Parasite counts or parasite incidences? Testing differences with four analyses of infracommunity modelling for seven parasite-host associations.

One of the challenges in studies of parasite community ecology is whether the input data for analyses should be parasite abundances/counts, i.e. count data (CD), or parasite incidences (presences/absences), i.e. incidence data (ID). We analysed species responses to environmental factors and species associations in the infracommunities of helminths and ectoparasites in four hosts from Europe (Sorex araneus and Myodes glareolus) and South Africa (Rhabdomys pumilio and Rhabdomys dilectus) and compared the results of four analyses [redundancy analysis (RD), RLQ analysis, joint species distribution modelling (JSDM) and Markov random fields (MRF)] that used either CD or ID as an input. In addition, we compared the differences between the CD and ID results of two analyses (JSDM and MRF) across parasite species between (a) host species within helminths and ectoparasites; (b) helminths and ectoparasites within a host species; and (c) parasite species with contrasting levels of intensity. The results of most analyses for the majority of parasite-host associations were qualitatively similar. However, models based on the ID input performed better than models based on the CD input in three out of four types of analyses (RDA, JSDM and MRF). The differences between the CD and ID models varied between host species (being the lowest in R. pumilio for JSDM and in S. araneus for MRF). However, they were not affected by the level of parasite intensity.