Cooperation by ant queens during colony-founding perpetuates alternative forms of social organization

Abstract Key social traits, like queen number in eusocial insect colonies, have long been considered plastic, but the recent finding that colony social organization is under strict genetic control in multiple ant lineages challenges this view. This begs the question of which hardwired behavioral mechanism(s) generate alternative forms of social organization during colony development. We addressed this question in the Alpine silver ant, Formica selysi, a species with two social forms determined by a supergene. Queens that carry exclusively the M haplotype are produced by and live in monogyne (= single-queen) colonies, whereas queens that carry at least one copy of the P haplotype are produced by and live in polygyne (= multiple-queen) colonies. With extensive field samplings and laboratory experiments, we show that both types of queens successfully establish colonies independently, without being accompanied by workers, but that they do so in contrasting ways. Monogyne queens were generally intolerant of other queens and founded colonies solitarily, whereas polygyne queens were mutually attracted to each other and mainly founded colonies cooperatively. These associations persisted for months after worker emergence, suggesting that cooperative colony-founding leads to permanent multiple queening. Overall, our study shows that queens of each social form found colonies independently in the field but that P-carrying queens are more likely to cooperate, thereby contributing to perpetuate alternative forms of social organization. Significance statement Understanding the genetic and behavioral underpinnings of social organization is a major goal in evolutionary biology. Recent studies have shown that colony social organization is controlled by supergenes in multiple ant lineages. But the behavioral processes linking the genotype of a queen to the type of colony she will form remain largely unknown. Here, we show that in Alpine silver ants, alternative supergene genotypes are associated with different levels of social attraction and tolerance in young queens. These hardwired differences in social traits make queens carrying the P supergene haplotype more prone to cooperate and form durable associations during independent colony-founding. These findings help explain how genetic variants induce alternative forms of social organization during the ontogeny of a colony. They also illustrate how simple phenotypic differences at the individual level can result in large differences at higher levels of organization. Supplementary Information The online version contains supplementary material available at 10.1007/s00265-021-03105-1.


Supplementary Information Figures
Size distribution of workers in incipient (green) and mature (orange) field colonies

Fig. S2
Brood size of incipient single-queen and multiple-queen field colonies. Box plots represent the median, the 1 st and 3 rd quartiles, and the non-outlier range. Each dot represents a colony

Fig. S3
Mortality of monogyne (M, blue) and polygyne (P, red) queens after one week in twoqueen (left) and three-queen (right) assays, according to the social origin of the other queen(s) in the assay (non-focal queens). Each non-focal queen is represented by one letter. Number of focal queens is displayed inside circles

Fig. S4
Failure of monogyne (M, blue) and polygyne (P, red) queens at starting a colony in two-queen (left) and three-queen (right) assays, according to the social origin of the other queen(s) in the assay (non-focal queens). Each non-focal queen is represented by one letter. Queens failed at starting a nest when they remained on sand surface throughout the experiment (or until their death). Number of focal queens is displayed inside circles

Fig. S5
Colony founding behavior exhibited by monogyne (M) and polygyne (P) queen in (a) two-queen and (b) three-queen assays. The social origin of the queens in the assays is indicated below the bars, with each letter representing one queen. Colony founding was characterized by observations of queens inside excavated sand cavities. Bars indicate the proportions of assays in which none of the queens founded a nest (black bars), at least one queen founded solitarily (grey bars) or queens founded cooperatively in association of two (white bar) or three queens (white bars with black dots). In 3 assays (2 P-P-M assays and 1 M-M-P assay), both solitary (1 queen) and cooperative (2 queens) colony-founding occurred. The outcome of these assays was categorized as 2-queen association. n refers to the number of assays

Fig. S6
Propensity of monogyne (M, blue) and polygyne (P, red) queens to gather in assays with two (left) and three (right) queens, according to the social origin of the other queen(s) in the assay (non-focal queens). Each non-focal queen is represented by one letter. Queens gathered when they were spatially close to another queen during the first observation. Number of focal queens is displayed inside circles  Tables   Table S1. Model 1: GLMM fitted to test whether the body size of workers differed between incipient and mature colonies. The colony of origin of the workers was included as a random factor.  Table S4. Model 9: GLMM fitted to test the mortality of queens after one week. The colony of origin of the queens and the assay id nested within the year were included as random factors. The model includes an interaction term for the two first factors. Estimates for main factors were calculated after removing the non-significant interaction term. Post hoc tests were corrected for multiple comparisons using the Benjamini-Hochberg procedure (FDR method).     Table S8. Model 7: GLMM fitted to test the probability that queens started their colonies solitarily or cooperatively. The colony of origin of the queens and the assay id nested within the year were included as random factors. Because the initial model encountered quasicomplete separation (see statistic section in the main text), the model was fit with a weak prior on the fixed effects. The model includes an interaction term for the two first factors. Estimates for main factors were calculated after removing the non-significant interaction term. Post hoc tests were corrected for multiple comparisons using the Benjamini-Hochberg procedure (FDR method).  Table S10. Model 10: GLMM fitted to test the probability that queens were alive with a brood at the end of the experiment. The colony of origin of the queens and the nest id nested within the assay id nested within the year were included as random factors. The model includes an interaction term for the two first factors. Estimates for main factors were calculated after removing the non-significant interaction term. Post hoc tests were corrected for multiple comparisons using the Benjamini-Hochberg procedure (FDR method).   Table S13. Survival probability after five months of monogyne (M) and polygyne (P) queens (focal queens) according to their mode of founding and to the social origin of the queens that co-founded with them (non-focal queens). Number of focal queens is indicated in parentheses.  Table S14. Model 12: GLMM fitted to test the probability that co-founding queens were dead at the end of the experiment. The colony of origin of the queens and the nest id nested within the assay id nested within the year were included as random factors. The model includes an interaction term for the two first factors. Estimates for main factors were calculated after removing the non-significant interaction term.