<p>Effective wetland management requires reliable indicator species and monitoring methods that minimize ecosystem disturbance. This study presents a non-invasive UAV-based approach for monitoring the distribution and habitat associations of <i>Nannophya pygmaea</i>, a wetland indicator dragonfly, in a protected floating-mat wetland in Japan. High-resolution aerial imagery enabled direct detection of individuals across the entire wetland, while multispectral and thermal data were used to characterize vegetation structure and surface temperature. Across eight surveys conducted during a single emergence season, 558 adult individuals were identified, allowing quantification of within-season dynamics and sex-specific spatial distribution. Logistic regression analyses revealed clear associations between occurrence probability and UAV-derived environmental variables, with surface temperature exerting the strongest influence. Vegetation structure and thermal conditions jointly explained fine-scale habitat differentiation between males and females. The approach allows repeated monitoring without entering fragile wetland habitats and provides spatially explicit information relevant to habitat condition assessment. Logistic regression models were used to quantify relationships between occurrence and environmental variables. However, the absence of ground-truth validation and the potential for re-sighting of individuals should be considered when interpreting the results. Logistic regression models were used to quantify relationships between occurrence and environmental variables. However, the absence of ground-truth validation and the potential for re-sighting of individuals should be considered when interpreting the results. These findings demonstrate that UAV-based monitoring of indicator species can support adaptive wetland management by linking species responses to microhabitat structure and environmental change.</p>