## Effect of increasing N limitation on leaf-level variables.

#### figure

Figures are generally photos, graphs and static images that would be represented in traditional pdf publications.

Relationship between leaf-level variables and experimental N limitation, where maximum N limitation (i.e., N limitation = 100%) means zero nitrogen addition in the N limitation treatment. Line shows modelled response (linear, except for N_{L} per unit mass, panel d, that also includes a quadratic term) and the shaded area indicates the 95% confidence interval. Each point represents one experiment. The individual panels show (a) photosynthesis (studies = 50, experiments = 182), (b) leaf area (21, 80), (c) N_{L} per unit area (23, 120), (d) N_{L} per unit mass (17, 95), (e) chlorophyll content (32, 116), and (f) Rubisco content (13, 50). See Table 3 for more details on model fit. We modelled a quadratic relationship for N_{L} per unit mass, as this model provided a better and more sensible fit to the data and was supported statistically (P<0.01, Table 3). A quadratic relationship was also weakly statistically supported for other variables (photosynthesis, leaf area, N_{L} per unit area; Table 3) but showed non-sensible fits driven by some data points at low N limitation and a linear fit was therefore preferred.