Graduation Date

Fall 2022

Document Type



Master of Science degree with a major in Natural Resources, option Fisheries

Committee Chair Name

Dr. Andre Buchheister

Committee Chair Affiliation

HSU Faculty or Staff

Second Committee Member Name

Dr. Andrew P. Kinziger

Second Committee Member Affiliation

HSU Faculty or Staff

Third Committee Member Name

Dr. Eric P. Bjorkstedt

Third Committee Member Affiliation

HSU Faculty or Staff

Fourth Committee Member Name

Dr. Darren Ward

Fourth Committee Member Affiliation

HSU Faculty or Staff

Subject Categories



Environmental DNA (eDNA) has developed into a useful tool for determining the distribution of rare aquatic species, but relatively few studies have directly compared the detection probabilities of this method with other conventional survey techniques. These comparisons can inform which method may be better suited to address study objectives. In this study, the overall goal was to compare the ability of eDNA and underwater visual count (UVC) surveys to detect juvenile coho salmon (Oncorhynchus kisutch), a species of conservation concern at the southern extent of its geographic range. Specifically, I address two objectives: (1) compare the ability of eDNA and UVC surveys to detect coho salmon and the influence of environmental covariates on detectability and (2) evaluate the utility of eDNA concentrations and habitat covariates to predict the count of coho salmon within small pools. Water samples for eDNA analysis and snorkel surveys were conducted at 96 pools across 25 stream reaches in the Smith River basin, California. I used multi-scale occupancy models to estimate method-specific detection probabilities (p) and the effect of habitat covariates, including basin area (as a proxy for discharge), residual pool depth, and large woody debris. Results showed that eDNA and UVC surveys had a high degree of agreement in detecting the presence of coho salmon at both the pool scale (93% agreement) and reach scale (80% agreement), however there were several occasions where only one method detected coho salmon. The top occupancy model, identified using Akaike’s information criterion, indicated that the detection probabilities were best predicted by method, basin area, residual pool depth, and an interaction between method and basin area. Under median habitat conditions, detection probabilities were similar and high for both methods (peDNA=91%, pUVC=89%). Residual pool depth had a slight positive effect on peDNA and pUVC. Detection probabilities for both methods were affected negatively by increasing basin size, but p declined more substantially for eDNA; at the highest basin areas, peDNA =40% compared to pUVC=78%. Finally, eDNA concentrations were a poor predictor of coho salmon count in small pools. The absence of a relationship between eDNA concentrations and fish counts is contrary to other studies and may have resulted as a consequence of the relatively small differences in counts observed between pools, which ranged from 0 to 210 individuals. Overall, this study illustrates that eDNA methods were as sensitive as UVC surveys for detecting coho presence under most conditions but could not be used to produce reliable estimates of the average observed count of the target species in this system. Therefore, these findings support the use of eDNA methods for monitoring the distributions of a rare species but indicate that implementation should be guided by study objectives and local environmental conditions.

Citation Style

American Fisheries Society


Thesis/Project Location