Introduction
Providing accurate biodiversity assessments is a critical goal in ecology and biogeography with estimations being constantly revised for some species-rich groups ((1) Costello et al 2017). This issue is increasingly important given the accelerating human footprint on Earth. The ongoing worldwide defaunation, characterized by massive population declines, may trigger the local or even global extinction of rare, elusive and cryptic species that are still unknown or poorly documented ((2) Barlow et al 2018, (3) Lees & Pimm 2015). Such biodiversity losses directly impact ecosystem functioning but also human health, well-being and livelihood ((4) Diaz et al 2018, (5) Duffy et al 2017). This urges scientists to improve the accuracy and to extend the breadth of biodiversity inventories and monitoring.
In the marine realm, the detection of species occurrences is particularly challenging due to the vast volume to monitor, the high diversity of habitats, the inaccessibility of some areas (e.g. deep sea) and the behavior of some species (cryptobenthic or elusive) ((6) Juhel et al 2019, (7) Brandl et al 2018). Environmental DNA (eDNA) metabarcoding is an emerging tool that can provide more accurate and wider biodiversity assessments than classical census methods particularly for rare and elusive species ((8) Garlapati et al 2019, (9) Boussarie et al 2018, (10) Fukumoto et al 2015). This non-invasive method is based on retrieving DNA naturally released by organisms in their environment, amplified by polymerase chain reaction (PCR) and then sequenced to ultimately identify corresponding species ((11) Ruppert et al 2019). However, inventorying and monitoring biodiversity using eDNA metabarcoding requires the completeness of a reference database to accurately assign each sequence to a given species (e.g. (9) Boussarie et al. 2018).
By now, only a minority of fish species are present in online DNA databases for mitochondrial regions targeted by metabarcoding markers, limiting the extent to which species diversity can be revealed by eDNA. This proportion of sequenced species is even lower in species-rich regions and poorly sampled habitats or taxa while the effort to complete genetic reference databases is long and costly. As an alternative, a diversity of Operational Taxonomic Units (OTUs) can be extracted from eDNA metabarcoding through filtering and clustering techniques ((12) Mahé et al 2014). However, the extent to which the diversity of OTUs from a limited number of eDNA samples can reveal or predict the diversity of vertebrate species in a given biodiversity hotspot has not yet been investigated. This is particularly challenging for cryptobenthic fish species that are key for reef ecosystems ((13) Brandl et al 2019) but usually missed by classical surveys ((7) Brandl et al. 2018).
The Bird’s Head Peninsula of West Papua (eastern Indonesia) is located in the center of the Coral Triangle (marine area extending from Malaysia to Solomon Islands, (14) Veron et al 2009) which is known to host the world’s richest marine biodiversity ((15) Allen & Erdmann 2012, (16) Mangubhai et al 2012). The current checklist of coastal fishes in the Bird’s Head Peninsula identifies 1,611 species belonging to 508 genera and 112 families ((17) Kulbicki et al 2013, (15) Allen & Erdmann 2012). This exceptional level of diversity and endemism is due to a combination of factors: the complex marine currents and history of sea level changes ((18) Mora et al 2003), vicariance and dispersal at various spatio-temporal scales ((19) Hubert et al 2017), a variety of habitats from karsts to many types of coral reefs ((16) Mangubhai et al 2012), the impact of geographic isolation ((19) Hubert et al 2017), the result of plates tectonic history ((20) Leprieur et al 2016, (21) Gaboriau et al 2019) and stable environmental conditions during the quaternary ((22) Pelissier et al 2014). As a result, Indonesia is endowed with an exceptional fish diversity which is still poorly known and under severe threats ((23) Exton et al 2019, (24) Jones et al 2019, (25) Ainsworth et al 2008). Predicting the level of vertebrate diversity from eDNA OTUs is thus a critical step in conservation, biogeography and ecology, particularly in marine biodiversity hotspots.
Here, using eDNA metabarcoding from 92 seawater samples across the Bird’s Head Peninsula, we (i ) assessed the diversity of coastal fish species based on an online reference database for the teleo primers region of the 12S mitochondrial rDNA gene ((26) Valentini et al 2016), (ii ) estimated the diversity of fish OTUs based on a custom filtering and clustering bioinformatic pipeline, and (iii ) tested the capacity of OTU accumulation curves to predict the regional fish diversity.