https://orcid.org/0000-0002-0535-7936
Figures
Abstract
High mountain lakes (HMLs) are considered unique and comparable ecosystems for monitoring global climate change. The food web structure can indicate the response of these ecosystems to ecological threats, such as fish introduction, by analyzing the trophic dynamics. Nonetheless, the food webs of tropical HMLs are less well-studied than temperate HMLs. The present study assessed the food webs of two neighboring (600 m apart) tropical HMLs, El Sol and La Luna, inside the crater of the Nevado de Toluca volcano, Mexico. It used stable isotopes (δ13C and δ15N) and Bayesian mixing models with different trophic discrimination factors and priors to assess the impacts of introduced rainbow trout, persisting only in the larger lake, El Sol. The food web in Lake El Sol was more complex than in Lake La Luna, mainly due to its larger size, extensive vegetated littoral zone, and being fueled by autochthonous primary production. In contrast, the smaller and fishless Lake La Luna has a reduced and bare littoral zone that harbored a simple food web substantially sustained by allochthonous carbon inputs. The persistence of introduced rainbow trout in Lake El Sol but not in Lake La Luna accentuated the differences between the lakes. The models suggested that rainbow trout fed on key consumers of littoral macroinvertebrates (70–80%) and pelagic zooplankton (20–30%), increasing the linkage between sub-networks. In both tropical HMLs, the species richness and herbivorous fraction were elevated compared with temperate HMLs, while the linkage density and omnivorous fraction were lower. Basal nodes dominated these tropical HMLs, and the vegetated littoral zone of Lake El Sol had more intermediate (omnivore) nodes. Our results showed the convenience of food web analysis to compare the effects of introduced fish in originally fishless lakes in different latitudes.
Citation: Jiménez-Seinos JL, Alcocer J, Planas D (2023) Food web differences between two neighboring tropical high mountain lakes and the influence of introducing a new top predator. PLoS ONE 18(6): e0287066. https://doi.org/10.1371/journal.pone.0287066
Editor: Frank O. Masese, University of Eldoret, KENYA
Received: January 8, 2023; Accepted: May 30, 2023; Published: June 13, 2023
Copyright: © 2023 Jiménez-Seinos et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: All relevant data are within the manuscript and its Supplementary Information.
Funding: The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. JA - 262970 - Fondo Sectorial de Investigación Ambiental SEMARNAT-CONACYT JA - IN105009, ES209301 - Universidad Nacional Autónoma de México/DGAPA/PAPIIT JA - PINCC 2020-2021 - Universidad Nacional Autónoma de México/Programa de Investigación en Cambio Climático.
Competing interests: The authors have declared that no competing interests exist.
Introduction
High mountain lakes (HMLs) are the highest water bodies in the world, located above the tree line, at more than 3000–4000 m a.s.l. at tropical and subtropical latitudes [1, 2]. The high altitude results in unique and harsh environmental features such as a high incidence of solar and UV radiation, oligotrophy, isolation, and marked habitat heterogeneity, leading to different patterns in local diversity among HMLs within the same district and the adaptations of organisms to altitude [1–3].
Although HMLs are generally remote, human activities also threaten these environments through air-borne pollution from local and long-range transport, pollution within their drainage basins, climate change, and the stocking of exotic fish in otherwise fishless lakes [2]. Fish stocking strongly influences food web patterns of inland water bodies, including HMLs, and is an important threat to, for example, endemic species [4]. Effects on food webs differ according to local conditions: food web topology may be simplified by a reduction in the diversity of top predators and their consumption of local species [4, 5] or may be made more complex by the addition of new trophic links if the introduced fish is an omnivore [6].
An introduced fish species could increase trophic interactions between littoral and pelagic food webs in a lake with a vegetated littoral shore by directly consuming benthic organisms [7, 8]. Concurrently, macrophytes and macroalgae provide habitat for primary and secondary producers, food for benthic invertebrates, and shelter for zooplankton [9, 10], promoting a foraging habitat for fish, especially in lakes with extended and gently sloped littoral zones [11]. The reliance of fish on zoobenthos is higher in small lakes where the littoral zone is the dominant habitat [8, 12].
The introduction of rainbow trout into natural fishless lakes creates a "cascade effect" in the pelagic food web by consuming herbivorous zooplankton, which reduces grazing pressure on algae and increases populations of inedible phytoplankton [4, 13]. Moreover, fish stocking adversely affects the ecological characteristics of herbivorous prey such as Daphnia (e.g., their size, phenology, morphology) [12, 14].
Analysis of food webs in HML ecosystems has used gut contents of consumers, meta-analyses of published literature, and, rarely, stable isotope analysis (e.g., [15–17]), but little of this information concerns tropical environments. More recently, analysis of tracer isotopes (δ13C and δ15N) and mixing models have been widely used to identify diet sources of consumers in natural environments [17–19], as well as autochthonous or allochthonous supplies [20], isotopic fractionation [21, 22], food chain length, and trophic position [23].
Mixing models based on a probabilistic Bayesian framework provide general explanatory information about the proportional contribution of different sources to a mixture. These models analyze the source and trophic discrimination factor (TDF) variability, the population structure of the consumer, and error assumptions [24–27]. Mixing models are sensitive to variation in TDF. Therefore, an incorrect TDF can lead to erroneous results in mixing models and a misunderstanding of trophic relationships between consumers and sources [28].
This study used a methodology that combines stable isotopes and ecological data to construct food webs. We aimed to understand the littoral and pelagic food web structure of two adjacent oligotrophic tropical HMLs with a similar origin, geology, oxygen concentration, and temperature but different community structure and the influence of fish introduction. Through visual prey selection, we hypothesize that rainbow trout introduction alters pelagic and littoral food webs in Lake El Sol (from now on, El Sol). In addition, the trout feed alternately on littoral and pelagic food sources and promote coupling between pelagic and littoral zones. In contrast, the homogeneous steep and naked littoral zone in Lake La Luna (after this La Luna) should limit the abundance and diversity of food sources, reflected in a simple food web with few nodes and trophic interactions and a shortened food chain.
Materials and methods
Study site
El Sol and La Luna are tropical HMLs within the Nevado de Toluca volcano crater (19°06’ N, 99°45’ W) at 4200 m a.s.l. (Fig 1). The Nevado de Toluca is the fourth-highest volcano in Mexico within the Trans-Mexican Volcanic Belt, the principal and highest volcanic mountain range in Mexico [29]. It has an alpine climate (Ew) [30, 31]. Despite their proximity (600 m) and the common origin and climate, the two lakes differ in morphometry, water chemistry [32], thermal regime, and, markedly, in their ecological communities [33–36]. In the 1950s, the Mexican government introduced rainbow trout (Oncorhynchus mykiss) into both lakes, but the species persisted only in El Sol [37]. Both lakes have marked differences in morphometry; El Sol is larger (surface: 19 ha versus 3 ha) and deeper (maximum depth: 12 m versus 10 m) with an extensive shallow, heterogeneous littoral zone, including vegetated and rock-dominated areas. La Luna has a small homogeneous steep non-vegetated littoral zone (bucket-shaped) with a small shore development. Both lakes are warm polymictic, discontinuous in El Sol and continuous in La Luna [38], and have no winter ice cover. Physicochemical and morphometric variables are summarized in (S1 Table).
Tropical high-mountain Lakes El Sol (front) and La Luna (back) inside the Nevado de Toluca volcano crater, Central Mexico. (Map data set obtained from open access topographic and toponymic vector data set, scale 1:100,000, National Institute of Statistics and Geography, Mexico at https://www.inegi.org.mx/app/mapas/).
Sampling
We used vertical hauls to collect phytoplankton (20 μm mesh) and zooplankton (70 μm mesh). Samples were sieved through 100 μm mesh to remove large zooplankton and retain filamentous algae. We removed the small zooplankton and macroalgae (≤100 μm) from the samples with a fine pipette and stereomicroscope. Isolated phytoplankton samples were then collected on pre-combusted GF/A glass fiber filters [39, 40]. Littoral benthic macroinvertebrates (BMIs) were collected with an acrylic corer (2.4 x 2.5 cm) at the vegetated and bare littoral zones and sieved through a 500 μm mesh. Using a stereomicroscope, we separated the dominant zooplankton and BMI species with a 500 μm pipette and entomology forceps. A sample of total zooplankton was also included in determining the isotopic concentration of bulk zooplankton (>70 μm). We maintained zooplankton and BMIs in pre-filtered lake water for 2 hours to allow gut evacuation [15, 40]. We collected macrophytes and macro-algae by hand, stored them in whirl-pack sample bags, and rinsed them three times with pre-filtered water. Epiphytes colonized macrophytes; then, we considered the macrophyte-epiphytes complex in the analysis. Rainbow trout were caught with a fishing rod, stored on the ice at 4°C, and frozen in the laboratory after sampling. For isotope analysis, we removed the transverse white dorsal muscle sample without skin or bones [41].
Statistical and stable isotope analysis
To calculate the isotope concentration (δ13C and δ15N) and the C:N ratio, all biological samples were frozen at -20°C, freeze-dried, and ground to a fine powder with an agate mortar and pestle. Samples were then analyzed in an IRMS (Isotopic Ratio Mass Spectrometer Isoprime 100TM, CF-EA) at the GEOTOP laboratory (Montreal, Canada) using two international references, VPDB for carbon and AIR for nitrogen. For δ13C isotopes, two internal reference materials (δ13C = -28.73±0.06‰ & -11.85±0.04‰) were used to normalize the results on the NBS19-LSVEC scale. A third reference material (δ13C = -17,04±0,11‰) was analyzed as an unknown to assess the exactness of the normalization. Results are given in delta units (δ) in ‰ vs. VPDB. The overall analytical uncertainty (1s) is better than ±0.1‰. For nitrogen isotopes (δ15N), two internal reference materials (δ15N = -0.10±0.24‰ & +14.95±0.09‰) were used to normalize the results on the AIR (IAEA-N1, IAEA-N2 & IAEA-N3) scale. A third reference material (δ15N = -0,1±0,15‰) was analyzed as an unknown to assess the exactness of the normalization. Results are given in delta units (δ) in ‰ vs. AIR. The overall analytical uncertainty (1s) is better than ±0.2‰.
We compared the isotopic values of the two lakes by the Mann-Whitney U test since the Shapiro-Wilk test showed that the δ13C of El Sol and the δ15N of La Luna had a non-normal distribution. The skewed δ15N distribution in La Luna was caused solely by the extremely low value of the dominant Zygnemataceae macroalga Temnogametum iztacalense. We also performed a dissimilarity test (ANOSIM) to compare the isotopic values of both lakes and zones (littoral and pelagic), considering δ13C and δ15N values. We analyzed the relationship between rainbow trout (adults: 33.48 ± 2.7 cm) weight (independent variable) and isotope values (dependent variable) with linear regression models.
We used a two-source food web equation to calculate the food chain length using the δ15N values of top consumers and baselines [23]. In El Sol, we used rainbow trout as the top consumer, the copepod Leptodiaptomus cuauhtemoci in the pelagic zone, and the gastropod Physa sp. for the littoral as a baseline. In La Luna, we used the cladoceran Daphnia ambigua as the top consumer and phytoplankton as the baseline. We calculated the trophic level of primary consumers from their average trophic discrimination factor (TDF) and the average δ15N value of their sources as baseline [23]. For the stocked lake, we used the average difference between δ13C and δ15N values of rainbow trout and its potential food sources to calculate the TDF [18, 22, 42]. We estimated the relationship between food sources and TDF with linear regressions, using δ13C and δ15N values of diet sources as independent variables and carbon and nitrogen TDFs as dependent variables [21].
Bayesian model construction for introduced rainbow trout diet
We constructed Bayesian mixing models (BMMs) with the MixSIAR R package using δ13C and δ15N as isotopic tracers to estimate the proportional contribution of potential food sources to the trout diet [25, 43]. We assumed a random effect for the consumer and did not account for population structure (differences in feeding behavior due to hierarchical organization, sex, or size) because all fish collected were adults capable of moving around the lake to feed. In addition, a multiplicative error was used to account for sampling bias and differences inherent in the predation process [27]. To measure the effect of TDF on food source contribution and model accuracy, we used four different published TDFs and one calculated TDF for δ13C and δ15N [15, 44–46]. We used two different priors: the first or generalist prior considered the same probability for fish to feed on all sources and was constructed with a uniform Dirichlet distribution using a hyperparameter alpha (α = 1) and divided by the number of sources (∑α) [25]; the second or biomass prior considered a differing probability of food sources to be eaten depending on their biomass within El Sol. We standardized the average annual biomass in mg C m- 2. For zooplankton biomass, we took the reported average biomass [35] and multiplied it by the zooplankton carbon percentage measured in this study. For benthic organisms, we used the reported biomass directly [47], except for gastropods, because this is the first study that sampled and reported gastropods (Physa sp.) in El Sol; for those, we used an average biomass value to calculate a neutral hyperparameter α and to avoid further bias in their contribution to the rainbow trout diet. Biomass values were standardized by the total number of sources (∑α) divided by the total biomass to build a less aggressive prior distribution and maintain the same mean (Fig 2) [25, 48].
Standardized Dirichlet distribution of macroinvertebrates. Source biomass was used as an informative prior distribution of the six possible rainbow trout diet sources within the littoral zone of Lake El Sol. Redlines show the median value of the distribution. (Dap = Daphnia ambigua, Gas = Phys sp., Lep = Leptodiaptomus cuauhtemoci, OliH = Limnodrilus hoffmeisteri, OliT = Tubifex tubifex, OliV = Lumbriculus variegatus).
We calculated statistical significance with the Gelman-Rubin convergence test; this analyzes Markov Chain Monte Carlo (MCMC) tested by the Bayesian method (JAGS) until the value is <1.1 and MCMC chains converge or where all desired posterior distributions have been already tested and are closer to 1. We also estimated convergence with the Gelweke test, which compared the means between the first (10%) and second (50%) parts of MCMC chains. We assumed convergence if their means were significantly similar [24, 25]. We used the widely applicable information criterion (WAIC) and leave-one-out (LOO) cross-validation to test out-of-sample prediction accuracy [49]. Finally, we validated the Bayesian mixing models using the in-polygon assumption, which was developed to determine whether a proposed mixing model is logical and includes all consumers within a Bayesian simulation of mixing polygons constructed with the isotopic values of food sources [50].
Community structure
We analyzed the community structure of each lake, using functional groups and trophic guilds to assign each taxon into ecological clusters at different trophic levels [51, 52]. We based a literature search on global databases and previous studies of the two lakes on proposing potential trophic interactions for fish, zooplankton, primary producers, and benthic macroinvertebrate taxa (S2 Table). We also added trophic links using Bayesian mixing models (rainbow trout) and stable isotope similarity between analyzed groups and species (primary consumers and detritivores). We assigned the trophic level of each node using the calculated food chain length and trophic level [23].
We calculated the food web topology and metrics with the Cheddar R package [43, 53] using ecologic groups as nodes. We proposed trophic links within a n x n matrix with predators as columns and prey as rows, where 1 means a trophic interaction and 0 a lack of it [6]. Finally, we used food web metrics to compare El Sol and La Luna lakes with each other and with temperate HMLs [6].
Results
Stable isotope analyses and trophic structure
El Sol had a more complex food web, where benthic consumers had higher δ13C values than pelagic ones (W = 125, p<0.001), and δ15N values were higher for pelagic consumers (W = 0, p<0.001), rainbow trout was in the middle of both zones, indicating a coupling role between them (Fig 3A). Differences between the δ13C values of the two oligochaetes in fish stocked El Sol indicated different food sources: L. hoffmeisteri (-22.57 ± 0.59‰) was associated with sediment organic matter (-21.55 ± 2.23‰) and L. variegatus (-12.35 ± 1.07‰) was associated with macrophytes with C4 metabolism (-11.17‰) and filamentous macroalgae (-10.32 ± 0.3‰) (Fig 3A). Among the pelagic components in El Sol, δ13C values of phytoplankton (-18.25 ± 3.17‰) and D. ambigua (-21.05 ± 0.1‰) showed a higher similarity compared with L. cuauhtemoci (-24.1 ± 3.17‰), suggesting that D. ambigua is the principal phytoplankton consumer in El Sol. The δ13C value of Physa sp. (-11.94 ± 0.14‰) was similar to those of C4-metabolism macrophytes (-11.17‰) and filamentous macroalgae (-10.32 ± 0.3‰), suggesting a littoral herbivorous feeding behavior in El Sol (S2 Table).
δ13C and δ15N values (mean and standard deviation) of single species and aggregated groups (zooplankton, phytoplankton) in Lakes El Sol (A) and La Luna (B). Dap = Daphnia ambigua, Gas = Physa sp., Lep = Leptodiaptomus cuauhtemoci, Malg = macroalgae (Oedogoniaceae and Zygnemataceae), MphyC3 = macrophytes with C3 metabolism, MphyC4 = macrophytes with C4 metabolism, Phyto = phytoplankton, SOM = sediment organic matter (L = littoral and P = profundal), OliH = Limnodrilus hoffmeisteri, OliT = Tubifex tubifex, OliV = Lumbriculus variegatus, Tro = Oncorhynchus mykiss, Zoo = bulk zooplankton.
Just six major groups represented La Luna. We considered SOM profundal because the littoral zone was not well delimited, and pelagic consumers had higher δ15N values than benthic. In comparison, benthic consumers had higher δ13C values than pelagic ones (Fig 3B). δ13C values were different by 1.12‰ between the two oligochaetes sampled in La Luna; Limnodrilus hoffmeisteri had δ13C values similar to those of SOM (△δ13C = 1.97‰), suggesting a detritivorous diet as reported in the literature (S2 Table) and Lumbriculus variegatus had values similar to those of phytoplankton (△δ13C = 1.52‰).
We obtained 14 trophic groups in El Sol and six in La Luna from stable isotope analysis (S3 Table); common groups in both lakes were bulk phytoplankton, cladocerans (Daphnia ambigua), BMI (Lumbriculus variegatus and Limnodrilus hoffmeisteri), benthic macroalgae and sedimentary organic matter (Fig 3). Isotopic δ13C values of littoral and pelagic common groups were different in both lakes (W = 119, p<0.001), and components in the fish-stoked El Sol were enriched compared to the values of the fishless La Luna (Table 1). δ15N had no significant difference between both lakes’ common groups (W = 310, p = 0.7). When comparing the two isotope tracers (δ13C and δ15N), ANOSIM showed differences between both lakes, indicating that isotopic values were more similar within each lake than between them (R = 0.18, p = 0.0017). The isotopic values of major comparable groups (Fig 3), considering both lakes, were also significantly different, independent of the lake (R = 0.56, p< 0.001), but no significant differences were found between zones (pelagic and littoral) (R = 0.03, p = 0.29). Phytoplankton δ13C values in El Sol were 7.78% higher than the phytoplankton δ13C values of La Luna.
Rainbow trout trophic discrimination factor (TDF)
The calculated TDF for rainbow trout white muscle, considering pelagic and littoral potential sources, was 0.49 ± 4.7‰ for δ13C and 4.43 ± 1.32‰ for δ15N. The difference in δ13C values between rainbow trout tissue and littoral sources was smaller (-1.6‰ ± 2.8‰) than for pelagic sources (7.25 ± 2.12‰). In rainbow trout tissues, 13C values were higher than in its diet sources except for Physa sp. and the oligochaete L. variegatus; consequently, both sources were loaded with a negative carbon TDF for the Bayesian models. As expected, δ15N values of rainbow trout (6.63 ± 0.37‰) were higher than in all potential food sources. In contrast, the difference in δ15N values between rainbow trout tissue and littoral sources (5 ± 0.85‰) was higher than for pelagic sources (2.6 ± 0.76‰). The littoral sources had higher δ13C values and then closer to rainbow trout δ13C values than pelagic sources, which in turn had higher δ15N values and closer to rainbow trout δ15N values.
Rainbow trout diet—Bayesian multivariate meta-analysis
Statistical tests performed after multivariate meta-analysis showed a convergence among MCMC chains for all models. All variables were < 0.01 in Gelman-Rubin and Gelweke tests, and the three chains were < 6 and closer to 1, showing a significant similarity between the means of tested MCMC chains. The better models explaining the isotope data values of rainbow trout diet using Leave-one-out comparison and meeting Bayesian mixing polygon assumption were Model 1 with biomass and uniform priors, Model 5 with biomass prior, and Model 2 with a uniform prior (Fig 4A–4D).
The proportional contribution of food sources to rainbow trout diet in El Sol. (A) Model 1 using TDF1 = (△δ13C = 0.49 ± 4.7‰, △δ15N = 4.43 ± 1.32‰) and biomass prior, (B) TDF1 = (△δ13C = 0.49 ± 4.7‰, △δ15N = 4.43 ± 1.32‰) and uniform prior, (C) Model 5, TDF5 = (△δ13C = 2.8 ± 1.5‰, △δ15N = 5.14 ± 1.35‰) and biomass prior, (D) Model 2, TDF2 = (△δ13C = 0.8 ± 1.1‰, △δ15N = 3 ± 2.6‰) and uniform prior.
According to Model 1, oligochaetes were the main prey for rainbow trout in El Sol using calculated TDF and biomass prior, with L. variegatus as the most important diet source (0.42 ± 0.13). In comparison, Tubifex tubifex (0.01 ± 0.035) and L. hoffmeisteri (0.81 ± 0.056) made a lower contribution (Fig 4A). Gastropods (Physa sp.) were the second most significant diet source (0.26 ± 0.14); these grazers, with a slightly higher δ15N and trophic position than detritivores (L. hoffmeisteri) or detritivore-herbivores (L. variegatus), would be a significant food source for rainbow trout. The biomass of gastropods is unknown, and their contribution is more uncertain than other sources. However, it was essential to incorporate the group in the analysis because of their high abundance in the vegetated littoral zone during the sampling. Cladocerans and copepods accounted for 0.023 of the adult rainbow trout diet. The main zooplanktonic sources were D. ambigua (0.15 ± 0.1) and L. cuauhtemoci (0.08 ± 0.07) (Fig 4A). For all models, the MIB L. variegatus and Physa sp. had a higher contribution to the rainbow trout diet, and D. ambigua had a higher contribution than L. cuauhtemoci, considering pelagic sources (Fig 4A–4D). The contribution of T. tubifex was higher in models with uniform priors, while L. hoffmeisteri contributed with a higher proportion in models using a biomass prior (Fig 4A–4D).
Food web structure
The littoral sub-network harbored the major differences in food web topology between the two lakes. Littoral links were higher in the stocked lake with an extended littoral vegetated zone (El Sol littoral 50, El Sol pelagic 37), and the number of links in both sub-networks, number of nodes, links, and species richness of BMI was higher in El Sol (Table 2). Pelagic links were higher than littoral links in the small fishless lake (La Luna pelagic 23, La Luna littoral 15). Littoral and pelagic sub-networks had more links between them in El Sol than in La Luna, revealing a higher coupling in the fish-stocked lake (El Sol 11, La Luna 6). Secondary consumer nodes were present only in El Sol (rainbow trout and Hydra vulgaris). In contrast, herbivorous nodes were present in both lakes and slightly higher within the pelagic zone, representing the higher fraction and contributing to the most significant number of links (Table 2).
Discussion
Despite their proximity and similar origin, El Sol and La Luna have marked food web differences at all trophic levels. A larger size, rainbow trout stocking, and an extensive vegetated littoral zone in the food web of the originally fishless tropical HML El Sol mirrored the results reported for other lakes, including HMLs in cold and temperate regions [4, 6] where active foraging by introduced fish has increased food web coupling. We observed a higher reliance of rainbow trout on littoral BMI, but zooplankton was also an important food source. The latter was evident in (Fig 3A) where rainbow trout and its diet sources draw an A-frame shape with pelagic sources on the left and benthic on the right considering top consumers as the top of the shape (Fig 3A), as observed in other omnivorous predators coupling pelagic and benthic pathways [54].
The food chain length in El Sol was longer (2.86), and its food web network was more complex, with higher taxonomic richness, number of links, nodes, and trophic levels (Table 2). The complexity was associated with the heterogeneous littoral zone of El Sol, which harbored a higher abundance and diversity of BMI. Littoral complexity affects invertebrate diversity and depends on the fish’s presence and food preferences [55, 56]. Littoral complexity can decrease zoo-planktivorous fish abundance by reducing the zooplankton (mainly filter feeders) food sources and increasing algae and periphyton, which compete with phytoplankton [55]. The effect of littoral complexity differs for zoobenthic and omnivorous fish, such as rainbow trout in El Sol, which use more littoral food sources when the habitat is more complex (higher presence of macrophytes) and zooplankton abundance is lower [54, 56]. The increase in BMI functional feeding groups is related to microhabitat complexity; BMI density increases with complexity but just with fish presence [56]. Both effects (fish presence and littoral complexity) could be the main drivers of a higher number of nodes and links in the littoral zone of El Sol. However, the few BMI predators can be associated with fish presence since their density positively relates to littoral complexity in lakes without fish [56].
Intermediate macrophyte cover can enhance the diversity and density of BMI compared with low vegetation or non-vegetated areas [56, 57]. The absence of macrophytes in La Luna (sparse filaments of Temnogametum iztacalense) could be associated with a lower density and diversity of BMI. Also, BMI with significative biomass were detritivores, and no path of autochthonous macroalgae as a food source was observed. The marked differences in the isotopic ratios, the short food chain (1.9), and the simple food web structure of La Luna suggest that the absence of omnivorous and secondary consumers and the low linkage between pelagic and benthic sub-networks intensify a separation in the trophic interactions and isotope values of its food web components, similar to the food web characteristics reported for ultra-oligotrophic lakes, acid lakes, and non-vegetated reduced littoral lakes [39, 58, 59].
The models indicated that the direct contribution of zooplankton to adult trout in El Sol ranges from 0.2 to 0.3 in diet proportion (Fig 4). Differently, the gut contents analysis of trout introduced into temperate Italian HMLs [13] showed a higher diet proportion (0.39). The present results show the importance of adult fish as consumers of the zooplankton population, as in other temperate mountain lakes [13]. Introduced fish can change the size structure of zooplankton from large-bodied cladocerans and copepods to smaller ones and rotifer-dominated communities through size-preferential consumption in temperate lakes [4, 13, 60] and HMLs [13, 61]. In the fish-stocked lake (El Sol), the most abundant zooplankton species, D. ambigua and L. cuauhtemoci were similar in size (948 ± 156 μm). On the contrary, in the fishless lake (La Luna), the cladoceran D. ambigua was notoriously larger (1,648 ± 277 μm) and pigmented, differing from D. ambigua in El Sol, which was hyaline. We suggest that this difference between lakes reflects the presence or absence of fish, as observed in other temperate lakes [5, 62].
The significantly lower δ13C of comparable major groups in La Luna was most likely associated with higher terrestrial inputs (allochthonous) from the surrounding catchment. Allochthonous inputs as a food source are frequent in unproductive lakes [63, 64]. In addition, the dominance of mixotrophic algae in the pelagic zone [36] and the dissimilarity between D. ambigua and phytoplankton δ13C values supported the importance of allochthonous organic matter as a carbon source in the food web of La Luna. In contrast, in El Sol, the δ13C values of herbivorous zooplankton (e.g., D. ambigua) and phytoplankton were similar, and food web components of all trophic levels were enriched in C13. This pattern in isotopic signals is probably associated with zooplankton reliance on autochthonous primary producers, higher primary production, and trophic status [40, 65].
The enriched phytoplankton δ13C values in El Sol suggest that epiphytic algae and unattached macroalgal filaments from the littoral zone (Oedogoniaceae and Zygnemataceae algae) were incorporated into the pelagic zone as suggested by previous studies. Cuna et al. (2022) [36] found filaments of littoral algae (Oedogoniaceae and Zygnemataceae) in El Sol’s plankton. Also, 87.5% of Bacillariophyceae morphospecies were tychoplanktonic.
Food web topology of tropical HMLs El Sol and La Luna versus temperate HMLs
Tropical HMLs El Sol and La Luna showed species richness at all trophic levels and functional groups that differed from temperate HMLs [6]. The species richness of phytoplankton and zooplankton was higher in the tropical HMLs, while benthic macroinvertebrates were more diverse in temperate ones. The omnivorous fraction was the most important in temperate HMLs and was associated with food web stability [6, 66]. With an extended littoral zone and introduced fish, the larger lake (El Sol) had a higher omnivorous fraction than the smaller fishless lake (La Luna) (Table 3). Omnivore nodes were associated with the littoral zone; hence, rainbow trout may decrease food web stability, as observed in temperate lakes [67] and HMLs [68]. Nevertheless, macrophyte beds in the stocked lake El Sol represented an important basal resource and habitat that would increase the littoral sub-web complexity by allowing more trophic interactions [59, 69].
(S = species richness, L = number of links, C = connectance, D = linkage density, Omn = omnivores fraction, Herb = herbivores fraction and B = basal species fraction, Lat = latitude, Tro = tropical, Tem = temperate, Fish = presence/absence of fish, Om = Oncorhynchus mykiss, Sf = Salvelinus fontinalis, St = Salmo trutta), No = fishless).
The herbivorous fraction was the most important in El Sol and La Luna. Among temperate HMLs, the herbivorous fraction was higher in the fishless lake Grande de Gredos (Table 3). Connectance has been reported as low for temperate and tropical HMLs, being equal in both tropical lakes and the temperate lake Grande de Gredos with native fish, while slightly higher in Caballeros (fishless) and Cimera (introduced trout) (Table 3). Considering that higher species richness and low connectance potentially decrease the vulnerability of food webs to disturbances [70], El Sol would be potentially more resistant to disturbance than La Luna, and both tropical HMLs would be potentially more resistant than their temperate counterparts. On the contrary, considering that lower vulnerability is also associated with a higher number of intermediate species [70], temperate HMLs are potentially less vulnerable, having an elevated number of omnivorous intermediate species, compared with the two tropical lakes (El Sol and La Luna) where herbivorous and basal species were high (Table 3).
Rainbow trout diet model and TDF comparison
We calculated a higher δ15NTDF (4.43 ± 1.32‰) and lower δ13CTDF (0.49 ± 4.7) than those previously published for rainbow trout [41, 71], but closer to the generally used δ13CTDF, 0.8 ± 1.1‰ [45]. These results, suggest that omnivorous fish like trout in tropical HMLs with a diet based mainly on benthic macroinvertebrates could have a higher TDF than commonly stated.
To improve accuracy, recent Bayesian mixing models of diet using stable isotope tracers have frequently included priors based on aspects of ecological data such as behavior, traditional knowledge of feeding ecology, stomach contents, or literature on food web topology [6, 17, 72]. The only way to find better prior assumptions and avoid those priors that introduce more uncertainty to the analysis is by comparison of Bayesian models with informative and uniform priors [73]. The fully Bayesian information criterion (leave-one-out validation) showed a higher weight (Loow) for Model 1 using the biomass prior and TDF calculated in this study (Table 4), corroborating the use of accurate priors [49]. Models 1, 4, and 5 had a higher weight for models calculated with the biomass prior, but models 2 and 3 had a higher weight for models calculated with a uniform prior (Table 4). The in-polygon Monte Carlo simulation [50] showed that all models except Model 4 (using the TDF reported by [41]) were accurate for rainbow trout diet estimation, where all sources were within the mixing polygon 95% area (S1 Fig).
OliV: Lumbriculus variegatus, OliT: Tubifex tubifex, OliH: Limnodrilus hoffmeisteri, Gas: Physa sp., Dap: Daphnia ambigua, Lep: Leptodiaptomus cuauhtemoci. TDF1 = (△δ13C = 0.49 ± 4.7‰, △δ15N = 4.43 ± 1.32‰), TDF2 = (△δ13C = 0.8 ± 1.1‰, △δ15N = 3 ± 2.6‰) [44, 45], TDF3 = (△δ13C = 0.91 ± 1.04‰, △δ15N = 3.41 ± 0.2‰) [46], TDF4 = (△δ13C = 1.85‰, △δ15N = 2.54‰) [41], TDF5 = (△δ13C = 2.8 ± 1.5‰, △δ15N = 5.14 ± 1.35‰) [15]. LOOic = Leave-one-out information criterion, se_LOOic = standard error of leave-one-out information criterion, dLOOic = difference leave one-out value compared with the best-fitted model, se_dLOOic = difference in the standard deviation of the leave-one-out value compared with the best-fitted model.
The results highlight the importance of estimating a correct TDF because the outcome of a mixing model depends on the mean and variance of TDF. Thus, accurate predictions on diet contributions to the consumer must be weighted on model design and cross-validation [61, 66]. The similar patterns in the contribution of possible sources to fish diet observed in this study, even with different model assumptions [74], strongly suggest that the use of different TDF values is a reliable alternative to Bayesian mixing methods in the analysis of fish diets in a tropical HML if isotopic data and fractionation studies are scarce.
Conclusions
The food web topology of the larger, vegetated, and fish-stocked El Sol was more complex than the smaller, non-vegetated, and fishless La Luna. Both the pelagic and littoral benthic food webs of El Sol were strongly linked with the C4 photosynthetic path. Fueled mainly by autochthonous primary production, El Sol exhibited a complex food web primarily associated with an extensive vegetated littoral zone. Differently, the smaller lake (La Luna), with a reduced and bare littoral zone, showed a simple food web and was linked mainly with allochthonous carbon paths.
The rainbow trout diet comprised predominantly (up to 0.8) littoral benthic macroinvertebrates and was complemented with herbivorous zooplankton in the pelagic zone. Rainbow trout introduction fostered the differences between the lakes by lengthening the food chain in El Sol and linking the pelagic and the benthic zones. At the same time, connectance and linkage density remained similar.
There were differences in the food webs between tropical HMLs El Sol and La Luna and temperate HMLs, with higher species richness and lower linkage density in the former. Also, the omnivorous fraction was markedly lower, and the herbivorous fraction was higher in tropical El Sol and La Luna than in other temperate HMLs.
Supporting information
S1 Fig. Mixing polygons calculated with Bayesian analysis for diet sources of rainbow trout.
https://doi.org/10.1371/journal.pone.0287066.s001
(TIFF)
S1 Table. Physicochemical, trophic, and morphometric variables of Lakes El Sol and La Luna.
https://doi.org/10.1371/journal.pone.0287066.s002
(DOCX)
S2 Table. Ecologic groups used for the food web construction of Lakes El Sol and La Luna.
https://doi.org/10.1371/journal.pone.0287066.s003
(DOCX)
S3 Table. Minimal data set of stable isotope data (δ13C and δ15N) for the major Lakes El Sol and La Luna groups.
https://doi.org/10.1371/journal.pone.0287066.s004
(DOCX)
Acknowledgments
The Comisión Estatal de Parques Naturales y de la Fauna (CEPANAF, Secretaría de Ecología, Gobierno del Estado de México) provided the permit for scientific research at the Área de Protección de Flora y Fauna Nevado de Toluca. We thank Agnieszka Adamowicz and Dr. Jean-François Hélie from the GEOTOP for the stable isotope analysis and Dra. Margarita Caballero Miranda from the Laboratorio de Paleolimnología, Instituto de Geofísica, UNAM, for sample processing. The authors recognize Anne Grant for the valuable scientific observations and linguistic adjustments that significantly improved this paper. We thank Mariana Vargas for outlining Fig 1.
References
- 1. Aguilera X, Lazzaro X, Coronel JS. Tropical high-altitude Andean lakes located above the tree line attenuate UV-A radiation more strongly than typical temperate alpine lakes. Photochem Photobiol Sci. 2013;12: 1649–1657. pmid:23722356
- 2. Catalan J, Donato Rondón JC. Perspectives for an integrated understanding of tropical and temperate high-mountain lakes. J Limnol. 2016;75.
- 3. Hamerlík L, Svitok M, Novikmec M, Očadlík M, Bitušík P. Local, among-site, and regional diversity patterns of benthic macroinvertebrates in high altitude waterbodies: do ponds differ from lakes? Hydrobiologia. 2014;723: 41–52.
- 4. Eby LA, Roach WJ, Crowder LB, Stanford JA. Effects of stocking-up freshwater food webs. Trends Ecol Evol. 2006;21: 576–584. pmid:16828522
- 5. Reissig M, Trochine C, Queimaliños C, Balseiro E, Modenutti B. Impact of fish introduction on planktonic food webs in lakes of the Patagonian Plateau. Biol Conserv. 2006;132: 437–447. https://doi.org/10.1016/j.biocon.2006.04.036
- 6. Sánchez-Hernández J, Cobo F, Amundsen P-A. Food Web Topology in High Mountain Lakes. PLoS One. 2015;10: e0143016. Available: pmid:26571235
- 7. Scheifhacken N, Fiek C, Rothhaupt K-O. Complex spatial and temporal patterns of littoral benthic communities interacting with water level fluctuations and wind exposure in the littoral zone of a large lake. Fundam Appl Limnol. 2007;169: 115–129.
- 8. Vander Zanden MJ, Vadeboncoeur Y. Fishes as integrators of benthic and pelagic food webs in lakes. Ecology. 2002;83: 2152–2161. https://doi.org/10.1890/0012-9658(2002)083[2152:FAIOBA]2.0.CO;2
- 9. Burks RL, Lodge DM, Jeppesen E, Lauridsen TL. Diel horizontal migration of zooplankton: costs and benefits of inhabiting the littoral. Freshw Biol. 2002;47: 343–365. https://doi.org/10.1046/j.1365-2427.2002.00824.x
- 10. James MR, Hawes I, Weatherhead M, Stanger C, Gibbs M. Carbon flow in the littoral food web of an oligotrophic lake. Hydrobiologia. 2000;441: 93–106. :1017504820168
- 11. Stoffels RJ, Closs GP. Abundance of brown trout as related to littoral zone gradient in Lakes Te Anau and Manapouri, New Zealand. New Zeal J Mar Freshw Res. 2002;36: 455–458.
- 12. Schindler DE, Scheuerell MD. Habitat Coupling in Lake Ecosystems. Oikos. 2002;98: 177–189. Available: http://www.jstor.org/stable/3547424
- 13. Tiberti R, Brighenti S, Iacobuzio R, Pasquini G, Rolla M. Behind the Impact of Introduced Trout in High Altitude Lakes: Adult, not Juvenile Fish are Responsible of the Selective Predation on Crustacean Zooplankton. J Limnol. 2014;73: 593–597.
- 14.
Walsh MR, Cooley 4th F, Biles K, Munch SB. Predator-induced phenotypic plasticity within- and across-generations: a challenge for theory? Proceedings Biol Sci. 2015;282: 20142205. https://doi.org/10.1098/rspb.2014.2205
- 15. Castro-Rebolledo MI, Muñoz-Gracia I, Donato-Rondón JC. Food web of a tropical high mountain stream: effects of nutrient addition. Acta Biológica Colomb. 2014;19: 33–41. https://doi.org/10.15446/abc.v19n1.38265
- 16. Sánchez-Hernández J., Cobo F., & Amundsen P. Food Web Topology in High Mountain Lakes. PLoS One. 2015;10. Available: pmid:26571235
- 17. Schmid-Araya JM, Schmid PE, Tod SP, Esteban GF. Trophic positioning of meiofauna revealed by stable isotopes and food web analyses. Ecology. 2016;97: 3099–3109. pmid:27870020
- 18. Phillips DL. Converting isotope values to diet composition: the use of mixing models. J Mammal. 2012;93: 342–352.
- 19. Porcel S, Fogel ML, Izaguirre I, Roesler I, Lancelotti JL. Effect of rainbow trout introductions on food webs in lakes of the arid Patagonia. Hydrobiologia. 2022;849: 2057–2075.
- 20. Doi H. Spatial patterns of autochthonous and allochthonous resources in aquatic food webs. Popul Ecol. 2009;51: 57–64.
- 21. Caut S, Angulo E, Courchamp F. Variation in discrimination factors (Δ15N and Δ13C): the effect of diet isotopic values and applications for diet reconstruction. J Appl Ecol. 2009;46: 443–453. https://doi.org/10.1111/j.1365-2664.2009.01620.x
- 22. Cherel Y, Hobson K, Hassani S. Isotopic Discrimination between Food and Blood and Feathers of Captive Penguins: Implications for Dietary Studies in the Wild. Physiol Biochem Zool Ecol Evol Approaches. 2005;78: 106–115. pmid:15702469
- 23. Post DM. Using stable isotopes to estimate trophic position: model, methods, and assumptions. Ecology. 2002;83: 703–718.
- 24. Parnell AC, Phillips DL, Bearhop S, Semmens BX, Ward EJ, Moore JW, et al. Bayesian stable isotope mixing models. Environmetrics. 2013;24: 387–399. https://doi.org/10.1002/env.2221
- 25. Stock BC, Jackson AL, Ward EJ, Parnell AC, Phillips DL, Semmens BX. Analyzing mixing systems using a new generation of Bayesian tracer mixing models. Nelson D, editor. PeerJ. 2018;6: e5096. pmid:29942712
- 26. Phillips DL, Gregg JW. Uncertainty in source partitioning using stable isotopes. Oecologia. 2001;127: 171–179. pmid:24577646
- 27. Moore JW, Semmens BX. Incorporating uncertainty and prior information into stable isotope mixing models. Ecol Lett. 2008;11: 470–480. pmid:18294213
- 28. Bond AL, Diamond AW. Recent Bayesian stable-isotope mixing models are highly sensitive to variation in discrimination factors. Ecol Appl. 2011;21: 1017–1023. pmid:21774408
- 29. Metcalfe SE. Late Quaternary Environments of the Northern Deserts and Central Transvolcanic Belt of Mexico. Ann Missouri Bot Gard. 2006;93: 258–273. Available: http://www.jstor.org/stable/40035725
- 30. Armienta MA, Vilaclara G, De la Cruz-Reyna S, Ramos S, Ceniceros N, Cruz O, et al. Water chemistry of lakes related to active and inactive Mexican volcanoes. J Volcanol Geotherm Res. 2008;178: 249–258. https://doi.org/10.1016/j.jvolgeores.2008.06.019
- 31. Peel MC, Finlayson BL, McMahon TA. Updated world map of the Köppen-Geiger climate classification. Hydrol Earth Syst Sci. 2007;11: 1633–1644.
- 32. Armienta MA, De la Cruz-Reyna S, Macías JL. Chemical characteristics of the crater lakes of Popocatetetl, El Chichon, and Nevado de Toluca volcanoes, Mexico. J Volcanol Geotherm Res. 2000;97: 105–125. https://doi.org/10.1016/S0377-0273(99)00157-2
- 33. Caballero-Miranda M. The Diatom flora of two acid lakes in central México. Diatom Res. 1996;11: 227–240.
- 34. Sarma SSS, Elías-Gutiérrez M, Serranía Soto C. Rotifers from high altitude crater-lakes at Nevado de Toluca Volcano, México. Hidrobiológica. 1969;6: 33–38. Available: https://hidrobiologica.izt.uam.mx/index.php/revHidro/article/view/592
- 35. Dimas-Flores N, Alcocer J, Ciros-Pérez J. The Structure of the Zooplankton Assemblages from Two Neighboring Tropical High Mountain Lakes. J Freshw Ecol. 2008;23: 21–31.
- 36. Cuna E, Alcocer J, Gaytán M, Caballero M. Phytoplankton Biodiversity in Two Tropical, High Mountain Lakes in Central Mexico. Diversity. 2022;14: 17.
- 37. Caballero M, Zawisza E, Hernández M, Lozano-García S, Ruiz-Cordova JP, Waters MN, et al. The Holocene history of a tropical high-altitude lake in central Mexico. The Holocene. 2020;30: 865–877.
- 38. Alcocer J, Roberson J, Oseguera LA, Lewis WM Jr. Rhythmic episodes of heating and cooling control thermal stratification of two tropical high mountain lakes. Aquat Sci. 2020;82: 58.
- 39. Bertolo A, Carignan R, Magnan P, Pinel-Alloul B, Planas D, Garcia E, et al. Decoupling of Pelagic and Littoral Food Webs in Oligotrophic Canadian Shield Lakes. Oikos. 2005;111: 534–546. Available: http://www.jstor.org/stable/3548645
- 40. Grey J, Jones RI, Sleep D. Stable isotope analysis of the origins of zooplankton carbon in lakes of differing trophic state. Oecologia. 2000;123: 232–240. pmid:28308728
- 41. Pinnegar JK, Polunin NVC. Differential fractionation of δ13C and δ15N among fish tissues: implications for the study of trophic interactions. Funct Ecol. 1999;13: 225–231. https://doi.org/10.1046/j.1365-2435.1999.00301.x
- 42. Newsome SD, Bentall GB, Tinker MT, Oftedal OT, Ralls K, Estes JA, et al. Variation in δ13C and δ15N diet–vibrissae trophic discrimination factors in a wild population of California sea otters. Ecol Appl. 2010;20: 1744–1752. https://doi.org/10.1890/09-1502.1
- 43.
R Core Team. R: A language and environment for statistical computing. Vienna, Austria: R Foundation for Statistical Computing; 2020.
- 44. Deniro MJ, Epstein S. Influence of diet on the distribution of nitrogen isotopes in animals. Geochim Cosmochim Acta. 1981;45: 341–351. https://doi.org/10.1016/0016-7037(81)90244-1
- 45. DeNiro M, Samuel E. Mechanism of Carbon Isotope Fractionation Associated with Lipid Synthesis. Science (80-). 1977;197: 261–263. pmid:327543
- 46. Zanden MJ Vander Rasmussen JB. Variation in δ15N and δ13C trophic fractionation: Implications for aquatic food web studies. Limnol Oceanogr. 2001;46: 2061–2066. https://doi.org/10.4319/lo.2001.46.8.2061
- 47. Oseguera LA, Alcocer J, Escobar E. Benthic macroinvertebrates of two tropical, high-mountain lakes at Nevado de Toluca Volcano at central region of Mexico. Hidrobiológica. 2016;26: 419–432. https://doi.org/10.24275/uam/izt/dcbs/hidro/2016v26n3/Oseguera
- 48. DeVries MS, Stock BC, Christy JH, Goldsmith GR, Dawson TE. Specialized morphology corresponds to a generalist diet: linking form and function in smashing mantis shrimp crustaceans. Oecologia. 2016;182: 429–442. pmid:27312263
- 49. Vehtari A, Gelman A, Gabry J. Practical Bayesian model evaluation using leave-one-out cross-validation and WAIC. Stat Comput. 2017;27: 1413–1432.
- 50. Smith JA, Mazumder D, Suthers IM, Taylor MD. To fit or not to fit: evaluating stable isotope mixing models using simulated mixing polygons. Methods Ecol Evol. 2013;4: 612–618. https://doi.org/10.1111/2041-210X.12048
- 51. Blondel J. Guilds or Functional Groups: Does It Matter? Oikos. 2003;100: 223–231. Available: http://www.jstor.org/stable/3548179
- 52. Simberloff D, Dayan T. The Guild Concept and the Structure of Ecological Communities. Annu Rev Ecol Syst. 1991;22: 115–143. Available: http://www.jstor.org/stable/2097257
- 53. Hudson LN, Emerson R, Jenkins GB, Layer K, Ledger ME, Pichler DE, et al. Cheddar: analysis and visualisation of ecological communities in R. Methods Ecol Evol. 2013;4: 99–104.
- 54. Stewart SD, Hamilton DP, Baisden WT, Dedual M, Verburg P, Duggan IC, et al. Variable littoral-pelagic coupling as a food-web response to seasonal changes in pelagic primary production. Freshw Biol. 2017;62: 2008–2025. https://doi.org/10.1111/fwb.13046
- 55. Carey MP, Maloney KO, Chipps SR, Wahl DH. Effects of littoral habitat complexity and sunfish composition on fish production. Ecol Freshw Fish. 2010;19: 466–476. https://doi.org/10.1111/j.1600-0633.2010.00433.x
- 56. Rennie MD, Jackson LJ. The influence of habitat complexity on littoral invertebrate distributions: patterns differ in shallow prairie lakes with and without fish. Can J Fish Aquat Sci. 2005;62: 2088–2099.
- 57. Beckett DC, Aartila TP, Miller AC. Contrasts in Density of Benthic Invertebrates Between Macrophyte Beds and Open Littoral Patches in Eau Galle Lake, Wisconsin. Am Midl Nat. 1992;127: 77–90.
- 58. Gaedke U, Kamjunke N. Structural and functional properties of low- and high-diversity planktonic food webs. J Plankton Res. 2006;28: 707–718.
- 59. He Y, Zhao K, Zhang H, He L, Niu Y, Zhang M, et al. Linking macrophyte community structure with food chain length: A case study in the largest freshwater lake in China and ecological restoration implications. Ecol Indic. 2021;123: 107363. https://doi.org/10.1016/j.ecolind.2021.107363
- 60. Izaguirre I, Lancelotti J, Saad JF, Porcel S, O´Farrell I, Marinone MC, et al. Influence of fish introduction and water level decrease on lakes of the arid Patagonian plateaus with importance for biodiversity conservation. Glob Ecol Conserv. 2018;14: e00391. https://doi.org/10.1016/j.gecco.2018.e00391
- 61. Brancelj A. Shifts in zooplankton communities in high-mountain lakes induced by singular events (fish stocking, earthquakes): evidence from a 20-year survey in Slovenia (Central Europe). Aquat Ecol. 2021;55: 1253–1271.
- 62. Sægrov H, Hobæk A, L’Abée-Lund JH. Vulnerability of melanic Daphnia to brown trout predation. J Plankton Res. 1996;18: 2113–2118.
- 63. Ask J, Karlsson J, Persson L, Ask P, Byström P, Jansson M. Terrestrial organic matter and light penetration: Effects on bacterial and primary production in lakes. Limnol Oceanogr. 2009;54: 2034–2040. https://doi.org/10.4319/lo.2009.54.6.2034
- 64. Karlsson J, Berggren M, Ask J, Byström P, Jonsson A, Laudon H, et al. Terrestrial organic matter support of lake food webs: Evidence from lake metabolism and stable hydrogen isotopes of consumers. Limnol Oceanogr. 2012;57: 1042–1048. https://doi.org/10.4319/lo.2012.57.4.1042
- 65. Woodland RJ, Magnan P, Glémet H, Rodríguez MA, Cabana G. Variability and directionality of temporal changes in δ13C and δ15N of aquatic invertebrate primary consumers. Oecologia. 2012;169: 199–209. pmid:22139449
- 66. Toro M, Granados I, Robles S, Montes C. High mountain lakes of the Central Range (Iberian Peninsula): Regional limnology & environmental changes. Limnetica. 2006;25: 217–252. 5.17
- 67. Nyström P, Svensson O, Lardner B, Brönmark C, Granéli W. The Influence of Multiple Introduced Predators on a Littoral Pond Community. Ecology. 2001;82: 1023–1039. https://doi.org/10.2307/2679900
- 68. Miró A, Ventura M. Introduced fish in Pyrenean high mountain lakes: impact on amphibians and other organisms, and conservation implications. Limnetica. 2020;39: 283–297.
- 69. Špoljar M, Lajtner J, Dražina T, Malekzadeh-Viayeh R, Radanović I, Zrinščak I, et al. Disentangling food webs interactions in the littoral of temperate shallow lakes. Nat Croat. 2017;26: 145–166.
- 70. Calizza E, Rossi L, Careddu G, Sporta Caputi S, Costantini ML. Species richness and vulnerability to disturbance propagation in real food webs. Sci Rep. 2019;9: 19331. pmid:31852953
- 71. Rounick JS, Hicks BJ. The stable carbon isotope ratios of fish and their invertebrate prey in four New Zealand rivers. Freshw Biol. 1985;15: 207–214. https://doi.org/10.1111/j.1365-2427.1985.tb00193.x
- 72. Kiszka JJ, Caputo M, Méndez-Fernandez P, Fielding R. Feeding Ecology of Elusive Caribbean Killer Whales Inferred From Bayesian Stable Isotope Mixing Models and Whalers’ Ecological Knowledge. Front Mar Sci. 2021;8.
- 73. Swan GJF, Bearhop S, Redpath SM, Silk MJ, Goodwin CED, Inger R, et al. Evaluating Bayesian stable isotope mixing models of wild animal diet and the effects of trophic discrimination factors and informative priors. Methods Ecol Evol. 2020;11: 139–149. https://doi.org/10.1111/2041-210X.13311
- 74. Bastos RF, Corrêa F, Winemiller KO, Garcia AM. Are you what you eat? Effects of trophic discrimination factors on estimates of food assimilation and trophic position with a new estimation method. Ecol Indic. 2017;75: 234–241. https://doi.org/10.1016/j.ecolind.2016.12.007