Organisms or societies are resource limited, causing important trade-offs between reproduction and defence. Given such trade-offs, optimal allocation theory predicts that, for animal societies with a soldier caste, allocation to soldiers should reflect local external threats.  Although both threat intensity and soldier allocation can vary widely in nature, we currently lack strong evidence that spatial variation in threat can drive corresponding variation in soldier allocation. The diverse guild of trematode parasites of the California horn snail provides a useful system to address this problem. Several of these species form colonies in their hosts with a reproductive division of labour including a soldier caste. Soldiers are non-reproductive and specialized on defence, attacking and killing invading parasites. We quantified invasion threat and soldier allocation for 168 trematode colonies belonging to 6 species at 26 sites spread among 10 estuaries in temperate and tropical regions. Spatial variation in invasion threat was matched as predicted by the relative number of soldiers for multiple parasite species. Soldier allocation correlated with invasion threat at fine spatial scales, suggesting that allocation is at least partly inducible. These results may represent the first clear documentation of a spatial correlation between allocation to any type of caste and a biotic selective agent.

Target species collection

We focused on six trematode species that have a soldier caste and infect California horn snails (Table 1) [16-18]. We collected over 300 snails from each of nine estuaries across California and two estuaries across Panama. We aimed to collect a minimum of 100 snails by hand during low tide from three sites per estuary. However, at one Panamanian estuary where snails were extremely patchy, we collected all snails from two sites, while at another, we opportunistically collected snails from eight sites to find our trematode species of interest. Additionally, in areas we expected to have low infection prevalence (given previous survey work), we collected more snails to get sufficient numbers of trematode colonies for dissection (Tables S3-S4). A total of 5520 snails were dissected to estimate infection prevalence, a proxy for invasion threat.

Identification, colony processing, index of invasion threat

In the laboratory, snail length was measured with Vernier callipers, processed following Torchin et al [25], and trematode species identified following Hechinger [31]. We assessed soldier allocation and deployment for trematodes colonies that were the only infection within a snail (i.e. no coinfection; Table 1). The deshelled snails were divided into three sections, and soldiers and reproductives were counted in each section (Figure 1B).

            As a metric for invasion threat, we summed the prevalences (proportion of hosts infected) of all trematode species [32]. This metric provides the cumulative risk of invasion by any trematode species at that site. We calculated infection prevalence for each site and each estuary (pooling all snails) (Table S3-S4).

See Read_Me.doc files for important information about raw data tables.