The response of forests to climate change depends in part on whether the photosynthetic benefit from increased atmospheric CO₂(∆C_a= future minus historic CO₂) compensates for increased physiological stresses from higher temperature (∆T). We predicted the outcome of these competing responses by using optimization theory and a mechanistic model of tree water transport and photosynthesis. We simulated current and future productivity, stress, and mortality in mature monospecific stands with soil, species, and climate sampled from 20 continental US locations. We modeled stands with and without acclimation to ∆C_aand ∆T, where acclimated forests adjusted leaf area, photosynthetic capacity, and stand density to maximize productivity while avoiding stress. Without acclimation, the ∆C_a-driven boost in net primary productivity (NPP) was compromised by ∆T-driven stress and mortality associated with vascular failure. With acclimation, the ∆C_a-driven boost in NPP and stand biomass (C storage) was accentuated for cooler futures but negated for warmer futures by a ∆T-driven reduction in NPP and biomass. Thus, hotter futures reduced forest biomass through either mortality or acclimation. Forest outcomes depended on whether projected climatic ∆C_a/∆T ratios were above or below physiological thresholds that neutralized the negative impacts of warming. Critically, if forests do not acclimate, the ∆C_a/∆T must be aboveca. 89 ppm⋅°C⁻¹to avoid chronic stress, a threshold met by 55% of climate projections. If forests do acclimate, the ∆C_a/∆T must rise aboveca. 67 ppm⋅°C⁻¹for NPP and biomass to increase, a lower threshold met by 71% of projections.

Summary: Optimal stomatal control models have shown great potential in predicting stomatal behavior and improving carbon cycle modeling. Basic stomatal optimality theory posits that stomatal regulation maximizes the carbon gain relative to a penalty of stomatal opening. All models take a similar approach to calculate instantaneous carbon gain from stomatal opening (the gain function). Where the models diverge is in how they calculate the corresponding penalty (the penalty function). In this review, we compare and evaluate 10 different optimization models in how they quantify the penalty and how well they predict stomatal responses to the environment. We evaluate models in two ways. First, we compare their penalty functions against seven criteria that ensure a unique and qualitatively realistic solution. Second, we quantitatively test model against multiple leaf gas‐exchange datasets. The optimization models with better predictive skills have penalty functions that meet our seven criteria and use fitting parameters that are both few in number and physiology based. The most skilled models are those with a penalty function based on stress‐induced hydraulic failure. We conclude by proposing a new model that has a hydraulics‐based penalty function that meets all seven criteria and demonstrates a highly predictive skill against our test datasets.