PMC:3216509 / 8352-19200 JSONTXT

{"project":"2_test","denotations":[{"id":"22355540-20701633-135776219","span":{"begin":6024,"end":6026},"obj":"20701633"}],"text":"Results\n\nThe effect of trophic level on carbon isotope values\nThe potential influence of trophic level on measured δ13C values was tested by comparison between carbon and nitrogen isotope compositions in scale tissue from salmon populations returning to the UK River Frome (RF) and North East Coast (NEC). The isotopic composition of carbon and nitrogen in fish tissues is controlled to different degrees by variations in isotopic composition at the base of the food web and by trophic discrimination. Nitrogen is more strongly fractionated than carbon during dietary assimilation, with a mean trophic increase of c.3‰ compared to c.1‰ in carbon262728. In our sampled scales, δ15N values correlate positively with size (RF salmon: n = 231, R2 = 0.41, p\u003c0.001; NEC salmon: n = 162, R2 = 0.17, p\u003c0.001), reflecting changes in trophic level (Fig. 1). Fish returning after more than one winter at sea clearly fed at higher trophic levels than the smaller 1SW returning fish. δ13C values also differ between 1SW and MSW returning fish, but show no positive relationship with either size (Fig. 1) or with δ15N values (Fig. 2), indicating that differences in the carbon isotope composition of scales cannot be explained by mass or trophic level. This strongly indicates that the carbon isotope composition of salmon collagen is dominated by variations in photosynthesis-related fractionation at the base of the food chain, and that salmon feeding in a common marine area are expected to have similar δ13C values45.\nDespite the large variation in plankton δ13C values across the North Atlantic Ocean142348, salmon returning to the River Frome maintain consistently different δ13C values to those returning to the NE Coast throughout the 18 year sampling period (ANOVA, with geographic origin as sole co-factor; n = 523, df = 1, F = 116.9, p\u003c0.001). In addition, year and sea age both significantly influence δ13C values (Year, n = 523, df = 17, F = 9.1, p\u003c0.001; sea age df = 1, F = 16, p\u003c0.001), with a significant interaction between all factors. These data strongly suggest that, in each sampled year, returning salmon from different natal origins fed in distinct locations. Stable isotope evidence for stock separation in feeding grounds is further explored in the supplementary information.\n\nLocating marine feeding grounds from stable isotope time series\nTo determine the likely location of open ocean feeding areas for our test salmon populations, we assessed the temporal covariance between scale δ13C values and SST in each one-degree grid square between 45–75°N latitude and 65°W–20°E longitude for the period 1985 to 20024950. Time series of tissue isotopes and SST are autocorrelated, violating the assumptions of serial independence demanded by most classical inference tests51. Autocorrelated time series are frequently de-trended to remove underlying low-frequency variations, however we are primarily interested in low frequency variations that are common to both time series, thus de-trending would be counter-productive51. Instead, the influence of high frequency fluctuations in the time series was reduced by applying LOESS smoothing with a span of 0.5 and polynomial order 2 (Fig. 3). Low frequency variations in the smoothed time series were correlated, and the influence of autocorrelation on significance tests was accounted for by adjusting the effective degrees of freedom for each time series pair using the modified Chelton method51. Areas with the highest correspondence between temporal variations in SST and measured δ13C values are suggested as the most likely marine feeding areas during deposition of scale collagen (Fig. 4).\nThe correlation approach taken here is based on the LOESS smoothed fit, and is sensitive to within-year variance, i.e. the assumption that all fish sampled return from the same location at sea. We tested the effect of within-year variance in δ13C values by removing all years where within-year δ13C standard deviations exceeded 5% of the mean value for that year. Two years were removed from the River Frome 1SW data, one from the River Frome MSW data, four from the NE Coast 1SW data and three from the NE Coast MSW data. Removing years of high variance did not have a dramatic effect on the locations of the proposed feeding grounds, but increased the strength of correlations.\nCorrelations between time series of scale δ13C values and SST vary spatially across the possible range for Atlantic salmon, reaching p values below 0.02 in MSW fish from both populations, with significant covariance (p values lower than 0.05) seen for all populations and sea ages. The spatial distribution of correlation coefficients is highly structured, and provides estimates of location that are consistent with known salmon feeding grounds and migration patterns3538475253. The largest contiguous areas of highest correlations, and thus most likely feeding areas, for both the 1SW and MSW components of the NE Coast mixed stock are found in the Norwegian Sea. Correlations for the 1SW component of the NE Coast stock are relatively low, perhaps reflecting a greater range of feeding locations within this population, but a region of high correlation is located within the suspected feeding ground north of the Faroe Islands. Two relatively small areas of high correlation are seen in the Grand Banks and Davis Strait areas. Based on information from tagging and likely swimming capabilities of 1SW returning fish415253, these can be identified as spurious correlations, and as with all such studies, the results of isotopic analyses should be viewed in the context of other available data to aid interpretation. Tissue isotope records for the MSW component of the NE Coast stock are strongly correlated with SST in the Norwegian Sea. These results are highly consistent with tag-recapture and return rate datasets3541475253, and suggest that salmon originating from the North East Coast of the UK generally follow the easterly branch of the North Atlantic current into the Norwegian Sea3747. Our results suggest that fish from the River Frome, in contrast, feed in more westerly regions, with 1SW returning fish occupying an area centred around the Faroe Islands and east Iceland (Fig. 4). The isotopic composition of MSW fish returning to the River Frome strongly implies marine feeding around the Icelandic shelf. Tagging databases show that southern European origin fish are caught more frequently in the west Greenland than the Faroese fisheries53, and limited fishing surveys suggest a salmon feeding ground east of Greenland, possibly in the Irminger Sea4653, both indicating an overall westerly migration route for salmon originating from southern European rivers. Our results support this, suggesting that fish originating from the River Frome follow the western branch of the North Atlantic current. The confidence associated with any location identified through correlation analyses will depend on the length of the time series and the degree of variability in both SST and measured δ13C values. The strong match between isotope derived feeding grounds, and known feeding areas strongly suggests that in this case the 18 year archive is sufficient to identity feeding areas at least to the same resolution as is possible through traditional tagging and capture fisheries.\nLarge multi-year variations in δ13C scale values are seen in the NE Coast salmon (Fig. 3), suggesting that their feeding areas are subject to higher inter-annual variability in oceanic conditions than those of the River Frome fish. This implies that variations in ocean climate have greater potential to influence growth or return rates for the NE Coast origin fish than for the Frome stock. A significant negative relationship has indeed been found between the numbers of fish returning to rivers in north east Scotland and SST in the Norwegian Sea35. Differences in marine feeding location may thus underpin the variation in relationships between growth indices, return rates and ocean climate indices observed between salmon populations3538.\nWhere autocorrelation adjusted correlations are significant at the 95% level, variations in SST explain between 43 and 97% of the observed variation in δ13C scale values. The unexplained variation may be attributed to a combination of variations in the isotopic composition of salmon diet (e.g. variations in trophic level, trophic fractionation, and plankton community composition), violations of the assumption of site fidelity within and between years, and possible contributions of later-formed collagen. The spatially constrained nature of the areas of high correlation, however, demonstrates that unexplained sources of isotopic variation are relatively minor compared to the effect of SST.\n\nImplications for models of salmon migration\nOur hypotheses of feeding ground location proposed from isotopic data are consistent with previous salmon migration models, In both the North East Coast and River Frome populations, proposed feeding areas for fish successfully returning after one winter at sea are closer to the river of origin than those proposed for MSW fish, and logical migration routes following the prevailing currents connect the proposed 1SW and MSW fish feeding areas. Our interpretations are based exclusively on fish that returned successfully to the home waters, and therefore we cannot rule out the possibility that fish from individual river stocks feed in a wider range of regions, but with differential rates of return. Tagged fish of UK origin (including one fish originating in the River Frome) have been recovered in the west Greenland fishery, but our isotopic data suggest that this is either a rare occurrence, or that River Frome origin fish feeding off west Greenland are relatively unlikely to return successfully.\nThe similar temporal trends, and thus relatively close geographic association between feeding grounds for 1SW and MSW fish for both test populations imply that stocks are spatially differentiated during their early outward migration. Hypotheses regarding stock origin within marine feeding grounds may be tested by genetically identifying the natal river origin of salmon caught in these feeding areas. Our interpretations are not biased by historic fishery areas, and identify a high degree of spatial segregation in marine feeding areas between regional populations and potentially, individual river stocks. This is similar to the strong coherence in regional populations noted for Pacific salmonids54, and supported by long term temporal records of catch and return rates in Norwegian and Scottish salmon38. Conservation of dwindling salmon stocks may require identification and management of open seas feeding grounds on a region- or river stock-specific basis, reflecting the different ocean conditions experienced by each stock."}