User manual | QUANTIFYING STREAMFLOW CHANGE FOLLOWING BARK BEETLE By

QUANTIFYING STREAMFLOW CHANGE FOLLOWING BARK BEETLE
OUTBREAK IN MULTIPLE CENTRAL COLORADO CATCHMENTS
By
Andrew J. Somor
_______________________
Copyright © Andrew J. Somor 2010
A Thesis submitted to the Faculty of the
DEPARTMENT OF HYDROLOGY AND WATER RESOURCES
In Partial Fulfillment of the Requirements
For the Degree of
MASTER OF SCIENCE
WITH A MAJOR IN HYDROLOGY
In the Graduate College
THE UNIVERSITY OF ARIZONA
2010
2
STATEMENT BY AUTHOR
This thesis has been submitted in partial fulfillment of 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.
Brief quotations from this thesis are allowable without special permission,
provided that accurate acknowledgment of source is made. Requests for permission for
extended quotation from or reproduction of this manuscript in whole or in part may be
granted by the copyright holder.
SIGNED: ______Andrew J. Somor______
APPROVAL BY THESIS DIRECTOR
This thesis has been approved on the date shown below:
____________________________________ ____08/05/10____
Dr. Peter A. Troch
Date
Professor of Hydrology & Water Resources
3
ACKNOWLEDGEMENTS
I would like to thank my family and thesis committee members for their support
and contributions to this research. Dr. Peter Troch encouraged thorough and sound
hydrologic methodology and analysis. Dr. Paul Brooks provided guidance on the
particularities of snow-dominated systems and challenged me to think in new and
different ways. Dr. David Breshears imparted the importance of effective and exciting
communication of scientific research. This work would not have been possible without
my family’s patience and encouragement. This research was funded through SAHRA
(Sustainability of semi-Arid Hydrology and Riparian Areas) and the National Science
Foundation. Any opinions, findings, and conclusions or recommendations expressed in
this material are those of the author and do not necessarily reflect the views of SAHRA or
NSF.
4
TABLE OF CONTENTS
LIST OF TABLES ...............................................................................................................7
ABSTRACT.........................................................................................................................8
1. INTRODUCTION ...........................................................................................................9
1.1 Relevance of Present Study .......................................................................................9
1.2 Recent Western US Bark Beetle Epidemic ..............................................................10
1.3 Subalpine Forests and the Catchment Water Budget ...............................................12
1.3.1 Transpiration .....................................................................................................13
1.3.2 Canopy Interception ..........................................................................................15
1.3.3 Snowpack Protection ........................................................................................16
1.4 Catchment Scale Studies of Bark Beetle Outbreak..................................................18
1.4.1 White and Yampa River Catchments ................................................................19
1.4.2 Jack Creek Catchment.......................................................................................21
1.4.3 Conceptual Model of Streamflow Response to Bark Beetle Outbreak.............22
1.5 Thesis Format...........................................................................................................24
2. PRESENT STUDY ........................................................................................................26
2.1 Description of Study Sites .......................................................................................26
2.2 Methodology ............................................................................................................28
2.3 Results ......................................................................................................................30
2.4 Conclusions and Implications ..................................................................................30
2.5 Future Research ........................................................................................................31
REFERENCES ..................................................................................................................32
5
TABLE OF CONTENTS – Continued
APPENDIX A – ELEVATED TEMPERATURES INHIBIT EXPECTED
STREAMFLOW RESPONSE TO TREE MORTALITY IN BEETLE-KILLED
SUBALPINE CATCHMENTS .........................................................................................38
Abstract ..........................................................................................................................40
1. Introduction ................................................................................................................42
2. Methods......................................................................................................................45
2a. Study Site Selection & Characterization ..............................................................45
2b. Streamflow and Climate Data ..............................................................................46
2c. Bark Beetle Outbreak Data ..................................................................................47
2d. Comparing Pre & Post-Outbreak Data ................................................................48
2e. Time-Trend Analysis ...........................................................................................49
3. Results ........................................................................................................................51
3a. Study Catchments Characteristics ........................................................................51
3b. Bark Beetle Outbreak Characteristics ..................................................................53
3c. Pre & Post Beetle Outbreak Streamflow ..............................................................54
3d. Pre & Post Beetle Outbreak Climate ...................................................................55
3e. Time-Trend Analysis ...........................................................................................55
4. Discussion ..................................................................................................................56
4a. Expected Post-Outbreak Streamflow Increase .....................................................56
4b. Post-Outbreak Streamflow Decrease ...................................................................59
4c. Updated Conceptual Model..................................................................................61
5. Conclusions ................................................................................................................62
6
TABLE OF CONTENTS – Continued
References ......................................................................................................................63
APPENDIX B – SUPPLEMENTAL TABLES & FIGURES ...........................................78
APPENDIX C – EVALUATING ASSUMPTIONS OF STATISTICAL TESTS FOR
COMPARING PRE & POST OUTBREAK DATA .........................................................82
1. Introduction ................................................................................................................82
2. Methods......................................................................................................................82
3. Results ........................................................................................................................83
4. Conclusions ................................................................................................................83
7
LIST OF TABLES
Table 1: Summary of Observed Post-Outbreak Streamflow Change ................................22
8
ABSTRACT
Over the last decade, millions of acres of western North American forest have
been reduced to areas of standing dead trees following eruptions in bark beetle
populations. This thesis provides up-to-date information on streamflow response to the
recent bark beetle outbreak in subalpine forests of the Colorado Rockies. Streamflow and
climate measures are evaluated in eight central Colorado catchments with long-term data
records and varying levels of beetle outbreak. No detectable streamflow change is
observed in 7 of 8 highly impacted catchments. A significant reduction in streamflow is
observed in 1 highly impacted catchment and is likely driven by tree mortality and record
warm temperatures. These findings deviate from expected results and have important
implications for vegetation and streamflow change under a warmer climate.
9
1. INTRODUCTION
1.1 Relevance of Present Study
In the arid western US, water is a scarce and valuable resource. Much of the
western population relies upon water which originates on high elevation (subalpine)
forested lands in the form of snow for domestic and irrigation supplies (Bales et al.
2006). Subalpine forests play a central role in snow accumulation, melt, and discharge
through complex ecohydrological interactions (Molotoch et al. 2009), and are therefore a
key component of the water supply system. These forests are not immune to change, as
they are subject to human manipulation and natural disturbance. The dynamic nature of
forest vegetation, and its importance in the subalpine water budget, requires a thorough
understanding of the hydrologic implications of forest change at the catchment scale for
improved water supply management and planning.
Western forests are becoming increasingly stressed through fire suppression,
drought, and warmer temperatures. Over the last decade, these stresses have facilitated
epidemic levels of bark beetle populations in forests of the western US and Canada,
resulting in tree mortality across millions of acres (Allen et al. 2010; Fettig et al. 2007;
Hicke et al. 2006; Jenkins et al. 2008; Raffa et al. 2008). Few studies have investigated
the effects of this type of forest disturbance on streamflow and the catchment water
budget, with only 3 sites examined in published literature (Love 1955; Bethlahmy 1974;
1975; Potts 1984). Results of these studies generally agree, each reporting an increase in
annual streamflow following bark beetle outbreak in subalpine forests. However, the
extensive record of paired catchment experiments in the US and abroad suggest
10
variability in the relationship between streamflow change and forest disturbance which is
largely a function of disturbance severity and climate (Bosch & Hewlett 1982; Brown et
al. 2005; Stednick 1996). The dated nature of existing beetle outbreak studies and
reported trends in temperature, precipitation, and streamflow (Groisman et al. 2004; Mote
et al. 2005; Regonda et al. 2005; Rood et al. 2005) in the mountain west necessitates
evaluation of streamflow change following recent bark beetle outbreaks.
The following sections summarize characteristics of the recent western US bark
beetle epidemic and its potential effects on streamflow timing and magnitude. The
current state of knowledge of the role of subalpine forests in the catchment water budget
is reviewed, and observed effects of bark beetle outbreak on streamflow, and their
underlying mechanisms, are examined.
1.2 Recent Western US Bark Beetle Epidemic
The bark beetle (family Curculionidae: Scolytinae) is a native inhabitant of
western US forests and has been described as the main agent of tree mortality in
coniferous systems (Berryman 1972; Fettig et al. 2007). The insects feed on, and
reproduce within, the water and nutrient transport system of a tree, and induce mortality
directly or through introduction of deadly fungi (Berryman 1972). Infestation of living
trees is visually conspicuous, as water stress causes foliage to turn yellow to red the year
following infestation, and needles fall from the tree crown in subsequent years.
Populations are generally present at low levels, with beetle attacks restricted to largediameter dead or dying trees (Jenkins et al. 2008), and are regulated by host tree
11
resistance, forest heterogeneity, and climate (Raffa et al. 2008). However, under the right
set of conditions, populations of certain beetle species can soar to epidemic levels and
cause widespread tree mortality. Ongoing and simultaneous outbreaks have affected
much of the western US over the last ten to fifteen years, extending from the pinyonjuniper woodlands of the southwest (Breshears et al. 2005) to high elevation pine, spruce,
and fir forests of the Rocky Mountains (Cain & Hayes 2009). In subalpine regions, the
majority of recent beetle induced tree mortality has been the result of increasing
populations of the mountain pine beetle (Dendroctonus ponderosae) and the spruce
beetle (Dendroctonus rufipennis).
The recent bark beetle outbreak is often labeled as unprecedented in terms of area
affected and number of trees killed (e.g. Fettig et al. 2007), though limited bark beetle
records add ambiguity to its historical context (Cain & Hayes 2009). What is becoming
increasingly clear is that recent outbreaks are partially a function of rising temperatures
and drought conditions across the west (Breshears et al. 2005; Lundquist & Bentz 2009).
Warm temperatures and drought act to increase water stress in host trees and increase
susceptibility to attack (Berg et al. 2006; Breshears et al. 2005; McDowell et al. 2008;
Raffa et al. 2008). Additionally, elevated temperatures reduce beetle mortality during the
winter months, and increase beetle growth and reproduction (Bentz et al. 1991; Lundquist
& Bentz 2009). Fire suppression has also been suggested as a driver of recent outbreaks,
as high forest stocking levels further contribute to resource stress in host trees (Fettig et
al. 2007; Jenkins et al. 2008). The response of individual bark beetle-tree species
combinations to the above stresses is variable, with some more sensitive to certain
12
stresses than other (McDowell et al. 2008; Negron et al. 2008). Models of bark beetle
outbreak under continued warming and drought have projected increased outbreak
severity, and migration of beetle populations to higher elevations over the next century
(Hicke et al. 2006; Logan et al. 2003; Williams et al. 2002).
1.3 Subalpine Forests and the Catchment Water Budget
Roughly 53% of the US water supply originates from forested lands (USDA
Forest Service 2007). The influence of forest vegetation on the partitioning of
precipitation among catchment water budget terms is significant and diverse. While this
influence mainly involves plant water uptake for growth and maintenance, and
interception of precipitation by the forest canopy, several other hydrological processes
are affected by forest cover. These include, but are not limited to, surface infiltration
(Neary et al. 2009), runoff redistribution (Breshears 2006), and subsurface macropore
flow through root channels (Beven & Germann 1982). In subalpine systems, effects of
the forest canopy on the surface energy budget are especially important due to the
accumulation of snow at the land surface, and forest canopy structure is a major control
on snow accumulation and melt (Mussleman et al. 2008; Stottlemyer &Troendle 2001;
Veatch et al. 2009). A more detailed discussion of subalpine forest-hydrology
interactions which may be affected by bark beetle induced tree mortality, including
transpiration, canopy interception, and snowpack protection, is provided in the following
sections.
13
1.3.1 Transpiration
Water is a necessary ingredient for life, and in particular, terrestrial vegetation.
Vascular plants use a portion of water taken up through their root systems in the
formation of plant tissues, though a much larger amount is lost to the atmosphere as a byproduct of carbon uptake for photosynthesis through leaf stomata. At the catchment scale,
the absolute magnitude of this transpirative loss is difficult to assess. A strict catchment
water budget approach does not allow for the separation of transpiration versus canopy
and soil evaporation/sublimation. Methods employed in studies of subalpine forest
transpiration over large spatial scales have included, but are not limited to, scaled-up sap
flow measurements and/or transpiration models (Kaufmann 1984; Kaufmann 1985;
Moore et al. 2008). Modeled transpiration in two catchments of the Fraser Experimental
Forest in the Colorado Rockies equaled approximately 40% of annual precipitation
(Kaufmann 1984). Modeled annual transpiration varied considerably among tree species,
with Engelmann spruce (Picea engelmanii) transpiring at a rate 72% greater than that of
lodgepole pine (Pinus contorta), a result of differences in leaf conductance, leaf-air
temperature gradient, and leaf area index (Kaufmann 1985). Modeled transpiration was
negligible between the months of November through March despite the dominance of
evergreen vegetation. Minimal winter transpiration was also observed in measured and
modeled values for a forest stand at nearby Niwot Ridge, CO (Moore et al. 2008), and
implied by no net winter CO2 uptake at a forested Niwot Ridge eddy covariance site
(Monson et al. 2002). In the Niwot Ridge carbon flux study, winter weather conditions
14
were periodically favorable for plant growth, and low soil moisture and/or temperature
were proposed as restrictions on forest growth during the winter months.
An obvious hydrologic effect of bark beetle outbreak is the cessation of
transpiration by beetle-killed trees. The work of Kaufmann (1985) discussed above
implies that the amount of water not transpired following outbreak may be largely
dependent on the bark beetle-host tree species combination. The proportion of this water
which becomes available for streamflow generation is unknown, as no studies have
directly measured soil moisture dynamics and transpiration of surviving vegetation
following beetle outbreak. However, water use in remaining healthy vegetation may be
indirectly assessed through observations of CO2 flux and dendrochronological methods at
outbreak sites. Brown et al. (2010) measured single tree and stand-level carbon uptake at
two mountain pine beetle outbreak sites in British Columbia, Canada. Outbreak sites
were net sinks of carbon during the growing season despite mortality of up to 95% of
canopy trees. Measurements at one study site took place during the first and second years
of beetle outbreak. The site was a greater carbon sink during the second year despite
infestation of 79% of canopy trees. The productivity increase was attributed to increased
growth in healthy canopy trees and understory vegetation. Similar conclusions have been
drawn from tree-ring measurements of surviving trees (Berg et al. 2006; Veblen et al.
1991) and point to increased transpiration by intact vegetation following bark beetle
outbreak.
15
1.3.2 Canopy Interception
Subalpine forests are generally dominated by evergreen tree species, allowing for
interception of precipitation throughout the entire year. The fate of intercepted
precipitation in part depends on its form. For the case of snow, the dominant form of
precipitation in high elevations areas, interception is followed by wind or gravity driven
redistribution, melt and subsequent drip, or sublimation (Pomeroy & Schmidt 1993).
Sublimation of intercepted snow is encouraged by its high surface area and atmospheric
exposure relative to the snowpack (Schmidt & Gluns 1991), and represents a potentially
significant loss of water from a catchment.
Estimation of the magnitude of interception loss through sublimation at the stand
to catchment scale is challenging due to the heterogeneous nature of the forest canopy
and variability of sublimation rates with elevation and aspect (Montesi et al. 2004).
Pomeroy and Schmidt (1993) estimated interception and sublimation from a spruce and
pine forest through the use of fractal geometry and found that 30% of annual snowfall
was lost through sublimation of intercepted snow. Pomeroy et al. (1998) applied a
process-based model to estimate snow interception and sublimation and provide similar
estimates for pine (31% of annual snowfall) and spruce (40% of annual snowfall) stands,
and much lower interception/sublimation in a mixed species evergreen-deciduous stand
(13% of annual snowfall).
Bark beetle outbreak would be expected to have no effect on canopy interception
and sublimation loss until a significant portion of needles fall from beetle-killed trees.
Complete canopy needle loss has been reported to occur 3-5 years after infestation, and is
16
followed by branch loss and tree blowdown (Mitchell & Preisler 1998). No studies have
directly quantified the expected reduction in canopy losses following beetle-induced
changes to canopy structure. Boon (2007) evaluated the effect by comparing snowpack
depth and snow water equivalent (SWE) beneath intact canopy, clearcut, and beetlekilled plots in British Columbia, Canada. Mountain pine beetle outbreak affected all trees
in the beetle-killed plot, with 70% showing at least partial needle drop, 30% with red
foliage, and some blowdown present. Lowest and highest snowpack depth and SWE were
observed in the intact and clearcut plots, respectively, with intermediate values in the
beetle-killed plot. Results parallel those reported for SWE differences in snowpacks
below intact conifer and leafless deciduous canopies, as peak snowpack SWE was found
to be 34-44% higher under the deciduous canopy due to reduced interception (LaMalfa &
Ryle 2008).
1.3.3 Snowpack Protection
As discussed in the above section, interception by the forest canopy can greatly
reduce the amount of snow reaching the land surface. However, once falling snow joins
the surface snowpack, the forest canopy encourages surface accumulation through its
influence on the snowpack energy budget. Energy budget terms most affected by the
presence of forest cover are those involving radiation and wind speed. Under forest
cover, the snowpack receives less solar radiation due to shading, and more longwave
radiation due to canopy emittance, relative to the snowpack of open areas (Boon 2009;
Musselman 2008; Harding & Pomeroy 1996; Veatch et al. 2009). Overall, the reduction
17
of incoming solar radiation dominates and net radiation at the snowpack surface is lower
under forest cover. An additional, often overlooked, factor influencing net radiation at the
snowpack surface is forest litter, which acts to reduce snowpack albedo (i.e. increase
absorption of solar radiation), and the incorporation of forest litter dynamics into
snowmelt models has provided improved predictions (Hardy et al. 2000).
The effects of wind on snowpack sublimation and melt (together referred to as
ablation) relate to its influence on temperature, humidity, and turbulent energy transfer at
the snow surface. Though effects on temperature and humidity gradients are likely
variable and highly dependent on surrounding conditions, higher wind speeds generally
act to increase sensible and latent heat fluxes into/out of the snowpack (Boon 2009).
Wind speeds under forest cover have been observed to be at least 2 times lower than in
open areas, contributing to reduced significance of turbulent fluxes in the snowpack
energy budget of forested areas (Boon 2009; Harding & Pomeroy 1996; Price & Dunne
1976). Other studies have shown that, despite reduced energy input, sublimation from the
below-canopy snowpack can be significant (Molotch et al. 2007), and equaled 20% of
peak snowpack water equivalent for two forested slopes at the Fraser Experimental Forest
(Schmidt et al. 1998).
Reduced canopy density following bark beetle outbreak should increase energy
input to the snowpack for sublimation and melt. Two studies have quantified energy
budget terms and modeled snowpack ablation in intact, beetle-killed, and clearcut plots at
separate locations in British Columbia, Canada (Boon 2007; 2009). In both cases, the
beetle-killed plot experienced higher net solar radiation and lower net longwave radiation
18
at the snowpack surface relative to intact canopy plots, though differences were not as
drastic as those between clearcut and intact canopy plots. Snowpack sublimation (as
measured by latent heat flux) was found to be negligible in intact canopy and beetlekilled stands due to low wind speed. The effect of beetle outbreak on ablation rate, which
relates to streamflow timing, varied. Boon (2007) found similar ablation rates in alive and
beetle-killed plots, with values 50% lower than the ablation rate of the clearcut plot. Boon
(2009) report a much higher ablation rate in the beetle-killed plot relative to the intact
canopy plot due to increased solar radiation reaching the snowpack.
1.4 Catchment Scale Studies of Bark Beetle Outbreak
Predicting hydrological change following bark beetle outbreak at the catchment
scale is complicated by the dependence of the processes discussed in the preceding
sections on canopy heterogeneity, species composition, understory characteristics, and
site microclimate. Further, canopy changes can have competing effects on the water
budget. For example, reduced transpiration by beetle-killed trees can increase
streamflow, while reduced snowpack protection can reduce streamflow through increased
snowpack sublimation. This section reviews previous studies of streamflow response to
bark beetle outbreak to assess the relative importance of changes to the aforementioned
processes at the catchment scale. Additionally, the conceptual model of streamflow
change following bark beetle outbreak derived from these studies is summarized.
19
1.4.1 White and Yampa River Catchments
In 1941, a spruce beetle outbreak began in the White River catchment in western
Colorado which killed overstory trees across 60% of the catchment’s 1880 km2 by 1946
(Love 1955). Streamflow change following this outbreak has been evaluated in multiple
publications (Love 1955, Bethlahmy 1974; 1975). Love (1955) used the pre-outbreak
empirical relationship between annual streamflow in the White River and annual
streamflow in the unimpacted Elk River, located 97 km northeast of the White River
catchment, to assess post-outbreak streamflow change. This provided an estimate of a 31
mm increase in White River annual streamflow during the outbreak period (1941-1946),
and a 60 mm increase following peak outbreak (1947-1951). The increase was attributed
to the combination of reduced transpiration and canopy interception loss from the
catchment, though no other data was collected to verify this.
Bethlahmy (1974) applied covariance analysis to White River and Elk River
annual streamflow data and found similar results. The White River streamflow increase
was lowest during the initial period of outbreak, and reached a maximum during the 5
year period from 1956-1960, 10-14 years after peak beetle outbreak. An additional highmortality, the Yampa River catchment, was investigated using the same method, and a
post-outbreak streamflow increase was also identified. Both catchments showed
variability in the magnitude of the streamflow increase on an individual year basis, and
this was assumed to be related to annual precipitation, with the highest streamflow
increase corresponding to wet years.
20
Streamflow change in the White River and Yampa River was further investigated
by Bethlahmy (1975). The post-outbreak annual streamflow increase over 5-year
intervals ranged from 12-21% in the White River and 11-28% in the Yampa River, with
the highest increase occurring 10-14 years after peak outbreak in both catchments. Postoutbreak monthly streamflow in October was significantly higher than expected in the
Yampa River only, and was assumed to reflect the reduction in growing season water
uptake by beetle-killed trees. The difference in response among the two impacted
catchments was attributed to differences in aspect, with the west-facing White River
catchment experiencing more growing season transpirative and evaporative loss (and
lower October streamflow increase) than the north –facing Yampa River catchment. The
influence of aspect was also evident in differences in changes to January streamflow,
maximum monthly (May or June) streamflow, and instantaneous peak flow following
beetle outbreak. All three streamflow measures were higher than expected following
beetle outbreak in both catchments, with the largest increases occurring on the westfacing White River catchment. The observed streamflow increases were suggested to be
the result of higher snow accumulation due to reduced interception loss. The change in
snowpack accumulation was assumed to be less significant in the Yampa River
catchment due to lower energy input and lower pre-outbreak interception loss. No
changes were identified in the timing of instantaneous peak flow as a result of beetle
outbreak at either site.
21
1.4.2 Jack Creek Catchment
Jack Creek lies among the subalpine forests of southwest Montana and drains a
133 km2 land area. From 1975-1977 the catchment experienced an outbreak of the
mountain pine beetle which killed 35% of mature timber uniformly across the
catchment’s forested area by 1977 (Potts 1984). Only one year of streamflow records
exist prior to the onset of outbreak, eliminating the possibility of applying linear
regression and covariance analysis to determine post-outbreak streamflow change, as in
the White River and Yampa River catchment studies discussed in the previous section.
Instead, Potts (1984) used streamflow data from 4 gaged catchments within 100 km and
double-mass analysis to quantify streamflow response. Results indicate a 15% increase in
annual streamflow over the post-outbreak period (1977-1982). This value was used in
conjunction with monthly streamflow distribution to generate average pre and postoutbreak monthly hydrographs. Hydrographs show increased post-outbreak fall and
winter streamflow, higher early snowmelt (April and May) streamflow, and similar peak
snowmelt (June) streamflow relative to pre-outbreak conditions. The short pre-outbreak
data record prevented statistical analysis of these changes. Streamflow changes were
attributed to reduced transpiration and interception loss following tree mortality and
needle drop. The absence of a change in peak monthly streamflow was attributed to the
increase in early snowmelt season streamflow and earlier snowmelt onset.
22
1.4.3 Conceptual Model of Streamflow Response to Bark Beetle Outbreak
The White River, Yampa River, and Jack Creek studies allow for the development
of a conceptual model of streamflow change following a high-severity bark beetle
outbreak episode in subalpine catchments based on empirical evidence. Expected postoutbreak changes in annual, seasonal, and peak streamflow magnitude and timing are
summarized in Table 1, and mechanisms driving these changes are reviewed below.
Table 1. Summary of Observed Post-Outbreak Streamflow Change
Site
Area
(km2)
Annual
Peak
SM3
Timing
Peak
Timing
Source
+18% +28%2
+27%
N/A
No
change
Bethlahmy
(1975)
+3% +25%*2
+4%*
N/A
No
change
Bethlahmy
(1975)
+15%
No
change
+2-3
wks
No
change
Potts
(1984)
Seas.
Low
Seas.
High
White R.,
CO
1880
+12% +21%
Yampa R.,
CO
N/A
+11% +28%
+3% +16%1
+28%
+39%1
Jack Cr.,
MT
133
+15%
+10%2
* Difference between observed & predicted values not significant at p < 0.05
1
Quantified from October streamflow
2
Quantified from peak monthly (May or June) streamflow
3
Snowmelt Timing
Annual streamflow is expected to increase by 10-30% following bark beetle
outbreak (Bethlahmy 1975; Potts 1984). The annual streamflow increase is driven by
reduced transpiration and snow interception loss due to cessation of transpiration by
beetle-killed trees and reduced canopy density following needle loss. The increase may
not be evident in individual years if annual precipitation is below average and soil
moisture storage requirements are not met. The streamflow increase may peak long after
the onset of beetle outbreak (15-20 years), though this is dependent on characteristics of
surviving vegetation and post-outbreak climate.
23
Late-summer, fall, and winter streamflow may increase following bark beetle
outbreak (Bethlahmy 1975; Potts 1984). Streamflow during these seasons is baseflow
dominated in subalpine catchments. If present, the seasonal low-flow increase is driven
by increased groundwater recharge due to reduced transpiration and snow interception
losses. The seasonal low-flow increase may not be observed in catchments with high
solar energy input (south and west facing catchments) due to higher growing season
transpiration by surviving vegetation and bare soil evaporation.
Snowmelt season streamflow is expected to increase following bark beetle
outbreak (Bethlahmy 1975; Potts 1984). Like annual and seasonal low-flow increases, the
increase in snowmelt season streamflow is driven by reduced transpiration and snow
interception losses. Reduced interception increases the amount of water which melts and
enters the soil, and reduced transpiration allows more of this water to enter to stream
relative to intact canopy conditions.
Peak monthly and instantaneous streamflow may increase following bark beetle
outbreak, though this increase is variable (Bethlahmy 1975; Potts 1984). Increased postoutbreak peak flow is encouraged by increased snowpack accumulation and reduced
transpiration of melt water. Whether this results in a sustained streamflow increase over
the entire snowmelt period or a “flashy” peak flow response is in part a function of
catchment aspect and snowmelt season climate. If snowmelt is delayed by climatological
conditions in south and west facing catchments, the combination of a rapid increase in
temperature and high solar energy input will produce a flashy response and peak
streamflow will increase.
24
Streamflow timing, as measured by the day of snowmelt onset and peak flow,
may be affected by bark beetle outbreak. The timing of snowmelt onset (determined from
a sustained streamflow increase) can occur 2-3 weeks earlier following outbreak if
supported by climatological conditions due to increased energy input to the snowpack.
Peak flow timing has not been found to be affected by bark beetle outbreak since any
increase in snowpack ablation rate is balanced by an increase in snowpack volume.
In the present conceptual model, the dominant effect of bark beetle outbreak on
the catchment water budget is reduced evapotranspiration (ET) and sublimation losses.
Any increase in snowpack sublimation or transpiration by remaining vegetation is
outweighed by reduced interception and cessation of transpiration by beetle-killed trees.
This model agrees with conclusions drawn from plot-scale snowpack data (Boon 2007)
and studies of streamflow response to forest harvest treatments in subalpine systems
(Troendle & King 1985; 1987), and is compared to observed streamflow response in 8
catchments affected by the recent bark beetle epidemic in central Colorado, USA, in the
study presented in Chapter 2.
1.5 Thesis Format
The format of this thesis follows that defined by the Manual for Theses and
Dissertations of the University of Arizona Graduate College. The preceding chapter
serves as a summary of the research topic, its relevance to science and society, and a
review of related published studies. The second chapter describes the methodology of the
current study and provides a brief review of results and implications. These are further
25
explored in Appendix A, a scientific manuscript prepared for submission to Water
Resources Research. Appendix B contains additional tables and figures summarizing data
and results, and Appendix C presents results of assessment of probability distribution and
autocorrelation analysis of hydrological and climate time series data used in the current
study.
\
26
2. PRESENT STUDY
This chapter provides a brief overview of the objectives, methods, results, and
conclusions of the present study. These subjects are covered in greater detail in Appendix
A.
2.1 Description of Study Sites
The study presented in this thesis was undertaken in order to answer the following
question: How have recent bark beetle outbreaks affected streamflow magnitude and
timing in impacted subalpine catchments? This required the collection and analysis of
long-term streamflow, climate, and bark beetle induced tree mortality data, and study
sites were selected according to data availability. Datasets made available for public use
by government agencies and academic institutions whose focus is on land and water
resource issues were reviewed and 8 catchments in central Colorado, USA, were
identified for study. Study catchments met the following criteria: 1) 20+ year record of
daily streamflow through water year 2008 or 2009, 2) no upstream flow diversions or
regulation by lakes/reservoirs, 3) 14+ year record of annual insect and disease induced
tree mortality through 2009, and 4) High forest cover (>50% of catchment area). Figure
A2 details the location of study catchments.
A summary of the physical properties of each study catchment is provided in
Table A1. Individual catchments are hereafter referred to by the three letter code listed in
column 2 of Table A1. All study catchments are located within the boundaries of the
USDA Forest Service National Forest system, and range in size from 15-71 km2. Mean
27
elevation of each catchment ranges from 3100-3300 meters. Dominant forest types in
each catchment are spruce-fir and lodgepole pine, with low amounts of aspen cover
present in some catchments. Four catchments (DAR, MID, BLG, RSS) have west or
south facing slopes over at least 65% of their area. North or east aspect predominates in
two catchments (SFW, WEA).
Mean annual hydrological and climatological data for each study catchment is
summarized in Table A1 and Figure A3. Mean annual precipitation over the streamflow
period of record ranged from approximately 600-800 mm. Mean annual temperature was
below 0°C for all study sites. Analysis of monthly climatological data revealed snowdominated conditions in all catchments, as approximately 65% of annual precipitation
falls in months with mean temperatures below 0.5°C (October – April). The proportion of
annual precipitation which becomes streamflow (annual runoff ratio) ranges from 37% in
KEY to 65% in BLG. Streamflow in all catchments is driven by snowmelt, with 70 –
90% of annual streamflow occurring during the snowmelt months (April-July) despite
only 30% of annual precipitation falling during this time.
Bark beetle outbreak extent in each catchment is illustrated Figures A2 & A5.
During the period of tree mortality record (1995-2009), mountain pine beetle and western
balsam bark beetle populations accounted for 99-100% of insect/disease induced tree
mortality in study catchments, with lodgepole pine and subalpine fir acting as host tree
species. Beetle outbreak pattern in each catchment was variable in time and space. A low
level outbreak episode occurred in select catchments in the early 2000’s and subsided by
2002. A second outbreak episode followed and peaked between the years 2004 and 2009.
28
Outbreak severity during this episode (as measured by impacted area and tree mortality)
varied by catchment. The year of outbreak onset was determined for each catchment as
the year in which impacted area exceeded 20% of the total catchment area. The earliest
beetle outbreak occurred in DAR and SFW (2004), while outbreak in KEY and RSS
lagged by several years. In all catchments, significant tree mortality was observed on at
least 35% of the catchment.
2.2 Methodology
Streamflow and climate data was obtained from the USGS National Water
Information System (http://waterdata.usgs.gov/nwis) and the PRISM Climate Group of
Oregon State University (www.prismclimate.org), respectively. The effect of bark beetle
outbreak on streamflow was evaluated through two methods: 1) comparison of pre and
post-outbreak data, and 2) the time trend analysis technique (Zhao et al. 2010). For each
water year, values of annual, seasonal, and peak streamflow were computed to evaluate
changes in streamflow magnitude. Changes in streamflow timing were determined from
calculated values of the date of the streamflow sprig pulse onset and the annual
hydrograph center of mass (Cayan et al. 2001; Stewart et al. 2005). Statistical analysis of
differences between pre and post-outbreak data was conducted using the non-parametric
Mann-Whitney U test. Similar analysis was applied to values of annual precipitation,
snowfall, seasonal distribution of precipitation, and minimum, maximum, and mean
temperature. Differences between pre and post-outbreak streamflow data were assessed
relative to changes in climatological conditions during the post-outbreak period.
29
The time trend analysis method represents a more rigorous method for
determining the effect of bark beetle outbreak on streamflow, though it is limited to
analysis of annual streamflow. Time trend analysis involves 3 main steps:
-1) Quantify the empirical relationship between pre-outbreak streamflow and
climate through multiple linear regression
-2) Use the empirical equation and post-disturbance climate data to predict postoutbreak streamflow
-3) Compare predicted and observed annual streamflow, any significant difference
is assumed to be the result of the bark beetle outbreak.
Here, the empirical relationship between annual streamflow (Q), annual precipitation (P),
and annual mean temperature (T) was determined from multiple linear regression. The
regression equation had the form
Q = a + bP + cT
where a, b, and c are fitted coefficients. Regression was performed on each catchment
individually using data from the start of the streamflow record to 5 years prior to the
onset of bark beetle outbreak, hereafter referred to as the calibration period. The 5 year
period between the calibration and post-outbreak periods was used to evaluate the
accuracy of regression equation (model) predictions, and is hereafter referred to as the
30
evaluation period. Evaluation and post-outbreak predicted and observed annual
streamflow values were compared using the paired t-test.
2.3 Results
Bark beetle outbreak did not produce a detectable increase in annual, seasonal, or
peak streamflow magnitude, or cause a significant shift in streamflow timing, in any
catchment (with the possible exception of SFW and BLG winter streamflow). A
significant reduction in annual runoff ratio and peak flow was observed in the highmortality DAR site following beetle outbreak. Annual streamflow in DAR was
significantly over-predicted during the post-outbreak period. The post-outbreak period
was characterized by wetter conditions in most catchments and warmer conditions in all
study sites.
2.4 Conclusions & Implications
In the conceptual model of streamflow response to bark beetle outbreak discussed
in Chapter 1, processes which drive ET and sublimation loss are diminished following
tree mortality, and annual streamflow is augmented during periods of average or above
average precipitation. Post-outbreak streamflow data and climatological conditions in the
study catchments indicate this conceptual model is incomplete. The absence of a postoutbreak streamflow increase in all catchments suggests that elevated temperatures
dampen or delay this increase. Previous studies have found minimal streamflow response
to beetle outbreak (Bethlahmy 1974) and forest harvest treatments (Troendle & King
31
1985) when annual precipitation is below average. The presence of wet post-outbreak
conditions in most catchments suggests streamflow response may be better predicted by
the annual ratio of water input to the catchment to energy available for ET/sublimation
(annual precipitation: annual potential ET, the aridity index). Furthermore, the apparent
streamflow reduction in DAR indicates that the combination of warm temperatures and
tree mortality can enhance processes driving ET/sublimation loss following outbreak.
The increase in ET/sublimation loss may be a result of increased water use by surviving
vegetation or increased sublimation of the snowpack. These findings reflect the potential
for complex and previously unobserved climate-vegetation-hydrology feedbacks under
temperature regimes predicted by global climate change models.
2.5 Future Research
Results of this study call for a thorough examination of water and energy fluxes in
beetle-killed areas at multiple spatial scales. At the plot scale, observations of soil
moisture and snowpack dynamics, the snowpack energy budget, and water use by
surviving vegetation are needed. At the stand to catchment scale, the spatial variability of
plot-scale measurements must be assessed. Additionally, streamflow change is likely
dependent on the spatial and temporal pattern of tree mortality across a catchment, and
application of advanced remote-sensing methods may provide invaluable information.
32
REFERENCES
Allen, C. D., 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(4), 660-684.
Bales, R. C., N. P. Molotch, T. H. Painter, M. D. Dettinger, R. Rice, and J. Dozier (2006),
Mountain hydrology of the western United States, Water Resources Research, 42,
W08432.
Bentz, B. J., J. A. Logan, and G. D. Amman (1991), Temperature-dependent
development of the mountain pine-beetle (Coleoptera, Scolytidae) and simulation of its
phenology, Canadian Entomologist, 123(5), 1083-1094.
Berg, E. E., J. D. Henry, C. L. Fastie, A. D. De Volder, and S. M. Matsuoka (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(3), 219-232.
Berryman, A. A. (1972), Resistance of conifers to invasion by bark beetle-fungus
associations, Bioscience, 22(10), 598-602.
Bethlahmy, N. (1974), More streamflow after a bark beetle epidemic, Journal of
Hydrology, 23(3-4), 185-189.
Bethlahmy, N. (1975), A Colorado episode: beetle epidemic, ghost forests, more
streamflow, Northwest Science, 49(2), 95-105.
Beven, K., and P. Germann (1982), Macropores and water-flow in soils, Water Resources
Research, 18(5), 1311-1325.
Boon, S. (2007) Snow accumulation and ablation in a beetle-killed pine stand in Northern
Interior British Columbia, BC Journal of Ecosystems and Management, 8(3), 1–13.
Boon, S. (2009), Snow ablation energy balance in a dead forest stand, Hydrological
Processes, 23(18), 2600-2610.
Bosch, J. M., and J. D. Hewlett (1982), A review of catchment experiments to determine
the effect of vegetation changes on water yield and evapo-transpiration, Journal of
Hydrology, 55(1-4), 3-23.
Breshears, D. D. (2006), The grassland-forest continuum: trends in ecosystem properties
for woody plant mosaics?, Frontiers in Ecology and the Environment, 4(2), 96-104.
33
Breshears, D. D., et al. (2005), Regional vegetation die-off in response to global-changetype drought, Proceedings of the National Academy of Sciences of the United States of
America, 102(42), 15144-15148.
Brown, A. E., L. Zhang, T. A. McMahon, A. W. Western, and R. A. Vertessy (2005), A
review of paired catchment studies for determining changes in water yield resulting from
alterations in vegetation, Journal of Hydrology, 310(1-4), 28-61.
Brown, M., T. A. Black, Z. Nesic, V. N. Foord, D. L. Spittlehouse, A. L. Fredeen, N. J.
Grant, P. J. Burton, and J. A. Trofymow (2010), Impact of mountain pine beetle on the
net ecosystem production of lodgepole pine stands in British Columbia, Agricultural and
Forest Meteorology, 150(2), 254-264.
Cain R.J., and J.L. Hayes (2009), Bark beetle conditions in western forests and formation
of the western bark beetle research group, In: Hayes, J.L., and J.E. Lundquist, comps.
Western Bark Beetle Research Group—a unique collaboration with Forest Health
Protection symposium, Society of American Foresters Conference, 23–28 October 2007,
Portland, OR, Gen. Tech. Rep. PNW-GTR-784, Portland, OR: U.S. Department of
Agriculture, Forest Service, Pacific Northwest Research Station: 1-23.
Cayan, D. R., S. A. Kammerdiener, M. D. Dettinger, J. M. Caprio, and D. H. Peterson
(2001), Changes in the onset of spring in the western United States, Bulletin of the
American Meteorological Society, 82(3), 399-415.
Fettig, C. J., K. D. Klepzig, R. F. Billings, A. S. Munson, T. E. Nebeker, J. F. Negron,
and J. T. Nowak (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(1-3), 24-53.
Groisman, P. Y., R. W. Knight, T. R. Karl, D. R. Easterling, B. M. Sun, and J. H.
Lawrimore (2004), Contemporary changes of the hydrological cycle over the contiguous
United States: Trends derived from in situ observations, Journal of Hydrometeorology,
5(1), 64-85.
Harding, R. J., and J. W. Pomeroy (1996), Energy balance of the winter boreal landscape,
Journal of Climate, 9(11), 2778-2787.
Hardy, J. P., R. Melloh, P. Robinson, and R. Jordan (2000), Incorporating effects of
forest litter in a snow process model, Hydrological Processes, 14(18), 3227-3237.
Hicke, J. A., J. A. Logan, J. Powell, and D. S. Ojima (2006), Changing temperatures
influence suitability for modeled mountain pine beetle (Dendroctonus ponderosae)
outbreaks in the western United States, Journal of Geophysical Research-Biogeosciences,
111(G2).
34
Jenkins, M. J., E. Hebertson, W. Page, and C. A. Jorgensen (2008), Bark beetles, fuels,
fires and implications for forest management in the Intermountain West, Forest Ecology
and Management, 254(1), 16-34.
Kaufmann, M. R. (1984), A canopy model (RM-CWU) for determining transpiration of
subalpine forests. 2. Consumptive water-use in 2 watersheds, Canadian Journal of Forest
Research, 14(2), 227-232.
Kaufmann, M. R. (1985), Annual transpiration in subalpine forests: large differences
among four tree species, Forest Ecology and Management, 13, 235-246.
LaMalfa, E. M., and R. Ryle (2008), Differential snowpack accumulation and water
dynamics in aspen and conifer communities: Implications for water yield and ecosystem
function, Ecosystems, 11(4), 569-581.
Logan, J. A., J. Regniere, and J. A. Powell (2003), Assessing the impacts of global
warming on forest pest dynamics, Frontiers in Ecology and the Environment, 1(3), 130137.
Love, L.D. (1955), The effect on stream flow of the killing of spruce and pine by the
Engelmann spruce beetle, Transactions – American Geophysical Union, 36(1), 113-118.
Lundquist, J.E. and B.J. Bentz (2009), Bark beetles in a changing climate, In: Hayes, J.L.,
and J.E. Lundquist, comps. Western Bark Beetle Research Group—a unique
collaboration with Forest Health Protection symposium, Society of American Foresters
Conference, 23–28 October 2007, Portland, OR, Gen. Tech. Rep. PNW-GTR-784,
Portland, OR: U.S. Department of Agriculture, Forest Service, Pacific Northwest
Research Station: 39-49.
McDowell, 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(4),
719-739.
Mitchell R.G., and H.K. Preisler (1998), Fall rate of lodgepole pine killed by the
mountain pine beetle in central Oregon, Western Journal of Applied Forestry 13, 23-26.
Molotch, N. P., P. D. Blanken, M. W. Williams, A. A. Turnipseed, R. K. Monson, and S.
A. Margulis (2007), Estimating sublimation of intercepted and sub-canopy snow using
eddy covariance systems, Hydrological Processes, 21(12), 1567-1575.
Molotch, N. P., P. D. Brooks, S. P. Burns, M. Litvak, R. K. Monson, J. R. McConnell,
and K. Musselman (2009), Ecohydrological controls on snowmelt partitioning in mixedconifer sub-alpine forests, Ecohydrology, 2(2), 129-142.
35
Monson, R. K., A. A. Turnipseed, J. P. Sparks, P. C. Harley, L. E. Scott-Denton, K.
Sparks, and T. E. Huxman (2002), Carbon sequestration in a high-elevation, subalpine
forest, Global Change Biology, 8(5), 459-478.
Montesi, J., K. Elder, R. A. Schmidt, and R. E. Davis (2004), Sublimation of intercepted
snow within a subalpine forest canopy at two elevations, Journal of Hydrometeorology,
5(5), 763-773.
Moore, D. J. P., J. Hu, W. J. Sacks, D. S. Schimel, and R. K. Monson (2008), Estimating
transpiration and the sensitivity of carbon uptake to water availability in a subalpine
forest using a simple ecosystem process model informed by measured net CO2 and H2O
fluxes, Agricultural and Forest Meteorology, 148(10), 1467-1477.
Mote, P. W., A. F. Hamlet, M. P. Clark, and D. P. Lettenmaier (2005), Declining
mountain snowpack in western north America, Bulletin of the American Meteorological
Society, 86(1), 39-49.
Musselman, K. N., N. P. Molotch, and P. D. Brooks (2008), Effects of vegetation on
snow accumulation and ablation in a mid-latitude sub-alpine forest, Hydrological
Processes, 22(15), 2767-2776.
Neary, D. G., G. G. Ice, and C. R. Jackson (2009), Linkages between forest soils and
water quality and quantity, Forest Ecology and Management, 258(10), 2269-2281.
Negron, J. F., et al. (2008), US Forest Service bark beetle research in the western United
States: Looking toward the future, Journal of Forestry, 106(6), 325-331.
Pomeroy, J.W., and R.A. Schmidt (1993), The use of fractal geometry in modeling
intercepted snow accumulation and sublimation, Proc. 61st Western Snow Conference,
Quebec City, QC, Western Snow Conference, 1-10.
Pomeroy, J. W., J. Parviainen, N. Hedstrom, and D. M. Gray (1998), Coupled modelling
of forest snow interception and sublimation, Hydrological Processes, 12(15), 2317-2337.
Potts, D. F. (1984), Hydrologic impacts of a large-scale mountain pine-beetle
(Dendroctonus-ponderosae Hopkins) epidemic, Water Resources Bulletin, 20(3), 373377.
Price, A. G., and T. Dunne (1976), Energy-balance computations of snowmelt in a
subarctic area, Water Resources Research, 12(4), 686-694.
Raffa, K. F., B. H. Aukema, B. J. Bentz, A. L. Carroll, J. A. Hicke, M. G. Turner, and W.
H. Romme (2008), Cross-scale drivers of natural disturbances prone to anthropogenic
36
amplification: The dynamics of bark beetle eruptions, Bioscience, 58(6), 501-517.
Regonda, S. K., B. Rajagopalan, M. Clark, and J. Pitlick (2005), Seasonal cycle shifts in
hydroclimatology over the western United States, Journal of Climate, 18(2), 372-384.
Rood, S. B., G. M. Samuelson, J. K. Weber, and K. A. Wywrot (2005), Twentiethcentury decline in streamflows from the hydrographic apex of North America, Journal of
Hydrology, 306(1-4), 215-233.
Schmidt, R. A., and D. R. Gluns (1991), Snowfall interception on branches of 3 conifer
species, Canadian Journal of Forest Researche, 21(8), 1262-1269.
Schmidt, R. A., C. A. Troendle, and J. R. Meiman (1998), Sublimation of snowpacks in
subalpine conifer forests, Canadian Journal of Forest Research, 28(4), 501-513.
Stednick, J. D. (1996), Monitoring the effects of timber harvest on annual water yield,
Journal of Hydrology, 176(1-4), 79-95.
Stewart, I. T., D. R. Cayan, and M. D. Dettinger (2005), Changes toward earlier
streamflow timing across western North America, Journal of Climate, 18(8), 1136-1155.
Stottlemyer, R., and C. A. Troendle (2001), Effect of canopy removal on snowpack
quantity and quality, Fraser experimental forest, Colorado, Journal of Hydrology, 245(14), 165-176.
Troendle, C. A., and R. M. King (1985), The effect of timber harvest on the fool creek
watershed, 30 years later, Water Resources Research, 21(12), 1915-1922.
Troendle, C. A., and R. M. King (1987), The effect of partial and clearcutting on
streamflow at deadhorse creek, colorado, Journal of Hydrology, 90(1-2), 145-157.
U.S. Department of Agriculture Forest Service (2007), Assessment of the status and
trends of natural resources from U.S. forest and range lands: 15 key findings, FS-875,
Washington, D.C., 12 pp.
Veatch, W., P. D. Brooks, J. R. Gustafson, and N. P. Molotch (2009), Quantifying the
effects of forest canopy cover on net snow accumulation at a continental, mid-latitude
site, Ecohydrology, 2(2), 115-128.
Veblen, T. T., Hadley, K.S., Reid, M.S., and A.J. Rebertus. The response of subalpine
forests to spruce beetle outbreak in Colorado. Ecology, 72(1), 213-231.
Zhao, F. F., L. Zhang, Z. X. Xu, and D. F. Scott (2010), Evaluation of methods for
estimating the effects of vegetation change and climate variability on streamflow, Water
37
Resources Research, 46, W03505.
38
APPENDIX A:
ELEVATED TEMPERATURES INHIBIT EXPECTED STREAMFLOW
RESPONSE TO TREE MORTALITY IN BEETLE-KILLED SUBALPINE
CATCHMENTS
39
Elevated temperatures inhibit expected streamflow response to tree mortality in
beetle-killed subalpine catchments
Andrew J. Somor, Paul D. Brooks, Peter A. Troch
Department of Hydrology and Water Resources, University of Arizona, Tucson, AZ,
USA
David D. Breshears
School of Natural Resources and the Environment, and Department of Ecology and
Evolutionary Biology, University of Arizona, Tucson, AZ, USA
40
Abstract
Understanding hydrologic response to vegetation change is imperative for longterm management of water resources because future changes to vegetative structure and
function are expected in response climate change. Of particular concern are high levels of
tree mortality, such as those observed following recent bark beetle outbreaks in western
North America, which are projected to become increasingly common. Historical
hydrological studies of bark beetle outbreak have reported a post-outbreak streamflow
increase which is in part dependent on post-outbreak climate. To date, no published
studies have investigated streamflow response to recent outbreaks under contemporary
climate conditions. This paper evaluates streamflow change in 8 subalpine catchments
during a period in which severe bark beetle outbreak and elevated temperatures coincide.
No significant increase in annual or peak streamflow, or annual runoff ratio (Q/P),
relative to pre-outbreak conditions was found in any study catchment. A significant
reduction in peak flow (-38±19%) and annual runoff ratio (-31±12%) was observed in the
catchment with the highest level of tree mortality and warming, and annual streamflow
was consistently lower (-30±15%) than that predicted from the pre-outbreak streamflowclimate relationship. Post-outbreak annual precipitation was average or above average in
all catchments, and pre-outbreak levels of ET and snow sublimation were likely
maintained or exceeded due to warm temperatures despite tree mortality. Post-outbreak
data suggests that the expected streamflow increase following forest disturbance is
delayed or deterred with a warmer temperature regime, and signals increased variability
41
in the relationship between vegetation and hydrologic change under global climate
change scenarios.
42
1. Introduction
Roughly 53% of the US water supply originates from forested lands (USDA
Forest Service 2007). In the arid west, water is drawn from rivers which surge in the
spring and summer as snow beneath high elevation (subalpine) forests melts. These
forests play a central role in regulating the partitioning of snow and snowmelt through
complex ecohydrological interactions (Molotch et al. 2009). During the winter months,
interception and sublimation from the forest canopy prevents an estimated 20-50% of
annual snowfall from reaching the land surface (Montesi et al. 2004). At the sub-canopy
level, shading and protection from wind reduce snowpack sublimation and delay melt
(Harding & Pomeroy 1996; Price & Dunne 1976; Molotch et al. 2009; Musselman et al.
2008). Additional field and modeling studies at the plant to plot scale have documented
the importance of canopy density and geometry, climate, and storm characteristics on
snowpack accumulation and melt (Musselman et al. 2008; Storck et al. 2002; Stottlemyer
& Troendle 2001; Veatch et al. 2009). Following melt and infiltration, a substantial
amount of soil water is taken up by forest vegetation, with transpiration losses estimated
as approximately 40% of annual precipitation in forested catchments of the Colorado
Rockies (Kaufmann 1984).
The forested landscape is a dynamic one, with change driven by human activity
and natural processes. Recently, western US forests have been subject to increased
stresses as a result of warmer temperatures, drought, and fire suppression (Allen et al.
2010). Over the last decade, these stresses have facilitated epidemic-level populations of
bark beetles throughout the forests of the western US and Canada (Breshears et al. 2005;
43
Fettig et al. 2007; Jenkins et al. 2008; Raffa et al. 2008). Bark beetles feed on the water
and nutrient transport system of a tree, inducing mortality directly or through introduction
of deadly fungi (Berryman 1972), and millions of acres of green forest have been reduced
to areas of standing dead trees following recent beetle outbreaks. Though previous
epidemics of the native insects have been documented (Berg et al. 2006), the
simultaneous and expansive nature of recent outbreaks has gained widespread attention,
and has been identified as an indicator of future high-severity forest disturbance under
global climate change (Allen et al. 2010; Logan et al. 2003).
The influence of subalpine forest vegetation on evapotranspiration (ET) and snow
sublimation processes portends catchment water budget change following severe bark
beetle outbreak and tree mortality. However, conflicting potential changes are evident.
For example, reduced transpiration by beetle-killed trees may act to increase streamflow,
while increased snowpack energy input under a beetle-killed canopy may increase
snowpack sublimation and reduce streamflow. Empirical evidence for the net streamflow
response to bark beetle outbreak is provided by studies of streamflow change following a
spruce beetle (Dendroctonus rufipennis) outbreak at two sites in central Colorado in the
1940’s (Love 1955; Bethlahmy 1974; Bethlahmy 1975), and following a mountain pine
beetle (Dendroctonus ponderosa) outbreak at one site in southern Montana in the
1970’s (Potts 1984). In all catchments, annual streamflow values were higher than
expected by an average of 10-30%, and corresponded to increased snowmelt season
streamflow. Post-outbreak increases in seasonal low-flow (fall and winter) and peak
streamflow, and earlier snowmelt timing, are also reported. These changes varied among
44
catchments and in time, and were assumed to be dependent on catchment aspect and postoutbreak climatological conditions, notably precipitation, though limited climate data was
available. Results provide a framework for expected water budget changes following
beetle outbreak, with decreased ET/sublimation losses and increased streamflow
dominating, and reduced transpiration and snow interception by beetle-killed trees
identified as the underlying streamflow-change mechanisms (Figure A1a). This
conclusion agrees with those of higher snowpack accumulation under beetle-killed plots
relative to intact forest plots (Boon 2007; 2009), and higher annual streamflow following
forest harvest treatments at the Fraser Experimental Forest in the Colorado Rockies
(Troendle & King 1985; 1987).
The prospect of increased streamflow following bark beetle outbreak may be
viewed as favorable as western US water supplies continue to dwindle (Rajagopalan
2009). However, despite the general agreement between previous studies, questions
related to streamflow response to recent outbreaks remain, notably those involving
streamflow change under varied climate. Observed and predicted trends in temperature
and precipitation in the mountain west (IPCC 2007; Groisman et al. 2004; Mote et al.
2005; Regonda et al. 2005; Rood et al. 2005) may act to amplify or dampen the expected
streamflow increase documented in historical studies following recent and future bark
beetle outbreak episodes. The objective of this paper is to quantify streamflow change
and climatological conditions following recent bark beetle outbreak in multiple subalpine
catchments, and to investigate the role of climate in regulating expected streamflow
changes.
45
2. Methods
2a. Study Site Selection & Characterization
This study used publicly available data from multiple federal land and water
resource agencies and academic institutions. The region of the central Colorado Rockies
was selected for investigation based on data availability and the severity of recent bark
beetle outbreaks in catchments with a suitable record of streamflow.
Eight catchments were selected for evaluation of bark beetle outbreak impacts
(Figure A2). Study catchments are located at the heart of the recent bark beetle outbreak
in the state of Colorado, USA, in Grand, Summit, and Eagle Counties. Catchment
selection was based on the availability of continuous, long-term (20+ years) streamflow
records, availability of tree mortality data, unimpaired flow status (no diversions or
upstream regulation), and high forest cover. All catchments lie within the USDA Forest
Service (USFS) National Forest System near the North American continental divide.
Catchment boundaries were delineated with the Basin Delineator Tool and spatial
data provided together as part of the National Hydrography Dataset Plus
(http://www.horizon-systems.com/nhdplus/index.php). Catchment elevation and
topographically-derived catchment properties were analyzed from US Geological Survey
(USGS) National Elevation Dataset 30-meter resolution data. Land cover in each
catchment was determined from land cover datasets generated by the USGS coordinated
Southwest Regional Gap Analysis Project (http://fws-nmcfwru.nmsu.edu/swregap). All
spatial data analysis was conducted in ArcMap 9.2.
46
2b. Streamflow and Climate Data
Daily streamflow data for each study catchment was obtained from the USGS
National Water Information System (http://waterdata.usgs.gov/nwis). Streamflow records
begin in most catchments in the mid-1960’s and extend through water year 2008 or 2009.
Stream condition was stated as unimpaired in each site’s most recent USGS Annual
Water Data Report. In all study catchments, at least 84% of reported daily streamflow
values are classified as “good” quality.
Annual (water year, Oct.-Sep.), seasonal, and 7-day peak streamflow values were
generated from daily data in order to evaluate changes in streamflow magnitude
following bark beetle outbreak. The water year was sub-divided into 3 seasons based on
average monthly streamflow magnitude; winter (October – March), snowmelt (April –
July), and fall (August – September). The winter and fall seasons are characterized by
baseflow dominated conditions, while the snowmelt season includes the peak streamflow
months. Changes in streamflow timing were determined from the date of the snowmelt
spring pulse onset (SPO) and annual hydrograph center-of-mass (COM). The timing of
SPO and COM was calculated according to methods described by Stewart et al. (2005)
and Cayan et al. (2001), respectively.
Like previous studies of streamflow effects on bark beetle outbreak, long-term
observations of climatological measures from within catchment boundaries were limited
to non-existent for all study sites. However, recently developed modeled climate datasets
provide a uniform method for estimating catchment-scale climatic conditions which were
unavailable for previous studies. Here, monthly gridded precipitation and
47
minimum/maximum temperature data for the contiguous US were obtained from the
PRISM Climate Group of Oregon State University (www.prismclimate.org). PRISM data
consists of 4.5 km resolution grids of monthly precipitation and temperature generated
from local weather station data and topographic information. Catchment-scale climate
data was calculated from gridded values by overlaying grid cells on a map of catchment
boundaries. An individual cell was selected if any portion it fell within catchment
boundaries. A weighted average was applied to selected grid cell data to determine
catchment-scale values, with weighting determined from the relative amount of
catchment area taken up by each grid cell. Monthly precipitation and minimum,
maximum, and mean temperature were calculated for each catchment using this method
over the entire range of the streamflow record. Monthly data was used to generate annual
precipitation and temperature values. Annual snowfall was estimated as the total amount
of precipitation falling in months with mean temperature below 0.5°C. The difference in
temporal resolution of climate and streamflow data should be noted, as climate measures
were developed from monthly data, while streamflow measures were developed from
daily measurements.
2c. Bark Beetle Outbreak Data
Bark beetle outbreak in each catchment was characterized from USDA Forest
Service insect & disease aerial survey data. Digitized annual aerial survey sketch maps
for the years 1995-2009 were obtained from the USFS Rocky Mountain Region
geospatial library (http://www.fs.fed.us/r2/gis). Aerial survey GIS data consists of annual
48
vector datasets with polygons representing the location of observed current-year tree
mortality and, for each polygon, an estimate of the number of trees killed, the tree
mortality agent, and host tree species. Bark beetle outbreak severity estimates were
supported by visual assessment of tree mortality in 2005 and 2007 USDA National Aerial
Imagery Program aerial photos.
For each year-catchment combination, values of impacted area (km2), currentyear tree mortality (# of trees), and associated mortality agent and host tree species were
extracted from annual aerial survey data. Bark beetle outbreak severity and timing was
quantified from three metrics: 1) cumulative impacted area, 2) cumulative tree mortality,
and 3) year of outbreak onset. Since study catchments vary in size and forest cover, tree
mortality estimates were normalized by catchment forested area to allow for comparisons
between catchments. Cumulative tree mortality and impacted are was considered over
single-year data since high tree mortality occurred during the years leading up to and
following peak mortality in most catchments. The year of outbreak onset was determined
as the first year in which cumulative impacted area exceeded 20% of total catchment
area. The 20% threshold has been reported as a minimum treatment area required for
detecting streamflow response following forest cover change in multiple studies (Brown
et al. 2005).
2d. Comparing Pre & Post-Outbreak Data
Initial evaluation of streamflow change following bark beetle outbreak was
conducted through statistical comparison of pre and post-outbreak streamflow and
49
climate data, where the post-outbreak period is defined as the year of outbreak onset
through the end of the data record. Statistically significant changes in streamflow
measures not expected from changes in post-outbreak climate were interpreted as a
potential bark beetle outbreak effect. For example, a significant increase in post-outbreak
annual streamflow without an accompanying increase in annual precipitation would
suggest that bark beetle outbreak acts to increase annual streamflow.
In all datasets, 2 assumptions underlying standard parametric statistical tests were
evaluated: 1) normal distribution, and 2) sample independence. Data probability
distribution was determined by viewing histograms and by applying the Lilliefors test for
normality (Lilliefors 1967). Sample independence was assessed from autocorrelation and
lag plots. The normality assumption did not hold for all datasets in all catchments. For
consistency, the non-parametric Mann-Whitney U test (Mann & Whitney 1947), which
includes no distribution assumption, was selected for determining significant differences
between pre and post-outbreak datasets, with α = 0.95. Examination of autocorrelation
plots indicated significant (p < 0.05) autocorrelation at lags of 1 and above in some
datasets. Lag plots revealed negligible to weak autocorrelation at these lags and no
corrective procedures to account for autoregressive error were applied.
2e. Time-Trend Analysis
A more rigorous and commonly-used method for quantifying the influence of
forest cover changes on streamflow is the time-trend analysis approach (Bosch & Hewlitt
1982; Zhao 2010). The time-trend analysis technique uses the pre-change relationship
50
between streamflow and climate in a single catchment to predict post-change streamflow
magnitude under intact forest cover. The method involves three main steps. First, an
empirical rainfall-runoff model is developed to predict annual streamflow using preoutbreak streamflow and climatological data. Next, post-outbreak streamflow is predicted
from the empirical model and post-disturbance climatological data. Last, observed (with
bark beetle outbreak) and predicted (without bark beetle outbreak) streamflow values are
compared, and the difference between observed and predicted streamflow values is
designated as the beetle outbreak effect. Here, the pre-outbreak empirical relationship
between annual streamflow (Q), annual precipitation (P), and annual mean temperature
(T) was quantified through multiple linear regression. The regression equation has the
form
Q = a + bP + cT
(1)
where a, b, and c are fitted coefficients. The streamflow-climate relationship was
calibrated to data from the start of the streamflow data record to 5 years prior to the onset
of bark beetle outbreak. The 5 year period between calibration and outbreak onset served
as an evaluation period where the ability of the empirical model to provide accurate preoutbreak streamflow predictions was assessed. Differences between predicted and
observed annual streamflow values for the evaluation and post-outbreak periods were
tested for significance using the paired t-test and α = 0.95. A significant difference
between predicted and observed streamflow during the post-outbreak period not present
during the evaluation period provides strong evidence that the difference is a result of
bark beetle outbreak.
51
Following model calibration, 4 assumptions of linear regression models were
tested for each catchment:
1) Linear relationship between dependent and independent variables
2) Normal distribution of residuals
3) Homoscedasticity of residuals
4) Independence of residuals
Plots of observed vs. predicted annual streamflow confirmed the linear relationship
between annual streamflow, precipitation, and mean temperature. Residual histograms
and the Lilliefors test for normality confirmed normal distribution of residuals. Residual
homoscedasticity was detected from plots of residuals vs. time and residuals vs. predicted
annual streamflow. Residual independence was assessed from autocorrelation and lag
plots. Weak lag-1 autocorrelation was evident in residuals of 1 study catchment (Middle
Creek). The addition of independent and/or dependent variable lag terms to the regression
equation did not improve regression error or remove residual autocorrelation and analysis
of regression output proceeded with the acknowledgement that the observed weak lag-1
autocorrelation may influence results.
3. Results
3a. Study Catchments Characteristics
Study catchments range in size from 15 to 71 km2 and mean elevation spans from
approximately 3200 to 3300 meters (Table A1). For brevity, individual catchments are
hereafter referred to using the 3 letter code listed in column 2 of Table A1. Aspect in 6
52
catchments (DAR, KEY, TUR, BLG, MID, RSS) is predominantly south and west, and is
predominantly north and east in SFW and WEA (Table A1). Forest covers 63-86% of
each catchment (Table A1) and dominant forest types are spruce-fir and lodgepole pine,
with apsen cover present in some catchments at low (<15%) levels. Additional land cover
types present in at least 10% of any catchment are dry tundra and riparian shrubland.
Catchments contain minimal man-made structures or development, though a portion of
KEY lies within the Keystone Ski Resort and a major highway (Interstate-70) runs
through BLG. A review of aerial photos revealed the presence of existing logging roads
in select catchments, though no indication of recent large-scale tree harvest activities
were evident.
Mean annual precipitation in the study catchments ranges from 600-800 mm
(Figure A3). Variability about mean annual values is approximately 20% (coefficient of
variation; CV) in all catchments. Catchments are characterized by long winters, with
mean annual temperatures below 0°C (Table A1) and mean monthly temperatures below
0.5°C from October through April. Precipitation during these months accounts for
approximately 65% of total annual precipitation. As temperatures warm in the late winter,
accumulated snowfall melts and drives streamflow. Streamflow during the snowmelt
months (April through July) makes up 70-90% of total annual streamflow despite only
30% of annual precipitation occurring during this time. Mean annual runoff ratio (the
ratio of annual streamflow to annual precipitation) ranges from 37% in KEY to 65% in
BLG (Figure A3). Catchments with high mean annual runoff ratio (>50%) generally have
lower forest cover (< 75%). The relationship between forest cover and annual runoff ratio
53
is significant (p < 0.01, R2 = 0.78) and illustrates the importance of forest vegetation in
facilitating water loss through evaporation and sublimation at the catchment scale. Interannual variability in annual streamflow is greater than annual precipitation variability
(CV = 30-35%), and points to additional influences on streamflow variability beyond
annual precipitation amount (temperature, snowfall, precipitation seasonality, etc.)
3b. Bark Beetle Outbreak Characteristics
The pattern of bark beetle outbreak in space and time was highly complex and
varied by catchment, though some consistencies were identified. In all study catchments
mountain pine beetle and western balsam bark beetle (Dryocoetes confuses) accounted
for 99-100% of the total number of trees killed from 1995-2009. Mountain pine beetle
was the dominant mortality agent (responsible for at least 80% of trees killed) in 6 study
catchments (DAR, SFW, KEY, BLG, MID, RSS). Host trees of mountain pine beetle and
western balsam bark beetle were lodgepole pine and subalpine fir (Abies lasiocarpa),
respectively.
Peak tree mortality and impacted area occurred in all catchments between the
years 2004 and 2009. A low-severity western balsam bark beetle outbreak episode
occurred in 4 catchments (SFW, WEA, TUR, BLG) from the late 1990’s through early
2000’s, and peaked in 2001. Peak current-year tree mortality during this episode
accounted for 1-14% of total 1995-2009 tree mortality. In contrast, peak tree mortality
during the later (2004-2009) outbreak episode accounted for 26-44% of total 1995-2009
tree mortality. A review of 2005 aerial photo imagery confirmed the low-severity nature
54
of the 2001 outbreak episode and further analysis was restricted to the years 2003-2009,
during which the highest levels of beetle-induced tree mortality were observed in all
catchments. Aerial photos of the DAR study site from 2005 and 2007 (Figure A4)
illustrate the widespread and severe nature of bark beetle outbreak during this period.
The onset of bark beetle outbreak was earliest in DAR and SFW (2004) and more
recent (2006 or 2007) in the remaining study catchments. Impacted area exceeded 30% in
all study catchments, and reached 50% of total catchment area in DAR and MID (Figure
A5). Tree mortality was highest in catchments in which mountain pine beetle was the
dominant mortality agent (DAR, SFW, KEY, BLG, MID, RSS).
3c. Pre & Post Beetle Outbreak Streamflow
No significant difference between pre and post-outbreak data was found for the
majority of annual, seasonal, and peak streamflow measures in all catchments.
Exceptions include DAR annual runoff ratio and peak flow, and SFW and BLG winter
streamflow (Table A2; represented by post outbreak minus pre-outbreak mean). Postoutbreak winter streamflow was above average in both SFW (p = 0.03; mean increase =
19%, standard deviation = 13%) and BLG (p = 0.02; mean increase = 22%, standard
deviation = 11%). Post-outbreak annual runoff ratio values were unexpectedly below preoutbreak (average) values in DAR (p = 0.03; mean decrease = 31%, standard deviation =
12%). Similarly, post-outbreak DAR peak flow was below average (p = 0.001; mean
decrease = 38%, standard deviation = 20%). In most catchments, the timing of the spring
55
pulse onset and hydrograph center of mass was advanced by 1 – 7 days, though these
differences were not statistically significant.
3d. Pre & Post Beetle Outbreak Climate
Post-outbreak climate was wetter and/or warmer relative to average conditions in
all catchments (Table A2). Post-outbreak annual precipitation was above average in 7 of
the 9 study catchments, with significant increases of 15-20% in DAR, SFW and BLG.
Annual snowfall during the post-outbreak period did not significantly differ from preoutbreak conditions in any catchment. A significant change in the seasonal distribution of
precipitation was observed in BLG only, which received a larger proportion of
precipitation during the winter months in post-outbreak years (Table A2).
Time series plots of annual temperature data reveal a period of sustained above
average temperatures which began in the early 2000’s in most catchments and lasted
through the post-outbreak period (Figure A6). Post-outbreak annual mean temperatures
were significantly higher than pre-outbreak values in 4 catchments (DAR, SFW, WEA,
BLG) by 1.0°C - 1.6°C (Table A2). Minimum or maximum annual temperatures were
significantly higher in the 4 remaining site (Table A2).
3e. Time-Trend Analysis
In all catchments the annual streamflow-climate relationship was significant (p <
0.001) and accounted for 60-80% of observed variability in annual streamflow during the
calibration period. Model coefficients (b and c in Equation 1) were positive for annual
56
precipitation and negative for annual mean temperature for all catchments, indicating that
increased precipitation drives increased streamflow while increased temperature acts to
reduce streamflow. Model error, as measured by mean absolute error (MAE), during the
evaluation period is similar to (within 10%) or lower than calibration MAE in 7 of 8
catchments, indicating acceptable model performance during the evaluation period (SFW
was the exception).
Post-outbreak annual streamflow is under-predicted in 3 catchments (WEA,
TUR, MID) and over-predicted in the remaining sites (Figure A7; negative residuals
(over-prediction) indicate beetle induced tree mortality acts to reduce annual streamflow,
and positive residuals (under-prediction) indicate mortality drives increased annual
streamflow). Annual streamflow residuals did not significantly differ from zero for the
evaluation period in any catchment. Post-outbreak residuals significantly differed from
zero in the high-mortality DAR site only (p = 0.02), with DAR annual streamflow
consistently over-predicted during the post-outbreak period. Observed DAR annual
streamflow was on average 147 mm (standard deviation = 99 mm), or 30% (standard
deviation = 15%) lower than expected during the post-outbreak period.
4. Discussion
4a. Expected Post-Outbreak Streamflow Increase
The expected increase in post-outbreak streamflow was observed in SFW and
BLG winter streamflow only (Table A2). An increase in post-outbreak winter low-flow
has been previously observed (Bethlahmy 1975; Potts 1984) and results from increased
57
groundwater recharge during previous water year snowmelt due to reduced transpiration
and increased snow accumulation (Figure A1a). The observed increase may have been
facilitated by bark-beetle induced tree mortality, though SFW and BLG each received
above average annual precipitation during the post-outbreak period (Table A2) and this
alone may have allowed for high groundwater recharge and release during the winter
months. The absence of a significant increase in other streamflow measures in these and
other catchments, and between predicted and observed annual streamflow (Figure A7),
suggest that a tree mortality driven streamflow increase, if present, only occurred during
the cold, non-growing season months in SFW and BLG.
Increased streamflow following bark beetle outbreak and other types of forest
disturbance is highly dependent on precipitation amount (Bethlahmy 1974; Brown et al.
2005), and is dampened if catchment water inputs do not satisfy soil moisture stores. The
absence of a significant increase in annual and peak streamflow in all catchments, and
seasonal streamflow in 6 catchments (as determined from pre/post-outbreak statistical
comparison and time-trend analysis) cannot be explained by low precipitation, as postoutbreak annual precipitation was average to above average in all catchments (Table A2).
Further, no significant decrease in post-outbreak annual snowfall relative to pre-outbreak
conditions, or changes in seasonal precipitation distribution which would discourage a
streamflow increase were found in any catchment.
The influence of temperature on the expected post-outbreak streamflow increase,
while not explicitly discussed in previous studies of streamflow response to bark beetle
outbreak (likely due to a lack of data), is distinct and straightforward. Warm temperatures
58
act to increase ET and snow sublimation losses from the catchment, thereby balancing
reductions in transpiration by beetle-killed trees and snow interception by a lower density
beetle-killed canopy. Warm post-outbreak conditions in study catchments (Figure A6;
Table A2) likely had a dampening effect on the expected post-outbreak streamflow
increase, and post-outbreak changes to the canopy structure and vegetative community
may have further encouraged ET/sublimation losses under warmer temperatures.
Potential changes to physical and biologically-mediated ET/sublimation processes which
may discourage a post-outbreak streamflow increase are reviewed in the following
paragraphs.
Pre-outbreak rates of sublimation and evaporation loss may have been maintained
despite reduced interception through snowpack sublimation and meltwater evaporation.
Needle loss and litter deposition by beetle-killed trees increases energy input to the
snowpack through reduced wind protection, increased shortwave radiation, and reduced
albedo (Winkler et al. 2010). Although plot-scale studies of the snowpack energy budget
in beetle-killed areas do not report significant latent heat flux (Boon 2007; 2009),
spatially-varied snowpack and latent heat flux measurements point to appreciable
snowpack sublimation with low canopy density and interception (Molotch et al. 2007;
Veatch et al. 2009) which is enhanced under warmer conditions (Molotch et al. 2007).
Post-outbreak increases in snowpack sublimation are likely variable across the catchment
and dependent on aspect and microclimate and further investigation of the spatial
variability of the snowpack energy budget in beetle-killed areas is needed.
59
High transpirative loss following bark beetle outbreak may have occurred in study
catchments through augmented water uptake by surviving canopy/sub-canopy trees or
understory vegetation released from competition. Although no studies have directly
monitored transpiration by surviving vegetation, measured increases in carbon uptake at
beetle-killed sites (Brown et al. 2010) and tree-ring growth of surviving trees (Berg et al.
2006; Veblen et al. 1991) point to the possibility of increased transpiration by surviving
vegetation following bark beetle outbreak. Photosynthetic activity in subalpine forest
vegetation has been shown to be highly dependent on seasonal temperatures and moisture
availability (Monson et al. 2002; Molotch et al. 2009; Sacks et al. 2007). Here, the
combination of warmer conditions, and increased precipitation in select catchments, may
have allowed for a sustained high rate of water uptake by surviving vegetation and
inhibited the expected streamflow increase.
4b. Post-Outbreak Streamflow Decrease
Several lines of evidence point to bark beetle outbreak and associated tree
mortality as a contributor to reduced streamflow in DAR. Notably, mean post-outbreak
annual runoff ratio and 7-day peak flow values were significantly below average (Table
A2) despite a significant increase in post-outbreak annual precipitation (Table A2) and no
significant change in annual snowfall or precipitation seasonality. Further, time-trend
analysis indicates that DAR annual streamflow values are significantly lower than
expected (Figure A7) from post-outbreak climate conditions. A comparison of pre and
60
post-outbreak hydrographs (Figure A8) reveals that the DAR post-outbreak streamflow
decrease generally occurs during and after the peak snowmelt month (June).
It should be noted that the average difference between DAR observed and
predicted post-outbreak annual streamflow (-147 mm; -30%) should not be entirely
designated as the “beetle outbreak effect”, but is likely due in part to temperature as well.
A review of the catchment’s temperature time series reveals few calibration period data
points in the post-outbreak temperature range, meaning the model is not “trained” to the
post-outbreak climate (see Figure A6). However, temperatures during the evaluation
period (mean = -0.8°C, standard deviation = 0.73°C) are slightly higher than the postoutbreak period (mean = -1.3°C, standard deviation = 1.0°C), and model error is
significant only under the combination of warmer temperatures and beetle outbreak. The
fact that some predictive ability is lost during the period of unprecedented warming is
problematic for quantification of streamflow response to natural forest disturbance under
extreme climate conditions using traditional, relatively simple, evaluation methods. The
alternative to time-trend analysis, the paired catchment approach (Zhou et al. 2010), is
generally not viable under such circumstances since streamflow data from a nearby
catchment with little/no tree mortality is required. Here, nearby low-mortality catchments
were located at higher elevations, contained low forest cover, and/or contained
streamflow diversions. Analysis of spatial and climate data from a low-mortality and
undeveloped catchment, Halfmoon Creek (USGS ID 07083000; located 80 kilometers
south of Darling Creek) showed that its mean elevation was at least 300 m higher than
study catchments included in this study, that it was only 30% forested, and that it
61
experienced a much larger temperature increase during the post-outbreak period
(+3.9°C), making it unsuitable for a paired catchment study. These issues call for
expanded streamflow monitoring and more sophisticated methods to obtain more
accurate estimates of streamflow response to vegetation change under future climate
change.
4c. Updated Conceptual Model
Few, if any, published studies have reported no change or a reduction in
streamflow as a result of widespread bark beetle outbreak in subalpine catchments, as
observed in the 8 study catchments presented here (with the possible exception of winter
streamflow in SFW and BLG) . Results call for an updated conceptual model of
streamflow response to bark beetle outbreak in which post-outbreak changes to
vegetative structure and function which act to maintain or increase catchment
ET/sublimation losses are potentially significant, especially under warmer conditions
(Figure A1b). An important point to consider is that the expected streamflow increase
may simply be delayed under the post-outbreak climate and outbreak pattern analyzed
here. A review of the time series of observed and predicted annual streamflow values (not
shown) reveals a streamflow increase in the last 1-2 years of WEA, TUR, and MID data
that is poorly predicted. The short record and inherent uncertainty of empirical modeling
make interpretation of this change as a bark beetle outbreak effect tenuous. However, the
general tendency of under-prediction of post-outbreak annual streamflow in these
catchments (Figure A3, positive mean post-outbreak residuals) points to differences
62
among catchments in the relative importance of processes outlined in Figure A1 which
act to increase vs. decrease streamflow. These differences are theorized to be dependent
on post-outbreak temperature, tree mortality location (aspect, proximity to channel),
canopy and subcanopy species diversity, and secondary/understory growth. Evaluation of
these characteristics, and less dramatic streamflow response, may be achieved with more
detailed remote sensing (Hicke & Logan 2009) and ground survey techniques, and
application of advanced ecohydrological modeling, and is beyond the scope of the
present study.
5. Conclusions
Eight catchments experienced tree mortality over 35-50% of their area from 20032009 following mountain pine beetle and western balsam bark beetle outbreak episodes.
The expected increase in streamflow following this level of tree mortality was not
detected in any study catchment, with the possible exception of winter streamflow in two
sites (SFW and BLG). Adequate water input was available for a streamflow increase, as
annual precipitation during the post-outbreak period was average to above average in all
catchments. The period of beetle outbreak was characterized by sustained above average
temperatures in all catchments, and the increase in energy available for ET and snow
sublimation likely inhibited the expected streamflow increase. The catchment
experiencing the highest level of tree mortality (DAR) exhibited a significant reduction in
annual runoff ratio and peak streamflow following outbreak, and post-outbreak annual
63
streamflow was poorly predicted, suggesting beetle induced tree mortality contributed to
the streamflow reduction.
Post-outbreak mean annual temperature was up to 1.6°C higher than the preoutbreak mean. The dataset presented in this paper provides a unique opportunity to gage
streamflow response to forest disturbance near temperatures predicted for the western US
by global climate change models (2-3°C increase over the next century; IPCC 2007). The
absence of detectable and consistent streamflow increase, and apparent streamflow
decrease, in these 8 catchments signify increasing complexity in the relationship between
climate, vegetation, and hydrology under altered climate conditions that is poorly
predicted from existing empirical knowledge. Results call for continued implementation
of catchment-scale vegetation-streamflow change studies and expanded catchment-scale
data collection under varied climatological conditions. Such studies may identify
additional aberrations from expected streamflow response and aid long-term water
management through improved understanding of the climate-vegetation-hydrology
change relationship.
References
Allen, C. D., 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(4), 660-684.
Berg, E. E., J. D. Henry, C. L. Fastie, A. D. De Volder, and S. M. Matsuoka (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(3), 219-232.
Berryman, A. A. (1972), Resistance of conifers to invasion by bark beetle-fungus
64
associations, Bioscience, 22(10), 598-&.
Bethlahmy, N. (1974), More streamflow after a bark beetle epidemic, Journal of
Hydrology, 23(3-4), 185-189.
Bethlahmy, N. (1975), A Colorado episode: beetle epidemic, ghost forests, more
streamflow, Northwest Science, 49(2), 95-105.
Boon, S. (2007) Snow accumulation and ablation in a beetle-killed pine stand in Northern
Interior British Columbia, BC Journal of Ecosystems and Management, 8(3), 1–13.
Boon, S. (2009), Snow ablation energy balance in a dead forest stand, Hydrological
Processes, 23(18), 2600-2610.
Bosch, J. M., and J. D. Hewlett (1982), A review of catchment experiments to determine
the effect of vegetation changes on water yield and evapo-transpiration, Journal of
Hydrology, 55(1-4), 3-23.
Breshears, D. D., et al. (2005), Regional vegetation die-off in response to global-changetype drought, Proceedings of the National Academy of Sciences of the United States of
America, 102(42), 15144-15148.
Brown, A. E., L. Zhang, T. A. McMahon, A. W. Western, and R. A. Vertessy (2005), A
review of paired catchment studies for determining changes in water yield resulting from
alterations in vegetation, Journal of Hydrology, 310(1-4), 28-61.
Brown, M., T. A. Black, Z. Nesic, V. N. Foord, D. L. Spittlehouse, A. L. Fredeen, N. J.
Grant, P. J. Burton, and J. A. Trofymow (2010), Impact of mountain pine beetle on the
net ecosystem production of lodgepole pine stands in British Columbia, Agricultural and
Forest Meteorology, 150(2), 254-264.
Cain R.J., and J.L. Hayes (2009), Bark beetle conditions in western forests and formation
of the western bark beetle research group, In: Hayes, J.L., and J.E. Lundquist, comps.
Western Bark Beetle Research Group—a unique collaboration with Forest Health
Protection symposium, Society of American Foresters Conference, 23–28 October 2007,
Portland, OR, Gen. Tech. Rep. PNW-GTR-784, Portland, OR: U.S. Department of
Agriculture, Forest Service, Pacific Northwest Research Station: 1-23.
Cayan, D. R., S. A. Kammerdiener, M. D. Dettinger, J. M. Caprio, and D. H. Peterson
(2001), Changes in the onset of spring in the western United States, Bulletin of the
American Meteorological Society, 82(3), 399-415.
Fettig, C. J., K. D. Klepzig, R. F. Billings, A. S. Munson, T. E. Nebeker, J. F. Negron,
and J. T. Nowak (2007), The effectiveness of vegetation management practices for
65
prevention and control of bark beetle infestations in coniferous forests of the western and
southern United States, Forest Ecology and Management, 238(1-3), 24-53.
Groisman, P. Y., R. W. Knight, T. R. Karl, D. R. Easterling, B. M. Sun, and J. H.
Lawrimore (2004), Contemporary changes of the hydrological cycle over the contiguous
United States: Trends derived from in situ observations, Journal of Hydrometeorology,
5(1), 64-85.
Harding, R. J., and J. W. Pomeroy (1996), Energy balance of the winter boreal landscape,
Journal of Climate, 9(11), 2778-2787.
Hicke, J. A., and J. Logan (2009), Mapping whitebark pine mortality caused by a
mountain pine beetle outbreak with high spatial resolution satellite imagery, International
Journal of Remote Sensing, 30(17), 4427-4441.
IPCC (2007), Climate change 2007. The physical science basis. Contribution of WG1 to
the Fourth Assessment Report of the Intergovernmental Panel on Climate Change,
Solomon S., Qin D., Manning M., Chen Z., Marquis M.C., Averyt K.B., Tignor M.,
Miller H.L. (eds.). Cambridge University Press: Cambridge; 889.
Jenkins, M. J., E. Hebertson, W. Page, and C. A. Jorgensen (2008), Bark beetles, fuels,
fires and implications for forest management in the Intermountain West, Forest Ecology
and Management, 254(1), 16-34.
Kaufmann, M. R. (1984), A canopy model (rm-cwu) for determining transpiration of
subalpine forests .2. Consumptive water-use in 2 watersheds, Canadian Journal of Forest
Research, 14(2), 227-232.
Lilliefors, H.W. (1967), On kolmogorov-smirnov test for normality with mean and
variance unknown, Journal of the American Statistical Association, 62(318), 399-402.
Logan, J. A., J. Regniere, and J. A. Powell (2003), Assessing the impacts of global
warming on forest pest dynamics, Frontiers in Ecology and the Environment, 1(3), 130137.
Love, L.D. (1955), The effect on stream flow of the killing of spruce and pine by the
Engelmann spruce beetle, Transactions – American Geophysical Union, 36(1), 113-118.
Mann, H.B., and D.R. Whitney (1947), On a Test of Whether one of Two Random
Variables is Stochastically Larger than the Other, The Annals of Mathematical Statistics,
18(1), 50-60.
Mitchell R.G., and H.K. Preisler (1998), Fall rate of lodgepole pine killed by the
mountain pine beetle in central Oregon, Western Journal of Applied Forestry 13, 23-26.
66
Molotch, N. P., P. D. Brooks, S. P. Burns, M. Litvak, R. K. Monson, J. R. McConnell,
and K. Musselman (2009), Ecohydrological controls on snowmelt partitioning in mixedconifer sub-alpine forests, Ecohydrology, 2(2), 129-142.
Monson, R. K., A. A. Turnipseed, J. P. Sparks, P. C. Harley, L. E. Scott-Denton, K.
Sparks, and T. E. Huxman (2002), Carbon sequestration in a high-elevation, subalpine
forest, Global Change Biology, 8(5), 459-478.
Montesi, J., K. Elder, R. A. Schmidt, and R. E. Davis (2004), Sublimation of intercepted
snow within a subalpine forest canopy at two elevations, Journal of Hydrometeorology,
5(5), 763-773.
Mote, P. W., A. F. Hamlet, M. P. Clark, and D. P. Lettenmaier (2005), Declining
mountain snowpack in western north America, Bulletin of the American Meteorological
Society, 86(1), 39-49.
Musselman, K. N., N. P. Molotch, and P. D. Brooks (2008), Effects of vegetation on
snow accumulation and ablation in a mid-latitude sub-alpine forest, Hydrological
Processes, 22(15), 2767-2776.
Potts, D. F. (1984), Hydrologic impacts of a large-scale mountain pine-beetle
(Dendroctonus-Ponderosae Hopkins) epidemic, Water Resources Bulletin, 20(3), 373377.
Powers, J. S., P. Sollins, M. E. Harmon, and J. A. Jones (1999), Plant-pest interactions in
time and space: A Douglas-fir bark beetle outbreak as a case study, Landscape Ecology,
14(2), 105-120.
Price, A. G., and T. Dunne (1976), Energy-balance computations of snowmelt in a
subarctic area, Water Resources Research, 12(4), 686-694.
Raffa, K. F., B. H. Aukema, B. J. Bentz, A. L. Carroll, J. A. Hicke, M. G. Turner, and W.
Romme, H. (2008), Cross-scale drivers of natural disturbances prone to anthropogenic
amplification: The dynamics of bark beetle eruptions, Bioscience, 58(6), 501-517.
Rajagopalan, B., K. Nowak, J. Prairie, M. Hoerling, B. Harding, J. Barsugli, A. Ray, and
B. Udall (2009), Water supply risk on the Colorado River: Can management mitigate?,
Water Resources Research, 45.
Regonda, S. K., B. Rajagopalan, M. Clark, and J. Pitlick (2005), Seasonal cycle shifts in
hydroclimatology over the western United States, Journal of Climate, 18(2), 372-384.
Rood, S. B., G. M. Samuelson, J. K. Weber, and K. A. Wywrot (2005), Twentiethcentury decline in streamflows from the hydrographic apex of North America, Journal of
67
Hydrology, 306(1-4), 215-233.
Sacks, W. J., D. S. Schimel, and R. K. Monson (2007), Coupling between carbon cycling
and climate in a high-elevation, subalpine forest: a model-data fusion analysis,
Oecologia, 151(1), 54-68.
Stednick, J. D. (1996), Monitoring the effects of timber harvest on annual water yield,
Journal of Hydrology, 176(1-4), 79-95.
Stewart, I. T., D. R. Cayan, and M. D. Dettinger (2005), Changes toward earlier
streamflow timing across western North America, Journal of Climate, 18(8), 1136-1155.
Storck, P., D. P. Lettenmaier, and S. M. Bolton (2002), Measurement of snow
interception and canopy effects on snow accumulation and melt in a mountainous
maritime climate, Oregon, United States, Water Resources Research, 38(11).
Stottlemyer, R., and C. A. Troendle (2001), Effect of canopy removal on snowpack
quantity and quality, Fraser experimental forest, Colorado, Journal of Hydrology, 245(14), 165-176.
Troendle, C. A., and R. M. King (1985), The effect of timber harvest on the Fool Creek
watershed, 30 years later, Water Resources Research, 21(12), 1915-1922.
Troendle, C. A., and R. M. King (1987), The effect of partial and clearcutting on
streamflow at Deadhorse creek, Colorado, Journal of Hydrology, 90(1-2), 145-157.
U.S. Department of Agriculture Forest Service (2007), Assessment of the status and
trends of natural resources from U.S. forest and range lands: 15 key findings, FS-875,
Washington, D.C., 12 pp.
Veatch, W., P. D. Brooks, J. R. Gustafson, and N. P. Molotch (2009), Quantifying the
effects of forest canopy cover on net snow accumulation at a continental, mid-latitude
site, Ecohydrology, 2(2), 115-128.
Veblen, T. T., Hadley, K.S., Reid, M.S., and A.J. Rebertus. The response of subalpine
forests to spruce beetle outbreak in Colorado. Ecology, 72(1), 213-231.
Winkler, R., S. Boon, B. Zimonick, and K. Baleshta (2010), Assessing the effects of postpine beetle forest litter on snow albedo, Hydrological Processes, 24(6), 803-812.
Zhao, F. F., L. Zhang, Z. X. Xu, and D. F. Scott (2010), Evaluation of methods for
estimating the effects of vegetation change and climate variability on streamflow, Water
Resources Research, 46, W0350.
68
Table A1. Summary of Catchment Properties
Site
Darling
Creek
South Fork
Williams Fork
Keystone
Gulch
Wearyman
Creek
Turkey
Creek
Black Gore
Creek
Middle
Creek
Red Sandstone
Creek
Code
Area
(km2)
Mean Elev.
(m)
S/W Aspect
(%)
Forest
(%)
Mean Ann. T
(°C)
DAR
23
3335
70
74
-2.7
SFW
71
3343
41
63
-2.6
KEY
24
3303
57
84
-1.8
WEA
25
3301
46
81
-2.4
TUR
62
3270
52
84
-1.9
BLG
33
3259
66
69
-1.8
MID
15
3187
73
86
-0.2
RSS
19
3169
68
85
-0.2
69
Table A2. Streamflow and Climate Variables With Statistically Significant Difference
Between Pre & Post-Outbreak Values1
Site
DAR
SFW
KEY
WE
A
TUR
BLG
MID
RSS
Winter
Q
(mm)
0
+11*
+12
Peak
Q
(mm)
-20*
-3
0
+3
+4
-0.17**
-0.06
+0.01
Annual
P3
(mm)
+132**
+111*
+81
%
Winter
P
+4
+4
+5
Ann.
Tmean
(°C)
+1.6*
+1.6**
+0.8
+0.7* +1.7**
+1.1** +1.0*
+1.1*
-0.6
+0.05
+10
+4
+1.4*
+1.4*
Ann.
RR2
Ann.
Tmin
(°C)
Ann.
Tmax
(°C)
+0.0
+1.2*
+0.1
+0.7
+0.5
-0.2
+1.1**
-0.1
+1.5*
1
Difference between pre & post-outbreak values expressed as post mean minus pre mean
2
RR = Runoff Ratio, 3P = Precipitation,
*Pre and post-outbreak data significantly different at p < 0.05
**Pre and post-outbreak data significantly different at p < 0.01
+2
+9*
+5
+4
+13
-4
-5
-27
+0.04
-0.07
+0.09
-0.04
+45
+139*
-89
-80
+4
+6*
0
0
+1.2
+1.0*
+0.4
+0.7
70
Figure A1. Conceptual model of streamflow change following bark beetle outbreak
expected from previous studies carried out under historical climate (a) and implied by
present-study results under elevated temperatures (b)
71
Figure A2. Map of study catchments with location of forest cover (green) and 2003-2009
beetle outbreak (red)
72
Mean Annual P, Q (mm)
1000
Precipitation
Streamflow
800
600
400
200
0
(0.54)
(0.64)
(0.37)
(0.39)
(0.41)
(0.65)
(0.40)
(0.52)
DAR
SFW
KEY
WEA
TUR
BLG
MID
RSS
Catchment
Figure A3. Mean annual precipitation (P), streamflow (Q), and runoff ratio (in
parentheses) for each study catchment (error bars = standard deviation)
73
Figure A4. Darling Creek study catchment in 2005 (left) and 2007 (right). Orange/red
patches are locations of severe mountain pine beetle outbreak.
74
Impacted Area (%)
100
80
60
40
20
0
DAR
SFW
KEY
WEA
TUR
BLG
MID
RSS
Catchment
Figure A5. Percentage of catchment in which bark beetle induced tree mortality was
observed over the period 2003-2009
2
o
Annual Mean Temperature ( C)
75
0
-2
-4
1965
1970
1975
1980
1985
1990
1995
2000
2005
2010
Year
Figure A6. Mean annual temperature averaged for all study catchments over shared
period of streamflow record
76
Mean Residuals (mm)
150
Evaluation
Post - Outbreak
50
-50
*
-150
DAR
SFW
KEY
WEA
TUR
BLG
MID
RSS
Catchment
Figure A7. Mean evaluation period and post-outbreak annual streamflow residuals
(* indicates difference between observed and predicted values significant at p < 0.05;
error bars = standard deviation)
150
Pre-Outbreak
Post-Outbreak
-
Streamflow (mm month 1)
77
100
50
0
Nov
Jan
Mar
May
Jul
Month
Figure A8. Pre and post beetle outbreak DAR hydrograph
Sep
78
APPENDIX B: SUPPLEMENTAL TABLES & FIGURES
This appendix includes tables and figures which summarize study catchment
properties, pre and post-outbreak streamflow and climate data, and details of the multiple
linear regression annual streamflow model. All catchments are referred to by their 3-letter
code listed in column 2 of Table A1. This information is intended to supplement that
discussed in Appendix A.
Table B1. Study Catchment Location and USGS Gage Information
1
Site
County
Ownership
USGS ID
Streamflow
Record Start
Streamflow
Record End
DAR
SFW
KEY
WEA
TUR
BLG
MID
RSS
Grand
Grand
Summit
Eagle
Eagle
Eagle
Eagle
Eagle
Arapaho & Roosevelt NF1
Arapaho & Roosevelt NF1
White River NF1
White River NF1
White River NF1
White River NF1
White River NF1
White River NF1
09035800
09035900
09047700
09063200
09063400
09066000
09066300
09066400
1966
1966
1958
1965
1964
1964
1965
1964
2009
2009
2009
2008
2008
2009
2009
2008
NF = National Forest
Table B2. Study Catchment Physical Properties and Forest Cover
Site
Min. Elevation
(m)
Max. Elevation
(m)
Mean Slope
(%)
Spruce-Fir
(%)
Lodgepole Pine
(%)
DAR
SFW
KEY
WEA
TUR
BLG
MID
RSS
2723
2716
2805
2758
2719
2781
2476
2805
3860
3998
3786
3700
3700
3865
3765
3769
36
40
34
33
33
31
35
29
58
50
56
65
64
45
52
53
16
11
29
7
9
16
20
29
79
Table B3. Bark Beetle Outbreak Characteristics
Site
DAR
SFW
KEY
WEA
TUR
BLG
MID
RSS
MPB Mortality1
(%)
99
99
83
11
30
97
96
99
WBB Mortality2
(%)
1
0
17
89
69
3
3
1
Outbreak
Onset
2004
2004
2007
2006
2006
2006
2006
2007
Impacted Area
(%)
50
42
49
40
41
44
50
35
Tree Mortality3
(TPH)
31
20
26
8
8
22
17
16
1
MPB mortality refers to the percentage of trees killed by mountain pine beetle from
2003-2009
2
WBB mortality refers to the percentage of trees killed by western balsam bark beetle
from 2003-2009
3
2003-2009 tree mortality expressed as trees killed per forested hectare
Table B4. Difference Between Mean Post and Pre-Outbreak Streamflow Data
1
Site
Annual
Q1
(mm)
Ann.
RR2
Winter
Q
(mm)
Snowmelt
Q
(mm)
Fall
Q
(mm)
Peak
Q
(mm)
SPO3
(days)
COM4
(days)
DAR
SFW
KEY
WEA
TUR
BLG
MID
RSS
-66
+26
+31
+36
+52
+26
+23
-71
-0.17**
-0.06
+0.01
+0.05
+0.04
-0.07
+0.09
-0.04
0
+11*
+12
+3
+2
+9*
+5
+4
-56
+17
+19
+33
+52
+19
+24
-70
-10
-1
0
0
-3
-3
-5
-6
-20*
-3
0
+4
+13
-4
-5
-27
-6
-4
-1
+1
-2
-1
-6
-1
-7
-5
-2
-1
-1
-4
-6
-5
Q = Streamflow, 2RR = Runoff Ratio, 3SPO = Spring Pulse Onset, 4COM = Hydrograph
Center of Mass
*Pre and post-outbreak data significantly different at p < 0.05
**Pre and post-outbreak data significantly different at p < 0.01
80
Table B5. Difference Between Mean Post and Pre-Outbreak Climate Data
Site
Annual
P1
(mm)
Annual
Snow
(mm)
%
Winter
P
%
Snowmelt
P
Winter
Tmean
(°C)
Snowmelt
Tmean
(°C)
Fall
Tmean
(°C)
DAR
SFW
KEY
WEA
TUR
BLG
MID
RSS
+132**
+111*
+81
+10
+45
+139*
-89
-80
+32
+21
+46
-50
-19
+75
-78
-92
+4
+4
+5
+4
+4
+6*
0
0
-3
-3
-2
-6
-6
-5
-3
-4
+1.3**
+1.2**
+0.4
+0.5
+0.4
+0.5
0.0
+0.1
+1.4*
+1.2*
0.0
+1.2
+1.1
+0.9
+1.2
+1.2
+0.5
+0.4
+0.3
+0.2
+0.2
+0.2
+0.4
+1.5*
1
P = Precipitation
*Pre and post-outbreak data significantly different at p < 0.05
**Pre and post-outbreak data significantly different at p < 0.01
Table B6. Summary of Multiple Linear Regression Equation Coefficients, Statistics &
Error
Site
DAR
SFW
KEY
WEA
TUR
BLG
MID
RSS
1
Coefficients1
Statistics
A
b
c
p-value
R2
Calibration
MAE
(mm)
-269
-201
-183
-209
-222
-231
-131
-54
0.87
0.94
0.61
0.61
0.64
0.97
0.55
0.59
-27
-10
-21
-21
-26
-15
-17
-17
1E-07
5E-11
1E-09
1E-10
4E-09
2E-09
1E-07
3E-09
0.65
0.80
0.64
0.75
0.69
0.69
0.62
0.67
53
39
38
42
51
57
58
56
Model Error2
Evaluation Post-Outbreak
MAE
MAE
(mm)
(mm)
55
76
32
39
22
59
41
60
147
79
38
58
54
101
80
26
Coefficients refer to coefficients of Equation 1 in Appendix A (Q = a + bP + cT)
Model error expressed as mean absolute error (MAE)
2
81
p = 0.02
p = 0.03
p = 0.03
p = 0.02
p = 0.02
Figure B1. Predictors of mean post-outbreak annual streamflow residuals with
statistically significant relationship (at p < 0.05)
82
APPENDIX C: EVALUATING ASSUMPTIONS OF STATISTICAL TESTS FOR
COMPARING PRE & POST OUTBREAK DATA
1. Introduction
The study presented in Appendix A used statistical comparisons of pre and post
bark beetle outbreak streamflow and climate data as a means to assess outbreak effects on
streamflow. Two underlying assumptions of parametric statistical tests for comparing 2
groups of samples are normal distribution and sample independence. These tests are not
appropriate for use if samples exhibit non-normal distribution or if the value of one
sample is highly dependent on the value of a previous sample. This appendix reviews
methods used for determining data probability distribution and sample independence in
streamflow and climate time series data and reports on the presence normal distribution
and autocorrelation in datasets.
2. Methods
Data probability distribution was assessed by applying the Lilliefors test for
normality using the MATLAB function ‘lillietest’. The Lilliefors test was applied to all
annual and seasonal streamflow and climate data included in the manuscript in Appendix
A. The test was only applied to pre-outbreak data since the small sample size of postoutbreak data precludes the evaluation of its probability distribution. The distribution of
post-outbreak data was assumed to correspond to that of pre-outbreak data. Normal
distribution was tested at the 95% confidence level.
83
Independence of climate and streamflow data was determined through visual
inspection of autocorrelation and lag plots. Significant autocorrelation (at the 95%
confidence level) indicated by autocorrelation plots was reviewed using lag plots.
3. Results
Results of the Lilliefors test for normality are provided in Table B1 for
streamflow data and Table B2 for climate data. Multiple streamflow and climate datasets
were non-normally distributed.
Results of autocorrelation analysis varied by catchment and variable tested. An
example of a typical autocorrelation plot is provided in Figure B1. Plots contain an
approximately exponential decay in autocorrelation function (ACF) from lags 0-5. A
review of the corresponding partial autocorrelation plot (Figure B2) contains a single
“spike” at lag 1, indicating the presence of autoregressive (AR) error at lag 1. ACF values
are outside 95% confidence intervals at various other lag values. In most cases, lag plots
generated for lags with ACF values outside 95% confidence intervals indicate negligible
autocorrelation and significant scatter (Figure B3). In some cases, weak autocorrelation is
evident (Figure B4).
4. Conclusions
The non-normal distribution of select datasets precludes the use of parametric
statistical tests for comparing pre and post-outbreak values. For consistency, the nonparametric Mann-Whitney U test was applied to all data, as it includes no probability
84
distribution assumption. Autocorrelation and lag plots suggest the presence of minor to
weak autocorrelation at multiple lags in select datasets. Autocorrelation was not
determined to be strong enough to warrant autoregressive error removal methods, and
analysis proceeded under the acknowledgement that dependence in samples may
influence results of statistical testing.
Table C1. Results of Lilliefors Test for Normality For Streamflow Data
(displayed as p-value for testing null hypothesis that data has normal distribution)
Fall
Peak
Annual Annual Winter Snowmelt
SPO
COM
Site
1
2
Q
RR
Q
Q
Q
Q
DAR
0.32
0.33
0.21
0.46
0.01* 0.50
0.50
0.50
SFW
0.05
0.08
0.50
0.22
0.04* 0.50
0.07
0.06
KEY
0.02*
0.50
0.10
0.14
0.005* 0.01* 0.003* 0.001*
WEA
0.37
0.16
0.50
0.50
0.001* 0.31
0.31
0.50
TUR
0.29
0.50
0.50
0.18
0.004* 0.01* 0.10
0.39
0.08
0.31
BLG
0.48
0.50
0.001*
0.50
0.002* 0.50
MID
0.02*
0.39
0.001*
0.02*
0.001* 0.49
0.50
0.43
RSS
0.50
0.43
0.07
0.50
0.001* 0.50 0.04*
0.41
1
2
Q = Streamflow
RR = Runoff Ratio
* indicates p < 0.05 and rejection of null hypothesis at 95% confidence level
85
Table C2. Results of Lilliefors Test for Normality for Climate Data
(displayed as p-value for testing null hypothesis that data has normal distribution)
% Winter
% Snowmelt
Annual
Annual Annual
Site
P
Snow
P
P
Tmean
DAR
0.20
0.50
0.42
0.37
0.05
SFW
0.20
0.50
0.42
0.23
0.02*
KEY
0.14
0.50
0.50
0.10
0.33
WEA 0.09
0.50
0.50
0.50
0.04*
TUR
0.06
0.50
0.50
0.50
0.28
BLG
0.26
0.50
0.50
0.49
0.35
MID
0.08
0.02*
0.50
0.29
0.50
RSS 0.03*
0.01*
0.50
0.50
0.44
* indicates p < 0.05 and rejection of null hypothesis at 95% confidence level
Figure C1. Example of typical autocorrelation plot
86
Figure C2. Example of typical partial autocorrelation plot
Figure C3. Example of typical lag plot exhibiting minor autocorrelation
87
Figure C4. Example of lag plot exhibiting weak autocorrelation

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