Parasites in space and time: a case study of haemosporidian spatiotemporal prevalence in urban birds.

Prevalence responses to anthropic factors differ across hosts and parasite species. We here analyzed the spatiotemporal variation of avian haemosporidian prevalence in bird assemblages of the Mooswald forest (i.e., urban greenspace; Freiburg, Germany), in response to local environmental features (e.g., water sources, human presence (visited)/absence (unvisited)) and bird-level traits (e.g., body condition, age, sex) in 2 years. We used a nested PCR protocol (mitochondrial (mt)DNA cytochrome b (cyt b) gene) and microscopy to determine haemosporidian infections. Prevalence was analyzed using a general linear multi-model (glmulti) approach with Akaike information criterion corrected for small samples (AICc), with subsequent model inferences using a GLMM on the best selected model, considering bird species as a random factor. Analyses were conducted for the main understory bird species (Blackcap - Sylvia atricapilla, Chaffinch - Coereba flaveola, Great Tit - Parus major, Blue Tit - Cyanistes caeruleus, European Robin - Erithacus rubecula, Blackbird - Turdus merula, Song Thrush - Turdus philomelos). We further conducted spatial autocorrelation analyses for all haemosporidian infections, and classification and regression trees (CARTs) for focal species. We analyzed a total of 544 samples of seven bird species. In 2011 prevalence for Haemoproteus/Plasmodium was 25.8% and 11.7% for Leucocytozoon. In 2013 prevalence for Haemoproteus/Plasmodium was 26.5% and 35.5% for Leucocytozoon. Haemosporidian prevalence was significantly different between some focal species. There was a negative association between distance to the nearest water source and prevalence in the year 2011, and the opposite pattern for the year 2013. However, when analyzed for the six focal species separately, such a relationship could change from a negative to a positive one, or there could be no relationship at all. For Leucocytozoon there was higher prevalence in the section of the forest visited by humans. We did not find spatial autocorrelation for prevalence across the study site, but there were statistically significant local spatial clusters in the visited section. Although there were similar responses of prevalence to some factors, infection patterns were generally bird species-specific. Thus, prevalence is a labile epidemiological parameter, varying spatiotemporally in an idiosyncratic way.