Insights into the representativeness of biodiversity assessment in large reservoir through eDNA metabarcoding

C11, representing the largest tributaries, experience much lower renewal rates. This should lead to the establishment of distinct communities, minimally influenced by eDNA from the reservoir’s main body. To assess the impact of sampling design with and without these stations, we constructed rarefaction curves to determine the number of samples to achieve our observed number of MOTUs. When any of the C01, C11, or C10 samples were excluded, the plateau area of the curve did not change significantly, remaining around 10 to 12 samples (Fig 6). However, when using only these three stations, the rarefaction curve almost reached the pla- teau shape in a smaller sample size (~8 samples) (S5 Fig). These data show that our sampling effort with 42 samples and 14 stations was excessive, resulting in a considerable amount of redundant information. For this reason, we suggest that for the TI reservoir, a sampling size with only three strategically positioned stations with more samples within each one would be capable of capturing the same number of MOTUs as 42 samples within 14 stations. The loca- tion with higher diversity was the one with higher current speed (C01), supporting that the observed diversity does not follow a geographic pattern per se and, instead, it can be a result from either abiotic or biotic factor iterations, still unknown. Either way, the observed diversity could be expected at random sampling with at least 12 sites across the entire reservoir. When considering each replicate as a sample, the diversity is high and multiple species

could be detected. However, the species distribution is very similar among them. Therefore, increasing the sampling effort by adding replicates in fewer locations could increase the num- ber of taxa. It is not the first time that increasing the number of replicates was suggested for river systems. For instance, Shaw et al. [19] already emphasized the same discussion when evaluating a river area in South Australia, where the authors recommend more replicates to increase the chance of detecting DNA fragments from taxa that are patchily distributed, with low abundance, low biomass or rare occurrence. Conclusions We constructed a large-scale biodiversity inventory to evaluate the diversity of a reservoir’s communities, focusing specifically on metazoans and ichthyofauna. Through the use of vari- ous descriptive and statistical metrics, we assessed the robustness of our biodiversity inventory and determined the optimal experimental design. We found that conducting consistent sam- pling at fewer sites with more replicates was more effective in capturing biodiversity. Metabar- coding proved to be a valuable tool, corroborating previous reports based on traditional monitoring surveys and identifying several previously unreported species. Supporting information S1 Fig. Metrics and spatial information regarding the Três Irmãos (TI) reservoir for hydrodynamic modeling. A) Bathymetry distribution across the reservoir. Darker shades show deeper sites. B) Biquadratic quadrangular finite elements for spatial discretization of the model. (PNG)

S2 Fig. Hydrodynamic modeling for the entire Três Irmãos (TI) reservoir. Maximum cur- rent speed as measured during the A) rainy and B) dry seasons. (PNG) S3 Fig. Computed water renewal rate for the Três Irmãos (TI) reservoir. The water renewal is shown moving from Nova Avanhandava to Três Irmãos reservoir during A) rainy and B) dry scenarios. (PNG) PLOS ONE eDNA metabarcoding reservoir PLOS ONE | https://doi.org/10.1371/journal.pone.0314210 January 24, 2025 17 / 22