13652311, 2002, 5, Downloaded from https://resjournals.onlinelibrary.wiley.com/doi/10.1046/j.1365-2311.2002.00441.x by Cochrane Mexico, Wiley Online Library on [03/01/2023].

See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Ecological Entomology (2002) 27, 565±573

Dynamics of springtail and mite populations: the role of

density dependence, predation, and weather
S T E V E N H . F E R G U S O N 1 and D A M I E N O . J O L Y 2 1
Faculty of Forestry and the Forest
2
Environment, Lakehead University and Department of Biology, University of Saskatchewan, Canada

Abstract. 1. Ecological theory suggests that density-dependent regulation of

organism abundance will vary from exogenous to endogenous factors depending
on trophic structure. Changes in abundance of soil arthropods were investigated
at three trophic levels, springtails (Collembola), predaceous mites (Acari), and
macro-arthropods (spider, adult and larval beetles, centipedes). Predictions were
that springtails are predator regulated and mites are food limited according to
the Hairston et al. (1960) model, which predicts alternating regulation by com-
petition and predation from fungi to springtails to mites to macro-arthropods.
The alternate hypothesis was based on the bottom-up model of trophic
dynamics, which predicts that each trophic level is regulated by competition for
resources.
2. The relative contributions to springtail and mite population dynamics of
endogenous (i.e. density-dependent population growth related to food availability)
and exogenous (i.e. predation and weather) factors were tested using time-series
analysis and experimental manipulation of water conditions. Box patterns were
distributed within an aspen forest habitat located in the Canadian prairies and
surveyed weekly from May to September 1997±1999. Each box depressed the leaf
litter, creating a microhabitat island for soil arthropods that provided counts of
invertebrates located immediately beneath the boxes.
3. Strong evidence was found for endogenous control of springtail and mite
numbers, indicated by a reduction in population growth related to density in the
previous week. Contrary to predictions, no evidence was found for regulation of
springtail numbers by mites, or for regulation of mite numbers by macro-arthro-
pods. Springtail population growth rate was related positively to current springtail
density (8 and 23% variation explained) and related negatively to 1-week lagged
density (85 and 58%), and related negatively to temperature (5 and 5%) for time-
series data and for experimental addition of water respectively. Mite population
growth rate was related positively to current mite density (54%) and temperature
(4%), and negatively to 1-week lagged mite density (20%) and precipitation (6%)
for time-series analysis. For experimental addition of water, mite growth rate
was related positively to current mite density (44%) and temperature (5%), and
negatively to 1-week lagged density (11%). Results differed from the Hairston
et al. (1960) model predictions but were consistent with a bottom-up view that
springtail and mite populations were regulated intrinsically by competition for
food and secondarily by temperature as a function of reproduction.

Correspondence: Steven H. Ferguson, Faculty of Forestry and

the Forest Environment, 955 Oliver Road, Lakehead University,
Thunder Bay, ON, P7B 5E1, Canada. E-mail: Steven.Ferguson@
lakeheadu.ca

# 2002 The Royal Entomological Society 565

 13652311, 2002, 5, Downloaded from https://resjournals.onlinelibrary.wiley.com/doi/10.1046/j.1365-2311.2002.00441.x by Cochrane Mexico, Wiley Online Library on [03/01/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
566 Steven H. Ferguson and Damien O. Joly

Key words. Acari, Collembola, humidity, life cycle, macro-arthropods,

numerical response, population dynamics, precipitation, regulation, reproduction,
time-series analysis.

Introduction food are less well known (Theenhaus et al., 1999). Preda-
ceous mites commonly feed on springtails (Berg et al., 1998;
A key question in soil biology and understanding the Ferguson, 2001) and both springtails and mites are common
trophic structure of the biota in the soil-litter system is prey of macro-arthropods, including spiders, beetles, and
how populations are limited relative to trophic position centipedes (Eisenbeis & Wichard, 1987; Ferguson, 2001).
(Pace et al., 1999; Walker & Jones, 2001). Although there This research was designed to investigate whether, and
is recent theoretical evidence to suggest that soil food-web by what means, springtail and mite numbers are regulated.
structure can influence population processes (Zheng et al., A regulatory factor is any density-dependent process that
1997), there is a dearth of experimental data on the eco- keeps populations within predictable density ranges by
system-level consequences of soil food-web architecture, affecting population growth quantifiably (Murray, 1982;
and much remains unknown about how soil food-web Fowler, 1987; Sinclair, 1989; Sinclair & Pech, 1996). First,
structure affects ecosystem productivity (Wardle, 1999). springtail and mite time-series data were examined for
Some general patterns in the trophic organisation of soil density-dependent population growth to determine whether
animal communities have been observed despite a high past densities regulate population growth rate. Population
degree of stochastic variation (Swift et al., 1979; Petersen regulation could result from a number of factors, including
& Luxton, 1982; Schaefer, 1995; Mikola & SetaÈlaÈ, 1998; density-dependent predation by mites on springtails (or
Ferguson, 2001): high species diversity within trophic macro-arthropods on mites) and intra-guild competition
groups, tendency to trophic generalism, high numbers and for resources via weather conditions (e.g. temperature and
fluctuating mesofaunal predators, and continuity in litter precipitation). For example, determining whether mites
decomposition. exert a negative influence on springtail population growth
An important question regarding soil community inter- rate tested mite predation as a regulatory factor of spring-
actions among trophic levels is whether populations are tails. If mites or macro-arthropods are shown not to be
regulated by competition for resource availability or by regulated by predation, then density dependence in spring-
density-dependent predation. The Hairston et al. (1960) tail and mite population growth arises from intra-guild
model predicts alternating regulation by predation and competition for resources. Finally, time-series analysis and
competition in the three-level chain such that micro-flora experimental manipulation were used to evaluate the limit-
(i.e. fungi and algae) are regulated by competition for food; ing effect of precipitation and temperature on springtail
microphytophages, such as springtails (Collembola), are and mite numbers. The experimental manipulation increased
predator regulated; and predators (e.g. predaceous mites) moisture by water addition, resulting in an increase in litter
are regulated by competition for prey (e.g. Schaefer, 1995). decomposition. Water addition should lead to no consistent
In contrast to the Hairston et al. (1960) model, however, the change in springtail biomass (Hairston et al., 1960) or lead
bottom-up view suggests that micro-phytophages (e.g. Col- to an increase of springtail and mite biomass if populations
lembola) and their predators (e.g. mites) are food controlled are food limited (bottom-up model).
in detritivore communities (e.g. Laakso & SetaÈlaÈ, 1999).
In the work reported here, the Hairston et al. (1960) and
bottom-up hypotheses were tested using time-series analysis Materials and methods
and a field experiment of springtail and mite populations in
a soil-leaf litter system. The experimental design to study Study organisms
trophic-level effects consisted of boxes distributed within
an undisturbed aspen Populus tremuloides forest. The weight In this analysis, micro-arthropods included groups of
of the box depressed the litter layer similar to a rock on the springtails (Insecta: Collembola) and mites (Acari: Oriba-
forest floor. Microclimate differences under the box relative tida; family Gamasides). In contrast, macro-arthropods
to exposed soil litter created a home to a diverse soil fauna included spiders, beetle larvae and adults, and centipedes.
community (Ferguson, 2000). Springtails are primarily Evidence of macro-arthropod groups capturing and hand-
fungivores or detritivores that play an important ecological ling micro-arthropods, probably as food, was gathered
role in preparing organic material for decomposition over the field research. Due to small numbers, spiders, beetles,
(Zinkler & Stecken, 1985). Populations of springtails are and centipedes were combined as one group. Other arthro-
considered to be controlled by exogenous factors such as pods were counted; these data are presented elsewhere
temperature and moisture (Klironomos & Kendrick, 1995; (Ferguson, 2001). Approximately five surface-dwelling
Hopkin, 1997) or predation (HaÊgvar, 1995; Schaefer, 1995), springtail species are characteristically found on xeric
however the influences of endogenous factors that act in a prairie (Jensen et al., 1973) and were placed in one group
density-dependent manner such as competition for limited as they occupy a similar niche, consuming detritus, fungi,

# 2002 The Royal Entomological Society, Ecological Entomology, 27, 565±573

 13652311, 2002, 5, Downloaded from https://resjournals.onlinelibrary.wiley.com/doi/10.1046/j.1365-2311.2002.00441.x by Cochrane Mexico, Wiley Online Library on [03/01/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Population dynamics of springtails and mites 567

and algae (Wolters, 1991; Hopkin, 1997). Surface-dwelling daylight. Counts of springtails and mites were an index of
springtails are soft-bodied plankton of the soil (Aoki, 1973; surface-dwelling numbers and are not considered an absolute
Johnston, 2000). Springtails are parthenogenetic; sexual estimate of abundance. In 1999, moisture levels were
maturity generally occurs at the sixth instar and oviposition manipulated under randomly chosen box sets (n ˆ 12) by
generally occurs within 12±48 h after moulting at 21  C pouring 250 ml of water under each box in the set (total
(Joosse, 1981; Axelsen et al., 1997). The size of springtails 1.25 l) after each survey. The 1997±1999 time-series data
averaged 1.5 mm (range: 0.2±2.6 mm, n ˆ 20) or 0.67 mg included data from the 12 plots in 1999 that were not
(range: 0.09±7.13 mg); mites averaged 1.7 mm including legs treated experimentally with water addition (i.e. controls).
(range: 1.1±2.9 mm) or 0.49 mg (range: 0.32 6.10 mg; Both time-series (1997±1999) and experimental (1999) data
S. H. Ferguson, unpublished). Predatory mites were repres- were used to determine factors affecting springtail and mite
ented by the genus Labidostomma and were placed in one ecology.
taxonomic group. Mites are protected by an exoskeleton and Weather parameters for Saskatoon were measured daily
are non-specific predators that commonly feed on springtails at the Kernen Prairie weather station by the Agriculture
(Berg et al., 1998; S. H. Ferguson, pers. obs.). The life cycle Department, University of Saskatchewan (12 km N of study
of the mite is comparatively simple, consisting of four site). Total precipitation (mm) and average weekly tempera-
gamasid stages: larva, protonymph, deutonymph, adult. ture ( C) at 5-cm soil depth were measured as the sum and
The length of the life cycle of mites is related indirectly mean of daily values between weekly surveys. As there were
to temperature within the range 15±28  C (Wright & weeks with no precipitation, 0.5 was added to precipitation
Chambers, 1994). before loge transformation (Yamamura, 1999).

Study site
Time-series analysis
2
The study area consisted of a 0.13-km stand of undis-
turbed trembling aspen in the prairie region of south-central First, the instantaneous rate of increase (hereafter
Saskatchewan, Canada (52 100 N, 106 410 W). Mean monthly referred to as population growth rate) of springtails and
daily temperature (1961±1990) varied from 3.9  C in April mites was calculated each year as:
to 18.6  C in July (annual mean ˆ 2.0  C) and precipitation
varied from 19.7 mm in April to 63.4 mm in June (annual Rt ˆ [ln(Nt ‡ T) ln(Nt)]/T
mean ˆ 28.9 mm) (Environment Canada, 2000).
where Nt is the population size at time t and T is the number
of days between consecutive surveys. A total of 61 surveys
Survey design was conducted, and 58 population growth rates (Rt) were
calculated from the results of the first (Nt) and second
Factors determining micro-arthropod population (Nt‡T) surveys for each year. Four sets of analyses were
dynamics were examined by counting springtails, mites, conducted: the 1997±1999 time-series data and the water
and macro-arthropods under artificial microhabitats addition experiment (1999) were analysed separately for
(n ˆ 24), created using contiguous sets of five 2-l plastic- both springtails and mites. Although springtails and mites
coated cardboard milk cartons (soil surface area 200 cm2). were analysed separately, densities of predators and prey
Boxes were filled with sand to weigh 2 kg, and depressed were included in general linear models to test for predation
the soil on average 1.3 cm. These boxes simulated a micro- effects. Auto-regression analysis of springtail and mite
habitat island for soil arthropods similar to a rock on the population growth rate with density and time (week, year)
forest floor (Ferguson, 2000). Boxes were overturned and as explanatory variables was used to produce residuals that
the numbers and size of springtails and mites were counted. removed temporal effects.
An effort was made to lift individual boxes without disturb- To test for density dependence, current and lagged (1, 2,
ing adjacent boxes and the pattern of lifting each of the five 3, 4, and 5 weeks) numbers of springtails and mites were
boxes varied among surveys. Twenty-three, 24, and 20 used to explain population growth rate (residuals). Negative
weekly surveys were conducted from May to October correlations between population growth rate and lagged
1997, 1998, and 1999 respectively. For 1997, the first six density would suggest density dependence in the springtail
survey results were excluded from data analysis due to a and mite time-series data. Auto-regression was used to
colonisation effect whereby few soil arthropods were explore time lags in the effect of temperature and precipita-
observed initially. Therefore, 17 surveys in 1997 were tion on springtail and mite population growth rate using
included in the time-series analyses for a total of 61 surveys correlograms.
across 3 years. Surveys began when the ground was no Growth rate residuals were used as the response variable
longer covered by snow and evidence of soil arthropods in regression analyses to determine which explanatory
under boxes was first observed. Surveys ended when few, factors influenced growth rate, with model acceptance set
if any, soil arthropods were observed under boxes. Surveys at P < 0.05. Explanatory variables included density depen-
were conducted between 11.00 and 17.00 hours during dence (current and lagged density), predation (predator

# 2002 The Royal Entomological Society, Ecological Entomology, 27, 565±573

 13652311, 2002, 5, Downloaded from https://resjournals.onlinelibrary.wiley.com/doi/10.1046/j.1365-2311.2002.00441.x by Cochrane Mexico, Wiley Online Library on [03/01/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
568 Steven H. Ferguson and Damien O. Joly

density), and weather (precipitation and temperature).

The complete set of explanatory variables was examined in
all combinations to determine the best model according to
Akaike information criteria (AIC). Thus, possible models
explaining variation in mite and springtail population rate
of growth (residuals) were determined based on the best
compromise between parsimony and bias (Burnham &
Anderson, 1998).
Mite and macro-arthropod predation as potential factors
limiting springtail and mite density respectively, was exam-
ined using partial correlation analysis (Sokal & Rohlf,
1981). For example, a negative correlation between preda-
ceous mite numbers and springtail population growth rate,
Rt, was tested while controlling statistically for springtail
numbers. Correlations with population growth rates are
mathematically equivalent to correlations with total mor-
tality when loss of breeding potential is treated as a killing
factor (Clutton-Brock et al., 1985). For example, a negative
correlation between mite numbers and springtail popula-
tion growth rate would indicate that mite predation acts as
a limiting factor, assuming that predation rate is linearly
proportional to the abundance of mites (Messier, 1991).

Results

Auto-regression of the 1997±1999 time-series data with

lagged precipitation, temperature, and density was used to
determine major explanatory variables to be included in
the final model describing springtail and mite population
growth rate. For precipitation, 1-week lagged precipitation
explained most (r2 ˆ 0.18 and 0.08) variation in springtail
and mite population growth rate respectively (Fig. 1a,b).
For temperature, 2- and 1-week lagged temperature
explained most (r2 ˆ 0.13 and 0.26) variation in springtail
and mite population growth rate respectively (Fig. 1c,d).
Most variation in population growth rate was explained
by current density (r2 ˆ 0.56 and 0.69 for springtails and
mites respectively) and 1-week lagged density (r2 ˆ 0.93
and 0.82) compared with 2- to 5-week lagged density (r2 ˆ
0.01±0.22) (Fig. 1e,f). Current density and 1-week lagged
density were therefore used as explanatory variables in the
subsequent analyses.
Next, auto-regression models were used to determine
regulation of springtail population growth rate. First for
springtails, analysis of the 1997±1999 time-series data failed
to find an effect of predation by mites (i.e. no negative Fig. 1. Autocorrelation function for soil arthropod time series
correlation between springtail population growth rate 1997±1999. (a) Springtail population growth rate as a function of weekly
and mite density) (Table 1). Only a small negative effect of time-lagged precipitation [loge transformed precipitation (mm) ‡ 0.5].
1-week lagged temperature (5% of variation explained) was (b) Mite population growth rate as a function of weekly time-
lagged precipitation [loge transformed precipitation (mm) ‡ 0.5].
estimated for springtail growth rate. Hot weather resulted
(c) Springtail population growth rate as a function of weekly time-
in a decrease in springtail numbers the following week. lagged temperature (loge transformed temperature,  C). (d) Mite
A strong negative correlation was detected between springtail population growth rate as a function of weekly time-lagged
population growth rate and 1-week lagged springtail density precipitation (loge transformed temperature,  C). (e) Springtail
(79%), indicating intra-guild density dependence (Table 1). population growth rate as a function of weekly time-lagged density
A significant, positive effect on springtail population growth (loge transformed springtail density). (f) Mite population growth
rate was estimated for current springtail density (8% rate as a function of weekly time-lagged precipitation (loge trans-
of variation explained). The strong negative association of formed mite density).

# 2002 The Royal Entomological Society, Ecological Entomology, 27, 565±573

 13652311, 2002, 5, Downloaded from https://resjournals.onlinelibrary.wiley.com/doi/10.1046/j.1365-2311.2002.00441.x by Cochrane Mexico, Wiley Online Library on [03/01/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Population dynamics of springtails and mites 569

Table 1. Auto-regressive coefficient estimates based on general linear model delineating factors affecting springtail population growth rate
1997±1999 time-series data (F2,57 ˆ 138.5, P ˆ 0.001)y. Detrending of time series was achieved by calculating residuals from regression over
time (year and weekly survey) and using the first difference for population growth rate (DXt ˆ Xt ± Xt 1, where Xt ˆ logeNt‡1 logeNt).
Variables considered most influential according to the most parsimonious model (Akaike information criteria) are delineated *.

Source Coefficient Partial r2 F P

Intercept ‡0.542  0.076 ± ± 0.001
*Current springtail density ‡0.0393  0.0054 0.070 (0.08)§ 52.2 0.001
*Lagged springtail density 0.141  0.0066 0.608 (0.85) 452.9 0.001
Mite density ‡0.0022  0.0058 0.0002 0.15 0.70
Precipitation (mm)z ‡0.0049  0.011 0.0003 0.21 0.65
*Temperature 0.0413  0.025 0.004 (0.05) 2.78 0.10

yR2 ˆ 0.93 (adjusted R2 ˆ 0.92).

zloge(precipitation ‡ 0.5).
§Partial r2 for best model *.

1-week lagged density and weaker positive association of For the 1999 water-addition experiment, no regulating
current density resulted in a cyclic pattern of striking effect of mites on springtail was established (Table 2).
increases and abrupt population decreases (Fig. 2). Again, springtail population growth rate was correlated
negatively with a 1-week lagged density (58% of variation
explained), and correlated positively with current springtail
density (23%; Table 2). A weak, negative relationship
between springtail growth rate and temperature (5%) was
detected. Also, water addition to the soil under boxes was
not a significant factor in describing springtail population
growth; however springtail numbers were correlated
strongly with precipitation (Fig. 3), and water addition
resulted in a significant increase in springtail numbers
(Fig. 4).
Analysing the 1997±1999 time-series data to determine
factors influencing mite population growth rate did not
find a significant effect due to macro-arthropod predators
or to springtail density (Table 3). A negative effect of
precipitation and a positive effect of temperature on mite
population growth rate were detected, however a negative
correlation was found between mite population growth rate
and 1-week lagged mite density (20% of variation
Fig. 2. Natural logarithm of the weekly numbers of springtails, explained), indicating intra-guild density dependence
mites, and macro-arthropods (spiders, beetle adults and larvae, (Table 3). Also, a significant, positive effect on mite popula-
and centipedes) for an aspen forest in Saskatchewan, Canada,
tion growth rate occurred for current mite density (54%).
1997±1999.

Table 2. Auto-regressive coefficient estimates based on general linear model delineating factors affecting springtail population growth rate
with experimental water treatment in 1999 (F6,37 ˆ 42.01, P ˆ 0.001)y. Detrending of time series was achieved by calculating residuals from
regression over time (weekly survey) and using the first difference for population growth rate (DXt ˆ Xt Xt 1). Variables considered most
influential according to the most parsimonious model (Akaike information criteria) are delineated *.

Source Coefficient Partial r2 F P

Intercept ‡0.254  0.053 ± ± 0.001
*Current springtail density ‡0.0418  0.0090 0.076 (0.23)§ 21.6 0.001
*Lagged springtail density 0.108  0.0098 0.430 (0.58) 121.7 0.001
Mite density 0.0024  0.0089 0.0003 0.07 0.79
Water 0.012  0.015 0.002 0.61 0.44
Precipitation (mm)z ‡0.000  0.005 0.0000 0.00 0.99
*Temperature 0.0072  0.0034 0.016 (0.05) 4.48 0.04

yR2 ˆ 0.89 (adjusted R2 ˆ 0.87).

zloge(precipitation ‡ 0.5).
§Partial r2 for best model *.

# 2002 The Royal Entomological Society, Ecological Entomology, 27, 565±573

 13652311, 2002, 5, Downloaded from https://resjournals.onlinelibrary.wiley.com/doi/10.1046/j.1365-2311.2002.00441.x by Cochrane Mexico, Wiley Online Library on [03/01/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
570 Steven H. Ferguson and Damien O. Joly

Fig. 4. Springtail numbers under treatment (250 ml water added

Fig. 3. The effect of precipitation on springtail numbers in 1999. after each survey) and control boxes (1999).

Table 3. Auto-regressive coefficient estimates based on general linear models delineating factors affecting mite population growth rate
1997±1999 time-series data (F5,57 ˆ 36.07, P ˆ 0.001)y. Detrending of time series was achieved by calculating residuals from regression over
time (year and weekly survey) and using the first difference for population growth rate (DXt ˆ Xt Xt 1). Variables considered most
influential according to the most parsimonious model (Akaike information criteria) are delineated *.

Source Coefficient Partial r2 F P

Intercept 0.0706  0.076 ± ± 0.36
*Current mite density ‡0.0948  0.010 0.371 (0.54)§ 86.12 0.001
*Lagged mite density 0.112  0.0091 0.658 (0.20) 152.9 0.001
Springtail density 0.0018  0.0064 0.0004 0.08 0.78
Macro-arthropods ‡0.00023  0.015 0.0000 0.00 0.99
*Precipitation (mm)z 0.0268  0.010 0.027 (0.06) 6.36 0.014
*Temperature ‡0.134  0.072 0.015 (0.04) 3.44 0.069

yR2 ˆ 0.78 (adjusted R2 ˆ 0.76).

§Partial r2 for best model *.
zloge(precipitation ‡ 0.5).

Table 4. Auto-regressive coefficient estimates based on general linear model delineating factors affecting mite population growth rate with
experimental water treatment in 1999 (F6,37 ˆ 9.41, P ˆ 0.001)y. Detrending of time series was achieved by calculating residuals from
regression over time (weekly survey) and using the first difference for population growth rate (DXt ˆ Xt ± Xt 1). Variables considered most
influential according to the most parsimonious model (Akaike information criteria) are delineated *.

Source Coefficient Partial r2 F P

Intercept ‡0.219  0.24 ± ± 0.37
*Current mite density ‡0.0824  0.016 0.311 (0.44)§ 27.18 0.001
*Lagged mite density 0.0721  0.014 0.309 (0.11) 27.01 0.001
Springtail density ‡0.0262  0.017 0.027 2.40 0.13
Water 0.0266  0.025 0.014 1.18 0.29
Precipitation (mm)z 0.00556  0.0092 0.004 0.37 0.55
*Temperature ‡0.146  0.099 0.025 (0.05) 2.18 0.15

yR2 ˆ 0.65 (adjusted R2 ˆ 0.58).

zloge(precipitation ‡ 0.5).
§Partial r2 for best model *.

This combination of a strong, positive relationship with A similar general linear model predicted mite population
current density and a weaker, negative relationship with growth rate using the 1999 water-addition experiment
1-week lagged density resulted in less erratic changes in (Table 4). No evidence for density-dependent predation
mite abundance relative to springtails (Fig. 2). effects on mite population growth rate was found for

# 2002 The Royal Entomological Society, Ecological Entomology, 27, 565±573

 13652311, 2002, 5, Downloaded from https://resjournals.onlinelibrary.wiley.com/doi/10.1046/j.1365-2311.2002.00441.x by Cochrane Mexico, Wiley Online Library on [03/01/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Population dynamics of springtails and mites 571

macro-arthropod numbers or springtail density (Table 4). and 1.9% of faunal biomass respectively (S. H. Ferguson,
Mite population growth rate was greater in areas with unpublished). In comparison, ants dominate, contributing
greater springtail density (5% of variation explained), and 30.5% of total biomass (S. H. Ferguson, unpublished).
was correlated negatively with 1-week lagged mite numbers Schaefer (1995) estimated that 1% of faunal biomass
(11%) and positively with current mite density (44%). consisted of springtails. Hence, the production of springtail
Water treatment and precipitation had no significant effect biomass is low compared with predation by a diverse and
on mite population growth rate. abundant predator community. Because some springtail
populations are prone to predation and live in a super-
abundance of detritus and fungal mass, some researchers
Discussion have argued that springtails endure high predation pressure
and tend to be predator regulated (Schaefer, 1995; Bilgrami,
Contrary to the top-down hypothesis (Hairston et al., 1997). For example, field experiments with artificial mani-
1960), evidence was found for bottom-up control of spring- pulations of spider and centipede density demonstrate the
tail and mite abundance that may relate to low productivity importance of these predatory populations for limiting
on the prairies (Knapp et al., 2001). Time-series analysis springtail density (Schaefer, 1995), although more systematic
suggests that seasonal population eruptions are the conse- and realistic experiments are required to provide convincing
quence of the interplay between the endogenous factors documentation. Also, simulations of the predator±prey
(density-dependent food limitation) and exogenous factors interactions of a predaceous mite and a springtail in a
(variation in temperature, e.g. Pech et al., 1992). Density microcosm experiment indicated sensitivity to changes in
dependence in springtail and mite population dynamics predator search rate, capture efficiency, habitat selection,
was demonstrated by a negative correlation between stored reserves, and springtail survival and reproduction
population growth rate and 1-week lagged density. Possible (Axelsen et al., 1997). Springtails, however, are a diverse
density-dependent mechanisms include intraspecific group of soil arthropods hosting a diverse predator com-
interactions or the effects of generalist predators (Hanski munity associated with the soil ecosystem.
et al., 1991). In order for mite predation of springtails, or Other researchers have noted weak or no effects of pre-
macro-arthropod predation of mites, to be density depend- dators on prey in soil systems (Laakso & SetaÈlaÈ, 1999).
ent, and therefore regulatory, there must be density depend- Although predators can have an indirect effect on the rate
ence in both the functional and numerical responses at which microbes are consumed, the micro-flora can
(Messier, 1996), except where the predator exhibits a compensate for biomass consumption by altering rate of
sigmoidal functional response. As mite numbers did not turnover (Mikola & SetaÈlaÈ, 1998). This should, in turn, help
correlate with springtail population growth rate and did not to buffer microbial-mediated processes against the effect of
exert a negative influence on springtail population growth top predators. Numerous researchers have reported a posi-
rate, it is argued that predation by mites did not regulate tive correlation between moisture content and density of
springtail numbers. Similarly, macro-arthropod predation soil-dwelling springtails (Erasmus & Ryke, 1970; Badejo &
of mites did not correlate with mite population growth rate. Van Straalen, 1993; Badejo et al., 1998). Precipitation is
Finally, analyses of time-series data and experimental related to primary productivity (Rosenzweig, 1968) and
addition of water indicated that springtail and mite has a strong effect on population dynamics of vertebrate
numbers were partly a function of temperature. Springtail populations (Swanson, 1998). A thin, chitinous exoskeleton
reproduction is probably temperature and ground moisture limits springtail distribution within ecosystems to sites with
dependent (Hopkin, 1997). Similarly, temperature influ- adequate moisture (Christiansen, 1992).
ences the life cycle of mites (Wright & Chambers, 1994). Most long-term census data on insect and vertebrate
Results from the addition of water generally support the species have been analysed on a yearly time scale so that
view that variation in rainfall and temperature regulate abundance in any given year is related to that in the pre-
springtail and mite densities through intra-guild competi- vious year (Turchin, 1990; Royama, 1992; Turchin & Taylor,
tion for resources (rainfall) and seasonal conditions (tem- 1992; Perry et al., 1993), yet dynamics of populations
perature) suitable for reproduction. The results of this study within years (e.g. seasonality) form the basis of many insect
suggest that changes in springtail and mite numbers may be population dynamics (Sequeira & Dixon, 1997). Clearly,
explained primarily on the basis of climate and competition important population regulatory processes, such as repro-
for food, and secondarily by predation. Although not ductive life cycles, may be operating on shorter time scales
investigated, other factors such as parasites or disease may than annual abundance. The multiple peak populations
influence springtails and mites through density-dependent of soil-dwelling springtails during the annual cycle arise
regulation. from the many generations of springtails during that cycle
This experiment was designed to determine the dynamic (Badejo & Van Straalen, 1993; Schaefer, 1995; Badejo et al.,
role of precipitation in affecting changes in numbers of 1998). The decline in populations after each peak may
litter-dwelling springtails and mites. Thus, which particular indicate the end of a generation as the pattern of fluctuation
enemies play an important role in causing springtail and in the populations differs with different genera (Badejo
mite oscillations is unknown. Springtails and mites are the et al., 1998). This implies that each genus has a particular
most numerous micro-arthropods but contribute only 3.9 life history and that the population dynamics of each genus

# 2002 The Royal Entomological Society, Ecological Entomology, 27, 565±573

 13652311, 2002, 5, Downloaded from https://resjournals.onlinelibrary.wiley.com/doi/10.1046/j.1365-2311.2002.00441.x by Cochrane Mexico, Wiley Online Library on [03/01/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
572 Steven H. Ferguson and Damien O. Joly

can be influenced by variation in micro-environmental Acknowledgements

attributes. For example, one genus (Rhodanela sp.) inhabit-
ing the litter layer produced large swarms after rainfall in Q. Ferguson helped to design the study and conduct the
the litter layer of an undisturbed forest soil (Lasebikan field surveys. D. O. Joly was supported by a University of
et al., 1985). Saskatchewan Graduate Scholarship and subsequently
Apparently, predation of springtails by mites has no a Natural Sciences and Engineering Council Postgraduate
dynamic feedback; contrary to speculation that predation Scholarship. S. H. Ferguson received financial support
by mites regulates springtails (HaÊgvar, 1995). Possibly, the from post-doctoral funding from the University of
low productivity of prairie ecosystems did not support Saskatchewan and the Government of the Northwest
a predator capable of regulating springtails and mites Territories. We benefited from the constructive comments
(Oksanen & Oksanen, 2000; Ferguson, 2001). For this and criticisms of B. Patterson and two anonymous
system, the average predator-induced mortality may be reviewers.
very high and the intrinsic rate of population increase low,
yet still predators had no dynamic impact. Hence, fluctua-
tions in predator-imposed mortality may affect springtail References
density in a stochastic manner but they cannot drive a
regular oscillation. It is worth examining further the lack Aoki, J. (1973) Soil Zoology. Hokuryu-kan, Tokyo.
of correlation between prey and predators. Note that the Axelsen, J.A., Holst, N., Hamers, T. & Krogh, P.H. (1997)
slopes of the relationship between springtail and mite Simulations of the predator±prey interactions in a two species
numbers and population growth rate were shallow (slope ecotoxicological test system. Ecological Modelling, 101, 15±25.
ˆ 0.11 and 0.09 for springtails and mites respectively), Badejo, M.A., Nathaniel, T.I. & Tian, G. (1998) Abundance of
suggesting that a large variation in numbers is possible. springtails (Collembola) under four agroforestry tree species
with contrasting litter quality. Biology and Fertility of Soils, 27,
For example, a four-fold increase in springtails occurred
15±20.
in 1999 with addition of water. By definition, a regulatory Badejo, M.A. & Van Straalen, N.M. (1993) Seasonal abundance
force increases in effect at high densities and decreases at of springtails in two contrasting environments. Biotropica, 25,
low densities. Regulation by mite predation cannot occur if 222±228.
mite population size does not respond to springtail density. Berg, M.P., Kniese, J.P., Bedaux, J.J.M. & Verhoef, H.A. (1998)
In fact, during late summer, mites outnumbered springtails. Dynamics and stratification of functional groups of micro- and
Feeding by mites on their own faecal matter (coprophagy) mesoarthropods in the organic layer of a Scots pine forest.
in the absence of prey suggests that adult mites can delay Biology and Fertility of Soils, 26, 268±284.
starvation in the absence of prey (2±5 weeks; Bilgrami, 1994, Bilgrami, A.L. (1994) Predatory behaviour of a nematode feeding
1997; Wright & Chambers, 1994), whereas adult mites can mite Tyrophagus putrescentiae (Sarcoptiformes: Acaridae).
Fundamental Applied Nematology, 17, 293±297.
live for 4±5 months with food (Wright & Chambers, 1994).
Bilgrami, A.L. (1997) Evaluation of the predation abilities of the
Mites generally do not show cannibalistic behaviour mite Hypoaspis calcuttaensis, predaceous on plant and soil
(Bilgrami, 1997) although under extreme conditions with- nematodes. Fundamental Applied Nematology, 20, 96±98.
out food for long periods, adults will cannibalise larvae and Burnham, K.P. & Anderson, D.R. (1998) Model Selection and
nymphs (Wright & Chambers, 1994). More research is Inference: a Practical Information-theoretic Approach. Springer-
needed to clarify the feeding behaviour of mites over time Verlag, New York.
(i.e. functional response) to address the possible implica- Christiansen, K. (1992) Springtails. Kansas School Naturalist, 39,
tions for regulation of their prey. 1±16.
In conclusion, springtail and mite population density Clutton-Brock, T.H., Major, M. & Guinness, F.E. (1985) Popula-
was determined primarily by endogenous factors (delayed tion regulation in male and female red deer. Journal of Animal
Ecology, 54, 831±846.
density-dependent regulation), which provided a reasonable
Eisenbeis, G. & Wichard, W. (1987) Atlas on the Biology of Soil
description of density changes without invoking top-down Arthropods. Springer-Verlag, Berlin.
predation. The predominant factor explaining springtail Environment Canada (2000) Canadian Climate Normals
and mite population growth was 1-week lagged density, [WWW document]. http://www.cmc.ec.gc.ca/climate/normals/
whereas including temperature in the model together with SASKS001.HTM
delayed density-dependent factors explained additional Erasmus, L.T. & Ryke, P.A.J. (1970) The soil mesofauna
variance in springtail and mite density. Regulation may occur associated with Eragrotis curvula. Pedobiologia, 10, 22±34.
by means of density-dependent processes acting within Ferguson, S.H. (2000) Predator size and distance to edge: is bigger
years whereas changes in year-to-year abundance may be better? Canadian Journal of Zoology, 78, 713±720.
due to differences in annual temperature and precipitation Ferguson, S.H. (2001) Changes in trophic abundance of soil
arthropods along a grass±shrub±forest gradient. Canadian
that may vary spatially with primary productivity (Knapp
Journal of Zoology, 79, 457±464.
& Smith, 2001). Future studies should use experimental Fowler, C.E. (1987) A review of density dependence in populations
approaches to understand the effects of delayed density of large mammals. Current Mammology, Vol. 1 (ed. by
dependence and rainfall on the demographic processes H. H. Genoways), pp. 401±441. Plenum, New York.
of survival, reproduction, immigration, and recruitment of HaÊgvar, S. (1995) Instability in small, isolated microarthropod
springtails and mites. communities. Bulletin of Entomology Pologne, 64, 123±133.

# 2002 The Royal Entomological Society, Ecological Entomology, 27, 565±573

 13652311, 2002, 5, Downloaded from https://resjournals.onlinelibrary.wiley.com/doi/10.1046/j.1365-2311.2002.00441.x by Cochrane Mexico, Wiley Online Library on [03/01/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Population dynamics of springtails and mites 573

Hairston, N.G., Smith, F.E. & Slobodkin, L.B. (1960) Community Petersen, H. & Luxton, M. (1982) A comparative analysis of soil
structure, population control, and competition. American fauna populations and their role in decomposition processes.
Naturalist, 94, 421±424. Oikos, 39, 287±388.
Hanski, I., Hansson, L. & Henttonen, H. (1991) Specialist Rosenzweig, M.L. (1968) Net primary productivity of terrestrial
predators, generalist predators, and the microtine rodent cycle. communities: prediction from climatological data. American
Journal of Animal Ecology, 60, 365±367. Naturalist, 102, 67±74.
Hopkin, S.P. (1997) Biology of Springtails (Insecta: Collembola). Royama, T. (1992) Analytical Population Dynamics. Chapman &
Oxford University Press, Oxford. Hall, London.
Jensen, P., Jacobson, G.L. & Willard, D.E. (1973) Effects of Schaefer, M. (1995) Interspecific interactions in the soil com-
mowing and raking on Collembola. Ecology, 54, 564±572. munity. Acta Zoologica Fennica, 196, 101±106.
Johnston, J.M. (2000) The contribution of microarthropods to Sequeira, R. & Dixon, A.F.G. (1997) Population dynamics of tree-
aboveground food webs: a review and model of belowground dwelling aphids: the importance of seasonality and time scale.
transfer in a coniferous forest. American Midland Naturalist, Ecology, 78, 2603±2610.
143, 226±238. Sinclair, A.R.E. (1989) Population regulation of animals. Ecolo-
Joosse, E.N.G. (1981) Ecological strategies and population gical Concepts (ed. by J. M. Cherrett), pp. 197±241. Blackwell
regulation of Collembola in heterogeneous environments. Science, Oxford.
Pedobiologia, 21, 346±356. Sinclair, A.R.E. & Pech, R.P. (1996) Density dependence,
Klironomos, J.N. & Kendrick, B. (1995) Relationships among stochasticity, compensation and predator regulation. Oikos,
microarthropods, fungi, and their environment. Plant Soil, 170, 75, 164±173.
183±197. Sokal, R. & Rohlf, F. (1981) Biometry. W. H. Freeman, New York.
Knapp, A.K., Briggs, J.M. & Koelliker, J.K. (2001) Frequency Swanson, B.J. (1998) Autocorrelated rates of change in animal
and extent of water limitation to primary production in a mesic populations and their relationship to precipitation. Conservation
temperate grassland. Ecosystems, 4, 19±28. Biology, 12, 801±808.
Knapp, A.K. & Smith, M.D. (2001) Variation among biomes in Swift, M.J., Heal, O.W. & Anderson, J.M. (1979) Decomposition in
temporal dynamics of aboveground primary production. Terrestrial Ecosystems. Blackwell, Oxford.
Science, 291, 481±484. Theenhaus, A., Scheu, S. & Schaefer, M. (1999) Contramensal
Laakso, J. & SetaÈlaÈ, H. (1999) Sensitivity of primary production to interactions between two collembolan species: effects on
changes in the architecture of belowground food webs. Oikos, population development and on soil processes. Functional
87, 57±64. Ecology, 13, 238±246.
Lasebikan, B.A., Badejo, M.A. & Prabhoo, N.R. (1985) The Turchin, P. (1990) Rarity of density dependence or population
collembolan fauna of two contrasting habitats at the University regulation with lags? Nature, 344, 660±663.
of Ife campus. Soil Fauna and Fertility (ed. by B. R. Striganova), Turchin, P. & Taylor, A.D. (1992) Complex dynamics in ecological
pp. 754±758. Proceedings of the 9th International Colloquium time series. Ecology, 73, 289±305.
on Soil Zoology, August 1985. Nauka, Moscow. Walker, M. & Jones, T.H. (2001) Relative roles of top-down and
Messier, F. (1991) The significance of limiting and regulating bottom-up forces in terrestrial tritrophic plant±insect herbivore±
factors on the demography of moose and white-tailed deer. natural enemy system. Oikos, 93, 177±187.
Journal of Animal Ecology, 60, 377±393. Wardle, D.A. (1999) How soil food webs make plants grow. Trends
Messier, F. (1996) On the functional and numerical responses of in Ecology and Evolution, 14, 418±420.
wolves to changing prey density. Ecology and Conservation of Wolters, V. (1991) Biological processes in two beech forest soils
Wolves in a Changing World (ed. by L. N. Carbyn, S. H. Fritts treated with simulated acid rain ± a laboratory experiment with
and D. R. Seip), pp. 187±197. University of Alberta Press, Isotoma tigrina (Insecta, Collembola). Soil Biology and Biochem-
Edmonton, Canada. istry, 23, 381±390.
Mikola, J. & SetaÈlaÈ, H. (1998) Productivity and trophic-level Wright, E.M. & Chambers, R.J. (1994) The biology of the
biomasses in a microbial-based soil food web. Oikos, 82, predatory mite Hypoaspic miles (Acari: Laelapidae), a potential
158±168. biological control agent of Bradysia paupera (Dipt.: Sciaridae).
Murray, B.G. (1982) On the meaning of density dependence. Entomophaga, 39, 225±235.
Oecologia, 53, 370±373. Yamamura, K. (1999) Transformation using (x ‡ 0.5) to stabilize
Oksanen, L. & Oksanen, T. (2000) The logic and realism of the the variance of populations. Researches in Population Ecology,
hypothesis of exploitation ecosystems. American Naturalist, 155, 41, 229±234.
703±723. Zheng, D.W., Bengtsson, J. & Agren, G.I. (1997) Soil food webs
Pace, M.L., Cole, J.J., Carpenter, S.R. & Kitchell, J.F. (1999) and ecosystem processes: decomposition in donor-control and
Trophic cascades revealed in diverse ecosystems. Trends in Lotka±Volterra systems. American Naturalist, 149, 125±148.
Ecology and Evolution, 14, 483±488. Zinkler, R.D. & Stecken, R. (1985) Digestion of leaf litter by the
Pech, R.P., Sinclair, A.R.E., Newsome, A.E. & Catling, P.C. (1992) springtail Tomocerus flavescens (Insecta, Collembola). Soil
Limits to predator regulation of rabbits in Australia: evidence Fauna and Soil Fertility (ed. by B. R. Striganova), pp. 730±734.
from predator-removal experiments. Oecologia, 89, 102±112. Nauka, Moscow.
Perry, J.N., Woiwod, I.P. & Hanski, I. (1993) Using response-
surface methodology to detect chaos in ecological time series.
Oikos, 68, 329±339. Accepted 13 January 2002

# 2002 The Royal Entomological Society, Ecological Entomology, 27, 565±573