Development of inland lakes as hubs in an invasion network.
The ability to predict spatially explicit dispersal by non-indigenous species is a difficult but increasingly important undertaking as it allows management efforts to be focused around areas identified as susceptible to invasion. Lakes may serve as useful models for these studies because the habitats are well defined, and vectors of spread may be readily identified and quantified. In this study, we examined patterns of spread of the non-indigenous spiny waterflea Bythotrephes longimanus to inland lakes in Ontario, Canada, to identify lakes for which management efforts to reduce traffic would be most effective. We surveyed people using lakes for recreational purposes to quantify movements of trailered boats and other risky activities, to model relative vector traffic from invaded lakes to non-invaded and other invaded lakes. Non-linear functions were developed to describe the cumulative number of invaded and non-invaded destination lakes visited by people leaving five important lakes already invaded by the spiny waterflea (Huron, Simcoe, Muskoka, Panache and Kashagawigamog). The relative difference in these functions was used to identify which lakes will develop into future invasion hubs and will therefore be most important to future dispersal of the species. In the recent past, Lake Muskoka has been an important hub from which the spiny waterflea has invaded other lakes. It is unlikely to continue to be a source for waterflea invasion as most outbound traffic is to previously invaded lakes. Conversely, most outbound traffic from Lakes Kashagawigamog and Simcoe is to non-invaded lakes and, therefore, these lakes are likely to develop into hubs in the future. Synthesis and applications. These data on zooplankton in lake systems and associated mechanisms of transport indicate patterns not only of intrinsic value to lake management, but also of potential importance in understanding invasions more generally. Frequency distributions of the number of outbound connections to both invaded and non-invaded destinations from invaded sources follow a power function, consistent with scale-free networks. These networks indicate that small proportions of sources function as hubs. Management efforts targeted to remove developing hubs from the invasion network, rather than equal effort applied to outbound vector traffic from all sources, may reduce the predicted rate of new invasions.