THE DEVELOPMENT OF HABITAT MODELS FOR STR EAM SALMONIDS, AND THEIR APPLICATION TO FISHERIES MANAGEMENT

The paper briefly describes the development of HABSCORE, a salmonid ha bitat assessment technique based on a series of empirical statistical models which relate salmonid abundance to observed habitat variables. Fisheries and habitat data for 602 notionally pristine sites throughou t England and Wales were used in the development of salmonid populatio n prediction models. These sites provided a total of 130 independent v ariables from which five regression models, predicting salmonid popula tion size, were produced. These models (for 0(+) salmon, > 0(+) salmon , 0(+) trout, > 0(+) [< 20 cm] trout and > 0(+) [> 20 cm] trout) expla ined between 28.7% and 46.2% of the total variance in population densi ties observed in the raw data. Partitioning of the variances within th e raw data suggested that the models account for between 45.1% and 86. 7% of the total spatial variation. Error associated with the measureme nt of the habitat variables used accounted for less than or equal to 1 .1% of the total variance in the five models. The quality of the model s is briefly discussed. The outputs of the models, and the potential m anagement applications of HABSCORE, are described.