Bergmann’s rule in amphibians: combining demographic and ecological parameters to explain body size variation among populations in the common toad Bufo bufo

Study species and study sites

The Common toad, B. bufo, has a wide distribution, from northwest Africa, over all of Europe (except Ireland and some Mediterranean islands) to the District of Irkutsk (108°30′E) in Russia (Frost 2008). In Europe, it occupies both lowland and mountainous habitats (Borkin and Veith 1997). Demographic parameters from populations of various parts of the European range are available (e.g. Hemelaar 1988; Reading 1991, 2007; Grossenbacher 2002), but data from southern part of the continent are still fragmentary (e.g. Cvetković et al. 2003, 2005).

The study sites were the small artificial lake Trešnja (44°36′26.6″N, 20°34′14″E, altitude 222 m a.s.l.) and the pond near Zuce village (44°40′55.9″N, 20°33′7.4″E, altitude 240 m a.s.l.). 240 adult toads (Trešnja, n = 148; Zuce, n = 92) were collected around the breeding sites from 2001 to 2003. Newly metamorphosed froglets (n = 25) were caught near water at Trešnja site in June 2004. (Detailed descriptions of sites and sampling procedures are given in Tomaševic et al. 2008). The impact of the sampling was expected to be small because minimum estimates of adult population sizes were 1396 and 1948 for Trešnja and Zuce respectively (J. Crnobrnja-Isailović, unpublished data).

Body size and associated life history traits

Body size of adults (sexed by the external characteristics and the presence of gonads) and newly metamorphosed froglets was measured as the length from snout to vent to the nearest 0.1 mm using dial caliper. Sexual size dimorphism was quantified by Sexual Dimorphic Index (Lovich and Gibbons 1992): SDI = (mean length of larger sex/mean length of smaller sex) − 1, with the result defined as positive when females are the larger sex.

Age was assessed by skeletochronology (e.g. Miaud 1992; Cvetković et al. 2005). As endosteal resorption can affect the accuracy by removing the first and sometimes even the second line of arrested growth (LAG) (e.g. Frétey and Le Garff 1996), the degree of resorption was estimated by osteometrical analysis (Tomaševic et al. 2008; Sagor et al. 1998).

The average annual body growth of toads after metamorphosis can be described by the von Bertalanffy (1938) growth equation:

where t = number of growing seasons experienced (age); t₀ = age at metamorphosis (proportion of the growing season already elapsed at metamorphosis); S_t = average body size after having experienced t growing seasons; S_m = average maximal body size; S₀ = average body size at metamorphosis; K = growth coefficient, defining the shape of the growth curve.

Growth rates at metamorphosis (R_init) and at the minimum age of maturation (R_mat) were calculated using equation for age specific growth rate, derived from the von Bertalanffy equation: r = dS/dt = K (S_m–S_t), where S_t = S₀ and S_mat = average body size at maturation. The von Bertalanffy growth model was fitted to the average growth curves using the least square procedure.

The average number of growing seasons to be experienced before S_mat is reached is the average minimum age of maturation (A_mat). The potential reproductive lifespan (PR) was defined as the difference between maximal and minimal recorded age of breeding adults; it represents the number of potential reproductive seasons.

Literature study and climatic variables

We compiled our field results and literature data on body size, age and growth patterns from 11 populations of B. bufo (Gittins et al. 1980, 1982; Hemelaar 1988; Kuhn 1994; Frétey and Le Garff 1996; Schabetsberger et al. 2000). These references were selected because they were comparable with respect to variables analysed and applied methods (e.g. age assessed by skeletochronology), given that populations were widely scattered geographically.

Climatic data (annual and monthly mean, minimum and maximum temperature, annual and monthly precipitations) were obtained from the Woldclim data set (http://www.worldclim.org), at a resolution of 2.5 × 2.5 min (Hijmans et al. 2005). The lengths of B. bufo activity period (AP) vary according to latitude and altitude. These AP were available in or were estimated from data in the original papers (Table 1). We recalculated the mean values of climatic variables for the specific months of the AP recorded in each of studied populations.

Statistical analyses

We used the analysis of covariance (ancova) to evaluate the variation of activity period and temperature with latitude and altitude. Variations in adult body size were also analysed with ancova. We included sex as a fixed factor and latitude, altitude, AP, temperature and age as covariates. In the first analysis, we used geographic features (altitude and latitude) as independent variables, to evaluate body size geographic variation along both gradients. Subsequently, we used climatic parameters (temperature and AP) as independent variables. As climatic parameters strongly depend on geographic features (see Results), this analysis was therefore performed a posteriori, to evaluate how the climatic features mediate the effect of geographic parameters.

Similarly, we performed multivariate analysis of covariance (mancova) to evaluate the effects of altitude and latitude on age parameters, as well as the effects of temperature and AP on the same parameters. Multivariate analyses were followed by univariate analyses to evaluate the effect of independent variables on each dependent parameter.

In all parametric tests, residuals were normally distributed (Kolmogorov–Smirnov tests, all p > 0.1). We used the Variance Inflation Factor (VIF) to evaluate issues because of collinearity among independent variables. Values of VIF > 5 indicate that collinearity can cause issues in the results of the analyses (Bowerman and O’Connell 1990). Statistical tests were performed using statistical packages Statistica v. 6 (Statsoft Inc., Tulsa, OK, USA) and spss 13.0. (SPSS Inc., Chicago, IL, USA).

Field study

Body size and age structure

Body size and age data for Trešnja and Zuce populations are presented in Table 1 (Trešnja = Serbia 2; Zuce = Serbia 1). Males were significantly smaller than females in both populations (t-test, p < 0.0001). The SDI was 0.380 and 0.368 for Trešnja and Zuce populations respectively. Body size of metamorphosed froglets averaged 8.6 ± 0.1 mm (range: 7.2–9.8 mm).

Skeletochronological analysis was performed on 187 individuals. LAGs in phalangeal cross sections were sharp and easily distinguishable. Endosteal resorption was often present, but did not affect the accuracy of age estimation (Tomaševic et al. 2008). The youngest age at which toads came to breed in both populations was 3 years for males and 4 years for females (Fig. 1). The oldest recorded ages were 9 and 10 years for males and 11 and 9 years for females, in Trešnja and Zuce populations respectively. Age distributions did not differ significantly between populations, neither for females nor for males (Fig. 1, Kolmogorov–Smirnov two-sample test, p > 0.10).

Body length and age were significantly correlated only in males from population Trešnja: r_s = 0.385, p = 0.002 (Trešnja females: r_s = 0.206, p = 0.16; Zuce males: r_s = −0.037, p = 0.84; Zuce females: r_s = 0.056, p = 0.71).

Patterns of growth

The growth of Common toad was well described by the von Bertalanffy growth model in Trešnja population (growth coefficient for Zuce population was not computed because of insufficient number of metamorphosed froglets found at this site). Females had a larger asymptotic size (S_m = 95.4) than males (S_m = 70.9), but the growth coefficients K were similar (males: 0.79 ± 0.11; females: 0.74 ± 0.08).

The growth rates were notably higher at metamorphosis (R_init) than at maturation (R_mat): for males R_init = 49.32 and R_mat = 8.55; for females R_init = 64.44 and R_mat = 12.54. In other words, R_mat is reduced to about 17% and 19% of the initial growth rates in males and females respectively. The calculated values of A_mat (2.5 years for males and 2.7 years for females) were somewhat lower than the recorded age of the youngest adults captured in both analysed populations.

Minimum recorded body size of breeding individuals was 55.9 mm and 60.4 mm in males and 72.2 mm and 78.3 mm in females, for Trešnja and Zuce populations respectively.

Literature survey

Climatic variables and the length of AP

The locations of studied populations (n = 13) are given in Table 1 and Fig. 2.

First, we evaluated the relationship among climatic variables: the maximum (T_max) and mean temperature (T_mean) during the AP were positively correlated (r_s = 0.92, p < 0.001). On the other hand, the minimum temperature (T_min) during the AP was not correlated with either T_max or T_mean (r_s = 0.07, p = 0.81 and r_s = 0.25, p = 0.41). T_min was similar in all populations, suggesting that about 3°C was the threshold minimum value over which the Common toad activity can start. The annual precipitation was correlated with the mean monthly precipitation during the AP (r_s = 0.97, p < 0.001). The mean monthly precipitation was strongly correlated with T_mean (r_s = −0.79, p = 0.001).

The length of AP at each location was strongly correlated with the mean annual temperature (r_s = 0.92, p < 0.001) and with the mean temperature during the AP (r_s = 0.675, p = 0.01). The AP in 13 studied populations varied from 12 to 33 weeks and T_mean during this period varied from 6.6 to 15.6°C (Table 1). Altitudes and latitudes of populations were not correlated (r_s = −0.44, p = 0.13). The ancova showed that AP was significantly shorter in high altitude and high latitude populations (latitude, F_1,10 = 26.45, p = 0.0004; altitude, F_1,10 = 73.30, p < 0.0001). Similarly, T_mean significantly decreased as latitude and altitude increased (ancova: latitude, F_1,10 = 13.41, p = 0.004, altitude, F_1,10 = 29.76, p = 0.0003).

Variation in body size with latitude, altitude, T_mean and AP

An overview of life history traits for 13 European B. bufo populations is given in Table 1. Mean adult body size varied from 52.8 to 72.4 mm and from 64.7 to 96.3 mm in males and females respectively. Body size was the parameter with the lowest coefficient of variation (CV = 9.7 and 12.3 %, in males and females respectively). Mean female body size was always larger than mean male size, the SDI ranging from 0.161 to 0.380 (CV = 27.9%).

Adult body size decreased as the latitude increased (ancova, F_1,12 = 10.16, p = 0.005; Fig. 3), but did not show a similar trend with respect to altitude (F_1,12 = 0.223, p = 0.642; Fig. 3). The significant sex effect confirmed that females were larger than males in all populations (F_1,12 = 25.68, p < 0.0001). Finally, the age effect was not significant (F_1,12 = 0.543, p = 0.47), and thus the observed latitudinal pattern of body size variation cannot be explained by a latitudinal variation in population age structure.

The analysis of body size variation with AP and T_mean (Fig. 3) showed that body size increased as T_mean increased (ancova, F_1,12 = 5.08, p = 0.036), but did not vary significantly with AP (F_1,12 = 0.14, p = 0.711).

The SDI did not vary significantly with latitude or altitude (ancova, p ≥ 0.15 for both variables), nor with AP or T_mean (ancova, p > 0.06 for both variables).

Patterns of variation with latitude

Controversial results for latitudinal body size variation in amphibians are available (Ashton 2002b; Olalla-Tárraga and Rodríguez 2007; Adams and Church 2008). Among Bufonidae, four of five species showed a converse Bergmann cline: Bufo viridis (Nevo 1972; Castellano and Giacoma 1998), B. calamita (Leskovar et al. 2006), B. hemiophrys (Eaton et al. 2005) and B. woodhousii (Kellner and Green 1995). In B. bufo, the pioneering study of Hemelaar (1988), which took into account the age of the individuals, showed that older age and larger maximal size where reached at northern than at southern latitudes. The exception was the southernmost population (43.47°N, France), and Hemelaar (1988) supposed it might belong to a subspecies B. bufo spinosus. However, more recent studies found little support for genetic differentiation of B. bufo spinosus (Lüscher et al. 2001; Kutrup et al. 2006).

Our analysis, which included 13 populations, did not confirm Hemelaar’s (1988) results: the Common toad exhibits a converse Bergmann-type response to latitude (Fig. 3), as observed in other Bufo species. Both sexes follow the same trend and, although the largest values were observed in the southernmost populations, the sexual size dimorphism did not vary significantly with latitude.

The first question to be asked is if the studied gradients are representative of the species range. Insufficient sampling of gradients may lead to erroneous conclusions, e.g. of simple trends where complex patterns exist (Gaston et al. 2008). Considering continental Europe, B. bufo is present from Finland (about 68°N) to southern Spain and Greece (about 37°N). Our study included populations from 43.47°N to 63.38°N (a 2240 km gradient, Fig. 2), thus representing the largest part of this range. Testing Bergmann’s rule with Rana temporaria along a 1600 km latitudinal gradient across Scandinavia, Laugen et al. (2005) found that body size decreased in the northernmost populations (65–67°N). In B. bufo, data are scarce for high latitudes, but mean body size of males was 65.8 mm (n = 47, SE = 0.82) at 64.10°N (J. Merilä, unpublished data), which is higher than the size observed in Norway at 63.38°N (Table 1). The Common toad clearly exhibits a converse Bergmann cline along the latitudinal gradient, but more data are needed to verify this trend at high latitudes.

Despite the body size variation along the latitudinal gradient, age (at maturity, mean age and longevity) did not vary significantly and growth did not differ among populations. A similar pattern was observed in R. temporaria (Laugen et al. 2005) and Rana septentrionalis (Leclair and Laurin 1996).

Patterns of variation with altitude

Bufo bufo from high altitude grew slower, and tended to reach older age and larger maximal size than lowland populations (Hemelaar 1988). Our review included populations from 15 to 1850 m a.s.l., rather close to the altitudinal limit (2300 m in the Alps, K. Grossenbacher, pers. comm.). Body size of both sexes did not vary significantly with altitude up to 1850 m; this observation was confirmed by the size of adults at 2110 and 2180 m, which was similar to the size of lowland individuals (K. Grossenbacher, unpublished data). Minimum and mean adult age significantly increased with altitude (this study), and Grossenbacher (2002) showed that the main effect of altitude on B. bufo life history traits was a strong influence on age (i.e. increased longevity). As adult body size did not increase with altitude while age did, growth was slowed down in highland populations: a low value of K (which decreased with increasing altitude) is associated with delayed maturation. Such strong delay in age at maturity, increased longevity and decrease of growth with altitude was observed in other amphibians as well (Tilley 1980; Berven 1982; Miaud et al. 1999, 2000).

Mechanisms

Several factors (i.e. climatic conditions, trophic resources, metabolic properties, interspecific competition and predator–prey interactions) may explain differences in body length and age among populations. Since these causes are not mutually exclusive, it is difficult to determine their individual influence on the optimal size for species or for particular location (Kozlowski 1992). Also, considerable variation in body size could be observed between neighbouring populations of amphibians (Reading 1988; Bruce and Hairston 1990; Augert and Joly 1993; Miaud et al. 2001).

Our study combined the analysis of geographic gradients with the study of biologically relevant climatic variables, and of demographic features of populations. Despite complex causes of size variation, our approach identified mechanisms that could influence the observed pattern of geographic variation – by coping with temperature conditions and length of activity period.

How can we explain the pattern of body size variation over the latitudinal gradient observed in B. bufo? The pattern of adult size variation could result directly from variation in population age structure, older individuals being larger. This was not the case with B. bufo, since age did not significantly increase with latitude. Variation in size among individuals of the same age may result from variation in size at metamorphosis (Berven 1990; Miaud et al. 1999; Morrison et al. 2004), differences in growth rates before maturity (Halliday and Verrell 1988; Ryser 1988; Miaud et al. 2000), and following maturity (Ryser 1989; Smirina 1994; Kellner and Green 1995).

Size at metamorphosis can vary with development temperature and a lower development temperature resulted in larger body size in numerous ectotherms (the temperature rule, Atkinson 1994) and, specifically, in amphibians (Laugen et al. 2005; Ficetola and De Bernardi 2006). In R. temporaria, higher developmental rates of northern tadpoles were promoted by selection stemming from seasonal time constraints rather than from low ambient temperature per se (Laugen et al. 2003). Data on size at metamorphosis among B. bufo populations are lacking and how adult size can be influenced by size at metamorphosis remains to be tested. However, if the recorded T_mean reflects mean water temperature of developing tadpoles, this temperature decreased as the latitude increased (this study) and the temperature rule cannot explain the observed correlative adult body size decrease.

A higher growth rate before maturity can lead to a larger size at the same age. The growth coefficient decreased as altitude increased and the same, although non-significant trend was observed with latitude. Moreover, this growth parameter increased as both T_mean and AP increased. Growth variation can, thus, partly explain adult body size variation among B. bufo populations.

It is clear that, generally, body size might also be strongly tied to environmental factors other than temperature or AP duration, such as precipitation and humidity (Ashton 2002b), or food availability (Miaud et al. 2001; Chown and Klok 2003; Meiri et al. 2007). The strong correlation between precipitation and temperature makes it difficult to perform analyses including the effects of each factor. In the newt Taricha granulosa, adults were larger in cooler, drier areas (Nussbaum and Brodie 1971), the growth of the salamander Salamandra lanzai strongly depended on rain levels (Miaud et al. 2001), and the toad Bufo viridis was larger in warmer, drier areas (Castellano and Giacoma 1998). Moisture in the environment can be extremely important, larger individuals having greater desiccation tolerance because of the relative decrease in surface area (Duellman and Trueb 1994). However, there was no significant correlation between B. bufo body size and the amount of precipitation (mean monthly precipitation during the AP, range 56–167 mm, males: r = 0.23, p = 0.50 and females: r = 0.01, p = 0.96).

The impact of ecological conditions along the altitudinal gradient differed from that observed along the latitudinal gradient: body size did not vary, but age of adulthood was strongly delayed as altitude increased, i.e. the growth coefficient decreased (flattened growth curve). Both AP and T_mean significantly decreased with increasing altitude, as well as with latitude. The difference in life history traits along these two gradients indicates that global ecological measures such as T_mean and AP are not sufficient to fully reflect differences between high latitudes and altitudes relevant for population biology. Ecological conditions obviously differ between high latitudes and altitudes (Bliss 1956): the amount of heat accumulated during the growing season is higher in temperate alpine environment than in the arctic, local contrasts (e.g. difference between exposed and non-exposed slopes) are lower in the arctic, with less incident energy and lower evapotranspiration (Billings 1973; Körner 2003).

Our analysis revealed that adult body size increased with T_mean during the active season, while age parameters were linked to the length of the activity season rather than to T_mean. This complex pattern can arise because amphibian body size depends on proximate factors acting on different phases of the life cycle (from eggs via maternal effects, tadpoles, to juvenile and adult life), and optimal factors (e.g. thermal optima for growth efficiency and growth rate) can vary as individuals grow (Angilletta and Dunham 2003). Moreover, such temperature thresholds can be different for development rate and for growth rate (Walters and Hassall 2006). Finally, adult body size is often correlated with fitness through fecundity and mating success, maturation depending on ecological conditions mediated by seasonal time constraints (Cabinita and Atkinson 2006).

Relations with other life history traits

With respect to patterns for allocating resources over the lifetime of amphibians (Bernardo 1994), B. bufo exhibited a typical S-shaped growth curve, i.e. decreased, but continuous growth after maturation. Growth and maturation are under the constraint of an annual environmental cycle, and an optimum growth rate would be a compromise between the benefit of attaining particular size in a given time and the cost incurred for rapid growth per se (Sibly and Atkinson 1994). This compromise would be observed as a trade-off between growth rate and other fitness traits (e.g. fecundity and juvenile mortality). Unfortunately, data on life history traits other than age and body size are sparse in B. bufo, as in many other amphibian species, and Bergmann’s clines have been rarely explained in the context of life-history traits so far.

Sexual size dimorphism in many amphibians follows the pattern of many other lineages of ectothermic vertebrates, i.e. female being larger than males, pattern usually explained by fecundity selection (Kupfer 2007). If latitude and altitude via correlated ecological conditions constrain growth, the effect could be larger on females that have to invest in fecundity more than males. Females were larger than males in all populations, and the sexual size dimorphism did not vary with latitude, altitude, AP or T_mean. There was insufficient data to analyse fecundity and egg size variation along latitudinal and altitudinal gradients. The number of eggs per female seems to be higher in southern populations (mean fecundity was more than 7000 egg/female/year (range: 2118–15050) in Serbia, Tomaševic et al. 2008), compared with Great Britain (range 425–4796, Gittins et al. 1984; 700–5000, Reading 1986), the Netherlands (1859–6305, Van Gelder 1995) and Germany (750–8100, Kuhn 1994), but comparisons must be treated with caution because of heterogeneity of studies. Fecundity variation probably results directly from female size variation, since body size is often associated with increased fecundity in amphibians (e.g. Gibbons and McCarthy 1986).

Another way to limit investment in fecundity is to reproduce, for example, biannually. There is strong evidence of intermittent breeding in female Common toads (e.g. Kuhn 1994; Schmidt et al. 2002; Frétey et al. 2004; K. Grossenbacher, pers. obs.), but it is not clear whether breeding frequency is lower in e.g. high altitude or latitude populations or in females compared with males. If these patterns exist, losing reproductive seasons with biannual reproduction can easily be counterbalanced by the strong increase in longevity in highland populations (Grossenbacher 2002; this study).

The question whether amphibians follow Bergmann’s rule remains particularly interesting because it requires identification of proximate causes acting on species with complex life cycles. If the rule appears to be a general trend for anurans when comparing between species (Lindsey 1966; Olalla-Tárraga and Rodríguez 2007), it is no longer the case when comparing body size variation within species (Ashton 2002b; this study). The relationship between intra- and interspecific patterns of spatial variation appears as one of the major open questions in the field (Gaston et al. 2008). Although it has often been assumed that patterns at both levels are similar (implying common underlying mechanisms), the differences may exist, as our study confirmed.

At the interspecific level, mean body size of anurans was negatively correlated with evapotranspiration, supporting the heat balance hypothesis (greater thermoregulatory abilities allow anurans to reach larger sizes in low energy areas, Olalla-Tárraga and Rodríguez 2007). As a converse pattern was clearly observed among populations of the Common toad, this hypothesis does not hold for intraspecific level. Studies are needed now that examine spatial patterns in different types of traits and their interactions (Gaston et al. 2008). The significant relation between B. bufo body size and mean temperature and duration of the activity period (and the varying patterns observed between latitude and altitude) prompts us to select precise ecological variables acting on each phase of the life cycle to infer causes of body size variation in anurans along gradients.