Critical thermal maxima in neotropical ants at colony, population, and community levels

Abstract

Global warming is exposing many organisms to severe thermal conditions and is having impacts at multiple levels of biological organisation, from individuals to species and beyond. Biotic and abiotic factors can influence organismal thermal tolerance, shaping responses to climate change. In eusocial ants, thermal tolerance can be measured at the colony level (among workers within colonies), the population level (among colonies within species), and the community level (among species). We analysed critical thermal maxima (CT_max) across these three levels for ants in a semiarid region of northeastern Brazil. We examined the individual and combined effects of phylogeny, body size (BS), and nesting microhabitat on community-level CT_max and the individual effects of BS on population- and colony-level CT_max. We sampled 1864 workers from 99 ant colonies across 47 species, for which we characterised CT_max, nesting microhabitat, BS, and phylogenetic history. Among species, CT_max ranged from 39.3 to 49.7°C, and community-level differences were best explained by phylogeny and BS. For more than half of the species, CT_max differed significantly among colonies in a way that was not explained by BS. Notably, there was almost as much variability in CT_max within colonies as within the entire community. Monomorphic and polymorphic species exhibited similar levels of CT_max variability within colonies, a pattern not always explained by BS. This vital intra- and inter-colony variability in thermal tolerance is likely allows tropical ant species to better cope with climate change. Our results underscore why ecological research must examine multiple levels of biological organisation.

Introduction

Temperature governs many biological processes, affecting different levels of biological organisation (Verberk et al., 2016). Changes in temperature can have profound effects on the functioning of cells, tissues, and organ systems (Brown et al., 2004). At the organismal level, changes in environmental temperature can directly affect reproduction, growth, and survival (Deutsch et al., 2008). Consequently, temperature can determine the abundance and dynamics of species populations within communities and, ultimately, species distribution ranges (Bujan et al., 2020). 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, 2022).

Ectothermic organisms are highly susceptible to temperature changes because their body temperatures completely depend on environmental thermal conditions (Béltran et al., 2021). In other words, temperature mediates all the physiological reactions of ectotherms, impacting their functional ecology (Angilletta, 2009). 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., 2016). An organism's critical thermal limits are defined by the minimum temperature (CT_min) and maximum temperature (CT_max) it can withstand without losing motor coordination (Lutterschmidt and Hutchison, 1997). 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 CT_max because it can indicate the vulnerability of organisms to present and future temperature increases (Diamond et al., 2012). However, CT_max can vary greatly among ectotherms and levels of biological organisation (i.e. individuals, populations, and communities; Verble-Pearson et al., 2015; Nascimento et al., 2022). Furthermore, this variation may be shaped by many biotic and abiotic factors (Sunday et al., 2014; Leiva et al., 2019).

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., 2012; Elizalde et al., 2020). Because they are social insects, ants are an excellent model for assessing differences in CT_max at different levels of biological organisation and for examining the factors behind these differences. Indeed, in ants, CT_max 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, 2020; Bujan et al., 2020; O'Donnell et al., 2020). That said, most research to date has focused on ant CT_max mainly at the community level; while some colony-level studies exist as well, population-level studies are scarce (Nascimento et al., 2022). Also rare is work that simultaneously analyses differences in CT_max at all three levels (but see Verble-Pearson et al., 2015). 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., 2021). It is necessary to consider multiple organisational levels in tandem because ant colonies may have subclasses of workers that differ in body size and CT_max (Baudier and O'Donnell, 2020). Ant body size may also vary among colonies as a result of environmental factors (Shik et al., 2019; Oliveira et al., 2022), which can lead to variation in CT_max within species (Baudier and O'Donnell, 2020). 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 CT_max 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., 2019) or brood development rates (Penick et al., 2017). While CT_max 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., 2022). If selection pressure is greater on more thermally vulnerable workers, it will take a toll on colony fitness and persistence (Baudier and O'Donnell, 2017). 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 CT_max can differ due to many biotic and abiotic factors (Roeder et al., 2021; Nascimento et al., 2022). Studies have looked at how ant CT_max relates to microhabitat use (e.g. Baudier et al., 2015, 2018; Kaspari et al., 2015), evolutionary history (e.g. Diamond et al., 2012; Arnan and Blüthgen, 2015), and body size (e.g. Kaspari et al., 2015; Verble-Pearson et al., 2015). 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 CT_max. For example, microhabitat appears to have a strong effect on ant CT_max (Baudier et al., 2015, 2018; Kaspari et al., 2015; Bujan et al., 2020) that could be underlain by species evolutionary history because microhabitat use is a highly conserved trait in ants (Lucky et al., 2013). Thus, it is only possible to determine each factor's contribution if both are analysed in tandem.

Simultaneously exploring how different factors drive CT_max 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., 2022). Such information is especially critical for tropical regions, where ants already live close to their CT_max values and studies remain infrequent (Diamond and Chick, 2018). 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., 2014), which will likely expose the region's ant fauna to severe heat stress.

Given this context, our study aims to characterise CT_max 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 CT_max within an ant community? (2) Are differences in community-level CT_max explained by species nesting microhabitat, body size, and/or evolutionary history? (3) Do colonies of the same species exhibit different CT_max values? (4) Do workers from monomorphic vs. polymorphic colonies exhibit different CT_max values? (5) Which level of organisation displays the greatest degree of variability in CT_max?

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., 2016), 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., 2014).

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., 2023). 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., 2019). 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 CT_max. No more than 4 h passed between worker collection and the beginning of the CT_max 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., 2012). 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., 2022). We used 38°C as the starting temperature because, in preliminary analyses, no species had a CT_max lower than this value. Although the rate of increase could affect absolute CT_max (Roeder et al., 2021), 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 CT_max (Diamond et al., 2012).

Body size

We measured mesosoma length (i.e. Weber's length) for all the ants used in the CT_max 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 CT_max by employing a time- and genus-calibrated ant phylogeny (Moreau and Bell, 2013). 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., 2022), Mycetomoellerius with Sericomyrmex (Hanisch et al., 2022), Mycocepurus with Myrmicocrypta (Hanisch et al., 2022), and Nylanderia with Paratrechina (Ward et al., 2016). 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., 2003) and Pagel's λ (Pagel, 1999) tests to quantify the phylogenetic signal in CT_max.

Statistical analyses

One of our goals was to compare variation in CT_max 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 CT_max deviated more than 3°C from the mean CT_max of colonies that had outliers.

Interspecific differences in CT_max were analysed using a general linear mixed model (GLMM), where the response variable was CT_max, 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 CT_max, and since we detected a phylogenetic signal in CT_max (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 CT_max (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 CT_max differed among colonies within species, GLMs were used; colony-specific CT_max 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 CT_max. 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 CT_max within colonies, we performed regression models where linear and quadratic terms for worker body size were the explanatory variables and worker CT_max 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 CT_max, we calculated two indices. First, we determined the range of CT_max at each level. At the community level, this metric was the difference between maximum and minimum CT_max across individuals. At the population level, it was the difference between maximum and minimum CT_max across individuals within species. At the colony level, it was the difference between maximum and minimum CT_max across individuals within colonies. Second, we calculated the mean and standard deviation of CT_max 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 CT_max for the community, a given species population, and a given colony as the standards of reference (i.e. for each level, we multiplied mean CT_max by CV to obtain variation in °C). For the community level, the standard of reference was community mean CT_max. For the population level, the two standards of reference were community mean CT_max and population mean CT_max (for the 16 species for which we had population-level data). For the colony level, there were three standards of reference: community mean CT_max, population mean CT_max, and colony mean CT_max.

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., 2021); and the PGLS were performed with the pgls function in the caper package (Orme et al., 2013).

Results

We collected CT_max 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 CT_max

Mean CT_max 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 CT_max (± SD) for the community was 44.8 ± 1.7°C.

Figure 1. Boxplots (horizontal bars) showing the species-level medians, means (red dots), and ranges of CT_max for the 47 ant species. The median and mean CT_max of the ant community are indicated by black and red dashed lines, respectively. Species were ordered by ascending mean CT_max.

Ability of nesting microhabitat, phylogeny, and body size to explain species CT_max

We detected a phylogenetic signal in CT_max 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 CT_max values than did less closely related species (Supplementary fig. S1).

Only body size had a significant effect on species CT_max, decreasing with increasing body size (Supplementary fig. S1; table 1).

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 CT_max

Population-level CT_max

Nine of the 16 species displayed significant differences in mean CT_max 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 CT_max but Tetramorium simillimum did. This result suggests that, within species, differences in CT_max among colonies were poorly explained by colony-specific differences in worker size.

Figure 2. Box plots showing the colony-level medians, means (red dots), and ranges of CT_max 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 CT_max

Nestmate workers typically exhibited different CT_max 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, CT_max 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 CT_max as colonies of polymorphic species (e.g. Atta sexdens and A. subterraneus) (figs 3 and 4, Supplementary table S3). For polymorphic species, worker CT_max 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: CT_max increased with body size only up to a certain point. In summary, in some polymorphic colonies, workers displayed variable CT_max values that were positively correlated with body size. In other polymorphic colonies, workers displayed variable CT_max values that did not correlate with body size. In yet other polymorphic colonies, there was no variability in CT_max. Finally, in monomorphic colonies, workers either did or did not display variability in CT_max (fig. 4).

Figure 3. Relationship between worker body size (mesosoma length) and CT_max 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 CT_max and/or body size. Abbreviation: A. subterraneus, Acromyrmex subterraneus; A. sexdens, Atta sexdens.

Figure 4. Within-colony variability in worker body size (mesosoma length) and CT_max for five species that displayed different relationships between these two variables: (A) monomorphic colony of Solenopsis sp.5 with highly variable in CT_max; (B) polymorphic colony of Atta sexdens in which body size and CT_max were positively correlated; (C) monomorphic colony of Odontomachus bauri with no variability in CT_max; (D) polymorphic colony of Camponotus blandus with variability in body size but not in CT_max; and (E) polymorphic colony of Camponotus atriceps with variability in both body size and CT_max without the variables being correlated. The number of workers per colony was 20. Each black dot represents one worker. Some workers overlapped in CT_max and/or body size.

Variability in CT_max at different levels of biological organisation

At the community level, the range of CT_max 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 CT_max 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 CT_max (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 CT_max (45°C) or population mean CT_max (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 CT_max 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 CT_max 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 CT_max (45°C) or population mean CT_max (46°C) was the standard of reference, respectively. It was as high as 3.3°C when colony mean CT_max (45°C) was the standard of reference.

Discussion

First, we examined community-level variability in CT_max. The difference between maximum and minimum species-specific CT_max was 12°C (n = 47 species), which is much lower than what has been found in other studies in tropical regions (Kaspari et al., 2015; Nowrouzi et al., 2018). For example, Kaspari et al. (2015) observed that the range of CT_max was 15°C for a rainforest ant community in Panama (n = 88 species). Australian rainforest ant communities (n = 20 species) had a CT_max range of 28°C (Nowrouzi et al., 2018). Similar results have been seen in temperate regions. For instance, for North American ants in different ecosystems (n = 132 species), Bujan et al. (2020) observed that the range of CT_max was 18.2°C. However, in a temperate ecosystem (Harvard Forest), Oberg et al. (2012) found that the ant community (n = 16 species) had a much lower value: 8°C. These results show that CT_max 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 CT_max values may be more resilient in the face of temperature changes (Arnan et al., 2015). The interspecific variation in CT_max that we observed also suggests that climate change, especially global warming, will affect species differently (Roeder et al., 2021; Nascimento et al., 2022).

Second, we explored whether three key factors explained CT_max within the ant community. We found that phylogeny and body size, but not nesting microhabitat, explained differences in CT_max among species. Our results underscore that species-specific CT_max is highly conserved in ants. Such has been seen in studies at larger spatial scales (Diamond et al., 2012; Diamond and Chick, 2018), 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 CT_max because of phylogenetic constraints (Diamond and Chick, 2018). With regards to body size, we found a negative relationship between CT_max and species body size: larger ants had lower CT_max values. This negative relationship has been observed in other studies (Verble-Pearson et al., 2015). Indeed, a recent review found that CT_max and ant body size are inconsistently related on small spatial scales – their association can be positive, negative, or non-existent (Roeder et al., 2021; Nascimento et al., 2022). 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., 2015). In addition, the ecological characteristics of the species studied may also have played a role (Verble-Pearson et al., 2015). For example, two of the larger species – Camponotus atriceps and Camponotus sp.2 – start foraging at nightfall, and nocturnal ants appear to have lower CT_max values (Esch et al., 2017). 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., 2022). Finally, contrary to previous studies that observed higher CT_max values in tree-nesting ants (Kaspari et al., 2015; Bujan et al., 2020), 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., 2013). Given that our analysis controls for phylogenetic relatedness, this effect may have disappeared.

Third, within species, colonies displayed different CT_max 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 CT_max within species. In previous research, colony-level differences in CT_max were observed for Messor arenarius (n = 3 colonies) and M. ebeninus (n = 5 colonies) (O'Donnell et al., 2020). In contrast, they were not seen in Pogonomyrmex barbatus (n = 10 colonies) (Roeder et al., 2022). All three species are polymorphic or variable in size. For M. arenarius and M. ebeninus, colony-level differences in CT_max were also unrelated to worker size (O'Donnell et al., 2020). There are a few hypotheses that could explain differences in CT_max among colonies. First, body size can affect CT_max at the individual level, and, consequently, this relationship may scale up to the colony level (Cerdá and Retana, 2000). Second, diet may play a role: access to higher levels of sucrose might increase worker CT_max (Bujan and Kaspari, 2017; Freires et al., 2023). Third, ants can display phenotypic plasticity in response to the environmental temperatures they experience (Nascimento et al., 2022), such that colonies of the same species can exhibit different CT_max 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., 2022). It is paramount that future work focus on the effects of diet and plasticity on ant CT_max among colonies within species (Roeder et al., 2021; Nascimento et al., 2022). Although past research has indicated that ectotherms seem unable to greatly increase CT_max (Bennett et al., 2021; Pottier et al., 2022), ants may be an exception. Our study showed that, among colonies within species, differences in mean CT_max could exceed 4°C.

Fourth, we discovered that worker CT_max was not influenced by the degree of colony polymorphism. Within colonies, CT_max could vary tremendously (e.g. up to 9°C) in a way that was not always associated with worker size. Indeed, the variability in CT_max within colonies was nearly equal to the variability within the community. Furthermore, some polymorphic colonies showed limited variability in CT_max, while some monomorphic colonies showed marked variability in CT_max. We observed a pronounced positive relationship between body size and CT_max 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 CT_max has yielded two contrasting results: (1) larger workers had higher CT_max than smaller workers (Cerdá and Retana, 2000; Baudier et al., 2015) and (2) larger and smaller workers had similar CT_max (Lytle et al., 2020; Yela et al., 2020). A positive relationship between worker size and CT_max has mainly been found in highly polymorphic species, such as those in the genus Atta (Ribeiro et al., 2012; Baudier and O'Donnell, 2020), those in the genus Cataglyphis (Cerdá and Retana, 2000), or those in army ant taxa (Baudier et al., 2015, 2018). The absence of a relationship between worker size and CT_max has been seen in species with a low degree of polymorphism, such as Ectatomma ruidum (Esch et al., 2017), Solenopsis invicta (Lytle et al., 2020), and Temnothorax curvispinosus (Yilmaz et al., 2019). In contrast to the above, there is little research looking at variation in CT_max among workers in monomorphic ants, which could possibly be explained by stress, age, or nutritional status (Nyamukondiwa and Terblanche, 2009). Regardless of the degree of morphological variation among workers, the presence of workers with different CT_max values allows colonies to explore microhabitats with different temperatures.

Fifth, our results showed that the greatest variability in CT_max existed at the community level (based on the CT_max ranges) and colony level (based on the CV for each level). Only one previous study had examined variability in ant CT_max at all three levels of biological organisation, and it found that variability was greatest at the community level (Verble-Pearson et al., 2015). Such is not uncommon. For instance, the variation in CT_max seen in an ant community in Panama (Kaspari et al., 2015) 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., 2012). Here, we also found dramatic variation at the community level (albeit less than in Kaspari et al., 2015). 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., 2020). Furthermore, genetic variation among populations can also explain variation in CT_max within and among colonies (reviewed in Perez and Aron, 2020). 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 CT_max in response to climate change.

Conclusion

Neotropical species in northeastern Brazil display differences in CT_max of up to ~12°C. Species-specific CT_max 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., 2022; Freires et al., 2023), which should allow them to navigate global temperature increases. Body size also explained differences in CT_max 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 CT_max 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 CT_max 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 CT_max among and within colonies.