Effect of energetic constraints on distribution and winter survival of weasel males


Corresponding author. E-mail: karolzub@zbs.bialowieza.pl


1. The absolute energy needs of small animals are generally lower than those of larger animals. This should drive higher mortality of larger animals, when the environmental conditions deteriorate. However, demonstration of the effect of energy constraints on survivals proved difficult, because the range of body mass within species is generally too small to produce enough variation for studying such an effect. An opportunity for an intraspecific study comes from weasels inhabiting the Białowieża Forest (north-eastern Poland), which are characterized by a threefold variation in body mass.

2. We assumed that in summer larger weasel males are favoured by sexual selection, because they are more successful when competing for mates. We then tested whether they suffer higher mortality in winter, because they have difficulty finding sufficient food to satisfy their energy needs and/or because the additional foraging time would result in increased exposure to predation.

3. We measured daily energy expenditures (DEE) of overwintering weasel males using the doubly labelled water (DLW) technique. We constructed an energetic model predicting how individuals of different size are able to balance their energy budgets feeding on large and small prey while minimizing time spent hunting, thereby reducing their own exposure to predation.

4. The range of body mass in overwintering weasels predicted by our model corresponded very well with the distribution of prey body mass in three different habitats within our study area. Larger individuals were able to compensate for higher food requirements by using habitats with larger prey species than those available to smaller male weasels. This effectively offset the expected negative association between body mass and winter survival predicted from considerations of energy balance.

5. Our results show how energetic constraints affect body mass and spatial segregation of a species at the intra-specific level not only across large geographical ranges, but also within a relatively small area.


Body mass is closely related to major life-history traits and is often suggested to be the result of compromises between the conflicting effects of sexual and other types of natural selection (Stearns 1992; Roff 2002). Larger individuals usually have higher fecundity, but they are also more likely to die when environmental conditions deteriorate (Tilley 1968; Boag & Grant 1981; Clutton-Brock et al. 1983; Promislov & Harvey 1990; Powell & King 1997; Wikelski & Trillmich 1997). There are many studies indicating that within populations, larger individuals are more fecund, whereas selection favouring smaller body mass is seldom reported (Blanckenhorn 2000). One of the most common factors thought to be involved in trade-offs selective for body mass is the resource necessary to satisfy energy demands (Schmidt-Nielsen 1984; Peters 1993). Two critical periods pose important energetic constraints on small mammals. The first is lactation where energy demands can exceed five times the basal metabolic rate (Speakman & McQueenie 1996; Johnson, Thomson & Speakman 2001; Speakman 2008). The second period is winter in the temperate and arctic regions, because declining temperatures increase the energy costs of thermoregulation (Jackson, Trayhurn & Speakman 2001) combined with declines in food availability. For herbivorous animals, the amount of available energy declines, because of reduced primary production, and in winter, their populations generally experience the greatest mortality (Pucek et al. 1993; Jędrzejewska & Jędrzejewski 1998). Predators feeding on rodents follow the same pattern, because decreasing density of prey in winter causes higher mortality (Jędrzejewski, Jędrzejewska & Szymura 1995).

Small mammals that remain active in the winter have numerous structural, physiological and behavioural adaptations that enhance survival (Chappell 1980; Merritt 1984). This acclimatization response to winter makes the association between body mass and winter survival far from simple. Interspecific comparisons have demonstrated that the absolute energy needs of larger animals are generally higher than those of small animals (but see for example Speakman et al. 2003). On the other hand, large animals are theoretically better able to tolerate food shortages because of greater fasting endurance (Schmidt-Nielsen 1984; Lindstedt & Boyce 1985; Millar & Hickling 1990). Therefore, some Arctic rodents develop larger body mass before winter (Gower, Nagy & Stetson 1994). However, many species of rodents maintain or reduce body mass in winter (Gliwicz 1996; Speakman 2000; Aars & Ims 2002). Small mammals also reduce locomotory activity and spend more time in warm nests (Hayes & O’Connor 1999; Humphries et al. 2005; Zub et al. 2009). Lower activity may also reduce exposure to predation risk (Lima 1998).

The lack of clear empirical evidence that energy expenditures may affect winter survival is related to the typically small range of body sizes within species, which is insufficient to produce enough variation for robust statistical analysis (but see Jackson, Trayhurn & Speakman 2001; Boratyński & Koteja 2009). The weasel Mustela nivalis is one of the few species expressing enormous variation in body mass, both biogeographically and locally (King 1989; Abramov & Baryshnikov 2000; Zub et al. 2009). The mating system is polygynous and intrasexual competition for mating opportunities is high (Moors 1980; King 1989). Weasels also have very high energy demands, mainly because of their resting metabolic rates (RMR) being twice as high as in other mammals of similar size (Casey & Casey 1979; Szafrańska, Zub & Konarzewski 2007). Thus, for males, sexual selection is expected to promote large body mass (Hunt et al. 2009), while higher survival rates due to lower energy needs may favour small body mass during non-reproductive periods.

Here, we explored the effect of body mass on energy expenditure and survival of male weasels during winter in the Białowieża Forest (NE Poland). Body mass of male weasels is very consistent over long periods (exceeding 1 year) and its variation is independent of time elapsed between consecutive measurements and season (Szafrańska, Zub & Konarzewski 2007). Weasels inhabiting the Białowieża Forest are characterized by very high individual variation of male body masses, ranging in adults from 45 g to almost 150 g (Zub 2006; Zub et al. 2009). The main assumption of our study is that in a polygynous mating system, large body mass is advantageous as larger males have greater fecundity. Therefore, we focused on factors that may constrain selection for larger body mass. Daily energy expenditures (DEE) of weasel males are directly proportional to their body mass (Zub et al. 2009). Thus, we hypothesized that the higher absolute winter energy expenditures incurred by large body mass force big males to spend more time hunting, which exposes them to higher predation risks or an increased risk of starvation. We modelled the time-energy budgets of overwintering weasels males based on DEE measured with the doubly labelled water technique (DLW) (Speakman 1997; Butler et al. 2004) and estimates of daily activity patterns obtained from radiotelemetry. In weasels, the duration of one successful hunting bout is independent of body mass (Sundell 2003). We demonstrated that small males can satisfy their daily energy needs during just one, approximately 1-h long hunting activity, whereas the largest males are forced to hunt for at least twice as long. They do not simply prolong hunting bouts, but instead they increase their number. We analysed the effect of prey mass on the time-energy budget of male weasels of different body mass. Finally, we compared our predictions from the time-energy model with observed patterns of activity, variation of body mass, and mortality rates of overwintering weasels.

Materials and methods

Study area

Our investigations were carried out in the central part of the Białowieża Forest (23·86°E, 52·70°N), NE Poland. The study area encompassed three distinct habitats: (i) woodlands, (ii) open meadows and (iii) river valleys with wet sedge meadows. These habitats differ with respect to the average body mass of resident rodent species as well as the body masses of male weasels. The woodland area was dominated by oak-lime-hornbeam forest, characterized by a high proportion of old growth and a diverse composition of tree stands (Faliński 1986). The two open habitat types (meadows and river valleys) were characterized by different water regimes owing to the persistence of floodwater in spring and autumn in the river valleys.

In our study area, weasels primarily hunt rodents, and alternative prey (insectivorous mammals, birds and amphibians) have limited importance only during the summer (Jędrzejewska & Jędrzejewski 1998). In woodlands, weasels capture mainly bank voles Myodes glareolus and yellow-necked mice Apodemus flavicollis, but in the river valleys and meadows they hunt root voles Microtus oeconomus (Jędrzejewska & Jędrzejewski 1998; Zub 2004).

Ambient temperature was recorded by a meteorological station located in the central part of the study area (Białowieża) and expressed as a daily mean value from four measurements, taken every six hours.

Live trapping of rodents and weasels

In 2003–2007, weasels were live-trapped every 2 months and during the breeding season (June–September), every month. Between 20 and 40 traps were located along transects (1·0–2·0 km in length), following fences, ditches and other linear features of the landscape, which are extensively used by weasels (Jędrzejewski, Jędrzejewska & Szymura 1995; Zub, Sönnichsen & Szafrańska 2008). Each trapping session lasted for 5–7 days. We distinguished seven different trapping sites (three in the forest, two in the meadows and two in the river valley) including different number of transects within each trapping site. In the forest habitat, the first transect was located near the edge (about 500 m away), the second one in the forest interior (more than 1 km from the forest edge), and the third one in the Palace Park – a woodland area surrounded by meadows and neighbouring the village of Białowieża to the south. The Palace Park was about 800 m from the nearest continuous forest area. In the meadow habitat to the south of the forest there were three transects – two in the western part and one in the eastern part. The two trapping sites in meadows were about 1 km apart and were partly isolated by young tree plantations and bushes. The trapping sites in the river valley habitat were about 600 m apart, but completely isolated by the village of Białowieża. One transect was located in the northern part of the river valley and two in the southern part of the river valley. Data on the approximate area of each trapping site and the number of weasels captured is presented in Table 1.

Table 1.   Characteristics of trapping sites, numbers and mean body masses of male weasels captured
HabitatSiteArea sampled (km2)Number of captured weasel malesMean body mass (g)95% CI for mean body mass
ForestForest 1 (interior)1·04078·672·9–84·4
Forest 2 (edge)1·02478·671·2–86·0
Forest 3 (park)0·51373·263·1–83·3
MeadowsMeadows 1 (west)1·04891·185·9–96·3
Meadows 2 (east)0·52688·381·2–95·4
River valleyRiver 1 (north)0·557104·699·8–109·4
River 2 (south)1·03697·291·1–103·2

To analyse between-habitat differences in mean body mass of potential prey of weasels, we used data from rodent trapping collected in the same study area over 9 years (1998–2007). Details about methodology used are presented by Jędrzejewska & Jędrzejewski (1998). Each trapping site used for weasel trapping was sampled separately for rodents. Rodent species in our study area maintain constant body mass until late winter (Hansson 1990; Gliwicz 1996); thus, we could use autumn body mass to estimate values for the winter season. To compare differences in body mass of weasel males between habitats, we used our own data, supplemented with data on weasels captured since 1960 from the archives of the Mammal Research Institute (MRI) in Białowieża. All weasels from the MRI data base were captured accidentally during rodent trapping. The total sample size used in analyses encompassed 244 weasel males, including 133 individuals captured in years 2003–2007.

Measurements of energy expenditures

For this study, we used the 19 DEE measurements made under winter conditions (from 1 November until 15 April), collected from different individuals during four seasons 2004/2005 (six individuals), 2005/2006 (nine individuals) and 2007/2008 (four individuals). Thirteen of them were captured in meadows, five in the river valley and one in the forest. In the Białowieża forest, winter period is characterized by mean daily temperatures below 10 °C and occurrence of snow cover. Captured animals were immediately transported to the laboratory, anaesthetized (with a mixture of ketamine and xylazine, Seal & Kreeger 1987), marked by ear punching and sexed. To allow for the restoration of body condition, captured weasels were provided with water ad libitum and 1–2 laboratory mice per day, depending on body mass. Weasels were fitted with radio-transmitters (AVM Instrument Company, Colfax, CA, USA and Biotrack, Dorset, UK) and kept in the laboratory to habituate to the collars. After 2 days, the animals were anesthetized, weighed and injected with 0·5 mL of doubly labelled water containing approximately 60%18Oxygen and 30%2Hydrogen. The weasel was returned to the cage for one hour to allow the isotopes to equilibrate in body water (Król & Speakman 1999) and then a 100-μL blood sample was drawn from the tail to estimate initial isotope enrichments (Speakman & Racey 1987). Upon completion of all procedures, the animal was released at the location of capture. Weasels were radiotracked according to the procedure described by Jędrzejewski, Jędrzejewska & Szymura (1995). We used radio-telemetry to check the position and activity of animal every 15 min. After approximately 48 h, we attempted to recapture the animals by setting several traps at the night resting site. Recaptured weasels were transported to the laboratory where we collected a final blood sample, usually by removing the scab left from the first bleeding. Capillaries that contained the blood samples were then vacuum distilled (Nagy 1983), and water from the resulting distillate was used to produce CO2 and H2 (methods in Speakman et al. 1990 for CO2, and Król, Murphy & Speakman 2007 for H2). The isotope ratios 18O: 16O and 2H: 1H were analysed using gas source isotope ratio mass spectrometry and were converted to values of daily energy expenditure using a single pool model as recommended for animals of this size by Speakman (1993). There are several alternative approaches for the treatment of evaporative water loss in the calculation. We chose the assumption of a fixed evaporation of 25% of the water flux (equation 7·17: Speakman 1997).

Radiotracking data collected during DLW studies were supplemented with the data on the activity of 33 males from the same study population collected during the winter season (November–April), over the years 2003–2007. For every individual, we calculated a mean value (for 2–4 days) of activity time and number of activity bouts. Weasels were captured with permission from the Polish nature conservancy authorities (permits ns. DOPweg-4201-04-6/03/jr, DOPog-4201-04-43/05/aj, LKE 2003/04 and LKE 2004/06).

Fat content

To estimate body composition of weasels, we used the total body electro-conductivity (TOBEC) technique (Walsberg 1988; Wirsing, Steury & Murray 2002). We used an EM-SCAN analyser (Springfield, IL, USA), calibrated with 28 weasels found dead (fresh road and predator kills, etc.). Carcasses were dried in still air at 60 °C until they reached constant mass (on average 72 h) and then ground to uniform consistency in a meat grinder. Lipids were extracted from a 2-g portion of each ground carcass using a Soxhlet extractor and petroleum ether as a solvent. Lean body mass was calculated as total fresh mass minus lipid content. We determined the relationship between EM-SCAN values and lean body mass using least squares regression. The best fit to the untransformed data was obtained with a second-order polynomial model (r2 = 0·92, P < 0·0001). For calculating lean body mass of live animals, we used the following equation:


where LBM – lean body mass, ILBM– lean body mass index (EM-SCAN readout). Before TOBEC measurements, animals were anesthetized with a ketamine-xylazine mixture to ensure standardized positions within the instrument (Robin et al. 2002).

In the laboratory, captured weasels were fed ad libitum. To estimate possible effect of food availability on fat reserve accumulation by weasels, we divided body composition data into three categories: measurements taken immediately after animals were captured and brought to the laboratory, measurements made the day after feeding in captivity and measurements made after more than 1 day in captivity.

Model of time-energy use

Mass-specific DEE of weasels is independent of total body mass (Zub et al. 2009; Szafrańska 2010, this study). Thus, DEE of weasels during 1 day can be described by following linear equation:

image(eqn 1)

where DEE is daily energy expenditure as measured by DLW [kJ], w is body mass [g], t is activity time [h], A are energetic costs of activity (including costs of thermoregulation) [kJ (g/h)−1] and R are energetic costs at resting [kJ (g/h)−1]. In the model, R is not equivalent of RMR, because it includes also costs of thermoregulation, whereas RMR is measured in thermoneutral zone. To find the values of A and R, equations for individual weasels have to be solved for two unknown variables. To minimize errors of estimation of A and RMR, the value of following function c2 should be minimized, using the principle of least square means over all individuals:


where i denote values for i-th individual. This can be done by calculating partial derivatives of c2 with respect to A and R, setting each derivative equal to zero. The solutions of the resulting equations are:

image(eqn 2)
image(eqn 3)

We then used equation (1) and values of A and R from equations (2) and (3) to calculate theoretical daily energy expenditures of overwintering male weasels weighing from 50 to 140 g, and engaging in 0 to 5 hunting bouts a day. Confidence intervals (CI) for mean A and R values were estimated by resampling individual data using 1000 repetitions.

To analyse time-energy budget, we used activity bouts lasting 0·85 h, calculated from radiotracking data (see results). We then estimated DEE of weasels of different body mass predicted by energy budget model (equation 1) at respective numbers (from 0 to 5) of activity bouts per day (solid lines Fig. 1a,b). Next, we superimposed the values on energy gain from different numbers of prey of two sizes: small (21 g) and large (28 g, dotted lines, Fig. 1a,b) – which represents the range of body mass of potential prey of weasels in our study area. Energy gain was computed using mass-specific energy contents of small rodents (Górecki 1965) and the assimilation efficiency of 0·74, calculated for weasels (Moors 1977). By comparing predicted DEE to these values, we were able to estimate the numbers of prey of different size, needed to satisfy the energy requirements of individuals of a given body mass.

Figure 1.

 Energy expenditures of weasel males as a function of body mass. Solid lines indicate expenditures predicted by energy budget model at respective numbers (from 0 to 5) of activity bouts per day, when one activity bout is equivalent of 0·85 h. Zero activity denotes expenditures equal to R (i.e. no costs of activity). The dotted lines denote energy gain from different numbers of large prey (28 g; a) and small prey (21 g; b).

Survival rates

We used the program MARK (White & Burnham 1999) to estimate survival rates of weasel males. We used the Cormack-Jolly-Seber (CJS) model as an adequate starting point. Then we compared different CJS models comprised of differing subsets of variables from a global CJS model, using the Akaike’s Information Criterion (Burham & Anderson 2002), with or without body mass as a covariate and trapping site locations as a categorical variable. Body mass was log transformed prior to analyses and standardized, using function implemented in MARK programme. When the model was constructed in MARK we used 3 months time intervals (January–March, April–June, July–September and October–December), which are equivalent of four seasons. Encounter history was constructed based on presence or absence of an individual in each season from spring 2003 until winter 2007/2008 (in total, 20 encounter occasions). During this period, weasels were trapped regularly in all habitats over a large area; thus, we reduced the likelihood of confusing mortality with emigration.

Statistical analyses

We used a General Linear Model (anova and ancova) to estimate the effect of habitat type on mean activity time, mean number activity bouts and mean body mass of weasels, and the effect on mean body mass and density of rodents. An analysis of covariance (ancova) was used when analysing the effect of body mass and laboratory conditions on the body composition of weasels, and the effect of rodent density on the mean body mass of male weasels. For a detailed analysis of activity patterns, we used a repeated measures general linear mixed model with habitat as a fixed factor, trapping site as a random factor (nested within habitat) and rodent abundance, body mass and ambient temperature as covariates. In this model, we used individual identity as an additional random factor, to account for repeated measures for some animals.

Rodent density was included in the models as a continuous variable and expressed as number of individuals captured per 100 trap-nights. Values expressed as percentages were arcsine transformed prior to analyses. The statistical analyses were performed using the R package (The R Foundation for Statistical Computing) and Resampling Stats for Excel. Results are presented as mean ± standard error, unless otherwise noted.


Between-habitat differences in body mass of weasels and its prey

In Tables 1 and 2, we present general characteristics of trapping sites, numbers and mean body masses of male weasels. Their body masses significantly differed between the three habitats (forest, meadows and river valleys, F2,207 = 40·66, P < 0·0001) as indicated by a general linear mixed model with habitat, year and season as fixed factors and trapping site as a random factor (nested within habitat). The largest weasels occurred in river valleys (101·7 g, 95% CI: 98·0–105·5, N = 93), they were slightly smaller in meadows (90·1 g, 95% CI: 85·9–94·3, N = 74) and smallest in the forest (77·7 g, 95% CI: 73·6–81·8, N = 77). Post-hoc tests (Tukey test for unequal sample sizes) revealed that all between-habitat differences in body masses were significant (P < 0·001).

Table 2.   Body masses of male weasels captured in the Białowieża Forest in 1961–2008
HabitatSeasonnMean−95%+95%MedianLower quartileUpper quartileVarianceSD
  1. Breeding season: June–September, non-breeding season: October–May.

River valleyAll94101·8997·93105·86103·5589·00115·00374·7019·36
River valleyBreeding62105·95101·49110·41107·5095·00118·20308·2917·56
River valleyNon-breeding3294·0386·64101·4396·2082·20112·09420·8120·51

Body mass variation was also significantly affected by year (F33,207 = 1·52, P = 0·04) and season (males captured during the breeding season (June–September) vs. animals captured after the breeding season, October–May, i.e. overwintering individuals, F1,207 = 16·92, P < 0·001). There was no detectable season × habitat interaction (P = 0·58) and across habitats overwintering males were significantly smaller (Fig. 2). The effect of trapping site was not significant (F5,207 = 1·06, P = 0·38).

Figure 2.

 Effect of habitat type and season on mean body mass of male weasels. Bars indicate 95% confidence limits. Breeding season: June–September, non-breeding season: October–May. Points in the frame indicate individuals captured at very high density of weasels in meadows.

To elucidate the possible effect of rodent density on between-habitat variation of weasel body mass, we used a smaller data set, as rodent densities were monitored in our study site only from 1998 onwards. A general linear mixed model described above with rodent density added as a covariate revealed that its effect was not significant (F1,150 = 0·66; P = 0·42).

Using a similarly structured general linear model, we found significant, between-habitat differences in body mass of the potential prey of weasels (F2,529 = 36·36; P < 0·0001). The smallest rodents occurred in the forest (mean body mass 21·4 g, 95% CI: 20·6–22·4 g), and larger prey were found in meadows and river valleys (mean body mass 25·2 g, 95% CI: 23·3–27·1 g and 28·5 g, 95% CI: 27·3–29·8 g, respectively). All means were statistically different from each other at P < 0·05 (post-hoc test Tukey test for unequal sample size). In contrast, we did not detect a significant effect of rodent density on their between-habitat body mass variation (F1,529 = 0·08; P = 0·77, rodent density included as a covariate).

Fat content

Weasels kept in the laboratory for more than a day had a fat content averaging 23·4 ± 1·44% of total body mass, whereas animals measured immediately after capture or on the following day had a fat content of 9·6 ± 1·4% and 14·3 ± 1·0%, respectively. The differences among these three groups of animals were significant (ancova, F2,37 = 5·83, P = 0·006, lean body mass was not significant as a covariate, F1,36 = 0·31, P = 0·58).

The proportion of fat accumulated by weasels was weakly related to the lean body mass (general linear model, R2 = 0·11, P = 0·052). In absolute values, larger individuals accumulated on average more fat than small ones (R2 = 0·36, P = 0·0004, time spend by animals in laboratory was significant as a categorical variable, F1,35 = 5·90, P = 0·02).

Assuming that 1 g of fat contains 40 kJ of energy (Peters 1993), and using R calculated from equation (3), we evaluated theoretical fasting time of weasels of different body mass. The smallest individuals (lean body mass 40 g) could survive for 2·3 days without feeding and still retain 25% of their initial fat reserves, but the largest individuals (lean body mass 120 g) could survive 3·2 days. The mean fasting time was 2·8 ± 0·2 days. In these calculations, we assumed that only lean body mass is metabolically active.

Pattern of hunting activity

The average duration (±SE) of a single activity bout of foraging weasels in winter was 0·85 ± 0·08 h (N = 33). The total daily activity time averaged 1·62 h (±0·19 h, N = 33), which corresponded to two activity bouts. We did not detect a significant effect of habitat (F1,24 = 0·02; P = 0·89), rodent abundance (F1,24 = 0·17; P = 0·68, Fig. 3), body mass (F1,24 = 0·10; P = 0·76) or mean daily temperature (F1,24 = 0·01; P = 0·91) on mean duration of one activity bout, as indicated by repeated measure ancova model. Also the interaction between body mass and ambient temperature was not significant (F1,24 = 0·01; P = 0·92).

Figure 3.

 Relationship between abundance index of rodents (N individuals captured per 100 trap-nights) and total daily activity time and duration of one activity bout.

Using a similarly structured general linear model, we also did not find any significant effect of habitat type (F1,24 = 0·001; P = 0·97), rodent abundance (F1,24 = 0·31; P = 0·58, Fig. 3), body mass (F1,24 = 0·07; P = 0·79), temperature (F1,24 = 1·97; P = 0·17), and interaction of body mass and temperature (F1,24 = 2·92; P = 0·10) on mean activity time of overwintering male weasels.

According to our observations and data from the literature (Jędrzejewski, Jędrzejewska & Szymura 1995; Sundell 2003), weasels finished their activity soon after making a kill. Accordingly, we deduced that the typical single 0·85-h-long activity bout was the equivalent of one prey item captured and therefore, weasels on average captured two prey items daily.

Daily energy expenditures

Effect of year, habitat type and ambient temperature on DEE measured in 19 overwintering individual weasels was not statistically significant (year: F2,12 = 1·99, P = 0·18, habitat type: F2,12 = 0·57, P = 0·58, ambient temperature: F2,12 = 1·04, P = 0·33). Only body mass used as a covariate significantly influenced DEE (F1,12 = 7·66, P = 0·02). The allometric scaling exponent of DEE was 0·84 (SE = 0·24), which, for the sake of simplicity, allowed us to assume proportionality of DEE to body mass in equations (1–3). Under this assumption, the energetic cost of activity (A) calculated from equation (2) was 0·220 kJ (g/h)−1 (95% CI: 0·075–0·295 kJ (g/h)−1). R calculated from equation (3) was three times lower than A and equalled 0·070 kJ (g/h)−1 (95% CI: 0·062–0·083 kJ (g/h)−1). On the other hand, R was significantly higher than RMR of overwintering weasels measured at 30 °C under laboratory conditions (0·042 kJ (g/h)−1, 95% CI: 0·034–0·051 kJ (g/h)−1, Szafrańska, Zub & Konarzewski 2007).

Winter time-energy budget

Equation (1) fitted to DEE and activity time of individual weasels explained a significant part of their overall variation (R2 = 0·57; P = 0·001). Mean DEE calculated from equations (1–3) increased from about 100 kJ for animals of 40 g to almost 300 kJ for the largest individuals (140 g; Fig. 1a). Our model predicted that one large prey (28 g) would satisfy the energy needs (R plus costs of activity during the 0·85 h of hunting that was necessary to capture one prey item) of weasels weighing less than 60 g (Fig. 1a). Larger individuals have to capture at least two prey items, and therefore must make two hunting trips (c. 1·6 h), because of higher energy expenditure during both rest and activity. Weasels exceeding 110 g would need to consume at least three large prey items per day, as even if they remained completely inactive their R equals the energy content of almost two large rodents (Fig. 1a). Our estimates indicate that one small prey item would not satisfy the daily energy needs of weasel males of any size, because such prey contain only enough energy to fuel the R of the smallest males (Fig. 1b). Large weasels feeding on small rodents (21 g) would need to capture one more prey item per day than if they fed on larger rodents (28 g) (Fig. 1a,b).

According to our predictions, the body mass of weasels that catch two large prey items daily should be limited to about 110 g; the corresponding mass limit for two small prey items is about 90 g. This pattern matches well the mean body masses of weasel males: both body mass of prey and body mass of male weasels are largest in the river valley and smallest in the forest (Fig. 2).

Survival rates

Seasonal differences in body mass (Fig. 2) suggest that winter survival of bigger males is lower than that of smaller ones. To test this idea more rigorously, we analysed capture–recapture data on 113 male weasels captured between 2003 and 2007 (mean ± SD, 22 ± 13 individuals per year). For all these individuals, we recorded 217 encounters, i.e. at least one capture during each 3-month time interval. There were on average two encounters per individual, maximum seven encounters. In Table 3, we present the results of survival analyses. Based on the model with four seasons, we found no differences between survival probability between winter and spring, and summer and autumn. We therefore reduced survival estimates to two time periods: the breeding season (April–September) and winter (October–March). This was our best model according to AIC values (AICc = 316·77, AICc weight = 0·56, Table 3). Survival of male weasels was significantly lower in winter (43%, 95% CI: 31–56%) than in the breeding season (68%, 95% CI: 52–81%). A model with body mass included as a covariate fitted the data slightly worse (AICc = 320·28, AICc weight = 0·10), and was significantly different from the best model not incorporating body mass (delta AICc = 3·51). A model with trapping site as an additional categorical variable fitted the data much worse (AICc = 325·55, delta AICc = 8·78). The function coefficients for the model with two seasons and body mass as a covariate, indicated that the effect of body mass on survival during winter was positive (coefficient = 0·26, 95% CI: −0·30 to 0·82) and during the breeding season negative (coefficient = −0·17, 95% CI: −0·85 to 0·52). Because the 95% confidence intervals for both coefficients included zero, this effect was not significant. Thus, our analyses do not support the existence of a negative relation between body mass and winter survival.

Table 3.   Results of survival analyses using Cormack-Jolly-Seber (CJS) model in MARK. Models are ordered according to AIC, starting from the model which best fitted the data
ModelAICcdelta AICcAICc weightModel likelihoodNumber of parameters
  1. AICc, Akaike’s Information Criterion corrected for sample size; delta AICc, the difference in the value of AIC c between a given model and the model having lowest AICc; AICc weight, proportion of data variation explained by the model.

  2. Four seasons: spring, summer, autumn, winter; two seasons: breeding season (April–September) and winter (October–March). In all models, recapture rates were constant.

Survival different in two seasons316·770·0000·5601·0003
Survival different in four seasons319·252·4730·1630·2905
Survival constant319·973·2020·1130·2022
Survival different in two seasons. Body mass included as covariate320·283·5090·0970·1735
Survival constant. Body mass included as covariate321·815·0360·0450·0813
Survival constant. Body mass included as covariate and trapping site as categorical variable324·017·2410·0150·0275
Survival different in two seasons. Body mass included as covariate and trapping site as categorical variable325·558·7790·0070·0129


Observations of captive weasels and the results of fieldwork indicate that the growth of male weasels ends when they become independent, aged about 3 months (East & Lockie 1964; Heidt, Petersen & Kirkland 1968; Frank 1985). Thus, most of the individuals captured did not significantly change their body mass as was indicated by the high long-term repeatability of this trait (Szafrańska, Zub & Konarzewski 2007). We therefore surmised that body-mass related energetic constrains, which result from availability of the prey of different sizes, may affect survival and distribution of male weasels. A model based on time-energy budgets predicted that larger individuals should suffer higher mortality in winter, because they would have difficulty finding sufficient food to satisfy their energy needs and/or because the additional foraging time would result in increased exposure to predation. However, body mass did not affect survival rates of male weasels, probably because they distribute themselves by size so that the larger individuals tend to occur in the habitats where larger prey is available.

Many species of mammalian predators demonstrate significant relationships between the quality of the environment, defined as the accessibility and the quality of the food, and their body mass. In richer environments, predators attain greater body mass (Cavallini 1995; Gortázar, Travaini & Delibes 2000; Kojola & Laitala 2001); however, this relationship is usually weakly expressed in small predators hunting on relatively small prey (Vézina 1985; Larivière & Crête 1993). Only in the stoat (Mustela erminea) is there the direct dependence between the index of mean body mass of the prey (calculated as the product of body mass and proportion of the potential prey in the prey community) and the average body mass of the predator (Erlinge 1987; King 1991). In Europe, this association coincides with the geographical variation of body mass of the stoat along north–south gradient. Weasels also follow the same pattern of geographical distribution of body mass with the smallest individuals found in the north and the northeast of the continent (King 1989; Abramov & Baryshnikov 2000; Yom-Tov, Yom-Tov & Angerbjorn 2009).

The present results demonstrated that the positive relationship between prey size and body size of small mustelids also holds at a local scale if the variation of body mass of potential prey and predator is big enough. Our results and time-energy model suggest that this relationship is driven by lack of association between the time required by the predator to hunt a single prey item and body mass of prey. This independence results in energy gain per time unit being higher when predators feed on larger prey. When energy gained from the kill does not balance energy expended during the hunt, the predator is forced to search for another prey, which on average, multiplies the time spent on hunting, and therefore, exposure to predation.

In small mammals, like weasels, every activity period outside the shelter is related to a high risk of predation (Korpimäki & Norrdahl 1989). In consequence, weasels have to minimize their activity, especially in winter when low temperatures outside their nests additionally increase their energy needs (Zub et al. 2009). Some previous authors have suggested the existence of an energetic ceiling constraining the total time animals can remain active and have expressed the maximum sustained working level as a multiple of resting metabolic rates measured at thermoneutrality (Sandell 1989; Peterson, Nagy & Diamond 1990; Daan, Deerenberg & Dijkstra 1996). One mechanism potentially mediating this ceiling is a limit on the maximum capacity to dissipate body heat (Król, Murphy & Speakman 2007; Speakman & Król, unpublished). Weasels, however, expend energy at a relatively low level during winter, generally not exceeding 2 × RMR (measured at 30 °C in the laboratory, Zub et al. 2009). Therefore, we suppose that energy intake by weasels is not limited by daily activity time, but rather that the animals minimize it and in this way decrease mortality related to predation.

The estimates of fat reserves obtained using the TOBEC method are in concordance with estimates of body composition using dilution space of labelled oxygen and hydrogen. According to equation given by Pace & Rathbun (1945), relating the body water pool and fat content, in the winter the mean fat reserves of weasel are equivalent of 16·1% of total body mass (Zub 2006). The ability of weasels to accumulate large amounts of subcutaneous fat enables them to survive periods of unfavourable weather conditions, when they are not able to hunt. Previous analyses reported a very low proportion of fat accumulated by this species (King 1989), but these results might be biased because individuals in poor condition were examined. Low fat content has also been reported for other species of small- and medium-size predators (Buskirk & Harlow 1989; Schoenemann 2004). Our results demonstrated that weasels are able to build up fat reserves in a very short period, but also use them very quickly. On the other hand, data for larger predators demonstrate that they are able to accumulate substantial amounts of fat (Huot, Poulle & Crête 1995; Winstanley, Buttemer & Saunders 1999) which do not interfere with their active way of hunting, as suggested by some authors (Lindstedt & Boyce 1985).

By relying on fat reserves, weasels can reduce additional energy costs related to thermoregulation when active at low ambient temperatures. Enclosure experiments indicate that these predators may fall into negative energy balance, when forced to hunt at low ambient temperatures, even if prey availability is very high (Jędrzejewska & Jędrzejewski 1989). According to available data (Millar & Hickling 1990; Schoenemann 2004), larger individuals (within a population) should be characterized by higher fasting endurance, because they are able to accumulate more fat reserves. Our analyses suggest that also in weasels, larger individuals are able to survive longer periods without hunting, which may help to explain why winter mortality is so weakly related to body mass. Of course, absolute food requirements of larger weasels are higher, and at some point, they have to satisfy their energy needs, but larger fat reserves may allow them far more flexibility with regard to the choice of the hunting time and co-ordinate hunting with the most favourable weather conditions.

According to our energetic model, the optimal overwintering strategy is to be a small individual feeding on large prey, because it requires only one activity bout daily, whereas larger individuals always have to capture more than one prey. This conclusion leads to two fundamental questions: why are some weasels larger than others, and why are individuals of different sizes spatially separated? The answer to the first question is probably because in polygamous species larger males are usually characterized by higher fecundity (Clutton-Brock 1983). This has not been confirmed for weasels, but the enormous sexual dimorphism and lack of male involvement in raising young observed in this species are strongly suggestive that larger body mass is advantageous for weasel males during reproduction (King 1989). This also leads to strong competition among males. Weasels are not territorial and the home ranges of males often extensively overlap (King 1989; Zub et al. 2009), but regular observations of aggressive encounters between radio-tracked male weasels indicate that smaller individuals are always forced to leave areas occupied by larger ones (Zub 2006). Observations of radio-collared weasel also suggest very high site fidelity, and most of the observed animals remain in the same type of habitat over long periods (Zub, Sönnichsen & Szafrańska 2008). We therefore suggest that smaller individuals are excluded by larger ones in open habitats, where the larger prey occur, whereas higher mortality driven by energy limitation prevent larger weasels to invade forest, inhabited by smaller species of rodents. Moreover, the open forest floor gives less protection against predators than dense vegetation in grasslands, which increases predation risk for larger individuals forced to hunt smaller prey during additional hunting bouts. This additionally amplifies the effect of spatial segregation of small and large weasels. Such a mechanism of spatial segregation explains the persistence of an enormous body mass range and is in agreement with high narrow sense heritability of body mass in the study population (K. Zub, S. Piertney, P. A. Szafrańska & M. Konarzewski, unpublished).

Further evidence supporting the suggestion that availability of larger prey promotes larger body mass in the weasel were provided by studies in Belarus, where introduction of the American mink (Mustela vison) and subsequent decline in populations of the stoat (the main competitor of the weasel) led to an increase in the mean body mass of weasels in most localities (Sidorovich & Solovej 2007). A mechanism potentially explaining this phenomenon is that weasels were able to exploit a higher availability of large prey (mainly the water vole Arvicola terrestris) in river valleys, which were previously dominated by stoat (Sidorovich & Solovej 2007).

A significant between-season variation in body mass found across habitats (Fig. 2) indirectly indicates that larger individuals may suffer higher winter mortality. However, spatial segregation of male weasels according to body mass reported here most likely weakens the effect of body mass of prey in the time-energy budget modelled here. We suggest that this accounts for the lack of significance of body mass in our analysis of mortality rates.

Our results show how body mass can be a key factor influencing the ecology of a species at the intra-specific level not only across large geographical ranges, but also within a relatively small area. According to our time-energy use model, large weasels must spend more time active per day than small individuals, if they both feed on prey of the same size. However, because large weasels occur in habitats where large prey is available and the distribution of small weasels is related to small prey, both groups of predators spend a similar amount of time hunting. Thus, spatial segregation of animals by body mass facilitates differential exploitation of their rodent prey, which also varies spatially in body mass. This undermines the expected association between body mass and survival predicted from considerations of energy balance. Selective processes on body mass during winter are likely to be more complex than the energetic models predict.


We are grateful to M. Chappell and Jeremy B. Searle for criticism and valuable comments. We would like to thank our numerous students for help in weasel trapping and technical assistance. This study was supported by the Polish Committee for Scientific Research (KBN) grants 3 P04F 05125 to K. Z. and 2 P04F 01329 to P. A. S.