Scottish Marine and Freshwater Science Vol 6 No 12: The demography of a phenotypically mixed Atlantic salmon (Salmo salar) population as discerned for an eastern Scottish river

Discussion

The modelling approach presented here relied on extending well-established protocols for whole-river salmon population assessments (eg Chaput 2012; Crozier et al. 2003a; Potter et al. 2003) to a more detailed sub-catchment and sub-stock level, and separating density-independent marine-mortality from density-dependent processes in freshwaters (Jonsson et al. 1998). Our results provide strong evidence that sub-stock structure affected productivities and trends in abundance at the North Esk. In general, if such sub-stock dynamics are not considered, the 'overall' population dynamics may be misunderstood, which might impair the estimation of reliable Biological Reference-Points ( BRPs). Additional data (see below) are needed to parameterise better models.

Single-stock catchment models assume either: that there are no sub-components (‑stocks) with different population parameters (including mortality regimes); and/or (as an approximation) that the proportions of any sub-components are constant over time (so that any small biases are constant). However, international salmon management ( ICES 2014) reports regional populations' statuses by the stock types MSW and grilse (having different oceanic feeding zones). Furthemore, Scottish salmon status assessments have treated early MSW (conveniently separable by return-date alone) differently for over twenty-five years (Youngson 2002). Our present findings add to the view that abundance trends and population parameters (especially fecundity aspects) potentially also differ between late MSW and grilse populations at the North Esk. Failure to account for such differences could potentially have deleterious consequences for weaker within-river stocks. Given the ova-fecundity differences reported by Bacon et al. (2012) and de Eyto et al. (2015), there is now greater cause for concern that population parameter, and/or population composition, differences within and between catchments may matter appreciably.

Single-river BRPs are commonly applied to other sites because of the significant challenges involved in developing within-river stock-specific BRPs (Crozier et al. 2003b). Our analysis is the first attempt to quantify whether sub-stock differences could affect reference points, and it strongly suggested they do. Interestingly our analysis (Fig2a, Fig4a) shows that in the North Esk, the estimated single-stock trend in abundance was, perhaps fortuitously, rather similar to the trend of the early MSW component that was the managers' main focus. However it was not similar to our inferred trends for late MSW salmon or grilse. Furthermore, there is no reason to believe that such similarity between single-stock trend and early MSW trends would apply at other sites.

The model presented here is a paradigm for Atlantic salmon populations that contain phenotypically diverse sub-stocks. It was parameterised for large east-coast Scottish catchments, using practicable, intensive, large-scale monitoring data, supplemented by a simple, generalised spatial-structure for sea-age and run-time distributions within such a (Scottish) catchment. The key findings of large differences in the population dynamics between different within-river sub-stocks were robust to quite extreme variations in the marine-mortality sub-models. They showed two important differences between the putative sub-stocks. First, that trends and annual/decadal fluctuations in returning adult numbers differed appreciably between sub-stocks over time. Secondly, the relative fecundities of putative sub-stocks (as deduced from SR analyses) were strongly affected by details of the (presently somewhat unrealistic) marine-mortality models. In short, it was both desirable and possible to decompose the North Esk's Atlantic Salmon stock into sub-stocks: however, detailed examination of the consequent productivity estimates showed the overall situation was inadequately described by the current combination of data and models.