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

Long-term Trends in Population Dynamics

Survival at Sea

An accurate assessment of marine-mortality is required to model sub-stock population dynamics of Atlantic salmon ('return-rates', sensu Chaput 2012). This paper inferred marine survivals indirectly from sub-process models of: (i) smolt-allocation between sub-stocks; and (ii) relative marine-mortalities, based on return-sizes or periods spent at sea (which are themselves correlated to size and so also loosely correlated with survival). Both these present process-models were imperfect: the sub-stock phenotypes are unlikely to be completely genetically heritable; the degrees of sub-stock mixing and assortative-matings are unknown; and the fishes' periods at sea and return-sizes are only loosely correlated with survival. But our finding that estimated sub-stock trends and parameters still differed, given any one of three widely differing variants of the marine-mortality sub-model, was persuasive evidence that the sub-stock structuring is indeed important.

Speculation about the effects of alternative marine-mortality model variants would be unconstrained and uninformative in the absence of suitable data. If observed changes in marine-mortality are either largely environmentally driven, or phenotypically plastic, then very detailed data on adult body-conditions and locations over time would be required to show this; such data are lacking and difficult to obtain. A better prospect lies in the discovery of suitable genetic markers for run-timing and sea-age. If found, and if those traits are reasonably heritable, such markers could allow heritability estimates of the phenotypic traits and, potentially the devising and parameterisation of better sub-stock models (by direct enumeration of smolt pheno-/genotypes). If the markers worked with DNA from fish-scales, such testing might well be possible from historical scale-collections which often include higher density situations than present populations.

In River ( PEA to Spawner) Survival

The in-river PEA to spawner survivals were largely derived from direct adult-count data. Survivals of grilse and late2 SW' from the North Esk's fisheries showed only small survival increases over the study period. In contrast, early2 SW' salmon, the prime focus of management attention, showed strong in-river survival increases (40% to virtually 100%,). The ameliorative fishery constraints introduced to conserve early MSW salmon may indeed have partly induced those different trends. However, despite the detailed North Esk data, it was not possible to quantitatively partition the improvements between different management actions and other indirect process changes that were coincident (eg environmental or economic).

Reproduction

At high spawner densities and at a whole-river level, the total North Esk productivity was around 200,000 smolts (Fig.2.d), and equated to some 20, 35 and 18 smolts/spawner (for early2 SW', late2 SW' and grilse respectively ). However, objective quantification of the smolt productivities per spawner at low population levels showed that the present over-simple reproductive model lead to unlikely sub-stock production estimates that varied too widely from one another. Proper incorporation of the true breeding structure (details of river-areas used by sub-stocks; sex-ratios; and assortative-mating patterns) and trait heritabilities (sea-age, run-timing and associated adult return-sizes), along with any consequent marine-mortality differences, would presumably do much to resolve these discrepancies (see Gurney et al. 2012). Lack of data currently prevents this.

Fecundities and Fecundity Anomalies

The maximum lifetime reproductive approach to productivity (Myers et al. 1999; Gibson et al. 2013) is powerful metric for comparing productivities across populations. Unfortunately lack of data precluded its use here.

The two most plausible candidate explanations for our observed productivity differences were either differing sex-ratios between sub-stocks (but sex-differences are far from able to bridge the gap - see Supplementary Material, part I, for an explanation of why) or else 'reproductive exchange' between the sub-stocks.

Reproductive exchange is most likely between grilse and late MSW salmon, which are thought to be predominantly co-located in the lowlands (Bacon et al. 2012 and references therein). Grilse X grilse adult phenotype matings might produce not just grilse offspring but also a nett surplus of late MSW offspring ( i.e. they produce more MSW offspring phenotypes than late MSW X late MSW matings produce grilse offspring phenotypes); as also might grilse X late MSW matings. Such explanation is entirely plausible. There is good evidence that the sea-age and within-year run-timing are heritable (Hansen and Jonsson 1991; Stewart et al. 2002; Vaha et al. 2010), as is development rate in both Atlantic salmon (Gjerde 1984) and other salmonids (Hankin et al. 1993). The likely degrees of heritability (eg Mousseau and Roff, 1987; Carlson and Seamons, 2008), combined with the sizes of the different North Esk sub-stocks (c. 6,000 grilse, 1,000 late MSW), could readily make good our estimated fecundity discrepancy. Moreover, a sound numerical basis for the necessary mechanisms has already been provided (Gurney et al. 2012), which includes an evolutionary argument, the necessary population dynamics and has findings that are robust to a variety of inheritance mechanisms (trait-dominances, for the simplifying one locus, two allele case examined). Its complete genetic determination (as discussed by Gurney et al. 2012) would be an oversimplification, as environmental effects are also known to play a role in sea-age determination (Jonsson et al. 2013). But so too does genetics: reality is likely to be somewhere between the extremes (of completely genetic or environmental determination). Unfortunately, the genetic mechanisms and quantitative heritability values, necessary for an improved understanding of processes, are presently unknown (see Gurney et al. 2012).

The idea that these three phenotypic sub-stocks are neither genetically fully mixed nor fully independent is, of course, not new (Taggart et al. 2001;Webb et al. 2007), although the corresponding genetic structuring has only recently been demonstrated (Vaha et al. 2010). The potentially important consequences for salmon management have often been ignored, given the general lack of appropriate information.

Managing a river like the North Esk as if the three phenotypic sub-stocks were fully independent will be slightly misleading (see next paragraph), at least until better data allow one to understand both the inter-breeding and the heritability consequences. However, it could also be argued that managing them as if they were a single stock would be much more misleading, given the very substantial differences in life history trends and parameters demonstrated here. Such single-stock approaches still occur: indeed a recent single-stock analysis (Massiot-Granier et al. 2014) uses much the same (but not all of) the North Esk data underlying the present analysis; but, as it has different assumptions, structure, data, data-corrections (no grilse-error adjustment) and emphases, and is based on so different a paradigm, it axiomatically reaches different conclusions.

Until the genetic interactions of the phenotypic sub-stocks are adequately understood, we note that treating a small late MSW sub-stock as if it was fully independent (as here) is probably conservatively prudent (one would treat a small independent (isolated) sub-stock more cautiously than one that was insulated from deleterious influences by the contribution of offspring arising from the much larger grilse sub-stock (should this latter hypothesis be correct)).