Subterranean ecosystems are understudied and challenging to conventionally survey given the inaccessibility of underground voids and networks. In this study, we conducted a eukaryotic environmental (eDNA) metabarcoding survey across the karst landscape of Christmas Island, (Indian Ocean, Australia) to evaluate the utility of this non-invasive technique to detect subterranean aquatic ‘stygofauna’ assemblages. Three metabarcoding assays targeting the mitochondrial 16S rRNA and nuclear 18S genes were applied to 159 water and sediment samples collected from 23 caves and springs across the island. Taken together, our assays detected a wide diversity of chordates, cnidarians, porifera, arthropods, molluscs, annelids and bryozoans from 71 families across 60 orders. We report a high level of variation between cave and spring subterranean community compositions which are significantly influenced by varying levels of salinity. Additionally, we show that dissolved oxygen and longitudinal gradients significantly affect biotic assemblages within cave communities. Lastly, we combined eDNA-derived community composition and environmental (water quality) data to predict potential underground interconnectivity across Christmas Island. We identified three cave and spring groups that showed a high degree of biotic and abiotic similarity indicating likely local connectivity. This study demonstrates the applicability of eDNA metabarcoding to detect subterranean eukaryotic communities and explore underground interconnectivity.

Six one-litre water replicates and one 50-ml sediment sample were collected from 23 cave and spring sites across Christmas Island in October 2018, totaling 159 samples across a 110km² area. DNA was extracted from half of each filter membrane and 250 mg of each sediment sample using a DNeasy PowerLyzer PowerSoil Kit (Qiagen; Venlo, Netherlands) following the manufacturer’s instructions. DNA was amplified using three previously published PCR assays (16S and 18S) to largely target bony fish, molluscs and arthropods (such as crustaceans and insects) from our mixed environmental samples (see Scientific Reports publication for more information). 

Libraries were sequenced on either a 300 cycle (for unidirectional sequencing of the 16S amplicons) or 500 cycle (for paired-end sequencing of the 18S amplicons) MiSeq® V2 Standard Flow Cell on an Illumina MiSeq platform (Illumina, San Diego, USA), housed in the TrEnD Laboratory at Curtin University, Western Australia. Unidirectional and unmerged paired-end sequencing reads were demultiplexed using OBITools (v1.2.9) and the insect package in RStudio (v1.1.423), respectively.

We have uploaded demultiplexed (unfiltered) data for public use. This is in a fastq format and corresponds to sample IDs (see Scientific Reports publication for more information). We have also uploaded a taxonomic (read abundance) matrix that has gone through our quality filtering (DADA2) pipeline and has been blasted against NCBI's GenBank (2019). This can be directly used for multivariate statistical analyses.