Predicting the outcomes of management strategies for controlling invasive river fishes using individual-based models

摘要

1. The effects of biological invasions on native biodiversity have resulted in a range of policy and management initiatives to minimize their impacts. Although management options for invasive species include eradication and population control, empirical knowledge is limited on how different management strategies affect invasion outcomes. 2. An individual-based model (IBM) was developed to predict how different removal (‘culling') strategies affected the abundance and spatial distribution of a virtual, small-bodied, r-selected alien fish (based on bitterling, Rhodeus sericeus) across three types of virtual river catchments (low/intermediate/high branching tributary configurations). It was then applied to nine virtual species of varying life-history traits (r-to K-selected) and dispersal abilities (slow/intermediate/fast) to identify trade-offs between the management effort applied in the strategies (as culling rate and the number of patches it was applied to) and their predicted effects. It was also applied to a real-world example, bitterling in the River Great Ouse, England. 3. The IBM predicted that removal efforts were more effective when applied to recently colonized patches. Increasing the cull rate (proportion of individuals removed per patch), and its spatial extent was effective at controlling the invasive population; when both were relatively high, population eradication was predicted. 4. The characteristics of the nine virtual species were the main source of variation in their predicted abundance and spatial distribution. No species were eradicated at cull rates below 70%. Eradication at higher cull rates depended on dispersal ability; slow dispersers required lower rates than fast dispersers, and the latter rapidly recolonized at low cull rates. The trade-offs between management effort and the outcomes of the invasion were, generally, optimal when intermediate effort was applied to intermediate numbers of patches. In the Great Ouse, model predictions were