Realistic simulation of photosynthesis is needed for accurate prediction of the global carbon cycle. In addition to the well-known instantaneous response to changes in temperature, photosynthetic processes acclimate to long‐term environmental changes by adjusting the maximum photosynthetic capacities (<em>V</em><sub>cmax</sub> and <em>J</em><sub>max</sub>) and stomatal behaviour The theoretical basis for acclimation can be understood as an outcome of eco-evolutionary optimality (EEO). Here, we have implemented an EEO‐based scheme to represent the universal acclimation of photosynthesis to environmental conditions independently of plant functional types in the JULES land surface model. We compare this implementation with two standard configurations of JULES (GL7.0, UKESM1). We evaluate model performance using daily, monthly and annual site-based observations of gross primary production (GPP) observations at flux towers from the PLUMBER2 data set. The EEO-based scheme, PJULES, produces better predictions of GPP than either GL7 or UKESM1, both of which substantially underestimate GPP at the higher end of the observed range. Comparison at individual sites shows that PJULES produces more realistic simulations of seasonal and diurnal cycles of GPP. The improvement in the estimation of GPP does not degrade simulated land-surface energy fluxes: predictions of the latent heat flux are similar to those from the UKESM1 and slightly better than those from GL7.0. The good performance of PJULES shows that a parsimonious model can offer a competitive alternative to more complex parameterisations for simulating land–atmosphere carbon and water exchanges.