In biomonitoring and impact assessments, the community’s metabolic capability may in fact be of more interest than conventional taxonomic identity. Taxonomic and functional profiles can respond differently to biogeography, abiotic environmental changes (e.g. organic content, metal concentration) and community processes and interactions, and exhibit different level of stochasticity and temporality. Having comprehension of both aspects can increase our understanding of community turnover and of its resilience to further changes via the computation of functional redundancy within the community (also referred to as contributional diversity).
Metagenomics has recieved considerable attention in the last few years for its ability to simultaneously determine which organisms are present, but also what metabolic capabilities they possess. However, it’s mainstream use is still limited by it’s relatively low sample throughput, cost-efficientcy, and heavy computational and data management requirements. To circumvent those issues, several programs have been developed to infer functional profiles from 16S eDNA metabarcoding (e.g. PICRUSt2). While not as accurate as metagenomics functional analyses, these methods can provide more complete functional profiles as they do not requiere high sequencing depth to assign functions, and can be particularly useful in situations where metagenomics would be prohibitively expensive, such as broad microbial routine monitoring surveys.
In this study, we aim to assess the performance of three of the most popular metabolic inference methods, namely paprica, Picrust2 and Tax4Fun2, against metagenomics and environmenntal data, in the context of salmon farm benthic surveys. In particular, we are interested in 1) comparing the taxonomic and functional diversity recovered from both eDNA metabarcoding and metagenomics, 2) assessing their sensitivity towards fish farm activities in terms of microbial turnover and correlation with environmental data, 3) comparing the accuracy of functional inference methods against metagenomics data, and 4) identifying poorly assigned functions and taxonomic groups with the aim to improve future surveys. While we anticipate the the presence/absence and relative abundance of the inferred gene families to approximate those observed from metagenomics, we expect taxonomic and functional diversity of the metabarcoding data to be substantially higher due to its sensitivity towards less abundant taxa, and therefore be at least as responsive as metagenomics to detect anthropogenic impacts.