Powered flight has evolved several times in vertebrates and constrains morphology and physiology in ways that likely have shaped how organisms cope with infections. Some of these constraints probably have impacts on aspects of immunology, such that larger fliers might prioritize risk reduction and safety. Addressing how the evolution of flight may have driven relationships between body size and immunity could be particularly informative for understanding the propensity of some taxa to harbor many virulent and sometimes zoonotic pathogens without showing clinical disease. Here, we used a comparative framework to quantify scaling relationships between body mass and the proportions of two types of white blood cells – lymphocytes and granulocytes (neutrophils/heterophils) – across 63 bat species, 400 bird species and 251 non-volant mammal species. By using phylogenetically informed statistical models on field-collected data from wild Neotropical bats and from captive bats, non-volant mammals and birds, we show that lymphocyte and neutrophil proportions do not vary systematically with body mass among bats. In contrast, larger birds and non-volant mammals have disproportionately higher granulocyte proportions than expected for their body size. Our inability to distinguish bat lymphocyte scaling from birds and bat granulocyte scaling from all other taxa suggests there may be other ecological explanations (i.e. not flight related) for the cell proportion scaling patterns. Future comparative studies of wild bats, birds and non-volant mammals of similar body mass should aim to further differentiate evolutionary effects and other aspects of life history on immune defense and its role in the tolerance of (zoonotic) infections.
Powered flight has evolved at least three times in the evolutionary history of vertebrates and yet is one of the most energetically costly modes of transportation (Rayner, 1988). Birds and bats experience a 6- to 14-fold and >25-fold increase over resting metabolic rate, respectively, in metabolic expenditure during flight, whereas a similarly sized mammal only experiences a 6- to 8-fold increase during sustained running (Schmidt-Nielsen, 1972; Thomas, 1975). Although there is some debate over whether bats or birds are more efficient fliers (Muijres et al., 2012; Swartz et al., 2007; Tian et al., 2006), there are clear functional and physiological constraints associated with this costly activity (Maurer et al., 2004; Muijres et al., 2012). One of the most evident constraints is body size. Exceptionally large and small body sizes have apparently been selected against in the evolution of flying vertebrates because of demands imposed by the physics of flight (Stanley, 1973); however, the constraining factors for bats and birds likely differ, as the largest bats are much smaller than the largest flying birds. The evolution of flight and body size constraints may have had numerous direct and indirect effects on evolution of the immune system in flying vertebrates. For example, evolution of a lightened skeleton (Feduccia and Feduccia, 1999; Dumont, 2010) may affect how immune cells are differentiated and distributed throughout the body. If larger fliers are not as efficient at circulating protective cells throughout their bodies, then they might require greater quantities of cells (Ruhs et al., 2020 preprint). It should be noted that the high energetic costs of flight have varying impacts on the immune system (Hasselquist et al., 2007; Voigt et al., 2020; Nebel et al., 2012). While birds and bats have much in common in terms of constraints that accommodate the ability to fly, the evolution of flight likely impacted the dynamics between body size, physiological traits and the exposure risk to pathogens relative to that of non-volant birds and mammals.
Although various hypotheses predict distinct forms of scaling for aspects of immunity (Cohn and Langman, 1990; Dingli and Pacheco, 2006; Wiegel and Perelson, 2004), there is strong evidence that particular immune cells [namely concentrations of granulocytes, such as the neutrophils of mammals (Downs et al., 2020a) and heterophils of birds (Ruhs et al., 2020 preprint)] scale hypermetrically with body size. Importantly, patterns for cell concentrations (and other aspects of immunity such as antimicrobial capacity) do not follow the hypoallometric pattern that we would expect based on metabolic rate (i.e. at a rate of −0.25) and instead support the Safety Factor Hypothesis. The Safety Factor Hypothesis proposes that larger animals favor infection risk reduction by investing heavily in baseline safety (Downs et al., 2020a; Harrison, 2017), namely by using a reserve pool of broadly protective granulocytes (e.g. neutrophils and heterophils). However, concentrations of granulocytes in birds scale at a steeper rate (b=0.19; Ruhs et al., 2020 preprint) than in mammals (b=0.11; Downs et al., 2020a). Although this difference could be an evolutionary artifact or driven by one of many other differences between birds and mammals, this steep allometry has been hypothesized to be related to flight, which may put larger birds at a higher risk of pathogen exposure (Downs et al., 2019; Ruhs et al., 2020 preprint). It should also be noted that larger birds and bats also have longer lifespans than similarly sized non-volant mammals (Munshi-South and Wilkinson, 2010; Healy et al., 2014; Wilkinson and Adams, 2019). As reaching a large body size generally involves a long maturation time, this puts larger animals at increased risk of exposure to or infection with pathogens over their long lifespans (Tian et al., 2015; Harrison, 2017). The potential for larger fliers to prioritize risk-reduction immunological strategies motivated our interest to investigate immune scaling in bats, and among bats, birds and other mammals.
Bats are a hyperdiverse taxon (Order Chiroptera, over 1400 species) with a nearly global distribution across habitats ranging from rainforests to deserts (Gunnell and Simmons, 2012; Simmons and Cirranello, 2020). Their unique habits and life histories (e.g. powered flight, echolocation, long lifespans despite a small body size) make bats a notable taxon for basic studies of ecology and evolution (Ingala et al., 2018; Jones and Teeling, 2006; Wilkinson and South, 2002). Bats have also been increasingly studied for their ability to harbor some viruses that are detrimental and often lethal to humans and domestic animals (Brook and Dobson, 2015; Guth et al., 2019). Bats are confirmed reservoir hosts for henipaviruses, Marburg virus, various lyssaviruses and most SARS-like coronaviruses (Amman et al., 2015; Banyard et al., 2011; Halpin et al., 2011; Li et al., 2005). Yet, with some exceptions (e.g. Rabies lyssavirus), these viruses appear to not kill and rarely cause clinical disease in bats (Williamson et al., 2000).
Whereas the high diversity of zoonotic viruses in Chiroptera might be partly driven by the speciose nature of this Order (Mollentze and Streicker, 2020), bat tolerance of particular viruses may be shaped by specialized immune mechanisms in these flying mammals (Brook et al., 2020; Zhang et al., 2013). Bat immunoglobulins and leukocytes are structurally similar to those of humans and mice (Baker et al., 2013), but bats also have unique immune system traits such as complement proteins robust to temperature change, lack of fever with bacterial (lipopolysaccharide) challenge, high constitutive expression of type I interferons, and dampened inflammation (Ahn et al., 2019; Hatten et al., 1973; Pavlovich et al., 2018; Stockmaier et al., 2015; Zhou et al., 2016). Flight may explain these distinctions, including increased metabolic rate that enables stronger immune responses and elevated body temperature that could mirror febrile responses to control infection (O'Shea et al., 2014; but see Levesque et al., 2021). However, the primary hypothesis for how bats can tolerate viruses is that they evolved mechanisms to minimize or repair the negative effects of oxidative stress generated as a consequence of flight (Zhang et al., 2013). For example, some bat species show resistance to protein oxidation and unfolding (Salmon et al., 2009), reduced lipid peroxidation (Wilhelm Filho et al., 2007), and lower hydrogen peroxide per unit oxygen consumed (Brunet-Rossinni, 2004). This propensity to resist acute oxidative stress and repair oxidative damage could have also helped bats cope with viral replication that would have otherwise caused cell damage (Kacprzyk et al., 2017; Xie et al., 2018; Zhang et al., 2013).
Here, we first asked whether leukocyte proportion scaling in bats is distinct from that of other taxa already described. Then, we asked whether the ability to fly (i.e. bats and birds) explains immune cell proportion allometry across extant vertebrate endotherms. Most studies assessing immunity in bats have been limited to few species (but see Schneeberger et al., 2013). We combined field-collected data from Neotropical bats with data from the primary literature to maximize sample sizes as well as phylogenetic and body size diversity. We then quantified scaling relationships for proportions of two primary leukocytes for which abundant data are available, lymphocytes and granulocytes. Lymphocytes include B and T cells, which provide specific, but time-lagged, protection through antibody production and coordination of cascading immune responses. Granulocytes (neutrophils in mammals and heterophils in birds) are phagocytes that rapidly protect against pathogens without education or much specificity (Lanier, 2013), although high concentrations of these cells can also promote tissue damage (Smith, 1994). Finally, we directly compared scaling relationships for cell proportions in bats with those of birds and non-volant mammals using an existing database (ZIMS, https://www.species360.org/products-services/zoo-aquarium-animal-management-software-2/, retrieved April 2019).
We predicted the forms of relationships between cell proportions and body size based on previously discovered scaling patterns among body size and cell concentrations (Downs et al., 2020a,b preprint; Ruhs et al., 2020 preprint). Proportions, however, sometimes pose difficulties for studies of allometry because they are bound rather than having no continuous upper limit and are inherently co-dependent (i.e. one cell type goes up, another goes down). As leukocyte concentration data for bats are extremely rare in the literature, and proportional data permitted comparisons that were otherwise presently impossible, we estimated scaling patterns using proportional data but encourage caution in comparing results from this analysis against prior scaling for leukocyte concentrations. We hypothesized isometry for lymphocytes in bats, as was observed previously for bird cell concentrations (Ruhs et al., 2020 preprint). We expected isometry to manifest because lymphocytes are a functionally diverse group of cells including both B and T cells (Lanier, 2013), the proportions of which could vary dramatically among species. By contrast, granulocyte functions are fairly homogeneous, so we predicted hypermetric granulocyte scaling in bats as was observed in other mammals and birds (Downs et al., 2020a; Ruhs et al., 2020 preprint). However, as larger fliers might overinvest in safety, we expected bat neutrophil proportions to scale hypermetrically, to the same degree (steeper than in non-volant mammals) as was observed for heterophil concentrations in birds (predictions based on whether flight influences scaling; Fig. 1; Ruhs et al., 2020 preprint). Alternatively, it is possible we could observe no impact of flight on cell proportion allometries, which could be due to life-history features (e.g. reproduction, sociality) or equal investment in risk-reduction strategy as a result of factors such as increased lifespan, regardless of body size.
MATERIALS AND METHODS
During April and May in 2017 and 2018, we sampled 160 bats from 26 species in the Orange Walk District of Belize (Herrera et al., 2018; Becker et al., 2020a, 2021). Bats were captured using mist nets (monitored continuously from approximately 19:00 h to 22:00 h) and harp traps (monitored every half-hour from 18:00 h to 22:00 h and then at 05:00 h the following morning); all individuals were identified to species based on morphology (Reid, 1997). We collected blood by lancing the propatagial vein with a sterile needle, followed by collection using heparinized capillary tubes. Thin blood smears were prepared and stained with buffered Wright–Giemsa (Astral Diagnostics Quick III). Most bats were sampled within 1–3 h of capture, and all bats were released after processing. Sampling followed guidelines for safe and humane handling of bats from the American Society of Mammalogists (Sikes and Gannon, 2011) and was approved by the Institutional Animal Care and Use Committees of the University of Georgia (A2014 04-016-Y3-A5) and American Museum of Natural History (AMNHIACUC-20170403). Sampling was authorized by the Belize Forest Department under permits WL/2/1/17(16), WL/2/1/17(19) and WL/2/1/18(16).
Bat leukocyte data
We used light microscopy (1000×) to quantify the proportion of neutrophils and lymphocytes from 100 leukocytes from each field sample (Schneeberger et al., 2013). As Neotropical bats are relatively limited in their range of body masses, we supplemented our leukocyte dataset with a systematic literature search (Fig. S1). We identified articles using Web of Science and the search terms TS=(bat OR Chiroptera OR flying fox) AND (hematology OR white blood cell OR leukocyte). For bat species sampled across multiple studies, we averaged cell proportions. When available, body mass of each bat species was extracted from EltonTraits (Wilman et al., 2014); however, for a few species (n=2), mass was averaged from the source paper. The literature search substantially increased our body mass range (from approximately 5–78 g to 4–804 g; see Fig. S2 for a comparison across all extant bat species; Wilman et al., 2014). In our dataset, our smallest bat species was the proboscis bat (Rhynchonycteris naso) at ∼3.8 g and our largest bat species was the large flying fox (Pteropus vampyrus) at 935.9 g. Within-species sample size ranged from 1 to 160 (mean±s.e.m. 21±4) but did not predict proportions of either cell type (neutrophils: ρ=−0.11, P=0.41; lymphocytes: ρ=0.09, P=0.48). We then extracted species means of lymphocyte and neutrophil proportions in whole blood from ZIMS (https://www.species360.org/products-services/zoo-aquarium-animal-management-software-2/, retrieved April 2019). ZIMS is a repository of veterinary data from captive, adult animals housed in facilities accredited by the Association of Zoos and Aquariums and considered healthy. As bats are rare in captivity, the ZIMS dataset only included nine species, six of which were species we had in our dataset from the literature. Therefore we calculated species averages for each cell type and the final bat dataset included 63 species.
Bird and non-volant mammal data
To compare bat leukocyte data with comparable data from birds and non-volant mammals, we extracted species means of lymphocyte and granulocyte (neutrophils/heterophils) proportions in whole blood from ZIMS (https://www.species360.org/products-services/zoo-aquarium-animal-management-software-2/, retrieved April 2019). We removed bat data (n=3) from the extracted ZIMS mammal database and any non-volant birds from the ZIMS bird database (n=14). When cleaning the data, we only included data from Global Species Reference Intervals. We compiled standardized species-level body mass data from the CRC Handbook of Avian Masses (Dunning, 2007) and/or publicly available databases such as AnAge (Tacutu et al., 2013), the Animal Diversity Website (Jones et al., 1997) and the Encyclopedia of Life (Parr et al., 2014). In our bird dataset, body mass ranged from a 13.8 g Gouldian finch (Erythrura gouldiae) to a 22,600 g Andean condor (Vultur gryphus). In our non-volant mammal dataset, body mass ranged from a ∼124 g Pygmy marmoset (Callithrix pygmaea) to a 3,824,540 g African elephant (Loxodonta africana).
Exercise 1: best-fit models for leukocyte proportion allometries in birds, bats and non-volant mammals
Our modeling progressed in two stages. First, to test hypotheses about allometric scaling of leukocytes in bats only, we used phylogenetic generalized mixed-effects models (GLMMs) with the ape and MCMCglmm packages in R (Hadfield, 2010; Paradis et al., 2004). All models included phylogenetic effects from a phylogeny produced in PhyloT using data from the National Center for Biotechnology Information (Letunic, 2015) and with resolved polytomies. We used that tree to create two phylogenetic covariance matrices, one for bat-only analyses and one that we used later for direct comparisons of scaling slopes across taxa. We set the inverse-gamma priors to 0.01 for the random effect of phylogenetic variance and default priors for the fixed effects in all models. All models were run for 260k iterations with 60k burn-in and a 200-iteration thinning interval (Downs et al., 2020a; Ruhs et al., 2020 preprint). For all models, we used Deviance Information Criterion (ΔDIC) to identify the best-fit GLMM. We defined the top model as that with the lowest DIC, and we considered models within ΔDIC<5 as having equivalent support (Richards, 2005). For all models, we also calculated Pagel's unadjusted λ and conditional and marginal R2 (Housworth et al., 2004; Nakagawa and Schielzeth, 2013). We then used this approach to determine the scaling relationship for lymphocyte and neutrophil proportions, separately, across 63 bat species. Also, because previously published slopes for mammal and bird leukocyte scaling used cell concentrations (Downs et al., 2020a; Ruhs et al., 2020 preprint), we determined the scaling relationships of cell proportion data for both birds (n=400) and non-volant mammals (n=251) independently to facilitate direct comparisons with bats. Results from bird and non-volant mammal-only models are presented in Tables S1 and S2. For each taxonomic group and each cell type, we produced two sets of a priori models:
Model 1: log10(leukocyte proportion)=log10a+ε,
Model 2: log10(leukocyte proportion)=log10a+b×log10(body mass) +ε.
Model 1 represents a null model with b=0, whereas model 2 estimates the scaling relationship between body mass and cell proportion.
Exercise 2: direct comparisons of allometries among taxa
Next, we directly compared the slopes of relationships between body mass and immune cell type in bats (n=63), birds (n=400) and non-volant mammals (n=251). Specifically, we fitted five models to the data and compared DIC scores to determine the best-fit versions:
Model 3: log10(leukocyte proportion)=log10a+ε,
Model 4: log10(leukocyte proportion)=log10a+b×taxon+ε,
Model 5: log10(leukocyte proportion)=log10a+b×log10(body mass) +ε,
Model 6: log10(leukocyte proportion)=log10a+b1×taxon+b2×log10(body mass)+ε,
Model 7: log10(leukocyte proportion)=log10a+b1×taxon+b2×log10(body mass)+b3 · log10(body mass)×taxon+ε.
Here, model 3 represents a null model with b=0, and model 4 only tests for mean differences in cell proportions per taxon (i.e. bat, bird, non-volant mammal), irrespective of body mass. Model 5 is analogous to model 2 from exercise 1, and it estimates a global scaling relationship between body mass and cell proportions across all taxa. Lastly, model 6 combines models 4 and 5 (i.e. mean differences between taxa and a global body mass slope), whereas model 7 explicitly tests whether scaling relationships between body mass and cell proportions differ among taxa.
Exercise 1: best-fit models for leukocyte proportion allometries in birds, bats and non-volant mammals
For bats, the intercept-only model (fitting b=0; model 1) and the mass model (model 2) received equivalent support for both lymphocytes and neutrophils (Table 1, Fig. 2). Slopes for lymphocytes (b=−0.06, credible interval CI −0.21:0.09) and neutrophils (b=0.06, CI −0.1:0.2) were indistinguishable from zero (Table 2, Figs 2 and 3). Although these results suggest isometry for lymphocytes and neutrophils, we encourage caution in their interpretation given that null models can represent either a true slope of zero or a lack of power to find allometry. For lymphocytes, phylogeny accounted for 34% of the variation and body mass accounted for <1% of the variation. For neutrophils, again, phylogeny accounted for 18% of the variation and body mass accounted for <1% of the variation. For both leukocyte types, the model fit of the mass model was 38–40% and did not increase much from the 27–33% of the intercept-only model. Collectively, our results suggest little allometric scaling of proportions of either cell type among bat species.
For birds, the mass model (model 2) was best supported (Table S1) for both cell types; lymphocytes scaled hypometrically (b=−0.07, CI −0.09:−0.06; Table S2; Fig. 3) and heterophils scaled hypermetrically (b=0.08, CI 0.07:0.1; Table S2). For non-volant mammals, the lymphocyte and neutrophil mass models were also the best supported (Table S1); lymphocytes scaled hypometrically (b=−0.08, CI −0.1:−0.06; Table S2; Fig. 3) and neutrophils scaled hypermetrically (b=0.04, CI 0.02:0.05; Table S2). Phylogeny explained between 61% and 63% of the variation in birds and 57–69% of the variation in non-volant mammals (Tables S2). Bird and non-volant mammal granulocytes and non-volant mammalian lymphocyte cell proportion scaling patterns were consistent in direction (but not magnitude) with previous analyses of cell concentrations (Downs et al., 2020a; Ruhs et al., 2020 preprint). However, bird lymphocyte proportions were hypometric here, whereas no evidence of allometry in lymphocyte concentrations was previously reported (Ruhs et al., 2020 preprint).
Exercise 2: direct comparisons of allometries among taxa
When bat lymphocyte proportions were directly compared with those of birds and non-volant mammals, we found equivalent support for the mass model (model 5; Table 3, Fig. 2), the model with independent effects of mass and taxon (model 6; ΔDIC=0.3) and the model in which allometries differed between taxa (model 7; ΔDIC=4.85). For the mass model (model 5), phylogeny explained a large proportion of the variance (∼74%), but the addition of mass increased explanatory power by 12% (marginal R2). For model 6, the addition of mass and taxon increased explanatory power by an additional 11% (marginal R2). When examining models that compared the lymphocytes among taxa (models 3, 4, 6, 7), bats were different from non-volant mammals but not birds (Table S3). Because the slope for bat lymphocyte proportions and body mass was indistinguishable from zero (from model 2, b=−0.06, CI −0.21:0.09; exercise 1), this effect was largely driven by differences in the intercept between taxa (−0.32 in bats; −0.17 in birds; −0.08 in non-volant mammals). In other words, mean lymphocyte proportions across all body sizes were lower in bats than in other mammals.
Granulocyte proportions were best explained by the model in which intercepts of allometries varied among taxa (model 7); this was driven by a universal hypermetric scaling slope across all species (b=0.06, CI 0.004:0.11). Phylogeny explained 57% of the total variation. Inclusion of mass and taxa increased explanatory power by 13% (marginal R2). However, there was equal support for model 5 (ΔDIC=3.29) and model 6 (ΔDIC=2.37). For the mass model (model 5), phylogeny explained a large proportion of the variance (∼78%), but the addition of mass increased explanatory power by 12% (marginal R2). For model 6, the addition of mass and taxon increased explanatory power by an additional 11% (marginal R2). When examining models comparing granulocyte proportions among taxa (models 3, 4, 6, 7), bat slopes and intercepts were not different from those of non-volant mammals or birds (Table S4). Therefore, phylogeny explains the majority of the variance in the proportions of both cell types, but body mass informs a moderate percentage of this variation, especially compared with taxon alone (marginal R2=0.03). The bat leukocyte data and endotherm metadata used in these analyses are given in Table S5.
Although bats and birds represent two independent evolutionary origins of flight (Rayner, 1988), both are flying endotherms. We therefore predicted they would be subject to similar selective pressures on their physiology that would ultimately impact the architecture of their immune system (McGuire and Guglielmo, 2009). Here, we quantified the scaling relationships for proportions of two leukocyte types across 63 species of bats and compared these patterns across other vertebrate endotherms. Broad, comparative analyses of immunity between bats and other taxa are generally rare (Becker et al., 2019; Shaw et al., 2017), and allometry provides a powerful analytic framework for systematically comparing such data across species (Downs et al., 2020a,b preprint; Ruhs et al., 2020 preprint). Therefore, our analyses aimed to shed light on variation in bat cellular immunity in general, and the role of flight in shaping immune scaling relationships.
When we examined body mass effects on bat cell proportions (posterior means from exercise 1; Fig. 2), we found little evidence for allometric scaling of either cell type across bat species. When comparing across taxa (exercise 2), however, bat lymphocyte proportions (represented by the intercept) more closely resembled those of birds than of non-volant mammals, as the credible intervals for bats overlapped with those for birds but not with those of non-volant mammals. However, bat neutrophil proportions were not distinguishable from those of birds or non-volant mammals, and all taxa tended to scale hypermetrically. Therefore, our results support the idea of bat immune systems having some distinctions from those of other endotherms but they also suggest that physiological alterations to facilitate flight may not explain the allometry of cell proportions. Further, because birds display a much steeper hypermetric slope for granulocytes (Ruhs et al., 2020 preprint), avian scaling patterns are likely driven by other physiological, life-history or exposure risk factors, rather than flight, that facilitated the need for larger birds to have disproportionately more circulating cells. We caution that our inability to distinguish bats from other endotherms in exercise 2 is likely driven by the high variation in bat data, thereby complicating the identification of allometric patterns among taxa. Additionally, and from an applied perspective, allometry of immune cells may also provide limited inference into differences in traits such as pathogen tolerance compared with other, more specific immune measures such as antimicrobial capacity and immune-associated gene expression. For leukocytes in particular, having more granulocytes may even promote pathology rather than host protection (Smith, 1994). However, comparative analyses such as these can still provide insights into how flight has shaped general immune scaling patterns and help generate predictions for future research into more functional immunological differences among taxa. Below, we focus on the results from exercise 1 to discuss taxon-specific scaling patterns and the (likely) lack of allometry in bats. We then address immunological similarities between bats, birds and non-volant mammals and what this might mean for pathogen tolerance to motivate future comparative research.
Are bats immunologically different?
As reaching a large body size requires a slow pace of life and involves greater resource turnover, larger animals are likely at greater risk of pathogen exposure (Lee, 2006; Tian et al., 2015; Harrison, 2017). Because of their long lifespans, increased energetic demands, and adaptive constraints associated with flight, large fliers likely have greater risk of pathogen exposure, even more so than large mammals. Thus, we predicted that bats, like birds (Ruhs et al., 2020 preprint), might need a disproportionately greater proportion of broadly protective cells than non-volant species, especially at larger body sizes. However, we found little evidence for intercept or slope differences between bats and non-volant species or for allometric scaling in bat cell proportions. Interestingly, while bats did not show any allometries when analyzed alone (exercise 1), in exercise 2, where cell proportions were compared across taxa, bat lymphocyte proportions more closely resembled those of birds than those of non-volant mammals (taxa models panel; Fig. 2). The (likely) lack of scaling, or possible isometry, in bats is intriguing given the evident allometric scaling patterns observed previously for birds and across primarily terrestrial mammals, for both cell concentrations and the proportions analyzed here (Ruhs et al., 2020 preprint).
Importantly, contrasting leukocyte patterns between bats and other taxa could be influenced by most bat data being from wild populations and the bird/mammal data being from captive populations. Wild populations are inherently more immunologically variable (Viney and Riley, 2014), which is reflected in the large credible intervals around our posterior mean estimates in bats. Many factors can influence wild-derived variation, but wild and captive animals can especially differ in pathogen exposure and stressors that can affect leukocyte composition (Davis et al., 2008; Johnstone et al., 2012; Herrera et al., 2019). For example, while captive animals are generally thought to have many constraints alleviated (e.g. ad libitum food, less pathogen exposure), which would lessen the need for investment in cellular defense, stress of captivity and handling can cause a decrease in lymphocytes and an increase in heterophils (Tian et al., 2015; Parker Fischer and Romero, 2019). Further, physiological adjustment to captivity and handling is likely species dependent and could take months (Parker Fischer and Romero, 2019). The vertebrate response to stress can alter white blood cell counts within minutes to hours, depending on the species and even across taxa and possibly with body size (Davis et al., 2008; Johnstone et al., 2012). In our wild-derived bat data from Belize and the broader literature, many blood samples were collected within 2 h of capture, although some studies did not include these details or had longer gaps between capture and processing. Importantly, exclusion of these studies or individuals known to be held for long durations did not influence the lack of scaling patterns of bat leukocytes (bat-only model 2; lymphocytes b=−0.09, CI −0.23:0.04; neutrophils b=0.07, CI −0.06:0.23).
The absence of allometry in our bat data (when analyzed alone in exercise 1), which are primarily from wild populations, might be more likely to reflect true developmental and environmental pressures on these species. In fact, another study demonstrated that the variation in cell proportions from wild species did not influence scaling patterns, as wild and captive birds displayed similar allometries (L.B.M., E.C.R., S.J.O. and C.J.D., in review). Further, mean heterophil proportions were lower in wild versus captive birds. Similarly, comparison of wild and captive rodents also revealed no differences between scaling patterns of total leukocytes or neutrophil counts, despite wild animals having lower mean lymphocyte counts than captive animals (Tian et al., 2015). Taken in sum, the lack of scaling patterns found here is unlikely to be driven by variation in wild bat cell proportions; however, data from wild populations are likely ideal to assess the drivers and consequences of variation in immunity (Tian et al., 2015; Becker et al., 2020b). Lastly, the relatively smaller sample size of bats (n=63) compared with birds and non-volant mammals (n=400 and 251, respectively) is also unlikely to drive the greater variance in our bat data. Randomly subsampling our bird and non-volant mammal data produced approximately equivalent estimates of intercepts (bird lymphocytes a=−0.09, CI −0.45:0.21; heterophils a=−0.6, CI −0.95:−0.28; non-volant mammal lymphocytes a=−0.08, CI −0.41:0.2; neutrophils a=−0.43, CI −0.63:−0.24) and slopes (bird lymphocytes b=−0.11, CI −0.14:−0.07; heterophils b=0.1, CI 0.06:0.13; non-volant mammal lymphocytes b=−0.09, CI −0.12:−0.04; neutrophils b=0.03, CI 0.01:0.06) for model 2, suggesting that the lack of allometry in bats is also unlikely to be driven by sample size.
The small differences in bat versus other taxa leukocyte proportion allometries support similar efforts to understand constitutive expression of other aspects of the bat immune system. For example, comparative genomic analyses show several unusual immunological aspects of bats, such as high constitutive expression of type I interferons and dampened inflammation (Ahn et al., 2019; Pavlovich et al., 2018; Zhou et al., 2016). The potential lack of cell allometry in bats suggests yet another distinct aspect of immunology in these flying mammals. However, our inability to allometrically distinguish bats from birds (for lymphocytes) and bats from all other taxa (for neutrophils) in our cross-taxa comparison also suggests there may be other ecological explanations rather than flight itself for the observed leukocyte proportion scaling patterns.
Although it is most likely that our results from exercise 1 support a lack of allometry in bats (as described above), it is impossible to entirely discount the alternative explanation for our results in that the bat data could instead represent isometry. Isometry would mean that bat species require the same proportions of leukocytes across body size, which could be driven by certain distinct aspects of bat biology such as their slow life history. Bats, even more so than birds, are long lived such that selection for safety and disease risk reduction is likely uniformly prioritized to accommodate longevity across body mass. Although body mass affects longevity in bats, other factors such as hibernation, cave use and latitude all have effects on lifespan of similar magnitude to those on mass (Wilkinson and Adams, 2019), which may complicate detecting mass effects on physiology or morphology linked to lifespan. For example, the high sociality and gregariousness of many bat species facilitates contact during roosting that could increase pathogen exposure (Kerth, 2008; Kunz, 1982; Webber et al., 2017) and would be generally equal across body size. To reiterate, although it is most likely that the bat data here represent a lack of scaling, these alternative explanations support the possibility that our bat data might instead represent isometry of lymphocyte proportions when directly comparing bats with other taxa.
Increased spillover of zoonotic viruses, such as henipaviruses and coronaviruses, has renewed public and scientific interest in whether bats are immunologically unique hosts (Brook and Dobson, 2015; Halpin et al., 2011; Li et al., 2005; Luis et al., 2013). Investigating allometric scaling patterns of immunological features, including but not limited to leukocytes, could shed light on the physiological traits that impact host ability to tolerate virulent pathogens. We here demonstrate some differences in the scaling patterns of innate immune cell proportions between taxa of endotherms; however, we did not observe substantial effects of body size on cell proportions in bats. It is important to note the difficulty in identifying allometric patterns of cell proportions (i.e. which are bound between 0% and 100%) compared with the previous discovery of hypermetric scaling of cell concentrations (Downs et al., 2020a; Ruhs et al., 2020 preprint). Future studies interested in immune allometry instead should measure cell concentrations, which are not bound by proportional limits and provide greater insight into the total stock of cellular immunological resources. Our sample also represents only a small fraction of bat diversity (about 4% of the >1400 species; Simmons and Cirranello, 2020), although our data do span the body mass continuum of extant bat species. To enhance our ability to examine relevant immunological patterns, we encourage greater quantification of immune components across the bat phylogeny, specifically within bat clades characterized by relatively larger body sizes (e.g. Pteropodidae).
Lastly, we focused on cell proportion allometry and the potential for body mass alone to explain immunological differences among species (Downs et al., 2019). Flying endotherms can vary in other ecological traits besides body mass that also shape pathogen exposure and immune investment, such as diet, coloniality and roost type (Minias et al., 2017; Schneeberger et al., 2013). To address such trait comparisons across equal ecological context (i.e. avoiding captive–wild contrasts), future comparative studies of wild bats, birds and non-volant mammals of similar body masses could help to confirm the patterns observed here. Such work could further differentiate evolutionary effects from those of flight and other aspects of life history on immune defense and provide insights into the traits that promote pathogen tolerance.
For assistance with field logistics and permits, we thank Mark Howells, Neil Duncan and the staff of the Lamanai Field Research Center. We also thank the many colleagues who helped net bats during 2017 and 2018 bat research in Belize as well as Hannah Droke, Ellen Chinchilli and Grace Carey for help compiling bat, bird and mammal data. Lastly, we thank members of the Martin lab at the University of South Florida, the Ketterson lab at Indiana University, and anonymous reviewers for constructive feedback.
Conceptualization: E.C.R., D.J.B., L.B.M., C.J.D.; Methodology: E.C.R., D.J.B., S.J.O., L.B.M., C.J.D.; Software: E.C.R., D.J.B.; Validation: E.C.R., D.J.B.; Formal analysis: E.C.R., D.J.B.; Investigation: E.C.R., D.J.B., S.J.O., O.O., M.B.F., N.B.S., L.B.M., C.J.D.; Resources: L.B.M., C.J.D.; Data curation: E.C.R., D.J.B.; Writing - original draft: E.C.R., D.J.B.; Writing - review & editing: E.C.R., D.J.B., S.J.O., O.O., M.B.F., N.B.S., L.B.M., C.J.D.; Visualization: E.C.R., D.J.B.; Supervision: L.B.M., C.J.D.; Project administration: L.B.M., C.J.D.; Funding acquisition: D.J.B., N.B.S., L.B.M., C.J.D.
This work was funded by the Achievement Rewards for College Scientists Foundation and The Explorers Club (D.J.B.), American Museum of Natural History (Theodore Roosevelt Memorial Fund to D.J.B., Taxonomic Mammalogy Fund to N.B.S.), and National Science Foundation (IOS 1656618 to L.B.M., IOS 1656551 to C.J.D.).
The authors declare no competing or financial interests.