In this study, we estimated the level of competition between the only known population of the Galápagos Pink Land Iguana, Conolophus marthae, and the syntopic population of the closely related and more widely distributed C. subcristatus on Wolf Volcano, Isabela Island. Our results indicate that C. marthae occupies a smaller isotopic niche than C. subcristatus, with 65% of the standard ellipse area (SEAc) falling within the SEAc of C. subcristatus (Fig. 1A). This overlap is reflected in the similar δ15N recorded for the two species (Table 1 – Model 1), which suggests that, overall, they occupy a similar trophic level, consistently with previous reports indicating that both species primarily feed on vegetation, as typical of most large iguanas [18, 19, 55, 56]. Importantly, males and females show larger overlap in C. marthae than in C. subcristatus (Fig. 1A-C), suggesting that higher differentiation between the isotopic niches of males and females contributes to the larger overall niche occupied by C. subcristatus. Moreover, comparing the standard ellipses obtained from raw data (Fig. 1A) and those from the residuals of a model accounting for size, productivity and space (SM Fig. 5), shows that, while the difference between sexes in residual δ13C of C. subcristatus is much smaller than in the raw data, the difference in δ15N is similar, implying that differences in trophic levels between male and female C. subcristatus are not explained by differences in size or occupied habitat and may directly reflect sex-specific resource selection.
Despite the broad overlap, our results showed that δ13C clearly differ between C. marthae and C. subcristatus at Wolf Volcano (Table 1 – Model 1), supporting the existence of resource partitioning mechanisms acting to reduce the competition between these species. Evidence in this regard is also provided by a recent study indicating that differences in chemical signatures of femoral pore secretions extracted from these two species could be associated with a difference in species diet . Niche-partitioning might be the result of two non-mutually exclusive mechanisms. First, individuals of the two species could feed on different trophic resources in the overlapping distribution area. The generalist habit of C. subcristatus [18, 19] could provide a competitive advantage to the species, allowing the use of resources that are present inside areas occupied by C. marthae individuals but not accessible to them perhaps due to some dietary or metabolic constraints. Second, they could feed in different habitats or microhabitats and consequently use different types of food . Recent field observations would support a differential microhabitat use for these two species, with C. marthae that seems to preferentially occupy more shaded areas. The potential different use of habitats or microhabitats by the two species has already been considered as a mechanism acting to enhance the avoidance of interspecific hybridization between these two populations on Wolf Volcano .
Both strategies are supported by our data. On the one hand, we found that i) the capture locations of the two species were spatially aggregated, ii) the two species were captured at locations differing in primary productivity (Table 2; SM Fig. 4) iii) productivity at capture points is a significant predictor of both δ13C and δ15N (Table 1) and iv) unmeasured features of the space (measured by the spatial tensor product) have a significant effect on δ15N (Table 1), indicating selection of different microhabitats as a likely niche partitioning mechanism. On the other hand, resource selection may also marginally contribute to differences in isotope niches. For example, while C3 plants have a δ13C ranging from − 22‰ to -30‰, C4 show values ranging from − 10‰ to -14‰ , and CAM (Crassulacean Acid Metabolism) from − 10 to -20‰ . The diet of C. subcristatus could therefore include a larger portion of C4 and/or CAM plants than the diet of C. marthae, translating to the higher values of δ13C observed in the former (Fig. 1A). Consistent with this view, Opuntia cactus, which were previously reported to constitute a large portion of the diet of C. subcristatus  and C. pallidus  during the dry season, is a constitutive CAM species , relatively rare inside the area of distribution of C. marthae, but abundant on the eastern side of Wolf Volcano. While this area is outside the core area of C. marthae, it is inhabited by C. subcristatus. It is not uncommon that when two species compete for common resources, the dominant species (often the larger one) could displace the less competitive one to a suboptimal trophic niche . Therefore, the current situation on Wolf Volcano could be the result of a competitive process that has led C. marthae to use a suboptimal array of trophic resources.
Our data hint at resource partitioning operating at an intraspecific level too, with males of both species showing higher δ13C than females (Fig. 1A) and C. subcristatus males showing higher δ15N than their conspecific females (Fig. 1A). Intraspecific niche partitioning mechanisms are widespread in animal populations and can be interpreted as a strategy to reduce intersexual competition for food resources . The isotopic pattern observed in our data is likely to reflect behavioral differences between sexes. It is indeed well known that, during the reproductive season, approximately between January and May, on Fernandina Island, C. subcristatus males exhibit a territorial behavior, defending areas that constitute the core part of the mating area, while the females of the species shift between territories defended by different males . Males reduce feeding during the mating season and food resources inside or near their territories are primarily necessary for females . After mating, females leave these areas to reach suitable nesting sites, while males remain in their territories if food is still abundant or they disperse, looking for alternative feeding areas . These observations, therefore, indicate that both temporal and spatial mechanisms act to reduce intersexual competition for trophic resources in C. subcristatus population on Fernandina Island. As δ15N of consumers generally increase with starvation , if the same behavioral mechanisms acted in C. marthae and C. subcristatus populations on Wolf Volcano, males would show higher δ15N during the period for which our data are informative (which can be very approximately estimated to encompass the months from January to June). Interestingly, our results suggest that this explanation might be reasonable for C. subcristatus population on Wolf Volcano as males of the species showed δ15N values higher than those of conspecific females (Fig. 1A), while this difference was not revealed for C. marthae (males and females were characterized by similar δ15N values; Fig. 1A), suggesting that this behavior may not be shared by the two syntopic populations.
A large part of the variance observed in δ13C and δ15N between species and sexes can be explained in terms of body size, productivity and relative location of the capture points (Fig. 2). Moreover, these ecological variables provided the highest independent contribution to explaining δ13C and δ15N recorded in the samples (Fig. 2). We found that, other predictors being equal, larger individuals show higher levels of both δ13C and δ15N (Table 1, Fig. 3), suggesting that they occupy higher trophic levels. It is not uncommon to find a positive relation between body size and trophic level in natural populations [66, 67]. Larger individuals often have a competitive advantage over smaller ones, allowing them to integrate their diet with trophic resources of animal origin that could not be available for smaller individuals. Accordingly, Hanson and colleagues  found a positive relationship between δ13C and SVL in Crocodylus porosus, suggesting that individuals of different body size are trophically linked to different primary resources.
We also found a negative relation between the NDVI of samples location and both δ13C and δ15N (Fig. 3). On the one hand, different availability of C3, C4 and CAM plant resources in areas with different productivity may contribute to the differences in δ13C. On the other hand, as animal tissues are enriched in both δ13C and δ15N relative to their diet [27–29], a direct explanation of this pattern may involve a higher consumption of animal food by those iguanas that occupy less productive microhabitats compared to those that were captured in more productive habitats. Interestingly, C. marthae individuals were mostly found in areas with higher levels of NDVI, if compared with C. subcristatus (SM Fig. 4). More vegetated areas also offer better shelter opportunities. Field observations suggested that C. marthae occupy more shaded areas than C. subcristatus , and this preference could be an ecological response to partial skin depigmentation of C. marthae . Therefore, this species might be found in more vegetated areas also because, here, iguanas might find a more shelters-enriched environment that could facilitate their basking-regulation behavior (Di Giacomo et al., submitted).
Our study outlines that small-scale spatial pattern, both in terms of habitat productivity and of an undetermined spatial effect revealed by the tensor product smooth on capture locations coordinates, did affect the isotopic composition of samples (Table 1 – Model 2; Table 1 – Model 3). On the one hand, we argued that this finding supports differential use of microhabitats by these two species. On the other hand, as the isotopic signature recorded for these species should reflect a time-scale of a few to several months [40, 41], this result strongly suggests that, at least in the time of the year for which our data may be informative, the feeding ranges of these two species may be extremely restricted. More in general, as our study only covered an area of ca. 2 km2, these results highlight the importance of considering the spatial dimension of processes when assessing ecological problems at any spatial scale.