We investigated the efficiency of the eDNA metabarcoding approach in estimating fish diversity of a tropical brackish environment and compared it to previous conventional surveys. The feasibility and vast potential of eDNA metabarcoding as a rapid tool in assessing fish diversity in Merbok Estuary was well demonstrated. Metabarcoding overcomes the limitations of conventional survey methods (e.g. the use of invasive/ lethal sampling gears that may disrupt habitats), resulting in a significant reduction of direct impact on the natural ecosystem41. In this regard, eDNA-based metabarcoding methods provide a non-invasive and cost-effective alternative for surveying aquatic ecosystems, and species with conservation interest. But its efficiency is contingent on the availability of a comprehensive and reliable reference taxonomic database. Furthermore, we also observed several conflicting findings; the detection of taxa unlikely to be present in the surveyed areas and the non-detection of previously recorded taxa (albeit a moderate proportion). We discuss all these points below.
Fish diversity detected via conventional surveys vs. eDNA metabarcoding
Our results lend further support to the use of an eDNA metabarcoding approach on water samples collected from the Merbok Estuary at capturing the diversity of its fish community over time-consuming conventional surveys; a finding noted in other similar studies29,31,42. As illustrated, the eDNA metabarcoding assays detected ~82% of the fish families previously recorded during the last decade in Merbok Estuary by traditional surveys and detected additional <100 species (i.e. residents, migrants, or frequenters) inhabiting Merbok Estuary, within just a two-day sampling period. Several were new records. Some of these species are regarded to be in larval stage and thus uncapturable using the sampling gears employed in previous surveys. Furthermore, taxonomic identification of larva is very challenging even when sampled17. With the capacity to differentiate species at any stage of development (i.e. larva, juvenile, or adults), the eDNA metabarcoding approach improves the accuracy of community composition estimation and provides additional information on species interactions within a community42. Thus, the eDNA assays provide a quick and powerful estimation of an aquatic ecosystem diversity through a more powerful detection efficacy28.
The species detected by eDNA metabarcoding reflect the estuarine-mangrove community in Merbok Estuary, including estuarine residents (e.g. Butis butis, Lates calcarifer, Epinephelus coioides), marine-estuarine-dependent species that include amphidromous species (e.g. Batrachomoeus trispinosus, Acentrogobius caninus, Siganus guttaus), marine species and marine migrants that use the estuary as a nursery and/or feeding grounds (e.g. Lutjanus johnii, Caranx ignobilis, Eleutheroma tetradactylum), and estuarine-freshwater-dependent species (e.g. Scatophagus argus)7,12,43.
Limitation one: absence of a comprehensive reference database
Despite its potential to uncover higher diversity in this estuary, we highlight here three limitations of the eDNA method from water samples. Firstly, the major challenge is the absence of a comprehensive reference fish database of the two genes in this area and in general, of tropical fish community. This is particularly true in megadiverse taxonomic and cryptic groups in tropical hotspot regions, such as Merbok. Based on historical literature review of formal and informal documentations (e.g.44-47), more than 500 species of fish could occur in Merbok region (coastal and brackish) with many of them being small and cryptic and not yet genetically examined. Furthermore, the significant number of misidentified species sequence entries in GenBank48 results in possible erroneous taxonomic assignment in the eDNA samples49.
In recent years, ecosystem managers are turning to eDNA metabarcoding to glean information about community composition and diversity. The target loci (i.e. COI and 12S) were purposely selected because they are frequently employed in DNA barcoding and phylogenetic studies7,50,51. However, there remain significant gaps in the reference database of several taxon groups which hampers the optimal utility of eDNA metabarcoding52. Thus, future monitoring surveys must prioritize the generation of reference sequences to include more annotated sequences in global databases (e.g. BOLD53 and NCBI GenBank48). Establishing high-quality reference databases of local biodiversity should be a prerequisite in DNA-based biodiversity monitoring. We have established a COI-based barcode of 134 fish species occurring in Merbok Estuary in a previous study7. More barcoding efforts on the region's ichthyodiversity are underway, including sequences generated from other mitochondrial markers (e.g. 12S and cytochrome b). This will elevate the efficacy of eDNA metabarcoding and offers the opportunity to re-analyse sequences generated in this study against updated databases. The eDNA dataset generated in this study will be valuable for estimating the Merbok fish community composition changes across time.
Limitation two: detection of metabarcodes from surrounding areas due to drifting tissue materials (not associated with the actual presence of whole fish)
Water movements could potentially transmit eDNA from adjacent waterways that flow into (freshwater streams) or out of (marine ecosystem) this estuary, extending the spatial coverage and hence taxonomic detection36,42. This factor needs to be considered when assessing the diversity of fish communities along a heterogeneous spatial gradient, such as estuaries and mangrove ecosystems54.
A few species from strictly freshwater families of Cyprinidae, Anabantidae, Channidae, Osphronemidae, Aplocheilidae or Poeciliidae were unexpectedly encountered through the eDNA assays. None of these have been recorded in previous surveys within the estuary but are known to occur upstream, in the freshwater part of the Merbok and Muda basins. For instance, 55 recorded the presence of the blue panchax, Aplocheilus armatus (Aplocheilidae), and the dwarf snakehead, Channa limbata (Channidae), in the streams of Gunung Jerai Forest Reserve, located ~15 km north of Merbok Estuary. We opine that the presence was due to downstream movement of DNA traces into the estuary and not the actual presence of the fish individual. A more recent diversity survey conducted in the streams of Ulu Muda Forest Reserve (located within the Muda basin and connected to the Merbok Estuary), located ~80 km northeast of our study area, recorded several taxa that correspond to our eDNA metabarcodes such as Barbodes binotatus, Mystacoleucus obtusirostris, Tor tambra, Anabas testudineus, and Channa striata56. Similarly, the detection of coral reef endemic species is likely due to eDNA transportation into the estuary by oceanic and tidal currents. This is supported by the timing of our sampling during high tide, when the ocean level reaches its maximum and seawater freely enters the estuary. 22 demonstrated that species from the surrounding marine and freshwater habitats are more easily detected during this period.
Our metabarcoding assays also detected a few taxa of conservation importance including the blacktail reef shark Carcharhinus amblyrhynchos, the blacktip reef shark Carcharhinus melanopterus, and the Indonesian shortfin eel, Anguilla bicolor, none of which had been previously recorded, thus inferring their occurrence in the study area or in nearby habitats. The discovery of the endangered Carcharhinus amblyrhynchos was unexpected. However, this species was as one of the common catches in Peninsular Malaysia a few decades ago based on a report on Malaysian shark fisheries57. Apparently, it still exists within the Merbok region and in all likelihood in several other places. Moving forward, more intensive studies to identify these areas are needed. The discovery of these threatened species highlights the value of eDNA metabarcoding as a tool for assessing biodiversity and represents a promising step towards a holistic conservation in estuaries, mangrove and coastal ecosystems20,23,41.
Limitation three: eDNA metabarcoding methods may not successfully capture whole species diversity
Despite its enormous potential, some taxa may still go undetected due to several inherent limitations of the metabarcoding assay58. The current study missed nine families previously identified by conventional surveys. This may be attributed to the following reasons: small number of individuals present in the environment, hence insufficient eDNA in the water sample59,60, rapid eDNA degradation61 , or due to primer specificity36. Often eDNA metabarcoding methods exhibit some degree of taxonomic selectivity due to primer specificity. To reduce the impact of selectivity, it is advised to use several DNA loci or primer sets62,63.
Two genetic markers were used in our study to compensate the drawbacks of using a single primer in the metabarcoding assays, such as weak taxonomic discriminatory power and primer binding biases62. We utilized COI and 12S markers in our eDNA assays, in reference to previous metabarcoding studies29,64,65. Our study showed that the COI assay produced higher reads with broader detections across taxa when compared to the 12S assay, which is similar to the findings of29,64. In comparison to the COI assay, 12S assay retrieved a higher number of species. The discrepancies in taxonomic detection between the COI and 12S assays are most likely attributable to the combination of few factors, including primer bias, differences in the completeness of the reference database, and the different taxonomic resolution of the primers21,66.
The 12S MiFish primers67 are reputed to be teleost specific and have been successfully employed in other eDNA surveys over a wide range of aquatic environments, including estuaries22,24, marine42,68, and freshwater ecosystems60,69. Despite being the standard barcode gene70, the COI marker appeared to be a less common choice as a fish metabarcoding primer. Notwithstanding the wider taxonomic coverage of COI sequences than other loci, its high variability within this standard barcode region makes it difficult to design universal fish-specific primers for short amplicons, hence reducing the COI optimal capacity in eDNA biodiversity assessment71.
Diversity patterns and composition
Biodiversity is inextricably linked to environment, especially in estuaries and mangrove ecosystems72. Based on the ecological characteristics of ecosystem landscapes in Merbok Estuary, there are three pre-defined zones (i.e. Zone A, Zone B, and Zone C). These three zones have different degrees of anthropogenic impact. Comparing the three zones: Zone B, the area with the lowest anthropogenic disturbances, harbours the highest diversity estimates based on the Chao1 and Shannon’s indices on the 12S assay (Figure 5b). Our findings agree with16 who likewise found the greatest abundance of fish species in the midstream of the Merbok Estuary. This is an indicator that optimal conditions to support high biodiversity warrants a pristine environment22. The NMDS ordination allied with PERMANOVA based on read counts of the two metabarcoding assays, confirming clustering results that three distinct fish communities were distinguishable by the designated zones. However, the results of the PERMANOVA showed that salinity measures at all sampling sites did not affect the community structuring of detected fish taxa, which appeared to minimally vary across sites in Merbok Estuary (Table 1).
Fish communities respond to many environmental variables such as salinity, water temperature, food availability, or sediment type to determine their dispersion throughout the aquatic habitats, especially in estuaries and mangrove ecosystems73. Apart from the variations observed in the zones, no other notable spatial trends in the MOTUs read abundance across the Merbok Estuary were observed. However, a detailed spatial and geographical comparison of fish communities requires a more extensive sampling coverage, a higher sample density, and comprehensive physicochemical data collection22,26, which was not the primary goal of our study. Even so, a recent study demonstrated that species assemblage composition differed noticeably across different environments on a small geographic scale, indicating the specificity of eDNA signals despite vast oceanic water flow26. Even though our study has shown that eDNA can provide valuable information on the community composition of different habitats within a specific region, we affirm that complementary conventional approaches with better spatial fidelity will allow us to assign species and attribute them to specific habitats with greater certainty supporting the eDNA method. Thus, we advocate incorporating eDNA metabarcoding sampling into traditional surveys as a best practice for performing whole biodiversity surveys in a complex ecosystem like the estuaries and the mangroves. 52 demonstrated that combining both eDNA and conventional approaches uncovers a broader taxa diversity and provides a more holistic perspective of species compositions in an aquatic environment.