Environmental conditions shape the global distribution of ant societies

Abstract

Sociality has evolved several times and is a key strategy for overcoming environmental challenges and promoting ecological success. Yet, it remains unclear how environmental conditions shape global variation in social traits of animals. With their diverse societies and global distribution, ants are ideal to test whether environmental conditions influence the distribution of animal social traits worldwide. Here, we used trait data for a total of 3,299 ant species to explore how three key social traits (reproductive structure, colony size, and worker polymorphism) vary with environmental conditions globally. We show that trait compositions are strongly structured by biomes, indicating that habitat types, as well as environmental factors like temperature and seasonality, influence sociality. Our findings highlight the crucial role of the environment in shaping the global distribution of sociality in ants, contributing to a better understanding of how complex animal societies evolved.

Sociality is thought to be key to the ecological success of ants, which dominate most terrestrial ecosystems (1). Although all ant species share similar fundamental social traits, such as a common breeding site, brood care, and division of reproductive labor (2–5), they exhibit diverse social structures. Within the over 14,000 described species of ants, tremendous variation is observed in characteristics such as colony size (from a dozen to several million individuals), colony structure (single nests to supercolonies), queen number, or worker polymorphism (morphologically differentiated worker subcastes) (1). Advantages of different forms of societies may depend on environmental conditions (4), such as temperature, humidity, elevation, and seasonality (4). However, previous work on this subject has focused on single species, single traits, or restricted geographic scales (4). Yet ant societies are not defined by single traits, but rather by suites of multiple coevolving traits (2, 3, 5) that may act together as adaptations to ecological pressures. Considering social trait compositions rather than individual traits may reveal “social strategies” adapted to specific ecological contexts.

To test if similar environmental contexts host similar ant societies, we considered not only individual environmental factors, but also biomes. Biomes are broad-scale ecological units, repeated across continents and defined by specific climate, soil, and vegetation types (6). There are 14 major terrestrial biomes worldwide, including tropical moist forests, deserts, temperate grasslands, and tundra (7). Biomes capture complex environmental combinations of selective pressures (6). Just as focusing on individual traits that have coevolved with others may not accurately reflect the organization of complex societies, relying on single environmental predictors may not adequately capture the broader ecological contexts that shape them. Thus, if ant societies are shaped by the environment, we would expect similar compositions of social traits to recur across similar biomes, even across distant geographic regions. Gaining insights into the types of societies that help ants thrive under specific conditions is important to better understand the role of sociality in determining species’ geographic range and responses to global change.

Insufficient data have long constrained global studies on the drivers of social traits. Here, we compiled and analyzed a unique global trait database to test links between global variation in ant societies and the environment for a total of 3,299 species. To achieve this, we focused on three social traits: colony size (number of workers in a single nest, for 1,682 species), worker polymorphism (presence of morphological polymorphism in the worker caste, for 2,087 species), and reproductive structure (monogynous colony with one single queen or polygynous with several queens, for 1,523 species). These traits cover different facets of sociality: group size, division of labor, and cooperative breeding. Together, they reflect how ant societies are organized.

Results and Discussion

To test if certain combinations of social traits (hereafter referred to as “trait compositions”) tend to occur in similar environments, we quantified the prevalence of the different modalities of three social traits across ant assemblages of 591 ecoregions (i.e., geographic subdivisions of biomes (7); Fig. 1). For each ecoregion, we quantified the “prevalence” of species exhibiting large colony size (>1,000 workers, the median value), worker polymorphism, and polygyny, corresponding to the proportion of species exhibiting these traits within an assemblage. We tested whether these trait compositions were linked to temperature, humidity, seasonality, topology, soil structure, resources, or ant diversity (Materials and Methods). Our modeling approach (redundancy analysis; RDA; Fig. 1) describes associations between multivariate trait data and environmental factors while accounting for spatial and phylogenetic autocorrelation. We found that temperature (partial R² = 0.11, permutation test: P = 0.015) and seasonality (partial R² = 0.07, permutation test: P = 0.005) affected social trait compositions of species assemblages, consistent with previous studies on colony size (4, 8) and worker polymorphism (9).

Fig. 1.

Associations of social trait compositions with (A) ant assemblages and (B) environmental factors. (A) Hierarchical clustering (Ward D2) of three groups of ant ecoregional assemblages based on their social trait prevalences (brackets on the Right). The first column shows colors of assemblages corresponding to their social trait compositions. Similar colors indicate similar trait compositions. The three following columns represent the associated prevalence values for social traits: large colony size, worker polymorphism, polygynous reproductive structure. (B) Associations of social trait compositions and environmental factors based on RDA. Significant factors (P < 0.05) are represented with red arrows. For readability, eigenvectors controlling for spatial and phylogenetic autocorrelation are not represented; axes are transformed with a sigmoid function.

Using a hierarchical clustering approach and estimating the ideal cluster number, we identified three groups of assemblages with distinct and recurring social trait compositions (Figs. 1 and 2). Group 1 corresponds to assemblages with prevalent worker polymorphism, small colony sizes, and monogyny (Wilcoxon tests: P < 0.0001). Group 2 exhibits prevalent worker polymorphism, polygyny (Wilcoxon tests; P < 0.0001), and large colony sizes (Wilcoxon test: P = 0.017). Finally, group 3 is characterized by prevalent polygyny and worker monomorphism (Wilcoxon tests: P < 0.0001). We then quantified how often assemblages with a given social composition occurred within the same biome (Materials and Methods), revealing that biomes predict the spatial distribution of social trait composition (Accuracy = 0.66; Binomial test: P < 0.0001; Fig. 2). This highlights that similar social trait compositions are repeatedly found across regions with comparable conditions on different continents. Group 1 had the largest number of assemblages (Fig. 1) and was particularly present in tropical regions, including the Neotropics and Afrotropics, Malagasy, Indomalayan, and Australian tropical forests (Fig. 2). It was associated with high temperatures and low seasonal variation (Fig. 1), typical of tropical forests and grasslands. High abundances of ants in these regions (10) could increase competition for nesting spaces, favoring species with small colonies using limited nesting spaces (11). High levels of competition and stable tropical climates with limited annual fluctuations could both favor worker polymorphism by promoting task specialization among subcastes and enabling the exploitation of a broader range of resources (12, 13). The prevalence of monogynous colonies also increases (Fig. 2), consistent with previous work hypothesizing polygyny to be advantageous under fluctuant conditions with high dispersal risks rather than stable ones (4, 14). Group 2 was predominantly found in deserts (Fig. 2), including the Sahara/Sahel, Western North American, Arabian, Central Asian, Australian, and Sechura deserts. There, high seasonality (Fig. 1) and scarce food resources may limit species richness (15) and lead to intense competition. To effectively compete and exploit limited and sporadic resource availability, worker polymorphism may be advantageous (9). Additionally, polygyny could reduce dispersal costs in extreme arid climatic conditions by enabling the foundation of new colonies near the parental nest via budding, decreasing mortality during postmating dispersal (4, 14). Under these conditions, large colony sizes could help limit the risks of predation and desiccation for foragers by limiting the amount of time they spent above ground (4).

Fig. 2.

Biomes predict distributions of social trait compositions. First column: three groups of assemblages with distinct trait compositions (detailed below each map). The color of each assemblage indicates relative distance in terms of social trait compositions (Fig. 1). Second column: biome groups associated with each respective social trait composition. Third column: assemblages that are present in the first two columns simultaneously. n_ecoregions: number of assemblages included in social trait compositions, biome groups, and in the two first columns simultaneously. n_species: number of species present in all assemblages of a social trait composition. S_1% is a support metric for each biome association with trait composition. Balanced Accuracy and F1 are metrics quantifying overall congruence between the first two columns’ classifications (Materials and Methods).

Finally, group 3 was dominant across temperate, Mediterranean, and boreal climates (Fig. 2). It had a mainly Holarctic distribution, ranging from the Iberian Peninsula to the Japanese archipelago, and temperate Southern Hemisphere regions, including New Zealand and Southern Chile. This group was associated with lower temperatures and medium to high seasonality (Fig. 1). Strong temperature fluctuations could limit worker polymorphism because the developmental and energetic costs of producing distinct castes are expected to be higher, favoring a more uniform worker caste (13). Additionally, the lower ant diversity compared to tropical regions (15) also implies less competition and fewer selective advantages of worker polymorphism for resource exploitation. Finally, cooler conditions may increase the risk of dispersal and single-queen foundations (14), favoring polygynous colonies.

Our findings reveal that the global distribution of ant social structures aligns with environmental conditions rather than occurring at random. This suggests that environmental filtering shapes the organization of ant societies, with similar social structures having evolved repeatedly across lineages. Considering their 140 My history spanning drifting continents and shifting climates, these global parallels are best explained by evolutionary convergence, revealing a broad ecological imprint on the evolution of ant societies.

Materials and Methods

We compiled a database of three key social traits (colony size, worker polymorphism, reproductive structure) for a total of 3,299 ant species. Species’ occurrences were aggregated into ecoregion-level assemblages using a multicriterion filtering procedure to correct for sampling bias. For each assemblage, we estimated the prevalence of social trait modalities and characterized multivariate social trait compositions, representing different types of societies. To assess how environmental factors structure global trait compositions, we used seven variables derived from high-resolution environmental datasets and applied a redundancy analysis while correcting for spatial and phylogenetic autocorrelation. Finally, we clustered assemblages by trait composition and identified biomes consistently associated with each social trait composition. Detailed material and methods are provided in the SI Appendix.

Supplementary Material

Appendix 01 (PDF)

Dataset S01 (XLSX)

Data, Materials, and Software Availability

Scripts, files, figures, and data have been deposited in GitHub (https://github.com/EddiePerochon/Ant_Sociality_Env) (16). Trait data are available in the supporting information (Dataset S01).