Spatial modelling of freshwater fish in semi-arid river systems: A tool for conservation

This paper examines the feasibility of using multivariate statistics to model fish species distribution and habitat requirements for intermittent streams in semi-arid regions, many of which are coming under increasing pressure from water resource development schemes. The assessment was based on the geographical distribution of six endemic fish species in the Guadiana river, a semi-arid river system in southern Iberia. Their presence was related to 20 environmental variables linked to climate, geomorphology, riparian vegetation and location in the drainage basin. These variables were collected in the field or from topographical maps, to evaluate habitat suitability and to predict the presence of the species according to season. Multivariate logistic regression in a geographic information system (GIS) environment was performed to identify regions with high probability of occurrence for each species. The variables that best explained the occurrence of the species were the sample location in the drainage basin, the geomorphology and the riparian vegetation. The models presented have a high predictive power and can be used in monitoring and predicting temporal changes caused by human activities. This modelling approach can be used to predict the areas that need to be conserved to protect or rehabilitate the endangered species. Armed with this information, managers can formulate conservation measures to prevent further degradation of the stocks and possibly enhance the populations. Copyright (C)2002 John Wiley Sons, Ltd.