Environmental drivers of body size in North American bats

Bergmann’s Rule—which posits that larger animals live in colder areas—is thought to influence variation in body size within species across space and time, but evidence for this claim is mixed. We tested four competing hypotheses for spatio-temporal variation in body size within bat species during the past two decades across North America. Bayesian hierarchical models revealed that spatial variation in body mass was most strongly (and negatively) correlated with mean annual temperature, supporting the heat conservation hypothesis (historically believed to underlie Bergmann’s Rule). Across time, variation in body mass was most strongly (and positively) correlated with net primary productivity, supporting the resource availability hypothesis. Climate change could influence body size in animals through both changes in mean annual temperature and in resource availability. Rapid reductions in body size associated with increasing temperatures have occurred in short-lived, fecund species, but such reductions likely transpire more slowly in longer-lived species.

, and 57 extinction risk (Brown 1995;Ripple et al. 2017). Understanding the factors that drive variation in 58 body size is thus among the most important goals in ecology (Kaspari 2005 (Bergmann 1847;Salewski & Watt 2017), which states that animals residing in colder climates 60 are larger than those residing in warmer climates, is a widely known macroecological pattern. 61 Although originally and primarily applied to differences in body size between closely related 62 species, Bergmann's Rule is often believed to extend to differences in body size within species as . The ratio between surface area and volume decreases with increasing body size, so while 71 absolute heat loss increases with increasing body size, smaller animals require greater metabolic 72 heat production to offset heat loss across their relatively large surface areas (Withers et al. 2016). 73 Larger body size could therefore be an adaptation to climates with cooler average temperatures. 74 Despite its intuitive appeal, empirical support for the heat conservation hypothesis within 75 species is mixed. Although ecologists have accumulated substantial evidence that individuals 76 within species tend to be larger in colder climates (e.g., Smith et al. 1995;Ashton 2002;Meiri & 77 Dayan 2003), more recent and more comprehensive tests have failed to find consistent 78 relationships between temperature and the body sizes of individuals within species (Riemer et al. 79 2018). Additionally, physiologists have questioned the validity of the heat conservation hypothesis 80 on physiological grounds (Scholander 1955;McNab 1971;Geist 1987   To evaluate the mechanistic underpinnings of Bergman's Rule, we tested whether spatial 128 and temporal variation in body mass of North American bats is best supported by the heat 129 conservation, heat mortality, resource availability, or starvation resistance hypotheses 130 (summarized in Table 1). As with many taxa, the intraspecific formulation of Bergmann's Rule is 131 exhibited by some species of bats (e.g., Burnett 1983;Bogdanowicz 1990;Lausen et al. 2008, 132 2019), but does not appear to be the norm among the clade as a whole (Riemer et al. 2018). 133 Critically, extensive records of bat captures permit a rare opportunity to test for Bergmann's Rule 134 and evaluate its associated hypotheses while accounting for other factors (e.g., sex, age, 135 reproductive condition, and time of year) that influence body size. We compiled 17 such data sets 136 and used Bayesian hierarchical models to weigh evidence for each hypothesis across both space 137 and time for 20 species of North American bats. We expected observed patterns of variation in 138 body mass to be driven by the same process or processes across both time and space. In other 139 words, if variation in body mass across space was best explained by one of our four hypotheses, 140 we also expected variation in body mass across time to be best explained by the same hypothesis. 141 This would provide strong evidence for a consistent selective force driving variation in body size. of year-to-year differences in mean temperatures. This represents roughly the period in which a 166 bat would be active in a given year (dates before 1 April are likely to be spent in hibernation or in 167 winter ranges). 168 To test the heat mortality hypothesis across space, we extracted data for each capture 169 location from the DAYMET daily climate summaries 1-km resolution data set (Thornton et al.  Table A1). Most species were larger at higher latitudes, but body size remained relatively 228 constant over our study period (Fig. A2). Significant spatial and temporal variation existed 229 among all predictor variables, enabling detection of meaningful relationships between body mass 230 and predictor variables (Fig. A3). Parastrellus hesperus). Most species exhibited minimal variation in body mass with respect to 239 maximum temperature (Fig. 2B), primary productivity (Fig. 2C), and spring/autumn temperatures 240 (Fig. 2D), suggesting a lack of support for the heat mortality, resource availability, and starvation 241 resistance hypotheses, respectively. For these three hypotheses, coefficients were relatively evenly 242 12 distributed around 0; 90% credible intervals overlapped 0 in most cases, and credible intervals that 243 did not overlap zero were distributed relatively evenly around zero.

246
Temporal variation in body mass most strongly supported the resource availability hypothesis, 247 with most species exhibiting greater body mass during years in which net primary productivity 248 was higher (Fig. 3C). For 14 of 20 species, body mass increased with increasing net primary 249 productivity (i.e., β > 0), and the probability that the coefficient was above zero was >95% for 7 to year-to-year differences in mean annual temperatures (Fig. 3A), maximum temperatures ( Fig.   253 3B), or spring/autumn temperatures (Fig. 3D), suggesting a lack of support for the heat 254 conservation, heat mortality, and starvation resistance hypotheses, respectively. For these 255 hypotheses, coefficients were relatively evenly distributed around 0, 90% credible intervals 256 overlapped 0 in most cases, and credible intervals that did not overlap zero were relatively evenly 257 distributed around zero or were distributed in the direction opposite most coefficients. Millien 2014). However, spatial patterns could take centuries or millennia to arise, even when they 332 are relatively clear-cut (and spatial patterns in body size are rarely so). This is especially true for 333 16 long-lived species, for which the pace of change is likely to be slower than for short-lived, more 334 fecund species. 335 Climate change will likely induce changes in body size for animals, but such changes may 336 be more complex than has been appreciated. Over the nearly 2 decades that we collected data, the 337 primary driver of short-term (annual) variation in body size was resource availability. Increases in 338 mean annual temperatures could make many ecosystems more productive for a longer portion of temperature-induced reductions in body size could take substantially longer to manifest than for 366 short-lived, more fecund species, and will be obscured by variation in resource availability.

Resource Availability
Because individuals living in more productive environments tend to be larger, individuals will be larger in areas where primary productivity is higher.
Net primary productivity during April -October (2000-2016; MODIS; Stockli 2020) Because individuals living in more productive environments tend to be larger, individuals will be larger after years in which primary productivity is higher.
Net primary productivity in months preceding capture, inclusive of month of capture (MODIS; Stockli 2020)

Starvation Resistance
Because larger body size increases an individual's ability to survive periods of resource scarcity, individuals will be larger in areas where periods of resource scarcity (e.g., winters) are most severe.