We have studied artificial bank vole populations in large outdoor enclosures (0.25ha) in Central Finland over many years. Voles were clearly territorial and density-limited in the enclosed system we used, similar to what is reported from wild populations [21
]. In the experimental populations, 3-4 females were breeding simultaneously, and we regularly started our experiments with populations of 5 females to buffer predation and natural mortality. To evaluate if here that breeding suppression is evident also in our enclosed system, we compiled a data base, consisting of 38 un-manipulated, experimental populations from different studies in the years 1997-2000. These populations, because they served as controls for various experimental treatments, did not undergo any kind of experimental treatment during the respective study periods.
In all populations of these earlier studies, 5 laboratory born females, descending from a colony of wild captured bank voles from the region were introduced to the enclosures. This resulted in 4.1 ± 1.2 (mean ± SD) potential breeders. In those of the studies that were conducted during summer (n = 27 populations) the females were adult and multiparous [5
] but not pregnant at the release to the enclosures. Three males were introduced 2-3 days after the females and remained in the enclosures for at least 1 week, but often as long as the females, depending on study design. Breeding and survival was monitored for one pregnancy cycle (= 3 weeks). In those of the studies which were carried over winter to monitor the onset of breeding and the first reproduction in spring (n = 11 populations) [25
] young immature females and males were introduced into the enclosures in October. Breeding rates of survivors were monitored in spring.
An experiment with increasing densities
All our earlier studies covered a density range of 8-20 females per hectare (i.e. 2-5 females/enclosure), but we had no information on breeding suppression, interaction and physiological responses in higher densities. We therefore conducted an experiment during the summers 2001 and 2002 using 12 enclosed experimental populations. We repeated, doubled and tripled the above described, initial density of 5 adult females. We used two enclosures for each density treatment in each of the summers, resulting in four replicate populations for 5, 10 and 15 females per enclosures, resembling 20, 40 and 60 females/hectare, which are comparatively high densities for potentially breeding females in open field populations [25
]. Females were transferred to the enclosures at day 0 of the experiment (see experimental schedule, Table ). Three, 6 and 9 males, respectively, were introduced in each population after 4 days of female habituation to ensure comparable operational sex ratios, and to produce synchronized pregnancies. Experimental voles were offspring from a permanent laboratory colony. Most were born during the preceding autumn and had over-wintered in the laboratory, as the majority of breeding females in early and mid summer in wild populations. All females had successfully bred in the laboratory and weaned litters. Females were not pregnant at the beginning of the experiment. At transfer to the enclosures, experimental populations consisted of non-related females of similar age composition among populations. Animals were ear-tagged for individual identification. They were removed from the enclosures shortly before giving birth (20 days ± 2 days length of pregnancy, compare Table ) and were returned to single cages. We measured survival rates, pregnancy rates, birth dates of litters and litter sizes.
Experimental schedule for 12 populations in two summers.
Towards the end of the field period (Table , day 18-22 after release to the field) we conducted live-trapping to estimate space use. Traps were set in the evening of the 18th
experimental day, controlled 3-times per day, and opened on the 22nd
day in the morning. The trap stations in the enclosures were permanently installed, well visited, and attractive sites for the animals with regular provisioning of bait during trapping. We therefore assumed that captures of several animals at the same location (not necessarily at the same time) were indicative for potential interaction among these individuals. Since we had no information on aggressive interactions, we used the number of exclusive trap locations, i.e. locations where exclusively only one female was captured, as an indicator of exclusive space without interaction. For territorially breeding bank vole females, exclusive space is an important prerequisite for breeding [43
During the first summer, physiological estimators of density stress were obtained by measuring faecal glucocorticoid metabolites (FGM), a non-invasive and non-terminal measurement. We followed a sampling, extraction and analysis protocol of Harper and Ausstad [44
] and slightly modified it as described in detail in our paper [45
]. In short, samples were boiled in ethanol and assayed with a commercial kit (ImmuChem Double Antibody Corticosterone 125
I RIA, MP Biomedicals, CA, USA) intended for the analysis of plasma and extracted urinary samples of rats and mice. Faeces of bank vole females was collected once a week (up to four measurements per female, Table ) and sampled during morning hours to minimize variation due to daily FGM fluctuations [46
]. Hormone metabolites from stressful events show in the faeces of rodents with similar body size and diet after 4-6 h [46
]. In order to sample only the pre-trapping stress hormone levels, we collected samples at latest 3 h after setting the trap, i.e. animals had been captured less than 3 h ago. Between the last space trapping (open traps 22nd
experimental day, morning) and the third FGM sampling (set traps 23rd
day, morning) elapsed a period of 24 h without trapping or handling of animals.
Comparison of FGM concentration sampled from the individuals before the onset of the field phase (still in single cages) showed, that the method a) reflects stressful events and b) reflects individual variation. a) Shortly after release to the enclosures, when the environment physical and social environment was unknown to the cage-bred females, FGM concentrations were on average 5 fold of the individual values sampled in the laboratory (laboratory: 1.4 ± 1.6, after release: 5.9 ± 8.3 ng FGM/mg faeces, paired t-test, t = -3.5, n = 37, p = 0.001). b) Individuals with relatively high FGM levels in the laboratory had also relative high FGM levels shortly after release (Pearson's rho = 0.410, n = 38, p = 0.011).
Density effects in breeding performance, spacing behaviour and physiology were analysed on the population level, using enclosure rates or enclosure means. This was done to avoid pseudo replication [48
]. We used the number of surviving females (as indicated during the space trapping) instead of the initial density treatment, because treatments later in the experiment overlapped: 3-6 females in the low density treatment, 7-11 in the double, and 10-13 in the triple density treatment. The use of a gradient instead of the initial treatment seemed therefore more appropriate. Effects of density and year were investigated with an analysis of covariance (ANCOVA, density as covariate, year as factor). Since neither year nor an interaction of year and density turned out to be significant in any of the tested variables, we here present the effects of density as regression models. We tested for both linear models (y = ax+b) and/or an inverse models (y = a/x+b). For significant regression models we give coefficients and constants in Table . The constant b in the inverse model indicates a threshold value for indefinitely high densities. All statistics were computed with SPSS (Version 17, SPSS Inc, Chicago Illinois).
Regression models for the effect of density (number of females per enclosure) on variables of reproduction, space use and physiology in enclosure experiments on bank voles.
The date reported here stems from several studies, which all were conducted under approval of the Committee for Animal Experimentation at the University of Jyväskylä. In all studies the set-up was such, that it caused no harm to the individuals of the experimental populations. The sampling for measurements of stress levels was carried out non-invasively by monitoring hormones in the faeces. The permission number for the last experiment was 34/31.5.2004, provided by the ethical committee named above.