Seaweed-eating sheep show that δS evidence for marine diets can be fully masked by sea spray effects

RATIONALE
Stable sulfur isotope compositions (δ34 S values) are a useful marker of terrestrial (lower δ34 S) versus marine (higher δ34 S) diets. In coastal areas, 34 S-enriched sea spray can obscure these marine/terrestrial differences. We sought to establish whether δ34 S values of sea spray-affected terrestrial fauna can be distinguished from those of marine-feeding terrestrial fauna.


METHODS
We measured bone and dentine collagen δ34 S values, as well as stable carbon (δ13 C) and nitrogen (δ15 N) isotope compositions via continuous flow-elemental analysis/isotope ratio mass spectrometry of 21 sheep (Ovis aries) raised on < 7 km2 island (North Ronaldsay, UK) that had widely divergent access to marine (seaweed) and heavily sea spray-affected terrestrial (grass) food sources. We also analyzed the bone collagen of marine and terrestrial fauna from this island.


RESULTS
Sheep bone collagen showed well-defined trends with highly significant correlations between δ13 C and δ15 N values indicative of feeding along a continuum of fully terrestrial to fully marine diets, consistent with other modern baseline data from marine and terrestrial animals in the same area. In contrast, the δ34 S value was generally elevated for all sheep and was not significantly correlated with either δ13 C or δ15 N.


CONCLUSIONS
Our findings demonstrate that δ34 S values are poorly suited to differentiate marine and terrestrial diets in terrestrial animals in areas with pronounced sea spray effects. Care must be taken to characterize the isotopic compositions of potential food items before the δ34 S value is used as a marker of the reliance on marine protein in modern and ancient contexts.


| INTRODUCTION
Stable sulfur isotope (δ 34 S) analysis of animal tissues can provide an independent line of evidence for exploring ecological relationships involving energy pathways (i.e., diets) and mobility (i.e., origin/provenience) in contemporary and ancient ecosystems. [1][2][3] This is because δ 34 S baselines may vary between regions (due to underlying differences in the isotopic compositions of sulfur sources and cycling 4 ) as well as the fact that, for many species, sulfur's isotopic composition is thought to undergo very little fractionation as it is passed between trophic levels. [5][6][7][8][9] In addition to exploring provenience and mobility, δ 34 S may provide an indicator for reliance of foods derived from ecosystems where the primary source of sulfur originated from isotopically distinctive sources such as marine organisms (very high δ 34 S values 6,10,11 ) or sulfide-adapted plants (very low δ 34 S values [12][13][14]. Use of δ 34 S as an indicator of marine-oriented diets is, however, complicated in coastal regions by the transport and deposition of 34 S-enriched sulfur from the marine to terrestrial environments via sea spray. 15 In this context, the extent to which the sulfur cycle at the base of coastal terrestrial food webs is supplied and driven by the local endogenous (e.g., from bedrock) versus exogenous (i.e., sea spray) sulfur sources is unclear. In some cases, terrestrial fauna from coastal areas may still possess relatively low δ 34 S values. 16 This is because even when sea spray supplies enough sulfur to meet the biological requirements for local flora, it remains possible that at least some of the sulfur taken up by plants will be derived from endogenous, terrestrial sources.
In the study reported here, we investigated the extent to which 34 S-enriched sulfur contributions from sea spray can "mask" the δ 34 S signal from marine foods by comparing the δ 34 S values of bone collagen from sheep (Ovis aries) raised on a small island (<7 km 2 ) in the North Atlantic, North Ronaldsay, with a wide range of diets along a marine-to-terrestrial continuum. North Ronaldsay (Figure 1) is an ideal location to explore this issue because its underlying geology (the Rousay Flagstone formation, a lacustrian Middle Devonian deposit) would contribute an endogenous δ 34 S value that is much lower than that of sea spray. Moreover, islanders of North Ronaldsay employ a specialized sheep husbandry strategy that, in concert with stable carbon (δ 13 C) and nitrogen (δ 15 N) isotope analyses, provides a "natural laboratory" for comparing the effects of marine-terrestrial dietary variation on the δ 34 S values of herbivores.

| Sheep husbandry on North Ronaldsay
Sheep husbandry on North Ronaldsay is renowned for its distinctive reliance on naturally abundant marine macroalgae (hereafter "seaweeds") that grow in or are washed up onto the local intertidal zone (hereafter "beaches"). 17-20 A stone dyke separating rocky coastal beaches from cleared farmland in the interior provides an effective barrier, allowing sheep husbandry to be selectively contained to either interior fields or beach areas where the primary food sources are grass and seaweed, respectively (although a comparatively small amount of unenclosed grasslands is also accessible outside the dyke, mostly at the island's far eastern tip). Typically, rams are left on the outside of the dyke year-round whereas ewes are brought into the interior in the spring (ca April) for lambing. After a period of several months (ca August), lambs and ewes are moved to outside the dyke to feed on seaweed. A small number of lambs are also born outside the dyke to ewes that have not been included in the spring roundup.
Based on this husbandry regime, depending on a sheep's age and sex, we can expect wide variation in the relative proportions of marine and terrestrial foods consumed. At the extreme ends will be lambs born outside the dyke, with fully marine diets (note also that they would be born of ewes that had only had access to seaweed over the gestation period), and lambs born inside the dyke, which died shortly after being moved from fields to beaches (and thus had diets largely based on terrestrial foods). Moreover, adult males, spending most of their lives outside the dyke, will have diets that are much more heavily and consistently influenced by marine foods than adult females, which will have spent a sizable portion (ca 40%) of each year grazing in the interior fields. The North Ronaldsay sheep bone samples included specimens from animals that died at a wide spectrum of ages (based on size and dentition) as well as both sexes (based on horn presence/absence) and we therefore expected that this sample would include a large range of dietary variation along the marine-to-terrestrial continuum.

| Sample collection and interpretive framework
In order to verify this, we also conducted δ 13 C and δ 15 N analyses, which, in the region's C 3 -dominated environment, provide an effective means of establishing the relative importance of terrestrial (producing lower δ 13 C and δ 15 N values) versus marine (producing higher δ 13 C and δ 15 N values) dietary inputs. 21,22 To assess the extent to which sheep diets vary through time, we also analyzed collagen from serially sampled tooth dentine from six sheep (five from North Ronaldsay and one from Westray). We sampled mandibular third molars which are thought to form (and will therefore reflect diet) over approximately a one-year period 23 beginning early in the second year of life. 24 Table S1 (supporting information). Replicate numbers, means (for check standards), and standard deviations for calibration standards (Table S2), check standards (Table S3) and sample replicates (Table S4) are also available in the supporting information. For δ 13 C, δ 15 N and δ 34 S values, respectively, the systematic errors (u (bias) ) were ±0.09‰, ±0.14‰, and ±0.31‰, the random errors (uR (w) ) were ±0.06‰, ±0.16‰, and ±0.24‰, and the standard uncertainties were ±0.11‰, ±0.21‰, and ±0.39‰. 30

| Collagen quality control
Collagen quality control was assessed using established criteria for δ 13 C, δ 15 N and δ 34 S including elemental composition thresholds (C > 13% and N > 4.8% for evaluating δ 13 C and δ 15 N and between 0.15% and 0.35% S for evaluating mammal δ 34 S) as well as elemental ratios (C:N Atomic between 2.9 and 3.6 for evaluating δ 13 C and δ 15 N and C: S Atomic and N:S Atomic within 600 ± 300 and 300 ± 100, respectively, for mammal δ 34 S). 31-33

| RESULTS
All samples had elemental compositions falling within the collagen quality control criteria for δ 13 C, δ 15 N, and δ 34 S (Tables 1 and 2; and   Table S5, supporting information).
Faunal baseline isotopic compositions from marine and terrestrial animals provide isotopic endpoints for assessing the importance of marine and terrestrial foods in sheep diets (Table 1 and  Because these data are not primarily derived from inshore species that would mainly consume seaweed (which could have higher δ 13 C values than offshore ecosystems [34][35][36] ), we expect that bone collagen from sheep consuming a fully marine-based diet could be slightly more enriched in 13 C relative to these marine taxa (for a review of aquatic carbon sources and cycling, see Guiry 37 ) Also, because this marine baseline is primarily composed of higher trophic level carnivores, we expect that a δ 15 N endpoint for bone collagen of sheep consuming primarily seaweed will be lower than that of our faunal baseline. 38,39 North Ronaldsay sheep bone collagen showed an extremely wide range of δ 13  spectrum from marine to terrestrial diets. This is also supported by a comparison of sheep bone collagen δ 13 C and δ 15 N values, which shows that these isotopic compositions are highly correlated (Spearman's ρ = 0.86, p < 0.001) as would be expected based on both the faunal baseline and the theoretical differences between marine and terrestrial carbon and nitrogen sources and cycling processes. 22,40 Isotopic compositions from tooth dentine collagen provide a diachronic perspective of diet over a period of approximately one year during early adult life ( Figure 5 and Table 2). All teeth showed pronounced isotopic patterning for both δ 13 C and δ 15  However, given the close proximity between the collection location for this sample and the Noup Head seabird colony, a plausible natural explanation for these δ 15 N fluctuations could be that they reflect occasional feeding in pastures influenced by seabird guano (which appeared to be common in the local vicinity). 43 On the whole, isotopic patterns in North Ronaldsay dentine profiles were more variable than in the individual from Westray. This is particularly the case for δ 13 [44][45][46][47] ) and mean dentine suggest that substantial isotopic variation can occur among years.

| SUMMARY AND CONCLUSIONS
In this study, we have provided the first systematic comparison of bone collagen δ 34 S in a domestic animal with a wide range of marineand terrestrial-oriented diets living in the same sea-spray-affected environment. Results show: (1) that there is no significant correlation between δ 34 S and diet along the terrestrial-to-marine continuum and, moreover, (2) that there is no significant difference between the δ 34 S of animals with the most and least marine-oriented diets. These findings provide further support for previous studies that have explored the effects of sea spray experimentally on other tissue such as hair. 15 In particular, our data experimentally demonstrate, for the first time, that sea spray effects can fully mask the δ 34  conjunction with a well-contextualized regional baseline, as a relative indicator for human or animal provenance along coast-to-inland sea spray gradients. 62