do predators limit the abundance of alternative prey? experiments with vole-eating avian and...

13
Predator-induced changes in population structure and individual quality of Microtus voles: a large-scale field experiment Kai Norrdahl, Henrik Heinila ¨, Tero Klemola and Erkki Korpima ¨ki Norrdahl, K., Heinila ¨, H., Klemola, T. and Korpima ¨ki, E. 2004. Predator-induced changes in population structure and individual quality of Microtus voles: a large-scale field experiment. / Oikos 105: 312 /324. In small mammal populations with multiannual oscillations in density, observational data have revealed cyclic changes in population structure, reproduction,and individual quality, but mechanisms inducing these changes have remained an open question. We analysed data collected during a 3-year predator reduction experiment to find out the effects of predators on population structure, reproductive parameters, and individual quality of Microtus voles (the field vole M. agrestis and the sibling vole M. rossiaemeridionalis ) in western Finland. Voles were collected by snap trapping in April, June, August, and October during 1997 /1999. The yearly reduction of predators from April to October had a clear positive effect on the abundance of sibling voles but did not significantly affect the densities of field voles. Predator reduction apparently also affected the age ratio and mean body size in late summer, as well as pancreatic weights of voles. However, all observed differences between predator reduction and control areas, except those in abundance, were small and may mainly reflect a generally higher survival leading to higher densities of voles in predator reduction areas. Our results also indicated a relative lack of high quality food at population peaks but not because of reduced foraging activity in the presence of predators. We conclude that the indirect effects of vole-eating predators on the population growth of main prey are small compared to the detrimental direct effects on prey survival. In the case of less preferred prey, indirect effects of predation through reduced interspecific competition may play a role at high densities. K. Norrdahl, H. Heinila ¨ , T. Klemola and E. Korpima ¨ ki, Dept of Biology, Univ. of Turku, FIN-20014 Turku, Finland ([email protected]). Multiannual population cycles of small rodent popula- tions at northern latitudes are well-known (Norrdahl 1995, Korpima ¨ki and Krebs 1996, Stenseth 1999) but the cause of these population cycles is still a debated question. Most researchers agree that trophic interac- tions drive multiannual cycles of northern rodent populations, although the relative importance of rodent /predator and plant /rodent interactions may vary between species and geographic areas (Jedrzejewski and Jedrzejewska 1996, Korpima ¨ki and Krebs 1996, Oksanen and Oksanen 2000, Oksanen et al. 2000, Turchin et al. 2000, Ergon et al. 2001, Hanski et al. 2001, Turchin and Batzli 2001, Klemola et al. 2003). Recent observational, theoretical and experimental re- sults suggest that predation drives the multiannual cycles of vole populations, at least in northern Europe, whereas food shortage and/or social factors may prevent un- limited population growth in vole populations that have escaped predator control (Stenseth et al. 1996, Korpi- ma ¨ki and Norrdahl 1998, Klemola et al. 2000, 2003, Turchin et al. 2000, Korpima ¨ ki et al. 2002, Norrdahl and Korpima ¨ki 2002a,b, Norrdahl et al. 2002, Huitu et al. 2003). In northern Europe, the 3 /5-year population cycle is basically a cycle of voles belonging to genus Microtus (Henttonen 1987, Hanski and Henttonen 1996, Oksanen Accepted 17 September 2003 Copyright # OIKOS 2004 ISSN 0030-1299 OIKOS 105: 312 /324, 2004 312 OIKOS 105:2 (2004)

Upload: independent

Post on 26-Nov-2023

0 views

Category:

Documents


0 download

TRANSCRIPT

Predator-induced changes in population structure and individual

quality of Microtus voles: a large-scale field experiment

Kai Norrdahl, Henrik Heinila, Tero Klemola and Erkki Korpimaki

Norrdahl, K., Heinila, H., Klemola, T. and Korpimaki, E. 2004. Predator-inducedchanges in population structure and individual quality of Microtus voles: a large-scalefield experiment. �/ Oikos 105: 312�/324.

In small mammal populations with multiannual oscillations in density, observationaldata have revealed cyclic changes in population structure, reproduction, and individualquality, but mechanisms inducing these changes have remained an open question. Weanalysed data collected during a 3-year predator reduction experiment to find out theeffects of predators on population structure, reproductive parameters, and individualquality of Microtus voles (the field vole M. agrestis and the sibling vole M.rossiaemeridionalis ) in western Finland. Voles were collected by snap trapping inApril, June, August, and October during 1997�/1999. The yearly reduction of predatorsfrom April to October had a clear positive effect on the abundance of sibling voles butdid not significantly affect the densities of field voles. Predator reduction apparentlyalso affected the age ratio and mean body size in late summer, as well as pancreaticweights of voles. However, all observed differences between predator reduction andcontrol areas, except those in abundance, were small and may mainly reflect a generallyhigher survival leading to higher densities of voles in predator reduction areas. Ourresults also indicated a relative lack of high quality food at population peaks but notbecause of reduced foraging activity in the presence of predators. We conclude that theindirect effects of vole-eating predators on the population growth of main prey aresmall compared to the detrimental direct effects on prey survival. In the case of lesspreferred prey, indirect effects of predation through reduced interspecific competitionmay play a role at high densities.

K. Norrdahl, H. Heinila, T. Klemola and E. Korpimaki, Dept of Biology, Univ. of Turku,FIN-20014 Turku, Finland ([email protected]).

Multiannual population cycles of small rodent popula-

tions at northern latitudes are well-known (Norrdahl

1995, Korpimaki and Krebs 1996, Stenseth 1999) but the

cause of these population cycles is still a debated

question. Most researchers agree that trophic interac-

tions drive multiannual cycles of northern rodent

populations, although the relative importance of

rodent�/predator and plant�/rodent interactions may

vary between species and geographic areas (Jedrzejewski

and Jedrzejewska 1996, Korpimaki and Krebs 1996,

Oksanen and Oksanen 2000, Oksanen et al. 2000,

Turchin et al. 2000, Ergon et al. 2001, Hanski et al.

2001, Turchin and Batzli 2001, Klemola et al. 2003).

Recent observational, theoretical and experimental re-

sults suggest that predation drives the multiannual cycles

of vole populations, at least in northern Europe, whereas

food shortage and/or social factors may prevent un-

limited population growth in vole populations that have

escaped predator control (Stenseth et al. 1996, Korpi-

maki and Norrdahl 1998, Klemola et al. 2000, 2003,

Turchin et al. 2000, Korpimaki et al. 2002, Norrdahl and

Korpimaki 2002a,b, Norrdahl et al. 2002, Huitu et al.

2003).

In northern Europe, the 3�/5-year population cycle is

basically a cycle of voles belonging to genus Microtus

(Henttonen 1987, Hanski and Henttonen 1996, Oksanen

Accepted 17 September 2003

Copyright # OIKOS 2004ISSN 0030-1299

OIKOS 105: 312�/324, 2004

312 OIKOS 105:2 (2004)

et al. 2000, Hanski et al. 2001, Norrdahl and Korpimaki

2002c), although the central role of Microtus voles may

be locally replaced by grey-sided voles (Clethrionomys

rufocanus ) in Lapland (Hansen et al. 1999a,b). Less

violent population oscillations in other sympatric small

mammals appear to be a mere reflection of the Microtus

vole cycle: varying predation pressure and possibly also

interspecific competition connect the population dy-

namics of sympatric species to the dominant Microtus

cycle (Hanski and Henttonen 1996, Norrdahl and

Korpimaki 2000a, Oksanen et al. 2000).

In addition to oscillations in population size, observa-

tional data have revealed significant cycle phase-related

changes in population structure, reproduction, and

individual quality (Krebs and Myers 1974, Norrdahl

and Korpimaki 2002a,b), but the role of predators as a

mechanism causing these phase-related changes has

remained an open question (Boonstra et al. 1998).

Theoretical and observational studies as well as small-

scale laboratory experiments have indicated that pre-

dators might affect the individual quality and reproduc-

tive output of prey by two mechanisms: (1) by selective

predation with respect to sex, age, or social status, or (2)

by triggering costly anti-predator adaptations in prey

(Ylonen 1994, Norrdahl and Korpimaki 1995a, 2000b,

Oksanen and Lundberg 1995). For example, a reduction

in foraging activity as a response to high predation risk

has been a common observation in previous studies

(Koskela and Ylonen 1995, Borowski 1998, Norrdahl

and Korpimaki 1998, Barreto and MacDonald 1999). If

the amount of high quality food is limited, reduced

foraging activity and/or redirection of foraging activity

to safer microhabitats where food is already depleted

may lead to lower body growth rates, delayed matura-

tion, deteriorated body condition, and lower immuno-

competence (Norrdahl and Korpimaki 2000b). Yet,

short-term experiments in the decline phase of the vole

cycle (Klemola et al. 1997b), or with fenced voles or

predators (Mappes et al. 1998, Jonsson et al. 2000,

Hellstedt et al. 2002), have concluded that selective

predation is a more likely explanation for the observed

patterns than behavioural changes in prey. However, the

conclusions made in these previous papers have suffered

from a lack of replicated experimental studies made in

proper spatial and temporal scale. Therefore, we ana-

lysed data collected during a 3-year predator reduction

experiment in western Finland. In this paper, we aim to

focus on changes in population structure, reproductive

parameters, and individual quality of Microtus voles (the

field vole M. agrestis and the sibling vole M. rossiaemer-

idionalis ). We also analysed species-level changes in the

relative abundance of the two Microtus species; the

impact of predator removal on the pooled abundance of

Microtus voles and on other voles species has been

published elsewhere (Korpimaki et al. 2002).

We made several predictions based on previous

research. If predators are driving the cycle phase-related

changes in population structure, reproduction, and

individual quality (see above), we predict that the

removal of predators should cause major changes in

the characteristics of vole populations. If the main

mechanism behind these changes is selective predation,

we predict marked changes in population structure but

not necessary in individual quality. If the main mechan-

ism would be anti-predator adaptations in prey, then we

predict marked changes in individual quality and/or

reproductive parameters but not necessary in sex ratio or

age structure. If the indirect effects of predators on prey

populations are food-mediated (altered foraging beha-

viour), we predict that the removal of predators should

improve the body condition of voles and affect the size

of internal organs reflecting the quality of forage and/or

the level of immunocompetence. As earlier studies of

herbivorous mammals have shown that liver and pan-

creas may respond to low-quality forage by hypertrophy

(enlargement) or hypotrophy (atrophy) (Smith et al.

1980, Bergeron and Jodoin 1982, Harju and Tahvanai-

nen 1994, Seldal et al. 1994, Harju 1996, Mora et al.

1996), we used these organs as bioassays of forage

quality (altered foraging behaviour). We included spleen

as the third internal organ measured, as hypotrophy of

the spleen may reflect low immunocompetence due to

the role of the spleen in lymphocyte production and

storage, whereas hypertrophy of the spleen may reflect a

prevailing infection by some bacteria or parasites (Chitty

and Phipps 1960, Tenora et al. 1979, Møller et al. 1998,

Møller and Erritzoe 2000). A previous study has shown

that passerine birds with a small spleen have an increased

risk to be depredated by cats (Møller and Erritzoe 2000),

but it is not known whether the same is true for small

mammals.

Methods

Study areas and predator reduction

The study was carried out between 1997 and 1999 in the

vicinity of Kauhava and Lapua, western Finland (638N,

238E). The study area consists mainly of agricultural

fields where small mustelids (the least weasel Mustela

nivalis and the stoat M. erminea ) and birds of prey (the

Eurasian kestrel Falco tinnunculus, the short-eared owl

Asio flammeus, the long-eared owl A. otus and Teng-

malm’s owl Aegolius funereus ) are the main predators of

voles (field vole, sibling vole, bank vole Clethrionomys

glareolus, and water vole Arvicola terrestris ) (Korpimaki

and Norrdahl 1991, Korpimaki et al. 1991, Norrdahl

and Korpimaki 1995c). The two Microtus voles are the

main prey of the above predators, whereas bank and

water voles are alternative prey items (Korpimaki and

Norrdahl 1991, Korpimaki et al. 1991). Multiannual

OIKOS 105:2 (2004) 313

cyclic fluctuations are most evident in Microtus voles

(Norrdahl and Korpimaki 2002c); therefore, we focused

on these two species.

We used four manipulation-control pairs of agricul-

tural fields (each 2.5�/3 km2) for the predator reduction

experiment. All experimental areas were within 180 km2

(12�/15 km). Each manipulation-control pair was as

similar as possible with respect to habitats inside and

surrounding the area. As the distance between all

manipulation and control areas was at least 5 km

(median of 6 km, range 5�/14 km), small mustelids and

breeding avian predators did not disperse from control

to manipulation areas during the yearly experimental

periods of six to seven summer months (Norrdahl and

Korpimaki 1996, Korpimaki and Norrdahl 1998). How-

ever, predators were free to move in and out of

experimental areas. We reduced densities of predators

over a 3-year period (1997�/1999) when vole populations

increased from a population low to a population peak

(Fig. 1). In the preceding year (1996), densities of vole

populations were moderate and declined to low numbers

towards the spring of 1997.

We removed stoats and least weasels by live-trapping

from April/May to October each year (for a detailed

description of the trapping procedure, Korpimaki et al.

2002). We removed 11�/32 mustelids from each area (2�/

11 in 1997, 1�/6 in 1998, and 4�/15 in 1999). Trapped

mustelids were transferred and released at least 30 km

from our study sites. The efficiency of the predator

reduction was confirmed by snow tracking in spring

(from late February to March, before predator reduc-

tion) and autumn (from late November to December,

after predator reduction). Six lines per area, each ca 1

km, were skied after a snowfall, so that tracks made by

animals only during the previous one or two nights were

visible. Identification of small mustelids was based on

track dimensions (Korpimaki et al. 1991). According to

snow-tracking, density indices (number of individuals

crossing track lines/km) did not differ between manip-

ulation and control areas in each spring but were

significantly lower in predator reduction areas in each

autumn (Korpimaki et al. 2002). This indicates that the

predator reduction was successful but that an influx of

new individuals to the reduction areas during winter

levelled differences in predator densities before next

spring.

In addition to removing mustelids, we reduced the

number of breeding avian predators by removing all

potential breeding sites (stick nests, natural cavities and

nest boxes) from the reduction areas in each spring of

1997�/1999. This procedure significantly reduced the

number of breeding territories of main vole-eating avian

predators in the manipulation areas relative to the

control areas: the yearly mean number of territories

varied between 0.2�/2 in reduction areas and 1.2�/3.6 in

control areas (Korpimaki et al. 2002).

Collection and examination of small mammals

To estimate vole abundance in the eight experimental

areas, a random sub-set of eight ditch lines and two

forest lines were chosen for each trapping session from

numbered ditches in agricultural fields and from num-

bered forest plots (i.e. the short line method; Norrdahl

and Korpimaki 1995b, Korpimaki and Norrdahl 1998).

Because one ditch or forest plot was only used once per

year, possible impacts of snap trapping on vole popula-

tions remained negligible and were relatively similar in

Fig. 1. The density index of (a) sibling voles and (b) field voles,and (c) the percentage of field voles among Microtus voles in thepredator reduction (dots) and control areas (squares) during1997�/1999. Values present the mean (9/SE) number of indivi-duals captured per trap line per area. Notes: trap indices werecollected in April, late June, August, and October each year.Shaded areas refer to snowy periods.

314 OIKOS 105:2 (2004)

the reduction and control areas. Ten Finnish metal

mouse snap traps (suitable for Microtus and bank voles)

and one Finnish metal rat snap trap (suitable also for

water voles) were set 10 m apart for two days in each

selected line and were checked once a day. Snap trapping

was performed simultaneously in each manipulation-

control area pair from late March to early April (before

manipulation), and again in late June, August and

October each year in 1997�/1999.

In 1997 and 1998, all trapped Microtus voles were

picked for analyses, whereas in the peak year 1999 we

selected either all Microtus voles (if the number captured

was less than 5 per species), or a random sample of at

least 5 field voles and 5 sibling voles from each trapping

occasion to the analysis. These animals were weighed

with a laboratory scale to the nearest 0.1 g, sexed, and

their reproductive status was checked using the mass of

testes, or the width and condition of uterus together with

signs of wear in nipples as a criterion for classification

(four categories: immature, mature, pregnant, lactating).

The number of embryos or, if the animal was not visibly

pregnant but had recently given birth, the number of

fresh scars in uterus were used as an index of litter size.

The body length of the animals (from snout to vent) was

measured to the nearest 1 mm. Animals were classified

as either overwintered (animals that had experienced the

previous winter) or young born in the year (hereafter:

young) by using information obtained from their pelage

(moulting pattern together with hair length and general

wear of the animal). In addition, we measured the mass

of the liver, pancreas and spleen to the nearest 1 mg

(spleen only in 1998�/1999). From autumn 1998 on-

wards, we also removed all fat (white adipose tissue)

situated subcutaneously around left thigh, and weighed

it to the nearest 1 mg.

We calculated the current increase rate of the popula-

tion for a one-month time interval using the formula:

rt�(1=T) ln (Nt�T=Nt)

where Nt is the catch index for a trapping occasion, and

Nt�T is the catch index for the following trapping

occasion in the same area, and T is the time (in months)

elapsed between the trapping occasions. To avoid divi-

sion by zero, zero values were replaced with a value

corresponding to a catch of 0.5 individuals in the

trapping occasion.

We tested the effect of predator reduction on popula-

tion structure (proportion of males, proportion of over-

wintered individuals, proportion of mature individuals,

proportion of pregnant or lactating females), litter size,

body size, body condition indices, and the relative mass

of organs with the analysis of variance (GLM procedure

of SAS statistical software, version 8.01). In these

analyses, the mean value for a trapping occasion (a 2-

night long trapping period in a 3-km2 area) was used as

an independent replicate, since trapped individuals were

removed from the population and no trapping site (i.e.

any given ditch or forest site) was used more than once a

year. Proportions were arcsine-transformed, and all

variables were weighed with sample size. As an index

of body condition we used the residuals of non-linear

regression (polynomial function) between log-trans-

formed body mass and body length (Norrdahl and

Korpimaki 2002a). Relative fat weight (residuals of a

linear regression between log-transformed fat weight and

body length) was used as a second index of body

condition. As the mass of internal organs is related to

body mass (Klemola et al. 1997a), we calculated relative

organ (liver, pancreas, spleen) mass as a residual of a

linear regression between log-transformed organ mass

and body mass. All residuals were calculated separately

for both genders and species.

The effect of the predator reduction on vole abun-

dance was tested with repeated measures ANOVA with

year and trapping time within year as repeated factors.

This statistical model views the experiment as a series of

three consecutive 6�/7-month long removal experiments

within the same areas. An alternative approach would

have been to view the experiment as a continuous 3-year

long experiment despite the fact that we removed

predators only from April to October. Although we

used the first model (year and month as repeated

factors) in our statistical analyses, we also present the

main results of the alternative model (time since the

start of the experiment as the repeated factor) to show

that the main results were not sensitive to model

selection. Direct density-dependence in the population

increase rate was estimated using linear regression

analyses (REG procedure of SAS), where the local

population increase rate was used as the response

variable and the local abundance of the species studied,

or the pooled abundance of all Microtus voles in the

beginning of the season was used as the explanatory

variables.

As we analysed several characteristics of vole popula-

tions, we had to make multiple statistical tests, which

may lead to an elevated risk of type I errors (false

rejection of the null hypothesis). However, as our

interpretations are based on the combination of results

from different statistical analyses rather than on results

of a single statistical test, an elevated risk of type II

errors can be regarded as an equally poor option than a

slightly elevated risk of type I errors. Experiment-wise

correction of P-values, such as in the standard Bonfer-

roni correction, lead to a high risk of type II errors

(Chandler 1995). ‘‘Family-wide’’ or ‘‘table-wide’’ correc-

tion of P-values reduces the probability of type II errors,

but the decision of what is a correct ‘‘family’’ of

statistical tests is not unambiguous (Rice 1989, Chandler

1995). The correct level of grouping of statistical tests

largely depends on the acceptable level of type I and type

II errors. In this study, we regarded type II errors as an

OIKOS 105:2 (2004) 315

equally serious problem as type I errors. Therefore, we

used ‘‘family-wide’’ Bonferroni corrections to adjust the

level of significance a with an intermediate level of

grouping of statistical tests. For example, analyses

testing the null hypothesis that reproductive parameters

(proportion of mature, proportion of pregnant or

lactating, and litter size) do not change as a result of

predator reduction formed one ‘‘family’’, and the

analyses of density-dependence in the reproductive

parameters another ‘‘family’’. This choice should in-

crease the probability that we captured biologically

important indirect effects of predators on prey popula-

tions without an unacceptable high risk of false rejection

of null hypotheses.

Results

Vole abundance

Predator reduction increased the abundance of sibling

voles but did not have an obvious effect on the

abundance of field voles [repeated measures ANOVA

with a�/0.025: field vole (treatment: F1,6�/0.23, P�/

0.65, treatment by year: F2,12�/0.24, P�/0.79, treatment

by month: F3,18�/1.83, P�/0.18), and sibling vole

(treatment: F1,6�/19.65, P�/0.004, treatment by year:

F2,12�/26.29, PB/0.001, treatment by month: F3,18�/

3.20, P�/0.048); Fig. 1)]. This result was not sensitive

to model selection in the statistical analysis: when the

experiment was analysed as a continuous 3-year long

experiment (repeated factor time instead of year and

month), the difference between the species became even

more evident [field vole (treatment: F1,6�/0.23, P�/0.65,

treatment by time: F11,66�/0.88, P�/0.57), and sibling

vole (treatment: F1,6�/19.65, P�/0.004, treatment by

time: F11,66�/4.53, PB/0.0001)].

The proportion of field voles among Microtus voles

appeared to be generally lower in the predator reduction

areas (treatment: F1,86�/5.57, P�/0.02), although both

in the experimental and control areas the proportion of

field voles among Microtus voles followed the same

general pattern: the proportion of field voles increased

during the low phase of the cycle, peaked at the end of

the low phase, and thereafter declined (Fig. 1c). Accord-

ingly, there was a clear negative relationship between the

abundance of Microtus voles and the proportion of field

voles (r�/�/0.45, PB/0.001, N�/88). Within months,

the relationship appeared to be negative from June to

October, although significantly so in June and October

only [June (r�/�/0.50, P�/0.012, a�/0.0125, N�/25),

August (r�/�/0.42, P�/0.06, N�/20), and October (r�/

�/0.58, P�/0.002, N�/25)], but not in April (r�/0.09,

P�/0.72, N�/18).

Although predator reduction did not have a clear

effect on the abundance of field voles, the growth rate of

field vole populations appeared to be higher in predator

reduction areas than in control areas during summer

1997 and 1998 (Fig. 2b). The same was not evident in the

peak year 1999. During most periods studied, spatial

variation in the population increase rates of voles

appeared to be negatively related to the local density

of voles (Table 1). However, in the peak phase (summer

and autumn 1999) the population increase rates of field

voles could not be explained by the abundance of the

same species, or the pooled abundance of voles (Table 1).

The spatial variation in the population increase rates

appeared to be highest at the population peak in

predator reduction areas (Fig. 2c).

Fig. 2. The mean (9/SE) population increase rate of (a) siblingvoles and (b) field voles, and (c) the coefficient of variation inthe population increase rates of Microtus voles in the predatorreduction (dots) and control areas (squares) during 1997�/1999.Population increase rate refer to the monthly populationincrease rate per area, measured from the change in the numberof trapped voles from one trapping occasion to the next one inthe same area. Notes as in Fig. 1.

316 OIKOS 105:2 (2004)

Sex ratio and age structure

The sex ratio appeared to be slightly more male biased in

control than in predator reduction areas, but this

difference was not statistically significant (treatment:

F1,100�/4.52, P�/0.036, a�/0.025). There were no

obvious differences between species in this respect

(interaction treatment by Species: F1,100�/1.7, P�/

0.19, Fig. 3a, b). Neither was there a relationship

between sex ratio and vole abundance [field vole (r�/

�/0.03, P�/0.80, N�/89), and sibling vole (r�/�/0.03,

P�/0.81, N�/75)].

There was no general difference in the proportion of

overwintered voles between predator reduction and

control areas (treatment: F1,131B/0.01, P�/0.99), but

the interaction between treatment and species (F1,131�/

5.78, P�/0.018, a�/0.025) was significant. The signifi-

cant interaction between treatment and species appeared

to be mainly due to interspecific differences in June (Fig.

3c, d). In August 1999, predator reduction areas had a

higher proportion of overwintered voles than control

areas (treatment: F1,11�/24.94, PB/0.001, a�/0.0125; no

overwintered voles were captured in August 1997 and

1998). In April and October, there were no significant

Table 1. Regression analyses of spatial variation in the population increase rate (per month) in relation to the abundance of thesame vole species, or the pooled abundance of both Microtus species during 1997�/1999. Of the two explanatory variables, thevariable giving a higher coefficient of determinant (�/best) was chosen to the final model. The population increase rate wasmeasured from April to June (�/spring), June to August (�/summer), August to October (�/autumn), and October to April(�/winter). Df�/1,6 in all cases.

Year Season Field voles Sibling voles

r2 P Best r2 P Best

1997 Spring 0.67 0.013 Pooled 0.61 0.023 OwnSummer 0.84 0.0012 Pooled 0.86 0.0009 PooledAutumn 0.20 0.27 Own 0.23 0.22 OwnWinter 0.61 0.022 Own 0.77 0.004 Own

1998 Spring 0.19 0.28 Pooled 0.50 0.049 OwnSummer 0.73 0.007 Own 0.56 0.034 OwnAutumn 0.86 0.0009 Pooled 0.12 0.40 OwnWinter 0.56 0.032 Pooled 0.48 0.06 Own

1999 Spring 0.60 0.024 Own 0.28 0.18 PooledSummer 0.09 0.48 Own 0.48 0.06 OwnAutumn 0.24 0.22 Own 0.55 0.034 Own

Fig. 3. The mean (9/SE)proportion of (a, b) males, and(c, d) overwintered individualsin predator reduction (dots) andcontrol areas (squares) during1997�/1999. Left panels presentdata for field voles and rightpanels for sibling voles. Notes asin Fig. 1.

OIKOS 105:2 (2004) 317

differences between treatment and control areas in this

respect.

Reproduction

We did not observe significant differences in the

proportion of mature voles or the proportion of

pregnant or lactating voles between the predator reduc-

tion and control areas (for treatment and its interactions,

all P�/0.034; a�/0.0167; Fig. 4). Also differences in the

litter size between the predator reduction and control

areas remained non-significant.

The proportion of mature voles was negatively related

to vole abundance [field vole (r�/�/0.33, P�/0.002, a�/

0.0125, N�/89), and sibling vole (r�/�/0.33, P�/0.004,

N�/75)]. The litter size of field voles appeared to be

negatively related to vole density, but this relationship

was not statistically significant (r�/�/0.36, P�/0.028, a�/0.0125, N�/37). The litter size of sibling voles was not

related to density (r�/�/0.01, P�/0.98, N�/39).

Body size and condition

The mean body size (body mass and body length) of field

voles was smaller in predator reduction areas than in

control areas in all trapping periods except April and

October 1997 (Fig. 5a, c). In sibling voles, the differences

were less obvious (Fig. 5b, d). Predator reduction had a

nearly significant negative effect on the mean body size

of Microtus voles in late summer [in August, body mass

(treatment: F1,37�/6.72, P�/0.014, a�/0.0125), and

body length (treatment: F1,37�/5.43, P�/0.025)],

whereas in April or October the differences between

the manipulation and control areas were small and

statistically non-significant.

Predator reduction did not have an obvious impact on

the body condition index (residual of polynomial

regression between body mass and body length) of

Microtus voles (Fig. 6a, b). Also the differences in the

fat index between the manipulation and control areas

remained non-significant (all P�/0.05; Fig. 6c, d).

The relationship between the body condition index

and vole abundance was negative in field voles (r�/

�/0.30, P�/0.004, a�/0.0125, N�/89) but not in sibling

voles (r�/�/0.09, P�/0.42, N�/75), whereas the rela-

tionship between the fat index and vole abundance

remained non-significant in both species [field vole

(r�/�/0.18, P�/0.17, N�/60), and sibling vole (r�/

�/0.07, P�/0.65, N�/46)]. However, the relationship

between the body condition index and vole abundance

varied between months. The negative relationship was

strongest in August (significant in both species) whereas

the same relationship appeared to be positive in April

(nearly significantly so in sibling voles: r�/0.60, P�/

0.006, a�/0.0031, N�/19).

Sizes of internal organs

There was a nearly significant interaction between

treatment and year in the relative mass of the pancreas

(F2,132�/3.66, P�/0.029, a�/0.0167). From 1997 to

Fig. 4. The mean (9/SE)proportion of (a, b) matureindividuals among all voles, and(c, d) the proportion of pregnantor lactating females among allfemales in predator reduction(dots) and control areas (squares)during 1997�/1999. Left panelspresent data for field voles andright panels for sibling voles.Notes as in Fig. 1.

318 OIKOS 105:2 (2004)

autumn 1998, predator reduction appeared to have little

impact on the relative pancreatic weights, whereas in

spring and summer 1999, field voles captured from

predator reduction areas appeared to have larger pan-

creas than those captured from control areas (Fig. 7a, b).

The relative mass of the liver and spleen did not differ

significantly between the predator reduction and control

areas (all P�/0.1; Fig. 7c�/f), although field voles

appeared to have relatively smaller spleens in predator

reduction areas during summer 1999 (Fig. 7e). The

relative mass of the liver declined towards the population

peak in autumn 1999 in both predator reduction and

control areas (Fig. 7c, d). The relationship between vole

abundance and the relative mass of liver was clearly

negative [field vole (r�/�/0.57, PB/0. 0001, a�/0.0071,

N�/84), and sibling vole (r�/�/0.33, P�/0.004, N�/

Fig. 5. The mean (9/SE) bodysize of field voles (left panels) andsibling voles (right panels) inpredator reduction (dots) andcontrol areas (squares) during1997�/1999. Body size wasmeasured by (a, b) body massand (c, d) body length. Notes asin Fig. 1.

Fig. 6. The mean (9/SE) bodycondition of field voles (leftpanels) and sibling voles (rightpanels) in predator reduction(dots) and control areas (squares)during 1997�/1999. Bodycondition was estimated by (a, b)the residual of the non-linearregression of log-transformedbody mass and body length, and(c, d) the residual of the linearregression of log-transformed fatweight and body mass. Fat weightrefers to the weight of whiteadipose tissue situatedsubcutaneously around left thigh.Notes as in Fig. 1.

OIKOS 105:2 (2004) 319

72)]. The relative masses of the pancreas and spleen were

not significantly related to vole abundance (all P�/0.05).

The relative masses of internal organs were not corre-

lated with each other, with the possible exception of liver

and spleen in sibling voles (r�/0.36, P�/0.009, N�/51;

all other P�/0.05).

Discussion

Predator reduction: strong response in prey quantity,

weak response in prey quality

The reduction of predators had a significant or nearly

significant effect on several of the population level

parameters studied: abundance, age ratio in late summer,

mean body size in late summer, and mean mass of the

pancreas. However, all observed differences between

predator reduction and control areas, except those in

abundance, were small and may mainly reflect a

generally higher survival of voles leading to higher

population densities in predator reduction areas.

The reduction of predators increased vole densities

(Korpimaki et al. 2002). However, the strength of this

positive response varied between species. Our results

show that the response was clear in sibling voles but non-

significant in field voles. In field voles, the predator

reduction only had a transient effect on population

growth during summers 1997 and 1998. There are at

least two alternative but not mutually exclusive explana-

tions for this result: (1) selective predation, and (2)

interspecific competition between Microtus voles. We

only removed small mustelids from the predator reduc-

tion areas, whereas the numbers of avian predators

where reduced by removing potential nest sites. If small

mustelids hunted selectively sibling voles, and the

numbers of hunting avian predators were effectively

reduced during the nesting period only, differences in

the predation pressure on field voles between the

predator reduction and control areas may have been

small outside the nesting period of birds. Interspecific

competition between sibling and field voles combined

with density-dependence in the reproduction of field

voles may have also levelled differences in field vole

Fig. 7. The relative mass of (a,b) pancreas, (c, d) liver, and (e, f)spleen of field voles (left panels)and sibling voles (right panels)in predator reduction (dots) andcontrol areas (squares) during1997�/1999. Figures present themean (9/SE) mass per area,measured as percentage of bodymass. Notes as in Fig. 1.

320 OIKOS 105:2 (2004)

densities between predator reduction and control areas.

Norrdahl and Korpimaki (1993) suggested that sibling

voles are competitively superior but more vulnerable to

predation than field voles because of the more aggre-

gated dispersion of the former species. The clumped

dispersion of sibling voles appears to be an advantage in

interspecific competition for space, but generally higher

vole densities in patches of sibling voles than in patches

of field voles attract patch-searching predators (Norr-

dahl and Korpimaki 1993). In addition, reproductive

output in late summer and autumn is sensitive to high

vole density in field voles: the proportion of mature

individuals and possibly also the litter size of field voles

were density-dependent. A prediction based on the

hypothesis of Norrdahl and Korpimaki (1993) was that

a removal of predators should lead to a rise in the

proportion of sibling voles, which proved to be true in

the present results. Accordingly, we propose that the lack

of a lasting positive response to predator reduction in

field voles is a consequence of selective predation

combined with density-dependence in the reproductive

output of field voles. Interspecific competition may have

also played a role in year 1999, when the mean number

of Microtus voles captured per 100 m of ditch (excluding

forest plots) ranged from 1.3 to 3.8.

The population increase rates of field voles were not

related to vole density at the population peak. One

possible explanation for this result is high spatial

variation in the occurrence of predators: spatial varia-

tion in predator-induced mortality may have masked

underlying density-dependence in population growth.

Although predator reduction areas had less predators

than control areas, predators were present in all areas at

least sporadically. The occurrence of predators was more

sporadic in predator reduction areas than in control

areas because of monthly removal trapping of small

mustelid predators and a paucity of suitable nest sites for

avian predators in the predator reduction areas.

In our study system, sibling vole appears to be the key

species in predator-prey interactions. However, the

geographic range of sibling voles is limited and hence

it cannot be the key species in most northern ecosystems

with cyclic vole populations. Nonetheless, our results

give rise to a more general conclusion by indicating that

even in multispecies assemblages vole-eating predators

mainly concentrate on a single prey species. The key

species should be the species which gives most energy for

patch-searching predators without risking the life of the

predator (as also suggested by conventional models of

optimal diet theory; Pyke 1984, Stephens and Krebs

1986). In northern Europe, Microtus voles and grey-

sided voles have been shown to be the most important

prey for vole-eating predators (Korpimaki 1988, 1992,

Korpimaki and Norrdahl 1991, Korpimaki et al. 1991,

Reif et al. 2001). Thus, in northern Europe the key

species is most likely one of these species. The abundance

of other similar-sized or smaller species in the assem-

blage may be determined as much by interspecific

competition as by predation. A prediction based on

this conclusion is that when the (local) key species is

removed, the profitability of other species from the

viewpoint of predators should determine which species

becomes the new key species, and whether the predator-

prey cycles will continue. If �/ and only if �/ the new key

species is abundant enough to sustain predator popula-

tions, predator-prey cycles may continue almost un-

changed although the former key species is absent.

Population structure and reproduction

The mean body size of voles, especially of field voles, was

smaller in the predator reduction areas than in the

control areas, which probably reflects a higher survival

of young individuals in the absence of predators. As a

consequence, there were more small-sized young indivi-

duals in the predator reduction areas. This interpretation

is supported by the fact that the differences in the mean

body size were non-significant during periods when the

population practically consists of a single age class

(overwintered in April and young in October). We

measured the age structure only on a crude level (over-

wintered vs young of the year) and cannot therefore

analyse the exact mean ages of the voles. However, it is

obvious that the strong positive response in vole

abundances to predator reduction was due to an increase

in the number of young voles. Although the proportion

of overwintered voles was higher in predator reduction

areas than in control areas in August 1999 suggesting

higher survival rates of overwintered voles in predator

reduction areas, practically all overwintered voles dis-

appeared before autumn. Previous research has also

shown that almost all overwintered voles die between

April and August (Norrdahl and Korpimaki 2002b),

mainly because of predation (Norrdahl and Korpimaki

1995c). The removal of weasels was a slow process, and

also the predator reduction areas had some hunting

predators especially from spring to summer. Thus, our

predator reduction procedure appeared to be too slow to

prevent the killing of most overwintered individuals,

particularly when vole densities were low (in years 1997

and 1998).

Despite the fact that vole densities were higher in

predator reduction than in control areas, we did not

observe significant differences in reproductive para-

meters between the areas. In summer, all overwintered

voles are reproductively active in our study area (Norr-

dahl and Korpimaki 2002b), and hence the proportion

of mature individuals or the proportion of pregnant or

lactating females reflect changes in the proportion of

reproductively active young voles. Apparently the pre-

dation pressures on mature and immature individuals

OIKOS 105:2 (2004) 321

did not deviate enough to cause observable changes in

the proportion of mature individuals. On the other hand,

our results also suggest that density-dependent changes

in the maturation of young voles played a minor role in

the density dynamics of the populations studied. Even

when differences in population densities were largest

(August to October 1999), the proportion of mature

individuals was fairly similar between the predator

reduction and control areas.

Individual quality

Predators appeared to have a minor impact on the body

condition of voles. Body condition index of field voles

was negatively related to vole abundance but at the

population peak in year 1999 the body condition indices

were very similar in predator reduction and control areas

despite differences in vole densities. In October, voles

captured from predator reduction areas appeared to be

in similar or better body condition than voles captured

from control areas despite higher population densities in

predator reduction areas.

Differences in the mass of internal organs between the

predator reduction and control areas were mainly small,

but indicated a relative lack of high quality food at high

population densities. In herbivorous mammals, a light

liver may reflect low protein gain (Harju 1996, Mora et

al. 1996), and an increase in pancreatic weight in small

rodents has been associated with high levels of protei-

nase inhibitors in food (Seldal et al. 1994). At high

population densities in 1999, the relative mass of the

liver was small but there were no differences between the

predator reduction and control areas in this respect. The

relative mass of the pancreas appeared to be larger in

predator reduction than in control areas in spring and

summer 1999, but even this difference was very small in

sibling voles. In predator reduction areas, pancreatic

weights appeared to be highest in spring 1999, while in

August and in October values were very similar between

years 1998 and 1999. As relative pancreatic weights were

similar in rapidly increasing (August 1998) and peak

(August 1999) populations, it is unlikely that the

differences in population increase rates between treat-

ments or between years would have been caused by

differences in the condition of the pancreas. Rather, this

difference may be interpreted as an epiphenomenon of

higher population densities and hence more intense

competition for high quality food in predator reduction

than control areas. Inter-areal differences in the relative

mass of the spleen were also non-significant, although

spleens of field voles appeared to be slightly smaller in

predator reduction areas. Overall, our results were in

disagreement with the predictions based on the

hypothesis that reduced foraging activity at high preda-

tion risk would have population-level consequences on

the quality of voles. Although our results indicate a

relative lack of high quality food at population peaks, it

should be noted that our data do not give support to the

hypothesis of Seldal et al. (1994) suggesting that a

grazing-induced action of trypsin inhibitors in food

plants is a cause for cyclic dynamics of mammal

populations.

Conclusions

Our results indicate that a reduction of vole-eating

predators has a strong impact on the densities of the

main prey (in our study system, sibling vole) but only a

minor impact on the structure of prey populations or the

quality of prey. The small changes in the structure of

prey populations, or quality of prey may be mainly

explained as a consequence of increased population

density (presumably higher survival of especially young

individuals). Density-dependence in reproduction may

partly compensate for changes in predator-caused mor-

tality, but the magnitude of these density-dependent

changes appears to be small compared to changes in

survival. Our results were in disagreement with the

predictions based on the hypothesis that behavioural

changes in prey as a response to high predation risk

might explain the cycle phase-related changes in repro-

ductive output and individual quality. Our results also

suggest that selective predation may operate more

strongly between prey species than between different

categories of adult voles. Thus, we conclude that the

indirect effects of vole-eating predators on the popula-

tion growth of main prey are small compared to the

detrimental direct effects on survival of prey. In the case

of less preferred prey, indirect effects of predation may

play a role through reduced interspecific competition at

high densities.

Acknowledgements �/ We thank O. Hemminki, S. Ikola and J.Koivisto for great help with the field work and Otso Huitu forcomments on the manuscript. The study was financiallysupported by the Academy of Finland (grants no. 63525,64542, 69014, 71110, 74131, 80696 and 202013 to E.K).

References

Barreto, G. R. and MacDonald, D. W. 1999. The responseof water voles, Arvicola terrestris , to the odours ofpredators. �/ Anim. Behav. 57: 1107�/1112.

Bergeron, J.-M. and Jodoin, L. 1982. Effets des composessecondaires des plantes sur le poids corporel et le poids decertains organes de la souris de laboratoire. �/ Can. J. Zool.60: 1855�/1866.

Boonstra, R., Krebs, C. J. and Stenseth, N. C. 1998. Populationcycles in small mammals: the problem of explaining the lowphase. �/ Ecology 79: 1479�/1488.

Borowski, Z. 1998. Influence of weasel (Mustela nivalisLinnaeus, 1766) odour on spatial behaviour of root voles(Microtus oeconomus Pallas, 1776). �/ Can. J. Zool. 76:1799�/1804.

322 OIKOS 105:2 (2004)

Chandler, C. R. 1995. Practical considerations in the use ofsimultaneous inference for multiple tests. �/ Anim. Behav. 49:524�/527.

Chitty, D. and Phipps, E. 1960. The effect of fleas on spleen sizein voles. �/ J. Physiol. 151: 27�/28.

Ergon, T., Lambin, X. and Stenseth, N. C. 2001. Life-historytraits of voles in a fluctuating population respond to theimmediate environment. �/ Nature 411: 1043�/1045.

Hansen, T. F., Stenseth, N. C. and Henttonen, H. 1999a.Multiannual vole cycles and population regulation duringlong winters: an analysis of seasonal density dependence.�/ Am. Nat. 154: 129�/139.

Hansen, T. F., Stenseth, N. C., Henttonen, H. et al. 1999b.Interspecific and intraspecific competition as causes of directand delayed density dependence in a fluctuating volepopulation. �/ Proc. Natl Acad. Sci. USA 96: 986�/991.

Hanski, I. and Henttonen, H. 1996. Predation on competingrodent species: a simple explanation of complex patterns. �/ JAnim Ecol. 65: 220�/232.

Hanski, I., Henttonen, H., Korpimaki, E. et al. 2001. Small-rodent dynamics and predation. �/ Ecology 82: 1505�/1520.

Harju, A. 1996. Effect of birch (Betula pendula ) bark and foodprotein level on root vole (Microtus oeconomus ). 2. Detox-ification capacity. �/ J. Chem. Ecol. 22: 719�/728.

Harju, A. and Tahvanainen, J. 1994. The effect of silver birch(Betula pendula ) powder on physiological performance offield voles (Microtus agrestis ). �/ Ann. Zool. Fenn. 31: 229�/

234.Hellstedt, P., Kalske, T. and Hanski, I. 2002. Indirect effects of

least weasel presence on field vole behaviour and demogra-phy: a field experiment. �/ Ann. Zool. Fenn. 39: 257�/265.

Henttonen, H. 1987. The impact of spacing behavior inmicrotine rodents on the dynamics of least weasels Mustelanivalis �/ a hypothesis. �/ Oikos 50: 366�/370.

Huitu, O., Koivula, M., Korpimaki, E. et al. 2003. Predationand winter food supply in population cycles of northernvoles: a two-factor experiment. �/ Ecology 84: 2108�/2118.

Jedrzejewski, W. and Jedrzejewska, B. 1996. Rodent cycles inrelation to biomass and productivity of ground vegetationand predation in the Palearctic. �/ Acta Theriol. 41: 1�/34.

Jonsson, P., Koskela, E. and Mappes, T. 2000. Does risk ofpredation by mammalian predators affect the spacingbehaviour of rodents? Two large-scale experiments.�/ Oecologia 122: 487�/492.

Klemola, T., Koivula, M., Korpimaki, E. et al. 1997a. Size ofinternal organs and forage quality of herbivores: are theredifferences between cycle phases in Microtus voles? �/ Oikos80: 61�/66.

Klemola, T., Koivula, M., Korpimaki, E. et al. 1997b. Smallmustelid predation slows population growth of Microtusvoles: a predator reduction experiment. �/ J. Anim. Ecol. 66:607�/614.

Klemola, T., Koivula, M., Korpimaki, E. et al. 2000. Experi-mental tests of predation and food hypotheses for popula-tion cycles of voles. �/ Proc. R. Soc. Lond. B 267: 351�/356.

Klemola, T., Pettersen, T. and Stenseth, N. C. 2003. Trophicinteractions in population cycles of voles and lemmings: amodel-based synthesis. �/ Adv. Ecol. Res. 33: 75�/160.

Korpimaki, E. 1988. Diet of breeding Tengmalm’s owls Aegoliusfunereus : long-term changes and year-to-year variationunder cyclic food conditions. �/ Ornis Fenn. 65: 21�/30.

Korpimaki, E. 1992. Diet composition, prey choice andbreeding success of long-eared owls: effects of multiannualfluctuations in food abundance. �/ Can. J. Zool. 70: 2373�/

2381.Korpimaki, E. and Norrdahl, K. 1991. Numerical and func-

tional responses of kestrels, short-eared owls, and long-earedowls to vole densities. �/ Ecology 72: 814�/826.

Korpimaki, E. and Krebs, C. J. 1996. Predation and populationcycles of small mammals. �/ BioScience 46: 753�/763.

Korpimaki, E. and Norrdahl, K. 1998. Experimental reductionof predators reverses the crash phase of small-rodent cycles.�/ Ecology 79: 2448�/2455.

Korpimaki, E., Norrdahl, K. and Rinta-Jaskari, T. 1991.Responses of stoats and least weasels to fluctuating volesabundances: is the low phase of the vole cycle due tomustelid predation? �/ Oecologia 88: 552�/561.

Korpimaki, E., Norrdahl, K., Klemola, T. et al. 2002. Dynamiceffects of predators on cyclic voles: field experimentationand model extrapolation. �/ Proc. R. Soc. Lond. B 269: 991�/

997.Koskela, E. and Ylonen, H. 1995. Suppressed breeding in the

field vole (Microtus agrestis ): an adaptation to cyclicallyfluctuating predation risk. �/ Behav. Ecol. 6: 311�/315.

Krebs, C. J. and Myers, J. H. 1974. Population cycles in smallmammals. �/ Adv. Ecol. Res. 8: 267�/399.

Mappes, T., Koskela, E. and Ylonen, H. 1998. Breedingsuppression in voles under predation risk of small mustelids:laboratory or methodological artifact? �/ Oikos 82: 365�/369.

Møller, A. P. and Erritzoe, J. 2000. Predation against birds withlow immunocompetence. �/ Oecologia 122: 500�/504.

Møller, A. P., Christe, P., Erritzoe, J. et al. 1998. Condition,disease and immune defence. �/ Oikos 83: 301�/306.

Mora, O., Shimada, A. and Ruiz, F. J. 1996. The effect of thelength and severity of feed restriction on weight, carcassmeasurements and body composition of goats. �/ J. Agric.Sci. 127: 549�/553.

Norrdahl, K. 1995. Population cycles in northern smallmammals. �/ Biol. Rev. 70: 621�/637.

Norrdahl, K. and Korpimaki, E. 1993. Predation and inter-specific competition in two Microtus voles. �/ Oikos 66:149�/158.

Norrdahl, K. and Korpimaki, E. 1995a. Does predation riskconstrain maturation in cyclic vole populations? �/ Oikos 72:263�/272.

Norrdahl, K. and Korpimaki, E. 1995b. Effects of predator-removal on vertebrate prey populations: birds of prey andsmall mammals. �/ Oecologia 103: 241�/248.

Norrdahl, K. and Korpimaki, E. 1995c. Mortality factors in acyclic vole population. �/ Proc. R. Soc. Lond. B 261: 49�/53.

Norrdahl, K. and Korpimaki, E. 1996. Do nomadic avianpredators synchronize population fluctuations of smallmammals?: a field experiment. �/ Oecologia 107: 478�/483.

Norrdahl, K. and Korpimaki, E. 1998. Does mobility or sexof voles affect risk of predation by mammalian predators?�/ Ecology 79: 226�/232.

Norrdahl, K. and Korpimaki, E. 2000a. Do predators limit theabundance of alternative prey? Experiments with vole-eatingavian and mammalian predators. �/ Oikos 91: 528�/540.

Norrdahl, K. and Korpimaki, E. 2000b. The impact ofpredation risk from small mustelids on prey populations.�/ Mammal Rev. 30: 147�/156.

Norrdahl, K. and Korpimaki, E. 2002a. Changes in indivi-dual quality during a 3-year population cycle of voles.�/ Oecologia 130: 239�/249.

Norrdahl, K. and Korpimaki, E. 2002b. Changes in populationstructure and reproduction during a 3-year population cycleof voles. �/ Oikos 96: 331�/345.

Norrdahl, K. and Korpimaki, E. 2002c. Seasonal changes in thenumerical responses of predators to cyclic vole populations.�/ Ecography 25: 428�/438.

Norrdahl, K., Klemola, T., Korpimaki, E. et al. 2002. Strongseasonality may attenuate trophic cascades: vertebrate pre-dator exclusion in boreal grassland. �/ Oikos 99: 419�/430.

Oksanen, L. and Lundberg, P. 1995. Optimization of reproduc-tive effort and foraging time in mammals: the influence ofresource level and predation risk. �/ Evol. Ecol. 9: 45�/56.

Oksanen, L. and Oksanen, T. 2000. The logic and realism of thehypothesis of exploitation ecosystems. �/ Am. Nat. 155:703�/723.

Oksanen, T., Oksanen, L., Jedrzejewski, W. et al. 2000.Predation and the dynamics of the bank vole, Clethrionomysglareolus. �/ Pol. J. Ecol. 48 (Suppl.): 197�/217.

Pyke, G. H. 1984. Optimal foraging theory: a critical review.�/ Annu. Rev. Ecol. Syst. 15: 523�/575.

OIKOS 105:2 (2004) 323

Reif, V., Tornberg, R., Jungell, S. et al. 2001. Diet variation ofcommon buzzards in Finland supports the alternative preyhypothesis. �/ Ecography 24: 267�/274.

Rice, W. R. 1989. Analyzing tables of statistical tests.�/ Evolution 43: 223�/225.

Seldal, T., Andersen, K.-J. and Hogstedt, G. 1994. Grazing-inducede proteinase inhibitors: a possible cause for lemmingpopulation cycles. �/ Oikos 70: 3�/11.

Smith, R. L., Hubartt, D. J. and Shoemaker, R. L. 1980.Seasonal changes in weight, cecal length, and pancreaticfunction of snowshoe hares. �/ J. Wildl. Manage. 44: 719�/

724.Stenseth, N. C. 1999. Population cycles in voles and lemmings:

density dependence and phase dependence in a stochasticworld. �/ Oikos 87: 427�/461.

Stenseth, N. C., Bjornstad, O. N. and Falck, W. 1996. Is spacingbehaviour coupled with predation causing the microtine

density cycle? A synthesis of current process-oriented andpattern-oriented studies. �/ Proc. R. Soc. Lond. B 263:1423�/1435.

Stephens, D. W. and Krebs, J. R. 1986. Foraging theory.�/ Princeton Univ. Press

Tenora, F., Wiger, R. and Barus, V. 1979. Seasonal and annualvariations in the prevalence of helminths in a cyclicpopulation of Clethrionomys glareolus. �/ Holarct. Ecol. 2:176�/181.

Turchin, P. and Batzli, G. O. 2001. Availability of food and thepopulation dynamics of arvicoline rodents. �/ Ecology 82:1521�/1534.

Turchin, P., Oksanen, L., Ekerholm, P. et al. 2000. Arelemmings prey or predators? �/ Nature 405: 562�/565.

Ylonen, H. 1994. Vole cycles and antipredatory behaviour.�/ Trends Ecol. Evol. 9: 426�/430.

324 OIKOS 105:2 (2004)