Haemosporidian parasites of birds are ubiquitous in terrestrial ecosystems, but their coevolutionary dynamics remain poorly understood. If species turnover in parasites occurs at a finer scale than species turnover in hosts, widespread hosts would encounter diverse parasites and potentially diversify as a result. Previous studies have shown that some wide-ranging hosts encounter varied haemosporidian communities throughout their range, and vice-versa. However, it remains difficult to test spatial patterns of diversity in this complex multi-host multi-parasite system because it remains inadequately surveyed. We sought to understand how and why a community of avian haemosporidian parasites varies in abundance and composition across an array of eight sky islands in southwestern North America. We tested whether bird community composition, aspects of the environment, or geographic distance explain parasite species turnover in a widespread, generalist host. We sampled 178 Audubon's Warblers (Setophaga auduboni) along elevational transects in eight mountain ranges and screened them for haemosporidian mtDNA. We tested predictors of infection using generalized linear models (GLMs) and we tested predictors of bird- and parasite-community dissimilarity using generalized dissimilarity modeling (GDM). Predictors of infection differed by genus: Parahaemoproteus was predicted by elevation and climate, Leucocytozoon varied idiosyncratically among mountain ranges, and Plasmodium was unpredictable, but rare. Parasite species turnover was nearly three-fold higher than bird species turnover and was predicted by elevation, climate, and bird community composition, but not by geographic distance. Haemosporidian communities vary strikingly at spatial scales of hundreds of kilometers, across which the bird community varies only subtly. The finer spatial scale of turnover among parasites species implies that their ranges tend to be smaller than those of their hosts. Avian host species should encounter different parasite species in different parts of their ranges, resulting in spatially varying selection on host immune systems. Furthermore, the fact that parasite turnover was predicted by bird turnover implies that different species within a host community affect each other's parasites, potentially facilitating indirect antagonistic effects.