Supervised forecasting of the range expansion of novel non‐indigenous organisms: Alien pest organisms and the 2009 H1N1 flu pandemic

Abstract

AbstractAimWe propose a simple supervised method for forecasting the expansion of the geographical range of non‐indigenous species without any detailed information on the pathway mechanisms. The models currently used, such as metapopulation models, population dynamics [e.g., the susceptible‐infected‐recovered (SIR) model] with diffusion process, and models using individual‐based mechanisms, require specific parameters according to the assumed mechanisms. These parameters are difficult to obtain immediately after the first infestation is detected, although subsequent pandemics of novel pathogens and non‐indigenous pest organisms can be fatal to humans and damage agriculture and native ecosystems.InnovationA simple machine‐learning approach was developed. The future distribution of the target species was predicted by composing the spreading patterns of the previous range expansions of various alien species similar to the target species. A matrix representing the early–late relationship of infestation between regions was used to compare and compose various range expansion patterns. The eigenvector of the composed matrix gives the predicted order of infestation. Spreading speed was considered using the standard deviation of infestation dates, and future infestation dates were calculated by a reduced major axis with the eigenvector. In the feasibility assessment, the study area (Japan) was divided into many small regions (prefectures), and the dates of infestation in these regions were predicted. Our method successfully predicted the infestation dates of various alien species and the novel H1N1 influenza pandemic in 2009.Main conclusionsOur method is less complex and requires simpler information than other methods and is applicable to all organisms. Constructing a database of the range expansions of alien species, including novel pathogens and agricultural and environmental pests, at the global level and with high spatial resolution, such as at the county or city level, will facilitate the prediction of future range expansions of hazardous organisms.