© B.I.C. Warren et al., Published by EDP Sciences 2026

1 Introduction

The trophic dynamics and competitive interactions of populations within communities are important in determining food web structure (Woodward and Hildrew, 2002; George and Collins, 2024) and has potentially important implications for individuals within migratory populations (Lizé et al., 2023). In environments where prey resources are limiting, the coexistence of closely related and morphologically similar species that share these resources compete intensely and where their competitive interactions are asymmetric, this can potentially result in competitive displacement (Cucherousset et al., 2012; Kakareko et al., 2013). Alternatively, these sympatric competitive interactions in resource limited systems can lead to niche partitioning, where the species exploit different food resources (MacArthur, 1965; Ross et al., 1985). However, trophic niche overlaps can remain between species where resources are unlimited and / or the species have insufficient adaptive capacity to switch to alternative prey (Abrams et al., 2006; Bauduin et al., 2013; Cucherousset et al., 2020). In allopatry, the absence of inter-specific competition can result in the species exploiting a wider range of resources than in sympatry, with reduced dietary specialisms (Tran et al., 2015). Where niches remain similar between sympatric and allopatric contexts, this can be due to functional constraints in the species, where the niche conservation hypothesis suggests species will retain their niche through space and time, irrespective of context (Comte et al., 2017).

Somatic growth rates and body sizes at age of fish are associated with prey resource quality and quantity, so inter-population differences in these metrics can reflect differences in the productivity of the ecosystems (Tanner et al., 2019; Lai et al., 2024). These differences in productivity could then be important in influencing how life history strategies are expressed within the coexisting species, especially if these are anadromous (Simmons et al., 2022). Atlantic salmon Salmo salar populations generally express obligate anadromy, where the majority of their growth and development occurs in marine environment, whereas brown trout Salmon trutta tend to express facultative anadromy, where within populations, some individuals express anadromy and others remain as river residents (Jonsson and Jonsson, 2021). These salmonid species coexist across much of their respective ranges, including across a wide range of river types, from nutrient rich chalk streams to nutrient poor upland streams (Crisp, 1993; Simmons et al., 2021). As they also have relatively similar functional morphologies and ecological requirements (Johnson and Johnson, 2011), sympatric populations have the potential to consume similar prey resources. However, communities of sympatric salmonid populations often express strong patterns of inter-specific ecological and trophic niche partitioning (Eloranta et al., 2013). This partitioning often becomes strongly apparent during ontogenetic development, when individuals within populations become increasingly specialised in their resource use (Jørgensen et al., 2000; Sánchez-Hernández and Cobo, 2016). Lake dwelling populations of Atlantic salmon and brown trout are often generalist foragers with a broad diet, but with individuals increasingly specialising on lower numbers of prey items as they grow, with inter-specific niche divergence becoming increasingly evident (Jørgensen et al., 2000). The opportunity for populations of the two species to compete for prey resources is elevated in streams where their spatial niche overlaps can be considerable, although juvenile salmon tend to be in faster flowing and shallower habitats than brown trout, with salmon habitat use also restricted by interactive segregation due to aggressive interspecific competitive interactions from the trout (Heggenes et al., 1999).

The aim of this study was to assess the interspecific trophic interactions between Atlantic salmon (“salmon”) and brown trout (“trout”) in two streams in Southern England that differed markedly in the somatic growth rates of both species using stable isotope analysis (SIA). The analyses of their growth rates and stable isotope ecology enabled testing of whether the sympatric populations expressed trophic (stable isotope) niche partitioning or overlap in both streams, and whether these patterns were influenced by their growth rates. In addition, samples were also collected from one of the streams from areas where either salmon or trout were in allopatry, enabling testing of the niche conservation hypothesis through comparison for each species of their stable isotope niche sizes and positions between sympatry and allopatry.

2 Materials and methods

2.1 Study areas and fish sampling

The two rivers were the River Frome and the River Teign in Southern England (Fig. 1). The Frome is a lowland chalk-stream in that flows for approximately 70 km from its source (50.50. 24°N; 02.36. 12°W) to its tidal limit (50.40. 38°N; 02.07. 30°W). There is a fish assemblage of approximately 15 species present, with non-salmonid species comprising of several species of cypriniforms that are numerically dominated by European minnow Phoxinus Phoxinus. However, the river is most recognised for its regionally significant populations of salmon and trout (Simmons et al., 2020; Marsh et al., 2020). The site sampled was on a millstream of approximately 2 km length with connections at the upstream and downstream end to the main river (Fig. 1), with the stream comprising of a series of pool/ riffle habitats where depths rarely exceeded 1 m and channel widths were to 6 m (Fig. 1). The fish assemblage was sampled using generator powered electric fishing, with samples collected in July 2008 and September 2008 (Cucherousset et al., 2011). The River Teign is an upland spate river in Devon, southern England (Fig. 1), which flows approximately 50 km from its source (50°38'19.0” N; 3°57'31.0” W) to its tidal limit below Newton Abbot (50°32'10.8”; N 3°35'13.7” W). Its fish assemblage is largely dominated by brown trout (anadromous and non-anadromous forms) and salmon (all assumed anadromous), but has low diversity and abundances of other fishes, with minnow Phoxinus phoxinus, bullhead Cottus gobio and stone loach Barbatuala barbatula mainly being present. Three focal reaches were used for repeated sampling that were located in the upper catchment: Blackaton Brook (Site 1), the North Teign (Site 2) and South Teign (Site 3) (Fig. 1). These sites were selected for the consistency in their relatively high populations of juvenile salmonids in historical surveys (WRT 2022). All the sites comprised of a series of pool and riffle sequences, with the major differences being in their physical characteristics. Site 1 was a relatively small tributary of maximum width 5m and depths to 0.8 m, Site 2 was up to 10 m wide with depths to 1.3 m, whereas Site 3 was more regulated in its flows due to the influence of a water supply reservoir upstream (Fernworthy Reservoir), and was of maximum width 6 m with depths to 1 m. These sites were sampled using back-pack electric fishing (SmithRoot, LR24) in September 2022, July 2023 and September 2023. Although in relatively close proximity, mark-recapture events using passive integrated transponder (PIT) tags revealed negligible mixing of fish between the sites across the sampling periods, so each site was considered independent. In addition, a further 7 sites across the River Teign catchment were sampled in late August 2023 using back-pack electric fishing, where 3 comprised of samples of allopatric salmon and 4 of allopatric trout (Fig. 1).

In all surveys on both rivers, following their capture, the fish were anaesthetised (M222), identified to species and measured (fork length, nearest mm). A sample of scales was taken from below the dorsal fin but above the lateral line and stored in a paper envelope for subsequent age analyses. A pelvic fin clip was taken for SIA. All fish were then held in clean river water for recovery to normal behaviour and were then returned to the area of river where they were captured. Fish sample sizes are provided in Table 1. Concomitantly, putative macro-invertebrate fish prey resources were then collected from sites using a sweep net (250 mm wide; 0.3 m bag depth; 250 μm mesh).

Fig. 1 Main: locations of the River Teign and River Frome in Southern England; bottom left: location of the Millstream study site on the River Frome; top right: sampling areas on the River Teign with sympatric populations (circles), allopatric salmon (diamonds) and allopatric trout (triangles) (cf. Tabs. 1 and 3); bottom right: core sympatric Teign sites.

2.2 Growth rate analyses

To assess differences in trout growth rates between the rivers, scales were viewed under a projecting microscope (x48 magnification), an age estimate was derived for each individual, and their lengths at the last annulus determined (‘length at age’ hereafter). The length-at-age trout data for all both rivers were then combined into a single dataset and a linear regression model fitted where age was the independent variable and length at the last annulus was the dependent variable. The standardised residuals of length at the last annulus of each individual trout were then determined from the predicted values derived from the model (Beardsley and Britton, 2012). These standardised residuals were then tested between the two rivers using ANOVA, with the mean standardised residuals (± 95% confidence limits) also plotted for comparative purposes (Fig. 2). For salmon, the presence of only 1 yr class present in the Frome meant that rather than plotting lengths at age, length at capture at age 0+ was used across both rivers only, with the mean length of all 0+ fish used to calculate the standardised residuals as outlined above, which were then tested in ANOVA. In all cases, negative standardised residuals from the regression indicate trout with their length at the last annulus being below the predicted length at age fitted by the regression model, and salmon that were smaller than the predicted 0+ fish length (i.e., the growth rates of these fish were slower than predicted by the model).

Fig. 2 Mean standardised growth residuals (clear circles) and their 95% CI (vertical lines) of (A) Brown trout and (B) Atlantic salmon for the Rivers Frome and Teign between samples collected in July and September. The dashed line separates the samples collected in July and September. Negative values indicate smaller lengths at the last annulus versus the predicted lengths of the regression model fitted using data from across both rivers and all sampling occasions.

2.3 Stable isotope analyses

All of the fish fin tissue and macroinvertebrate samples were dried to constant mass at 60 °C within four weeks of sampling. These samples were then analysed at the Cornell University Stable Isotope Laboratory (New York, USA) for δ¹³C and δ¹⁵N in a Thermo Delta V isotope ratio mass spectrometer (Thermo Scientific, USA) interfaced to a NC2500 elemental analyser (CE Elantach Inc., USA), with this completed within six months of the fish being sampled. Analytical precision of the δ¹³C and δ¹⁵N sample runs was estimated against an internal standard sample of animal (deer) material every 10 samples, with the overall standard deviation estimated at 0.08 and 0.04 ‰ respectively. Ratios of C:N indicated no requirement for lipid normalisation (generally 3.5 to 3.9) (Winter and Britton, 2021). As the putative macroinvertebrate prey of the River Frome were all from the same reach and direct comparisons in stable isotope metrics were not being completed with the Teign fish then the Frome SI data were used in their original form. In the River Teign, however, there were temporal and spatial differences in the SI data that meant the fish data were corrected (De Santis et al., 2021). δ¹⁵N were converted to trophic position (TP) (Olsson et al., 2009):

$$\text{TP}=\left({\delta }^{15}{\text{N}}_{\text{fish}}–{\delta }^{15}{\text{N}}_{\text{prey}}/3.4\right)+2$$(1)

where TP and δ¹⁵N_fish are the trophic positions and the nitrogen ratios of each individual fish, δ¹⁵N_prey is the mean nitrogen ratio of the putative macroinvertebrate prey resources, 2 is the trophic position of these prey resources (as primary consumers) and 3.4 is the generally accepted fractionation factor between adjacent trophic levels (Post, 2002). The δ¹³C data were converted to corrected carbon (δ¹³C_corr) (Olsson et al., 2009):

$${\delta }^{13}{\text{C}}_{\text{corr}}=\left({\delta }^{13}{\text{C}}_{\text{fish}}–{\delta }^{13}{\text{C}}_{\text{meanMI}}\right)/{\text{CR}}_{\text{MI}}$$(2)

wherein δ¹³C_fish is the δ¹³C value of each fish, δ¹³C_meanMI is the mean δ¹³C of the macroinvertebrate prey and CR_MI is the carbon range (δ¹³C_max − δ¹³C_min) of the same macroinvertebrates (Olsson et al., 2009). An important aspect in the application of Equation (1) and (2) was the putative prey macroinvertebrate resource data used in the calculations. Sampling of the macroinvertebrates in the focal Teign sites revealed a relatively species poor community of low abundance, where the only taxonomic group sufficiently present to provide at least triplicate samples for SIA across all three sites was Gammarus pules. Thus, it was Gammarus pulex that were used exclusively for the macro-invertebrate SI data n Equation (1) and (2). Functionally, gammarids are detritivores that often consume some terrestrial resources, with these resources tending to be less variable between sites than aquatic resources, meaning the full extent of isotopic variability in the putative prey resources might not have been captured (i.e., underestimated). However, MacNeil et al. (1997) suggested that Gammarus pulex has a wider food base than is often acknowledged, which would increase their isotopic variability and reducing the effect of this limitation on calculations in Equation (1) and (2). In addition, a visual inspection on the original SI data at the species/ site/ date level was also completed to identify whether there were any major discrepancies between the patterns between the original and corrected data that could indicate an issue with the application of these macroinvertebrate data.

The fish original SI data for the Frome, and corrected SI data for focal Teign sites were then used to calculate the trophic niche sizes (as the stable isotope niche) of salmon and trout for each location and sampling date (Tab. 1). These calculations used the stable isotope niche as a proxy of the trophic niche (Jackson et al., 2011). Whilst closely related to the trophic niche, the stable isotope niche is also influenced by factors including growth rate and metabolism (Jackson et al., 2011). These niches were calculated using standard ellipse areas (SEA) in SIBER (Jackson et al., 2011; Jackson et al., 2012), where SEAs are a bivariate measure of the distribution of individuals in stable isotope space. The ellipses enclose the core 40% of data and so they represent the typical resource use of the analysed population (Jackson et al., 2011). The Bayesian estimate of SEA (SEA_B) tested differences in niche sizes between the species in the different sites and was calculated using a Markov chain Monte Carlo simulation (10⁴ iterations per category) (Jackson et al., 2011; Jackson et al., 2012). Differences in the sizes of stable isotope niches (as SEA_B) of the species were evaluated in SIBER by calculating the probability that the relative posterior distributions of the niche sizes were significantly smaller or larger between the species (α = 0.05). Thus, no overlap in the 95% confidence intervals of SEA_B indicated a significantly smaller/ larger niche size (depending on the direction of the difference between the species). The SI data were then used to calculate stable isotope niche overlap (%) between the species using SEA_c in SIBER (subscript ‘_c’ indicates a small sample size correction was used; Jackson et al., 2012).

Differences in TP and δ¹³C_corr between the allopatric and sympatric contexts in the River Teign were tested for each species in linear mixed effects models (LMEM), where prior to model construction, assumptions of normality of residuals and homoscedasticity were checked. The structure of the data used in the model combined the site level data into the two contexts of allopatry and sympatry, with each site within the contexts representing a replicate. Using data from each individual fish per site and per context in the model as true replicates would inflate the residual degrees of freedom and so the models fitted site as a random effect (Dossena et al., 2012). The model structure used TP and δ¹³C_corr as the dependent variable, context and fish length as fixed effects, and site as the random variable. Each model was fitted using restricted maximum likelihood to determine the parameter estimates, with differences in TP and δ¹³C_corr by context determined using estimated marginal means and multiple comparison post-hoc analyses (general linear hypothesis test). The LMEMs were then complemented by calculation of the stable isotope metrics as outlined above.

All data analyses were completed in R Development Core Team, (2022). Where error is provided around mean values, it represents 95% confidence limits unless otherwise stated.

3 Results

3.1 Lengths and growth rates of sympatric salmonids

Salmon present in the River Frome samples were between 60 and 110 mm and between 53 and 147 mm in the Teign (Tab. 1), with the latter being significantly larger (F_1,224 = 36.57, P < 0.001). Trout in the River Frome samples were between 73 and 361 mm versus 48 and 325 mm in the Teign (Tab. 1), with these differences not significant (ANOVA: F_1,239 = 0.22, P = 0.64). All Frome salmon were age 0+ at the time of sampling versus between 0+ and 2+ in the Teign, whereas trout in both rivers were present between age 0+ and 3+. The standardised residuals of the lengths at the last annulus of Teign trout, and of 0+ salmon length, were all negative (i.e., slower growing than predicted), whereas both metrics for the Frome in September were positive (i.e., faster growing than predicted), with these growth differences being significant (ANOVA: trout F_3,209 = 55.72, P < 0.001; salmon F_3,126 = 26.55, P < 0.001; Fig. 2).

3.2 Stable isotope analyses of sympatric salmonids

The relationships between fish length and δ¹³C, δ ¹³C_corr, δ¹⁵N and TP were generally significant for each species and river, except Teign trout versus TP and Frome salmon versus δ¹³C (Tabs. 1 and 2). In all significant regressions, the relationship between length and the stable isotope metric was positive, with increasing values of δ¹³C, δ¹³C_corr, δ¹⁵N and TP as length increased. The stable isotope metrics for salmon and trout in the River Frome revealed a consistent pattern of trophic niche partitioning between the species in samples collected in both July and September, with no overlap in standard ellipse areas (Tab. 1; Fig. 3). The general patterns observed in the uncorrected data in the focal Teign sites were broadly similar to those of the corrected data (Figs. 3 and 4). The corrected data enabled appropriate comparison in isotopic niche metrics between the species and sites, with trout and salmon in Sites 1 and 2 being strongly divergent in their trophic niches, with similar niche positions and sizes in both species on all sampling occasions (Tab. 1; Figs. 4 and 5). In Site 3, the trophic niches of the two species less diverged, but still had relatively limited overlap (Figs. 4 and 5).

Fig. 3 Standard ellipse areas of sympatric Atlantic salmon (grey ellipse) and brown trout (black ellipse) for samples collected from the River Frome in July and September 2008. Note that these stable isotope data were used without correction.

Fig. 4 Stable isotope plots (using uncorrected values of δ13C and δ15N) of Atlantic salmon (filled circles) and brown trout (clear circles) for Site (S) 1, 2 and 3 in September (9) 2022 and 2023, and July (7) 2023. Note the differences in scales on the x and y axes for the plots between S1 to S3, which was done data comparison purposes between the two species.

Fig. 5 Stable isotope niches using corrected stable isotope data (δ13C to δ13Ccorr (‰); Eq. (2; δ15N to trophic position (TP); Eq. (1) for Atlantic salmon (grey ellipses) and brown trout (black ellipses) in the River Teign sites (cf. Tab. 1 details and sample sizes, Fig. 3 for uncorrected data), where solid lines are fish sampled in September 2022, long dashes in July 2023 and short dashes in September 2023.

3.3 Stable isotope analyses of salmonids in allopatry vs sympatry

The LMEMs revealed that in both brown trout and salmon, differences in TP and δ¹³C_corr were not significant between allopatry and sympatry in the River Teign, with fish length being the only significant variable for δ¹³C_cor in both species (Tabs. 3 and 4). Moreover, comparison of the stable isotope niche positions of the allopatric and sympatric contexts by species revealed that their overall niche positions in stable isotope space were relatively similar between the two contexts. For the isotopic niche, SEAc indicated larger niche sizes in sympatry, but with the 95% credible intervals of SEA_B overlapping between the contexts for both species (Tab. 5). Within each species, there was relatively high niche overlap at allopatric versus sympatric contexts (>90%; Fig. 6; Tabs. 3 and 5).

Fig. 6 Stable isotope niches according to δ13Ccorr (‰; Eq. (2) and trophic position (TP; Eq. (1) of Atlantic salmon (top) and brown trout (bottom) in allopatry (grey ellipse) and sympatry (black dashed ellipse) in the River Teign (cf. Tab. 3 for sample sizes and fish lengths).

4 Discussion

The sympatric brown trout and Atlantic salmon populations of these two rivers in Southern England revealed consistent patterns of stable isotope niche partitioning. The positions of the stable isotope niches of trout and salmon were similar in both allopatry and sympatry and, while the singular estimates of the standard ellipse areas (SEA, SEAc) indicated larger niche sizes in sympatry, the 95% credible intervals of SEA_B indicated these niches did overlap in their sizes. Accordingly, with the minimal differences in niche position and overlaps in the 95% credible intervals of SEA_B suggest alignment to the niche conservation hypothesis, with the species in allopatry not consuming a wider range of prey resources than those in sympatry. This pattern of stable isotope niche partitioning is consistent with some other salmonid fish communities, such as coexisting brown trout, Arctic charr Salvelinus alpinus and brook trout Salvelinus fontinalis (Eloranta et al., 2013; Cucherousset et al., 2007).

The standardised growth residuals of both salmonid species indicated relatively slow growth in the River Teign, with some salmon being in their third year of life without yet running to sea as smolts, whereas most salmon in the Frome emigrate to sea at age 1+ years (Simmons et al., 2020, 2022). Although prey availability was not measured quantitively in either the Frome or Teign, the Frome and its connected side-streams are recognised for their high macroinvertebrate diversity and abundances, with it being a chalk stream of high water quality (Cannan and Armitage, 1999; Armitage et al., 2003). The Teign is comparatively species poor compared to the Frome and has generally lower macroinvertebrate abundances, but also has high water quality in the study reaches (Gravelle et al., 1997). Despite the suggested lower productivity in the Teign versus the Frome, there was no evidence to suggest the two salmonid fishes were competing for shared prey resources in the Teign. The partitioning in their stable isotope niches indicated some differentiation in their resource use, with presumably sufficient prey diversity and abundance in the river to support both fish populations (MacArthur, 1965; Ross, 1985). Nevertheless, a limitation of the study here was its inability to develop greater understandings in the differences in resource use between the two species in the Teign (as the less productive system), at least in part to the low taxonomic diversity that was captured in the macroinvertebrate samples. This resulted in the sole use of gammarids as putative prey resources in Equation (1) and (2), which has already been outlined as a limitation due to their detritivory of allochthonous terrestrial resources (cf. Methods and Materials). Although the datasets for two rivers were collected approximately 15 yr apart, this was not considered a limitation due to the stable isotope processing occurring in the following months after sampling. However, these results do indicate that to understand the differences in prey between the species would need to use complementary methods such as stomach content analyses (Amundsen and Sánchez‐Hernández, 2019) and/ or DNA based approaches (Traugott et al., 2021).

When species are present in allopatric contexts then the lack of inter-specific competition can enable the population to exploit a wider range of resources versus their populations within communities where other species are present, with allopatric diets thus often being more generalised than those in sympatry (Tran et al., 2015). This was not, however, evident in the River Teign, where the stable isotope values and niche sizes and positions of the allopatric trout and salmon were similar to those in sympatry − and where there were differences in metrics (e.g., SEAc), they indicated larger niches in sympatry than allopatry. Where niche sizes are larger in sympatry, it suggests some intra-specific resource partitioning and/ or greater opportunistic foraging in the presence of another salmonid fish. While not tested directly, the stable isotope niche partitioning evident between the two species in sympatry was interpreted as being due to some subtle differences in their functional morphologies and habitat use, which were then conserved when the fish were in allopatry, with this retention of their trophic niche characteristics across these contexts being consistent with the niche conservation hypothesis (Comte et al., 2017). These salmonid populations were thus consistent with other communities of sympatric salmonids that often exhibit strong ecological and trophic niche partitioning (Eloranta et al., 2013). This has already been detected in lake dwelling populations of salmon and trout where dietary specialisations were enabled through piscivory (Jørgensen et al., 2000). In stream populations, the more limited space available to the fishes potentially increases the potential for niche overlap to occur, and thus the differences in habitat use that generally occur between these species (e.g., juvenile salmon use faster flowing and shallower habitats versus trout) were assumed as being an important factor in driving this niche conservatism (Heggenes et al., 1999).

The coexistence of fishes can, however, be driven more by differences in their dietary resource use rather than habitat selection (Ross, 1986), where interspecific stable isotope niche partitioning reduces the competitive interactions between species by promoting niche complementarity (Larocque et al., 2020). Within salmonids, rainbow trout Oncorhynchus mykiss and brook trout can shift their diets towards resources of higher δ¹³C values as they grow (Ciancio Blanc et al., 2008; Jardine et al., 2005), with this ontogenetic diet shift often representing a shift from aquatic to terrestrial prey, and thus this diversification of diet reduces the intensity of competition for aquatic prey resources (Sánchez-Hernández and Cobo, 2018). In addition, the competitive interactions and trophic relationships of populations of coexisting salmonid fishes potentially influences the expression of their life history strategies (Simmons et al., 2022). In both rivers, salmon are considered to express obligate anadromy, where the only major difference between the rivers was their age of emigration, with this influenced by their somatic growth rates and thus, most likely, differences in prey abundances or fish density, but not the trophic niche size or position of the population, or the niche overlap with trout. The brown trout populations of both rivers express facultative anadromy, where within populations, some individuals express anadromy and others remain as river residents (Jonsson and Jonsson, 2021). However, the results present here do not provide evidence that the decision of whether to stay within the river or to emigrate to sea was driven by issues relating to their trophic niche size, position or overlap with salmon (Ferguson et al., 2019).

In conclusion, despite contrasting somatic growth between these salmonid fish populations across the two rivers, the general patterns detected in their stable isotope niches were similar, with stable isotope niche partitioning evident and with these niches being largely conserved when the species were in allopatry. These results suggest that differences in individual and population level growth rates were not important factors in driving their trophic niche sizes and were considered to instead relate to differences in functional morphology and habitat use. These results thus emphasise the importance of identifying the differences in trophic ecology between closely related coexisting fish species for developing new understandings of their inter-specific interactions.

Acknowledgments

BW was funded through a studentship partnership between Bournemouth University, Environment Agency, the Teign Anglers Conservation Association and the West Country Rivers Trust. We especially thank Geoff Stephens for all his help in organising site access and electric fishing.

References

All Tables

All Figures

	Fig. 1 Main: locations of the River Teign and River Frome in Southern England; bottom left: location of the Millstream study site on the River Frome; top right: sampling areas on the River Teign with sympatric populations (circles), allopatric salmon (diamonds) and allopatric trout (triangles) (cf. Tabs. 1 and 3); bottom right: core sympatric Teign sites.
In the text

Fig. 2 Mean standardised growth residuals (clear circles) and their 95% CI (vertical lines) of (A) Brown trout and (B) Atlantic salmon for the Rivers Frome and Teign between samples collected in July and September. The dashed line separates the samples collected in July and September. Negative values indicate smaller lengths at the last annulus versus the predicted lengths of the regression model fitted using data from across both rivers and all sampling occasions.

In the text

	Fig. 3 Standard ellipse areas of sympatric Atlantic salmon (grey ellipse) and brown trout (black ellipse) for samples collected from the River Frome in July and September 2008. Note that these stable isotope data were used without correction.
In the text

	Fig. 4 Stable isotope plots (using uncorrected values of δ13C and δ15N) of Atlantic salmon (filled circles) and brown trout (clear circles) for Site (S) 1, 2 and 3 in September (9) 2022 and 2023, and July (7) 2023. Note the differences in scales on the x and y axes for the plots between S1 to S3, which was done data comparison purposes between the two species.
In the text

Fig. 5 Stable isotope niches using corrected stable isotope data (δ13C to δ13Ccorr (‰); Eq. (2; δ15N to trophic position (TP); Eq. (1) for Atlantic salmon (grey ellipses) and brown trout (black ellipses) in the River Teign sites (cf. Tab. 1 details and sample sizes, Fig. 3 for uncorrected data), where solid lines are fish sampled in September 2022, long dashes in July 2023 and short dashes in September 2023.

In the text

	Fig. 6 Stable isotope niches according to δ13Ccorr (‰; Eq. (2) and trophic position (TP; Eq. (1) of Atlantic salmon (top) and brown trout (bottom) in allopatry (grey ellipse) and sympatry (black dashed ellipse) in the River Teign (cf. Tab. 3 for sample sizes and fish lengths).
In the text