Methods
Our samples come from archived collections of scales taken from wild Atlantic salmon returning to two distinct regions of the UK: the North East Coast drift net fishery, which samples a mixed stock originating in rivers from the North East coast of the UK55, around 55°N and 1°W, and the Dorset River Frome stock from the south coast of England, around 50.5°N and 2°W. Both sampled stocks belong to the proposed southern European population. For both archives, we determined δ13C and δ15N values of scales from approximately 10 1SW and 10 MSW fish for each available year between 1985 and 2002 (see supplementary information for a statistical justification of the sample numbers).
Scales are composite bioapatite-collagen structures. The mineralised component contains a minor inorganic carbonate component, but acid pre-treatment of scales results in biologically insignificant changes in the bulk isotopic composition56. δ13C values in the collagen component of fish scales are consistently enriched in δ13C relative to muscle values reflecting the high proportion of glycine in collagen57. Scale δ13C values are, however, linearly related to the isotopic composition of other body tissues across multiple genera, under differing diet conditions and growth rates565859, indicating a consistent relationship between scale collagen and diet δ13C values.
Scales are incremental tissues growing allometrically with body size increase. Scale collagen is deposited only during growth, and is not turned over metabolically. Occasionally, scales are resorbed, but this is easily identified, and any scales showing resorption are not used for analyses. We therefore target collagen formed during spring-summer growth. Scales were briefly (c. 2–5 minutes) soaked in de-ionised water, manually cleaned using forceps and a scalpel to remove adherents such as lipids and guanine, and dissected under a transmitted light microscope. The last summer of growth at sea (indicated by widely spaced circuli) was excised to obtain a temporally distinct sample60. In 1SW fish, the summer section from the edge of the scale was sampled, while in MSW fish the summer immediately before the final winter at sea was sampled. Because of the three dimensional nature of growth of collagen in scales60 both 1SW and, to a lesser extent, 2SW samples will include a minor contribution of collagen laid down during the return migration. This additional collagen may contribute to the relatively high within year variance in seen in δ13C values in 1SW fish. The measured values represent the isotopic composition of fish tissues averaged over the last full season of marine growth (approximately 8 months, March – October). Samples were weighed to ∼0.60mg, and the isotope ratios determined by elemental analysis isotope ratio mass spectrometry (EA-IRMS), using L-glutamic acid as an in-house calibration standard. Measurement precision assessed as 2× standard deviation of 16 replicate analyses of USGS40 glutamic acid for δ15N and δ13C is, respectively, 0.7‰ and 0.1‰. Comparisons of carbon isotope values between populations were performed using ANOVA with geographic origin, year and sea age as factors. All statistical analyses were performed using the R statistical language61.