Using chemical structure and inocula characteristics to predictively model biodegradation rate

Abstract

Predictive biodegradation models [i.e. Quantitative Structure Biodegradation Relationship (QSBR) models] might be used as an alternative to current regulatory biodegradation tests to predict chemical persistence. Current models are mostly based on the results derived from regulatory Ready Biodegradability Tests (RBTs), which are highly variable and were not designed to provide half-life data and therefore fundamentally undermines efforts to reliably predict chemical persistence. Improvement to existing approaches for developing and verifying predictive models and their reliability, respectively, have been proposed, and the use of functional gene and 16S rRNA amplicon sequencing techniques towards identifying and quantifying the putative chemical degraders have been studied. Several QSBR models for aromatic chemicals were developed according to OECD principles. Models for mono-aromatic chemicals were verified and calibrated with experimentally determined rates (both from pure culture and natural mixed communities). Traditional test methods were combined with functional genes and 16S amplicon sequence analyses to develop a relationship between rate, chemical concentration and competent putative chemical degrader abundance. QSBR models for mono-aromatic chemicals were stable (R2 = 0.8924), robust (Q2LOO = 0.8718) and had good predictive ability (Q2F1 = 0.8829, Q2F2 = 0.8835, and Q2F3 = 0.9178). In these models, biodegradation rates were associated with electronic, lipophilic and steric descriptors, and thus provided information on the mechanisms of different rate-limiting steps associated with the biodegradation process. However, all the variation in biodegradation rates cannot be explained by the structure alone, the prevailing environmental conditions have a significant role in determining the extent of chemical degradation. Biodegradation rates (k) of chemicals in natural mixed communities were significantly correlated with the ratio of abundance of initial putative degrader abundances (X0) and the starting chemical concentration (C0) (Pearson correlation coefficient (r) > 0.9 and p-value < 0.05). Predictive models developed by relating k with X0 and C0 reliably predicted the rate of studied chemicals. Experimentally determined rates further formed the basis towards calibrating the developed QSBR models. The molecular analysis revealed that majority of identified putative chemical degraders were rare taxa, and their enrichment did not necessarily influence the overall biomass count of the microbial community, and therefore biodegradation models that only consider the overall biomass would not account for the kind of relationships found in this study. Application of 16S amplicon sequencing and functional gene analyses techniques in biodegradation studies will help in depth screening of diversity and function of microbial community in an inoculum and enables better understanding of biodegradation outcomes.