A framework to study the context-dependent impacts of marine invasions

The ecological impacts of marine invasive species vary according to the spatial and temporal scale of analysis, thereby challenging the extraction of generalities about underlying mechanisms. Here, we applied a broad impact framework that addresses this scale-dependency, to test if general drivers of impacts can be identified and quantified from marine invasion experiments. This framework explains variability in impacts according to the unique (specific) and universal (general) attributes of the (1) invasive organism, (2) resident biota, (3) resource levels, and (4) abiotic conditions. In this framework, unique and universal attributes encompass the properties that are either ecologically relevant to only a few specific invasions (e.g. a unique toxin) or to most invasions (e.g. invader density, size, age or longevity), respectively. We reviewed 88 published marine invasion impact experiments, where 18 tested for effects of universal and 11 for unique attributes of the invasive organism (63 tested for presence-absence effects, where these attributes are confounded). A meta-analysis confirmed that the species identity and density (representing a unique and universal attribute, respectively) of the invader significantly predicted impacts. These attributes should, therefore, whenever possible, be treated as separate impact-modifiers. By contrast, very few experiments have tested if universal or unique attributes of the resident biota, the resource levels or the abiotic conditions modify invasion impact. This highlights a major research gap: quantitative syntheses cannot be undertaken until more factorial experiments have manipulated the invasive species and habitat-associated drivers in concert (with treatments per test factor). In conclusion, to facilitate a broader understanding of marine invasion impacts, we advocate that universal and unique impact-components, whenever possible, are treated as separate test entities that should be examined for each of the four impact drivers. (C) 2011 Elsevier B.V. All rights reserved.