Abstract
The phenological timing of leaf out in temperate forests is a critical transition point each year that alters the global climate system, which in turn, feeds back to plants, driving leaf out to occur nearly 3 d earlier per decade as temperatures rise. To improve predictions of leaf out timing, urban heat islands (UHIs) or densely developed areas that are hotter than surrounding undeveloped regions are often used to approximate warming via space-for-time substitutions (i.e., rural-to-urban temperature gradients). However, more than just environment changes along these gradients-urban regions are highly managed systems with limited-to-no within species diversity. We demonstrate here that recent observations that UHI gradients underpredict leaf out response to temperature when compared to temperature gradients through time is likely because both genetics and environment are changing across rural-to-urban gradients, whereas only environment is changing through time. We tested this hypothesis using genomic, phenological, and temperature data of northern red oak (Quercus rubra) over several years between an urban and rural site. Across our gradient, models that included just temperature predicted moderate advancement of leaf out. However, if we account for the genetic diversity of our trees in our model, leaf out phenology is predicted to advance significantly more in response to temperature. We demonstrate that this stronger relationship between phenological timing and climate is because urban trees have reduced genetic diversity as they are planted from limited stock by humans and, moreover, are most closely related to individuals at the rural site that leaf out later on average.