Environmental DNA (eDNA) analysis offers a promising tool for rapid and early detection of aquatic plant invasive species, but currently suffers from substantial unknowns that limit its widespread use in monitoring programs. We conducted the first study to test the factors related to eDNA-based detectability of 2 invasive aquatic plants, Egeria densa and Myriophyllum spicatum, over extended periods of time. Specifically, we examined how plant growth stage and abundance relate to detection in semi-natural and natural conditions. We conducted a mesocosm experiment over a 10-wk period to assess changes in eDNA detection as a function of plant growth and changing biomass. We also sampled lakes with varying species abundances and resampled a subset of lakes to test temporal variability in detection.

We used multilevel occupancy modeling to determine factors associated with detection and generalized linear mixed effects modeling to assess important predictors of eDNA concentration. In mesocosm experiments, we found that detection was less reliable while plants were actively growing but improved as a function of increasing senescence. Plant abundance in tanks was a poor predictor of detection in water samples. These findings were supported by field sampling, which resulted in higher detections for E. densa during senescence periods and only weak or ambiguous relationships between eDNA and total plant abundance in lakes for both species. Within lakes, proximity to shallow photic zones and discrete plant patches were associated with increased detections and concentrations of eDNA. However, detection at the lake scale (based on 4 sampling stations) was typically successful only at the highest levels of plant abundance. Detection and concentrations of eDNA were consistently lower for M. spicatum than for E. densa in the mesocosm experiment and field sampling, suggesting that overall detectability of aquatic invasive plants varies by species.

Our results support sampling during senescence periods to improve detection, but generally low levels of detection and weak relationships with plant abundance indicate that substantial hurdles remains to implement eDNA analysis for early detection of, and rapid response to, aquatic invasive plants.