Beyond canopy cover: UAV-LiDAR reveals three-dimensional structural divergence and parsimonious candidate diagnostic traits for mangrove monitoring

Canopy cover is widely used in mangrove monitoring, yet its ability to represent three-dimensional (3D) community structure has rarely been quantified. We used UAV-LiDAR to survey 38 10 × 10 m plots across embanked (BW), tidal-creek (TC), and tidal-flat (TF) habitats on Jinniu Island, China, with TC and TF further pooled as tidally connected habitats (TCH). Seventeen structural traits spanning five dimensions were derived to compare the habitat-response strength of 2D canopy coverage and 3D structure and to identify a parsimonious set of candidate diagnostic traits. Habitat differentiation was concentrated in 3D structure: none of the three 2D coverage traits differed significantly among habitats after false discovery rate (FDR) correction, whereas 13 of 14 3D traits did. TC and TF showed limited overall differentiation, while BW exhibited directionally consistent departures from TC, TF, and TCH across most 3D traits. For the BW–TCH comparison, the median absolute Cliff’s δ was 0.017 for 2D traits and 0.731 for 3D traits, yielding a 2D–3D effect-size contrast of Δ E = 0.714 (95% CI: 0.258–0.828). Multi-evidence screening identified maximum vegetation height (Hmax) or the 95th percentile of vegetation-point heights (H P95 ) and the interquartile range of vegetation-point heights (IQR) as core candidate diagnostic traits representing canopy development and height heterogeneity, with the proportion of vegetation points between 0.5 m and half of the plot-specific HP95 (rLPR 50 ) retained as a complementary trait for vertical configuration. These results show that similar canopy cover can conceal marked differences in vertical community structure. Integrating a small number of ecologically interpretable 3D traits with coverage-based monitoring provides a more informative and parsimonious basis for mangrove structural assessment.