Introduction

Trophic competition, along with other types of competition and predator–prey interactions, is one of the main ways in which an invasive species might modify the trophic web of a newly invaded ecosystem (David et al. 2017). Competitive interactions between native and invasive species have been repeatedly studied in animal ecology, highlighting the ability of an invasive species to overlap with, modify or displace the trophic niche of native competitors (Blanchet et al. 2007; Pérez-Santigosa et al. 2011; Bertolino et al. 2014; David et al. 2017). However, only a few of these studies have assessed possible geographical differences or evolutionary changes in the effects of an introduced competitor on native species, hence leading to the contentious assumption that the impact of an invasive species is similar along time and space (but see Stuart et al. 2014). Theoretically, once a new competitive relationship is created, selective pressures should favour individuals that better withstand it. Thus, invasive competitors influence natural selection of native species and vice versa, possibly inducing changes in the usual trophic position of a species or its habitat-linked morphology (Crowder 1986; Bourke et al. 1999; Stuart et al. 2014). Indeed, since evolutionary changes in response to predator–prey interactions with an invasive species can occur very rapidly (Trussell and Nicklin 2002; Phillips and Shine 2006; Nunes et al. 2014), similar evolutionary rates may also occur with trophic competition. Therefore, the competitive roles of native and invasive species might change, with the effects of long-term invasion largely depending on the capacity of both species to adapt (Cox 2004).

Anuran tadpoles have long been used as models for competition studies, including those on interactions between native and invasive species (Kupferberg 1997a; Kiesecker et al. 2001; Smith 2005; Cabrera-Guzmán et al. 2013; San Sebastián et al. 2015b). Most tadpole species from temperate areas are generalists that feed mainly on detritus, algae and phanerogams, also incorporating small animals and carcasses into their diets (Diaz-Paniagua 1985; Campeny 2001; Caut et al. 2013; Arribas et al. 2014). Several studies have suggested that trophic competition, and the quantity or quality of available resources, can influence growth and development, affecting individual fitness before and after metamorphosis (Brown and Rosati 1997; Kupferberg 1997b; Morey and Reznick 2001; Álvarez and Nicieza 2002; Enriquez-Urzelai et al. 2013; Martins et al. 2013; Richter-Boix et al. 2013; Pujol-Buxó et al. 2017). Methods examining gut content to analyse diet are not precise or reliable enough to detect subtle changes in the diet of tadpoles (Altig et al. 2007). Thus, stable isotope analysis (SIA) has become increasingly common in trophic studies on species whose general trophic ecology is already known (Whiles et al. 2006; Schiesari et al. 2009; Caut et al. 2013; Arribas et al. 2015; San Sebastián et al. 2015a). For instance, in the case of the inhabitants of small and enclosed systems comprising relatively simple and short food webs, such as ephemeral ponds, trophic changes might consist of modifications in the proportion rather than type of items consumed, making SIA especially useful.

Using SIA, San Sebastián et al. (2015a) reported that tadpoles of the invasive Discoglossus pictus (Anura: Discoglossidae) affected the diet of the tadpoles of the native competitor Epidalea calamita (Anura: Bufonidae), possibly displacing them from their preferred trophic niches in ponds. The invasive species, D. pictus, is continuously expanding (Geniez and Cheylan 2012; Llorente et al. 2015), which creates a gradient of time of coexistence with the native competitor. Using this natural experiment (HilleRisLambers et al. 2013), we attempted to evaluate possible modifications in the trophic ecology of both species in relation to the time since invasion. We examined their trophic niches in 16 ponds in four different areas with distinct co-evolutionary histories, which differed in the amount of time the native and invasive species had coexisted. If the native species is not displaced right at the onset of the invasion, competitors might progressively adapt to exploit different resources, leading to larger interspecific differences as the amount of time since invasion increases. By contrast, processes like phenotypic plasticity might have allowed large dietary differences from the onset of the invasion. Under this assumption, differences in the direction of this trophic partitioning or in the niche widths of each competitor might still be noticeable. Theoretically, interspecific competition should constrain niche width (Van Valen 1965; Araújo et al. 2008; Bolnick et al. 2010). However, several studies have shown that the dominant competitor can force the subordinate competitor to increase their niche width when both competitors have very similar food preferences (Codron et al. 2011; Abbey-Lee et al. 2013; San Sebastian et al. 2015a). Thus, it might also be possible to detect differences in the niche width along with different degrees or types of trophic segregation—i.e., using the same, or different, trophic levels.

Based on the results of several studies (Crowder 1986; Bourke et al. 1999; Codron et al. 2011; Abbey-Lee et al. 2013; Stuart et al. 2014; San Sebastian et al. 2015a), we hypothesised that interspecific differences in the mean trophic niche should widen with increasing time since the invasion (Fig. 1a–d). We also assumed that the invasion had initially forced wider niche widths, and therefore we expected these to become progressively narrower and more specialized (Fig. 1a–d). This would embody a case in which an initial native-invasive trophic overlap is eased after several generations of coexistence, possibly diminishing deleterious effects for the native species in the long term.

Fig. 1
figure 1

Schematic representation of expected results according to our hypothesis (a–d), compared to the obtained results (e–h). Circles represent trophic niche in a theoretical δ13C/δ15N space (continuous: invasive D. pictus, discontinuous: native E. calamita), its size and position corresponding to trophic width and mean diet. Plots are placed from left to right according to time since invasion: a, e represent the invasion onset at the expansion front; b, f represent the far expansion, in which our results coincide with San Sebastián et al. (2015a); c, g represent the mid expansion; and d and h represent the origin area

Full size image

Materials and methods

Study species

The Mediterranean Painted Frog (D. pictus) was introduced from Algeria to mainland Europe (Banyuls de la Marenda, SE France) approximately 110 years ago (Wintrebert 1908; Zangari et al. 2006). These populations became invasive, currently occupying a coastal strip stretching from Montpellier (SE France) to Sant Celoni (Catalonia, NE Spain) (Geniez and Cheylan 2012; Llorente et al. 2015). The native competitor chosen for this study, the Natterjack Toad (E. calamita), occurs throughout the invasive range of D. pictus. It breeds after heavy rainfall in ephemeral or temporary ponds. Tadpoles from both species have a benthic morphology and are often found in syntopy in ephemeral and temporary ponds (Escoriza and Boix 2012; Richter-Boix et al. 2013; authors, in prep.). However, while the invasive tadpoles have greater consumption rates, the native species bases its growth more on efficiency (Pujol-Buxó et al. 2016). When they compete, the native species acts as the subordinate competitor and shows reduced fitness both in the laboratory and in the field (Richter-Boix et al. 2013; San Sebastian et al. 2015a, b; authors, in prep.).

Field sampling and pond characterization

During the spring of 2016 and 2017, we monitored 69 ephemeral and temporary ponds (sensu Richter-Boix et al. 2013) in which the two study species were expected to co-occur (authors, in prep.). All the ponds were located along the southern expansion of D. pictus, clearly grouped into the following four different areas with distinct co-evolutionary histories: (1) the “origin area”, near (approximately 20 km) the site where the invasive species was first introduced and in which the native and invasive species have coexisted for approximately 110 years; (2) the “mid expansion”, 60 km from the origin area and in which the two species have coexisted for 30–40 years; (3) the “far expansion”, 80 km from the origin area and in which the species have coexisted for approximately 20 years; and finally (4) the “expansion front”, less than 3 km from the current southern distributional limit of the invasive species (100 km from the origin area) and in which the native and invasive species have co-occurred for less than 5 years. Assuming a mean generation time of 3 years (Oromi et al. 2012), 30–35 generations of the native toad have had experience competing with the invasive species in the origin area, with fewer generations of the native species having been exposed to the invasive frog in the other areas. Meanwhile, the invasive species, with a similar generation time (Oromi et al. 2016), has had equal experience with the native species in all the areas because the native toad has always been present. This has given the invasive species a clear “evolutionary advantage” in the expansion front, where the native species is naïve to the alien species. We sampled 16 ponds in total, as follows: (a) three to five ponds had to be sampled per study area—origin area: 4, mid expansion: 3, far expansion: 5, expansion front: 4; (b) only ponds with no other tadpole species were sampled; (c) both study species had to be at least at Gosner stage 25 (Gosner 1960); and (d) the whole tadpole guild—i.e., most tadpoles of both species—had already attained medium or large sizes. For SIA, we randomly sampled 10 tadpoles per species and pond, plus a small collection of the main accompanying species and trophic resources—algae, plants—when possible. All samples were euthanised with 70% ethanol in the field, which was substituted with clean ethanol 70° in the lab.

When ponds were sampled, they were also characterized (Table S1). We obtained total tadpole density and the ratio of D. pictus to E. calamita tadpoles as the mean of counts in five randomly placed square meters in the pond. The count was made at the whole pond if it had less than five square meters of surface. Pond surface was calculated using in situ field measures—as many necessary to collect the shape of the pond—and we obtained the maximum pond depth as the largest value obtained after probing with a vertical pole. We also noted the mean surface of the pond under direct sunlight—accounting for the natural movement of the sun—, and we noted all accompanying species of macroscopic vegetal and animal communities that could be detected by observation and dip-netting.

Laboratory procedures

As a surrogate for the trophic niche of each species, we obtained their carbon (δ13C) and nitrogen (δ15N) isotopic signatures, which are the most commonly used in similar studies (Whiles et al. 2006; Schiesari et al. 2009; Caut et al. 2013; Arribas et al. 2015; San Sebastián et al. 2015a). We also obtained δ13C and δ15N of the sampled vegetal and animal communities. According to previous studies, δ13C values provide information on the source of carbon, while δ15N values indicate the trophic level of the organism due to its high fractionation rates (Tieszen et al. 1983; Minagawa and Wada 1984; Ambrose and DeNiro 1986; Griffith 1992; Koch et al. 1995; Gannes et al. 1997). Isotopic signatures reflect the diet over the period during which the analysed tissue is formed or turned over (Layman et al. 2007; Caut et al. 2008), which in our case was the whole tadpole and its entire life (San Sebastián et al. 2015a). To avoid biases arising from the food remains in the digestive tract of the tadpoles, these were extracted before SIA. After this, samples were dried in a Memmert heater at 60 °C for 3 days and homogenised into fine powder using a manual grinder. As the lipid content in these tadpole species is low (San Sebastián et al. 2015a), we did not remove the lipids before SIA. The powdered samples were then weighed (0.25–0.3 mg for tadpoles and other animals and 1.2–1.4 mg for algae) and placed in tin capsules for mass spectrometry using an elemental analyser (Flash EA 1112) coupled to a stable isotope ratio mass spectrometer (CF-IRMS). The laboratory (Scientific-Technical Services of the University of Barcelona) uses international standards for δ15N and δ13C, atmospheric nitrogen and Pee Dee Belemnite, respectively, which are run after every 12 samples: IAEA CH7 (87% C), IAEA CH6 (42% C) and USGS 24 (100% C) for δ13C; and IAEA N1, IAEA N2 (21% N) and IAEA NO3 (13.8% N) for δ15N. Accuracy was ± 0.1% and ± 0.2% for δ13C and δ15N, respectively.

Statistical analyses

To perform interspecific comparisons, we corrected the isotopic signatures of each species using the species-specific fractionation values obtained in San Sebastián et al. (2015a). Henceforth only these δ13C and δ15N values are used. We first explored interspecific differences by measuring the mean and dispersion of the isotopic signatures for each species and pond. We used the methodologies described in Turner et al. (2010) to test for interspecific differences in the mean δ13C, δ15N and bivariate isotopic signatures for each pond. We used SIBER (Jackson et al. 2011) to obtain and test for interspecific differences in the corrected standard ellipse area (SEAc) and the total area (TA) (Layman et al. 2007).

After detecting widespread interspecific differences in most ponds (see Results), we aimed to determine the environmental variables that might have influenced the signatures. Based on the composition of the vegetal and animal communities, we created two distance matrices among the ponds, using the simple matching coefficient for the macroscopic vegetal community (as we had not overlooked any species), but using the Jaccard index for the animal community (as we could have overlooked some species during sampling). We applied a multidimensional scaling to both distance matrices, keeping two axes for the macroscopic vegetal community and three axes for the animal community, representing 69% and 60% of the variability.

We then fitted a linear mixed model (LMM) for each isotope, respectively using the tadpole’s individual values of δ13C and δ15N as response variable (N = 320), and the following as the explanatory terms: species; distance from the origin of the invasion (in km); the ratio of D. pictus tadpoles to E. calamita tadpoles in the pond; the mean proportion of the pond under direct sunlight; and the obtained variables for the vegetal and animal communities. We allowed for first order interactions between the species and the rest of the terms, always including both pond and study area as random intercepts to account for location- and area-specific changes in isotopic baselines. From these full models, including all the possible terms, we conducted an exhaustive model selection using the corrected Akaike information criterion (AIC, Burnham and Anderson 2003) implemented in glmulti (Calcagno 2013), allowing for the presence of interactions without the need for both single effects to be also present, i.e., not applying the marginality rule. Once the best set of models was found, we conducted likelihood ratio tests (LRTs) to decide which elements had to be kept in the consensus.

We obtained the direction of interspecific segregation for each pond as the angle present between species means in bivariate space. To explore the changes in the direction of the interspecific segregation in ponds, we tested if the angle between the segregation directions of a pair of ponds was higher than would be expected by chance using a self-written routine [Script S1, roughly based on the ideas in Turner et al. (Turner et al. 2010)]. The test consisted of randomising the angle between the segregation directions of each pair of ponds 999 times, posteriorly obtaining the probability that each value was due to chance. A similar test, adding the calculation of the mean for each study area, was used to determine if the four study areas significantly differed in their mean angle of segregation between the species (Script S2).

We studied the dispersion of the isotopic signatures using the following: the range of δ15N values (the δ15N range) and δ13C values (the δ13C range) in the pond to measure the diversity in the trophic level and origin of the organic matter in the pond, respectively; the mean nearest neighbour distance (MNND), which is the distance between the bivariate centroids of the isotopic signatures of both species in a community of two species, used as a surrogate for the magnitude of interspecific trophic segregation (Layman et al. 2007); and the SEAc of each species in each pond. The following were possible explanatory variables: the distance from the origin of the invasion (in km); the ratio of D. pictus to E. calamita tadpoles in the pond; the mean tadpole density in the pond; the pond surface; the maximum pond depth; the mean surface of the pond under direct sunlight; and the variables of vegetal and animal communities. For these community-level measures, the ratio among the number of ponds (N = 16) and the number of potential explanatory variables (11) did not enable proper modelling. Thus, we had to explore possible relationships by separately running regressions of each dispersion measure on each explanatory variable, posteriorly correcting p values using the false discovery rate (FDR, Benjamini and Hochberg 1995). All statistical analyses were performed in R (R Core Team 2015).

Results

The mean bivariate isotopic signature differed between the native and invasive species in all the ponds, with a few exceptions when isotopic signatures were tested individually (Table 1). In general, D. pictus showed higher δ15N and lower δ13C values. By contrast, interspecific differences in trophic niche width were only detected in one pond out of 16 (Table 1 and Table S2).

Table 1 Mean δ13C and δ15N values, plus the corrected standard ellipse area (SEAc) for each pond and species and the p values for interspecific differences
Full size table

The first variable defining the animal communities (An1) correlated negatively with the presence of most groups, but positively with the presence of cladocerans and ostracods (Table S3), thus creating an axis that ranged from small filtering animal communities to more complex assemblages containing more predators (Boix et al. 2004). The community-level indicators for the other two variables (An2 and An3) were unclear. The first variable defining the macroscopic vegetal communities (Veg1) correlated negatively with all items, creating an axis that demonstrated vegetal diversity. The second axis (Veg2) provided similar information, but positively correlated with the presence of chlorophytes (Table S4).

The best model for δ15N included the distance from the origin of invasion plus its interaction with species, Veg1 without any interactions, and Veg2 plus its interaction with species (Table S5). Several of the best ten models included other terms, namely tadpole density, species, pond depth and surface, and An1 plus its interaction with species, which were tested using LRTs. The interaction of species with the distance from the origin of invasion, Veg2 and An1 were highly significant (distance*species: LRT = 9.02, p = 0.003; Veg2*species: LRT = 48.46, p < 0.001; An1*species: LRT = 40.98, p < 0.001), as were its interaction with the dimensions of the ponds (pond depth: LRT = 3.85, p = 0.049; pond surface: LRT = 4.79, p = 0.029). Tadpole density (LRT = 5.53, p = 0.137) and Veg1 (LRT = 2.46, p = 0.119) did not give significant results. In the case of Veg1, although the term was included in the best model, it was absent from the others in the best 10 models. We concluded that the factors affecting the δ15N values of tadpoles were: (1) the interaction between each species and Veg2, with the δ15N values decreasing for D. pictus and increasing for E. calamita when chlorophyta dominated the vegetal community (Fig. S1); (2) the interaction between each species and An1, with the δ15N values increasing for D. pictus and decreasing for E. calamita when the animal community was mainly dominated by Cladocera and Ostracoda (Fig. S2); and (3) the time since the invasion of the alien species, with both species displaying segregation in their δ15N values in the expansion front, but hardly in the origin area (Fig. 2 and Fig. S3). Values of δ15N of both species increased with the depth of the pond or decreasing vegetation diversity, and slightly decreased with increasing surfaces of the ponds.

Fig. 2
figure 2

Isotopic segregation between the two species according to the distance from the origin of the invasion: a δ15N and the regression lines for each species; b δ13C and the regression lines for each species; and c the distance between bivariate centroids for each studied pond and regression line (not significant)

Full size image

The best LMM for δ13C included the proportion of the pond under direct sunlight, Veg1, and their interactions with species. The second best model was similar, but included species plus its interaction with distance instead of sunlight and its interaction (Table S6). The remaining models were weakly supported (ΔAICc > 10), but pond depth plus its interaction with species appeared in several of them (Table S6) and were also tested using LRTs. All interactions, tested against a model including species, distance from the origin of invasion, proportion of the pond under direct sunshine, pond depth and Veg1, were significant (species*distance: LRT = 9.85, p = 0.002; species*sunshine: LRT = 17.70, p < 0.001; species*Veg1: LRT = 14.90, p < 0.001), except for the interaction between species and pond depth (LRT = 3.41, p = 0.065). Pond depth gave no significant results (LRT = 2.19, p = 0.139). We concluded that the factors determining the δ13C values of tadpoles were: (1) the interaction between each species and the effects of sunshine, with direct sunlight increasing δ13C values, especially in E. calamita (Fig. S4); (2) the interaction between each species and Veg1, with the mean δ13C values decreasing and interspecific segregation increasing with greater vegetal diversity (Fig. S5); and (3) the time since the invasion of the alien species, with segregation in the δ13C values being more marked when more time had elapsed (Fig. S3). The remaining variables and interactions did not have statistical significance.

Two ponds located at opposite extremes of the gradient of coexistence—one from the origin area and one from the expansion front—gathered all significant differences in angles of segregation among ponds (this includes the comparison between the angles of segregation of these two ponds, Table S7). Segregation of the two species in one of the ponds from the origin area was different from that observed in two ponds in the expansion front and two in the far expansion, while segregation in one of the ponds in the expansion front was different from that observed in one of the ponds in each of the other three study areas (Table S7). Thus, we never found significant differences among ponds from the same geographic area. Between the areas, changes in the angles between the segregation directions were not statistically significant (Fig. 3, Table S8).

Fig. 3
figure 3

Trophic position of the native tadpoles (E. calamita) in relation to the invasive tadpoles (D. pictus) in each pond. Each dot represents the interspecific difference between the two bivariate centroids in a pond and is coloured according to the study area that the pond belongs to. The arrow follows the means for each study area, from the expansion front (native species are naïve to the invasive species) to the origin area (approximately 110 years of coexistence). Vertical and horizontal lines at zero indicate no differences between the native and invasive species

Full size image

The few instances in which regressions of dispersion measures were significant disappeared after FDR correction (Table S9). Both the MNND and the range of δ13C values of the tadpoles (dX_range) increased slightly when the ponds were deeper, the vegetation diversity was greater and more time had elapsed since invasion. The SEAc of E. calamita tended to increase slightly with increasing vegetal diversity.

Discussion

Previous studies using SIA have already listed several factors that affect the trophic ecology of tadpole guilds (Whiles et al. 2006; Schiesari et al. 2009; Caut et al. 2013; Arribas et al. 2015; San Sebastián et al. 2015a). We aimed to explore possible geographical variations in interspecific competition by studying the same competitive relationship across populations with different co-evolutionary histories. As expected, trophic segregation was observed in ponds where the two tadpole species co-occurred (Table 1, San Sebastián et al. 2015a). However, the nature of this segregation changed between populations and ponds in predictable ways. We found several ecological factors that promoted or obstructed segregation, with segregation occurring more in the carbon isotopic signature (δ13C) than in the nitrogen one (δ15N) when more time had elapsed since the invasion (Figs. 1, 2).

The δ15N signature is used to infer trophic level because it has high isotopic fractionation, its values increasing with higher trophic levels (Minagawa and Wada 1984; Ambrose and DeNiro 1986). We noted great variation in the δ15N values of the native and invasive tadpole species between areas and ponds (Table 1). However, matching variations also occurred in the δ15N values of accompanying species (Figs. S6 to S9). Thus, after correcting for fractionation, comparisons of the δ15N values of the tadpoles with those of other items in the same ponds revealed that both tadpole species were low-level consumers (Diaz-Paniagua 1985; Campeny 2001; Caut et al. 2013; Arribas et al. 2014). This implies the marked changes in the δ15N of tadpoles between areas and ponds are a consequence of the natural variation in isotopic baselines occurring along time and space (Syväranta et al. 2006; Woodland et al. 2012). We detected significant differences between the δ15N values of the two species in 14 of the 16 ponds studied, with the native species occupying a higher trophic level in only three of them (Fig. 3, Table 1). The tendency of the native species to occupy lower trophic levels (San Sebastian et al. 2015a) was still present after incorporating ecological variables and the time since invasion, which also greatly affected the trophic level. For instance, both species increased their trophic level when vegetation diversity was reduced. However, variations in the composition of the vegetal community affected each species differently (Figs. S1 and S5), indicating different preferences for food items between the two species (San Sebastián et al. 2015b). Differences in the trophic level of the two species were more marked when the accompanying animal community contained only small filtering animals and no predators (Fig. S2), probably due to interspecific differences in the effects of predation risk (Pujol-Buxó et al. 2017). Most interestingly, however, the overlap in trophic level between the native and invasive species was significantly higher when more time had elapsed since the invasion (Fig. 2), contradicting our hypotheses.

The carbon isotopic signature (δ13C) has been traditionally used to indicate the origin of the consumed organic matter because it has low fractionation values and there are differences in carbon assimilation between vegetation typologies and habitats (Tieszen et al. 1983; Ambrose and DeNiro 1986; Griffith 1992; Koch et al. 1995; Gannes et al. 1997). We found great variation in the δ13C values, mainly between the ponds and study areas (Figs. S6 to S9), which might have been due to changes in the carbon resources available, or spatial and temporal changes in isotopic baselines as well (Syväranta et al. 2006; Woodland et al. 2012). We observed differences between the mean δ13C values of the two species in 14 of the 16 ponds, the native species showing lower values in only two cases (Table 1). This is consistent with the results of San Sebastian et al. (2015a). Interestingly, the interspecific difference in the δ13C values was significantly altered by several factors. For instance, interspecific δ13C segregation increased with greater vegetal diversity (Fig. S5). Moreover, all δ13C values increased with greater sunlight and displayed larger interspecific differences (Fig. S4), possibly due to changes in the microscopic algal communities (which we did not study). Finally, the native and invasive species showed more segregation in their δ13C values when more time had elapsed since the invasion (Fig. 2), which fits our hypothesis.

Therefore, with an increasing amount of time elapsing since the invasion, differences in the items consumed became larger, while differences in the trophic level became smaller (Figs. 1, 2), resulting in a very slight increase in the magnitude of the trophic segregation between the native and invasive species. Although these geographical patterns were always the same regardless of the analyses performed and the ecological factors considered, significant results were only detected in models in which each tadpole was used as a data unit (N = 320). A larger number of ponds could be studied to clearly determine the extent to which the changes here detected are biologically important regardless of its statistical significance.

Interspecific differences in the dispersion measures were infrequent and lacked a common direction or tendency (Table 1 and Table S2). We hypothesised that the trophic niches had to be wider when trophic overlaps were greater (Codron et al. 2011; Abbey-Lee et al. 2013; San Sebastian et al. 2015a), expecting this to occur in the expansion front. However, we did not find any clear changes in the trophic widths, possibly because an increasing amount of time since the invasion had affected the nature, but not magnitude (i.e., the bivariate distance) of the segregation. Our data did not support the hypothesis of San Sebastian et al. (2015a) that the invasive species has a more flexible trophic width than E. calamita because only the native species slightly increased its trophic width when vegetation diversity increased. The trophic width of the native species might be modified in the presence of the invasive competitor (San Sebastián et al. 2015a); however, we did not observe this change during the invasion. Thus, based on our observations (Fig. 1e–h), evolutionary changes in the competitive interaction between these native and invasive species do not act on niche width or the magnitude of mean differences in the diet, but on how they segregate in terms of their trophic level and items consumed.