Finding the appropriate variables to model the distribution of vector‐borne parasites with different environmental preferences: climate is not enough

Abstract

Understanding how environmental variation influences the distribution of parasite diversity is critical if we are to anticipate disease emergence risks associated with global change. However, choosing the relevant variables for modelling current and future parasite distributions may be difficult: candidate predictors are many, and they seldom are statistically independent. This problem often leads to simplistic models of current and projected future parasite distributions, with climatic variables prioritized over potentially important landscape features or host population attributes. We studied avian blood parasites of the genera Plasmodium, Haemoproteus and Leucocytozoon (which are viewed as potential emergent pathogens) in 37 Iberian blackcap Sylvia atricapilla populations. We used Partial Least Squares regression to assess the relative importance of a wide array of putative determinants of variation in the diversity of these parasites, including climate, landscape features and host population migration. Both prevalence and richness of parasites were predominantly related to climate (an effect which was primarily, but not exclusively driven by variation in temperature), but landscape features and host migration also explained variation in parasite diversity. Remarkably, different models emerged for each parasite genus, although all parasites were studied in the same host species. Our results show that parasite distribution models, which are usually based on climatic variables alone, improve by including other types of predictors. Moreover, closely related parasites may show different relationships to the same environmental influences (both in magnitude and direction). Thus, a model used to develop one parasite distribution can probably not be applied identically even to the most similar host-parasite systems.

Diverse avian malaria and other haemosporidian parasites in Andean house wrens: evidence for regional co‐diversification by host‐switching

“…This pattern was not driven by elevational sampling bias because nearly half of our samples were from below this elevation. This pattern agrees with a recent study of haemosporidian richness and prevalence across the Iberian Peninsula (Pérez‐Rodríguez et al ). Remarkably, two Leucocytozoon lineages (L1, L5) were found at exceptionally high elevations (4123 and 4384 m a.s.l, respectively), as were single Haemoproteus (H9, 4300 m a.s.l.)…”

Habitat fragmentation, not habitat loss, drives the prevalence of blood parasites in a Caribbean passerine

“…in temperate regions with a very different climate (Pérez-Rodríguez et al 2013): in both regions Haemoproteus parasites are more frequent where yearly rainfall is relatively high and temperatures are stable year-round. In turn, it stands out how H-LOXPOR01 prevalence is determined, in our tropical study site, by the same climatic constraints that determine prevalence of Haemoproteus spp.…”

“…the proportion of infected hosts), as well as to decouple the relative importance of the metrics of habitat destruction versus other variables (Pérez-Rodríguez et al 2013). the proportion of infected hosts), as well as to decouple the relative importance of the metrics of habitat destruction versus other variables (Pérez-Rodríguez et al 2013).…”

“…At the landscape level, we also considered the composition and the spatial configuration of both forests and the habitat matrix, which can affect host and/or vector mobility among patches. Finally, and since climatic variables seem to play a key role in explaining the diversity and abundance of these parasites (Sehgal et al 2011, Loiseau et al 2013, Pérez-Rodríguez et al 2013, we also consider the climatic environment of each of the sampled locations. We use PLS regression, a technique rarely used on ecological studies (Carrascal et al 2009), as a convenient tool to overcome these statistical problems.…”

When and how should biotic interactions be considered in models of species niches and distributions?

“…Various correlative approaches for modelling the distributions of (or suitability for) individual species have seen widespread application, typically only using abiotic predictors (generally termed ecological niche modelling or species distribution modelling; Guisan & Thuiller, ; Elith & Leathwick, ; Peterson et al ., ; Anderson, ). Recently, many studies have integrated biotic interactors into mechanistic niche models (Kearney & Porter, ) as well as correlative ones (Meier et al ., ; Godsoe & Harmon, ; Jaeschke et al ., ; Linder et al ., ; Blois et al ., ; Giannini et al ., ; González‐Salazar et al ., ; Pellissier et al ., ; Pérez‐Rodríguez et al ., ). Various other approaches consider multiple species simultaneously (Richmond et al ., ; Kissling et al ., ; Clark et al ., ; Pollock et al ., ).…”