Network-Based Biomonitoring:Exploring Freshwater Food Webs With Stable Isotope Analysis and DNA Metabarcoding

Compson, Z.G. and Monk, W.A. and Hayden, B. and Bush, A. and O'Malley, Z. and Hajibabaei, M. and Porter, T.M. and Wright, M.T.G. and Baker, C.J.O. and Al Manir, M.S. and Curry, R.A. and Baird, D.J. (2019) Network-Based Biomonitoring:Exploring Freshwater Food Webs With Stable Isotope Analysis and DNA Metabarcoding. Frontiers in Ecology and Evolution, 7. ISSN 2296-701X

Abstract

Threatened freshwater ecosystems urgently require improved tools for effective management. Food web analysis is currently under-utilized, yet can be used to generate metrics to support biomonitoring assessments by measuring the stability and robustness of ecosystems. Using a previously developed analysis pipeline, we combined taxonomic outputs from DNA metabarcoding with a text-mining routine to extract trait information directly from the literature. This pipeline allowed us to generate heuristic food webs for sites within the lower Saint John/Wolastoq River and the Grand Lake Meadows (hereafter called the “GLM complex”), Atlantic Canada's largest freshwater wetland. While these food webs are derived from empirical traits and their structure has been shown to discriminate sites both spatially and temporally, the accuracy of their properties have not been assessed against other methods of trophic analysis. We explored two approaches to validate the utility of heuristic food webs. First, we qualitatively compared how well-trophic position derived from heuristic food webs recovered spatial and temporal differences across the GLM complex in comparison to traditional stable isotope approaches. Second, we explored how the trophic position of invertebrates, derived from heuristic food webs, predicted trophic position measured from δ 15N values. In general, both heuristic food webs and stable isotopes were able to detect seasonal changes in maximum trophic position in the GLM complex. Samples from the entire GLM complex demonstrated that prey-averaged trophic position measured from heuristic food webs strongly predicted trophic position inferred from stable isotopes (R 2 = 0.60), and even stronger relationships were observed for some individual models (R 2 = 0.78 for best model). Beyond their areas of congruence, heuristic food web and stable isotope analyses also appear to complement one another, suggesting a surprising degree of independence between community trophic niche width (assessed from stable isotopes) and food web size and complexity (assessed from heuristic food webs). Collectively, these analyses indicate that trait-based networks have properties that correspond to those of actual food webs, supporting the routine adoption of food web metrics for ecosystem biomonitoring.