Distinguishing Signal from Noise: Understanding Patterns of Non-Detections to Inform Accurate Quantitative Metabarcoding

Abstract

Correcting for amplification biases in genetic metabarcoding data can yield quantitative estimates of template DNA concentrations. However, a major source of uncertainty in metabarcoding data is the presence of non-detections, where a technical PCR replicate fails to detect a species observed in other replicates. Such non-detections are an important special case of variability among technical replicates in metabarcoding data, particularly in environmental samples. While many sampling and amplification processes underlie observed variation in metabarcoding data, understanding the causes of non-detections is an important step in distinguishing signal from noise in metabarcoding studies. Here, we use both simulated and empirical data to 1) develop a qualitative understanding of how non-detections arise in metabarcoding data, 2) outline steps to recognize uninformative data in practice, and 3) identify the conditions under which amplicon sequence data can reliably detect underlying biological signals. We show in both simulations and empirical data that, for a given species, the rate of non-detections among technical replicates is a function of both the template DNA concentration and species-specific amplification efficiency. Consequently, we conclude metabarcoding datasets are strongly affected by (1) deterministic amplification biases during PCR and (2) stochastic sampling of amplicons during sequencing - both of which we can model - but also by (3) stochastic sampling of rare molecules prior to PCR, which remains a frontier for quantitative metabarcoding. Our results highlight the importance of estimating species-specific amplification efficiencies and critically evaluating patterns of non-detection in metabarcoding datasets to better distinguish environmental signal from the noise inherent in molecular detections of rare targets.

Towards a framework to unify the relationship between numerical abundance, biomass, and quantitative eDNA

“…In effect, metabarcoding read count is a variable that is correlated with, but ultimately a proxy index of, template eDNA concentrations in environmental samples, in much the same way that metrics such as CPUE or BPUE are indicators of N and biomass in natural ecosystems. Research is progressing on how to disentangle relationships between quantitative read count data and original template eDNA concentrations (35, 36), and we are confident that the utility of applying our framework to metabarcoding data will improve as methods progress to improve the fidelity of metabarcoding data to template concentrations. However, we nevertheless suspect that currently applying allometric corrections to metabarcoding datasets will likely be challenging in communities with low species numbers because stochastic primer amplification bias may mask or overwhelm potential allometric effects.…”

“…It is important to draw a distinction between factors influencing the interpretation of quantitative eDNA data that can affect its distribution in nature, vs. issues associated with molecular methods that can affect the assessment of that quantification. Allometry represents an 'ecology-side' variable impacting the pseudo steady-state distribution of eDNA observed in environments, similar to the impacts of factors such as transportation (17,23,24), temperature (25,26), pH (27), activity (28,29) (35,36), and we are confident that the utility of applying our framework to metabarcoding data will improve as methods progress to improve the fidelity of metabarcoding data to template concentrations. However, we nevertheless suspect that currently applying allometric corrections to metabarcoding datasets will likely be challenging in communities with low species numbers because stochastic primer amplification bias may mask or overwhelm potential allometric effects.…”

“…Metabarcoding methods, however, are likely to generate substantially more 'noise' in the relationship between quantitative eDNA data (i.e., species read counts) and template eDNA concentrations because issues like primer bias (and resulting preferential amplification) as well as community composition, can add substantial 'noise' to the relationship between quantitative metabarcoding eDNA data and species abundance (38)(39)(40). Research is progressing on how to disentangle such effects (41,42), and we are confident that the utility of applying our framework to metabarcoding data will improve as methods progress to improve the fidelity of quantitative metabarcoding data to template eDNA concentration. However, we nevertheless suspect that currently applying allometric corrections to metabarcoding datasets will likely be challenging in communities with low species numbers because stochastic primer amplification bias may mask or overwhelm any potential allometric effects.…”

“…As discussed previously, these amplification biases can limit the fidelity of relative read count data to original template concentrations in eDNA samples (32–34). This inherently limits (to a degree) the application of our framework to infer N and biomass from quantitative metabarcoding data, which can only serve as a proxy for the ‘true’ template concentrations (as discussed above in detail), although we again note that substantial progress is being made to ‘de-noise’ read-count and template eDNA concentration relationships, e.g., (34–36).…”

Archived DNA reveals marine heatwave‐associated shifts in fish assemblages

“…To this end, recent work has begun to characterize the effects of additional mechanistic processes prior to amplification, particularly the effect of subsampling processes on observed read abundances and non-detections (Egozcue et al, 2020;Gold, Shelton, et al, 2022).…”