Understanding seedling growth relationships through specific leaf area and leaf nitrogen concentration: generalisations across growth forms and growth irradiance.

Seedling relative growth rate (RGR) achieved under favourable growth conditions can be thought of as a useful bioassay of the potential ability of species to take advantage of favourable growth opportunities; that is, of a species' growth strategy. The consistency of relationships between RGR and its component attributes leaf nitrogen productivity (LNP), leaf N per area (LNCa), specific leaf area (SLA) and leaf mass ratio (LMR) was assessed across 12 datasets comprising three growth forms (grasses, herbaceous dicots and woody plants; 250 species in total). These relationships were characterised in terms of scaling slopes (regressions on log-log axes, the slopes giving the proportional relationship between the variables). Mathematically, the expected scaling slope between RGR and each component is 1.0, giving an appropriate null hypothesis to test against (whereas the widely used null hypothesis of zero correlation is in fact inappropriate for this situation). Deviations below 1:1 scaling slopes indicate negative covariance between the components. Consequently, the correlation structure between the components of RGR should also be investigated. Biologically, RGR should scale 1:1 with SLA at a given LNCa and somewhat more weakly with LNCa at a given SLA. SLA and LNCa should themselves scale with a slope of between 0 and -1, with the actual slope indicating the extent to which between-species variation in SLA dilutes leaf N on an area basis versus the ability of species to maintain LNCa at a given growth irradiance. On average, across the 12 datasets RGR scaled close-to-proportionally with SLA, and 1:1 with SLA at a given LNCa. RGR scaled with LNCa with null or negative slopes, since SLA and LNCa scaled negatively (with slopes generally shallower than -1); however, RGR scaled positively (but less than proportionally) with LNCa at a given SLA. For these key relationships there were no qualitatively different conclusions with respect to the growth form under consideration or the growth irradiance at which the seedlings were grown. RGR also scaled close-to-proportionally with LNP, while LNP and LNCa were negatively associated. These relationships involving LNP are difficult to interpret since it can be shown that they are, at least potentially, the result of the interactions between RGR, SLA and LNCa, as well as reflecting intrinsic differences in the efficiency of nitrogen use in the growth process.