., Merry, F., 2008. Interactions among
Amazon land use, forests and climate: prospects for a near-term forest tipping
point. Philosophical Transactions of the Royal Society B-Biological Sciences 363,
1737–1746.
Newbery, D.M., Lingenfelder, M., 2009. Plurality of tree species responses to drought
perturbation in Bornean tropical rain forest. Plant Ecology 201, 147–167.
Nishimua, T.B., Suzuki, E., Kohyama, T., Tsuyuzaki, S., 2007. Mortality and growth of
trees in peat-swamp and heath forests in Central Kalimantan after severe
drought. Plant Ecology 188, 165–177.
O’Connor, T.G., 1999. Impact of sustained drought on a semiarid Colophospermum
mopane savanna. African Journal of Forage Science 15, 83–91.
Oak, S.W., Steinman, J.R., Starkey, D.A., Yocey, E.K., 2004. Assessing oak decline
incidence and distribution in the southern U.S. using forest inventory
and analysis data. In: Spetich, M.A. (Ed.), Upland Oak Ecology Symposium:
History, Current Conditions, and Sustainability, Gen. Tech. Rep. SRS-73,
USDA Forest Service, Southern Research Station, Asheville, NC, pp. 236–
242.
Oberhuber, W., 2001. The role of climate in the mortality of Scots pine (Pinus
sylvestris L.) exposed to soil dryness. Dendrochronologia 19, 45–55.
Ogaya, R., Penuelas, J., 2007. Tree growth, mortality, and above-ground biomass
accumulation in a holm oak forest under a five-year experimental field drought.
Plant Ecology 189, 291–299.
Ogle, K., Whitham, T.G., Cobb, N.S., 2000. Tree-ring variation in pinyon predicts
likelihood of death following severe drought. Ecology 81, 3237–3243.
Olano, J.M., Palmer, M.W., 2003. Stand dynamics of an Appalachian old-growth
forest during a severe drought episode. Forest Ecology and Management 174,
139–148.
Ogibin B.N., Demidova, N.A. Successional dynamics of old-growth spruce forests in
the watersheds of the rivers Northern Dvina—Pinega in the Arkhangelsk Region.
In: Kauhanen, H., Neshataev, V., Vuopio, M. (Eds.), Northern Coniferous Forests—From Research to Ecologically Responsible Forestry. Finnish Forest
Research Institute, Helsinki, in press.
Ollinger, S.V., Goodale, C.L., Hayhoe, K., Jenkins, J.P., 2008. Effects of predicted
changes in climate and atmospheric composition on ecosystem processes in
northeastern U.S. forests. Mitigation and Adaptation Strategies for Global
Change 13, 467–485.
Parmesan, C., 2006. Ecological and evolutionary responses to recent climate change.
Annual Review of Ecology, Evolution, and Systematics 37, 637–669.
Payette, S., Fortin, M.J., Morneau, C., 1996. The recent sugar maple decline in
southern Quebec: probable causes deduced from tree rings. Canadian Journal
of Forest Research-Revue Canadienne De Recherche Forestiere 26, 1069–
1078.
Pedersen, B.S., 1998. The role of stress in the mortality of Midwestern oaks as
indicated by growth prior to death. Ecology 79, 79–93.
Pedersen, B.S., 1999. The mortality of midwestern overstory oaks as a bioindicator
of environmental stress. Ecological Applications 9, 1017–1027.
Peñuelas, J., Lloret, F., Montoya, R., 2001. Severe drought effects on mediterranean
woody flora in Spain. Forest Science 47, 214–218.
Petercord, R., 2008. Zukünftige gefährdung der Rotbuche durch rinden- und holzbrütende Käfer in Baden-Württemberg. Mitteilungen der Deutsche Gesellschaft
für Allgemeine und Angewandte Entomologie 16, 247–250.
Phillips, O.L., Lewis, S.L., Baker, T.R., Chao, K.-J., Higuchi, N., 2008. The changing
Amazon forest. Philosophical Transactions of the Royal Society B-Biological
Sciences 363, 1819–1827.
Phillips, O.L., Aragão, L.E.O.C., Lewis, S.L., Fisher, J.B., Lloyd, J., López-González, G.,
Malhi, Y., Monteagudo, A., Peacock, J., Quesada, C.A., van der Heijden, G.,
Almeida, S., Amaral, I., Arroyo, L., Aymard, G., Baker, T.R., Bánki, O., Blanc, L.,
Bonal, D., Brando, P., Chave, J., Alves de Oliveira, Á.C., Dávila Cardozo, N.,
Czimczik, C.I., Feldpausch, T.R., Freitas, M.A., Gloor, E., Higuchi, N., Jiménez,
E., Lloyd, G., Meir, P., Mendoza, C., Morel, A., Neill, D.A., Nepstad, D., Patiño, S.,
Peñuela, M.C., Prieto, A., Ramı́rez, F., Schwarz, M., Silva, J., Silveira, M., Sota
Thomas, A., ter Steege, H., Stropp, J., Vásquez, R., Zelazowski, P., Alvarez Dávila,
E., Andelman, S., Andrade, A., Chao, K., Erwin, T., Di Fiore, A., Honorio, C., Keeling,
E., Killeen, H., Laurance, T.J., Peña Cruz, W.F., Pitman, A., Núñez Vargas, N.C.A.,
Ramı́rez-Angulo, P., Rudas, H., Salamão, A., Silva, R., Terborgh, N., TorresLezama, J.A., 2009. Drought sensitivity of the Amazon rainforest. Science
323, 1344–1347.
Potts, M.D., 2003. Drought in a Bornean everwet rain forest. Journal of Ecology 91,
467–474.
Poupon, H., 1980. Structure et dynamique de la strate ligneuse d’une steppes
Sahelienne au nord du Senegal. Travaux et documents del’ O.R.S.T.O.M. no.
115, O.R.S.T.O.M., Paris.
79
C.D. Allen et al. / Forest Ecology and Management 259 (2010) 660–684
Raffa, K.F., Aukema, B.H., Bentz, B.J., Carroll, A.L., Hicke, J.A., Turner, M.G., Romme,
W.H., 2008. Cross-scale drivers of natural disturbances prone to anthropogenic
amplification: the dynamics of bark beetle eruptions. Bioscience 58, 501–517.
Raftoyannis, Y., Spanos, I., Radoglou, K., 2008. The decline of Greek fir (Abies
cephalonica Loudon): Relationships with root condition. Plant Biosystems
142, 386–390.
Rennenberg, H., Loreto, F., Polle, A., Brilli, F., Fares, S., Beniwal, R.S., Gessler, A., 2006.
Physiological responses of forest trees to heat and drought. Plant Biology 8,
556–571.
Rice, K.J., Matzner, S.L., Byer, W., Brown, J.R., 2004. Patterns of tree dieback in
Queensland, Australia: the importance of drought stress and the role of resistance to cavitation. Oecologia 139, 190–198.
Rich, P.M., Breshears, D.D., White, A.B., 2008. Phenology of mixed woody-herbaceous ecosystems following extreme events: net and differential responses.
Ecology 89, 342–352.
Richardson, D.M., et al., 2009. Multidimensional evaluation of managed relocation.
Proceedings of the National Academy of Sciences of the United States of America
1062 (24), 9721–9724.
Rigling, A., Cherubini, P., 1999. Wieso sterben die Waldföhren im Telwald bei Visp?
Schweizerische Zeitschrift für Forstwesen 150, 113–131.
Rigling, A., Dobbertin, M., Bürgi, M., Feldmeier-Christe, E., Gimmi, U., Ginzler, C.,
Graf, U., Mayer, P., Zweifel, R., Wohlgemuth, T., 2006. Baumartenwechsel in den
Walliser Waldföhrenwäldern. In: Wohlgemuth, T. (Red.), Wald und Klimawandel. Forum für Wissen 2006, 71 pp.
Robitaille, L., Allard, G., Bordeleau, M., Dessureault, M., Gagnon, F., Lachance, D.,
Picher, R., Roberge, M., 1982. Mortalité dans les érablières: tournée du 12, 13 et
14 octobre dans les régions de Québec (Beauce), de l’Estrie et de Trois-Rivières.
Gouv. Du Québec, Min. de l’énergie. et des ressour., Dir. de la rech. for. Rapport
interne n8 227, 33 p.
Rolim, S.G., Jesus, R.M., Nascimento, H.E.M., do Couto, H.T.Z., Chambers, J.Q., 2005.
Biomass change in an Atlantic tropical moist forest: the ENSO effect in permanent sample plots over a 22-year period. Oecologia 142, 238–246.
Rouault, G., Candau, J.N., Lieutier, F., Nageleisen, L.M., Martin, J.C., Warzee, N., 2006.
Effects of drought and heat on forest insect populations in relation to the 2003
drought in Western Europe. Annals of Forest Science 63, 613–624.
Roy, G., Larocque, G.R., Ansseau, C., 2004. Retrospective evaluation of the onset
period of the visual symptoms of dieback in five Appalachian sugar maple stand
types. Forestry Chronicle 80, 375–383.
Ryan, M.G., Phillips, N., Bond, B.J., 2006. The hydraulic limitation hypothesis
revisited. Plant, Cell and Environment 29, 367–381.
Sala, A., Hoch, G., 2009. Height-related growth declines in ponderosa pine are not
due to carbon limitation. Plant, Cell and Environment 32, 22–30.
Sarris, D., Christodoulakis, D., Körner, C., 2007. Recent decline in precipitation and
tree growth in the eastern Mediterranean. Global Change Biology 13 (6), 1187–
1200.
Savage, M., 1997. The role of anthropogenic influences in a mixed-conifer forest
mortality episode. Journal of Vegetation Science 8, 95–104.
Schilli, S., Dobbertin, M., Rigling, A., Bucher, H.U., in press. Waldföhrensterben im
Churer Rheintal – ein Vergleich zum Wallis. Bündnerwald.
Scholze, M., Knorr, W., Arnell, N.W., Prentice, I., 2006. A climate-change risk analysis
for world ecosystems. Proceedings of the National Academy of Sciences of the
United States of America 103, 13116–13120.
Schutt, P., Cowling, E.B., 1985. Waldsterben, a general decline of forests in central
Europe: symptoms, development, and possible causes. Plant Disease 69, 548–558.
Seager, R., Ting, M., Held, I., Kushnir, Y., Lu, J., Vecchi, G., Huang, H.-P., Harnik, N.,
Leetmaa, A., Lau, N.-C., Li, C., Velez, J., Naik, N., 2007. Model projections of an
imminent transition to a more arid climate in southwestern North America.
Science 316, 1181–1184.
Seidel, D.J., Fu, Qiang, Randel, W.J., Reichler, T.J., 2008. Widening of the tropical belt
in a changing climate. Nature Geoscience 1, 21–24.
Semerci, A., Şanlı, B.N., Şahin, Ö., Çelik, O., Balkız, G.B., Ceylan, S., Argun, N., 2008.
Examination of tree mortalities in semi-arid central Anatolian region of Turkey
during last six-year period (2002–2007). Poster presentation at: International
Conference ‘‘Adaptation of Forests and Forest Management to Changing Climate
with Emphasis on Forest Health: A Review of Science, Policies, and Practices’’,
Umeå, Sweden, FAO/IUFRO, 25–28 August 2008.
Seppala, R., Buck, A., Katila, P. (eds.), 2009. Adaptation of Forests and People to
Climate Change—A Global Assessment Report. IUFRO World Series Vol. 22.
International Union of Forest Research Organizations, Helsinki, 224 pp.
Shaw, J.D., Steed, B.E., DeBlander, L.T., 2005. Forest Inventory and Analysis (FIA)
annual inventory answers the question: What is happening to pinyon–juniper
woodlands? Journal of Forestry 103, 280–285.
Shtrakhov, 2008. Forest health and protection in Russia. In: Oral Presentation At:
International Conference ‘‘Adaptation of Forests and Forest Management to
Changing Climate with Emphasis on Forest Health: A Review of Science,
Policies, and Practices’’, Umeå, Sweden, FAO/IUFRO, 25–28 August 2008.
Siwecki, R., Ufnalksi, K., 1998. Review of oak stand decline with special reference to
the role of drought in Poland. European Journal of Forest Pathology 28, 99–112.
Skelly, J.M., Innes, J.L., 1994. Waldsterben in the forests of Central Europe and
Eastern North America—fantasy or reality? Plant Disease 78, 1021–1032.
Slik, J.W.F., 2004. El Nino droughts and their effects on tree species composition and
diversity in tropical rain forests. Oecologia 141, 114–120.
Soja, A.J., Tchebakova, N.M., French, N.H.F., Flannigan, M.D., Shugart, H.H., Stocks,
B.J., Sukhinin, A.I., Varfenova, E.I., Chapin, F.S., Stackhouse Jr., P.W., 2007.
Climate-induced boreal forest change: predictions versus current observations.
Global and Planetary Change 56 (3–4), 274–296.
683
Solberg, S., 2004. Summer drought: a driver for crown condition and mortality of
Norway spruce in Norway. Forest Pathology 34, 93–107.
Starkey, D.A., Oak, S.W., Ryan, G.W., Tainter, F.H., Redmond, C., Brown, H.D., 1989.
Evaluation of oak decline areas in the south. Protection Report R8-PR 17, US
Forest Service, 36 pp.
Starkey, D.A., Oak, S.W., 1989. Site factors and stand conditions associated with
oak decline in southern upland hardwood forests. In: Rink, G., Budelsky, C.A.
(Eds.), Proceedings of the Seventh Central Hardwood Conference, Carbondale, IL, 5–8 March 1989. Gen. Tech. Rep. NC-132. USDA Forest Service, North
Central Forest Experiment Station, St. Paul, MN, pp. 95–102.
Starkey, D.A., Oliveria, F., Mangini, A., Mielke, M., 2004. Oak decline and red oak
borer in the interior highlands of Arkansas and Missouri: natural phenomena,
severe occurrences. In: Spetich, M.A. (Ed.), Upland Oak Ecology Symposium:
History, Current Conditions, and Sustainability, Gen. Tech. Rep. SRS-73, USDA
Forest Service, Southern Research Station, Asheville, NC, pp. 217–222.
Sterl, A., et al., 2008. When can we expect extremely high surface temperatures?
Geophysical Research Letters 35, L14703 doi:10.1029/2008GL034071.
Stringer, J.W., Kimmer, T.W., Overstreet, J.C., Dunn, J.P., 1989. Oak mortality in
eastern Kentucky. Southern Journal of Applied Forestry 13, 86–91.
Suarez, M.L., Kitzberger, T., 2008. Recruitment patterns following a severe drought:
long-term compositional shifts in Patagonian forests. Canadian Journal of Forest
Research 38, 3002–3010.
Suarez, M.L., Ghermandi, L., Kitzberger, T., 2004. Factors predisposing episodic
drought-induced tree mortality in Nothofagus: site, climatic sensitivity and
growth trends. Journal of Ecology 92, 954–966.
Swaty, R.L., Deckert, R.J., Whitham, T.G., Gehring, C.A., 2004. Ectomycorrhizal
abundance and community composition shifts with drought: predictions from
tree rings. Ecology 85, 1072–1084.
Swetnam, T.W., Betancourt, J.L., 1998. Mesoscale disturbance and ecological
response to decadal climatic variability in the American southwest. Journal
of Climate 3128–3147.
Tafangenyasha, C., 1997. Tree loss in the Gonarezhou National Park (Zimbabwe)
between 1970 and 1983. Journal of Environmental Management 49, 355–366.
Tafangenyasha, C., 1998. Phenology and mortality of woody plants during and after
a severe drought in southeastern Zimbabwe. Transactions of Zimbabwe Scientific Association 72, 1–6.
Tafangenyasha, C., 2001. Decline of the mountain acacia, Brachystegia glaucescens in
Gonarezhou National Park, southeast Zimbabwe. Journal of Environmental
Management 63, 37–50.
Tainter, F.H., Williams, T.M., Cody, J.B., 1983. Drought as a cause of oak decline and
death on the South Carolina coast. Plant Disease 67, 195–197.
Thabeet, A., Vennetier, M., Gadbin-Henry, C., Denelle, N., Roux, M., Caraglio, Y., Vila,
B., 2009. Response of Pinus sylvestris L. to recent climate change in the French
Mediterranean region. Trees, Structure and Functions 23 (4), 843–853.
Thuiller, W., Albert, C., Araujo, M.B., Berry, P.M., Cabeza, M., Guisan, A., Hickler, T.,
Midgley, G.F., Paterson, J., Schurr, F.M., Sykes, M.T., Zimmermann, N.E., 2008.
Predicting global change impacts on plant species’ distributions: future challenges. Perspectives in Plant Ecology, Evolution and Systematics 9, 137–152.
Touchan, R., Anchukaitis, K.J., Meko, D.M., Attalah, S., Baisan, C., Aloui, A., 2008. Long
term context for recent drought in northwestern Africa. Geophysical Research
Letters 35, L13705.
Trotter, R.T.I., 2004. Linking Climate Change and Community Dynamics: Pinyon Pine
Stability and Sensitivity in a Heterogeneous Landscape. Northern Arizona
University.
Tsopelas, P., Angelopoulos, A., Economou, A., Soulioti, N., 2004. Mistletoe (Viscum
album) in the fir forest of Mount Parnis, Greece. Forest Ecology and Management
202, 59–65.
Tsvetkov, V.F., Tsvetkov, V.I., 2007. The problem of spruce forests—mortality in the
Arkhangelsk Region. In: Dying Spruce Forests of Arkhangelsk Region. Problems
and Means of their Solution, Department of Forest Complex of Arkhangelsk
Region, Arkhangelsk, Russian Federation, pp. 20–30.
van Mantgem, P.J., Stephenson, N.L., 2007. Apparent climatically induced increase of
tree mortality rates in a temperate forest. Ecology Letters 10, 909–916.
van Mantgem, P.J., Stephenson, N.L., Byrne, J.C., Daniels, L.D., Franklin, J.F., Fulé, P.Z.,
Harmon, M.E., Larson, A.J., Smith, J.M., Taylor, A.H., Veblen, T.T., 2009. Widespread increase of tree mortality rates in the western United States. Science 323,
521–524.
van Nieuwstadt, M.G.L., Sheil, D., 2005. Drought, fire and tree survival in a Borneo
rain forest, East Kalimantan, Indonesia. Journal of Ecology 93, 191–201.
Vennetier, M., Vila, B., Liang, E.Y., Guibal, F., Thabeet, A., Gadbin-Henry, C., 2007.
Impact of climate change on pine forest productivity and on the shift of a
bioclimatic limit in a Mediterranean area. Options Méditerranéennes, Série A, n8
75, CIHEAM/IAMB, Bari, Italy, pp. 189–197.
Vennetier, M., Cecillon, L., Guénon, R., Schaffhauser, A., Vergnoux, A., Boichard, J.-L.,
Bottéro, J.-Y., Brun, J.-J., Carrara, M., Cassagne, N., Chandioux, O., Clays-Josserand, A., Commeaux, C., Curt, T., Czarnes, S., De Danieli, S., Degrange, V., Di Rocco,
R., Domeizel, M., Doumenq, P., Doussan, C., Estève, R., Faivre, N., Favier, G.,
Gaudu, J.-C., Gros, R., Guiliano, M., Guillaumaud, N., Hoepffner, M., Juvy, B., Le
Roux, X., Lebarriller, S., Malleret, L., Martin, W., Mas, C., Masion, A., Massiani, C.,
Mermin, E., Mille, G., Morge, D., Pignot, V., Poly, F., Renard, D., Ripert, C., Ruy, S.,
Tardif, P., Tatoni, T., Théraulaz, F., Vassalo, L., Asia, L., 2008. Etude de l’impact
d’incendies de forêt répétés sur la biodiversité et sur les sols: recherche
d’indicateurs. Rapport final. Cemagref, Ministère de l’Agriculture et de la pêche,
Union Européenne, Aix en Provence, 236 p.
Vertui, F., Tagliaferro, F., 1998. Scots pine (Pinus sylvestris L.) die-back by unknown
causes in the Aosta Valley, Italy. Chemosphere 36, 1061–1065.
80
684
C.D. Allen et al. / Forest Ecology and Management 259 (2010) 660–684
Viljoen, A.J., 1995. The influence of the 1991/92 drought on the woody vegetation of
the Kruger National Park. Koedoe 32, 85–97.
Villalba, R., Veblen, T.T., 1998. Influences of large-scale climatic variability on
episodic tree mortality in northern Patagonia. Ecology 79, 2624–2640.
Voelker, S.L., Muzika, R., Guyette, R.P., 2008. Individual tree and stand level
influences on the growth, vigor, and decline of Red Oaks in the Ozarks. Forest
Science 54, 8–20.
Walther, G.-R., Berger, S., Sykes, M.T., 2005. An ecological ‘‘footprint’’ of climate
change. Proceedings of the Royal Society B 272, 1427–1432.
Wang, H.B., Zhang, Z., Kong, X.B., Lui, S.C., Shen, Z.R., 2007. Preliminary deduction of
potential distribution and alternative hosts of invasive pest, Dendroctonus
valens (Coleoptera: Scolytidae). Scientia Silvae Sinicae 143, 71–76.
Waring, R.H., 1987. Characteristics of trees predisposed to die. Bioscience 37, 569–
577.
Wermelinger, B., Seifert, M., 1999. Temperature-dependent reproduction of the
spruce bark beetle Ips typographus, and analysis of the potential population
growth. Ecological Entomology 24, 103–110.
Wermelinger, B., Rigling, A., Schneider, M., Dobbertin, M., 2008. Assessing the role of
bark- and wood-boring insects in the decline of Scots pine (Pinus sylvestris) in
the Swiss Rhone valley. Ecological Entomology 33, 239–249.
Werner, W.L., 1988. Canopy dieback in the upper montane rain forests of Sri Lanka.
GeoJournal 17, 245–248.
Williamson, G.B., Laurance, W.F., Oliveira, A.A., Delamonica, P., Gascon, C., Lovejoy,
T.E., Pohl, L., 2000. Amazonian tree mortality during the 1997 El Nino drought.
Conservation Biology 14, 1538–1542.
Woo, S.-Y., Lim, J.H., Je, S.M., Lee, D.K., Kwon, M.J., Ryang, S., 2007. Decline in Mt.
Halla-A Linkage with Physiological Changes Caused by Climate Change.In:
Fourth USDA Greenhouse Gas Conference, Baltimore, MD, 6 February 2007.
Woods, P., 1989. Effects of logging, drought, and fire on structure and composition
of tropical forests in Sabah, Malaysia. Biotropica 21, 290–298.
Worrall, J.J., Egeland, L., Eager, T., Mask, R.A., Johnson, E.W., Kemp, P.A., Shepperd,
W.D., 2008. Rapid mortality of Populus tremuloides in southwestern Colorado,
USA. Forest Ecology and Management 255, 686–696.
Würth, M.K.R., Peláez-Riedl, S., Wright, S.J., Körner, C., 2005. Non-structural carbohydrate pools in a tropical forest. Oecologia 143, 11–24.
Zweifel, R., Zeugin, F., 2008. Ultrasonic acoustic emissions in drought-stressed
trees—more than signals from cavitation? New Phytologist 179, 1070–1079.
Zweifel, R., Rigling, A., Dobbertin, M., 2009. Species-specific stomatal response of
trees to microclimate—a link to vegetation dynamics? Journal of Vegetable
Science 20, 442–454.
81
APPENDIX B: GROWTH-MORTALITY RELATIONSHIPS IN PIÑON PINE (PINUS
EDULIS) DURING SEVERE DROUGHTS OF THE PAST CENTURY: SHIFTING
PROCESSES IN SPACE AND TIME
This paper was published in the journal PLoS ONE.
This is an open-access article distributed under the terms of the Creative Commons
Attribution License version 4.0, which permits unrestricted use, distribution, and
reproduction in any medium, provided the original author and source are credited.
© 2014 A. Macalady and H. Bugman
82
Growth-Mortality Relationships in Piñon Pine (Pinus
edulis) during Severe Droughts of the Past Century:
Shifting Processes in Space and Time
Alison K. Macalady1*, Harald Bugmann2
1 University of Arizona, School of Geography and Development and Laboratory of Tree-Ring Research, Tucson, Arizona, United States of America, 2 Forest Ecology,
Department of Environmental Systems Science, ETH Zürich, Zürich, Switzerland
Abstract
The processes leading to drought-associated tree mortality are poorly understood, particularly long-term predisposing
factors, memory effects, and variability in mortality processes and thresholds in space and time. We use tree rings from four
sites to investigate Pinus edulis mortality during two drought periods in the southwestern USA. We draw on recent sampling
and archived collections to (1) analyze P. edulis growth patterns and mortality during the 1950s and 2000s droughts; (2)
determine the influence of climate and competition on growth in trees that died and survived; and (3) derive regression
models of growth-mortality risk and evaluate their performance across space and time. Recent growth was 53% higher in
surviving vs. dying trees, with some sites exhibiting decades-long growth divergences associated with previous drought.
Differential growth response to climate partly explained growth differences between live and dead trees, with responses
wet/cool conditions most influencing eventual tree status. Competition constrained tree growth, and reduced trees’ ability
to respond to favorable climate. The best predictors in growth-mortality models included long-term (15–30 year) average
growth rate combined with a metric of growth variability and the number of abrupt growth increases over 15 and 10 years,
respectively. The most parsimonious models had high discriminatory power (ROC.0.84) and correctly classified ,70% of
trees, suggesting that aspects of tree growth, especially over decades, can be powerful predictors of widespread droughtassociated die-off. However, model discrimination varied across sites and drought events. Weaker growth-mortality
relationships and higher growth at lower survival probabilities for some sites during the 2000s event suggest a shift in
mortality processes from longer-term growth-related constraints to shorter-term processes, such as rapid metabolic decline
even in vigorous trees due to acute drought stress, and/or increases in the attack rate of both chronically stressed and more
vigorous trees by bark beetles.
Citation: Macalady AK, Bugmann H (2014) Growth-Mortality Relationships in Piñon Pine (Pinus edulis) during Severe Droughts of the Past Century: Shifting
Processes in Space and Time. PLoS ONE 9(5): e92770. doi:10.1371/journal.pone.0092770
Editor: Hormoz BassiriRad, University of Illinois at Chicago, United States of America
Received October 10, 2013; Accepted February 25, 2014; Published May 2, 2014
Copyright: ß 2014 Macalady, Bugmann. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits
unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: AKM was supported by a US Department of Energy Graduate Research Environmental Fellowship (GREF) and the GK-12 Graduate Fellows Program of
the National Science Foundation. HB was supported by a Haury Fellowship from the University of Arizona, Laboratory of Tree-Ring Research. This work is also a
contribution to the Swiss National Center of Competence in Research (NCCR) ‘Climate’, funded by the Swiss National Science Foundation. The funders had no role
in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing Interests: The authors have declared that no competing interests exist.
* E-mail: [email protected]
off insects and pathogens; and (2) hydraulic failure, i.e., the
collapse of the water-conducting system in the xylem [16,21].
Much recent drought-mortality research has built on this
framework in an experimental context to detect failure points
within the linked hydraulic, photosynthetic and carbon transport
systems of trees subject to drought, with the aim of clarifying
mechanisms and identifying physiological and climatological
thresholds beyond which death occurs (cf. [20,22,23]). However,
our understanding of tree mortality remains inadequate for
projecting the impacts of climate change on forests [6]. An
important set of knowledge gaps relates to understanding variability
in interacting drought-mortality processes and thresholds across
landscapes and through time, and determining the time scales that
are most important for understanding mortality risk, including the
influence of previous events on the status and resistance of trees
[22,24,25].
Tree ring studies offer an excellent way to complement more
physiologically detailed but temporally and spatially limited
Introduction
Climate-related tree mortality has been documented in forests
around the world and may be intensifying in some regions due to
rising temperature and enhanced drought [1–6]. Forests play an
important role in regulating the earth’s energy, carbon and water
cycles [7], and increases in tree mortality rates or rapid collapses in
forest cover could have major implications for ecosystems and
ecosystem services [8–10]. Yet, our ability to predict future forest
dynamics is limited by gaps in our understanding of tree death,
and associated uncertainty about how represent tree mortality in
vegetation models [6,11–15].
The processes underlying drought-associated tree mortality are
particularly unclear [16–20]. Leading hypotheses suggest that tree
mortality may arise from two interrelated mechanisms: (1) carbon
starvation, i.e., water stress causes trees to close their stomata, thus
reducing photosynthesis and constraining the availability of
carbon necessary for maintaining metabolic functions or fending
PLOS ONE | www.plosone.org
1
May 2014 | Volume 9 | Issue 5 | e92770
83
Tree Growth and Drought-Associated Mortality Risk
WRK, SEV) is located on relatively gentle terrain at the middle-tohigh end of the local elevation range of piñon, where the species
co-occurs with one or more Juniperus species (J. monosperma and J.
scopulorum). We refer to the study areas by acronym, and add 2000
or 1950 after each to distinguish between the sampling representing the two drought-mortality events.
experimental studies of drought-associated tree mortality. Radial
stem growth is, at least in the short term, a low priority for a tree’s
allocation of available carbon [26,27]. Thus tree growth is often a
sensitive indicator of changes in a tree’s carbon balance due to
environmental or tree intrinsic factors. In the context of droughtassociated mortality, for example, if trees close their stomata to
reduce the risk of desiccation and hydraulic failure, carbon uptake
and presumably radial growth are reduced. Hypothesized
relationships between growth, a tree’s carbon budget, and
mortality are reflected by mortality algorithms in many forest
models, where recent (1–3 year) growth is a basis for determining
the risk of death in a given time-step [11,15]. Tree rings have been
used to test and improve the empirical basis for such algorithms
(cf. [28–32]). However, these studies have focused mostly on
sporadic, individual-tree mortality, rather than on tree die-off
associated with severe droughts. Furthermore, a paucity of longterm datasets has hindered the evaluation of the temporal and
spatial generality of relationships between growth and the
likelihood of death, although stable relationships are critical to
the incorporation of growth-mortality algorithms in dynamic
vegetation models [24,33]. A number of studies have explored
growth rates in trees that eventually died during drought to test
hypotheses about the physiological mechanisms of mortality [34–
43], but to our knowledge no studies have quantitatively assessed
the importance of multiple growth variables and time scales for
shaping mortality risk during drought-related die-off.
In the semi-arid conifer forests of western North America,
prolonged drought and heat have interacted with bark beetles to
produce mortality across millions of hectares during the last 20
years [5,44–46]. In the piñon-juniper woodlands of the southwestern USA (SW), drought and the activity of bark beetles led to
the widespread death of piñon pines (Pinus edulis Engelm. and P.
monophylla Torr. & Frem) [5,44,47]. It has been suggested that
several aspects of the 2000s drought and associated die-off were
novel, and that unusually warm conditions caused elevated
mortality rates that anticipate ‘global change’ conditions
[5,44,48,49]. However, tree mortality was also widespread in the
SW during the 1950s [50,51]. Although cooler than the 2000s
drought across much of New Mexico [49], instrumental and treering based records indicate that the 1950s drought was one of the
most severe and protracted of the past 500 years [5]. Exceptional
preservation of long-dead trees in these landscapes provides a
unique opportunity to compare tree-ring growth and mortality
patterns during the two mortality episodes, allowing not only for
an evaluation of long- and short-term factors operating during
widespread drought-associated die-off, but also an assessment of
the generality of mortality processes and growth-mortality
relationships during droughts with distinct climatic patterns.
In this study we draw on recent and archived tree-ring
collections and stand structural measurements from sites spanning
a latitudinal gradient in New Mexico, USA, to investigate droughtrelated mortality of piñon pine (P. edulis Engelm.). Specifically, we
(1) analyze P. edulis growth patterns and mortality during the 1950s
and 2000s droughts; (2) determine the influence of climate and
competition on growth in trees that died vs. those that survived;
and (3) derive regression models of growth-mortality risk and
evaluate their performance across space and time.
Ethics Statement and Data Availability
This research was performed at the Sevilleta National Wildlife
Refuge, Bandelier National Monument and the Carson National
Forest. Necessary permissions were obtained from the National
Park Service, National Fish and Wildlife Service and National
Forest Service. All data are available upon request.
Field Sampling and Laboratory Methods
2000s mortality. In 2008–2010, we selected 10–30 recently
dead (bark and fine branches remaining) ‘‘target’’ trees at each of
the four study areas (Tables 1, 2). Living trees were selected as
‘‘control cases’’ to compare with each dead tree based on
proximity, similarity in micro-topography, tree diameter and
overall stature [29,31]. Mortality processes can differ between
adult and juvenile trees [52]; we selected trees greater than 9cm
diameter at root collar (DRC) in order to focus our study on the
mortality of mature trees [53]. Very few mature piñon trees were
alive at BNM in 2010, thus 30 dead target trees were selected at
this site for comparison to live and dead trees at the other sites.
Two increment cores were extracted from each living target tree
at breast height (135 cm), and a cross-section was taken from each
paired dead tree. In the laboratory, cores and cross-sections were
prepared, crossdated and ring widths were measured using
standard dendrochronological techniques [54]. We confirmed
visual crossdating statistically using the computer program
COFECHA [55]. In total, 167 (n = 98 dead and n = 69 live)
target trees were sampled and successfully crossdated. One live
target tree at TRP and two dead trees at BNM could not be
crossdated, and were dropped from the study. For the characterization of stand structure and spatial patterns of mortality, we
measured the diameter and noted the status (live or dead) and
species of each tree (.1 cm DBH) within a 7.5 m radius plot
centered on each live and dead target tree.
1950s mortality. SEV and BNM are associated with a
previous study for which long-dead trees were measured and
sampled along with living neighbors (Allen, Betancourt and
Swetnam, unpublished data). All living trees, snags and downed
remnants within two 0.5 ha plots at BNM and SEV were
measured, and each piñon tree was sampled for dendroecological
purposes. These 0.5 ha plots fall within the larger sampling areas
for each site described above. Each core and cross section was
prepared and visually crossdated as above in order to determine
inside and outside ring dates for each tree. To investigate growthmortality relationships associated with the 1950s drought, we
selected target trees from the archived specimens of this sampling
campaign (all housed at the Laboratory of Tree-Ring Research,
University of Arizona). We identified dead specimens with good
preservation (e.g. where the last year of radial growth could
reliably be determined) that were $9 cm DRC and had outer ring
dates between 1940 and 1960. By many definitions, the 1950s
drought actually stretched from the mid-1940s through the
beginning of the 1960s [5], and so we allowed selection of trees
with outer ring dates slightly preceding the first significant drought
year in the 1940s (1946). Dead tree samples were measured and
checked for dating errors as described above. We then identified
and measured cores or cross-sections from trees $9 cm DRC that
survived through the early 1960s. Growth increments from these
Materials and Methods
Study Areas
We assessed P. edulis mortality close to lower treeline at four
study areas that span gradients in climate and stand composition
in New Mexico (Tables 1, 2; Fig. S1). Each area (TRP, BNM,
PLOS ONE | www.plosone.org
2
May 2014 | Volume 9 | Issue 5 | e92770
84
23
30
26
22
27
28
0
30
30
10
29
10
BNM1950
SEV2000
SEV1950
1950s
2000s
1950s
Woodland Structure and Spatial Patterns of Mortality
387
2050
34.346N 106.556W
Sevilleta Natl.
Wildlife Refuge, NM
We calculated density and basal area of live and dead trees
within neighborhood plots and used quasi-binomial regression to
assess relationships between woodland structure and mortality
severity (the percentage of recently dead trees) during the 2000s
mortality event. We assessed fine-scale spatial patterning of recent
mortality by testing for differences in stand composition and
mortality-severity around live versus dead target trees.
It was not possible to make the same assessments of woodland
structure prior to the 1950s mortality episode because of unknown
tree locations within the 0.5 ha plots, but we made conservative
estimates of piñon mortality severity using the dendrochronologically determined birth and death dates of the trees sampled on the
0.5 ha plots (cf. above).
The rings of juniper trees at our sites cannot be reliably
crossdated due to many false and missing rings and lack of circuit
uniformity, and therefore no assessments of juniper size, structure
and mortality were made.
Growth and Growth Indices
We calculated basal area increments (BAI), relative basal area
increments (RelBAI) and ring width indices (RWI) from raw ringwidths (RW) for use in subsequent analyses. Basal area increments
(cm2 yr21) were calculated from ring widths (mm yr21) for each
tree radius using the inside-out method:
doi:10.1371/journal.pone.0092770.t001
BNM2000
2000s
340
1940
35.766N 106.276W
Bandelier Natl.
Monument, NM
PLOS ONE | www.plosone.org
Outside ring dates were recorded as the best-available
approximation of the death date for each dead tree. Direct
observations of dying trees at BNM2000 suggest that tree-ring
estimates and actual death dates agree within a year or two (C.
Allen, personal communication). Many insects and diseases are known
to affect piñon pine, but Ips confusus LeConte – the piñon Ips bark
beetle – has been associated with the most severe damage, and is
known to attack both living and recently dead trees [56,57]. We
thus documented evidence of Ips attack for each dead target tree
by noting whether an individual contained Ips beetle galleries on
the sampled portion of the tree bole and/or whether the sample
contained blue stain – a fungal pathogen introduced by bark
beetles – in the sapwood [58].
10.6
WRK2000
10.1
TRP2000
2000s
340
1990
10.1
312
2100
35.816N 106.246W
White Rock, NM
Tres Piedras, NM
36.346N 105.936W
8.7
2000s
Dead Target
Trees
Site Acronym
Annual
Temp. (C6)
Annual Precip.
(mm)
Elevation (m)
Location
Study Area
Table 1. Location, characteristics and sample sizes for each study site.
samples were used to estimate tree diameter in the year 1960.
Trees from this pool were matched as much as was possible to
dead ‘‘case’’ trees based on the 1960-diameter estimate and 0.5 ha
study plot. The pool of suitable survivor trees was limited by the
fact that larger trees appear to have been preferentially killed
during the 1950s drought at these sites. Ultimately we selected 25
dead and 26 surviving target trees for SEV1950, and 23 dead and
22 surviving trees for BNM1950, for a total of 96 trees, with the
1950s dataset containing slightly larger dead than surviving trees
(Table 2). However, all trees met our size criteria (DRC $9 cm)
and were estimated to be at least 65 years old by the 1950s
drought.
Target Tree Characteristics
Mortality event
Live Target
Trees
Tree Growth and Drought-Associated Mortality Risk
p
BAIt ~
i
P
2
RW zd
{p t
i
P
t{1
100
2
RW zd
ð1Þ
where d is an estimate of the distance from the first measured ring
to the pith [59], i is the first year of growth in the time series, and t
is the current year of growth.
RelBAI (cm2 yr21 yr21) was calculated by dividing basal area
increments for each year by the previous year’s total basal area
3
May 2014 | Volume 9 | Issue 5 | e92770
85
Tree Growth and Drought-Associated Mortality Risk
Table 2. Mortality severity and stand characteristics for each study site.
Site
Piñon
mortality (%)
Juniper
Mortality (%)
Tree
Density (stems/ha)
Piñon
Density (stems/ha)
Total
Basal Area
(m2/ha)
Piñon
Basal Area
(m2/ha)
TRP2000
64.0 (4.1)
2.9 (1.7)
426.8 (36.8)
333.8 (36.9)
8.8 (0.9)
6.1 (0.7)
WRK2000
82.1 (7.5)
0.2 (0.2)
605.5 (78.1)
206.6 (34.4)
7.0 (1.1)
3.0 (0.6)
BNM2000
99.6 (0.4)
2.6 (1.0)
808.4 (71.5)
282.9 (41.0)
9.7 (1.1)
3.5 (0.7)
SEV2000
19.9 (3.2)
3.1 (1.0)
726.2 (49.0)
331.1 (31.3)
11.2 (0.8)
3.1 (0.3)
BNM1950
64.5 (7.5)
-
-
262.1 (26.5)
-
-
SEV1950
46.5 (8.5)
-
-
432.0 (61.0)
-
-
Standard errors for each measurement are indicated in parentheses. For 2000s sites, the mean and standard error were calculated based on 7.5m neighborhood plot
data. For 1950s sites, the mean and standard error are from measurements at two separate 0.5-hectare (ha) plots. Tree density and basal area reflect pre-mortality
conditions. No estimates of juniper mortality, total tree density or basal area were made for the 1950s sites due to lack of dendroecological data for juniper and lack of
tree size reconstructions for all trees within each plot.
doi:10.1371/journal.pone.0092770.t002
abrupt growth changes. Average growth [16,26] was calculated as
the mean of annual growth measurements over k = 3, 5, 7, 10,
15…50 years. Growth variance has been documented as a factor
that influences predisposition to mortality in semi-arid woodlands
[35,36]. We chose mean sensitivity, a statistic of year-to-year
growth variability that reflects both the variance and the first-order
autocorrelation of the time series [63], because compared to
standard deviation (Table 2) or first-order autocorrelation (not
shown) it differed more strongly between living and dead trees.
Mean sensitivity was calculated from growth time series according
to Eq. 2 in Biondi and Quedan [64], where k = the length of the
tree-ring series t = 1,2,…,k = year in the tree-ring series:
[60]. We utilized the C-method to generate ring-width indices
[61]. This method transforms ring widths by dividing individual
series by a curve that reflects the biological expectation of constant
annual basal area increment for each tree. The C-method thus
standardizes individual ring width series to a common mean and
variance, but unlike other standardization methods, it allows
individual index series to retain low-frequency variability and
trends due to, for example, injury, senescence, competition, and
climatic influences. Measurements from multiple radii were
averaged to generate single records of RW, BAI, RelBAI and
RWI for each tree.
Although BAI is often considered to be a more biologically
meaningful growth metric than raw ring-widths or ring-width
indices [61,29,62,39], we utilized RW for building models of
growth-based mortality risk and assessing growth relationships to
climate and competition, and RWI for the calculation of average
growth chronologies. RWI was used in chronologies in order to
minimize the influence of particular trees with high mean growth
and variance, and to highlight changes in the trajectory versus the
average growth rate in dead versus surviving trees. We utilized
RW in quantifying growth-mortality relationships because, unlike
RWI, it retains gross differences in growth rates between live and
dead trees, and contrary to expectation, we found less pronounced
size-related trends in piñon RW compared to BAI and RelBAI
(Fig. S2). Although data from our neighborhood plots indicate that
tree size was likely a predisposing factor for mortality at our 2000s
sites (Fig. S3), our study design is not well-suited to quantify the
combined influence of tree size and growth on mortality risk,
because (1) the 1950s dataset is slightly biased towards larger dead
trees due to preservation issues; and (2) average sampled tree size
(and age) was not stratified between study sites (Table 2). Thus
even though tree size and age matched relatively well between
living and dead trees and target trees shared similar dominant or
co-dominant status, we sought to choose a growth metric that is
least sensitive to tree size in order to make a conservative
estimation of the growth-mortality relationships. For the sake of
comparison, we also analyzed basic growth differences between
live and dead trees using all metrics, and generated growthmortality models using BAI, RelBAI and RWI as outlined below.
These models had slightly different predictor variables and
performance, but did not lead to different conclusions and thus
are not shown.
We generated four types of indices from growth time series to
develop a pool of predictor variables of mortality risk: average
growth, growth variability, growth trend and the frequency of
PLOS ONE | www.plosone.org
k
P
jRWt {RWt{1 j
k t~2
MS~
k
k{1
P
RWt
ð2Þ
t~1
Growth trend was calculated as the slope of the linear regression of
growth fitted across all years within different time intervals [29].
Both mean sensitivity and growth trend were calculated over k = 5,
10, 15…50 years.
Several studies have noted a preponderance of abrupt growth
declines in trees predisposed to die [31,34], and we also noted
differences in the frequency of growth releases in our dataset. We
defined abrupt growth changes as 50% reductions or increases in
growth averaged over a 10 year period as compared to average
growth over the previous 10 years. We counted the number of
such changes over k = 5, 10, 15…50 years.
For the calculation of all predictor variables, live tree growth
was truncated at the last year of the corresponding dead tree pair.
Mortality Modeling
We used linear mixed effects logistic regression to relate growth
indices to tree status (live or dead) [65,66]. We followed the
general procedure of Das et al. [31], modified for mixed effects
modeling, to identify the most parsimonious model structure. We
designated study site/period as a random effect. For the fixed
effects, we: (1) generated models with only one covariate using a
temporal range of growth indices from each growth category
(average growth, growth trend, mean sensitivity, and abrupt
growth changes). We did not include indices after 35 years because
multiple trees with shorter crossdated growth records dropped out
of the predictor pool after this point, making it difficult to compare
4
May 2014 | Volume 9 | Issue 5 | e92770
86
Tree Growth and Drought-Associated Mortality Risk
models; (2) used Akaike’s Information Criterion (AIC) to compare
the support for each model [67]; and (3) created a suite of models
with multiple fixed effect predictors using the variables from the
three single-variable models with the lowest AIC score in each
growth category and/or with differences in AIC scores below 2
[67]. Independent variables were transformed if Wald tests
indicated non-linear relationships [68]. Random effects were
dropped from each model if AIC scores and likelihood ratio tests
on nested models indicated that a simpler model structure was
more parsimonious [69]. The three models with the lowest AIC
scores overall are presented along with the best-ranked single
variable models for comparison.
.
2
n DBHj
X
DBHi2
CI~
DISTij
j~1
where j = 1,…,n are competitor trees, i is the target tree, DBH is
tree diameter and DIST is the distance between target and
competitor trees [29]. For analyses of the 1950s sites, competition
was not included as a covariate because neighborhood data were
not available. Growth, climate and competition variables were
converted to z-scores specific to site and the modeled period.
Model fitting reflected our goals of assessing the effects of
climate and competition on tree growth, and testing whether trees
destined to die and survive responded differently to climate and
competition in the years leading up to mortality. First, through
exploratory analyses, we noted that the relationship between ring
width and climate is sometimes slightly curvilinear, and models
without higher-order climate terms contained highly skewed
residuals. Thus, following [69] we started with a ‘beyond-optimal’
model that included as fixed effects both linear and quadratic
climate variables along with tree status, competition, and site.
Because of the potential complexity of interactions in a model with
a high number of predictors, we restricted the ‘beyond-optimal’
model to 2 and 3-way interactions that represented interpretable
biological processes [69]. We did not include interactions between
climate variables, as our initial analyses indicated that they were
small. Interactions between quadratic climate terms and other
predictors were not included, either. Tree ID was added as a
random effect to account for non-independence of growth within
individual trees. More complicated random effects structures (for
example, allowing different growth trends over years for different
trees, or nesting Tree ID within site) were rejected, as they did not
improve the full model. Residual autocorrelation of growth
between years and heterogeneity of variance in residuals were
accounted for by adding model correlation structures and variance
weights [69,65,41].
We used an iterative, AIC-based backwards selection to
sequentially drop terms from the ‘beyond-optimal’ model [69].
We developed a final model structure using growth data that
started in 1960 and 1910 for the 2000s and 1950s datasets,
respectively. We retained this structure for models of growth over
different time periods to allow for the straightforward comparison
of coefficients. Residuals were inspected to ensure that assumptions about residual independence, heterogeneity and normality
were adequately met. Final models were fit using the restricted
maximum likelihood criteria. Growth data from BNM2000 were
not included in model fitting, as there were no living trees.
Unless otherwise noted, all statistical tests were performed in R
v3.0.1 [73]. We used the dplR library v1.5.6 for BAI and RWI
calculations and dendrochronology statistics [74]. Generalized
linear mixed modeling was performed using functions from the
lme4 library v0.99999911-6 [75]. Linear mixed models of tree
growth were fit using the library nlme v3.1-110 [65].
Model Diagnostics, Validation and Interpretation
We computed a variety of diagnostic and validation statistics to
aid in the interpretation of best-ranked logistic models. Correct
classification rates were calculated from confusion matrices
generated by a bootstrapped internal validation routine (1000
iterations) in which models were fit repeatedly with a random subsample containing 60% of the data and validated on the remaining
40% [29]. Trees were classified as living if their survival
probability was above an empirically determined threshold [70].
We also externally validated best-ranked models on the dataset
from BNM2000, which contained only dead trees, did not match
the case-control study design, and thus could not be used in model
building. However we expected that the best models would
correctly classify the majority of dead trees at BNM2000 if
mortality processes and thresholds were similar here compared
with other sites. The Area Under the Receiver Operating
Characteristic curve (ROC) is a threshold-independent measure
of model discrimination, where a value of 0.5 suggests no
discrimination and values above 0.8 suggest excellent discrimination between live and dead trees [71]. Odds ratios were calculated
from regression coefficients to assess changes in relative mortality
risk associated with changes in growth. Odds ratios indicate a
change in the likelihood of mortality given a meaningful change in
the predictor variables. For example, an odds ratio of 2.0
associated with a 0.1 mm increase in average growth can be
interpreted as a doubling of the likelihood of survival with each
0.1 mm growth increase, all else being equal.
Effects of Climate and Competition on Tree Growth and
Mortality
We fit a separate set of linear mixed-effect models to make posthoc assessments of how two factors - climate and competition influenced growth in trees destined to die and survive droughtmortality events [41]. For target trees at TRP, WRK and SEV, we
modeled ring width as a function of cool season precipitation
(previous September through May) (PPTcool), early summer (May–
July) average vapor pressure deficit (VPDMJJ), a continuous index
of competitive pressure (CI), and a categorical variable representing tree status (Live or Dead). The seasonal climate variables were
chosen based on initial comparisons between mean-value chronologies and PRISM climate model output (4 km resolution) [72]
(Text S1). PRISM data were chosen instead of weather station
data because PRISM data better explained the variability in tree
growth chronologies (not shown).
Indices of competitive pressure (CI) were derived from the
neighborhood plot data (cf. above). A distance- and size-weighted
index, calculated using only conspecific (e.g. piñon) neighbors was
used for modeling growth, as it yielded the strongest and most
consistent correlations with recent growth:
PLOS ONE | www.plosone.org
ð3Þ
Results
Spatial and Temporal Mortality Patterns
The severity of piñon mortality during the recent (2000s)
drought ranged from 20% to 99% across our study sites, and was
least severe at SEV2000 (Table 2). Piñon mortality associated with
drought in the 1940s and 1950s was also severe at both SEV1950
and BNM1950 (45%–65%), with slightly higher mortality at
BNM1950 (Table 2). These measurements are based on 7.5 m
neighborhood plot data around each target tree (2000s drought) or
5
May 2014 | Volume 9 | Issue 5 | e92770
87
Tree Growth and Drought-Associated Mortality Risk
dead and live trees in 0.5 ha plots (1950s drought), and thus reflect
mortality severity at our study sites only. However, for the 2000s
drought the patterns in our data conform to other published
studies with more extensive sampling, in which the 2000s mortality
was found to be greater in northern versus south-central New
Mexico [76].
Outside ring dates correspond well with periods of decadal
drought, but mortality was more or less synchronous depending on
site and mortality episode (Fig. 1). Fine-scale spatial patterning of
mortality at the 2000s study sites varied along the latitudinal
gradient, with a non-significant positive relationship between tree
density and mortality severity at TRP2000 grading into a weakly
significant negative relationship between density and mortality
severity at SEV2000 (Fig. S4). Finer-scale clumping of dead piñon
was also characteristic of mortality at the northern sites in the
2000s (TRP2000 and WRK2000), with more piñon trees and
more dead piñon trees around dead versus living target trees (not
significant at the 0.05 level at WRK2000). No such clustering
existed at SEV2000 (Table S1). Drought-associated mortality was
generally concentrated in medium-sized to larger trees at the
2000s sites, based on size-class distributions of living and dead
piñon trees measured in neighborhood plots (Fig. S3).
Evidence of Insect Attack
Evidence on sampled cross-sections points to the almost
ubiquitous presence of Ips beetles during both mortality events.
Seventy-nine percent of dead trees in the 2000s dataset contained
Ips galleries on sampled bole cross-sections, and 94% had sapwood
colored by blue stain fungus. At least one type of evidence was
present on 97% of dead samples. All samples with no evidence of
successful bark beetle attack were from the SEV2000 site, where
10% of dead trees had no indication of activity. In our 1950s
dataset, 67% of samples exhibited Ips galleries on sampled crosssections, and 100% contained blue stain fungus.
Differences in Growth between Dead and Surviving Trees
Averaged across all sites, trees that survived drought had higher
recent growth rates (53% higher 3-yr average RW, p,0.001) and
lower average mean sensitivity (18% lower 20-yr mean sensitivity,
p,0.001) than dead trees. Average recent growth and mean
sensitivity exhibited the same direction of difference between live
and dead trees at all sites where both live and dead trees were
sampled, although differences were only significant at SEV2000
and the 1950s sites (Table 3; Figs. S5–S6). Significant differences
in growth trends and abrupt growth changes were present between
live and dead trees at some sites, but they were less consistent in
magnitude and direction (Table 3; Figs. S7–S8). Differences in
average growth and mean sensitivity extended back many decades
at some sites, yet they were generally minor and insignificant
earlier in the life of trees (Figs. S5–S6). At some sites with stronger
growth differences between live and dead trees, a divergence in
growth occurred after previous severe and/or protracted drought
intervals (e.g., after the 1950s for 2000s sites, and after a drought at
the turn of the century for the 1950s sites; cf. Fig. 1).
Figure 1. Growth chronologies and death dates from piñon
target trees. Live (black) and dead (grey) tree ring-width index
chronologies for TRP2000 (A), WRK2000 (B), BNM2000 (C), SEV2000 (D),
BNM50 (E), and SEV50 (F). Tukey’s biweight robust mean was used to
calculate chronology values from individual index series. A smoothing
spline (df = 40) (thicker lines) is overlain on the annual mean value
chronologies (thinner lines). A horizontal dashed line indicates the
number of trees contributing to chronologies in each year. Bar plots of
outside ring dates for dead trees at each site are shown in the small
panels within each larger time series panel. The transparent grey boxes
show SW drought events (as defined in [5]) preceding the 2000s (A–D)
and 1950s (E–F) mortality events. The period 1945–1964 was the sixth
strongest drought event since 1000 A.D., and the period 1899–1904 was
the seventeenth strongest.
doi:10.1371/journal.pone.0092770.g001
Growth-based Models of Mortality Risk
The best logistic growth-mortality models were highly significant and resulted in good discrimination between live and dead
trees, with ROC scores above 0.80 and correct classification rates
slightly above 70% (Table 4). Models containing multiple and
longer-term growth variables featured substantially lower AIC
scores and higher discrimination statistics than models that
included only recent growth as a predictor. The best-ranked
model included an average growth variable, a measure of year-toPLOS ONE | www.plosone.org
year growth variability, and a term characterizing abrupt growth
increases (Table 4). The relative survival probability associated
with each predictor varied depending on site (Fig. 2). The
direction of growth-mortality relationships was consistent across
sites included in the model-building dataset. However, the strength
6
May 2014 | Volume 9 | Issue 5 | e92770
PLOS ONE | www.plosone.org
1.771
0.009
3yr BAI (cm2/yr)
3yr RelBAI (cm2/yr/yr)
2.840
0.020
20yr BAI (cm2/yr)
20yr RelBAI (cm2/yr/yr)
7
0.515
0.523
0.310
0.783
1.401
20yr Mean Sensitivity (RWI)
20yr Mean Sensitivity (BAI)
20yr Standard Deviation (RW)
20yr Standard Deviation (RWI)
20yr Standard Deviation (BAI)
20.04
20yr Trend (BAI)
7.41
8.34
20yr Increases (RWI)
20yr Increases (BAI)
6.70
6.10
5.43
20.06
20.06
20.02
1.252
0.898
0.282
0.577
0.569
0.563
0.017
2.290
1.614
0.514
0.006
1.284
0.765
0.234
16.1
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
8.18
5.18
3.68
20.07
20.07
20.04
1.489
1.014
0.432
0.515
0.508
0.501
0.037
2.958
1.954
0.859
0.016
1.948
1.186
0.476
13.6
Dead
WRK2000
5.70
5.30
4.70
20.04
20.04
20.01
1.132
0.828
0.229
0.588
0.581
0.578
0.013
2.286
1.668
0.460
0.010
2.005
1.337
0.357
17.4
Live
5.20
5.00
3.90
20.05
20.04
20.01
0.947
0.724
0.178
0.598
0.596
0.591
0.009
1.915
1.406
0.348
0.006
1.431
1.036
0.248
17.2
Dead
SEV2000
0.602
1.184
0.540
0.205
0.426
0.295
2.80
3.43
7.43
2.27
20.07
20.08
20.02
0.879
4.60
3.70
20.05
20.06
20.02
0.944
0.742
0.544
0.915
0.536
0.423
0.012
0.419
0.027
1.738
0.319
1.341
0.578
0.006
1.805
0.017
1.389
0.157
0.686
0.412
11.9
Dead
1.329
11
Live
3.50
1.41
1.00
0.00
20.01
20.01
0.589
0.651
0.232
0.538
0.528
0.521
0.020
1.015
1.200
0.442
0.017
1.162
1.146
1.87
1.65
0.91
0.00
0.00
0.00
0.596
0.739
0.158
0.733
0.732
0.729
0.008
0.886
1.087
0.236
0.007
0.927
1.102
0.239
12.1
8.8
0.439
Dead
BNM1950
Live
Significant differences between live and dead trees within sites are indicated by boldface type (p,0.05, Student’s t-test). No live trees were sampled for this study at BNM2000.
doi:10.1371/journal.pone.0092770.t003
6.41
20yr Increases (RW)
Abrupt growth changes
20.02
20.03
20yr Trend (RW)
20yr Trend (RWI)
Growth trend
0.509
20yr Mean Sensitivity (RW)
Growth variability
0.609
1.488
20yr RW (mm/yr)
20yr RWI
Long-term average growth
0.328
0.824
3yr RW (mm/yr)
15.9
3yr RWI
Recent average growth
Diameter (cm dbh)
BNM2000
Live
Dead
TRP2000
Live
Table 3. Average size and growth characteristics of live and dead target trees.
SEV1950
4.15
1.52
0.70
20.03
20.04
20.02
0.568
0.667
0.261
0.475
0.477
0.478
0.031
1.103
1.366
0.521
0.020
0.830
1.013
0.358
7.7
Live
2.12
1.54
1.35
20.03
20.03
20.01
0.482
0.669
0.139
0.692
0.693
0.694
0.008
0.784
1.077
0.220
0.005
0.466
0.662
0.131
10.7
Dead
88
Tree Growth and Drought-Associated Mortality Risk
May 2014 | Volume 9 | Issue 5 | e92770
PLOS ONE | www.plosone.org
log(RW25) + Sens15+ AbruptIncrease10
8
-
-
-
-
1+ log(RW20) | site
20.90
73.00
71.90
31.20
12.90
5.90
2.80
2.20
0.00
ROC
0.773
0.512
0.584
0.746
0.816
0.834
0.842
0.846
0.847
ROCboot
0.782
0.517
0.580
0.736
0.812
0.826
0.838
0.848
0.844
72.4%
53.4%
48.9%
67.3%
71.4%
71.6%
73.5%
71.5%
72.6%
Dead
Trees
CCR
72.9%
43.3%
57.5%
67.5%
69.4%
73.4%
71.5%
71.4%
72.7%
Live
Trees
CCR
72.6%
48.4%
53.2%
67.4%
70.4%
72.5%
72.5%
71.5%
72.6%
All
Trees
CCR
kappa
0.453
20.032
0.064
0.348
0.408
0.450
0.450
0.429
0.453
Variables include average growth (RW), mean sensitivity (Sens), growth trend (Trend), and abrupt growth changes (AbruptIncreases), with the number of years over which variables were averaged indicated after variable type. The
best single-variable models in different growth categories, along with a model containing recent average growth as the only predictor variable (log(RW3)), are shown for comparison. DAIC is the difference in AIC between the bestranked model and the model shown in each table row, with smaller values indicating more parsimonious model fit. ROC is a threshold independent measure of model discrimination, where 0.5 suggests no discrimination and
values above 0.8 suggest excellent discrimination. Correct classification rates (CCR) are based on a bootstrapped internal validation with 1000 iterations in which 60% of the data was used for model fitting and 40% was used for
model validation. Trees were classified as living if model output was greater than the empirically defined threshold [70]. ROCboot is an average of the ROC statistics generated in the model-fitting portion of the bootstrapping
routine. The kappa statistic measures the proportional improvement of the model classification over a random assignment of tree status [99], and was also estimated by taking an average of kappa statistics generated in the
bootstrapping routine.
doi:10.1371/journal.pone.0092770.t004
log(RW3)
Recent Average Growth
AbruptIncrease10
Abrupt growth changes
Trend15
Growth trend
Sens15
Growth sensitivity
log(RW20)
Average growth
Best 1-variable
log(RW30) + Sens15
1+ log(RW30) | site
1+ log(RW25) | site
log(RW30) + Sens20+ AbruptIncrease10
Best 2-variable
1+ log(RW30) | site
log(RW30) + Sens15+ AbruptIncrease10
D AIC
Random Effects
1+ log(RW30) | site
Fixed Effects
Model Diagnostics and Validation
Model Form
Table 4. Best-ranked growth-mortality models.
89
Tree Growth and Drought-Associated Mortality Risk
May 2014 | Volume 9 | Issue 5 | e92770
90
Tree Growth and Drought-Associated Mortality Risk
Figure 2. Predicted probabilities of mortality associated with the best-ranked logistic regression model. The figure shows predicted
survival probabilities associated with the best-ranked growth-mortality model in Table 4. Values for mean sensitivity over 15 years (Sens15) and the
count of abrupt growth increases over 10 years (AbruptIncreases10) are held at their mean for the dataset in (A); AbruptIncreases10 and log(RW30)
(average 30-year growth rate) are held at their mean in (B); log(RW30) and Sens15 are held at their mean in (C).
doi:10.1371/journal.pone.0092770.g002
of growth-mortality relationships was weaker overall for the 2000s
sites (and at TRP2000 in particular), and the threshold of mortality
varied. Specifically, trees at the 1950s sites exhibited lower growth
and higher growth variability than trees at the 2000s sites while
still featuring survival probabilities above the empirically determined mortality threshold in our model (0.497) (Fig. 2).
Odds ratios for the best-ranked model indicate that, averaged
across sites, a 0.1 mm increase in RW averaged over 30 years
increases the relative odds of survival by 1.32, all else being equal.
Likewise, a 0.1 increase in mean sensitivity over a 15-year period
decreases survival odds by 0.72, and one additional growth
increase leads to an increase in survival probability by a factor of
1.2. Model coefficients and 95% confidence intervals generated in
the bootstrapping routine indicates that all terms with the
exception of the random slope term for average growth are
significant at the 95% level (Table S2). We nonetheless retained
the random slope term, given consistent improvements in AIC and
significant increases in the log-likelihood compared to models
without this term [69].
Although the best-ranked models produced satisfactory correct
classification rates when evaluated internally, they performed
poorly when externally validated on the BNM2000 dataset,
significantly under-predicting mortality (only 12% of dead trees
were classified correctly) (Table 5). Best-ranked models also
performed unevenly when internally validated at the site level
(Table 5). Site-specific correct classification rates indicate good
model performance at the 1950s sites and SEV2000, but correct
classification rates were not much better at TRP2000 and
WRK2000 than if trees were classified by chance. Fitting
growth-mortality models only on data from those individual sites
did not dramatically improve this outcome (not shown).
Effects of Climate and Competition on Growth and
Mortality
We used linear mixed-effects models to make post-hoc assessments of the influence of climatic and competitive factors on radial
growth in trees that survived and eventually died. These models
confirm that growth over the decades prior to drought-mortality
events was different depending on eventual tree status (live/dead),
significantly and positively related to precipitation (PPTcool), and
negatively related to growing-season VPD (VPDMJJ), though the
effect of tree status and the slope of the growth response to climate
varied across sites (Tables 6–7; Figs. 3–4, S9–S10). A tree’s growth
was related to eventual tree status more strongly and consistently
in 1950s and SEV2000 trees, confirming a generally stronger
growth-mortality signal when compared to WRK2000 and
TRP2000.
Interaction terms between tree status, PPTcool and VPDMJJ
provide evidence of a differential response to climate amongst
surviving and dying trees (Tables 6–7). Survivors from both the
2000s and 1950s exhibited a generally greater response (steeper
slope) to precipitation than dying trees, driven by an enhanced
growth response during wet years (Tables 6–7; Figs. 3A,C,E,
4A,C). Interactions between tree status and VPDMJJ indicate that
surviving trees also usually had a greater growth response to
VPDMJJ, driven by enhanced growth during years when
Table 5. Site-specific correct classification rates for best-ranked growth-mortality models.
Correct Classification Rates
Site
Dead Trees
Live Trees
All Trees
TRP2000
60.0–63.3%
62.1%
61.0–62.7%
WRK2000
60.0%
60.0–70.0%
60.0–65.5%
BNM2000
12.0%
-
12.0%
SEV2000
73.3%
80.0–83.3%
76.7–78.3%
BNM1950
82.6%
95.5%
88.8%
SEV1950
90.1%
77.3%
84.1%
Each column shows the range of correct classification rates for the three best-ranked general mortality models. Bold typeface highlights correct classification rates that
are consistently above 70%.
doi:10.1371/journal.pone.0092770.t005
PLOS ONE | www.plosone.org
9
May 2014 | Volume 9 | Issue 5 | e92770
4.29
22.50
0.01
0.01
0.04
0.28
0.03
20.13
0.02
PPTCool
(PPTCool)2
VPDMJJ
(VPDMJJ)2
PLOS ONE | www.plosone.org
20.38
23.01
0.06
0.01
0.02
0.01
20.02
20.02
0.03
0.01
0.01
0.03
PPTCool:CI
PPTCool:Live Status
VPDMJJ:CI
VPDMJJ:Live Status
VPDMJJ:CI:Live Status
2.40
0.27
1.26
1.38
0.017
0.790
0.207
0.169
0.003
0.707
0.124
0.014
,0.001
,0.001
,0.001
,0.001
,0.001
P-value
WRK2000
20.03
0.03
0.01
0.05
0.01
0.05
0.01
0.03
0.10
0.06
0.24
0.11
0.01
0.03
0.01
0.03
0.15
SE
20.02
20.02
0.30
20.10
0.06
20.17
20.07
0.26
20.47
b
2.40
20.60
1.26
2.00
20.73
20.38
1.24
20.91
4.58
25.49
26.95
9.20
23.19
t-value
0.017
0.549
0.207
0.045
0.466
0.706
0.217
0.362
,0.001
,0.001
,0.001
,0.001
0.002
P-value
SEV2000
0.03
20.10
0.01
0.08
20.02
20.02
2.40
23.73
0.03
0.01
1.26
2.65
21.73
20.38
4.59
22.47
23.34
25.51
25.01
14.86
25.95
t-value
0.01
0.03
0.01
0.06
0.06
0.14
20.15
0.01
0.01
0.01
0.02
0.08
SE
0.63
20.03
20.08
20.04
0.24
20.49
b
0.017
,0.001
0.207
0.008
0.084
0.707
,0.001
0.015
0.001
,0.001
,0.001
,0.001
,0.001
P-value
Summary of the linear mixed-effects model with the formula RW,PPTCool + (PPTCool)2+ VPDMJJ + (VPDMJJ)2+ CI + Tree Status + Site + CI:Tree Status + PPTCool:CI + PPTCool:Tree Status + VPDMJJ:CI + VPDMJJ:Tree Status + PPTCool:Site +
(PPTCool)2:Site + VPDMJJ:Site + (VPDMJJ)2:Site + CI:Site + Tree Status:Site + PPTCool:CI:Site + PPTCool:Tree Status:Site + VPDMJJ:Tree Status:Site + VPDMJJ:CI:Tree Status, random = (,1 | TreeID). A correlation term and variance weights
were also included in the model in order to account for residual autocorrelation of growth between years and variance heterogeneity of residuals by TreeID and across fitted values [65]. Growth was modeled from 1980, with the
end of the modeled period varying depending on the outer growth year of the dead tree in each pair. Model parameters with estimates of (p,0.05) are in boldface type. The reference level for Tree Status is Dead. Contrasts were
applied to calculate coefficients and significances associated with each site.
doi:10.1371/journal.pone.0092770.t006
0.01
0.02
1.55
CI:Live Status
0.09
20.11
0.14
CI
28.89
0.01
Live Status
4.05
0.01
19.09
27.84
0.06
20.51
Intercept
t-value
SE
b
Model Terms
TRP2000
Table 6. The modeled relationship between tree growth, status, competition and climate in trees from 2000s sites.
91
Tree Growth and Drought-Associated Mortality Risk
10
May 2014 | Volume 9 | Issue 5 | e92770
92
,0.001
24.45
0.02
20.10
,0.001
23.65
0.03
20.10
VPDMJJ:Live Status
Summary of the linear mixed-effects model with the formula RW,PPTCool + (PPTCool)2+ VPDMJJ + (VPDMJJ)2+ Tree Status + Site + PPTCool:Tree Status + VPDMJJ:Tree Status + PPTCool:Site + (PPTCool)2:Site + (VPDMJJ)2:Site + Tree
Status:Site + PPTcool:Tree Status:Site + VPDMJJ:Tree Status:Site, random = (,1 | TreeID). A correlation term and variance weights were also included in the model in order to account for residual autocorrelation of growth between
years and variance heterogeneity of residuals by TreeID and across fitted values [65]. Growth was modeled from 1930, with the end of the modeled period varying depending on the outer growth year of the dead tree in each pair.
Model parameters with estimates of (p,0.05) are in boldface type. The reference level for Tree Status is Dead. Contrasts were applied to calculate coefficients and significances associated with each site.
doi:10.1371/journal.pone.0092770.t007
0.6774
,0.001
7.41
0.42
0.07
0.53
2.53
0.03
0.08
PPTCool:Live Status
PLOS ONE | www.plosone.org
0.01
,0.001
0.0116
7.41
0.07
0.53
Live Status
0.02
,0.001
0.320
20.99
20.01
0.2149
21.24
0.01
20.02
(VPDMJJ)2
0.01
,0.001
26.10
20.09
,0.001
28.79
0.01
20.13
VPDMJJ
0.01
,0.001
213.48
0.01
,0.001
213.48
0.01
20.10
20.10
,0.001
(PPTCool)2
8.33
0.01
0.11
,0.001
20.31
0.02
0.40
29.10
0.05
PPTCool
t-value
SE
20.43
,0.001
25.47
0.06
20.34
(Intercept)
b
P-value
t-value
SE
b
BNM50
Model Terms
Table 7. The modeled relationship between tree growth, status, and climate in trees from 1950s sites.
SEV50
P-value
Tree Growth and Drought-Associated Mortality Risk
Figure 3. The predicted effects of precipitation (PPTcool), vapor
pressure deficit (VPDMJJ) and competition (CI) on growth in
surviving and dying trees from 2000s sites. The relationships
reflect the model shown in Table 6. High and low competition levels are
set to 75th and 25th percentiles of CI, respectively, with the predicted
effects shown separately for TRP2000 (A, B), WRK2000 (C, D) and
SEV2000 (E, F).
doi:10.1371/journal.pone.0092770.g003
atmospheric vapor demand was low (Tables 6–7; Figs. 3B,D,F,
4B,D). However, the interactions between tree status and climate
were weaker and less consistent at TRP2000 (Table 6; Figs. 3B,
S9).
Competition played a complex yet significant role in modulating
tree growth. Growth was negatively affected by the presence of
conspecific neighbors, and competition also reduced the ability of
trees to grow well during years with abundant PPTcool, although
this effect was weaker at WRK2000 and SEV2000 than at
TRP2000 (Table 6; Figs. 3A,C,E, S9). Competition also modulated the response of trees to VPDMJJ, but this effect was
contingent on tree status, with live trees with low CI best able to
exploit years with low atmospheric demand, and the response of
dying trees to VPDMJJ not significantly affected by competition
(Table 6; Figs. 3B,D,F, S9).
The magnitude and significance of the growth predictors shifted
slightly through time, but the direction of effects remained
generally consistent (Figs. S9, S10).
11
May 2014 | Volume 9 | Issue 5 | e92770
93
Tree Growth and Drought-Associated Mortality Risk
droughts by shedding leaf or root mass [21,79,80], other
physiological adjustments [21], and/or they may have sustained
injuries such as loss of xylem conductance [25,81,82]. As a result of
such responses, some trees became more vulnerable to mortality
during subsequent drought [78]. This is consistent with findings on
much shorter time scales in P. sylvestris, where drought and
herbivory-associated reductions in leaf area reduced carbon
uptake and reserves and influenced mortality risk during a
subsequent drought year [21,83]. It is also consistent with the
inclusion of the abrupt growth increase term in our best growthmortality model (Table 4), which reflects how recovery from periods
of lower growth (and presumably higher stress) is important to
mortality risk beyond the influence of average growth rate alone.
Second, the mortality of neighbors during previous decadal
drought may have freed survivors from competition, allowing
some to recover faster and/or boost their productivity. The lack of
significant differences between spatial neighborhoods around dead
and surviving trees at SEV2000 and WRK2000 (Table S1) – sites
that exhibited the strongest growth divergences – along with weak
relationships between mortality severity and tree density or basal
area in neighborhood plots (Fig. S4), suggests that overall tree
density was not the most important factor driving mortality risk
among the trees in our study. However, the negative influence of
conspecific neighbors on piñon growth, and the negative
interaction between tree climate response and competition
suggests that, although the effect may be complex, competition
contributes to growth trajectories, and by extension, likely
influences drought-mortality risk (Fig. 3).
The decades-long divergences in growth between dying and
surviving trees that we observed are in agreement with those
documented in a few other long-term studies [34,36–38], and
suggest that understanding tree recovery after drought may be
critical to understanding the full impacts of drought-mortality
events and anticipating future tree mortality. For example, would
the 2000s die-off have been worse if the drought had occurred a
decade sooner, when fewer trees had sufficiently recovered from
the 1950s drought? Was the severity of mortality during the 2000s
drought contingent upon the character and timing of the 1950s
drought and the climate in the following years? Past divergences in
growth also suggest that there is now a new pool of vulnerable
trees that were injured but not killed during the early 2000s
drought [25], further highlighting that potential changes in the
frequency of drought may dictate the severity of future die-off events,
along with changes in drought intensity and duration.
These results and interpretations are consistent with the view of
extreme ecological events put forth by Gutchick and BassiriRad
[84], in which the consequences of such events are hypothesized to
become most evident during a long recovery period. Among our
trees, structural adjustments or injuries caused by previous severe
drought (as hypothesized above) might have had a genetic
component [84], and the associated fitness costs in terms of lost
growth potential proved fatal, even if decades later. Thus,
drought-mortality processes may be more fully understood if, in
addition to quantifying the instantaneous effects of climate on tree
carbon and hydraulic dynamics, we expand the consideration of
the controls on prolonged recovery from severe drought events,
which may determine how individuals respond to environmental
variability years to decades later [21,84].
Figure 4. The predicted effects of precipitation (PPTcool) and
vapor pressure deficit (VPDMJJ) on growth in surviving and
dying trees from 1950s sites. The relationships reflect the model
shown in Table 7. Predicted effects are shown separately for BNM50 (A,
B) and SEV1950 (C, D).
doi:10.1371/journal.pone.0092770.g004
Discussion
Long-term Factors Predispose Trees to Die during
Drought Events
Radial tree growth serves as a proxy for tree status in the years
preceding death, integrating the effects of drought, injuries, and
tree structural characteristics on the carbon dynamics that can
lead to mortality during prolonged and severe droughts. We
observed consistently lower average growth in trees that died
versus trees that survived, and significantly greater year-to-year
growth variability. The latter may be related to carbon reserves
and thus the relative capacity of trees to buffer themselves against
inter-annual swings in resource availability [39,77], although we
cannot prove a physiological link via our data. Lower average
growth and higher mean sensitivity among dying piñon are
consistent with previous studies in other semi-arid forests
[35,36,42], and with the hypothesis that piñon death during the
droughts of the 1950s and 2000s was related at least in part to
constraints on carbon uptake and/or storage, leading to lower
growth and, ultimately, the inability to meet metabolic requirements or repel attacking insects [15,16,39].
More surprising than the observed growth differences per se is
the fact that at some sites these differences extend over multiple
decades, and yet were not present early in the growth records of
trees (Figs. 1, S5–S8). This is consistent with the decline-disease
theory of tree death [52,78], and suggests that long-term processes
or the contingent effects individuals’ response to previous events
are underlying at least a significant portion of the mortality during
two large, seemingly sudden die-off events. We propose two nonexclusive processes that are consistent with our observations.
First, growth in surviving and dying trees appears to diverge
most strongly after previous record-setting decadal droughts, at
least at some sites (Fig. 1). Some trees may have reacted to these
PLOS ONE | www.plosone.org
Dying Trees Exhibit a Differential Response to Climate
The significantly different growth rates of dying vs. surviving
trees leading up to the 2000s and 1950s droughts are partly due to
differential responses to precipitation and VPD (Tables 6–7;
Figs. 3–4, Figs. S9–S10). McDowell et al. [39] found that growth
12
May 2014 | Volume 9 | Issue 5 | e92770
94
Tree Growth and Drought-Associated Mortality Risk
of P. ponderosa that died during the 2000s drought was more
responsive to a drought index than growth of surviving trees, with
dying trees growing less during the drier years leading up to
mortality. They suggest that this is consistent with trees
predisposed to die by low leaf-level gas exchange and carbon
uptake driven by chronic water stress. However, Ogle et al. [35]
found that growth in mature drought-killed piñon in Arizona was
less responsive to drought variability than in survivors, and Millar
et al. found that limber pine (P. flexilus) [42] and whitebark pine (P.
albicaulis) [43] that died were less responsive to decreasing water
deficit than survivors, at least at high temperatures.
We found that trees predisposed to die exhibited higher mean
sensitivity (e.g. higher year-to-year growth variability, Table 2).
Previous researchers have suggested that high mean sensitivity
reflects greater limitation by inter-annual swings in climate or
other environmental variables (e.g. [77] and see above). Although
dying trees grew less than survivors during hotter, drier years
(Figs. 3–4), our models suggest that growth in dying piñon was
generally less responsive to the overall range of PPTcool than in
survivors (Tables 6–7; Figs. 3–4, S9–S10). The response of dying
trees to VPDMJJ was also generally less pronounced than among
survivors, though this effect was reduced or negligible at
TRP2000. Importantly, these differences were driven by enhanced
growth of surviving trees during wet or cool, rather than dry or hot
years (Figs. 3–4). This suggests that, in addition tree response to
drought stress per se, the ability to maximize photosynthesis and
growth during years with abundant water supply and low VPD
may be an important aspect of tree survival during subsequent
severe or prolonged drought. Recent evidence generated by
precision dating of 14C in carbon within and respired by trees
points to the utilization of years-to-decades-old stored carbohydrates for functions such as dormant-season metabolism, defense
and repair [85–88]. Trees that survived may have been able to
store excess carbon from enhanced photosynthesis during wet
periods, and to use this carbon for defense and metabolism during
subsequent drought periods when growth and photosynthesis were
severely constrained [89].
Modeling Drought-associated Tree Mortality Using
Growth-based Predictor Variables
Do simple metrics of tree productivity and carbon balance
reflect the complex physiological, structural and life history aspects
of individual trees that lead to their mortality during severe
drought events? To our knowledge, our study represents the first to
comprehensively assess and quantify growth-mortality relationships in the context of widespread drought-associated die-off, and
it is one of only a few studies to look at the stability of growthmortality relationships across space and time (but see [24]). Our
models correctly classified ca. 70% of the trees across sites and
drought events, which is comparable to or slightly below correct
classification rates in other growth-mortality studies [29,31]. Thus,
there is promise in using relatively simple, growth-based empirical
approaches for assessing drought-mortality risk, at least for certain
species or functional groups, even when drought and associated
insect activity causes rapid and widespread mortality.
However, we found variable growth-mortality relationships and
thresholds between sites and drought events, and uneven model
performance at the site level. Although individual sites did not
exhibit opposite relationships between growth and mortality, as
with background mortality in some European forests [96], the
northern sites (TRP, WRK, BNM) were characterized by weaker
associations between growth and mortality (Table 5; Fig. 2). This
may represent a shift from mortality factors associated with
chronic constraints on overall tree growth, to factors associated
with shorter-term and/or exogenous factors. These factors include
rapid hydraulic failure or carbon starvation in some more vigorous
trees due to acute drought stress [40,97], and/or the build-up of
bark beetle populations that were able to overcome the resistance
of trees with relatively higher growth rates [98]. Many bark beetles
favor more vigorous trees with larger food stores, but they are only
able to overcome the defenses of such trees at higher population
densities [98]. The build-up of relatively large Ips populations may
have been favored at some sites during the comparatively hot
2000s drought [49], as beetle development is accelerated by warm
conditions [45].
Regardless of the underlying physiological causes, growthmortality models calibrated on empirical data from the 1950s
drought alone would have under-predicted mortality during the
2000s drought, thus underscoring the problems with projecting
future mortality rates using empirical relationships established
based on one drought event or experiment (see also [23]). If bark
beetle attack is an underlying driver, integrating information on
bark beetle dynamics with empirical indicators of tree physiological stress will be important for improving the predictive capacity
of mortality models. Furthermore, our best growth-mortality
models included longer-term, less simplistic growth metrics,
suggesting that the time scales and/or cumulative processes
considered by many current models should be extended (cf. [31]).
Competition Reduces Tree Growth and Modulates Trees’
Ability to Respond to Favorable Climate
High stand density has been found to increase the likelihood of
mortality via competition in many forests [29,52,78]. However,
recent studies of drought-associated mortality patterns in southwestern woodlands have documented variable relationships
between tree density (or basal area) and the severity of mortality
within a stand or site [53,76,90–92]. The proportion of dead trees
in the neighborhood plots at our study sites was also inconsistently
related to tree density or basal area (Fig. S3). However, growth of
piñon trees across sites was negatively influenced by the presence
of conspecific neighbors, with CI also reducing the ability of trees
to take advantage of wet and/or cool conditions over the decades
before drought (Table 6; Figs. 3, S9). Similar models fit with a CI
that was calculated using neighbors of all species produced less
consistent responses across sites (not shown), suggesting differences
in the competitive effects of conspecific vs. heterospecific
neighbors. Thus our results suggest that managing competition
in forests is likely to be important to promote resistance to
mortality during drought (cf. [93,94]). At the same time, it is
important to note the apparent complexity involved, with species
mix in addition to overall tree density needing consideration, as
well as the potential role of stand structural characteristics beyond
their influence on tree vigor (cf. [95]). Ultimately, further study is
required to resolve the role of competition and tree density on
mortality.
PLOS ONE | www.plosone.org
Conclusions
Understanding the processes that underlie drought-related tree
mortality is critical for anticipating future forest dynamics and
associated feedbacks to the earth system, and for developing
management plans that enhance the robustness and resilience of
forests to climate change. Our study documented high levels of
piñon mortality during both the 2000s and 1950s droughts, with
almost ubiquitous evidence of bark beetle activity on dead trees.
More synchronous mortality was observed at northern sites in
New Mexico during the 2000s event. Dying trees generally had
lower average growth rates and greater year-to-year growth
variability than trees that survived, but early in their life, these
13
May 2014 | Volume 9 | Issue 5 | e92770
95
Tree Growth and Drought-Associated Mortality Risk
different overall (p,0.0001, Kolmogorov-Smirnov test), for
TRP2000 (p = 0.0313), WRK2000 (p = 0.0006), and SEV2000
(0.0014). There were not enough living trees to test for differences
at BNM2000.
(EPS)
differences were not evident. Instead, decades-long growth
divergences between surviving and dying trees suggest that recent
growth differences are related to the response and recovery of trees
to previous severe droughts, at least at some sites. This pattern
further suggests that a pool of trees that survived the early 2000s
drought may now be particularly vulnerable during future
droughts. These trees should be investigated in more detail to
reveal the processes that influence their recovery [84].
We show that tree growth response to climate is an important
predisposing factor underlying mortality during widespread,
drought- and insect-related mortality events. In particular, our
results suggest that tree response to wet/cool years, in addition to
the response to drought years, may be an important aspect of
vulnerability. The growth response of surviving trees during very
wet/cool years in the decades preceding mortality events likely
enhanced their carbon reserves, which was important for
withstanding subsequent drought and insect attack. The competitive environment also influenced tree growth and the ability of
trees to respond favorably to wet conditions, suggesting that
controlling tree density is likely to enhance tree resistance to
mortality during drought. However, conspecific vs. heterospecific
competitive effects appear to be different, and should therefore be
considered in detail.
The discriminatory ability of logistic growth-mortality functions
underscores the potential of simple empirical approaches to
represent mortality risk in models of vegetation dynamics, even in
the context of widespread mortality events associated with drought
and bark beetles. However, incorporating multiple and longerterm aspects of tree growth and life history is important for fully
capturing mortality risk. Furthermore, shifting growth-mortality
relationships across space and time point to the challenges
associated with calibrating mortality algorithms. Although we
cannot fully explain the weakening of growth-mortality relationships at the northern study sites during the 2000s drought, one
consistent hypothesis is that bark beetle dynamics played a more
important role in the recent die-off event, shifting the physiological
basis for mortality and highlighting the need for further study of
tree-insect dynamics to improve the prediction of tree mortality
during drought.
Figure S4 The relationship between tree density and
mortality severity. Each point represents the ratio of dead Pinus
edulis (PIED) versus total tree density in 7.5m neighborhood-plots
surrounding target trees at 2000s study sites ((A) TRP2000
(n = 60), (B) WRK2000 (n = 20) and (C) SEV2000 (n = 60)). A
linear regression line is shown to provide a visual estimate of the
relationship. Quasi-binomial regression was used to statistically
assess the direction, magnitude and significance of depicted
relationships. For TRP2000 (A), 0.00091x + 0.18240, p = 0.19.
For WRK2000 (B), y = 0.00246x+0.06733, p = 0.17. For
SEV2000 (C), y = –.00107x-– 0.22008, p = 0.021.
(EPS)
Figure S5 Box and whisker plots of ring widths
averaged over different time intervals. Live tree growth is
truncated at the outside ring date of the dead tree in the tree pair.
Boxes drawn around time intervals on the x-axis denote significant
differences between live and dead trees (p,0.05, Student’s t-test).
(EPS)
Figure S6 Box and whisker plots of growth mean
sensitivity averaged over different time intervals. Live
tree growth is truncated at the outside ring date of the dead tree in
the tree pair. Boxes drawn around time intervals on the x-axis
denote significant differences between live and dead trees (p,0.05,
Student’s t-test).
(EPS)
Box and whisker plots of tree growth trends
averaged over different time intervals. Live tree growth is
truncated at the outside ring date of the dead tree in the tree pair.
Boxes drawn around time intervals on the x-axis denote significant
differences between live and dead trees (p,0.05, Student’s t-test).
(EPS)
Figure S7
Figure S8 Box and whisker plots of the number of
USA. The distribution of P. edulis is shown in light gray.
(EPS)
abrupt growth increases averaged over different time
intervals. Live tree growth is truncated at the outside ring date of
the dead tree in the tree pair. Boxes drawn around time intervals
on the x-axis denote significant differences between live and dead
trees (p,0.05, Student’s t-test).
(EPS)
Figure S2 Relationships between tree diameter and
Figure S9 Coefficients from models relating growth to
radial growth as represented by different growth
metrics. Only live tree data is shown. A linear regression line
with 95% confidence intervals is plotted over the raw data for
relative basal area increments (RelBAI) (A), raw ring widths (RW)
(B), basal area increments (BAI) (C), and ring width indices (RWI)
(D). Trends are significant at the 95% level for (A) and (C), but not
significant for (B) and (D). Growth is represented by an average of
the most recent 3-years in each growth record.
(EPS)
climate, tree status and competition for trees that died
and survived the 2000s drought. Coefficients are from linear
mixed-effects models relating growth (RW) to precipitation
(PPTcool), vapor pressure deficit (VPDMJJ), competition (CI), and
tree status (L/D). Coefficients were calculated separately for
TRP2000 (A), WRK2000 (B), and SEV2000 (B). All predictor
variables were converted to z-scores prior to modeling, allowing
for a direct comparison of coefficients between models. Growth
was modeled starting from five different dates leading up to the
2000s drought-mortality event (bar colors). The end of the
modeled period varied depending on tree pair, with growth in
surviving trees truncated at the outer year of growth in the
corresponding dead tree.
(EPS)
Supporting Information
Figure S1 Study area locations within New Mexico,
Figure S3 Size-class distributions of Pinus edulis that
died and survived the 2000s drought. Distributions are
derived from measurements made in 7.5m neighborhood plots
around each target tree. Bars in each histogram represent 2.5cm
size classes, based on tree diameters at breast height (DBH). Data
are pooled across all 2000s study sites in (A), and shown separately
for TRP2000 (B), BNM2000 (C), WRK2000 (D) and SEV2000
(E). Size-class distributions for live and dead trees are significantly
PLOS ONE | www.plosone.org
Coefficients from models relating growth to
climate and tree status for trees that died and survived
the 1950s drought. Coefficients are from linear mixed-effects
Figure S10
14
May 2014 | Volume 9 | Issue 5 | e92770
96
Tree Growth and Drought-Associated Mortality Risk
models relating growth (RW) to precipitation (PPTcool), vapor
pressure deficit (VPDMJJ), and tree status (L/D). Coefficients were
calculated separately for BNM1950 (A) and SEV50 (B). All
predictor variables were converted to z-scores prior to modeling,
allowing for a direct comparison of coefficients between models.
Growth was modeled starting from five different dates leading up
to the 1950s drought-mortality event (bar colors). The end of the
modeled period varied depending on tree pair, with growth in
surviving trees truncated at the outer year of growth in the
corresponding dead tree.
(EPS)
type. Bootstrapped estimates were generated by fitting models to
1000 samples drawn from the calibration data. The Estimates
columns represent model coefficients for fixed effects and standard
deviations for random effects.
(DOCX)
Table S1 Fine scale spatial patterning of mortality at
We thank Craig D. Allen, Julio L. Betancourt and Thomas J. Swetnam for
access to archived samples, constructive discussions and review of an earlier
draft of the manuscript. The comments of Vincent Gutschick and an
anonymous reviewer greatly improved the manuscript. We appreciate
project support, feedback and review of an earlier draft of the manuscript
from Nate G. McDowell and Connie Woodhouse. Christof Bigler and
Dean Billiheimer helped with statistical methods. We are grateful for field
assistance from Kay Beeley, Collin Haffey, Matthias Kläy, Derek Murrow
and Greg Pederson.
Text S1 Justification for the choice of climate predictors in linear mixed-effects models.
(DOCX)
Acknowledgments
2000s sites. Significant differences in tree density and basal area
in neighborhood plots surrounding dead versus living target trees
are in boldface type (p,0.05, Student’s t-test). PIED is Pinus edulis.
JUMO is Juniperus monosperma.
(DOCX)
Table S2 Bootstrapped estimates and confidence intervals for model terms in the best-ranked growthmortality model. The model formula is Tree Status
,log(RW30) + Sens15 + AbruptIncreases10 + (1 + log(RW30) |
Site), with validation statistics shown in Table 4. Variables include
average growth (RW), mean sensitivity (Sens), and the number of
abrupt growth increases (AbruptIncreases), with the number of
years over which variables were averaged indicated after variable
Author Contributions
Conceived and designed the experiments: AKM HB. Performed the
experiments: AKM HB. Analyzed the data: AKM HB. Contributed
reagents/materials/analysis tools: AKM HB. Wrote the paper: AKM HB.
References
1. Van Mantgem PJ, Stephenson NL, Byrne JC, Daniels LD, Franklin JF, et al.
(2009) Widespread Increase of Tree Mortality Rates in the Western United
States. Science 323: 521–524. doi:10.1126/science.1165000.
2. Phillips OL, Aragão LEOC, Lewis SL, Fisher JB, Lloyd J, et al. (2009) Drought
Sensitivity of the Amazon Rainforest. Science 323: 1344–1347. doi:10.1126/
science.1164033.
3. Peng C, Ma Z, Lei X, Zhu Q, Chen H, et al. (2011) A drought-induced
pervasive increase in tree mortality across Canada’s boreal forests. Nature
Climate Change 1: 467–471. doi:10.1038/nclimate1293.
4. Carnicer J, Coll M, Ninyerola M, Pons X, Sánchez G, et al. (2011) Widespread
crown condition decline, food web disruption, and amplified tree mortality with
increased climate change-type drought. Proceedings of the National Academy of
Sciences: 201010070. doi:10.1073/pnas.1010070108.
5. Williams AP, Allen CD, Macalady AK, Griffin D, Woodhouse CA, et al. (2013)
Temperature as a potent driver of regional forest drought stress and tree
mortality. Nature Climate Change 3: 292–297. doi:10.1038/nclimate1693.
6. Allen CD, Macalady AK, Chenchouni H, Bachelet D, McDowell N, et al. (2010)
A global overview of drought and heat-induced tree mortality reveals emerging
climate change risks for forests. Forest Ecology and Management 259: 660–684.
doi:10.1016/j.foreco.2009.09.001.
7. Bonan G (2008) Carbon cycle: Fertilizing change. Nature Geoscience 1: 645–
646. doi:10.1038/ngeo328.
8. Kurz WA, Dymond CC, Stinson G, Rampley GJ, Neilson ET, et al. (2008)
Mountain pine beetle and forest carbon feedback to climate change. Nature 452:
987–990.
9. Breshears DD, López-Hoffman L, Graumlich LJ (2011) When ecosystem
services crash: preparing for big, fast, patchy climate change. Ambio 40: 256–
263.
10. Reichstein M, Bahn M, Ciais P, Frank D, Mahecha MD, et al. (2013) Climate
extremes and the carbon cycle. Nature 500: 287–295. doi:10.1038/nature12350.
11. Bugmann H (2001) A review of forest gap models. Climatic Change 51: 259–
305. doi:10.1023/A:1012525626267.
12. Keane RE, Austin M, Field C, Huth A, Lexer MJ, et al. (2001) Tree mortality in
gap models: application to climate change. Climatic Change 51: 509–540.
13. Fisher R, McDowell N, Purves D, Moorcroft P, Sitch S, et al. (2010) Assessing
uncertainties in a second-generation dynamic vegetation model caused by
ecological scale limitations. New Phytologist 187: 666–681.
14. Seidl R, Fernandes PM, Fonseca TF, Gillet F, Jönsson AM, et al. (2011)
Modelling natural disturbances in forest ecosystems: a review. Ecological
Modelling 222: 903–924. doi:10.1016/j.ecolmodel.2010.09.040.
15. McDowell NG, Beerling DJ, Breshears DD, Fisher RA, Raffa KF, et al. (2011)
The interdependence of mechanisms underlying climate-driven vegetation
mortality. Trends in Ecology & Evolution 26: 523–532. doi:10.1016/
j.tree.2011.06.003.
16. McDowell N, Pockman WT, Allen CD, Breshears DD, Cobb N, et al. (2008)
Mechanisms of plant survival and mortality during drought: why do some plants
survive while others succumb to drought? New Phytologist 178: 719–739.
PLOS ONE | www.plosone.org
17. Sala A, Piper F, Hoch G (2010) Physiological mechanisms of drought-induced
tree mortality are far from being resolved. New Phytologist 186: 274–281.
doi:10.1111/j.1469–8137.2009.03167.x.
18. McDowell NG (2011) Mechanisms linking drought, hydraulics, carbon
metabolism, and vegetation mortality. Plant Physiology 155: 1051.
19. Choat B, Jansen S, Brodribb TJ, Cochard H, Delzon S, et al. (2012) Global
convergence in the vulnerability of forests to drought. Nature 491: 752–755.
doi:10.1038/nature11688.
20. Zeppel MJ, Adams HD, Anderegg WR (2011) Mechanistic causes of tree
drought mortality: recent results, unresolved questions and future research
needs. New Phytologist 192: 800–803.
21. Bréda N, Huc R, Granier A, Dreyer E (2006) Temperate forest trees and stands
under severe drought: a review of ecophysiological responses, adaptation
processes and long-term consequences. Annals of Forest Science 63: 625–644.
22. Anderegg WR, Kane JM, Anderegg LD (2013) Consequences of widespread tree
mortality triggered by drought and temperature stress. Nature Climate Change
3: 30–36.
23. McDowell NG, Fisher RA, Xu C, Domec JC, Hölttä T, et al. (2013) Evaluating
theories of drought-induced vegetation mortality using a multimodel–experiment framework. New Phytologist 200: 304–321.
24. Wunder J (2007) Conceptual advancement and ecological applications of tree
mortality models based on tree-ring and forest inventory data [Ph.D.
dissertation]. Zurich, Switzerland: ETH Zurich.
25. Anderegg WRL, Plavcová L, Anderegg LDL, Hacke UG, Berry JA, et al. (2013)
Drought’s legacy: multiyear hydraulic deterioration underlies widespread aspen
forest die-off and portends increased future risk. Global Change Biology 19:
1188–1196. doi:10.1111/gcb.12100.
26. Waring R (1987) Characteristics of trees predisposed to die. BioScience 37: 569–
574.
27. Dobbertin M (2005) Tree growth as indicator of tree vitality and of tree reaction
to environmental stress: a review. European Journal of Forest Research 124:
319–333. doi:10.1007/s10342-005-0085-3.
28. Wyckoff PH, Clark JS (2002) The relationship between growth and mortality for
seven co-occurring tree species in the southern Appalachian Mountains. Journal
of Ecology 90: 604–615.
29. Bigler C, Bugmann H (2003) Growth-dependent tree mortality models based on
tree rings. Canadian Journal of Forest Research 33: 210–221.
30. Bigler C, Bugmann H (2004) Assessing the performance of theoretical and
empirical tree mortality models using tree-ring series of Norway spruce.
Ecological Modelling 174: 225–239.
31. Das AJ, Battles JJ, Stephenson NL, Van Mantgem PJ (2007) The relationship
between tree growth patterns and likelihood of mortality: a study of two tree
species in the Sierra Nevada. Canadian Journal of Forest Research 37: 580–597.
32. Wunder J, Reineking B, Bigler C, Bugmann H (2008) Predicting tree mortality
from growth data: how virtual ecologists can help real ecologists. Journal of
Ecology 96: 174–187.
15
May 2014 | Volume 9 | Issue 5 | e92770
97
Tree Growth and Drought-Associated Mortality Risk
33. Hawkes C (2000) Woody plant mortality algorithms: description, problems and
progress. Ecological Modelling 126: 225–248.
34. Pedersen BS (1998) The role of stress in the mortality of midwestern oaks as
indicated by growth prior to death. Ecology 79: 79–93.
35. Ogle K, Whitham TG, Cobb NS (2000) Tree-ring variation in pinyon predicts
likelihood of death following severe drought. Ecology 81: 3237–3243.
36. Suarez ML, Ghermandi L, Kitzberger T (2004) Factors predisposing episodic
drought-induced tree mortality in Nothofagus– site, climatic sensitivity and
growth trends. Journal of Ecology 92: 954–966. doi:10.1111/j.13652745.2004.00941.x.
37. Bigler C, Bräker OU, Bugmann H, Dobbertin M, Rigling A (2006) Drought as
an inciting mortality factor in Scots pine stands of the Valais, Switzerland.
Ecosystems 9: 330–343.
38. Bigler C, Gavin DG, Gunning C, Veblen TT (2007) Drought induces lagged
tree mortality in a subalpine forest in the Rocky Mountains. Oikos 116: 1983–
1994.
39. McDowell N, Allen CD, Marshall L (2010) Growth, carbon-isotope discrimination, and drought-associated mortality across a Pinus ponderosa elevational
transect. Global Change Biology 16: 399–415.
40. Levanic T, Cater M, McDowell NG (2011) Associations between growth, wood
anatomy, carbon isotope discrimination and mortality in a Quercus robur forest.
Tree physiology 31: 298–308. doi:10.1093/treephys/tpq111.
41. Hereş A-M, Martı́nez-Vilalta J, López BC (2012) Growth patterns in relation to
drought-induced mortality at two Scots pine (Pinus sylvestris L.) sites in NE
Iberian Peninsula. Trees 26: 621–630. doi:10.1007/s00468-011-0628-9.
42. Millar CI, Westfall RD, Delany DL (2007) Response of high-elevation limber
pine (Pinus flexilis) to multiyear droughts and 20th-century warming, Sierra
Nevada, California, USA. Canadian Journal of Forest Research 37: 2508–2520.
doi:10.1139/X07-097.
43. Millar CI, Westfall RD, Delany DL, Bokach MJ, Flint AL, et al. (2012) Forest
mortality in high-elevation whitebark pine (Pinus albicaulis) forests of eastern
California, USA; influence of environmental context, bark beetles, climatic
water deficit, and warming. Canadian Journal of Forest Research 42: 749–765.
doi:10.1139/x2012-031.
44. Breshears DD, Cobb NS, Rich PM, Price KP, Allen CD, et al. (2005) Regional
vegetation die-off in response to global-change-type drought. Proceedings of the
National Academy of Sciences of the United States of America 102: 15144.
45. Bentz BJ, Régnière J, Fettig CJ, Hansen EM, Hayes JL, et al. (2010) Climate
change and bark beetles of the western United States and Canada: direct and
indirect effects. BioScience 60: 602–613.
46. Anderegg WR, Berry JA, Smith DD, Sperry JS, Anderegg LD, et al. (2012) The
roles of hydraulic and carbon stress in a widespread climate-induced forest dieoff. Proceedings of the National Academy of Sciences 109: 233–237.
47. Raffa KF, Aukema BH, Bentz BJ, Carroll AL, Hicke JA, et al. (2008) Cross-scale
Drivers of Natural Disturbances Prone to Anthropogenic Amplification: The
Dynamics of Bark Beetle Eruptions. BioScience 58: 501. doi:10.1641/B580607.
48. Adams HD, Guardiola-Claramonte M, Barron-Gafford GA, Villegas JC,
Breshears DD, et al. (2009) Temperature sensitivity of drought-induced tree
mortality portends increased regional die-off under global-change-type drought.
Proceedings of the National Academy of Sciences 106: 7063.
49. Weiss JL, Castro CL, Overpeck JT (2009) Distinguishing Pronounced Droughts in
the Southwestern United States: Seasonality and Effects of Warmer Temperatures. Journal of Climate 22: 5918–5932. doi:10.1175/2009JCLI2905.1.
50. Allen CD, Breshears DD (1998) Drought-induced shift of a forest–woodland
ecotone: rapid landscape response to climate variation. Proceedings of the
National Academy of Sciences of the United States of America 95: 14839.
51. Swetnam TW, Betancourt JL (1998) Mesoscale disturbance and ecological
response to decadal climatic variability in the American Southwest. Journal of
Climate 11: 3128–3147.
52. Franklin JF, Shugart HH, Harmon ME (1987) Tree Death as an Ecological
Process. BioScience 37: 550–556. doi:10.2307/1310665.
53. Floyd ML, Clifford M, Cobb NS, Hanna D, Delph R, et al. (2009) Relationship
of stand characteristics to drought-induced mortality in three Southwestern
piñon-juniper woodlands. Ecological Applications 19: 1223–1230.
54. Stokes MA, Smiley TL (1968) An introduction to tree-ring dating. Tucson, AZ:
University of Arizona Press. 73 p.
55. Holmes RL (1983) Computer-assisted quality control in tree-ring dating and
measurement. Tree-Ring bulletin 43: 69–78.
56. Wood S (1982) The Bark and Ambrosia Beetles of North and Central America
(Coleoptera: Scolytidae), a Taxonomic Monograph. Great Basin Naturalist
Memoirs, Brigham Young University Number 6.
57. Rogers TJ (1993) Insect and disease associates of the piñon-juniper woodlands.
Gen. Tech. Rep. RM-236. Santa Fe, NM: US Department of Agriculture,
Forest Service, Rocky Mountain Forest and Range Experiment Station. 42–62.
58. Perkins DL, Swetnam TW (1996) A dendroecological assessment of whitebark
pine in the Sawtooth-Salmon River region, Idaho. Canadian Journal of Forest
Research 26: 2123–2133.
59. Applequist M (1958) A simple pith locator for use with off-center increment
cores. Journal of Forestry 56: 141.
60. Wunder J, Reineking B, Matter JF, Bigler C, Bugmann H (2007) Predicting tree
death for Fagus sylvatica and Abies alba using permanent plot data. Journal of
Vegetation Science 18: 525–534.
PLOS ONE | www.plosone.org
61. Biondi F, Qeadan F (2008) A theory-driven approach to tree-ring standardization: defining the biological trend from expected basal area increment. TreeRing Research 64: 81–96.
62. Bigler C, Bugmann H (2004) Predicting the time of tree death using
dendrochronological data. Ecological Applications 14: 902–914.
63. Strackee J, Jansma E (1992) The statistical properties of mean sensitivity–a
reappraisal. Dendrochronologia 10: 121–135.
64. Biondi F, Qeadan F (2008) Inequality in paleorecords. Ecology 89: 1056–1067.
doi:10.1890/07-0783.1.
65. Pinheiro JC, Bates DM (2000) Mixed-Effects Models in S and S-Plus. Springer.
560 p.
66. Agresti A (2013) Categorical data analysis. 3rd ed. Hoboken, NJ: John Wiley &
Sons. 744 p.
67. Burnham KP, Anderson D (2002) Model Selection and Multi-Model
Inference: A Practical Information-Theoretic Approach. 2nd ed. New York,
NY: Springer-Verlag New York. 496 p.
68. Harrell FE (2001) Regression modeling strategies: with applications to linear
models, logistic regression, and survival analysis. New York, NY: SpringerVerlag New York. 568 p.
69. Zuur AF, Ieno EN, Walker NJ, Saveliev AA, Smith GM (2009) Mixed effects
models and extensions in ecology with R. Ney York, NY: Springer-Verlag New
York. 580 p.
70. Fawcett T (2003) ROC graphs: Notes and practical considerations for data
mining researchers. Technical Report HPL-2003-4. Palo Alto, CA, USA: HP
Laboratories. Available: http://www.hpl.hp.com/techreports/2003/HPL-20034.pdf.
71. Hosmer DW, Lemeshow S, Sturdivant RX (2013) Applied logistic regression.
3rd ed. Hoboken, NJ: John Wiley & Sons. 528 p.
72. Daly C, Neilson RP, Phillips DL (1994) A Statistical-Topographic Model for
Mapping Climatological Precipitation over Mountainous Terrain. Journal of
Applied Meteorology 33: 140–158. doi:10.1175/1520-0450(1994)033,
0140:ASTMFM.2.0.CO;2.
73. R Core Team (2013) R: A language and environment for statistical computing.
Vienna, Austria: R Foundation for Statistical Computing. Available: http://
www.R-project.org/.
74. Bunn A (2008) A dendrochronology program library in R (dplR). Dendrochronologia 26: 115–124. doi:10.1016/j.dendro.2008.01.002.
75. Bates DM, Maechler M, Bolker B (2013) lme4: Linear-mixed Effects Models
Using S4 Classes.
76. Clifford MJ, Royer PD, Cobb NS, Breshears DD, Ford PL (2013) Precipitation
thresholds and drought-induced tree die-off: insights from patterns of Pinus
edulis mortality along an environmental stress gradient. New Phytologist 200:
413–421. doi:10.1111/nph.12362.
77. Fritts HC, Smith DG, Cardis JW, Budelsky CA (1965) Tree-ring characteristics
along a vegetation gradient in northern Arizona. Ecology: 394–401.
78. Manion PD (1991) Tree disease concepts. 2nd ed. Prentice Hall. 422 p.
79. Poyatos R, Aguadé D, Galiano L, Mencuccini M, Martı́nez-Vilalta J (2013)
Drought-induced defoliation and long periods of near-zero gas exchange play a
key role in accentuating metabolic decline of Scots pine. New Phytologist 200:
388–401. doi:10.1111/nph.12278.
80. Landhäusser S, Lieffers V (2012) Defoliation increases risk of carbon starvation
in root systems of mature aspen. Trees 26: 653–661. doi:10.1007/s00468-0110633-z.
81. Plaut JA, Wadsworth WD, Pangle R, Yepez EA, McDowell NG, et al. (2013)
Reduced transpiration response to precipitation pulses precedes mortality in a
pinon–juniper woodland subject to prolonged drought. New Phytologist 200:
375–387.
82. Hacke UG, Stiller V, Sperry JS, Pittermann J, McCulloh KA (2001) Cavitation
Fatigue. Embolism and Refilling Cycles Can Weaken the Cavitation Resistance
of Xylem. Plant Physiology 125: 779–786.
83. Galiano L, Martı́nez-Vilalta J, Lloret F (2011) Carbon reserves and canopy
defoliation determine the recovery of Scots pine 4 yr after a drought episode.
New Phytologist 190: 750–759. doi:10.1111/j.1469-8137.2010.03628.x.
84. Gutschick VP, BassiriRad H (2003) Extreme events as shaping physiology,
ecology, and evolution of plants: toward a unified definition and evaluation of
their consequences. New Phytologist 160: 21–42.
85. Guérard N, Maillard P, Bréchet C, Lieutier F, Dreyer E (2007) Do trees use
reserve or newly assimilated carbon for their defense reactions? A 13C labeling
approach with young Scots pines inoculated with a bark-beetle-associated fungus
(Ophiostoma brunneo ciliatum). Annals of Forest Science 64: 601–608.
doi:10.1051/forest:2007038.
86. Carbone MS, Czimczik CI, Keenan TF, Murakami PF, Pederson N, et al.
(2013) Age, allocation and availability of nonstructural carbon in mature red
maple trees. New Phytologist: n/a–n/a. doi:10.1111/nph.12448.
87. Richardson AD, Carbone MS, Keenan TF, Czimczik CI, Hollinger DY, et al.
(2013) Seasonal dynamics and age of stemwood nonstructural carbohydrates in
temperate forest trees. New Phytologist 197: 850–861.
88. Gaylord ML, Kolb TE, Pockman WT, Plaut JA, Yepez EA, et al. (2013)
Drought predisposes piñon-juniper woodlands to insect attacks and mortality.
New Phytologist 198: 567–578. doi:10.1111/nph.12174.
89. Breshears DD, Myers OB, Meyer CW, Barnes FJ, Zou CB, et al. (2008) Tree
die-off in response to global change-type drought: mortality insights from a
decade of plant water potential measurements. Frontiers in Ecology and the
Environment 7: 185–189.
16
May 2014 | Volume 9 | Issue 5 | e92770
98
Tree Growth and Drought-Associated Mortality Risk
95. Fettig CJ, Klepzig KD, Billings RF, Munson AS, Nebeker TE, et al. (2007) The
effectiveness of vegetation management practices for prevention and control of
bark beetle infestations in coniferous forests of the western and southern United
States. Forest Ecology and Management 238: 24–53.
96. Wunder J, Brzeziecki B, Żybura H, Reineking B, Bigler C, et al. (2008) Growth–
mortality relationships as indicators of life-history strategies: a comparison of
nine tree species in unmanaged European forests. Oikos 117: 815–828.
97. Sthultz CM, Gehring CA, Whitham TG (2009) Deadly combination of genes
and drought: increased mortality of herbivore-resistant trees in a foundation
species. Global Change Biology 15: 1949–1961. doi:10.1111/j.13652486.2009.01901.x.
98. Boone CK, Aukema BH, Bohlmann J, Carroll AL, Raffa KF (2011) Efficacy of
tree defense physiology varies with bark beetle population density: a basis for
positive feedback in eruptive species. Canadian Journal of Forest Research 41:
1174–1188.
99. Fleiss JL, Levin B, Paik MC (2013) Statistical methods for rates and proportions.
3rd ed. Hoboken, NJ: John Wiley & Sons. 800 p.
90. Negron JF, Wilson JL (2003) Attributes associated with probability of infestation
by the piñon ips, Ips confusus (Coleoptera: Scolytidae), in piñon pine, Pinus
edulis. Western North American Naturalist 63: 440–451.
91. Greenwood DL, Weisberg PJ (2008) Density-dependent tree mortality in
pinyon-juniper woodlands. Forest Ecology and Management 255: 2129–2137.
92. Santos MJ, Whitham TG (2010) Predictors of Ips confusus Outbreaks During a
Record Drought in Southwestern USA: Implications for Monitoring and
Management. Environmental Management 45: 239–249. doi:10.1007/s00267009-9413-6.
93. Linares JC, Camarero JJ, Carreira JA (2010) Competition modulates the
adaptation capacity of forests to climatic stress: insights from recent growth
decline and death in relict stands of the Mediterranean fir Abies pinsapo. Journal
of Ecology 98: 592–603. doi:10.1111/j.1365-2745.2010.01645.x.
94. Ruiz-Benito P, Lines ER, Gómez-Aparicio L, Zavala MA, Coomes DA (2013)
Patterns and Drivers of Tree Mortality in Iberian Forests: Climatic Effects Are
Modified by Competition. PLoS ONE 8: e56843. doi:10.1371/journal.
pone.0056843.
PLOS ONE | www.plosone.org
17
May 2014 | Volume 9 | Issue 5 | e92770
Supporting Information
Figure S1.
99
100
0.15
A
B
2.0
3-Yr RW (mm yr-1)
3-Yr RelBAI (cm2 yr-1 yr-1)
1.5
0.10
1.0
0.05
0.5
0.00
5
15
20
25
30 5
0.0
10
15
20
25
30
D
C
2.5
2.0
6.0
3-Yr RWI
3-Yr BAI (cm2 yr-1)
8.0
10
1.5
4.0
1.0
2.0
0.5
0.0
5
Figure S2.
10
15 20 25
DBH (cm)
30 5
10
15 20 25
DBH (cm)
30
0.0
101
A
75
Tree Status
Dead
Live
Count
50
25
0
0
20
25 C
20
15
10
5
0
60
40 B
30
20
Count
10
0
20 D
E
15
40
10
20
5
0
40
0
Figure S3.
20
40
0
0
DBH (cm)
20
40
102
1.0
A
B
C
Ratio of Dead PIED
0.8
●
0.6
0.4
0.2
0.0
0
Figure S4.
500 1000 1500 2000
0
500 1000 1500 2000
Tree Denisty (stems/ha)
0
500 1000 1500 2000
Figure S5.
50
18751900
Lifetime
40
30
20
10
C
50
18251850
Lifetime
40
30
20
10
E
5
50
18751900
Lifetime
40
30
20
10
B
5
50
18251850
Lifetime
Years
40
30
20
10
D
5
50
18751900
Lifetime
40
30
20
10
A
5
50
18751900
Lifetime
40
30
2.0
20
5
2.0
10
5
Ring Width (mm2 yr-1)
103
Tree Status
3.0
D
L
1.0
2.0
1.0
0.0
F
0.0
2.0
1.0
1.0
0.0
0.0
Figure S6.
Years
50
18251850
Lifetime
50
18751900
Lifetime
40
30
20
10
1.0
0.5
0.5
0.0
0.0
C
40
30
20
10
E
5
50
18751900
Lifetime
40
30
20
10
5
50
18751900
Lifetime
40
30
20
1.0
B
5
50
18251850
Lifetime
40
30
20
D
10
5
1.5
Tree Status 2.0
D
L
1.5
A
5
50
18751900
Lifetime
40
30
20
2.0
10
2.0
10
5
Mean Sensitivity
104
F
2.0
1.5
1.5
1.0
1.0
0.5
0.5
0.0
0.0
105
0.2
A
B
C
0.0
0.1
-0.2
Growth Trend (mm2 yr-1)
0.0
-0.1
-0.2
0.2
5
10
20
30
40
50
D
5
10
20
30
40
5
50
E
10
20
Tree Status -0.4
D
L
-0.6
30 40 50
F
0.2
0.1
0.1
0.0
0.0
-0.1
-0.1
-0.2
5
10
20
30
40
50
5
10
20
30
Years
Figure S7.
40
50
5
10
20
30
40
50
-0.2
106
30
A
B
C
20
Tree Status 30
D
25
L
20
15
15
No. of Abrupt Growth Increses
25
10
10
5
5
0
30
5
10
20
30
40
50
D
5
10
20
30
40
50
E
5
10
20
30
40
50
0
30
F
25
25
20
20
15
15
10
10
5
5
0
0
5
10
Figure S8.
20
30
40
50
5
10
20 30
Years
40
50
5
10
20
30
40
50
Figure S9.
0.4
VPDMJJ:CI:
Live Status
Model coefficients
A
VPDMJJ:
Live Status
VPDMJJ:
CI
PPTcool:
Live Status
PPTcool:
CI
CI:
Live Status
Live
Status
CI
0.6
(VPDMJJ)2
0.6
VPDMJJ
Model coefficients
0.6
(PPTcool)2
PPTcool
Model coefficients
107
1950
1960
1970
1980
1990
0.2
0.0
-0.2
B
0.4
0.2
0.0
-0.2
C
0.4
0.2
0.0
-0.2
-0.4
Figure S10.
A
0.4
VPDMJJ:
Live Status
PPTcool:
Live Status
Live Status
(VPDMJJ)2
0.6
VPDMJJ
Model coefficients
0.6
(PPTcool)2
PPTcool
Model coefficients
108
1900
1910
1920
1930
1940
0.2
0.0
-0.2
-0.4
B
0.4
0.2
0.0
-0.2
-1
Density (trees ha )
Total
PIED
JUMO
Dead PIED
Dead JUMO
Basal Area (m2 ha-1)
Total
PIED
JUMO
Dead PIED
Dead JUMO
Table S1.
511.2
369.7
84.9
279.2
3.8
9.9
7.1
2.8
5.3
0.0
454.7
296.6
101.5
171.7
2.0
7.6
5.0
2.6
2.4
0.3
TRP2000
Live
Dead
0.218
0.218
0.982
0.005
0.619
0.124
0.036
0.437
0.005
0.591
P-value
5.4
2.1
3.0
1.9
0.0
582.4
158.5
384.8
147.1
5.7
8.6
3.8
4.7
2.4
0.0
628.1
254.7
345.2
220.7
0.0
WRK2000
Live
Dead
0.149
0.158
0.211
0.555
0.331
0.791
0.189
0.761
0.292
0.331
P-value
11.3
3.2
8.2
0.6
0.4
775.3
318.8
399.9
54.7
11.3
11.1
3.1
8.0
0.8
0.4
790.4
343.3
390.5
60.4
13.2
SEV2000
Live
Dead
0.820
0.674
0.654
0.443
0.801
0.790
0.829
0.661
0.688
0.758
P-value
109
110
Fixed Effects
(Intercept)
log(RW30)
Sens15
AbruptIncreases10
Random Effects
Site: (Intercept)
Site: log(RW30) (Intercept)
Site: log(RW30)
Table S2.
Estimate
SE
Bootstrapped
Estimate
Bootstrapped
Range
(95% CI)
4.437
2.799
-3.26
0.208
1.171
0.893
1.238
0.07
4.851
3.089
-3.46
0.224
(3.231, 6.947)
(2.037, 4.640)
(-6.046, -0.859)
(0.078, 0.379)
1.958
1.62
0
-
2.364
1.955
0.042
(1.167, 4.128)
(0.941, 3.550)
(0.000, 0.415)
111
Text S1.
Justification for the choice of climate predictors in linear mixed-effects models
Climate covariates in linear mixed-effects models were chosen based on initial
comparisons between RWI mean-value chronologies and PRISM climate model output
(4km resolution) [1]. We downloaded monthly (1898-2010) PRISM temperature,
precipitation, and dew point data for each study site, and from these data calculated
monthly precipitation minus potential evapotranspiration (P-PET) according to [2] and
vapor pressure deficit (VPD) according to [3]. Monthly climate variables and RWI
chronologies were entered as inputs to the program Seascorr, which utilizes correlation
and bootstrapped estimations of correlation significance to assess the strength of
relationships between growth chronologies and monthly, seasonal, and annual climatic
variables [4]. RWI correlated most strongly with PRISM-derived precipitation and P-PET
averaged over the nine-month period ending in May or June of the growth year,
depending on site. May-July or May-August and previous September-November average
temperature and VPD consistently resulted in the highest correlations between annual
growth and temperature-related variables. These correlation windows are consistent with
findings from other studies of conifer climate response in the region [3]. For ease of
interpretation, we used 9-month cool season precipitation ending in May and May-July
average VPD as the climatic predictors in mixed effects models presented here. Models
using slightly different seasonal combinations of precipitation and VPD, as well average
temperature and P-PET covariates, were also evaluated but are not shown because results
were very similar.
112
Literature Cited
1. Daly C, Neilson RP, Phillips DL. A Statistical-Topographic Model for Mapping
Climatological Precipitation over Mountainous Terrain. Journal of Applied
Meteorology. 1994;33: 140–158.
2. Hamon WR. Estimating potential evapotranspiration. Proceedings of the American
Society of Civil Engineers. 1961. pp. 107–120.
3. Williams AP, Allen CD, Macalady AK, Griffin D, Woodhouse CA, Meko DM, et al.
Temperature as a potent driver of regional forest drought stress and tree mortality.
Nature Climate Change. 2013;3: 292–297.
4. Meko DM, Touchan R, Anchukaitis KJ. Seascorr: A MATLAB program for
identifying the seasonal climate signal in an annual tree-ring time series. Computers
& Geosciences. 2011;37: 1234–1241.
113
APPENDIX C: MORTALITY RISK OF AN ARIDLANDS CONIFER DURING SEVERE DROUGHT DEPENDS ON RADIAL GROWTH AND INVESTMENT INTO
DEFENSE
This paper was prepared to submit to Oecologia.
Alison K. Macalady, Matthias Kläy, Monica L. Gaylord, Nathan B. English, Craig D.
Allen, Thomas W. Swetnam, Nate G. McDowell, Harald Bugmann
Abstract
Drought and insects frequently interact to produce widespread tree mortality in
forests. However, our ability to project future forest dynamics is limited because the processes leading to tree death during drought are poorly understood. Tree defenses may be
a key element for survival, but there are few empirical tests of their importance. Using
Pinus edulis as a case study, we measured properties of resin ducts – a critical component
of the defense system against bark beetles – and growth in tree rings from trees that died
and survived two severe droughts (1950s, early 2000s) at four sites in New Mexico,
USA. Recent resin duct number, size, and the ratio of resin duct to xylem area were significantly higher in trees that survived, with ~58% larger resin ducts in surviving trees.
Radial growth was also higher on average in trees that survived, but was not as consistently different across sites and drought events. Statistical models that consider both
growth and resin ducts best reflected mortality risk overall, with correct classification
rates exceeding 80%. However, defense attributes were most important to predicting tree
114
survival during the most recent drought, which is likely to reflect amplified bark beetle
pressure, particularly in the more recent drought (2002/2003). These results suggest that
accounting for defense traits and carbon allocation priorities will be important for improving predictions of future drought-associated mortality.
Key words: Tree mortality, drought, tree defense, Ips confuses, Pinus edulis, tree rings.
Introduction
The sensitivity of tree mortality to drought and high temperature has been documented across many ecosystems, and mortality may be increasing globally with increasing temperature (Allen et al. 2010). The physiological mechanisms and thresholds that
lead to the death of individual trees or to widespread mortality during drought remain unclear, however, and thus an accurate representation of tree mortality in vegetation models
remains challenging (Keane et al. 2001; Seidl et al. 2011; McDowell et al. 2011).
Our understanding of carbon and hydraulic constraints within trees that die during
hot and dry conditions is fast improving. Symptoms of dying trees include limited carbon
uptake (Breshears et al. 2009; McDowell et al. 2010), increased carbon metabolism (e.g.,
respiration) (Adams et al. 2009; Zhao et al. 2013), hydraulic failure (Martínez-Vilalta et
al. 2002; Anderegg et al. 2012), and interactions between these processes (Plaut et al.
2012; Hartmann et al. 2013; McDowell et al. 2013; Sevanto et al. 2014). However, the
allocation of available resources, e.g., to growth, defense, storage, reproduction, has received relatively little attention, and the importance of prior tree carbon allocation strate-
115
gies for tree survival during drought have not been thoroughly investigated (Seidl et al.
2011; McDowell et al. 2013).
Tree allocation to defense may be particularly important in the context of droughtassociated conifer mortality. Biotic agents are often involved in drought-mortality events
(Allen et al. 2010), and in western North America, bark beetle activity has been widely
associated with conifer die-off (Raffa et al. 2008; Bentz et al. 2009; Williams et al. 2013;
Gaylord et al. 2013). Conifers have evolved carbon-rich, resin-based defense systems to
fend off attacking insects (Franceschi et al. 2005), but the importance of tree resource allocation to defense vs. other carbon pools for determining conifer resistance to the combined effects of drought and insect attack is largely unknown (Koricheva et al. 1998;
Stamp 2003).
Two hypotheses of tree defense against insect attack may be particularly relevant
in the context of drought-and bark-beetle related mortality. The ‘tree stress hypothesis’
(TSH) emphasizes that trees subject to low resource availability should be more susceptible to mortality from insect attack or drought. The rationale here is that lower overall tree
resource availability results in proportional decreases in carbon uptake, growth, stored
carbon, and capability to defend against insects. Thus any of these pools should be positively related to mortality due to drought and/or bark beetle attack (White 1969; Waring
and Pitman 1983; Waring and Pitman 1985; Waring 1987). Support for the TSH comes
from studies that show positive relationships between growth rates and survival during
drought and beetle infestations (Bigler et al. 2006; Dobbertin et al. 2007; McDowell et al.
2010; Williams et al. 2013; Macalady and Bugmann 2014), reports of fewer bark beetle
116
attacks in faster-growing trees and stands (Raffa and Berryman 1983; Waring and Pitman
1983; Christiansen et al. 1987; Negron 1997), and of greater flow and/or yield of defensive resins in stands with higher resource availability (Kolb et al. 1998; McDowell et al.
2007).
Despite evidence in support of the TSH, other studies have long hypothesized
tradeoffs between different internal demands for tree carbon (Loomis 1953; Rhoades
1979; Bryant et al. 1983; Coley et al. 1985; Mattson and Haack 1987; Bazzaz et al. 1987;
Berryman 1988; Herms and Mattson 1992; Tuomi 1992; Stamp 2003). In this view, preferential allocation of available carbon to growth may promote survival during drought by
increasing trees’ ability to acquire and store more resources, whereas increased allocation
to defense may favor survival if herbivory is high. The ‘growth-differentiation balance
hypothesis’ (GDBH) postulates that this evolutionary tradeoff plays out in the context of
carbon sink-driven relationships between growth and defense that vary across resource
gradients (Loomis 1932; Lorio Jr. 1986; Herms and Mattson 1992; Matyssek et al. 2005).
At very low resource availability (e.g., during severe drought stress), the GDBH predicts
decreased growth and defense due to constraints on both growth and carbon uptake. During moderate drought stress, growth is limited more strongly than photosynthesis, freeing
up carbohydrates for defense. With abundant water, growth is favored over secondary
metabolism, reducing the relative allocation to defense. Thus the TSH and the GDBH are
not always mutually exclusive, but represent different views on the importance of overall
tree carbon economy (TSH) versus carbon allocation (GDBH) for defense capacity and
survival.
117
In Pinus sp., resin ducts are formed as part of the constitutive (pre-formed) defense system (Bannan 1936; Wu and Hu 1997). Oleoresin, a carbon-based defense which
is critical to conifer defense against bark beetles, is produced and stored in resin ducts
(Münch 1919; Bannan 1936; Christiansen et al. 1987). Quantifying the resin duct system
yields an important window into pre-formed defense, particularly since other variables
such as resin flow rate, yield, or pressure measurements are highly variable (Gaylord et
al. 2013), are strongly correlated with sampling temperature (Gaylord et al. 2007), and
are difficult to measure repeatedly on the same tree (Perrakis and Agee 2006) or across
large spatial scales (Kane and Kolb 2010). Resin ducts and growth can be easily measured in tree rings, allowing for comparisons between tree allocation to growth and resin
defenses on annual to centennial time scales (Münch 1919; Mergen and Echols 1955;
Reid and Watson 1966; DeAngelis et al. 1986; Blanche et al. 1992; Kane and Kolb
2010).
Piñon-juniper woodlands of the southwestern United States (SW) have recently
been subject to widespread tree mortality associated with drought and high temperature
(Breshears et al. 2009; Williams et al. 2013). Piñon mortality in this most recent drought
has been associated with Ips confusus LeConte (Raffa et al. 2008; Gaylord et al. 2013), a
temperature-sensitive bark beetle which attacks weakened and recently dead trees (Wood
1982). In SW woodlands, a previous episode of tree mortality during the 1950s drought
has also been documented (Allen and Breshears 1998; Swetnam and Betancourt 1998).
This drought was substantially cooler than the 2000s drought but still ranks among the
most severe of the past 500 years (Williams et al. 2013).
118
Here we compare tree-ring growth and resin duct (i.e., defense) records from Pinus edulis trees that survived and died during the cooler 1950s drought and the warmer
2000s drought across sites in central New Mexico, USA. We seek to clarify the relationships between radial stem-growth, investment into defense anatomy and mortality risk.
Our study is not the first to investigate the importance of resin ducts for tree survivorship
during insect outbreaks (Kane and Kolb 2010; Ferrenberg et al. 2014), but our study is
unique in that we: (i) consider the importance of resin duct defenses in a species for
which drought impacts are emphasized as the major mortality agent (i.e., tree defenses
have not been considered explicitly yet); (ii) address how growth and defense traits and
their association with mortality vary across landscapes and drought events. We address
four specific questions:
(1) Are xylem resin duct and growth attributes significantly different in those P.
edulis trees that survived vs. died during drought?
(2) Is the prediction of drought-related mortality improved by considering resin
duct attributes in addition to radial stem growth alone?
(3) How consistent is the relationship between resin duct defenses, growth and
tree survivorship across sites and drought-mortality events?
(4) Is there evidence for internal tradeoffs between allocation to growth and defense in the tree-ring record of growth and resin ducts?
119
Methods
Study Sites
We included data from four sites in central New Mexico: Tres Piedras (TRP,
36.34 ˚N 105.93 ˚W, elevation 2100 m), White Rock (WRK, 35.81 ˚N 106.24 ˚W,
1990 m), Bandelier National Monument (BNM, 35.76 ˚N 106.27 ˚W, 1940 m), and Sevilleta National Wildlife Refuge (SEV, 34.34 ˚N 106.55 ˚W, 2050 m). The sites span gradients in climate, stand composition and density, soil substrate, as well as mortality severity
during the 2000s and 1950s droughts. Detailed site descriptions can be found in
(Macalady and Bugmann 2014), and are summarized briefly below. Mixtures of Pinus
edulis and Juniperus spp. are characteristic of all sites, with juniper dominating basal area
at SEV, piñon and juniper components being more equal at BNM and WRK, and piñon
dominating at TRP. Site-level mortality (>1 cm DBH) for P. edulis during the 2000s
drought was highest at BNM (>95%), and 82%, 64% and 20% mortality measured in
plots at WRK, TRP and SEV, respectively. At two sites, BNM and SEV, previous collections of dead remnant wood from trees that died during the 1950s drought, along with
cores from living neighbors, allowed for assessment of mortality processes during the
1950s drought. Piñon mortality during the 1950s drought event as measured at our plots
amounted to 47% and 65% at SEV and BNM, respectively (Macalady and Bugmann
2014). We refer to sites and time periods below by the site acronym followed by 2000 or
1950 (i.e., BNM1950 refers BNM in 1950).
120
Sampling of live-dead tree pairs
We sampled pairs of living and recently dead mature, dominant or co-dominant
trees that were similar in diameter, stature and micro topographical position. One or two
cores were extracted from each live tree and a cross-section was taken from each dead
tree. Samples from twenty live-dead pairs were selected at TRP and SEV each, from ten
pairs at WRK, and 3 pairs at BNM, where exceedingly high piñon mortality limited the
availability of live adult trees. At BNM, we selected an additional 20 dead trees beyond
the dead trees in the 3 pairs already selected to bolster our sample size (n = 23 dead, n = 3
live). Pairs of live and dead trees from archived collections of long-dead, remnant material were selected for the 1950s mortality at SEV and BNM. We sought to create pairs of
trees that were similarly sized prior to the 1950s drought based on variables measured in
the field by (Swetnam and Betancourt 1998). We assembled 20 pairs of trees at both
BNM50 and SEV50, for a total of 206 trees (n=113 dead, n=93 live) across all sites. This
sample set represents a subset of 263 trees sampled for a related study in which we evaluated predisposal to drought based on long-term growth patterns measured in ring widths
(Macalady and Bugmann 2014). All dead trees had outer ring dates that overlap or slightly pre-date the droughts of the 2000s and 1950s, and 97% of the dead trees had at least
one form of evidence of insect attack (blue stain fungus or Ips galleries), suggesting the
widespread presence of bark beetles during both mortality events (Macalady and
Bugmann 2014). Table 1 contains sampled tree and radii numbers for each site, along
with average tree diameters.
121
Dendroecological methods
Standard tree-ring data – Samples from each tree were prepared and cross-dated
using standard dendrochronology techniques (Stokes and Smiley 1968). All cores and
two radii from each cross-section were dated and measured. Basal area increments (BAI)
and standardized ring width indices (RWI) were calculated from raw ring widths (RW)
and distance-to-pith was estimated (Applequist 1958) where cores did not intersect the
center of the tree. BAI was calculated using the inside-out method (Biondi and Qeadan
2008). Growth measurements from the two radii per tree were averaged to create treelevel time series, which were used in all analyses. We show basic results using both RW
and BAI, but the bulk of analyses used raw ring widths to preserve gross differences in
growth between live and dead trees and because RW is less sensitive to differences in
tree size than BAI (Macalady and Bugmann 2014).
Resin duct variables – Expanding on Kane and Kolb (2010), a highly polished
surface (progressively sanded, ending with a 9 µm finishing paper) in one or, where
available, two 5 mm-wide windows from each tree was scanned to 2,400 dpi with a distortion-free scanner (Color Scanner Epson Expression 10000XL). Images were imported
into ImageJ (Rasband 1997). Within each window, resin ducts (consisting of duct lumen
and epithelial cells surrounding the lumen) were counted and traced using a digital stylus.
Xylem area within each ring was also measured. The number of ducts per ring (NUM) as
well as the area of each duct and total xylem area were recorded for 20 years prior to the
outside ring date of each dead tree, with the first year of duct measurements in live trees
matched to the earliest measurement in the corresponding dead tree in each pair. Mean
122
duct size (SIZE) was calculated from the areas of individual ducts within each ring. Annual duct density (DEN) was calculated by dividing resin duct counts by the measured
xylem area of each ring. Relative duct area (RelArea) was calculated by dividing the total
measured area of resin duct tissue by the measured xylem area. In years with no xylem
growth within the measurement window, zero was assigned to NUM. However SIZE,
DEN and RelArea were coded as missing (‘NA’ in the software R; see below), because
no measurements could be made. In years with measurable radial growth, but no resin
ducts, zero was assigned to all duct variables except SIZE, which could not be calculated
and was therefore coded as missing (‘NA’). As with growth, we averaged measurements
from individual radii to create tree-level resin duct time series. Raw measurements and
mean-value chronologies from the raw time series were used in the subsequent analyses.
Statistical analyses of growth and duct variables and their relationship
We assessed differences in the resin duct attributes and growth of live vs. dead
trees across the four sites and the two time periods using analysis of variance (ANOVA).
Post-hoc pairwise t-tests, corrected for multiple comparisons using the Holm method,
were made to assess differences in growth and resin duct attributes between specific sites.
Simple Pearson’s correlations and general additive mixed models (GAMMs) were used
to assess relationships between radial growth and duct variables. Box plots of correlations
within individual trees demonstrate the range of observed growth-defense relationships.
GAMM models were fit to assess curvilinear relationships between resin ducts and
growth. GAMM models accommodate the non-independence of growth and resin duct
123
measurements within trees and sites, as well as the residual variance heterogeneity and
autocorrelation of tree-ring time series (Wood 2006; Zuur et al. 2009). Duct attributes
were modeled as dependent variables and ring width as the independent variable. TreeID
and Site were modeled as nested random effects to account for the non-independence of
measurements within these groups. Variance and correlation structures were added to the
models to account for the remaining autocorrelation and residual variance homogeneity
(Wood 2006; Zuur et al. 2009).
Models of mortality risk
We created logistic regression models to determine which aspects of growth
and/or resin duct defenses were most associated with the risk of mortality. Averages of
growth and resin duct variables were generated over different time periods to use as predictors in the models. Prior to the calculation of all variables, live tree time-series were
truncated at the last year of the corresponding dead tree of the pair. Duct NUM, SIZE,
DEN and RelArea were averaged over 3, 5, 10, 15 and 20 years. Average growth (RW)
and growth variability were nearly equally important mortality predictors in previous
studies investigating growth-mortality relationships in southwestern conifers (Ogle et al.
2000; Macalady and Bugmann 2014; Kane and Kolb 2014), and thus these two aspects of
growth were included as potential predictors. Average growth was calculated as the mean
of annual growth measurements over k = 3, 5, 10, 15 and 20 years. Mean growth sensitivity (Strackee and Jansma 1992), a metric of year-to-year growth variability, was calculat-
124
ed from growth time series according to Eq. 2 in (Biondi and Qeadan 2008), where
t = 1,2, …, and k = year in the interval of the tree-ring series:
𝑀𝑆 = !
!!!
!
!!!
!"! !!"!!!
! !"
!
!!!
(Eq. 1)
Mean sensitivity was calculated over k = 5, 10, 15 and 20 years.
Linear mixed effects logistic regression was used to relate growth and duct indices to mortality (Live or Dead) (Agresti 2013). Study site was entered as a random intercept variable in all models, in order to account for potential variability in model parameters over space and time. Models were compared using an information theoretic approach. Akaike’s Information Criteria (AIC) scores were used to select the most parsimonious single-variable models in categories reflecting average growth rate, growth variability, and each resin duct attribute (Burnham and Anderson 2002). Predictor variables
from the two single-variable models with the lowest AIC score in each category and/or
with AIC scores less than 2 from the model with the lowest score were used to construct
multiple-variable models with up to four variables. We excluded from consideration
models containing two variables from the same category due to concerns about collinearity. We present the two multiple variable models with the lowest AIC scores in each of
three categories: growth-only models, resin duct-only models, and models containing
both resin duct and growth variables. Internal and external correct classification rates are presented as indicators of
model performance (Hosmer et al. 2013). Internal validation was done using a bootstrapped internal validation routine on all data used in model construction. Models were
fit with a random sub-sample of data containing 60% of trees from the dataset (1000 iter-
125
ations), and then validated on the remaining 40% of the data (Bigler and Bugmann 2003;
Macalady and Bugmann 2014). Models were also applied to the data from the BNM2000
site that were not used in model building (because of the lack of surviving trees) (Hosmer
et al. 2013).
The Area Under the Receiver Operating Characteristic curve (ROC) is reported as
a threshold-independent measure of model discrimination, with values >0.7 suggesting
adequate discrimination and values >0.8 suggesting excellent discrimination between live
and dead trees (Hosmer and Lemeshow 2000). An AIC-based Evidence Weight is shown
to quantify the degree to which resin duct and combined resin duct/growth models were
better or worse than the best-ranked growth-only model (Burnham and Anderson 2002;
Das et al. 2008). For example, a model with an Evidence Weight of 10 indicates that it is
ten times more likely than the model to which it is compared, with scores above 7.4 considered as strong evidence of improvement.
We used R v3.0.1 for all statistical calculations (R Core Team 2014). The dplR library v1.5.6 was used for dendrochronological statistics (Bunn 2008), the nlme library
v3.1-118 was used for modeling of growth and resin duct relationships (Pinheiro et al.
2014), the mgmv package v1.7-29 for GAMM (Wood 2006), and the lme4 library
v0.99999911-6 for logistic mixed models of mortality risk (Bates et al. 2013).
126
Results
Resin duct and growth characteristics across space and time
Vertical resin ducts appeared throughout piñon annual growth rings, typically
concentrated in the later portion of the earlywood and in the latewood (though this was
not quantified). We did not observe tangential lines of ducts within individual rings – socalled ‘traumatic’ ducts, which can indicate an induced response to injury or insect attack
(Arbellay et al. 2014). For all trees at all sites, the number of vertical resin ducts per annual ring (NUM) averaged 3.34 per year (range 0–20). Mean duct size (SIZE) averaged
0.025 mm2 (range 0.004–0.074 mm2). DEN, i.e., the number of ducts per unit xylem area,
and relative area (RelArea), i.e., the percentage of annual ring comprised of duct tissue,
averaged 1.73 (ducts/mm2) and 4.00%, respectively. Duct SIZE was the most consistent
(i.e., least variable) among all measured attributes, with a coefficient of variation (CV) of
33.2%. In contrast, the CV was 85.5% for RW, 82.4% for NUM, 75.7% for DEN and
65.6% for RelArea, respectively.
Recent average radial growth (RW, BAI) as well as NUM, SIZE and RelArea
were higher on average in surviving (live) compared to dead individuals (Figure 1; Tables 1-2). Growth in the three years prior to mortality was 24-52% higher in surviving
trees, with the significance of the difference dependent on the growth metric. Duct SIZE
was most different between live and dead trees, with 58% larger ducts on average in surviving trees. Only DEN was higher on average in dead trees, but the difference was small
and insignificant (p = 0.9, Student’s t-test).
127
NUM, SIZE, and RelArea – as well as RW – varied significantly between study
sites (Table 2). Post-hoc pairwise t-tests on the ANOVA showed that, for RW and NUM,
site differences are most pronounced between BNM2000 and all other sites (Table S1).
For SIZE and RelArea, values from the 1950s were often significantly lower compared to
those from the 2000s, but differences across sites during the same mortality event usually
were negligible (Table S1). Overall differences across sites were not significant for DEN.
Inserting a Site:Tree Status interaction term in the ANOVA was not significant for duct
variables SIZE, DEN or RelArea, indicating that differences in duct attributes between
live and dead trees were fairly consistent, even if average values were not the same across
sites (Table 2). In contrast, a significant Site:Tree Status interaction for RW and NUM
(which is the duct variable that correlates most strongly with RW – cf. below), indicates
significant variability in RW and NUM differences between live and dead trees across
sites.
Relationships between growth and duct attributes
Mean-value chronologies of duct variables and growth co-vary to a degree on an
inter-annual basis, suggesting that high-frequency variability common to each of these
metrics is due to an external forcing such as climate (Figure 2). Drought is strongly associated with inter-annual variability in piñon growth (Williams et al. 2013), and the correspondence between low growth and low resin duct defenses in regional drought years indicates that both growth and defense are compromised in severe drought years (Figure 2).
The relationship between growth and defense during wetter years is less pronounced
(Figure 2).
128
Differences in resin duct attributes between live and dead trees were either more
constant through time than differences in RW – as for duct SIZE (Figure 2 c,h) – and/or
they began to diverge in the years before each mortality period, as for RelArea and SIZE
(Figure 2 b,e,f,j). Live and dead SIZE and RelArea chronologies diverge most strongly
after single-year droughts (1938, 1946; 1996), i.e., in the years leading up to the severe
droughts of the 2000s and 1950s (Figure 2).
Within individual trees, Pearson’s correlations revealed that the relationship between annual growth and duct measurements is highly variable (Figure 3a-d). On average, NUM and SIZE were positively correlated with RW within trees, with a much wider
spread of correlation values for SIZE. In contrast, the median correlation coefficient is
close to zero for RelArea, and slightly negative for DEN. For DEN and RelArea, withintree correlations depended on overall growth rate, with more positive correlations among
trees with lower than median (10-year) growth rates, and weaker or negative correlations
among trees at higher growth rates (DEN: p = 0.00017; RelArea: p = 0.00035). For SIZE,
this pattern was also evident, but not significant (p = 0.298).
Statistical (GAMM) models fitted with knotted cubic splines allowed us to better
estimate the shape of relationships between growth and duct variables within and across
trees (Figure 3b-e). These models had significantly more support than linear models with
the same form (p < 0.0001 for NUM, SIZE, DEN and RelArea), indicating that the relationship between annual growth and resin duct attributes is best described as non-linear.
RW was a significant predictor of all duct variables in the models, with adjusted R2 values for full models ranging from 0.08 for RelArea, 0.26 for DEN, 0.34 for SIZE, to 0.60
129
for NUM. Models with splines fit separately by site were not significantly better than
models fit with only one smoother (p < 0.0001 for all variables), indicating basic conformity in the shape of relationships across the different tree populations.
The basic shape of relationships between resin ducts and growth were also similar
between live and dead trees, however absolute differences between live and dead trees
were most pronounced at medium levels of growth. For example, differences in 3-year
SIZE were most pronounced and significant at low-medium levels of 3-year average
growth (Figure 5).
Models of mortality risk
Both resin duct and growth attributes were significant predictors of mortality risk
(Table 3). However, models with both growth and resin duct variables had substantially
more support than the most parsimonious growth-only or resin duct-only models (Table
3, Figure 4a). The most parsimonious model overall contained longer-term average ring
width and mean sensitivity along with recent (3-yr) duct SIZE, and had ≈1019 times more
support than the best-performing growth-only model, and still ≈150 times more support
than the best resin duct-only model (Table 3, Figure 4a). Models that combined resin duct
and growth variables had ROC >0.94 indicating excellent model discrimination, and correctly classified over 82% of trees in the bootstrapped internal validations and 70% of the
unpaired dead trees from BNM2000 (these were not included in model building).
Growth-only models contained average recent RW and longer-term mean sensitivity, had adequate discrimination (ROC ≥0.75), and correctly classified around 70% of
130
trees in the internal validation. However, only 25% of dead trees from BNM2000 (Table
3) were correctly classified, consistent with a previous study on a larger dataset
(Macalady and Bugmann 2014). The most parsimonious resin duct-only models contained recent RelArea and SIZE, and represented a substantial improvement over growthonly models (Figure 4a). Interestingly, the improvement in discrimination gained by including resin duct along with growth variables in models was substantially larger for the
2000s versus 1950s datasets, and it was especially pronounced at the sites in northern
New Mexico (Figure 4b).
Discussion
Defense anatomy is strongly associated with drought-mortality risk in P. edulis
Insect attack and associated tree defenses have been suggested as being relevant
for the drought-related mortality of P. edulis (Breshears et al. 2005; Shaw et al. 2005;
McDowell et al. 2008; Anderegg et al. 2015; Meddens et al. 2015). However, research to
date has focused on the physiological response to drought and heat stress alone (Adams et
al. 2009; Breshears et al. 2009; Plaut et al. 2012; Sevanto et al. 2014). We showed that
pre-formed defense anatomy is strongly associated with P. edulis survivorship over a
wide range of site and stand conditions, and that models of mortality risk that account for
resin duct attributes have orders of magnitude more support than models that contain only
tree growth. Thus our results suggest that tree defenses and associated insect attacks are
critical for understanding the mechanisms of piñon mortality during drought (e.g., Gay-
131
lord et al. 2013).
Resin ducts have also been associated with tree survivorship during drought and
outbreaks of Ips spp. in a ponderosa pine (P. ponderosa) stand in Arizona (Kane and
Kolb 2010), and outbreaks of Mountain pine beetle (Dendroctonus ponderosae) among
lodgepole (P. contorta) and limber pine (P. flexilis) in a stand in Colorado (Ferrenberg et
al. 2014). Additional studies have linked xylem, phloem and/or bark resin duct attributes
to vulnerability to attack by other insects, though not explicitly to mortality (e.g., Baier
1996; Alfaro et al. 1997; Moreira et al. 2012). Our study builds on these results by highlighting the importance across space and time of defense anatomy for the droughtmortality risk of a widespread aridlands conifer that is attacked by bark beetle species
that are generally considered less aggressive (Raffa et al. 2008). The range of species,
landscapes and droughts where preformed resin defenses have now been associated with
die-offs provides strong evidence that accounting for defense is likely to be important for
improving predictive models of drought-mortality risk (cf. McDowell et al 2013), certainly among the genus Pinus.
Spatial and temporal variability in the importance of growth and resin duct attributes
The degree to which simple growth-mortality models were improved by including
resin duct attributes varied depending on site and period. The most striking pattern is the
much larger improvement of mortality models for the 2000s compared to the 1950s. Although we do not provide direct evidence, the pattern suggests that defense or defense allocation may have been particularly important during the recent drought. We suggest two
132
(non-exclusive) explanations: 1) relatively higher temperatures during the 2000s versus
the 1950s drought, especially in northern New Mexico (Breshears et al. 2005; Weiss et al.
2009; Williams et al. 2013), may have amplified insect pressure, thus emphasizing the
importance of preformed defense for tree survivorship. This is also consistent with the
especially large improvement for sites in northern New Mexico, where higher mortality
rates and more synchronous mortality dates are consistent with the greater importance of
exogenous mortality factors (Macalady and Bugmann 2014); 2) higher average growth
rates in the years preceding the 2000s drought may have accentuated the importance of
growth-defense allocation patterns and trade-offs in the preformed defense system. The
period between 1976 and 1995 was among the wettest of the past thousand years, and
was associated with an anomalous growth surge in trees across the region (Swetnam and
Betancourt 1998). We found that growth and resin ducts are not as positively correlated
when growth rates are relatively high, and that resin duct and growth chronologies are
more similar during drought years versus wet years. Accounting for resin ducts may thus
have been more important for accurately reflecting trees’ constitutive defenses leading up
to the 2000s drought, when resources were more abundant and trees’ internal allocation
patterns may have become more limiting of defense levels than resource availability per
se. Regardless of the cause, variability in the importance of defense versus growth over a
spectrum of insect pressures and growth rates deserves direct investigation.
133
The importance of resin duct size for mortality risk
Recent duct SIZE was the best single predictor of mortality in statistical models
of mortality risk. Duct radius was a significant predictor of mortality during the 2000s
drought in a stand of P. ponderosa in Arizona (Kane and Kolb 2010), and among beetleattacked P. contorta (Ferrenberg et al. 2014), but it was not related to mortality in P. flexilis (Ferrenberg et al. 2014). In both studies, a stronger relationship between duct density
and mortality was found, whereas our sites and time periods featured a lower correspondence between duct density and death. A number of factors may explain why duct SIZE
and, by extension, RelArea, relate more strongly to mortality than duct NUM or DEN in
our study. First, as pointed out by Kane and Kolb (2010), duct size may be more closely
related to resin flow rate and volume than other aspects of the resin duct system because
of the exponential relationship between potential flow and radius (Schopmeyer et al.
1954; Baier 1996; Lombardero et al. 2000; Baier et al. 2002). Second, epithelial cells in
preformed resin ducts may become activated to produce de novo resins in an induced defense reaction (Walter et al. 1989; Hudgins and Franceschi 2004). Third, SIZE differences between live and dead trees did not vary significantly across sites, indicating more
stable mortality thresholds when compared with other variables. Finally, P. edulis appears to have a substantially higher density of resin ducts than the other conifers. In our
study, DEN averaged ≈1.8 per mm2 vs. between 0.4 and ~1.0 for P. contorta (Reid and
Watson 1966), 0.4 and 0.8 for P. sylvestris (Rigling et al. 2003) and 0.8 for P. ponderosa
(Kane and Kolb 2010). We thus suggest that constitutive defense anatomy in this relatively slow-growing, drought-prone species is quite important, and that the high resin-duct
134
density in P. edulis implies that small changes in SIZE, summed over all ducts, result in
larger differences in potential resin production, storage and exudation pressure than the
addition of a few additional resin ducts.
The more stable differences in SIZE between live and dead trees may also indicate a genetic basis for mortality susceptibility. Aspects of both growth and defense anatomy have been shown to be heritable in spruce and pine species (e.g., Bridgen and
Hanover 1982; Roberds et al. 2003; Romanelli and Sebbenn 2004; Rosner and Hannrup
2004; Westbrook et al. 2013). The stronger divergence of duct SIZE (and RelArea) records after 1938/1946 and 1996, respectively, could then reflect a gene x environment interaction, with up-regulation of resin defense in certain trees as a reaction to non-lethal
drought (Westbrook et al. 2013). Alternatively, this divergence may represent systemic
induced resistance to low-level insect attacks during previous single-year droughts. Systemic induced resistance is hypothesized to have a prolonged, plant-wide effect on chemical and anatomical defenses (Bonello et al. 2006). For example, prolonged increases in
xylem resin duct size and resin production have been shown in reaction to resin tapping
in P. pinaster (Rodríguez-García et al. 2014). In either case, the recent divergence likely
explains the entry of recent versus longer-term resin duct size into the best regression
models.
Relationships between growth, defense and mortality - are tradeoffs evident?
Current plant defense theories, including the GDBH, posit non-linear relationships and trade-offs between growth and defense that may influence plant fitness (Herms
and Mattson 1992). Many studies of resin ducts in conifers have noted positive correla-
135
tions between duct attributes and growth and thus concluded that no tradeoff between
growth and defense is evident (Mergen and Echols 1955; Fahn and Zamski 1970; Kane
and Kolb 2010; Ferrenberg et al. 2014). Still, other studies have documented weak or
negative relationships between growth and some resin duct variables (Reid and Watson
1966; Stephan 1967; Wimmer and Halbwachs 1992; Wimmer and Grabner 1997; Rigling
et al. 2003).
We found that the means of all correlations between ring width and resin duct variables are either positive or close to zero, and that surviving trees had on average both
higher growth rates and more numerous/larger resin ducts. In the context of mortality
prediction, regression models that contain only growth variables and only resin duct variables as predictors have reasonable discriminatory ability (~70% and ~80% of trees correctly classified, respectively), and the coefficients in these models reflect the overall
positive relationship between growth, defense and survival. This is consistent with the
tree stress hypothesis (TSH, cf. Waring and Pitman 1983) and highlights the importance
of the overall tree carbon budget in determining basic levels of allocation to both growth
and defense in this system.
However, our findings are not inconsistent with the GDBH, and we suggest the
GDBH may better explain nuances in our data. First, we note generally more positive
correlations between growth and resin duct attributes in trees with lower average growth
rates, consistent with the prediction that growth and defense should be more positively
correlated at low levels. Second, the shape of fits from regressions (GAMMs) relating
growth and resin duct attributes are curvilinear, with positive correlations between
136
growth and ducts weakening or tapering to negative relationships as growth rates increase. This is consistent with the prediction of the GDBH that allocation to growth is
favored over defense at higher levels of resource availability. The tendency for correlations to be positive overall may reflect the both the interconnection between xylem production and resin duct formation (e.g., growth and defense are not entirely distinct in our
case), and the prevalence of years in which strong resource limitation is experienced
among trees in our drought-prone study area.
Defense attributes became most variable and most different between living and
dead trees at medium levels of growth (e.g., Figure 5), which is also consistent with the
evolutionary tradeoff suggested by the GDBH. At very low resource levels, trees may be
more vulnerable to mortality during drought or bark beetle attack due to limits on the
overall availability of carbon. At high levels of resources, growth may be favored over
defense because growth may improve acquisition of resources, especially when competition is strong. Under medium resource levels, many allocation strategies may exist, and
what strategy is favored may depend on levels of herbivory. In our dataset, the importance of allocation to defense versus growth is reflected most strongly in the many
relatively faster-growing trees that died during the 2000s drought/insect outbreak because
the rate of allocation to resin defenses was much lower than those that survived.
It is also interesting to note that recent average growth remained as a predictor in
the best overall models of mortality risk, but that the coefficient switched from positive to
negative when both growth and resin duct size variables were considered. This may reflect the importance of the rate of allocation to growth versus defense for survivorship,
137
but it may also reflect bark beetle preference for more vigorous trees, all else being equal
(Baier 1996; Boone et al. 2011). In other words, proportionally larger resin ducts required
for survival in trees that were fast-growing may indicate trees with higher growth rates
prior to prolonged drought experienced relatively more beetle attacks during drought and
thus required greater defenses for survival.
Conclusion and Implications
Both growth and resin duct attributes were on average greater in P. edulis that
survived versus died during droughts, but differences in resin duct attributes are stronger
and less variable across sites and drought events than differences in growth. Furthermore,
models of mortality risk that account for tree defense anatomy had more support than
models that considered growth alone, highlighting the importance of tree defenses to survivorship during drought and associated insect attack in this species.
The fact that both recent growth and resin duct defenses are higher on average in
surviving trees is consistent with the tree stress hypothesis (e.g., White 1960, Waring and
Pitman 1983), which emphasizes the importance of overall tree carbon balance versus
allocation strategies for determining susceptibility to drought and herbivory. However,
the GDBH, which is not inconsistent the TSH at high levels of environmental stress, may
better explain variability in growth-defense relationships we observed across years and
sites. Indeed, we show that relationships between radial growth and resin duct attributes
are non-linear across years that represent a broad spectrum of environmental conditions,
with the shape of the relationships largely consistent with the predictions of the GDBH.
138
Furthermore, we show that despite being relatively fast growing, many trees that died
during the 2000s drought became vulnerable to mortality because they maintained lower
rates of allocation to defense.
Our results have implications both for projecting the fate of forests under climate
change, and for forest management in a warming world. The fact that resin ducts are
more important for mortality prediction than growth suggests that tree defenses should be
accounted for in models of forest dynamics. The spatial and temporal variability in
growth-defense relationships further suggests that accounting for how trees allocate
available carbohydrates will be important for improving our understanding of tree mortality during drought. The strong and potentially increasing role of tree defenses in droughtmortality risk of P. edulis suggests that a useful line of future tree mortality research
could focus on defense-associated life history traits, trade-offs with growth and/or other
plant functions, and the sensitivity of these variables to environmental stress on a variety
of timescales, along with direct assessments of the how accounting for different traits improves mortality prediction in a range of settings.
If resin duct traits such as mean duct size are heritable (Rosner and Hannrup
2004), our results further suggest defense anatomy as one basis for the selection of genotypes with resistance to mortality during bark beetle activity coupled with drought. The
genetics of tree resin defenses, and how this links to mortality risk, warrants further research, especially among conifers such as piñon pine that are highly vulnerable to die-off.
139
Acknowledgements
NE was supported by Los Alamos National Laboratory (LANL)-LDRD; NE was supported by DOE-BER; CDA acknowledges the United States Geological Survey (USGS)
Ecosystems and Climate and Land Use Change mission areas. AKM and TWS were supported by LANL-IGPP, and AKM was additionally supported by a US DOE Graduate
Environmental Research Fellowship, and a Marshall Dissertation Fellowship from the
University of Arizona. Any use of trade names is for descriptive purposes only and does
not imply endorsement by the U.S. Government.
References
Adams HD, Guardiola-Claramonte M, Barron-Gafford GA, et al (2009) Temperature
sensitivity of drought-induced tree mortality portends increased regional die-off
under global-change-type drought. Proc Natl Acad Sci 106:7063–7066.
Agresti A (2013) Categorical data analysis, 3rd edn. John Wiley & Sons, Hoboken, NJ.
Alfaro RI, He F, Tomlin E, Kiss G (1997) White spruce resistance to white pine weevil
related to bark resin canal density. Can J Bot 75:568–573.
Allen CD, Breshears DD (1998) Drought-induced shift of a forest–woodland ecotone:
rapid landscape response to climate variation. Proc Natl Acad Sci 95:14839–
14842.
Allen CD, Macalady AK, Chenchouni H, et al (2010) A global overview of drought and
heat-induced tree mortality reveals emerging climate change risks for forests. For
Ecol Manag 259:660–684.
Anderegg WR, Berry JA, Smith DD, et al (2012) The roles of hydraulic and carbon stress
in a widespread climate-induced forest die-off. Proc Natl Acad Sci 109:233–237.
Anderegg WRL, Hicke JA, Fisher RA, et al (2015) Tree mortality from drought, insects,
and their interactions in a changing climate. New Phytol.
140
Applequist M (1958) A simple pith locator for use with off-center increment cores. J For
56:141.
Arbellay E, Stoffel M, Sutherland EK, et al (2014) Resin duct size and density as ecophysiological traits in fire scars of Pseudotsuga menziesii and Larix occidentalis.
Ann Bot 114:973–980.
Baier P (1996) Defence reactions of Norway spruce (Picea abies Karst.) to controlled
attacks of Ips typographus (L.)(Col., Scolytidae) in relation to tree parameters. J
Appl Entomol 120:587–593.
Baier P, Führer E, Kirisits T, Rosner S (2002) Defence reactions of Norway spruce
against bark beetles and the associated fungus Ceratocystis polonica in secondary
pure and mixed species stands. For Ecol Manag 159:73–86.
Bannan MW (1936) Vertical Resin Ducts in the Secondary Wood of the Abietineae. New
Phytol 35:11–46.
Bates DM, Maechler M, Bolker B (2013) lme4: Linear-mixed Effects Models Using S4
Classes.
Bazzaz FA, Chiariello NR, Coley PD, Pitelka LF (1987) Allocating Resources to Reproduction and Defense. BioScience 37:58–67.
Bentz B, Logan J, MacMahon J, et al (2009) Bark beetle outbreaks in western North
America: causes and consequences. University of Utah Press, Salt Lake City, UT,
p 42.
Berryman AA (1988) Towards a Unified Theory of Plant Defense. In: Mattson WJ, Levieux J, Bernard-Dagan C (eds) Mechanisms of Woody Plant Defenses Against Insects. Springer New York, pp 39–55.
Bigler C, Bräker OU, Bugmann H, et al (2006) Drought as an inciting mortality factor in
Scots pine stands of the Valais, Switzerland. Ecosystems 9:330–343.
Bigler C, Bugmann H (2003) Growth-dependent tree mortality models based on tree
rings. Can J For Res 33:210–221.
Biondi F, Qeadan F (2008) Inequality in paleorecords. Ecology 89:1056–1067.
Blanche CA, Lorio Jr. PL, Sommers RA, et al (1992) Seasonal cambial growth and development of loblolly pine: Xylem formation, inner bark chemistry, resin ducts,
and resin flow. For Ecol Manag 49:151–165.
141
Bonello P, Gordon TR, Herms DA, et al (2006) Nature and ecological implications of
pathogen-induced systemic resistance in conifers: A novel hypothesis. Physiol
Mol Plant Pathol 68:95–104.
Boone CK, Aukema BH, Bohlmann J, et al (2011) Efficacy of tree defense physiology
varies with bark beetle population density: a basis for positive feedback in eruptive species. Can J For Res 41:1174–1188.
Breshears DD, Cobb NS, Rich PM, et al (2005) Regional vegetation die-off in response
to global-change-type drought. Proc Natl Acad Sci 102:15144–15148.
Breshears DD, Myers OB, Meyer CW, et al (2009) Tree die-off in response to global
change-type drought: mortality insights from a decade of plant water potential
measurements. Front Ecol Environ 7:185–189.
Bridgen MR, Hanover JW (1982) Genetic variation in oleoresin physiology of Scotch
pine. For Sci 28:582–589.
Bryant JP, Chapin FS, Klein DR (1983) Carbon/Nutrient Balance of Boreal Plants in Relation to Vertebrate Herbivory. Oikos 40:357–368.
Bunn A (2008) A dendrochronology program library in R (dplR). Dendrochronologia
26:115–124.
Burnham KP, Anderson D (2002) Model Selection and Multi-Model Inference: A Practical Information-Theoretic Approach, 2nd edn. Springer-Verlag New York, New
York, NY.
Christiansen E, Waring RH, Berryman AA (1987) Resistance of conifers to bark beetle
attack: Searching for general relationships. For Ecol Manag 22:89–106.
Coley PD, Bryant JP, Chapin FS (1985) Resource Availability and Plant Antiherbivore
Defense. Science 230:895–899.
Das A, Battles J, van Mantgem PJ, Stephenson NL (2008) Spatial elements of mortality
risk in old-growth forests. Ecology 89:1744–1756.
DeAngelis JD, Nebeker TE, Hodges JD (1986) Influence of tree age and growth rate on
the radial resin duct system in loblolly pine (Pinus taeda). Can J Bot 64:1046–
1049.
Dobbertin M, Wermelinger B, Bigler C, et al (2007) Linking increasing drought stress to
Scots pine mortality and bark beetle infestations. TheScientificWorld J 7:231–
239.
142
Fahn A, Zamski E (1970) The influence of pressure, wind, wounding and growth substances on the rate of resin duct formation in Pinus halepensis wood. Isr J Bot
19:429–46.
Ferrenberg S, Kane JM, Mitton JB (2014) Resin duct characteristics associated with tree
resistance to bark beetles across lodgepole and limber pines. Oecologia
174:1283–1292.
Franceschi VR, Krokene P, Christiansen E, Krekling T (2005) Anatomical and Chemical
Defenses of Conifer Bark against Bark Beetles and Other Pests. New Phytol
167:353–375.
Gaylord ML, Kolb TE, Pockman WT, et al (2013) Drought predisposes piñon-juniper
woodlands to insect attacks and mortality. New Phytol 198:567–578.
Gaylord ML, Kolb TE, Wallin KF, Wagner MR (2007) Seasonal dynamics of tree
growth, physiology, and resin defenses in a northern Arizona ponderosa pine forest. Can J For Res 37:1173–1183.
Hartmann H, Ziegler W, Kolle O, Trumbore S (2013) Thirst beats hunger – declining hydration during drought prevents carbon starvation in Norway spruce saplings.
New Phytol 200:340–349.
Herms DA, Mattson WJ (1992) The Dilemma of Plants: To Grow or Defend. Q Rev Biol
67:283–335.
Hosmer DW, Lemeshow S, Rodney X. Sturdivant (2013) Applied logistic regression, 3rd
edn. John Wiley & Sons, Hoboken, NJ.
Hudgins JW, Franceschi VR (2004) Methyl jasmonate-induced ethylene production is
responsible for conifer phloem defense responses and reprogramming of stem
cambial zone for traumatic resin duct formation. Plant Physiol 135:2134–2149.
Kane JM, Kolb TE (2010) Importance of resin ducts in reducing ponderosa pine mortality
from bark beetle attack. Oecologia 164:601–609.
Kane JM, Kolb TE (2014) Short- and long-term growth characteristics associated with
tree mortality in southwestern mixed-conifer forests. Can J For Res 44:1227–
1235.
Keane RE, Austin M, Field C, et al (2001) Tree mortality in gap models: application to
climate change. Clim Change 51:509–540.
Kolb TE, Holmberg KM, Wagner MR, Stone JE (1998) Regulation of ponderosa pine
foliar physiology and insect resistance mechanisms by basal area treatments. Tree
Physiol 18:375–381.
143
Koricheva J, Larsson S, Haukioja E, Keinänen M (1998) Regulation of woody plant secondary metabolism by resource availability: hypothesis testing by means of metaanalysis. Oikos 212–226.
Lombardero MJ, Ayres MP, Lorio Jr PL, Ruel JJ (2000) Environmental effects on constitutive and inducible resin defences of Pinus taeda. Ecol Lett 3:329–339.
Loomis WE (1953) Growth and differentiation - an introduction and summary. In:
Loomis WE (ed) Growth and differentiation in plants. Iowa State College Press,
Iowa, pp 1–17.
Loomis WE (1932) Growth-differentiation balance vs. carbohydrate-nitrogen ratio. Proc
Am Soc Hortic Sci 29:240–245.
Lorio Jr. PL (1986) Growth-differentiation balance: A basis for understanding southern
pine beetle-tree interactions. For Ecol Manag 14:259–273.
Macalady AK, Bugmann H (2014) Growth-Mortality Relationships in Piñon Pine (Pinus
edulis) during Severe Droughts of the Past Century: Shifting Processes in Space
and Time. PloS One 9:e92770.
Martínez-Vilalta J, Piñol J, Beven K (2002) A hydraulic model to predict droughtinduced mortality in woody plants: an application to climate change in the Mediterranean. Ecol Model 155:127–147.
Mattson WJ, Haack RA (1987) The role of drought in outbreaks of plant-eating insects.
BioScience 37:110–118.
Matyssek R, Agerer R, Ernst D, et al (2005) The plant’s capacity in regulating resource
demand. Plant Biol 7:560–580.
McDowell N, Allen CD, Marshall L (2010) Growth, carbon-isotope discrimination, and
drought-associated mortality across a Pinus ponderosa elevational transect. Glob
Change Biol 16:399–415.
McDowell NG, Adams HD, Bailey JD, Kolb TE (2007) The role of stand density on
growth efficiency, leaf area index, and resin flow in southwestern ponderosa pine
forests. Can J For Res 37:343–355.
McDowell NG, Beerling DJ, Breshears DD, et al (2011) The interdependence of mechanisms underlying climate-driven vegetation mortality. Trends Ecol Evol 26:523–
532.
McDowell NG, Fisher RA, Xu C, et al (2013) Evaluating theories of drought-induced
vegetation mortality using a multimodel–experiment framework. New Phytol
200:304–321.
144
McDowell N, Pockman WT, Allen CD, et al (2008) Mechanisms of plant survival and
mortality during drought: why do some plants survive while others succumb to
drought? New Phytol 178:719–739.
Meddens AJ, Hicke JA, Macalady AK, et al (2015) Patterns and causes of observed piñon pine mortality in the southwestern United States. New Phytol 206:91–97.
Mergen F, Echols RM (1955) Number and Size of Radial Resin Ducts in Slash Pine. Science 121:306–307.
Moreira X, Alfaro RI, King JN (2012) Constitutive defenses and damage in Sitka spruce
progeny obtained from crosses between white pine weevil resistant and susceptible parents. Forestry 85:87–97.
Münch E (1919) Naturwissenschaftliche Grundlagen der Kiefernharz nutzung. Biol Reich
Land- Forstwirtsch 10:1–140.
Negron J (1997) Estimating probabilities of infestation and extent of damage by the
roundheaded pine beetle in ponderosa pine in the Sacramento Mountains, New
Mexico. Can J For Res 27:1936–1945.
Ogle K, Whitham TG, Cobb NS (2000) Tree-ring variation in pinyon predicts likelihood
of death following severe drought. Ecology 81:3237–3243.
Perrakis DD, Agee JK (2006) Seasonal fire effects on mixed-conifer forest structure and
ponderosa pine resin properties. Can J For Res 36:238–254.
Pinheiro JC, Bates DM, DebRoy S, et al (2014) nlme: Linear and Nonlinear Mixed Effects Models.
Plaut JA, Yepez EA, Hill J, et al (2012) Hydraulic limits preceding mortality in a piñon–
juniper woodland under experimental drought. Plant Cell Environ 35:1601–1617.
Raffa KF, Aukema BH, Bentz BJ, et al (2008) Cross-scale Drivers of Natural Disturbances Prone to Anthropogenic Amplification: The Dynamics of Bark Beetle Eruptions. BioScience 58:501–517.
Raffa KF, Berryman AA (1983) The role of host plant resistance in the colonization behavior and ecology of bark beetles (Coleoptera: Scolytidae). Ecol Monogr 27–49.
Rasband WS (1997) ImageJ, US National Institutes of Health, Bethesda, Maryland, USA.
R Core Team (2014) R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria
145
Reid RW, Watson JA (1966) Sizes, distributions, and numbers of vertical resin ducts in
lodgepole pine. Can J Bot 44:519–525.
Rhoades DF (1979) Evolution of plant chemical defense against herbivores. In: Rosenthal
GA, Janzen DH (eds) Herbivores: Their Interaction with Secondary Plant Metabolites. Academic Press, New York, pp 1–55.
Rigling A, Brühlhart H, Bräker OU, et al (2003) Effects of irrigation on diameter growth
and vertical resin duct production in Pinus sylvestris L. on dry sites in the central
Alps, Switzerland. For Ecol Manag 175:285–296.
Roberds JH, Strom BL, Hain FP, et al (2003) Estimates of genetic parameters for oleoresin and growth traits in juvenile loblolly pine. Can J For Res 33:2469–2476.
Rodríguez-García A, López R, Martín JA, et al (2014) Resin yield in Pinus pinaster is
related to tree dendrometry, stand density and tapping-induced systemic changes
in xylem anatomy. For Ecol Manag 313:47–54.
Romanelli RC, Sebbenn AM (2004) Parâmetros genéticos e ganhos na seleção para
produção de resina em Pinus elliottii var. elliottii, no Sul do Estado de São Paulo.
Rev Inst Flor 16:11–23.
Rosner S, Hannrup B (2004) Resin canal traits relevant for constitutive resistance of
Norway spruce against bark beetles: environmental and genetic variability. For
Ecol Manag 200:77–87.
Schopmeyer CS, Mergen F, Evans TC (1954) Applicability of Poiseuille’s law to exudation of oleoresin from wounds on slash pine. Plant Physiol 29:82.
Seidl R, Fernandes PM, Fonseca TF, et al (2011) Modelling natural disturbances in forest
ecosystems: a review. Ecol Model 222:903–924.
Sevanto S, Mcdowell NG, Dickman LT, et al (2014) How do trees die? A test of the hydraulic failure and carbon starvation hypotheses. Plant Cell Environ 37:153–161.
Shaw JD, Steed BE, DeBlander LT (2005) Forest Inventory and Analysis (FIA) annual
inventory answers the question: what is happening to pinyon-juniper woodlands?
J For 103:280–285.
Stamp N (2003) Out of the Quagmire of Plant Defense Hypotheses. Q Rev Biol 78:23–
55.
Stephan G (1967) Untersuchungen über die Anzahl der Harzkanäle in Kiefern (Pinus sylvestris). Arch Forstwes 16:461–470.
146
Stokes MA, Smiley TL (1968) An introduction to tree-ring dating. University of Arizona
Press, Tucson, AZ.
Strackee J, Jansma E (1992) The statistical properties of mean sensitivity—a reappraisal.
Dendrochronologia 10:121–135.
Swetnam TW, Betancourt JL (1998) Mesoscale disturbance and ecological response to
decadal climatic variability in the American Southwest. J Clim 11:3128–3147.
Tuomi J (1992) Toward integration of plant defence theories. Trends Ecol Evol 7:365–
367.
Walter J, Charon J, Marpeau A, Launay J (1989) Effects of wounding on the terpene content of twigs of maritime pine (Pinus pinaster Ait.). Trees 3:210–219.
Waring R (1987) Characteristics of trees predisposed to die. BioScience 37:569–574.
Waring RH, Pitman GB (1983) Physiological stress in lodgepole pine as a precursor for
mountain pine beetle attack. Z Für Angew Entomol 96:265–270.
Waring R, Pitman G (1985) Modifying lodgepole pine stands to change susceptibility to
mountain pine beetle attack. Ecology 889–897.
Weiss JL, Castro CL, Overpeck JT (2009) Distinguishing Pronounced Droughts in the
Southwestern United States: Seasonality and Effects of Warmer Temperatures. J
Clim 22:5918–5932.
Westbrook JW, Resende MFR, Munoz P, et al (2013) Association genetics of oleoresin
flow in loblolly pine: discovering genes and predicting phenotype for improved
resistance to bark beetles and bioenergy potential. New Phytol 199:89–100.
White TCR (1969) An Index to Measure Weather-Induced Stress of Trees Associated
With Outbreaks of Psyllids in Australia. Ecology 50:905–909.
Williams AP, Allen CD, Macalady AK, et al (2013) Temperature as a potent driver of
regional forest drought stress and tree mortality. Nat Clim Change 3:292–297.
Wimmer R, Grabner M (1997) Effects of climate on vertical resin duct density and radial
growth of Norway spruce [Picea abies (L.) Karst.]. Trees 11:271–276.
Wimmer R, Halbwachs G (1992) Holzbiologische Untersuchungen an fluorgeschädigten
Kiefern. Holz Als Roh- Werkst 50:261–267.
Wood S (2006) Generalized additive models: an introduction with R. CRC Press, Boca
Raton, FL.
147
Wood S (1982) The Bark and Ambrosia Beetles of North and Central America (Coleoptera: Scolytidae), a Taxonomic Monograph.
Wu H, Hu Z (1997) Comparative anatomy of resin ducts of the Pinaceae. Trees 11:135–
143.
Zhao J, Hartmann H, Trumbore S, et al (2013) High temperature causes negative wholeplant carbon balance under mild drought. New Phytol 200:330–339.
Zuur AF, Ieno EN, Walker NJ, et al (2009) Mixed effects models and extensions in ecology with R. Springer-Verlag New York, Ney York, NY.
20
39
38
15.4
0.354
1.805
2.38
2.52
0.031
0.032
1.87
1.58
5.4%
4.7%
Total no. of trees
No. of radii (Ring Width)
No. of radii (Resin Ducts)
Tree diameter (cm DBH)
Short-term average growth
3 yr RW (mm)
3 yr BAI (cm2)
Number of Ducts (#/year)
3 yr
5 yr
Mean Duct Size (mm2/yr)
3 yr
5 yr
Duct Density (#/mm2/yr)
3 yr
5 yr
Duct Relative Area (%/yr)
3 yr
5 yr
Live
2.7%
2.6%
1.44
1.26
0.020
0.022
1.52
1.99
0.256
1.373
20
40
40
15.8
Dead
TRP2000
3
6
6
9.4
7.67
8.53
5.3%
5.0%
1.60
1.41
2.7%
2.7%
1.05
1.02
0.037 0.026
0.037 0.027
10.33
11.00
1.726 1.427
3.913 3.978
3
3
3
7.8
Live
3.8%
4.2%
2.04
2.05
0.020
0.022
3.01
3.81
0.411
1.731
20
40
40
13.9
Dead
Dead (external
(pairs) validation)
BNM2000
6.6%
6.4%
2.13
2.12
0.032
0.032
3.47
3.54
0.370
2.005
10
20
20
17.3
Live
4.6%
4.7%
2.44
2.36
0.020
0.022
2.22
2.47
0.248
1.444
10
20
20
17.4
Dead
WRK2000
20
40
40
11.9
Dead
1.68
1.96
5.2%
4.7%
2.15
1.94
3.5%
3.6%
2.63
2.40
0.026 0.016
0.026 0.017
3.35
3.25
0.375 0.175
1.295 0.641
20
39
35
11.0
Live
SEV2000
20
35
40
12.5
Dead
20
20
20
7.0
Live
20
40
40
10.1
Dead
SEV1950
1.86
1.51
2.07
2.41
1.03
1.41
4.9%
4.0%
1.94
1.63
2.8%
2.7%
2.04
1.92
3.7%
3.7%
1.70
1.65
1.8%
2.3%
1.49
1.74
0.026 0.016 0.025 0.013
0.025 0.017 0.023 0.014
3.04
2.35
0.425 0.257 0.346 0.132
1.217 1.055 0.763 0.443
20
27
25
8.9
Live
BNM1950
urements of radii lengths (for live trees, this length was truncated at the date of the outer ring in the paired dead tree).
between live and dead trees (Student’s t-test, p < 0.05). Tree diameters reported here are under-bark estimates, based on meas-
Table 1. Sample sizes and average growth and resin duct measurements. Numbers in bold represent significant differences
Tables
148
2.874
1.954
1.917
12.688
Sum Sq.
18.610
0.060
5.090
349.380
Site
Tree Status
Site:Tree Status
Residuals
Sum Sq.
Site
Tree Status
Site:Tree Status
Residuals
shown).
F
2.035
0.030
0.557
DEN3
8.652
29.412
5.772
F
RW3
p-value
0.076
0.862
0.733
<0.0001
<0.0001
<0.0001
p-value
Sum Sq.
0.012
0.020
0.000
0.068
117.260
92.960
54.020
433.200
Sum Sq.
F
6.807
56.029
0.272
RelArea3
10.341
40.989
4.764
F
NUM3
p-value
<0.0001
<0.0001
0.928
<0.0001
<0.0001
<0.0001
p-value
0.001
0.006
0.000
0.006
Sum Sq.
7.305
170.551
1.044
F
SIZE3
<0.0001
<0.0001
0.393
p-value
sults were similar for longer averages, and consistent with more complex repeat measures analyses of values across all years (not
ducts, and RelArea is the annual relative area of duct tissue (%). Tests displayed here are for recent (3-year average) values. Re-
nual ring width (mm/year), NUM is annual number of ducts, SIZE is annual average duct size (mm2), DEN is annual density of
of the differences in recent average growth and resin duct parameters between study sites and tree survivorship status. RW is an-
Table 2. Differences in resin ducts and growth across sites and between live and dead trees. Two-way Analysis of Variance
149
Reference model
Top model overall
log(RW3) + Sens20
log(RW3) + Sens15
Sens20
log(rw3)
40.9
56
366
RelArea3 + SIZE3
RelArea5 + SIZE3
Size3
0.79 -4.44
0.8 -3.39
-5.84
1.23
347
353
-3.54 -7.06
-3.6 -7.01
410
405
Coefficeints
RW15 + Sens15 + SIZE3
RW20 + Sens15 + SIZE3
Fixed + (1 | site)
which were not used in model building.
87.4
92.1
92.9
99
5.5
5.8
17.7
0
0.5
0.78
0.76
0.74
0.73
0.94
0.94
0.92
0.94
0.93
∆ AIC ROC
1.00E+00
9.76E-02
6.43E-02
3.02E-03
6.11E+17
5.34E+17
1.40E+15
9.68E+18
7.50E+18
Evidence
Weight
Model Diagnostics
70.2%
70.3%
66.7%
66.5%
85.6%
84.8%
81.8%
83.5%
83.3%
Dead
Trees
71.5%
69.1%
63.8%
69.8%
80.7%
83.2%
74.9%
81.7%
81.7%
70.8%
69.7%
65.2%
68.2%
83.2%
84.0%
78.4%
82.6%
82.5%
Internal
Live
All
Trees Trees
25.0%
30.0%
35.0%
40.0%
65.0%
65.0%
56.5%
70.0%
70.0%
External
Dead
Trees
Correct Classification Rates
amongst trees from the model-building dataset. The external validation dataset consisted of dead trees from the BNM2000 site,
the reference model is more likely. Internal validation statistics are based on a bootstrapped estimate of correct classification
model versus a reference model. Scores above 7 indicate a substantial improvement, whereas weights below one indicate that
values above 0.8 suggest excellent discrimination. The Evidence Weight is an AIC-based indicator of the likelihood of a given
parison to top-ranked models. ROC is a measure of model discrimination, where a value of 0.5 suggests no discrimination and
of years over which variables were averaged indicated after variable type. The best single-variable models are shown for com-
mean sensitivity (Sens), duct number (NUM), average size (SIZE), density (DEN) and relative area (RelArea), with the number
Table 3. Best-ranked logistic mixed effect regression models of mortality risk. Variables include average growth (RW),
150
151
Figure Legends
Figure 1. Differences in growth and resin duct attributes between live (blue) and
dead (red) trees. Box and whisker plots show the median and quantiles for recent (3year average) growth and resin duct attributes across all sites and trees. Ring Width (A),
Basal Area Increment (B), number of ducts (C), duct mean size (D) duct density (E) and
duct relative area (F). Differences in means between live (blue) and dead (red) trees were
significant for ring width (RW, t = 3.103, p = 0.0022), number of ducts (t = 2.5368, p =
0.012), average duct size (t = 10.934, p < 0.0001), and duct relative area (t = 6.9357, p <
0.0001). Differences were not significant for basal area increment (BAI, t = 1.5201, p =
0.1303) or duct density (t = -0.1198, p = 0.9048) (Student’s t-test).
Figure 2. Resin duct and growth time series. Biweight robust mean-value chronologies
of RW and RD attributes from live (blue) and dead (red) trees across 2000s (A-E) and
1950s (F-J) sites. The number of trees entering into chronologies is shown by dotted lines
plotted on the right-hand y-axes. Although climate patterns varied between sites, which
may mask site-specific trends in the chronologies, regional climate coherence is generally
strong. Regionally wet and dry years are indicated by arrows (wet=solid, dry=dashed) in
the top panels.
Figure 3. Relationships between resin duct variables and ring width. Box-plots show
the distribution of Pearson’s correlations between resin duct (RD) variables and ring
widths (RW) within all individual live (blue) and dead (red) trees (A). Predicted values
152
with 95% confidence intervals of GAMM relationships between RW and RD number
(B), mean size (C), density (D), and relative area (E). Rug plots on A-E show the density
of measurements along a spectrum of RW levels. The dashed horizontal lines show
mean-values for each RD metric.
Figure 4. Comparisons of models including defense characteristics over models containing only growth predictors. The AIC-based Evidence Weight is plotted to show the
relative statistical support associated with combined growth and defense models (black),
defense-only models (blue), compared to growth-only models (green) (A). Site-specific
correct classification rates associated the most parsimonious growth-only (green) versus
growth and defense model (black) show how model improvement varies by site and
drought-mortality event (B).
Figure 5. The relationship between recent radial growth (RW3) and recent average
duct size (SIZE3) in surviving (blue) and dead (red) trees. A LOESS regression fit
with 95% confidence intervals (grey ribbon) is plotted over the data points from individual trees. Recent average resin duct size increases more rapidly with ring width in tress
that survived (L) versus died (D).
153
A
B
●
1.5
●
1.0
●
●
Ring Width (mm)
●
●
●
●
●
●
●
●
0.5
●
Basal Area Increment (cm2)
2.0
●
8
●
6
●
4
●
●
●
●
●
●
●
2
C
D
●
10
●
0.04
●
8
●
●
0.05
Mean Duct Size (mm2)
●
12
Number of Ducts
●
0
0.0
●
●
0.03
6
●
●
4
0.02
2
0.01
0
E
●
0.10
0.08
F
●
10
●
●
●
0.06
0.04
0.02
0.00
8
6
●
●
●
●
4
●
●
2
0
D
Figure 1.
12
Duct Density (no. mm2-1)
0.12
Duct Relative Area
●
L
Tree Status
D
L
1975
Figure 2.
E
60
1985
1995
Year
2005
0
10
20
20
40
60
D
F
1925
30
0
I
30
10
20
20
40
60
H
30
0
0
10
20
20
40
60
G
30
Ring Width (mm)
0.5
1.0
1.5
0
10
20
0 20 40 60 80 100
Number of Ducts
2
4
6
8 0.0
30
40
F
1935
1945
Year
1955
Number of Trees
0
0
C
20
40
0
B
10
20
Duct Density (# / mm2)
Duct Size (mm2)
0.02
0.04
0 1 2 3 4 5 0.00
A
0
0
Duct Relative Area
0.00
0.04
0.08
154
0.00
Relative Area
0.02
0.04
Duct Density (# / mm2)
0.0 0.5 1.0 1.5 2.0
15
Pearson's correlation
−0.5
Number
B
0
0
Figure 3.
1
1
2
2
3
3
4
D
4
Mean Duct Size (mm2)
0.00
0.02
0.04
0
Number of Ducts
5
10
155
A
1.0
0.5
0.0
Tree Status
Dead
Live
−1.0
Mean Size
Density
Ring Width (mm)
Relative Area
C
E
0
1
2
3
4
0
1
2
3
4
156
A
1E+19
Model Type
Resin Ducts
+ Growth
Resin Ducts Only
Evidence Weight
1E+17
1E+15
1E+13
1E+11
Growth Only
1E+9
1E+7
1E+5
1E+3
log(rw3)
Sens20
log(RW3) + Sens15
Size3
RelArea5 + Size3
RelArea3 + Size3
RW15 + Sens15
+ Size3
1E-3
RW20 + Sens15
+ Size3
1
-1
B
Correct Classification Rate
100%
80%
60%
40%
20%
0%
Figure 4.
TRP2000 WRK2000 BNM2000 SEV2000
BNM1950 SEV1950
157
0.05
SIZE3 (mm2)
0.04
0.03
0.02
Tree Status
D
L
0.01
0.0
Figure 5.
0.5
1.0
RW3 (mm)
1.5
2.0
158
Supplementary Materials
Table S1. Post-hoc tests of growth and resin duct differences between specific sites.
Results of the overall ANOVA are shown in Table 2. Adjustments for multiple comparisons were made using the Holm method. P-values less than 0.05 are in boldface type. RW
is annual ring width (mm/year), NUM is annual number of ducts, SIZE is annual average
duct size (mm2), DEN is annual density of ducts, and RelArea is the annual relative area
of duct tissue (%). Tests displayed here are for recent (3-year average) values.
RW3
SEV50
BNM2000
WRK2000
TRP2000
SEV2000
BNM50
1
0.001
1
1
1
NUM3
SEV50
BNM2000 WRK2000 TRP2000
2.70E-06
1
0.00224
1
0.00015
1
1
2.70E-05
1
1
BNM50
0.1495
0.002
1
0.9601
1
SEV50
2.00E-07
6.75E-02
1
0.1208
SIZE3
SEV50
BNM2000
WRK2000
TRP2000
SEV2000
1
1
0.1466
0.1572
1
0.6238
0.0135
0.0098
1
SEV50
BNM2000
WRK2000
TRP2000
SEV2000
0.156
1
0.039
1
1
0.379
2.30E-05
0.109
0.011
1
1
1
DEN3
1
0.1831
0.2083
0.054
0.242
1
RelArea3
0.051
1
1
BNM2000 WRK2000 TRP2000
0.1208
1.80E-05
0.4114
3.10E-03
1
0.8574
1
1
1
1
1
1
0.8
1
0.13
1
1
1
1
1
0.21
159
APPENDIX D: CLIMATE AND TREE INFLUENCES ON XYLEM RESIN DUCTS IN
PIÑON PINE
This paper was prepared to submit to Dendrochronologia.
Alison K. Macalady, Matthias Kläy, Monica L. Gaylord, Harald Bugmann, Nathan E.
English, Craig D. Allen, Thomas W. Swetnam, Nate G. McDowell, Connie Woodhouse
Abstract
Resin ducts are an important component of conifer defense against insects and
pathogens. Low levels of tree investment into xylem resin ducts, for example, relates
strongly to the risk of mortality during drought and bark beetle attack in a number of
species. Tree internal and climatic controls on the levels of such defenses, however, are
poorly quantified, limiting understanding of how defense against insects may vary with
climate, tree ontogeny and in relation to radial growth. Here we present results from the
first known dendroclimatological investigation of vertical xylem resin ducts in piñon
pine, a dominant conifer in woodlands of the southwestern USA that has been subject to
widespread tree mortality associated with severe drought and attack by various bark
beetles. Resin duct chronologies developed for four sites exhibit higher levels of intrinsic
variability and lower inter-series correlations compared to ring width chronologies, which
may reflect the greater dominance of tree internal versus environmental control on
defense anatomy. However, resin duct chronologies contain a coherent response to
160
climate that is distinct from the response of ring widths. Specifically, while ring width
chronologies respond positively to previous fall, winter and current spring precipitation,
and negatively to early summer temperature, resin duct number, density, size and relative
area (% of xylem area containing resin duct tissue) exhibit positive correlations with
previous year winter and early spring precipitation, and resin duct number, density, and
relative area respond positively to mid-late summer precipitation and drought. These
findings have implications for understanding tree vulnerability to drought and bark beetle
outbreaks due to climate variability and change. Results also suggest the possibility that
some types of resin duct chronologies may be useful as climate proxies, improving
reconstructions of especially summer climate in the SW USA.
Introduction
The influence of climate on tree secondary growth has been documented
thoroughly in tree-ring studies, especially in western North America (cf. Fritts 2001;
Wettstein et al. 2011; St George and Ault 2014). Among conifers in the semi-arid
southwestern US, the strong influence of drought on tree-ring widths, especially in the
seasons preceding the onset of spring cambial activity, is well established (Fritts, 1976;
Fritts et al., 1965a; Williams et al., 2013). These and related modeling studies (cf.
Vaganov et al 2006a) have improved understanding of how climatic factors control plant
growth, and bolster climatic reconstructions using the extensive network of tree-ring
width chronologies in the region (e.g., Cook et al. 2007).
161
Tree-ring widths are only one of many characteristics of tree secondary growth
that reflect the influence of abiotic factors on tree physiology, however. Xylem plays a
role in many essential physiological functions in woody plants, from water transport and
structural support to food storage and transport, chemical signaling, and the synthesis,
storage and transport of secondary metabolites and defense compounds. Environmental
variability is known to influence different xylem functions across many timescales. For
example, there is considerable phenotypic plasticity in how xylem tracheid dimensions
and vessel diameters respond to climate stress on intra-annual to decadal timescales
(Bryukhanova and Fonti, 2013; Deslauriers and Morin, 2005; DeSoto et al., 2011;
Eilmann et al., 2009; Fonti et al., 2009b; Sperry et al., 2006; Vaganov et al., 2006;
vonArx et al., 2012). Ray parenchyma - essential for tree carbon storage and mobilization
- also appear sensitive to inter-annual climate variability (Olano et al., 2013).
Dendrochronological approaches hold great potential for understanding how
climate influences xylem anatomical features and associated tree physiological functions,
including defense against insects and pathogens. Such information may be especially
useful for assessing forest responses to global change (Fonti et al., 2010; Wimmer, 2002),
and could accelerate development of new tree-ring based environmental proxies (e.g.,
George and Nielsen 2000; Panyushkina et al. 2003; Liang et al. 2013). Tree-ring studies
of plant defenses are surprisingly rare. In many conifers, resin ducts in the secondary
plant body originate in the cambium and form a network of living parenchyma and
epithelial cells responsible for producing, storing and transporting oleoresins in the xylem
and phloem (Bannan, 1936; Münch, 1919; Wu and Hu, 1997). These secondary
162
compounds form the primary conifer defenses against bark beetles and other stem-boring
insects (Franceschi et al., 2005), which are major agents of tree mortality and disturbance
in coniferous forests globally (Allen et al., 2010). Resin ducts can occur as a regular part
of xylem development, contributing to the constitutive or first-line defense system, or can
be formed as a part of an induced defense response to mechanical disturbance or insect
attack (Bannan, 1936; Franceschi et al., 2005; Swetnam et al., 2009).
Factors controlling the formation of induced or parallel lines of so-called
‘traumatic’ resin ducts has been studied more extensively (Arbellay et al., 2014a, 2014b;
Franceschi et al., 2002; Hudgins and Franceschi, 2004), but information on how external
factors influence formation of regularly-forming resin ducts is lacking. Some researchers
suggest a possible sensitivity of resin duct density to summer temperature or drought
(Kilpeläinen et al., 2007; Reid and Watson, 1966; Rigling et al., 2003; Wimmer and
Grabner, 1997). However, dendroclimatological approaches are uncommon, and those
that do exist have produced contrasting conclusions. Vertical xylem resin ducts appear to
respond to high summer temperatures and drought in Norway Spruce (Picea abies)
(Wimmer and Grabner, 1997) and Pinus sylvestris (Rigling et al., 2003), but no distinct
climate signal was found in Pinus nigra (Levanič, 1999) and P. leicodermis (Todaro et
al., 2007). Developing a better understanding of the climatic and internal influences on
xylem resin ducts is important for advancing our understanding of tree vulnerability to
insect attack, as xylem resin duct attributes affects resin production (cf. Blanche et al.
1992; Rodríguez-García et al. 2014) and tree predisposition to mortality from droughtbark beetle interactions (Ferrenberg et al., 2014; Kane and Kolb, 2010; Macalady et al.,
163
in preparation).
Xylem resin ducts in species common to the southwestern USA, where bark
beetle outbreaks and droughts have resulted in widespread tree mortality across millions
of hectares (Williams et al., 2013), have never been targeted for dendroclimatological
study. Here, we present a tree-ring based investigation of multiple aspects of vertical
resin ducts (RD) and ring widths (RW) in piñon pine (P. edulis Engelm.), a droughttolerant, pigmy conifer with a wide distribution across the region (Romme et al., 2009).
Specifically, we: (1) document variations in resin duct number, size, density (number of
resin ducts per unit xylem area) and relative area (ratio of xylem area to resin duct area in
each ring) with tree cambial age and tree diameter; (2) assess statistical properties of RD
and RW chronologies; and, (3) determine how monthly and seasonal climate variables
affects interannual variability in RD attributes, especially relative to radial growth.
Methods
Study sites
Increment cores for this study were taken from mature dominant or sub-dominant
living trees from four sites in central New Mexico: Sevilleta National Wildlife Refuge
(SEV), Tres Piedras (TRP), White Rock (WRK), and Bandelier National Monument
(BNM). Climate at these sites is semi-arid, with the scant precipitation arriving in two
distinct seasonal pulses (Sheppard et al., 2002). Widespread and spatially homogenous
snow and rain are typical in the late fall, winter and early spring, and are delivered via
regional frontal systems that are often associated with the winter jet stream. Late spring
164
and early summer are marked by a distinct hot and dry period. Summer rains, which
constitute over half of total precipitation in some areas, are delivered via more localized,
spatially heterogeneous convective storms associated with the North American Monsoon
(Sheppard et al., 2002). Monthly average precipitation and temperature data from SEV
are shown in Figure 1. More detailed descriptions of each site can be found in (Macalady
and Bugmann, 2014).
Field and laboratory methods
Two cores were extracted from 20 living trees at SEV and TRP, and from 10 trees
at WRK. One or two cores from 20 additional live trees from SEV and BNM (which is
located only a few kilometers from WRK) were available from earlier collections
archived at the Laboratory of Tree-Ring Research in Tucson, Arizona (Table 1). All cores
were prepared using standard dendrochronological techniques (Stokes and Smiley, 1968),
and were crossdated visually based on ring widths. Ring widths were measured to the
nearest micron, with the measured interval spanning the period 1975-2010 and 19261960, depending on the dataset (Table 1). Crossdating was verified statistically using the
COFECHA program (Grissino-Mayer, 2001; Holmes, 1983). A few sampled cores could
not be crossdated reliably, or if dated, could not be measured reliably for resin ducts;
these were excluded from subsequent analyses (see Table 1 for details on the number of
retained cores).
A cambial age was assigned to each ring by counting years starting from the pith
of each core. Where cores did not intersect the pith, years to pith were determined using
165
a visual estimator (Applequist, 1958). Vertical (axial) xylem resin ducts were measured
using a variation of the method detailed in Kane and Kolb (2010). Core surfaces were
sanded and then polished progressively, ending with 9 micron finishing paper. Due to
variability in the width of the increment cores, sample windows exactly 5-mm wide were
scanned to 2400 dpi with a distortion-free scanner (Color Scanner Epson Expression
10000XL). Images were imported into ImageJ (Rasband, 1997). Resin ducts in each ring
within the 5-mm wide windows (consisting of the duct lumen and surrounding epithelial
cells) were counted (resin duct number or RD NUM) and traced using a digital stylus.
Total measured xylem area was also traced and measured. The mean size of ducts (RD
Size) was calculated from the areas of all individual ducts within each ring. Duct density
(RD DEN) was calculated by dividing RD NUM by the measured xylem area in each
year, and duct relative area (RD RelArea) was calculated by dividing the total measured
resin duct area by the measured xylem area. In years with no xylem growth within the
measurement window (locally absent rings), a zero was assigned for RD NUM, but RD
Size, DEN and RelArea were assigned NAs, as no measurements could be made of those
attributes. When a ring was present but did not contain resin ducts within the
measurement window, a zero was assigned for all RD variables except RD Size, which
could not be calculated and was therefore also assigned an NA.
Data from SEV contained more replication and more temporal coverage
compared to other sites, and were used in the majority of analyses. Data from other sites
were collected as part of different study using shorter time series (Macalady et al., in
preparation), and were used in assessing RD relationships with cambial age and tree size,
166
and to check the generality of tree-ring statistics and responses of RD chronologies to
climate. Data from archived SEV and BNM samples contain measurements that do not
overlap in time with measurements from the recently collected material from SEV and
WRK, and were used to check the temporal stability of relationships (Table 1).
Statistical analyses
Raw, core-level measurements were detrended (to remove age-related and bole
geometry effects) using a cubic smoothing spline with a frequency response of 0.5 at a
wavelength equal to 0.67 of the length of the measured time series (Cook and Peters,
1981). In the case of RD time series that contained NAs, the cubic smoothing spline was
fit to a time series with the NAs removed, and then the NAs were re-inserted into the
indexed series. Core indices were calculated by dividing raw measurements by the fitted
smoothing spline. Pooled autocorrelation was modeled and removed from detrended
index series in order to produce residual indices (Meko, 1981). Robust bi-weight mean
chronologies of RW, NUM, Size, RelArea and DEN were computed for each site by
averaging over all residual indices (Cook and Kairiukstis, 1990).
We report the mean, average mean sensitivity and first-order autocorrelation of all
raw core-level data from SEV, and characterize the strength of the common signal in RW
and RD chronologies from SEV by calculating the effective chronology signal (rbareff)
(Cook and Kairiukstis, 1990) and the expressed population signal (EPS) from index
series (Wigley et al., 1984). Rbareff values above 0.5 indicate a strong common signal
167
between cores and trees, and EPS values above 0.85 are considered best for climate
reconstruction.
We used Pearson’s correlation to assess the relationship between indexed RW and
RD mean-value chronologies. Some resin duct attributes exhibited a strong dependence
on radial growth, consistent with the fact that resin ducts occur within the xylem (Table
3). We were interested in understanding both raw relationships between resin duct
attributes, radial growth and climate, and relationships between resin duct chronologies
and climate that are independent from the influence of radial growth on resin ducts.
Therefore, we also calculated ‘adjusted’ resin duct chronologies in which the statistical
dependence of RD measurements on RW was removed (Griffin et al., 2011; Olano et al.,
2013). We made this adjustment at the core versus the tree or chronology level because
relationships at the core level more closely reflect physiological links between xylem
radial growth and resin duct formation. Following Griffin et al. (Griffin et al., 2011),
core-level RD index series were regressed onto core-level RW index series, and residuals
from this regression were averaged into ‘adjusted’ resin duct (RDa) chronologies.
Climate response of ring widths and resin duct attributes
We assessed relationships between tree-ring variables and climate by comparing
tree-ring width and resin duct mean and adjusted mean value chronologies to monthly
and seasonal climate data from the Parameter-elevation Regressions on Independent
Slopes Model (PRISM) gridded dataset (Daly et al., 1994). We downloaded monthly
temperature, dew point, and precipitation data and averaged across the nearest four 4 km-
168
square grid cells to create a monthly climate record for each site. We calculated
bootstrapped monthly correlations and partial correlation for the tree-ring data from SEV
using the Seascorr routine (Meko et al., 2011), modified for use in the R-based package
Treeclim (Zang and Biondi, 2013). Seascorr requires that primary and secondary climate
variables are assigned, and then the relationship between tree ring chronologies and the
secondary climate variable is assessed using partial correlations, e.g., after taking into
account the correlations with the primary climate variable and between climate variables.
Based on preliminary analyses, we assigned accumulated precipitation as the primary
correlate, and mean temperature as the secondary variable.
We used multiple stepwise regressions on ring width and adjusted resin duct
chronologies from SEV to determine seasonal climate variables that most strongly
control radial growth and resin duct attributes. Three-month climate seasons included the
months starting with the August prior to the current growth year and ending with October
of the current growth year, for a total of ten possible 3-month seasonal climate predictor
variables. These ten variables were entered into the multiple regression in the order of
highest to lowest correlation with tree-ring chronologies, with partial correlations
considered after each entered variable (see Williams et al. 2013). We assessed spatial and
temporal variability in climatic sensitivity of different growth and resin duct chronologies
by comparing simple Pearson’s correlations between ring width and adjusted resin duct
chronologies and seasonal PRISM climate data from all sites.
We performed all analyses using R, version 3.1.0 (R Core Team, 2014). The dplR
library v. 1.5.9 was used for calculating tree-ring statistics (Bunn, 2008). DetrendeR v.
169
1.0.4 was used for autoregressive modeling and detrending of tree-ring time series (Filipe
Campelo, 2012). Treeclim v. 1.0.6 was used for bootstrapped estimates of correlation
between climate and tree-ring variables (Zang and Biondi, 2013).
Results
Age and size-related trends in resin duct attributes
The number of resin ducts (RD NUM) in each annual ring decreases precipitously
with cambial age; however, RD Size remains fairly constant throughout the life of a tree
(Figure 2). RD RelArea is also fairly constant with respect to cambial age (Figure 2).
With respect to tree size, there is no consistent pattern between recent ring widths or RD
NUM and tree diameter, whereas RD Size increases strongly and RD DEN decreases
significantly with tree size (Figure 3). RD RelArea also increases with tree size but the
relationship is not significant (α = 0.05; Figure 3).
Statistical properties of tree-ring width and resin duct chronologies
Despite being contained within the xylem, raw RD time series from SEV2000
exhibit statistical properties that are very different from the ring-width chronology (Table
2). Whereas RW tends to be positively autocorrelated, resin duct attributes exhibit
negative first order autocorrelation, with autocorrelation coefficients particularly negative
for RD NUM, DEN and RelArea. It follows that mean sensitivity is high in RD NUM,
DEN and RelArea compared to RW. Resin duct Size exhibits relatively low mean
170
sensitivity as a result of low relative standard deviation (not shown) and much less
negative AR1 than other duct variables (Table 2).
The amount of common variability in RD time series is considerably lower than
in RW (Figure 4), with series inter-correlations (expressed as rbareff) that are considerably
higher than 0.5 in the RW chronology, but much lower among resin ducts, especially RD
Size (Table 2). Adjusted RD chronologies, which have had the influence of RW on RD
formation removed at the core level, exhibited even lower rbareff values. Despite low
inter-series correlations, EPS values for entire chronologies were above 0.8 with the
exception of the RD Size chronology, which exhibited lower inter-series correlations in
addition to smaller average sample sizes in each year due to the more common presence
of NA values in core-level chronologies. Although absolute values of statistics differed
slightly, the patterns across chronologies at SEV2000 are consistent with those among
SEV1950 chronologies (Table S1, Figure S1), as well as with patterns at the other sites
listed in Table 1 (not shown).
Correlations among growth and resin duct mean-value chronologies
Mean-value index chronologies from SEV2000 are all positively correlated, with
correlations between RW and RD NUM and RD Size particularly large (Table 3). Despite
being metrics that reflect the amount of resin duct tissue per unit of xylem area, the
common high-frequency variability contained in RD RelArea and DEN chronologies is
still positively correlated with RW (Table 3; Table S2). Time series plots of mean value
chronologies (Figure 4) reflect the overall positive correlations between chronologies,
171
though “pointer years” in the RW chronology (i.e., consistently very narrow or wide rings
among most trees in a site; also knows as “signature years” sensu Douglass 1941) are not
always reflected equally in RD chronologies. Specifically, wide ring-width pointer years
appear less consistently associated with large RD chronology variables versus narrow
pointer years (Figure 4, S1). When adjusted chronologies are considered, RD Size
remains strongly and positively associated with RW, but is less well correlated with RD
NUM, DEN and RelArea. Adjusted RD RelArea and DEN mean-value chronologies are
very highly and positively associated, reflecting the dominance of variability in the
number of resin ducts over resin duct size in determining shared interannual variability in
the RelArea adjusted chronology.
Relationships between climate and tree-ring chronologies
The RW index chronology from SEV2000 exhibits strong positive correlations
with precipitation in the autumn prior to current growing season, and through the winter
and spring of the current year (Figure 5). Average temperature correlates negatively with
RW especially in the previous autumn, current spring and early summer. Despite
relatively inter-series correlations that are much lower than that of RW chronologies
(Table 2), RD mean-value chronologies also exhibit significant correlations with monthly
precipitation and mean temperature. Correlations with unadjusted RD chronologies show
similarity with RW correlations, especially in regard to positive correlation with
precipitation in the previous autumn and winter months. However, whereas correlations
between RW and current summer precipitation and temperature are weak, RD NUM,
172
DEN and RelArea all exhibit strong negative associations with later summer and early
autumn precipitation, and positive associations with temperature.
Slight differences in the timing of maximum correlation with precipitation
between RD and RW chronologies, along with the strong influence of late growing
season conditions on RD NUM, DEN and RelArea, are more pronounced in correlations
between monthly climate variables and RDa chronologies (Figure 6). Once the influence
of RW on RD chronologies is taken into account, NUM, DEN and RelArea appear less
influenced by previous autumn and winter precipitation, but remain strongly and
positively influenced by precipitation in the current year’s late winter and early spring
The negative RD correlations with June, July and August precipitation and positive
partial correlations with temperature late into the current year’s growing season are
persistent and even more pronounced.
In contrast to the other RD metrics, RD Size appears to have a relationship to
climate that is most similar to RW, although strong positive associations with FebruaryMarch precipitation are not present in the adjusted chronology, and the positive
association with early summer precipitation has shifted from May to June (Figure 6). As
with RW, the RD Size chronology also lacks the strong positive association with mean
summer and autumn temperature that characterized other RD chronologies.
Climatic responses vary slightly between SEV2000 and SEV1950 tree-ring
chronologies, however adjusted RD NUM, DEN and RelArea chronologies maintain
strong positive correlations with mid-winter and early spring precipitation (especially
February), along with mostly negative correlations with summer precipitation, if during a
173
more restricted number of months (Figures S2, S3). In contrast to the marked positive
partial correlations with late season temperature for the SEV2000s trees, the partial
correlations between adjusted resin duct chronologies and summer temperature is muted
in the SEV1950s trees (Figures S2, S3).
Stepwise multiple regressions between the combined chronologies from SEV2000
and SEV1950 and 3-month seasonal climate variables are all highly significant, though
the overall association between climate and RW is stronger than between climate and RD
chronologies (Table 3). Differences in seasonal climate sensitivities are also apparent.
RW is strongly and positively associated with precipitation in the previous autumn,
winter, spring, and early summer seasons, and negatively associated with late spring and
early summer temperature. RDa chronologies are also strongly and positively related to
FMA precipitation, but with the exception of RD Size, they are not strongly related to
previous autumn and winter precipitation. Furthermore, with the exception of RD Size,
associations between RDa chronologies and MJJ-ASO precipitation are negative, with
individual seasonal predictors significant (p < 0.01) in many cases. Additional
association with summer and autumn mean temperature are positive, when present,
although weaker than the association with precipitation. Correlations with seasonal
climate variables were similar when vapor pressure deficit (VPD) was used instead of
temperature in the regressions (Table S3). Regressions with VPD explained more of the
variability in RW, but slightly less of the variability in RDa chronologies. The pattern of
correlations between seasonal climate variables and tree-ring chronologies varies slightly
174
across sites (Figure 7). However, the main effects outlined in Table 3 are generally
conserved.
Discussion
Cambial age and tree size-related changes in resin defenses
Declining ring widths with tree cambial age is a common phenomenon that may
be driven by physiological changes, changes in tree competitive status or changes in bole
geometry (Cook and Kairiukstis, 1990). Cambial age-related trends in piñon RD NUM
mirrors the expectation of exponentially declining RW due to changes in bole geometry,
and underscores the tight linkages between xylem formation and the number of resin
ducts formed per annual year in P. edulis (Table 3). However cambial age-related trends
in RD attributes may also reflect species-specific physiology. In Picea abies, for
example, RD NUM declined precipitously in the first few years from the pith and then
stabilized, and RD density followed a similar trend as RD NUM, indicating declines in
resin duct frequency independent of xylem formation (Wimmer and Vetter, 1999). Flat or
very slight increases in piñon RD Size, DEN, and RelArea with distance to pith is more
consistent with findings in Pinus radiata, where RD NUM and Size also increased
slightly with distance to the pith (Cown et al., 2011). In relation to tree size, it may be
that in pine species, constitutive resin duct defenses actually increase slightly, perhaps
reflecting the greater vulnerability of larger trees to bark beetle attack (Amman and Baker
1972). For example piñon RD Size, the most important resin duct attribute for
175
determining tree susceptibility to mortality during drought (Macalady et al., in
preparation), increases strongly with tree size, all else being equal (Figure 3).
Statistical properties of RD chronologies
One of the more striking differences between RW and RD chronologies is the
tendency for negative autocorrelation structure and high mean sensitivity in RD NUM,
DEN and RelArea chronologies, relative to RW (Table 2, S1). Although negative
autocorrelation is a statistical property that is apparently rare in tree-ring growth
chronologies (Bunn et al 2013), negative first-order autocorrelation (AR1) has also been
documented in RD density measurements in P. contorta (Reid and Watson, 1966), and
Picea abies (Wimmer and Grabner, 1997), much lower positive AR1 characterize resin
duct versus RW chronologies in P. sylvestris (Rigling et al., 2003), and negative AR1
characterizes some vessel diameter chronologies (Fonti et al. 2007). Mean sensitivity
(MS), a metric of year-to-year variability that is not independent from AR1, is also nearly
double that of RW among our piñon RD NUM, DEN and RelArea chronologies, with
absolute values exceeding the top 95% of values from 11 piñon RW chronologies
analyzed by (Fritts and Shatz, 1975). In Picea abies, MS values were not as high as in our
study, but were ~5 times higher than in corresponding RW chronologies (Wimmer and
Grabner, 1997). The reasons for these statistical properties are not entirely clear, but high
MS and low AR1 suggests that the formation of resin ducts and other anatomical features
is less influenced by factors that are known to increase autocorrelation in ring widths,
such as needle retention and carbon storage from previous years. Structural properties of
176
wood anatomical measurements may also be a factor: for example, zeros are more
common in the record of resin duct counts (e.g., the number of resin ducts can be zero
both when RW is zero and when there are no resin ducts within a larger xylem ring),
which can influence MS values in particular.
The other consistent difference between RD and RW chronologies is the low
signal strength especially in the adjusted RD chronologies (Tables 2, S1). A similar
tendency is apparent among other RD and non-RD tree-ring anatomical variables (Fonti
et al., 2009a; Olano et al., 2013; Wimmer and Grabner, 1997). Mean RD interseries
correlations were close to half that of corresponding RW chronologies in P. abies
(RW=0.38-0.42, RD=0.24-0.28; Wimmer and Graber 1997), and mean between-tree
correlations (Rbar) of ray paryenchma chronologies ranged from -0.005-0.030 versus
0.295 for corresponding ring widths (Olano et al., 2013). Greater intrinsic control on
anatomical features and limits to anatomical variability imposed by the structures’
functionality are both likely underlying reasons for lower common variability among
wood anatomy chronologies. While the latter may strongly limit RD Size in particular,
we suggest that low signal strength in resin duct parameters likely reflects strong genetic
controls on resin ducts and tree resin defenses generally (Westbrook et al., 2013). Studies
in other species, notably Picea abies, indicate stronger genetic control over xylem RD
attributes than radial growth (Rosner and Hannrup, 2004).
177
Climatic controls on radial growth, resin duct frequency and size
Despite the relatively high levels of intrinsic variability in P. edulis RD
chronologies, the shared variability among the records is strongly associated with climate
(Figures 5-7, S1-S2; Tables 4, S3). Co-variability in resin duct chronologies across sites
(not shown) and their similar climatic responses (Figure 7) give us confidence in the
validity of this interpretation. The response of RD attributes to monthly and seasonal
climate variables is similar and yet distinct from climatic influences on interannual
variations in radial growth. RW correlates positively and most strongly with precipitation
in the autumn and winter prior to the growing season, and also exhibits positive
precipitation and negative temperature relationships during late spring/early summer
(Figures 5-7, S1-S2; Tables 4, S3). This is a well-established response among conifer
total ring width chronologies in the southwestern U.S. (Fritts et al. 1965a; Fritts 2001;
Williams et al. 2013), and reflects the influence of 1) soil water recharge during previous
autumn and winter months on plant available water during the growing season; and 2)
production of photosynthates during wetter, cooler months that are stored and then
available for growth after the activation of the cambium in the spring (Fritts, 1976).
Temperature during the arid early summer, and to some degree during the previous
autumn, also correlates negatively with RW due to a tight relationship between
temperature, atmospheric moisture demand, and plant water stress during and prior to and
during the growing season (Williams et al., 2013).
Similar to RW, P. edulis RD chronologies exhibit positive relationships to
previous autumn and winter precipitation (Figures 5,7, S1), reflecting at least in part the
178
fact that resin ducts are contained within the xylem. However, with the exception of RD
Size, the relationship between RD chronologies and precipitation during previous autumn
and early winter months weakens considerably once the influence of RW on RD
chronologies is removed. Significant positive relationships with precipitation in February,
March and April are persistent, however, pointing to the strong sensitivity of RD
formation and size to late winter and early spring moisture availability, above and beyond
the influence of FMA precipitation on RW (Figures 6-7, S2; Table 5). Positive
relationships between winter and early spring precipitation and RD formation and size in
P. edulis may reflect the particular importance of winter and spring drought to limiting
the amount of stored photosynthates available for investment into both growth and
defense. Due to direct climatic controls on cambial activity (Vaganov et al 2011) winter
soil moisture recharge may also be important simply for maintaining cambial activity
through the latter part of the growing season, when the formation of resin ducts is
apparently concentrated (Blanche et al., 1992; Rigling et al., 2003; Wimmer and Vetter,
1999)).
The development of a greater number, density and relative area of resin duct
tissue is also consistently associated with dry conditions in July and August. This is in
stark contrast to the negative response of RW to early summer drought and the lack of
consistent influence of mid-late summer and early autumn climate on current year
growth. Warm/dry September and October conditions may have an additionally positive
effect on resin duct formation, as is apparent in some of our chronologies (Figure 7).
Radial growth among trees in southern Colorado had largely ceased by August (Fritts et
179
al., 1965b), but our results suggest that in New Mexico, P. edulis cambium may remain
active through September.
RD positive response to summer drought is consistent with the response of RD
DEN documented in dendroclimatological study of Picea abies and Pinus sylvestris
(Rigling et al., 2003; Wimmer and Grabner, 1997), as well as with anecdotal studies of
RD density in P. contorta . Experimental studies on seedlings of P. halpensis and P.
sylvestris also revealed a positive association between RD formation and temperature
(Kilpeläinen et al., 2007; Zamski, 1972). The physiology underlying such associations
remains obscure, but there are several plausible hypotheses. The growth differentiation
balance hypothesis, or the GDBH (Herms and Mattson, 1992; Loomis, 1953; Lorio Jr.,
1986), is a leading theory of plant defense that suggests moderate, short-term drought
such as that which occurs annually in early summer in the SW, may curtail cell division
and elongation but not photosynthesis, due a tighter and more direct control of growth by
water availability. This has the effect of freeing up carbon for cell differentiation
processes, such as those associated with the formation of resin and resin ducts.
Warm temperatures in early autumn may also prolong cambial activity (Peltola et
al. 2002), allowing for continued cell differentiation and possibly, the formation of new
resin ducts by differentiation of xylem mother cells. Due to the minimal impact of
latewood on final radial increment (Fritts et al., 1965b), any such prolongation may have
minimal effect on RW chronologies, but significant effects on the formation of resin
ducts. Hormonal cues linking resin duct formation directly to drought or temperature
stress may also play an important role. Hormones are known to trigger the rapid, induced
180
formation of ‘traumatic’ ducts after mechanical disturbance (Franceschi et al., 2002;
Hudgins and Franceschi, 2004; Werker and Fahn, 1969), and production of ethylene, one
of the hormones involved in traumatic duct formation, has also been documented
following climate excursions such as temperature extremes and drought (Abeles et al.,
1992).
The climate response of RD Size chronologies stands in contrast to the response
of all other resin duct attributes. RD Size exhibits weaker and more variable responses
across sites than other RD attributes (Figure 7), consistent with much lower inter-series
correlations (Table 2, S1). Low RD Size inter-series correlation may also be influenced
by changes in the amount of resin contained in ducts, which is not necessarily stable from
year-to-year (Hudgins and Franceschi, 2004; Münch, 1919). Thus any intrinsic influence
of climate on resin duct size may be blurred by the influence of climate or other stimuli
on resin production during the multiple years across which resin duct cells remain active.
Despite the high level of ‘noise’, the RD Size climate response is still significant in some
months and seasons at SEV, though the response is largely similar to the response of RW
even in the adjusted chronology (Figures 5-7, S1, S2). It is unclear why the response of
RD Size tracks the RW response more closely other RD attributes, when theory and
experimental findings imply that resin formation should also be influenced by stressors
such as drought, as well as by any increase in available carbohydrates late in the growing
season when growth sinks have diminished.
181
Implications for dendroclimatology
There are limitations to fully characterizing the climatic response of any tree-ring
attribute in short time series such as ours. However, we have confidence in our basic
interpretations given that we found consistent patterns across a number of sites, and
similar responses during two distinct temporal periods that each span ~30 years (Figure
7). Although more study may be especially useful in characterizing any unique climatic
response of RD Size, our results reveal a distinct climate response of other RD attributes
that may be useful for improving seasonal climate reconstructions. In a stepwise
regression, including resin duct attributes along with ring widths substantially improved
the characterization of climate response variables including FMA precipitation, as well as
late growing season precipitation and temperature (not shown). The potential for
additional summer climate information may be particularly important in the SW US,
where summer rains play an essential role in regional water supplies and ecosystem
function (Ray et al., 2007). Summer monsoon variability is poorly captured in total ring
width chronologies, but is reflected in latewood chronologies of ponderosa pine and
Douglas-fir (Griffin et al. 2013). Latewood in species such as P. edulis, which is
widespread at lower elevations and whose geographic distribution closely mirrors the
footprint of the NAM, does not appear to be a reliable indicator of summer moisture (C.
Woodhouse, unpublished data). RD chronologies might therefore be useful for further
characterizing summer climate, especially in areas where opportunities for the
development of summer-sensitive latewood chronologies are more limited.
182
Improvements in digital imaging could improve the speed with which resin duct
chronologies are produced, helping to overcome low signal strength. Currently, the most
tedious portion of the process relates to measuring resin duct size, but this variable may
not need to be measured to capture the most relevant information. If our current results
for P. edulis remain robust to further study, resin duct size may not be important as a
separate climate proxy, and the common signal and associated climate response in RD
DEN and RD RelArea chronologies appears to be quite similar (Figure 7).
Implications for tree defenses under climate change
Low xylem resin duct density, size and/or area have been shown to relate strongly
to mortality risk during droughts and associated bark beetle outbreaks in a number of
species (Ferrenberg et al., 2014; Kane and Kolb, 2010; Macalady et al., in preparation).
Here we show that xylem resin duct formation is influenced by interannual changes in
seasonal and monthly climate. This suggests that aspects of tree vulnerability to and
defense against bark beetles may be altered by ongoing climate change. Recent drought
in the region may be an early manifestation of climate change (Garfin et al., 2014; Seager
et al., 2007; Williams et al., 2013), and the SW is likely to become even hotter and drier
in the 21st century (Cayan et al., 2010; Garfin et al., 2014). Although predictions for
changes in summer precipitation are less robust than those for winter precipitation (Cayan
et al., 2010), anticipated increases in temperature during late summer, along with any
increases in summer drought, could enhance some aspects tree defenses in P. edulis, all
else being equal. However, two additional findings are likely to negate or reverse any
183
potential benefit of summertime warming and drying. These include the fact that copious
winter precipitation – which is very likely to decline under climate change - appears to be
equally important for resin duct formation. Furthermore, the metric most important for
survivorship in P. edulis – RD Size - appears least well controlled by climate, with the
weak climate response we observed limited to a positive relationship to winter and spring
precipitation.
Conclusions
Despite high levels of intrinsic variability and low interseries correlations, resin
duct chronologies from P. edulis across New Mexico contain distinct information about
climate. The climate response of RD NUM, DEN and RelArea is characterized by
positive relationships with winter precipitation and a positive response to summer
drought and temperature. A coherent climate response of P. edulis RD attributes stands in
contrast to observations in P. nigra (Levanič, 1999) and P. leicodermis (Todaro et al.,
2007), in which no climatic response was found, but confirms observations of a positive
relationship to summer temperature and/or drought in P. sylvestris and Picea abies
(Rigling et al., 2003; Wimmer and Grabner, 1997). The positive response of P. edulis RD
attributes to winter precipitation has not been observed previously in other species. The
patterns documented here provide evidence about climatic influences on the level of
conifer resin defenses in the SW US, which can in turn inform understanding of seasonal
and year-to-year vulnerability of forest trees to insect attack and mortality during
drought. The current study may also serve as a basis for improving reconstructions of
184
climate in the SW USA, especially summer climate, through the development of longer
and more replicated RD chronologies.
Acknowledgements
AKM was supported by a US Department of Energy Graduate Environmental Research
Fellowship and a grant from Los Alamos National Laboratory, IGPP. MK was supported
by a Haury Fellowship from the University of Arizona, LTRR. NE was supported by a
grant from DOE-Office of Science. NE was supported by Los Alamos National Lab’s
Lab Directed Research and Development program.
Citations
Abeles, F.B., Morgan, P.W., Saltveit Jr, M.E., 1992. Ethylene in plant biology. Academic
press.
Allen, C.D., Macalady, A.K., Chenchouni, H., Bachelet, D., McDowell, N., Vennetier,
M., Kitzberger, T., Rigling, A., Breshears, D.D., Hogg, E.H. (Ted), Gonzalez, P.,
Fensham, R., Zhang, Z., Castro, J., Demidova, N., Lim, J.-H., Allard, G.,
Running, S.W., Semerci, A., Cobb, N., 2010. A global overview of drought and
heat-induced tree mortality reveals emerging climate change risks for forests. For.
Ecol. Manag. 259, 660–684.
Amman, G.D., Baker, B.H., 1972. Mountain pine beetle influence on lodgepole pine
stand structure. J. For. 70, 204–209.
Applequist, M., 1958. A simple pith locator for use with off-center increment cores. J.
For. 56, 141.
Arbellay, E., Stoffel, M., Sutherland, E.K., Smith, K.T., Falk, D.A., 2014a. Resin duct
size and density as ecophysiological traits in fire scars of Pseudotsuga menziesii
and Larix occidentalis. Ann. Bot. 114, 973–980.
185
Arbellay, E., Stoffel, M., Sutherland, E.K., Smith, K.T., Falk, D.A., 2014b. Changes in
tracheid and ray traits in fire scars of North American conifers and their
ecophysiological implications. Ann. Bot. 114, 223–232.
Bannan, M.W., 1936. Vertical Resin Ducts in the Secondary Wood of the Abietineae.
New Phytol. 35, 11–46.
Blanche, C.A., Lorio Jr., P.L., Sommers, R.A., Hodges, J.D., Nebeker, T.E., 1992.
Seasonal cambial growth and development of loblolly pine: Xylem formation,
inner bark chemistry, resin ducts, and resin flow. For. Ecol. Manag. 49, 151–165.
Bryukhanova, M., Fonti, P., 2013. Xylem plasticity allows rapid hydraulic adjustment to
annual climatic variability. Trees 27, 485–496.
Bunn, A., 2008. A dendrochronology program library in R (dplR). Dendrochronologia
26, 115–124.
Cayan, D.R., Das, T., Pierce, D.W., Barnett, T.P., Tyree, M., Gershunov, A., 2010.
Future dryness in the southwest US and the hydrology of the early 21st century
drought. Proc. Natl. Acad. Sci. 107, 21271–21276.
Cook, E.R., Kairiukstis, L.A., 1990. Methods of dendrochronology: applications in the
environmental sciences. Springer.
Cook, E.R., Peters, K., 1981. The smoothing spline: a new approach to standardizing
forest interior tree-ring width series for dendroclimatic studies.
Cook, E.R., Seager, R., Cane, M.A., Stahle, D.W., 2007. North American drought:
reconstructions, causes, and consequences. Earth-Sci. Rev. 81, 93–134.
Cown, D.J., Donaldson, L.A., Downes, G.M., 2011. A review of resin features in radiata
pine. N. Z. J. For. Sci. 41, 41–60.
Daly, C., Neilson, R.P., Phillips, D.L., 1994. A Statistical-Topographic Model for
Mapping Climatological Precipitation over Mountainous Terrain. J. Appl.
Meteorol. 33, 140–158.
Deslauriers, A., Morin, H., 2005. Intra-annual tracheid production in balsam fir stems and
the effect of meteorological variables. Trees 19, 402–408.
DeSoto, L., la Cruz, M. De, Fonti, P., 2011. Intra-annual patterns of tracheid size in the
Mediterranean tree Juniperus thurifera as an indicator of seasonal water stress.
Can. J. For. Res. 41, 1280–1294.
186
Eilmann, B., Zweifel, R., Buchmann, N., Fonti, P., Rigling, A., 2009. Drought-induced
adaptation of the xylem in Scots pine and pubescent oak. Tree Physiol. 29, 1011–
1020.
Ferrenberg, S., Kane, J.M., Mitton, J.B., 2014. Resin duct characteristics associated with
tree resistance to bark beetles across lodgepole and limber pines. Oecologia 174,
1283–1292.
Filipe Campelo, 2012. detrendeR: Start the detrendeR Graphical User Interface (GUI).
Fonti, P., Eilmann, B., García-González, I., von Arx, G., 2009a. Expeditious building of
ring-porous earlywood vessel chronologies without loosing signal information.
Trees 23, 665–671.
Fonti, P., Treydte, K., Osenstetter, S., Frank, D., Esper, J., 2009b. Frequency-dependent
signals in multi-centennial oak vessel data. Palaeogeogr. Palaeoclimatol.
Palaeoecol. 275, 92–99.
Fonti, P., von Arx, G., García‐González, I., Eilmann, B., Sass‐Klaassen, U., Gärtner, H.,
Eckstein, D., 2010. Studying global change through investigation of the plastic
responses of xylem anatomy in tree rings. New Phytol. 185, 42–53.
Franceschi, V.R., Krekling, T., Christiansen, E., 2002. Application of methyl jasmonate
on Picea abies (Pinaceae) stems induces defense-related responses in phloem and
xylem. Am. J. Bot. 89, 578–586.
Franceschi, V.R., Krokene, P., Christiansen, E., Krekling, T., 2005. Anatomical and
Chemical Defenses of Conifer Bark against Bark Beetles and Other Pests. New
Phytol. 167, 353–375.
Fritts, H.C., 1976. Tree rings and climate. Academic Press, London, U.K.
Fritts, H.C., Shatz, D.J., 1975. Selecting and characterizing tree-ring chronologies for
dendroclimatic analysis.
Fritts, H.C., Smith, D.G., Cardis, J.W., Budelsky, C.A., 1965a. Tree-ring characteristics
along a vegetation gradient in northern Arizona. Ecology 394–401.
Fritts, H.C., Smith, D.G., Stokes, M.A., 1965b. The biological model for paleoclimatic
interpretation of Mesa Verde tree-ring series. Mem. Soc. Am. Archaeol. 101–121.
Garfin, G., Blanco, H., Comrie, A., Piechota, T., Waskom, R., 2014. Ch. 20: Southwest.,
in: Climate Change Impacts in the United States: The Third National Climate
Assessment. U.S. Global Change Research Program, pp. 462–486.
187
George, S.S., Nielsen, E., 2000. Signatures of high-magnitude 19th-century floods in
Quercus macrocarpa tree rings along the Red River, Manitoba, Canada. Geology
28, 899–902.
Griffin, D., Meko, D.M., Touchan, R., Leavitt, S.W., Woodhouse, C.A., 2011. Latewood
chronology development for summer-moisture reconstruction in the US
Southwest. Tree-Ring Res. 67, 87–101.
Griffin, D., Woodhouse, C.A., Meko, D.M., Stahle, D.W., Faulstich, H.L., Carrillo, C.,
Touchan, R., Castro, C.L., Leavitt, S.W., 2013. North American monsoon
precipitation reconstructed from tree‐ring latewood. Geophys. Res. Lett. 40, 954–
958.
Grissino-Mayer, H.D., 2001. Evaluating crossdating accuracy: a manual and tutorial for
the computer program COFECHA. Tree-Ring Res. 57, 205–221.
Herms, D.A., Mattson, W.J., 1992. The Dilemma of Plants: To Grow or Defend. Q. Rev.
Biol. 67, 283–335.
Holmes, R.L., 1983. Computer-assisted quality control in tree-ring dating and
measurement. Tree-Ring Bull. 43, 69–78.
Hudgins, J.W., Franceschi, V.R., 2004. Methyl jasmonate-induced ethylene production is
responsible for conifer phloem defense responses and reprogramming of stem
cambial zone for traumatic resin duct formation. Plant Physiol. 135, 2134–2149.
Kane, J.M., Kolb, T.E., 2010. Importance of resin ducts in reducing ponderosa pine
mortality from bark beetle attack. Oecologia 164, 601–609.
Kilpeläinen, A., Gerendiain, A.Z., Luostarinen, K., Peltola, H., Kellomäki, S., 2007.
Elevated temperature and CO2 concentration effects on xylem anatomy of Scots
pine. Tree Physiol. 27, 1329–1338.
Levanič, T., 1999. Vertical resin ducts in wood of black pine (Pinus nigra Arnold) as a
possible dendroecological variable. Phyton Austria 39, 123–127.
Liang, W., Heinrich, I., Simard, S., Helle, G., Liñán, I.D., Heinken, T., 2013. Climate
signals derived from cell anatomy of Scots pine in NE Germany. Tree Physiol. 33,
833–844.
Loomis, W.E., 1953. Growth and differentiation - an introduction and summary, in:
Loomis, W.E. (Ed.), Growth and Differentiation in Plants. Iowa State College
Press, Iowa, pp. 1–17.
Lorio Jr., P.L., 1986. Growth-differentiation balance: A basis for understanding southern
pine beetle-tree interactions. For. Ecol. Manag. 14, 259–273.
188
Macalady, A.K., Bugmann, H., 2014. Growth-Mortality Relationships in Piñon Pine
(Pinus edulis) during Severe Droughts of the Past Century: Shifting Processes in
Space and Time. PloS One 9, e92770.
Macalady, A.K., Kläy, M., Bugman, H., Gaylord, M.L., Allen, C.D., Swetnam, T.W.,
McDowell, N.G., in preparation. Mortality risk of an aridlands conifer during
severe drought depends on radial growth and investment into defense.
Meko, D.M., 1981. Applications of Box-Jenkins methods of time series analysis to the
reconstruction of drought from tree rings (Ph.D. dissertation). University of
Arizona, Tucson, AZ.
Meko, D.M., Touchan, R., Anchukaitis, K.J., 2011. Seascorr: A MATLAB program for
identifying the seasonal climate signal in an annual tree-ring time series. Comput.
Geosci. 37, 1234–1241.
Münch, E., 1919. Naturwissenschaftliche Grundlagen der Kiefernharz nutzung. Biol
Reich Land- Forstwirtsch. 10, 1–140.
Olano, J.M., Arzac, A., García-Cervigón, A.I., von Arx, G., Rozas, V., 2013. New star on
the stage: amount of ray parenchyma in tree rings shows a link to climate. New
Phytol. 198, 486–495.
Panyushkina, I.P., Hughes, M.K., Vaganov, E.A., Munro, M.A., 2003. Summer
temperature in northeastern Siberia since 1642 reconstructed from tracheid
dimensions and cell numbers of Larix cajanderi. Can. J. For. Res. 33, 1905–1914.
Rasband, W.S., 1997. ImageJ, US National Institutes of Health, Bethesda, Maryland,
USA.
Ray, A.J., Garfin, G.M., Wilder, M., Vásquez-León, M., Lenart, M., Comrie, A.C., 2007.
Applications of monsoon research: Opportunities to inform decision making and
reduce regional vulnerability. J. Clim. 20, 1608–1627.
R Core Team, 2014. R: A language and environment for statistical computing. R
Foundation for Statistical Computing, Vienna, Austria.
Reid, R.W., Watson, J.A., 1966. Sizes, distributions, and numbers of vertical resin ducts
in lodgepole pine. Can. J. Bot. 44, 519–525.
Rigling, A., Brühlhart, H., Bräker, O.U., Forster, T., Schweingruber, F.H., 2003. Effects
of irrigation on diameter growth and vertical resin duct production in Pinus
sylvestris L. on dry sites in the central Alps, Switzerland. For. Ecol. Manag. 175,
285–296.
189
Rodríguez-García, A., López, R., Martín, J.A., Pinillos, F., Gil, L., 2014. Resin yield in
Pinus pinaster is related to tree dendrometry, stand density and tapping-induced
systemic changes in xylem anatomy. For. Ecol. Manag. 313, 47–54.
Romme, W.H., Allen, C.D., Bailey, J.D., Baker, W.L., Bestelmeyer, B.T., Brown, P.M.,
Eisenhart, K.S., Floyd, M.L., Huffman, D.W., Jacobs, B.F., Miller, R.F.,
Muldavin, E.H., Swetnam, T.W., Tausch, R.J., Weisberg, P.J., 2009. Historical
and Modern Disturbance Regimes, Stand Structures, and Landscape Dynamics in
Piñon–Juniper Vegetation of the Western United States. Rangel. Ecol. Manag. 62,
203–222.
Rosner, S., Hannrup, B., 2004. Resin canal traits relevant for constitutive resistance of
Norway spruce against bark beetles: environmental and genetic variability. For.
Ecol. Manag. 200, 77–87.
Seager, R., Ting, M., Held, I., Kushnir, Y., Lu, J., Vecchi, G., Huang, H.-P., Harnik, N.,
Leetmaa, A., Lau, N.-C., Li, C., Velez, J., Naik, N., 2007. Model Projections of
an Imminent Transition to a More Arid Climate in Southwestern North America.
Science 316, 1181–1184.
Sheppard, P.R., Comrie, A.C., Packin, G.D., Angersbach, K., Hughes, M.K., 2002. The
climate of the US Southwest. Clim. Res. 21, 219–238.
Sperry, J.S., Hacke, U.G., Pittermann, J., 2006. Size and function in conifer tracheids and
angiosperm vessels. Am. J. Bot. 93, 1490–1500.
St George, S., Ault, T.R., 2014. The imprint of climate within Northern Hemisphere
trees. Quat. Sci. Rev. 89, 1–4.
Stokes, M.A., Smiley, T.L., 1968. An introduction to tree-ring dating. University of
Arizona Press, Tucson, AZ.
Swetnam, T.W., Baisan, C.H., Caprio, A.C., Brown, P.M., Touchan, R., Anderson, R.S.,
Hallett, D.J., 2009. Multi-millennial fire history of the Giant Forest, Sequoia
National Park, California, USA. Fire Ecol. 5, 120–150.
Todaro, L., Andreu, L., Alessandro, C.M. D’, Gutiérrez, E., Cherubini, P., Saracino, A.,
2007. Response of Pinus leucodermis to climate and anthropogenic activity in the
National Park of Pollino (Basilicata, Southern Italy). Biol. Conserv. 137, 507–
519.
Vaganov, E.A., Anchukaitis, K.J., Evans, M.N., 2011. How Well Understood Are the
Processes that Create Dendroclimatic Records? A Mechanistic Model of the
Climatic Control on Conifer Tree-Ring Growth Dynamics, in: M.K. Hughes,
T.W. Swetnam, H.F. Diaz (Eds.), Dendroclimatology. Springer Netherlands, pp.
37–75.
190
Vaganov, E.A., Hughes, M.K., Shashkin, A.V., 2006. Growth Dynamics of Conifer Tree
Rings: Images of Past and Future Environments. Springer.
von Arx, G., Archer, S.R., Hughes, M.K., 2012. Long-term functional plasticity in plant
hydraulic architecture in response to supplemental moisture. Ann. Bot. 109,
1091–1100.
Werker, E., Fahn, A., 1969. Resin ducts of Pinus halepensis Mill.–Their structure,
development and pattern of arrangement. Bot. J. Linn. Soc. 62, 379–411.
Westbrook, J.W., Resende, M.F.R., Munoz, P., Walker, A.R., Wegrzyn, J.L., Nelson,
C.D., Neale, D.B., Kirst, M., Huber, D.A., Gezan, S.A., Peter, G.F., Davis, J.M.,
2013. Association genetics of oleoresin flow in loblolly pine: discovering genes
and predicting phenotype for improved resistance to bark beetles and bioenergy
potential. New Phytol. 199, 89–100.
Wettstein, J.J., Littell, J.S., Wallace, J.M., Gedalof, Z., 2011. Coherent Region-, Species-,
and Frequency-Dependent Local Climate Signals in Northern Hemisphere TreeRing Widths. J. Clim. 24, 5998–6012.
Wigley, T.M., Briffa, K.R., Jones, P.D., 1984. On the average value of correlated time
series, with applications in dendroclimatology and hydrometeorology. J. Clim.
Appl. Meteorol. 23, 201–213.
Williams, A.P., Allen, C.D., Macalady, A.K., Griffin, D., Woodhouse, C.A., Meko,
D.M., Swetnam, T.W., Rauscher, S.A., Seager, R., Grissino-Mayer, H.D., Dean,
J.S., Cook, E.R., Gangodagamage, C., Cai, M., McDowell, N.G., 2013.
Temperature as a potent driver of regional forest drought stress and tree mortality.
Nat. Clim. Change 3, 292–297.
Wimmer, R., 2002. Wood anatomical features in tree-rings as indicators of environmental
change. Dendrochronologia 20, 21–36.
Wimmer, R., Grabner, M., 1997. Effects of climate on vertical resin duct density and
radial growth of Norway spruce [Picea abies (L.) Karst.]. Trees 11, 271–276.
Wimmer, R., Vetter, R.E., 1999. Tree-ring analysis: biological, methodological and
environmental aspects. CABI Publishing, Berkeley, CA.
Wu, H., Hu, Z., 1997. Comparative anatomy of resin ducts of the Pinaceae. Trees 11,
135–143.
Zamski, E., 1972. Temperature and photoperiodic effects on xylem and vertical resin duct
formation in Pinus halepensis Mill. Isr. J Bot 21, 99–107.
191
Zang, C., Biondi, F., 2013. Dendroclimatic calibration in R: The bootRes package for
response and correlation function analysis. Dendrochronologia 31, 68–74.
Site Name
Sevilleta National Wildlife Refuge
Sevilleta National Wildlife Refuge
White Rock
Bandelier National Monument
Tres Piedras
Dataset
Acryonm
SEV2000
SEV1950
WRK2000
BNM1950
TRP2000
possible.
36.34˚N 105.93˚W
35.76˚N 106.27˚W
35.81˚N 106.24˚W
34.34˚N 106.55˚W
34.34˚N 106.55˚W
Location
2100
1940
1990
2050
2050
Tree
Measured Analysis
Diameter
Interval
Type
(cm)
1926-1955 Primary
1929-1960 Ancillary
20 39 (RW), 38 (RD) 15.4 (5.6) 1982-2005 Ancillary
20 27 (RW), 25 (RD) 9.0 (5.2)
10 20 (RW), 20 (RD) 17.3 (6.0) 1975-2010 Ancillary
20 20 (RW), 20 (RD) 7.0 (3.9)
20 38 (RW), 35 (RD) 11.2 (3.3) 1975-2008 Primary
EleNo.
No. cores
vation
trees
(m)
in good enough shape to prepare the surface well enough to make resin duct measurements, but ring width measurements were still
The number of measured cores differs between radial growth (RW) and resin duct (RD) measurements because a few cores were not
Table 1. Site locations and sample descriptions. Mean tree diameters are under-bark estimates at breast height (SD in parentheses).
Tables
192
EPS
n
Rbar eff
Mean Measurement
Mean Sensitivity
AR1
0.975
19.97
0.66
0.525
0.467
0.310
RW
0.961
19.91
0.551
3.995
0.907
-0.294
NUM
0.781
17.06
0.173
0.028
0.225
-0.092
Size
0.876
19.41
0.268
1.707
0.845
-0.296
DEN
0.898
19.41
0.313
0.043 -0.844 --0.329 --
RelArea
0.921
19.91
0.369
NUM a
---0.388
16.91
0.036
Size a
---0.819
18.74
0.194
DEN a
---0.809
18.50
0.186
RelAreaa
DENa, and RelAreaa), in which resin duct indices have had the dependence of RW indices removed statistically.
chronology (n) are measured on residual index series. Statistics are also reported for adjusted resin duct chronologies (NUMa, Sizea,
(AR1) represent the means of raw core-level measurements. Rbareff, EPS and the average number of trees in each year of the
(RelArea) chronologies at SEV2000 for the period 1975-2008. Mean measurement, mean sensitivity and first order autocorrelation
Table 2. Statistics for ring width (RW), resin duct number (NUM), mean size (Size), density (DEN), and relative area
193
194
Table 3. Pearson’s correlations between ring width (RW) and resin duct number
(NUM), mean size (Size), density (DEN) and relative area (RelArea) chronologies
from SEV2000. Correlations with adjusted chronologies are in bold and on the bottom
left, whereas correlations with non-adjusted chronologies are in regular type and to the
top right. Correlations between adjusted and non-adjusted chronologies of the same type
are in italics.
RW
NUM
Size
DEN
RelArea
RW
1.000
0.410
0.561
0.291
0.289
NUM
0.836
0.665
0.163
0.291
0.289
Size
0.881
0.781
0.840
0.161
0.188
DEN
0.640
0.761
0.522
0.895
0.961
RelArea
0.707
0.867
0.606
0.962
0.851
195
Table 4. Results of multiple stepwise regressions between ring width and adjusted
resin duct chronologies and seasonal precipitation and temperature. Chronologies
from both SEV2000 and SEV50 are included in the regressions. Standardized regression
coefficients are reported, with associated p-values indicated by type facing. P-values
above 0.01 are printed in regular typeface, p-values below 0.01 are italicized, p-values
below 0.001 are in bold, and p-values below 0.0001 are bold and underlined. Seasonal
climate variables are consecutive 3-month sums (precipitation) or averages (temperature),
with possible predictors for annual chronology values starting in the previous year’s
autumn and extending through autumn of the current year. Months included in the
seasonal averages are abbreviated, with previous years’ values indicated by lowercase
lettering (e.g., aso represents the previous year’s August September and October). All
models had p-values under 0.0001, with corresponding model R2 values reported at the
bottom of the table.
Precipitation
aso
ndJ
FMA
MJJ
ASO
Temperature
aso
ndJ
FMA
MJJ
ASO
R2
Adjusted R 2
RW
0.328
0.295
0.412
0.333
-0.17
NUM a
Size a
0.522
0.295
0.412
0.262
-0.336
DEN a
RelAreaa
0.552
-0.292
-0.232
0.645
-0.144
-0.295
-0.165
-0.189
-0.212
0.153
0.201
0.584
0.54
0.406
0.377
0.331
0.2977
0.476
0.441
0.519
0.478
196
Figures Captions
Figure 1. Climatogram for SEV. Box and whisker plots of monthly precipitation (a),
and average monthly temperature (b) for the period 1926-2008. Boxes represent the
interquartile range with the median depicted as a horizontal line within each box.
Whiskers represent the extent of the data that fall above and below 1.5 times the
interquartile range. Black dots are outliers.
Figure 2. Cambial age-related trends in resin duct (RD) attributes. Data are from
measurements pooled across all sites listed in Table 1. When it was feasible, up to sixteen
additional non-consecutive measurements per core, drawn from 4 dry, 4 wet and 8
average years, are included in the plots in order to better represent the younger cambial
ages. The total number of measurements in each plot varies slightly due to differences in
rules for assigning NAs or zeros in years with no resin ducts or no radial growth (see text
for details) (n=4398 for RW, n=2683 for RD NUM, n=2156 for RD Size, n=2497 for RD
DEN and RD RelArea). The black line shows a loess regression (and grey 95%
confidence interval ribbons) fit to the data points.
Figure 3. Tree size-related trends in growth (RW) and resin duct (RD) attributes.
Data are from measurements across all sites listed in Table 1. Tree diameters are underbark estimates at breast height for the year of the last measurement in each site’s
chronology. Each point represents an average of the previous ten years of measurements
in each tree, in order to emphasize the influence of tree size on average tree-ring
characteristics. Black lines show loess regressions (and grey 95% confidence interval
197
ribbons) fit to data points. Linear regressions between RD Size and RD DEN and tree
size are significant at the 0.05 level.
Figure 4. Time series plots of residual chronologies from SEV2000. Individual
residual time series are shown in grey, and biweight mean chronology for each resin duct
(RD) and growth (RW) variable is shown in black. The vertical lines correspond to
marker years in the RW chronology, with 1981, 1989, 2002, 2006 corresponding to
narrow rings (solid lines) and 1986, 1992, 1998 corresponding to wide rings (dashed
lines).
Figure 5. Bootstrapped correlations between SEV2000 ring width (RW) and RD
chronologies (NUM, Size, DEN and RelArea), monthly precipitation (blue bars) and
mean temperature (red bars) for 1975-2008. Red bars represent partial correlations, in
which the influence of precipitation on correlations with temperature has been removed
(see text). Black outlines around bars represent significant correlations and partial
correlations at the 0.05-level, as evaluated using exact bootstrapping. Lowercase and
uppercase letters indicate months of the previous and current growth year, respectively.
Figure 6. Bootstrapped correlations and partial correlations between SEV2000 ring
width (RW) and adjusted resin duct chronologies (NUMa, Sizea, DENa and RelAreaa)
and monthly precipitation (blue bars) and mean temperature (red bars) for 19752008. Red bars represent partial correlations, in which the influence of precipitation on
198
correlations with temperature has been removed (see text). Black outlines around bars
represent significant correlations and partial correlations at the 0.05 level, as evaluated
using exact bootstrapping. Lowercase and uppercase letters indicate months of the
previous and current growth year, respectively.
Figure 7. Pearson’s correlations between RW and adjusted resin duct chronologies
(NUMa, Sizea, DENa and RelAreaa) and seasonal climate variables across sites.
Three-month seasons of accumulated precipitation (Precip) and average temperature
(Temp) are the same as those in Table 4, with lowercase and uppercase letters
representing months from the previous and current growth year, respectively. The
timespan for each site’s chronologies is listed in Table 1.
199
●
●
●
●
Precip. (mm)
150
●
●
●
●
●
●
●
●
●
●
●
●
100
●
●
50
●
●
●
●
●
●
●
●
●
F
M
●
●
●
●
●
●
●
●
●
●
●
●
N
D
N
D
●
●
●
●
●
●
●
0
J
A
M
J
J
A
S
O
●
●
Mean Temp. (Deg C)
●
●
20
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
10
●
●
●
●
●
0
●
●
J
Figure 1.
F
M
A
M
J
J
Month
A
S
O
200
No. Ducts/year
Ring Width (mm)
10
3
2
1
0
100
200
Cambial Age
10
5
0
0
Figure 2.
100
200
Cambial Age
5
0
RD Relative Area (mm2/mm2)
RD Density (No. /mm2/year)
0
RD Size (mm2/year)
15
4
0
100
200
Cambial Age
0.20
0.15
0.10
0.05
0.00
0
100
200
Cambial Age
0.06
0.04
0.02
0.00
0
100
200
Cambial Age
201
2.0
No. Ducts/year
1.0
0.5
0.0
3.5
10
2.5
2.0
1.5
1.0
Figure 3.
2
0
20
3.0
0
4
Tree Diameter (cm)
10
20
Tree Diameter (cm)
RD Relative Area (mm2/mm2)
RD Density (No. /mm2/year)
0
RD Size (mm2/year)
Ring Width (mm)
6
1.5
0
10
20
10
20
Tree Diameter (cm)
0.08
0.06
0.04
0.02
0
Tree Diameter (cm)
0.04
0.03
0.02
0
10
20
Tree Diameter (cm)
RD RelArea Index
RD DEN Index
RD Size Index
RD NUM Index
RW Index
202
4
3
2
1
0
3
2
1
0
−1
1.5
1.0
0.5
0.0
3
2
1
0
3
2
1
0
1980
Figure 4.
1990
2000
2010
203
RW
NUM
0.6
0.3
0.0
−0.3
−0.6
s o n d J FMAM J J A SO s o n d J FMAM J J A SO
Correlations and Partial Correlations
Size
DEN
0.6
0.3
0.0
−0.3
−0.6
s o n d J FMAM J J A SO s o n d J FMAM J J A SO
RelArea
0.6
0.3
0.0
−0.3
−0.6
s o n d J FMAM J J A SO
Month
Figure 5.
204
NUMa
Sizea
0.50
0.25
Correlations and Partial Correlations
0.00
−0.25
s o n d J FMAM J J A SO s o n d J FMAM J J A SO
DENa
RelAreaa
0.50
0.25
0.00
−0.25
s o n d J FMAM J J A SO s o n d J FMAM J J A SO
Month
Figure 6.
205
SEV2000
SEV1950
RW
WRK2000
BNM1950
TRP2000
SEV2000
SEV1950
NUM
WRK2000
BNM1950
TRP2000
Correlation
Coefficients
0.6
SEV2000
Size
Site
SEV1950
WRK2000
BNM1950
0.0
−0.3
TRP2000
−0.6
SEV2000
SEV1950
DEN
WRK2000
BNM1950
TRP2000
SEV2000
RelArea
SEV1950
WRK2000
BNM1950
TRP2000
aso ndJ FMA MJJ ASO aso ndJ FMA MJJ ASO
Precip.Precip.Precip.Precip.Precip. Temp. Temp. Temp. Temp. Temp.
Climate Variable
Figure 7.
0.3
206
Supplemental Material
Table S1. Statistics for ring width (RW), resin duct number (NUM), mean size
(Size), density (DEN), and relative area (RelArea) chronologies at SEV1950 for the
period 1923-1955. Mean measurement, mean sensitivity and first order autocorrelation
(AR1) represent the means of raw core-level measurements. Rbareff, EPS and the average
number of trees in each year of the chronology (n) are measured on residual index series.
Statistics are also reported for adjusted resin duct chronologies (NUMa, Sizea, DENa, and
RelAreaa), in which resin duct indices have had the dependence of RW indices removed
statistically.
SEV1950
Mean Measurement
Mean Sensitivity
AR1
Rbar eff
EPS
n
RW
0.443
0.547
0.248
0.727
0.981
19.00
NUM
2.751
1.005
-0.164
0.51
0.951
18.70
Size
0.023
0.230
-0.153
0.085
0.568
14.09
DEN
1.839
0.911
-0.265
0.221
0.827
16.91
RelArea
0.039 -0.896 --0.256 -0.256
0.854
16.91
NUM a
Size a
---0.227
0.831
16.70
DEN a
---0.017
0.186
12.97
RelAreaa
---0.159
0.745
15.46
0.154
0.736
15.30
207
Table S2. Pearson’s correlations between ring width (RW) and resin duct number
(NUM), mean size (Size), density (DEN) and relative area (RelArea) chronologies
from SEV1950. Correlations with adjusted chronologies are in bold and on the bottom
left, whereas correlations with non-adjusted chronologies are in regular type and to the
top right. Correlations between adjusted and non-adjusted chronologies of the same type
are in italics.
RW
NUM
Size
DEN
RelArea
RW
1.000
0.096
0.439
0.099
0.212
NUM
0.635
0.822
0.429
0.537
0.624
Size
0.780
0.712
0.816
0.267
0.411
DEN
0.373
0.615
0.345
0.942
0.929
RelArea
0.391
0.667
0.393
0.844
0.922
208
Table S3. Results of multiple step-wise regressions between ring width and adjusted
resin duct chronologies and seasonal precipitation and vapor pressure deficit.
Chronologies from both SEV2000 and SEV50 are included in the regressions.
Standardized regression coefficients are reported, with associated p-values indicated by
type facing. P-values above 0.01 are printed in regular typeface, p-values below 0.01 are
italicized, p values below 0.001 are in bold, and p-values below 0.0001 are bold and
underlined. Seasonal climate variables are consecutive 3-month sums (precipitation) or
averages (VPD or vapor pressure deficits), with possible predictors for annual chronology
values starting in the previous year’s autumn and extending through autumn of the
current year. Months included in the seasonal averages are abbreviated, with previous
years’ values indicated by lowercase lettering (e.g., aso represents the previous year’s
August September and October). All models had p-values under 0.0001, with
corresponding model R2 values reported at the bottom of the table.
Precipitation
aso
ndJ
FMA
MJJ
ASO
RW
0.350
0.162
0.225
0.334
-0.148
VPD
aso
ndJ
FMA
MJJ
ASO
-0.219
-0.244
R2
Adjusted R 2
0.617
0.569
NUM a
Size a
0.558
0.295
0.412
0.262
-0.453
DEN a
RelAreaa
0.583
-0.249
-0.269
0.656
-0.347
-0.167
0.209
0.377
0.357
0.331
0.2977
0.448
0.4203
0.507
0.4732
209
Figure S1. Time series plots of residual chronologies from SEV1950. Individual
residual time series are shown in grey, and the biweight mean chronology for each resin
duct (RD) and growth (RW) variable is shown in black. The vertical lines correspond to
marker years in the RW chronology, with 1934, 1943, 1946 and 1951 corresponding to
narrow rings (solid lines) and 1929, 1933, and 1941 corresponding to wide rings (dashed
lines).
4
3
2
1
0
−1
RD Size Index
RD NUM Index
RW Index
2.5
2.0
1.5
1.0
0.5
0.0
−0.5
RD RelArea Index RD DEN Index
1.4
1.2
1.0
0.8
0.6
3
2
1
0
4
3
2
1
0
1925
1935
1945
1955
210
Figure S2. Bootstrapped correlations and partial correlations between SEV1950
ring width (RW) and resin duct chronologies (NUM, Size, DEN and RelArea) and
monthly precipitation (blue bars) and mean temperature (red bars) for 1923-1955.
Red bars represent partial correlations, in which the influence of precipitation on
correlations with temperature has been removed (see text). Black outlines around bars
represent significant correlations and partial correlations at the 0.05 level, as evaluated
using exact bootstrapping. Lowercase and uppercase letters indicate months of the
previous and current growth year, respectively.
211
RW
NUM
0.75
0.50
0.25
0.00
−0.25
−0.50
s o n d J FMAM J J A SO s o n d J FMAM J J A SO
Correlations and Partial Correlations
Size
DEN
0.75
0.50
0.25
0.00
−0.25
−0.50
s o n d J FMAM J J A SO s o n d J FMAM J J A SO
RelArea
0.75
0.50
0.25
0.00
−0.25
−0.50
s o n d J FMAM J J A SO
Month
212
Figure S3. Bootstrapped correlations and partial correlations between SEV1950
ring width (RW) and adjusted resin duct chronologies (NUM, Size, DEN and
RelArea) and monthly precipitation (blue bars) and mean temperature (red bars)
for 1923-1955. Red bars represent partial correlations, in which the influence of
precipitation on correlations with temperature has been removed (see text). Black outlines
around bars represent significant correlations and partial correlations at the 0.05 level, as
evaluated using exact bootstrapping. Lowercase and uppercase letters indicate months of
the previous and current growth year, respectively.
0.8
NUMa
Sizea
Correlations and Partial Correlations
0.4
0.0
−0.4
s o n d J FMAM J J A SO s o n d J FMAM J J A SO
0.8
DENa
RelAreaa
0.4
0.0
−0.4
s o n d J FMAM J J A SO s o n d J FMAM J J A SO
Month
Was this manual useful for you? yes no
Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project

Download PDF

advertisement