Figures
Abstract
Environmental DNA (eDNA) is an effective approach for detecting vertebrates and plants, especially in aquatic ecosystems, but prior studies have largely examined eDNA in cool temperate settings. By contrast, this study employs eDNA to survey the fish fauna in tropical Lake Bacalar (Mexico) with the additional goal of assessing the possible presence of invasive fishes, such as Amazon sailfin catfish and tilapia. Sediment and water samples were collected from eight stations in Lake Bacalar on three occasions over a 4-month interval. Each sample was stored in the presence or absence of lysis buffer to compare eDNA recovery. Short fragments (184–187 bp) of the cytochrome c oxidase I (COI) gene were amplified using fusion primers and then sequenced on Ion Torrent PGM or S5 before their source species were determined using a custom reference sequence database constructed on BOLD. In total, eDNA sequences were recovered from 75 species of vertebrates including 47 fishes, 15 birds, 7 mammals, 5 reptiles, and 1 amphibian. Although all species are known from this region, six fish species represent new records for the study area, while two require verification. Sequences for five species (2 birds, 2 mammals, 1 reptile) were only detected from sediments, while sequences from 52 species were only recovered from water. Because DNA from the Amazon sailfin catfish was not detected, we used a mock eDNA experiment to confirm our methods would enable its detection. In summary, we developed protocols that recovered eDNA from tropical oligotrophic aquatic ecosystems and confirmed their effectiveness in detecting fishes and diverse species of vertebrates.
Citation: Valdez-Moreno M, Ivanova NV, Elías-Gutiérrez M, Pedersen SL, Bessonov K, Hebert PDN (2019) Using eDNA to biomonitor the fish community in a tropical oligotrophic lake. PLoS ONE 14(4): e0215505. https://doi.org/10.1371/journal.pone.0215505
Editor: Hideyuki Doi, University of Hyogo, JAPAN
Received: August 30, 2018; Accepted: April 4, 2019; Published: April 22, 2019
Copyright: © 2019 Valdez-Moreno 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: Data for BOLD datasets are available at: dx.doi.org/10.5883/DS-EBACALAR, dx.doi.org/10.5883/DS-EBACAMPH, dx.doi.org/10.5883/DS-EBACAVES, dx.doi.org/10.5883/DS-EBACREPT, dx.doi.org/10.5883/DS-EBACMAMM. Raw Sequence Data – ENA are available at: http://www.ebi.ac.uk/ena/data/view/PRJEB25950.
Funding: DNA extraction and NGS were carried out at the Centre for Biodiversity Genomics (CBG), University of Guelph through funding to PDNH from Ann McCain Evans and Chris Evans. MVM and MEG thank CONACYT for support through the Sabbatical Stays Program (Grants 261790 and 262267). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
Introduction
Environmental DNA (eDNA) has gained popularity for biomonitoring, especially for the detection of invasive species and for baseline surveys of animal and plant communties [1,2]. This eDNA derives from cells shed into the environment as mucus, urine, feces, or gametes [2–7]. More than 120 articles have now considered eDNA, including a special issue on the topic [8]. The rapid adoption of eDNA-based bioassessments reflects two factors: improved access to the reference sequences [9,10] required to link the short reads from eDNA to their source species, and the availability of high-throughput sequencers that can generate large volumes of data at modest cost.
Conventional biomonitoring programs require extensive effort with sampling repeated over time and space [11]. Because these methods utilize nets and electrofishing, they are invasive and can damage habitats [3,12,13]. Furthermore, they rarely reveal all species in an environment, and each specimen that is collected must be identified to a species, requiring access to taxonomic experts [14,15]. Given these limitations, eDNA-based methods are an attractive approach for the quantification of biodiversity required for conservation and management [8,16].
Most prior eDNA studies have focused on the detection of invasive fish species [2,3,17] or rare/ endangered aquatic species [18,19] although they have also been used to examine the abundance and biomass of fish species [20,21]. Early results revealed complexities in data acquisition and interpretation that have provoked studies to improve sampling and DNA extraction while also minimizing methodological biases [16,22,23].
Varied protocols have been used to preserve eDNA, but they usually involve the immediate storage of samples at low temperature [21], precipitation [24], the addition of lysis buffers to filters [25], and the inclusion of surfactants [26]. While eDNA has traditionally been assumed to rapidly degrade, this is not always the case. For example, Deiner et al. [27] employed long-range PCR (>16 kb) to recover whole mitochondrial genomes from the six fish species in a mock community, and from 10 of 12 species represented in water samples from a stream. Prior studies have also demonstrated that eDNA metabarcoding is an effective alternative to traditional biomonitoring as it has revealed the species composition of fish communities in various ecosystems including a species-rich coastal sea in Japan [28] and a river in Indiana, USA [29]. In fact, in both cases, it revealed species that were overlooked by other sampling methods.
Although most past eDNA studies have examined temperate ecosystems, there is increasing interest in the use of this method in the tropics [30–32]. The present study sought to develop an eDNA protocol for biomonitoring the native and non-native fish fauna of oligotrophic tropical Lake Bacalar and adjacent habitats, with a secondary goal to verify its capacity to detect an invasive fish, the Amazon sailfin catfish, Pterygoplichthys sp., which represents a serious threat in this region, for future biomonitoring applications.
Material and methods
Ethics statement
Water and sediments were obtained from Lake Bacalar and surrounding habitats; the samples did not require any permit. As well, no approval was needed from the Ethics Committee on Research for El Colegio de la Frontera Sur in accord with Mexican Animal Welfare Regulation NOM-062-ZOO-1999 because we only collected water samples and sediments, not vertebrates. Laboratory experiments on fish held in aquaria also do not require approval under the Mexican Animal Welfare Regulation NOM-062-ZOO-1999 [33,34] so no permits were required for the aquarium study on Amazon catfish. A letter from the Ethics Committee on Research for El Colegio de la Frontera Sur confirmed this study met all standards for experimentation and research integrity.
Study site
Lake Bacalar, the largest freshwater habitat in the Yucatan Peninsula, is renowned for its striking blue color, the clarity of its water, and for the world’s largest occurrence of microbialites. With a length of more than 10 km [35–37], it occupies a larger fault basin [38] (50 km long, 2 to 3 km wide). With sediments derived from karst limestone, it represents the world’s largest freshwater lens system [39]. Lake Bacalar is not directly connected to the sea, but it shows a high rate of groundwater flow [40]. The northern part is connected to a complex system of lagoons including Laguna Chile Verde and Laguna Guerrero. The southern part of the lake has an indirect connection to the sea via the Hondo River that enters Chetumal Bay. There are three cenotes (sinkholes) in the lake whose depths range from 35–55 m (pers. unp. data).
Water temperatures in Lake Bacalar range from 25–28°C, while the Secchi transparency averages 10.3 m. The lake is unstratified [35,41] with a water temperature of 27°C or higher throughout the year. Its surface waters are slightly alkaline (pH = 7.8) with a conductivity of 1220 μS/cm [41] and HCO3 values higher than in marine environments [35].
All samples were taken from the littoral zone at a maximum depth of 0.5 m, except La Unión (2 m) and Xul-Ha (4 m). Eight sampling sites were examined along the main axis of Lake Bacalar and associated systems in the Hondo River (Fig 1). Samples were collected in December 2015, January 2016, and April 2016.
The map was prepared using QGIS v 2.8.2 under the INEGI (Instituto Nacional de Geografía Estadística e Informática) free documentation license from topographic chart E16A66 (Scale 1:50,000) (www.inegi.org.mx/app/areasgeograficas/?ag=23).
Sampling
Eight localities (one with three points) (Fig 1) were sampled and most were examined on more than one occasion, producing a total of 14 sampling events. Three sites (Cocalitos, Huay Pix and Xul-Ha) were sampled in December 2015, four sites in January 2016 (Alvaro Obregón Viejo, Huay Pix, La Unión and Pedro A. Santos), and seven sites in April 2016 (Cacao, Cocalitos, Huay Pix, Juan Sarabia, La Unión, Pedro A. Santos and Xul-Ha).
All sampling bottles and equipment were handled with gloves to minimize contamination with human DNA. At each site, three replicate samples of water and sediment were taken. Each 1 L water sample was placed in a sterilized CIVEQ bottle, while each sediment sample was placed in a separate Ziplock bag (100 cm3). Each water and sediment sample was collected using a new turkey baster. Sediment samples were obtained from the upper 1 cm. All samples were immediately placed on ice, and then transported to the lab in Chetumal for processing.
Processing water samples
To minimize eDNA degradation, water samples were filtered within 7 hours of collection. Prior to processing, all lab surfaces and materials were sterilized with 10% bleach, followed by 70% ethanol; gloves were changed between each sample. Because samples had to be transported to Canada for processing, we evaluated sample transportation in the presence or absence of lysis buffer. Each sample was split into two 0.5 L subsamples that were filtered through separate 0.22 μm filters. One filter was stored with 1 ml of PW1 solution with grinding media while the other was placed with grinding media in a dry tube covered by aluminum foil. This approach resulted in six samples per locality for each collection event (Fig 2). All filters were stored at -18°C before being transported on ice from Chetumal to Guelph where DNA extraction was undertaken. The interval between filtration and DNA extraction was always less than 48 hours.
Processing sediment samples
During the first collection trip, three replicate sediment samples (20 cm3 each) were collected at each site. Similar to water samples, sediments were preserved in the presence or absence of lysis buffer for transportation: each sample was split into two 10 cm3 subsamples; 2 ml of 5M GuSCN buffer was either used as a preservation buffer or was added during lysis. For the second and third collection events, one pooled sample (60 cm3) was collected at each site and split into two replicates of 30 cm3; 6 ml of 5M GuSCN buffer [42] was used as a preservation buffer or was added during lysis step (Fig 2). Sediments were kept at -18°C before transportation on ice to Guelph.
DNA extraction from water samples
Prior to DNA extraction, all lab surfaces and pipettors were sterilized using 10% bleach, followed by 70% ethanol. Pipettors were repeatedly cleaned with 10% bleach, followed by 70% ethanol during extraction and gloves were frequently changed. Centrifuge rotors, adapters, and tube racks were washed with diluted ELIMINase (Decon Labs) (1:10) and rinsed with deionized water.
DNA was extracted from the filters from the water samples using a PowerWater DNA extraction kit (MoBio) following a modified lysis protocol (tubes were incubated at 65°C prior to the grinding stage) with minor modifications to avoid cross-contamination between samples; all steps requiring 650 μl were reduced to 625 μl. Columns were incubated at 56°C for 15 min prior to DNA elution in 100 μl of PW6 buffer and quantified using Qubit 2.0 and HS dsDNA kit.
DNA extraction from sediment samples
All lab surfaces and equipment were decontaminated as described above. The DNA extraction protocol employed ProK digestion in the presence of chaotropic salts followed by the use of magnetic beads to selectively bind and then release DNA. Sera-Mag SpeedBead Carboxylate-Modified Magnetic Particles (Hydrophilic) from GE Healthcare were employed to bind/release DNA. They were prepared as described in [43] in a polyethylene glycol/NaCl buffer (0.1% Sera Magnetic Particles (w/v) in 18% PEG-8000, 1M NaCl, 10 mM Tris-HCl, pH 8.0, 1 mM EDTA pH 8.0, 0.05% Tween 20). Extraction was initiated by adding a 1/5 volume of 5M GuSCN lysis buffer [42] to non-treated samples and a 1/10 volume of ProK (20 mg/ml). Each tube was vortexed and incubated for 2 hours at 56°C on an orbital shaker, then held for 30 min at 65°C before being centrifuged at 2000 g for 2 min. DNA was extracted from 500 μl of the resultant lysate after its transfer to a 2 ml Eppendorf LoBind tube. 500 μl of prepared magnetic beads was added to each lysate, and a pipettor was used to mix the beads and lysate before incubation at room temperature for 10 min. Tubes were then placed on a magnet (DynaMag) and incubated until the supernatant was clear. Beads were thoroughly washed once with 1 ml of PWB buffer [44] to remove contaminants and twice with 1 ml of 70% ethanol. Magnetic beads were then dried at room temperature for 10–15 min before DNA was eluted into 100 μl of 10 mM Tris-HCl, pH 8.0 and quantified using Qubit 2.0 and HS dsDNA kit.
PCR amplification
To avoid PCR inhibition, each DNA extract from sediments was diluted 1:10 with molecular grade water prior to PCR amplification, while extracts from water were used without dilution.
To increase species recovery and overcome possible PCR bias introduced by UMI-labelled fusion primers [45], we followed a two-step PCR approach [45] with the first round employing conventional primers, while the diluted PCR product serves as a template for a second round of PCR with fusion primers containing sequencing adapters and UMI-tags.
A 184–187 bp segment of the barcode region of COI was amplified with two primer sets (AquaF2/C_FishR1, AquaF3/C_FishR1) known to be effective for vertebrates (Table 1). The PCR reactions employed the master mix described in [46] and Platinum Taq. The first round of PCR employed the following thermal regime: 94°C for 2 min, 40 cycles of 94°C for 40 s, annealing at 51°C for AquaF2 or 50°C for AquaF3 for 1 min, and 72°C for 1 min, with a final extension at 72°C for 5 min. Following the first round, PCR products were diluted 2-fold with molecular grade water, and 2 μl was transferred to a new tube for a second round of PCR to create amplicon libraries tagged with unique molecular identifiers (UMIs) with fusion primers containing Ion Xpress MID tags and Ion Adapters (Table 1). The same UMI-tag was applied to all three replicates for each treatment from a particular collection event (Fig 2). The PCR regime for the second round consisted of 94°C for 2 min, 20 cycles of 94°C for 40 s, annealing at 51°C for 1 min, and 72°C for 1 min, with a final extension at 72°C for 5 min. PCR products were visualized on an E-Gel96 (Invitrogen, Thermo Fisher Scientific).
Libraries for AquaF2/C_FishR1 and AquaF3/C_FishR1 were purified separately using magnetic beads prepared as described in [43], quantified using Qubit 2.0 and HS dsDNA kit, and normalized to 13 pM prior to sequencing. Most amplicon libraries were sequenced on an Ion PGM using the 400 Template and Sequencing kits following manufacturer’s instructions with either a 316 or 318 chip. Water samples from April 2016 were also analyzed on an Ion S5 using the Chef for templating and an Ion 520 and 530 kit with a 530 chip. Fig 2 summarizes the treatment for each water or sediment sample. HTS data (FASTQ files) are available in the European Nucleotide Archive (ENA) under submission number: PRJEB25950.
Detection of an invasive fish
We conducted a mock eDNA experiment to ensure our methods could detect species of Pterigoplichthys. Two of these fishes were kept in a 10 L aerated aquarium at the Chetumal laboratory for 10 days at 25°C and fed algal pellets. One liter of water was collected, and two 0.5 L subsamples were filtered through separate 0.22 μm filters; filters were stored with buffer and no buffer as described above. Upon completion of the experiment, fish were kept at the ECOSUR for educational purposes.
Bioinformatics workflow
We created a reference dataset (DS-EBACALAR) on BOLD with sequences from 3,534 Mexican fishes representing 576 BINs, including 219 records from Lake Bacalar and surrounding habitats (Hondo River, and adjacent lakes–Huay Pix, Xul-Ha). This reference barcode library included sequences for 53 of the 57 fish species (Table 2) known from these habitats (see Fig 1). As well, four other datasets with all public data on BOLD for Amphibia, Mammalia, Aves, and Reptilia were constructed. These five datasets were employed as a basis for assigning a source species to each OTU recovered from the present eDNA analyses.
The following procedure was used to process the raw NGS reads: Cutadapt (v1.8.1) was used to trim primer sequences; Sickle (v1.33) was used for size filtering (sequences from 100–205 bp were retained), while Uclust (v1.2.22q) was employed to recognize OTUs based on the >98% identity and a minimum read depth of 2 thresholds. The Local Blast 2.2.29+ algorithm was then used to compare each OTU to the reference sequences in five datasets: DS-EBACALAR Bacalar Fish Dataset for eDNA Detection (3,534 sequences), and public BOLD data for Amphibia, Aves, Mammalia, and Reptilia represented by the following datasets: DS-EBACAMPH (11,018 sequences), DS-EBACAVES (28,914 sequences), DS-EBACMAMM (39,890 sequences), DS-EBACREPT (5,424 sequences). Raw Blast output results were parsed using a custom-built python script OTUBlastParser.py, and concatenated using ConcatenatorResults.py available at https://gitlab.com/biodiversity/OTUExtractorFromBLASToutput, before the results in tab-delimited format were imported to MS Excel, filtered by min score of 250, and 97–100% percent identity range. Filtered results were further processed and visualized in Tableau software 10.4. All reads derived from humans were excluded from the final results. The OTUBlastParser script works best for curated reference databases with flagged closely related species or species complexes lacking resolution. In this scenario, adjustments can be made to the metadata file containing taxonomy information for each Sample ID, where a species complex name or a genus name can be listed under ‘Identification’ column, enabling recognition of these problematic taxa at the parsing stage.
Species accumulation curves were calculated in EstimateS software available at http://purl.oclc.org/estimates [48], with the following settings: 1000 randomizations; do not extrapolate rarefaction curves; classic formula for Chao 1 and Chao 2. Results were exported to Microsoft Excel and visualized in Tableau software 10.4.
Sequences with 97–100% identity to fish species represented in the DS-EBACALAR dataset were mapped onto a neighbor-joining Taxon ID tree generated in BOLD for these taxa, and visualized in iTOL [49].
The eDNA analyses recovered sequences from 75 species of vertebrates including 47 fishes (Table 2), 15 birds, 7 mammals, 5 reptiles and 1 amphibian, most previously known from this region. Although substantially more species were recovered from the water than from the sediments (Table 3, Fig 3), five species were only recovered from sediments–racoon (Procyon lotor), pig (Sus scrofa), meso-american slider turtle (Trachemys venusta), brown jay (Psilorhinus morio), and great blue heron (Ardea herodias). However, all fish species detected in sediment samples were also recovered from the water samples (Table 3, S1 Table).
Species accumulation curves were calculated in EstimateS software with 1000 randomizations and classic formula for Chao 1 and Chao 2.
The three fish orders with highest number of detected species were Cichliformes (12 species), Cyprinodontiformes (11 species), and Gobiiformes (5 species) (S2 Table). Although Pterygoplichthys sp. was not detected in water and sediment samples, it was recovered from the mock eDNA study (S1 and S2 Tables) with 6,031 reads on PGM and 165,709 reads on S5 platform.
In total, 272,610 reads obtained by the PGM from the water and sediment samples showed a close match to a vertebrate species in one of the five reference libraries. The three sites with the most fish species recovered were Alvaro Obregón Viejo, Cocalitos and Xul-Ha (Fig 4, Table 4), with 24–25 fish species per site. On average, the sediment and water samples from a particular site on a particular date generated eDNA sequences for 45% (mean = 20.9; range = 15–25) of the 47 fish species detected in the survey (Table 4). Although nearly 2/3 of these reads derived from sediments (177,077 reads versus 95,533 from water), more species of vertebrates were recovered from the water samples (Fig 4). In fact, no species were recovered from the sediment samples taken at three sites (Cocalitos, Huay Pix, Xul-Ha) in December 2015 or from those preserved without buffer from Alvaro Obregón Viejo in January 2016. Although water samples revealed more species, there was evidence of temporal shifts in species recovery and DNA concentration with the lowest value in December 2015 (Figs 4 and 5A). Best results were obtained with filtered water rather than sediments for both treated and untreated samples (Figs 3, 4 and 5A).
A–Number of vertebrate species recovered for each sampling event and treatment (only PGM data for fishes are shown, circle size indicates average read count per sampling event). B–DNA concentration of sediment and water samples (one replicate per sampling event) measured on Qubit. Whiskers correspond to data within 1.5x of interquartile range (IQR).
With two exceptions, negative PCR and DNA extraction controls did not produce any vertebrate reads aside from those derived from humans (S2 Table). Two reads of Trichromis salvini were recovered from a negative control on the S5, and two reads of Mayaheros urophthalmus were obtained from a negative control on the PGM. Because both cases involved a species common in samples from the same run, they likely reflect tag-switching due to a misread UMI [60,61] or cross-contamination. Given their rarity, it is unlikely such analytical errors significantly impacted our overall conclusions.
Comparison of results from two sequencing platforms
Two sequencing platforms (PGM, S5) were used to analyze the same set of water samples from April 2016 and from the mock experiment. The PGM generated 57,689 reads that matched vertebrates while the S5 generated 1,106,574 vertebrate reads. Reflecting its 20-fold higher sequence count, the S5 recovered more fish species than the PGM (41 vs. 34). However, large shifts in the relative number of read counts for the component species were also detected. For example, Trichromis salvini comprised 59.1% of the S5 reads, but just 5.9% of those from PGM (Fig 6). This could be due to different conditions for emulsion PCR generation: OneTouch2 instrument in the PGM workflow utilizes a reagent cartridge where oil and water phases are mixed through microchannels before the resultant emulsion is pushed into a special amplification plate with a closed channel, while emulsion for the S5 was generated on the Ion Chef instrument via pipetting and transferred to a regular microplate for amplification.
Our bioinformatics pipeline for parsing blast search results enabled fast and efficient data processing, but assignments to closely related species or species complexes always require further validation. To verify the accuracy of the identifications assigned to the sequences recovered from eDNA, we generated a Neighbor-Joining tree in BOLD for detected fish species and mapped corresponding top-hit process IDs (Fig 7). In our eDNA data Cyprinodon simus/beltrani and Bramocharax-Astyanax complex represent species complexes lacking resolution with COI.
Blue–species previously reported from the Lake Bacalar region; red–species new to Lake Bacalar); orange–species complexes lacking resolution with COI: 1 –Bramocharax-Astyanax complex, 2 –Cyprinodon simus/beltrani; green–Pterygoplichthys from mock eDNA experiment.
Discussion
The major goal of this study was to evaluate the applicability of eDNA for biomonitoring fishes in a tropical oligotrophic lake. Although eDNA has often been thought to degrade rapidly, factors such as temperature, alkalinity, and trophic state [31,62] affect its stability. For example, cold temperatures, low UV-B levels, and high pH slow eDNA degradation [62], while acidity promotes it [62,63]. The overall probability of eDNA detection also depends on its production which may vary by species, by season, by density, and diet, and its loss from the study system via water discharge or diffusion [62]. The impacts of increased temperature on eDNA recovery is inconsistent. For example, a laboratory study showed that eDNA degradation increased with rising temperature, particularly in water samples from an oligotrophic lake [31]. By contrast, Robson et al [32] evaluated effects of high water temperature and fish density on the detection of invasive Mozambique tilapia in ponds via eDNA protocols and found that increased water temperatures did not affect degradation rates. However, they did detect increased rates of eDNA shedding at 35°C. In our study we detected two non-native tilapia species (O. mossambicus and O. niloticus) in Lake Bacalar (S1 and S2 Tables). Sampling methodology and sample handling are crucial for eDNA preservation. Therefore, samples were preserved in the presence or absence of buffer to compare overall DNA yield and eDNA recovery.
The present study represents the first time that the effectiveness of eDNA has been tested in an oligotrophic tropical lake where high temperatures (>27°C year-round) should speed DNA degradation. However, its water is slightly alkaline (pH 7.8) [41] which may aid eDNA preservation, perhaps explaining the better recovery of eDNA from water samples than sediments (Figs 2 and 5A). The recovery of species from sediments was low (0–8 species per site) despite the higher DNA concentration in most samples treated with buffer (Fig 5B). The low recovery may reflect the presence of inhibitors or nuclease activity although best practices were followed to minimize DNA loss [64,65]. In particular, samples were transferred onto ice immediately following collection, and DNA was promptly extracted from the samples.
Higher sequence output on the Ion S5 makes it possible to process more samples in a single run or to increase detection sensitivity. Unlike the ‘hands-on’ PGM workflow, requiring high quality Milli-Q water for sequencing and qualified personnel for operation, fully automated Ion Chef-S5 workflow with pre-filled reagent cartridges facilitates easy setup of the Chef and S5 instrument in any laboratory. The vertebrate primer sets utilized in our study can be also be used on Illumina platform.
Most prior eDNA studies have examined vertebrates in temperate freshwater streams [66], lakes [67] or sea water [68,69]. However, Lopes et al. [30] recently demonstrated that eDNA surveys were effective in revealing anurans in streams in the Brazilian Atlantic forest, while Robson et al. [32] developed eDNA protocols that employ qPCR to detect invasive fish in tropical environments. Our results further validate the effectiveness of eDNA surveys in these settings.
There are three major approaches for of eDNA detection: the use of qPCR to detect one or a small number of target species [70,71], shotgun sequencing of aquatic metagenomes [72] and PCR-based metabarcoding [28,29,66–69]. Metabarcoding studies cannot deliver information on species composition without a reliable reference database, such as that available for Mexican freshwater fishes which includes records for 93% of the 70 species known from Lake Bacalar and associated wetlands [50,53–56,73]. The present study affirmed the effectiveness of eDNA analysis as a tool for rapidly assessing the species composition of fish communities. The extraction of eDNA from just 42 liters of water collected on three dates at 14 sampling points revealed 41 of 57 of previously recorded species and added 6 potentially new records or species to the fauna. The present results mirror those obtained in a study which used eDNA to examine the composition of fish communities a coastal sea [28]. The analysis of 94 water samples obtained by sampling 47 sites in six hours revealed 40 of the 80 species known from 14 years of underwater surveys, as well as 23 new species.
The present study recovered sequences records from six fish species new to the Lake Bacalar region (Bardiella ronchus, Ctenogobius fasciatus, Gobiosoma sp., Dorosoma anale, Oligoplites saurus, and one Engraulidae). Two of the new species (B. ronchus, O. saurus) detected in Juan Sarabia and Alvaro Obregón Viejo are known from the adjacent Chetumal Bay, but have not been reported from inland waters (Valdez-Moreno, pers. obs.). The new gobiid, C. fasciatus, was detected in four localities (Xul-Ha, Pedro A. Santos, Cocalitos, Huay Pix), but was overlooked in prior field surveys because of its morphological similarity to other gobids in Lake Bacalar, especially Lophogobius cyprinoides. The other two new records require confirmation–the presence of a Gobiosoma in Huay Pix (51 reads) and a member of the family Engraulidae in Pedro A. Santos (>7000 reads).
A few sequences (13) of Lachnolaimus maximus, a commercially fished marine species, were recovered from Huay Pix, but the presence of its DNA was almost certainly mediated by human activity as fish were prepared for consumption.
As reported previously, species in two characid genera (Astyanax and Bramocharax) are difficult to discriminate using DNA barcodes as they show low divergences between species and genera [53,74]. Moreover, in a study utilizing three mitochondrial genes (Cyt b, 16S, COI) and a single nuclear gene (RAG1), Bramocharax was found to be polyphyletic, with species in this genus being sisters to different clades of Astyanax, making Astyanax paraphyletic [75]. In fact, all Bramocharax species grouped with sympatric Astyanax lineages (or even with allopatric Astyanax populations), with less than 1% divergence. In our dataset, Astyanax aeneus, Astyanax mexicanus, and Bramocharax caballeroi also formed an intermixed cluster (Fig 7) in the Taxon ID tree generated (see the dataset DS-EBACALAR in www.boldsystems.org). Moreover, recently A. aneus from Lake Bacalar was described as a new species (Astyanax bacalarensis Schmitter-Soto, 2017) based on traditional taxonomy [76]. Among the species detected with eDNA, Astyanax mexicanus was the least certain as it was only represented by 7 reads from the S5 with an average identity of 0.97 (S2 Table), which may be low quality reads.
Although Dorosoma anale was only represented by 13 reads from water at La Unión station, this species is known to occur at sites in northern Belize close to the Hondo River (Valdez-Moreno pers. obs.).
The presence of two other species (Cyprinodon beltrani (two PGM reads) and Vieja fenestrata (five S5 reads) was less certain. C. beltrani is native to Chichancanab lagoon, and cannot be distinguished from C. simus with DNA barcodes (see DS-EBACALAR in www.boldsystems.org), while V. fenestrata is native to the Papaloapan River [54] so its presence in the study area is unlikely.
Our results also revealed good eDNA recovery for other vertebrates, including rare species such as Tamandua mexicana, which was represented in both PGM and S5 sequences from a river sample (Juan Sarabia). Another interesting record involved Oreothlypis peregrina, a migratory warbler known to overwinter in the Yucatan Peninsula, but that is very morphologically similar to other species.
Biodiversity in freshwater ecosystems in undergoing losses as a result of landscape transformation, pollution, and biological invasions. In fact intentional or accidental introduction of invasive alien species is the third leading cause of global biodiversity loss [77]. In 2008 it was estimated that nearly 40% of the freshwater fish species in North America, including Mexico, were threatened by invasive species [78]. The Lake Bacalar water basin has been recognized as one of the hydrological basins with highest priority for conservation by the Mexican National Commission for the Use and Knowledge of the Biodiversity (CONABIO) [79]. Fortunately, most wetlands and lakes in this region are relatively pristine, excepting the Hondo River on the border between Mexico and Belize, which has been heavily impacted by the discharge of organic waste and pesticides, by vegetation clearing, and by the introduction of invasive such as tilapia [80], and the Amazon sailfin catfish (Pterygoplichthys pardalis), which was recently detected [59,81,82] near La Unión [59]. The impact of this declining water quality and the rising incidence of invasive species on the native fish fauna needs to be carefully monitored and eDNA-based studies could provide a cost-effective way to meet this need.
Conclusions
We developed field sampling protocols and a HTS pipeline which enabled the efficient recovery of eDNA from several tropical aquatic ecosystems. Water samples consistently revealed more vertebrate species than sediment samples although about 10% of the species were only recovered from sediments.
eDNA sequences were recovered from 75 species of vertebrates including 47 species of fishes, with six new records for the Lake Bacalar region and two other species whose detection is likely to be due to human activity. Sequences were also detected from another 28 vertebrate species including 15 birds, 7 mammals, 5 reptiles, and 1 amphibian, all species known from the watershed.
This study indicates that eDNA can be successfully applied to biomonitor vertebrates in a tropical oligotrophic lake and can aid conservation and monitoring programs in tropical areas by improving our capacity to map occurrence records for resident and invasive species. The current results should help to convince both regulatory agencies and the public that this approach can provide the detailed information on species composition needed to underpin conservation policy for tropical aquatic ecosystems.
Supporting information
S1 Table. S1 Table.
Summary of coverage for vertebrate reads for all collection events and for positive controls.
https://doi.org/10.1371/journal.pone.0215505.s001
(XLSX)
S2 Table. S2 Table.
Taxonomy, top hit Process IDs, average overlap, average score, coverage, and average identity for vertebrate species detected by eDNA in environmental samples, positive and negative controls.
https://doi.org/10.1371/journal.pone.0215505.s002
(XLSX)
Acknowledgments
We thank Jose Angel Cohuo, Miguel Valadez, and Rosaura Castro from the Instituto Tecnológico de Chetumal and Fernando Cortés Carrasco from El Colegio de la Frontera Sur for help with collecting and processing samples and Janneth Padilla Saldívar for assistance with map preparation. DNA extraction and NGS were carried out at the Centre for Biodiversity Genomics (CBG), University of Guelph through funding to PDNH from Ann McCain Evans and Chris Evans. We thank Thomas Braukmann from CBG, University of Guelph for designing the bioinformatics pipeline for processing the raw data. This paper represents a contribution from the Chetumal node of the Mexican Barcode of Life (MEXBOL) network. MVM and MEG thank CONACYT for support through the Sabbatical Stays Program (Grants 261790 and 262267).
References
- 1. Deiner K, Bik HM, Mächler E, Seymour M, Lacoursière-Roussel A, Altermatt F, et al. Environmental DNA metabarcoding: Transforming how we survey animal and plant communities. Mol Ecol. 2017;26: 5872–5895. pmid:28921802
- 2. Barnes MA, Turner CR. The ecology of environmental DNA and implications for conservation genetics. Conserv Genet. 2015;17: 1–17.
- 3. Thomsen PF, Willerslev E. Environmental DNA–an emerging tool in conservation for monitoring past and present biodiversity. Biol Conserv. 2014;
- 4. Mächler E, Deiner K, Spahn F, Altermatt F. Fishing in the water: effect of sampled water volume on environmental DNA-based detection of macroinvertebrates. Environ Sci Technol. 2016;50: 305–312. pmid:26560432
- 5. Taberlet P, Coissac E, Hajibabaei M, Rieseberg LH. Environmental DNA. Mol Ecol. 2012;21: 1789–1793. pmid:22486819
- 6. Dejean T, Valentini A, Duparc A, Pellier-Cuit S, Pompanon F, Taberlet P, et al. Persistence of environmental DNA in freshwater ecosystems. PLoS One. 2011;6: 8–11. pmid:21858099
- 7. Schmelzle MC, Kinziger AP. Using occupancy modelling to compare environmental DNA to traditional field methods for regional-scale monitoring of an endangered aquatic species. Mol Ecol Resour. 2016;16: 895–908. pmid:26677162
- 8. Goldberg CS, Strickler KM, Pilliod DS. Moving environmental DNA methods from concept to practice for monitoring aquatic macroorganisms. Biol Conserv. 2014;
- 9. Hebert PDN, Cywinska A, Ball SL, deWaard JR. Biological identifications through DNA barcodes. Proc Biol Sci. 2003;270: 313–21. pmid:12614582
- 10. Ratnasingham S, Hebert PDN. BOLD: The Barcode of Life Data System (www.barcodinglife.org). Mol Ecol Notes. 2007;7: 355–364. pmid:18784790
- 11. Hajibabaei M. The golden age of DNA metasystematics. Trends Genet. 2012;28: 535–537. pmid:22951138
- 12. Lodge DM, Turner CR, Jerde CL, Barnes MA, Chadderton L, Egan SP, et al. Conservation in a cup of water: estimating biodiversity and population abundance from environmental DNA. Mol Ecol. 2012;21: 2555–2558. pmid:22624944
- 13. Goldberg CS, Turner CR, Deiner K, Klymus KE, Thomsen PF, Murphy MA, et al. Critical considerations for the application of environmental DNA methods to detect aquatic species. Methods Ecol Evol. 2016;7: 1299–1307.
- 14. Sigsgaard EE, Carl H, Møller PR, Thomsen PF. Monitoring the near-extinct European weather loach in Denmark based on environmental DNA from water samples. Biol Conserv. 2014;183: 46–52.
- 15. de Carvalho MR, Bockmann FA, Amorim DS, Brandão CRF, de Vivo M, de Figueiredo JL, et al. Taxonomic impediment or impediment to taxonomy? A commentary on systematics and the cybertaxonomic-automation paradigm. Evol Biol. 2007;34: 140–143.
- 16. Rees HC, Maddison BC, Middleditch DJ, Patmore JRM, Gough KC. The detection of aquatic animal species using environmental DNA–a review of eDNA as a survey tool in ecology. Journal of Applied Ecology. 2014.
- 17. Comtet T, Sandionigi A, Viard F, Casiraghi M. DNA (meta)barcoding of biological invasions: a powerful tool to elucidate invasion processes and help managing aliens. Biol Invasions. 2015;17.
- 18. Bergman PS, Schumer G, Blankenship S, Campbell E. Detection of adult green sturgeon using environmental DNA analysis. PLoS One. 2016;11: 1–8. pmid:27096433
- 19. Eva B, Harmony P, Thomas G, Francois G, Alice V, Claude M, et al. Trails of river monsters: Detecting critically endangered Mekong giant catfish Pangasianodon gigas using environmental DNA. Glob Ecol Conserv. 2016;7: 148–156.
- 20. Doi H, Inui R, Akamatsu Y, Kanno K, Yamanaka H, Takahara T, et al. Environmental DNA analysis for estimating the abundance and biomass of stream fish. Freshw Biol. 2016; 30–39.
- 21. Takahara T, Minamoto T, Yamanaka H, Doi H, Kawabata Z. Estimation of fish biomass using environmental DNA. PLoS One. 2012;7: 3–10. pmid:22563411
- 22. Deiner K, Walser J-C, Mächler E, Altermatt F. Choice of capture and extraction methods affect detection of freshwater biodiversity from environmental DNA. Biol Conserv. 2014;
- 23. Turner CR, Barnes MA, Xu CCY, Jones SE, Jerde CL, Lodge DM. Particle size distribution and optimal capture of aqueous macrobial eDNA. Methods Ecol Evol. 2014;5: 676–684.
- 24. Ficetola GF, Miaud C, Pompanon F, Taberlet P. Species detection using environmental DNA from water samples. Biol Lett. 2008;4: 423–5. pmid:18400683
- 25. Wegleitner BJ, Jerde CL, Tucker A, Chadderton WL, Mahon AR. Long duration, room temperature preservation of filtered eDNA samples. Conserv Genet Resour. 2015;7: 789–791.
- 26. Yamanaka H, Minamoto T, Matsuura J, Sakurai S, Tsuji S, Motozawa H, et al. A simple method for preserving environmental DNA in water samples at ambient temperature by addition of cationic surfactant. Limnology. 2017;18: 233–241.
- 27. Deiner K, Renshaw MA, Li Y, Olds BP, Lodge DM, Pfrender ME. Long-range PCR allows sequencing of mitochondrial genomes from environmental DNA. Methods Ecol Evol. 2017;
- 28. Yamamoto S, Masuda R, Sato Y, Sado T, Araki H, Kondoh M, et al. Environmental DNA metabarcoding reveals local fish communities in a species-rich coastal sea. Sci Rep. 2017;7: 40368. pmid:28079122
- 29. Olds BP, Jerde CL, Renshaw MA, Li Y, Evans NT, Turner CR, et al. Estimating species richness using environmental DNA. Ecol Evol. 2016;6: 4214–4226. pmid:27516876
- 30. Lopes CM, Sasso T, Valentini A, Dejean T, Martins M, Zamudio KR, et al. eDNA metabarcoding: a promising method for anuran surveys in highly diverse tropical forests. Mol Ecol Resour. 2017;17: 904–914. pmid:27987263
- 31. Eichmiller J, Best SE, Sorensen PW. Effects of temperature and trophic state on degradation of environmental DNA in lake water. Environ Sci Technol. 2016; acs.est.5b05672. pmid:26771292
- 32. Robson HLA, Noble TH, Saunders RJ, Robson SKA, Burrows DW, Jerry DR. Fine-tuning for the tropics: application of eDNA technology for invasive fish detection in tropical freshwater ecosystems. Mol Ecol Resour. 2016;16: 922–932. pmid:26849294
- 33. Rivera E, Hernandez R, Carissimi AS, Pekow C. Laboratory animal legislation in Latin America. ILAR J. 2017;57: 293–300. pmid:29117400
- 34.
NORMA Oficial Mexicana NOM-062-ZOO-1999, Especificaciones técnicas para la producción, cuidado y uso de los animales de laboratorio [Internet]. 2001. Available: https://www.aaalac.org/intlRefs/SENASICA_NORM_143.PDF
- 35. Gischler E, Gibson MA, Oschmann W. Giant Holocene freshwater microbialites, Laguna Bacalar, Quintana Roo, Mexico. Sedimentology. 2008;55: 1293–1309.
- 36. Castro-Contreras SI, Gingras MK, Pecoits E, Abert NR, Petrash D, Castro-Contreras SM, et al. Textural and geochemical features of freshwater. Palaios. 2013;29: 192–209.
- 37. Centeno CM, Legendre P, Beltrán Y, Alcántara-Hernández RJ, Lidström UE, Ashby MN, et al. Microbialite genetic diversity and composition relate to environmental variables. FEMS Microbiol Ecol. 2012;82: 724–735. pmid:22775797
- 38. Bauer-Gottwein P, Gondwe BRN, Charvet G, Marín LE, Rebolledo-Vieyra M, Merediz-Alonso G. Review: The Yucatán Peninsula karst aquifer, Mexico. Hydrogeol J. 2011;19: 507–524.
- 39. Moore YH, Stoessel RK, Easley D. Fresh-water/sea-water relationship within a groundwater system, northeastern coast of the Yucatán Peninsula. Gr Wat. 1992;30: 343–350.
- 40. Perry E, Velazquez-Oliman G, Marin LE. The hydrogeochemistry of the karst aquifer system of the northern Yucatan Peninsula, Mexico. Int Geol Rev. 2002;44: 191–221.
- 41. Pérez L, Bugja R, Lorenschat J, Brenner M, Curtis J, Hoelzmann P, et al. Aquatic ecosystems of the Yucatan Peninsula (Mexico), Belize, and Guatemala. Hydrobiologia. 2011;661: 407–433.
- 42. Ivanova NV, Fazekas AJ, Hebert PDN. Semi-automated, membrane-based protocol for DNA isolation from plants. Plant Mol Biol Report. 2008;26: 186–198.
- 43. Rohland N, Reich D, Rohland and Reich , Rohland N, Reich D. Cost-effective, high-throughput DNA sequencing libraries for multiplexed target capture. Genome Res. 2012;22: 939–946. pmid:22267522
- 44. Ivanova N V., Dewaard JR, Hebert PDN. An inexpensive, automation-friendly protocol for recovering high-quality DNA. Mol Ecol Notes. 2006;6: 998–1002.
- 45. Berry D, Ben Mahfoudh K, Wagner M, Loy A. Barcoded primers used in multiplex amplicon pyrosequencing bias amplification. Appl Environ Microbiol. 2011;77: 7846–9. pmid:21890669
- 46.
Ivanova N V, Clare EL, Borisenko A V. DNA barcoding in mammals. In: Kress J, Erickson D, editors. Analytical protocols In: DNA barcodes, methods in molecular biology. Humana Press; 2012. pp. 153–182. https://doi.org/10.1007/978-1-61779-591-6
- 47. Ivanova NV, Zemlak TS, Hanner RH, Hebert PDN. Universal primer cocktails for fish DNA barcoding. Mol Ecol Notes. 2007;7: 544–548.
- 48.
Colwell RK. EstimateS: Statistical estimation of species richness and shared species from samples. Version 9. User’s guide and application. [Internet]. 2013. Available: http://purl.oclc.org/estimates
- 49. Letunic I, Bork P. Interactive tree of life (iTOL) v3: an online tool for the display and annotation of phylogenetic and other trees. Nucleic Acids Res. 2016;44: W242–5. pmid:27095192
- 50. López-Vila JM, Valdéz-Moreno ME, Schmitter-Soto JJ, Mendoza-Carranza M, Herrera-Pavón RL. Composition and structure of the ichthyofauna in the Hondo River, Mexico-Belize, using a harpoon. Rev Mex Biodivers. 2014;85: 866–874.
- 51.
Schmitter-Soto JJ. Catalogo de los peces continentales de quintana roo [Internet]. 1996. Available: http://www.nativefishlab.net/library/textpdf/20569.pdf
- 52. Schmitter-Soto JJ. Distribution of continental fishes in northern Quintana Roo, Mexico. Southwest Nat. 1999;44: 166–172.
- 53. Valdez-Moreno M, Ivanova N V., Elías-Gutiérrez M, Contreras-Balderas S, Hebert PDN. Probing diversity in freshwater fishes from Mexico and Guatemala with DNA barcodes. J Fish Biol. 2009;74: 377–402. pmid:20735566
- 54.
Miller RR. Peces dulceacuícolas de México. México, D.F.: Comisión Nacional para el Conocimiento y Uso de la Biodiversidad, Sociedad Ictiológica Mexicana A. C., El colegio de la Frontera Sur y Consejo de los Peces del Desierto México-Estados Unidos; 2009.
- 55. Neil SJ. Field studies of the behavioral ecology and agonistic behavior of Cichlasoma meeki (Pisces: Cichlidae). Environ Biol Fishes. 1984;10: 59–68.
- 56. Schmitter-Soto JJ, Gamboa-Pérez HC. Composición y distribución de peces continentales en el sur de Quintana Roo, Península de Yucatán, México. Rev Biol Trop. 1996;44: 199–212.
- 57. Gamboa-Pérez H and Schmitter-Soto J.J. Distribution of cichid fishes in the littoral of Lake Bacalar, Yucatan Peninsula. Environ Biol Fishes. 1999;54: 35–43.
- 58. Miller RR. Composition and derivation of the freshwater fish fauna of Mexico. An la Esc Nac Ciencias Biológicas. 1986;30: 121–153.
- 59. Schmitter-Soto JJ, Quintana R, Valdéz-Moreno ME, Herrera-Pavón RL, Esselman PC. Armoured catfish (Pterygoplichthys pardalis) in the Hondo River basin, Mexico-Belize. Mesoamericana. 2015;19: 9–19.
- 60. Palmer JM, Jusino MA, Banik MT, Lindner DL. Non-biological synthetic spike-in controls and the AMPtk software pipeline improve mycobiome data. PeerJ. 2018;6: e4925. pmid:29868296
- 61. Esling P, Lejzerowicz F, Pawlowski J. Accurate multiplexing and filtering for high-throughput amplicon-sequencing. Nucleic Acids Res. 2015;43: 2513–24. pmid:25690897
- 62. Strickler KM, Fremier AK, Goldberg CS. Quantifying effects of UV-B, temperature, and pH on eDNA degradation in aquatic microcosms. Biol Conserv. 2014;
- 63. Seymour M, Durance I, Cosby BJ, Ransom-Jones E, Deiner K, Ormerod SJ, et al. Acidity promotes degradation of multi-species environmental DNA in lotic mesocosms. Commun Biol. 2018;1: 4. pmid:30271891
- 64. Prosser S, Martínez-Arce A, Elías-Gutiérrez M. A new set of primers for COI amplification from freshwater microcrustaceans. Mol Ecol Resour. 2013;13: 1151–1155. pmid:23795700
- 65. Elías-Gutiérrez M, Valdez-Moreno M, Topan J, Young MR, Cohuo-Colli JA. Improved protocols to accelerate the assembly of DNA barcode reference libraries for freshwater zooplankton. Ecol Evol. 2018;8: 3002–3018. pmid:29531713
- 66. Vences M, Lyra ML, Perl RGB, Bletz MC, Stanković D, Lopes CM, et al. Freshwater vertebrate metabarcoding on Illumina platforms using double-indexed primers of the mitochondrial 16S rRNA gene. Conserv Genet Resour. 2016;8: 323–327.
- 67. Sato H, Sogo Y, Doi H, Yamanaka H. Usefulness and limitations of sample pooling for environmental DNA metabarcoding of freshwater fish communities. Sci Rep. 2017;7: 14860. pmid:29093520
- 68. Port JA, O’Donnell JL, Romero-Maraccini OC, Leary PR, Litvin SY, Nickols KJ, et al. Assessing vertebrate biodiversity in a kelp forest ecosystem using environmental DNA. Mol Ecol. 2016;25: 527–541. pmid:26586544
- 69. Andruszkiewicz EA, Starks HA, Chavez FP, Sassoubre LM, Block BA, Boehm AB. Biomonitoring of marine vertebrates in Monterey Bay using eDNA metabarcoding. PLoS One. 2017;12: e0176343. pmid:28441466
- 70. Dougherty MM, Larson ER, Renshaw MA, Gantz CA, Egan SP, Erickson DM, et al. Environmental DNA (eDNA) detects the invasive rusty crayfish Orconectes rusticus at low abundances. Journal of Applied Ecology. 2016. pmid:27773942
- 71. Klymus KE, Richter CA, Chapman DC, Paukert C. Quantification of eDNA shedding rates from invasive bighead carp Hypophthalmichthys nobilis and silver carp Hypophthalmichthys molitrix. Biol Conserv. 2015;183: 77–84.
- 72. Cowart DA, Murphy KR, Cheng C-HC. Metagenomic sequencing of environmental DNA reveals marine faunal assemblages from the West Antarctic Peninsula. Mar Genomics. 2018;37: 148–160. pmid:29223543
- 73.
Espinoza Pérez H, Gaspar Dillanes MT, Fuentes Mata P. Listado faunísticos de México. III. Los peces dulceaciucolas Mexicanos. México, D. F.: Universidad Nacional Autonoma de Mexico; 1993.
- 74. Rossini BC, Oliveira CAM, Melo FAG de, Bertaco V de A, Astarloa JMD de, Rosso JJ, et al. Highlighting Astyanax species diversity through DNA barcoding. PLoS One. 2016;11: e0167203. pmid:27992537
- 75. Ornelas-García C, Domínguez-Domínguez O, Doadrio I. Evolutionary history of the fish genus Astyanax Baird & Girard (1854) (Actinopterygii, Characidae) in Mesoamerica reveals multiple morphological homoplasies. BMC Evol Biol. 2008;8: 340. pmid:19102731
- 76. Schmitter-Soto JJ. A revision of Astyanax (Characiformes: Characidae) in Central and North America, with the description of nine new species. J Nat Hist. Taylor & Francis; 2017;51: 1331–1424.
- 77.
Sarukhán J, Whyte A, editors. Millenium ecosystem assessment. Ecosystems and human well-being: biodiversity synthesis [Internet]. Ecosystems. Washington, DC: Island Press; 2005. Available: https://www.millenniumassessment.org/documents/document.356.aspx.pdf
- 78. Jelks HL, Walsh SJ, Burkhead NM, Contreras-Balderas S, Diaz-Pardo E, Hendrickson DA, et al. Conservation status of imperiled North American freshwater and diadromous fishes. Fisheries. 2008;33: 372–407.
- 79.
Arriaga Cabrera L, Aguilar Sierra V, Alcocer Durán J, Jiménez Rosenberg R, Muñoz López E, Vázquez Domínguez E, editors. Regiones hidrológicas prioritarias: fichas técnicas y mapa (escala 1:4,000,000) [Internet]. Mexico: Comisión Nacional para el Conocimiento y Uso de la Biodiversidad; 1998. Available: https://books.google.ca/books?id=b116YgEACAAJ
- 80. Schmitter-Soto JJ, Ruiz-Cauich LE, Herrera RL, González-Solís D. An index of biotic integrity for shallow streams of the Hondo River basin, Yucatan Peninsula. Sci Total Environ. 2011;409: 844–852. pmid:21146199
- 81. Wakida-Kusunoki AT, Luis Enrique Amador-del Ángel. Aspectos biológicos del pleco invasor Pterygoplichthys pardalis (Teleostei: Loricariidae) en el río Palizada, Campeche, México. Rev Mex Biodivers. 2011;82: 870–878.
- 82.
Alfaro REM, Fisher JP, Courtenay W, Ramírez Martínez C, Orbe-Mendoza A, Escalera Gallardo C, et al. Armored catfish (Loricariidae) trinational risk assessment guidlines for aquatic alien invasive species. Test cases for the snakeheads (Channidae) and armored catfishes (Loricariidae) in North American inland waters. Montreal, Canadá: Commission for Environmental Cooperation; 2009. pp. 25–49.