Under the karst: detecting hidden subterranean assemblages using eDNA metabarcoding in the caves of Christmas Island, Australia

Subterranean ecosystems are significantly understudied due to the inherent difficulty in accessing underground environments. It's estimated that over 80% of Australia's subterranean fauna remain undiscovered. These ecosystems, including caves, aquifers, and anchialine systems, harbor diverse taxa adapted to challenging conditions like low light, variable temperature, salinity, and dissolved oxygen. Many species are endemic due to habitat fragmentation. Traditional biospeleological surveys rely on capture-based sampling, which is often limited by inaccessibility. Environmental DNA (eDNA) metabarcoding offers a non-invasive alternative for assessing biodiversity and distribution in these challenging environments. While eDNA metabarcoding has proven effective in various environments, its application to subterranean eukaryotic communities remains underexplored. Christmas Island, with its extensive karst system and unique biodiversity, provides an ideal location to evaluate this technique. The island's karst landscape is used for phosphate mining and water supply, making a comprehensive biodiversity audit crucial. This study aimed to: (1) identify subterranean stygofauna using eDNA metabarcoding, (2) assess community composition variation, and (3) investigate potential underground interconnectivity by integrating biotic and abiotic data.

Previous research on subterranean fauna has largely relied on single-source barcoding or genome building, requiring specimen collection. This approach is limited by the difficulty of accessing subterranean environments. Studies incorporating genetic methodologies have focused on individual specimen sequencing, which is hindered by the inaccessibility of many subterranean environments. Existing research demonstrates the potential of eDNA metabarcoding for detecting microbial communities in underground water samples, but its application to eukaryotic stygofauna requires further investigation. The existing limited reference databases for subterranean fauna may hinder the accuracy of species-level identifications based on eDNA metabarcoding. Therefore, this study aimed to address the gap in research on the application of eDNA metabarcoding to survey eukaryotic stygofauna in subterranean aquatic ecosystems, particularly in the context of limited reference databases.

Field sampling was conducted during the late dry season in October 2018 across 23 cave and spring sites on Christmas Island. Six 1-liter water replicates and one 50-ml sediment sample were collected from each site, totaling 159 samples. Environmental parameters (pH, temperature, conductivity, salinity, dissolved oxygen, air saturation) were measured at each site. DNA was extracted from filter membranes (water samples) and sediment samples using a DNeasy PowerLyzer PowerSoil Kit. Three PCR assays targeting bony fish, molluscs, and arthropods were used for amplification. qPCR was used to quantify the amplicons, libraries were constructed and sequenced on an Illumina MiSeq platform. Bioinformatics analysis involved demultiplexing, quality filtering using DADA2, and taxonomic assignment using BLASTn against NCBI GenBank and a curated 16S rDNA database. Taxonomic assignments were curated, and ASVs detected in blanks were removed. Presence/absence data were analyzed using PERMANOVA, DistLM, SIMPER, and hierarchical clustering to examine community composition variation and potential underground interconnectivity.

The three metabarcoding assays yielded 35,698,221 sequencing reads. After filtering, 115 identifiable taxa were detected, representing 71 families across 60 orders from seven phyla (Chordata, Cnidaria, Porifera, Arthropoda, Mollusca, Annelida, and Bryozoa). Taxa were largely associated with marine environments (53.9%). Thirteen bony fish taxa were identified, including three putative new occurrence records. Forty-seven arthropod taxa were detected, predominantly terrestrial, but also including aquatic species. PERMANOVA analysis showed significant variation in community composition between water and sediment samples (P=0.000). DistLM analysis revealed that site type (cave/spring) and salinity explained the highest proportion of variance in community composition across all sites (9% and 6.1%, respectively). Analysis of cave sites showed longitudinal transitions and dissolved oxygen influenced community dissimilarity. Hierarchical clustering (Jaccard similarity) identified eight significantly different groups based on community composition and three groups based on environmental parameters. Three groups exhibited both biotic and abiotic clustering, suggesting potential underground interconnectivity: Whip Cave and The Grotto, Jones Spring and Waterfall Spring, and Lost Lake Cave sites 1 and 2.

This study successfully applied eDNA metabarcoding to detect a wide range of eukaryotic taxa in the subterranean environments of Christmas Island. The detection of both aquatic stygofauna and terrestrial troglofauna highlights the method's versatility. The strong influence of salinity and dissolved oxygen on community composition reflects the unique environmental conditions of these habitats. The identification of potential underground connectivity based on biotic and abiotic similarities provides valuable insights into the hydrological dynamics of the island's karst system. The results update distribution records for several species and potentially resolve unidentified specimens from previous studies. Although the reference database was incomplete, this study demonstrated the power of eDNA metabarcoding as a non-invasive method for surveying subterranean biodiversity.

This study demonstrates the efficacy of eDNA metabarcoding for characterizing multi-trophic eukaryotic subterranean diversity, overcoming limitations of traditional biospeleological surveys. The findings reveal a rich biodiversity influenced by salinity, dissolved oxygen, and spatial gradients. Future research should focus on expanding subterranean reference databases to improve taxonomic resolution and enable more detailed ecological assessments. This approach can revolutionize subterranean ecosystem research and conservation efforts by providing a cost-effective, non-invasive way to monitor biodiversity and connectivity.

The study's sampling design might not fully capture the total taxonomic richness at each site, as indicated by taxa accumulation curves. The incomplete reference database for subterranean fauna may have led to some taxonomic assignments being less precise than desired, particularly for species-level identifications. Future studies should address these limitations by increasing sampling effort and expanding reference databases, including sequences from formalin-preserved specimens.