Disentangling physiological and physical explanations for body size-dependent thermal tolerance Free

Many common responses to anthropogenic warming are body size dependent, including reductions in mean body size within populations and the extirpation of large species within communities (Forster et al., 2012; Gardner et al., 2011; Ohlberger, 2013; Salvatteci et al., 2022; Verberk et al., 2021). Size-dependent thermal tolerance (SDTT; see Glossary), the propensity for organisms of different body size to have different heat or cold tolerance limits, could help explain these patterns. Thermal tolerance (see Glossary) can change ontogenetically and vary among individuals of different size at the same life stage (Bowler and Terblanche, 2008; Dahlke et al., 2020; Di Santo and Lobel, 2017; Gilbert and Lattanzio, 2016; Gunderson et al., 2019; Heatwole et al., 1969; Illing et al., 2020; Kingsolver and Buckley, 2020; Klockmann et al., 2017; Liwanag et al., 2018; Peck et al., 2009; Roeder et al., 2021; Ruthsatz et al., 2022; Smith and Ballinger, 1994; Fig. 1A). Thermal tolerance can also differ among species of different size (Brusch et al., 2016; Leiva et al., 2019; Peralta-Maraver and Rezende, 2021; Rubalcaba and Olalla-Tárraga, 2020; von May et al., 2019; Fig. 1B). Generally, larger body size correlates with lower thermal tolerance (Peralta-Maraver and Rezende, 2021). However, positive associations between body size and tolerance have also been observed (Baudier et al., 2015; Clark et al., 2017). The direction of the size effect on thermal tolerance may be mediated by habitat (i.e. terrestrial versus aquatic) and the time scale of temperature exposure (Leiva et al., 2019).

But why does thermal tolerance vary with body size? Physiologists typically think of variation in thermal tolerance through the lens of evolved or plastic differences at the biochemical, cellular and organ/system level. For example, if two species or individuals differ in thermal tolerance, we assume it is because they differ in the temperature dependence of factors such as enzyme function, gene regulation, aerobic scope, etc. (Angilletta, 2009; Bodensteiner et al., 2021; Chung and Schulte, 2020; Ern et al., 2023; Portner, 2010; Somero et al., 2017). Hypothesized explanations for SDTT follow this general model, and often involve size-based differences in oxygen delivery and energy consumption. Variation in traits such as absolute and tissue-specific metabolic rate, oxygen transport cost, diffusion distance and rates of stored energy depletion are all associated with differences in size (Gillooly et al., 2001, 2016; Schmidt-Nielsen, 1997; Verberk et al., 2011). As a result, organisms of different size may reach energetic limits at different tissue temperatures (Leiva et al., 2019; Leiva et al., 2023; McKenzie et al., 2021; Messmer et al., 2017; Peck et al., 2009; Peralta-Maraver and Rezende, 2021). Throughout, I will refer to such mechanisms as generating physiological SDTT (see Glossary).

However, large and small organisms can differ in thermal tolerance even if the temperature dependence of their physiological processes is the same. Body size affects heat exchange between tissues and the environment, and that variation in heat exchange alone can cause what I will refer to as physical SDTT (see Glossary). Physical SDTT is rooted in two observations that will be discussed in greater detail in the following section. First, body size dictates rates of body temperature change, and second, rates of body temperature change influence thermal tolerance limits. Physical and physiological SDTT are not mutually exclusive, but they do make different predictions (Fig. 2). Physiological SDTT predicts that thermal tolerance differences occur because the physiological processes of large and small organisms respond differently when they experience the same body temperature dynamics. Physical SDTT predicts that thermal tolerance differences occur because large and small organisms experience different body temperature dynamics when they experience the same abiotic environment.

In both terrestrial and aquatic habitats, large individuals usually warm and cool more slowly than small individuals under a given set of physical conditions (Bilyk and DeVries, 2011; Gates, 1980; Porter and Gates, 1969; Stevenson, 1985; see below for exceptions). The effect of body size on the rate of body temperature change can be large (Fig. 3A), even among taxa of relatively small size (Fig. 3B). For example, among three lizard species, the heating rate of a 0.5 g individual was approximately six times faster than the heating rate of a 12 g individual (∼15°C min⁻¹ versus 2.5°C min⁻¹) in the same thermal environment (Herczeg et al., 2007; Fig. 3B).

Size-dependent temperature change is usually thought to occur for two primary reasons. First, large individuals have more mass, so more energy must be exchanged to change their temperature. Second, large animals typically have a smaller surface area to volume ratio, reducing the rate of heat transfer between the body core and the surrounding environment (Gates, 1980; Porter and Gates, 1969; Stevenson, 1985). However, other factors can also mediate size-dependent body temperature fluctuations. These factors relate to the relative positions that organisms of different size occupy within an environment. For example, large and small individuals differ in their distance from the substrate, and the physical conditions that influence both body temperature equilibria and rates of body temperature change shift with substrate proximity (Campbell and Norman, 2012). Among relatively small animals such as insects, body size affects how much of the body is within the static air of the boundary layer around the substrate (Kaspari et al., 2015; Stevenson, 1985; Woods, 2013). Large insects on leaves reside outside of the leaf boundary layer and are predicted to experience higher equilibrium body temperatures and higher rates of body temperature change than smaller species within the boundary layer (Pincebourde et al., 2021; Fig. 3C). For larger taxa such as lizards, size additionally influences heat exchange between the body and environment by affecting the depth of the boundary layer around the integument and the speed of air moving past the body (Kearney et al., 2021).

Crucially, the rate of body temperature change affects thermal limits. This has been demonstrated in invertebrate and vertebrate taxa from both aquatic and terrestrial habitats (Allen et al., 2016; Chown et al., 2009; Claunch et al., 2021; Heatwole et al., 1968; Illing et al., 2020; Leong et al., 2022; Mora and Maya, 2006; Moyen et al., 2019; Overgaard et al., 2012; Terblanche et al., 2007; Vinagre et al., 2015). For example, the heat and cold tolerance limits of Drosophila melanogaster and the Argentine ant (Linepithema humile; Fig. 3D,E) differ based on how fast the temperature changed during experiments (Chown et al., 2009). The differences observed were sometimes large: L. humile warmed at 0.05°C min⁻¹ had a heat tolerance limit 7°C lower than those warmed at 0.5°C min⁻¹ (Chown et al., 2009). Similarly, thermal limits were positively associated with the observed rates of individual body temperature change in four species of Sceloporus lizard (Claunch et al., 2021) and in the mussel Mytilus californicus (Moyen et al., 2019).

Rates of body temperature change probably influence thermal limits because they affect the amount of time that organisms experience harmful temperatures (Moyano et al., 2017; Rezende et al., 2011; Terblanche et al., 2007). It is well established that the ability to survive a given stressful temperature is conditional upon how long that temperature is experienced (Rezende et al., 2014). This concept extends to fluctuating thermal environments, where accumulated damage is related to the integral of time spent beyond the body temperature threshold at which physiological damage occurs and how far beyond that threshold the body temperature gets (Jørgensen et al., 2021, 2019; Kingsolver and Woods, 2016; Rezende et al., 2020; Siegle et al., 2018; Sørensen et al., 2013; Williams et al., 2016). For example, D. melanogaster heat tolerance limits were lower when the temperature increased at a slow (0.06°C min⁻¹) versus a fast (0.1°C min⁻¹) rate, and slower temperature change was also associated with greater heat shock protein gene expression, potentially indicative of greater molecular damage (Sørensen et al., 2013).

Slow rates of temperature change are usually associated with lower tolerance for heat or cold, but slow thermal ramping (see Glossary) sometimes leads to greater tolerance (Chidawanyika and Terblanche, 2011; Chown et al., 2009; Vinagre et al., 2015). If a thermal ramp is slow enough, acclimatory processes that compensate for temperature change can be activated, such as the remodeling of cell membranes and the expression of thermo-protective proteins (Havird et al., 2020; Peralta-Maraver and Rezende, 2021; Rezende et al., 2011). The rates of temperature change that produce positive versus negative effects on thermal tolerance differ among species (see citations above), and could be at least partially explained by species differences in how quickly acclimation occurs (Einum and Burton, 2023).

All thermal tolerance limits, whether mediated by physical or physiological SDTT, are ultimately due to some form of compromised physiological function. Yet the pathway to physical SDTT is fundamentally different from how we traditionally conceptualize thermal tolerance differences in physiology: it requires no difference among individuals in the temperature dependence of physiological processes per se at the molecular, tissue or organ/system levels. This divergence from the classical physiological point of view is meaningful and has implications for how we design and interpret thermal tolerance experiments, and also for our understanding of thermal ecology and thermal adaptation.

If SDTT is observed experimentally, how do we know whether the explanation is physiological or physical? Thermal limits are usually measured with one of two approaches: dynamic or static experiments (Lutterschmidt and Hutchison, 1997). Dynamic experiments (see Glossary), where an individual is warmed or cooled until a physiological temperature limit is reached, are probably the most common. The approach reflects the fact that organisms experience fluctuating temperatures within their habitats. In static experiments (see Glossary), each individual experiences a single temperature, and the experiment ends when the individual can no longer maintain physiological function (different endpoints can be used) at that temperature. The longer an organism can function at a given high or low temperature, the more tolerant they are. There has been considerable discussion of the relative merits of dynamic versus static experiments, but they can be unified under a time-integrated exposure framework (Jørgensen et al., 2021; Ørsted et al., 2022; Peralta-Maraver and Rezende, 2021), and SDTT has been demonstrated in both experiment types (Leiva et al., 2019; Peralta-Maraver and Rezende, 2021). In this Commentary, I focus mostly on dynamic experiments because of their commonality, but how static experiments can be influenced by physical SDTT is also discussed.

Dynamic thermal tolerance experiments can be placed into two broad methodological categories, and the distinction is crucial for our understanding of SDTT. I refer to the different categories as single-rate and variable-rate dynamic experiments (see Glossary). The rate being referenced here is the rate of body temperature change of individuals, not the rate of change in the temperature of their surrounding environment (Figs 2 and 4). In single-rate experiments, an individual's body temperature is monitored during the thermal ramp and heat flux is adjusted such that all individuals experience the same rate of body temperature change (e.g. Gunderson et al., 2018; Fig. 2A). In contrast, variable-rate experiments apply methods that allow individuals of different size to experience different rates of body temperature change. In these experiments, thermal ramping is accomplished either by placing individuals in a static thermal environment in which body temperature is out of thermal equilibrium (such as under a heat lamp or in a chilled water bath), or by placing them in a dynamic thermal environment in which environmental temperature (e.g. air or water) changes at a constant rate (e.g. Gunderson et al., 2019; Figs 1A and 2B). In either case, the rate of change in body temperature can differ as a function of body size owing to thermal inertia or any of the physical mechanisms described above. Variable-rate experiments are probably more common than single-rate experiments, especially in relatively small and/or aquatic organisms – such as insects and fish – for which core body temperatures are difficult to directly measure in real time during experiments.

Single- and variable-rate experiments tell us different things about SDTT. The interpretation of single-rate experiments is straightforward. If SDTT is found, then physiological effects are responsible because body temperature changed at the same rate among all individuals regardless of size. If SDTT is not found, there are no differences in temperature-dependent physiology based on size. Physical SDTT cannot be tested with a single-rate experiment. However, if SDTT is found in a variable-rate experiment, it is not possible to determine whether it is due to physiological or physical explanations, or both. If no SDTT is detected using a variable-rate experiment, it is possible that there are no physical or physiological effects (perhaps the most likely explanation), or, alternatively, the physiological and physical effects could cancel each other out (see below).

Neither single- nor variable-rate experiments are inherently better than the other. Their value depends on the hypothesis being tested. If one is interested in classically defined, inherent physiological differences in thermal tolerance between individuals or species of different size, then single-rate experiments that control for physical effects of size should be used. However, if one is interested in size-dependent variation in overheating risk as a result of natural temperature dynamics – such as diel temperature fluctuations or those caused by movement from one microclimate to another – variable-rate experiments should be used. In those cases, the experimental protocols should be calibrated such that the variable ramping rates of differently sized individuals reflect those expected within the habitat(s) of interest (Allen et al., 2016; Chown et al., 2009).

As an example of the application of single-rate experiments to test for physiological SDTT, I have re-analyzed published heat tolerance data on Anolis lizards collected by myself and collaborators (Deery et al., 2021; Gunderson et al., 2018; Leal and Gunderson, 2012; Table S1). None of these data were collected to test for SDTT, but the goal in all cases was to understand variation in physiological heat limits, so dynamic single-rate methods that control for confounding physical effects of body size were always applied. Furthermore, the same rate of body temperature ramping (2°C min⁻¹) was always used, allowing for comparison across studies.

I performed two types of analysis. First, I tested for physiological SDTT at the intraspecific level, focusing among individuals within species. Second, I tested for SDTT among species at the macroevolutionary scale using phylogenetic comparative methods. I restricted individual-level tests to taxa with a sample size of at least N=10, with the largest sample size of N=52. This resulted in Pearson product moment correlation tests within 14 populations representing 12 species (Table S1). In 13 of the 14 tests there was no relationship between body size and physiological heat tolerance among individuals (P>0.05; Table S1). Data for the two species with the largest sample sizes, A. carolinensis and A. sagrei, are shown in Fig. 4A and B, respectively. Body size was associated with physiological heat tolerance in only one species (A. poncensis), with larger individuals being less heat tolerant (Table S1).

To test for physiological SDTT at the species level, I used phylogenetic generalized least squares regression (Grafen, 1989) and included data from all species (Table S1; phylogenetic information from Mahler et al., 2010). I modeled heat tolerance as a function of species body size and allowed residual error to vary with relatedness by including Pagel's λ as a parameter (Revell and Harmon, 2022). There was no relationship between body size and physiological heat tolerance across species (P=0.144; Fig. 4C). In summary, there is little evidence for physiological SDTT in Anolis lizards with respect to heat limits. Whether physical mechanisms of SDTT affect anoles and their overheating risk in their natural habitats remains to be seen.

To comprehensively test for both physiological and physical SDTT, both variable- and single-rate experiments could be applied and compared. There are many possible outcomes, a few of which are shown in Fig. 5. If SDTT is physical but not physiological, there will only be an effect of size in the variable-rate experiment (Fig. 5B,C). If SDTT is physiological but not physical, the SDTT relationship will be the same in both the variable- and single-rate experiments (Fig. 5D,G). If both physiological and physical SDTT are present, the two effects could act additively or synergistically in the same or opposing directions. For example, if both effects are in the same direction, a size relationship will be present with both experimental designs, but the slope will be steeper in the variable-rate experiment because both effects manifest simultaneously (Fig. 5E,I). In contrast, if the physiological and physical effects are in opposite directions, then the slope will be smaller in the variable-rate experiment (Fig. 5F,H).

The effect of single- versus variable-rate dynamic experiments on SDTT is nicely demonstrated by a study of heat tolerance in tropical ants (Kaspari et al., 2015). Thermal tolerance was measured on dozens of species ranging in mass from 0.01 to 57 mg. In one set of experiments, the authors found SDTT, with large ant species significantly more heat tolerant. However, they determined that large ants experienced significantly cooler temperatures during the experiments because their long legs keep them above the boundary layer. They subsequently conducted another set of experiments in which a new heating method was used that removed body temperature differences based on size. In those experiments, SDTT was no longer present (Kaspari et al., 2015). In effect, the authors conducted and compared variable- and single-rate experiments and found results tantamount to those shown in Fig. 5B.

Static thermal tolerance experiments are not dependent on thermal ramping but can still be confounded by physical SDTT. First, the equilibrium temperatures that animals experience in experimental and natural environments can differ with body size (Kaspari et al., 2015; Kearney et al., 2021; Pincebourde et al., 2021; Rubalcaba and Olalla-Tárraga, 2020; Stevenson, 1985; Woods, 2013). Second, the temperatures organisms experience in a static experimental treatment are not truly static. There is a thermal equilibration period when individuals are placed into and taken out of treatments, and equilibration time is likely to differ by size (Heatwole et al., 1969). Together, differences in equilibration temperature and equilibration time can cause large and small individuals to experience different magnitudes of thermal exposure in a given static treatment, affecting the measured thermal tolerance. Care should be taken in static experiments to ensure that physical size issues do not confound results when physiological variation in thermal tolerance is of interest (Kaspari et al., 2015).

Organisms live in spatially and temporally complex abiotic environments where physical interactions between body size and thermal conditions influence body temperature fluctuations. The implications of this have been best explored with respect to behavioral thermoregulation. Because size affects equilibrium body temperatures and rates of body temperature change, the body temperatures available within a habitat (i.e. the operative thermal environment) and the costs and benefits of moving between microclimates are size dependent (Christian et al., 2006; Kearney et al., 2021; Pincebourde and Casas, 2015; Sears and Angilletta, 2015; Seebacher and Shine, 2004; Stevenson, 1985). As a result, whether individuals need to behaviorally thermoregulate and the net fitness effect of doing so also depend on size.

Physical SDTT in natural environments is similarly mediated by differences in equilibrium body temperature and rate of body temperature change. In habitats where equilibrium body temperature is size dependent, the likelihood of organisms reaching thermal limits will differ by size. A recent global analysis of lizards predicted that large individuals in the tropics equilibrate to higher body temperatures, and therefore large individuals are more vulnerable to anthropogenic warming (Rubalcaba and Olalla-Tárraga, 2020). Furthermore, if thermal limits change with rate of body temperature change, then overheating or cooling risk will change with size. Although not explicitly analyzed with respect to body size, a macrophysiological study with insects illustrates the concept nicely (Allen et al., 2016). The study showed that higher heating rates lead to greater predicted warming tolerance owing to rate effects on realized heat limits. The preceding examples are consistent with large individuals being more vulnerable to temperature extremes owing to physical effects. However, that could be counterbalanced by size differences in the time it takes to reach thermal limits. Although slow heating rates can indeed lower realized thermal tolerance, they can also provide more time to move before a thermal limit is reached.

Ultimately, interactions between physiological and physical SDTT in nature will be context dependent, complicating the ease with which we can make predictions based on simple experiments or mechanistic models. If no physiological SDTT is present, organisms of different size could still differ in susceptibility to heat or cold owing to physical SDTT. Those effects could favor large or small individuals depending on the specifics of the organism and the habitat. If physiological SDTT is present in a system, it could either be enhanced or negated by physical SDTT effects that manifest within the habitat (Fig. 5). Analyses that integrate physical and physiological SDTT to derive generalizable predictions across organisms and habitats will be extremely valuable for understanding the mechanisms by which organisms of different size are affected by changing thermal conditions.

The thermal tolerance limits of ectotherms are often size dependent. In both experimental and natural settings, this size dependence could be due to differences in underlying physiology, physical interactions between the organism and environment or both. Disentangling these effects is important for a fundamental mechanistic understanding of thermal tolerance variation among individuals and species of different size, and for predicting how organisms will respond to changing temperatures.

I would like to thank Eric Riddell, Michael Logan, Art Woods, an anonymous reviewer, the Editors and members of my lab group for helpful feedback on earlier versions of this paper.