Long-term study shows that increasing body size in response to warmer summers is associated with a higher mortality risk in a long-lived bat species

Abstract

Change in body size is one of the universal responses to global warming, with most species becoming smaller. While small size in most species corresponds to low individual fitness, small species typically show high population growth rates in cross-species comparisons. It is unclear, therefore, how climate-induced changes in body size ultimately affect population persistence. Unravelling the relationship between body size, ambient temperature and individual survival is especially important for the conservation of endangered long-lived mammals such as bats. Using an individual-based 24-year dataset from four free-ranging Bechstein's bat colonies (Myotis bechsteinii), we show for the first time a link between warmer summer temperatures, larger body sizes and increased mortality risk. Our data reveal a crucial time window in June–July, when juveniles grow to larger body sizes in warmer conditions. Body size is also affected by colony size, with larger colonies raising larger offspring. At the same time, larger bats have higher mortality risks throughout their lives. Our results highlight the importance of understanding the link between warmer weather and body size as a fitness-relevant trait for predicting species-specific extinction risks as consequences of global warming.

1. Introduction

Many animal species respond to global warming with changes in body size [1]. As body size is a critical trait in many species, deviations from an optimal body size can impact population persistence [2–4]. Within species, the size of individuals can influence population growth via reproduction and survival, with larger individuals typically having higher fitness [5,6]. By contrast, in inter-species comparisons, it is the large species that are typically characterized by long generation times, and, consequently, low population growth rates and low resilience to disturbance [7,8]. It is unclear, therefore, how changes in body size at the population level translate into population growth rate. Most of the documented changes in body size in response to global warming pertain to ‘global shrinking’, i.e. a reduction in body size, as predicted by Bergmann's rule in endotherms [1]. By contrast, increases in body size at the population level are rare, particularly in mammals. One exception are alpine marmots, a social and hibernating mammalian species, which show increases in body size and population growth in response to climate warming [9]. To our knowledge, however, no studies have yet reported negative fitness effects of increasing body size in response to warmer weather.

Bats have extraordinarily long lifespans and low annual reproductive output compared to other mammals of similarly small size [10]. Being distinctively on the slow side of the slow–fast life-history continuum, bats are expected to face a high extinction risk if their populations decline [11]. Indeed, bats are of global conservation concern, with endangered populations in many parts of the world [12]. How body size might change in response to global warming, and what consequences such changes would have for bat population dynamics, is unknown. In this context, it is important to note that bats do not comply well with expectations regarding optimal body size [13,14] and show no evidence for Bergmann's rule [15]. This impedes predictions about the link between global warming, changes in body size and population dynamics in bats. With a recent study reporting an increase in mortality with large body sizes in female Bechstein's bats (Myotis bechsteinii) [16], exploring the link between body size and weather during the period of juvenile growth becomes crucial for predicting climate change-related extinction risks in bats [17].

Ambient temperature is a key driver of behavioural and physiological responses that have evolved in temperate zone bats. For example, in times of cold temperatures, bats can enter torpor, a state of reduced body temperature and low metabolic rate, which provides significant energy savings [18,19]. Costs of entering torpor during the reproductive period are prolonged gestation and reduced milk production [20,21]. As a result, body size in bats has been linked to warm temperatures during the postnatal development, as some short-term studies suggest [22,23]. In addition, biological parameters such as colony size can also affect juvenile growth. By forming larger roosting groups, colony members can save energy required for thermoregulation [19], which could potentially translate into faster growth of juveniles.

Here, we examine the effect of ambient temperature on the development of adult body size in Bechstein's bats, and assess subsequent mortality costs. Bechstein's bat is a long-lived, forest-living species of high conservation concern that is strictly protected by European law [24]. Like all European bats, females form maternity colonies during spring and summer to communally raise their offspring. Maternity colonies are closed societies due to the strict natal philopatry of females (adult males are solitary [25]). The absence of dispersal in females, in combination with 24-year-long monitoring of bats individually marked with radio-frequency identification (RFID) tags, allowed for an accurate assessment of mortality. We focus on body size and mortality of adult females, as the latter is the most sensitive demographic parameter for population growth in long-lived mammals, including bats [26]. Incorporating records from four free-ranging Bechstein's bat colonies with variable size ranges over time allows us to produce robust estimates of the development of body size in relation to ambient temperatures, birth timing and colony size. In addition, our colonies live in an area highly relevant for global warming, as it shows an above-average increase of summer temperatures, in regional as well as global comparisons [27].

Specifically, we asked the following questions: (i) is there a time period when juvenile growth is most sensitive to ambient temperature? As bats have low prenatal growth rates [28], we expect the sensitive time window affecting growth to last for several weeks after birth. (ii) Does warm weather influence body size? We expect an increase in body size with warmer summer temperatures, as bats may avoid torpor, a physiological state that impedes development in juveniles. In addition, with higher temperatures resulting in better prey availability, juvenile growth could be promoted by an improved body condition of the mothers. (iii) Are there colony-specific traits that affect body size? We suspect that colonies with more females might have larger offspring, as social thermoregulation might be more effective in larger roosting groups. In addition, the timing of birth might have an effect on the body size reached by the juveniles before hibernation. We hypothesize that bats born earlier in summer have a larger time window available for growth during favourable summer conditions and might thus be larger than late-born bats.

2. Material and methods

(a) . Study site and data collection on colony sizes, body sizes and survival

We analysed a dataset spanning 24 years (1996–2019) on body size and survival of RFID-tagged female Bechstein's bats living in four colonies of (BS, GB2, HB, UA). All colonies inhabit forests near the city of Würzburg, Bavaria, Germany [29]. At least twice a year, the colony members were captured from their day roosts (bat boxes) to take biometric measurements. Previously unmarked females were individually marked with subcutaneously implanted RFID-tags (Trovan, Germany) [30]. With no immigration occurring in these colonies [29], unmarked females were surviving young from the previous year. The forearm length (FAL), a proxy of body size [31], was measured to the nearest 0.1 mm using callipers. We only included measurements of fully grown females recruiting into the population in spring after their first hibernation. Using hand-held and automatic RFID-readers attached to the bat boxes, we recorded the presence of individuals from mid-April to September on a nearly daily basis in order to assess colony sizes and monitor the bats' roosting behaviour [29]. We defined the colony size as the number of bats that were logged and/or caught in a year. The percentage of females that were not recorded in a given year, but known to be alive from records in later years, was negligible (0.09% over all colonies and years; one animal in one year).

(b) . Assessment of birth date

The timing of births in three of the colonies (BS, GB2, UA) was inferred either directly by means of infrared video recordings inside of bat boxes [32] or, in most cases, indirectly via RFID-based monitoring of nightly arrivals of bats in boxes The first birth events were notable by a persistent increase in nightly arrivals of adult females in the box, which is indicative of nursing [33] (as examples, see electronic supplementary material, figure S7). We chose the first day of the sustained, increased nightly activity as the birth date. As births are highly synchronized ([32]; electronic supplementary material, figure S7), we were able to adopt a single day of birth per colony and year. In HB, it was impossible to determine births dates as this colony was monitored with hand-held RFID-loggers only. For monitoring methods and data availability, see electronic supplementary material, table S1.

(c) . Meteorological data

Daily data on ambient temperatures were provided by the Bayerische Landesanstalt für Wald und Forstwirtschaft (LWF) from the meteorological station ‘Waldklimastation’ (WKS; ID WUE [34]) which is situated on a forest meadow at a distance of 2 km from the colony GB2, and 10 km from the farthest one, UA. We used 24 h mean, minimum and maximum temperature as well as mean temperature during night as aggregate measures. To fill data gaps, we used hourly temperature data obtained from the DWD Climate Data Center (meteorological station ‘Würzburg’; ID 05705 [35]), averaged to fit the WKS data format. Temperature data were nearly identical at both stations (see electronic supplementary material, figure S1).

(d) . Data analysis and model building

All analyses were performed in R (R Core Team, 2018 and 2020 [36]). To explore the link between body size and mortality, we repeated the discrete time survival analysis described in Fleischer et al. [16], using an updated dataset including 363 adult females (248 of which had already been used by Fleischer et al., [16]). Data analysis and model selection were performed using the gam function in the mgcv package [37]. The parameter year entered the models as a factor, whereas age and size were included as smooth functions, covering the nonlinear aspect of fits.

To find the period (Climwin_Summer) during which ambient temperature had the largest impact on the timing of birth and FAL, we applied the R package climwin [38]. The package uses a systematic approach to detect a time window over which a biological response variable is sensitive to a weather variable. Periods of climate sensitivity, as well as the best identifying predictors, are found by using a sliding window approach, where the start and end time of a climate window are varied and all combinations are modelled. All possible models are then compared using the information-theoretic model selection criteria AICc [38,39].

Climatic variables tested were the aforementioned aggregated data for temperature. For FAL as response variable, we employed the sliding window approach over daily increments lasting from 1 May until 30 September, in order to include both the pre- and postnatal period of the bats. For the timing of birth as response variable, the time range tested lasted from 15 March, the time when the first bats are expected to start to emerge from their hibernacula, to 15 July, the time when all birth events had occurred.

To avoid the pitfalls of multiple comparisons, we followed the recommended procedures [38,39] and created randomized datasets without weather signal (n = 100) and compared those with the best model. This allowed us to quantify the likelihood of obtaining strong model support by chance simply due to the high number of models created and tested [39]. Furthermore, we explored possible autocorrelations of the candidate weather signals and possible interactions between them, using the ‘autowin’ and ‘crosswin’ functions of the package, respectively.

Integrating the observed dates of births directly into the analysis would have reduced the sample size considerably, as it was impossible to obtain birth dates for all years in all colonies (see electronic supplementary material, table S1 for an overview of the data). We, therefore, established the use of a proxy variable for which data were available for the entire period under review. From previous studies in other bat species [40,41], we learned that spring temperatures strongly affected the timing of births. Therefore, we first determined the appropriate temperature aggregate and sensitive time window relevant for the timing of births in Bechstein's bat, using the climwin-package as described above. We found a linear effect of temperature on the timing of births in a regression analysis (see electronic supplementary material, figure S2). In a next step, we built generalized additive models (GAM; [42]) including factors such as colony size, with colony ID as a random factor. Model comparison confirmed that the maximum temperature for the responsible Climwin_Spring period was the most decisive predictor of birth events (see electronic supplementary material, table S2 and figure S2). Consequently, spring temperature (Climwin_Spring) was established as a proxy for birth date. Correlation analysis confirmed that Climwin_Spring was independent from Climwin_Summer (r₄₃ = −0.09, p = 0.55), which was important in order to consider both parameters independently in further analyses.

To test for nonlinear interactions between our response variable (FAL) and the explanatory variables, we applied GAMs using the function ‘gam’ from the mgcv package [37]. All models were based on an identity link function fitted with the restricted maximum likelihood (REML) method and a γ-value of 1.4. Models were then compared using the Akaike information criterion (AIC) [43]. Colony ID entered the model as a factor, controlling for possible differences in the local habitat of the four study sites. Following the climwin analysis for the most sensitive time windows (Climwin_Spring and Climwin_Summer), we selected the most relevant weather parameter (temperature) to be included using the AICc criterion. Temperature as well as colony size were first centred (by subtracting the mean) to reduce multicollinearity, then modelled as smooth functions to capture the potential nonlinear impacts of these variables. We further tested for an interaction between colony size and temperature using a tensor product smooth in the gam function [42], as during cold weather periods, the colony size might have been more important. For more details on the data used for the GAM, see electronic supplementary material, table S3.

To model responses of body size to different environmental factors, we built predictions based on fitted gams using a ‘response’ type in the ‘predict.gam’ function of the package that returns predictions on the scale of the response.

3. Results

(a) . Changes in body size and temperature over the study period

For a total of 358 adult females, FAL was available for 24 years (1996–2019), ranging from 38.9 to 45.9 mm with a mean of 42.7 mm (±1.19 s.d.). For a given cohort, the mean FAL values fluctuated between 41.4 and 44.3 mm (figure 1). Despite considerable oscillations, body size (FAL) increased significantly over the 24-year period (R = 0.13, p = 0.011). During the same period, spring temperature increased by 2.19°C and summer temperature by 2.36°C (figure 1; see also electronic supplementary material, table S5).

Figure 1.

Boxplots depict the variation of forearm length (mm) in adult females Bechstein's bats over the time period 1996–2019 (left axis). Scatterplots show the overview of the mean monthly air temperature (°C) for the summer months June–August (right axis).

(b) . Survival and body size

Larger bats had a higher risk of mortality (as shown by Fleischer et al. [16] on a subset of the current data). To assess the relative impact of age, individual body size and environment on mortality, several models were run and compared. In the best model (mortality ∼ year + s(age) + s(size)) both body size (p = 0.040) and age (p = 0.012) were retained as significant factors explaining mortality. Mortality rate increased notably with FAL of 44.5 mm and larger. The effect of size on mortality was the same in a year with average mortality (2014) compared to the most extreme year (2010; figure 2). In contrast with Fleischer et al. [16], the factor ‘age’ now entered as significant factor with older bats showing higher mortality rates, most likely a consequence of a higher number of ‘old’ individuals due to the increased length of the study (resulting in 46% more individuals compared to [16]).

Figure 2.

Mortality increase (as smooth function from a gam) with forearm length (for environmental conditions fixed at an average year = 2014 and a year with exceptionally high mortality (2010)). Ninety-five per cent confidence intervals added.

(c) . The influence of temperature on body size and the critical time window

Of all tested temperature aggregates, minimum temperature had the strongest influence on body size (see electronic supplementary material, table S6). The warmer it was during a critical time window, the larger the bats grew (by +0.49 mm per °C). The critical time window in which body size was most susceptible to external influences ranged from 22 June until 16 July (p < 0.001; figure 3). With births usually occurring around 16 June, the sensitive time period began roughly a week after birth events, and ended 25 days later, at about the time when the juveniles were fully fledged and the lactation period ended [32]. No other time periods, either before or after, had as great as an impact as the one in June–July (figure 3) and we found no autocorrelations and correlations between weather variables (see electronic supplementary material, figure S4).

Figure 3.

Heat map of ΔAICc values for all fitted climate windows, which displays one clear peak determining the sensitive time window: 76–100 days before our reference date (30 September). (Online version in colour.)

(d) . The influence of temperature on the birth date

Variation in colony birth date was best explained by average daily maximum spring temperature from 3 April until 26 May (see electronic supplementary material, table S4). Births occurred earlier the warmer it was during this sensitive time period (by −2.62 days per °C, p < 0.001, see electronic supplementary material, figure S3 and tables S2 and S4).

(e) . Other parameters influencing body size

The best model (no. 5; table 1) explained 30.9% of the deviance in body size, with summer temperature (p < 0.001) and colony size (p < 0.001) both affecting body size. The warmer it was during the identified sensitive summer window (figure 4a) and, independently, the more females were present in a colony (figure 4b), the larger the body size of juveniles. An interaction between colony size and temperature was not supported by the data (see electronic supplementary material, figure S5). This implied that colony size was not of more importance when it was cold during the development of juveniles. Colony ID as a factor provided only a small improvement (model no. 6). Spring temperature, our proxy for the timing of birth, was not included in the two best models.

Table 1.

Summary for the model comparison for body size. Summer temperature = Climwin_Summer minimum temperature, spring temperature = Climwin_Spring maximum temperature. In bold: the best model, selected for its minimal AIC value in combination with smallest degrees of freedom.

model	terms	d.f.	AIC
1	colonyID	5.00	1122.41
2	colony size	5.66	1088.12
3	spring temperature	9.61	1116.65
4	summer temperature	3.00	1074.13
5	summer temperature + colony size	11.11	1026.89
6	summer temperature + colony size + colony ID	14.71	1027.38
7	summer temperature + colony size + spring temperature	7.52	1030.25
8	summer temperature + colony size + spring temperature + colony ID	10.97	1031.65
9	interaction(summer temperature + colony size) + spring temperature	9.02	1032.94
10	interaction(summer temperature + colony size) + colony ID + spring temperature	12.57	1033.83

Figure 4.

Predicted response of body size (forearm length (mm)) in dependence of minimum summer temperature (°C) (a) and colony size (b). All underlying GAMs are calculated with k = 8, γ = 1.4 and method = REML.

4. Discussion

In various taxa, a decline of the average body size within populations in response to global warming has been observed [1]. In contrast with this ‘global shrinking’ of many species, we found that warmer summers resulted in larger body sizes in our population of Bechstein's bats. Due to elevated mortality costs of large-sized females, this specific effect of global warming carries a strong concern with respect to the conservation of this species. In addition, the number of adult females within a maternity colony also had a positive effect on body size of juveniles. As summer temperatures in Germany are predicted to further increase by 1.6–3.8°C until 2080 [44], Bechstein's bats may further increase in body size, with potential negative effects on population persistence during ongoing global warming.

(a) . Effect of spring temperature on parturition

Our finding that warm spring temperatures lead to earlier parturition dates corresponded to the findings of previous studies in other bat species. While Lučan et al. [40] reported only April temperatures to be significant for parturition in Daubenton's bats (Myotis daubentonii), in greater horseshoe bats (Rhinolophus ferrumequinum), both April and May weather conditions affect birth timing [45]. The latter corresponds well with the sensitive time window found in our study, which ranged from the beginning of April to the end of May. It has been suggested [40,46] that spring temperature might have an indirect effect on bats' reproductive timing via prey availability and thermoregulation. In addition, increased spring temperatures may shorten hibernation length and thereby advance the onset of pregnancy [40]. This explains well the start of the sensitive time window for Bechstein's bats, since they typically leave their hibernacula around mid-April.

What defines the end of the sensitive period for birth timing is, however, less clear. In Bechstein's bats, the sensitive window is comparatively long. This suggests that environmental conditions might affect both the early stage of ovulation/impregnation and the later stage of fetal development. It has been shown that cold weather spells delay parturition dates in pipistrelle bats (Pipistrellus pipistrellus) [47] due to the extended use of torpor at low temperatures. In the little brown bat (Myotis lucifugus), pregnant females have been shown to make less use of torpor, probably to avoid its negative effect on fetal growth [48]. Thus, the closure of the sensitive period before late pregnancy seems sensible. During late pregnancy, females are likely to avoid using torpor as much as possible, and consequently, cold weather would no longer impact parturition dates.

(b) . Effect of summer temperature on body size

We showed that the most sensitive time period for weather effects on body size began with the second half of June, shortly after the typical onset of birth event. This agrees well with previous reports in another bat species, which found a correlation between juvenile's growth and roost temperatures at the end of June [23]. Since bats have very low fetal growth rates [28], an effect of ambient temperature on body size before birth seems unlikely. After birth, however, juvenile growth rates speed up rapidly: for similar sized Myotis species, fully grown juveniles are typically recorded 6 to 8 weeks after birth [41], which fits well to our own unpublished data on Myotis bechsteinii. In conclusion, our sensitive time window, which lasted around four weeks, encompassed predominantly the period of the rapid increase in FAL [49] during nursing.

Most importantly, we could clearly demonstrate that warm summer temperatures led to large body sizes. While previous work by other authors described the positive effect of roost temperature [22,23], no other study examined ambient temperatures (warm weather) and their effect on body size in bats. Ambient temperature is, in our eyes, crucial, as it not only strongly influences roost temperature, and, therefore, bats’ resting metabolic rate [50], but also prey availability via insect activity [51]. The effect of weather on the foraging activity of insectivorous bats is well known [52]. The observed increase in body size during warm summers could, therefore, be the result of higher prey availability triggered by warm temperatures, and thereupon better body condition of nursing females. In addition, high ambient temperatures leading to warmer roosts allow bats to significantly reduce energy expenditure [19], and this surplus of energy can directly be invested into the offspring size. As juvenile growth is impeded by torpor [53], the ability to stay out of torpor in warmer temperatures might result in larger body size because of uninterrupted growth. This is corroborated by our finding that minimum temperature had a larger impact on body size than maximum or mean temperature.

(c) . Effect of colony size on body size

Apart from the aforementioned temperature effect, we found that larger colony sizes were associated with larger body sizes of fully grown juveniles. By contrast, Zahn [23] found no effect of colony size on body size in the greater mouse-eared bat (Myotis myotis) [23]. The difference between these two species could be explained by their different colony sizes and roost type they use for raising offspring. Bechstein's bat colonies comprise 10 to about 40 adult females that roost in small bat boxes or tree cavities. By contrast, maternity colonies of greater mouse-eared bats often number several hundred females that roost in spacious roofs of buildings where variation in colony size affects thermoregulation less than variation in ambient temperature.

It is well documented that female bats living together in maternity colonies can actively adjust their roosting behaviour to their metabolic needs. Besides choosing roosts with respect to microclimatic conditions [54,55], in bat species such as Bechstein's bats that live in fission–fusion societies, roosting group size is readily adjusted depending on the reproductive status, and thus, different metabolic needs [30,56]. During lactation, Bechstein's bats remain in relatively large daily groups, which is associated with significant energy savings [19]. This fits well with our finding that larger colonies tend to produce larger offspring. By socially aggregating and actively choosing warm roosts, bats create a favourable environment that promotes growth in juveniles. At the same time, increasing summer temperatures lead to larger body sizes that result in higher mortality rates once certain limits are reached (44–44.5 mm). While the aggregative behaviour of the females and their preference for warm roosts offers a plastic and fast response to cold weather, in times of global warming these behavioural adaptations might become maladaptive. Reproducing females could run into an ecological trap by choosing an overly warm roosting environment, where their fitness is lower.

It seems plausible that larger colonies occur in areas with better food resources, which also may have produced the observed positive association between colony size and body size of the offspring. However, we do not think that this scenario applies to the colonies in our study because the number of females differed between colonies, but also varied considerably over time, with ample overlaps among colonies. One sudden reduction in colony size (2011) was followed by a slow recovery due to a low rate of population growth in connection with a lack of movements between colonies. Hence, the size of a colony does not seem to be indicative of the quality of local food resources, rather than mirror population dynamics of the immediate past. In addition, colony ID as a proxy for local environmental conditions did not enter as a significant factor in the model further indicating that local resource availability did not affect body size.

(d) . The fitness effect of large body size in bats

There is some evidence that being larger is beneficial for female bats during reproduction. Ransome [57] surmised that reduced skeletal size is a long-term handicap to successful breeding in greater horseshoe bats. Williams & Findley [58] argued that larger female bats might have a reproductive advantage as they can more easily maintain homeothermy during gestation. Additionally, for larger females, the proportional load of carrying young is reduced, as is the relative cost of producing milk [59]. Other than Bechstein's bats ([16] and our study with a larger sample size), however, to the best of our knowledge, no study yet found a negative fitness impact of large body size in female bats. Currently, it is unclear why larger females in Bechstein's bats face a higher mortality risk, but one plausible explanation is that, compared to smaller individuals, they need more food to build up the fat reserves necessary for successful hibernation. Moreover, FAL may be more than a mere proxy for body size, as it is a functionally important morphological feature that is directly linked to flight performance [60]. It seems possible that at times of low insect availability, larger females face more problems to meet their energy requirements.

In Bechstein's bats, as in most European bat species, females are slightly larger than males [58]. We focused in this study on the link between ambient temperature, female body size and female mortality, as female mortality is the crucial demographic parameter for population growth in bats. In greater horseshoe bats, paternity success was correlated positively with male size [61], whereas female reproductive success was independent of body size. Whether body size of male Bechstein's bats affects survival or reproductive success remains to be studied, but we expect little or no impact for population growth due to the promiscuous mating system with low reproductive skew in this species [62].

(e) . Implications of our results on the future of Bechstein's bat populations

It might be suspected that the observed positive relationship between summer temperature and body size would result in strong directional changes in body size over time with increasing temperature. Although a positive trend of body size over time is discernible (Pearson correlation coefficient R = 0.13; p = 0.011), average body size of offspring of the year fluctuated considerably over the years. Figure 5 shows that on average warmer temperatures led to on average larger body sizes. But body size repeatedly fell to low levels whenever summer temperatures were below average, for example, in 2007, 2011 and 2014. The strongest drop in body size occurred after a population crash in the winter 2010/2011 where 60% of individuals died (electronic supplementary material, figure S6; [16]). Small colony size, together with the cold 2011 summer, then led to the observed sudden drop in body size. Other effects, such as mismatches in food availability, which result from decoupling across environments and trophic levels in the course of climate change [63], could also be the source of mitigating effects of temperature on body size.

Figure 5.

Direction of change in the deviation from the mean per year for the mean minimum summer temperatures (dots) and forearm length (boxplots). Boxplots depict the centred values of forearm length (mm) for all cohorts (centred around the overall mean of 42.69 mm forearm length). The summer temperature (°C) is the minimum temperature during the critical Climwin_Summer period, centred around the mean across all years for this time window (11.59°C). The horizontal line depicts the exact mean, values above the line are above average and values below the line are below average.

Taking an evolutionary perspective, the observed temperature–body size relationship can be seen as a reaction norm that may differ across individuals. If the slope of this reaction norm is heritable, larger individuals would carry the genetic basis for larger offspring, given further increasing summer temperatures. Any resulting runaway process, however, might be counteracted by the observed mortality cost of large body size (this study and [16]). The outcome of this scenario is far from clear, as yet unidentified positive effects of large body sizes on reproductive success or, potentially, more females giving birth in warmer summers might counterbalance the increase in mortality with body size. Dangers are that further increase in ambient temperatures might tip this balance. It is, therefore, important for future work to also assess the heritability of body size in bats to better understand the potential and limits for evolutionary adaption in this trait.

5. Conclusion

Our long-term study on female Bechstein's bats demonstrates, for the first time, a link between warmer summer temperatures during nursing of young, adult body size and an increased mortality risk. Understanding the relationship between ambient temperature and body size as one mechanism shaping population dynamics is crucial for predicting the consequences of global warming on species survival. This is especially true for long-lived mammals such as bats, which are dependent on high adult survival and longevity to maximize population growth.

Ethics

Handling, tagging and monitoring of the bats were conducted under permits for species protection (55.1.-8642.01-2/00) and animal welfare (55.2-DMS 2532-2-20) that had been issued by the government of Lower Franconia.

Data accessibility

Climate and bat data are available from the Dryad Digital Repository: https://dx.doi.org/10.5061/dryad.s7h44j16k [64].

Authors' contributions

C.M.: data curation, formal analysis, visualization, writing—original draft; A.S.: conceptualization, methodology, supervision, validation, writing—review and editing; G.K.: conceptualization, funding acquisition, resources, supervision, validation, writing–review and editing.

All authors gave final approval for publication and agreed to be held accountable for the work performed therein.

Competing interests

We declare we have no competing interests.

Funding

This work profited strongly from the financial support of the German Research Foundation (DFG RTG 2010) ‘Biological Responses to Novel and Changing Environments’.