Introduction
Environmental DNA (eDNA) is the total pool of DNA isolated from environmental samples (Pawlowski et al., 2020). Macroorganisms such as fish are believed to produce eDNA in the form of epidermis, mucus, and feces into their outer environments (Barnes & Turner, 2016; Rodriguez-Ezpeleta et al., 2021). Recently, the eDNA-based monitoring of species distribution, abundance, and composition has been developed to target a variety of taxa in aquatic and terrestrial environments (e.g., Ficetola et al., 2008; Yamanaka & Minamoto, 2016; Ushio et al., 2018; Valentin et al., 2020). The target eDNA, which is detected by a polymerase chain reaction (PCR), allows to evaluate species presence and relative abundance, making eDNA analysis a non-disruptive, cost-effective, and high-sensitivity monitoring tool compared with traditional capture-based surveys (e.g., Thomsen et al., 2012; Miya et al., 2015; Jo et al., 2020a; Lopes et al., 2021). Although there is potential for eDNA-based biomonitoring to become an essential approach for biodiversity and ecosystem conservation, knowledge on the production source and persistence state of eDNA, as well as its transport and degradation, is lacking (Barnes & Turner, 2016; Harrison et al., 2019; Jo et al., 2021a). The lack of or scarcity in understanding the characteristics and dynamics of eDNA causes uncertainty in eDNA-based species inferences, often resulting in ecological interpretation difficulty (Hansen et al., 2018).
Specifically, the estimation of species abundance is a significant challenge in eDNA analysis (Roussel et al., 2015; Harper et al., 2018). Although the eDNA concentration shows a positive correlation with target species abundance for multiple taxa (Takahara et al., 2012; Pilliod et al., 2013; Uthicke et al., 2018; Wu et al., 2018; Ponce et al., 2021), in natural environments, these relationships are weakened compared with those in controlled laboratory conditions (Yates et al., 2019). Numerous environmental factors complicate the diffusion, retention, and degradation of eDNA in natural environments (e.g., temperature fluctuation, water chemistry, and hydrogeographic conditions), which renders eDNA quantification and its relationship with species abundance in the field unclear. A few studies previously addressed this challenge by correcting the eDNA concentration with a river flow (e.g., eDNA flux), mathematically modeling the processes of eDNA transport and degradation (Carraro et al., 2018; Levi et al., 2019; Fukaya et al., 2021). Nevertheless, the practical application of eDNA-based abundance estimation with high reliability in natural environments is still in its infancy.
Given that eDNA degradation significantly depends on the complex interactions between the eDNA state and multiple abiotic factors (Jo & Minamoto, 2021), the cellular and molecular structures of eDNA function as important factors to better understand the relationship between eDNA concentration and species abundance. Furthermore, a recent review discussed the benefit of targeting various types of eDNA beyond short mitochondrial DNA fragments (nuclear eDNA, longer eDNA fragments, and larger eDNA particles) for reliable species detection and accurate abundance estimation (Jo et al., 2021). Larger eDNA particles (eDNA detected in more significant size fractions) can more frequently possess longer eDNA fragments (Jo et al., 2020b) that persist for a shorter duration in water because of the inflow of degraded eDNA from larger to smaller size fractions (Jo et al., 2019). This finding suggests that the larger eDNA particles, likely derived from intra-cellular DNA (e.g., cell and tissue fragments), are released and assumed to be fresher and more precise biological signals than their smaller counterparts (e.g., organelle and extra-cellular DNA) (Jo et al., 2019; 2021). Therefore, the larger eDNA particles collected using a larger pore filter may accurately imitate the neighboring species abundance.
Conversely, we suspected that the larger eDNA particles do not continuously improve the species abundance estimation accuracy. Although much of macrobial eDNA is concentrated in a 1–10-µm size fraction (Turner et al., 2014; Jo et al., 2019; Zhao et al., 2021), it exists in water at various size fractions (<0.2 to >180 μm in diameter). Among them, eDNA particles collected by a filter with a considerable pore size (tens or hundreds of micrometers, e.g., tissue aggregation and scale) are clumped and distributed heterogeneously in water (Furlan et al., 2016). The stochasticity of retrieving such “huge” eDNA particles in a water sample could worsen the relationship between eDNA concentration and species abundance. Hence, the eDNA particle size and size fraction may associate nonlinearly with the relationship between eDNA concentration and species abundance.
Additionally, it is believed that longer eDNA fragments are degraded rapidly (e.g., Jo et al., 2017; Shogren et al., 2018), whereas agreement is not usually reached in previous studies (e.g., Bylemans et al., 2018). The longer eDNA fragments can persist for a shorter duration in water owing to its rapid degradation (Jo et al., 2017). Given its characteristics, the longer eDNA fragments also reflect more recent biological information than the shorter fragments. However, compared with shorter fragments, longer eDNA fragments are expected to be less frequently detected, specifically in natural environments (Jo et al., 2021b). The stochasticity and heterogeneity of longer eDNA fragments lead to higher variances of eDNA yields across replicated samplings, showing their poor applicability for accurate abundance estimation. Altogether, it should be evaluated how target marker length and particle size of target eDNA could influence the relationship between eDNA concentration and species abundance.
This study compared the relationships between eDNA concentration and species abundance among different eDNA size fractions and target marker lengths and proposed the characteristics of eDNA suitable for accurate abundance estimation. To address the issue, an aquarium experiment was performed using zebrafish (Danio rerio ). We reared the fish in experimental tanks with different individual densities, filtered the rearing water samples using other pore size filters, and quantified eDNA concentrations targeting different fragment lengths. In this study, we focused on the R2 values and slopes of the linear regressions as the parameters showing the relationship between eDNA concentration and species abundance; the former parameter represented the accuracy of the relationship, and the latter the sensitivity of eDNA concentration in response to changes in species abundance (Eichmiller et al., 2016). Furthermore, we reanalyzed the dataset from a previous study that measured the particle size distribution of Japanese jack mackerel (Trachurus japonicus ) eDNA (Jo et al., 2019), and assessed the importance of the eDNA particle size and size fraction for the abundance estimation performance.