Distribution and habitat use by juvenile Atlantic salmon (Salmo salar) at multiple spatial scales, and implications for habitat modelling and fish-habitat management

Implicit assumptions of micro-habitat models are that (1) habitat limits population levels and small-scale information on habitat selection behaviours of individuals can be used to manage populations at large spatio-temporal scales (scale-up); (2) the single or few measurement scales used in habitat models are appropriate for identifying important habitats; and (3) "better" habitats are characterised by a higher density or frequency-of-use, i.e. density can be used as an indicator of habitat quality.; (1) Based on scope- and rate-diagrams from field-data and theoretical scenarios of movement and mortality, I concluded that salmonid habitat models operate in the context of processes that my not be important to the problems we would like to address. I suggested survey designs that allow problem associated with scale-up to be overcome.; (2) I evaluated distributions of juvenile Atlantic salmon (Salmo salar) over a range of spatial scales based on a stream-tank study (spatial scales 1 cm to 3 in) and field data (spatial scales I cm to 15 m), to determine whether patchiness of fish distributions or associations with depth, water velocity and substrate depended on spatial scale, to determine scales most appropriate to habitat models, mid to compare multi-scale versus single-scale habitat modelling approaches. Results indicated associations with conspecifics, substrate, water velocity and depth changed with spatial scale and direction relative to water flow. Associations were most different from random at small spatial scales (ambit radius < 50 cm). Both studies indicated that single- and multi-scale habitat selection models were equally able to describe fish densities at small spatial scales (ambit radius < 4 m). The field-based study indicated that single- and multi-scale models often failed to describe fish densities at scales larger than used in the model (scale-up).; (3) I studied density-dependent habitat use by Atlantic salmon parr based on experimental riverine enclosures and field data. Results from the experimental study indicated that habitat use changed with population density. Results from the field-based study were less clear with some of the results suggesting density-dependent distribution processes. I concluded that habitat selection by salmon parr was density-dependent and highly variable Changes in habitat use with density were most likely due to small-scale spacing behaviour or territoriality.; I concluded that quantitative multi-scale approaches are important to habitat modelling, identified important research questions, presented some novel techniques for scaling analyses and made suggestions to improve habitat modelling and resource management.