Estimates of integration, mean trait correlation, constraints, and flexibility may limit pupfish diversification on San Salvador Island
Our results suggest that greater independence in trait variation is associated with diversification in this system. The P matrix of the SSI generalist-only group had significantly greater estimates of integration and mean squared correlation between craniofacial traits, higher estimates of constraints, and lower estimates of flexibility– all measurements that can be interpreted as proxies for independence between traits (Table 2; Figure 6). For example, we observed the highest correlation between cranial traits in the SSI generalist-only group, which indicates that the mechanisms producing shifts in trait values for one trait are likely to affect the values for many others, which ultimately may limit the phenotypic space that these populations can access (Goswami & Polly, 2010). Furthermore, the constraint and flexibility estimates for the SSI generalist-only group suggest that they are limited, specifically in comparison to Caribbean and SSI radiating group, in their ability to respond to and align with a wide range of selection gradients. This limited ability to respond to a range of selection gradients, may explain why these populations are exposed to similar environmental conditions, yet have not evolved the snail- and/or scale-eater specialists found in other lakes.
Other empirical studies also suggest that independence across many biological levels may be positively associated with diversification. Greater amounts of gene duplication, larger ratios of nonsynonymous:synonymous mutations, and lower levels of covariation between morphological traits, have all been suggested as integral to diversification in some African cichlid radiations (Brawand et al., 2014; Machado et al., 2014; Selz et al., 2014), and represent independence at both the genetic and morphological levels. Similarly, Ravinet et al. (2014) suggest that populations of three-spined stickleback from the Japan Sea have not colonized freshwater habitats and subsequently diversified, as many of the closely related Pacific freshwater populations have done, specifically due to the increased correlation between their morphology and dietary niche. Lastly, the flexible stem hypothesis suggests that the divergence within adaptive radiations emerges from plastic phenotypic variation in an ancestral population (West-Eberhard, 2003). Through this lens, the properties of the Caribbean P-matrix could be viewed as being plastic (i.e., low levels of constraints, high levels of flexibility, and low respondability)—estimates which remain largely the same in radiating groups and which have shifted in non-radiating groups. Other empirical studies, however, have found mixed results regarding the relationship between plasticity and diversification. Navolón et al. (2020) observed that higher integration, which may imply lower levels of plasticity, likely helped produce the diversity seen in the adaptive radiations of Darwin’s finches and Hawaiian honeycreepers. On the other hand, lower levels of integration in brown trout allowed them to diversify into a new ecological niche when faced with a new selective pressure (Závorka et al., 2017). These results support that independence and flexibility, either within or between, molecular, morphological, and/or behavioral traits contributes to a group’s ability to diversify. The findings presented in this study further support this conclusion.