Incorporating estimates of capture probability and river network covariance in novel habitat – abundance models: Assessing the effects of conservation stocking on catchment-scale production of juvenile Atlantic salmon (Salmo salar) from a long-term electrofishing dataset

There are increasing calls for “conservation stocking” to counter declines in Atlantic salmon (Salmo salar) populations, the assumption being that stocking can bypass population bottlenecks and increase recruitment over natural processes. However, there are too few quantitative studies with sufficient data to assess the efficacy of conservation stocking. The Girnock Burn is a unique long-term monitoring site where adult and juvenile salmon numbers have been monitored for over 50 years, including 11 years with conservation stocking. Adults were monitored at a fixed trap and juveniles were monitored by electrofishing. In stocked years, ova were incubated in surface water to reduce density-independent over-winter mortality. In eight years, eyed ova were stocked at uniform densities to reduce local density-dependence. In three years, stocking replicated natural spatial variability in ova deposition removing any potential benefits of reduced local density-dependence. Juvenile production was estimated by summing the product of reach-scale density estimates and river area obtained from a novel spatial statistical river network model that incorporated the effects of capture probability, habitat and stock level. Capture probability varied with life-stage (age 0+ fry or ≥1+ parr), electrofishing pass and day of the year, but importantly also exhibited a positive temporal trend across years. Survival from ova to fry was density-independent and higher under uniform stocking than natural spawning or simulated natural spawning. Under uniform stocking, fry densities varied smoothly with altitude, while under natural spawning and simulated natural spawning, fry exhibited a more patchy distribution. Increased fry production did not translate to increased parr production, which was strongly density-dependent. This likely reflected the inability of fry to move between stocked locations and suitable overwintering habitat, decreasing survival between fry and parr life-stages. Consequently, there was no overall benefit of stocking. The modelling framework used in this study provides a valuable approach for interpreting long-term datasets where site locations, equipment and staffing vary over time. The long-term Girnock dataset was valuable in separating management action from natural population regulation and permitting understanding of ecological processes. The study indicates that conservation stocking can be ineffective, even where implemented to best scientific standards. It is therefore recommended that a detailed understanding of local population dynamics is obtained, and a realistic appraisal of the expected benefits of stocking is undertaken, before management actions are considered.