The effect of reproductive occurrences and human descriptors on invasive pet distribution modelling: Trachemys scripta elegans in the Iberian Peninsula

Abstract

Pet release is important for the introduction of invasive species but propagule pressure from this pathway is higher in areas with intense human presence. In fact, potential invaders are frequently released into areas with sub-optimal environmental conditions, and are not able to reproduce or establish viable populations. SDMs are mainly focused on determining species environmental suitability and in the case of invasive species modelling these do not integrate the invasion as a process, missing the crucial human contribution. Our model aims to introduce novelty by incorporating variables representing human activities such as pet shop densities or human areas. These are directly related with the studied species invasion pathway. Here we investigate the binomial influence of reproductive occurrences and human-related variables on invasive pet distribution modelling. As our model species, we used the red-eared slider turtle, Trachemys scripta elegans (T.s.e.), an iconic invasive species with worldwide dissemination. Ten variables (predictors) were selected for T.s.e. modelling. These were four bioclimatic (BIO1=Annual Mean Temperature, BIO10=Mean Temperature of Warmest Quarter, BIO11=Mean Temperature of Coldest Quarter and BIO12=Annual Precipitation), three topographic (altitude, slope and the compound topographical index (CTI)), and three human variables (pet shop density, distance to urban areas and human density). To model suitability for T.s.e. establishment we tested four combinations of three sets of descriptor variables (human, environmental and both), with two types of occurrence records (reproductive or all records), at a global (worldwide) or regional scale (Iberian Peninsula (IP)). For each combination, an ensemble of niche-based models (NBMs) was built using nine different types of models implemented in BIOMOD2.In worldwide modelling, the use of reproductive occurrence records enables the most accurate predictions. At a regional scale, the combination of environmental and human variables promotes higher accuracy than using human variables alone. For worldwide and regional modelling, the most important variables are associated with temperature (Mean Temperature of Coldest Quarter; Annual Mean Temperature; Mean Temperature of Warmest Quarter). For regional modelling the human variable with highest importance was human density.Our approach may be useful for the management of released pet invaders, namely identifying high invasion risk areas where management efforts should be prioritized. From our models we recommend that T.s.e. management efforts should be prioritized to coastal and lowland areas. This study highlights that when modelling invasive species distributions, it is very important to use records that accurately represent species establishment (e.g. species breeding). Moreover, it clarified that variables associated to propagule pressure can improve model performance, even though these are much less used than bioclimatic variables. This modelling framework may be applied to other invasive species with different invasion pathways, by using different variables, directly linked with the respective invasion vectors and taking into account anthropogenic pressure. These modelling procedures should provide a better view of areas at risk of invasion, should improve management efforts and should also provide important insights on the role of human vectors on the shaping of invasive species distribution.