Species traits influence their response to environmental conditions and the match between phenotypes and environment mediates spatial changes in species composition. These trait-environment linkages can be disrupted in human-modified landscapes. Human land-use creates habitat fragments where dispersal limitation or edge effects can exclude species that may otherwise suit a given macro-scale environment. Furthermore, stressful micro-environments in fragments may limit viable trait combinations resulting in stronger trait covariance compared to contiguous forest, especially in harsher macroenvironments. In a wet tropical forest landscape in the Western Ghats Biodiversity Hotspot of peninsular India, I compared fragments with adjacent contiguous forest for signatures of trait-mediated assembly of tree communities along macroenvironmental gradients. Using four key plant traits—seed size, specific leaf area (SLA), wood density, and maximum height—I evaluated changes in trait-mediated abundances and trait covariance across environmental gradients. Trait-mediated abundances primarily changed along the elevation gradient in contiguous forest, smaller-seeded, shorter, thinner-leaved species increased at higher elevations. In fragments, higher SLA species increased in more seasonal climate and decreased with higher precipitation, and larger seeds decreased at warmer sites. However, traits only weakly predicted abundances and only contiguous forests experienced significant compositional change via traits, driven by trait syndromes varying along a composite environmental gradient defined by elevation, water deficit, and soil C:N ratio. Covariance of seed size and maximum height along gradients of precipitation and temperature revealed divergent constraints on viable phenotypes in fragments and contiguous forest. Notably, local biotic conditions (functional diversity) had stronger effects than environment in explaining trait covariance. Overall, the results imply that trait syndromes and trait covariance, rather than single traits, determine the phenotypes best suited to different macroenvironmental conditions and should inform management or restoration goals in fragments.