Causes and consequences of natal dispersal in root voles,Microtus oeconomus

ANIMAL BEHAVIOUR, 1998, 56, 1355–1366 Article No. ar980911

Causes and consequences of natal dispersal in root voles, Microtus oeconomus GRY GUNDERSEN & HARRY P. ANDREASSEN

Department of Biology, Division of Zoology, University of Oslo (Received 30 July 1997; initial acceptance 14 November 1997; final acceptance 23 April 1998; MS. number: 5601)

ABSTRACT To test the causes and consequences of variation in natal dispersal in root voles we released 53 matrilines (mothers with newly weaned litters) separately in field enclosures, during nine consecutive periods. The matrilines could disperse and distribute themselves among three pre-emptied habitat patches. Two dispersal measures were recorded: short-distance dispersal defined as individuals immigrating to a neighbouring patch, and long-distance dispersal defined as unsettled individuals captured along the fence of the enclosures. We analysed the role of social factors (i.e. maternal and litter characteristics), habitat quality (i.e. seasonal effect) and experimentally manipulated shape of the natal patch in dispersal. The consequences of dispersal were analysed with respect to the spatial distribution of kin, and to pregnancy in females and sexual maturation in males. Dispersal was unrelated to patch shape. In agreement with the inbreeding avoidance hypothesis, long-distance dispersal was male biased and philopatric males were most frequently reproductively inactive. Whilst young males avoided their mother, they seemed to disperse, settle and mature sexually independently of their sisters. In agreement with the resource competition hypothesis, young females avoided their mother and were most frequently reproductively inactive when residing in their mother’s patch. We conclude that inbreeding avoidance was underlying the male dispersal pattern. For females, long-distance dispersal was most in agreement with the inbreeding avoidance hypothesis while short-distance dispersal could be explained by the resource competition hypothesis. 

strategy), and as a process with significant population level consequences. In particular, the spatial distribution of kin resulting from dispersal is thought to be a basic determinant of vole population dynamics (Charnov & Finerty 1980; Lambin & Krebs 1991, 1993; Wolff 1995). A multitude of proximate factors of dispersal have been proposed (see Lidicker & Stenseth 1992), mainly associated with individual or litter characteristics (Bekoff 1977; Ims 1987, 1989, 1990; Bondrup-Nielsen 1993; Jacquot & Vessey 1995), local social factors (Lidicker 1975; McShea 1990; Wolff 1992, 1993; Bollinger et al. 1993; McGuire et al. 1993) and habitat quality (Lidicker 1975; Wolff & Lidicker 1980). It has also been proposed, however, that spatially explicit aspects of the environment, such as habitat patch shape, enhance dispersal (Stamps et al. 1987; Harper et al. 1993; Ims 1995). The individual’s responses to some of these proximate factors may have an adaptive basis (e.g. social factors); others may not (e.g. habitat patch shape). To disentangle which are adaptive responses and which are not, controlled experiments are needed in which both the triggering factors, and the fitness consequences, of dispersal are analysed.

Natal dispersal has attracted considerable attention from both behavioural ecologists and population biologists (see e.g. Stenseth & Lidicker 1992). While behavioural ecologists have been concerned with understanding the proximate and ultimate causes of dispersal and its fitness consequences (e.g. Bengtsson 1978; Greenwood 1980; Liberg & von Schantz 1985; Pusey 1987), population biologists have considered natal dispersal to be a central process for the understanding of demography, population dynamics and genetics (Krebs 1979; Johnson & Gaines 1990; Stenseth & Lidicker 1992). Thus, the understanding of dispersal as a behavioural strategy and as a process influencing population level phenomena requires that the causes and consequences of dispersal must be linked. The connection between the causes and the consequences of dispersal is, however, rarely addressed in the same study. Dispersal has been extensively studied in voles, Arvicolidae, both as a topic by itself (i.e. as a behavioural Correspondence: H. P. Andreassen, Department of Biology, Division of Zoology, University of Oslo, P.O. Box 1050 Blindern, N-0316 Oslo, Norway (email: [email protected]). 0003–3472/98/121355+12 $30.00/0

1998 The Association for the Study of Animal Behaviour

1355



1998 The Association for the Study of Animal Behaviour

1356 ANIMAL BEHAVIOUR, 56, 6

To reveal the adaptiveness of natal dispersal, we designed a study on root vole litters where we were able to infer both the causes and consequences of dispersal. We replicated introductions of root vole mothers with newly weaned litters into separate study plots, each consisting of three habitat patches which facilitated the observation of dispersal between patches and the resultant spatial association between kin members. We tested the following proximate causes of dispersal: (1) social factors, where we analysed the effects of natural variability in the characteristics of the individuals and of the matriline (i.e. presence of the mother and her body mass, offspring body mass, litter size, litter sex ratio and sex); (2) habitat quality, where we analysed the effects of the natural variation in habitat quality as a seasonal effect; (3) patch shape, where we analysed the effects of experimentally manipulated patch shapes (circular-, cross- and linearly shaped). We followed each experimental trial long enough to analyse the spatial relations between postdispersal individuals and to compare reproduction of dispersers and residents, that is, to test the consequences of dispersal. Predictions with regard to the causes and consequences of dispersal may be derived from the two main hypotheses on ultimate factors for dispersal in mammals: inbreeding avoidance (e.g. Pusey 1987) and resource competition (e.g. Liberg & von Schantz 1985). (1) Inbreeding avoidance hypothesis. If dispersal is an adaptation against inbreeding, it should be sex biased, caused by social factors such as the presence of the mother or opposite-sex kin (cf. Wolff 1992). The consequences of inbreeding avoidance for the spatial distribution of postdispersal animals would be a unisexual colonization of patches by littermates. An inhibition of sexual maturation of philopatric sons would also support the inbreeding avoidance hypothesis even in the absence of sex-biased dispersal. (2) Resource competition hypothesis. If dispersal is an adaptation against resource competition, it should be triggered by social factors such as (a) litter size, since competition for resources will increase with increasing litter size, and (b) presence of mothers, since daughters would compete with their mother for breeding space. The vole habitat in this study consisted principally of perennial meadow vegetation, which would gradually decrease in quality as it wilted towards the end of the study period. Hence, under the resource competition hypothesis, we also predicted that dispersal rates would increase towards the end of the season given that the young attempted to reproduce. The consequences of competition for resources would be an even dispersion of animals within the plot, and reproductive inhibition of philopatric daughters. Predictions with regard to the effect of patch shape on dispersal rates may be made either on the basis of resource competition (i.e. have an adaptive basis), or the movement characteristics of individuals relative to patch edges (i.e. a nonadaptive, mechanistic consequence). The perception of resources may vary with patch shape so that resource competition is enhanced when the perimeter/ area ratio of a patch increases (Fauske et al. 1997). If so, we expected to find a statistical interaction between litter

size or presence of mothers and patch shape affecting dispersal rate and settlement pattern. If dispersal is only a passive side-effect of individual movement pattern, we expected dispersal to be caused by the probability of reaching and crossing the border of the natal patch. This would give two outcomes: (1) dispersal rates increase with increasing perimeter/area ratio of the patches (Okubu 1980; Kareiva 1985; Stamps et al. 1987); or (2) dispersal rates increase with increasing number of corners in the patches (Forman 1995). The latter prediction is based on the assumption that individuals follow habitat edges and cross them when they suddenly reach a new edge, in contrast to curved edges which the individuals are more likely to follow. METHODS

Field Procedures Experimental area and animals This study took place between 1 July and 12 November 1994, at Evenstad Research Station, southeast Norway. The study area, consisting of six vole-proof plots, each measuring 5016.7 m, was sown with a mixture of grass and clover in 1989, and has thereafter been burned and fertilized yearly to fulfil the habitat requirements of root voles (Tast 1966; Ims et al. 1993). The entire area was surrounded with a chicken mesh fence 1.5 m high supplied with an electric wire to prevent mammalian predators from entering. Because we saw a nesting pair of long-eared owls, Asio otus, hunting voles in the plots in the first period (see below), we set up a net extending 1.8 m above ground to cover the entire area before the second period started, to prevent further avian predation. The root voles used in the present experiment originated from a southern Norwegian strain (Ims 1997), which were outbred for three to four generations at the Animal Division, Department of Biology, of the University of Oslo. The voles were kept in a room used only for root voles, with separate cages (603030 cm; water and food ad libitum) for each pair of root voles. Photoperiod (16:8 h light:dark), temperature (14C), food (laboratory pellets, oats and hay) and daily care were standardized according to the procedures used by the Animal Division.

Treatment Three habitat patches of tall and dense meadow vegetation were linearly arranged in each of the six plots. The surroundings were nonhabitat areas in which we removed the vegetation with herbicides (Fig. 1). Each habitat patch was 50 m2, the same size as a home range core area of a reproducing female root vole (Andreassen et al., 1998), and was separated from other patches by 5 m of nonhabitat. This interpatch distance is large enough to reduce movement of resident animals between patches (Berg 1995). The central patch, hereafter called the release patch, was the patch on to which the animals were introduced to the plots (see below), whereas the other two patches in each plot, hereafter called immigration

GUNDERSEN & ANDREASSEN: DISPERSAL IN ROOT VOLES 1357

50 m A

5m

16.7 m B

5m

C

100 m

5m

D

E

F

Nonhabitat

Habitat

Figure 1. Habitat configurations, and the location of trap stations (). Habitat patches in the centre of the plots are the release patches; the two circular patches on either side of the release patches are the immigration patches.

patches, established dispersal events followed by settlement (see below). The experimental treatment consisted of varying the shape of the release patch as either circular, cross- or curvilinearly shaped (hereafter called circle, cross and linear plots). Owing to the different shapes, the number of corners and the perimeter/area ratio of the central patches varied: 0.5/m for circles, 0.75/m for crosses and 1.0/m for the linear patches. The sequence of the three treatment levels was chosen at random for the first three plots (A, B, C), and then repeated for the next three plots (D, E, F; Fig. 1), to ensure maximum spatial interspersion of the treatments (Hurlbert 1984).

(but see below). Before each period we removed any voles already present. Each experimental trial started with the release of a matriline (a mother with her newly weaned litter), together with an unfamiliar sexually mature (scrotal) male, in the central patch of each of the six plots. We released the animals from the cages in which they had been brought up by opening one of the walls, so that the animals could move freely in and out of the cages. The cages stayed at the release point during the whole trial. We recorded weight and sex for each individual before release, and individually marked all the animals by toe clipping. An unfamiliar scrotal male was released with the matriline to ensure the young females matured and reproduction started early. Neither the mother nor the young females were pregnant, and none of the young males was scrotal at release. For each period, the six available matrilines were randomly allocated between the plots, and thus between the patch shape treatments. The animals were left undisturbed during a settlement period of 12 days, whereafter we removed them by using Ugglan multiple capture live traps before the start of the next period. Two live traps were situated inside each patch and two by the fence of each enclosure (hereafter termed fence traps; see Fig. 1). These live traps were active for 3 whole days and were checked every 12th hour. All traps were baited with potatoes and wholegrain oats. In addition, in each patch we placed four snap traps in runways for the last 24 h of the removal trapping to ensure that all animals were removed. We recorded individual identity and trap station for each trap event whereafter the animals were kept in separate cages (food and water ad libitum) until the last trapping day. Fifteen days after release the animals were killed by cervical dislocation and dissected. The dissection included recordings of weight and reproductive status, that is, number and size of embryos, and presence of visible tubili in the cauda epididymus, for females and males, respectively. Because one whole matriline and one litter died just before release in period 4, one replicate of the linear treatment was lacking, and one mother was released solitarily with the mature male in a circle treatment in period 4. Thus, we performed a total of 53 trials (six plotsnine periods
one matriline), consisting of 53 mothers, 168 young females and 152 young males. The total number of replicates for the circle, cross and linear treatments were thus 17–18–17 for the litters, and 18–18–17 for the mothers, respectively. We used 17 scrotal males during the study. To ensure independence between the treatment and the males, we randomized males between the plots in each period. Before the experiment started (period 1), each plot had been colonized by three root vole pairs during 20 days in June to establish runway systems and scent marks so that the initial conditions for the experiment did not differ substantially from those later on.

Ethical Note The experimental trial We conducted the experiment during nine consecutive periods, each of which included six trials, one per plot

We used the minimum sample size required to answer the questions we addressed: the nine periods were needed

1358 ANIMAL BEHAVIOUR, 56, 6

Analysis Definition of spatial status Voles disperse shortly after weaning, males a few (2–7) days later than females (Boonstra et al. 1987; Ims 1990; Lambin 1994). The 12-day settlement period of each trial should therefore suffice to observe all natal dispersal events triggered by factors early in the life history of the individuals. Some individuals do not disperse (i.e. residents; Boonstra et al. 1987), others disperse short distances and establish themselves in the neighbourhood of the natal area (i.e. within the maternal home range; Lambin et al. 1992), and others disperse over several kilometres (Liro & Szacki 1987; Steen 1994). To distinguish between residents, short-distance dispersers and long-distance dispersers within the extent of an enclosed plot, we used the location of individuals at removal as a criterion. Individuals that were removed from the release patch were defined as ‘residents’; individuals removed from the immigration patches dispersed a short distance and were termed ‘immigrants’; and we presumed that individuals *Editorial footnote: toe clipping is a technique that has been widely used for marking small vertebrates for individual identification. There are now, however, several, more benign, alternative marking techniques that are suitable for short- and/or long-term marking (see for example Poole 1994, ASAB Newsletter, 20, 7–8, copies obtainable from the Ethical Committee Secretary; see also McGregor & Peake, in press, The role of individual identification in conservation biology. In: Behavioural Ecology and Conservation Biology (Ed. by T. M. Caro), Oxford: Oxford University Press). As part of the continually developing ethical policy of the journal, therefore, the editors will in future require convincing justification for the use of toe clipping rather than one of the alternatives.

Settled

Unsettled Immigrant

Resident

Immigrant

Nonphilopatric offspring Philopatric offspring (mother absent) (mother present)

Unsettled

to include the whole reproductive season, and in each period we had two replicates of each of the three treatment levels which is a minimum to obtain some variation in each treatment level in each period. We killed the voles after a trial, in part so we could assess their reproductive status. In addition, because we used laboratory-reared litters under seminatural conditions we could not release them into natural populations, or bring them back to the laboratory. The alternative of studying natural populations was not an option as we needed to manipulate habitats, and to control populations and individual life histories. The voles were killed by cervical dislocation because we found it to be the most humane way of killing voles in the field. This method does not require the transport of animals to institutions with special equipment. We also used snap-trapping which may not always kill animals outright but we needed to use this method to ensure that the plots were empty before the start of a new period. Only one individual was snap-trapped, and it was killed outright. We marked individuals by toe clipping. We clipped a minimum number of toes (maximum two) to distinguish between all individuals within a period. We have never observed detrimental effects on the health or breeding performance of root voles or other vole species that could be connected to toe clipping.*

Figure 2. The types of spatial status of individuals at removal. The distinction between philopatric and nonphilopatric is done according to the presence of the mother and was thus not specific to patch type (i.e. release and immigration patch).

removed from the fence traps were destined for longdistance dispersal and were termed ‘unsettled’. Thus, we assume that residents and immigrants had settled themselves within the plots, whereas unsettled individuals had not (see Fig. 2 for a description of spatial status). The spatial status of individuals depends not only on residency versus dispersal, but also on whether offspring establish themselves together with their mother (i.e. philopatry; Boonstra et al. 1987). Thus, we defined individuals as ‘philopatric’ if they were found in the same patch as their mother at the time of removal, and ‘nonphilopatric’ otherwise. Philopatry was thus not patch specific, but depended on maternal settlement (see Fig. 2). We determined the spatial status of the individual, as defined above, based on the capture location at removal which may be insufficient to establish the home range of an individual (Andreassen et al. 1993). To validate the assumption that individuals used mainly the patch from which they had been removed, we placed two dishes with dyed oat-porridge in each patch 24 h before we started the removal trapping. The dyes (fluorescent colour) were patch specific, and if the porridge was ingested by the voles, we could establish in which patch an animal had foraged by examining the faeces (see Hovland & Andreassen 1995). The mean proportions of settled and unsettled individuals that had been eating dyed bait were 0.87 (95% CI 0.82–0.91) and 0.24 (95% CI 0.11–0.38), respectively. The mean proportion of cases for which the colour in faeces matched the inhabited patch (as decided by removal trapping) for settled individuals was as high as 0.85 (95% CI 0.82–0.88). Another source of error with the identification of the spatial status of individuals could be that vole movement and settlement were disturbed by the sequential removal of animals during the 3 days of removal trapping. However, 93% of the individuals were removed on the first day of trapping. We therefore conclude that these two sources of error are likely to have only a minor influence on the spatial status results.

Testing causes of dispersal To test for the causes of dispersal in root voles we analysed the two dispersal categories (immigrated and unsettled) by logistic models (Hosmer & Lemeshow 1989) using predictors describing social factors (maternal presence in the release patch, maternal and offspring body mass at release, litter size and sex ratio, and sex), habitat

GUNDERSEN & ANDREASSEN: DISPERSAL IN ROOT VOLES 1359

quality (as a linearly decreasing seasonal variable, i.e. periods 1–9) and patch shape. In addition, as some unknown factors could be associated with the individual plots influencing dispersal rate, a plot identity factor was included in the full model. For each response we used a backward selection procedure to select a statistical model including only significant (P<0.05) terms, from a full model containing all predictor variables and two-way interactions. We did all analyses separately for the mothers and the litters. We did sex-specific analyses of the litters whenever the response was affected by sex, or when embryo size was needed as a covariable in the analysis. We did not attempt to assess any effects of the unfamiliar mature male as adult males use large areas and therefore roamed over the entire plot areas (Ims et al. 1993; Andreassen et al. 1998).

Testing consequences of dispersal We did three analyses to test for different aspects of the postdispersal spatial relations among settled kin members (unsettled individuals were removed from these analysis). (1) We analysed the spatial association between the mother and her offspring by estimating the proportion (binomial 95% CI) of daughters and sons that were philopatric (i.e. inhabited the same patch as their mother). If individuals in a litter distributed themselves randomly among the three patches available we expected the mother to live together with one-third of her daughters and sons. If the mother lived with a smaller proportion than one-third of her offspring, this would indicate that daughters and sons avoided their mother or were expelled by her, whereas a higher proportion would indicate an attraction between mothers and daughters/ sons. (2) We analysed the spatial association between young females and young males at the patch level (linear regression with litter as a random factor) by correlating the logit transformed proportion of females in the litter with the logit transformed proportion of males in the litter inhabiting the same patch. A positive correlation would indicate an attraction, a negative correlation would indicate an avoidance, whereas no correlation would indicate that the settlement of the two sexes was independent of each other. (3) To see if young females or young males aggregated within the patches we analysed young females and males separately with Monte Carlo tests (Manly 1991). We did not use parametric methods (e.g. based on the Poisson distribution; see e.g. Krebs 1989) as they are not suitable when as few individuals as there are within a litter are distributed in only three patches (see e.g. Green 1966). The number of females/males in a litter were randomly allocated to the three patches. This was done for all litters with more than one established young female or male. The number of patches inhabited by 0, 1, 2, . . ., largest litter size, for the whole sample of litters, was counted. We repeated this procedure 5000 times to establish a null distribution of the frequency of patches inhabited by a given number of individuals, assuming random settlement (median; 2.5 and 97.5 percentiles). The null

distribution was then compared with the empirical data. A significant difference between the observed and the expected distribution was obtained when the observed value was above or below the 2.5 and 97.5 percentiles of the expected distribution. Prior to these inferential analyses of the distribution pattern of kin, we did an initial correspondence analysis (Jongman et al. 1987) to make a graphical representation of the dependence between the location of settlement and animal category (i.e. mothers, daughters and sons). We did the correspondence analysis at the patch level, by entering the number of animals belonging to each category inhabiting a patch as responses. As measures of components of fitness we used the recorded reproductive parameters, that is, for females, the presence of embryos, and for males, the presence of tubili epididymus. We analysed the fitness consequences of dispersal in terms of these parameters of sexual maturity by comparing sexual maturation between immigrants and residents, and between unsettled and established individuals (see Fig. 2). We analysed the fitness consequences of the spatial composition of kin members by using the same reproductive parameter. We tested the consequences of philopatric behaviour on offspring by comparing philopatric and nonphilopatric offspring. In addition, we tested the consequences for young females and young males of sharing a patch with their brothers or sisters. In all these analyses on the fitness consequences of dispersal, we compared only animals that had settled in a patch (i.e. excluded the unsettled individuals, Fig. 2). We used logistic models for all these statistical tests. RESULTS

Demographic Development Fifty of the 53 released mothers and 262 of the 320 released offspring survived the field period. All three mothers that died during the field period inhabited linear plots. There was no significant treatment effect on maternal survival, however (Fisher’s exact test: P=0.104). For the analysis of litter survival, we excluded the first period without the predator net, because the survival rate in the first period (0.57, 95% CI 0.39–0.73) was much lower than in the last eight periods (0.86, 95% CI 0.81–0.89; logistic model: ÷21 =12.70, P<0.001). For the last eight periods litter survival decreased over the season (Fig. 3a). The growth of all animal categories (Fig. 3b), and the reproductive activity of mothers and sexual maturation of young males (Fig. 3c) also decreased throughout the season. The proportion of pregnant females in each litter did not correlate with any of the predictor variables (all ÷21 <2.26, P>0.132).

Causes of Dispersal The proportion of young animals that emigrated from the release patch (both immigrants and unsettled) averaged 0.75 (95% CI 0.71–0.79). This proportion did not differ between young females and young males

1360 ANIMAL BEHAVIOUR, 56, 6

(÷21 =0.21, P=0.651; Table 1). When considering the proportion that settled down in an immigration patch, however, the proportion was higher for females than for males (÷21 =23.77, P<0.001). Consequently, the probability of becoming unsettled was clearly highest for young males. Although 28 out of the 50 surviving mothers left the release patch (and settled in the immigration patch), none of them attained unsettled status (Table 1). Because of the differential dispersal pattern between young females and young males, we did sex-specific analyses of the proximate factors for dispersal. The probability of a mother leaving the release patch and becoming an immigrant decreased with increasing litter size (Fig. 4a). The proximate factors for immigration among young animals depended on maternal factors as a higher proportion of young females immigrated if their mother resided in the release patch (Fig. 4b), whereas male immigration was negatively correlated with maternal body mass (Fig. 4c). With regard to the probability of becoming unsettled, there were no significant predictors (÷21 <5.10 and P>0.078). (a)

1.0

Survival

0.8 0.6 0.4 0.2 0.0

1

2

3

4

5

6

7

8

16

(b)

14 12 Growth (g)

9

Young females

10

Young males

8 6

Mothers

4 2 0 –2

1

2

3

4

5

6

7

8

1.0

Mothers

9 (c)

Reproduction

0.8 0.6

Young females

0.4

Young males

0.2 0.0

1

2

3

4

5 6 Period

7

8

9

Table 1. Number of mothers, young females and young males trapped at three locations at removal: the release patch (residents), the two immigration patches combined (immigrants) and the fence traps (unsettled) Category

Mothers Young females Young males

Residents

Immigrants

Unsettled

22 38 27

28 96 56

0 10 35

Spatial Consequences of Dispersal The initial correspondence analysis of spatial dispersion indicated an overall avoidance between mothers and their offspring with regard to settlement (dimension 1), and also a tendency for the sexes to be located in different patches (dimension 2, Fig. 5). This lack of association between the sexes, however, may be due to the higher proportion of unsettled males than unsettled females. The correspondence analysis also illustrates that a higher proportion of mothers tended to reside in the release patch, whereas the young animals tended to emigrate (i.e. immigrate and unsettle; Fig. 5, see also Table 1). The proportions of philopatric daughters (mean=0.17, 95% CI 0.12–0.26) and sons (mean=0.21, 95% CI 0.13–0.30) were well below the expectation from a random dispersion (i.e. 0.33) indicating an avoidance between mothers and their offspring. The predicted proportions of philopatric daughters and sons were statistically indistinguishable (÷21 =0.47, P=0.493). The proportion of daughters in a litter that had settled in a patch did not correlate significantly with the proportion of their brothers that had settled in the same patch, indicating that the settlement of one sex was independent of the settlement of the other sex (Fig. 6). The Monte Carlo test of the spatial distribution of young animals showed that patches inhabited by several (>3) or no sisters/brothers were frequently observed, and hence patches inhabited by one or two sisters/brothers rarely observed, compared with

Figure 3. (a) The proportion of the litter that survived the field period throughout the season.  (period 1, excluded from the statistical analysis) and  indicate the observed proportions. The line depicts the predicted litter survival probabilities over the season (logistic model: χ21 =8.85, P=0.003) with the first period excluded. Two litters in period 4 died before release and are thus not in the plot (see Methods). (b) Predicted growth (i.e. change in body mass from release until killed) over the season for mothers (slope= −1.00±0.31 (SE), F1.47 =10.30, P=0.003), young females (slope= −0.67±0.12 (SE), F1.92 =28.90, P<0.001) and young males (slope= −1.42±0.20 (SE), F1.69 =52.49, P<0.001). Embryo size is included as a covariable in the models for mothers and young females, and body mass at release and litter (random factor) are included in the models for young females and males. (c) Predicted probabilities of becoming reproductively active throughout the season. Pregnancy of mothers (logistic model: slope= −0.41±0.22 (SE), χ21 =4.25, P=0.039), pregnancy of young females (slope= −0.05±0.09 (SE), χ21 =0.30, P=0.586) and presence of tubili epididymus in young males (slope= −0.06±0.20 (SE), χ21 =53.51, P<0.001). Litter average of male body mass is included in the model for males.

GUNDERSEN & ANDREASSEN: DISPERSAL IN ROOT VOLES 1361

Unsettled

0.6 0.4 0.2

–0.5 3

2

4

5 6 Litter size

7

1

1.5

2

Resident –0.5 Female

Immigrant

(b) 0.9

–1

0.8

Figure 5. Results from the correspondence analysis of the association between different groups of animals at removal, presented by the locations of mothers, daughters and sons along the first two dimensions (Dim 1 and Dim 2). Mean dimension values based on all individuals in the three spatial status classes (with 95% confidence squares) are illustrated with boxes.

0.7 0.6 0.5 0.4

Mother resided

Mother immigrated Proportion of males in a patch

Young female immigration

Mother

Dim 2 0.5

9

8

1.0

1.0 Young male immigration

0.5 Dim 1

Maternal immigration

(a)

0.8

0.0

1

Male

1.0

(c) 0.8 0.6 0.4 0.2 0.0 30

35

40 45 50 Maternal body mass

55

60

Figure 4. (a) Maternal immigration probability dependent on litter size (χ21 =3.93, P=0.048). Circles present the observations (1= emigrated). (b) Young female immigration probability (and 95% CI) dependent on maternal settlement (resided or immigrated) (χ21 =9.34, P=0.002). (c) Young male immigration probability dependent on maternal body mass (χ21 =5.15, P=0.023). Different symbols denote different weight on the observation in the model due to different binomial denominators (i.e. litter size). : Proportion of immigrated males in each litter with large weight on the logistic regression line (i.e. litter size of five males); : small weight (litter size of one male); : the intermediate levels (litter size of two, three and four males).

the number expected given a random settlement pattern (Fig. 7). These results suggest that both young females and males had a tendency to cluster intrasexually.

1 7

2

2

2

6

0.8 2 0.6

4

2

0.4 0.2

0

2 10 2 5 2 0.2 0.4 0.6 0.8 Proportion of females in a patch

6 1

Figure 6. Proportion of young females in a litter inhabiting a patch plotted against the proportion of young males inhabiting the same patch. Numbers indicate the number of overlapping data points. The line depicts the estimated regression line (F1.38 =0.38, P=0.542; litter is included as a random factor).

of dispersal on young male sexual maturation, whereas established young females tended to have a higher probability of becoming pregnant than unsettled young females. Philopatric daughters tended to have a lower probability of becoming pregnant than nonphilopatric daughters. Also, philopatric sons tended to show less sexual maturation than nonphilopatric males, whereas daughters and sons experienced no fitness consequences of sharing a patch with their sisters or brothers. DISCUSSION

Fitness Consequences of Dispersal Table 2 gives the results of the analyses of fitness consequences of dispersal and spatial composition of kin members (i.e. spatial status). There were no consequences

At first animal dispersal was thought to be nonadaptive, that is, dispersing individuals were sick, weak and social outcasts and forced to leave their home area (Elton 1927, 1930). With the emerging disciplines of behavioural and

No. of patches

1362 ANIMAL BEHAVIOUR, 56, 6

70 60 50 40 30 20 10

(a)

40

(b)

30 20 10 0

1

2

3

4

5 6 0 1 Number in patch

2

3

4

5

Figure 7. Expected number (median with 2.5% and 97.5% percentiles) of patches inhabited by (a) 0, 1, 2, . . ., 6 young females and (b) 0, 1, 2, . . ., 5 young males given a random settlement. The maximum number of females and males that had established in a litter were 6 and 5 for females and males, respectively. The lines (——) show the observed numbers.

evolutionary ecology, however, emphasis was put on natal dispersal as an adaptive behavioural strategy. Empirical information on small mammals, which have served as prominent model organisms in the study of animal dispersal (Stenseth & Lidicker 1992), has suggested that natal dispersal is a complex phenomenon that may be proximately triggered by multiple factors (Johnson & Gaines 1990). The apparent multitude of triggering mechanisms, however, has complicated the interpretation of which ultimate factors underlie the proximate responses. A few recent studies (Ims 1987, 1989; McShea 1990; Wolff 1992; Bollinger et al. 1993; Bondrup-Nielsen 1993) have successfully used an experimental approach to test ultimate hypotheses of small mammal natal dispersal, and we also used this approach.

Experimental Setting Some of the control measures applied in experimental studies, such as enclosures, may limit the generality of the results; however, experiments on dispersal do have some useful properties. By using a combined laboratory and field experimental protocol (see Ims 1990) it is possible to know the study animals’ life histories. Moreover, with enclosures one can cancel out nonfocal factors, and distingish between mortality and dispersal which is often not possible in open populations (Gaines & McClenaghan 1980; Getz & Hofmann 1986). The main drawback of enclosure studies is the limits they impose on the spatial scale. The scale of our study was large enough, however, to incorporate some essential aspects of the root vole’s natural environment. Root voles typically inhabit a highly patchy habitat for which patch sizes may approximate the size of the home range core area of breeding females (Tast 1966; Lambin et al. 1992; Viitala 1994; Gliwicz 1997). In such a setting solitary monogamous pairs may form (Viitala 1994; R. A. Ims, unpublished data). Thus, the patch size and configuration of our experiment simulate important aspects of the natural habitat. Nevertheless, our results still need to be interpreted with the scale constraints in mind. A major obstacle in studies of small mammal dispersal, both in natural, observational settings and in enclosure experiments, is to obtain operational definitions of dispersal (e.g. Wolff et al. 1996 and references therein). Commonly, dispersers have been defined as individuals that have moved a certain distance to establish their own breeding home range in continuous habitats (e.g. Boonstra et al. 1987;

Table 2. Estimates and statistics (chi-square values and two-tailed P values) of reproduction (i.e. fitness consequences) for the different types of spatial status and animal categories Category

Dispersal Daughters Resident/Immigrant Settled/Unsettled Sons Resident/Immigrant Settled/Unsettled Spatial composition of kin Daughters With/without mother With/without sisters With/without brothers Sons With/without mother With/without sisters With/without brothers

Predicted values and 95% CI

χ21

P

0.57 (0.41–0.72)/0.53 (0.43–0.63) 0.61 (0.52–0.70)/0.12 (0.03–0.27)

0.13 22.97

0.718 0.001

0.45 (0.26–0.66)/0.35 (0.24–0.48) 0.37 (0.26–0.50)/0.31 (0.20–0.45)

1.29 1.84

0.256 0.174

0.26 (0.10–0.52)/0.60 (0.49–0.71) 0.58 (0.48–0.68)/0.49 (0.34–0.64) 0.55 (0.42–0.68)/0.55 (0.44–0.66)

5.52 1.02 0.21

0.019 0.312 0.650

0.07 (0.00–0.28)/0.42 (0.30–0.54) 0.33 (0.21–0.46)/0.46 (0.29–0.65) 0.39 (0.26–0.54)/0.35 (0.21–0.52)

3.28 0.20 1.43

0.070 0.656 0.231

Covariance

S, W S, W

S, W S, W S, W

The categories are the consequences of dispersal for fitness (resident versus immigrant and settled versus unsettled) and the consequences of the spatial composition of kin (inhabiting a patch with versus without a mother; with versus without a sister; with versus without a brother). The predicted values with 95% CI were estimated from the model including only the main factor. The covariables season (S) and weight at release (W), are included in some of the models and are referred to in the last column (Covariance). A Bonferroni modification of the two-tailed significance level (α level=0.050) due to multiple comparisons resulted in a one-tailed α level of 0.020 for the first category and an α level of 0.012 for the second.

GUNDERSEN & ANDREASSEN: DISPERSAL IN ROOT VOLES 1363

Sandell et al. 1991; Jacquot & Vessey 1995). This critical distance has been more or less arbitrarily chosen. The patchy nature of our experimental setting facilitated two definitions of dispersal. Supported by previous radiotelemetric studies on the present strain of root voles at Evenstad research station (Berg 1995), we assumed that the interpatch distance of 5 m was sufficiently large to inhibit movement of established individuals between patches. Thus, settlement in the immigration patches was defined as true dispersal from the release patch. Furthermore, we assumed that the nonhabitat area surrounding the patches, as well as the relatively long distance from the nearest habitat patch to the edge traps (minimum 9 m), was large enough to indicate that individuals caught in the edge traps had not settled in the habitat patches (but see Wolff et al. 1996). Strictly speaking, our study included only one experimental factor from which causal inferences could be drawn, namely patch shape. Because of the highly controlled setting and the large number of replicates, however, we also consider inferences from the other focal factors analysed to be strong, although they are in a strict sense correlational. We consider the likelihood of hidden confounding effects making these inferences invalid to be low.

Predictions Patch shape We manipulated patch shape to test whether patch edges could affect dispersal rates but it proved to be an unimportant factor. Studies conducted on larger spatial scales and thus including population level processes have provided disparate results regarding edge effects on dispersal rates (for a review see Ims 1995), probably because of the different species and different spatial scales studied. With respect to spatial scale, it may matter at which stage in the dispersal process a patch border is encountered. For instance, since patches in the present study were small, voles crossed the border of the patch and the natal area simultaneously. In larger habitat patches the dispersing individuals might have traversed unfamiliar ground before reaching the patch border. Since animals typically make different movement decisions on unknown and known (natal) ground (Bell 1991), different patch emigration rates may result. Thus, with the scale limitation of the present study in mind, we suggest that patch shape is an unimportant determinant of dispersal at the spatial scale of a female’s home range core area.

Shields 1983) and that philopatric offspring become sexually inhibited by the opposite-sex parent (Dobson 1979; Brody & Armitage 1985; Holzenbein & Marchinton 1992; Wolff 1992, 1993). Contrary to our predictions based on the inbreeding avoidance hypothesis, however, dispersal was not triggered by the opposite sex kin, nor did dispersal result in a unisexual colonization of patches.

Resource competition In accordance with the resource competition hypothesis, the dispersal of daughters seemed to be triggered by maternal residency and philopatric daughters tended to become reproductively inhibited. As both daughters and mothers are limited by the same resource, food (Ostfeld 1985, 1990; Ims 1987), these results are in concordance with a strategy to avoid competition for local resources. Although female–female interactions are generally supposed to suppress female sexual maturation in voles (Kawata 1987) it has, as far as we know, never been shown that inhibition of the daughter’s reproduction may result from philopatry in the absence of the father. It is possible that the small patch size used in our study could have induced resource competition that is not usually realized on larger patches, because the daughters then have the opportunity to avoid overlapping with their mother’s home range. Indeed, earlier radiotelemetric studies of breeding females of the same root vole strain have shown that home range core areas rarely overlap (Andreassen et al. 1998). In addition, in a continuous population daughters will also experience competition from nonkin females. This could modify the cost to benefit ratio of philopatry versus dispersal and make the option of philopatry more attractive. The triggering factors for the immigration pattern of sons may also be interpreted as support for the resource competition hypothesis. The sons’ immigration rate increased with decreasing maternal body mass. If habitat quality is reflected by maternal body mass offspring from mothers with low body mass are likely to disperse in search of better habitat. The transfer of food stress from mother to offspring as a maternal effect has previously been shown to affect offspring behaviour and dispersal tendency in microtines (Andreassen & Ims 1990; Wong & Bondrup-Nielsen 1992; Bondrup-Nielsen 1993). In disagreement with our predictions derived from the resource competition, hypothesis, however, dispersal rates were not associated with litter size or season, nor did dispersal result in an even dispersion of animals. On the contrary, there was an intrasexual aggregation of siblings (see also Lambin 1994).

Inbreeding avoidance In accordance with the inbreeding avoidance hypothesis we found that dispersal was sex biased. More males than females moved long distances (i.e. they became unsettled). Furthermore, philopatric sons tended to have a lower probability of becoming sexually mature than nonphilopatric sons. These two results conform with the general mammalian dispersal pattern attributed to inbreeding avoidance, namely that males are the dispersing sex (Greenwood 1980; Dobson 1982; Horn 1983;

Synthesis The spatial status of the daughters was connected to reproductive status. Daughters that moved long distances were most often not pregnant; those that moved short distances were most often pregnant; and philopatric daughters seemed to become reproductively inhibited. To avoid inbreeding, females may be expected to disperse long distances before mating. By contrast, to avoid

1364 ANIMAL BEHAVIOUR, 56, 6

resource competition, females should move short distances to the first uncontested site, and no further, owing to the mortality risk related to travel (Murray 1967; Waser 1985; Steen 1994). If the ultimate cause of long-distance dispersal was local resource competition, the dispersing females would save time by mating before leaving the natal patch, and colonizing an empty patch as a pregnant female. Although the adult males used in our study were unrelated to the matriline, they might have been perceived as the father by those females dispersing long distances, as Microtus spp. lack kin recognition mechanisms (Ferkin 1988; Lambin & Mathers 1997). With respect to the settlement pattern, we had expected that if short-distance dispersal was connected to resource competition, females would be evenly dispersed among the three patches. Daughters tended to aggregate spatially, however. An assumption underlying the expectation of even spatial distribution of females is that the resource requirements of mothers and daughters was equal and that competition for food was symmetric. Compared with daughters, however, mothers are larger and socially dominant (Ims 1987; Irgens 1996; Moe 1997) and have larger litters (Ims 1997). Furthermore, in the present study a higher proportion of the mothers became pregnant than their daughters. Mothers might thus have had higher resource requirements or resource-holding potential than their daughters. Thus, avoidance between daughters and mothers due to resource competition would result in aggregation of daughters in patches not occupied by their mothers. The long-distance dispersal was male biased as expected according to the inbreeding avoidance hypothesis. Shortdistance dispersal in males may be related to inbreeding, as sons seemed to avoid their mother, and philopatric sons were frequently reproductively inactive and thus possibly inhibited by their mother.

Conclusion We have shown that natal dispersal in root voles structured the spatial relations among kin within matrilines and that the spatial status of postdispersal individuals was associated with individual fitness consequences. Natal dispersal seemed adaptive with elements compatible with both the inbreeding avoidance and resource competition hypotheses. Specifically, male dispersal and long-distance dispersal by females seemed to be mechanisms to avoid inbreeding, whereas short distance dispersal by females seemed to be a mechanism to avoid resource competition. Acknowledgments We thank J. Aars, H. Gundersen, R. A. Ims, R. G. Karlsen and J. A. Moe and the personnel at the animal division for field and laboratory assistence. R. A. Ims contributed during the planning and writing of the study. X. Lambin, H. Lampe, T. Storaas and L. Wauters commented on the manuscript. This study was supported financially by the Nansen endowment foundation, and is part of the

project: Habitat fragmentation: implications for population dynamics, supported by The Research Council of Norway. References Andreassen, H. P. & Ims, R. A. 1990. Responses of female greysided voles Clethrionomys rufocanus to malnutrition: a combined laboratory and field experiment. Oikos, 59, 107–114. Andreassen, H. P., Ims, R. A., Stenseth, N. C. & Yoccoz, N. G. 1993. Investigating space use by means of radiotelemetry and other methods: a methodological guide. In: The Biology of Lemmings (Ed. by N. C. Stenseth & R. A. Ims), pp. 590–618. London: Academic Press. Andreassen, H. P., Hertzberg, K. & Ims, R. A. 1998. Space use responses to habitat fragmentation and connectivity in the root vole Microtus oeconomus. Ecology, 79, 1223–1235. Bekoff, M. 1977. Mammalian dispersal and the ontogeny of individual phenotypes. American Naturalist, 111, 715–732. Bell, W. J. 1991. Searching Behaviour. London: Chapman & Hall. Bengtsson, B. O. 1978. Avoiding inbreeding: at what cost? Journal of Theoretical Biology, 73, 439–444. Berg, K. W. 1995. Space use responses of root voles (Microtus oeconomus) to a habitat fragmentation gradient. Candidatus Scientiarum thesis, University of Oslo. Bollinger, E. K., Harper, S. J. & Barrett, G. W. 1993. Inbreeding avoidance increases dispersal movements of the meadow vole. Ecology, 74, 1153–1156. Bondrup-Nielsen, S. 1993. Early malnutrition increases emigration of adult female meadow voles, Microtus pennsylvanicus. Oikos, 67, 317–320. Boonstra, R., Krebs, C. J., Gaines, M. S., Johnson, M. L. & Craine, I. T. M. 1987. Natal philopatry and breeding systems in voles (Microtus spp.). Journal of Animal Ecology, 56, 655–673. Brody, A. K. & Armitage, K. B. 1985. The effects of adult removal on dispersal of yearling yellow-bellied marmots. Canadian Journal of Zoology, 63, 2560–2564. Charnov, E. L. & Finerty, J. P. 1980. Vole population cycles: a case for kin-selection. Oecologia, 45, 1–2. Cockburn, A. 1988. Social Behavior in Fluctuating Populations. London: Croom Helm. Dobson, F. S. 1979. An experimental study of dispersal in the California squirrel. Ecology, 60, 1103–1109. Dobson, F. S. 1982. Competition for mates and predominant juvenile male dispersal in mammals. Animal Behaviour, 30, 1183– 1192. Elton, C. S. 1927. Animal Ecology. New York: MacMillan. Elton, C. S. 1930. Animal Ecology and Evolution. Oxford: Oxford University Press. Fauske, J., Andreassen, H. P. & Ims, R. A. 1997. Spatial organisation in a small population of the root vole Microtus oeconomus in a linear habitat. Acta Theriologica, 42, 79–90. Ferkin, M. H. 1988. The effect of familiarity on social interactions in meadow voles, Microtus pennsylvanicus: a laboratory and field study. Animal Behaviour, 36, 1816–1822. Forman, T. T. 1995. Land Mosaics. The Ecology of Landscapes and Regions. Cambridge: Cambridge University Press. Gaines, M. S. & McClenaghan, L. R. 1980. Dispersal in small mammals. Annual Review of Ecology and Systematics, 11, 163–196. Getz, L. L. & Hofmann, J. E. 1986. Social organization in free-living prairie voles, Microtus ochrogaster. Behavioral Ecology and Sociobiology, 18, 275–282. Gliwicz, J. 1997. Space use in the root vole: basic pattern and variability. Ecography, 20, 383–389. Green, R. H. 1966. Measurements of non-randomness in spatial distributions. Research in Population Ecology, 8, 1–7.

GUNDERSEN & ANDREASSEN: DISPERSAL IN ROOT VOLES 1365

Greenwood, P. J. 1980. Mating systems, philopatry and dispersal in birds and mammals. Animal Behaviour, 28, 1140–1162. Harper, S. J., Bollinger, E. K. & Barrett, G. W. 1993. Effects of habitat patch shape on population dynamics of meadow voles (Microtus pennsylvanicus). Journal of Mammalogy, 74, 1045– 1055. Holzenbein, S. & Marchinton, R. L. 1992. Spatial integration of maturing-male white-tailed deer into the adult population. Journal of Mammalogy, 73, 326–334. Horn, H. S. 1983. Some theories about dispersal. In: The Ecology of Animal Movement (Ed. by I. R. Swingland & P. J. Greenwood), pp. 54–62. Oxford: Clarendon Press. Hosmer, D. W. & Lemeshow, S. 1989. Applied Logistic Regression. New York: J. Wiley. Hovland, N. & Andreassen, H. P. 1995. Fluorescent powder as dye in bait for studying foraging areas in small mammals. Acta Theriologica, 40, 315–320. Hurlbert, S. H. 1984. Pseudoreplication and the design of ecological field experiments. Ecological Monographs, 54, 187–211. Ims, R. A. 1987. Responses in the spatial organization and behavior to manipulations of the food resource in the vole Clethrionomys rufocanus. Journal of Animal Ecology, 56, 485–596. Ims, R. A. 1989. Kinship and origin effects on dispersal and space sharing in Clethrionomys rufocanus. Ecology, 70, 607–616. Ims, R. A. 1990. Determinants of natal dispersal and space use in the grey-sided vole, Clethrionomys rufocanus: a combined laboratory and field experiment. Oikos, 57, 106–113. Ims, R. A. 1995. Movement patterns related to spatial structures. In: Mosaic Landscapes and Ecological Processes (Ed. by L. Hansson, L. Fahrig & G. Merriam), pp. 85–109. London: Chapman & Hall. Ims, R. A. 1997. Determinants of geographic variation in growth and reproductive traits in the root vole. Ecology, 78, 461–470. Ims, R. A., Rolstad, J. & Wegge, P. 1993. Predicting space use responses to habitat fragmentation: can voles Microtus oeconomus serve as an experimental model system (EMS) for capercaillie grouse Tetrao urogallus in boreal forest? Biological Conservation, 63, 261–268. Irgens, E. G. 1996. Effect of habitat destruction on female root vole (Microtus oeconomus) social behaviour. Candidatus Scientiarum Thesis, University of Oslo. Jacquot, J. J. & Vessey, S. H. 1995. Influence of the natal environment on dispersal of white-footed mice. Behavioral Ecology and Sociobiology, 37, 407–412. Johnson, M. L. & Gaines, M. S. 1990. Evolution of dispersal: theoretical models and empirical tests using birds and mammals. Annual Review of Ecology and Systematics, 21, 449–480. Jongman, R. H., Ter Braak, C. J. F. & von Tongeren, O. F. R. 1987. Data Analysis in Community and Landscape Ecology. Wagenengen: Pudoc. Kareiva, P. 1985. Finding and losing host plants by Phyllotreta: patch size and surrounding habitat. Ecology, 66, 1809–1816. Kawata, M. 1987. Pregnancy failure and suppression by female– female interaction in enclosed populations of the red-backed vole, Clethrionomys rufocanus bedfordiae. Behavioral Ecology and Sociobiology, 20, 89–97. Krebs, C. J. 1979. Dispersal, spacing behaviour, and genetics in relation to population fluctuations in the vole Microtus townsendii. Fortschritte der Zoologie, 25, 61–77. Krebs, C. J. 1989. Ecological Methodology. New York: Harper & Row. Lambin, X. 1994. Natal philopatry, competition for resources and inbreeding avoidance in Townsend’s voles (Microtus townsendii). Ecology, 75, 224–235.

Lambin, X. 1994. Territory acquisition and social facilitation by litter-mate Townsend’s voles (Microtus townsendii). Ethology, Ecology and Evolution, 6, 213–220. Lambin, X. & Krebs, C. J. 1991. Can changes in female relatedness influence microtine population dynamics. Oikos, 61, 126–132. Lambin, X. & Krebs, C. J. 1993. Influence of female relatedness on the demography of Townsend’s vole populations in spring. Journal of Animal Ecology, 62, 536–550. Lambin, X. & Mathers, C. 1997. Dissipation of kin discrimination in Orkney voles, Microtus arvalis orcadensis: a laboratory study. Annales Zoologici Fennici, 34, 23–30. Lambin, X., Krebs, C. J. & Scott, B. 1992. Spacing system of the tundra vole Microtus oeconomus during the breeding season in Canada’s western Arctic. Canadian Journal of Zoology, 70, 2068– 2072. Liberg, O. & von Schantz, T. 1985. Sex-biased philopatry and dispersal in birds and mammals: the Oedipus hypothesis. American Naturalist, 126, 129–135. Lidicker, W. Z., Jr. 1975. The role of dispersal in the demography of small mammals. In: Small Mammals: Their Productivity and Population Dynamics (Ed. by F. B. Golley, K. Pertrusewicz & L. Ryszkowski), pp. 103–128. London: Cambridge University Press. Lidicker, W. Z., Jr & Stenseth, N. C. 1992. To disperse or not to disperse: who does it and why? In: Animal Dispersal: Small Mammals as a Model (Ed. by N. C. Stenseth & W. Z. Lidicker, Jr), pp. 21–36. London: Chapman & Hall. Liro, A. & Szacki, J. 1987. Movement of field mouse Apodemus agrarius (Pallas) in a suburban mosaic of habitats. Oecologia, 74, 438–440. McGuire, B., Getz, L. L., Hofmann, J. E., Pizzuto, T. & Frase, B. 1993. Natal dispersal and philopatry in prairie voles (Microtus ochrogaster) in relation to population density, season, and natal social environment. Behavioral Ecology and Sociobiology, 32, 293– 302. McShea, W. J. 1990. Social tolerance and proximate mechanisms of dispersal among winter groups of meadow voles, Microtus pennsylvanicus. Animal Behaviour, 39, 346–351. Manly, B. F. J. 1991. Randomization and Monte-Carlo Methods in Biology. London: Chapman & Hall. Moe, J. A. 1997. Does social behaviour among female root voles (Microtus oeconomus) depend on familiarity? A combined laboratory and field experiment. Candidatus Scientiarum Thesis, University of Oslo. Murray, B. G., Jr. 1967. Dispersal in vertebrates. Ecology, 48, 975–978. Okubu, A. 1980. Diffusion and Ecological Problems: Mathematical Models. New York: Springer-Verlag. Ostfeld, R. S. 1985. Limiting resources and territoriality in microtine rodents. American Naturalist, 126, 1–15. Ostfeld, R. S. 1990. The ecology of territoriality in small mammals. Trends in Ecology and Evolution, 5, 411–415. Pusey, A. E. 1987. Sex-biased dispersal and inbreeding avoidance in birds and mammals. Trends in Ecology and Evolution, 2, 295–299. Sandell, M., Agrell, J., Erlinge, S. & Nelson, J. 1991. Adult philopatry and dispersal in the field vole Microtus agrestis. Oecologia, 80, 153–158. Shields, W. M. 1983. Optimal inbreeding and the evolution of philopatry. In: The Ecology of Animal Movement (Ed. by I. R. Swingland & P. J. Greenwood), pp. 132–159. Oxford: Clarendon Press. Stamps, J. A., Buechner, M. & Krishnan, V. V. 1987. The effects of edge permeability and habitat geometry on emigration from patches of habitat. American Naturalist, 129, 533–552. Steen, H. 1994. Low survival of long distance dispersers of the root vole (Microtus oeconomus). Annales Zoologici Fennici, 31, 271–274. Stenseth, N. C. & Lidicker, W. Z., Jr. 1992. Animal Dispersal: Small Mammals as a Model. London: Chapman & Hall.

1366 ANIMAL BEHAVIOUR, 56, 6

Tast, J. 1966. The root vole, Microtus oeconomus (Pallas), as an inhabitant of seasonally flooded land. Annales Zoologici Fennici, 3, 127–171. Viitala, J. 1994. Monogamy in free living Microtus oeconomus. Annales Zoologici Fennici, 31, 343–345. Waser, P. M. 1985. Does competition drive dispersal? Ecology, 66, 1170–1175. Wolff, J. O. 1992. Parents suppress reproduction and stimulate dispersal in opposite-sex juvenile white-footed mice. Nature, 359, 409–410. Wolff, J. O. 1993. What is the role of adults in mammalian juvenile dispersal? Oikos, 68, 173–176.

Wolff, J. O. 1995. Friends and strangers in vole population cycles. Oikos, 73, 411–414. Wolff, J. O. & Lidicker, W. Z., Jr 1980. Population ecology of the taiga vole, Microtus xanthognathus, in interior Alaska. Canadian Journal of Zoology, 58, 1800–1812. Wolff, J. O., Schauber, E. M. & Edge, D. 1996. Can dispersal barriers really be used to depict emigrating small mammals? Canadian Journal of Zoology, 74, 1826–1830. Wong, K. L. & Bondrup-Nielsen, S. 1992. Long-term effects of infant malnutrition on the behaviour of adult meadow voles, Microtus pennsylvanicus. Canadian Journal of Zoology, 70, 1304– 1308.