Skin samples were collected from six species of stranded dead dolphins (Atlantic white-sided; white-beaked; short-beaked common; striped; Risso’s; and bottlenose) found on the Scottish coastline between 2015 and 2021. Bottlenose dolphins were assigned a “local” or “non-local” classification based on their presence in photo identification databases for Great Britain and Ireland. Associated stranding event data are reported in Table 1 . Skin samples are routinely collected during post-mortem examinations conducted by the Scottish Marine Animal Stranding Scheme (IJsseldijk et al., 2019; Davison and ten Doeschate, 2020) and stored at –20°C in dedicated SMASS and National Museums Scotland (NMS) tissue archives until preparation for analysis. Calves and very young animals primarily reliant on nursing for nutrition were excluded from the analysis based on their body length measurements (Jefferson et al., 1993; Kinze et al., 1997; Kastelein et al., 2003; Jefferson et al., 2008; Meissner et al., 2012; Peters et al., 2020) ( Supplementary Material Table 1 ). Therefore, the animals included in this study are a combination of both adults and nutritionally independent juveniles. Stranding events were selected from the north, east, and west coasts of Scotland, including the Orkney and Shetland Isles. To control for degradation effects, specimens were selected based on carcass condition at time of tissue collection. The majority of samples were classified as 2a to 2b: “freshly dead” to “slight decomposition” ( Table 1 ) (Kuiken, 1991; Payo-Payo et al., 2013; IJsseldijk et al., 2019).

Lipids have a strong influence on tissue δ ¹³C (DeNiro and Epstein, 1977; Post et al., 2007). Lipids are ¹³C-depleted relative to protein and should be removed to reduce variability caused by differing lipid content among individuals and species (McConnaughey and McRoy, 1979; Post et al., 2007). While dual analysis of samples is recommended, this can be labour intensive and cost-prohibitive with large sample sets. One aliquot of each sample (n = 57) underwent no lipid extraction to avoid the deleterious effect of chemical extraction on collagen amino acids and its impact onresultant δ ¹⁵N (Smith et al., 2020). A subset of dolphin skin samples (n = 37) was separated into two aliquots, where one aliquot was chemically lipid extracted to accurately calculate skin lipid content and account for its impact on the proteinaceous component of the remaining non-lipid extracted samples (McConnaughey and McRoy, 1979; Post et al., 2007). Following a modified Bligh and Dyer (1959) method, the lipid extracted aliquot underwent three rinses of 30 minutes each in 10 mL of 2:1 chloroform:methanol, prior to drying at 60°C. Dolphin skin samples were desiccated at 60°C, powdered, and weighed into pressed-tin capsules.

The following equation, adapted to this dataset and following Post et al. (2007), was used to correct the remaining non-lipid extracted (NLE) dolphin δ ¹³C values ( Figure 1 and Supplementary Material Table 2 ):

where α is the intercept of regression between Δ ¹³C_LE-NLE and C:N_NLE, and β is the slope ( Supplementary Material Table 2 ):

The carbon and nitrogen stable isotope compositions (δ ¹³C and δ ¹⁵N) of dolphin skin (n = 50) were measured using an Elementar Pyrocube elemental analyser (EA), coupled to a ThermoScientific™ Delta Plus XP isotope mass spectrometer via a ConFlo IV system, using helium as the carrier gas at the National Environmental Isotope Facility (NIEF) isotope ecology laboratory. Additional (n = 7) bottlenose dolphin samples were analysed using a Costech elemental analyser coupled to a ThermoScientific™ Delta V isotope ratio mass spectrometer via a ConFlo IV system, using helium as the carrier gas at the Laboratory for Stable Isotope Science (LSIS) at the University of Western Ontario. Sample duplicates were included for every ten samples. All isotope results are reported in δ-notation in per mil (‰) relative to international standards calibrated to VPDB and AIR, respectively.

At the NEIF laboratory, carbon and nitrogen stable isotope compositions were calibrated using Fluka gel (porcine gelatine; n = 29; 1σ standard deviation [SD] measured δ ¹³C = 0.15 ‰, accepted δ ¹³C = –20.10 ‰ and 1σ SD δ ¹⁵N = 0.14 ‰, accepted δ ¹⁵N = +5.73 ‰) and USGS40 (L-glutamic acid; n = 4; 1σ SD δ ¹³C = 0.2 ‰, accepted δ ¹³C = –26.39 ± 0.04 ‰ and 1σ SD δ ¹⁵N = 0.2 ‰, accepted δ ¹⁵N = –4.52 ± 0.06 ‰). Fluka gel, Alagel (alanine and gelatine; n = 11; 1σ SD δ ¹³C = 0.09 ‰, accepted δ ¹³C = – 8.98 ‰ and 1σ SD δ ¹⁵N = 0.12 ‰, accepted δ ¹⁵N = +2.29 ‰), and Glygel (glycine and gelatine; n = 11; 1σ SD δ ¹³C = 0.07 ‰, accepted δ ¹³C = –38.47 ‰ and 1σ SD δ ¹⁵N = 0.11 ‰, accepted δ ¹⁵N = +23.12 ‰) were used to monitor instrument accuracy, precision, and drift. Fluka gel of different weights were used to monitor instrument linearity. Standards were repeated every ten samples. Analytical error was <0.2 ‰ for both δ ¹³C and δ ¹⁵N.

At LSIS, standards were analysed at the start and end of each analytical session and after every five samples; no instrumental drift was detected in the analytical sessions. The carbon and nitrogen stable isotope compositions were calibrated using USGS40 (L-glutamic acid; n = 4; 1σ SD δ ¹³C = 0.04 ‰, accepted δ ¹³C = –26.39 ± 0.04 ‰; 1σ SD δ ¹⁵N = 0.04 ‰, accepted δ ¹⁵N = –4.52 ± 0.06 ‰) and USGS41a (L-glutamic acid; n = 5; 1σ SD δ ¹³C = 0.04 ‰, accepted δ ¹³C = +36.55 ± 0.08 ‰; 1σ SD δ ¹⁵N = 0.22 ‰, accepted δ ¹⁵N = +47.55 ± 0.15 ‰). The accuracy of the calibration curve was tested using the LSIS internal standard (keratin, MP Biomedicals Inc., Cat No. 90211, Lot No. 9966H; n = 20) (measured δ ¹³C = –24.09 ± 0.05 ‰ [1σ SD]; measured δ ¹⁵N = +6.45 ± 0.13 ‰ [1σ SD]; n = 20 for both), which compared well with its accepted values (δ ¹³C = –24.05 ‰; δ ¹⁵N = +6.40 ‰; n = 1999 for both). Accuracy was also assessed using P. Szpak’s SRM-14 (polar bear bone collagen; measured δ ¹³C = –13.75 ± 0.03 ‰; measured δ ¹⁵N = +21.65 ± 0.06 ‰; n = 2 for both), which compared well with its accepted values (δ ¹³C = –13.63 ± 0.09 ‰; accepted δ ¹⁵N = +21.62 ± 0.28 ‰; n = 794 for both). Sample duplicate analyses differed by an average of 0.08 ‰ for δ ¹³C and 0.20 ‰ for δ ¹⁵N (n = 7 for both).

Data analyses were performed using R version 4.2.1 (R Core Team, 2022). Dolphin data distribution was assessed using a Shapiro-Wilk normality test. A PERMANOVA was used to identify differences in mean δ ¹³C and δ ¹⁵N among species. A post hoc pair-wise comparison (with Bonferroni correction to produce adjusted significance levels) was used to identify which species differed. Welch’s Two Sample t-test (assuming unequal variance) was used to assess intraspecific differences between local and non-local bottlenose dolphins. Dolphin isotopic niches, as represented by skin carbon and nitrogen stable isotope compositions (lipid-corrected δ ¹³C and δ ¹⁵N), were evaluated using the package SIBER (Stable Isotope Bayesian Ellipses in R [Jackson et al., 2011]). SIBER creates isotopic niche models of consumers using Bayesian multivariate normal distributions, calculating core (ellipses constraining 40% of data per species - Standard Ellipse Area corrected for small sample size [SEAc] and Bayesian Standard Ellipse Area [SEA_B]) and total (convex hull containing 100% of data per species - Total Area [TA]) isotopic niches, in addition to interspecific core niche overlap.

Northeast Atlantic fish, cephalopod, and crustacean muscle δ ¹³C and δ ¹⁵N (n = 1964 specimens) were compiled to provide additional context for interpreting dolphin diet (Das et al., 2003; Schaal et al., 2010; Chouvelon et al., 2012; Varela et al., 2013; Jennings and Cogan, 2015; Louzao et al., 2017; Madgett et al., 2019; Olivar et al., 2019; Baltadakis et al., 2020) ( Figure 2 , Table 2 ; Figure S2 and Supplementary Material Table 3 ). Prey muscle δ ¹³C_cor has been normalised for lipid content following Post et al. (2007), and Suess Effect-corrected (0.015 ‰ per year) to 2021 (Keeling, 1979; Sonnerup et al., 1999). Information on prey quality (determined by energy density in kJ/g^-1) are also presented ( Table 2 ) (Lawson et al., 1998; Andersen, 1999; Pedersen and Hislop, 2001; Spitz et al., 2010a).

Collectively, dolphin skin δ ¹³C ranged from –19.4 to –16.7 ‰ and δ ¹⁵N ranged from +10.3 to +14.5 ‰ ( Tables 1 , 3 ). White-beaked dolphin (n = 7) had the highest measured δ ¹³C and δ ¹⁵N, with a mean of –17.3 ± 0.4 ‰ and +13.4 ± 0.8 ‰, respectively. Atlantic white-sided dolphin (n = 5) had the lowest measured δ ¹³C with a mean of –18.7 ± 0.5 ‰, whereas striped dolphin (n = 10) had the lowest δ ¹⁵N, with a mean of +10.7 ± 0.3 ‰. The data were checked for normality with a Shapiro-Wilk test (δ ¹³C W = 0.96242, p = 0.07395; δ ¹⁵N W = 0.97528, p = 0.2922). Interspecific variation was identified among dolphin species (PERMANOVA: F = 97.005, p = 0.001) ( Supplementary Material Table 4 ). Pair-wise comparison results with Bonferroni correction are reported in Supplementary Material Table 5 . Mean δ ¹³C and δ ¹⁵N of warm-water adapted species (short-beaked common and striped dolphin) were significantly different from one another (adjusted p-value = 0.015). Likewise, mean δ ¹³C and δ ¹⁵N of cold-water adapted species (white-beaked and Atlantic white-sided dolphin) were also significantly different (adjusted p-value = 0.03). Striped dolphin mean δ ¹³C and δ ¹⁵N was significantly different from both white-beaked and Atlantic white-sided dolphin (adjusted p-value = 0.015), whereas common dolphin was not significantly different from either cold-water adapted species due to isotopic overlap.

Intraspecific isotopic variation is driven by a number of factors. Within this strandings dataset, there was a higher proportion of male animals for all species. Stranding season also varied by species, where Atlantic white-sided, white-beaked, and bottlenose dolphins stranded most during the summer and autumn months (peak July-September), while striped, short-beaked common, and Risso’s dolphins stranded primarily during the fall and winter months (peak November-March) ( Figure 1 , Table 1 ).

Within this dataset, no significant differences (Welch’s Two Sample t-test, short-beaked common dolphin: δ ¹³C t = 1.45, df = 15, p = 0.167, δ ¹⁵N t = -0.16, df = 16, p = 0.875; bottlenose dolphin: δ ¹³C t = 0.16, df = 8, p = 0.873, δ ¹⁵N t = -0.44, df = 7, p = 0.671) were found between the δ ¹³C and δ ¹⁵N of male and female animals for species where n >10. No significant difference (Welch’s Two Sample t-test; t = -0.28, df = 16, p = 0.783; t = -0.65, df = 16, p = 0.522) was observed between δ ¹³C and δ ¹⁵N of short-beaked common dolphin stranded in the Spring-Summer versus the Fall-Winter periods.

Northeast Atlantic dolphin species stranded in Scottish waters possessed distinct isotopic niches ( Figure 2 ; Supplementary Material Figure 3 ). Short-beaked common dolphin had the largest total isotopic niche (TA), followed by bottlenose dolphin. Risso’s dolphin had the smallest TA, followed by striped dolphin. Bottlenose dolphin had the largest core isotopic niche (SEA_C and SEA_B), while striped dolphin had the smallest. Non-local bottlenose dolphins had significantly lower δ ¹³C than known local animals (Welch’s Two Sample t-test; t = 9.68, df = 5.07, p < 0.001) ( Figure 3 ). Five dolphin species displayed some degree of interspecific core niche (SEAc) overlap ( Table 4 , Figure 2 ). Short-beaked common dolphin SEAc overlapped with every species, except striped dolphin. Short-beaked common SEAc overlapped with bottlenose dolphin by 51%, with white-beaked dolphin by 30%, and with Atlantic white-sided dolphin by 7%. Striped dolphin was the only species with no interspecific core niche overlap.

Dolphin species isotopic niche confers information about dietary niche, relative trophic level, and habitat use ( Figure 2 ). The two cold-water adapted species (Atlantic white-sided and white-beaked dolphin) have non-overlapping TA and SEAc ( Figure 2 , Table 3 ; Supplementary Material Figure 3 ). White-beaked dolphins target moderate quality, high trophic level prey, while Atlantic white-sided dolphins target high quality, low trophic level prey. Additionally, there is minimal overlap between TA and no overlap among SEAc of warm-water adapted short-beaked common and striped dolphins.

Short-beaked common and striped dolphin core isotopic niches do not overlap in the Northeast Atlantic (Mèndez-Fernandez et al., 2012; this study). The overlap in δ ¹³C, but separation in δ ¹⁵N in Scottish waters may indicate that they target similar habitat and species, but feed at different trophic levels. Common dolphins may be feeding on many of the same high quality prey species (e.g., Myctophidae, herring, mackerel) as striped dolphins, but simply targeting larger individuals. In contrast, short-beaked common and striped dolphin populations in the north- and southwest Mediterranean Sea exhibit a high degree of isotopic overlap (Giménez et al., 2017; Borrell et al., 2021). However, spatial segregation (deep water versus coastal water use) was observed between the two species, which allowed them to co-exist while exploiting adjacent habitats to avoid competition for prey (Giménez et al., 2017). This spatial and trophic partitioning presumably decreases competition between cetacean species evolved to occupy similar thermal ranges (see also MacKenzie et al., 2022).

Dolphin isotopic niche size and range are correlated with species foraging strategy. Large TA and SEAc typically indicate a consumer with a generalist diet and a high degree of dietary plasticity, whereas small TA and SEAc indicates specialist feeding with minimal plasticity. Striped, white-beaked, and Risso’s dolphins possess the smallest SEAc, indicating low dietary plasticity. Risso’s dolphins’ small SEAc and lack of interspecific isotopic overlap ( Table 3 ) highlight how their specialisation in cephalopod prey help them avoid competition with the other predominantly piscivorous dolphin species (Pusineri et al., 2007; Meynier et al., 2008; Luna et al., 2022). Likewise, striped dolphin SEAc does not overlap with any other species, indicating their specialisation in small, low trophic level (yet high energy density) mesopelagic and pelagic prey. Short-beaked common dolphins are generalists, and their extensive SEAc overlap with bottlenose, white-beaked, and Atlantic white-sided dolphin is likely driven by interspecific consumption of Gadiformes and high energy density pelagic schooling fish. The range and abundance of prey species (notably pelagic schooling fish and cod) in the North Atlantic are highly reactive to ocean warming and nutrient availability (Rose, 2005; van der Kooij et al., 2016; Olafsdottir et al., 2019), with knockdown effects on predator distribution. The high degree of dietary plasticity and increasing abundance of warm-water adapted short-beaked common dolphins in Scottish waters could create future competition for prey among cold-water adapted white-beaked and Atlantic white-sided dolphins. The white-beaked dolphin is considered a food specialist (Jansen et al., 2010), where increasing dietary overlap may pose future challenges as it faces contracting and displaced habitat.

Based on combined isotopic niche and stomach content records, medium sized Gadiformes and shoaling pelagic fish are the most highly consumed resources among the six Northeast Atlantic dolphin species, eliciting the highest degree of potential interspecific dietary overlap. White-beaked and Atlantic white-sided dolphin preferred prey are also economically important species to United Kingdom fisheries, where pelagic fish (e.g., herring and mackerel) followed by demersal fish (e.g., Gadidae species such as cod and haddock) are the most frequently landed species by UK fisheries (2008-2018) (Elliott and Holden, 2018). There is heavy commercial fishing in Scotland, with a strong impact on regional prey stocks (sometimes resulting in stock crashes), where the highest takes come from areas with the highest observed abundance of white-beaked and Atlantic white-sided dolphins (west of Scotland and the northern North Sea) (Elliott and Holden, 2018). As such, competition for prey from both ecological and anthropogenic sources should be considered when assessing cumulative stressors acting on cold-water adapted dolphin populations.

Isotopic variation exists within the data of each analysed dolphin species. This expected variation is driven by a variety of regional, behavioural, and ontogenetic factors. While no significant difference in δ ¹³C and δ ¹⁵N was observed based on sex or stranding season for short-beaked common and bottlenose dolphins (where n >10), the small sample size of this exploratory study limits further in-depth analyses of other dolphin species. Strandings in cold-water and warm-water adapted dolphin species peaked during the warmer and colder months, respectively ( Figure 1 ). Water temperature significantly impacts dolphin stress response (Houser et al., 2011). The interaction between dolphin thermal preference and thermal stress may play a role in their overall health, prey choice, and stranding incidence in Scottish waters. Future work with an expanded sample set is required to address intraspecific differences and the impact of sex, age class, stranding season, and prey quality in regional populations.

Most intraspecific isotopic variation is likely due to seasonal differences between cetacean stranding events and seasonal changes in diet or prey δ ¹³C and δ ¹⁵N (for example, Troina et al., 2020a; Troina et al., 2020b). Dolphin habitat choice and foraging patterns can vary with seasonal changes in prey abundance, body size, and quality (McCluskey et al., 2016). Male-female pod segregation and sex-specific caloric requirements and movement patterns during mating and calving seasons may also produce different diets (Canning et al., 2008). Species that experience large-scale seasonal mobility (as documented in Atlantic white-sided dolphins) or a high degree of dietary plasticity (for example, short-beaked common and bottlenose dolphins) consume a wider variety of prey with differing isotopic compositions (Northridge et al., 1997; Santos et al., 2001; Pusineri et al., 2007; Meynier et al., 2008). Adult and juvenile animals of the same species (e.g., white-beaked dolphin) can also consume different prey (Ringelstein et al., 2006; Hernandez-Milian et al., 2015), or prey size may be correlated with body length (as seen in bottlenose dolphins) resulting in clear ontogenetic shifts in diet (Knoff et al., 2008).

Use of tissue samples obtained from stranded animals is a low-cost and opportunistic way to monitor wild cetacean populations. We acknowledge the caveats associated with utilising these samples and their associated data, where cetacean health frequently plays a role in stranding events (Arbelo et al., 2013). The reporting of stranding cases can be biased by a variety of physical and social factors (for example: coastal geography, direction of ocean currents, human population density in coastal area, ease of reporting) (ten Doeschate et al., 2018). In addition, we assume carcass decomposition has not altered the original skin isotopic composition. Payo-Payo et al. (2013) reported no significant change in dolphin skin δ ¹³C and δ ¹⁵N after 62 days of non-submerged decomposition, applicable to carcass condition code 4 and above. The pressing demand for cetacean ecological information, however, coupled with the extensive challenges associated with collecting live biopsy samples (or tissue from by-caught animals) highlight the continual utility and importance of opportunistic samples taken from stranded animals.