Can we predict where to expect host-associated differentiation?

Background/Question/Methods: When reproductive isolation evolves between parasite populations living on different host species, each population can independently accumulate genome-wide genetic differentiation due to their host-specific associations. This phenomenon is called host-associated differentiation (HAD) and has been proposed as a source for parasite biodiversity. Although many parasite-host case study systems have been shown to involve HAD, it is uncertain which ecological and biological traits best explain the general occurrence of HAD. Still, several factors have been proposed within verbal models aimed to explain the evolution of HAD in specific case study systems. The proposed factors that best explain the occurrence of HAD involve mechanisms that facilitate reproductive isolation among parasites on different host species (e.g., endophagy, asexual reproduction, or allochrony). However, these factors have yet to be quantitatively correlated with the presence of HAD. In this study, we performed a meta-analysis that evaluated the correlation between potential explanatory factors and the general occurrence of HAD.

Results/Conclusions: A discriminant analysis of HAD occurrence based on the incidence of explanatory factors demonstrated that parasite-host case study systems involving HAD can successfully be distinguished from case studies without HAD. Based on this, an infinite random forest analysis was used to generate a hierarchy of conditional probabilities based on the incidence of significant explanatory factors. The hierarchy of factor-based conditional probabilities provides a tool for making reliable predictions about the likelihood of a given parasite-host system to show HAD and indicates distinct routes toward its acquisition. We discovered that the incidence of immigrant inviability, gall-making, and their combination best explain the presence of HAD in a given parasite-host system. Also, contrary to expectation, endophagous feeding best explains the absence of HAD, indicating that the majority of endophagous feeders are not expected to show HAD.