Group size rather than social status influences personal immune efficacy in a socially polymorphic bee
Abstract
The evolution of group living is associated with increased pressure from parasites and pathogens. This can be offset by greater investment in personal immune defences and/or the development of cooperative immune defences (social immunity). An enduring question in evolutionary biology is whether social-immune benefits arose in response to an increased need in more complex societies, or arose early in group living and helped facilitate the evolution of more complex societies. In this study, we shed light on this question through investigating how immunity varies intraspecifically in a socially polymorphic bee. Using a novel immune assay, we show that personal antibacterial efficacy in individuals from social nests is higher than that of solitary individuals, but that this can be explained by higher densities in social nests. We conclude that personal immune effects are likely to play a role in the social/solitary transition in this species. These patterns are consistent with the idea that social immunity evolved secondarily, following the evolution of group living. The flexibility of the individual immune system may have favoured a reliance on its use during the facultative phase early in social evolution.
1. Background
Living in societies precipitates advantages through specialization and cooperation, but comes at the cost of increased resource competition and a heightened threat from pathogens. Pathogens (defined here broadly to include viruses, bacteria, protozoans, helminths and fungi) pose a greater threat to groups because transmission is more likely at higher densities and among genetically related individuals [1–6]. To combat these heightened risks, animals living in groups have been reported to increase investment in personal immunity (i.e. density-dependent prophylaxis or DDP; [7,8]). Ant queens cooperatively founding nests, for example, have higher investment in personal immunity than solitary foundresses [9], and cross-species studies of wasps, bees and thrips, have shown that the efficacy of cuticular antimicrobials increases with group size ([10–12], but see [13]). Group living species can also employ a diverse array of cooperative immune defences which extend beyond the actor to benefit other group members. These defences are collectively known as ‘social immunity’ [14] and include mutual grooming, exclusion of infected individuals, and use of antimicrobial secretions [5,14–19]. Social immunity can ameliorate the need for personal immunity, and investment in social immunity may trade-off against investment in personal immunity because of the cost of mounting immune responses [5,20].
Social immunity was initially considered a characteristic of highly eusocial societies with obligatory group living. However, this perspective has since been questioned on the grounds that it is based on several misconceptions, including mounting evidence for social-immune defences in non-eusocial organisms [21]. This raises the question of whether social immunity can be considered a secondary trait that arose following the evolution of complex obligatory social systems (the ‘eusociality framework’ [14,22,23]), or an ancestral trait which may have facilitated the transition from solitary to group living, and which can thus be considered an important factor in the evolution and elaboration of social behaviour (‘group living framework’ [5,20,21]).
Further elucidation of the relationship between social evolution and immune function has been constrained by the limitations of appropriate model systems, including potential confounding effects arising from cross-species comparisons [10–12,24], transient life-history stages [9], the use of sub-social species, which we might expect to have limited social-immune behaviours [25], and use of highly eusocial species, which are not suitable for studies of early stages in social evolution [5,20]. In this study, we provide a new perspective on this question by exploring the relationship between social context and personal immune efficacy in a facultatively social organism: the small carpenter bee Ceratina okinawana. In this species, both social and solitary forms exist in the same population, allowing us to compare social context without interspecific cofounding effects. We use a novel immune assay to explore how personal immunity varies between individuals from solitary and social colonies, allowing us to shed new light on the relative importance of personal and social immunity in the transition between solitary and social behaviour.
2. Methods
(a) Model system
Entire colonies of Ceratina okinawana were collected in October 2021 in stems of Miscanthus sinensis from Tokashiki island, Japan (26.19611°N, 127.3606°E). Nests were opened in the laboratory and adults were maintained in identical laboratory conditions in artificial nests until immune assays (electronic supplementary material, figure S1). Colonies were classified as solitary (one adult female) or social (more than one adult female; electronic supplementary material, data file S1). We measured size characteristics (weight, wing length and head width) of bees and quantified mandibular wear and wing wear as proxies of age (electronic supplementary material, figure S2). In the absence of information on how the presence of different brood stages or sexes of adult bees might influence immune investment, we use total ‘group size’ (number of all brood and adults in the nest) as a measure of the density of individuals in the nest.
(b) Immune assays
For immune testing, we used a single adult female bee from each nest (n = 24 social and n = 32 solitary). For social nests, we selected a random individual or the larger individual when there was a clear size difference, under the assumption this was more likely to be the original founding queen and thus more directly comparable to solitary nesting bees (see also electronic supplementary material). We quantified personal immune efficacy by assessing the capacity for bee haemolymph to suppress bacterial growth, using a novel immune assay (electronic supplementary material, methods). Briefly, whole-bee haemolymph was extracted into a collection buffer by centrifuging bees after removing antennae. To this mix, we added Escherichia coli modified to grow in antibiotic medium, allowing us to avoid contamination from microorganisms naturally present in bees. This mixture was incubated for 15 h in antibiotic medium, and growth of E. coli quantified each hour over the subsequent 6 h as the change in the absorbance in this mixture using a spectrophotometer at 600 nm. As a control, we used a mix of buffer and bacteria (n = 10).
(c) Statistical analysis
Social and solitary colony bee size and wear characters, brood numbers and nest lengths were compared using Wilcoxon's signed-rank tests, while brood composition of nests was compared using Fisher's exact test. Immune efficacy was quantified as the total change in absorbance over the experimental duration. This was compared between treatments (social and solitary) and the control using the Kruskall–Wallace test and Dunn post hoc test using the dunn.test package of R [26]. We explored relationships between the log-transformed values of the change in absorbance and demographic and individual factors using a generalized linear model (GLM), which included social status, individual size, mandibular wear and number of group members (adults and brood) as factors and the interaction between group size and social status. We report test statistics from type-II analysis of variance on the output from the GLM, generated with the Anova function of the car package for R [27]. All analyses were performed in R v. 4.0 [28].
3. Results
(a) Contrasts between social and solitary nests
Individuals from social and solitary nests did not differ in size or wear characteristics (figure 1a), and there was no difference in the number of brood between social types (figure 1b). The nests of social colonies were longer (figure 1c; Wilcoxon's W = 614, p = 0.001), and social and solitary colonies differed in brood composition (Fisher's exact test, p = 0.033), though both types of colonies had representatives of all brood stages (figure 1d). Measures of adult size (wing length, head width and weight) were correlated with each other (Pearson's ρ > 0.91, p < 0.001) as were measures of wear (ρ = 0.55, p < 0.001), though there was no relationship between size and wear (electronic supplementary material, figure S3). We thus selected head width and mandibular wear as representative measures of size and age for use in GLM analysis.
Figure 1. Comparative characteristics for (a) adult females, (b) brood number, (c) nest length and (d) brood composition, from social (red) and solitary (blue) nests. Boxes indicate interquartile range, solid horizontal lines medians, whiskers minimum and maximum values, and points indicate outliers. Head width, wing length and nest length are in millimetres, wing wear and total brood are counts; mandibular wear is a ratio of width to height, and brood composition is the proportion of total brood.
(b) Immune efficacy in social and solitary individuals
Bacterial growth was suppressed by bee haemolymph, and this effect was greater in individuals from social nests than those from solitary nests (figure 2a). Control samples showed relatively low variance, indicating the assay approach yielded consistent results where expected, and that immune efficacy varied considerably among individuals from both solitary and social nests. However, differences between treatments were consistent with a density effect, as GLM analyses indicated that immune efficacy increased with group size (figure 2b; χ2 = 6.071, p = 0.013) while there was no interaction between group size and social status (χ2 = 0.179, p = 0.672) and no significant effect of other factors (electronic supplementary material, table S5).
Figure 2. (a) Boxplot of total change in absorbance for the two treatments and the control group. Sample sizes are included above the x-axis. Letters above bars indicate significant differences among groups based on Kruskal–Wallace and Dunn post hoc tests. (b) Jitter plot of relationship between overall change in absorbance and group size (number of adults and brood in the nest). Blue dots and line indicate solitary colonies and red dots and line indicate social colonies. Shaded areas denote 95% confidence intervals of the regression lines. Change in absorbance was quantified as the change in optical density at 600 nm of the medium containing haemolymph and bacteria, with higher values indicating greater bacterial growth and thus lower immune efficacy.
4. Discussion
We use a novel assay technique to demonstrate that bees from social and solitary nests in a facultatively social bee exhibit patterns consistent with differences in personal immune efficacy, with social nesting bees having higher antibacterial activity than solitary nesters. However, bees from larger colonies (more adults and brood) also had higher immune efficacy than those in smaller colonies, suggesting that differences in immune efficacy among social and solitary nests can be explained by a density-dependent prophylactic effect [7] rather than an influence of social status per se. We interpret this pattern as indicative of an increase in personal immunity in response to the greater risk of infection posed by larger groups. Previous studies have shown that lower immune efficacy in solitary individuals can result from the stress of being alone [29] (though this can also arise from stress associated with group living [4]). Additionally, while our experimental protocol effectively precluded of social-immune effects in our immune assay, this did not exclude the influence of social-immune effects occurring prior to experimentation, such as immune-priming through trophallaxis [30,31]. However, the above alternatives are unlikely given that (i) the costs of mounting an immune response mean that investment in personal immunity can generally be expected to limit investment in social immunity [32,33], and (ii) the relationship between immune efficacy and group size showed no evidence of a marked shift across the social/solitary boundary and was largely linear, as expected from density-related effects.
Density-dependent prophylactic effects have been documented in a wide range of non-social insects, but are also found in social species (reviewed in [34]). Similarly, social immunity, once considered the hallmark of eusocial organisms, has been found in sub-social species [21], blurring the expected association between social behaviour and social immunity. Comparative studies across different kinds of social systems have shown increases in individual immune efficacy with increasingly complex social groups of related species ([10–12], but see [25]), but cannot disentangle the effects of group size from those of social context, and are complicated by life-history differences among study species. The present study is the first study to our knowledge to investigate changes in immune efficacy in a socially polymorphic species, allowing us to investigate the role of social immunity in the early stages of social evolution without the potential confounding effects of interspecific comparisons. Our results indicate a reliance on personal immune systems over the transition from solitary to social behaviour in this species. These findings are consistent with the idea that social immunity evolved secondarily, following the evolution of group living (the ‘eusociality framework’) rather than being an ancestral feature which facilitated the evolution of sociality (group living framework) [21].
While we are able to demonstrate an overall effect of group size on personal immune efficacy across social and solitary colonies in C. okinawana, some limitations of our study await further investigation. First, our immune assay approach does not permit disentangling effects of haemolymph quality and quantity, and the relative importance of these factors deserves further investigation. Second, the haemolymph extraction protocol precludes an assessment of the reproductive status of adults, which may be important given the possible trade-off between investment in immunity and reproduction [35]. Therefore, although we targeted larger (and thus more likely to be reproductive [36]) individuals in social nests, the potential influence of this trade-off remains unclear. Third, we show that brood number and composition do not differ markedly between social and solitary colonies, but cannot rule out the possibility that brood of different developmental stages influence immune investment differently. Experimental manipulation of brood composition and/or comparative study of immune expression during colony ontogenetic development are thus needed. Finally, our findings are correlative and would benefit from experimental validation, such as through manipulations of group size.
DDP is a form a phenotypic plasticity [34], and as such may befit socially polymorphic species. The degree of flexibility offered by this mechanism is evidenced by studies of bumblebee workers, which can rapidly shift immune function in response to change in social context [37]. As transitions from solitary to social life are likely to progress through a facultative stage [38], a single system offering flexibility may be more adaptive at this stage than switching between different forms of immunity. Further studies of socially polymorphic species will help establish the generality of these patterns.
Data accessibility
The data are provided in the electronic supplementary material [39].
Authors' contributions
T.T.H.N.: conceptualization, data curation, formal analysis, investigation, methodology and writing—original draft and writing—review and editing; T.A.: conceptualization, formal analysis, investigation, methodology, project administration, resources, supervision, validation and writing—review and editing; A.L.C.: conceptualization, data curation, formal analysis, funding acquisition, methodology, project administration, resources, supervision, validation, visualization, writing—original draft and writing—review and editing.
All authors gave final approval for publication and agreed to be held accountable for the work performed therein.
Conflict of interest declaration
We declare we have no competing interests.
Funding
This study was funded by Tokyo Metropolitan University and Japan Society for the Promotion of Science (grant no. Kakenhi 21K06292).
Acknowledgements
We are grateful to the editor and two anonymous referees for constructive comments which helped improve the manuscript. We thank Seth Barribeau for helpful discussion on immune testing techniques and Vu Nguyen Thanh Binh for help collecting bees.