Karst regions are characterized by exposed rock, shallow soil, and poor water retention. A comparative analysis of tree traits in exposed versus covered karst habitats is essential for understanding species-specific ecological adaptation strategies in heterogeneous karst environments. Previous studies have primarily focused on contrasts between karst and non-karst ecosystems, with limited attention given to habitat heterogeneity within karst landforms. This study aims to elucidate the differences in economic, hydraulic, and anatomical traits of trees in exposed and covered karst habitats. We selected five dominant tree species from typical subtropical exposed karst (bedrock-exposed) and covered karst (soil-covered) areas: Cinnamomum burmannii, Liquidambar formosana, Osmanthus fragrans, Eriobotrya japonica, and Cinnamomum camphora. The measured traits included: (1) leaf traits, which are comprised of Huber value (Hv), leaf dry matter content (LDMC), specific leaf area (SLA), stomatal density (SD), guard cell length (GCL), stomatal area index (SAI), upper epidermis thickness (UET), palisade tissue thickness (PT), spongy tissue thickness (ST), lower epidermis thickness (LET), total epidermis thickness (ET), and leaf thickness (LT); (2) branch traits, which consist of wood density (WD), sapwood water content (SWC), vessel density (VD), hydraulically weighted vessel diameter (D_h), theoretical maximum branch hydraulic conductivity (K_th), and Carlquist vulnerability index (VI). Key findings include: (1) Variation in the 18 measured traits differed substantially. For leaf traits, coefficients of variation (CV) ranged from 16.9% (leaf dry matter content) to 97.4% (leaf area). For woody tissue traits, CVs ranged from 24.4% (sapwood density) to 90.9% (Carlquist fragility index). (2) Trait variability was generally higher in trees from covered karst habitats (CVs: 17.1% to 101.4%) compared to those from exposed karst habitats (CVs: 14.2% to 69.0%). Significant differences (P < 0.05) between habitat types were found for specific leaf area, guard cell length, stomatal area index, spongy tissue thickness, upper epidermis thickness, and leaf thickness. (3) Leaves exhibited divergent trait covariation strategies between the two habitats, whereas branches showed convergent patterns. Branch-leaf trait covariation was closely coupled in covered karst trees but significantly decoupled in exposed karst trees. (4) Principal component analysis (PCA) revealed distinct trait integration patterns. In exposed karst habitats, leaf traits were functionally integrated along an economic-hydraulic dimension. In contrast, leaf traits in covered karst habitats achieved multidimensional differentiation across economic, hydraulic, and anatomical structure dimensions. Branch traits in both habitats exhibited a consistent two-dimensional structure, with the hydraulic dimension prioritized over the anatomical structure dimension. This study demonstrates that trees in exposed karst habitats predominantly adopt a conservative survival strategy, while those in covered karst habitats favor a resource-acquisition strategy. This divergence is reflected in significant differentiation in branch and leaf traits between the two habitat types.