Methodological caveats in the environmental modelling and projections of climate niche for ticks, with examples for Ixodes ricinus (Ixodidae).

Interest is increasing in inferring the climate niche of health-threatening arthropods and projecting such inferences onto a territory. This approach is intended to predict the range of tick distribution and to elucidate tick responses to climate scenarios, using so-called correlative models. However, some methodological gaps might prevent achieving an adequate background for hypothesis testing. We explore, describe, and illustrate these procedural inaccuracies with examples focused on the tick Ixodes ricinus and examine how these factors might affect modelling outcomes. Our aim was to develop a backdrop of rules for developing reliable models for these parasites. The use of partial sets of tick occurrences might produce unreliable associations with climate because the algorithms cannot capture the complete niche with which the tick is associated. Reliability measures of the model cannot detect these inaccuracies, and undesirable estimations of the niche will prevail in the chain of further calculations. The use of inadequate environmental variables (covariates) may lead to inflation of the results of the model through two statistical processes, autocorrelation and colinearity. We demonstrate the high colinearity existing in climate products derived from interpolation of climate recording stations. Our explicit advice is to focus on the training of climate models with satellite-derived information of climate, from which colinearity of the time series has been removed through a harmonic regression. We also emphasize the high uncertainty if inference about the climate niche is expanded into different time slices, like projected climate scenarios.