How, When, and Where Relic DNA Affects Microbial Diversity

Abstract

Extracellular or "relic" DNA is one of the largest pools of nucleic acids in the biosphere. Relic DNA can influence a number of important ecological and evolutionary processes, but it may also affect estimates of microbial abundance and diversity, which has implications for understanding environmental, engineered, and host-associated ecosystems. We developed models capturing the fundamental processes that regulate the size and composition of the relic DNA pools to identify scenarios leading to biased estimates of biodiversity. Our models predict that bias increases with relic DNA pool size, but only when the species abundance distributions (SADs) of relic and intact DNA are distinct from one another. We evaluated our model predictions by quantifying relic DNA and assessing its contribution to bacterial diversity using 16S rRNA gene sequences collected from different ecosystem types, including soil, sediment, water, and the mammalian gut. On average, relic DNA made up 33% of the total bacterial DNA pool but exceeded 80% in some samples. Despite its abundance, relic DNA had a minimal effect on estimates of taxonomic and phylogenetic diversity, even in ecosystems where processes such as the physical protection of relic DNA are common and predicted by our models to generate bias. Our findings are consistent with the expectation that relic DNA from different taxa degrades at a constant and equal rate, suggesting that it may not fundamentally alter estimates of microbial diversity.IMPORTANCE The ability to rapidly obtain millions of gene sequences and transcripts from a range of environments has greatly advanced understanding of the processes that regulate microbial communities. However, nucleic acids extracted from complex samples do not come only from viable microorganisms. Dead microorganisms can generate large pools of relic DNA that distort insight into the ecology and evolution of microbial systems. Here, we develop a conceptual and quantitative framework for understanding how relic DNA influences the structure of microbiomes. Our theoretical models and empirical results demonstrate that a large relic DNA pool does not automatically lead to biased estimates of microbial diversity. Rather, relic DNA effects emerge in combination with microscale processes that alter the commonness and rarity of sequences found in heterogeneous DNA pools.

Keywords: biodiversity; ecology; extracellular DNA; mathematical modeling; phylogenetic analysis; sampling theory.

FIG 1

Modeling relic DNA dynamics. (a) The amount of relic DNA in a microbial environment is determined by inputs associated with the mortality of viable individuals with intact DNA and by losses associated with the degradation of relic DNA. If the diversity of sequences contained in the relic DNA pool is sufficiently different from that in the intact DNA pool, then relic DNA may bias estimates of microbial biodiversity (as indicated by different colored boxes) when sampling from the total (intact + relic) DNA pool. (b) We developed a sampling-based simulation model to explore the effects of mixing intact and relic DNA on estimates of diversity. We populated intact and relic communities with individuals from a lognormal species abundance distribution (SAD). We altered the diversity (i.e., evenness, E) of the relic community by changing the scale parameter of the lognormal distribution describing the SAD. We then sampled and mixed the intact and relic communities so that the relic contribution to total community ranged from 0.01 to 0.96. (c) To gain mechanistic insight into how bias arises, we created a stochastic process-based model that captures features that influence relic DNA dynamics, including immigration, birth, death, and degradation (a). We simulated a range of degradation rates to achieve relic DNA pool sizes with proportions ranging between 0.05 and 0.95. To explore how degradation alters the SAD of the relic community, we explored three scenarios. First, we simulated a neutral scenario where relic DNA sequences produced by different species degrade at the same rate. Second, we simulated conditions under which physical, chemical, or biological processes reduce the degradation rate of relic DNA belonging to some species via protection. Third, we simulated “hot spots” where more abundant relic DNA sequences experience higher rates of relic DNA degradation, a condition that may arise in structured habitats where there are patchy distributions of individuals and their metabolic products (i.e., enzymes). We ran simulations for 10,000 time steps and then sampled the intact and relic communities. To quantify bias in diversity (b and c), we calculated “richness ratios” which reflect the number of species in the total DNA pool (intact + relic) divided by the number of species in the intact DNA pool in a simulation. When values for richness ratios equal 1, relic DNA has no effect on estimates of diversity; when richness ratios are >1, relic DNA overestimates true diversity; and when richness ratios are <1, relic DNA underestimates true diversity.

FIG 2

Proportion of bacterial relic DNA in different ecosystem types. We quantified the amount of intact DNA in a sample after removing relic DNA with a DNase treatment. We then estimated the proportion of relic DNA as 1 − (intact DNA/total DNA), where the total DNA concentration was quantified without DNase treatment. Relic DNA constituted an appreciable portion of the total DNA pool but was not affected by the ecosystem type from which the sample was collected (gut, soil, sediment, and water). Gray symbols are the observed data, and black symbols represent means ± 95% confidence intervals.

FIG 3

Effect of relic DNA on within-sample (alpha) bacterial diversity in different ecosystem types. We tested for the effects of bias caused by relic DNA by calculating diversity ratios for richness (a), evenness (b), and phylogenetic diversity (c). The ratios reflect the diversity of the total DNA pool (intact + relic) divided by the diversity of the intact DNA pool. Relic DNA did not bias any measures of diversity in any of the ecosystem types. Richness was calculated as the number of operational taxonomic units (97% sequence similarity of the 16S rRNA gene), evenness was calculated using Simpson’s evenness index, and phylogenetic diversity was calculated using Faith’s D index. Gray symbols are the observed data, and black symbols represent means ± 95% confidence intervals.

FIG 4

Linear regressions testing for the effect of relic DNA on measures of alpha-diversity. See the text for descriptions of how diversity metrics and diversity ratios were calculated. When included in the regression model as an indicator variable, ecosystem type had no effect on the slopes or intercepts of the regressions (P ≥ 0.26).

FIG 5

Effect of relic DNA on the among-sample (beta) bacterial diversity in different ecosystem types. (a and b) We tested for the effects of bias caused by relic DNA by calculating a beta-diversity ratio based on centroid distances. Centroid distances were estimated after performing principal-coordinate analyses (PCoAs) using taxonomic (a) and phylogenetic (b) distance metrics (Bray-Curtis and UniFrac, respectively). The centroid distance ratio was calculated on each sample within an ecosystem type and reflects the composition of the total DNA pool (intact + relic) relative to the intact DNA pool. Relic DNA had no effect on beta-diversity for any of the ecosystem types sampled. Gray symbols are the observed data, and black symbols represent means ± 95% confidence intervals.

FIG 6

Linear regression testing for effect of relic DNA on measures of beta-diversity using taxonomic distances calculated using the Bray-Curtis method (a) and phylogenetic distances calculated using UniFrac (b). See Materials and Methods and Fig. S6 for a description of how centroid distance ratios were calculated. When included in the regression model as an indicator variable, ecosystem type had no effect on the slopes or intercepts of the regressions (P ≥ 0.29).