Arctic vegetation is characterized by high spatial variability in plant functional type (PFT) composition and gross primary productivity (P). Despite this variability, the two main drivers of P in sub-Arctic tundra are leaf area index (L) and total foliar nitrogen (N ). L and N have been shown to be tightly coupled across PFTs in sub-Arctic tundra vegetation, which simplifies up-scaling by allowing quantification of the main drivers of P from remotely sensed L . Our objective was to test the L -N relationship across multiple Arctic latitudes and to assess Las a predictor of P for the pan-Arctic. Including PFT-specific parameters in models of L -N coupling provided only incremental improvements in model fit, but significant improvements were gained from including site-specific parameters. The degree of curvature in the L-N relationship, controlled by a fitted canopy nitrogen extinction co-efficient, was negatively related to average levels of diffuse radiation at a site. This is consistent with theoretical predictions of more uniform vertical canopy N distributions under diffuse light conditions. Higher latitude sites had higher average leaf N content by mass (N ), and we show for the first time that L-N coupling is achieved across latitudes via canopy-scale trade-offs between N and leaf mass per unit leaf area (L ). Site-specific parameters provided small but significant improvements in models of P based on L and moss cover. Our results suggest that differences in L -N coupling between sites could be used to improve pan-Arctic models of P and we provide unique evidence that prevailing radiation conditions can significantly affect N allocation over regional scales.