Introduction
Temperature governs many biological processes, affecting different levels of biological organisation (Verberk et al., Reference Verberk, Overgaard, Ern, Bayley, Wang, Boardman and Terblanche2016). Changes in temperature can have profound effects on the functioning of cells, tissues, and organ systems (Brown et al., Reference Brown, Gillooly, Allen, Savage and West2004). At the organismal level, changes in environmental temperature can directly affect reproduction, growth, and survival (Deutsch et al., Reference Deutsch, Tewksbury, Huey, Sheldon, Ghalambor, Haak and Martin2008). Consequently, temperature can determine the abundance and dynamics of species populations within communities and, ultimately, species distribution ranges (Bujan et al., Reference Bujan, Roeder, de Beurs, Weiser and Kaspari2020). Given that global temperature has increased considerably in recent decades as a result of climate change (IPCC, 2021), it is paramount to understand the possible impacts of temperature increases at different levels of biological organisation (Parr and Bishop, Reference Parr and Bishop2022).
Ectothermic organisms are highly susceptible to temperature changes because their body temperatures completely depend on environmental thermal conditions (Béltran et al., Reference Béltran, Houzel-Herculanos, Sinervo and Whiting2021). In other words, temperature mediates all the physiological reactions of ectotherms, impacting their functional ecology (Angilletta, Reference Angilletta2009). Thermal performance curves (TPCs) are commonly used to assess how body temperature affects performance and fitness in ectotherms, giving rise to predictions about how these organisms may be affected by climate change (Sinclair et al., Reference Sinclair, Marshall, Sewell, Levesque, Willett, Slotsbo, Dong, Harley, Marshall, Helmuth and Huey2016). An organism's critical thermal limits are defined by the minimum temperature (CTmin) and maximum temperature (CTmax) it can withstand without losing motor coordination (Lutterschmidt and Hutchison, Reference Lutterschmidt and Hutchison1997). These limits are often incorporated into TPCs as benchmarks to improve understanding of relevant physiological limits and to frame the response of key biological traits. Special attention has been paid to CTmax because it can indicate the vulnerability of organisms to present and future temperature increases (Diamond et al., Reference Diamond, Sorger, Hulcr, Pelini, Del Toro, Hirsch, Oberg and Dunn2012). However, CTmax can vary greatly among ectotherms and levels of biological organisation (i.e. individuals, populations, and communities; Verble-Pearson et al., Reference Verble-Pearson, Gifford and Yanoviak2015; Nascimento et al., Reference Nascimento, Câmara and Arnan2022). Furthermore, this variation may be shaped by many biotic and abiotic factors (Sunday et al., Reference Sunday, Bates, Kearney, Colwell, Dulvy, Longino and Huey2014; Leiva et al., Reference Leiva, Calosi and Verberk2019).
Among ectotherms, ants stand out because they are ubiquitous and abundant in almost all terrestrial ecosystems, where they mediate many ecological processes and provide essential ecosystem services such as seed dispersal, protection against herbivores, and nutrient cycling (Del Toro et al., Reference Del Toro, Ribbons and Pelini2012; Elizalde et al., Reference Elizalde, Arbetman, Arnan, Eggleton, Leal, Lescano, Saez, Werenkraut and Pirk2020). Because they are social insects, ants are an excellent model for assessing differences in CTmax at different levels of biological organisation and for examining the factors behind these differences. Indeed, in ants, CTmax can be measured at the colony level (among nestmate workers), population level (among colonies of the same species), and community level (among different species) (Baudier and O'Donnell, Reference Baudier and O'Donnell2020; Bujan et al., Reference Bujan, Roeder, de Beurs, Weiser and Kaspari2020; O'Donnell et al., Reference O'Donnell, Bulova, Caponera, Oxman and Giladi2020). That said, most research to date has focused on ant CTmax mainly at the community level; while some colony-level studies exist as well, population-level studies are scarce (Nascimento et al., Reference Nascimento, Câmara and Arnan2022). Also rare is work that simultaneously analyses differences in CTmax at all three levels (but see Verble-Pearson et al., Reference Verble-Pearson, Gifford and Yanoviak2015). However, taking a more holistic approach is crucial in better understanding how species may be affected under future conditions of climate change (Roeder et al., Reference Roeder, Roeder and Bujan2021). It is necessary to consider multiple organisational levels in tandem because ant colonies may have subclasses of workers that differ in body size and CTmax (Baudier and O'Donnell, Reference Baudier and O'Donnell2020). Ant body size may also vary among colonies as a result of environmental factors (Shik et al., Reference Shik, Arnan, Oms, Cerdá and Boulay2019; Oliveira et al., Reference Oliveira, Lima, Leal and Arnan2022), which can lead to variation in CTmax within species (Baudier and O'Donnell, Reference Baudier and O'Donnell2020). While a few studies have looked at a small number of highly polymorphic species (i.e. those displaying marked within-colony variation in worker size), we do not know how much CTmax varies among workers across a broader range of polymorphic species or in monomorphic species (i.e. those displaying limited within-colony variation in worker size). At the colony level, thermal performance can be better assessed by using the temperature-dependent rates of key biological processes, such as metabolic rates (Shik et al., Reference Shik, Arnan, Oms, Cerdá and Boulay2019) or brood development rates (Penick et al., Reference Penick, Diamond, Sanders and Dunn2017). While CTmax reflects just one aspect of performance, it is also an important thermal trait because it affects worker foraging and, consequently, colony energy supply (Arnan et al., Reference Arnan, Lázaro-González, Beltran, Rodrigo and Pol2022). If selection pressure is greater on more thermally vulnerable workers, it will take a toll on colony fitness and persistence (Baudier and O'Donnell, Reference Baudier and O'Donnell2017). In such cases, temperature increases could have negative effects at the colony level, an impact that would be obscured when exclusively analysing community-level responses.
Ant CTmax can differ due to many biotic and abiotic factors (Roeder et al., Reference Roeder, Roeder and Bujan2021; Nascimento et al., Reference Nascimento, Câmara and Arnan2022). Studies have looked at how ant CTmax relates to microhabitat use (e.g. Baudier et al., Reference Baudier, Mudd, Erickson and O'Donnell2015, Reference Baudier, D'Amelio, Malhotra, O'Connor and O'Donnell2018; Kaspari et al., Reference Kaspari, Clay, Lucas, Yanoviak and Kay2015), evolutionary history (e.g. Diamond et al., Reference Diamond, Sorger, Hulcr, Pelini, Del Toro, Hirsch, Oberg and Dunn2012; Arnan and Blüthgen, Reference Arnan and Blüthgen2015), and body size (e.g. Kaspari et al., Reference Kaspari, Clay, Lucas, Yanoviak and Kay2015; Verble-Pearson et al., Reference Verble-Pearson, Gifford and Yanoviak2015). However, the results of this research have been inconsistent. Furthermore, such work has rarely examined the combined effects of these factors and their ability to explain variation in ant CTmax. For example, microhabitat appears to have a strong effect on ant CTmax (Baudier et al., Reference Baudier, Mudd, Erickson and O'Donnell2015, Reference Baudier, D'Amelio, Malhotra, O'Connor and O'Donnell2018; Kaspari et al., Reference Kaspari, Clay, Lucas, Yanoviak and Kay2015; Bujan et al., Reference Bujan, Roeder, de Beurs, Weiser and Kaspari2020) that could be underlain by species evolutionary history because microhabitat use is a highly conserved trait in ants (Lucky et al., Reference Lucky, Trautwein, Guenard, Weiser and Dunn2013). Thus, it is only possible to determine each factor's contribution if both are analysed in tandem.
Simultaneously exploring how different factors drive CTmax in ants is the foundation for understanding how ants will respond to future temperature increases, including which species may be more susceptible or possess greater adaptive potential (Nascimento et al., Reference Nascimento, Câmara and Arnan2022). Such information is especially critical for tropical regions, where ants already live close to their CTmax values and studies remain infrequent (Diamond and Chick, Reference Diamond and Chick2018). In the dry tropical forest of northeastern Brazil, for example, climate models predict a 3–6°C increase in temperature by 2100 (Magrin et al., Reference Magrin, Marengo, Boulanger, Buckeridge, Castellanos, Poveda, Scarano and Vicuña2014), which will likely expose the region's ant fauna to severe heat stress.
Given this context, our study aims to characterise CTmax in ants inhabiting a semiarid neotropical region and to understand the underlying factors (i.e. phylogeny, nesting microhabitat, and body size) operating at different levels of biological organisation – the community level (among species), the population level (among colonies within species), and the colony level (among workers within colonies). We addressed the following five questions: (1) How variable is CTmax within an ant community? (2) Are differences in community-level CTmax explained by species nesting microhabitat, body size, and/or evolutionary history? (3) Do colonies of the same species exhibit different CTmax values? (4) Do workers from monomorphic vs. polymorphic colonies exhibit different CTmax values? (5) Which level of organisation displays the greatest degree of variability in CTmax?
Materials and methods
Study area
This study was conducted in the city of Garanhuns (Latitude: 8°53′27″ South, Longitude: 36°29′48″ West), located in the rural region of Pernambuco, northeastern Brazil. The mean annual temperature is 20°C; temperatures decrease to a minimum of 15°C in the winter and increase to a maximum of 30°C in the summer. The climate is hot, tropical, subhumid, and dry (Barbosa et al., Reference Barbosa, Souza, Galvíncio and Costa2016), a result of the city's location in a transitional zone between Zona da Mata and Sertão, where climates converge. The area is characterised by semideciduous seasonal vegetation, where there is ecological interplay with natural humid forests; phytogeographically, it is classified as Atlantic Forest and Caatinga (Costa et al., Reference Costa, Lima, França, Lima and Gomes2014).
Ant sampling
Ants were randomly sampled at several locations in Garanhuns from December 2020 to November 2021. We targeted tree trunks, soil, garbage, and lawns in city squares and parks. Sampling took place in the morning, in the afternoon, and at night. Carbohydrate- and protein-based baits were used to attract ants when necessary. The baits were solely used to attract the ants; none of the ants collected had actually consumed the bait. This measure was taken because carbohydrate consumption can increase ant heat tolerance (Freires et al., Reference Freires, Ferreira, Nascimento and Arnan2023). Using an entomological aspirator, 20 workers were collected per colony for 3–4 colonies per species; these colonies were separated by at least 100 m. For a given species, colonies were sampled at the same time of day within a 4-month period to minimise the effects of seasonality. That said, for some species, particularly arboreal species, it was impossible to sample 20 workers. In such cases, we collected at least six individuals, which is the standard minimum sample size when characterising species traits (Gaudard et al., Reference Gaudard, Robertson and Bishop2019). To ensure that the workers belonged to the same colony, we first checked if the ants were following the same trail. After sampling the workers, we immediately placed them in 50-ml Falcon tubes; each tube contained a small cotton ball soaked in water to prevent ant desiccation. Within the tubes, the workers did not display aggressive behaviour towards each other, which reinforces the idea that they were nestmates. The workers were then identified to species or morphospecies using the ant collection at the laboratory of Ecology, Botany, and Ethnobiology, University of Pernambuco (LEBE) –Garanhuns Campus. We also sent samples of all the ants collected to the Ant Biology and Systematics Laboratory (director: Dr Rodrigo M. Feitosa) at the Federal University of Paraná. An ant taxonomist confirmed the identities of the species and morphospecies. Hereafter, for simplicity's sake, we will use the term species to refer to both species and morphospecies.
Maximum thermal tolerance assays
The workers we sampled were immediately taken to the laboratory to measure CTmax. No more than 4 h passed between worker collection and the beginning of the CTmax measurements; the minimum time elapsed was 1.5 h. In the laboratory, workers were first transferred from Falcon tubes to 7-l plastic trays. Next, the ants were placed in 1.5-ml microcentrifuge tubes (1 worker per tube) plugged with cotton, preventing access to any thermal refuges (Oberg et al., Reference Oberg, Del Toro and Pelini2012). Workers that were visibly injured (e.g. limping or moving very slowly) were removed and discarded. Each tube was placed in a randomly chosen locule of a dry bath heater (8 × 6 Thermal-Lok Dry Heat Bath, USA Scientific, Orlando, Florida) that had been preheated to 38°C. Heating block temperature was increased by 1°C every 3 min (Arnan et al., Reference Arnan, Lázaro-González, Beltran, Rodrigo and Pol2022). We used 38°C as the starting temperature because, in preliminary analyses, no species had a CTmax lower than this value. Although the rate of increase could affect absolute CTmax (Roeder et al., Reference Roeder, Roeder and Bujan2021), such is not a concern given our interest in comparing workers experiencing the same protocol. At the end of each 3-min period, we checked levels of ant movement, observing whether each ant could reorient itself after being disturbed by us lightly tapping on the tube. During testing, at least three individuals from each colony were kept in tubes outside of the device as controls to observe whether any of these individuals died from stress. No mortality was observed. The temperature at which an ant lost muscle coordination was defined as its CTmax (Diamond et al., Reference Diamond, Sorger, Hulcr, Pelini, Del Toro, Hirsch, Oberg and Dunn2012).
Body size
We measured mesosoma length (i.e. Weber's length) for all the ants used in the CTmax assays. This metric serves as a proxy of total body size. Measurements took place using a 1-mm paper placed over the stage of a dissecting microscope. The ants were placed in profile on the graph paper, and their legs were stretched out, so that we could clearly see the point where the pronotum met the cervical shield. We measured the distance between that point and the posterior basal angle of the metapleuron.
Nesting microhabitat
We identified the nesting microhabitat used by each of the 47 species we had sampled utilising the literature, field observations, and consultations with experts. Ants were classified as either ground-nesting species or tree-nesting species.
Phylogenetic history
We tested for the presence of a phylogenetic signal in CTmax by employing a time- and genus-calibrated ant phylogeny (Moreau and Bell, Reference Moreau and Bell2013). This phylogeny was then pruned to retain a single species per genus and thus generate a genus-level phylogeny. Four genera represented within our samples (Holcoponera, Mycetomoellerius, Mycocepurus, and Nylanderia) were not present in the original phylogeny and were thus added as sibling species to their closest relatives: Holcoponera with Gnamptogenys (Camacho et al., Reference Camacho, Franco, Branstetter, Pie, Longino, Schultz and Feitosa2022), Mycetomoellerius with Sericomyrmex (Hanisch et al., Reference Hanisch, Sosa-Calvo and Schultz2022), Mycocepurus with Myrmicocrypta (Hanisch et al., Reference Hanisch, Sosa-Calvo and Schultz2022), and Nylanderia with Paratrechina (Ward et al., Reference Ward, Blaimer and Fisher2016). Then, by manually editing the NEWICK tree, the species represented in our samples were inserted into the tree as polytomies at the basal genus level (Supplementary fig. S1). Next, we used Blomberg's K (Blomberg et al., Reference Blomberg, Garland and Ives2003) and Pagel's λ (Pagel, Reference Pagel1999) tests to quantify the phylogenetic signal in CTmax.
Statistical analyses
One of our goals was to compare variation in CTmax at different levels of biological organisation. We thus graphed all the colony-level data using boxplots to identify and exclude potential outliers that could lead us to spurious results. Several colonies showed outliers, which were always on the low side of values (i.e. with very low heat tolerance), suggesting that they might be somehow weakened individuals. Our approach was to eliminate data (35) for workers whose CTmax deviated more than 3°C from the mean CTmax of colonies that had outliers.
Interspecific differences in CTmax were analysed using a general linear mixed model (GLMM), where the response variable was CTmax, the fixed factor was species, and the random factors were colony within species. The individual workers were the replicates (n = 1864). Data were checked for normality and homoscedasticity.
To identify the factors that best explained differences in species-level CTmax, and since we detected a phylogenetic signal in CTmax (see results), we ran a phylogenetic generalised least squares (PGLS) model utilising a variance-covariance matrix structured by the species' inferred phylogenetic relationships. The response variable was CTmax (species-specific mean), and the fixed factors were body size (species-specific mean) and nesting microhabitat. The species sampled were the replicates (n = 47).
To test whether CTmax differed among colonies within species, GLMs were used; colony-specific CTmax was the response variable, and colony was the fixed factor. One model was conducted for each species, for which three or four colonies had been sampled (total of 16 models). Mean body size can vary intraspecifically among colonies and could thus explain colony-level differences in CTmax. Consequently, the models were re-run after adding mean colony-level body size as a covariate.
To test whether there was a relationship between worker size and CTmax within colonies, we performed regression models where linear and quadratic terms for worker body size were the explanatory variables and worker CTmax was the response variable. We ran one model per colony. When the quadratic term was not significant, we reran the model without it.
To determine which level of biological organisation displayed the most variability in CTmax, we calculated two indices. First, we determined the range of CTmax at each level. At the community level, this metric was the difference between maximum and minimum CTmax across individuals. At the population level, it was the difference between maximum and minimum CTmax across individuals within species. At the colony level, it was the difference between maximum and minimum CTmax across individuals within colonies. Second, we calculated the mean and standard deviation of CTmax for each level; the standard deviation was then divided by the mean to obtain the level's coefficient of variation (CV). Using the CV, we determined how much each level varied in °C using mean CTmax for the community, a given species population, and a given colony as the standards of reference (i.e. for each level, we multiplied mean CTmax by CV to obtain variation in °C). For the community level, the standard of reference was community mean CTmax. For the population level, the two standards of reference were community mean CTmax and population mean CTmax (for the 16 species for which we had population-level data). For the colony level, there were three standards of reference: community mean CTmax, population mean CTmax, and colony mean CTmax.
All the analyses were performed using R software (R Development Core Team, 2019). The GLMs were performed using the glm function in the stats package; the GLMMs were performed with the lme function in the nlme package (Pinheiro et al., Reference Pinheiro, Bates, DebRoy and Sarkar2021); and the PGLS were performed with the pgls function in the caper package (Orme et al., Reference Orme, Freckleton, Thomas, Petzoldt, Fritz, Isaac and Pearse2013).
Results
We collected CTmax data for 1864 workers from 99 colonies. Altogether, 47 species belonging to 6 subfamilies and 26 genera were sampled (table S1). The best represented subfamilies were Myrmicinae (23 species) and Formicinae (10 species) followed by Pseudomyrmecinae (6 species), Dolichoderinae (4 species), Ectatomminae (2 species), and Ponerinae (2 species). There were 29 ground-nesting species and 18 tree-nesting species. Species-specific body size varied, ranging from 0.4 to 3.43 mm in mesosome length (Supplementary table S1).
Community-level CTmax
Mean CTmax differed significantly among the 47 ant species (GLMM: χ246 = 295.1, p < 0001). Two ground-nesting, fungus-growing ants displayed the lowest values (mean ± SD): 39.3°C ± 1.3 for Mycocepurus smithii and 40.6°C ± 0.6 for Sericomyrmex mayr (fig. 1). Two arboreal species displayed the highest values: 49.7°C ± 0.5 for Pseudomyrmex termitarius and 48.9°C ± 0.4 for Camponotus blandus (fig. 1; Supplementary table S1). Mean CTmax (± SD) for the community was 44.8 ± 1.7°C.
Boxplots (horizontal bars) showing the species-level medians, means (red dots), and ranges of CTmax for the 47 ant species. The median and mean CTmax of the ant community are indicated by black and red dashed lines, respectively. Species were ordered by ascending mean CTmax.
Ability of nesting microhabitat, phylogeny, and body size to explain species CTmax
We detected a phylogenetic signal in CTmax based on both Blomberg's K (K = 0.002, p = 0.001) and Pagel's λ (λ = 0.77, p = 0.003). More specifically, more closely related species had more similar CTmax values than did less closely related species (Supplementary fig. S1).
Only body size had a significant effect on species CTmax, decreasing with increasing body size (Supplementary fig. S1; table 1).
Results of the phylogenetic generalised least square models exploring the effects of worker body size (mesosoma length) and nesting microhabitat on species-specific CTmax
Population-level CTmax
Nine of the 16 species displayed significant differences in mean CTmax among colonies (fig. 2; Supplementary table S2). When body size was included as a covariate, the results were generally similar, except that Pheidole sp.1 no longer exhibited a significant difference in among-colony CTmax but Tetramorium simillimum did. This result suggests that, within species, differences in CTmax among colonies were poorly explained by colony-specific differences in worker size.
Box plots showing the colony-level medians, means (red dots), and ranges of CTmax for the 16 ant species (n = 3 or 4 colonies sampled) in the analysis. Significance values: *, P < 0.05; **, P < 0.01; and ***, P < 0.001. Each colour represents one colony per species. Abbreviations: A. subterraneus, Acromyrmex subterraneus; A. sexdens, Atta sexdens; B. patagonicus, Brachymyrmex patagonicus; C. arboreus, Camponotus arboreus; C. atriceps, Camponotus atriceps; D. diversus, Dolichoderus diversus; N. fulva, Nylanderia fulva; O. bauri, Odontomachus bauri; P. longicornis, Paratrechina longicornis; T. simillimum, Tetramorium simillimum.
Colony-level CTmax
Nestmate workers typically exhibited different CTmax values (figs 3 and 4), as observed in 92 colonies (n = 99 colonies across 47 species; Supplementary table S3). In one colony, workers even had differences of up to 9°C (Solenopsis sp.5). In a few other colonies, CTmax was the same for all workers (Acromyrmex subterraneus, C. blandus, Crematogaster sp.4, Dolichoderus diversus, Holcoponera pernambucana, and Odontomachus bauri). Interestingly, colonies of monomorphic species (e.g. Nylanderia fulva and Brachymyrmex patagonicus) could vary as much in CTmax as colonies of polymorphic species (e.g. Atta sexdens and A. subterraneus) (figs 3 and 4, Supplementary table S3). For polymorphic species, worker CTmax was positively related to body size in some colonies, notably in three colonies of A. sexdens and one colony of A. subterraneus (fig. 3; Supplementary table S3); for the A. subterraneus colony and one of the A. sexdens' colonies, there was a positive linear relationship. For the other two A. sexdens colonies, the relationship was quadratic: CTmax increased with body size only up to a certain point. In summary, in some polymorphic colonies, workers displayed variable CTmax values that were positively correlated with body size. In other polymorphic colonies, workers displayed variable CTmax values that did not correlate with body size. In yet other polymorphic colonies, there was no variability in CTmax. Finally, in monomorphic colonies, workers either did or did not display variability in CTmax (fig. 4).
Relationship between worker body size (mesosoma length) and CTmax for the four colonies with significant regression results. The number of workers per colony was 20. Each black dot represents a worker. Some workers overlapped in CTmax and/or body size. Abbreviation: A. subterraneus, Acromyrmex subterraneus; A. sexdens, Atta sexdens.
Within-colony variability in worker body size (mesosoma length) and CTmax for five species that displayed different relationships between these two variables: (A) monomorphic colony of Solenopsis sp.5 with highly variable in CTmax; (B) polymorphic colony of Atta sexdens in which body size and CTmax were positively correlated; (C) monomorphic colony of Odontomachus bauri with no variability in CTmax; (D) polymorphic colony of Camponotus blandus with variability in body size but not in CTmax; and (E) polymorphic colony of Camponotus atriceps with variability in both body size and CTmax without the variables being correlated. The number of workers per colony was 20. Each black dot represents one worker. Some workers overlapped in CTmax and/or body size.
Variability in CTmax at different levels of biological organisation
At the community level, the range of CTmax was 12°C (38–50°C). The CV was 3.8%, which means variation was as high as 1.7°C. At the population level, B. patagonicus had the largest range (10°C; 38–48°C), while O. bauri had the smallest range (1°C; 42–43°C). The mean range (±SD) of CTmax for colonies of the same species was 5.3 ± 2.4°C across all the species sampled. Brachymyrmex patagonicus had a CV of 5.7%, which means variation was as high as 2.5°C when using community and population mean CTmax (both 45°C) as standards of reference. Odontomachus bauri had a CV of 0.8%, which means variation was as high as 0.4 or 0.3°C when using community mean CTmax (45°C) or population mean CTmax (43°C) as the standard of reference, respectively. At the colony level, the greatest variability was seen among workers of Solenopsis sp.5, for which the range of CTmax was 9°C (40–49°C). In contrast, no variability was observed among workers of single colonies of A. subterraneus, C. blandus, Crematogaster sp.4, D. diversus, and H. pernambucana or of three colonies of O. bauri. The mean (±SD) range of CTmax among workers of the same colony across all colonies was 2.7 ± 1.8°C. The highest colony-level CV (7.4%) was seen in a colony of Solenopsis sp. 5. Thus, variation was as high as 3.3 or 3.4°C when community mean CTmax (45°C) or population mean CTmax (46°C) was the standard of reference, respectively. It was as high as 3.3°C when colony mean CTmax (45°C) was the standard of reference.
Discussion
First, we examined community-level variability in CTmax. The difference between maximum and minimum species-specific CTmax was 12°C (n = 47 species), which is much lower than what has been found in other studies in tropical regions (Kaspari et al., Reference Kaspari, Clay, Lucas, Yanoviak and Kay2015; Nowrouzi et al., Reference Nowrouzi, Andersen, Bishop and Robson2018). For example, Kaspari et al. (Reference Kaspari, Clay, Lucas, Yanoviak and Kay2015) observed that the range of CTmax was 15°C for a rainforest ant community in Panama (n = 88 species). Australian rainforest ant communities (n = 20 species) had a CTmax range of 28°C (Nowrouzi et al., Reference Nowrouzi, Andersen, Bishop and Robson2018). Similar results have been seen in temperate regions. For instance, for North American ants in different ecosystems (n = 132 species), Bujan et al. (Reference Bujan, Roeder, de Beurs, Weiser and Kaspari2020) observed that the range of CTmax was 18.2°C. However, in a temperate ecosystem (Harvard Forest), Oberg et al. (Reference Oberg, Del Toro and Pelini2012) found that the ant community (n = 16 species) had a much lower value: 8°C. These results show that CTmax range at the community scale can vary greatly across biogeographical regions, within which there exists habitat-related heterogeneity. Understanding these dynamics is particularly important because communities containing species with a broader range of CTmax values may be more resilient in the face of temperature changes (Arnan et al., Reference Arnan, Blüthgen, Molowny-Horas and Retana2015). The interspecific variation in CTmax that we observed also suggests that climate change, especially global warming, will affect species differently (Roeder et al., Reference Roeder, Roeder and Bujan2021; Nascimento et al., Reference Nascimento, Câmara and Arnan2022).
Second, we explored whether three key factors explained CTmax within the ant community. We found that phylogeny and body size, but not nesting microhabitat, explained differences in CTmax among species. Our results underscore that species-specific CTmax is highly conserved in ants. Such has been seen in studies at larger spatial scales (Diamond et al., Reference Diamond, Sorger, Hulcr, Pelini, Del Toro, Hirsch, Oberg and Dunn2012; Diamond and Chick, Reference Diamond and Chick2018), but our study shows that the same is true at smaller spatial scales. In the context of climate change, these results are concerning because they suggest that the ants in our study region may be limited in their capacity to increase CTmax because of phylogenetic constraints (Diamond and Chick, Reference Diamond and Chick2018). With regards to body size, we found a negative relationship between CTmax and species body size: larger ants had lower CTmax values. This negative relationship has been observed in other studies (Verble-Pearson et al., Reference Verble-Pearson, Gifford and Yanoviak2015). Indeed, a recent review found that CTmax and ant body size are inconsistently related on small spatial scales – their association can be positive, negative, or non-existent (Roeder et al., Reference Roeder, Roeder and Bujan2021; Nascimento et al., Reference Nascimento, Câmara and Arnan2022). Our results could have been influenced by the fact that we conducted our study in an urban area, where large ants may be less common as a consequence of anthropogenic pressures (Gibb et al., Reference Gibb, Sanders, Dunn, Watson, Photakis, Abril, Andersen, Angulo, Armbrecht, Arnan, Baccaro, Bishop, Boulay, Castracani, Del Toro, Delsinne, Diaz, Donoso, Enríquez, Fayle, Feener, Fitzpatrick, Gómez, Grasso, Groc, Heterick, Hoffmann, Lach, Lattke, Leponce, Lessard, Longino, Lucky, Majer, Menke, Mezger, Mori, Munyai, Paknia, Pearce-Duvet, Pfeiffer, Philpott, de Souza, Tista, Vasconcelos, Vonshak and Parr2015). In addition, the ecological characteristics of the species studied may also have played a role (Verble-Pearson et al., Reference Verble-Pearson, Gifford and Yanoviak2015). For example, two of the larger species – Camponotus atriceps and Camponotus sp.2 – start foraging at nightfall, and nocturnal ants appear to have lower CTmax values (Esch et al., Reference Esch, Jimenez, Peretz, Uno and O'Donnell2017). To understand how body size may interact with thermal tolerance in ants, more studies are needed, notably those comparing how this relationship manifests itself under different environmental conditions and/or across varying pools of species that differ in biogeographical origin or evolutionary history (Nascimento et al., Reference Nascimento, Câmara and Arnan2022). Finally, contrary to previous studies that observed higher CTmax values in tree-nesting ants (Kaspari et al., Reference Kaspari, Clay, Lucas, Yanoviak and Kay2015; Bujan et al., Reference Bujan, Roeder, de Beurs, Weiser and Kaspari2020), our study found no significant differences between ground-nesting and tree-nesting species. Microhabitat use is a highly conserved trait in Formicidae (Lucky et al., Reference Lucky, Trautwein, Guenard, Weiser and Dunn2013). Given that our analysis controls for phylogenetic relatedness, this effect may have disappeared.
Third, within species, colonies displayed different CTmax values (9 of the 16 species studied); body size had no effect. Although we only tested 3–4 colonies per species, this was enough to find differences in CTmax within species. In previous research, colony-level differences in CTmax were observed for Messor arenarius (n = 3 colonies) and M. ebeninus (n = 5 colonies) (O'Donnell et al., Reference O'Donnell, Bulova, Caponera, Oxman and Giladi2020). In contrast, they were not seen in Pogonomyrmex barbatus (n = 10 colonies) (Roeder et al., Reference Roeder, Paraskevopoulos and Roeder2022). All three species are polymorphic or variable in size. For M. arenarius and M. ebeninus, colony-level differences in CTmax were also unrelated to worker size (O'Donnell et al., Reference O'Donnell, Bulova, Caponera, Oxman and Giladi2020). There are a few hypotheses that could explain differences in CTmax among colonies. First, body size can affect CTmax at the individual level, and, consequently, this relationship may scale up to the colony level (Cerdá and Retana, Reference Cerdá and Retana2000). Second, diet may play a role: access to higher levels of sucrose might increase worker CTmax (Bujan and Kaspari, Reference Bujan and Kaspari2017; Freires et al., Reference Freires, Ferreira, Nascimento and Arnan2023). Third, ants can display phenotypic plasticity in response to the environmental temperatures they experience (Nascimento et al., Reference Nascimento, Câmara and Arnan2022), such that colonies of the same species can exhibit different CTmax values if they live under different microclimatic conditions. The first hypothesis seems to have little support; the second has not yet been tested; and the third is generally supported by past research (Nascimento et al., Reference Nascimento, Câmara and Arnan2022). It is paramount that future work focus on the effects of diet and plasticity on ant CTmax among colonies within species (Roeder et al., Reference Roeder, Roeder and Bujan2021; Nascimento et al., Reference Nascimento, Câmara and Arnan2022). Although past research has indicated that ectotherms seem unable to greatly increase CTmax (Bennett et al., Reference Bennett, Sunday, Calosi, Villalobos, Martínez, Molina-Venegas, Araújo, Algar, Clusella-Trullas, Hawkins, Keith, Kühn, Rahbek, Rodríguez, Singer, Morales-Castilla and Olalla-Tárraga2021; Pottier et al., Reference Pottier, Burke, Zhang, Noble, Schwanz, Drobniak and Nakagawa2022), ants may be an exception. Our study showed that, among colonies within species, differences in mean CTmax could exceed 4°C.
Fourth, we discovered that worker CTmax was not influenced by the degree of colony polymorphism. Within colonies, CTmax could vary tremendously (e.g. up to 9°C) in a way that was not always associated with worker size. Indeed, the variability in CTmax within colonies was nearly equal to the variability within the community. Furthermore, some polymorphic colonies showed limited variability in CTmax, while some monomorphic colonies showed marked variability in CTmax. We observed a pronounced positive relationship between body size and CTmax in the polymorphic species A. sexdens (n = 3 colonies) and A. subterraneus (n = 1 colony). However, no such pattern was seen in the colonies of other polymorphic species (i.e. Mycetomoellerius urichii and Camponotus spp.) or in those of less polymorphic species (i.e. Pheidole spp.). Previous research analysing the effect of body size on within-colony CTmax has yielded two contrasting results: (1) larger workers had higher CTmax than smaller workers (Cerdá and Retana, Reference Cerdá and Retana2000; Baudier et al., Reference Baudier, Mudd, Erickson and O'Donnell2015) and (2) larger and smaller workers had similar CTmax (Lytle et al., Reference Lytle, Costa and Warren2020; Yela et al., Reference Yela, Calcaterra and Aranda-Rickert2020). A positive relationship between worker size and CTmax has mainly been found in highly polymorphic species, such as those in the genus Atta (Ribeiro et al., Reference Ribeiro, Camacho and Navas2012; Baudier and O'Donnell, Reference Baudier and O'Donnell2020), those in the genus Cataglyphis (Cerdá and Retana, Reference Cerdá and Retana2000), or those in army ant taxa (Baudier et al., Reference Baudier, Mudd, Erickson and O'Donnell2015, Reference Baudier, D'Amelio, Malhotra, O'Connor and O'Donnell2018). The absence of a relationship between worker size and CTmax has been seen in species with a low degree of polymorphism, such as Ectatomma ruidum (Esch et al., Reference Esch, Jimenez, Peretz, Uno and O'Donnell2017), Solenopsis invicta (Lytle et al., Reference Lytle, Costa and Warren2020), and Temnothorax curvispinosus (Yilmaz et al., Reference Yilmaz, Chick, Perez, Strickler, Vaughn, Martin and Diamond2019). In contrast to the above, there is little research looking at variation in CTmax among workers in monomorphic ants, which could possibly be explained by stress, age, or nutritional status (Nyamukondiwa and Terblanche, Reference Nyamukondiwa and Terblanche2009). Regardless of the degree of morphological variation among workers, the presence of workers with different CTmax values allows colonies to explore microhabitats with different temperatures.
Fifth, our results showed that the greatest variability in CTmax existed at the community level (based on the CTmax ranges) and colony level (based on the CV for each level). Only one previous study had examined variability in ant CTmax at all three levels of biological organisation, and it found that variability was greatest at the community level (Verble-Pearson et al., Reference Verble-Pearson, Gifford and Yanoviak2015). Such is not uncommon. For instance, the variation in CTmax seen in an ant community in Panama (Kaspari et al., Reference Kaspari, Clay, Lucas, Yanoviak and Kay2015) accounted for 74% of the variance in a global data set that included 269 ant populations found from 0 to 66° North in latitude (Diamond et al., Reference Diamond, Sorger, Hulcr, Pelini, Del Toro, Hirsch, Oberg and Dunn2012). Here, we also found dramatic variation at the community level (albeit less than in Kaspari et al., Reference Kaspari, Clay, Lucas, Yanoviak and Kay2015). However, it is worth mentioning that a single colony or population can display almost as much variation as an entire community (40–49 and 38–48°C compared to 38–50°C, respectively). It is unlikely that our colony- and population-level results were influenced by the number of workers sampled, given that sample sizes were largest for the species with the highest and lowest degree of variability. However, it could be that our study area – an urban environment – mainly contains thermal generalist species and that colony-level values thus more closely reflect community values (Franzén et al., Reference Franzén, Betzholtz, Pettersson and Forsman2020). Furthermore, genetic variation among populations can also explain variation in CTmax within and among colonies (reviewed in Perez and Aron, Reference Perez and Aron2020). Given the limited ability of morphology to explain the striking variability that we observed, future work must elucidate the underlying factors at play and identify the species that have a greater capacity to increase CTmax in response to climate change.
Conclusion
Neotropical species in northeastern Brazil display differences in CTmax of up to ~12°C. Species-specific CTmax is best explained by phylogeny, which could clearly constrain the ability of ants to deal with climate change. However, ant colonies can often cope with challenging temperatures via acclimatisation, thermal plasticity and/or improved nutrition (Nascimento et al., Reference Nascimento, Câmara and Arnan2022; Freires et al., Reference Freires, Ferreira, Nascimento and Arnan2023), which should allow them to navigate global temperature increases. Body size also explained differences in CTmax among species but not among colonies within species. That said, a positive correlation did exist in some colonies of highly polymorphic species; it may play an important role in colony division of labour and in the context of climate change. The presence of differences in colony-level CTmax within species, whether monomorphic or polymorphic, demonstrates that ants possess the ability to adapt to different thermal conditions, although this issue has rarely been explored. We wish to emphasise that we observed at least as much variability in CTmax within colonies as within the community, a fact that could help shape the adaptive potential of ant populations as they face changes in environmental temperature. It is urgent to conduct new research that clarifies the factors affecting variability in CTmax among and within colonies.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/S0007485324000567.
Acknowledgements
We thank J Pearce-Duvet for editing the manuscript's English. This study received financial support from the University of Pernambuco (191_APQ 2020, Process 427). XA extends his appreciation to the Brazilian National Council for Scientific and Technological Development (CNPq) for his productivity grant (PQ-2, Process 307385/2020-5). T. C. and G. N. are grateful to FACEPE for its financial support (a postdoctoral grant [BFP – 0157-2.05/20] and a research grant [IBPG-0649-2.05/20], respectively).
Competing interests
None.
