AimEven successful invaders are abundant only in a fraction of locales they inhabit. One of the main challenges in invasion ecology is explaining processes that drive these patterns. We investigated recruitment of a globally invasive fish, common carp (Cyprinus carpio), across three ecoregions to determine the role of environmental characteristics, predatory communities and propagule pressure on the invasion process at coarse and fine spatial scales.LocationLakes across Northern Forest, Temperate Forest and Great Plains ecoregions of North America.Methods
We used data from 567 lakes to model presence or absence of carp recruitment using environmental conditions (lake clarity, area, maximum depth), native predatory fishes (micropredators, mesopredators, large predators) and propagule pressure (abundance of adult carp). We formed a set of alternative models and evaluated their support using an information theoretic approach. Once most supported models were identified, we used classification tree to determine how variables included in these models interacted to affect carp recruitment. Finally, we conducted a field experiment to test the predictions of the classification tree analysis.ResultsCarp recruitment was strongly regulated by processes associated with water clarity, which appeared to function as a broad-scale ecological filter. Carp were unlikely to recruit in clear, oligotrophic lakes (Secchi depth > 2 m) despite the presence of adults in many such systems. Recruitment was more likely to occur in regions with turbid lakes, but abundant micropredators could inhibit it there.Main conclusionsCarp recruitment and invasions across large geographic areas are attributable to a two-layer ecological filter with lake clarity/productivity acting as a coarse-scale filter and micropredators acting as a fine-scale filter. This two-layer filter might explain the complex patterns of carp invasions among and within different ecoregions. Ecological filters may also explain the success of other aquatic invaders that show similarly patchy spatial distribution patterns.