Abstract
High seedling mortality is a key bottleneck in tropical forest restoration, but the mechanisms governing the survival–growth trade-off remain unclear. We evaluated seedling performance in a high-diversity restoration planting in Goytacazes National Forest (Tableland Atlantic Forest), monitoring 771 individuals from 54 native species over three years (2012–2015). Using Stratified Cox Proportional Hazards models for survival and Linear Mixed Models for relative growth rates (RGR), we also quantified spatial variance (ICC). Wood Density (WD) was the strongest predictor of survival: high-WD species had lower mortality risk (HR = 0.82), independent of spatial heterogeneity. The study period included a flood (2013) and a drought (2015), and this context—together with the very low herbivory—suggests that extreme climatic events drove the survival advantage of high-WD species. Contrary to expectations from the Plant Economics Spectrum, high WD did not reduce vertical or radial growth. Furthermore, we found a non-significant negative trend ( p = 0.146) between potential maximum height (H max ) and early growth, indicating that initial performance was independent of the species’ eventual vertical niche. Our results suggest that high-WD species show higher survival without early-growth costs, highlighting their strategic value for restoration in high-stress environments.