Environmental DNA-based biomonitoring of Cuban Crocodylus and their accompanying vertebrate fauna from Zapata Swamp, Cuba

Crocodylians globally face considerable challenges, including population decline and extensive habitat modification. Close monitoring of crocodylian populations and their habitats is imperative for the timely detection of population trends, especially in response to management interventions. Here we use eDNA metabarcoding to identify the Critically Endangered Crocodylus rhombifer and the Vulnerable C. acutus, as well as vertebrate community diversity, in Cuba’s Zapata Swamp. We tested four different primer sets, including those used previously in Crocodylus population genetic and phylogenetic research, for their efficiency at detecting crocodylian eDNA. We detected C. rhombifer eDNA in 11 out of 15 sampled locations within its historical geographic distribution. We found that data analyses using the VertCOI primers and the mBRAVE bioinformatics pipeline were the most effective molecular marker and pipeline combination for identifying this species from environmental samples. We also identified 55 vertebrate species in environmental samples across the four bioinformatics pipelines— ~ 85% known to be present in the Zapata ecosystem. Among them were eight species previously undetected in the area and eight alien species, including known predators of hatchling crocodiles (e.g., Clarias sp.) and egg predators (e.g., Mus musculus). This study highlights eDNA metabarcoding as a powerful tool for crocodylian biomonitoring within fragile and diverse ecosystems, particularly where fast, non-invasive methods permit detection in economically important areas and will lead to a better understanding of complex human-crocodile interactions and evaluate habitat suitability for potential reintroductions or recovery programs for threatened crocodylian species.

Biodiversity conservation faces multiple challenges caused by anthropogenic habitat degradation, the reduction of natural land due to agricultural expansion, and the introduction of invasive alien species 1 .In addition to the human-driven causes, the lack of information on species distribution and habitat requirements limits our ability to mitigate conservation threats 2 .Reptiles are a significant component of biodiversity, and despite the global decline in their natural populations, our understanding of their conservation status is largely unknown 3 .Living crocodylians are among the most prominent apex predators and also among the most threatened groups of vertebrate taxa, with 26 currently recognized species-46% (12 out of 26) of them belonging to the genus Crocodylus 4,5 , including four Critically Endangered and two Vulnerable species 6 .All of the most threatened crocodylian species are targeted by dedicated conservation efforts, often including conservation breeding for reintroduction purposes, translocations, and increased law enforcement 7 .Among the species that are not considered threatened, most are managed as part of sustainable use programs and are harvested for their meat and skins in a considerable global trade 7 .As a result, integral population monitoring programs, that include crocodylians and their accompanying fauna (e.g., diversity of potential vertebrate prey), are critical for evaluating the success of intense conservation interventions or ensuring sustainable offtake and trade.Unfortunately, crocodylians are Table 1.Representation of C. rhombifer eDNA sequence detection at each surveyed locality in each technical and biological replicate sampled when using VertCOI primers and after data analysis in mBRAVE.The Table did not include localities where no C. rhombifer eDNA detection was found in any replicate.Average sequence length (ML), average mean similarity (MS%) to the reference haplotypes, and the total number of reads (Reads) generated from each technical replicate are included.

Discussion
Environmental (e)DNA has become a standard non-invasive technique for rapidly surveying single-species or multispecies communities around the world 19,32 .However, only one study published to date has assessed the potential for monitoring crocodylians using eDNA 31 .The mentioned study used species-specific qPCR assays to amplify crocodylian eDNA fragments from water samples collected in a controlled environment.Here, we show for the first time the potential of an eDNA metabarcoding approach to rapidly detect Cuban crocodiles and describe species richness (i.e., potential prey) found in their habitats.Our results provide both proof of concept that eDNA in natural settings detects crocodylians (in our case, mainly Cuban crocodiles), and that well-established metabarcoding assays and bioinformatic pipelines could be an immediately accessible solution.www.nature.com/scientificreports/Among the primer sets and bioinformatic pipelines we evaluated for the detection and identification of the threatened crocodylians of Cuba, the VertCOI primers analyzed in mBRAVE performed the best.The VertCOI primer cocktail amplifies a ± 185 bp sequence within the standard COI barcode region, which has several advantages.First, eDNA is a degraded DNA approach, meaning that shorter fragments are more likely to successfully amplify 19 .Second, as the mitochondrial COI region (5ʹ) has been the standard animal barcoding marker for over 20 years 33,34 , a substantial reference database facilitates species identification in the Barcode of Life Data System (BOLD 35 ).Using these primers and analytical pipeline, we successfully amplified eDNA sequences from 11 of 15 sites and identified them as C. rhombifer.However, none of the amplified C. rhombifer sequences could be matched to previously identified COI haplotypes 36,37 .This is easily explained by the lack of overlap between the fragment amplified by VertCOI in the BOLD database, which targets the 5ʹ COI region 35 , compared to that amplified by Croc_COI, which targets the 3ʹ COI region 36,37 .These previously reported haplotype-specific sequences are also not present in the mBRAVE reference datasets.We also failed to amplify sequences that could be assigned to C. acutus despite its known presence in at least seven of our 15 sampled sites 15,16 .This is most likely because the only 5ʹ COI haplotypes in the BOLD reference database represent continental C. acutus, which are known as highly divergent from the Antillean evolutionary lineage 14,38,39 .It also highlights the need for DNA reference database completeness, particularly representing Crocodylus haplotype diversity to enhance species identifications based on the target molecular region.
This lack of specific reference sequences is a limiting factor for the VertCOI primers, at least for New World crocodylians.While this issue may be resolved using species-specific primers and customized reference databases, the existing Croc_COI and Croc_CR failed almost entirely to amplify crocodylian eDNA amongst our samples.Confirmation of their ability to amplify crocodylian DNA in environmental samples would have enabled future researchers to conduct eDNA metabarcoding in the range of virtually any crocodylian species globally without the need to develop species-specific primers or assays.However, their lack of success is likely because the target fragment sizes are too long, which results in unmerged end reads, lack of primers expected specificity when used in environmental samples or, more likely, with fragments too small for the primers to bond due to the degraded state of the DNA in the environment.As a result, either novel species-specific primers targeting smaller fragments and/or species-specific qPCR/dPCR assays may be more successful.Either way, a higher representation of Crocodylus haplotypes in standard DNA reference databases, such as that utilized by mBRAVE, will be needed.
Crocodiles spend most of their time in the water, only going ashore to bask, find new water reservoirs during the dry season, capture prey near the shore, or nest 7 .Additionally, courtship, mating, seasonal movements, resting, and feeding behaviors are carried out in the water, specifically on the surface 15 .Crocodiles also move between sites by swimming on the water's surface 9 .During these behaviors, a significant amount of eDNA might be expected to be shed in the water column, which could increase the probability of finding eDNA fragments at the top of the water column and explains why the samples were collected on the water's surface of the crocodiles' ecosystem contrary to sample in depth.However, further research will be needed to evaluate whether crocodylians' shed eDNA rapidly settles and, contrary to our assumption, is more likely to be found at greater depths.
One of the main limitations of using eDNA to detect threatened crocodylians is that low population abundance leads to low DNA concentrations in the environment 40 .In these cases, the probability of detecting rare DNA can be augmented by increasing the number of samples collected and the number of PCR replicates and/ or cycles 41,42 .Although targeted detection sensitivity and specificity could also be boosted by alternative molecular techniques such as digital PCR 43 , it lacks the capacity for multispecies detection (e.g., Crocodylus and their diet) and was considered out of the scope of the present study.Other factors, such as DNA degradation 44 and low DNA concentration caused by flowing water, can also lead to misrepresentation of species occurrence in surveyed ecosystems 20,45,46 .During sampling, for example, four crocodiles (± 1.5 m total length) were observed in Estamento, presumably all C. rhombifer.In spite of this visual confirmation, only two eDNA fragments were recovered in one biological replicate at this site.The Estamento ponds are considered shallow and have very little vegetation cover compared to the other sampled sites, suggesting that higher temperatures and increased UV exposure may result in increased eDNA degradation.Three other sampled areas (Majá Parado, Zona de Liberación, and Canal) are lotic ecosystems where no eDNA fragments were amplified.These sites are located within a canal system through which a large volume of water is carried from the eastern to the western region of Zapata Swamp.The movement of water, in this case, may simply over disperse and dilute the already rare crocodylian eDNA, decreasing its probability of being sampled.Future eDNA studies of threatened crocodylians should consider increasing sampling efforts (multiple points per site), sampling from different depths, and collecting larger volumes of water to overcome these issues.
In spite of these limitations, we detected C. rhombifer at 11 of our 15 sampled localities, including Estamento, Punta Arena, and Estero de Punta Arena.These latter sites are reported as low-density areas of crocodiles within the geographic range 9 , highlighting the utility of the current approach to detecting Cuban crocodiles at sites with low density and low observation probability using traditional methods like nocturnal spotlight surveys.Despite the historical reports of the presence of crocodiles in all the sampled localities, four of them had no detections of Crocodylus eDNA fragments.These four sites correspond to those with the lowest density of crocodiles traditionally observed 9 , but also the higher anthropogenic pressures such as illegal hunting and habitat modification due to economic activities.Consequently, it is as yet unclear if crocodiles remain at these sites.In either case, our eDNA metabarcoding surveys are providing needed, updated information on the distribution of threatened Cuban Crocodylus in this critical ecosystem after nearly 30 years of no population updates.

Biodiversity within the historical distribution areas of the genus Crocodylus at Zapata Swamp
A secondary advantage of using eDNA metabarcoding to survey rare and threatened species dependent on conservation intervention is the concurrent ability to rapidly assess biological communities, including the presence www.nature.com/scientificreports/and diversity of key resources (e.g., prey or habitat) needed by these species 2,22,44,46 or the presence of potentially injurious or competing species, including invasives 28,47,48 .To date, the biodiversity of the ecosystems within the Zapata Swamp has been explored using only traditional methods (e.g., Targarona 2013), limiting the chances of detecting slight shifts in the biological community structure and cryptic or rare species.Through eDNA, we were able to update our understanding of Zapata's biodiversity, including through the detection of eight as yet reported species in the sampled sites, such as goldbelly topminnow (Girardinus falcatus).In total we detected 55 species, 16 more than the highest number identified using a single bioinformatics platform (Fig. 1).Our use of multiple primer sets and bioinformatic pipelines allowed us to overcome limitations inherent in any one reference database.For example, the MetaWorks trained datasets do not contain any C. rhombifer or C. acutus reference sequences, thus necessitating additional databases for full and accurate species identification.The 55 detected species are all potential prey for Cuban Crocodylus 9,49 .A better understanding of the distribution of important prey, especially for C. rhombifer, could facilitate decision-making for reintroduction program success.We know already that our results are likely to be painting an incomplete picture of prey availability.First, our exclusively aquatic sampling regime may limit our ability to detect terrestrial species, also known as part of the C. rhombifer diet 15,16 .Second, the primer sets/cocktails used for COI DNA amplification here are not versatile enough to amplify mollusk and amphibian DNA effectively, despite their observed presence in the sampled sites.Additional work is needed, which may also aid in the search for elusive endemics, such as the Critically Endangered dwarf hutia (Mesocapromys nanus) and Zapata Rail (Cyanolimnas cerverai).eDNA metabarcoding has allowed us to confirm the presence of three species within the genus Eucinostomus (Eucinostomus argenteus, Eucinostomus gula, and Eucinostomus havana), which are not reported beyond the genus level in Zapata Swamp regular species inventories, due difficult identification using traditional morphological methods.Altogether, our study reinforces the utility of eDNA-based biomonitoring in ecosystems across the Cuban archipelago.
Exotic and/or invasive species are increasingly problematic for conservation management, especially where they have a high potential to negatively impact native species 50 , and alien species are among the most critical threats to reptile conservation in many regions 3 .Our eDNA results detected at least three species that have never been observed or detected in Cuba to date, much less the Zapata Swamp area, including little egret (Egretta garzetta) and korogwe tilapia (Oreochromis korogwe).These would be considerable extralimital observations for these species, including their known extralimital range, and thus, assignment to them at the species level may be because of issues with the reference databases or lower marker resolution to distinguish species within these genera.The other species are the red-eared slider (Trachemys scripta) and fish of the genus Clarias (from 10 of the 15 sampled localities).The latter is a known invader in Cuba 51 , and this genus is a known predator of hatchling crocodiles, in addition to competing with young crocodiles for food resources (personal observations).In the case of the former, our approach did not allow us to distinguish the confidence of species assignment-a known limitation with eDNA studies that must refer to BOLD and GenBank reference sequences, among others.In the case of closely related species where one or more are absent from reference databases, sequences can be assigned to erroneous species or to more than one species, reducing confidence in reporting biodiversity accurately.For example, Trachemys scripta has been reported as introduced in the Cayman Islands and Puerto Rico 52 , but never in Cuba.Though 2958 eDNA fragments amplified using VertCOI primers had 99.11% mean similarity with the reference sequence BOLD:AAK2050; this sequence has been variably assigned to four different Trachemys species from the Caribbean region, including the endemic T. decussata.It is unclear at this time if our results signal a new species introduction or, most likely, if only the 185 bp COI molecular fragment lacks enough resolution to distinguish closely related Trachemys species unambiguously.

Conclusions for conservation of Cuban Crocodylus
Understanding how the environmental variables affect the use and selection of the habitat by crocodiles is of particular interest for improved management of the genus Crocodylus in Cuba.In the case of the Critically Endangered C. rhombifer, it permits the identification of appropriate locations for recovery or reintroduction initiatives.The benefits of eDNA metabarcoding are significant in both large spatial-scale studies and in fragile ecosystems.For instance, we were able to rapidly detect crocodiles at sites that are difficult to access and would normally require days of searching due to historically low abundance.Our inability to detect crocodiles at other sites may be due to true absences or may be due to false negatives for species resulting from partial PCR inhibition or DNA degradation due to environmental factors 2,40 .We also have shown that eDNA metabarcoding allows for quick identification of invasive species to inform appropriate management and mitigation decisions.A better understanding of invasive species distribution in Zapata Swamp will facilitate focusing conservation efforts and limited resources on the most sensitive areas, including C. rhombifer breeding sites.
The application of multi-marker approaches for eDNA metabarcoding, such as the one conducted in the present study, allowed for the detection of Crocodylus, as well as a comprehensive list of aquatic prey and invasive species in their habitat.Moving forward, Crocodylus molecular biomonitoring programs will benefit from building custom COI reference databases.In the future, it may also be possible to increase the target eDNA fragment length within the standard animal barcode region to better reconstruct specific haplotypes using second-generation HTS technology.Unfortunately, hybrid identification will not be possible using standard mitochondrial fragments.It would eventually require nuclear marker combinations, which are still undescribed in this system 39 and may ultimately not be feasible using an eDNA approach.Altogether, the eDNA metabarcoding approach has proven helpful for detecting crocodiles in the wild, with a limited Cuban C. acutus detection rate due to the incomplete DNA reference databases.The multispecies identification approach has also allowed for identifying crocodiles' prey and invaders impacting the natural ecosystems they inhabit.As a result, this approach facilitates a more rapid update of the geographic range of Cuban Crocodylus and for focusing conservation management efforts on more sensitive crocodile populations or suitable sites for recovery or reintroduction programs.

Study area
The Zapata Swamp, located in the southern region of Matanzas province, Cuba (Fig. 2), is the largest wetland in the insular Caribbean 53 .This area hosts high levels of biodiversity and endemism in several ecosystems, including freshwater lagoons and channels, grasslands, brackish water zones, estuaries, forests, and coral reefs 54 .Two hydro-climatic seasons regulate the water levels and the periods of flooding of these habitats that cover 4250 km 2 .We selected 15 sampling localities to collect water samples within the historical distribution of the genus Crocodylus at Zapata Swamp (Figs. 2, 3).A brief description of those areas is offered below:

Zanja del Diez
Canal (~ 9 km length) located in the southwestern region of the Zapata Swamp.Its average width is 2 m, and its depth ranges between 0.3 and 1 m.Salinity values depend on the distance from the sea, the tides, and the hydroclimatic season.This channel is considered the main genetic exchange route between crocodile populations of the inland freshwater ecosystems (typically C. rhombifer), and the coastal populations (typically C. acutus), so hybrids between Cuban Crocodylus are also expected 15 .

Zanja del Nueve
This canal differs from the rest by its construction on limestone rock following the terrain's depressions and being narrower (~ 1 m).Grasslands of swamps and mangroves dominate this channel.Salinity values are subject to the same factors as in the Zanja del Diez.Mainly C. rhombifer and hybrids are expected in the site 15 .

Estamento
Calcic savannah with small lagoons with variable water levels and salinity associated with rainfall.The vegetation consists primarily of xeromorphic coastal and mangrove forests around the lagoons.Both Cuban Crocodylus and their hybrids are expected to be present 15 .www.nature.com/scientificreports/methods (e.g., spotlight surveys) in this area 15,16 .

Estero de Punta Arena
System of canals through which the water of the northern region of Zapata Swamp drains towards the sea.The margins are covered with mangrove forests.Salinity values depend on the season of the year and the tides.Mainly crocodile hybrids are observed in this site 15,16 .

Laguna de Vitorino
Shallow pond with a surface area of 0.9 km 2 covered mostly with mangroves and swamp forests.This lagoon is located within a savannah ecosystem, consisting of the only water reservoir during the dry period.American crocodiles and hybrids are typically observed in this site 15,16 .

Lagunitas
Small freshwater pond of approximately 0.03 km 2 surrounded by swamp forest that floods seasonally, allowing the aquatic fauna to move to other ecosystems within the Zapata Swamp.Hybrid crocodiles are mainly expected in this area 15,16 .
Zanja Santo Tomás Canal (12.8 km length) built in the early 1900's to transport wood and charcoal.This channel crosses several ecosystems of swamp grassland and mangrove forests.Crocodile hybrids are frequently observed in this area 15 .

Canal de los Patos
Canal system built in the 1960s to facilitate the water runoff from the eastern swamp to the western swamp.Grasslands and swamp thickets dominate this area, with characteristic savannah vegetation, evergreen forests, and occasional mangroves.This locality is a frequent fishing spot, and there are reports of a few Cuban crocodile individuals, as well as hybrids 15,16 .

Laguna del Tesoro
One of Cuba's most important natural lakes, with a surface area of 9 km 2 and a depth of 10 m.Swamp grasslands surround it, showing a high degree of human impact due to tourism and fishing.It represents a typical habitat for the Cuban crocodile, although hybrids have also been found 15,16 .

Laguna Nueva
Pond spanning approximately 0.4 km 2 , and with a maximum depth of 4 m, connected to Laguna del Tesoro and Canales de Hanábana throughout channels.The fauna and flora composition are similar to that of Laguna del Tesoro, including Cuban crocodiles and hybrids 15,16 .

Fauna refuge "Canales de Hanábana"
Canal systems where Swamp grasslands cover 87% of the total area, and swamp forests and freshwater vegetation occupy the rest of the ecosystem.Currently, there are 175 species of terrestrial vertebrates reported in the area, 34 of which are endemic 54 .Our study includes three localities within this area: Majá Parado, Zona de Liberación, and Canal.Cuban crocodiles are mainly expected in this area, which have mainly been released as part of a reintroduction program (Figs. 2, 3).

Sample collection, filtration, and storage
We collected 2 L of water from the surface from a single point in each locality selected randomly (Fig. 3) using two sterilized 1 L glass bottles, each of which was processed as a separate biological replicate.We also collected two samples of water (1 L each) from one pond (~ 9 m 3 ) where 11 captive C. rhombifer (~ 2 m long) are kept at the Zapata Swamp Crocodile Farm (ZSCF), representing a positive control.Prior to sampling, we cleaned containers and lids with 50% bleach, rinsed 3× with distilled water, and sterilized them in an autoclave (121 °C and 1 atm) for 30 min.We fully immersed the containers in the water and put the lid right after it filled up wearing single-use gloves to collect the sample and avoid cross-contamination.We transported samples at environmental temperature in a box to limit UV exposure to a field lab for further processing.At the field lab, sampling personnel changed into clean clothing to prepare the work area.We filtered samples using a vacuum pump and magnetic filtration cups with a three-piece manifold connected to a receiver container and a nitrocellulose mixed ester membrane filter (diameter 47 mm, pore size 1 μm) mounted between the mating surfaces of the filtration cups.We filtered each 1 L bottle per sampling locality separately, and the number of filters per 1 L sample differed due to the presence of suspended particulates.Filtering resulted in 83 total membranes, including the field negatives (2 L of distilled water filtered for each collection site), lab negatives (unused filter as a DNA extraction negative control), and positive controls (2 L of water filtered from the ZSCF breeding pen).We sprayed filtration cups and forceps with DNA-away surface decontaminant (Molecular Bioproducts, USA) and rinsed 3× with distilled water between samples.Once out of the field, we stored filters in a cool, dry room in a sterilized pack with silica gel.

DNA extractions
We extracted DNA from every filter membrane independently following a modified Qiagen DNeasy Blood and Tissue Kit specifically for eDNA 55 .We transferred each membrane into a sterile Petri dish using single-use pipette www.nature.com/scientificreports/tips and cut it into two halves using a sterile razor blade.We further cut each half into smaller pieces (~ 1 cm) before transferring them into separate 2.0 mL microcentrifuge tubes (one per half) containing approximately 250 g of 1 mm diameter glass beads.We then added 380 μL of ATL buffer (Qiagen) and put them into a Tissue-Lyser (Qiagen) for 5 min at 25 Hz, after which they were spun down for 30 s at 11,000×g.We repeated the tissue disruption process twice for each subsample after inverting the tube holders.Next, we added 20 μL of Proteinase K, vortexed for 10 s, and spun down for 30 s at 11,000 g before incubating at 56 °C at 700 rpm overnight in a light duty orbital shaker (Ohaus).We vortexed for 15 s and centrifuged for 30 s at 11,000×g again.We added 400 μL of buffer AL, vortexed for 10 s, and incubated at 56 °C for 10 min.Following incubation, we added 400 μL of 95-100% ethanol and briefly vortexed.We transferred 640 μL of the mixture into a DNeasy Spin Column, settled in a 2 mL collection tube, centrifuged for 1 min at 11,000×g and repeated this step twice until transferring all the lysate.We added 500 μL of buffer AW1 (Qiagen) to wash the material retained on the silica membrane of the DNeasy Spin Column, and replaced the collection tube with a new 2 mL tube, after centrifuging for 1 min at 11,000×g.We washed the membrane again with 500 μL of buffer AW2 (Qiagen) and centrifuged for 5 min at 17,000×g.We carefully transferred the Spin Columns to a LoBind microcentrifuge tube (LoBind Eppendorf), added 200 μL of 70 °C prewarmed buffer AE (Qiagen), incubated for 15 min at room temperature and centrifuged for 5 min at 11,000×g.We stored the LoBind tubes with DNA extract at − 20 °C.We performed a second elution step using 100 μL of prewarmed buffer AE to collect a reserve elution, which we stored in LoBind tubes at − 20 °C.Before amplification, we created a master DNA extract tube per sample by pooling across the extract products of each filter membrane that resulted from each 1 L sample.This resulted in 49 master extracts being amplified (two for each of the 15 sampling sites plus the positive control pond, one for each field negative control, and one lab negative control).

PCR amplification, library preparation, and MiSeq sequencing
We amplified all extract products in triplicates with four different PCR primer sets (Table 2).First, we tested Croc_COI 36 and Croc_CR 14,56 , which amplify 548 and 458 bp fragments, respectively, of the mitochondrial cytochrome c oxidase subunit I (COI) and control region (CR).These markers were developed and used as part of several prior crocodylian population genetic and phylogenetic studies and represent the most common haplotype reference sequences in public databases for Neotropical Crocodylus species 14,36,[56][57][58][59] .Second, we tested VertCOI 60,61 , a combination of versatile primer cocktails that amplify a 185 bp fragment of the COI mtDNA region with seven nucleotide substitutions between reported C. rhombifer and Cuban American crocodile haplotypes and that has been used widely in our lab in other eDNA metabarcoding studies (unpublished data).We additionally used VertCOI to simultaneously detect and describe the complement of biodiversity to be found at these sites, including available crocodylian prey.Finally, we used the MiFish primers 25 , which amplify a 220 bp fragment of the mitochondrial 12S gene region, to increase the success of fish detection as part of the available diet to crocodylians in Zapata Swamp.
From each master extract tube, we amplified three technical replicates.We conducted PCRs in 25 μL reaction volumes containing 12.5 μL of 2× KAPA HiFi HS ReadyMix (Roche), 0.2 μM of primers with Illumina adaptors, and 2.5 μL of eDNA extract.We included a negative field control for every batch of amplified samples, and a sequencing blank (20 μL of molecular-grade water) was added to each plate previous to pooling and normalization of the libraries.Amplifications were accomplished with the thermocycling conditions indicated in Table 2.We amplified each sample a second time using the first PCR products as template after a clean-up using 1× NGS magnetic beads (Macherey-Nagel) following the manufacturer's protocol.In this round, we amplified the target regions using dual-index primer combinations for each sample where the sequence of index primers was equivalent to the Nextera XT Index Kit (Illumina).We performed round two PCRs in 50 μL reaction volumes, including 5 μL of cleaned PCR product, 5 μL of each index primer (10 μM), 25 μL of 2× KAPA HiFi, and 10 μL of molecular biology grade water.Cycling conditions were as follows: 95 °C (180 s), 12 cycles at 95 °C (30 s), 55 °C (30 s), 72 °C (30 s), and a final extension at 72 °C (300 s).We visualized second-round PCR products on Table 2. List of primers used in the present study (without including the Illumina adapters 67 ) and thermocycling conditions followed during the first PCRs for eDNA metabarcoding library preparation for each primer set.

Figure 1 .
Figure 1.List of identified taxa from the eDNA fragments amplified from the water samples collected in this study after using two mitochondrial regions (COI and 12S) and three bioinformatic pipelines for data analysis.The taxa within the boxes represent species for which the markers used resulted in lower species resolution.*Newly recorded species, **Invasive taxa.