ABSTRACT Climate change is altering forest structure and function globally, directly affecting trees and seedlings through warming and drought, and indirectly through increased mortality and changing understory conditions. However, forest community responses also depend on local characteristics, including species‐specific sensitivities and site conditions like soil moisture. Because seedling recruitment is a key demographic bottleneck, evaluating recruitment across local environmental gradients is essential for predicting shifts in forest composition. We hypothesized that warmer, drier conditions would limit recruitment, especially for species associated with mesic sites, whereas xeric species may be pre‐adapted to drier conditions. Alternatively, xeric species‐site complexes might be more susceptible because they are already under drought stress, whereas mesic sites could buffer drought impacts. We further predicted that forest overstories would buffer understory conditions and ameliorate climate effects on seedlings, particularly in mesic sites with higher canopy cover. We tested these hypotheses by evaluating soil moisture and temperature effects on seedling abundances of 10 tree species, using 26 years of annual seedling censuses from 12 sites spanning soil resource gradients in temperate hardwood forests of Lower Michigan, USA. Recruitment of mesic species‐site complexes was generally more sensitive to warmer, drier conditions than that of xeric complexes, with future climates likely to favor recruitment of Acer rubrum , Ostrya virginiana , Prunus serotina , Quercus rubra , and Q. alba . Overstory buffering effects were strong: seedling abundances were negatively related to canopy openness for mesic and widespread species, but not for xeric complexes. Our results demonstrate that canopy cover can buffer mesic species‐site complexes from drought, and that fine‐scale environmental variation strongly mediates seedling responses to climate. These findings reveal strong interspecific variation in recruitment sensitivity, emphasizing the importance of local conditions and early life stages in determining forest responses to climate change.
McNichol et al. (Thu,) studied this question.