The modification of natural habitats requires behavioural plasticity, which may be challenging for ‘specialist’ species. Quantifying habitat requirements and behavioural responses of specialists to landscape transformation is thus a priority for baseline data to inform conservation practices. Using camera-trap surveys of the forest-dependent Lemon Dove Aplopelia larvata in conjunction with detailed microhabitat-scale covariates, we assessed habitat use during two periods in the year: autumn—winter and spring—summer (which largely encompassed peak breeding). We used occupancy modelling of forest-structural covariates to produce models of the probability of occupancy and detection of Lemon Doves in patches of the Indian Ocean Coastal Belt Forest of South Africa. The average occupancy and detection probability as indicated by the top-performing models was 0.39 ± 0.08 and 0.26 ± 0.05, respectively, during autumn—winter, and 0.37 ± 0.08 and 0.25 ± 0.04, respectively, during spring—summer. Although occupancy and probability of detection remained relatively constant between seasons, there was seasonal variation in the influence of individual covariates for both measures. The overall trend of positive influences on Lemon Dove occupancy was that of complex and diverse habitat structures and high plant species richness. The specific covariates that influenced occupancy positively during spring—summer may reflect the ecological requirements for nestling provisioning for both dietary needs and an avoidance of potential disturbance. Thus, while Lemon Doves may be less habitatspecific during autumn—winter, conservation management plans for safeguarding the breeding success of the species are advised to ensure adequate protection of large forest patches with complex and diverse interior structures and minimal disturbance.