Costs of coexistence along a gradient of competitor densities: an experiment with arvicoline rodents



    1. Department of Biological and Environmental Science and Konnevesi Research Station, University of Jyväskylä, PO Box 35, FIN-40014 University of Jyväskylä (Finland);
    2. Department of Animal Behaviour, University of Bielefeld, PO Box 10 01 31, D-33501 Bielefeld, Germany
    Search for more papers by this author

    1. Department of Biological and Environmental Science and Konnevesi Research Station, University of Jyväskylä, PO Box 35, FIN-40014 University of Jyväskylä (Finland);
    Search for more papers by this author

Jana Anja Eccard, Department of Animal Behaviour, University of Bielefeld, PO Box 10 01 31, 33501 Bielefeld, Germany. Tel.: +49 521 106 2723. Fax: +49 521 106 2998. E-mail:


  • 1Costs of coexistence for species with indirect resource competition usually increase monotonically with competitor numbers. Very little is known though about the shape of the cost function for species with direct interference competition.
  • 2Here we report the results of an experiment with two vole species in artificial coexistence in large enclosures, where density of the dominant competitor species (Microtus agrestis) was manipulated. Experimental populations of the subordinate vole species (Clethrionomys glareolus) were composed of same aged individuals to study distribution of costs of coexistence with a dominant species within an age-cohort.
  • 3Survival and space use decreased gradually with increasing field vole numbers. Thus, responses to interference competition in our system appeared to be similar as expected from resource competition. The total number of breeders was stable. Reproductive characteristics such as the timing of breeding, and the litter size were not affected. In the single species enclosures a proportion of surviving individuals were not able to establish a breeding territory against stronger conspecifics. Under competition with heterospecifics such nonbreeders suffered high mortality, whereas the breeders survived.
  • 4Combined interference of dominant conspecifics and heterospecifics probably increased the frequency of aggressive interactions, social stress and mortality for the weaker individuals within a homogeneous age cohort of the subordinate competitor population.
  • 5Our results suggest, that in open systems where bank voles are outcompeted over the breeding season by faster reproducing field voles, animals able to establish a territory may be able to withstand competitor pressure, while nonbreeding bank vole individuals are forced to emigrate to suboptimal forest habitats.


Interference competition, interspecific territoriality and aggressiveness occur in a wide range of animal communities. Theories on competition have mainly focused though on indirect exploitative competition for a common resource. Interference competition has received relatively little attention (Amarasekare & Nisbet 2001; Amarasekare 2002). In species with indirect competition for resources, the density of a subordinate competitor may gradually decline with increasing density of a dominant competitor. With interference competition, however, dominant competitors may be able to inflict a damage disproportionately exceeding their numbers. Competing predators, for example, may prey on each other or on the others juveniles (Polis, Myers & Holt 1989 for review). House wrens actively destroy nests and broods of Bewick's wrens (Kennedy & White 1996) and thus reduce the recruitment rate of populations of subordinate competitors. In hermit crabs, dominant competitors monopolize intact shells and leave broken or eroded shells to the subordinate crab species, thus increasing vulnerability to predations (Turra & Denadai 2004). Without direct predatory interaction or facilitated predation by the competitor, most interspecific interference competitors act through direct aggressive behaviour (e.g. Walls 1990). The shape of the decline function in systems with interference interaction remains largely unstudied.

In animal communities species often avoid detrimental effects of competition by segregation either in time or in space (Connel 1980; Rosenzweig 1995; Morris 1999). Consequently, fitness costs of coexistence in mixed species communities that may have historically led to segregation are difficult to study. Experimental studies of situations in forced coexistence of probable competitor species may offer a tool to study both mechanisms of competition and density-dependent processes. Population-level responses to competition are the cumulative effects of individual responses. Life-history traits of individuals that describe growth, maturation, reproductive events and mortality (Stearns 1992) can be used to explain mechanisms of population density changes and the interaction of species within communities (e.g. Wauters, Lurz & Gurnell 2000; Eccard & Ylönen 2003a; Gurnell et al. 2004).

In a series of studies we investigated the effects of the interspecific competition on life-history variables of a subordinate competitor, using a system with two boreal arvicoline rodent species, the bank vole Clethrionomys glareolus and the field vole Microtus agrestis. Interference competition for space between the species is likely as food niches of the species have little overlap (Hansson 2002) and interspecific aggression increases from nonbreeding to breeding season (Fey, Ylönen & Eccard, unpublished data) despite increasing resource levels. Home range size and survival rate of the subordinate species are reduced in the presence of the dominant competitor, while measures depending on resource availability such as litter size and body weight were not affected (Eccard & Ylönen 2002). In northern Europe populations fluctuate in synchrony (Huitu et al. 2003; Huitu, Norrdahl & Korpimäki 2004) increasing competition during peak densities. Populations of Clethrionomys and Microtus voles are thought to typically exploit either forests or fields, respectively. Bank voles, however, can regularly be found in both habitats (Hansson 2002; Huitu et al. 2003) and prefer field habitats if not competitively excluded (e.g. Ylönen, Kojola & Viitala 1988). Over winter, the two genera can coexist in the same habitat (Iverson & Turner 1972 and own observations). As field vole densities in early summer grow faster than bank vole densities, the latter might face a trade-off situation of increasing competition in the field vole habitats over the summer (Henttonen & Hansson 1984). The two species may therefore offer a good system to study the mechanisms and fitness effects of interspecific competition.

Bank voles can tolerate many fold density variations (e.g. Alibhai & Gipps 1985; Henttonen 2000; Ivanter & Osipova 2000). Intraspecific competition among females is severe and regulation of maturation and of breeding is density-dependent (e.g. Prevot-Julliard et al. 1999; Yoccoz et al. 2001) as an exclusive territory is a prerequisite for breeding (Bujalska 1985). Thus, when total population density increases, the breeding density can only increase until all territories are occupied (e.g. Crespin et al. 2002) and remain stable on the saturation level (Bujalska & Grüm 1989; Eccard & Ylönen 2001). Consequently, the proportion of breeders is at maximum until all territories are occupied, and falls with increasing densities above this point (Fig. 1a, point a). If interspecific competition would act similarly, we would expect the same picture with the total rodent density (sum of individuals of both species) on the x axis. Meanwhile, as we will show in this study, mechanism of intraspecific competition and interspecific competition probably differ. A graphical model on interspecific density dependence based on the results of this study will be developed in the Discussion (Fig. 1b).

Figure 1.

A graphical model of density dependence of the number of females and of the proportion of breeding females with exclusive breeding territories, depending on the number of conspecifics (a) or on the number of heterospecific, dominant competitors (b). Note that the curves depicting number of individuals share the same theoretical scale (left axis), different from the scale of proportion (right axis). Point ‘a’ indicates the number conspecifics at which all breeding territories are occupied and the number of breeders remains stable over increasing total density. An increase of heterospecific density (b, this study) affects at first the survival of the nonbreeding females while the number of breeders stays constant over low ranges of increase of heterospecifics. Survival of breeders is reduced only beyond ‘b’ number of heterospecifics, which is reached at relative higher numbers compared with ‘a’ number of conspecifics. Thus, the density dependence observed in increasing density of conspecifics (a) is reversed with increasing number of heterospecifics.

In the present study, we manipulated the population sizes of field voles as the dominant competitor in large outdoor enclosures. With this experiment we (1) investigate the shape of the cost function of the superior competitor's population size on survival, reproduction, and space use of the subordinate competitor. In the past we had exposed bank voles to low numbers of competitors, here we expand the exposition from low over moderate and medium to high numbers of competitors. (2) In earlier experiments we had found, that bank vole mortality was increased by competition especially for younger and smaller breeders compared with older and larger breeders (Eccard & Ylönen 2003b). In this experiment we investigate which individuals within the cohort of overwintered individuals would bear the costs of coexistence. In short-lived species in seasonal environments, the cohort of overwintered animals is rather uniform in age and size by the onset of breeding. However, within the cohort intraspecific competition can be high after good local winter survival. Individuals differ in competitive strength, as not all of them are able to establish territories (Eccard & Ylönen 2001). We would expect an increased mortality from interspecific interference competition in those intraspecifically subordinate individuals.


The experiment was conducted in outdoor enclosures (0·25 ha) build in an old field near Konnevesi Research Station, Central Finland, at 62°37′ N, 26°20′ E. The experiment consisted of five runs of parallel enclosure replicates during the summers of 1999–2001. Vegetation consisted of grasses and forbs, with successional growth of willow, occasional young birch, spruce and elder trees. Each enclosure had a permanent grid of 5 × 5 Ugglan life traps with 10 m spacing.

Experimental bank vole populations consisted of five bank vole females and three to five bank vole males. Animals were born in autumn of the previous year and had overwintered in the laboratory under winter light conditions (6 h light for 2–3 months). During spring, the bank vole females had reproduced at least once in the laboratory prior to the experiment to ensure individuals’ fertility. Females were not pregnant or lactating when released to the enclosures.

Experimental bank vole populations were exposed to field voles in an additive set up, spanning a continuum from 0 to 30 field vole individuals (1 : 1 sex ratio) per enclosure. Total vole densities thus varied fourfold between controls (eight to 10 bank voles per 0·25 ha) and high competitor density treatment (eight to 10 bank voles plus 30 field voles per 0·25 ha). If an effect of competition is detected with this experimental design, the interpretation is open to the alternative hypothesis of being caused by the higher density in the competition treatment rather than interspecific competition. We nevertheless favoured the additive design for two reasons: the density range from which on interspecific competition occurs can only be determined with this design (Connell 1983) and furthermore, the design is appropriate to detect which life-history variables can be affected at all by the competition. We distinguish the effects of intra- and interspecific competition in our discussion with the help of the available literature on density dependence in bank vole populations (Fig. 1). Bank voles at peak density years are considered a pest for the forestry in many regions. Population dynamics are therefore well studied. The experiment was repeated in five temporal runs over three summers, with three to four simultaneous enclosure replicates (Table 1). Each run included a control without field voles. All voles were individually marked using small fingerling ear tags (C. glareolus) or fur clippings (M. agrestis).

Table 1.  Schedule of enclosure experiment over 3 years with 19 experimental bank vole populations that were exposed to different densities of competing field voles in 0·25 ha outdoor enclosures
Season and yearNo. of enclosures Field vole density (individuals per 0·25 ha)
Control (0)Moderate (8–11)Medium (16–20)High (26–30)
June 19991 11
June/July 1999121 
July 19991111
May/June 20001111
June/July 20012  2

Bank vole females were the first to be introduced into the enclosures. After two nights for habituation we added bank vole males and field voles simultaneously. After 2 weeks we live trapped for 10 trap checks over three consecutive days to determine space use of individuals. We started to remove the animals from the enclosures the 17th day after release of the bank vole males and field voles, which was also the last of the 10 trap checks. Bank vole females were observed in single cages in the laboratory to determine the accurate date of birth and the litter size. Nineteen days after the release of bank vole males to the enclosures, the first litters were born.

Numbers of bank voles introduced to the enclosures and the number of surviving females were not always the same in all enclosures. We therefore analysed proportions of survival out of introduced bank voles, and proportions of breeding females out of survived females with probit regression models in SPSS (SPSS Inc., Chicago, IL, USA). We also tested for the influence of season (start of the replicate) on these variables in separate models. Experiments were conducted in the first half of the summer where conditions were continuously improving with the date, therefore we were able to use the season as a continuous variable. Continuous response variables (averaged over enclosures) were analysed with linear regression models. Space use was calculated for 100% Minimum Convex Polygons using RANGES V software (Kenward & Hodder 1997).


Proportion of surviving bank voles decreased with increasing competitor numbers in the same enclosure (Probit regression, inline image = 31·2, P = 0·019, Fig. 2A) and increased with the advancing season (inline image = 39·3, P = 0·002, Fig. 2b). The number of bank vole females surviving until the end of the experiment could be predicted by the number of field voles released to the same enclosure (linear regression with a constant and competitor individuals as predictors: R2 = 0·336, anova F (d.f. 1, 17) = 8·6, P = 0·009; y = −0·07x + 4·3, for predictions by both competition and season see Table 2). Thus, the addition of a single field vole slightly decreased bank vole females’ chance of survival (Factor −0·07), i.e. the addition of 14 field voles reduce survival chances for one bank vole female.

Figure 2.

Proportion of surviving bank voles out of the introduced animals (upper panels a and b) and proportion of pregnant females out of surviving females (lower panels c and d) in 19 enclosed populations. Explanatory was either the field vole density (left panels a and c) or season (right panels b and d). Open symbols depict the observed values, dots the values predicted by probit models with significance levels for: (a) P < 0·05; (b,c) P < 0·01; (d) not significant.

Table 2.  Linear regression models for response variables of 19 enclosed bank vole populations predicted by both the number of field vole individuals in the same enclosure and the season (start of enclosure replicate). Statistics for the model (model R2, Analysis of variance: F-values, degrees of freedom (regression, residual), significance levels) and regression coefficients (unstandardized B, significance level) for the predictors are given. Models include a constant. Degrees of freedom among models may differ if data were not available for all enclosures, e.g. if no litters were born in an enclosure, timing of births or litter size could not be recorded. MCP: Minimum convex polygon
 Model R2Fanova d.f.PField vole numberSeason (days)
No. of females alive0·4827·42, 160·005−0·070·0090·090·050
No. of females pregnant0·3855·02, 160·021−0·0250·2730·1100·013
Mean timing of births (days after release of males)0·2011·82, 140·2070·020·7160·160·092
Mean litter size (no. of pups)0·1401·12, 140·3470·140·4730·400·227
Mean space use (100% MCP calculated in m2)0·5387·72, 110·008−17·760·0035·720·551
Mean trap use (no. of traps)0·2451·82, 110·214−0·020·6040·100·104

The total number of pregnant bank vole females per enclosure was 2·7 ± 1·3 females per enclosure and stayed rather stable among enclosures with different field vole densities (Fig. 3a) and increased with season (Table 2). The proportion of pregnant females out of the surviving females had therefore increased with field vole density (inline image = 38·8, P = 0·002, Fig. 2c) and the proportion of pregnant females was independent of season (inline image = 16·6, P = 0·483, Fig. 2d). Litters were born 19–29 days (22 ± 2 days, mean ± SD of the enclosure averages) after the release of bank vole males to the enclosures (Fig. 3b) and counted 5·6 ± 0·8 pups (mean ± SD of enclosures averages). Litter size and timing of litters born to bank vole females were neither related to field voles in the same enclosure nor to the season (Table 2).

Figure 3.

Effects of interspecific competition with field voles (x-axis) and season (different lines) on 19 enclosed bank vole populations. (a) The number of bank vole pregnancies per enclosure, (b) the timing of litters (mean number of days since the release of bank vole males to the enclosure), and (c) mean space use of bank vole females as indicated by live trapping (100% minimum convex polygon). For statistics please see Table 2.

Space use per bank vole female decreased with increasing competitor density (Fig. 3c) and was independent of season (Table 2). The predicted space use of a single bank vole female was reduced to nearly a third in the presence of 30 field voles (330 m2) compared with enclosures without field voles (870 m2). The number of captures per bank vole female (6·7 ± 1·2 mean captures per enclosure), a measure that could have contributed the observed differences in calculated home range size (Harris et al. 1990) was not affected by competitor numbers or season (Table 2). The addition of a single field vole would reduce the predicted mean space use of a bank vole female by 18 m2, a 50th part of the entire home range without field voles (linear regression: R2 = 0·569, anova F (d.f. 1, 12) = 15·8, P = 0·002; territory size = −18 m2 × number of field voles + 870 m2, for predictions by both competition and season see Table 2. As both the number of surviving bank voles and the space use per bank vole decreased, the total space use of the bank vole population declined with the increase of field vole numbers.


In this study the increasing strength of interference competition by field voles gradually reduced the numbers of surviving bank voles. In enclosures without field voles, both breeding and nonbreeding bank vole females survived, while in enclosures with field voles survivors were mostly breeders. The observed size reduction of territories in the presence of field voles and no apparent effect of field voles’ presence on reproduction in this study support our earlier results on the nature of competition between this two species (Eccard & Ylönen 2002; Eccard et al. 2002) even though competitor numbers were higher. Generally, territory size in female arvicoline rodents is affected by food availability (Ostfeld 1985). Experimental food addition can decrease the territory size of breeding females also in Clethrionomys voles (Ims 1987). If field voles would reduce bank voles’ food resources without interacting directly with the bank voles, territory size should have expanded. In our system we observed the opposite: with increasing field vole densities, surplus bank voles disappeared and mean territory size shrank. Furthermore, litter size, a measure responding to food shortages, was not affected by the competitors. We can therefore conclude that field voles have not limited bank voles’ food resources. Space use of bank voles shrank under intruder pressure, indicating direct behavioural interference by field voles. Agonistic interactions were reported from other Clethrionomys–Microtus species pairs (Grant 1970). Further, in the last replicate of this study (Summer 2001) we were able to collect evidence for agonistic interaction in staged dyadic encounters (Fey, Eccard, Ylönen, unpublished).

distribution of costs within a single age cohort

As not all surviving bank vole females were breeding in the single-species enclosures we think that the initial density of all experimental bank vole populations was above their maximum breeding density (Fig. 3a) for the enclosed system. Similarly, in open bank vole populations, not all overwintered females are able to establish breeding territories if local densities are high (Eccard & Ylönen 2001) and territory holders are probably the strongest of the competing bank vole females. Meanwhile, in the enclosures an increasing proportion of the surviving bank voles were breeding with the increase of field vole numbers. This could be a result of either only the breeders surviving, or of the territorial breeders having an increased mortality that would then reduce competition among the bank vole females and allow the initially nonbreeding females to utilize the available space for own breeding. We used the timing of breeding to distinguish between the two scenarios. If we assume, that territory owners would die through the interference with field voles, then we would expect later deliveries in enclosures with field voles than in enclosures without field voles, where the initial territory owners were able to keep their territories. As we did not measure a delay in the delivery date of bank vole females in higher field vole densities (Table 2, Fig. 3b) this scenario was not likely. Interspecific interference probably regulated bank vole numbers by cutting the ‘surplus’ bank voles, i.e. the nonbreeding individuals. The strongest bank vole females, those which were able to occupy breeding territories in competition among conspecifics, survived and bred successfully also in high densities of field voles. Thus competition with field voles may act as a selective force between stronger breeders and weaker nonbreeders, adding low survival probability to the latter. We had found a similar mechanism earlier in age-structured populations, where cohorts of younger breeders were bearing the costs of competition (Eccard & Ylönen 2003b).

Bank voles can be regularly found in the field vole habitats in winter (Iverson & Turner 1972) and spring (Hansson 2002). During summer they are outcompeted by the faster increasing field vole population and migrate to forest habitats (Ylönen et al. 1988). As nonbreeding subordinate were the first to suffer survival costs from field vole competition this study, they are probably the first to emigrate in an open system.

intraspecific and interspecific competition

An addition of field voles reduced survival rate of bank voles and increased the proportion of breeding females to nonbreeding females. With the additive experimental set-up of our study, we cannot rule out that the same results could be obtained with increased densities of conspecifics, i.e. we cannot directly distinguish between results of interspecific and intraspecific competition. However, as bank voles are considered pest rodents in many parts of their range, numerous long-term data sets on population dynamics have been collected (e.g. Alibhai & Gipps 1985; Henttonen 2000; Ivanter & Osipova 2000) and density dependence of bank vole population dynamics has been analysed with respect to survival (e.g. Crespin et al. 2002; Stenseth et al. 2002) and maturation of females (e.g. Prevot-Julliard et al. 1999). With increasing conspecific density, the proportion of breeders decreases due to severe intraspecific competition for territories (e.g. Bujalska & Grüm 1989; Prevot-Julliard et al. 1999; Yoccoz et al. 2001; and Fig. 1a). Meanwhile, the addition of heterospecifics as in our study decreased the number of conspecific bank voles. As nonbreeders within a cohort were the first suffering survival reduction, the conspecific density dependence of the breeding rate described in Fig. 1(a) reversed. Field vole addition cuts the surplus bank voles until only breeders are left (at point b Fig. 1b). Up to this point the breeding proportion of the subordinate competitor steadily increases and stabilizes at maximum when all survivors are breeders. Thus, the total density of a mixed community increased the proportion of breeding to nonbreeding bank vole females. Density dependence of bank voles’ survival and breeding in a mixed-species community was thus fundamentally different from conspecific density dependence.

shape of the cost function

Survival in response to field vole numbers declined gradually (Fig. 2a). The costs of coexistence from interference thus appear to be similar to a monotone decline function expected from resource competition. This indicates a direct translation of individual competitor numbers into strength of competition, probably due to rising numbers of agonistic interactions between individuals of the two species. Field vole individuals are larger and usually dominance is size related in rodents (Hanski & Henttonen 1996). Similar to the increased mortality of male bank voles due to increased frequencies of aggressive encounters with conspecific males (e.g. Smyth 1968; Bujalska & Grüm 1989) an increased frequencies of aggressive encounters with heterospecifics may increase mortality for the bank voles. Meanwhile, the effect size of a single field vole on bank vole individuals was relatively small (Figs 2 and 3) compared with density-dependent effects of conspecifics. This was probably due to the social system of field voles that allows high densities with overlapping territories and kin clustering (Pusenius & Viitala 1993; Pusenius et al. 1997) and little extra space per individual compared with territorial bank vole females.

In conclusion, interference competition in our study system increased costs of coexistence gradually in a similar manner as expected from exploitation competition. Within an age cohort, subordinate individuals of the subordinate species were bearing the cost of competition. Similar numbers of subordinate competitor were able to breed in all field vole densities, thus the reproductive output of the population seemed to remain stable over the investigated gradient of competitor density. The apparent survival reduction of weaker members of the subordinate species in the hermetic study system would probably force weaker individuals in open populations to emigrate and to utilize suboptimal habitats when densities of intra- and interspecific competitor populations are high.


We want to thank Tero Klemola, Heikki Henttonen, Joel Brown, Harry Andreasson and two anonymous referees who kindly advised earlier revisions. We also thank our field helpers over the years for patience with angry field voles and stuck Lada seat belts: Aira Aalto, Dorota Dudek, Karen Fey, Otso Huitu and Chris Madden. This study was financed by the Academy of Finland through the ‘Center of Excellency in Evolutionary Ecology’ at the University of Jyväskylä. Animal keeping and experiments were conducted according to institutional guidelines and approved by the University Board for Animal Experiments.