Improving Salmon Management with Automated Habitat Characterization and Functional Habitat Connectivity

English Abstract:

Currently, Atlantic salmon conservation requirements and management targets established by the Quebec Wildlife Service rely on a model linking a physical habitat quality index and the number of migrating smolts monitored for several rivers. The production capacity of other rivers is then extrapolated using estimates of habitat quality based on channel width, substrate and morphohydraulic reach types (pools, riffles, meanders, channels, rapids, lakes). However, accumulating evidence suggest that besides habitat quality, the spatial arrangement of functional habitats patches (i.e. specific to life stages (fry, parr, adult) or function (feeding, shelter, spawning)) affects river production capacity.

Characterizing river habitat and keeping data up to date over decades represent a considerable challenge for provincial wildlife managers, considering the large extent to cover, including thousands of river kilometers remote areas in the North. In this context, increasingly available high resolution areal remote sensing imagery appears as a cost effective opportunity providing continuous data over large extents.

In this study, we developed an automated method to delineate reach types and functional habitat patches using hyperspectral areal remote sensing imagery and digital elevation model analyses. The classification of homogeneous reaches was based on channel width, slope, radius of curvature and relative depth classes. Then, we modeled the connectivity between functional habitats using a least cost approach integrating life-stage specific fish mobility. Our results present quantitative estimates of habitat probability of use based on river habitat spatial arrangement and illustrate the management functionality of the method in improving the estimation of river production capacity.