Environmental DNA highlights the influence of salinity and agricultural run-off on coastal fish assemblages in the Great Barrier Reef region

Agricultural run-off in Australia’s Mackay-Whitsunday region is a major source of nutrient and pesticide pollution to coastal and inshore ecosystems of the Great Barrier Reef. While the effects of run-off are well documented for the region’s coral and seagrass habitats, the ecological impacts on estuaries, the direct recipients of run-off, are less known. This is particularly true for fish communities, which are shaped by the physico-chemical properties of coastal waterways that vary greatly in tropical regions. To address this knowledge gap, we used environmental DNA (eDNA) metabarcoding to examine fish assemblages at four locations (three estuaries and a harbour) subjected to varying levels of agricultural run-off during a wet and dry season. Pesticide and nutrient concentrations were markedly elevated during the sampled wet season with the influx of freshwater and agricultural run-off. Fish taxa richness significantly decreased in all three estuaries (F = 164.73, P = <0.001), along with pronounced changes in community composition (F = 46.68, P = 0.001) associated with environmental variables (largely salinity: 27.48% contribution to total variance). In contrast, the nearby Mackay Harbour exhibited a far more stable community structure, with no marked changes in fish assemblages observed between the sampled seasons. Among the four sampled locations, variation in fish community composition was more pronounced within the wet season (F = 2.5, P = 0.001). Notably, variation in the wet season was significantly correlated with agricultural contaminants (phosphorus: 6.25%, pesticides: 5.22%) alongside environmental variables (salinity: 5.61%, DOC: 5.57%). Historically contaminated and relatively unimpacted estuaries each demonstrated distinct fish communities, reflecting their associated catchment use. Our findings emphasise that while seasonal effects play a key role in shaping the community structure of fish in this region, agricultural contaminants are also important contributors in estuarine systems.

Methods
This dataset comprises eDNA and chemical data obtained from water samples collected at four locations in Queensland, Australia in 2018 and again in 2020. Three of the locations are estuaries, and the fourth is the Mackay Harbour. 10 sites were selected at each estuary, at approximately 1 km intervals from the mouth of the river going upstream, and four sites were selected in the harbour. 

For chemical analyses, water samples were collected from each site following Queensland Government protocols. Samples were collected at different depths to establish a profile along the water column and measurements were averaged for each location. All samples for nutrient and pesticide analyses were placed on ice in the field and immediately refrigerated (pesticides) or frozen (nutrients) upon returning to the field laboratory until analysis. Nutrient analyses were carried out by the Chemistry Centre of the Department of Environment and Science, Queensland Government (Queensland, Australia), a NATA (National Association of Testing Authorities) accredited laboratory. Chlorophyll a concentrations were determined by spectrophotometric procedures. Dissolved organic carbon was measured using an automated carbon analyser. Total Kjeldahl values for nitrogen and phosphorus were measured via catalysed acidic block digestion with colorimetric segmented flow analyser finish. Targeted pesticide analyses were conducted by the Queensland Government Forensic and Scientific Services (Queensland, Australia) and a NATA accredited library through direct injection using Liquid Chromatography with tandem mass spectrometry (LC-MS-MS). Water samples were filtered through a 0.2 mm membrane prior to LC-MS-MS, and the concentrations measured were of soluble pesticides (see Supplementary Material Table S2 for limits of reporting). Chemical analysis results may be found in Supplementary-Workbook-1_Env-Data-2018 and Supplementary-Workbook-2_Env-Data-2020.

For eDNA analyses, three 2 L water samples were collected at each site. Samples were filtered at the field laboratory within 16 hours of collection and stored frozen in liquid nitrogen. Further processing took place in the eDNA and Biomonitoring Lab of Macquarie University. This included eDNA extraction using Qiagen PowerWater kits (Qiagen, Netherlands), DNA amplication via PCR using two fish-specific primer sets targeting teleost and elasmobranch fishes respectively, and pooling and purification of amplified samples. Final products were sequenced at the Ramaciotti Centre for Genomics, University of New South Wales. Sequencing results were processed through the Greenfield Hybrid Analysis Pipeline (GHAP), with operational taxonomic units matched to the MitoFish reference database for taxonomic assignment with the cutoff of 0.8 or higher match similarity. Prior to statistical analyses, datasets were cleaned based on contamination from positive controls, replicates were combined, low abundance reads (<0.0001 proportion for each sample) were removed, and all the read counts were converted into presence-absence. The presence-absence OTU table may be found in Supplementary-Workbook-3_Datasets-Tele01-Elas02.