Evaluation of a diatom eDNA-based technique for assessing water quality variations in tropical lakes and reservoirs

Abstract

Freshwater bodies, both lotic (canals, rivers) and lentic (lakes and reservoirs), in the tropics have only rarely been subjected to intense monitoring regimes owing to limited expertise and resources, despite tropical habitats being among the most dynamic and threatened globally. Diatoms are an important part of assessing aquatic environmental changes in parts of the world where monitoring is well-established because of their sensitivity to a range of indicators of water quality. In such cases, diatoms have traditionally been identified based on morphology. Morphological information on tropical diatom taxa and the expertise to assemble it are sparse. However, DNA-based assessment is less reliant on taxonomic expertise and has the potential to be more cost-efficient. The application of DNA-based technology could thus expedite the incorporation of diatoms into routine monitoring in the tropics. Here we evaluate an environmental DNA (eDNA) metabarcoding workflow for diatom identifications based on 109 water samples from six freshwater reservoirs in Singapore, a highly urbanised part of tropical Southeast Asia. The six reservoirs span a range of sizes and catchment settings. We generate baseline information for lentic systems by targeting planktonic assemblages and including two commonly used barcodes, rbcL and 18S, to assess detection differences. Congruence with morphological identifications is generally low and each technique reveals distinct species lists, but overall congruence between methods improves when comparisons are at the genus level. We discuss the probable methodological and analytical causes for this incongruence and conclude that both approaches be used concurrently for generating reference databases of relatively poorly-known diatom diversity in tropical freshwater habitats. More promisingly, we find that eDNA datasets were sufficient to reveal both spatial and temporal patterns in diatom assemblages across the six study sites. We also show that regardless of the detection method used, diatom assemblages correlated with similar environmental variables. Results highlight the potential feasibility of eDNA-based diatom monitoring of freshwater habitats in Southeast Asia and in the tropics more broadly.