Estimating occupancy and detection probability of the Amazonian manatee (Trichechus inunguis), in Central Amazon, Brazil

Abstract Understanding the distribution and abundance of threatened species is crucial to elaborate effective management plans for wild populations; however, elusive species prove difficult to detect. To support conservation strategies for the Vulnerable Amazonian manatee (Trichechus inunguis), the only freshwater sirenian, we analyzed presence/absence data with hierarchical models based on imperfect detection to assess T. inunguis occupancy in a Sustainable Development Reserve, Brazilian Central Amazon. In parallel, we compared the effectiveness of direct and indirect sampling methods to provide occupancy (ψ) and detection (p) estimates. Combining both sampling methods’ presence datasets provided higher accuracy estimates. The Amazonian manatee’s detection probability had never been estimated before: surprisingly, it was high (p = 0.50, SD = 0.05) and positively related with macrophyte coverage. Results suggest that the studied communities resident impact is not affecting the manatee occupancy, with greatest probabilities closer to human settlements. The final occupancy estimate obtained (ψ = 0.85, SD = 0.12) can be a baseline to Amazonian manatee long-term monitoring studies, and provide support for decision makers and local communities to establish effective protection zones for the species. Our approach highlights the potential of hierarchical models to understand the distribution not only of T. inunguis in different habitats, but also of other threatened Amazonian aquatic mammals.