Predicting substrate size at a watershed scale to inform conservation planning for a declining salmonid species

Kyleisha J. Foote , Shawn J. Leroux, Ava J. Hart, Nick C. Murphy, Craig F Purchase

Good quality spawning habitat is critical for fish embryo development, survival, and overall population productivity. Appropriate riverbed substrate size is particularly important for riverine-spawning salmonids but the availability of suitable substrate may vary across a watershed. Predicting substrate size at watershed extents may therefore be critical to inform management and conservation of salmonid populations. Using a 30 m Digital Elevation Model (DEM), we modelled riverbed substrate size using the Shields equation at stream segments across a large watershed (~11,200 km2) in Newfoundland, Canada. We use this model to predict the distributions of substrate suitable for Atlantic salmon (Salmo salar) spawning. We confront model predictions with field data on substrate size at 149 sites across the watershed. Our model predicted that 58% of segment length contained suitable (16-35 mm) or marginal (35-64 mm) substrate for salmon spawning. At these classes our model precision was between 24 and 39%. Overall, our model predicted the correct field-measured substrate class for 42% of segments. This was bolstered by higher precision (77%) at larger substrate sizes (greater than 64 mm) possibly due to this class being over-represented in field data. Model predictions of substrate size increased alongside field measurements but generally under-estimated substrate size. Our model provides an important step toward watershed-scale assessments of potential salmon spawning habitat to help guide more efficient restoration and conservation planning. We surmise that higher resolution DEM data would allow us to fine-tune model predictions to areas more relevant to salmon spawning.