Rainfall can explain adaptive phenotypic variation with high gene flow in the New Holland Honeyeater (Phylidonyris novaehollandiae)

Identifying environmentally driven changes in traits that serve an ecological function is essential for predicting evolutionary outcomes of climate change. We examined population genetic structure, sex-specific dispersal patterns, and morphology in relation to rainfall patterns across an island and three peninsulas in South Australia. The study system was the New Holland Honeyeater (Phylidonyris novaehollandiae), a nectarivorous passerine that is a key pollinator species. We predicted that rainfall-related mechanisms would be driving local adaptation of morphological traits, such that in areas of lower rainfall, where nectar is less available, more insectivorous traits - shorter, deeper bills, longer tarsi, and longer wings - would be favored. The study populations differed in phenotype across the Eyre, Yorke, and Fleurieu Peninsulas and Kangaroo Island despite high gene flow (single continuous population) and sex-biased dispersal (males were philopatric and females dispersed). We tested the role of rainfall in shaping the observed phenotypic differences, and found strong support for our predicted relationships: birds in areas of higher rainfall had higher condition indices, as well as longer bill-head length, deeper bills, and shorter tarsi. Bill depth in males in high-rainfall sites showed signals of stabilizing selection, suggesting local adaptation. In addition to these local indications of selection, a global pattern of directional selection toward larger size for bill-head length, bill-nostril length, and wing length was also observed. We suggest this pattern may reflect an adaptive response to the relatively dry conditions that South Australia has experienced over the last decade. We conclude that rainfall has shaped aspects of phenology in P. novaehollandiae, both locally, with different patterns of stabilizing and directional selection, and globally, with evidence of adaptive divergence at a landscape scale.