Premium
Modelling spatial exclusion of a vulnerable native fish by introduced trout in rivers using landscape features: a new tool for conservation management
Author(s) -
Woodford Darragh J.,
Cochrane Thomas A.,
McHugh Peter A.,
McIntosh Angus R.
Publication year - 2011
Publication title -
aquatic conservation: marine and freshwater ecosystems
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.95
H-Index - 77
eISSN - 1099-0755
pISSN - 1052-7613
DOI - 10.1002/aqc.1209
Subject(s) - trout , electrofishing , brown trout , habitat , streams , predation , salmo , ecology , environmental science , salvelinus , fishery , biology , fish <actinopterygii> , computer science , computer network
Habitat variability can affect the ability of invasive predators to exclude native prey species from parts of a landscape. If habitat consistently mediates predator impacts, the outcome of predator–prey interactions could be modelled using habitat data alone. In New Zealand streams, invasive predatory trout create source–sink dynamics in populations of the native fish Galaxias vulgaris . Within trout‐invaded sink habitat, G. vulgaris are excluded from small, stable streams that occur far from galaxiid demographic sources (streams above barriers to trout). A GIS‐based spatial model using stream size and distance‐to‐source data was developed to predict where trout would exclude G. vulgaris in a river network, and to detect undiscovered trout‐free source populations. The maximum distance from source streams at which G. vulgaris could co‐occur with trout was modelled in two trout‐invaded riverscapes using a quantile limit function in GIS. The predictive ability of the model was tested using training datasets and data from two independent catchments where fish occurrence patterns were quantified from electrofishing surveys. The model correctly predicted G. vulgaris absence in the first test catchment, but failed to predict absence in the second test dataset because of an initial lack of knowledge of demographic sources and insufficient survey data. This demonstrated that G. vulgaris exclusion by trout could be predicted by the model, provided it is constructed and validated in an iterative fashion. The current model provides a useful tool both for conserving critical source populations, as well as for selecting streams for rehabilitation through trout eradication. Since species interaction models based on landscape data are likely to be generally useful in guiding conservation management, researchers should investigate whether other biotic interactions that can be consistently predicted by habitat exist in nature. Copyright © 2011 John Wiley & Sons, Ltd.