Thomas J. Kwak W. Gregory Cope Ryan J. Heise Robert B. Nichols Krishna Pacifici Tamara J. Pandolfo 2017 <p>Mussel monitoring data are abundant, but methods for analyzing long-term trends in these data are often uninformative or have low power to detect changes. We used a dynamic occurrence model, which accounted for imperfect species detection in surveys, to assess changes in species occurrence in a longterm data set (1986–2011) for the Tar River basin of North Carolina, USA. Occurrence of all species decreased steadily over the time period studied. Occurrence in 1986 ranged from 0.19 for <i>Utterbackia imbecillis</i> to 0.60 for <i>Fusconaia masoni</i>. Occurrence in 2010–2011 ranged from 0.10 for<i> Lampsilis radiata</i> to 0.40 for<i> F. masoni</i>. The maximum difference between occurrence in 1986 and 2011 was a decline of 0.30 for <i>Alasmidonta undulata</i>. Mean persistence for all species was high (0.97, 95% CI ¼ 0.95–0.99); however, mean colonization probability was very low (,0.01, 95% CI ¼ ,0.01–0.01). These results indicate that mussels persisted at sites already occupied but that they have not colonized sites where they had not occurred previously. Our findings highlight the importance of modeling approaches that incorporate imperfect detection in estimating species occurrence and revealing temporal trends to inform conservation planning.</p> application/pdf 10.31931/fmbc.v20i1.2017.13-19 en Freshwater Mollusk Conservation Society Declining occurrence and low colonization probability in freshwater mussel assemblages: A dynamic occurrence modeling approach article