Modelling occupancy and abundance of Eastern Sand Darter ( Ammocrypta pellucida) while accounting for imperfect detection.

摘要

Accurate representations of occupancy, abundance, and habitat associations are particularly important for the conservation and management of imperilled species. When species are detected imperfectly, assessments of these important attributes are likely to be biased. Occupancy and abundance models that account for imperfect detection were developed to examine factors that affect the distribution and abundance of the Threatened Eastern Sand Darter (Ammocrypta pellucida), a small benthic fish with a discontinuous and declining range in eastern North America. Models were developed based on repeated field surveys conducted in the Grand and Thames rivers, Ontario, Canada. Candidate models included physical habitat, biotic, and anthropogenic covariates that were hypothesized to affect Eastern Sand Darter distribution and abundance. Occupancy models based on imperfect detection and those based on naïve detections had similar outcomes in the Grand River where detection probability was high, but these two modelling approaches produced different results for the Thames River where detection probability was low. Despite the different detection probabilities between the two Ontario study streams, factors affecting site occupancy were similar. The most important factor affecting site occupancy was the proportion of sand and fine gravel, but water clarity and predator and competitor indices had support in additive models. Life-stage models suggest that adult and young-of-the-year Eastern Sand Darter occur in similar habitats although young of the year appear to be more tolerant of silt substrates. At broader scales, Eastern Sand Darter occupancy was more closely related to substrate at the site level than factors measured at the reach and valley segment levels, lending support to the substrate specificity of this species. Multi-state and zero-inflated negative binomial models that accounted for imperfect detection showed that factors affecting site abundance differed between the Grand and Thames rivers and were different from factors affecting occupancy. Factors important in determining occupancy were the most important covariates in abundance models based on naïve counts that ignore imperfect detection. Model predictions suggest that formerly occupied watersheds in Ontario (Ausable River, Big Otter Creek) have the potential to support Eastern Sand Darter and should be considered for repatriation efforts.