Results

The two DNA metabarcoding assays (fwh and Crust16S) generated a combined total of 81.3 million reads across both experiments. For Experiment 1, the fwh assay produced 10.1 million raw reads, of which 2.7 million reads and 23 ZOTUs remained after removal of unassigned, contaminant, and non‐target sequences (Table S3). The Crust16S assay yielded 8.2 million raw reads, resulting in 7.4 million reads and eight ZOTUs retained after all filtering stages (Table S3). For Experiment 2, the fwh assay produced 37 million reads, which, following decontamination and removal of non‐target sequences, was reduced to 14.5 million reads and 151 ZOTUs (Table S3). The Crust16S assay generated 25.9 million raw reads, with 23.4 million reads and 43 ZOTUs retained post‐filtering (Table S3).

Experiment 1: Unaltered Sediment vs. Sugar‐Flotation

Using the fwh assay, the final dataset comprised 17 samples with an average of 152,138 (SE ± 35,411) reads per sample and a total of 23 ZOTUs. For the Crust16S library, the final dataset comprised nine samples with an average read depth of 827,550 (SE ± 235,282) reads and a total of eight ZOTUs. Essentially all samples prepared with the sugar‐flotation method identified at least one ZOTU from one of the two assays (fwh = 17/18, Crust16S = 9/18 samples). Except for a single sample from the fwh dataset, samples extracted using the standard sediment DNA extraction method failed to detect any target ZOTUs (i.e., those assigned to aquatic macroinvertebrates).

The 23 ZOTUs detected in the fwh dataset were assigned to five aquatic macroinvertebrate species, while the eight ZOTUs from the Crust16S assay matched five taxa. Only crustacean species were recorded, with Parartemia laticaudata and Triops sp. BV‐2012 detected by both fwh and Crust16S, while the remaining six species were detected by a single assay (Table S4).

Using the combined fwh and Crust16S dataset, the sugar‐flotation method significantly outperformed the standard method. All taxa were detected exclusively using the sugar‐flotation method except for a single Patcypris outback detection in one standard sample (Figure 2; Table S4). Patcypris outback was detected in all sites, and Triops sp. BV‐2012, Parartemia laticaudata , and Eocyzicus careyensis were also common. The detections of the remaining species were sporadic, with four taxa (Reticypris ‘sp. Biologic‐OSTR106’, Branchinella nichollsi , Branchinella sp., and Schizopera ‘sp. Biologic‐HARP080’) only identified from one site (Table S4).

FIGURE 2: Percentage of sites in which each taxon was detected using metabarcoding following standard sediment DNA extraction (red) and sugar‐flotation (blue) treatments. Detection frequencies are shown for each taxon as the percentage of the six sites that had at least one positive detection, with results from both metabarcoding assays combined into a single presence–absence record per taxon.

Experiment 2: Sediment Mass for Sugar‐Flotation

Of the 48 samples sequenced, 45 fwh samples (mean read depth = 323,289 ± 22,547 SE) and 35 Crust16S samples (mean read depth = 668,332 ± 72,283 SE) passed all bioinformatic and quality filtering criteria (Table S3). The breakdown of reads per sediment‐weight category for each assay is in Table S5.

Across all samples, the fwh assay detected 151 ZOTUs assigned to 24 aquatic macroinvertebrate taxa, while the Crust16S assay yielded 43 ZOTUs assigned to 11 taxa (Table S6). Across both assays, a total of 27 taxa were detected, comprising 16 taxa unique to fwh, three unique to Crust16S and eight taxa shared between the two assays. Similar to Experiment 1, most species detected were crustaceans (n = 24), although three dipterans from the Ceratopogonidae and Chironomidae were also identified. Diplostraca (seven taxa) and Ostracoda (six taxa) were the two most speciose groups, with Anostraca (four taxa), Cladocera (three taxa), Calanoida (two taxa), Cyclopoida (one taxon) and Notostraca (one taxon) also detected. Branchinella nichollsi , Daphnia queenslandensis, Eocyzicus armatus, Eocyzicus sp., Parartemia laticaudata , Patcypris outback and Triops sp. BV‐2012 were detected in all sites at least once, while Branchinella australiensis , Copepoda ‘sp. Biologic‐CALA001’, Meridiecyclops ‘sp. Biologic‐CYCL093_’, Ozestheria beleriandensis_, O. matuwa, Ozestheria sp., Tanytarsus bispinosus and Nilobezzia sp. were detected in a single site (Figure 3).

FIGURE 3: Taxa detected across the four sugar‐floated sediment weight categories used in this study. Colour of the square reflects the percentage of sites where the taxa were detected.

Using the combined dataset of both assays, the 250 g (n species = 22), 500 g (n species = 22) and 1000 g (n species = 25) sediment treatments detected essentially the same total number of species and were all higher than the 50 g treatment (n species = 13; Figure 3). Average species richness increased significantly with sediment weight (LMM; F 1,43 = 29.7, p < 0.001), showing a strong positive linear trend (t 43 = 5.08, p < 0.001). After accounting for site‐level variation, mean species richness increased from an average of 3 ± 0.8 species at 50 g to 9.5 ± 1.5 species per sample at 1000 g of sediment (Figure 4). Post hoc Tukey tests revealed that 50 g samples contained significantly fewer species than 250 g (p = 0.001), 500 g (p < 0.001), and 1000 g (p < 0.001) treatments, whereas differences among the higher sediment weights were not significant.

FIGURE 4: Dot plot displaying the mean (± standard error) species richness detected from increasing sediment weights (50, 250, 500, and 1000 g) following sugar‐flotation. Bars and asterisks between dots represent statistical significance (**p < 0.01, *** = p < 0.001). Data from fwh and Crust16S have been pooled with co‐detected species recorded only once.

Community composition analyses based on Bray–Curtis dissimilarities revealed significant effects of sediment weight (PERMANOVA; F 3,29 = 2.37, R 2 = 0.08, p = 0.008) and site (F 3,29 = 15.94, R 2 = 0.51, p = 0.001). The interaction between sediment weight and site was not significant (F 8,29 = 1.1, R 2 = 0.09, p = 0.32). Differences between sites explained a substantially greater proportion of the variation in community composition (51%) than different sediment weights (7.6%), indicating that differences among sites were the dominant driver of community structure. This trend is reflected in Figure 5 where samples from the same site tended to cluster together with less separation between the different sediment categories within each site.

FIGURE 5: nMDS plot displaying community compositional differences between sites (colours) and sediment levels (shapes) in metabarcoding data generated from Lake Way sediment samples. Shaded ellipses represent 95% confidence regions for each site.

We detected no reduction in the variability of community composition among replicate samples within weight classes as sediment weight increased. PERMDISP analyses showed no significant differences in within‐group dispersion for any site (permutation p: Site 7 = 0.2; Site 2 = 0.3; Site 3 = 0.3; Site 5 = 0.9). Three sites (e.g., Site 7, Site 3 and Site 5) showed non‐significant trends of declining variance with increasing sediment weight while one site (e.g., Site 2) displayed the reverse (Figure 6).

FIGURE 6: Site‐level patterns of community dispersion (Bray‐Curtis distance to group centroid) across sediment weight classes. Each panel shows a separate site: (a) Site 2, (b) Site 3, (c) Site 5, and (d) Site 7. Points represent individual replicate samples, and lines show linear regressions with 95% confidence intervals.

Method Comparison: Metabarcoding vs. Sediment Rehydration

The metabarcoding approach (median = 9.5 taxa, range = 8–15) identified significantly more taxa (χ 2 = 4.3, p = 0.037) than the rehydration trials (median = 5.5 taxa, range = 4–6; Figure 7a). The two methods shared most taxa (9 of 16), though the metabarcoding detected more unique taxa (7 unique taxa) than the rehydration trials (1 unique taxon; Figure 7b). Two dipterans (Tanytarsus and Nilobezzia) and one ostracod (Cyprididae), cladoceran (Macrothricidae), calanoid (Boeckella), anostracan (Branchinella), and Spinicaudata (Ozestheria) were detected in the metabarcoding data but not present in the rehydration trials, while one cladoceran (Macrothrix) hatched but was not identified by metabarcoding.

FIGURE 7: (a) Boxplot showing the difference in the number of taxa identified between the metabarcoding (blue) and rehydration trials (orange). Metabarcoding data are from the 1000 g treatment in Experiment 2. * indicates that difference in taxon richness between the two methods was statistically significant (p = 0.037). (b) Venn diagram comparing the taxa identified via only metabarcoding (blue) and rehydration trials (orange) and taxa identified by both methods (green) from dry sediments collected from Lake Way. N = total number of taxa identified by either method. To account for differences in taxonomic resolution between the two methods, all species level identifications have been raised to genus.