Meta‐analysis supports further refinement of eDNA for monitoring aquatic species‐specific abundance in nature

Abstract

The use of eDNA to detect the presence/absence of rare or invasive species is well documented and its use in biodiversity monitoring is expanding. Preliminary laboratory research has also shown a positive correlation between the concentration of species‐specific eDNA particles and the density/biomass of a species in a given environment. However, the extent to which these results can be extended to natural environments has yet to be formally quantified. We collated data from experiments that examined the correlation between eDNA and two metrics of abundance (biomass and density) and, using mixed‐effects meta‐analysis, quantified the strength of that correlation across laboratory and natural environments. We found that eDNA particle concentration was more strongly correlated with abundance in laboratory environments compared to natural environments, accounting for approximately 82% and 57% of the observed variation in abundance, respectively. We found some evidence of potential publication bias that may have impacted the estimation of the correlation in natural environments; after smaller studies were removed from the dataset, eDNA particle concentration accounted for approximately 50% of the observed variation in abundance in natural environments. No effect of abundance metric was found on the strength of correlation between eDNA particle concentration and abundance. Despite a weaker general correlation in natural environments, eDNA concentration often still explained substantial variation in abundance. eDNA research is still an emergent field of study; with only moderate improvements in technology or technique, it could represent a powerful new tool for quantifying abundance.

The relationship between eDNA particle concentration and organism abundance in nature is strengthened by allometric scaling

“…Molecular techniques that quantify 57 the concentration of environmental DNA (eDNA) particles represent a promising tool, with 58 recent studies demonstrating support for a correlation between eDNA concentration and 59 abundance (Pilliod, Goldberg, Arkle, & Waits, 2013; Takahara, Minamoto, Yamanaka, Doi,& 60 Kawabata, 2012; Thomsen et al, 2012). For example, laboratory studies have demonstrated a 61 strong correlation between eDNA concentration and abundance (Eichmiller,Miller,& Sorensen,62 2016; Klymus, Richter, Chapman, & Paukert, 2015), exhibiting a mean correlation coefficient of 63 0.9 (R 2 = 0.81) (Yates, Fraser, & Derry, 2019). Studies in nature, however, have generally found 64 weaker correlations than laboratory studies, with a mean correlation coefficient of 0.71-0.75 (R 2 65 = 0.51-0.57) (Yates et al, 2019).…”

“…For example, laboratory studies have demonstrated a 61 strong correlation between eDNA concentration and abundance (Eichmiller,Miller,& Sorensen,62 2016; Klymus, Richter, Chapman, & Paukert, 2015), exhibiting a mean correlation coefficient of 63 0.9 (R 2 = 0.81) (Yates, Fraser, & Derry, 2019). Studies in nature, however, have generally found 64 weaker correlations than laboratory studies, with a mean correlation coefficient of 0.71-0.75 (R 2 65 = 0.51-0.57) (Yates et al, 2019). Although correlations remain moderately strong in nature, 66 much of the variation in eDNA particle concentration across environments often remains 67 unexplained.…”

“…Wild populations 81 often exhibit substantial inter-population variation in the distribution of individual biomass 82 (Donald, Anderson, Mayhood, Anderson, & Correlations, 1980;Guernon, Yates, Fraser, & 83 Derry, 2018; Millien et al, 2006;Sebens, 1987), which may in turn scale to affect overall 84 population-level rates of eDNA production (Maruyama et al, 2014) and partially account for the 85 substantial unexplained variation observed between eDNA concentration and traditional metrics 86 of abundance (e.g. density and biomass) in nature (Yates et al, 2019). 87…”

The utility of a metagenomics approach for marine biomonitoring

“…Alternatively, rather than amplifying and sequencing standardized DNA barcode regions from samples, it is possible to sequence random segments of DNA directly from the sample-essentially the whole genome shotgun sequencing of environmental samples, which is a technique commonly referred to as "metagenomics" (Handelsman et al 1998;Chen and Pachter 2005;Tringe and Rubin 2005). Because there is no amplification of the DNA, the read counts should be directly proportional to the proportion of starting material, which itself is correlated to biomass (Yates, Fraser, and Derry 2019;Salter et al 2019;Bista et al n.d.). Curiously, although this was the first genomics approach used to characterize ocean biodiversity (Venter et al 2004), it has since fallen out of favour except in the microbial sphere (Gregory et al 2019) or in small systems such as the analysis of gut contents (Srivathsan et al 2015) or small mock communities (Bista et al n.d.).…”

Environmental DNA facilitates accurate, inexpensive, and multiyear population estimates of millions of anadromous fish

“…Recent research suggests that eDNA quantified with real-time quantitative polymerase chain reaction (PCR) or digital-droplet PCR can provide a proxy for actual abundance in controlled experiments (Rees et al, 2014), in ponds (Lacoursière-Roussel, Côté, Leclerc, & Bernatchez, 2016;Takahara et al, 2012) in streams (Doi et al, 2015;Levi et al, 2019;Lodge et al, 2012;Tillotson et al, 2018;Wilcox et al, 2016) and in marine bays (Plough et al, 2018). However, the efficacy of environmental DNA based indices of abundance in natural settings have produced mixed results (Yates, Fraser, & Derry, 2019) and have not yet been assessed in a management context for long-term population monitoring.…”