Online Appendix III: NTI results
As explained in Appendix II, the Nearest Taxon Index (NTI) is sensitive to changes in tree topology, particularly to the resolution of clades at the intrafamilial level, when the phylogeny is large. With this dataset, therefore, the NTI results are less robust than the NRI results. I have included the NTI results here, but I caution against extensive interpretation. The NTI results are likely to change as better phylogenetic resolution becomes available.
Patterns of NTI during succession are much less clear than the patterns observed in NRI. With NRI, overdispersion was evident at multiple scales: the species found in a successional age category were a non-random, overdispersed subset of species in the regional pool, and species at a site were overdispersed both with respect to the regional species pool and to the pool of species found in sites of that age category (Fig. 1). With NTI, no consistent successional trends appear (Fig. S1). Most of the values are close to zero, indicating random phylogenetic structure as assessed by this metric. Positive NTI values, indicating clustering, were detected in intermediate-aged forests (21-30 yrs), while negative values indicative of overdispersion were found both in younger forests (16-20 yrs) and old-growth forests (Fig. S1).
Fig. S1. The nearest taxon index (NTI) at three different scales: 1) the NTI of the species pool in each age category compared to the regional supertree (white bars); 2) the average NTI of sites within an age category compared to the regional supertree (pale gray bars); 3) the average NTI for sites within an age category compared to a supertree constructed from species present in that age category (dark bars). Error bars represent ±1 standard error.
In comparing NTI at different size classes, also, the pattern of community phylogenetic structure is not as pronounced as with NRI. The pattern of decreasing phylogenetic overdispersion in larger size classes is similar to the NRI results, though with NTI the pattern breaks down in forests of 31-44 yrs. With NTI, trees in the largest size class (≥10 cm dbh) show evidence of clustering at all forest ages, whereas with NRI the only positive values were for large trees in young (<21 yr) forests (Fig. 2).
Fig. S2. The average NTI for stems within a diameter class in each age category, based on comparisons to the regional supertree. Error bars show ±1 standard error.