Body-mass or sex-biased tick parasitism in roe deer (Capreolus capreolus)? A GAMLSS approach

Macroparasites feeding on wildlife hosts follow skewed distributions for which basic statistical approaches are of limited use. To predict Ixodes spp. tick burden on roe deer, we applied Generalized Additive Models for Location, Scale and Shape (GAMLSS) which allow incorporating a variable dispersion. We analysed tick burden of 78 roe deer, sampled in a forest region of Germany over a period of 20 months. Assuming a negative binomial error distribution and controlling for ambient temperature, we analysed whether host sex and body mass affected individual tick burdens. Models for larval and nymphal tick burden included host sex, with male hosts being more heavily infested than female ones. However, the influence of host sex on immature tick burden was associated with wide standard errors (nymphs) or the factor was marginally significant (larvae). Adult tick burden was positively correlated with host body mass. Thus, controlled for host body mass and ambient temperature, there is weak support for sex-biased parasitism in this system. Compared with models which assume linear relationships, GAMLSS provided a better fit. Adding a variable dispersion term improved only one of the four models. Yet, the potential of modelling dispersion as a function of variables appears promising for larger datasets.