Terrestrial ecosystems store about one third of anthropogenic CO₂ emissions through differences between their photosynthetic and respiratory fluxes. Approximately half of ecosystem respiration comes from plants, and of that approximately half is leaf respiration. Despite its importance for determining global carbon fluxes and pools, leaf respiration remains poorly constrained in land surface models (LSMs). Next generation LSMs increasingly incorporate Vegetation Demography Models (VDMs) with multi-layered canopies which require an improved understanding of how leaf respiration varies vertically through the canopy. Most LSMs assume a constant ratio between leaf maintenance respiration (R_dark) and the maximum capacity for carboxylation by the enzyme Rubisco (V_cmax) from the top of the canopy to the understory. However, recent observational evidence suggests the ratio of R_dark to V_cmax decreases by almost 50% through the canopy. We tested the sensitivity of the VDM ELM-FATES (Energy Exascale Earth System Model Land Model - Functionally Assembled Terrestrial Ecosystem Simulator) to canopy gradients of R_dark. In global simulations, we find that steeper canopy gradients of R_dark lead to an increase in understory survival and an increase in leaf biomass driven by an optimality-based allocation scheme. As a result, leaf area index (LAI) increased by as much as 43% in tropical regions compared with the default parameterization, leading to improvements in model comparisons with remotely sensed benchmarks. Across the ensemble, vegetation carbon varied from 308 Pg C to 448 Pg C, with the steepest canopy gradient of R_dark leading to total global vegetation carbon 38% higher than the default parameterization. In simulations at a single tropical forest site with competing plant functional types, we find that the canopy gradient of R_dark alters competitive dynamics and coexistence, with steeper gradients of R_dark favoring shade tolerant PFTs. Our results show the importance of canopy gradients in leaf traits and fluxes for determining whole plant carbon budget and emergent ecosystem properties such as competitive dynamics, LAI, and vegetation carbon. The high model sensitivity to canopy gradients in R_dark highlights the need for more observations of how leaf traits and fluxes vary along light micro-environments to inform critical dynamics in LSMs.