Disentangling the spread dynamics of insect invasions using spatial networks

Describing and understanding spatiotemporal spread patterns in invasive species remains a longstanding interdisciplinary research goal. Here we show how a network-based top-down approach allows the efficient description of the ongoing invasion by Drosophila suzukii in Chile. To do so, we apply theoretical graph methods to calculate the minimum cost arborescence graph (MCA) to reconstruct and understand the invasion dynamics of D. suzukii since the first detection in 2017. This method estimates a directed rooted weighted graph by minimizing the total length of the resulting graph. To describe the temporal pattern of spread, we estimate three metrics of spread: the median dispersal rate, the median coefficient of diffusion, and the median dispersal acceleration. The estimated MCA shows that over four years, D. suzukii colonized a ~1,000 km long strip in the central valley of Chile, with an initial phase with long paths and connections and no clear direction pattern, folowed by a clearer north-east propagation pattern. The median dispersal rate for the entire period was 8.8 [7.4 - 10.6,95% CI], while the median diffusion coefficient was 19.6 meters2/day [13.6 - 27.9, 95% CI]. The observed spread dynamics and the log-normal distribution of accelerations are consistent with longdistance dispersal events. The complexities of real landscapes cannot be summarized in any model, but this study shows how an alternative top-down approach based on graph theory can facilitate the ecological analysis of the spread of an invasive species in a new territory.