Canopy photosynthesis models have predicted an optimal leaf area index (LAI; leaf area per unit surface area) and leaf nitrogen distribution at which whole-plant carbon gain per unit N is maximized. In this study we experimentally tested these models, using transgenic P SAG12 -IPT tobacco (SAG; Nicotiana tabacum L.) plants with delayed leaf senescence and therefore a greater LAI and more uniform N distribution than the wild type (WT). In a competition experiment, the increased density of surrounding WT plants caused a greater reduction in dry mass of mature SAG target plants than in that of WT target plants, indicating negative effects of delayed leaf senescence on performance at high canopy density. Vegetative SAG plants achieved a lower calculated daily carbon gain than competing WT plants because the former retained leaves with a negative carbon gain in the shaded, lower part of the canopy. Sensitivity analyses showed that the carbon gain of SAG plants would increase if these lower leaves were shed and the N reallocated from these