African ecosystems remain underrepresented in the plant trait datasets used to parameterise land surface models (LSMs), limiting our confidence in simulations of African landscapes. We assessed whether African plant trait data can improve parameterisation of plant functional types (PFTs) in the Joint UK Land Environment Simulator (JULES vn 7.1) model by updating two key photosynthetic traits: leaf mass per area (LMA) and leaf nitrogen content (Nmass). First, we evaluated African-derived trait estimates relative to default JULES parameters. We then analysed impact of mean annual temperature, mean annual precipitation and an aridity index on variability in these traits using generalized additive mixed models (GAMM), and tested the effect of updated trait parameter values on simulated gross primary productivity (GPP) against data from 16 African eddy covariance sites spanning different biomes. Comparison with JULES default parameters showed that African trait values differed for several PFTs, highlighting the value of trait-informed regional parameterization for representing African vegetation more realistically. GAMM analysis showed that LMA and Nmass varied nonlinearly across climate space, indicating that African leaf trait variation is shaped by interacting environmental gradients rather than single climate drivers. Incorporating the revised trait values into JULES generally produced modest improvements in simulated GPP. Improvements were expressed primarily through reduced bias and root mean square error, with little effect on temporal correlation. Performance gains were most evident at grassland sites, highlighting the limitations of representing diverse vegetation using single PFT-level trait values. This study demonstrate that regionally derived plant trait data can refine LSM parameterisation and can improve aspects of model performance in African ecosystems. They also show that regionally derived trait data provide a practical basis for evaluating and updating default model parameters, while highlighting the need for approaches that better represent within-PFT trait variability across environmental gradients and biome types.