Addressing the importance of constraint and adaptation on the ecology and evolution of host specificity in parasites is an ongoing challenge. Adaptive explanations for the evolution of host specificity predict that specialists are derived from generalists and experience greater extinction rates because of limited ability to adapt to changing conditions. Non-adaptive explanations predict that opportunities to parasitize hosts are limited -- geographic barriers and endemism limit the diversity of organisms to parasitize. Here I explore these predictions by (1) developing statistics that allow for the synthesis of published information on host specificity; (2) compiling published phylogenies of parasites to synthesize macroevolutionary patterns of host specificity; (3) using comparative methods to evaluate constraints between the richness and phylogenetic relatedness of host species; and (4) using biogeographic information to test non-adaptive explanations for variation in host specificity. Analysis of published information on host specificity requires a synthesis of comparative and meta-analytical methods. I develop a statistical framework based on the generalized theory of least squares that integrates phylogenies into meta-analysis, and outline a protocol for testing evolutionary hypotheses by contrasting neutral and adaptive models of evolution based on an Ornstein-Uhlenbeck process. Using evolutionary meta-analysis, I synthesize 43 published phylogenies to show that macroevolutionary predictions on host specificity -- where specialist parasites are continually evolving from generalists and going extinct -- are not universal for parasites. The exceptions were many, and phylogenies where generalists are derived from specialists occur as frequently as lineages with the predicted generalist-to-specialist trajectory.
Focusing on 70 species of obligate avian brood parasites, I further show that the host range (richness of host species) of brood parasites predicts host phylogenetic diversity (average shared phylogenetic distance of host species) worse than random, and that 96% of broad host ranges are invariant in host phylogenetic diversity; indicating that generalists can be as constrained as specialists in the phylogenetic diversity of hosts.
Finally, I show that geographic range size is a good predictor of host specificity in avian brood parasites -- where specialist brood parasites occupy smaller geographic ranges that generalists with many host species. This relationship appears to be mediated by the broad continental endemism of independently evolved clades of brood parasites.