This is an open access article under the terms of the Creat ive Commo ns Attri butio n-NonCo mmerc ial-NoDerivs License, which permits use and distribution in any medium, provided the original work is properly cited, the use is non-commercial and no modifications or adaptations are made.

Phenotypic distribution within species can vary widely across environmental gradients but forecasts of species' responses to environmental change often assume species respond homogenously across their ranges. We compared predictions from species and phenotype distribution models under future climate scenarios for Andropogon gerardii, a widely distributed, dominant grass found throughout the central United States. Phenotype data on aboveground biomass, height, leaf width, and chlorophyll content were obtained from 33 populations spanning a ~1000 km gradient that encompassed the majority of the species' environmental range. Species and phenotype distribution models were trained using current climate conditions and projected to future climate scenarios. We used permutation procedures to infer the most important variable for each model. The species-level response to climate was most sensitive to maximum temperature of the hottest month, but phenotypic variables were most sensitive to mean annual precipitation. The phenotype distribution models predict that A. gerardii could be largely functionally eliminated from where this species currently dominates, with biomass and height declining by up to ~60% and leaf width by ~20%. By the 2070s, the core area of highest suitability for A. gerardii is projected to shift up to ~700 km northeastward. Further, short-statured phenotypes found in the present-day short grass prairies on the western periphery of the species' range will become favored in the current core ~800 km eastward of their current location. Combined, species and phenotype models predict this currently dominant prairie grass will decline in prevalence and stature. Thus, sourcing plant material for grassland restoration and forage should consider changes in the phenotype that will be favored under future climate conditions. Phenotype distribution models account for the role of intraspecific variation in determining responses to anticipated climate change and thereby complement predictions from species distributions models in guiding climate adaptation strategies.

In the study of geographic range boundary development, the focus has been on leading rather than on trailing edge dynamics. This is an important caveat as trailing edge dynamics will be critical for an understanding population level persistence. Our study begins to fill this knowledge gap and extends the conceptual framework of the field by focusing on trans-generational environmental effects. We found that while these effects may overcome some constraints on stress tolerance evolution and range expansion, other constraints may be created to limit range.

Offspring phenotypes may be altered by environments that their parents lived in. These environmentally-induced trans-generational effects may be mediated by epigenetic mechanisms such as DNA methylation. Little is known about the role of such epigenetic effects in evolution; however, it is expected to facilitate evolution. To expand geographic range, it is thought that most species would have to adapt via evolution by natural selection to stressful environments beyond range boundaries. Contrary to expectations, we show that DNA methylation in an upland mustard species may underlie a drought-induced trans-generational tradeoff that may constrain the process of adaptation to stressful environments at lower elevations.