The study was performed in El Cobre stream, located in the Valparaiso administrative region of Chile and forms part of the Río Aconcagua watershed (33°34.027'S, -70°37.896'W; 623 m elevation). The stream has an intermediate flow, depending on winter precipitation, although some pools persist during the dry season. The high part of the system is drained by the El Gallo and El Sauce streams, in whose watersheds are located “El Soldado” and “Los Navíos” copper mines. The vegetation is dominated by spiny and sclerophyllous shrubs typical of the semi-arid Mediterranean region of Chile (Di Castri 1968). This water system is one of the few that retains surface water during the dry season (summer), and it is common for it to flow for approximately 5 km before infiltrating the subsoil. It connects with the rest of the hydrological network only in heavy rainfall. In October 2021 (austral spring), two transects of 100 m length and 10 m width in the stream were performed, separated by 3400 m (called sites CO-1 and CO-2). Fish and anurans were captured at both sites using a 12 V battery-operated SAMUS 725 MD electrofisher unit (600 W), which generates a DC pulse of up to 1000 V, with a current intensity of 20 to 60 A, and a pulse duration of 10 seconds (Pottier et al. 2020). Captured individuals were examined for visible injuries prior to release, confirming that the procedure did not harm the collected animals. Individuals were identified by species – fish according to Dyer (2000) and amphibians following Correa et al. (2011) – and weighed to the nearest gram. Calipers were used to measure the snout ventral length (SVL) of amphibians and total length of fish. We collected aquatic invertebrates with a Surber net, sampling an area of 0.09 m² with three replicates at each site (Ramírez 2010). We sampled the native macrophyte Myriophyllum aquaticum, the only species in CO-1 (no macrophytes were found at CO-2). To analyze heavy metals in the environment, surface sediments (a compound sample from 10 sampling points in each site, taken from the top 2 cm) and surface water from the stream (a compound sample of 3 sampling points at each site) were examined. Additionally, the results of 15 previous sampling sessions conducted between 2018 and 2022 are reported here to assess the historical diversity of aquatic vertebrates in this stream. These campaigns were part of the Chilean Environmental Assessment System.

Ten individuals of X. laevis (all from site CO-1) were collected and used for diet analysis (stomach contents); animals were euthanized using immersion in a 0.2% tricaine methanesulfonate (MS-222) bath. Stomach contents were identified under a stereoscopic microscope to the minimum possible taxonomic level (Fernández and Domínguez 2001). The length and width (mm) of each prey item were measured to estimate the volume and percentage contribution of each prey taxon (Barreto-Lima 2009). Other estimations included the percentage of each prey category, i.e., the proportion (%N) of each prey of the total prey found. The prey percentage, frequency, and volume data were integrated into a relative importance index of prey RII (Pinkas et al. 1971).

Two samples of M. aquaticum were used to analyze the isotopes of macrophytes. For invertebrates, a pooled compound sample of each taxon was used due to the small size of these organisms after removing hard tissues (mollusk shells, exoskeletons). For fish and amphibians (adults only), a 4 mg sample of muscle tissue (dorsal muscle in fish and thigh muscle in anurans) was obtained. Samples were ground, dried, and lipids were extracted in petroleum ether using a Soxhlet extractor (Lobos et al. 2022). The isotopic composition of carbon and nitrogen (δ¹³C and δ¹⁵N) was determined with a continuous flow isotope ratio mass spectrometer (Thermo Finnigan Delta V Advantage) coupled to a Carlo Erba NC 2500 elemental analyzer via continuous flow in the Laboratory of Biogeochemistry and Applied Stable Isotopes (LABASI) at the Pontifical Catholic University of Chile. Stable isotope ratios are reported in δ notation, expressed as deviations from international standards: Pee Dee Belemnite for δ¹³C and atmospheric N₂ for δ¹⁵N. Values were normalized to internal standards calibrated against International Atomic Energy Agency (IAEA) reference materials (IAEA-N1, IAEA-N2, and IAEA-NO₃ for nitrogen; NBS22, CH6, and NBS19 for carbon, respectively). The ratio of ¹³C/¹²C and ¹⁵N/¹⁴N was expressed as relative difference per thousand (‰) using the equation:

δ13Corδ15 N=RSample RStandard -1·103 Equation 1

R_sample and R_standard are the corresponding ratios of heavy to light isotopes (¹³C/¹²C and ¹⁵N/¹⁴N) in the sample and the standard, respectively. Typical precision of the analyses was ± 0.23‰ for δ¹⁵N and 0.25‰ for δ¹³C.

To perform niche comparisons between X. laevis and the collected fishes, analyses were conducted using R (SIBER package; Jackson et al. 2011) to obtain four quantitative Bayesian metrics (mean and 95% confidence intervals) adapted from Layman et al. (2007): nitrogen range (NR), carbon range (CR), Euclidean distance of each individual to the centroid (CD), and standard deviation of the nearest neighbor distance (SDNND). A fifth metric was the corrected standard ellipse area (SEAc expressed in ‰²). SEAs are comparable to the univariate SD and contain ca. 40% of the data, providing a better description of the isotope niche of a population. For a posteriori comparison of the ellipses, the Bayesian standard ellipse areas (SEAb) were estimated according to Jackson et al. (2011). Then, the proportions of prey that contributed to the assimilated diet of X. laevis were determined using the SIAR mixing model (Parnell and Jackson 2013). To minimize the parameters to be estimated in the model (Phillips et al. 2014), prey were grouped into three trophic groups: carnivores, detritivores, and herbivores, an approach previously used by Molina-Burgos et al. (2018). In this study, these three groups are isotopically distinct in δ¹⁵N (ANOVA, F_2,5 = 9.43; p = 0.04), but not in δ¹³C (F_2,5 = 1.81; p = 0.31). The parameters of the model were estimated with 300,000 Markov chain iterations, and 1.3 ± 0.3 δ¹³C and 2.3 ± 0.18 δ¹⁵N were used as the trophic enrichment factors (McCutchan et al. 2003).

The trophic level of organisms was estimated as follows:

NTorganism =NTbase +δ15 NConsumer -δ15 NBase TDF Equation 2 (Post 2002).

Where NT is the trophic position n of an organism, δ¹⁵N consumer is the value of the organism evaluated, and δ¹⁵N base is the nitrogen value of a basal organism (in this case the invertebrate with the lowest position; Baetidae). It was assumed that the NT base in this case occupies position 2 (over the producers of the trophic web). For the trophic discrimination factor (TDF), the reference value 2.3‰ was used (McCutchan et al. 2003).

Cu, Mn, As, and Zn levels were evaluated in sediments, water, and aquatic biota. For the aquatic biota, a variable number of replicates for each species collected was obtained (Appendix 1: Table A1); these were homogenized to obtain a humid paste of 0.5–1 mg. Samples were dried (60 °C for 48 h) before metal analysis. The US EPA 3050B protocol (US EPA 1996) was used, which involves repeated digestion with HNO₃ and hydrogen peroxide (H₂O₂) to recover the metals present in the organic fraction (where metals are bioavailable). Metal concentrations for biological samples and sediments were determined using an atomic absorption spectrophotometer (ICP-OES) by the flame technique (Thermo Scientific X series 2, at the Laboratory of Agronomy and Natural Systems of the Pontifical Catholic University of Chile). The analytic procedure was verified using the Dorm-3 reference material (biological samples) and Mess-3 (sediments) obtained from the National Research Council Canada (NRC). The analytical error in both cases was less than 5%. Concentrations were expressed as mg Kg^-1 of dry weight. The reference values used for sediments were those of the “interim sediment quality guideline” (ISQG) and “probable effect level” (PEL); the former is the concentration below which adverse effects are not expected, while the latter is the concentration over which adverse biological effects are expected to appear (CEQG 2003). The levels of heavy metals in water samples (mg L⁻¹) were measured after storage in cold conditions, followed by laboratory analysis using the flame technique (APHA 2017).

To integrate the information on isotopes and metals, the biomagnification factor (BMF) for X. laevis and the bioaccumulation factor (BAF) for the community (Dehn et al. 2006) were calculated as follows:

BMF=CPredator CPrey δ15 NPredator δ15 NPrey  Equation (3)

BAF=CBiota CSediment  Equation (4)

Where C_Predator and C_Prey are the concentrations of metals in the predator and the prey, respectively, in mg K^-1, and the δ¹⁵N_Predator and δ¹⁵N_Prey are the ‰ of the ¹⁵N content, respectively. Values of BMF greater than 1 are interpreted as a signal of a possible biomagnification process of the metal. In contrast, values lower than 1 are considered as biodilution of the metal in the predator. In the case of BAF, C_Biota and C_Sediment represent the metal concentrations in biota and sediment in mg K^-1. BAF values greater than 1 could be interpreted as a signal of bioaccumulation (Dehn et al. 2006; Mortuza and Al-Misned 2015).

Three fish species (Cnesterodon decenmaculatus invasive; Basilichthys microlepidotus and Cheirodon pisciculus native) and one invasive amphibian (Xenopus laevis) were collected from the two sampling sites (Table 1). Sixteen aquatic invertebrate taxa were found, 13 of which (81%) were of the class Insect. Richness was 14 taxa for CO-1 and 7 for CO-2 (Table 2). The most represented families were Chironomidae (larvae, 46.63% in CO-1 and 92.07% in CO-2), Coenagrionidae nymphs (15.4% in CO-1), and Physidae (14.9%) in CO-1.

The diet of X. laevis was composed of 41 prey items from seven taxa, including invertebrates (insects and mollusks) and fish (Table 3). The relative importance index RII shows that the diet was mainly snails of the genus Physa (44.78%), dragonfly nymphs of Coenagrionidae (19.06%), and the aquatic coleopteraDytiscidae (17.16%). The native fish (Cheirodon pisciculus) was important in the percentage of volume consumed (28.50%; due to their size), but less important in terms of its low RII value (present in only two of the 10 frogs analyzed).

The isotope comparison of the community analyzed is shown in Fig. 1 and Table 4. Nitrogen varied from a basal value of 2.60 (macrophyte) to 13.52 (the fish B. microlepidotus). Carbon ranged from -16.36 (B. microlepidotus) to -32.46 (Hydrophilidae). The trophic grouping of invertebrates is well supported in terms of isotope ratios for δ¹⁵N (ANOVA, F_2,5 = 9.43; p = 0.04), not for δ¹³C (F_2,5 = 1.81; p = 0.31). There were at least five trophic levels; the highest of which were occupied by the native fish (C. pisciculus, B. microlepidotus) and the invasive vertebrates (X. laevis, C. decenmaculatus).

The nitrogen range in the community was 1.86‰ (Cis: 1.84–1.88), and the carbon range was 13.64‰ (Cis: 13.62–13.65). The trophic diversity, measured as the mean distance to the centroid CD, was 4.87‰ (Cis: 4.86–4.87), and the trophic evenness SDNND was 3.27‰ (Cis: 3.25–3.28). SEAc was different between species evaluated (SEAc C. decenmaculatus = 1.02‰², SEAc B. microlepidotus = 0.79‰², SEAc C. pisciculus = 4.49‰², SEAc X. laevis = 6.10‰²), and position in the δ ¹³C-δ ¹⁵N biplot (Fig. 2). The only overlap seen between vertebrate species was between C. decenmaculatus and C. pisciculus at 4.61%. The probability that C. decenmaculatus had lower SEAb than B. microlepidotus was 42% and was 99% for C. pisciculus and X. laevis, respectively. The likelihood that B. microlepidotus had lower SEAb than C. pisciculus and X. laevis was 99%; the probability that C. pisciculus had lower SEAb than X. laevis was 73%. Despite the similarity in carbon values between food resource groups (herbivores, detritivores and carnivores), the distinct nitrogen values between groups permitted the determination that the greatest dietary contribution of prey by X. laevis was detritivore invertebrates (62.36±2.96%), followed by carnivorous organisms (36.54±2.82) and herbivorous invertebrates (1.1±0.99) (Fig. 3).

The concentrations of heavy metals varied among the taxa analyzed (Appendix 1: Table A1). All the metals present in the sediments of the El Cobre stream were above the ISQG and PEL reference norms. In contrast, the metal concentrations in water had values under the detection limits. As shown in Fig. 4, all the analyzed elements, except Zn, decreased as the trophic level of the organisms increased (except the Baetidae).

Biomagnification was found for Cu and Zn in X. laevis (Appendix 1: Table A2). The increase in Cu was noted for the prey C. pisciculus, and in Zn for the Physidae and the odonate Coenagrionidae. Only Baetidae was near the threshold (≥1) of the bioaccumulation factor (Appendix 1: Table A3) for Zn.

This study was financed by a fund for freshwater studies of Anglo American. The funders had no role in study design, data collection and analysis, decision to publish or preparation of the manuscript.

GL, CG and JS conceived the idea; GT, AA, JT, NR, HS and GL collected most data; GT, CG and VG organized the dataset and analyses; GL created the figures; GL and CG wrote the first draft.

This project was performed with number permit 448 Chilean Fishing and Aquiculture Subsecretary. All research pertaining to this article did not require any ethics committee approval.

The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.