Molecules to management: Navigating uncertainty in environmental DNA biomonitoring of coral reefs

Abstract

Environmental DNA (eDNA) tools are transforming biodiversity assessments, offering a powerful, minimally intrusive way to obtain comprehensive biological information from genetic traces in the environment. While the promise of guiding conservation decision-making across disciplines is immense, its reliability for routine monitoring remains uncertain, constrained by biological complexity and methodological limitations that influence detection dynamics. This dissertation evaluated the performance of eDNA tools across layered sources of uncertainty, from ecological variability to methodological biases, to assess their effectiveness for coral reef monitoring. First, eDNA surveys were evaluated for their ability to capture temporal variation at multiple reef sites in Hawai‘i and to identify the seasonal drivers that lead to community-wide shifts in biodiversity. Next, the impact of different eDNA collection methods on the detection of cryptic reef taxa that are otherwise difficult to survey was assessed. Finally, the sensitivity of eDNA for detecting rare species was tested by validating a novel assay for the nuisance macroalga Chondria tumulosa. To interpret inherently uncertain molecular detections, site-occupancy detection models calibrated with visual survey data were used. These models quantify false positives and false negatives, and estimate key parameters for integrating eDNA into applied conservation and management decisions. By addressing spatial, ecological, and technical sources of variability, this work offers an assessment of eDNA reliability and its readiness for integration into real-world coral reef monitoring efforts, including long-term ecological assessments, early detection of invasive species, and tracking biodiversity responses to management interventions. While the transformative potential of eDNA is clear, successful implementation depends on rigorous testing, strategic trade-offs, and transparent accounting for uncertainty.