Forecasting dispersal of nonindigenous species

A leading conceptual model of the invasion process suggests that nonindigenous species (NIS) must pass through a series of ‘filters’ when dispersing from colonized to non-colonized regions. These steps include the initial dispersal of propagules, survival of these propagules upon encountering the new physicochemical environment, and biological integration into the new community. Here, I forecast invasions for two aquatic NIS, the spiny waterflea Bythotrephes longimanus and zebra mussel Dreissena polymorpha based on the entire invasion sequence using gravity models to assess movement of propagules, data on lake morphometry and physicochemistry, and data on fish community composition. The gravity models included information on movement patterns of recreationalists and life-history characteristics of the NIS that may facilitate invasions. I also contrast the abilities of a hierarchical approach to a single ‘all-in-one’ model that considered all variables simultaneously in detecting actual invasions versus false alarms. Here, the ‘all-in-one’ model was better at predicting invasions if they had, in fact, occurred. Next, I compare predictions of Bythotrephes invasions for three types of gravity models: total-flow-, production- and doubly-constrained. These models differ in the type of information required to parameterize the model. The Production-constrained model was most likely to detect actual invasions relative to false alarms, and the total-flow-constrained model was least likely to predict false positives. I also compare backcast patterns of propagule pressure for two groups of related species: one group comprising the spiny waterflea and the fishhook waterflea Cercopagis pengoi; and the other, the zebra mussel and quagga mussel Dreissena rostriformis bugensis. Differences in species\u27 life-histories may interact with various transport mechanisms to produce highly dissimilar levels of propagule pressure to inland lakes. Species with the broadest distribution had the highest propagule pressure scores. Finally, I examine the attributes of an invasion network formed by lakes invaded by spiny waterfleas connected by recreational traffic. I was interested in whether specific lakes served as ‘hubs’, and whether the network of lakes exhibited a scale-free topology. Management implications for a scale-free invasion network include a potential decrease in the overall rate of NIS spread if propagule flow from ‘hubs’ is reduced