SEARCH

SEARCH BY CITATION

Appendix S1. Supplementary analyses.

Just as in classical seedling growth analysis (equation 1, main text), the relative growth rate of a shoot (RGRshoot) can be broken down into several components:

RGRshoot = NARshoot • LA/SM = NARshoot • LM/SM • SLA    (A1)

where LA/SM refers to total leaf area per shoot dry mass and LM/SM the total leaf dry mass per shoot dry mass. Although we were primarily interested in the relationships between LA/SM and each of leaf size and wood density (WD), by considering those between these traits and each of LM/SM and SLA we gain a deeper understanding of how the LA/SM-related trends come about.

Trends in LA/SM and SLA with leaf size

LA/SM is the product of LM/SM and SLA; consequently, leaf size-LA/SM relationships reflect the combined effects of relationships between leaf size and each of LM/SM and SLA. Considered either at the standard stem cross-section or at the standard distance, LM/SM increased strongly as leaf size increased, both across all species or at individual sites, whereas leaf size and SLA were either unrelated or negatively associated (Fig. S1). That is, the leaf size-SLA relationships largely counteracted the positive LM/SM-leaf size relationships, resulting in LA/SM and leaf size being positively correlated at only a subset of sites, all forests in fact (GHP, Ash and TRF).

Trends in LA/SM and SLA with wood density

Considered either at the standard stem cross-section or at the standard distance, LM/SM showed mixed within-site relationships with WD (Fig. S2). However, in both cases the individual slopes did not differ significantly (both p > 0.1), with the common fitted slopes clearly negative (both p < 0.002). Individual WD-SLA relationships were also mixed, considered at either the standard distance or stem cross-sectional area (Fig. S2). In both cases the individual slopes were deemed heterogeneous (both p < 0.005), although the traits were negatively associated when a single regression line was fitted across all species, ignoring site structure (at 250 mm, r2 = 0.07, p = 0.010; at 10 mm2, r2 = 0.28, p < 0.001).

Fig S1. Relationships between log10 leaf size and each of total leaf mass per shoot mass (LM/SM) and specific leaf area (SLA). Regression lines are shown for significant within-site relationships only. (A, B) LM/SM10mm2 increased with increasing leaf size both across all species (r2 = 0.63, n = 116) and within each of the seven sites (r2 values ranging between 0.22 and 0.67; for six of the seven sites P < 0.030; at the seventh, P = 0.087). SLA and leaf size were negatively correlated at GLP and RHM (r2 = 0.30, 0.58 respectively, both p < 0.01) and unrelated at each of the other sites (all P ≥ 0.30). Symbols: Ash, black circles; TRF, grey circles; GHP, light green triangles; GLP, blue triangles; Myall, pink squares; RHM, dark green squares; RHW, red diamonds. (C, D) LM/SM10mm2 increased with increasing leaf size both across all species (r2 = 0.46, n = 96) and within each of the four sites (r2 values ranging between 0.24 and 0.65, all P < 0.025). SLA and leaf size were negatively correlated at GLP (r2 = 0.31, P = 0.017) and unrelated at each of the other sites (all P ≥ 0.16). Symbols: Ash, red circles; TRF, blue circles; GLP, grey triangles; Myall, black triangles.

Fig S2. Relationships between wood density (WD) and each of LM/SM and SLA. Regression lines are shown for significant within-site relationships only. Symbols as for Fig. S1. (A, B). WD250mm and LM/SM250mm were negatively correlated at GLP and Myall (r2 = 0.29, 0.50; both P ≤ 0.02) but unrelated at the other two sites (both P > 0.120). The common fitted slope was negative (ANCOVA; P < 0.001, model r2 = 0.23). WD250mm and SLA were negatively correlated at Ash (r2 = 0.46, P = 0.001) and across all species (r2 = 0.07; P = 0.010), unrelated at TRF (r2 = 0.05, P = 0.123), positively correlated at GLP (r2 = 0.29, P = 0.021) and showed a tendency towards a positive relationship at Myall (r2 = 0.20, P = 0.148). (C, D) Negative relationships were observed between WD10mm2 and LM/SM10mm2t GLP and GHP (r2 = 0.33 to 0.45, both P < 0.022); at the other two sites the traits were unrelated. The common fitted slope was negative (ANCOVA; P < 0.001, model r2 = 0.40). WD10mm2 and SLA were negatively correlated at GHP (r2 = 0.29, P = 0.031) and across all species (r2 = 0.28, P < 0.001), positively correlated at GLP (r2 = 0.38, P =0.004), and unrelated elsewhere (both P > 0.25).

FilenameFormatSizeDescription
PPL_699_sm_FigS1.pdf45KSupporting info item
PPL_699_sm_FigS2.pdf42KSupporting info item

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.