Towards Defining The Ideal Sample Size For Spatial Distribution Modeling Using A Virtual Vector

The objective of this work is to use a virtual disease vector at the scale of the European continent for the empirical testing of the number of spatial sampling points that are needed to develop stable spatial models to reliably map the distribution of disease pests: How many sampling points should one need to take to build stable spatial models? Despite the fact that this is crucial information to enable the cost-efficient planning of spatial sampling campaigns little has been published on this topic. To achieve this objective, we first designed a ‘virtual vector’ with a known distribution based on a set of climatic and environmental data at a 1X1 km grid. We then selected 10 areas of 400X400 km representative of different ecoclimatic settings of our virtual
vector in Europe. In each of these areas we computed 500 models with sample sizes varying from 10 to 5000 randomly selected sampling points. In this paper we present preliminary results and discuss how these results may contribute to improve the quality and reduce the cost of spatial distribution models and guide our future research.