POPULATION ECOLOGY
The Effect of Winter Temperature on Forest Tent Caterpillar
(Lepidoptera: Lasiocampidae) Egg Survival and Population Dynamics
in Northern Climates
BARRY J. COOKE1
AND
JENS ROLAND
University of Alberta, Department of Biological Sciences, Edmonton, AB, Canada, T6G 2E9
Environ. Entomol. 32(2): 299Ð311 (2003)
ABSTRACT Overwintering mortality of forest tent caterpillar [Malacosoma disstria (Hübner)] eggs
was estimated over a 360 km2 grid of 83 plots in north-central Alberta over the period 1992Ð1996 during
a local outbreak. Egg mortality in the trembling aspen (Populus tremuloides Michaux) canopy was
generally low; however, 20% of the eggs laid in the summer of 1995 failed to hatch in the spring of
1996. In the shrub layer, 70% failed to hatch. In both the canopy and shrub layers, the spatial pattern
of mortality was density-independent, with high mortality occurring in low-lying areas. Daily temperature records suggested that the proximal cause of death was freezing during mid-winter. Caterpillar populations peaked in 1995, before perturbation, and collapsed during the summer of 1996,
largely as a result of larval parasitism. The timing of this perturbation-assisted population collapse
coincided loosely with the penetration of larvae down into the shrub layer. We illustrate how winter
temperature, albeit a density-independent factor, probably acts in a partially density-dependent
manner through interactions with density-dependent behavioral and physiological processes that
inßuence spatial variation in vulnerability and susceptibility to winter cold. We argue that cold winter
temperatures are an important factor inßuencing the long-term dynamics of forest tent caterpillar
populations in northern climates.
KEY WORDS periodic insect outbreaks, population cycles, density-dependence, extreme weather,
stochastic perturbation, nonlinear dynamics
THE CENTRAL AIM of population ecology has been to
identify the biological mechanisms that inßuence population ßuctuations and to describe precisely how
these forces balance one another to achieve a state of
population regulation. Over the decades, differing
views have emerged regarding the importance of (1)
density-dependent versus density-independent survival and reproduction (Nicholson 1954, Andrewartha
and Birch 1954), (2) environmental stochasticity
(May 1973, Reeve 1988, Wilson et al. 1998), (3) nonlinear dynamics and complex attractors (May 1976,
Schaffer and Kot 1985, Berryman and Millstein 1989,
Turchin and Taylor 1992, Logan and Allen 1992, Hastings et al. 1993), (4) spatial heterogeneity (Hilborn
1975, Hastings 1977), (5) top-down versus bottom-up
processes (Hunter and Price 1992), and (6) nonequilibrium dynamics (Hastings and Higgins 1993, Hastings 1998, Kaneko 1998). Disparate views on the origin
of animal population cycles continue to be argued
even today (Kendall et al. 1999).
The debate has been especially intense regarding
the occurrence of periodic insect outbreaks (e.g.,
Martinat 1987, Ginzburg and Taneyhill 1994, Berry1
E-mail: [email protected].
man 1996, Myers 1998). While there is an emerging
consensus that no one view is likely to prevail in all
systems, synthesis has not yet been achieved (c.f.
Turchin 1995). Even among the best-studied systems there is substantial debate about the nature of
population regulation and the cause of outbreaks.
For example, despite decades of research on the
spruce budworm, Choristoneura fumiferana Clemens
(Lepidoptera: Tortricidae), it is still unclear whether
outbreaks are caused by an eruptive mechanism
whereby regulation about a low-density equilibrium
state occasionally fails (Morris 1963, Ludwig et al.
1978, Clark and Holling 1979, Blais 1983, Campbell
1993), or by a noneruptive mechanism whereby populations cycle loosely about a single equilibrium state
because of lagged effects of parasitism (Royama 1984,
1992, Régnière and Lysyk 1995, Berryman 1996).
The case of the forest tent caterpillar, Malacosoma
disstria (Hübner) (Lepidoptera: Lasiocampidae), is
remarkably similar in this regard. A defoliator of
broad-leaved trees, especially trembling aspen (Populus tremuloides Michaux), the forest tent caterpillar
exhibits occasional population eruptions that are synchronized into large-scale outbreaks (Hildahl and
Reeks 1960, Sippell 1962, Hodson 1977, Daniel and
300
ENVIRONMENTAL ENTOMOLOGY
Vol. 32, no. 2
Fig. 1. Aspen stands sampled across the study grid, located near Cooking L. in central Alberta. Plots, shown as black
circles, are overlaid onto a classiÞed photo-mosaic indicating forest (light gray), clearing (white), and large water bodies (dark
gray).
Myers 1995). In the heavily forested parts of North
America population densities may span four orders of
magnitude over the course of an outbreak (Witter
et al. 1975).
The natural history and population ecology of
M. disstria in northern forests is well known and
has been summarized before (Hodson 1941, Witter
1979, Ives and Wong 1988, Fitzgerald 1995). Brießy,
pharate larvae pass the winter inside their eggs,
which are aggregated into a single band on the tips of
twigs. Although cold tolerant, they are susceptible to
freezing by extreme cold (Wetzel et al. 1973), with the
risk of mortality varying seasonally with changes in
temperature and physiology (Hanec 1966). Larvae
emerge in early spring, around the time of aspen budbreak. If foliage is not available, young larvae are
vulnerable to freezing and starvation (Blais et al. 1955,
Raske 1975). Larvae feed colonially for the Þrst four
stages, during which time they are preyed upon by
spiders, ants, pentatomids, and birds (Green and Sullivan 1950), and parasitized by numerous species of
wasps and tachinid ßies (Williams et al. 1996). During
the Þfth larval stage, entire stands of trees can be
completely defoliated, and larvae may wander in
search of food and/or pupation sites (Batzer et al.
1995). Many of these larvae act as hosts to Dipteran
parasitoids. Upon completing larval development,
these parasitoids drop to the ground and pupate in the
litter, where they pass the winter underneath the
snow, and emerge the following spring. Nonparasitized larvae spin protective silken cocoons in foliage,
and pupate. A few weeks later, sexually mature moths
emerge, mate, and the females lay eggs.
Little is known about the factors that lead to the
initiation of forest tent caterpillar outbreaks, although
warm spring weather (Ives 1973), phenological synchrony with host-plants (Parry et al. 1998), and reduced efÞciency of natural enemies (Parry et al. 1997)
are all plausible mechanisms. Population decline is
often associated with high rates of parasitism, although
sudden collapse is sometimes associated with cold
winter weather or viral epizootics (Hodson 1977, Witter 1979). Despite considerable efforts to model the
dynamics of this system (Rose and Harmsen 1978,
1981), it remains unclear to what extent population
cycles are driven by periodic, density-dependent collapse and whether top-down or bottom-up processes
are responsible.
The Þrst objective of this paper is to present original
Þeld data from a recent outbreak of forest tent caterpillar, near Cooking Lake, Alberta. These data show
how winter egg mortality varies both temporally, as a
function of density and temperature, and spatially, as
a function of topography and microclimate. A second
objective is to outline a conceptual model of forest
tent caterpillar egg survival that is consistent with
these observations and which may help to explain the
quasi-periodic nature of outbreaks.
Methods
Study Area. The 360 km2 study area was located
in the Ministik Hills of north-central Alberta, near
Cooking Lake, 40 km east of Edmonton (Fig. 1). The
area is characterized by a rolling morainal deposit
with knob-and-kettle topography. The landscape is
mostly forested aspen parkland dominated by
trembling aspen (P. tremuloides), with balsam poplar
(P. balsamifera L.) occurring on wetter, low-lying
sites. Regionally, the study area is embedded in a
April 2003
COOKE AND ROLAND: TENT CATERPILLAR EGG SURVIVAL AND POPULATION DYNAMICS
matrix of fragmented aspen parkland dominated by
agriculture and cattle ranching.
An irregular grid of 83 plots was established by
choosing stands of aspen that were easily accessible
and were ⬇1.6 km apart from one another (Fig. 1).
Plots varied in terms of forest cover, elevation, and soil
moisture. Those on the western and far eastern portions of the grid were located in residential and drier,
agricultural areas. Those in the central portion of the
grid were wetter and heavily forested. Elevations for
each sample plot on the study grid were obtained by
choosing the nearest 25-foot contour line or altitudinal
landmark shown on 1:50 000 National Topographic
Survey maps for Cooking Lake-ToÞeld. Elevation varied from 2300 to 2600 feet (700 Ð 800 m) with the
lowest elevations occurring around major water bodies such as Cooking Lake and Hastings Lake on the
north part of the grid, the Ministik Lakes on the western part of the grid, and Beaverhill Lake, located 10 km
to the east of the grid.
Population Data. Estimates of M. disstria cocoon
density were obtained for each stand over the years
1993Ð1998. Cocoons were collected using a timed
Þxed-count, such that the area sampled varied among
plots according to cocoon density and the ease of
search and discovery. Plot size was typically larger
than that used in the shrub-layer egg survey and
smaller than that used in the canopy-layer egg survey.
Detailed methods of population sampling are described in a related study (Roland and Taylor 1997).
Winter Egg Mortality. In early April 1996, two
M. disstria egg masses were collected from each of the
plots in the study area, usually within a few meters of
plot centers. These living egg masses, which were laid
in the summer of 1995, were taken from shrubs and
trees ⬍2 m in height and stored in a shelter outdoors.
A second sample was collected two weeks later and
processed in the same manner. Eggs were allowed to
hatch naturally. Unhatched eggs from both samples
were individually dissected to determine the various
causes of failure and to calculate hatching success.
During the summer of 1997 the canopies of all 83 of
those same plots were resampled to obtain old egg
masses. Minimal effort was required to sample treetops because a late, wet snow fell on 21 May 1997,
weeks after aspen buds had developed into leaves,
causing many stems to bend and break under the
heavy load. An average of 7.34 ⫾ 0.28 trees were
sampled at each plot. The origin of the downed treetops could usually be ascertained so that we were able
to avoid sampling the same tree twice. The pattern of
sampling at each plot varied according to the pattern
and scale of damage. The sampled area was usually
elliptical in shape and did not exceed ⬎100m in any
direction. The canopy-layer plots were located within
100 m of the plot centers, and were much larger than
the shrub-layer plots.
A total of 4,528 egg masses were collected from 609
tree-tops and large branches from damaged aspen
stems. These old egg masses included cohorts laid in
the summer of 1996 and years previous because old
egg masses persist for several years on twigs whose
301
radial growth has ceased because of severe defoliation.
Egg masses were indexed in clear plastic vials so that
they could be inspected, handled, and sorted.
Egg masses were arrayed on a large table and sorted
according to their apparent age. To do this, each egg
mass was compared with every other egg mass in terms
of (1) weathering of the spumaline covering the egg
mass, (2) erosion around the edges of individual shells
of hatched eggs, and (3) fungal development. The
difference between 1996 egg masses and older egg
masses was immediately obvious: 1996 egg masses had
substantial spumaline and older egg masses did not.
After several days of sorting, major differences between 1995 egg masses and older egg masses became
apparent: (1) 1995 egg masses were usually white, not
gray from weathering, and had only traces of spumaline left; (2) edges of hatched eggshells were smoothly
circular, not at all frayed from weathering; (3) 1995
egg masses had no fungi or lichens growing on them.
Thus sorted, groups could be distinguished on the
basis of the nature and degree of weathering. Each
group of egg masses was assigned a likely cohort year,
from 1992 to 1996, old egg masses being more weathered. Differences between earlier cohort years earlier
than 1994 were less discernible.
A sub-sample of unhatched eggs from the 1995 cohort was dissected under a stereoscopic microscope to
check for the presence of pharate M. disstria larvae to
determine the proportion which were either infertile
or parasitized. Eggs that failed to hatch as a result of
factors other than infertility and parasitism had a
blackish appearance because of the head capsules of
the pharate larvae showing through the translucent
egg shell. This macroscopic feature was found to be a
reliable indicator of pharate larval mortality. Consequently, each egg mass was grossly categorized as
having either greater or less than half of its eggs winter-killed. This method is faster than counting eggs
individually, and is almost as accurate for large samples. For each stand, the proportion of egg masses with
⬎50% egg mortality was computed for use in analysis.
Weather Data. Daily minimum temperatures for
Edmonton Municipal Airport (113.52⬚ W, 53.57⬚ N)
and Edmonton International Airport (113.58⬚ W,
53.30⬚ N) were obtained for the periods 1967Ð1998 and
1995Ð1996, respectively. The municipal airport is located 30 km northwest of the study area, within the
heat island of the Edmonton urban core and the international airport is situated rurally, 30 km west of the
study area, outside the urban heat island (Hage 1972).
Advanced Very-High Resolution Radiometer
(AVHRR) thermal infrared satellite images were obtained from the United States National Oceanic and
Atmospheric Administration (NOAA) to diagnose airßow patterns over northwestern Canada over the period 10 Ð21 January 1996. The goal here was to determine whether the meteorological conditions that led
to winter egg mortality were highly anomalous.
Statistical Analysis. Two-way analysis of variance
(ANOVA) was used to determine if winter mortality
of M. disstria eggs differed signiÞcantly among plots
and cohort years. The dependent variable was the
302
ENVIRONMENTAL ENTOMOLOGY
Vol. 32, no. 2
stand-level proportion of egg masses with 50% egg
mortality or higher. Linear regression was used to
determine the effect of elevation, vegetation stratum,
and cocoon density on spatial variation in winter mortality of the 1995 cohort. For each vegetation stratum,
nonlinear regression with a four-parameter logistic
model (Pinheiro and Bates 2000) was used to describe
the dependence of mortality on elevation. Correlograms were calculated for model residuals to detect
large-scale spatial autocorrelation that might have prevented the detection of spatially density-dependent
mortality. All analyses were performed with the R
statistical package, version 1.4.1 (Ihaka and Gentleman 1996).
Results
Winter Egg Mortality. M. disstria egg hatch differed
signiÞcantly between plots (F ⫽ 1.7; df ⫽ 82, 198; P ⫽
9.1 ⫻ 10⫺4) and between years (F ⫽ 27.1; df ⫽ 4, 198;
P ⫽ 2.2 ⫻ 10⫺12). The only cohort year in which winter
egg mortality in the canopy was signiÞcantly different
from zero was 1995 (mortality ⫽ 20.3 ⫾ 3.3%, t ⫽ 7.3;
df ⫽ 81; P ⫽ 6.06 ⫻ 10⫺12)Ñthe same cohort year for
which shrub-layer egg mortality was estimated at
70.3 ⫾ 2.9%.
On those 1995 canopy-layer egg masses that exhibited partial hatch, the unhatched eggs were often
clustered together along the length of one side of
the egg mass. The pattern of egg hatch within egg
masses was not recorded, nor was the orientation
of twigs. However, an independent sub-sample of
50 egg masses from one stand 100 m south of Hastings Lake conÞrmed that, for egg masses on twigs
oriented horizontally, hatch rates were greater for
eggs on the top side of the twig. No clustering of
hatch was observed for egg masses on twigs oriented
vertically.
The spatial pattern of winter mortality from gridwide canopy- and shrub-layer samples was remarkably similar, with high mortality in the north and
south and low mortality in the east and west (Fig. 2).
Mortality was generally higher at lower elevations,
although this relationship might have been exaggerated by a single large patch of low mortality on some
high land just south and east of Hastings Lake (Fig. 1).
Linear regression showed that 49% of the variation
in percent winter mortality of the 1995 cohort could
be attributed to the vegetation stratum sampled
(higher in shrub layer by 46.7 ⫾ 4.2%: F ⫽ 122.7; df ⫽
1, 142; P ⫽ 2.2 ⫻ 10⫺16) and elevation (19.1 ⫾ 4.9%
higher for each 100-foot drop in elevation: F ⫽ 14.2;
df ⫽ 1, 142; P ⫽ 2.3 ⫻ 10⫺4). Mortality was unrelated to spatial variation in cocoon density (F ⫽ 0.7;
df ⫽ 1, 142; P ⫽ 0.39). When canopy mortality was
computed using all cohort-years, instead of just
the 1995 cohort, the proportion of explained variation remained high, at 66%, suggesting that that our
analysis was insensitive to possible cohort classiÞcation errors.
Logistic regression indicated that the relationship
between winter egg mortality and elevation was sig-
Fig. 2. Spatial patterns in percent winter egg mortality in
the canopy (a) and shrub (b) layer, compared with elevation
(c) and 1995 cocoon density (d). Surfaces created by Þtting
anisotropic exponential variograms to the data and kriging
with a 0.5 km resolution grid. Contours generated using the
R base package contour algorithm (Ihaka and Gentleman
1996).
niÞcantly nonlinear (Fig. 3). One plot had a particularly strong inßuence on the poor Þt of the canopylayer regression. This plot, located on the far eastern
end of the grid at a low elevation of 2,350 feet, appeared to be an outlier, with 0% mortality in the
canopy and 77% mortality in the shrub layer. The
landform here is not knob-and-kettle morainal deposit, but a ßat plain that gradually drops in elevation
eastward to Beaverhill Lake. Correlogram analysis of
residuals from the logistic regression of mortality on
elevation indicated signiÞcant spatial autocorrelation
April 2003
COOKE AND ROLAND: TENT CATERPILLAR EGG SURVIVAL AND POPULATION DYNAMICS
303
Fig. 3. Winter egg mortality (y) as a function of elevation
(x) and vegetation stratum. Regression lines: ycanopy ⫽ 34.00Ð
27.81/(1 ⫹ exp [(2516 ⫺ x)/7.912]); yshrub ⫽ 76.60/(1 ⫹ exp
[(2560 ⫺ x)/⫺11.94]).
that might have prevented the detection of spatially
density-dependent mortality (Fig. 4).
Egg mass counts suggested that populations increased steadily from 1992 to 1995 and then declined
in 1996 (Fig. 5a, bars). These estimates of egg population densities are biased, and the bias probably increases with cohort age because old egg masses have
a greater probability of falling off twigs than recently
laid egg masses. Therefore the increase in densities
from 1992Ð1995 was probably not as dramatic as egg
mass counts suggest, and the drop in density from
1995Ð1996 was probably larger. Cocoon counts (Fig.
5a, circles) and aerial defoliation maps (J. R., unpublished data)are more reliable and conÞrm that populations changed little from 1994 to 1995, but declined
sharply in 1996.
A statistical test of association between egg survivorship and cocoon density through time should reveal a density-dependent relationship; however, there
were only Þve temporal observations (Fig. 5b). Moreover, these are confounded by temporal variation in
minimum winter temperatures, which happened to be
lowest in the year of highest density. Thus, it is unclear
from these limited data whether winter egg mortality
is related to population density.
Weather Patterns during the Winter of 1995–1996.
There were approximately eight cold periods lasting
more than two days during the interval from 1 November 1995Ð30 April 1996 (Fig. 6). On 18 Ð19 January
1996 rural temperatures dropped several degrees
below ⫺41⬚C, which is commonly cited as the midwinter cold tolerance threshold of forest tent caterpillar eggs (Hanec 1966). The cold tolerance threshold, however, changes with the seasonal biology of
the insect (Hanec 1966), so it possible that anomalously cold weather on 7 March or 24 March might
have killed these eggs. Notably, January temperatures
Fig. 4. Spatial structure of residuals from logistic regression of canopy- and shrub-layer mortality on elevation,
shown in Fig. 3. Bottom correlograms indicate gradients in
residuals. Filled squares denote signiÞcant autocorrelations
at the 95% conÞdence level, judged using a permutational test
with 100 simulations. Surfaces and contours generated as in
Fig. 2.
within the urban heat island of Edmonton did not drop
below ⫺41⬚C.
The sudden transitions between extreme warm and
extreme cold were associated with severe latitudinal
304
ENVIRONMENTAL ENTOMOLOGY
Fig. 5. Annual ßuctuations in population density, mortality, and temperature: (a) number of canopy egg masses
assigned to each cohort year (shaded bars) and cocoon
density estimates for each cohort year (solid circles); (b)
winter mortality of canopy-dwelling eggs laid in 1992 through
1996 (bars) and minimum urban temperature recorded each
winter (open circles, labeled as to date on which low temperature occurred).
shifts in jet stream ßow over central Alberta (Fig. 7).
On 11 January warm air was drawn northward up the
eastern slopes of the Rocky Mountains in association
with a meridional pattern of jet stream ßow. That
moist air cooled 12 January and the jet stream shifted
to a southern position, establishing a zonal ßow pattern on 13 January. A second, smaller meridional
blocking pattern began developing on 14 January in
Vol. 32, no. 2
association with the freshly cooled, moist air. The jet
stream shifted on 15 January, bringing in additional
moisture from the PaciÞc. This meridional block climaxed 16 January, bringing snow to the Edmonton
area on both days. The blocking pattern degenerated
on 17 January as the jet stream shifted toward a deep
southern trajectory. On 18 January extremely cold
arctic air moved in from the north, and persisted for
several days. The net change in temperature over that
time was more than ⫺42⬚C.
The Winter of 1995–96 in Context. How unusual
was the winter of 1995Ð1996? The pattern of daily
minimum temperatures differed substantially from
those in the three previous years (Fig. 8). During the
winter of 1995Ð1996 temperatures in Edmonton ßuctuated sharply between warm and cold episodes, leading to unfavorable conditions for egg survival through
three possible mechanisms: acute cold, prolonged exposure, and premature physiological changes associated with a warm spell before a cold snap. The coldest
temperature of ⫺37.8⬚C occurred 8 d after the minimum temperature had reached ⫹4.1⬚C. For 14 d thereafter the minimum did not exceed ⫺24.0⬚C. Under
these conditions, it may be that pharate larvae, before
being subjected to extreme and prolonged cold, underwent physiological changes in cold tolerance normally associated with spring time (Hanec 1966). Or
perhaps the snow that fell on 15Ð16 January moistened
the underside of egg masses sufÞciently to facilitate
freezing. It may be that the previous warm spell did
not alter egg physiology, but only served to enhance
the moisture-holding capacity of the outer spumaline,
making eggs more vulnerable to freezing.
In contrast, although the minimum daily temperature during the winter of 1993Ð1994 was nearly as cold
(⫺37.3⬚C, 7 February 1994), the warm phase that
preceded this minimum was not as warm (⫺4.1⬚C,
29 January 1994) and the transition to cold was less
abrupt (requiring 9 d) (Fig. 8). Temperatures also
rebounded upward immediately after the cold snap.
By all counts (acute effects, exposure effects, and
Fig. 6. Daily minimum temperatures from 1 November 1995Ð30 April 1996 for Edmonton International Airport (rural,
solid line) and Edmonton Municipal Airport (urban, dotted line) (courtesy Environment Canada). Arrows indicate prolonged
cold events. Horizontal lines indicate winter low temperatures at each location.
April 2003
COOKE AND ROLAND: TENT CATERPILLAR EGG SURVIVAL AND POPULATION DYNAMICS
305
Fig. 7. Advanced Very-High Resolution Radiometer (AVHRR) thermal infrared satellite images of northwestern Canada
indicating cloud cover from 10 to 21 January 1996 (courtesy United States National Oceanic and Atmospheric Administration,
NOAA). Dotted line indicates position of jet stream, which ßows West to East (left to right). Note: projection varies among
frames.
physiological preconditioning effects) the winters of
both 1993Ð1994 and 1994 Ð1995 were less risky for tent
caterpillar egg survival than the winter of 1995Ð1996.
Discussion
Winter Egg Mortality. The reason for clustered
patterns of hatch within M. disstria egg masses is not
known. Given that minimum temperatures were
above freezing on 11 January, and heavy snow fell on
15Ð16 January, it may have been a result of excess of
moisture clinging to the underside of the egg mass
before the deep freezeÑan effect that could be exacerbated by the hygroscopic nature of spumaline
(Hodson and Weinman 1945). Notably, there was also
substantial snowfall associated with the drop in temperature 18 Ð24 March (Fig. 8), making it unclear
whether mortality occurred in mid- or late-winter. A
critical question is whether the translucent eggshells
of unhatched eggs were caused by winter weathering
of eggshell exteriors or by pharate larvae chewing
from the inside in preparation for springtime hatch.
Vertical differences in egg mortality within and between plots suggest that cold air-pooling was likely a
contributing factor to the geographic pattern of hatch.
When elevations differ by hundreds of meters, temperatures are typically cooler at higher altitudes; however, when elevations differ by only tens of meters,
306
ENVIRONMENTAL ENTOMOLOGY
Vol. 32, no. 2
Fig. 8. Daily minimum urban temperatures (circles) from 1 December to 31 March for the winters of 1992Ð1993 through
1995Ð1996, compared with the long-term extreme high/low daily minimum over the period 1967Ð1996 (lines). Shaded
segments indicate largest, most sudden dips in temperature posing the greatest threat to tent caterpillar eggs. The most severe
and persistent of these arctic anomalies occurred in 1996. Daily snowfall (bars) shown in upper portion of each panel.
spatial patterning of temperature is largely determined by patterns of cold air drainage over the topographical landscape, especially during still nights under stable thermal inversions. High mortality at low
elevations was therefore probably a result of cold air
draining from high ground and pooling in sloughs and
ponds. Cold air perhaps did not pool on the east end
of the grid, but drained eastward toward Beaverhill
Lake. This would explain anomalously high egg survival at those low elevations.
Correlogram analysis indicated that even when the
effects of elevation were accounted for, there remained substantial spatial structure in residual patterns of egg hatch (Fig. 4). This is the sort of pattern
one might expect from thermal inversion breakdown.
Hage (1972), documenting one such event near Edmonton, explained how surface-level patchiness in
night-time air temperature can be caused by vertical
wind-shear. If a thermal inversion breakdown occurred on the western part of our study grid, causing
cold air to drain south and east, this could explain why
shrub-layer mortality was high in the southeast while
canopy-layer mortality was low. Temperature measurements throughout the canopy and shrub layer and
would be needed to test this possibility.
Temporal variation in the effect of temperature on
mean egg survival cannot be entirely because of differences in M. disstria population density because
population densities were just as high in 1993 as they
were in 1995 while egg mortality was higher in 1995
than in 1993 (Fig. 5). Although possible, it seems
unlikely that such a large difference in mortality could
be a result of observed differences in minimum temperature. The low daily minimum of ⫺37.8⬚C observed
April 2003
COOKE AND ROLAND: TENT CATERPILLAR EGG SURVIVAL AND POPULATION DYNAMICS
in January 1996 was only a little colder than the low
daily minimum of ⫺36.5⬚C observed in February 1994
(Fig. 6). One possibility is that the actual on-site temperature difference might have been larger than these
urban measurements indicate. Alternatively, winterkill may be inßuenced not only by the acute effect of
low daily temperatures, but also by the degree of
physiological preconditioning before exposure to abnormal cold and by the length of exposure. These
effects, which are not captured in a single physical
parameter such as a daily minimum temperature,
could vary considerably between years.
Another possibility is that cold-weather anomalies
may select against the most susceptible or the most
vulnerable individuals such that a surviving population might be more cold-adapted than the original
population. This could explain why hatch rates were
lower for the 1995 cohort than for the 1996 cohort
(Fig. 5). Such a mechanism would lead to important
nonstationary dynamic effects (Royama 1992).
Stochasticity of Winter Temperatures in Alberta.
The winter of 1995Ð96 was more extreme than either
of the previous three winters. However, 4 yr of daily
temperature data may be insufÞcient to evaluate the
importance of arctic anomalies to the long-term dynamics of tent caterpillar populations. Arctic anomalies might only have to occur once a decade to cause
decadal oscillations in forest tent caterpillar abundance.
It is noteworthy that a cold-air outbreak occurring
in February 1989 developed in much the same way as
the January 1996 anomaly (Hartjenstein and Bleck
1991). Daily minimum temperatures in 1989 followed
a pattern remarkably similar to the 1996 sequence, as
did the pattern of snowfall (Fig. 8). This arctic intrusion is particularly noteworthy because it may have
assisted in the collapse of the 1987Ð1988 tent caterpillar outbreak that occurred across central Alberta.
Unfortunately, we know of no data addressing this
possibility.
Cold-air outbreaks are, in fact, a common feature
of Albertan weather, especially in mid-winter. These
typically occur when arctic air intrudes from the
northeast and is blocked by the Rocky Mountains
(Colle and Mass 1995). As a result, temperatures typically drop by 20 Ð30⬚C while a high-pressure ridge
develops along the eastern slopes, frequently in association with a low pressure trough along the western
slopes. The meridional ßow that develops along the
western slopes tends to divert warm, moist PaciÞc air
away from Alberta and toward the American PaciÞc
NorthwestÑa synoptic pattern that occurs frequently
in winter (Changnon et al. 1993). Viewed over decadal
timescales, such sudden transitions from paciÞc to
arctic anomalies may be rather common. It has been
suggested that this is, in fact, a characteristic feature
of Rocky Mountain-induced atmospheric cyclogenesis (Keshishian et al. 1994, Davis 1997).
Because of the relatively high frequency of these
cold arctic anomalies, forest tent caterpillar populations in Alberta may be particularly vulnerable to
winter-kill. The risk may be further heightened by the
307
fact that cold-air outbreaks sometimes occur shortly
after intrusions of warm, moist air from the American
PaciÞc Northwest.
Density-Dependent Vulnerability and Susceptibility to Cold. Based on spatial and temporal patterns
of mortality, there are two reasons to suspect that
M. disstria eggs in 1996 might have been predisposed
to winter-kill through density-effects. First, high egg
mortality occurred only after populations had been at
epidemic levels for several years, from 1993 through
1995. With so few observations, this may be mere coincidence. However, similar observations have been
made regularly (Prentice 1954, Gautreau 1964, Witter
et al. 1975), suggesting the temporal association between winter mortality and density may be more than
coincidental. Second, it seems unlikely that temperature variation alone could produce patches of mortality varying from 0 Ð100%.
There are two good mechanistic reasons to suspect
that winter temperature acts as a partially densitydependent perturbation on M. disstria egg survival.
First, the vertical distribution of M. disstria egg masses
is density-dependent, and this inßuences the degree of
exposure to winter cold. Although moths prefer to
oviposit in upper branches of the tree crown, and
especially on the terminal shoot (Batzer et al. 1995),
they will oviposit in low trees and shrubbery when
populations are at extremely high densities, as observed in this study. Because of cold-air pooling, egg
masses laid in shrubbery are more vulnerable to perturbation by winter cold. Unfortunately, without precise annual estimates of egg densities in the canopy
and the shrub layers, it is difÞcult to assess the strength
of this density-dependence.
Second, the quality of foliage consumed by M. disstria larvae is also density-dependent, and this inßuences a populationÕs susceptibility to winter cold.
When M. disstria densities are high and aspen foliage
is in short supply, larvae will feed on other, less-preferred species of Populus, Salix, Betula, and Acer (Nicol
et al. 1997). When densities are extremely high and
overstory hosts become severely defoliated, fourth
and Þfth stage larvae are often forced to forage on
nonhost understory shrub species (Batzer et al. 1995),
including Corylus, Alnus, Amelanchier, and even Rosa.
Diet quality thus depends on larval population density. The reason we think larval food competition results in partially density-dependent susceptibility to
winter-kill is because of the effect on egg hatch in the
following generation. Experimental studies show that
M. disstria larvae fed a diet rich in nonhost foliage tend
to produce offspring that have lower overwintering
survival rates (J.R. Spence, University of Alberta, unpublished data).
We introduce the term susceptibility to describe the
conditional probability of winter-kill resulting from
varying intrinsic levels of cold tolerance. We use the
term vulnerability to describe the conditional probability of mortality resulting from varying exposure to
a given cold temperature. Vulnerability and susceptibility to winter cold are thus deterministic, densitydependent functions that help determine the level of
308
ENVIRONMENTAL ENTOMOLOGY
density-independent mortality caused by a given cold
temperature treatment. They are expected to lead to
complex patterns of mortality because they vary in
time - as a function of population densityÑand in
spaceÑas a function of extrinsic levels of exposure and
intrinsic levels of cold tolerance.
One problem is apparent. If winter egg mortality is
density-dependent, why was density unrelated to
mortality in the spatial data set? There are three possible explanations: (1) large-scale gradients in winter
temperatures may have obscured any effects of density; (2) cocoon densities measured over several hundred square meters may be a poor indication of the
degree of larval penetration into the understory; (3)
cocoon counts and hatch rate estimates could be subject to different sources of error because plot sizes and
locations were not rigidly deÞned. As all three are
likely to be true to some degree, it is not fair to
conclude from the spatial data set that winter-kill is
density-independent.
Long-Term Dynamics. Cold temperatures during
the winter of 1995Ð1996 killed 20% of the 1995 cohort
of M. disstria eggs residing in the canopy, and 70% of
those residing in the shrub layer. The population-level
mortality rate was probably in between these values,
and closer to 20%, as most eggs occur in the canopy
(Batzer et al. 1995). A 20% loss on its own is insufÞcient
to induce population collapse. However, we argue it is
the suddenness, timing, and regularity of such lossesÑ
and not so much the magnitudeÑthat is important to
the quasi-periodic dynamics of the forest tent caterpillar.
If canopy-dwelling tent caterpillar populations are
less subject to perturbation by winter cold, it is fair to
ask: why did canopy populations fail to rebound after
the cold snap in January 1996? To understand the
effect of topographical, density-dependent perturbation impacts on tent caterpillar dynamics, one must
consider the role of parasitoids in population regulation. Overwintering in snow-covered soil, parasitoids
were probably buffered from cold weather. Upon
emerging the following spring, they likely moved to
the canopy to attack the only available hosts, namely
those that were not killed by cold. The result was
prolonged suppression of both canopy- and shrubdwelling host sub-populations. Notably, it is the mobility of parasitoids and the ability to aggregate to high
host densities that couples the fate of the canopydwelling hosts to the state of the shrub-dwellers.
According to this scheme, the dynamic impact of a
given perturbation on the collective host population
depends on parasitoid density as much as on host
density. When host populations are high, they are
vulnerable to perturbation. When parasitoid populations are also high, then the system is fragile and host
populations are vulnerable to perturbative collapse. If,
instead, parasitoid populations are low, the system
would be resilient and host populations would rebound. A mild perturbation would therefore have a
relatively large impact if the system were fragile, while
a severe perturbation would have a relatively small
impact if the system were resilient. This explains how
Vol. 32, no. 2
a sudden 20% loss of eggs at the start of a season could
have a large impact on dynamics.
The hypothesis is a compelling one. If winter mortality is partially density-dependent, this could explain
why statistical evidence of an association between
winter temperature and egg survival or winter temperature and generational survival has been so elusive
in temporal analyses (e.g., Ives 1973, Witter et al. 1975,
Daniel and Myers 1995), yet so clear in spatial analyses
(e.g., Roland et al. 1998, Cooke and Roland 2000). It
could also explain why winter egg mortality exhibits
Þne-scaled spatial variation (e.g., Witter and Kulman
1972). Finally, a topoclimatic mechanism could explain why (1) forest tent caterpillar egg mortality is
sometimes higher in uplands (e.g., Gautreau 1964);
(2) defoliation tends to be most severe along river
valleys; and (3) outbreaks are less common and less
severe at northern latitudes (e.g., Daniel and Myers
1995).
In theory, density-independent perturbations can
have a marked inßuence on the temporal dynamics of
a predator-prey interaction if the impact of perturbations is partially density-dependent (Royama 1992).
For a jointly regulated system, the collapse of an outbreak may require several years for the build-up of
parasitoid populations and the arrival of a suitable
climatic perturbation. Density-dependent impacts of a
perturbing agent thus provides a mechanism whereby
stochastic factors may act as regulatory forces.
In conclusion, there is excellent evidence that (1)
forest tent caterpillar eggs cannot tolerate winter temperatures much below a certain threshold and (2) this
tolerance threshold is occasionally exceeded in Canada. There is good evidence that (3) vulnerability of
forest tent caterpillar eggs to winter freezing depends
on small-scale topoclimatic effects, and (4) susceptibility is in part determined by parental larval hostplant quality. There is some evidence of (5) a preference for canopy oviposition and foraging which
cannot be fulÞlled at very high population densities.
Collectively, this suggests winter mortality of eggs
may be driven by partially density-dependent perturbation effects.
It is unclear to what extent the periodic nature of
forest tent caterpillar outbreaks is attributable to partially density-dependent winter mortality of eggs;
however, the possibility exists. Certainly, deterministic processes, such as disease transmission, parasitism,
and predation cannot be ignored. At the same time,
our research suggests one should also not ignore the
stochastic elements. The dynamic interplay between
deterministic and stochastic processes is a rich source
of dynamics that could be especially relevant for insects living in extreme climates.
Our model of partially density-dependent perturbation effects provides a synthetic framework that
uniÞes the seemingly disparate views of Nicholson
(1954) and Andrewartha and Birch (1954). It is also
consistent with the ideas of density-vague population
change (Strong 1986) and imperfect density-dependence (Milne 1957), which have been so roundly
criticized (Berryman 1992). The concept of partially
April 2003
COOKE AND ROLAND: TENT CATERPILLAR EGG SURVIVAL AND POPULATION DYNAMICS
density-dependent perturbations is not especially
new (Royama 1992). However, it has, so far, received
little attention from empiricists, theoreticians, and
analysts.
In empirical population modeling it is typically assumed that stochastic perturbations are density-independent and that their impacts on survival are also
density-independent. But if perturbation effects are
actually partially density-dependent, then the regulatory, density-dependent structure of the plant-hostparasitoid interaction would be misrepresented. One
therefore runs the risk of falsely concluding that stochastic climatic effects are a result of deterministic,
density-dependent processes. Models so parameterized may Þt the sample data, but perform poorly in
out-of-sample validation tests and in experimental
trials.
A thorough analysis of the statistical properties of
theoretical systems governed by partially densitydependent stochastic perturbations is required. Particularly intriguing is the idea that partially densitydependent stochastic perturbation could lead to patterns of ßuctuations that closely resemble chaotic
dynamics. Distinguishing chaos from density-independent stochasticity is already quite challenging.
Evaluating the general relevance of partially density-dependent perturbation effects will be difÞcult.
Analytical models designed to detect the perturbing
inßuence of density-independent factors are presently incapable of detecting partially density-dependent stochastic effects. The statistical problem of incorporating density into the analysis is no simple
matter. It may be that analytical approaches are not
feasible and that the only alternative is experimentation and process-oriented simulation. This is an open
question that should interest ecological statisticians
and quantitative ecologists.
Acknowledgments
Helena Bylund provided the shrub-layer egg mortality
data. John Spence kindly provided access to unpublished
experimental data. Chris Schmidt and Lorne Louden assisted
with Þeld collections. This research was supported Þnancially
by the Canadian Circumpolar Boreal Institute, the Sustainable Forest Management Network of Centres of Excellence,
and the University of Alberta Biodiversity Challenge Grants
program funded by the Alberta Conservation Association.
References Cited
Andrewartha, H. G., and L. C. Birch. 1954. The Distribution
and Abundance of Animals. University of Chicago Press,
Chicago.
Batzer, H. O., M. P. Martin, W. J. Mattson, and W. E. Miller.
1995. The forest tent caterpillar in aspen stands: distribution and density estimation of four life stages in four
vegetation strata. For. Sci. 41: 99 Ð121.
Berryman, A. A. 1992. Vague notions of density-dependence. Oikos 62: 252Ð254.
Berryman, A. A. 1996. What causes population cycles of
forest Lepidoptera? TREE 11: 28 Ð32.
Berryman, A. A., and J. A. Millstein. 1989. Are ecological
systems chaoticÑand if not, why not? TREE 4: 26 Ð28.
309
Blais, J. R. 1983. Trends in the frequency, extent and severity of spruce budworm outbreaks in eastern Canada. Can.
J. For. Res. 13: 539 Ð547.
Blais, J. R., R. Prentice, W. L. Sippell, and D. Wallace. 1955.
Effects of weather on the forest tent caterpillar Malacosoma disstria Hbn. in central Canada in the spring of 1953.
Can. Entomol. 87: 1Ð 8.
Campbell, R. W. 1993. Population dynamics of the major
North American needle-eating budworms. USDA For.
Serv., PaciÞc Northwest Research Station, Res. Pap. PNW
RP-463, 222 p.
Changnon, D., T. B. McKee, and N. J. Doesken. 1993. Annual snowpack patterns across the Rockies: long-term
trends and associated 500-mb synoptic pattern. Mon.
Wea. Rev. 121: 633Ð 647.
Clark, W. C., and C. S. Holling. 1979. Process models, equilibrium structures, and population dynamics: on the formulation and testing of realistic theory in ecology.
Fortschr. Zool. 25: 29 Ð52.
Colle, B. A., and C. F. Mass. 1995. The structure and evolution of cold surges east of the Rocky Mountains. Mon.
Wea. Rev. 123: 2577Ð2610.
Cooke, B. J., and J. Roland. 2000. Spatial analysis of largescale patterns of forest tent caterpillar outbreaks. Ecoscience 7: 410 Ð 422.
Daniel, C. J., and J. H. Myers. 1995. Climate and outbreaks
of the forest tent caterpillar. Ecography 18: 353Ð362.
Davis, C. A. 1997. The modiÞcation of baroclinic waves by
the Rocky Mountains. J. Atmos. Sci. 54: 848 Ð 868.
Fitzgerald. T. D. 1995. The Tent Caterpillars. Cornell University Press, Ithaca, NY.
Gautreau, E. J. 1964. Unhatched forest tent caterpillar egg
bands in northern Alberta associated with late spring
frost. Can. Dep. For., For. Entomol. and Pathol. Branch,
Bi-monthly Prog. Rep. 20: 3.
Ginzburg, L. R., and D. E. Taneyhill. 1994. Population cycles of forest Lepidoptera: a maternal effect hypothesis.
J. Anim. Ecol. 63: 79 Ð92.
Green, G. W., and C. R. Sullivan. 1950. Ants attacking larvae
of the forest tent caterpillar, Malacosoma disstria Hbn.
(Lasiocampidae, Lepidoptera). Can. Entomol. 82: 194 Ð
195.
Hage, K. D. 1972. Nocturnal temperatures in Edmonton,
Alberta. J. Appl. Meteor. 11: 123Ð129.
Hanec, A. C. 1966. Cold-hardiness in the forest tent caterpillar, Malacosoma disstria Hübner (Lasiocampidae,
Lepidoptera). J. Insect Physiol. 12: 1443Ð1449.
Hartjenstein, G., and R. Bleck. 1991. Factors affecting cold
air outbreaks east of the Rocky Mountains. Mon. Wea.
Rev. 119: 2280 Ð2292.
Hastings, A. 1977. Spatial heterogeneity and the stability of
predator-prey systems. Theor. Pop. Biol. 12: 37Ð 48.
Hastings, A. 1998. Transients in spatial ecological models. In
J. Bascompte and R. V. Solé [eds.], Modeling Spatiotemporal Dynamics in Ecology. Springer, New York, pp. 189 Ð
198.
Hastings, A., C. L. Hom, S. Ellner, P. Turchin, and H.C.J.
Godfray. 1993. Chaos in ecology: is mother nature a
strange attractor? Annu. Rev. Ecol. Syst. 24: 1Ð33.
Hastings, A., and. K. Higgins. 1993. Persistence of transients
in spatially structured ecological models. Science 263:
1133Ð1136.
Hilborn, R. 1975. The effect of spatial heterogeneity on the
persistence of predator-prey interactions. Theor. Pop.
Biol. 8: 346 Ð55.
Hildahl, V., and W. A. Reeks. 1960. Outbreaks of the forest
tent caterpillar, Malacosoma disstria Hbn., and their ef-
310
ENVIRONMENTAL ENTOMOLOGY
fects on stands of trembling aspen in Manitoba and
Saskatchewan. Can. Entomol. 92: 199 Ð209.
Hodson, A. C. 1941. An ecological study of the forest tent
caterpillar, Malacosoma disstria Hbn., in northern Minnesota. Minn. Tech. Bull. 148.
Hodson, A. C., and C. J. Weinman. 1945. Factors affecting
recovery from diapause and hatching of eggs of the forest
tent caterpillar, Malacosoma disstria Hbn. Univ. of Minn.
Agric. Exp. Stn., Tech. Bull. 170.
Hodson, A. C. 1977. Some aspects of forest tent caterpillar
population dynamics. In: H. M. Kulman and H. M. Chiang
[eds.], Insect Ecology: Papers Presented in the A. C.
Hodson Lecture Series. Agric. Exp. Stn., Univ. of Minn.,
St. Paul, MN. Tech. Bull. 310, pp. 5Ð16.
Hunter, M. D., and P. W. Price. 1992. Playing chutes and
ladders: heterogeneity and the role of bottom-up and
top-down forces in natural communities. Ecology 73:
724 Ð732.
Ihaka, R., and R. Gentleman. 1996. R: a language for data
analysis and graphics. J. Comp. Graph. Stat. 5: 299 Ð314.
Ives, W.G.H. 1973. Heat units and outbreaks of the forest
tent caterpillar, Malacosoma disstria (Lepidoptera:
Lasiocampidae). Can. Entomol. 105: 529 Ð543.
Ives, W.G.H., and H. R. Wong. 1988. Tree and shrub insects
of the prairie provinces. Can. For. Serv., North. For. Res.
Cent. Inf. Rep. NOR-X-292.
Kaneko, K. 1998. Diversity, stability, and metapopulation
dynamics: remarks from coupled map studies. In J. Bascompte and R. V. Solé [eds.], Modeling Spatiotemporal
Dynamics in Ecology. Springer, New York, pp. 27Ð 45.
Kendall, B. E., C. J. Briggs, W. M. Murdoch, P. Turchin, S. P.
Ellner, E. McCauley, R. M. Nisbet, and S. M. Wood. 1999.
Why do populations cycle? A synthesis of statistical
and mechanistic modeling approaches. Ecology 80: 1789 Ð
1805.
Keshishian, L. G., L. F. Bosart, and W. E. Bracken. 1994.
Inverted troughs and cyclogenesis over interior North
AmericaÑa limited regional climatology and case studies.
Mon. Wea. Rev. 122: 565Ð 607.
Logan, J. A., and J. C. Allen. 1992. Nonlinear dynamics
and chaos in insect populations. Annu. Rev. Entomol. 37:
455Ð 477.
Ludwig, D., D. D. Jones, and C. S. Holling. 1978. Qualitative
analysis of insect outbreak systems: the spruce budworm
and the forest. J. Anim. Ecol. 47: 315Ð332.
Martinat, P. J. 1987. The role of climatic variation and
weather in forest insect outbreaks. In P. Barbosa and
J. Schultz [eds.], Insect Outbreaks, Academic Inc., New
York, pp. 241Ð268.
May, R. M. 1973. Stability in randomly ßuctuating versus
deterministic environments. Am. Nat. 107: 621Ð 650.
May, R. M. 1976. Simple mathematical models with very
complicated dynamics. Nature (Lond.) 261: 459 Ð 467.
Milne, A. 1957. The natural control of insect populations.
Can. Entomol. 89: 193Ð219.
Morris, R. F. 1963. The dynamics of epidemic spruce budworm populations. Mem. Entomol. Soc. Can. 31.
Myers, J. H. 1998. Synchrony in outbreaks of forest Lepidoptera: a possible example of the Moran effect. Ecology
79: 1111Ð1117.
Nicholson, A. J. 1954. An outline on the dynamics of animal
populations. Aus. J. Zool. 2: 9 Ð 65.
Nicol, R. W., J. T. Arnason, B. Helson, and M. M. Abou-Zaid.
1997. Effect of host and nonhost trees on the growth and
development of the forest tent caterpillar, Malacosoma
disstria (Lepidoptera: Lasiocampidae). Can. Entomol.
129: 991Ð999.
Vol. 32, no. 2
Parry, D., J. R. Spence, and W.J.A. Volney. 1997. Responses
of natural enemies to experimentally increased populations of the forest tent caterpillar, Malacosoma disstria.
Ecol. Entomol. 22: 97Ð108.
Parry, D., J. R. Spence, and W.J.A. Volney. 1998. Budbreak
phenology and natural enemies mediate survival of Þrst
instar forest tent caterpillar (Lepidoptera: Lasiocampidae). Environ. Entomol. 27: 1368 Ð1374.
Pinheiro, J. C., and D. M. Bates. 2000. Mixed Effects Models
in S and S-PLUS. Springer, New York
Prentice, R. M. 1954. Decline of populations of the forest
tent caterpillar in central Saskatchewan. Can. Dep. Agric.,
For. Biol. Div., Bi-monthly Prog. Rep. 10: 5.
Raske, A. G. 1975. Cold-hardiness of Þrst instar larvae of the
forest tent caterpillar, Malacosoma disstria (Lepidoptera:
Lasiocampidae). Can. Entomol. 107: 75Ð 80.
Reeve, J. D. 1988. Environmental variability, migration, and
persistence in host-parasitoid systems. Am. Nat. 132: 810 Ð
836.
Régnière, J., and T. J. Lysyk. 1995. Population dynamics of
the spruce budworm, Choristoneura fumiferana. In Armstrong, J. A. and W.G.H. Ives [Eds.], Forest Insect Pests
in Canada. Nat. Res. Can., Can. For. Serv., Ottawa, Ont.,
pp. 95Ð105.
Roland, J., B. G. Mackey, and B. Cooke. 1998. Effects
of climate and forest structure on duration of forest
tent caterpillar outbreaks across central Ontario. Can.
Entomol. 130: 703Ð714.
Roland, J., and P. D. Taylor. 1997. Insect parasitoid species
respond to forest structure at different spatial scales.
Nature (Lond.) 386: 710 Ð713.
Rose, M. R., and R. Harmsen. 1978. Using sensitivity analysis
to simplify ecosystem models: a case study. Simulation 31:
15Ð26.
Rose, M. R., and R. Harmsen. 1981. Ecological outbreak
dynamics and the cusp catastrophe. Acta Biotheoretica.
30: 229 Ð253.
Royama, T. 1984. Population dynamics of the spruce budworm Choristoneura fumiferana. Ecol. Monogr 54: 429 Ð
462.
Royama, T. 1992. Analytical Population Dynamics. Chapman & Hall, New York.
Schaffer, W. M., and M. Kot. 1985. Do strange attractors
govern ecological systems? Bioscience 35: 342Ð350.
Sippell, L. 1962. Outbreaks of the forest tent caterpillar,
Malacosoma disstria Hbn., a periodic defoliator of broadleaved trees in Ontario. Can. Entomol. 94: 408 Ð 416.
Strong, D. R. 1986. Density-vague population change.
TREE 1: 39 Ð 42.
Turchin, P. 1995. Population regulation: old arguments and
a new synthesis, pp. 19 Ð 40. In N. Cappuccino and P. Price
[eds.], Population Dynamics: New Approaches and Synthesis. Academic, Toronto.
Turchin, P., and A. D. Taylor. 1992. Complex dynamics in
ecological time-series. Ecology 73: 289 Ð305.
Wetzel, B. W., H. M. Kulman, and J. A. Witter. 1973. Effects
of cold temperatures on hatching of the forest tent caterpillar, Malacosoma disstria (Lepidoptera: Lasiocampidae). Can. Entomol. 105: 1145Ð1149.
Williams, D.J.M., D. Parry, and D. W. Langor. 1996. Sampling and identiÞcation of forest tent caterpillar parasitoids in the prairie provinces. Canadian Forest Service,
Northwest Region, Northern Forestry Centre, Information Report NOR-X-345.
Wilson, H. B., C. Godfray, M. P. Hassell, and S. Pacala. 1998.
Deterministic and stochastic host-parasitoid dynamics in
spatially extended systems. In J. Bascompte and R. V. Solé
April 2003
COOKE AND ROLAND: TENT CATERPILLAR EGG SURVIVAL AND POPULATION DYNAMICS
[eds.], Modeling Spatiotemporal Dynamics in Ecology.
Springer, New York, pp. 63Ð 81.
Witter, J. A. 1979. The forest tent caterpillar (Lepidoptera:
Lasiocampidae) in Minnesota: a case history review.
Great Lakes Entomol. 12: 191Ð197.
Witter, J. A., and H. M. Kulman. 1972. Mortality factors
affecting eggs of the forest tent caterpillar, Malacosoma
disstria (Lepidoptera: Lasiocampidae). Can. Entomol.
104: 705Ð710.
311
Witter, J. A., W. J. Mattson, and H. M. Kulman. 1975. Numerical analysis of a forest tent caterpillar (Lepidoptera:
Lasiocampidae) outbreak in northern Minnesota. Can.
Entomol. 107: 837Ð 854.
Received for publication 19 February 2002; accepted 3 October 2002.