. Population Cycles in Small Mammals and

. Population Cycles in Small Mammals  and
Population Cycles in Small Mammals
CHARLES J. KREBS
and
JUDITH
H. MYERS
Institute of Animal Resource Ecology, University of British Columbia,
Vancouver, C a d
268
I. Introduction
.
268
11. Historical Perspective
270
111. Definition of the Problem .
270
A. What Prevents Unlimited Increase?
272
B. What Causes the Cyclic Periodicity?
.
272
c. What Produces Synchrony? .
272
D. What determines the Amplitude of the Cycle? .
273
IV. Population-Density Changes
273
A. Techniques of Estimating Density .
275
B. Do Population Cycles Really Occur?
278
C. Structure of Population Fluctuations in Microtines
279
1. Increase Phase
283
2. Peak Phase
284
3. Decline Phase
289
4. Phsse of Low Numbers
29 1
V. Demographic Machinery
.
291
A. Reproduction
291
1. Litter Size
293
2. Pregnancy Rate
.
294
3. Length of Breeding Season
296
4. Age a t Sexual Maturity
299
5. Sex Ratio
299
6. Summary
300
B. Mortality
300
1. Adult Mortality
.
305
2. Juvenile Mortality
.
310
3. Prenatal Mortality
.
310
4. Summary
31 1
C. Dispersal
314
D. Growth.
320
VI. Hypotheses to Explain Microtine Cycles
320
A. Food
321
1. Selectivity of Microtine Food Habits and Habitats
326
2. The Effect of Microtine Grazing on the Food Supply
3. The Influence of Food Quality and Quantity on Microtine Numbem 331
337
B. Predation
.
347
C. Weather and Synchrony
363
D. Stress Hypothesis
363
E. Behavior
.
373
F. Genetics
267
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
268
CHARLES J. KREBS AND JUDITH H. MYERS
VII. Evolution of Microtine Cycles
VIII. Summary .
.
Acknowledgements
References .
.
.
.
.
.
384
386
389
389
I. INTRODUCTION
Vole and lemming populations undergo fluctuations which comprise
one of the classic unsolved problems of animal ecology. Two facets of
these fluctuations have interested ecologists: their cyclic periodicity
and their amplitude. Many small rodents reach high densities every
three to four years, and the word cycle has been used in a loose sense
to describe the alternating sequence of high and low populations. The
amplitude of fluctuation is great. Fields which were essentially empty
one year may be scarred with microtine runways the next year.
Lemmings which were hard to catch one fall may be swarming everywhere the next spring. Rodent populations thus seem to incorporate
fluctuations (common in many organisms) and a cyclic rhythm longer
than a year (uncommon).
Voles and lemmings have always been an enigma to population
ecologists because they have never seemed to fit into the current
orthodoxy. Population changes in voles and lemmings cannot be
explained or predicted. We cannot apply current theories on population
control to these rodents and this lack of understanding may have three
explanations: (1) these small rodents are poor subjects in which to
study population ecology; (2) ecologists working on these animals
have been particularly inept; or (3) our current views on population
control may be incomplete. Each of these explanations may be supported by a survey of the literature on this problem.
The purpose of this review is to summarize current information on
population cycles in small rodents. We will look first at some general
questions about cycles, then discuss the demographic machinery which
drives the changes in numbers, and finally analyze the current theories
which purport to explain population cycles in rodents.
11. HISTORICAL
PERSPECTIVE
Outbreaks or plagues of voles and mice have been described in the
Old Testament, and Charles Elton has gathered much of this historical
material into his classic book “Voles, Mice and Lemmings” (1942).
Elton describes (p. 3) the response of people to plagues of mice:
The affair runs always along a similar course. Voles multiply. Destruction reigns. There is dismay, followed by outcry, and demands to
POPULATION CYCLES IN SMALL MAMMALS
269
Authority. Authority remembers its experts or appoints some: they
ought to know. The experts advise a Cure. The Cure can be almost
anything: golden mice, holy water from Mecca, a Government Commission, a culture of bacteria, poison, prayers denunciatory or tactful,
a new god, a trap, a Pied Piper. The Cures have only one thing in
common: with a little patience they always work. They have never been
known entirely to fail. Likewise they have never been known to prevent
the next outbreak. For the cycle of abundance and scarcity has a rhythm
of its own, and the Cures are applied just when the plague of voles is
going to abate through its own loss of momentum.
Although outbreaks of voles were known for thousands of years, the
cyclic periodicity of high population levels was not recognized until
the 1920’s. Collett (1895), for example, discussed in detail the natural
history of the Norwegian lemming (Lemmus lemmus), but in concentrating on years of very high density and the “migrations” which accompanied them, he concluded that “prolific years” recurred at
irregular and unpredictable intervals. Hewitt (1921) was one of the
first to quantify the popular idea of cycles in wildlife species in Canada,
and he showed that lynx and red fox populations reached peaks at
regular intervals of nine to ten years while the arctic fox fluctuated
more rapidly with peaks at regular intervals of four years.
I n 1923 Charles Elton read Collett’s book and Hewitt’s book and
realized that the “migration years” of the Norwegian lemming might
be a reflection of a regular fluctuation in populations of these small
rodents. Little was known of population changes in animals in the
1920’5, and the prevailing belief was that populations were stable in
size and that all outbreaks of species were due to man’s interference
with nature (Egerton, 1968). Elton recognized that regular fluctuations
in populations of rodents in arctic regions would challenge the simple
“balance of nature” idea, and would open up a field of research on the
causes of periodic fluctuations and their evolutionary consequences
(Elton, 1924).
If periodic fluctuations in small rodents were to be understood,
detailed population data would be needed. But in 1923 there was not a
single census of a rodent population to show the changes from year to
year, nor was there much information on birth and death rates. Elton
organized at Oxford a group of biologists to study fluctuations, and
this group evolved into the Bureau of Animal Population.
Interest and research on rodent populations has increased greatly
since the 1930’s. The importance of voles as pests of farmers and
orchard growers and their role in the spread of diseases stimulated work
in the U.S.S.R. and in the U.S.A., and we now turn to a review of the
modern work on periodic fluctuations.
270
CHARLES J.
KREBS AND JUDITH IT. MYERS
111. DEFINITION
OF
THE
PROBLEM
The general problem of understanding cyclic fluctuations in small
rodent populations can be subdivided into four specific problems:
A. What prevents unlimited increase in the population?
B. What causes the cyclic periodicity of three to four years?
C. What produces synchrony of populations over large areas?
D. What determines the amplitude of the fluctuation?
A.
WHAT PREVENTS UNLIMITED I N C R E A S E ?
This is a general problem which is not specific to rodents and does not
involve a necessary cyclic periodicity. For any fluctuation in a population, we can arbitrarily recognize four phases (Fig. 1). The increase
phase is adopted as a standard of reference to which we can compare the
YEARS
FIG.1. Schematic diagram of the four phases of the population cycle in small
rodents. The phase of low numbers may not be present in all cycles.
other phases. To determine what prevents continual population
increase we look for differences between the increase phase and the
peak phase which follows it. We might find, for example, a higher
predation mortality in the peak than in the increase. Similarly we look
for differences between the increase and decline phases, then attempt to
discover if the differences described are universal to all population
cycles, and to categorize differences which are universal as necessary or
suficient conditions.
necessary condition: if the population is to increase geometrically,
this condition must be satisfied
suficient condition: if this condition is satisfied, the population
will enter the increase phase
POPULATION CYCLES IN SMALL MAMMAL8
271
For example, amume that the increase phase may occur only when
disease x is not present in the population. The absence of disease x
would thus be t i necessary condition for population increase. However,
if the disease is absent but the females are not in reproductive condition, increase will not ensue and hence the absence of disease x is
not a sufficient condition for population increase. What we are actually
looking for is the set of necessary conditions which together are
sufficient to cause the population to increase. We thus attempt to sort
out necessary factors for further analysis.
We begin by attempting to specify simple conditions but if we are
unable to explain population cycles we will try to specify more complex
conditions. At present no one can specify the necessary conditions for
population growth in any species of vole or lemming. Why do we not
abandon this approach? Can we not be more precise by recognizing
the multiplicity of causes of population cycles? Since there is disagreement about the causes of population cycles, perhaps we should abandon
the search for a universal explanation and be content with more restricted hypotheses for single species in particular communities.
The multiple-factor hypothesis is particularly dangerous as a
methodological argument. If taken at its face value as a vague armchair
theory, the multiple-factor hypothesis is certainly true. The factors
which affect a lemming population in Alaska are certainly different
from those affecting a vole population in Kansas. But if we adopt this
hypothesis as our research strategy, we lose one of the most important
checks on scientific speculation-the testability of hypotheses. Suppose
that we adopt the hypothesis that the absence of disease x is a sufficient
condition for population growth for Microtus ochrogaster in Kansas.
We cannot test this hypothesis on Microtus ochrogaster in Nebraska
because this is a different situation. Nor can we test the hypothesis on
M. pennsylvanicus in New York. If carried to an extreme, we cannot
even test the hypothesis on the next cycle of M. ochrogaster in Kansas
because multiple factors may intervene over time as well as space.
We believe that a universal explanation must be sought for population fluctuations in voles and lemmings until we have evidence that two
or more distinct explanations are required. With this approach we can
test alternative hypotheses on any species in any location.
It must be pointed out that we are not interested simply in a list of
factors which influence rodent populations. Abundant food and cover
will certainly harbor populations capable of reaching higher densities,
and predation can sometimes remove a large portion of the population.
Food, weather, and predation are factors which are acting on all
populations. We accept this generalization, but we wish to ask if these
factors are necessary or sufficient causes of population cycles. Often a
272
CHARLES J. KREBS AND JUDITH H. MYERS
single mechanism hypothesis is criticized simply because observations
have been made in which a number of factors have been shown to
influence populations. This is not a valid criticism of our search for a
single underlying mechanism for microtine cycles. Many environmental
factors affect the average density level of rodent populations, and we
recognize this as a second problem different from the question of what
prevents unlimited increase (Chitty, 1960). Other environmental
factors will affect population density in a sporadic manner, and not
play a necessary role in every cycle. I n any single study of one species
over one fluctuation, the different roles of these factors may be impossible to disentangle. The whole concept of necessary and sufficient
conditions precludes generalizations from single observations.
Most of the research we will summarize here is concerned with the
question of what prevents unlimited increase (although often phrased as
its converse, what causes population declines).
B.
WHAT CAUSES THE CYCLIC PERIODICITY?
This question is logically secondary to the first question, and yet
many biologists have been concerned with the three to four year
periodicity. The cause of the particular periodicity will very much
depend on the driving force behind the population fluctuations. The
periodicity might be imposed by the physical or biotic environment, or
it might be internally generated as some function of generation time
of the rodents.
c. W H A T
PRODUCES SYNCHRONY?
Vole or lemming populations over thousands of square miles may
reach peak numbers in the same year, and this is another aspect of
population cycles that must somehow be explained. Again, this problem
is logically secondary to the first problem, and only when we can explain
the fluctuations will we be able to study the causes of synchrony.
D.
WHAT DETERMINES THE AMPLITUDE O F THE CYCLE?
Some years of peak numbers are much higher than others, and the
same is true of the low points of the cycle. Variations in amplitude have
been especially noted in the Norwegian lemming in which “lemming
years” occur at irregular intervals. Very few attempts have been made
to determine why the amplitude is larger in some years or why it
varies from one habitat to another.
POPULATION CYCLES I N SMALL M A M M A L S
273
IV. POPULATION
DENSITY
CHANGES
A.
TECHNIQUES O F ESTIMATING D E N S I T Y
Progress in defining the phenomenon of population cycles has been
limited by inadequate density data. I n the simplest case we recognize
only two density states: “high” density and “low” density. Next we
can obtain an index of density by the use of sampling with traps,
surveys for runways or fecal pellets, or visual sightings. Much of the
work on small rodents has utilized kill traps of various sorts to provide
an index of population density. Trap catches are a function both of
density and of activity patterns. If voles have large home ranges one
year and small home ranges the next, a trapping index will decrease
even if actual densities are the same in the two years. Individual
trappers vary greatly in the ability to set traps in good locations,
and this factor can add to the variance in trap catches. I n general,
indices of density obtained by trap sampling will show trends of
density changes but cannot be interpreted quantitatively.
Absolute density estimates can also be obtained by removal trapping.
This method was first developed by Leslie and Davis (1939) and independently derived by DeLury (1947). As animals are removed from
an area, the catch per unit of trapping effort will fall off and reach
zero at the point where the whole population has been removed.,If we
assume constant trappability of the whole population and no immigration, we can use linear regression techniques to estimate the size of the
population being trapped. Unfortunately, two serious problems have
plagued this approach (Smith et al., 1971). First, the probability of
capture is not constant for the whole population (Tanaka, 1960).
And second, immigration occurs once the removal trapping begins.
This forces one to try to measure the area depopulated by the kill
traps, an area which may be several times greater than the actual area
occupied by traps (Smith et al., 1971). The area affected by trapping is
difficult to determine in removal studies. A more basic limitation of
this approach is that it destroys the population we should be trying to
study, and consequently mark-and-release techniques have been
utilized for long-term studies.
Mark-and-recapture techniques permit an accurate measurement of
density. Since the pioneering work of Leslie et al. (1953), there has
been available a continuously improving series of statistical techniques
for this estimation problem (Cormack, 1968). The application of markand-recapture techniques requires an assumption of randomness of
capture of marked and unmarked voles. The randomness of capture
assumption has been tested on only a few vole populations, and in no
274
CHARLES J. KREBS AND JUDITH H. MYERS
case has it been shown to be a valid assumption. Leslie et al. (1953)
showed that Microtus agrestis were not sampled randomly between the
marked and unmarked segments of the population. Some voles are
trap-prone and others are trap-shy. The same results were obtained for
M . californicus by Krebs (1966). Tanton (1965, 1969) showed a seasonal
change in probability of capture for Apodemus sylvaticw and Clethrionomys glareolus. Tanaka (1963, 1972) has shown that the probability
of capture is different for unmarked and for marked voles of Microtus
montebelli, Clethrionomys rufocanus, and C . smithi.
One way to provide randomness of capture might be to prebait
animals for several days or weeks before trapping begins. Tanaka
(1970) prebaited C . rufocanus for three days and showed that this
amount of prebaiting increased the probability of capture of unmarked
voles slightly. Andrzejewski et al. (1971) showed that C. glareolus
which were trap-shy were caught more readily in permanent trap sites
than in random trap sites which were not prebaited. Capture was not,
however, at random in either of these studies, and Krebs (1966) found
that continuous prebaiting at permanent trap sites was not sufficient
to provide random sampling in Microtus californicw. Even though
prebaiting does not equalize the probability of capture over all individuals, it may still improve census estimates. We have found that
M. townsendi cannot be live-trapped even at high densities without
prebaiting (Krebs, unpublished). The same problem is found in N.
pennsylvanicus during the summer (Krebs et al., 1969).
If the assumptions of standard capturerecapture analysis cannot be
met in rodent populations, two courses of action are available. First,
recent techniques for mark and recapture estimation with unequal
catchability can be utilized (Marten, 1970; Seber, 1970). The difficulty
is that again some uniformity assumptions must be made (e.g. that an
individual has a fixed probability of capture throughout its life).
No one to date has used these techniques on a long-term field study.
Second, one can attempt to enumerate the population by saturation
live-trapping at frequent intervals and hope that the errors involved are
relatively small. This approach was adopted by Chitty and Phipps
(1966) and has been used by Krebs (1964a, et seq.). If permanent
trapping stations are used, the enumeration approach seems to provide
the best technique for studying population processes in small rodents.
Unfortunately, many workers on small rodent populations do not
appreciate the problems of density estimation, and the literature is
filled with examples of population estimates derived from the Lincoln
Index with no attempt to satisfy the assumptions, examples of indices
of density such as snap-trap catches being interpreted quantitatively,
and sampling techniques applied with no appreciation of sampling
276
POPULATION CYCLES IN SMALL MAMMALS
theory. For some purposes these faults are not critical, but when
information about rates of change in density is required, the proper
techniques should be used.
B.
D O P O P U L A T I O N C Y C L E S R E A L L Y OCCUR?
There are few long-term data on vole and lemming populations and
the longer the time series, the more unreliable the data. Elton (1942)
summarized the bulk of the historical data on voles and lemmings.
Table I gives the peak years for the Norwegian lemming in south
Norway for almost 80 years. Peak years tend t o recur at three- or
four-year intervals. Figure 2 shows fur returns for the arctic fox
TABLEI
Peak years for the Norwegian Lemming i n South Norway, 1862-1938.
(After Elton, 1942)
1862-3
1866
1868-9
1871-2
1876-6
1879-80
1883-4
1887-8
1890-1
189P5
1897
1902-3
1906
1909-10
1918
1920
1922-3
1926-7
1930
1933-4
1938
I"'""'""'""'""'""'""'""'""'""'""'""'
1870
PEAK
YEARS
1875
1880
1885
1890
1895
1900
1905
1
1910
* * * * * + * * * * * *
1915
w
9
1920
*
1925
FIG.2. Fur return statistics for the arctic fox in Ungava District, 1868-1924.
(Data from Elton, 1942, pp. 415-416.)
276
CHARLES J. KREBS AND JUDITH H. MYERS
(Alopex lagopus) in Ungava from 1867 to 1924. Fur returns are unreliable indicators of absolute population changes but do tend to reflect
the observations of trappers and naturalists (Elton, 1942). These data
show a three- or four-year cycle in arctic fox populations, which
follow the abundance of lemmings.
Koshkina (1966) reports data from a standard kill-trap census of
voles in the boreal forest of the Kola Peninsula (Fig. 3). Thirty years of
1935
PEAK
YEARS
*
1940
*
1945
*
1950
*
1955
8
*
1960
*
1965
FIG 3. Autumn population densities in the red-grey vole, Clethrionomys rwfocanws,
from the central Kola Peninsula. (After Koshkina, 1966.)
observations cover seven population cycles with a period of four or five
years between peak numbers. These records comprise one of the longest
runs of quantitative information on vole numbers. Chitty and Chitty
(1962) report population trends in Microtus agrestis from Lake Vyrnwy,
Wales, from 1932-1960 (Table 11). Qualitative assepsment of the phase
of the population cycle was obtained from a mixture of snap-trapping
and live-trapping studies over this 28-year period (except for World
War 11). Peak populations recur at intervals of four years usually,
although three- and five-year cycles were found. Similar observations
have been made on the brown lemming at Barrow, Alaska (Fig. 4).
Many other studies of shorter duration could 'be cited here. There
are 18 genera and 105 species of voles and lemmings (Arata, 1967), and
perhaps only one-fifth of these species has been studied in depth. We
will assume here that the species studied have been representative of
the group, and will draw our conclusions from an incomplete sample.
Populations of voles and lemmings thus fluctuate with a period
between peaks of three to four years usually, although two-, five- and
six-year cycles are not uncommon. We know of no microtine data
277
POPULATION CYCLES IN SMALL MAMMALS
TABLEI1
Population trends among the volea (Microtus agrestis) at Lake Vyrnwy, Wales,
from 1932 to 1960.* P b e of the cycle waa judged from a mixture of map-trap
and live-trap samples. (After Chitty and Chitty, 1962)
Phase of population cycle
Increase
1932
31936
1947
1951
1954
1969
Decline or scarcity
Peak
1933
1937
1948
1952
1956
1955
1960
1934
1938
1946
1949
1953
1958
1957
1935
1939
1947
1950
1954
1959
91936
1951
* Data for 1932-39 from Marohnant area; 1946-54 from miscellaneous ereas; 1964-60
from Old Road area.
70
60
g
50
b
40
P
u,
.-F 30
E
E
3
20
10
0
PEAK
YEARS
46
*
48
*
50
52
*
54
56
*
YEAR
58
60
*
62
64
* *
66
FIQ.4. Population densities in summer of brown lemming, Lemmus trimucrmtus,
at Point Barrow, Alaska. (After Schultz, 1969.)
K
278
CHARLES J. KREBS AND JUDITH R. MYERS
gathered quantitatively over a three- or €our-year period which fails
to show a population cycle. W e conclude that microtine rodent populations
normally undergo population cycles with a period of three to four years
and this density pattern should be assumed to be the normal conjiguration.
Microtine populations that do not fluctuate cyclically are the unusual
situation and if any can be located they would be exceptionally
important to study. We feel that the burden of proof should be shifted
to those who would claim to have a non-cyclic population.
One of the dogmas about population cycles is that-they are more
pronounced in arctic regions (Odum, 1971, p. 193). A corollary of this
dogma is that southern populations of voles should have reduced
amplitudes of cycles and ultimately reach a point of having no cyclic
fluctuations at all. We have been unable to trace the origin of this
dogma; perhaps it was first mentioned by Howell (1923). Dymond
(1947, p. 14) also reports the dogma: “It has long been recognized that
periodic fluctuations in animal populations are virtually confined to the
northern part of the northern hemisphere and are especially characteristic of the Arctic, not only in America but also in Europe and Asia”.
Keith (1963, pp. 67-68) reports that the ten-year cycle is absent from
some southern populations of snowshoe hare (Lepus americanus) and
ruffed grouse (Bonasa umbellus), but the available data are poor and
this conclusion uncertain. We can find no quantitative evidence that
vole and lemming cycles are more pronounced in arctic regions than
they are farther south. Wildhagen (1952) states that lemming fluctuations are more pronounced in northern Norway than in southern
Norway, and Kalela (1962) supports this statement. But no rodent
census data are available and one may be comparing the “visibility” of
peak populations in northern and southern habitats rather than the
cyclic amplitude.
We conclude that vole and lemming populations go through regular
cycles of abundance everywhere they have been studied. We view these
cycles as a special type of population fluctuation, and most of the
discussion to follow is independent of whether -the rodent fluctuations
are regular or irregular. Even if one denies that microtine populations
cycle regularly, one must still explain their fluctuations.
c. S T R U C T U R E O F P O P U L A T I O N F L U C T U A T I O N S I N M I C R O T I N E S
One of the first steps toward understanding a population fluctuation
is to describe it in some detail. Part of the lack of progress in explaining
microtine cycles is due to the fact that emphasis has been on identifying
cycles and determining relative densities. But to understand cycles we
279
POPULATION CYCLES IN SMALL MAMMALS
must describe them in detail. A t what season does the increase begin?
How jong is the peak phase? When does the decline begin and how
rapid is it? We now attempt to answer some of these questions.
1. Increase phase
The increase phase is defined as a period of large increase in numbers
from one spring to the next (Chitty and Chitty, 1962). There are two
views on the structure of the increase phase. The increme phase might
be a gradual, exponential build-up from low numbers over two or even
three years. Koshkina (1966) suggests that the number of Clethrionomys
M
M
J
S
N
~
M
M
J
S
N
J
M
M
J
S
N
~
M
M
J
\
S
N
/
1933
1934
1935
1936
FIG.5. A population cycle of Microtw, penmylvankua on two areas near Ithaca,
New York. Winter months are shaded. (After Hamilton, 1937.)
on the Kola Peninsula gradually increases over three summers to a
peak. Pitelka (1958) states that brown lemming cycles in northern
Alaska have two successive winters of rapid population growth so that
numbers build up gradually over two years. Fuller (1969) found that
Clethrionomys gapperi and C . rutilus in northern Canada increased from
an extreme low in 1964 to a peak in 1966. Hamilton (1937) described a
population cycle of Microtus pennsylvanicus in New York in which the
increase occurred gradually over two years (Fig. 5). Populations
increased in the summer and dropped back during the winter months,
so that the net annual increase was relatively small from 1933 to 1935.
Bodenheimer (1949) states that populations of H.guentheri in Israel
increase gradually over two years to reach a peak.
An alternative view is that the increase phase is a rapid explosion
280
CHARLES J. KREBS AND JUDITH H. MYERS
which occupies one year or less. Table I1 shows that a number of
populations studied by Chitty and Chitty (1962) went through the
increase phase in one year. Our studies of M . ochrogaster and M .
pennsylvanicus in Indiana have provided several examples of rapid
increases; Fig. 6 gives one example. We never found in the Indiana
Microtus a gradual increase of the type Hamilton (1937) observed
(cf. Fig. 5). Newson (1963) describes a period of increase in Clethrionomys glareolus near Oxford that occupied one year.
60
40
9i
a
30
20
0
z
10
7
4
2
1
1968
1969
1970
FIG.6. A decline and subsequent increase in Microtus ophrogaster on the Carlson
Farm area in southern Indiana. Winter months are shaded. Vertical lines delimit
breeding period. (From Myers and Krebs, 1971.)
In Table I11 we present data on the instantaneous rate of population
growth ( r ) for the increase phase of the population cycle. Data are
presented only for populations trapped intensively at monthly intervals
(or less); we include some winter estimates derived from an accurate
fall sample and a spring sample. Some of the rates of increase in Table
I11 are unusually high. The three high values for Clethrionomys glareolus
POPULATION CYCLES I N SMALL MAMMALS
281
TABLEI11
Measured rates of population growth in the increase phme of the population cycle
f o r several vole species. Geometric increase is assumed; r is measured as
instantaneow rate per week
Species
Time period
Mean r
Reference
Ckthrionomys
gbreolus
March-Dec. 1958
March 1958Jan. 1959
April-July 1966
AprilJuly 1968
April-July 1970
0.074 Newson (1963)
0.050 Newson (1963)
0.132 Petrusewicz et al. (1971)
0-135 Petrusewicz et al. (1971)
0.129 Andrzejewski and
Rajska (1972)
Microtw
pennsy lvanicus
June-Sept. 1933
June-Sept. 1933
March-Sept. 1933
March-Sept. 1934
March-Sept. 1934
March-Sept. 1934
June-Nov. 1965
Dec. 1965-Feb. 1966
Aug.-NOv. 1967
Aug.-Nov. 1969
Aug.-Nov. 1967
Dec. 1967Jan. 1968
M a y J u l y 1967
Aug.-Nov. 1967
Dec. 1967-Jan. 1968
Aug.-Nov. 1969
Dec. 1969-Feb. 1970
0.035
0.022
0.051
0.035
0.041
0.040
0.100
0.046
0.063
0.089
0-096
0.029
0.059
0.134
0.032
0.036
0-032
Microtus
ochrogaster
June-Oct. 1965
Nov. 1965-March 1966
Feb.-Oct. 1968
June-Oct. 1967
Aug.-Oct. 1969
NOV.1969-Feb. 1970
0-037 Krebs et al. (1969)
0.036 Krebs et al. (1969)
0.041 Krebs, unpublished
0-031 Gaines and Krebs (1971)
0.231 MyersandKrebs (1971b)
0.041 Myersand Krebs (1971b)
Microtw
californicus
Nov. 1962-March 1963
Nov. 1962July 1963
Jan.-April 1964
May-Oct. 1964
Aug. 196Wune 1967
0.088
0.091
0.042
0.070
Krebs (1966)
Krebs (1966)
Krebs (1966)
Krebs (1966)
Batzli and Pitelka (1971)
Oct. 1952-May 1963
Sept. 1951-May 1952
Sept. 1959-May 1960
0.051
0.115
0.083
Thompson (1955a)
Thompson (19554
Krebs (1964a)
Lemmw
trimucronatus
0.040
Hamilton (1937)
Hamilton (1937)
Hamilton (1937)
Hamilton (1937)
Hamilton (1937)
Hamilton (1937)
Krebs et al. (1969)
Krebs et al. (1969)
Gaines and Krebs (1971)
Gaines and Krebs (1971)
Gaines and Krebs (1971)
Gaines and Krebs (1971)
Gaines and Krebs (1971)
Gaines and Krebs (1971)
Gaines and Krebs (1971)
Gaines and Krebs (1971)
MyersandKrebs (1971b)
282
CHARLES J. KREBS AND JUDITH H. MYERS
TABLE111 (contd.)
Species
Time period
Mean r
Reference
Dicrostonyx
groenlandiczls
Sept. 1959-May 1960
0.042
Krebs (1964a)
Clethrionomys
mfocanzls
June-Sept. 1954
0.146
Kalela (1957)
are from an island population (see later discussion). One very high
r value for Microtus ochrogmter (0.231) was probably enhanced by
immigration into the area. Table I V presents one aid t o interpreting
Table 111; it gives doubling times for a range of r-values and the rate of
growth over six months and one year.
Since Leslie and Ranson (1940) calculated that M. agrestis might
increase ten-fold over a six-month breeding season, there have been few
attempts to analyze the increase phase quantitatively. Table I11
indicates that for the few cases we have measured, the increase observed
is more typically three-fold to six-fold over a six-month period.
TABLEIV
Table of instantaneous population growth rates, the corresponding doubling time in
weeks, and the number of animals that would be present for every starting individual
after six months and one year had elapsed at the indicated growth rate
Instantaneous
growth rate
per week
Doubling time
in weeks
No. of times popln. has
multiplied a t the end of:
6 months
0.02
0.03
0.04
0-05
0.06
0.07
0-08
0.10
0.12
0.14
34.7
23-1
17-3
13.9
11.6
9.9
8.7
6.9
5.8
5.0
1.68
2.18
2.83
3.67
4.76
6.17
8.00
13.46
22-65
38.09
1 year
2.83
4.76
8.00
13.46
22.65
38.09
64.07
181.3
512.9
1451.0
POPULATION CYCLES IN SMALL MAMMALS
283
2. Peuk phase
The peak phase is defined as a period of little change in numbers from
one spring to the next (Chitty and Chitty, 1962). The peak phase is
usually obvious, since population densities are typically much higher
than they are in other phases of the cycle. Some species, however, do
not have a well-defined peak phase. Microtus californicus is one example
(Fig. 7 ) ;M. ochroguster is another (Krebs et al., 1969). I n these populations there is typically an increase phase, followed by a brief period of
high numbers, and then a decline phase.
1962
1963
1964
FIG. 7. A population cycle in Microtus calqornicw at Berkeley, California.
(After Krebs, 1966, and Pearson, 1971.)
The peak phase in other species is well-defined and may last for a
year (or rarely two years). Chitty and Chitty (1962) show that the peak
year in M. agrestis begins with a spring decline in numbers that may
come at slightly different times in the two sexes. This spring decline is
followed by a more or less rapid rise in numbers so that in the fall of
the peak year numbers are roughly the same as they were in the spring.
Thompson (1955a) described a spring decline in the brown lemming
during the peak year, and Krebs (1964a) also observed this drop in
lemming populations in northern Canada. Figure 8 shows a spring
decline in a peak phase of M. pennsylvanicus in 1968. I n this particular
case both males and females declined from February to early May and
the population then recovered to high numbers in late summer. Half
284
CHARLES J. KREBS AND JUDITH H. MYERS
1967
1968
1969
FIU. 8. A population cycle of Microtus pennsylvanicue in southern Indiana.
(After Gaines and Krebs, 1971.)
of the population may disappear during this spring decline of the peak
year.
3. Decline phase
The decline phase of the cycle seems especially variable. Chitty
(1955) recognized three types of decline (Fig. 9). The most gradual type
of decline is the Type H. Numbers fall gradually over one to two years
with some recovery during the breeding season. Type G declines are
gradual declines in which there is no recovery during the breeding
season; numbers fall over one year or less. Type M declines are “crash”
declines in which numbers fall to a low during the winter and early
spring after a peak year. Of ten declines studied in Microtus agrestis,
Chitty and Chitty (1962) classed three as Type M “crashes”, four as
Type G or intermediate to M and G, and three as Type H declines.
There are few examples in the literature of Type M “crash” declines
that have been monitored accurately. Some of the brown lemming
declines at Barrow, Alaska, have probably been of this type (see Fig. 4).
Zejda (1967) studied a peak and decline of a Clethrionomys glareolus
population. The population peaked in September 1964, gradually
’OPULATION CYCLES I N SMALL MAMMALS
,
I
285
I
FIG.9. Hypothetical diagram of the three types of population declines recognised by Chitty (1955).
declined through December, then dropped very rapidly and completely
disappeared by mid March 1965. Krebs et al. (1969) monitored a
population of Microtus ochrogaster (Fig. 10) which began declining in
October 1966, fell rapidly through December, and then more gradually
until completely disappearing by April 1967. A population of M .
californicus which showed a Type M “crash” in 1963 was studied by
Krebs (1966).
The Type G decline in which numbers fall continuously through a
breeding season was first described by Godfrey (1955) for two populations of M . agrestis. A Type G decline was found in the lemmings
Lemmus trimucronatus and Dicrostonyx groenlandicus in northern
Canada by Krebs (1964a). Figure 6 shows a Type G decline in Microtus
ochrogaster from Indiana. Many of the declines described by Krebs
et al. (1969) and Gaines and Krebs (1971) for M . penwylvanicus were
probably Type G declines since they occurred during the breeding season,
but they were followed very quickly by a return to the phase of increase.
Type H declines were first described by Hamilton (1937) for M .
pennsylvanicus. Figure 7 shows a Type H decline in M . californicus.
Kalela (1957) studied a population cycle of Clethrionomys rufocanus in
Finnish Lapland; some recovery of the population was indicated after
the initial decline, and hence a Type H decline occurred (Fig. 11).
Koshkina (1965) presents data from two declines of C. rutilus in the
boreal forest of the U.S.S.R.; both declines fit the Type H classification.
Gaines and Krebs (1971, p. 709) show a Type H decline for Microtus
ochrogaster in Indiana.
The recovery of the population during a Type H decline may be
substantial, and this has caused much confusion about cyclic fluctuations in the literature. Chitty and Chitty (1962) observed that autumn
population densities in Microtus agrestis could be nearly equal for
286
CHARLES J. KREBS AND JUDITH H. MYERS
I
I
9
II
I
I
b
- -
1
1
1
1
1
A M J
1967
A M J J A S
1966
1965
FIG.10. A population cycle of Microtw ochrogaater in southern Indiana. A Type
M decline occurred in the fall of 1966. (After Krebs et al., 1969.)
2 5 1
I
I
I
I
1
1
1
1
1
I
I
I
I
I
I
I
I
I
c
In
.-
-c
0F 2 0 n
e
J--'\.,
;-;-
0
-
L
J@lt\\\,\
\
\
P
z
\\\' \
I\\/
5-
Z
9
0
O
!
!
!
!
!
I
I
I
I
I
I
FIG. 11. A population cycle of Clethrionomye rufocanue in northern Finland.
A Type H decline occurred in 1956. (After Kalela, 1957.)
287
POPULATION CYCLES IN SMALL MAMMALS
several years in a row during the increase phase, the peak phase, and a
Type H decline. Superficial observations on autumn densities thus
might lead one to conclude that M. agrestis populations do not fluctuate
in cycles.
Table V attempts to summarize the available information on rates of
change in declining populations. Overwinter (or dry season) declines are
separated from summer (or wet season) declines. Some of these data are
only rough estimates available from kill trapping. Two points can be
noted from Table V: (1) populations decline at rates which are usually
TABLEV
M e w r e d rates of population chunge in the decline phase of the poplation cycle for
several species of lemmings and voles. Geometric change is aaszcnzed; r is maacncrerl
aa instcmtaneoua rate per week
Species
Time period
Type of
decline
Meanr
Reference
OVERWINTER DECLINES
Aug. 1963-May 1964
- 0.061
C~ethrionomys
gkcreolua
Dec. 1964-Feb. 1965
- 0.275 Zejda (1967)
Clethrwnomys
rufocanua
Sept. 1955-June 1956
July 1966-April 1967
Aug. 1968-April 1969
- 0.063
- 0.039
- 0.054
Kalela (1957)
Petrusewicz et al. (1971)
Petrusewicz et al. (1971)
Microtua
pennsylvanicua
Dec. 19354une 1936
Dec. 1935June 1936
Dec. 193Wune 1936
Nov. 1966-Feb. 1967
Nov. 1968-Feb. 1969
- 0.053
- 0-078
- 0.048
- 0.012
- 0.023
Hamilton (1937)
Hamilton (1937)
Hamilton (1937)
&ebS et al. (1969)
Gaines and Krebs (1971)
Microtus
ochrogaater
Oct. 1966-March 1967
NOV.1968-Feb. 1969
Nov. 1967-Feb. 1968
Nov. 1968-Feb. 1969
Nov. 1968-Feb. 1969
- 0.065
-0.183
-0.058
Krebs et al. (1969)
Krebs, unpublished
Gaines and Krebs (1971
GainesandKrebs (1971
MyeraandKrebs(197lb)
Microtua
ca1qOrnicua
Sept. 1963Jan. 1964
Sept. 1963-Feb. 1964
- 0.156 Krebs (1966)
- 0.084 Krebs (1966)
Lemmua
Sept. 196&May 1961
- 0.064
KrebS (1964~)
Sept. 196&May 1961
- 0.032
Kpbs (1964a)
- 0.165
- 0.073
Koshkina (1965)
trimucromtua
Dicrostunyx
groenhndicw
288
CHARLES J. KREBS AND JUDITH H. MYERS
TABLEV (contd.)
Species
Time period
Type of
decline
Mean r
Reference
SUMMER DECLINES
Clethrimomys
rmtilus
June-Aug. 1964
H,
+0-180t Koshkina (1965)
Clethrionomys
glareolw
April-Oct. 1967
HI
+ 0.042
Petrusewicz et al. (1971)
Clethrionomys
rufocanw,
JuneSept. 1956
H,
+ 0.062
Kalela (1957)
Microtus
pennsylvanicus
MarchJuly 1967
Feb.July 1969
Feb.-July 1969
Feb.July 1969
G
G
G
G or HI?
- 0.029
- 0.061
- 0.1 19
- 0.017
Krebs et al. (1969)
Gaines and Krebs (1971)
Gaines and Krebs (1971)
MyersandKrebs (1971b)
Microtw
ochrogaster
March-Sept. 1967
Feb.-Oct. 1968
March-Aug. 1969
M
H,
G
-0.083
0.023
- 0.142
+
Krebs et al. (1969)
Gaines and Krebs (1971)
MyersandKrebs (1971b)
Microtw
ugreatis
May-Aug. 1951
May-Aug. 1952
G
G
- 0.107
- 0.089
Godfrey (1955)
Godfrey (1955)
Microtw,
calqmicus
Feb.-July 1964
MarchJuly 1963
Feb.-May 1964
HI
M
H,
+0.062
-0.185
+0-076
Krebs (1966)
Krebs (1966)
Krebs (1966)
Lemmus
trimwronatw
June-Aug. 1961
G
-0.124
Krebs (1964e)
Dicrostonyx
gr0enlaTldicw
June-Aug. 1961
G
- 0.131
Krebs (19644
*
H, refers to the first year of a Type H decline.
Note that populations can increase in the summer breeding season of the decline
phaae of Type H.
t
greater than rates at which they increase (cf. Table 111). This is
particularly true of Type G and M declines, and (2) during the summer
breeding season of Type H declines, populations may increase at rates
which equal those of the increase phase (e.g. Fig. 7). This observation is
particularly important because it shows the complex nature of population cycles. Populations do not simply increase to high densities and
decline to low densities. High population density is not suficient to
produoe a decline, and low density is not suflcient to stop a decline.
289
POPULATION CYCLES IN SMALL MAMMALS
4. Phase of low numbers
Populations may fall to low numbers and remain there for one to
three years, but in some cycles this phase is absent and populations go
directly from the decline phase to the increase phase (e.g. Fig. 6). Very
little is known about the phase of low numbers in voles or lemmings.
Koshkina (1966) suggested that populations of Clethrionmys rufocanus
on the Kola Peninsula did not have a phase of low numbers but
after declining began to increase gradually over two or three years.
Norwegian lemming populations on the Kola Peninsula, however, did go
through phases of scarcity for several years.
S
O
N
1957
D
J
F
M
A
M
J
J
A
S
1958
FIG 12. Annual cycle in the phase of low numbers for Microtwr (penmylvanicus
in southern Michigan. Winter months are shaded. (After Getz, 1960.)
Getz (1960) studied a Michigan population of Microtw pennsylvanicus
that was apparently in the phase of low numbers (Fig. 12). I n both
marsh and old field habitats voles showed an annual cycle with little
net change in numbers. During the spring and summer increase the
population grew at 7% per week, but this was not sustained. Krebs
(1966) described a similar sequence in M. culifornicus in the low phase
(Fig. 13); numbers rose rapidly for a short time but then fell back
during the breeding season to the low density at which they started.
We do not have a sufficient number of descriptions of low populations
290
CHARLES J. D E B S AND JUDITH H. MYERS
20.0
15.0
-
-
E 10.0g 8.01
5.0-
>”
v)
0
3.0-
Y
0
0
2.0-
Z
1.0
0.8
O
J
!
A
!
S
!
O
1963
!
N
!
D
!
J
!
F
!
!
!
!
M A M J
J
! I
1964
FIG.13. Annual cycle in the phase of low numbers for Microtus cali,fornicua at
Berkeley, California. (After Krebs, 1966.)
of any vole species to say if the patterns shown in Figs. 12 and 13
are general. Pearson (1963), for example, shows a three-year period of
great scarcity in M . californicus but his data are not sufficient to
determine whether the sequence of density change displayed in Fig. 13
applied to the three years.
Until there are more data on the phase of low numbers we will not be
able to distinguish two quite different interpretations of this phase:
1. that the population declines to a level below our accuracy of
.
measurement and then begins to grow geometrically back to the
next peak; the early stages of this geometric growth we call
the “phase of low numbers” but such a name reflects more our
inability to measure changes in low density populations than
the biological reality;
2. that the population declines and remains low for a long period;
brief spurts of population growth may occur but numbers quickly
fall back to a low level; this “start-stop” type of population curve
persists until the phase of increase occurs, and the net population
growth is zero in spite of low densities.
Biologists typically stop working on a population once it gets sparse
and the critical turnaround from the low phase to the phase of increase
has rarely been studied.
POPULATION CYCLES IN SMALL MAMMALS
291
V. DEMOGRAPHIC
MACHINERY
A.
REPRODUCTION
Populations rise and fall because of changes in birth, death and
dispersal rates, and we now turn to consider these three. Birth rates in
polyoestrous mammals are a function of six components (Fig. 14), and
we must analyze each component separately.
Total yearly reproduction
/ \
Total p r l y cmliryo praluction
(per niature female)
/\si/l\
Alonthly cnhryo ntcs
Litter sire
Nunibcr of animals breeding
Population
Sex ratio
Age at
maturity
Lengrh of breeding season
Yrcgnancy rare
FIG. 14. Components of reproduction in polyoestrous mammals. (After Krebs,
1964.)
1. Litter size
One way in which to encourage population growth is to have larger
litters, and we now enquire whether the number of embryos per pregnant
female changes in relation to cyclic phase. We will not review here the
statistical problems of estimating and comparing litter size in voles and
lemmings (see Zejda, 1966; Keller and Krebs, 1970). Litter size may be
affected by season of year, body weight of female, age and parity, and
one must control for these variables if comparisons are to be valid.
We have been forced to disregard a significant fraction of the data
in the literature because of this problem.
Several authors have claimed that litter size does not change from
phase to phase in the cycle. Kalela (1957) found no evidence that
Clethrionomys rufocanus populations had lower litter sizes in the peak
or decline phase compared with the phaee of increase. Thompson
(195Sa)reported no change in litter size over a brown lemming cycle at
Barrow, Alaska. Table V I gives average litter sizes for C. rutilus
studied by Koshkina (1965), and illustrates the fact that litter size is
unaffected by the cyclic phase. Krebs (1964a) could find no significant
changes in litter size over a cycle of the lemmings Lemmus trimwronatus
292
CHARLES J . KREBS AND JUDITH H. MYERS
TABLEVI
Average number of embryos in overwintered females of Clethrionomys rutilus
during a cyclic Juctuation in numbers. Sample size in parentheses.
(After Koshkina, 1965)
Year and cyclic phase
May
June
July
Average
May to July
1962
1963
1964
Increase
Peak
Type H decline
7.8
(17)
7.5
(43)
6-2
(20)
7.11
8.0
(52)
6.5
(42)
5.7
(49)
6.76
(13)
7.6
(39)
6.9
(21)
7.25
6-6
and Dicrostonyx groenlandicus. Stein (1957) could find no significant
year-to-year changes in litter sizes of Microtus arvalis in an intensive
six-year study covering two declines. Keller and Krebs (1970) could
find no changes in litter size related to cyclic phase in M . pennsylvanicus
in a three-year study. Hoffmann (1958) found no change in litter size
in M . californicus in relation to phase of the cycle.
Several authors have claimed, to the contrary, that litter size is
higher in increasing and peak populations and lower during the decline
phase. Hamilton (1937) reported that litter size was higher during the
increase phase of a cycle of Af. pennsylvanicus and lower during the
decline phase. His data unfortunately do not support this conclusion.
The largest litter sizes were recorded in the peak year of 1935, and the
increase years of 1933 and 1934 had essentially the same litter sizes as
the decline year of 1936. Bodenheimer (1949) stated that M . guenthri
had a higher average litter size in the increase phase, but no data are
presented t o substantiate this claim. We have been unable to find any
good quantitative evidence that litter size is lower during the decline
phase. Koshkina and Khalansky (1962) present data on Lemmus
lemmus that are difficult to interpret. Litter size was highest in one
year of increase and one peak year, and lowest in another peak year,
and two decline years. Unfortunately data are grouped over all size
classes for all months and parity classes, so it is impossible to determine
if these trends are valid. If they are valid, these data would be the first
to show a depressed litter size in the decline phase.
POPULATION CYCLES IN SMALL MAMMALS
293
Finally, several authors have suggested that litter size is reduced in
peak populations but essentially the same during the other phases of
the cycle. Hoffmann (1958) reported a 1 6 2 5 % drop in litter size during
the peak summer for a Microtus montanus population. Tanaka (1964)
shows a 17% drop in litter size during the peak summer for Clethrionomys smithi in Japan. Patric (1962) claims to have shown a 1 6 3 7 %
drop in litter size for C. gapperi populations; his estimates, however,
assume no significant effects of season, weight of female, or parity on
litter size. Zejda (1964) found a slight depression in litter size of C.
glareolus in the peak summer. Keller and Krebs (1970) found that litter
size was depressed 25% during the peak summer in Microtus ochrogaster.
Koshkina (1966) states that Clethrionomys rufocunw has minimal litter
size in the peak year, compared with the increase phase, but also
minimal in the decline and low phases. Unfortunately the data presented
are not sufficient to verify these claims because only mean values for
the whole year are given. Reichstein (1964) shows that litter sizes of
Microtus arvalis are reduced by 17-22% in peak years.
If the changes in litter size are to be an important driving force in the
population cycle, litter size should be depressed in the decline phase and
enhanced in the increase phase. We can find little evidence of these
trends in the populations studied to date, and we conclude that this
component of reproduction is not a critical link in the demographic
machinery.
2. Pregnancy rate
We next consider the percentage of mature females which are
pregnant during the breeding season. Note that we are not concerned
here with the age of sexual maturity or the length of the breeding
season.
Most workers seem to agree that the pregnancy rate does not vary in
relation to the population cycle. Figure 15 illustrates the similar
proportions of pregnant females in three years for Clethrionomys
rufocanus in Finland. Mullen (1965) found the same result in Lemmw
trimucronatus a t Barrow, Alaska over four years. Krebs (19644 found
no cyclic variation in the percentage of females pregnant for either
Dicrostonyx groenlandicus or Lemmus trimucronatw in northern
Canada. Keller and Krebs (1970) found no significant differences in
percentage of mature females pregnant during the summer months for
Microtus pennsylvanicus and M . ochrogaster.
A few authors have suggested that the pregnancy rate goes up in the
phase of increase. Hamilton (1937) measured the fraction of female M .
pennsylvanicus in New York that were both pregnant and lactating and
294
CHARLES J. KREBS AND JUDITH H. MYERS
I
I
100-
I
I
I
1
I
I
I
I
-
t
c
-
80-
t
2
a
n
-
-
60-
t n al
0
40
2
20-
G
0
1956
-
2
I
I
10
20
JUNE
I
I
I
10
20
JULY
I
I
10
20
10
AUGUST
20
SE PT.
FIG. 15. Percentage of mature females pregnant in Clethrionomys rufocanus in
northern Finland. The population increased in 1954, peaked in 1955, and declined
in 1956. (After Kalela, 1957.)
claimed that the breeding rate accelerated from the beginning of the
increase phase until the decline. He presents these data:
yo pregnant and
1934 (increase)
1935 (peak)
1936 (decline)
Sample size
lactating
107
223
282
47.65
53.65
55.66
None of these differences are statistically significant (x2 = 2.00, df = 2).
Nor are any of the individual months of May to August significantly
different between years. We conclude that there is no evidence of a
change in the breeding rate in Hamilton’s data. Bodenheimer (1949)
also suggested an increased pregnancy rate in increasing populations of
Microtus guentheri, but he presents no data to substantiate this claim.
3. Length of breeding semon
The breeding season of most voles and lemmings is very elastic in
length, and changes in the length of the breeding season are a major
driving force in causing the population cycle.
Winter breeding is the most spectacular illustration of the reproductive abilities of microtines. Voles and lemmings can breed during
some winters but not in others, and we need to know if this is related
to the phase of the population cycle. Several authors have described
winter breeding in lemmings. Sutton (Sutton and Hamilton, 1932)
found winter breeding in both Dicrostonyx groenlandicus and Lernrnus
POPULATION CYCLES I N SMALL MAMMALS
295
trimucronatus in the Canadian Arctic during a period of increase.
Krebs (1964a) found winter breeding in both these lemming species
during a phase of increase and no winter breeding during a decline
phase. Mullen (1965) shows the same result for Lemmus at Barrow,
Alaska. Soviet workers have recognized for many years the importance
of winter breeding in lemmings. Dunaeva and Kucheruk (1941) found
winter breeding in both Dicrostonyx torquatus and Lernrnus sibiricus
during a period of increase. Nasimovich et al. (1948) believed that
winter breeding of the Norwegian lemming was limited to the phase of
increase. Koshkina and Khalansky (1962) review winter breeding in
the Norwegian lemming and conclude that it plays a significant role in
the rapid population growth of this species.
Winter breeding has been noted in many vole species but there is
conflicting evidence of its relation to cyclic phases. After a favorable
summer and autumn Khlebnikov (1970) observed winter reproduction
in Clethrionomys rutilus. He interpreted this as being related to the
increase phase. Zejda (1962) analyzed winter breeding in the bank vole,
C. glareolus, and observed two successive winters of breeding. Neither
of these episodes of winter breeding led to population growth, and he
concluded that winter breeding was affected by the availability of food
(principally acorns) but did not lead to an outbreak. Smyth (1966)
argued that the relationship between winter breeding in voles and acorn
crops is not a simple one. A good food supply, such as a heavy acorn
crop, may be necessary for winter breeding but not sufficient. Newson
(1963), for example, found C . glareolus breeding in the winter of 19581959 (phase of increase) but not breeding in the winters of 1957-1958
or 1959-1960. There was a good acorn crop in fall 1958, but none in
1957 or 1959. But Newson noted that voles in grassland where there were
no acorns also bred during the winter of 1958-1959. Some aspect of
population density may interact with the available food supply and this
question awaits an experimental attack.
Winter breeding has been noted in Nicrotus by many workers, and
in many cases it occurs during the increase phase of the cycle and is
absent in the winter following the peak (reviewed in Keller and Krebs,
1970). This association, however, is not perfect. Chitty (personal
communication) has recorded winter breeding in Microtus agrestis in
the increase phase of the cycle, but some cycles occurred in which no
winter breeding was evident.
More evidence is available on the length of the summer breeding
period. I n the phase of increase the summer breeding often starts
early and ends late (or carries on through the winter), while in the peak
year the breeding season often ends abnormally early. Koshkina and
Khalansky (1962) pointed out that the Norwegian lemming stops
296
CHARLES J. KREBS AND JUDITH H. MYERS
breeding early during the peak year, and Thompson (1955a) also
observed this in the brown lemming. Krebs (1964a) observed an early
stop to summer breeding in a peak year for the brown lemming and the
varying lemming. Kalela (1957) found a shortened summer breeding
period in both the peak year and in the decline year for Clethrionomys
rufocanus (Fig. 15). Zejda (1967) pointed out that a very short reproductive season was a characteristic feature of the peak year. He observed
a peak population of C. glareolus that stopped breeding in June.
Koshkina (1966) states that the summer breeding season is one month
shorter in the peak year for C. rufocanus.
I n declining populations the breeding season often starts later than
usual and may also end early. Godfrey (1955) observed a three-week
delay in onset of summer breeding in Microtus agrestis. Chitty (1952)
also reported a delay in the summer breedingseason in declining populations of M . agrestis. Declining populations of M . calijornicus may delay
breeding for one to two months (Krebs, 1966). A slight delay in the
start of summer breeding was observed for M. ochrogaster and M .
pennsylvanicus by Keller and Krebs (1970).There are few data available
for northern species with respect to possible delays in the start of
summer breeding. Kalela’s (1957) observation on C l e t h r i o m y s
rufocanus is one which has been noted: a declining population started
breeding late in 1955 even though the spring came early. I n northern
species which typically begin breeding in spring when the snow melts,
it may be difficult to detect any delay independent of spring weather
variations. Further consideration of the interaction of weather and
breeding season is discussed later.
We conclude that the phase of increase in many voles and lemmings is
associated with an extended summer breeding season and possibly
winter breeding. I n the peak phase the summer breeding season is
shortened and winter breeding is absent. The decline phase often
resembles the peak phase, and may show a delay in the onset of summer
reproduction.
4 . Age at sexual maturity
The age at which an organism reaches sexual maturity has a critical
impact on its potential for population growth (Cole, 1954). At present
there are no good ways of aging living small rodents and we must rely
on weight as an index of age. Several methods have been suggested
for aging dead microtines. For Clethrionomys wear of the rooted
molars can be taken as an indication of age, and Lidicker and MacLean
(1969) suggest an aging method for Microtus californicus based on
relative cranial and body measurements. However, both of these
techniques are influenced by the environmental conditions to which the
POPULATION CYCLES IN SMALL MAMMALS
297
individual is exposed. Therefore, slow growth during the summer
causes underestimation of the age using the Lidicker and MacLean
(1969) technique, and so this technique has the same biases as the use
of body weight as an estimator of age. A new method for aging wild
rodents based on the fractions of soluble and insoluble proteins in the
eye lens is described by Otero and Dapson (1972). As the individual
ages a larger portion of the lens protein becomes insoluble in water.
This method is supposed to be less influenced by environmental factors
than other aging techniques. If one has detailed knowledge of a population’s breeding seasons and mortality rates, one can use weight as a
reasonable index of age, particularly for young animals.
Natural history observations have established that age at sexual
maturity is variable in microtines and that changes in the rate of
sexual maturation of young voles and lemmings are a major driving
force behind population cycles. Young Norwegian lemmings about
20 days old (25 g) were found pregnant in the summer of increase by
Koshkina and Khalansky (1962), while almost none of the summerborn young lemmings became mature in the following year of peak
density. Mullen (1965) records delayed maturation of male brown
lemmings in a peak summer. Kalela (1957) in a detailed investigation
of Clethrionomys rufocanus in Finland found that the maturation rate
of the early summer young was strongly affected by population density:
Proportion of early summer young mature
Males
Females
almost all
almost all
1954 (increase)
almost none
majority
1955 (peak)
majority
majority
1956 (decline)
Koshkina (1965) showed that maturation of Clethrionomys rutilus was
inversely related to population density (Fig. 16). Note that there were
always fewer males maturing then females. The same observation was
made by Zejda (1967) for C. glareolus.
Few studies on weight at sexual maturity have utilized the quantitative techniques of Leslie et al. (1945) to estimate the median body
weight at sexual maturity. Figure 17 shows changes in size at sexual
maturity in the brown lemming over a cycle in numbers. Note that
males are more strongly depressed in maturation than females. I n the
decline year of 1961 no young males matured, although young females
did mature at about four weeks of age. Keller and Krebs (1970) show
that the median weight at sexual maturity was higher in peak populations of Microtus pennsylvanicus and M . ochrogaster but equal in
increasing and declining populations. This work illustrates some of the
problems of using weight as an index of age. We know that growth
298
CHARLES J. KREBS AND JUDITH H. MYERS
60
-
50
-
I
' " I
1964
I
I
[
I
I
I
I
,
l
l
~
l
a
-l
-
~
-
?
-
c
0
E 40-
-x
-
-
3
30-
-
0
2
9)
WJ
0
c
-
0
20-
5
Y
-
2
10-
1961
O
a o
1958
-
1962
0
1959
0
1
0
1
"
3
'
1
"
~
"
'
1
S
6
'
1960
1963
15
18
b-b-
'
12
FIG. 16. Percentage of young Ckthrionomye rmtilw becoming sexually mature
in their first summer in relation to the population density. Young sampled from
June 1 5 J u l y 31 each year; density measured by snap-trap catch per 100 trap
nights in May. 0 = males;
1959
1960
Low
Peak
1961
Decline
1962
Low
FIG.17. Median body weight a t sexual maturity for brown lemming (Lemmus
trimucronatua) summer-born young, Baker Lake, Canada. Only the first summer
litter is included. (After Krebs, 1964%)
POPULATION CYCLES IN SMALL MAMMALS
299
rates of individuals tend to be low in declining populations (Krebs et al.,
1969). Consequently, if weight at sexual maturity is equal in increasing
and declining populations, age at sexual maturity must be greater in
declining populations. If we had a reliable indicator of age in voles, we
could investigate this deduction directly.
We conclude that the age of sexual maturity is an important variable
in the reproductive strategy of microtine rodents. Age a t sexual
maturity is increased in peak populations and perhaps also in declining
populations. More work is required to quantify these trends in species
which can be aged.
5 . Sex ratio
If the sex ratio is sufficiently disturbed from the typical 50% males
two things may happen. First, if there are too few males, females might
go unmated and consequently the pregnancy rate would decrease.
We have not been able to find any evidence from voles or lemmings
that females ever experience such limitation. Second, a shift in the
population sex ratio might regulate its density. Increasing populations
might have a higher percentage of females to increase the reproductive
output, while peak populations might equalize the sex ratio or even
favor males (Williams, 1966, p. 148). Unfortunately, natural selection
does not seem to operate in such a way to maximize population fitness.
Many studies on microtines have commented on sex ratios, but few
have analyzed the variables which affect the observed sex ratios.
Males are typically less abundant than females, but the sex ratio does
not correlate with population density in Microtus pennsylvanicus or
M . ochrogaster (Myers and Krebs, 1971a).
Abnormal sex ratios (20-30y0 males) in the wood lemming, Myopus
schisticolor, occur in field populations because some females produce
only female offspring and other females produce both sexes (Kalela and
Oksala, 1966). There is as yet little information on how this abnormal
sex ratio varies in relation to population density. Kalela and Oksala
(1966) describe a declining population of the wood lemming in which
the sex ratio increased from about 20-30y0 males in the peak year to
about 47% males in the year of decline. Whether this change was caused
by movement of animals or differential mortality is not known.
Except for the interesting case of the wood lemming, there is no
indication that variations in sex ratios are associated with population
fluctuations in voles and lemmings.
6. Summary
Reproductive changes are part of the machinery which drives the
population cycle. Not all components of reproduction are involved,
300
CHARLES J. KREBS AND JUDITH H. MYERS
however. Litter size does not change over the cycle, except in some
species in which it is lower in the peak year. Females in declining
populationshave normal litter sizes. The percentage of females pregnant
during the breeding season also seems to be independent of the population cycle. Length of breeding season is highly variable. Winter breeding
and extended summer breeding seasons occur during the increase
phase of many species. The peak year often has a shortened summer
breeding season, and the decline phase may also have a restricted
breeding season. Age at sexual maturity is the second component of
reproduction to change during a population cycle. Animals in peak
populations reach maturity at older ages and heavier weights, and
young voles and lemmings may not mature at all in their f i s t summer
if born into a peak population. Age a t maturity may also be delayed in
declining populations. Finally, sex ratios do not seem to vary in any
systematic way over the cycle in numbers.
B. M O R T A L I T Y
The reproductive changes discussed in the previous section are
sufficient to generate a population cycle even if the mortality schedule
were constant and independent of density. We here investigate the
mortality schedules of voles and lemmings and attempt to see if there
are patterns of change in mortality which reinforce or cancel the changes
in reproduction.
Mark-and-recapturework is necessary for the estimation of mortality
rates, and relatively little of this has been done on microtine populations
throughout a cycle in numbers. Sampling problems, discussed above
for the estimation of population size, plague estimation of mortality
rates. Marked animals may not respond to traps in the same way.
Juvenile animals are hard to catch in live traps. Animals may move
off the trapping area and since disappearance is equated to death,
measured mortality rates are really “loss rates”.
Mortality rates are very labile in lemming and vole populations.
Mortality varies with age but, given the accuracy of present methods,
we recognize only three age categories: adult, juvenile and nestlings.
We will now discuss each of these three age groups and then discuss
prenatal mortality.
1. Adult mortality
Most of the available data on mortality rates comes from the live
trapping of adult animals. Chitty (1952) estimated mortality rates in
peak and declining populations of Microtus agrestis. Figure 18 shows
some of the earliest data on survival rates in a declining population.
301
POPULATION CYCLES IN SMALL MAMMALS
- I
PEAK PHASE
A
M
J
J
A
1937
S
O
L
N
D
DECLINE PHASE
J
F
M
A
M
J
1938
FIG. 18. Minimum survival rates per 28 days for Microtwr agrestia during a
peak and subsequent decline. Adults in 1937 gradually disappeared over the
summer and the young of 1937 overwintered and declined in the spring of 1938.
= adults of 1937, 0 = young of 1937. Two areas were live-trapped to obtain
these estimates. (After Chitty, 1952.)
Adult voles in the peak year survive well (probability of survival per
28 days about 0.7) but gradually disappear through the summer to
be replaced by their young. These young voles survived very poorly
in the peak summer until August when survival rates improved. The
young voles overwintered with good survival until January or February
1938, when survival rates dropped and the population almost disappeared. Further detailed observations were made on several declining
populations of M . agrestis (Chitty and Chitty, 1962; Newson and Chitty,
1962). I n some declines survival rates deteriorate through the winter,
but in other declines survival remains good during the winter and
deteriorates only in the spring. A second peculiarity of the spring
deterioration in survival was noted: the two sexes may experience poor
survival at different times (Chitty and Phipps, 1966). Figure 19 shows
one example in which male M. agrestis declined about eight weeks
before the females. Chitty and Phipps (1966) concluded that M.
agrestis suffered two different types of losses: a steady drain on numbers
during most of the year, and sudden severe losses, especially in spring.
Spring losses occurred in peak and decline phases for M. agrestis and
z
.
',
i t
2-
b.-e-----~-..o.-.a.-4
I
I
t
I
I
, I ,
I
I
I
1
1
1
1
1
I
I
L
FIG.19. Cohort survivorship curves for overwintered adults of Microtw, agrestie
in the spring of the decline phase. Note that heavy losses of males occurred a t
the end of March, while females survived well until the end of May. (After
Chitty and Phipps, 1900.)
Adult losses are also related to cyclic phase in Microtus californicus
(Batzli and Pitelka, 1971; Krebs, 1966). Figure 20 shows that the
expectation of life is higher for voles in expanding populations. The
range of average life expectation for eight populations (Krebs, 1966)
was:
Adult males
Adult females
Expanding
populations
8-12 weeks
12-13 weeks
Declining
populations
3-6 weeks
2-7 weeks
The poor survival characteristic of low and declining populations was
manifest even in very sparse populations. For example, the RFS 6
area reached a "high" of about 20 per acre (see Fig. 13), which is about
one-tenth the density of a normal peak population, but the survival
rate per two weeks was only 0.45 in males and 0.36 in females. Survival
rates often differed in the two sexes in M . californicus. The simplest
interpretation of these episodes is that they are sampling artifacts from
small populations, but not all episodes can be explained away so
simply. Figure 21 shows a portion of the increase phase for one population of M . californicus. From mid March to early June 1963 males were
at a plateau in numbers (r = +0.006 per week) while females on the
same area almost doubled their numbers (r = +0.047). Part of the
303
POPULATION CYCLES IN SMALL MAMMALS
FIG.20. Survivorship curves for Mkrotua californicua from time of first live-trap
capture for animals from two expanding populations and two low or declining
populations. Adult males only. (After Krebs, 1966.)
difference between the sexes was caused by lower survival rates in the
males:
Survival rates per 14 days
Males
November 1962-March 1963
MarchJune 1963
0.84
0.74
Females
0.90
0.91
Males always survive less well than females during the breeding season,
but beyond this normal difference, some mortality factor affected the
males but not the females in this population from March to June.
Further details of the pattern of survival changes in cyclic vole
populations were obtained on Microtuspennsylvanicus and M. ochrogaster
by Krebs et al. (1969). Figure 22 illustrates one series of survival
estimates for M. penmylvanicus. Survival rates differed little in
304
CHARLES J. KREBS AND JUDITH H. MYERS
0.6
n
MALES
Y
Q
p:
UJ
I
FEB.
I
MAR.
I
APR.
I
MAY
I
JUNE
I
JULY
AUG.
1963
FIG.21. An episode in the increase phase of a Microtus californicus population
in which the two sexes behaved differently. Male survival rates were significantly
below female rates for 12 weeks in a row from March to June (hatched bar).
(After Krebs, 1966.)
increasing and in peak populations, but deteriorated in the decline
phase, following the pattern of changes in adult survival which was
described above. Survival rates do not correlate well between the two
sexes on the same area, which is another way of saying that males may
be surviving poorly while females survive quite well. One new observation was contributed by the study: survival rates do not correlate well
305
POPULATION CYCLES IN SMALL MAMMALS
GRID I
MlCROTUS PENNSYLVANICUS
gn
1.0
4
n
.8
W
a
w
l-
.6
a
a
_I
.4
3
.2
s5
u)
z
a
0
A
J
A
1967
O
D
F
A
J
1968
A
O
D
F
A
J
1969
FIU. 22. Minimum survival rates obtained by bi-weekly live-trapping of a
Microtus pennqjlvanicue population in Indiana. (Density data for this population
in Fig. 8.) Winter months are shaded. Mean survival rates for winter and summer
periods shown a t bottom. Horizontal line marks survival rate at which one half
of the population disappears per month. (Krebs, unpublished data.)
between two species living together on the same area. For example,
M . ochrogmter suffered high mortality and declined in numbers in fall,
1966, when M . pennsylvanicus on the same field were surviving very
well (Krebs et al., 1969, p. 599). Similarly Tast and Kalela (1971)
report the increase of a Lemmus lemmus population occurring simultaneously with the decline of a Microtus agrestis population.
Getz (1960) estimated a mean lifespan of about eight weeks for a low
population of M . pennsylvanicus in Michigan, and this seems to be
another example of a population at low density suffering a high rate of
loss. There is little information on survival rates of voles in the phase
of low numbers.
There is an unfortunate shortage of quantitative data on survival
changes in vole and lemming populations. The available evidence
suggests that survival of adults is nearly the same in the increase and
peak phases, but deteriorates in the decline phase and the phase of low
numbers.
2. Juvenile mortality
Juvenile mortality rates are particularly difficult to estimate. Only a
few juveniles are caught in live traps so mark-and-recapture techniques
are only slightly useful for sampling this segment of the population.
In most cases we can only estimate juvenile mortality indirectly by
determining the number of pregnancies in the population, estimating
the number of young born and then determining what fraction of these
306
CHARLES J. KREBS AND JUDITH H. MYERS
reach the trappable population of adult voles. Sampling techniques that
could catch large numbers of small juveniles would be most useful,
but at present none exist (cf. Andrzejewski and Rajska, 1972). '
Chitty (1952) reported that juvenile losses in Microtus agrestis were
high during the first half of the peak summer breeding season but were
reduced in the late summer and early fall. Godfrey (1955)reported high
juvenile losses in this species during the summer of two decline years.
Chitty and Phipps (1966) showed that young M . agrestis born between
March and June of a decline phase survived very poorly, while young
born from July to November survived well (Fig. 23). Less than one
young per pregnancy was recruited from March to June, even though
mean litter size was 4-6.
0
10
20
30
40
50
60
CUMULATIVE NO. ADVANCED PREGNANCIES
FIG.23. Cumulative plot of number of advanced pregnancies in Microtus agrestis
and the number of young entering the live traps four weeks later, decline phase,
1960. (After Chitty and Phipps, 1960.)
Summer mortality rates of juvenile lemmings were estimated by
knowledge of litter size, number of adult females breeding on the
trapping area, and subsequent number of juveniles that appeared in
live traps (Krebs, 1964a). Table VII gives these data for peak and
declining populations of the brown and varying lemmings in northern
Canada. The results are only approximate but suggest that survival
rates in the decline were only about half those in the peak.
307
POPULATION CYCLES I N SMALL MAMMALS
TABLEVII
Estimated early juvenile survival ratesfor the brown lemming Lemmus trimucronatus
and the varying lemming Dicrostonyx groenlandicus during the peak Bummer of
1960 and the decline summer of 1961. Juveniles were caught in live traps anywhere
from two to five weeks of age; estimated survival rates are corrected for the age at
first capture. (After Krebs, 1964a.)
Brown lemming
Total no. of litters
Calculated no. of young
lemmings born
No. of juveniles later
caught in traps
Estimated survival rate
from birth to 14 days
Varying lemming
1960 peak
1961 decline
1960 peak
1961 decline
summer
summer
summer
summer
25
148
2
15
3
18
9
60
59
4
7
8
0.62
0.29
0.64
0.28
By comparing the number of active mammae of female Microtus
montanus to the number of placental scars, Hoffmann (1958) attempted
to measure nestling mortality. The assumption here is that nursing
young are discriminate in which nipples are suckled, so that not all
nipples are developed. The frequency distribution of active mammae
indicates that this may be the case since females were observed with
one to seven mammae developed (median four). The inclusion of
placental scars remaining after prenatal losses would increase the index
of nestling mortality measured in this way. Hoffmann’s data showed
decreased nestling mortality in the summer of the population decline.
This technique merits further testing. If Hoffmann’s conclusion is
correct and if juvenile losses are high in declining populations, then the
major losses would have to occur after weaning.
For voles with overlapping generations Krebs and DeLong (1965)
proposed an index of early juvenile survival:
no. of juveniles recruited at time t
index of early juvenile survival at time t
no. of lactating females at time t 4 weeks
This index was used to investigate the association between early
juvenile survival and rate of population growth in the California vole
(Krebs, 1966). We wish to determine which of four independent
variables-male
survival rate, female survival rate, percentage of
females lactating and index of early juvenile survival-are
most
useful for predicting the mean rate of population growth. All variables
308
CHARLES J. KREBS AND JUDITH H. MYERS
used were mean values covering the “summer” or “winter” portions
of the year (corresponding approximately to the breeding season and
the non-breeding season). Table VIII gives the results of a multiple
regression analysis of these five variables in the California vole. From a
statistical point of view, female survival rate is the most important
determinant of population growth in Microtus californicus and the
survival rate of young juveniles is second in importance. Neither the
male survival rate nor the percentage of females which are lactating
are needed to predict the rate of population growth.
TABLEV I I I
Multiple regression of mean rate of population growth ( Y )in Microtus californicus’
on mule survival rate (XI),femule survival rate (XJ, percentage of lactating females
( X 8 ) ,and index of early juvenile survival (X4).The best equation to describe this
relationship is Y = 0.4393 X,+O.O498 X,-O.3941 which has R = 0.88. (Data
from Krebs, 1966.)
Male survival rate
Female survival rate
Percentage lactating females
Index of juvenile survival
Multiple
regression
coefficient
Partial
correlation
coefficient
Relative
importance*
n.s.
0.4393
n.8.
0.0498
0.14
0-86
0.11
0.74
1 -0
0.66
* Relative importance measured by the ratio of standardized partial regression
coefficients.
A similar analysis was carried out by Krebs (1972) on a more extensive
set of data on Microtus pennsylvanicus and M . ochrogaster from Indiana.
For M . ochrogaster early juvenile survival was the most important
determinant of population growth, and female survival and the percentage of lactating adults were of secondary importance. For M . pennsylvanicus female survival rate and the percentage of lactating adults were
most important and early juvenile survival was of secondary importance.
Juvenile survival was thus important in affecting population growth in
both vole species.
Very few data are available concerning juvenile survival of lemmings
and voles. It appears that juvenile survival is particularly low in
declining populations and can also be low in peak populations, but we
have no details of how this loss is distributed by age or between different
litters in a population. Techniques for marking and recapturing
juveniles would aid in overcoming these problems.
We can explore the relationships between population growth and
POPULATION C Y C L E S IN SMALL MAMMALS
309
survival of adults and juveniles in a simplified life-table computation.
The basic variables are the adult survival rate per 14 days and the
index of early juvenile survival. The constants in the life table are the
litter size at birth (4.54 in Microtus pennsylvanicus), the age a t sexual
maturity and the age at first capture. For a series of simplified calculations with M . pennsylvanicus we have assumed the age at f i s t capture
and the age at sexual maturity both to average five weeks. The index
of early juvenile survival in association with the litter size at birth
and the age a t first capture determines the survival between birth and
SlJRVlVAL RATE
PER 14 DAYS
FIG.
24. Isopleths for the instantaneous rate of population growth (r) as a function
of survival rate of subadult and adult voles and number of juveniles recruited
into the trappable population per pregnancy. Microtwr pennsylvanicwr parameters were used in this simple life table calculation: litter size, 4.54; age a t
maturation, 5 weeks; age at first capture, 5 weeks.
recruitment. We assume in the simple calculations that the survival
rate of adults is constant from first capture onward, and that litters are
produced continually at three-week intervals throughout life. Figure
24 shows the results of this simplified life-table model for Microtus
pennsylvanicus. I n most natural populations less than two recruits are
obtained from each pregnancy, and consequently the rate of population
growth is very dependent on the survival rate of reproducing females.
Table IX gives the average parameters of reproduction and survival for
M . pennsylvanicus in Indiana, and illustrates the general decline in reproduction and juvenile survival from the increase phase to the peak,
and the drop in subadult and adult survival from the increase and peak
phases to the decline phase. Note that the survival of adult females need
L
310
CHARLES J. KREBS AND JUDITH H. MYERS
TABLEIX
Demographic parameters for Microtus pennsylvanicus populations live-trapped in
southern Indiana from 1965 to 1970. Early juvenile survival was measured by the
number of recruits entering the trappable population per lactating female. The
aurvival rate of adults was estimated by simple enumeration every 14 days.
Survival
Reproduction
yo Adults
Increase phase
Peak phase
Decline phase
lactating
Early
juvenile
45
29
27
1-31
0.96
0.88
Subadult and adult
Males
Females
0.78
0.79
0-71
0.86
0.85
0.72
deteriorate only 0.10 to 0.15 per 14 days in order to produce a population decline. Figure 24 illustrates this also.
3. Prenatal mortality
Embryos might be lost either before implantation or after, and this
mortality in utero could be an additional driving force behind rodent
cycles. Prenatal losses are assessed by the differences in counts between
corpora lutea in the ovaries, implanted foetuses in the uterus and resorbing embryos which appear in mummified form as pregnancy continues. Since most prenatal losses are small (often lessthan5-10%), large
sample sizes are needed to achieve statistical precision, and consequently
few data are available for fluctuating populations of voles and lemmings.
Kalela (1957) reported no obvious increase in prenatal mortality in a
declining population of Clethrionomys rufocanus. Hoffmann (1958)
reported only a slight change in prenatal mortality between peak and
declining populations of Microtus montanus. Krebs (1964a) found no
increase in prenatal mortality in declining populations of Lemmzls
trimucronatus and Dicrostonyx groenlandicus in Canada, and Mullen
(1965) described the same finding for Lemmus trimucronatus in Alaska.
Keller and Krebs (1970) found no changes in prenatal mortality over a
population cycle in Microtus ochrogaster and M . pennsylvanicus. Stein
(1957) reported only 3.6% resorptions in 1513 embryos of M . arvalis.
Thus, prenatal mortality does not seem to be related to the population cycles of small rodents. No one has yet found a population
declining because of excessive prenatal losses.
4. Summary
Mortality changes are part of the syndrome of demographic events
which drive population cycles in rodents. Adult mortality rates are low
POPULATION CYCLES IN SMALL MAMMALS
311
in the increase and peak phases, and are high in the decline phase and
also in the phase of low numbers. Juvenile losses are high in the peak
phase and in the decline phase. Prenatal mortality does not seem to
vary systematically during the population cycle.
c. D I S P E R S A L
Population densities can change because of variations in birth,
death, or dispersal rates, and almost all population studies on lemmings
and voles have been concerned only with b i r t h and deaths. The
simplest dynamic assumption is that immigration cancels emigration
and the population changes are solely a function of birth and death
rates. This simple view would be adequate if there were no spatial
heterogeneity in nature and no marginal habitats for small rodents
(Anderson, 1970).
The importance of dispersal in population regulation of voles W&B
first shown by studies on enclosed populations. Clarke (1955) showed
that Microtus agrestis populations in large cement cages (67 m2)would
increase to numbers far in excess of those ever found in natural areas.
He obtained a population “high” at 58 individuals, which is equivalent
to 3500 per acre (8657per ha), about ten times higher than ever occurs in
nature. Van Wijngaarden (1960) obtained densities of M . arvalis up to
7.25 voles per m2 in 100 m2 pens, which is equivalent to 29 300 voles
per acre (72 500 per ha), approximately 100 times higher than natural.
The same results were reported by Louch (1956) for M . pennsylvanicus,
Frank (1953) for M . arvalis, and Houlihan (1963) for M . californicus.
These studies on enclosed vole populations are difficult to interpret
because animals are maintained on artificial food with little or no
predation, and dispersal is prevented. The next step was to study
an enclosed population in a natural habitat with a normal complement
of predators and natural forage. Krebs et aZ. (1969) studied populations
of Microtus pennsylvanicus and M . ochrogmter enclosed in a two-acre
(0.8ha) grassland surrounded by a wire fence extending two feet above
ground. These populations behaved in the same way as the confined
laboratory populations-they increased to abnormally high densities
and then decimated the natural forage (Fig. 25). Since severe overgrazing is rarely seen in natural grasslands, we concluded that by
preventing dispersal we had destroyed the ability of the population to
regulate its density at a level below that of gross starvation.
Few studies have been made of vole populations in large enclosures
in the field. Gentry (1968) observed that M . pinetorum in a two-acre
enclosure reached densities far above those in natural habitats. He
312
CHARLES J. KREBS AND JUDITH H. MYERS
1965
1916
1967
FIG.25. “Fence-effect’’ in Microtus ochrogaster. Grid A is an unfenced control
population, grid D is a two-acre fenced enclosure, which increased to a maximum
of 411 voles, 5.5 times the highest control density. The habitat on grid D was
destroyed by overgrazing during the winter of 1966-1967. (After &ebs et al.,
1969.)
attributed this “fence-effect” to a restriction of dispersal and the
addition of food as trap bait.
Studies of enclosed populations suggest that dispersal may be a key
factor in determining population trends in voles and lemmings. We can
envisage two ways in which dispersal might be important to a population. First, dispersal may act as a safety valve for the population, and
dispersing voles may normally be killed by one hazard or another.
When population density becomes high more and more animals might
emigrate and die, so that the decline phase might be associated with
much emigration. Second, dispersal might act selectively in such a way
that the quality of dispersing voles differs from the quality of residents.
A number of relevant qualities might be involved: ability to avoid
predators, ability to utilize certain food plants, or aggressiveness,
Selectivedispersal would be more important early in the population cycle
and, in contrast to the first mechanism, dispersal in the increase phase
might be most important for its qualitative effects. Instead of dispersers
representing a random sample of the population which wilI be eliminated
selective dispersal during the population increase could change the
quality of the population resident at peak densities. I n order to obtain
POPULATION CYCLES IN SMALL MAMMALS
313
some information on these possible mechanisms, we must measure
dispersal rates in fluctuating populations.
The measurement of dispersal rates is relatively simple in principle
but few workers have tried to monitor dispersal during a population
cycle. The “death rate” measured in live-trapping studies is more
properly called a loss rate, since individuals which emigrate are counted
in the same way as ones which die. Only one study has attempted to
separate loss-by-emigration from loss-by-death i n situ. Myers and
Krebs (1971b) maintained two grassland areas free of voles for two
years and measured the amount of colonization which occurred from
adjacent control areas. Some voles (Microtus pennsylvanicus and
M . ochroguster) which disappeared from the control areas turned up as
immigrants on the vole-free areas and hence we could obtain a minimum
estimate of the proportion of the mortality given in Table I X which was
TABLEX
Percentage of losses known to be due to dispersal for two control popdations of
Microtus pennsylvanicus in southern Indiana. Dhpersing voles were picked u p a8
they colonized vole-free arem. Total number lost in parentheses. (After Myers and
Krebs, 1971b.)
Phase of cycle
Males
Females
Increase phase
Peak phase
Decline phase
56% (32)
33% (157)
15% (53)
69% (16)
25% (127)
12% (42)
loss-by-emigration. Table X gives these results, and shows that lossesby-emigration are proportionally largest in the increase phase and
smallest in the decline phase. Consequently, the high mortality rates of
adult voles in the decline are associated with death i n situ rather than
with dispersal.
There are several criticisms which can be made of this single study,
and more attempts must be made to measure dispersal losses before
we can reach any general conclusions about the relationship between
mortality and dispersal losses. First of all, in this study it was necessary
for the dispersing voles to remain in the vacant habitat a sufficient
length of time so that they could be caught in traps (maximum two
weeks). It is possible that other dispersers existed which were not
attracted to the vacant habitat and therefore could not be monitored.
Thus there may have been a set of dispersers which were influenced
by the peak population densities and emigrated but they were not in
search of a new, less crowded, suitable habitat. These may be thought
of as “pathological” dispersers most certainly to suffer high mortality
314
CHARLES J. KREBS AND JUDJTH H. MYERS
rates. I n order to identify this potential type of disperser i t would be
necessary to catch every animal leaving a population. This could be
done by monitoring egress from a semi-enclosed population. No one
has done this yet.
Immigration is more difficult to measure than emigration because of
the difficulties of live-trapping voles and lemmings. It is impossible to
know that you have trapped and removed every single individual from
an area. Thus, new individuals which appear on a live-trapping area
may have been the offspring of females which had avoided being
trapped. On the other hand, they may have moved in from adjacent
areas. Some method of radioactive marking of pregnant females might
be used to get at this problem. If it were possible to label radioactively
all young being produced in surrounding areas, dispersers from these
areas could be positively identified. Genetic markers might be used in a,
similar way. What we would like to determine is the exact source
area of each immigrant. So far nothing has been done along these lines.
We expect that the results of measuring immigration would not be the
converse of those for measuring emigration. Vole populations should be
closed to most immigrants, at least after the increase phase is over, so
that emigrants from one area will not usually be able to colonize an
adjacent area, except if population densities are low or if it is a marginal
habitat or is otherwise vacant for historical reasons. This discussion
leads to another suggestion for studying dispersal: artificial immigration
could be used to measure the ability of a population to absorb immigrants at different phases of the population cycle.
I n summary, dispersal is the least studied process in the population
equation for voles and lemmings. Studies on enclosed populations
indicate that numbers reach abnormally high levels when dispersal is
stopped. Almost no one has attempted to measure dispersal rates over a
population cycle. A single study showed highest dispersal during the
increase phase and almost no dispersal during the population decline.
Dispersal may change a population qualitatively as well m quantitatively. This idea will be elaborated in a later section discussing the
possible role of genetic changes in causing microtine cycles.
D.
GROWTH
The growth of individual animals in cyclic populations is important
because it is tied to the primary processes of birth, death and dispersal.
The size at sexual maturity is the most direct linkage between individual
growth and the reproductive rate of a population.
Growth can be measured in many ways but the simplest measurements are changes in weight or length. Length is a good measure of
315
POPULATION CYCLES IN SMALL MAMMALS
size since it is a measure of skeletal development, while weight is a
measure of robustness and a relatively poor measure of size. Weight
is easy to determine for live animals in the field and it is also easily
standardized among different observers. Length, by contrast, is more
difficult to measure on live animals and almost impossible to standardize among observers (Jewel1 and Fullagar, 1966). The ideal study
would consist of one observer measuring weights and lengths on all
individuals, but in most cases only weight data are collected.
One of the generalized features of population cycles of rodents is
that animals in peak populations are much larger than those in other
phases of the cycle. This feature was first recognized by Chitty (1952)
for Microtus agrestis. Table XI gives some representative figures for
TABLEX I
Mean body weight ( f 1 standard error) of adult m l e volea and lemminga at the
atart of the breeding season in different phaeea of the cycle
Microtua
agreetial
(May)
Increase phase
Peak phase
Decline phase
1
*
4
28.3 f 1.0
34-1 f 1.2
18.9 f 0.9
Microtw
Lemmw
trimucronatua2 arvaliaa
(spring)
(1&30 June)
50-2 f 1.8
79.3 f 2.5
6 1 . 6 k 2.1
22-6
24.4
22-4
Microtw
calqoornicu+
(October)
52+2
64k 1
60+ 1
Area 0, 1957-1960, from Chitty and Chitty (1962, Table 4 ) .
1959-1961, from Krebs (1964a, Table 45).
1951-1953, from Stain (1957, Table 10).
1966-1968, from Batzli and Pitalka (1971, Fig. 3).
changes in mean body weight with changes in density for these microtine species, and Fig. 26 illustrates changing body weight distributions
for a M . ochrogmter population. There are three ways in which this
change in body size associated with density could be produced. First,
voles may simply live longer in the increase and peak phases and
consequently achieve the maximum of their growth potential. Second,
voles may grow faster in the increase and peak phases then in the
decline phase, so that animals of equal age are larger in increasing
and peak populations. Third, growth rates of juvenile and subadult
voles may be the same in all years of the population cycle but asymptotic
weights of adults may vary with cyclic phase. Any one or a combination
of these three mechanisms could produce the observed heavy-weight
individuals of peak populations.
We can eliminate the first explanation as a sufficient one. I n Microtzls
agrestis Chitty (1952) has shown that larger voles of the peak phase are
316
CHARLES J. RREBS AND JUDITH H. MYERS
1961
1966
1967
FIG.26. Body weight distributions for snap-trapped samples of Microtus ochrogaster from southern Indiana. These populations increased in 1965, peaked in
1966, and declined in 1967. Winter months are shaded; one small square equals
one vole. (After Keller and Krebs, 1970.)
the same ages as smaller voles of the decline phase. Zimmermann
(1955) has shown that size changes in M . arvalis populations are not
simply changes in age composition. Krebs (1964a) reported that in the
lemmings Lemmus trimucronatus and Dicrostonyx groenlandicus the
heavy animals of the peak year were on the average younger than the
light animals of the decline phase.
The second and third explanations are difficult to separate with the
available data. Growth rates are higher in increasing and peak populations of Microtus pennsylvanicus (Fig. 27) and M . ochrogaster (Krebs
et al., 1969), and these results support the second explanation. The
same relationship was found in M . californicus but was confounded
with seasonal and reproductive effects on growth (Krebs, 1966). Unfortunately there are no data available on growth rates for species of
Clethrionomys or Microtus which do not breed during winter and yet
fluctuate cyclically.
The third explanation of a variable asymptotic weight could be
investigated if a sufficient number of measurements on individuals
taken over time were available. Several authors have recognized that
the growth curves of spring-born voles differ from those of autumnborn voles. Reichstein (1964) recognized two patterns for Microtus
arvalis. Voles born from March to June increase rapidly in weight
(to a maximum of 47 g) and become sexually mature. Voles born from
POPULATION CYCLES I N SMALL MAMMALS
317
June to October increase in weight only to 15-22 g and remain all
winter at these low weights. Chitty (1952) observed the same general
pattern for M . agrestis, and Kalela (1957) reported it for Clethrionomys
rufocanus. But while these seasonal variations have been clearly
described, few have tried to relate individual growth curves to density
changes.
Anderson (1970) emphasizes the distinction in growth and maturation
between spring- and fall-born animals, and refers to them as nearly
separate “generations”. There is little justification for such a clear-cut
1
$
-01I1
0
1
1
1
10
1
20
1
1
30
BODY ‘WEIGHT
t
1
40
1
,
50
-G
FIG.27. Instantaneous relative growth rates of Microtus pennsylvanicus males
from southern Indiana in relation to body weight. For increase phase, n = 691;
for peak phase, n = 1898; for decline, n = 333. The slopes of the three regression
lines are significantly different (p < 0.01). (After Krebs et al., 1973.)
distinction between “spring and summer generations” and “autumn
generations”. Clarke and Forsyth (1964), for example, document large
differences in sexual activity among fall-born Microtus agrestis of
different years. I n M . ochrogaster and M . pennsylvanicus in Indiana
the breeding season continues most of the year and there are several
generations in the spring and summer. We do not see how the weight
changes associated with population cycles can be explained by the
seasonal growth patterns associated with spring-born or fall-born
young. To discuss the population increase of cyclic rodents as instances
of exceptional years in which the spring and summer generation
survives and continues to the fall generation (Anderson, 1970) merely
begs all the questions we have been trying to answer. We still have to
ask why in some years the first summer generation is able to survive
while in others it is not.
318
CHARLES J. KREBS AND JUDITH H. MYERS
Differences in body weight have not been the only criterion by
which one could recognize rodents from peak populations. Zimmermann
(1955) found that mandible lengths in Microtus arwalis changed over
the population cycle in the same way that body weight changed.
Krebs (1964b) investigated the relationships between body size and
skull size in brown and varying lemming populations from northern
Canada. Lemmings were larger in peak populations, when measured
by body weight, body length, or skull dimensions. But surprisingly
the relationships between skull and body measurements changed
systematically in relation to population density. Lemmings of a given
I8O
1
160
-
LEMMUS
Males
E
E
f
D
140-
c
0)
_I
0 120c
P
100 -
8
0
1
,
,
22
I
,
24
I
,
1
26
Condylobosal
I
,
,
30
28
Length
,
,
I
32
I
,
34
mm
FIG. 28. Relationship between skull length and body length in male brown
lemmings from northern Canada. The regression line which fits the measurements
from the peak phase of summer 1960 did not fit the measurements from the
decline phase of summer, 1961. The position of the regression line moves up and
down the graph as the population density fluctuates. (After Krebs, 1964b.)
body size did not have the same skull size at different phases of the
population cycle (Fig. 28). These changes in skull-body relationships
are significant because they might be evidence for genotypic changes
over the population cycle. No one has repeated these observations for
any other rodent species, and we do not know how general such a
pattern might be.
High body weights in the peak breeding season were considered to
be characteristic of all rodent cycles by Krebs (1964a).A single exception
has been found. Fuller (1969) followed a population fluctuation in
POPULATION CYCLES IN SMALL MAMMALS
319
Clethrionomys rutilus and C . gapperi and found no change in mean
body weight. Fuller presents weight data for two years only, the peak
and decline summers for C. gapperi and two apparent peak summers
for C. rutilus. Further data are needed for increasing and low populations,
but it is puzzling that he found no differences in the declining population of C. gapperi. Elliott (1969) claimed that C. gapperi populations do
not fluctuate cyclically in most of their distributional range. If high
populations of C. gapperi represent irregular irruptions rather than
regulm cycles, we might use the high body weight criterion to distinguish
these two classes of population fluctuations. Regardless of our classification scheme, however, it would seem important to find other microtine
populations which do not behave as predicted.
Another approach to the study of growth over a rodent cycle is
to compute “indices of condition”. LeCren (1961) proposed a relative
condition factor obtained as the ratio
observed weight
weight predicted from body length
Condition factors of this type have been widely used in fish population
studies. We have tried to use LeCren’s index of condition to investigate
fluctuations of Microtus pennsylvanicus and M . ochrogaster in southern
Indiana (unpublished data). Snap-trap samples were obtained over six
years, and body length and weight were taken during standard
autopsies (Keller and Krebs, 1970). We pooled all the data to calculate
the body weight (Y)-body length (X) regression for each species, and
then referred individual voles to this common regression to get the
predicted weight. Figure 29 shows our results for M . ochrogaster,
which reached peak densities in 1966 and 1969 in our study areas.
It is apparent that there are large changes in “condition’) of voles from
year to year. We could detect no clear trends related to density,
however. Condition was “poor” in increasing populations in 1965 and
<<
average’’ to “good” in declining populations in 1967. There was no
clear seasonal trend, and this may reflect the relatively mild winters
of southern Indiana. We concluded from our analysis that there were
real differences from year to year in relative condition but these differences were not related to cyclic density changes.
There is an array of more sophisticated methods for determining
relative condition of rodents. Krebs (1964a) used an arbitrary fat
index to judge the amount of stored fat on lemmings (Lemmw and
Dicrostonyz). There was no relation between the amount of fat which
lemmings had stored and population density. Batzli and Pitelka (1971)
extracted the fat from Microtw californicus carcasses and also were
not able to associate levels of stored fat with population changes.
320
CHARLES J. KREBS AND JUDITH H. MYERS
1.10
z
f
lc5
0
0
U
O 100
Ez
09S
--
I
1
1965
~~
1966
1967
1868
1969
1970
FIG. 29. Index of relative condition (observed weight/predicted weight) for
Microtus ochroguster in southern Indiana. Both sexes combined, snap trap
samples. Populations peaked in 1966, declined in 1967, and peaked again in 1969.
There is no apparent relation between relative condition and density changes.
(Krebs and Myers, unpublished data.)
Therefore, the increased body weights characteristic of individuals
from peak populations are not simply the result of additional stored
fat.
We will not try here to review possible explanations for the growth
differences described. They are part of the syndrome of changes for
which any satisfactory theory must somehow account.
To summarize, high body weights have been associated with peak
population densities for a variety of voles and lemmings. These large
animals are not simply older animals. Growth rates are higher for
individuals in increasing and peak populations for the few voles for
which we have detailed data. Whether individuals from different
phases of the population cycle have different asymptotic weights
remains unclear. Growth differences occur not only in body size but
also in skeletal proportions.
VI. HYPOTHESES
T O EXPLAIN
POPULATION
CYCLES
We have described the demographic characteristics of microtine
cycles. We will next review hypotheses which have been proposed to
explain these demographic and density changes.
A.
FOOD
The lemming cycle, according to Lack (1954), was due to the overexploitation by the lemmings of their habitat with destruction of the
food and cover resulting in greater exposure to predators. Thus, the
POPULATION CYCLES IN SMALL MAMMALS
32 1
lemmings were thought to bring about their own demise by eating the
vegetation which provided cover and protection while the predators
acted as the agents of mortality. Pitelka (1958) reformulated the food
hypothesis and suggested that the lack of food brought about by high
density lemming populations led to malnutrition and reduced reproduction, and thus a population decline.
To analyze the relationship of microtines to their habitat we will
consider three questions: (1) Are microtines selective in their food
choices? (2) What is the effect of microtine grazing on the habitat?
and (3) Does the quantity or quality of the food supply become limiting
to increasing microtine populations?
1. Selectivity of microtine food habits and habitats
Microtines live in a variety of habitats from woodland (Clethrionomys)
to meadows, grasslands and old fields (Microtus, Symptomys, Pitymys)
and to alpine meadows and tundra (Microtus, Lemmus, Dicrostonyx).
For the grassland and tundra dwellers food and cover are provided by
the same plants.
Various workers have raised the question of the selectivity of
microtines in choosing food plants from their habitat. The preferred
food plants of Microtus pennsylvanicus in Minnesota were found to be
clover and dandelion, neither of which are common in the natural
habitat of this animal (Thompson, 1965). Introduced grasses are
readily eaten by M . ochrogaster, M . pennsylvanicus (Thompson, 1965;
Zimmerman, 1965) and M . californicus (Batzli and Pitelka, 1971).
Therefore, microtines appear to be very catholic in their food habits
and they generally take what is most rea’dily available (Martin, 1956),
often introduced grass species. However, Godfrey (1953) reported that
Helicotrichon pubescens occurred more often in fecal pellets of Microtus
agrestis than would be predicted from its abundance in the habitat,
and Batzli and Pitelka (1971) also had some evidence of food preferences
of M . californicus. Fleharty and Olson (1969) found that, although
availability of food types was an important factor, M . ochrogmter
showed some selectivity based on the growth stage and the palatability
of the food plants. Kalela and Koponen (1971) state that lemmings
show preference in the types of mosses they eat, particularly favoring
those of the genus Dicranum.
Thompson’s studies of food preference of Microtus pennsylvanicus
showed that plants from old field habitats were more acceptable than
those from marshes, tall grass prairie or boreal forest, and the developmental environment of the individual Microtus did not influence its
food preference. Also Thompson found that the quality of the soil
on which a plant was grown did not influence its acceptability to M.
322
CHARLES J. KREBS AND JUDITH H. MYERS
pennsylvanicus. Poa pratensis from gravel, clay and silt loam areas
was judged by Thompson to be unfavorable, suboptimal, and optimal
based on the color, vigor and succulence of the plants. Pieces of sod
with grass from these three areas were presented simultaneously
to Microtus pennsylvanicus. The percentages of the stems which were
clipped by the animals were 63, 63 and 58%. This test suggests that
microtines may not be very selective in choosing the quality of their
food.
Microtus ochrogaster in Indiana takes a wider variety of plant
species as food than does M . pennsylvanicus, and this is correlated with
the greater diversity of plant species in habitats where M . ochrogmter
are common (Zimmerman, 1965). Batzli and Pitelka (1971) confirmed
that for M . californicus the species common in the habitat were also
common in the diet.
If food supply is a critical influence on the dynamics of rodent
cycles, we might look for differences between microtine species with
different food habits. An attempt at assessing the relation of voles
and their habitats to population phenomena can be made by comparing
Microtus, which are grassland dwellers, to Clethrionmys, which live
in scrubby or wooded areas. Comparisons of food habits of C. rutilus
and M . oeconmus dwelling in a white spruce forest near College,
Alaska were made by Grodzinski (1971). Berries, fruits, tree seeds,
fungus and lichens were preferred by the red-backed voles (C. rutilus),
while greens were most preferred by the tundra vole ( M . o e c o n m w ) ,
although berries, fruits and seeds were also taken to a large degree.
The results of laboratory preference tests agreed well with those of
stomach content analyses. But all species of Clethrionomys are not
similar in their feeding habits. Tast and Kalela (1971) state that
C. rufocanus is a greens eater while C. rutilus is a seed eater. All species
of Microtus seem to feed on greens.
That Microtus and Clethrionomys compete where sympatric has been
demonstrated by observations that when only one species is present,
for example on an island, it will invade the habitat usually occupied
by the other species (review in Morris and Grant, 1972). Also experimental manipulations have been carried out to show that M.pennsy1vanicus tends to exclude C. gapperi from grasslands and C. gapperi
tends to exclude M . pennsylvanicus from wooded areas. Therefore,
although their preferred habitats are different, there is some overlap
in the ecological requirements of these two species.
Clethrionomys populations tend to exist at lower densities than
Microtus populations (Table XII). Some but not all species of Clethrionomys seem to fluctuate in regular cycles, but whether fluctuations are
related to feeding habits is not clear. One of the longest series of popda-
POPULATION CYCLES IN SMALL MAMMALS
323
TABLEXI1
Peak population densities in Clethrionomys
Density
30/acre
1O/acre
Approx. 30/acre
Approx. 30/acre
lO/acre
14/acre
Approx. 30/acre
Approx. 70/acre
20-40/acre
Species
Author
C. rutilus
C. rutilus
C. rutilus
C. rutilus
C. gapperi
C. gapperi
C. glareolus
C . glareolus
C. glareolus
Whitney (unpublished)
Fuller (1969)
Koshkina (1965)
Pruitt (1968)
Fuller (1969)
Elliott (1969)
Ashby (1967)
Newson (1963)
Petrusewicz et al. (1971)
tion data on a microtine is that of Koshkina (1966) shown in Fig. 3.
These data are for C. rufocanus (a greens eater) and show regular
population fluctuations. Density data for C. rutilus (a seeds eater) in
the Russian taiga are included in Fig. 16, based on the data of Koshkina
(1965). Of the seven years of the study (1958-1964) two years of peak
abundance occurred, 1960 and 1963. Pruitt (1968) collected density
data for C. rutilus for seven years near Fairbanks, Alaska and observed
peak populations in 1954 and 1959 (Fig. 30). Others have found population fluctuations of Clethrionomys but with no clear cyclic pattern. For
YEAR
FIG.30. The number of Clethrionomys mrtilus trapped on a one-acre plot of taiga
forest near Fairbanks, Alaska. These estimates come from a single trapping in
September of each year. Data from Pruitt (1968).
324
CHARLES J. KREBS AND JUDITH H. MYERS
example, Whitney (unpublished) (Fig. 31) had peak densities of about
20 to 30 C. rutilus per acre in each autumn of three years of his study,
but during one summer densities were very low until late August while
in other summers densities were somewhat higher. Elliott’s (1969)
study of C. gapperi showed one year in which the population maintained very low densities through the summer. An island population
1968
1969
I970
1971
1968
1969
1970
1971
FIG.31. Density and survival rates of Clethrionomgs rutilus in College, Alaska.
Density declines and poor survival occur in the fall before freezing temperatures
and in the spring in association with the beginning of the breeding eeason. Data
from Whitney (unpublished).
of C. glareolus was followed by Petrusewicz et al. (1971) for three years.
Densities in the first and third summers were markedly higher than that
in the second. Tast and Kalela (1971) claim that C. rufocanus cycles in
synchrony with Microtus oeconomus, M . agrestis, and Lemmus lemmus
in Finnish Lapland, and provide combined data for the densities of all
four species for a nine-year period, with peak densities in 1964 and
1969 (Fig. 32).
325
POPULATION CYCLES IN SMALL MAMMALS
All studies indicate that Clethrionomys reaches the annual peak
late in the fall. For this reason we must be suspicious of data which
are obtained from only one or two trappings a year or from field
samples collected for only a short summer period.
There are only three records of winter breeding in Clethrionomys,
and this genus would seem to differ from lemmings and Microtus in that
winter breeding is rare. No one has investigated why Microtus is able
to breed in some winters while Clethrionomys usually does not. Evernden
1963
1964
1965
1966
1967
1968
1969
1970
1971
YEAR
FIU. 32. The relation of microtine densities to plant production (number of
Eriophoruwa angustifoliuna shoots per 75 sq. m) in Finnish Lapland. Data from
Tast and Kalela (1971).
and Fuller (1972) have indicated that the beginning of the breeding
season in C. gapperi is controlled by day length. Snow is a very effective
light filter, and we do not know whether C. gapperi monitors the light
levels in spring by short excursions above the snow surface. But as
shown previously (p. 295), Clethrionomys are capable of winter breeding.
I n Whitney’s (unpublished) study breeding began in a C. rutilus
population in March 1971 when the snow depth was 120 cm and it was a
month before the snow melted to a depth of 20 cm. Also Whitney’s
study showed great variability in the date of onset of the breeding
season as follows: April 1969; June 1970; and March 1971.
Pinter and Negus (1965) review studies of photoperiod and reproduction on Microtus species. I n their own work they showed that the
food quality, specifically young growing shoots of oats, had a greater
influence on reproduction in M . montanus than did photoperiod.
326
CHARLES J. KREBS AND JUDITH H. MYERS
Data are not available for a detailed comparison of Microtus and
Clethrionomys. We might predict that Clethrionomys with only rare
winter breeding would fluctuate less than Microtus which can breed all
year. Also Clethrionomys should have a lesser impact on its woodland
habitat than the grassland-dwelling Microtus, which again might act
to buffer fluctuations. There are no data available to test these simple
suggestions. Clethrionomys show many of the demographic characteristics of fluctuating microtines (see section V). Whitney’s work shows
that in the same study area, C . rutilus reached similar densities in the
autumns of the three years of the study, while Microtus oeconomus
reached peak densities in one summer and declined to extremely low
densities thereafter. Studies comparing sympatric and allopatric
populations of Microtus and Clethrionomys may help us to determine
more specifically those characteristics which distinguish cycling microtine populations and to determine the influence of habitat on population
phenomena.
2. The effect of microtine grazing on the food supply
Do foraging activities of microtines remove a significant proportion
of the food supply? Thompson (195513) and Pitelka (1958) report
extensive forage utilization preceding declines of brown lemming
populations at Barrow, Alaska. However, Chitty (1960) and Krebs
(1964a) reviewed studies in which neither habitat destruction nor
starvation were observed to be associated with declining vole and
lemming populations.
A number of studies have been carried out to measure the percentage
of the available food energy which is consumed by microtines (Table
XIII). With two exceptions the percentage of available energy conTABLEXI11
Percent of available food energy consumed per year by small mammals in a variety of habitata
Ecosystem
Pinewood
Mazury Lakeland,
Poland
Oak-pine forest
Mazury Lakeland,
Poland
Mixed and deciduous
Kompinos Forest,
Poland
Principal
small mammals
yo Available
food consumed
(kcal)
Reference
Clethrionomys glareolus
Apodemus Jlavicollis
0.6-1.9
Ryszkowski (1969)
C. glareolus
A . Jlavicollis
0+0+3
Ryszkowski (1969)
C. glareolus
A . Jlavicollis
A . agrarius
0.6
Ryszkowski (1969)
POPULATION CYCLES IN SMALL MAMMALS
327
TABLEXI11 (contd.)
Ecosystem
Principal
small ma&als
yo Available
food consumed
(kcal)
Reference
-
Oak-hornbeam
Cracow, Poland
Beechwood
Ojc6w, Poland
Spruce plantation
Bjornstorp,
South Sweden
Forest plantation
on peat bog August6w
Forest, Poland
Alpine meadow
Bieszczady Mountains,
Poland
Cultivated plants
South-easternPoland
Old field
Ingels, South Finland
Cultivated fields
and refuge habitats,
Poland
Old field
Alaskan Taiga
Forest
C . glareolus
A . jeavicollis
C . glareolua
A . Jlavicollb
Microtus agreati8
4.6
Grodzinski (1971)
3.9
Grodzinski et al. (1969)
16-24
Hansson (1971)
M . oeconomus
3.1
Gebczynska (1970)
Pitymy8 subterranew,
M . agreatis
1.03
Grodzinski et al. (1966)
M . arvalb
1-20
Migula et al. (1970)
M . agreatb
3-14
Myllymiiki (1969)
M . arvalb
0-6-1-3
Trojan (1969)
M . penmylvanicw,
C . rmtilw, M . omnomua
Tamkzaciecrme
hudaonicua, ~ h w o m y s
aabrinw,, Sorex
cinereua
04-1.6
3-47
Golley (1960)
Grodzinski (1971)
sumed by the microtines has been estimated to be less than 5% of the
available net primary production. The opposite side of the coin is to
determine the influence of microtines on the production of the habitat.
Schultz (1965) measured grass production in a tundra area from which
lemmings were excluded and found that over seven years of observation,
the production by vegetation in the exclosures declined (Fig. 33).
Outside in grazed areas the production waa almost always greater than
in the exclosure although there was some vaziation from year to year.
However, this relationship waa not simple since the exclosures were
actually built in 1949 and during the fist four years (1960-1963) the
measured production of the vegetation averaged 600 pounds per acre.
The standing crop of the tundra vegetation was measured by
Schultz (1965) and these measurements showed the same variation as
did the production data. While years of high lemming density seemed
328
CHARLES J. KREBS AND JUDITH H. MYERS
75
2
U
0
&
50
P
P
.-E
P
3
0
n
..- 251
C
52
c
36
U
ii
3
2
B
.-C
3
L
c
20
-al
YEAR
FIG.33. Primary production in exclosures (dotted line) decreased as a result of
no lemming grazing. Primary production in open tundra fluctuates and is
moderately related (r = -0.69) to lemming density (indicated by bars). Production data from Schultz (1966) density data are maximum average number of
lemmings caught per trapline of the two or three trapping periods per summer
from Pitelka (1972).
to decrease the standing crop there was no significant relationship
between the measured standing crop of the vegetation and the density
of lemmings (r = -0.25, n.8.).
Estimates of the proportion of the standing crop taken by peak
microtine populations have varied from low values of 15% reported by
Krebs (1964a) for lemmings and 38% by Batzli and Pitelka (1970)
for Microtw californicus, to high values of 50% by Thompson (1955b)
and 50-90% by Schultz (1964) for lemmings,
Grazing by microtines has been shown to influence the plant species
composition of the habitat. Summerhayes’ (1941) study of M . agrestis
showed that grazing caused reduction in the grasses and allowed
flowering plants and mosses to be established. Exclosures which
prevented grazing by the California vole permitted Batzli and Pitelka
(1970) to observe an 85% reduction in the volume of major M .
californicus food plants by grazing of a peak population (160 per acre),
POPULATION CYCLES IN SMALL MAMMALS
329
and a 70% decrease in the seed fall of preferred grasses in grazed
areas.
Kalela (1962) suggested that microtine density fluctuations were
coupled to variation in plant production resulting from weather
conditions. The emphasis on the climate as the underlying variable
leading to microtine cycles arises from the geographically widespread
synchrony observed in small mammal density fluctuations. Rodent
cycles are viewed by Kalela as being the result of random oscillations of
weather conditions which influence the flowering frequency of food
plants and variations in their nutritional state, as well as inherent
rhythms of the food plants and rodent populations. To investigate this
association Tast and Kalela (1971) accumulated density data on four
species of microtines (Lemmus lemmus, Microtus agrestis, M . oeconomus,
Clethrionomys rufocanus) and two of their preferred food plants,
narrow leaved cotton grass, Eriophorum angustifolium, and golden-rod,
Solidago virgaurea.
The microtine populations, estimated by two trappings a year with
snap-traps, showed peak densities in the fall of 1964 and again in the
fall of 1969 (Fig. 32). While the number of shoots of the cotton grass
was highest in 1964 in synchrony with the peak microtine density, in
1969, the year of the second microtine peak population, the number of
cotton grass shoots was no higher than it had been during the intervening years of low microtine densities. The number of shoots of
Solidago showed little change between 1965, when quantification of this
plant began, and 1969, but the year after the microtine peak there was
a small increase in the number of shoots of this plant.
Exclosures were used by Tast end Kalela t o show that microtines
had no influence on the number of flowering Eriophorum plants, and
through the study there was a decrease in the number of flowering
shoots of Eriophorum because of human disturbance. The microtines
still reached peak densities in 1969 even though the number of flowering
shoots of Eriophorum was about one-Hth that of the 1963-1964 peak.
The number of flowering shoots of Solidago was the highest in 1966, L
year of very low microtine density, and 1969, a year of peak microtine
density. While Tast and Kalela (1971) report positive correlations
between food conditions and rodent population fluctuations, none of
the correlations were significant.
A more recent paper (Tast, 1972) has related mean winter weights
of Microtus OeconOmuS to the food supply as indicated by the density
of sterile shoots of Eriophorum (Fig. 34). A single point gives the
impression that there might be a relation between these factors. The
sample size on which this point is based is four animals. More data are
obviously required.
330
CHARLES J. KREBS AND JUDITH H. MYERS
10
20
30
Mean Body Weight
FIQ.34. The mean winter weights of Mkrotu8 oeconomu8 in relation to available
food (no. of Eriophorum shoots the previous autumn). This apparent relationship
is due to a single point (starred) for which the sample size was 4 animals. Data
from Tast and Kalela (1971) and Tast (1972).
Vole and lemming populations can affect the amount of vegetation
and its species composition, but they might also affect the nutrient
composition of the vegetation. Relatively little work has been done
on this aspect, and most of the available data is from the Barrow,
Alaska studies of the brown lemming.
The levels of phosphorus, nitrogen, calcium, magnesium, potassium
and sodium in the forage of lemmings in Barrow, Alaska were monitored
by Pieper (1964) to elucidate the relationship between the lemmings
and their habitat. Forage, clipped from ungrazed exclosures, provided
controls for determining the effect of the lemmings on the nutrient
levels of their habitat. Pieper found that in exclosures without lemmings
there was little fluctuation from one year to the next in the nutrient.
levels in the forage, and that, as Schultz had found with the production
of the vegetation, prevention of grazing lowered the nutrient content
of the plants. Lemmings obviously stimulate the nutrient cycles on the
tundra, and hence there is a correlation between lemming density and
nutrient levels (Fig. 36). With the exception of magnesium, all other
nutrients were at highest concentrations in the forage in the summer
of the year of peak lemming density. Pieper (1964) interprets this
h d i n g as follows. I n the winter preceding the summer of peak lemming
density there was a rapid build-up in lemming numbers. Under the
snow these lemmings were depositing quantities of fecal pellets and
POPULATION CYCLES IN SMALL MAMMALS
331
urine. With the coming of the spring melt-off the nutrients contained
in the excretory products were readily made available to the plants:
thus the high nutrient levels. This relationship between grazing,
nutrient availability, and nutrient cycling explains the synchronous
cycles between lemming density and nutrient levels in Fig. 35. Lemmings
clearly do influence plant nutrient composition. Whether the variation
in the nutrient levels influences lemming numbers will be considered
in the next section.
Unfortunately there are no comparable sets of data for voles from
more temperate climates, and we do not know whether Microtus cycles
in temperate grasslands are affecting the nutrient composition of the
forage in the same way that lemmings obviously do.
YEAR
FIG. 35. Percentage phosphorus in the summer forage and brown lemming
densities (bars) in Barrow, Alaska. The horizontal line indicates the similarity
between phosphorus levels preceding and following years of peak population
density. Data for phosphorus levels from Schultz (1969), lemming density estimates are the maximum average number of lemmings caught per trapline of
the two or three trapping periods per summer from Pitelka (1972).
3. The injluence of food quality and quantity on microtine numbers
Four experiments have now been carried out in an attempt to test
the hypothesis that either the quality or the quantity of food available
to microtines is a factor responsible for their population fluctuations.
I n the first of these Hoffmann (1958) added ammonium nitratephosphorus fertilizer to a meadow in the Sierra Nevada inhabited by
332
CHARLES J. KREBS AND JUDITH 11. MYERS
Microtus montanus. At the time of the fertilizer addition the Microtus
population on the experimental area was at a lower density than the
control population, but while the control population declined through
the summer, the experimental population maintained its lower density
and declined during the next winter. The fertilizer did not increase
the amount of protein in one of the favored food plants of the voles,
Carex sp., and there was no relation between the protein content of
Carex and the density of voles supported, in various meadows studied by
Hoffmann. Thus, no relation between quality of food and vole density
could be shown and the addition of fertilizer, while possibly delaying
the vole decline for a few weeks, did not prevent it.
A six-acre fertilized plot with three to four times the net primary
productivity of a control plot was established in the tundra by Schultz
(1969)and followed from 1961 to 1965. Protein, phosphorus and calcium
levels in the vegetation were raised four to five times by this experimental treatment. Only during one winter (1963), following a year of
relatively high lemming numbers, was the lemming population higher
on the fertilized plot and these animals were rapidly taken by predators
d b r the spring melt-off. A higher lemming density was not reported
for the “improved” habitat in years of high lemming density and so this
experiment does not provide evidence that the quantity or quality of
tundra vegetation is a limiting factor to lemming populations.
Both fertilization of the grassland and supplemental feeding were
tested by Krebs and DeLong (1965) as techniques for preventing
the decline of Microtus californicus. While the animals on the experimental area demonstrated high growth rates, good recruitment, and
increasing densities for the first five months of the experiment, all
of these factors suddenly began to decline. The control population
continued to increase to high numbers while the population with
supplemental food declined to a very sparse density. This experiment
was criticized by Batzli and Pitelka (1971) for the failure to take into
account the quality of the natural food which was available to the
voles, the nutritive quality of the supplemental food, or the possible
effect of feeding stations on the social structure of the population.
Finally, biological assay of the amount of food available to populations of short-tailed voles, Microtus agrestis and bank voles, Clethrionomys glareolus, showed that just previous to the population decline
there was sufficient food to maintain the populations for a year (Chitty
et al., 1968). A Microtus population given supplemental oats and
carrots did not increase in density and body growth rates remained
low, typical of declining vole populations. An interesting observation
in this study was that it was impossible to predict either from the body
weight or the condition of voles kept in 11-ft2enclosures when they had
POPULATION CYCLES I N SMALL MAMMALS
333
eaten out the natural food supply and were on the brink of starvation.
Three animals died without losing weight. This points out the fallacy
of observations that symptoms of starvation were not observed in
declining populations (Rauch, 1950, p. 176) and suggests that observations of this sort are not useful in judging the relation between an
animal and its food supply.
An alternative to providing supplemental food to a microtine
population is to determine if a habitat can in fact support higher
microtine densities than it naturally does. Two-acre enclosures of
grassland in southern Indiana supported densities of Microtus ochrogaster
and M . pennsylvanicus several times greater than did a similar habitat
just outside the fence. Although eventual habitat destruction resulted,
immediate recovery occurred at the beginning of the next growing
season, and introduced Microtus always responded with rapid population growth (Krebs et al., 1969, 1973). These experiments showed
that higher densities of voles could be supported by the Indiana
habitat than normally were. Even after extensive habitat utilization
there was no delay in recovery to forage vegetation sufficient for
growth and reproduction of Microtus populations.
A further test of the influence of standing crop, amount of food
available and percentage cover on microtine populations is provided by
the work of Batzli and Pitelka (1970, 1971). I n this study two populations of Microtus californicus were monitored, one in Richmond,
California bordering the San Francisco Bay and the second in the
Briones Hills, ten miles east of Richmond. Peak population densities
of M . californicus in these two areas were almost the same in 1967-1968
when the study was conducted and both populations showed similar
declines in the fall of 1968. However, these workers report that the
standing crop, average height of vegetation and the percentage cover
were all greater in the Briones study area, and the volume of Bromus
rigidus, a major component of the food of M . californicus, was ten
times greater at this area than in the Richmond plot. Unfortunately
the study was terminated just as the Microtus population decline began,
but it is obvious that in this case even ten times the amount of one
food plant did not allow a greater peak density nor prevent the
beginning of the population decline.
Field experiments have so far suggested that food limitation is not
the factor stopping the increase phase of microtine populations. But we
still must analyze the nutrient and lemming cycles described by
Schultz (1969) in terms of what is cause and what is effect. The
nutritional threshold hypothesis of Schultz is as follows: During the
summer of peak lemming population the nutrients in the forage
are at high concentrations, and high production and consumption
334
'
CHARLES J. KREBS AND JUDITH H. MYERS
concentrates nutrients in organic material so that they are unavailable
the next year to the growing plants. After the peak year, nutrient
levels in the forage are low-too low to permit adequate reproduction
by the lemmings. I n the next two years greater amounts of nutrients
are released and by the fourth year both the nutrient levels in the forage
and the densities of lemmings have recovered. The building block of
this hypothesis is that at some times the nutritional levels (particularly
phosphorus and calcium) are below the thresholds necessary for
reproduction by the lemmings.
Let us investigate this hypothesis by looking at the data on which
it is based (Fig. 35). The two years of peak lemming density have
corresponding peak phosphorus levels. But what is contrary to the
hypothesis is that phosphorus levels are the same in the years before
and after peak population densities. That is, the same levels of phosphorus give rise to increasing lemming population densities as are
supposed to be limiting to the reproduction of the lemming in the year
following the peak. Therefore, although data indicate synchronous
cycles of nutrients and lemmings, there is no indication of nutrient
levels falling below those which can give riee to increasing lemming
densities.
To determine if nutrients are related to reproduction in lemmings
we have plotted the proportion of females pregnant and the percentage
phosphorus in the forage for July and August of the three years for
which these data are available (Fig. 36). As is typical of cycling
microtines (see page 296), the breeding season in the year of peak
densities is shortened so that in August of 1960 when the nutrient
levels are high, the proportion of pregnant females is low. There is no
indication from these data that phosphorus deficiency limits reproduction in lemmings.
Calcium is another nutrient which is stressed by Schultz (1969)
as being important, and in this case it is the necessity of threshold
levels of calcium for adequate lactation by the lemmings which is
thought to be important. Mullen (1968) in his study of lemmings at
Barrow, Alaska monitored the onset and rate of mammary gland
development in pregnant females. He found that in 1960, when Pieper's
(1964) data show high concentrations of calcium and other nutrients in
the forage, the development of the mammary glands began later and
proceeded more slowly during pregnancy than in 1962 when calcium
concentration was low. So there is no indication of calcium limiting
lactation, at least as measured by mammary gland development.
One of the characteristics which seems almost always to be associated
with increasing lemming populations is that the lemmings breed in the
winter preceding the peak. From this observation one would predict
POPULATION CYCLES IN SMALL MAMMALS
335
that if winter breeding were prevented in most years because of
nutrient deficiency, the most important nutrient would have been
strikingly higher in the autumn preceding the summer of peak lemming
density (e.g. 1960). The concentration of sodium was twice aa high in
August of 1969 as it was in August of other years but this waa the
only nutrient to show this trend (Pieper, 1964). However, Schultz
(1969, p. 88) dispenses with sodium as a nutrient possibly important
to lemming cycles by the statement " . . .sodium show(s)no relationship
to lemming numbers at all, nor were the data cyclic". I n August,
@ 1960
-16
-24
-32
-40
-48
-56
Proportion Phosphorus in Vegetation
FIG.36. The relationship between the available nutrient and reproduction in the
brown lemming is shown from the proportion of females pregnant (Mullen, 1968)
and the percent phosphorus in the vegetation (Pieper, 1964). July data are solid
dots (e),August data (0).
1962, preceding a summer of intermediate lemming density, the sodium
level of the forage was the lowest ever recorded by Pieper, suggesting
that the high sodium before the 1960 peak lemming density may have
been merely a chance association. And the importance of sodium to
microtine cycles may be further questioned by our failure to find a
correlation between soil sodium and densities of Microtus ochrogaster
and M . pennsylvaniczls in southern Indiana (Krebs et al., 1971).
Lemmings, like other herbivores, probably select their forage,
and this complicates the comparison of lemming needs and nutrient
content of random forage samples. The nutrient levels reported in
these studies are for all the plant parts extending above the surface
of the soil. It is possible that nutrient levels in specific plant parts are
336
CHARLES J. EREBS AND JUDITH H. MYERS
much different than those recorded for the complete plant. Whether
lemmings can select plants or plant parts of high nutrient content,
even when average nutrient content is low, is unknown. However,
Thompson’s (1965) test of the selectivity of Microtus pennsylvanicus
indicated that they did not seem to recognize the quality of the food
plant.
To summarize, we must conclude that, contrary to the statement of
Schultz (1969, p. 86), the biology of the lemming is a very important
element for understanding the tundra ecosystem. Evidence confirms
Pieper’s (1964) hypothesis that lemming grazing actually increases
the nutrients available to plants, and that synchronous cycles between
lemming density and nutrient content in plants are the result of the
effect of lemming density on the availability of nutrients to the plants.
Thus lemmings affect the tundra plants but the plants do not seem to
determine lemming density changes. Data published thus far seem to
contradict the nutritional threshold hypothesis.
What other experiments might be done to test the food hypothesis?
Schultz (1969) has begun experiments by fertilizing six-acre plots in
the tundra. There are two problems with this experimental design.
As well as changing the quality and quantity of food, the amount of
cover is also altered dramatically by fertilization. This confounds any
interpretations unless cover can be changed on other areas without
changing the food supply. Secondly, six acres is a very small part of
the tundra. Immigration to the optimal habitat may be great, thereby
complicating the understanding of the demography of lemmings in the
fertilized area.
Studies using islands or enclosed populations could be very revealing.
Can the tundra, like a southern Indiana grassland, support more
animals than it presently does? Can an increasing lemming population
be established on habitat from which a peak population has just been
removed? One thing is clear: measuring lemming density alone is
not sufficient. We must have detailed information about birth, death,
and growth rates, or any environmental measurements are useless.
Furthermore, we need a greater understanding of how the microtine
actually interacts with its environment. We still have little idea of
what the nutrient requirements of microtines are, or how selective
they are in choosing the quality of their food. The nutrient levels
reported thus far are values for complete plants and the nutrients have
only been measured during the summer.
Nutrient cycling in the tundra is exaggerated by the thin layer
of soil above the permafrost and the slow rates of decomposition.
Therefore, the interactions between the lemmings and their environment are likely to be quite different from those of the Microtus in
POPULATION CYCLES I N SMALL MAMMALS
337
grasslands in California, England or New York (Haynes and Thompson,
1965). If we are seeking generalities of microtine cycles, as pointed out
earlier in this article, we must not overlook the uniqueness of the
tundra and we must avoid overemphasis of characteristics specific to
this situation.
To summarize, microtine rodents take only a small fraction of the net
primary production. Microtine grazing affects the plant species composition of the habitat, and at least in tundra ecosystems also affects
the nutrient composition of the forage. No one seems to suggest that
voles or lemmings are limited by the amount of food available, and
attention has turned to the quality of the food available. The nutrient
threshold hypothesis has been suggested as an explanation for lemming
cycles, but we can find no evidence that nutrient levels cause any of
the characteristic features of cyclic lemming populations.
B.
PREDATION
“Foxes also hunt them, and the wild ferrets in particular destroy
them; but they make no way against the prolific qualities of the animal
and the rapidity of its breeding.’’ These are the words of Aristotle
quoted by Elton (1942, p. 3) and the animal to which he was referring
was the field mouse. However, even though this opinion was expressed
in the early stages of the history of small mammal cycles, the question
whether predation might be the driving force behind population
fluctuations was by no means put to rest.
Two aspects must be separated in discussing predator-prey questions:
(1) How do the predators respond to variations in prey abundance? and
(2) Are these responses sufficient to explain population changes in the
prey? As with the food hypothesis, one of these two elements may be
present without the other.
Early writers were most taken by the correspondence between
fluctuations in populations of herbivores such as lemmings, voles and
rabbits, and of carnivores, such as foxes, weasels and raptors. Data
revealing oscillations of predators and their cycling prey was reviewed
by Elton (1942) and Lack (1954, pp. 204-226). Shelford (1943) considered the cycle of the varying lemming as being typical of predatorprey oscillations described by Lotka and Volterra. The prey species
increases in density providing more food for the predator, which responds with increased reproduction and a build-up in its own population to a level larger than can be supported by the prey population. A
decline in the prey population ensues, followed by a fall in the predator
population through death or emigration.
An extensive literature exists documenting the numerical response
338
CHARLES J. KREBS AND JUDITH H. MYERS
of predators to microtine populations. Gross (1947) summarizes the
history of snowy owl migrations from the Arctic into New England.
These migrations followed years of high lemming population densities.
The subsequent movement southward was triggered as food became
scarce when the lemming population declined.
Avian predators can be categorized as restricted feeders which depend
predominantly on one prey species, and general feeders which take a
variety of prey species (Craighead and Craighead, 1969, p. 182).
Restricted feeders travel about looking for high prey densities, and
during a year of high Microtus numbers a great influx of restricted
feeders into the Craigheads’ study area occurred, while populations
of general feeders remained about constant. Other reports of predator
movements in response to fluctuating vole and lemming populations
were made by Honer (1963) for the barn owl, Maher (1970) and Pitelka
et al. (1955) for the jaeger and snowy owl, Mysterud (1970) for the
boreal owl, and Stendell (1972) for the white-tailed kite. Some raptors
are only able to breed when the population of their prey is relatively
high (jaegers and snowy owls, Maher (1970) and Pitelka et al. (1955);
rough-legged hawk, Hagen (1969); and the white-tailed kite, Stendell
(1972)).
But given that predator species can respond to high densities of
voles and lemmings by increasing their own numbers, is mortality
caused by predation sufficient to regulate vole and lemming populations? Opinions are divided on this question. Some studies claim that
predators took a high proportion of the prey population and therefore
me a necessary factor in preventing population increase. Others claim
that predators had very little influence on the prey population (Table
XIV).
A modified version of the predator hypothesis was suggested by
Lack (1954). Lack concluded that predators are too scarce in proportion
to the prey at the time of lemming or vole peak densities, and hence
predators cannot stop the increase of breeding prey populations.
Therefore, it was necessary for another factor, such as food limitation or
intraspecific competition, to stop the microtine increase by shutting
off reproduction before the predators could catch up and exert any
influence on prey population density.
The first serious attempt to quantify the influence of predation on a
population of Microtus was made in 1951 by Brant (1962). This was
done by analyzing raccoon scats which were deposited on his study
area. Brant surprisingly found that twice as many M . californicus were
taken by predators than he had estimated to be present on the study
area. The source of error could be either poor estimation techniques
for Microtus 0; the concentration of predator scats from a larger
339
POPULATION CYCLES IN SMALL MAMMALS
TABLEXIV
The in$wnce of p&rs
on jluctwting populatiolzg of microtinee. The acoo(112t8
vary from impeaaiona of the in$wnce of the p&rs
to attempts to quantify the
percentage of the vole population a c t d y taken. The values given for the peddur
i n $ w m are minimum v a l w since genercclly a worker wa8 concerned with only one
type of predator. The studiee of mrid p & h generally do not c&r
terreatrid
p&h and wice versa
Prey species
Predator species
Snowy owl,
Pomerine jaeger
Snowy owl,
Pomerine jaeger
Snowy owl,
Pomarine jaeger,
Least weasel
Predator influence
Author
Extensive
Pitelka et al. (1966)
varying
Pitelka (1968)
Weasels and raptors
Density prey > %/acre.
Maher (1970)
No influence
Density prey < 26/mcre.
Depressing influence
Important during decline. Thompson (1966b)
Weasel influence during
period of low numbers
Ate 8-20y0 spring popln Watson (1967)
Ate 20-3 1yo summer popln
Little effect
Krebs (1964a)
Raccoons
Caused decline
Brant (1962)
Feral cats, raccoons
Ate 88, 26 and 33%
declining popln
Ate 28% declining popln
Pearson (1964, 1966,
Snowy owl,
Pomerine jaeger,
Least weasel
Snowy owl
Kites and shorteared owls
1971)
Stendell
(unpublished)
Microtus
motatanus
Weasels and ermine
Ate 40% decline
2 1 4 7 % low numbers
8% increasing
(winter poplns)
Fitzgerald (1972)
Microtus
Raptors
Little effect
Chitty (1962)
Elton (1942, p. 192)
Microtus
pmnsylvanicus
Raptors
Ate 26% high density
Ate 22% low density
Craighead and
Craighead (1969,
p. 321)
Microtus
Short-eared owl
Ate 8-61 yo declining
Lockie (1966)
POPln
No influence on breeding
POPh
Ogreeti.3
340
CHARLES J. KREBS AND JUDITH H . MYERS
hunting area to Brant’s study area. If the latter were the case, more
Microtus remains would be deposited on the area than were removed
from the area.
The most intensive work on predation has been done on Microtus
californicus by Pearson (1964, 1966, 1971). The technique of scat
analysis was used by Pearson to measure the impact of terrestrial
predators on M . californicus populations for parts of three population
fluctuations. An assumption underlying this technique is that predator
droppings are not concentrated on the study area. The best estimates
of the proportion of the Microtus population taken by predators was
obtained during the population decline, when reproduction by the
Microtus population was reduced or had ceased altogether. The vole
population can be estimated at the end of the breeding season and the
number of individuals taken from this “standing crop” by the predators
determined. As indicated in Table XIV, the percentage of the Microtus
population which was taken by terrestrial predators during the population decline was 88, 25 and 33% for the three cycles studied by Pearson.
Why was the predation pressure during the first Microtus cycle so much
greater than that of the next two cycles? One possible explanation
is that a program of feral cat eradication was carried out in the study
area after the first period of decline. Therefore, the predator population
was smaller during the last two population cycles (Table XV). The
rate of decrease of the vole population was very similar regardless of
the proportion of the population taken by predators.
The relation of predation to microtine populations proposed by
Pearson (197 1) contains the following points. First, avian predators
are not sufficiently intensive to determine population trends in rodents
since they leave the area when prey abundance is low. Mammalian
predators are the important agents of mortality because they are less
mobile, and therefore, when microtines are at low densities, they
TABLEX V
Comparative predation pressure by terrestrial predators (mostlyferal cats) on declining
Microtus californicus population8 for three periods of population decline. Data front
Pearson (1966, 1971).
1961
Microtusl
No.
carnivore carnivores
Mid-decline
Greatest predation
pressure
yo Loss to predation
130
72
88
8
10
1963
Microtwl
No.
carnivore carnivores
800
500
25
3.5
4-5
1965
Microtwl
No.
carnivore carnivores
600
224
33
2
2.3
POPULATION CYCLES IN ‘SMALL MAMMALS
34 1
supplement their diets with other species but remain on the area
exerting predation pressure. Secondly, predators do not stop the
increase of a breeding microtine population. Pearson proposes that
predators can be responsible for the amplitude of microtine cycles by
their ability to decrease the populations to very low levels through
continued predation pressure when the microtine populations are low.
Predators can also influence the periodicity of the cycle by prolonging
the period of low numbers. Note that according to Pearson’s ideas, the
critical period to study predation on microtine rodents is during the
late decline and the phase of low numbers. Unfortunately almost no
one seems to do this.
The influence of predation by ermine and weasel on a Microtus
montanus population was studied by Fitzgerald (1972) a t the University
of California research station on Sagehen Creek in the Sierra Nevada.
Vole populations and winter predation by weasels and ermine were
estimated by counting the number of winter nests made by voles (an
estimate of the vole population). The number of Microtus nests which
had been invaded by ermine and weasels as well as the remains of
Microtus left near these nests were counted as an indication of the
predators’ activities. A crucial assumption of this study is that one
vole nest found by Fitzgerald in the spring, indicates the presence of
one vole in the overwintering populations. However, more direct
estimates made by Fitzgerald in a live-trapping study, indicated a
possible ratio of one to three voles per nest. This ratio varied among
years. The formation in the autumn of “great families”, a living unit
composed of parents and several litters in a single nest, is discussed by
Frank (1957) for M . arwalis. This also suggests that one vole per winter
nest may not always be valid.
However, if we assume that one winter nest indicates at least one
overwintering vole, the maximum percentages of M . montanus taken
for the four years of the study are shown in Table XVI. I n addition
to the 40% of the Microtus population which were calculated to have
been taken by predators during the winter of the population decline,
9% of the population were found dead in their nests the next spring.
In many cases more than one dead Microtus was found in a nest, which
suggests that the population estimate based on the number of nests
may be an underestimate. There is no way to determine how many of
the deaths in the decline would have occurred if predators had been
absent.
Another study recently completed (Stendell, unpublished) quantifies
the predation by raptors, particularly the white-tailed kite, Elanus
leucurus, on Microtus californicus populations. I n this case populations
of M . californicus on Grizzly Island, Solano Co., California were
hl
342
CHARLES J. KREBS AND JUDITH H. MYERS
TABLEXVI
Data from Fitzgerald (1972) vrmwuring winter predation on Microtus montanus
populations by ermine and weasel. The number of nests is considered to be equivalent
to the number of voles i n the autumn and the percent eaten is calculated from the
number of Microtus nests occupied by weasels and ermine and count8 of remaim
of eaten Microtus. Only in 1968-69 did Fitzgerald jind evidence for vole mortality
other than by ermine and weasel predation. I n this year 9% of the autumn vole
population was found dead in their nests from unexplained cauaes.
1965-6 6
No. vole nests/34 acres
Population phase
yovoles eaten
No. ermine
191
Low
21.2
?
Winters
1966-67
1967-68
292
Low
56.6
4P
783
Increase
7.8
1
1968-69
793
Decline
39.7
4
estimated by live-trapping, snap-trapping and runway transect
analysis. The populations of aerial predators were counted, and pellets
containing identifiable remains of prey were analyzed to estimate the
proportion of the prey eaten. During the eight months of vole population decline avian predators took approximately 28% of the prey
population. However, during the first three months of the population
decline only 8% of the original Microtw population were taken by the
kites and owls (Fig. 37). As the Microtus population declined the predation pressure increased, as Pearson suggested. But this predation
pressure was not continued because the dense kite population left the
island in February, eight months after the beginning of the vole decline.
Another study of Microtw californicus on Grizzly Island documents
the predation by terrestrial predators during a time when aerial
predation was light (Myers, unpublished). On a study plot of approximately four acres the California vole population increased abruptly
from May to July and then decreased between July and August
(Table XVII). If we make the assumption that the number of Microtus
remains in scats deposited on the study area equals the number of
Microtus removed by predators, we find that during the population
increase (May to June) twice as many remains of Microtus occurred in
scats as were trapped in the area. The next month predation also
remained high with the number of Microtus remains in scats equal to
half the population trapped in June. Thirty-five percent of the loss
during the first month of the decline could be accounted for by
predators, as was 43% of the loss during the next month (August).
House mice, which are not preyed upon to the same extent, also
declined at this time, indicating that a mortality factor other than
POPULATION CYCLES IN SMALL MAMMALS
200
100
-
343
\.
0
0
0-
-X
C
.-0
-0
4-
2
g
10-
n
Q,
9
J
A
S
O
N
-
D
-
J
F
MONTH
Fro. 37. As the population of Microtua d$n-nkua decreased from the peak of
214 200 voles in the 2380 acres of suitable habitat on Grizzly Island, the percentage of the vole population which wm taken m h month by the whitetailed kite, short-eared owl and the barn owl increased. Data from Stendell
(unpublished).
predation was working (Table XVII). For 11 months (September to
July) no Microtw were trapped but scats deposited on the study area
still contained Microtw remains. This shows that terrestrial predators,
in this case primarily feral cats, were able t o find Microtus even when
their numbers were low and their distribution patchy. What we need
to know is whether predation pressure is strong enough to prevent
the build-up of Microtw populations when they are at low numbers.
For the present it must be concluded that feral cats are more persistent
at searching for Microtw when they are at low numbers than are
students of small mammal populations. These observations are consistent with Pearson's hypothesis.
344
CHARLES J. KREBS AND JUDITH H. MYERS
TABLEXVII
Predation pressure by terrestrial predators on increasing and declining Microtus
californicus populatibn compared to predation pressure on a sympatric house
mouse population. Estimates are conservative since the densities of the small
mammals are taken to be the number trapped in 600-900 trap nights on an area
of approximately 4 acres (1.6 hectares). Data from Myers (unpublished).
Microtus californicus
No.
No.
yo Loss due to No.
captured in scats
predation
captured
May 1971
19
40
June 1971
46
July 1971
155
Aug. 1971
28
23
3
-
1
-
180
35%
4
5%
1
1%
111
12
0
0
-
120
197
45
Sept. 1971
Sept. 1971July 1972
-
Mus muaculus
No.
yo Loss due to
in scats
predation
27
43 %
37
720
41
The trapping area in this study was very small (four acres) and so
any conclusions are subject to errors from edge effects. However, there
are several points which can be made. As predicted by Pearson, heavy
predation did not stop the increase of the Microtus population; however,
when the prey population was very low the terrestrial predators were
still able to find them. The distribution of Microtus during the phase
of low numbers might be very patchy. If this is the case, large areas
of the habitat would have no Microtus, while in limited areas voles
would exist. The probability of a predator dropping a scat on some
section of the large area lacking Microtus would be greater than that
of his dropping scats on the restricted areas of Microtus habitation.
This situation would bias the data towards the appearance of high
predation pressure during periods of low prey population densities in
a, trapping study of the sort described.
There is similarity among many studies of predation on microtine
populations in regard to the proportion of the prey population taken
by predators, with values usually ranging from 25 to 40%. The estimate
of 88% of declining M . californicus population removed by terrestrial
predators (Pearson, 1964) is outstandingly high. However, all estimates
are conservative since usually only one type of predator, either aerial
or terrestrial, is considered in an individual study (Table XIV).
POPULATION CYCLES IN SMALL MAMMALS
345
Whether or not aerial and terrestrial predators compete for the same prey
has not been investigated. I n the M . californicw declines studied by
Stendell (1972) and Myers (unpublished) and the M . montanus decline
studied by Fitzgerald (1972), predation, while possibly accentuating
the decline, was not sufficient to cause it. A prerequisite is the cessation
of reproduction and the presence of other mortality factors which as yet
are not identified.
Two important questions remain: ( 1 ) what is the predation pressure
during the period of low numbers, and (2) how much loss can a breeding
microtine population sustain? Unfortunately data regarding the first
question are limited. Maher (1970) claims that in Point Barrow,
Alaska the lemming population is free of avian predation for the first
1.75 to 2.75 years after the decline. While no M . californicus were
trapped on the Grizzly Island study area for over 11 months, cat scats
found on the study area during this time still contained vole remains.
Most of the aerial predators left Grizzly Island following the vole
decline, but kites and marsh hawks still hunted in some parts of the
island (Stendell, 1972; Myers, personal observations).
What experiments can we devise to test the suggestion of Pearson
that predation in the phase of low numbers delays the start of the next
cycle? One experiment which should be done is to add predators to a
population to test whether the periodicity of the cycle is lengthened, or
to remove predators to determine if the cycle is shortened. Since the
reduction of cats on the Pearson study area, the M . californicus have
been exhibiting a two-year cycle (Pearson, 1971) and the populations
have shown type H declines with a slight build-up of the population
after the initial decline. Several replicates of this experiment will be
necessary before conclusive results can be obtained. It has previously
been stated that a population of M . californicus on Brooks Island in the
San Francisco Bay, where there are no terrestrial predators, did not
cycle (Pearson, 1966) but Lidicker (1973) has found evidence of a twoyear cycle. However, densities of voles on this island are almost almost
higher than on the mainland. The addition of a terrestrial predator to
this island might be used to test the influence of predation on the
amplitude and timing of a microtine cycle.
How much loss can small mammal populations sustain? No one has
applied the techniques of optimum yield analysis originally developed
for fisheries (Krebs, 1972, Ch. 16) to microtine rodent populations.
We know that at some loss rate, a population must be driven to extinction, and that populations of different species vary enormously in
their ability to withstand sustained cropping. A few experiments have
been done on rodents.
I n a recent study with house mice, M u s musculus, Adamczyk and
346
CHARLES J. KREBS AND JUDITH H. MYERS
Walkowa (1971) removed 32% of the population every month and were
not successful in decreasing the size of the population. They in fact
raised the standing crop. This increase was not due to immigration
but was the result of increased survival of young born into the population and longer residency of mice not removed by “artificial predation”.
House mice have a larger litter size and higher reproductive potential
than microtines, but these experiments show that moderate mortality
does not always cause a decline in a small mammal population.
Krebs (1966) removed all Microtus californicus weighing more than
40 g, in many cases removing more than half the number of animals
trapped on the study area every two weeks, but still was not able to
prevent the population from increasing. However, there was considerable immigration of animals from surrounding areas into this population. Krebs et al. (1969) cropped from fenced populations of M .
ochrogaster and M . pennsylvanicus one-third of the adult population
every two weeks and found that these populations still maintained
higher instantaneous rates of population increase than unfenced
control populations. This suggests that a predation rate of approximately 2% per day will not stop the increase of a breeding Microtus
population. A population of M . californicus in a 120 ft2 outdoor pen
had to be cropped at a rate of over 50% a month to maintain a maximum
population of 40 individuals (Houlihan, 1963).
If predation is an important mortality agent in rodent populations,
we should be able to correlate demographic events with predation
pressure. One important aspect of the mortality which occurs during a
microtine decline is that it can be very selective. While two species of
microtines often cycle in phase (Krebs, 1964a; Tast and Kalela, 1971),
sometimes the population decline of one species will precede that of
the other by several months (Krebs et al., 1969; Tast and Kalela, 1971).
Survival of male microtines frequently decreases before that of the
females during the population decline or the mortality on the two
sexes can vary sporadically (Krebs, 1966; Krebs et al., 1969, 1973).
If predators are causing these changes in mortality, they must be highly
selective in their action. We find no support for such selectivity in the
literature. Stendell (unpublished) found that kites took age categories
and sexes in proportion to what was available in the trappable population of Microtus californicus.
To summarize, the role of predation in microtine cycles is limited
to the mortality component of the demographic machinery, and consequently other factors must be invoked to explain reproductive and
growth changes. No one seems to believe that predation can stop a
breeding population in the increase phase, and the major function of
predation is postulated to be in reducing the peak population to low
POPULATION CYCLES I N SMALL MAMMALS
347
numbers and then holding numbers low so as to delay the next cyclic
build-up. I n some declines only a small fraction of the loss can be
attributed to predation, and the evidence suggests that predation is not
necessary to cause the decline phase. Predation may contribute to the
rate of decline of a population, and this seems to be its major role in
many populations. Whether predators hold prey numbers down in the
phase of low numbers is an interesting question on which few data can
be cited. Experiments manipulating predator numbers will be necessary
t o answer this question.
C . W E A T H E R A N D SYNCHRONY
Weather can affect microtine populations, and Fuller (1967, 1969)
has suggested that weather effects are one explanation of microtine
cycles. Because Fuller’s work has been concerned with high latitude
microtines, he was particularly interested in winter weather conditions,
the critical periods being at the time of the fall freeze and the spring
thaw (Fuller, 1967). A study of Clethrionomys gapperi, C. rutilus and
the cricetine Peromyscus maniculatus in the vicinity of Great Slave
Lake, N.W.T. was undertaken by Fuller (1969) to compare the demographic characteristics of these three species living in the same habitat
and under the same general weather conditions. Populations of all
three rodent species were high in the summer of 1966. The spring of
1967 was the coldest and wettest, and was followed by low summer
population densities of C. gapperi and P. maniculatus. Because data
were not collected during the winter we do not know exactly when the
populations declined. However, C. rutilus was undaunted by the severe
winter and late spring in 1967 and remained at peak densities during the
summer of 1967. Fuller proposed that the difference in the reaction of
the three species to the “hard” winter of 1966-1967 was due to greater
cold-tolerance of C. rutilus.
Another study of Clethrionomys gapperi was undertaken by Elliott
(1969) in the vicinity of Edmonton, Alberta. This study covered the
years 1965-1968. The winter of 1967-1968 was judged most severe by
Elliott because of its thin and unstable snow cover and the greatest
amounts of rain during weeks with freezing temperatures. C. gapperi
densities were the lowest observed in the spring of 1968 for any of the
four years of the study and there was almost no recovery of the population during the summer of 1968. Thus a severe winter was clearly
associated with a population decline.
Fuller andElliott couldonly conjecture what was happening to thevoles
during the winter because their data consisted only of density estimates
in fall and in spring, and survival estimates from animals marked in the
348
CHARLES J. KREBS AND JUDITH H. MYERS
fall and recaptured in the spring. However, Whitney (unpublished)
collected survival data for Clethrionomys rutilus and Microtus oeconomus
during three years, a t the same time that he was monitoring winter
climatological features, Two winters were classified by Whitney as
poor for voles even though the conditions were quite dissimilar. The
winter of 1969-1970 began with a rapid freeze but temperatures were
higher than average and ten times during the winter there were thaws.
Snow conditions were poor which resulted in colder than normal ground
temperatures. This would probably be judged by Fuller to be a harsh
winter for voles. There were two periods of loss in numbers for the
C. rutilus population (Fig. 31). The first of these was from September
to November. Survival of females was particularly low in September;
this was approximately six weeks before the winter freeze and the
first snows. For the remainder of the winter the percentage of voles
surviving every two weeks ranged from 70 to 90%. The second period
of poor survival was between February and March, over a month before
the little snow present melted. I n general, winter survival by C.
rutilus was superior to summer survival. The period of poor survival
at the start of the breeding season in the spring is typical of microtine
populations (see p. 283) and could not be tied to specific weather
factors.
Whitney reported a similar picture in a sympatric population of
Microtus oeconomus. The survival of males was better during the winter
than in the summer. The deterioration of survival in the spring occurred
a month earlier in this species than in Clethrionomys rutilus on the
same study area.
Summer survival of C. rutilus was low in 1970 (Fig. 31); however, it
improved in August and except for a short period in September
continued to be extremely high during the winter. This winter was not
judged by Whitney to be good for voles either. The fall freeze was
gradual, the snow fall heavy and the subnivean space was not developed
until late in the winter,
Whitney’s study is extremely important because it shows the
necessity of obtaining detailed demographic data in judging the influence
of extrinsic factors on the population. We do not yet know what
characteristics of winters are most stressful for voles. But Whitney’s
data, in agreement with that reviewed earlier, indicates that the
crucial period is really in the fall, with a period of poor survival often
preceding freezing weather and snow, and in the spring, in association
with the beginning of the breeding season. No simple association
could be made between winter conditions and survival of Clethrionomys
rzctilus and Microtus oeconomus.
The spring melt-off might be a time period which is critical to
POPULATION CYCLES IN SMALL MAMMALS
349
arctic microtines. There is great variation from year to year in the
timing of the snow melt, and this might affect summer population
growth. Without winter sampling it is difficult to judge accurately the
beginning of the summer breeding season of arctic microtines. However,
Mullen (1968) attempted to do this by looking for signs of recent
reproduction in female brown lemmings taken in June. From this
analysis he proposed that the date of beginning of the summer breeding
season is crucial to the demographic performance of the population
that summer. His data are presented in Table XVIII. We only have a
verbal description of the demographic changes available from Mullen’s
work and these descriptions do not always agree with trapping data
from the same vicinity given in Pitelka (1972). However, there are
TABLEXVIII
The data for the beginning of the summer breeding season and the subsequent change
in the population density are recorded for five years by Mullen (1968)
Year
~~
a
8
Popln density and
summer popln change
~
1960
1961
1962
1963
1965
1
Beginning of
breeding
17 June
27 May
7 June
Late May
Late June
June peakdeclinedl
Low-no change
Low-no change
Moderate--declineda
Highdeclined8
Population remained constant according to Pitelka (1972).
Population low and remained constant according to Pitelka (1972).
Population change from Pitelka (1972).
three summers of declining density. I n two of these the breeding season
began late and in the other it started quite early. Both the breeding
seasons of 1961 and 1962 began relatively early and neither resulted in
increasing lemming populations. We can find no evidence in these
data or in the data of Krebs (1964a) that the timing of the snow melt
strongly affects subsequent population trends in lemmings.
Climatic factors and the abundance of Microtus arvalis were investigated by Straka and Gerasimov (1971) in Bulgaria. I n this case
summer drought was the stressful weather condition under consideration. I n the areas of Bulgaria where rain was not limiting and conditions were optimal for M . arvalis the populations demonstrated three
cycles during the nine years of the study, and there were no correlations
with temperature or rainfall variations. Similarly, at the southern edge
of the species range, while densities were generally low and fluctuations
erratic, there were no correlations with rainfall or temperature. However,
350
CHARLES J. KREBS AND JUDITH H. MYERS
in northern Bulgaria, an area subject to periodic summer droughts,
M . arvalis populations sometimes reached high densities. The determining factor seemed to be the amount of rain during the second half of
the summer. Good rains and associated forage growth permitted
breeding by the voles into the autumn, and therefore dense overwintering populations.
The area near Berkeley, California is characterized by summer
droughts of varying lengths. The study of Batzli and Pitelka (1971)
of M . californicus included two years in which rainfall was below
normal. I n one of these years the voles were at low densities and in
the other at peak density. Thus summer drought does not appear to
determine population trends in the California vole.
As part of a study of M . penmylvanicw and M . ochrogaster in
southern Indiana solar radiation, rainfall, soil temperature, and
humidity were recorded for the five years of the study (Krebs, unpublished). No correlations could be found between these weather
characteristics and Microtus population fluctuations. Furthermore,
the two species living under the same weather regime did not always
fluctuate in synchrony. Weather appears to exert little limitation on
vole populations in Indiana, where extremes of drought or cold do not
occur.
While it is impossible to vary winter weather experimentally and
thus to test its influence on microtine cycles, populations which are
out-of-phase could be used for this test. This experiment requires an
increasing population contiguous with a peak or declining population.
One means for accomplishing this is to introduce microtines to enclosures, where they respond by increasing in numbers. The influence
of a particular winter on an enclosed and an adjacent free-living
population could then be observed. The relation of drought to population phenomena of southern microtines could be tested by observing
populations in irrigated areas. For example, we could ask whether
irrigation abolishes the cycling of M . californicus populations.
One of the conceptual difficulties in recognizing the role of weather in
microtine cycles is the diversity of populations in which cycles occur.
Voles from arctic to temperate areas seem to go through population
cycles which have many demographic attributes in common. If we
explain a fluctuation in M . oeconomus by winter snow conditions, must
we seek another explanation for M . pennsylvanicus in areas where
snow is rare?
We recognize that weather has received too little attention from
students of microtine rodents. We need to look both at the destructive
aspects of weather on survival rates and at the permissive aspects of
weather in allowing reproduction. Because many of these effects can
POPULATION CYCLES IN SMALL MAMMALS
351
be transmitted indirectly via the food supply or cover available, the
role of weather may be most difficult to untangle in natural populations, even with field experimentation.
Regular fluctuations in populations presumably require a regular
stimulus, and few would claim that weather variables change in a
regular three- to four-year pattern. However, the imagination is pressed
to conceive of any factor other than widespread weather conditions
which might a c t as the cue for synchronizing fluctuations of microtines
over broad geographic areas. There are numerous accounts of populations of microtines which are out of phase, but synchrony seems to be
the usual case. Chitty (1952, 1960) and Chitty and Chitty (1962)
record asynchronous populations of Microtus agrestis. I n southern
Indiana, Krebs et al. (1969) found some sympatric populations of M .
ochrogaster and M . pennsylvanicus which fluctuated in synchrony, but
other asynchronous populations were also monitored (Keller and Krebs,
1970). Pitelka (1961) reports asynchronous brown lemming populations
in northern Alaska, while Watson (1956) and Krebs (1964a) found
sympatric populations of Lemmus and Dicrostonyx to fluctuate in
phase. Both Mullen (1968)and Krebs (1964a)report peak brownlemming
populations in the summer of 1960. Mullen’s study was done in Barrow,
Alaska and Krebs’ in Baker Lake, N.W.T. over 2000 miles away. The
microtines of Finnish Lapland appear generally to fluctuate in synchrony, but cases of asynchrony have also been observed here (Tast and
Kalela, 1971).
For weather to act as a synchronizing factor it must be postulated
that similar weather conditions can have contrary effects on populations
in different phases of the population fluctuation (Frank, 1957). Chitty
(1967, 1969)considers this to be quite possible if the quality of microtine
populations varies with the density. Therefore peak populations might
be severely influenced by a bout of poor weather while an expanding
population would be hardly affected (Fig. 38).
Leslie (1959) proposes a model which describes how an external
random factor (such as weather) acting on populations which are
geographically isolated brings the oscillations of the population
densities into phase. So we have some reassurance that, theoretically,
weather could be a synchronizing element, but we still lack the biological understanding to determine if it is acting in this way.
A factor which has some bearing on this topic is the degree of
geographical isolation of populations. I n the tundra there are vast
areas of suitable lemming habitat. However, in the temperate zone,
Microtus habitat is largely a relic of farming practices. Fields in
different stages of succession provide habitats of changing suitability.
For this reason we might expect more out-of-phase populations in arem
362
CHARLES J . KREBS AND JUDITH H. MYERS
70.
60
-
I
50
I
I
I
I
I
8
-
I
-
POOR WEATHER
A
1
7060
50
2
-
3
4
5
-
7
FAVORABLE W E A THER
-
I
2
3
4
5
1954
1955
1956
1957
1958
C
YEAR
FIG.38. Parts A and B demonstrate possible situations in which weather might
act to bring two asynchronous vole populations into synchrony. I n A poor
weather conditions have a strong influence on a peak population while having
little influence on a population in the phase of increase. In B favorable weather
extends the duration of peak population density in one population and accelerates
POPULATION CYCLES I N SMALL MAMMALS
353
of patchy habitat than in the tundra. Whether this is the case we
do not know. It is not possible to sample lightly from a number of arem
to determine if populations are fluctuating together. Each population
must be studied over time. Pruitt (1968) sampled populations across
the North American Arctic and while he could not verify synchronous
cycles in small mammal population densities, he claimed to fhd a
cycle in the small mammal biomass. The sizes of most samples were
quite small. Haynes and Thompson (1965)found positive relationships
in the amount of vole activity among areas within 20 miles of each
other, but it is not possible from these data to determine whether
asynchronous populations were observed.
Out-of-phase populations are hard to locate and hence rarely
studied. Until we know what weather conditions are most important
to microtines, we shall be unable to determine how synchrony is
brought about between scattered rodent populations.
I n summary, weather must be important to microtine populations if
synchrony occurs. We do not know how weather acts to synchronize
cycles. Few studies have been made of arctic microtines during the
winter, and we have found no simple associations between weather and
population events. Weather effects do not seem to explain vole and
lemming cycles, and the main driving forces must be sought elsewhere.
D.
STRESS H Y P O T H E S I S
High animal densities increase the probability that individuals will
interact. If these interactions are disturbing to the animals, ‘‘social
stress” may be expected to rise with increasing population densities.
The stress hypothesis of Christian (1950) is probably one of the most
widely known theories of population regulation, even reaching the
every-day world of analogies with human populations.
The stress hypothesis is an outcrop of the work of Selye (1946) on the
response of the pituitary and adrenal to stress. I n the extreme case the
increased activities of the pituitary and adrenals cause exhaustion,
low resistance and general susceptibility of the individual to a variety
of potential mortality factors. I n addition a corresponding inhibition
of the pituitary-gonadal function can decrease reproduction (Christian
the rate of increase in an expanding population. Part C plots data taken from
Chitty and Chitty (1962) and shows the number of Microtus agrestis trapped
per hundred trap nights. Because of favorable weather in the winter of 1956-67
one population remained at peak densities while the other increased to peak
densities. The two populations declined in synchrony. The two curves represent
two different populations. Bee discussion in Chitty (1967).
354
CHARLES J. KREBS AND JUDITH H. MYERS
et al., 1965). Aggressive behavior of individuals is an intimate part of
this theory, with stress and aggressiveness increasing in a spiral.
We will consider behavioral characteristics of cycling rodents in another
section of this article, but here we want to consider evidence which has
been gathered relevant to the search for characteristics of physiological
stress associated with high microtine densities.
First let us reiterate the characteristics of cycling microtines which
must be explained if we wish to support the stress hypothesis. The
impairment of reproduction occurs at peak density in the form of a
shortened reproductive season, but other factors of natality such as
litter size and prenatal loss do not consistently vary with microtine
density. Therefore, we might expect stress to shorten the reproduction
season but not to cause increased prenatal deaths. Mortality is higher
during the population decline and is particularly severe in very young
animals. Males and females may suffer poor survival at different times.
From this we might predict that the social climate of males and females
and of young animals is different.
One of the &st steps toward testing the stress hypothesis is to
look for evidence of hyperactivity of the adrenals among rodents of
high density populations. Secondly, we would look for poor physiological
conditions among animals from declining populations. Christian et al.
(1965) cite a number of examples of rodents under abnormally high
densities in the laboratory which show characteristics consistent with
what would be predicted from the stress hypothesis. An extensive
review of work using caged laboratory animals to elucidate the effects
of isolation and grouping on brain chemistry and the functioning of
the endocrine glands is available in the review by Brain (1971a) and will
not be dealt with here.
For years adrenal function has been assayed by the weight of the
adrenal glands. Several attempts at finding relations between adrenal
weights and microtine population densities have failed (Christian, 1961;
H. Chitty, 1961; Krebs, 1964a). A primary drawback has been the
analysis of adrenal weight data. The adrenal weight changes with the
body weight but it also varies with reproductive condition, age and sex
of the individuals and season of the year. Furthermore, the relation
between adrenal weight and body weight is most likely not linear
(Krebs, 1964a), and it is not valid to compare adrenal weights by using
values which are given in mg adrenal wt/gm body wt. Chitty (1961)
and Krebs (1964a) used the technique of standardized means (Hill,
1959)to correct for body weight so that comparisons of adrenal weights
could be made independent of body weight.
While adrenal weight may not be a good measure of adrenal activity
(Christian and Davis, 1964; Andrews and Strohbehn, 1971), this was
POPULATION CYCLES IN SMALL MAMMALS
355
. the technique used by Christian and Davis (1966) in their most recent
investigation. Since these authors conclude a “marked parallelism”
between adrenal weights and population density for Microtus pennsylvanicus, this paper deserves further consideration. Overlooking previous
discussions of appropriate methods for analysis of adrenal-body weight
relations (Chitty, 1961; Krebs, 1964a), Christian and Davis used mg
adrenal w t / l O O gm body w t for their comparative index. By doing this
they are assuming a relationship between adrenal weight and body
weight which is linear with a slope of 1, an unlikely situation. The
population index was the number of voles caught/1000 trap nights in
one 24-hour period. This technique of population estimation is not
reliable. Population density declined from its highest point in September
1960 until March 1961. The values are as follows:
Date
voles/1000
trap nights
24 September 1960
2 December 1960
20 January 1961
2 March 1961
100
28
103
21
As pointed out by Christian and Davis, the low value for December
must underestimate the population density since it occurred during
the non-breeding season and the population could not have increased to
such an extent between December and January as indicated by the
data. While Christian and Davis (1966) claim “a striking degree of
positive correlation” between adrenal weight and the index of population size (voles/1000 trap nights) this correlation was not significant
(T = 0.59; 0.10 > P > 0.05).
The relationship between adrenal weight and population density
(Christian and Davis, 1966) was found only among mature females.
However, as reviewed earlier it is the younger animals which seem to
suffer the heaviest mortality during the population peak and decline.
This high mortality among the young would not be explained by social
stress based on Christian and David results. Later work of Christian
(1971b)indicates that social stress as measured by wounding is strongest
among mature male Microtus pennsylvanicus. Combining the results
from Christian’s two studies leads to the conclusion that there is no
correspondence between endocrinological stress measured by adrenal
weight and aggressiveness shown by wounding in mature male voles.
New approaches to the assay of adrenal activity have been used by
Andrews (1968, 1970)and Andrews and Stohbehn (1971). Corticosteroid
secretion rates can be measured on adrenals maintained in tissue culture.
356
CHARLES J. KREBS AND JUDITH H. MYERS
I n addition, respiratory rates of cultured adrenals are indicators of
adrenal activity and the response of adrenal glands to ACTH is another
measure of the state of the glands.
Andrews (1968) studied adrenal activity in brown lemmings from
Barrow, Alaska. His &st observation was that cultured adrenals
display a circadian fluctuation in secretory activity. Therefore, comparisons among groups of individuals must be made over standard
lengths of time, and during the same time of the day. The whole process
is complicated by the fact that Andrews (1968) stored animals for
varying lengths of time in the laboratory and this changed the timing
of the circadian secretory rhythm of the adrenals. Comparison among
groups in this study is most difficult.
A second complicating factor in the interpretation of Andrew’s
data is that his three measures of adrenal activity are not correlated.
The highest respiratory rates, and secretory rates measured by the
conversion of C1“acetate into corticosteroids, occurred in a sample
of lemmings collected from a peak population in July and killed for
analysis three weeks later in early August. However, the highest rate of
corticosteroid secretion measured, using fluorometric determination of
ethylacetate extractable steroids, occurred in a sample of lemmings also
collected from a peak density population in July but maintained in the
laboratory until February when they were analyzed. These results
indicate that different biochemical pathways for the production of
corticosteroids are being used under varying conditions. The biological
significance of these results was not interpreted by Andrews (1968).
There are few statistics used in Andrews’ (1968) study and so it
is difficult to make comparisons between samples. However, all of his
lemming samples were collected in July 1965, a year of peak lemming
density a t Barrow, Alaska (Pitelka, 1972). The factor which varied
among samples was the length of time the lemmings were maintained in
the laboratory. Therefore, when Andrews (1968, p. 91) talks about
“glands obtained during summer 1965 and winter 1966, following a
lemming population crisis”, in fact he has collected animals from the
same peak population and merely analyzed the adrenals after varying
lengths of time. It is doubtful whether there is any correspondence
between the physiological state of peak density lemmings maintained
in the laboratory for seven months, and lemmings which remain in the
natural population after the population decline. The only conclusion
that can be drawn is that maintaining lemmings in the laboratory seems
to change the secretion of adrenosteroids, but we have not advanced in
understanding the biology of lemmings.
The respiration rate of adrenals of lemmings collected during the
summer of population increase (1964) and the summer of the peak (1965)
POPULATION CYCLES IN SMALL MAMMALS
357
are not significantly different. However, the respiratory rate of adrenals
from animals captured in July 1965 and analyzed approximately one
month later was significantly less than those of lemmings caught in
July 1965 and analyzed in February 1966, seven months after capture.
While ACTH stimulated steriod production in adrenals of lemmings
maintained in the laboratory for seven months, it decreased steriod
production of adrenals from lemmings collected and analyzed during
the summer of 1965. This is interpreted as an indication that adrenals
of animals collected from populations of peak densities were secreting
at the maximum rate.
According to Andrews’ description, the second year of his work
(1969) was one of low lemming density. Sample sizes were quite small
but the following trends were observed during the summer: (1) The
content of ACTH in male pituitaries remained high through the summer
but decreased in female pituitaries from early July through August.
(2) The responsiveness of adrenals to exogenous ACTH measured as
corticosteroid production was similar between males and females and
seemed to increase in the August sample. (3) Andrews and Strohbehn
claim a decreased sex ratio a t the end of the summer, but there are no
statistics to verify this and the sex ratio is 60% females (N = 13).
This does not indicate strong differential mortality between males and
females as stated by these authors, and leaves in doubt their conclusion
that higher male mortality is associated with higher male adrenocortical and pituitary function.
Comparisons of data from the 1969 study of low population density
to earlier data on peak populations were not made by Andrews and
Strohbehn (197 1). However, adrenals from low-density populations
were stimulated by ACTH while those from a peak-density population
were not.
Analyses of adrenal activity obtained by Andrews using the more
elaborate methods are going to be even more complicated than were
analyses which utilized adrenal weight. Along with the variations
associated with body weight, sex, reproductive condition, and season
of the year (similar to those observed in earlier studies of adrenal
weights), there are circadian rhythms of adrenal secretion and respiration, and complex pictures which result from measuring steroid production with several techniques.
For any progress to be made in determining the role of the adrenal
in affecting population processes, it will be necessary for proper
demographic, physiological, and statistical procedures to be employed.
The behavior of microtines is most certainly mediated through the
endocrine system. So far no progress has been made toward correlating
endocrinological changes to the early termination of the breeding
358
1
CHARLES J. RREBS AND JUDITH H. MYERS
season during periods of peak lemming densities, or to high mortality
in the decline.
It is possible that the adrenal steroid levels could be experimentally
increased in a wild population by the injection of hormones into
microtines. However, we understand so little of how these hormones
work that analysis of the results of such an experiment could be too
complicated to reveal any insight. The alternative might be to use
sedatives to decrease social stress, but again manipulations of body
chemistry can have undesirable side-effects.
I n addition to all this work on the adrenal gland, various other
physiological indices of condition have been studied by some workers.
Hematological characteristics might be used as indicators of the
physiological condition of animals. Newson and Chitty (1962) searched
for an association between reticulocytosis and anemia and population
decline in Microtus agrestis. While they found seasonal variation in
hemoglobin levels of voles, the animals were not anemic during the
population decline and they failed to find a physiological abnormality
which might explain the decline.
An extensive investigation of the blood of brown lemmings was undertaken by Mullen (1965) in a search for physiologicalchanges which might
be associated with density. Blood glucose levels were measured to test
for hypoglycemia. Hypoglycemia (low blood sugar) has had a part in
€he history of small mammal cycles since the report by Green and
Larson (1938) that snowshoe hares in Minnesota suffered a decline
because of “shock disease” characterized by low blood sugar, degeneration
of the liver, and failure to store glycogen. Chitty (1959) refuted this
work and recorded his own failure to fhd “shock disease” in Microtus
agrestis. Also Chitty (1959) found that low glycogen reserves occurring
in experimental laboratory populations of M . agrestis did not increase
mortality. Houlihan (1963) found that blood sugar of an enclosed
population of M . californicus decreased when the food supply was low
but during a period of sharp decline (37% lost in two weeks) the blood
sugar level of individuals in this population was the same as in a
control population of moderate density which did not decline.
Similarly, while Mullen (1965) found variation in the blood sugar
level of brown lemmings, he was not able to associate this variation
with population density or density changes.
It has been suggested that there is a negative relationship between
the number of circulating eosinophils and adrenocortical activity in
rodents (Speirs and Meyer, 1949). If this is so, determination of
eosinophil levels can be used as a means of assaying adrenal activity.
Houlihan (1963) studied eosinophil levels in two enclosed (120 ft2
outdoor pens) populations of Microtus californicus. The density of one
POPULATION CYCLES IN SMALL MAMMALS
359
population was maintained at approximately 40 individuals while the
other population increased, with provision of additional food, to a
peak density of 158 individuals. After maintaining high densities for
several months the uncropped population declined. Houlihan ( 1963)
compared eosinophil levels between these two populations. I n the
months before and after the decline of the uncropped population,
eosinophil counts were higher than those of individuals from the
moderate density control population. This indicates increased adrenal
activity if eosinophil is a valid measure. However, in the month of the
decline of the high density population, the eosinophil counts were
significantly higher than in the other two months. Therefore, it was not
possible to interpret these results in a simple manner, and Houlihan
concluded that eosinophil levels should not be used as an indication of
adrenal stress.
Mullen (1965) comes to the same negative conclusion as Houlihan in
regard to the use of eosinophil levels as a measure of adrenal activity.
Rather than using the number of eosinophils as his index, Mullen
determined the percent of total leucocytes represented by eosinophils.
Therefore, although there is a trend for the eosinophil level to be low
in individuals taken from the 1960 peak population, this was found to
be the result of a greater number of leucocytes all together, which
lowers the percent of eosinophils, although the absolute number of
eosinophils was actually higher in 1960 than in other years. Crowding
lemmings in the laboratory did not influence eosinophil levels.
Determination of non-protein nitrogen in the blood is another physiological index in microtines. Houlihan (1963) used this measure on his
artificial high and low density Microtus populations. The trends in this
measure were the same in both populations. Houlihan did find several
indications of physiological derangement in his declining population.
These were: (1) a 30% decrease in thyroid activity, (2) an increase in
the time required for blood clotting and a change in the quality of the
blood clots, and (3) an increased susceptibility to the blood sampling
procedures resulting in more deaths, particularly among males. At the
time of the decline there was severe fighting and wounding. Although
these enclosed populations were maintained outdoors, they were
artificial in that additional food had to be added for the high density
(equivalent to 47701acre) to be reached. The degree of similarity of this
microtine decline to a natural decline still remains a question.
Other physiological assays performed by Mullen (1965) on brown
lemmings were leucocyte, erythrocyte and reticulocyte counts, and
hematocrit values. Two factors complicate the interpretation of these
data. Pirst, experiments in which these characteristics were determined
for lemmings kept under crowded conditions and kept singly in the
360
CHdRLES J. KREBS AND JUDITH H. MYERS
laboratory, showed that none of these hematological factors seemed to
vary with artificially imposed density levels. Secondly, the densities of
field populations given by Mullen (1965) do not agree with the trapping
data of Pitelka (1972) for the same vicinity at Barrow, Alaska. Mullen
refers to the summer of 1963 as a peak year while Pitelka’s data indicate
that it was a year of low numbers. This conflict might be explained by
the fact that Mullen trapped periodically during the winter of 19621963. Therefore, numbers may have been high and declined before
8000
1960
1961
1962
1963
PEAK
LOW
LOW
LOW
LAB
FIG.39. Leucocyte levels of lemmings collected in the field compared to those
maintained in the laboratory. The leucocyte counts for both males and females
in 1960 are significantly higher than in other years. Data from Mullen (1966).
While Mullen regards 1963 a year of peak lemming density, Pitelka’s trapping
data for the same vicinity indicates that this was a year of low density (Pitelka,
1972).
Pitelka’s June trapping period. Without adequate demographic data
it is impossible to judge what phase of the population fluctuation was
represented in 1963. However, there is general agreement that 1960
was a year of peak density and we can use these data to indicate the
condition of lemmings from peak population densities.
The number of circulating leucocytes was significantly higher among
both males and females during the peak year of 1960 (Fig. 39). High
leucocyte levels are often an indication of infection or disease, but
Mullen (1965) had no data on possible pathologies associated with these
POPULATION CYCLES IN SMALL MAMMALS
36 1
elevated leucocyte levels. The number of circulating leucocytes declined
in animals brought into the laboratory from the 1960 peak population.
While there was no variation in the number of circulating erythrocytes in brown lemmings during the summers of 1961 to 1963 the number
of reticulocytes was higher in lemmings collected in June 1960 (peak
density) and June 1963 than were those of lemmings from the intervening years of 1961 and 1962. This may indicate that during the early
summer of years of high population density the production of red blood
cells is greater. However, if the total number of red blood cells does not
change it must also mean that the survival time of red blood cells is
higher. The decline in the lemming population at Barrow did not occur
until autumn or Winter, 1960. Late summer reticulocyte levels were not
significantly different in 1960 than in other years. It is interesting
that lemmings with high levels of circulating reticular erythrocytes
TABLEXIX
Percent circulating reticular erythrocyte%in the blood of lemmings collected f r m the
field and those collected from the field and maintained in the laboratory for two to
three month. Data;from Mullen (1965).
~~~~~~~~
~
Year
1960
1961
1962
1963
*
yo Reticulocytes-Laboratory
1.91 f 0.27
2.26 0.17
3.37 f 0-36*
yo Reticulocytes-Field
2.48 k 0.16
2.06 f 0.65
1.69 f 0.26
4.32 k 0.46*
Signirioently higher then other years, P < 0.01.
from the June 1963 population, maintained these high levels after two
to three months in the laboratory (Table XIX). I n the natural population the level of reticulocytes declined during the summer, indicating
that there .was selective mortality against individuals with high
reticulocyte levels. Again artificially imposed high densities did not
influence the number of circulating reticular erythrocytes.
Hematocrit values did not show meaningful variation among
lemmings for the years 1961-1963. The volume of blood cells remained
constant although the composition of types changed.
We are left with a contradiction after an analysis of physiological
indicators of stress. The strength of the stress hypothesis has been
the results of experimental crowding of small mammal populations
in the laboratory. Mullen (1965) has reported two hematological
variations which seem to be associated with density of natural lemming
populations, and yet he waa not able to mimic these changes in crowded
laboratory populations. Mullen’s conclusion was that his data do not
362
CHARLES J. KREBS AND JUDITH H. MYERS
support the social stress theory for population control but that they do
suggest that disease or metabolic disorder may be associated with the
population decline. The question remains as to whether the susceptibility to disease of individuals in high-density populations is greater.
Further investigation of the comparative survival of individuals with
high and low leucocyte counts and reticulocyte counts should be done.
After the decline, the levels of these two blood parameters returned to
normal (Mullen, 1965). Did individuals with high leucocyte and
reticulocyte levels have higher mortality or did the physiology of
surviving individuals change? Studies of this sort could be done by
taking blood samples from individuals at intervals in the field.
Hematological tests of adrenal activity have been unsuccessful and
the search for hypoglycemia associated with declining microtine
populations seems to have been futile. However, the use of blood cell
counts as physiological indicators deserves further consideration.
The extent of ectoparasitism in populations of Microtus californicus
was found to vary with population density (Batzli and Pitelka, 1971).
More animals were infested by fleas, lice and mites in the autumn
following the peak population density than in the previous autumn of
population increase. Ectoparasitism may be a useful characteristic for
assay of the general condition of voles. Because there is most likely a
relationship between the condition of the animals and the ectoparasite
load this approach to the investigation of stress and population density
deserves further consideration.
While dead animals are rarely found following a decline in microtine
populations, Fitzgerald (1972) reports finding dead M . montanus in
nests the spring after a winter decline. Dead and dying M . califomicus
were observed on Grizzly Island, California during the decline in
population during the late summer of 1971 (Myers, personal observation), and Rauch (1950) observed dying lemmings during the 1949
decline in Alaska. Voles from declining populations which are brought
into the laboratory seem usually to survive well (Newson and Chitty,
1962; Krebs, 1966; Andrews and Strohbehn, 1971). Because we know
almost nothing of the characteristics of dying microtines or if, in fact,
the animals are dying in situ during the decline, we cannot judge the
relation of the stress theory to mortality in natural populations of
cycling rodents.
There is one set of observations which remains difficult to explain
with the “stress hypothesis”: that enclosed populations are able to
reach much higher densities than are observed in natural populations.
Microtines can both live and reproduce a t densities much higher than
occur naturally. Proponents of the stress hypothesis argue that social
stress is not a simple function of population density, and consequently
POPULATION CYCLES IN SMALL MAMMALS
363
the “stress” level of a natural population at 100 per acre could be equal
to the stress level of an artificial population at a caged density of
10 000 per acre. Whether or not one accepts this argument, the point
we wish to make is that more attention should be paid to the behavioral
interactions which must cause social stress. Excessive preoccupation
with physiological measurements may have sidetracked us from more
relevant behavioral aspects of social stress.
In summary, the stress hypothesis suggests that microtine populations
peak and decline because of physiological deterioration of adrenal
functions. Although this theory is popular even in the communications
media, we can find no evidence from natural populations to support it.
Few studies have discovered any apparent relationships between
adrenal functions and population changes, and none of the characteristic features of reproduction, mortality, dispersal, or growth which we
discussed previously have been associated with physiological measurements of adrenal changes.
E.
BEHAVIOR
The behavior hypothesis suggests that interactions between individual animals are critical in causing population fluctuations in
small rodents. The behavior hypothesis is an intrinsic hypothesis
since it states that a necessary factor preventing unlimited increase is
a change in the behavior of individuals in the population. We will
consider this theory separately from the stress hypothesis and the
genetic hypothesis because no mechanisms of behavioral change or
inheritance are specified. We inquire, aa a first approximation, only
whether behavioral interactions change during a population cycle.
The primary behavior which might be involved in population events
is spacing behavior. There has never been any doubt that voles and
lemmings show hostility toward one another, and space themselves over
the habitat; people differ greatly in how to interpret such a fact.
Watson and Moss (1970) have carefully reviewed how population
limitation can be achieved by means of aggression in vertebrates, and
they cite three conditions which are necessary to show that aggressive
behavior limits the density of breeding animals:
1. there must be a substantial “surplus” population which does not
breed;
2. these surplus animals must be capable of breeding if the more
dominant animals are removed;
3. the breeding animals must not be completely depleting some
resource, such as food or nesting sites.
364
CHARLES J . KREBS AND JUDITH H. MYERS
How can one demonstrate that a surplus population of microtines
exists? The best way to do this is to crop a resident population of
breeding animals, and to see if new animals take up the positions
vacated by removals. This experiment was first done by Smyth (1968)
on Clethrionomys glareolus at Oxford, and he found that extensive
immigration offset his removals. Krebs (1966) cropped a two-acre
grassland of 1758 Microtus californicus over one year and yet found
little difference in density between the cropped population and its
control. Myers and Krebs (1971b) described a substantial influx of M .
pennsylvanicus into cropped areas in Indiana. Watts (1970) cropped a
population of Clethrionomys gapperi of adult males during a phase of
increasing density, and found little difference in the rate of growth of
the control and experimental populations. Elliott ( 1969) removed
males from another population of C. gapperi and found that they were
replaced by other adult males. He concluded that spacing behavior
was important in determining breeding densities. Dahl (1967) did the
converse experiment of adding M . pennsylvanicus to a resident population and found that he could not increase density by adding voles.
These cropping experiments can be criticized because the “surplus”
voles moving into the experimental areas might be the resident breeding
animals from surrounding areas. Hence the breeding density of the
whole area might be depressed by filling in the evacuated habitat, and
no truly “surplus” voles might exist. This did not appear to be the case
in the Krebs (1966) experiment, since not a single marked individual
was drawn from the control area to the removal area 300 f t away.
Surplus animals in rodent populations presumably disperse and
are largely lost to various agents of mortality. Consequently the
study of dispersal in vole and lemming populations is also a study of
the movements of “surplus” animals. Because of the general habits of
rodents, we are not able to study their social organization in the same
way we can for territorial birds. Hence the demonstration of surplus
voles or lemmings will probably never be as elegant in experimental
terms as similar experiments on birds.
The surplus population of microtine rodents always seems to be
capable of breeding, thus satisfying condition (2) above. The immigrants
of Microtus californicus which colonized the removal area were active
breeders (Krebs, 1966), and the same was generally true in the study
by Myers and Krebs (1971b), although some young animals not yet
sexually mature also dispersed. Cyclic rodents seem to have adopted
the general strategy of decreasing the population of breeding individuals
as they go from the increase phase to the peak phase. Reproductive
changes, such as increase in the age at sexual maturity, all seem to
force more of the population into a “surplus”, non-breeding category.
365
POPULATION CYCLES IN SMALL MAMMALS
The third criterion of Watson and Moss (1970) is that the breeding
animals are not completely depleting some resource, such as food. This
particular criterion brings together most of the controversy over
microtine cycles. We have argued above that the present evidence does
not indicate that populations of cyclic rodents are completely using
some resource such as food. But it is possible to argue that we do not
yet appreciate what a “resource” is to a vole, and consequently we
should consider other types of evidence.
If hostility is part of the mechanism for generating population cycles
in rodents, we must expect that the aggressive behavior of individuals
changes with density. Two types of data are available on this question.
First, relatively crude data can be obtained from observing skin
TABLEXX
Proportion of skins of male brown lemmings (Lemmus trimucronatus) showing
wounds on inner surface during a population cycle in northern Canada. Overwintered adults are all sexually mature;summRr-bornyoungare all sexually immature.
Sample size in parentheses. After Krebs (1964a).
June
Increase phase (1959)
Peak phase (1960)
Decline phase (1961)
0.29
(7)
0.52
(99)
0.18
(33)
Overwintered adults
JdY
August
0.20
(5)
0.69
(39)
0-65
(26)
0-71
(7)
0.36
(14)
-
Summer-born
Young
0.09
(11)
0.33
(255)
0.24
(45)
wounds on animals from different phases of the cycle. Krebs (1964a)
scored wounds from flat skins of brown and varying lemmings, and
found that wounding was more severe in peak populations than in
increasing ones, but that wounding remained severe in the decline
phase even though density had fallen (Table XX). Summer-born
young lemmings had some wounds even though most of them were
sexually immature. Some suggestion of increased wounding associated
with sexual activity of summer-born lemmings was obtained from one
sample in the increase phase in which one of eleven immature males
was wounded compared with six of eight mature males (Krebs, 1964a,
p. 49). Young females did not show this effect, and almost none of
them showed wounds.
Christian (1971a) made similar observations on a peak and declining
population of Microtus pennsylvanicus. Mature males showed more
wounding in spring of the peak year than in the spring of the decline
366
CHARLES J. KR.EBS AND JUDITH H. MYERS
year. He found that wounding was confined almost entirely to adult
males; females had few signs of wounds on the skin. Immature young
males were not subject to wounding either. Christian suggested that
when the age at sexual maturity is low, fighting will be more prevalent
because a larger proportion of the male population will be mature.
He therefore predicts more fighting a t low densities because of early
maturation, but thinks that the influence of the larger portion of
belligerent males is offset by the lower density.
Wounding was not greater in higher density populations of M .
californicus (Batzli and Pitelka, 1971). Again the amount of wounding
observed in males was greater than that in females.
Aggression, fighting and wounding may be greater in Clethrionomys
than in Microtus. I n a year of low abundance of C. rytilus Koshkina
(1965) observed badly wounded individuals, particularly among those
who were sexually mature. She judged that the wounding was sufficient
to kill the animals, and badly injured voles were not often recaptured.
There are two difficulties in interpreting data on skin wounds.
First, the amount of wounding can only be a very crude behavioral
index of aggressiveness. Many of the techniques by which animals
space themselves do not involve physical aggression (Lorenz, 1963).
Second, wounding indices confound the two variables of population
density and aggressiveness. We cannot decide from wounding data
whether (1) the average level of individual aggressiveness is constant
and independent of the cycle, and all the changes in wounding are a
result of changes in density and therefore more numerous interactions,
or (2) whether the behavior of individuals also changes in relation to
density.
We know little of the factors affecting contact rates of individual
voles or lemmings in the field. Pearson’s (1960) elegant photographic
work on runway usage should caution us that contact rates may not be
a simple function of population density. There are indications that
the area traversed by individuals is smaller when population densities
are high (Krebs, 1964a, 1966, and unpublished; Koshkina, 1965).
By restricting its area of activity as the population density increases,
a vole may be successful at maintaining an almost constant rate of
intraspecific interaction regardless of population density. Because we
cannot be certain of the social climate in field populations we need to
obtain behavioral measurements on individual animals, and to look at
some of the details of social organization in microtine rodents.
Techniques for measuring aggressiveness on live rodents have been
used in psychological research for a long time, but few attempts have
been made to apply these techniques to voles and lemmings. The
simplest procedure is to observe two individuals in a fighting arena.
POPULATION CYCLES IN SMALL MAMMALS
367
The arena may be a neutral arena or a home cage of the animals.
Often only males have been tested, in the possibly mistaken belief that
important aspects of spacing behavior were restricted to males (the
bird analogy), and because reproductive cycles of females complicate
the analysis of their behavior.
The first attempt to see if aggressiveness varied with density was
made by Tamura (1966), working with Microtus californicus. She ran
bouts between 167 adult males brought in from a fluctuating population and tested in home cages in the laboratory. Thirty-three behavioral
variables were recorded from these male-male interactions, and over the
two years of study Tamura could detect no significant patterns of
change in the aggressive components of behavior.
A second attempt to see if male aggressiveness varied with density
was made by Krebs (1970), working with M . ochrogmter and M .
pennsylvanicus. Two classes of behavioral measurements were made.
Exploratory activity was measured in an open field, and aggressiveness
was measured in paired encounters in a neutral arena. These measurements were made on field animals brought into the laboratory for two
days and then returned to the field. Exploratory activity showed some
relationship to population changes, particularly in M . ochrogmter,
but the exploratory behavior scores of individuals were not useful
in predicting either the duration of life or the home range size of the
individual males. We would expect that, if we could measure behavior
accurately and if the measured type of behavior is related to the
demographic machinery of density changes, we could predict individual
attributes such as length of life from the behavioral data. Aggressiveness scores changed significantly in both species such that voles in peak
populations were most aggressive. Aggressive behavior profiles were
obtained for voles from increasing, peak, and declining populations.
Figure 40 illustrates profiles for M . pennsylvanicus. These profiles were
obtained from data collected from 1965 to 1967, and we used them to
predict population changes from 1967 to 1970 (Krebs, 1971). The
attempt to predict population parameters from aggressive behavior
data was only partly successful, and the relations between demography
and aggressive behavior were weak. Krebs (1971) suggested that there
might be three reasons for this: (1) aggressive behavior of female voles
might be more important than male aggressiveness; (2) the set of
behaviors measured might be a poor index of the important spacing
behaviors in social groups in nature; or (3) aggressive behavior may not
be an important factor in causing population changes in voles.
There has been relatively little work on the techniques of measuring
aggressive behavior in wild rodents. We are inclined to think that our
present techniques do measure something which is important in
368
CHARLES J. KREBS AND JUDITH H. MYERS
INCREASE PHASE
l4-d4+
‘1 1 , g ,
EXTREME
* o
W
sa o
3
LL
.
,
T
I
84
.
2
,
3
,
4
,
5
.
6
4
L
1
2
3
4
5
6
PEAK PHASE
EXTREME
0
2
W
8
4
W
0
3
0
a
LL
DECLINE PHASE
i4
4L
1
2
3
4
5
6
FIG. 40. Sample aggressive behavior profiles for 12 Microtw pennaylvanicus
males showing behavior typical of increase, peak, and decline phases, with
extreme cases shown at right. Frequency recorded in 10 min bouts. (After
Krebs, 1970.)
POPULATION CYCLES IN SMALL MAMMALS
369
nature. For example, Fig. 41 shows the trends in aggressive and
avoidance behavior of M . ochrogaster through a population decline and
subsequent increase. There was a marked change as the population
finished the decline and then began to increase, but we cannot explain
these changes in behavior because they are largely statistical effects.
When we can obtain accurate measurements on single individuals, it
PEAK
,
DECLINE
I
-1.4
!
I
PEAK
I
I
WINTER
WINTER I SUMMER
SUMMER
1968
1969
I
SUMMER WINTER
1968
INCREASE
I
I
SUMMER
,
I
1970
I
I
SUMMER
1969
WINTER
SUMMER
1970
FIG.41. Changes in the aggressiveness factor score and the avoidance factor
score (defined in Krebs, 1971) for Microtus ochrogmter through a population
cycle from 1968 to 1970 in southern Indiana. (Krebs and Myers, unpublished
data.)
may be possible to determine the dynamics of the changes which seem
t o occur in the different phases of the cycle.
Turner (1971) has made a detailed study of the annual cycle of
aggression in male M . pennsylvanicus from Manitoba. Aggressive
acts increased in frequency at the onset of reproduction and decreased
at the end of the breeding season. Turner was able to show that an
individual vole’s aggressive score was positively correlated with his
chances of surviving and negatively correlated with home range size.
370
CHARLES J. KREBS AND JUDITH H . MYERS
Heavier voles were usually more dominant in aggressive encounters. It
is tempting to interpret these data in terms of population fluctuations.
Two characteristics of peak populations are smaller home ranges
and larger-sized animals. Both these characteristics were found by
Turner (1971) to be related to aggressiveness in male voles. Turner’s
study is a model of the type of careful analysis of aggresssive behavior
that must be done if we are to understand the role of intraspecific
strife in affecting numbers. He is presently continuing observations
to determine whether aggressiveness is related to population density.
Conley (1971) measured aggressive behavior of male and female
M . longicaudus during a peak summer and a decline summer in the
mountains of New Mexico. He found that voles from a peak population
were more aggressive toward each other than were voles from a declining
population. Conley also studied a sympatric population of M . mexicanus
which was at low density during the two years. This population showed
no difference in aggressiveness between years. Aggressive reactions of
male and female voles were the same.
Social antagonism was analyzed by Getz (1972) by the use of multiplecapture live-traps for M . pennsylvanicus. Getz showed that many
multiple captures of two adult males occurred during his five-month
study, and he could see no evidence of great antagonism between males
in his declining population. However, the number of double captures of
adult females was significantly less than expected, which might suggest
some antagonism among females.
The movements of Norwegian lemmings, which are associated with
high densities, may be triggered by the aggressiveness of both male and
female residents (Clough, 1968). Unfortunately, no detailed studies
have been made on changes in agonistic behavior through a population
fluctuation of this legendary lemming.
Frank (1957) has argued that behavioral mechanisms are critical
in determining population changes in M . arvalis. He refers to these
behavioral mechanisms collectively as the “condensation potential”.
There are three major elements in the condensation potential. First,
home ranges can be reduced to small sizes. Both males and females
are territorial, and females drive off their male offspring and allow
the young females to settle in or near their home range. Second,
females can form mother-families and even “great families” at the end
of the breeding season, so that groups of immature animals can overwinter together. These great families break up once breeding begins
in the spring. I n some cases related females remain together on a
common territory and bring up their litters in a common nest. Third,
males of M . arvalis show spacing behavior at all times when they are
breeding, and consequently many more females than males occur in
POPULATION CYCLES IN SMALL MAMMALS
371
peak populations. Male mortality is thus much greater than female
mortality, and males fight vigorously for territories and show more
wounding.
The pattern of changes described by Frank (1957) for M . arvalis
should be evident in a changing sex ratio over a population fluctuation.
There has been no evidence from any of the studies on M . agrestis
(Chitty, 1952), M . ochrogaster and M . pennsylvanicus (Myers and
Krebs, 1971a) and M . californicus (Krebs, 1966) that sex ratios change
systematically over the population cycle. There is often a surplus of
females in the trappable population, but this seems to be a constant
feature of all the phases of the cycle. The absence of any sex ratio
change may mean that M . arvalis has a different pattern of social
structure to these other Microtus species.
More detailed studies are needed of the social structure of vole
and lemming populations. Some aspects of social behavior can be
studied in enclosed pens, but we badly need ways of studying field
populations in situ. Perhaps the simplest approach we can take at
this stage is to study male-male, male-female and adult-young interactions in a laboratory system, and then to use these observations as a
basis for designing field experiments. For example, we might manipulate
social organization by artifically changing the sex ratio in natural
populations. Alternatively, by cropping one sex and age group in a
population we might gain some insight on how social organization
influences population density.
An alternative series of experiments might involve the modification
of behavior with drugs. Implants of testosterone, for example, might
be used to make selected males aggressive. Nothing has been done so
far on vole and lemming populations to alter behavior with drugs.
We have said little so far about the possibility that behavioral
interactions can cause the mortality associated with declining populations. Except for the observations of Koshkina on Clethrionomys
rutilus, there is no evidence that behavioral interactions between voles
or lemmings are lethal, and it seems highly unlikely that adult animals
often die as a direct result of fighting. Some fraction of the loss of small
juveniles might be due to aggressive interactions with adults, and more
evidence is needed on this point. But we have studied fenced populations of Microtus ochrogaster and M . pennsylvanicus at very high
densities and not found any significant mortality of subadult and adult
voles (Krebs et al., 1969). If behavioral interactions are important
causes of population declines, they must act by forcing individuals to
succumb to other agents of destruction. A model for this suggestion
can be found in the muskrat (Errington, 1967) and the red grouse
(Jenkins et al., 1964). Socially inferior voles (non-aggressive animals)
372
C H a L E S J. m E B S AND JUDITH H. MYERS
might be more subject to loss by predation, bad weather, or stress
diseases. Note that only a small mortality change is necessary to
cause a population decline (cf. Fig. 24). A 10-15% drop in the probability of surviving per month would be sufficient to account for most
of the losses in declining populations. We are thus looking for a steady
mortality factor of small magnitude rather than a catastrophic mortality
factor of large magnitude.
We have shown previously that there seems to be little dispersal
during the decline phase, and most of the losses seem to be deaths in
situ. This result, if it is a general one, is difficult to reconcile with the
results of behavioral studies. The muskrat and red grouse models
would suggest that social intolerance should produce considerable
dispersal, and these dispersing animals (or those pushed into marginal
habitats) should suffer high mortality. But with voles we must, at
present, postulate that social intolerances in the decline phase somehow
lead to mortality but without much dispersal movement.
We cannot specify now what the behavioral attributes of dispersing
voles and lemmings should be. Most of the studies cited above suggest
that animals in peak populations are the most aggressive. We might
guess that more aggressive populations would produce more dispersal,
but this is not the case since dispersal rate is highest during the phase
of increase. Myers and Krebs (1971b) compared the aggressive behavior
profiles of male Microtus pennsylvanicus which had dispersed with
those of resident males, but found no clear and consistent differences
in the types of variables scored by Krebs (1970). Overall the dispersing
males tended to show characteristics of aggressiveness, and in one case
the “increase” behavior type (Krebs, 1970) was over-represented in
dispersing M . ochrogaster. We feel there must be some behavioral
reason why some individuals disperse and others remain as residents.
Perhaps the important traits which we would like to measure are so
transitory they disappear after dispersal has occurred. More work is
clearly required on these behavioral problems.
Behavioral interactions between individual rodents have been shown
to have strong effects on reproduction and growth in confined populations (review by Christian, 197lb). The physiological and endocrinological mechanisms involved in the suppression of reproduction and
growth have been particularly well studied in rodents by Christian and
his co-workers. The main gap at present is the specific application of
these findings to the details of changes found in field populations.
For example, we need to know whether there is a certain behavioral
milieu which permits winter breeding in some years and another
milieu which stops the breeding season early.
If behavioral changes are important in population cycles, we must
POPULATION CYCLES I N SMALL MAMMALS
373
determine the mechanisms behind the behavioral shifts. On the one
hand, behavioral changes can be caused by physiological shifts in
brain chemistry or hormonal balances caused by isolation or grouping
(reviewed by Brain, 1971a, b). These changes are usually considered as
phenotypic and are often tied in with the stress hypothesis. On the
other hand, behavioral changes could have a genetic basis and be tied
in with the genetic hypothesis of Chitty (1967). Obviously behavioral
changes will still have physiological and endocrinological mechanisms
even if they are genetically influenced. No one knows whether the
behavioral changes described above are phenotypic or genotypic, and
we suggest that these questions should be tackled in the logical order:
(1) does spacing behavior change over the population cycle? (2) can the
behavioral changes be shown to be heritable by the standard techniques
of quantitative genetics? and (3) what are the physiological pathways
by which the relevant spacing behaviors are iduenced? At present
most effort is being expended on question (3), even though we have
little data on question (1) and no data at all on question (2).
I n summary, the behavior hypothesis seems to hold the best possibility for explainjng the changes in reproduction, mortality, dispersal,
and growth which drive the population cycle. Spacing behavior, or
hostility, seems to produce a ccsurplus”population of animals which
move into vacant areas. If spacing behavior causes the population
fluctuations, the aggressive behavior of individuals must change with
density. This hypothesis has been verified both by examining skin
wounds and by paired-encounters of males in arenas. Male voles and
lemmings are most aggressive in peak populations. Little work has
been done on female aggressive behavior, and some workers suggest
that females may be even more aggressive than males. It is not clear
how behavioral interactions can account for the mortality changes
found in declining populations, since fighting itself rarely leads to
deaths. No one knows whether the behavioral changes that have been
found in microtine rodents are phenotypic or genotypic, and the
behavioral hypothesis could be subsumed under either the stress hypothesis or the genetic hypothesis once this is known.
F.
GENETICS
When Chitty (1958, 1960) first proposed his hypothesis that the
quality of microtines changed with density as a result of selection on
genetically determined behavioral types, two immediate objections
arose. The first of these was that the hypothesis was overly complex.
As stated by Pitelka (1958), “it may be a strain on Occam’s razor to
N
374
CHARLES J. RREBS AND JUDITH H. MYERS
suggest genetical hypotheses regarding fluctuations as long as more
directly ecological explanations can be invoked and tested”. The
second criticism was that selection could not be sufficiently strong
to bring about such dramatic genetic change over several generations
as to account for population fluctuations (Christian and Davis, 1964).
Both of these criticisms are examples of how only a short time ago, it
was the rule for the population ecologist to view populations largely as
genetically homogenous aggregates of individuals. While R.A. Fisher
(1930) considered that selective advantages of approximately 1 % per
generation were acting on natural populations, Ford (1964) reviews
cases demonstrating selective advantages of 20-30% per generation.
Ford concluded in his summary of “Ecological Genetics” that “unexpectedly great selective forces are normally operating” (p. 296). If
strong selection is possible in natural populations, we cannot automatically disregard the role of selection in population regulation.
Chitty’s genetic behavioral hypothesis to explain the cycling of
rodents began from a negative basis. All of the “simple” hypotheses
proposed t o explain microtine cycles had been unsatisfactory (Chitty,
1960). The positive basis for the behavioral-genetic polymorphism idea
was the observation that a t the time of the population decline, Microtus
agrestis populations are composed of some individuals with high growth
potentials and others with low growth rates (Newson and Chitty, 1962).
Therefore, the population is a composite of individuals of different
phenotypes. The presence of large voles in peak populations suggests
that the composition of the population at the peak is different than that
during the decline. Krebs (1964b) observed a change in the relationship
of skull measurements with body measurements associated with density
changes in brown and varying lemmings (see Fig. 28). This again
suggests the possibility of selection varying the phenotypic composition
of microtine populations during different phases of the population cycle.
Whether the basis for the phenotypic change is genetic could only be
suggested.
A starting point for investigation of the genetic-behavioral hypothesis
is to inquire if genetic changes do occur as a microtine population
undergoes fluctuations in density. Gershenson (1945) and Voipio (1969)
have suggested that shifts in genetically controlled coat-color morph
frequencies occur in association with density changes in rodents, but
in microtines individuals with coat-color variations are scarce
(Semeonoff, 1972), Another technique for elucidating genetic systems
is to monitor changes in the frequency of alleles at polymorphic loci,
coding proteins with differing structure (Semeonoff and Robertson,
1968; Canham, 1969; Tamarin and Krebs, 1969; Gaines and Krebs,
1971). The technique of electrophoresis enables one t o demonstrate
POPULATION CYCLES IN SMALL MAMMALS
375
structurally-varying proteins. Genetic polymorphisms in albumins and
transferrins (both serum proteins) and esterases and leucine aminopeptidase (both enzymes occurring in serum) have now been investigated
in microtines.
The h s t study of this type was that of Semeonoff and Robertson
(1968) who discovered a change in the gene frequency of an esterase
locus during a population decline of Microtus agrestis in Scotland.
Canham (1969) monitored albumins and transferrins in Clethrionornys
rutilus and C. gapperi populations in the Northwest Territories and
Qlberta and found a correlation between density and heterozygote
fitness. The studies of Tamarin and Krebs (1969) and Gaines and Krebs
(1971) covered five years’ observation of gene frequencies of Microtus
ochrogaster and M . pennsylvanicus. Periods of very strong selection
occurred on the two loci under study, leucine aminopeptidase and
transferrin (Fig. 42). Therefore, dramatic genetic changes can occur
in association with density changes; but how repeatable are these
observations? Many of the findings of Tamarin and Krebs (1969) were
verified in the continuing study of Gaines and Krebs (1971), which
included more populations over another population cycle. In both
studies of M . ochrogaster there was a positive correlation between the
frequency of the TfE allele and changes in population density (Fig. 43).
This relationship was found to be largely the result of better survival
of the heterozygote (TfE/TfF) during the population decline (Table
XXI). I n the six populations studied, male M . ochrogaster heterozygous
at the transferrin locus had either better or equal survival than the
TfE/TfE homozygote.
In M . pennsylvanicus, Tamarin and Krebs (1969) observed a negative
correlation between density and frequency of the TfE allele in females,
hile Gaines and Krebs (1971) found this relationship for males
Table XXII). As shown in Table XXIII, the survival differences
among genotypes in the two studies are not always consistent but tend
to show the same trends. Certainly the relationship between gene
frequency and density change is not a perfect one (Fig. 43) but there is a
definite statistical trend.
Not all local populations are genetically similar, however. One
population studied by Gaines and Krebs (1971) (Grid I) had frequencies
of the TfE allele which were much lower than the other populations.
It was most common for gene frequencies for the TfE allele to range
between 0.40 and 0.60, but in this one population the frequencies rarely
exceeded 0.40 and were most often between 0.20 and 0.40. Unlike the
other populations in which the TfE/TfE homozygote and the Tfc/TfE
heterozygote had superior survival rates, in this population the Tfc/Tfc
homozygote had a survival advantage. However, this population still
P
376
CRARLES J. KREBS AND JUDITH H. MYERS
00
&‘A.
.A-
34
JAN.
FEB.
i
I
MARCH
I
I
APRIL
MAY
FIG. 42. Genetic changes in association with density changes for Microtus
pennsylvanicus males on one area during the spring of 1969. (After Gaines and
Krebs, 1971.)
demonstrated a population fluctuation and was not demographically
distinctive from other populations.
Is this inconsistency contradictory to the genetic-behavioral
hypothesis? To answer this question we must consider what we are
doing by monitoring genetic changes using marker alleles. When this
To
study began the question was simply “do genetic changes OCCUT?’’
answer this we needed a genetic trait which was easily scored for
individuals without having to remove them from the populations.
Electrophoretic variants were the obvious tool. We did not predict that
the arbitrarily picked genetic trait would be the driving force behind
POPULATION CYCLES I N SMALL MAMMALS
TfE
frequency
decreasing
( 56.5 )
78
43
(64.5
Tf
frequency
increasing
35
( 56.5 )
86
(64.5 1
377
X 2= 30.70
~(0.005
FIG.43. Gene frequency changes and density changes for Microtus ochrogaster
in southern Indiana were scored for every bi-weekly trapping period. There is an
apparent relationship between increasing frequency of the TfE allele and increasing
density in Microtus ochrogaster (Gaines and Krebs, unpublished data).
cycles. But we did hope that, if strong selection were associated with
fluctuating density, it would change gene frequencies in the marker
alleles as part of the total genome. We can only guess that in a population such as that on Grid I, which had the “abnormally” low TfE
frequency, the transferrin locus is associated in a slightly different
linkage group so that selection in that population had different results
than in the other populations. This population was not geographically
isolated from other populations, and we would expect these differences
to be smoothed out over time.
If the marker alleles can be used as indicators of the genetic types
which change in frequency over the cycle, then we might be able to use
these genetic types further to investigate the demographic characteristics of cycling microtines. For example, we observed that in Microtus
ochrogaster the TfF allele is maintained at a very low frequency, but that
selection favors heterozygotes during the population decline (Table
XXI). With the beginning of the increase phase the TfE/TfE genotype
regains its selective advantage and the frequency of the TfE allele rises.
We predicted therefore that beginning with a low-density population,
TfE/TfE animals should respond with a greater rate of increase than
TfE/TfF individuals. The other homozygote W / T F rarely occurs in
nature and we predict that a population composed of all individuals of
this genotype should also do poorly.
37 8
CHARLES J . KREBS AND JUDITH H. MYERS
TABLEXXI
Comparison of the survival rates per 14 days for the transferrin genotypes of Microtus
ochrogaster males in s i x populations studied. A1 includes data collected during the
first population cycle on area A , and A 2 is from the second cycle on the same area.
Underlined values are the higher for that set of data.
Area
Population
change
Genotype
F
H
I
A1
EE
0.74
0.87
-
0.81
0.89
-
-
0.69
0.64
0.59
0.75
Increase
EF
EE
Decrease
EF
-
0.65
0.78
-
0.81
~
0.84
~
0.84
0.76
A2
C
-
0.85
0.85
-
0.68
0.75
0.79
0.68
-
0.70
0.75
TABLEXXII
Comparison of the relation between the frequency of the transferrin TfE allele and
density in Microtus pennsylvanicus in the studies of Tamarin and Krebs (1969)
and Gaines and Krebs (1971).
Correlation coefficients
Tamarinl
Geines2
Males
Fem a1es
0.06
- 0.67**
-0*18*
0.03
Relationship of TfE frequency and density (based on one population).
Relationship between change in TfE frequency and change in density (based on four
pOpUI8t~S).
* P<*05
** P<.Ol
TABLEXXIII
Compariaon of the survival rates per 14 days for the transferrin genotypes of Microtus
pennsylvanicus revealed in the studies of Tamarin and Krebs (1969)and Gaines
and Krebs (1971).Underlined zcaluea are the highest for that set of data.
Genotype
Tfc/Tfc
Tfc/TfE
TfE/TfE
Tamarin
Increasing
Decreasing
0.82
0.95
0.90
0.76
0.77
0.78
-
Gaines
Increasing
Decreasing
0.84
0.84
0.94
-
0.66
0.75
0.75
-
POPULATION CYCLES I N SMALL MAMMALS
379
To test these predictions, field populations of M. ochrogaster were
established in two-acre enclosures in Indiana with three founder
populations of 20 individuals all of one genotype (Gaines et al., 1971).
Three such introductions were made in one year so that by the end of
the experiment we had data for three populations of each of the three
transferrin genotypes. The surprising result was that we could find no
statistical differences between populations established with different
genotypes. Populations of each of the genotypes were capable of rapid
population increase. I n all three populations started with heterozygotes
there was an increase in the frequency of the TfE allele by the end of the
experiment, which covered four months and included two generations.
Because of this we suggest that when the TfF/TfF genotypes are free
from competition with the other genotypes they are able to do well,
but in mixed populations they are at a disadvantage.
The failure to verify predictions about the effect of selection of
M. ochrogaster transferrin genotypes might be explained by the fact
that the enclosures, which prevented dispersal, disrupted the normal
demographic machinery. We suggest that perturbations of field
populations may be more meaningful (Gaines et al., 1971; Krebs, 1971).
This could be accomplished by switching homozygotes of two types
between populations and observing the demographic response.
Genetic markers can be used as the basis for classification of microtines into increase, peak or decline categories, but there will be considerable noise in the classification system. On the basis of this work with
transferrin in M. ochrogaster and M . pennsylvanicus, Tamarin, Gaines
and Krebs were able to distinguish increase and decline types based on
the comparative success of the genotypes during these population
phases. We would like to be able more directly to distinguish increase,
peak and decline individuals based on those attributes which give them
the advantage during these population phases. A pertinent question is:
Is there a rapid growth genotype which is favoured during the increase
phase or is there a genotype which is capable of withstanding crowded
conditions of peak populations but is less resistant to mortality factors?
The problem is to devise a system in which the heritability of the factors
can be scored.
Heritability studies of ecologically important attributes would seem
to offer a promising area for future work. For example, heritability of
growth could be determined for microtine rodents. Few studies have
attempted to determine the growth potential of animals in cycling
populations. Newson and Chitty (1962) found animals of high and low
growth potentials in declining populations. Krebs (1966) showed that
some voles from a declining population of M. californicus would grow
to large sizes if they were removed to the laboratory, although little
380
CHARLES J. KREBS AND JUDITH H. MYERS
growth was found in field animals. Myers and Krebs (1971b) with a
very small sample showed that male M . ochrogaster dispersing from an
increasing population had a low growth potential. No one has monitored,
under standard conditions in the laboratory, growth of microtines taken
from various phases of the population fluctuation.
If specific behavior patterns which were measurable in the laboratory
could be found to be associated with changes in population density,
these might serve as more meaningful genetic markers. We were largely
unsuccessful in our attempts to find a laboratory measurement which
would distinguish dispersing and non-dispersing M . ochrogaster and
M . penmylvanicus (Myers and Krebs, 1971b).Although there is evidence
for genetically determined behavioral types in house mice (Van
Oortmerssen, 1970), no work has begun to determine the genetic
component of behavior in microtines.
There may be ways to obtain individuals of different types if we
let the field situation act as the selection process. As discussed earlier,
dispersal is a particularly important process during the phase of
population increase. A possible explanation for the higher density
of enclosed populations is that the prevention of dispersal has forced
individuals which would normally have left to stay in the population.
Consequently, the quality of individuals of high populations in enclosures should be different from that of natural peak populations.
Our data on dispersal in M . pennsylvanicus (Myers and Krebs, 1971b)
indicated that those individuals with the highest reproductive potential
were dispersing. These were young females which had just become
sexually mature (Fig, 44). Using marker alleles at the transferrin
locus we found that there was a genetic component to dispersal among
females during the phase of population increase.
We suggest that instead of comparing the reproductive potential
of transferrin genotypes, we should look a t the reproduction potential
of dispersers and non-dispersers, of animals from peak populations and
declining populations, of animals from high-density populations in
enclosures (those which have not been selected by dispersal) and from
high-density natural populations. If we are to test the Chitty hypothesis,
we must be concerned with demographically pertinent genotypes rather
than marker loci which may be variously linked with loci under
density selection.
I n summary, data collected on the genetics of cycling vole populations
by the use of electrophoretic variants as genetic markers demonstrate
dramatic genetic changes associated with density changes. I n some
cases ( M . ochrogaster and M . penmylvanicus) there are significant
correlations between changes in gene frequency and changes in density,
and these changes show some consistency over two population cycles
POPULATION CYCLES IN SMALL MAMMALS
381
and among different local populations. Furthermore, characteristics of
fitness such as survival and growth show generally consistent patterns,
although there is some variation.
, While there is now general agreement that populations are composites of organisms with considerable genetic variation, and that the
genetic make up of populations can change in association with density
q
MICROTUS PENNSYLVANICUS
20
10
>
0
VAGINAL
ORIFICE
PERFORATE
GRID
F
GRID GRID
I
K
FIG.44. Grids I and F are control populations of resident animals and Grid K
represents a dispersal population which moved into an area of vacant habitat.
The proportion of young females in the dispersal population is greater than that
in either of the control populations and the proportion of these females which are
sexually mature (vaginal orifice perforate) is also greater among the young
dispersing females. (Myers and Krebs, unpublished.)
fluctuations, the question still remains as to whether genetic changes are
the driving force behind demographic changes or whether density
fluctuations and variation in natality and mortality associated with
them cause the fluctuations in gene frequencies. This problem has been
discussed in depth elsewhere (Gaines and Krebs, 1971; Krebs, 1971;
Krebs et al., 1973) but we will reiterate it here.
There is no simple way to separate cause and effect in this situation.
If genetic changes were the driving force, it might be expected that
382
CHARLES J. KREBS AND JUDITH H. MYERS
there would be a lag between the gene frequency change and the density
change as occurs in the age-specificselection model of King and Anderson
(1971). This does not seem to be the case (Fig. 42). However, if there are
genotypes which grow faster and reproduce earlier during the phase of
population increase as suggested by the work of Gaines and Krebs
(1971), the population increase will result in a higher proportion of this
genotype unless mortality compensates for the greater reproductive
potential .
Charlesworth and Giesel (1972) propose a model which considers the
influence of demographic structures, particularly the age structure,
on changes in gene frequency of polymorphic systems. The basis of the
model is that if a genotype produces early in its life, there will be little
advantage if the population is primarily composed of older individuals
which are already reproducing. However, if the age structure favors
growing animals, those individuals which are able t o reproduce early
will have a greater advantage. Therefore, if there is a change in the age
structure of the population, a fluctuation in gene frequency follows.
While these authors claim that their model produces a population cycle
and gene-frequency change which resembles the observations of
Tamarin and Krebs (1969), we see some differences. First, the genefrequency changes produced by the model were only in the order of
6-9% while observed changes in Microha populations are more often
in the vicinity of 20% (Fig. 42).
Furthermore, if it is assumed that a cycle of the model is equal to
the period of time for an individual to reproduce (about three weeks for
microtines), the rate of the gene-frequency change is very slow as
compared with those of field studies. With the model a 9% genefrequency change required 40 cycles. This would be the equivalent of
over two years on a microtine time scale which clearly doesn’t apply
to field observations.
One of the observations arising from the field data is that gene
frequencies in males and females are often quite different and it is not
clear that the model would allow this result. The Charlesworth-Giesel
model showed changing death rates to have little influence on gene
frequencies. We have found that differential survival particularly
during the decline is a major factor in gene-frequency changes.
Finally we find it impossible to interpret the correspondence between
simulated population declines caused by decreasing m(x) functions and
declines of field populations which arise both from decreased natality
and from increased mortality. We feel that the responsibility lies with
the authors of theoretical models to interpret the biological consequences
of assumptions made in their models, for without them statements
such as “The resultant population cycle resembles vaguely the pattern
POPULATION CYCLES IN SMALL MAMMALS
383
observed in oscillating vole populations” (Charlesworth and Giesel,
1972) are very misleading.
One of the criticisms made by Charlesworth and Giesel (1972) of
the results showing genetic changes in fluctuating populations is that
the several systems arbitrarily chosen as genetic markers have all
shown a relation to population density. This criticism is not a serious
one if in fact periods of dramatic density changes are associated with
strong selection. In this case the whole genome would be expected to be
under selection, and a large proportion of individual loci would most
likely change in frequency. The females of both Microtus ochrogaster
and M . pennsylvanicus demonstrated significant correlations between
change in gene frequency and change in density for one of the two
systems studied by Gaines and Krebs (1971) but not for the other.
Significant correlations between gene frequency and density occurred
for both loci in males of the two species. We interpret the general trend
for all genetic systems which have been studied to show a relation
between genetic change and density change, to indicate strong selection
occurring with population fluctuations. However, we cannot explain
the exceptions which occurred in female M . ochrogaster and M .
pennsylvanicus which failed to show a relationship for one of the
two loci studied (Gaines and Krebs, 1971).
In conclusion, data have not yet contradicted the Chitty geneticbehavioral hypothesis of microtine cycles, but we have only accomplished
the preliminary steps in testing the hypothesis. Genetic changes seem
to be a part of population fluctuations in microtines, and this aspect
of genetic heterogeneity will have to be taken into consideration in
future studies of microtine cycles. Our studies have indicated that
dispersal, particularly during the phase of increase, is more important
than originally thought by Chitty. We propose a modified version of
the Chitty Hypothesis in Fig. 45. Two variations are included in this
version which were not included in the model diagrammed by Krebs
(1964a). The first of these is the loop which allows for emigration
and the colonization of new habitats and the possible establishment
of refuge populations. The second change is that rather than emphasizing
aggressive behavior, we feel that other forms of spacing behavior
could work in a similar way. For example, very docile individuals
might be at a selective advantage as the population increases. Selection
for aggressive behavior could also lead to susceptibility to other
selective factors. Originally, it was thought that aggressive behavior
among individuals might be the direct cause of the population decline.
However, until we know the nature of the mortality factors acting
during the decline we will not; be able to interpret this aspect of the
hypothesis.
384
CHARLES J. KREBS AND JUDITH H. MYERS
INCREASE I N NUMBERS
/”
MUTUALjNTERFEREN\
REDUCED’ GROWTH OF
EMIGRATION^ COLONIZATION
RESIDENT POPULATION
OF VACANT
AREAS
BECAUSE OF:
R E LOWER
PRODUCT
\
j O R INCREASED
TALlTY
OTHER
CONTINGENCIES
SELECTION FOR
SPACING BEHAVIOR AND
J
REFUGE
POPULATIONS
IN MARGINAL
HABITATS
AGAINST OTHER
ATTRIBUTES
REDUCED POPULATION SIZE
1
\
I
J.
REDUCED INTERFERENCE
/
IMMIGRATION
1
SELECTION FOR H I G H
REPRODUCTIVE POTENTIAL
INCREASE IN NUMBERS
FIG.46. Modified version of the Chitty behavioral-genetic hypothesis to explain
microtine cycles. Dispersal is viewed as being a more important aspect than
originally proposed by Chitty. Central to the hypothesis is selection acting
through behavioral interactions and changing the genetic composition of the
population with fluctuating densities.
V I I . E V O L U T I OON
F M I C R O T I N EC Y C L E S
Are microtine cycles an adaptation? This question has been mentioned
only peripherally in the literature. We should note that if we support
the extrinsic factor hypotheses-weather, food shortage, predation, or
disease-this question is relatively meaningless, since in these cases
population fluctuations are in effect forced on the population by outside
agents. However, if we support the intrinsic factor hypotheses-stress,
behavior, genetics-this question is important because populations
must have evolved a mechanism of self-regulation to arrest population
growth below the limits set by starvation. We therefore continue this
discussion on the assumption that some mechanism of self-regulation
occurs in voles and lemmings.
We might imagine the following scenario in the evolution of microtine
rodents. These small animals are subject to an array of hazards, from
bad weather to predators and diseases. If they had only bad weather
to cope with, they could no doubt survive by having a good reproductive potential and behavioral adaptations for tunneling and nest
POPULATION CYCLES I N SMALL MAMMALS
385
building. To cope with predators was perhaps a more serious challenge,
particularly with ground predators such as weasels, and we might
imagine that even higher reproductive potentials would be necessary
to keep from going extinct. Those rodent species which could overcome
these limitations now found themselves up against another problem.
Excessive reproductive capacity carries with it the seeds of habitat
destruction and starvation, particularly in areas where predators are
less common. From this challenge rodent species must have evolved
some form of self-regulation. This could have been achieved by
individual selection or by group selection. There are obvious advantages
to a group of rodents from not destroying the habitat, and we can
easily see that group selection would favor the evolution of selfregulation. The more dificult question is whether individual selection
would lead in the same direction.
Spacing behavior can evolve readily on the basis of individual
selection. A vole which begins to drive away his neighbours by physical
aggression would be at a selective advantage when crowding became
serious. Brown (1964) has discussed the evolution of territoriality in
birds, and has pointed out that aggressiveness is primarily a behavioral
response to competition for resources which are in short supply. Since
food for voles and lemmings cannot be defended readily, we would
suggest that breeding space is probably the resource in short supply
as density rises.
Once spacing behavior begins to evolve, the next problem is to set
the upper limits to such behavior. If some aggressiveness is good in
high-density populations, is not more aggressiveness even better?
There are some obvious limits to excessive aggressiveness. Very
aggressive rodents might find it impossible to mate, or they might
destroy their own offspring. There are at least some vague upper limits
to aggressiveness. But why does the population not stabilize at some
intermediate level of aggressiveness? I n order to fluctuate periodically,
a microtine population must continually be overshooting the"optimum"
level of aggressiveness.
The solution to this dilemma may lie in the fluctuating seasonal
environments in which microtines live. Two extreme end-points can
be selected toward. At the low density end of the scale there is no
premium on aggressive behavior, and the selective premium falls on
genotypes capable of breeding late into the fall and winter, maturing
early, and growing rapidly. At the high density end of the scale, there
is a great premium on aggressive behavior, and reproductive behavior
is shifted toward deferring maturity until a later date when density has
fallen. SchaEer and Tamarin (1973) have analyzed how reproductive
effort should be expended in fluctuating populations in order to
386
CHARLES J. KREBS AND JUDITH H. MYERS
maximize individual fitness. They concluded that the observed patterns
of reproductive changes in cyclic rodents were consistent with the
hypothesis that individuals attempt to maximize fitness in all phases
of the cycle.
We can thus see how a polymorphism in spacing behavior might be
maintained in a population by means of time lags in adjusting to
density changes. This, however, leads us to a further question of
whether the behavioral and reproductive changes need be genetically
determined or could be the plastic properties of a flexible phenotype.
We do not know the relative advantages and disadvantages of adopting
a phenotypic self-regulating mechanism or a genotypic mechanism.
We see no way of determining a priori which way evolution should move
within the general scope of self-regulatory mechanisms.
Why does natural selection not damp out population fluctuations in
rodents? There are some genetic advantages to population fluctuations
that might answer this question. Carson (1968) has discussed the
genetic consequences of the “population flush” in insects. The low
density after the decline allows both the testing of new genotypes and
the possibility of rapid spread of new genetic combinations because of
founder effects. If this is true, we would predict a more rapid rate of
evolutionary change in species of voles and lemmings than in species
of rodents which maintain stable numbers. If we can postulate that
there are genetic advantages to population fluctuations, we could then
see why evolutionary changes have reinforced the fluctuations rather
than damping them away.
In summary, population regulation mechanisms must be under some
evolutionary control, if self-regulatory hypotheses are correct. Periodic
fluctuations in microtine rodents could be generated by individual
selection toward two extreme morphs: a low-density, docile, reproductive form; and a high-density, aggressive, less reproductive form.
Time-lags in responding to density changes could generate a cycle.
Such a self-regulatory system would be preserved by natural selection
if there are some genetic advantages to population fluctuations.
V I I I . SUMMARY
Population cycles in voles and lemmings are accompanied by a
series of changes which we summarize here in point form:
1. The increase phase is the least variable phase and may be over
very quickly.
2. The decline phase is the most variable phase; it may be very
rapid or may be slow and prolonged over two years. ((Crash”
declines over one or two weeks are not typical.
POPULATION CYCLES IN SMALL MAMMALS
387
3. The periodicity is variable, three to four years is typical, but
some cycles may be two or five years in length.
4 . Some populations may not fluctuate, but none of this type has
been stddied carefully.
5 . Fluctuations occur in a variety of genera and species from arctic
to temperate areas, from Mediterranean to continental climates,
from snowy areas to snow-free areas.
6. Fluctuations sometimes occur in synchrony ovkr large geo-
graphical areas (thousands of square miles). Synchrony is
seldom absolute, however, and local out-of-phase populations
occur. Synchrony is not continental or world-wide.
7. The amplitude of these fluctuations is not necessarily larger
in more northern populations.
8. Populations living in a wide variety of plant communities in a
small geographic area all fluctuate in the same way, often in
phase.
9. Reproductive#rate is highest in the increase phase, owing to ( 1 )
longer breeding season, including winter breeding in some
species, and (2) lower age at sexual maturity. I n the peak and
decline phases reproductive rate is reduced.
10. There is no systematic difference between increasing populations
and declining populations in (1) litter size, (2) percentage of
adult females pregnant during the breeding season, or (3) sex
ratios.
11. Mortality rates in all sex and age groups are lowest in the increase
phase.
12. Adult and subadult mortality rates are low during the peak,
and high in the decline phase.
13. Juvenile mortality may be high in the peak summer and may
remain high until the end of the peak breeding season, when
it is suddenly reduced. Juvenile losses are very high in the
decline phase.
14. Survival of adult males fluctuates independently of that of
adult females, when viewed on a weekly time scale. Males may
suffer heavy losses in the decline for a few weeks when females
are surviving very well, and vice versa.
388
CHARLES J. KREBS AND JUDITH H. MYERS
15. Prenatal mortality may vary slightly over the cycle but is not a
serious loss even in declining populations.
16. Dispersal is most frequent from increasing populations, and
relatively infrequent from declining populations.
17. Two closely related species may live in the same habitat and
reach peak numbers in the same year, but declines in one
species may occur while the other species in the same area
remains a t high numbers for several months.
18. Voles brought into the laboratory from field populations which
are declining will live for a very long time as isolated pairs or
individuals.
19. Populations kept in small cages in the laboratory or room-sized
cages outdoors increase to unnaturally high densities, several
to many times that found in nature, and do not cycle.
20. Populations in two-acre enclosures in the field increase to
unusually high numbers and may reach the limit of their food
supply and starve.
21. Adult animals in peak populations are typically larger (by
20-50%) than those at other times in the cycle. This increase in
body size is reflected in changed skeletal proportions. These
large animals may occur in the late increase phase and early in
the decline phase when the decline is very gradual.
22. Changes in gene frequency at marker loci occur in association
with density fluctuations.
23. Aggressiveness of male and female microtines increases and
home range size decreases with increasing population density.
We have attempted throughout this article to point out experiments
which might be done to test various hypotheses and elucidate certain
aspects of microtine cycles. Two of the most important questions which
remain to be answered are: (1) what permits longer reproductive seasons
in some years? and (2) what is the nature of the mortality occurring
during the decline? Studies of either of these questions should not
overlook the quality of the individuals in the population. While it may be
possible to show what nutritional factors permit microtines to reproduce, the crucial aspect to relate nutrition to population cycles might
be to demonstrate that individuals in increasing populations have lower
nutritional thresholds or are more efficient in their use of nutritional
factors. Similarly, there might be a number of causes of mortality
POPULATION CYCLES I N SMALL MAMMALS
389
during the population decline, with the important variable being that
the overall resistance of individuals from peak populations is lower
than that of individuals at low or increasing densities. We emphasize
that we should be concerned with the variability demonstrated by
individuals composing populations.
We think that enough is now known about the natural history of
microtine fluctuations for us to be able to devise some simple experiments to test alternative hypotheses. Field experiments in ecology
are particularly difficult to replicate properly, and part of our effort
must be geared to replicating experiments over several populations of
several species. If we spend more effort on devising, executing and
interpreting experiments, we may find less time devoted to specsc
schools defending one hypothesis at all costs. We would discourage
simple descriptive studies of microtine populations, even of species
yet unstudied, unless they are coupled with some experimental analysis.
Finally, we cannot resist making a prediction about the future.
We feel that studies of the heritability of reproductive capabilities,
growth potentials, and behavior of microtines will be the key to
unlocking the mystery of rodent cycles.
ACKNOWLEDGEMENTS
We wish to acknowledge our colleagues formerly at Indiana University, Michael Gaines, Barry Keller and Robert Tamarin, whose
efforts contributed so substantially to much of our own work reviewed
in this article. Rey Stendell and Paul Whitney made available to us
as yet unpublished results from their Ph.D. studies and we are very
grateful to be able to include their findings in this paper. The support
of the National Institutes of Health, the National Science Foundation,
the National Research Council of Canada and the Miller Institute for
Basic Research of the University of California, Berkeley made possible
our studies reported and reviewed here, and for this support we are
most appreciative.
REFERENCES
Adamczyk, K. and Walkowa, W. (1971). Ann. 2002.Pennici 8, 146-163. Compensation of numbers and production in a Mus muaculua population aa a
result of partial removal.
Anderson, P. K. (1970).Symp. Zool. SOC.Lond.26, 299-326. Ecological structure
and gene flow in small mammals.
Andrews, R. V. (1968). Phyeiot. Zool. 41, 8&94. Daily and seasonal variation in
adrenal metabolism of the brown lemming.
Andrews, R. V. (1970). Acta Endocrinologica 65, 639-644. Effects of climate
and social pressure on the adrenal response of lemmings, voles and mice.
0
390
CHARLES J. RREBS AND JUDITH H. MYERS
Andrews, R. V. and Strohbehn, R. (1971). Comp. Biochem. Physiol. 38A, 183-201.
Endocrine adjustments in a wild lemming population during the 1969
summer season.
Andrzejewski, R. and Rajska, E. (1972). Acta Theriol. 17, 41-56. Trappability of
bank vole in pitfalls and live traps.
Andrzejewski, R., Fejgin, H. and Liro, A. (1971). Acta Theriol. 16, 401-405.
Trappability of trap-prone and trap-shy bank voles.
Arata, A. A. (1967). In “Recent Mammals of the World” (S. Anderson and
J. K. Jones Jr., Eds). Ronald Press, New York. Muroid, Gliroid, and
Dipodoid Rodents.
Ashby, K. R. (1967). J . 2001.Lond. 152, 389-613. Studies on the ecology of
field mice and voles (Apodemw aylvaticw, Clethrimomys glareolw and
Microtw agrmtk) in Houghall Wood, Durham.
Batzli, G. 0. and Pitelka, F. A. (1970). Ecology 51, 1027-1039. Influence of
meadow mouse populations on California grassland.
Betzli, G. 0. and Pitelka, F. A. (1971). J. Mamma?. 52, 141-163. Condition and
diet of cycling populations of the California vole, Microtw cal$~rnicu8.
Bodenheimer, F. S . (1949). Problems of vole populations in the Middle East.
Report on the population dynamics of the Levant Vole (Microtus guentheri
D. et A.). Res. Council of Ierad, Jerusalem.
Brain, P. F. (1971a). Communs. Behawl. Biol. 6 , 116-123. The physiology of
population limitation in rodents-A review.
Brain, P. F. (1971b). Communa. Behavl. Biol. 6, 7-18. Some physiological and
behavioral consequences of hormonal modifications in the early life of
rodents-A review.
Brant, D. H. (1962). Univ. Calif. Pubk. 2001.62, 105-184. Measures of the movements and population densities of small rodents.
Brown, J. L. (1964). Wikon Bull. 76, 160-169. The evolution of diversity in
avian territorial systems.
Canham, R. P. (1969). Ph.D. Thesis, Univ. Alberta. 121 pp. Serum protein
variation and selection in fluctuating populations of cricetid rodents.
Carson, H. L. (1968). In “Population Biology and Evolution” (R. C. Lewontin,
Ed.), pp. 123-137. Syracuse Univ. Press, Syracuse, New York. The population flush and its genetic consequences.
Charlesworth, B. and Giesel, J. T. (1972). Am. Nat. 106, 402-411. Selection in
populations with overlappinggenerations. IV.Fluctuations in gene frequency
with densitydependent selection.
Chitty, D. (1962). Phil. Tram. €3. SOC.Ser. B 236, 505-552. Mortality among
voles (Microtw agrmtk) a t Lake Vyrnwy, Montgomeryshire in 193639.
Chitty, D. (1965). In “The Numbers of Man and Animals” (J. B. Cragg and
N. W. Pirie, Eds),pp. 57-67. Edinburgh. Adverseeffects of population density
upon the viability of later generations.
Chitty, D. (1968). Cold Spring Harb. Symp. Qwcnt. Biol. 22, 277-280. Selfregulation of numbers through changes in viability.
Chitty, D. (1969). Ecology 40, 728-731. A note on shock disease.
Chitty, D. (1960). Can. J . ZooZ. 38, 99-113. Population processes in the vole and
their relevance to general theory.
Chitty, D. (1967). Proc. ecol. SOC.A w t . 2, 61-78. The natural selection of selfregulatory behaviour in animal populations.
Chitty, D. (1969). Am. 2002.9, 400. Regulatory effects of a random variable.
POPULATION CYCLES IN SMALL MAMMALS
391
Chitty, D. and Chitty, H. (1962). Symp. Therwlogicum Bmo, 1960, 67-76.
Populations trends among the voles a t Lake Vyrnwy, 1932-1960.
Chitty, D. and Phipps, E. (1966).J. Anim. Ecol. 35,313-331. Seasonal changes in
survival in mixed population of two species of vole.
Chitty, D., Pimentel, D. and Krebs, C. J. (1968).J. Anim. Ecol. 37, 113-120.
Food supply of overwintered voles.
Chitty, H. (1961).J. Endocr. 22, 387-397. Variations in the weight of the adrenal
glands of the field vole (Microtw, agreatie).
Chitty, H. and Chitty, D. (1962).Symp. Theriologicum, B m ,1960, 77-86. Body
weight in relation to population phase in Microtus agretb.
Christian, J. J. (1950).J . Mammal. 31, 247-259. The adreno-pituitary system
and population cycles in mammals.
Christian, J. J. (1961).Proc. natn. Acad. Sci. U.S.A. 47, 428-449. Phenomena
associated with population density.
Christian, J. J. (1971a). Biol. of Reprod. 4, 248-294. Population density and
reproductive efficiency.
Christian, J. J. (1971b).J . Mammal. 52,556-567.Fighting, maturity, and population density in Microtus pen.nsylvanicuS.
Christian, J. J. and Davis, D. E. (1964).Science 146, 1550-1560. Endocrines,
behavior and population.
Christian, J. J. and Davis, David E. (1966).J. M a m d . 47,l-18. Adrenal glands
in female voles (Microtwr pennqlvanicw) as related to reproduction and
population size.
Christian, J. J., Lloyd, J. A. and Davis, D. E. (1965).Recent Prog. HOWL Rea.
21, 501-578. The role of endocrines in the self-regulation of mammalian
populations.
Clarke, J. R. (1955).Proc. R. SOC.Ser. B 144, 68-85. Influence of numbers on
reproduction and survival in two experimental vole populcttions.
Clarke, J. R. and Forsyth, I. A. (1964). Ben. Comp. Endocr. 4, 233-242.
Seasonal changes in the gonads and accessory reproductive organs of the
vole (Microtwr uqreatk).
Clough, G . C. (1968).Papere of the Norwegian State Game Rea. Inst. Ser. 2, No.
28, 1-49. Social behaviour and ecology of Norwegian Lemmings during a
population peak and crash.
Cole, L. C. (1954).&. Rev. Biol. 29, 103-137. The population consequences of life
history phenomena.
Collett, R. (1895). “Myodea lemmw: Its Habits and Migrations in Norway”.
Christiania Videnskabs-Selskabs Forhandlinger 1895, No. 3.
Conley, W. H. (1971).Ph.D. Thesis, Texas Tech. Univ. Behavior, demography
and competition in Microtwr longicaudw and M . mexicanus.
Cormack, R. M.(1968).Oceanogr.Mar. Biol. Ann. Rev. 6,455-506. The statistics of
capture-recapture methods.
Craighead, J. J. and Craighead, F. C. (1969).“Hawks, Owls and Wildlife”, 433 pp.
Dover Publications, Inc., New York, 2nd Edition.
Dahl, A. E. (1967).M.Sc. Thesis, UNv. of Calgary. “Responses of free-living vole
populations to experimental modifications in density.”
DeLury, D. B. (1947). Biometrice 3, 145-167. On the estimation of biological
populations.
Dunaeva, T. N. and Kucheruk, V. V. (1941).Material on the Fauna and Flora
of the U.S.S.R. No. 4 (19),1-80. Material on the ecology of the terrestrial
vertebrates of the tundra of South Yamal. (In Russian.)
392
CHARLES J. KREBS AND JUDITH R. MYERS
Dymond, J. R. (1947). Tram. R.SOC.Can. 41, (5), 1-34. Fluctuations in animal
populations with special reference to those of Canada.
Egerton, F. N. (1968). J . Hiat. Biol. 1, 225-259. Studies of animal populations
from Lamarck to Darwin.
Elliott, P. W. (1969). Ph.D. Thesis, Univ. of Alberta. Dynamics and regulation
of a Clethrionomya population in Central Alberta.
Elton, C. S. (1924). Br. J. exp. Biol. 2, 119-163. Periodic fluctuations in the
numbers of animals: their causes and effects.
Elton, C. S. (1942). “Voles, Mice and Lemmings”. 496 pp. Clarendon Press,
Oxford.
Errington, P. L. (1967). “Of Predation and Life”. 277 pp. Iowa State Univ.
Press, Ames.
Evernden, L. N. and Fuller, W. A. (1972). Can. J . Zool. 50, 1023-1032. Light
alteration caused by snow and its importance to subnivean rodents.
Fisher, R. A. (1930). “The Genetical Theory of Natural Selection”. 291 pp.
Dover Publications, New York.
Fitzgerald, B. M. (1972). Ph.D. Thesis, Univ. California, Berkeley. 143 pp. The
role of weasel predation in cyclic population changes of the montane vole
(Microtua montanm).
Fleharty, E. D. and Olson, L. E. (1969). J . Mammal. 50, 475-486. Summer food
habits of Microtua ochrogaeter and Sigmodon hiapidua.
Ford, E. B. (1964). “Ecological Genetics”. 335 pp. Broadwater Press, Welwyn
Garden City, Herts.
Frank, F. (1953). Zool. J b . (Syat.) 82, 95-136. Untersuchungen uber den Zusammenbruch von Feldmausplagen (Microtua amtalia Pallas).
Frank, F. (1957). J . Wildl. Mgmt. 21, 113-121. The causality of microtine cycles
in Germany.
Fuller, W. A. (1967). Terre Vie 114, 97-115. Ecologie hivernale des lemmings et
fluctuations de leurs populations.
Fuller, W. A. (1969). Ann. 2001.Fennici 6, 113-144. Changes in numbers of
three species of small rodent near Great Slave Lake, N.W.T. Canada,
1964-1967, and their significance for general population theory.
Gaines, M. S. and Krebs, C. J. (1971). Evolution 25, 702-723. Genetic changes in
fluctuating vole populations.
Gaines, M. S., Myers, J. H. and Krebs, C. J. (1971). Evolution 25, 443-450.
Experimental analysis of relative fitness in transferrin genotypes of Microtua
ochrogaater.
Gebczynska, Z. (1970). Acta Theriol. 15, 33-66. Bioenergetics of a root vole
population.
Gentry, J. B. (1968). Rea. Popul. Ecol. 10, 21-30. Dynamics of an enclosed
population of pine mice, Microtua pinetorum.
Gershenson, S. (1945). Genet& 30,207-251. Evolutionary studies on the distribution and dynamics of melanism in the hamster (Cricetua crketuo L.).
Getz, L. L. (1960). Am. M a l . Nat. 64, 392-405. A population study of the vole,
Microtua pennaylvanicua.
Getz, L. L. (1972). J . Mammal. 53, 310-317. Social structure and aggressive
behaviour in a population of Microtw pennaylvanicua.
Godfrey, G. K. (1953). Saugetierk. Mitt. 1, 148-151. The food of Microtua w r a t h
L. in Wytham, Berkshire.
Godfrey, G. K. (1955). J . Mammal. 36, 209-214. Observations on the nature of
the decline in numbers of two Microtzce populations.
POPULATION UYCLES IN SMALL MAMMALS
393
Golley, F. B. (1960). Ecol. Monogr. 30, 187-206. Energy dynamics of a food chain
of an old-field community.
Green, R. G. and Larson, C. L. (1938). Science 87, 298-299. Shock disease in
the snowshoe hare cycle.
Grodzinski, W. (1971). Acta Theriol. 16, 231-275. Energy flow through populations of small mammals in the Alaskan taiga forest.
Grodzinski, W., Bobek, B., Drozdz, A. and Gorecki, A. (1969). I n “Energy
Flow Through Small Mammal Populations” (K. Petrusewicz and L.
Ryszkowski, Eds), pp. 291-298. Energy flow through small rodent populations in a beech forest.
Grodzinski, W., Gorecki, A., Janas, K. and Migula, P. (1966). Acta Theriol. 11,
419431. Effect of rodents on the primary production of alpine meadows
in Bieszczady mountains.
Gross, A. 0. (1947). Auk 64, 684-601. Cyclic invasion of the snowy owl and the
migration of 1945-1946.
Hagen, Y. (1969). Fauna (Oelo) 22, 73-126. Norwegian studies on the reproduction of birds of prey and owls in relation to micro-rodent population
fluctuations. (English summary)
Hamilton, W. J. Jr. (1937). J. agric. Res. 54, 779-790. The biology of microtine
cycles.
Hansson, L. (1971). Ann. 2001.Fennici 8,118-126. Estimates of the productivity
of small mammals in a south Swedish spruce plantation.
Haynes, D. W. and Thompson, D. Q. (1965). Tramactions of the Thirtieth North
American Wildl. and Nat. Reeources Conf. 393-400. Methods for estimating
microtine abundance.
Hewitt, C. G. (1921). “The Conservation of the Wild Life of Canada”. 344 pp.
Scribner’s, New York.
Hill, A. B. (1959). “Principles of Medical Statistics”. 314 pp. Lancet Ltd., London.
Hoffmann, R. S. (1958). Ecol. Monogr. 28, 79-109. The role of reproduction and
mortality in population fluctuations of voles (Microtw).
Honer, M. R. (1963). Ardea 51, 168-195. Observations on the barn owl (Tyto
alba guttata) in the Netherlands in relation to its ecology and population
fluctuations.
Houlihan, R. T. (1963). Univ. Calq. PubL 2001.65, 327-362. The relationship of
population density to endocrine and metabolic changes in the California vole,
Microtwr Caliifornicw.
Howell, A. B. (1923). J. Mammal. 4, 149-165. Periodic fluctuations in the numbers
of small mammals.
Jenkins, D., Watson, A. and Miller, G. R. (1964). J. appl. Ecol. 1, 183-195.
Predation and red grouse populations.
Jewell, P. A. and Fullagar, P. J. (1966). J. Zool., London 150, 501-509. Body
measurements of small mammals: sources of error and anatomical changes.
Kalela, 0. (1957). Ann. Acad. Sci. Fenn. Ser. A., IV, 34, 1-60. Regulation of
rufocanw
reproduction rate in subarctic populations of the vole C~kthrionomyo
(Sund.).
Kalela, 0. (1962). Ann. A d . Sci. Fenn. Ser. A., JY,66, 1-38. On the fluctuations
in the numbers of arctic and boreal small rodents as a problem of production
biology.
Kalela, 0. and Koponen, T. (1971). Ann. 2002.FenniCi. 8, 80-84. Food consumption and movements of the Norwegian lemmings in areas characterized by
isolated fells.
394
CHARLES J. KREBS AND JUDITH H. MYERS
Kalela, 0. and Oksala, T. (1966). Ann. Univ. Turkuensis, Ser. A, 11, Bio1.-Geogr.
37, 5-24. Sex ratio in the wood lemming, Myopus schistiwlor (Lilljeb.), in
nature and in captivity.
Keith, L. B. (1963). “WildlifeTs Ten-Year Cycle”. 201 pp. Univ. Wisconsin Press,
Madison.
Keller, B. L. and Krebs, C. J. (1970). Ecol. Monogr. 40, 263-294. Microtus
population biology. 111. Reproductive changes in fluctuating populations
of M . ochrogaster and M . pennsylvanicus in southern Indiana, 1965-1967.
Khlebnikov, A. I. (1970). 2001.Zh. 42, 801-2. Winter reproduction of the
northern redbacked vole (Clethrionomys rutilus) in the dark-coniferous
taiga of the west Sayan mountains. (English summary).
King, C. E. and Anderson, W. W. (1971). Am. Nat. 105, 137-156. Agespecific selection. 11. The interaction between r and K during population
growth.
Koshkina, T. V. (1965). Bull. Mom. SOC.N d . , Biol. Sect. 70, 5-19. Population
density and its importance in regulating the abundance of the red vole.
(In Russian, transl. by W. A. Fuller.)
Koshkina, T. V. (1966). BUZZ. Moec. SOC.Nat., Biol. Sec. 71, 14-26. On the
periodical changes in the numbers of voles (as exemplified by the Kola
Peninsula). (In Russian, transl. by W. A. Fuller.)
Koshkina, T. V. and Khalansky, A. S. (1962). Zool. Zh. 41,604-615. Reproduction
of the Norwegian lemming (Lemmus lemmus L.) on the Kola Penninsula.
(In Russian, transl. by W. A. Fuller.)
Krebs, C. J. (1964a). Arctic Inst. N . Amer. Tech. Paper 15, 104 pp. The lemming
cycle a t Baker Lake, Northwest Territories, during 1959-62.
Krebs, C. J. (1964b). Can. J . Zool. 42, 631-643. Cyclic variation in skull-body
regressions of lemmings.
Krebs, C. J. (1966). Ecol. Monogr. 36, 239-273. Demographic changes in fluctuating populations of Microtus mlifornicus.
Krebs, C. J. (1970). Ecology 51, 34-62. Microtus population biology: behavioural
changes associated with the population cycle in M . ochrogaster and M .
pennaylvanicus.
Krebs, C. J . (1971). Proc. Adv. study Inst. Dynamics Numbers Popul. (Oosterbeek,
1970), pp. 243-256. Genetic and behavioural studies on fluctuating vole
populations.
Krebs, C. J. (1972). “Ecology: The Experimental Analysis of Distribution and
Abundance”. 694 pp. Harper and Row, New York.
Krebs, C. J. and DeLong, K. T. (1966). J . Mammal. 46, 566-573. A Microtus
population with supplemental food.
Krebs, C. J., Keller, B. L. and Tamarin, R. H. (1969). Ecology 50, 587-607.
Microtw, population biology: demographic changes in fluctuating populations
of M . ochrogmter and M . pennaylvanicus in southern Indiana.
Krebs, C. J., Keller, B. L. and Myers, J. H. (1971). Ecology 52, 660-663. Microtus
population densities and soil nutrients in southern Indiana grasslands.
Krebs, C. J., Gaines, M. S., Keller, B. L., Myers, J. H. and Tamarin, R. H. (1973).
Science 179, 35-41, Population cycles in small rodents.
Lack, D. (1954). “The Natural Regulation of Animal Numbers”. 279 pp. Oxford
Univ. Press, Oxford.
LeCren, E. D. (1951). J . Anim. Ecol. 20, 201-219. The length-weight relationship
and seasonal cycle in gonad weight and condition in the perch (Perca
jzuviatilis)
.
POPULATION CYCLES IN SMALL MAMMALS
395
Leslie, P. H. (1959). Physiol. Zool. 32, 151-159. The properties of a certain lag
type of population growth and the influence of an external random factor
on a number of such populations.
Leslie, P. H. and Davis, D. H. S. (1939). J . Anim. Ewl. 8,9&113. An attempt to
determine the absolute number of rats on a given area.
Leslie, P. H. and Ranson, R. M. (1940). J. Anim. Ewl. 9, 27-52. The mortality,
fertility and rate of natural increase of the vole (Microtus w r a t h ) as
observed in the laboratory.
zool. SOC.Lo&. 115,
Leslie, P. H., Perry, J. S. and Watson, J. S. (1945). PTOC.
473-488. The determination of the median body-weight a t which female rats
reach maturity.
Leslie, P. H., Chitty, D. and Chitty, H. (1953). Biometrika 40, 137-169. The
estimation of population parameters from data obtained by means of the
capture-recapture method. 111. An example of the practical applications of
the method.
Lidicker, W. Z. Jr. (1973). Ecol. Monogr. 43,271-302. Regulation ofnumbersin an
island population of the California vole, a problem in community dynamics.
Lidicker, W. Z. and MacLean, S. F. (1969). Am. Midl. Nat. 82, 450-470. A
method for estimating age in the California vole, Microtus d i f m i c u s .
Lockie, J. D. (1955). Bird Study, 2, 53-69. The breeding habits and food of
short-eared owls after a vole plague.
Lorenz, K. (1963). “On Aggression”. 273 pp. Methuen and Co. Ltd., London.
Louch, C. D. (1966). Ecology 37, 701-713. Adrenocortical activity in relation to
the density and dynamics of three confined populations of MiCTOtU8
pennsylvanicus.
Maher, W. J. (1970). Wibon Bull. 82, 130-157. The pomarine jaeger as a brown
lemming predator in northern Alaska.
Marten, G. G. (1970). Small M a m d Newsletter 4, 31-44. A regression method for
mark-recapture estimation of population size with unequal catchability.
Martin, E. P. (1956). Univ. Kans. Publs Mwr. nat. Hist. 8 (6), 361-416. A population study of the prairie vole (Microtu8 ochrogaster) in northeastern Kansas.
Migula, P., Grodzinski, W., Jasinski, A. and Musialek, B. (1970). Acta Them’ol. 15,
233-252. Vole and mouse plagues in southeastern Poland in the years
1945-1967.
Morris, R. D. and Grant, P. R. (1972). J. Anim. Ewl. 41, 275-290. Experimental
studies of competitive interaction in a two-species system. Iv. Mkrotua
and Clethrionomys species in a single enclosure.
Mullen, D. A. (1965). Ph.D. Thesis, Univ. California, Berkeley. 173 pp. Physiologic
correlations with population density and other environmental factors in the
brown lemming, Lemmus trimwronatus.
Mullen, D. A. (1968). Univ. Calif. Pubb Zool. 85, 1-24. Reproduction in brown
lemmings (Lemmw trimucronatwr) and its relevance to their Cycle of
abundance.
Myers, J. H. and Krebs, C. J. (1971a). Am. Nat. 105, 325-344. Sex ratios in open
and enclosed vole populations: demographic implications.
Myers, J. H. and Krebs, C. J. (1971b). Ecol. Momgr. 41, 53-78. Genetic,
behavioural, and reproductive attributes of dispersing field voles Mkrotus
pennsylvanicus and Mkrotus ochrogaster.
Myllymiiki, A. (1969). I n “Energy FIow Through Small Mammal Populatione”
(K. Petrusewicz and L. Ryszkowski, Eds). Productivity of a free-living
population of the field vole, Mkrotus q,watis (L.).
396
CHARLES J. KREBS AND JUDITH H. MYERS
Mysterud, I. (1970). Nytt Mag. 2001.18,49-74. Hypotheses concerning characteristics and causes of population movements in Tengmalm’s owl (Aegoliua
funereus (L.)).
Nasimovich, A., Novikov, G. and Semenov-Tyan-Shanski, 0. (1948). Materialy
PO gryzunam 3, 203-262. The Norwegian lemming: its ecology and role in the
nature complex of the Lapland Reserve. (In Russian.)
Newson, J. and Chitty, D. (1962). Ecology 43, 733-738. Haemoglobin levels,
growth and survival in two Mkrotus populations.
Newson, R. (1963). Ewlogy 44, 110-120. Differences in numbers, reproduction
and survival between two neighboring populations of bank voles (Cleth&nomy8 glareolua).
Odum, E. P. (1971). “Fundamentals of Ecology”. 3rd Edition. 574 pp. Saunders,
Philadelphia.
Otero, J. G. and Dapson, R. W. (1972). Rm. Popul. Ecol. 13, 162-160. Procedures in the biochemical estimation of age of vertebrates.
Patrio, E. F. (1962). J . Mammal. 43, 200-205. Reproductive characteristics of
red-backed mouse during years of differing population densities.
Pearson, 0. P. (1960). Ewl. Monogr. 30, 231-249. Habits of Microtus ccclgornicw,
revealed by automatic photographic records.
Pearson, 0. P. (1963). Ecology 44, 540-649. History of two local outbreaks of
feral house mice.
Pearson, 0. P. (1964). J. M a m m l . 45, 177-188. Carnivore-mouse predation:
an example of its intensity and bioenergetics.
Pearson, 0. P. (1966). J. Anim. Ecol. 85, 217-233. The prey of carnivores during
one cycle of mouse abundance.
Peamon, 0. P. (1971). J. Mammal. 52, 41-49. Additional measurements of the
impact of carnivores on California voles (Microtus ca1;ifornicua).
Petrusewicz, K., Bujalska, G., Andrzejewski, R. and Gliwicz, J. (1971). Ann.
2001.Fennici 8, 127-132. Productivity processes in an island population of
Cleth?$onomye ghreolua.
Pieper, R. D. (1964). Ph.D. Thesis, Univ. California, Berkeley. Production
and chemical composition of arctic tundra vegetation and their relation to
the lemming cycle.
Pinter, A. J. and Negus, N. C. (1965). Am. J. Physiol. 208, 633-638. The effects
of nutrition and photoperiod on reproductive physiology of Microtw
mtcmzce.
Pitelka, F. A. (1958). Cold Spring Harb. Symp. quant. Biol. 28, 237-261. Some
aspects of population structure in the short-term cycle of the brown
lemming in northern Alaska.
Pitelka, F. A. (1961). Progress report for 1960-61 submitted to Arctic Institute
of North America. 18 pp. Ecology of lemmings and other microtines in
Northern Alaska.
Pitelka, F. A. (1972). Proc. 1972 Tundra Biome Symp. Univ. Wash., pp. 132-135.
Cycle pattern in lemming populations near Barrow, Alaska.
Pitelka, F. A., Tomich, P. Q. and Treichel, G. W. (1965). Ecol. Monogr. 25,
85-117. Ecological relations of jaegers and owls as lemming predators near
Barrow, Alaska.
Pruitt, W. 0. (1968). Mammlia 32, 172-191. Synchronous biomass fluctuations
of some northern mammals.
Rauch, R. (1960). Arctic a, 166-177. Observations on a cyclic decline of lemmings
(Lemmua) on the arctic coast of Alaska during the spring of 1949.
POPULATION CYCLES IN SMALL MAMMALS
397
Reichstein, H. (1964). 2. wigs. Zool. 170, 112-222. Untersuchungen zum
Korperwachstum und zum Reproduktionspotential der Feldmaua, Microtw,
arvalia (Pallaa, 1779).
Ryszkowki, L. (1969). I n “Energy Flow Through Small Mammal Populations”
(K. Petrusewicz and L. Ryszkowski, Eds), pp. 281-289. Estimates of
consumption of rodent populations in different pine forest ecosystems.
Schaffer, W. M. and Tamarin, R. H. (1973). Evolution 27, 111-124. Changing
reproductive rates and population cycles in lemmings and voles.
Schultz, A. M. (1964). I n “Grazing in Terrestrial and Marine Environments”
(D. Crisp, Ed.), pp. 57-68. Blackwells, Oxford. The nutrient-recovery
hypothesis for arctic microtine cycles. 11.Ecosystem variables in relation to
the arctic microtine cycles.
Schultz, A. M. (1965). The Tundra as a Homeostatic System. Presented a t
A.A.A.S. Meetings, Dec. 1965. (Mimeo).
Schultz, A. M. (1969). I n “The Ecosystem Concept in Natural Resource Management’’ (G. M. Van Dyne, Ed.), pp. 77-93. Academic Press, New York. A
study of an ecosystem: the arctic tundra.
Selye, H. (1946). J. clin. Endocr. 6, 117-230. The general adaptation syndrome
and the diseases of adaptation.
Semeonoff, R. (1972). J. H e r d . 63,48-52. Two coat color variants in the prairie
vole.
Semeonoff, R. and Robertson, F. W. (1968). Biochem. Genet. 1, 20S227. A
biochemical and ecological study of plasma esteraae polymorphism in
natural populations of the field vole, Microtw, ugreatis L.
Seber, G. A. F. (1970). Biometrioo 26, 13-22. The effects of trap response on tag
recapture estimates.
Shelford, V. E. (1943). Ecology 24,472-484. Abundance of the collared lemmings
in the Churchill area, 1929-1940.
Smith, M. H., Blessing, R., Chelton, J. G., Gentry, J. B., Golley, F. B. and
McGinnis, J. T. (1971). Acta Theriol. 16, 105-126. Determining density for
small mammal populations using a grid and assessment lines.
Smyth, M. (1966). J. Anim. Ecol. 35, 471-485. Winter breeding in woodland
mice, Apodemw, sylvaticw, and voles, Clethrionomys glareolus and Microtw,
agrmtis, near Oxford.
Smyth, M. (1968). J. Anim. E d . 37, 167-183. The effects of the removal of
individuals from a population of bank voles, Clethrionomys glareolus.
Speirs, R. S. and Meyer, R. K. (1949). Endocrinology 45, 403-429. The effects of
stress, adrenal and adrenocorticotrophichormones on circulating eosinophils
of mice.
Stein, G. H. W. (1957). 2. Stiugetierk. 22, 117-135. Materialien zur Kenntnis der
Feldmaus, Microtus arvalis P.
Stendell, R. (1972). Ph.D. Thesis, Univ. California, Berkeley. The occurrence,
food habits, and nesting strategy of white-tailed kites in relation to a fluctuating vole population.
Straka, F. and Gerasimov, S. (1971). Ann. Zool. Fennici 8, 113-116. Correlations
between some climatic factors and the abundance of Microtus arvalh in
Bulgaria.
Summerhayes, V. S. (1941). J . Ecology 29, 1-48. The effect of voles (Microtw,
agreatk) on vegetation.
Sutton, G. M. and Hamilton, W. J. (1932). Mem. Carneg. Mus. 12, (Pt. 2), 1-111.
The mammals of Southampton Island.
398
CHARLES J . KREBS AND JUDITH H. MYERS
Tamarin, R. H. and Krebs, C. J, (1969). Evolution 23, 183-211. Microtua population biology. 11. Genetic changes at the transferrin locus in fluctuating
populations of two vole species.
Tamura, M. (1966). M. A. Thesis, Univ. California, Berkeley. Aggressive
behaviour in the California meadow mouse (Microtus californicus).
Tanaka, R. (1960). Jap. J . Ecol. 10, 102-106. Evidence against reliability of the
trap-night index as a relative measure of population in small mammals.
Tanaka, R. (1963). Res. Popul. Ecol. 5, 139-146. On the problem of trap-response
types of small mammal populations.
Tanake, R. (1964). Ree. Popul. Ecol. 6 , 5 6 6 6 . Population dynamics of the Smith’s
red-backed vole in highlands of Shikoku.
Tanaka, R. (1970). Rea. Popul. Ecol. 12, 111-125. A field study of the effect of
prebaiting on censusing by the capture-recapture method in a vole population.
Tanaka, R. (1972). Rea. Popul. Ecol. 13, 127-151. Investigation into the edge
effect by use of capture-recapture data in a vole population.
Tanton, M. T. (1965). J . Anim, Ecol. 34, 1-22. Problems of live-trapping and
population estimation for the wood mouse, Apodemus aylvaticus (L.).
Tanton, M. T. (1969). J . Anim. Ecol. 38, 511-529. The estimation and biology of
populations of the bank vole (Ckthrionomys glareolua (Schr.)) and wood
mouse (Apodemw, aylvaticw (L.)).
Tast, J. (1972). Ann. Zool. Fennici 9, 116-119. Annual variations in the weights
of wintering root voles, Microtus oecmomus, in relation to their food conditions.
Tast, J. and Kalela, 0. (1971). Ann. Acad. Sci. Fenn. Ser. A . Iv.Biol. 186, 1-14.
Comparisons between rodent cycles and plant production in Finnish Lapland.
Thompson, D. Q. (1955a). Arctic I m t . N . Am., Final Report, Proj. ONR-133,
63 pp. Ecology of the Lemmings.
Thompson, D. Q. (195513). Trans. 20th N . Am. Wildl. Conf., pp. 166-178. The
role of food and cover in population fluctuations of the brown lemming at
Pt. Barrow, Alaska.
Thompson, D. Q. (1965). Am. Midl. Nat. 74, 76-86. Food preference of the
meadow vole (Microtua pennaylvanicua) in relation to habitat affinities.
Trojan, P. (1969). I n “Energy Flow Through Small Mammal Populations.”
(K. Petrusewicz and L. Ryszkowski, Eds). Energy flow through a population
of Microtua arvcclia (Pall.) in an agrocenosis during a period of maw occurrence.
Turner, B. N. (1971). M.Sc. Thesis, Univ. North Dakota. The annual cycle of
aggression in male Microtw, penneylvanicus, and its relation to population
parameters.
Van Oortmerssen, G. A. (1970). Behaviour 38, 1-92. Biological significance,
genetics, and evolutionary origin of variability in behaviour within and
between inbred strains of mice (Mua muaculua).
Voipio, P. (1969). Oikos, 20, 101-109. Some ecological aspects of polymorphism in
the red squirrel Sciurua vulgaris L. in northern Europe.
Watson, A. (1956).J . Anim. Ecol. 25, 289-302. Ecological notes on the lemmings
Lemrnua trimuwonatua and Dicrostonyx groenlandicua in B f f i Island.
Watson, A. (1957). Ibia 99, 419-462. The behaviour, breeding and food-ecology
of the snowy owl Nyctea scandiaca.
Watson, A. and Moss, R. (1970). I n “Animal Populations in Relation to their
Food Resources” (A. Watson, Ed.), pp. 167-220. Blackwell, Oxford. Dom-
POPULATION CYCLES IN SMALL MAMMALS
399
inance, spacing behaviour and aggression in relation to population limitation
in vertebrates.
Watts, C. H. S. (1970). J . Mammal. 51, 341-347. A field experiment on intraspecific interactions in the red-backed vole, Clethrionomys 9appel.i.
Whitney, P. (1973). Ph.D. Thesis, in preparation. Inst. Arctic Biology, Univ.
Alaska.
Wijngwden, A. van (1960). Versl. ladbouwk. Onderz. No. 66, 22, 1-28. The
population dynamics of four confined populations o f the continental vole
Microtw arvalis (Pallas).
Wildhagen, A. (1952). “Om Vekslingene i Bestanden av Smiignagere i Norge
1871-1949”. 192 pp. Statens Viltundersekelser, Drammen.
Williams, G. C. (1966). “Adaptation and Natural Selection”. 307 pp. Princeton
Univ. Press, Princeton, N. J.
Zejda, J. (1962). 2001.Lzkty 11, 309-321. Winter breeding in the bank vole,
Clethrionomy8 ghreolw Schreb.
Zejda, J . (1964). Zool. L&y 13, 16-30. Development of several populations o f the
bank vole C1kthrionomy8 glareolw Schreb., in a peak year.
98
Zejda, J. (1966). Zool. L k t y 15, 193-206. Litter size in ~ h % & J n O W &ghreolw
Schreber 1780.
Zejda, J. (1967). Zool. Lbty 16,221-238. Mortality ofe population ofClethrionomy8
glareolw Schreb. in a bottomland forest in 1964.
Zimmerman, E. G. (1965). J . Mammal. 46,606-612. A comparison of habitat and
food of two species of Microtw.
Zimmermann, K. (1955). Zeit. SGtqetierk. 20, 114-118. Korpergroese und
Bestandsichte bei Feldmiiusen (Microtw arvdzk).
Was this manual useful for you? yes no
Thank you for your participation!

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

Download PDF

advertising