Across-family analyses of ERR Slopes with respect to ecological, evolutionary, and biogeographic correlates
For across-family analyses using phylogenetic independent contrasts (PIC), we found that Malesian richness was not associated with ERR slopes (Table 1). However, clustering of richness, based on thelocalized richness ratio , was positively associated with ERR slopes (Table 1, Fig. 3a). Families with high localizedrichness ratio metrics and steeply positive ERR slopes included Dipterocarpaceae (ratio of 0.82 and ERR slope of 0.53) and Anacardiaceae (ratio of 0.84 and slope of 0.45). In contrast, families with low localized richness ratio metrics and weaker positive ERR slopes included Begoniaceae (ratio of 0.34 and slope of 0.24) and Gesneriaceae (ratio of 0.35 and slope of 0.14). Similarly, the elevation with peak in richness had a negative relationship with ERR slopes, with higher ERR slopes when richness peaked at low elevations as compared to high elevations (Table 1, Fig. 3b). For instance, the first quartile of families (15 total) with low elevation peaks in richness (< 200 m a.s.l.) had an ERR slope average of 0.48, whereas those with peaks within the last elevation peak quartile (avg. 1295 m a.s.l.) had an ERR slope average of 0.09. Familial elevational extent also had a negative relationship with ERR slopes (Table 1, Fig. 3c). Families restricted to a narrow section of the gradient, either from ecological specialization or geometric constraints (e.g. Burseraceae, familial extent of ~ 2000 m and ERR slope of 0.55), had steeper ERR slopes as compared to families with broad elevational occurrence (e.g. Asteraceae, familial extent of ~ 4500 m and ERR slope of -0.13). Average lower elevation limit and average higher elevation limit metrics both had negative relationships with ERR slopes (Table 1, Fig. 3d), with higher ERR slopes for families with occurrence boundaries constrained to lower elevations. Counter to our predictions, neither evolutionary age nor latitudinal extent had a direct relationship with ERR slopes (results insignificant). Consistent with our predictions, biogeographical affiliation strongly influenced ERR slopes. Families with greater westwardly skewed species occurrence (towards Sundaland) displayed a strongly positive relationship with ERR slopes, whereas a negative relationship emerged eastwardly (towards Sahul), with ERR slopes becoming shallower, neutral, or negative (Table 1, Fig. 4a-d). The spatially intermediate location, Wallacea, had mixed results when comparing metrics of endemism or overlap (Table 1).Cosmopolitanism did not directly influence ERR slopes (Table 1, Fig. 4d). The first pPCA component (pPC1, representing 43% of the variance) had a positive relationship with ERR slopes (d.f. = 47, F-statistic = 141.8, Adj. R2 = 0.75, P -value < 0.001, Fig. 5). The observed dichotomy of range-size trends for lowland Sundaland flora with concentrated centers of richness, with low pPC1 scores and strongly positive ERR slopes, compared to montane Sahul flora with distributed richness, with high pPC1 scores and weaker ERR slopes (Table 1; see Supporting information). The second component (pPC2, 19% of the variance) was non-significant.