Intraspecific variation in leaf functional traits has gained widespread attention as a means to evaluate, predict, and manage plant responses to environmental conditions, however there are considerable uncertainties regarding the extent and drivers of intraspecific trait variation (ITV) in domesticated plants. In a coffee (Coffea arabica) agroforestry system, we quantified ITV in seven leaf traits [i.e. area (LA), mass per area (LMA), dry matter content (LDMC), thickness (Lth), nitrogen concentrations (LNC), maximum photosynthetic rate on area and mass bases (Asat, Amass, respectively)] across managed gradients of soil fertility and light availability. Leaf physiological traits (Asat, Amass), as well as LA, showed the greatest extent of variation within coffee, while morphological traits (LMA, LDMC, Lth) and leaf N were less variable. All traits differed significantly as a function of light and fertilization treatment, however light was more influential in driving ITV in coffee leaves. Low light availability resulted in greater ITV for physiological leaf traits (Asat and Amass), while high light constrained ITV in most morphological-(LA, LMA, LDMC), physiological-(Asat, Amass) and chemical-(LNC) traits. Fertilization treatments did not induce systematic shifts in the extent of ITV. In addition, shade management treatments explained 9.2% of the variation in multivariate trait syndromes, while nutrient management regimes explained only 2.9%. Our results indicate that highly heterogeneous aboveground resource environments such those created by agroforestry, results in greater ITV for key crop physiological parameters. Based on ecological theory, such patterns indicate that management systems promoting resource heterogeneity should promote higher rates of resource partitioning, and greater resource-use efficiency in agroecosystems.