Artificial Selection and the Skin Microbiome Independently Predict Parasite Resistance

It is well established that host genetics determine much of hosts response to parasites, but recent research has highlighted that the host microbiome plays a role in defense against parasites. Interactions between host resistance to parasites and the microbiome are widely acknowledged; for example, the microbiome can modulate the host immune response, and vice versa. However, it remains unclear how the host immune system and the microbiome may together influence the host's overall response to parasites. In many species males and females differ in both their genetically-based parasite resistance and the composition and structure of their associated microbiomes. We may therefore expect interactions between host resistance and the microbiome to differ between males and females, potentially explaining the often-observed variation in their response to parasites. To investigate how host genetics and the microbiome interact to shape host response to parasites, we imposed truncation selection on Trinidadian guppies (Poecilia reticulata) for their response to the specialist ectoparasite Gyrodactylus turnbulli. The 30% of guppies with the lowest number of parasites, 'parasite load', during experimental infection founded the resistant line, and the 30% with the highest parasite load founded the susceptible line, while randomly chosen uninfected fish founded the control line. After 3-6 generations of breeding in the absence of parasites, we sampled the skin-associated microbiome of fish from these lines and then infected them with G. turnbulli. We used Dirichlet multinomial modeling (DMM) machine learning to identify the bacterial community types across lines and evaluated the importance of selection line and bacterial community type in explaining variation between fish in parasite load over time 'infection severity'. Among females, the resistant line developed significantly lower infection severity, and the susceptible line developed significantly higher infection severity, than any other treatment. Among males, however, while the susceptible line developed significantly higher infection severity than the other lines, there was no difference between the resistant and control lines. We also found that female fish in the resistant line had significantly lower tolerance compared to the control or susceptible line females. This apparent tradeoff between resistance and tolerance was also supported by analysis of the individual-level data. We found that host-associated microbiomes explain as much of the variation in infection severity as host genetics, and that these two factors appear independent and additive in their effects on infection severity. Furthermore, a post-hoc test showed that a model including both artificial selection line and bacterial community type is significantly better at explaining variation in infection severity than models with either variable alone. Our results therefore suggest that the microbiome plays an active role in the interaction between host and parasite and is not a passive reflection of host genetics.