Phylogenetic beta diversity reveals historical effects in the assemblage of the tree floras of the Ryukyu Archipelago

Authors


Abstract

Quantifying the roles of historical versus contemporary constraints in determining species diversity is a central issue in island biogeography, and the phylogenetic beta diversity between islands is an essential measure specifying the influence of historical barriers on insular assemblages. In this study, using phylogenetic information for 513 tree species on 26 islands in the subtropical Ryukyu Archipelago, phylogenetic beta diversity between islands was calculated, and effects of historical factors (gaps as surrogate measures of historical barriers) and current ones (distance, area and elevation) on the phylogenetic structure of tree assemblages were examined. The pattern of phylogenetic beta diversity demonstrated that the Tokara Gap and geographical distance were consistently important for characterizing tree assemblages in the Ryukyus relative to other historical and current factors, which suggests that the Tokara Gap and distance-limited dispersal from the two adjacent source islands have left a deep imprint on the phylogenetic structure of the current tree flora of the islands.

In a recent study in the Journal of Biogeography, Nakamura et al. (2009) reported that historical barrier effects of two deep-sea channels, together with current geographical factors such as geographical distance and area, gave rise to floristic dissimilarity between islands in the Ryukyu Archipelago, an area that has had a unique geographical history. The Tokara and Kerama gaps divided the Ryukyu Archipelago into the northern, central and southern Ryukyus during the Pliocene (Nakamura et al., 2009). These deep-sea barriers have retained the disconnection between regions in the Ryukyu Archipelago, although land-bridge connections between Taiwan and the southern Ryukyus and between Kyushu and the northern Ryukyus were present intermittently because of sea-level changes throughout the Pleistocene (Ota, 1998). Therefore, the Tokara and Kerama gaps have historically played an important role in the assembly of the heterogeneous contemporary flora and fauna in the Ryukyus (Chiang & Schaal, 2006). Nakamura et al. (2009) utilized this biogeographical setting and tested geohistorical and current influences on floristic differentiation between islands in the Ryukyu Archipelago.

Quantifying the roles of historical versus contemporary constraints in structuring species diversity is one of the central issues in island biogeography, and historical constraints on biota are often examined based on the influences of palaeogeographical factors on differences in species composition between islands (Price, 2004; Hausdorf & Hennig, 2005). However, such an approach is indirect (or surrogate) when quantifying the historical effects. Lasting effects of historical biogeographical events may best be examined using phylogeny-based analyses (Crisci et al., 2003). Because species traits such as dispersal ability and environmental tolerance are phylogenetically determined (i.e. niche conservatism; Pearman et al., 2008), the effects of historical constraints are expected to leave a deep trace in the phylogenetic structures of assemblages (Wiens & Donoghue, 2004; Emerson & Gillespie, 2008; Graham & Fine, 2008). If geohistorical processes have had significant influences on insular assemblages in the Ryukyu Archipelago, detectable effects of the Tokara and Kerama gaps should be observed in phylogenetic beta diversity of the flora between islands. Unfortunately, Nakamura et al. (2009) overlooked the phylogeographical considerations in their paper.

The present study therefore focused on phylogenetic beta diversity between islands, and examined the effects of historical factors (gaps) and current factors (distance, area and elevation) on the phylogenetic structure of tree assemblages in the Ryukyus. The 513 tree species were extracted from the presence–absence matrix of 1815 seed plants on 26 islands in the Ryukyu Archipelago (Appendix S1 in Nakamura et al., 2009), and the phylogenetic tree was constructed from an APG3-derived megatree (reference tree: R20100318.new) using Phylomatic (Webb & Donoghue, 2005), choosing the maximally resolved seed plant tree option. Although the constructed tree was resolved to genus level for most families, genera that could not be resolved were placed as basal polytomies within their families. Branch lengths in the phylogenetic tree were adjusted using the Bladj application in Phylocom (Webb et al., 2008), whereby the nodes having a fossil calibration were fixed, and other branch lengths were set by placing them evenly between the dated nodes (Wikström et al., 2001). The phylogenetic beta diversity between islands was calculated with UniFrac (Lozupone & Knight, 2005; Faith et al., 2009), which measures the phylogenetic distance between two assemblages as a unique (unshared) fraction of the branch length, i.e. branches found either in one island or the other, but not in both. The pairwise phylogenetic beta diversity between islands was compiled, together with differences in geographical factors between islands: the matrices of the Tokara Gap (TGapM) and Kerama Gap (KGapM), pairwise differences in area (km2, AreaM) and maximum elevation (m, EleM) between islands, and pairwise geographical distance (km, DistM) between islands. The matrices of the Tokara and Kerama gaps had a dummy variable, in which a value of 0 was given to matrix elements of two islands that neither crossed over the Tokara Gap nor the Kerama Gap, and a value of 1 was given for matrix elements of two islands traversing the Tokara Gap or the Kerama Gap. The matrices of pairwise differences in area and maximum elevation were provided as the absolute values of differences in area and maximum elevation between two islands.

The effects of the Tokara Gap (TGapM) and Kerama Gap (KGapM) and geographical distance (DistM), difference in area (AreaM) and difference in maximum elevation (EleM) between islands on phylogenetic beta diversity between islands were estimated using multiple regression models on distance matrices (MRM), an extension of the Mantel test, using permutation tests of significance for regression parameter estimates and R2 (Legendre et al., 1994). The MRM is a multiple linear regression of a response matrix on two or more explanatory matrices, whereby the parameter of an explanatory variable is estimated by adjusting for the parameter of the others (Lichstein, 2007; Goslee, 2010). In the multiple regression analysis, current climatic factors were not included to avoid multicollinearity between variables; the distance matrices of temperature in the coldest month and annual precipitation were significantly correlated with those of geographical distance and maximum elevation distance, respectively (Nakamura et al., 2009). Three models were examined following Nakamura et al. (2009). The first model included the two historical explanatory matrices: the Tokara and Kerama gaps. The second model was examined with three explanatory matrices: the Tokara and Kerama gaps and geographical distance. The final model included all five explanatory variables. The analyses using the MRM were conducted with 10,000 permutations. All analyses were performed in the R environment for statistical computing (R Development Core Team, 2010), with the ecodist (Goslee & Urban, 2007) and picante (Kembel et al., 2010) packages.

The effect of the Tokara Gap on the phylogenetic beta diversity of tree assemblages was consistently significant in all models of two, three and five variables (Table 1). The effect of geographical distance on the phylogenetic beta diversity was also important in all models that included geographical distance as the explanatory variable, despite the marginal significance. In contrast, the influence of the Kerama Gap on phylogenetic beta diversity was not significant in any model, although Nakamura et al. (2009) found that the effect of the Kerama Gap on floristic beta diversity was significant in their first model, which included two variables, and that the effect of the Kerama Gap was parallel to that of geographical distance in their three- and five-variable models. Therefore, in the analyses of Nakamura et al. (2009), the correlation between historical and contemporary geographical factors may raise a sensitive issue in considering historical barrier effects on insular assemblages: namely, the inter-island distance may result in floristic beta diversity through dispersal limitation (Hubbell, 2001), irrespective of the existence of gaps. The analysis of phylogenetic beta diversity successfully disentangled the confounding effects of the Kerama Gap and geographical distance. In this study, the pattern of phylogenetic beta diversity showed that the Tokara Gap and geographical distance were crucially important for characterizing tree assemblages in the Ryukyus, relative to other historical and current factors.

Table 1.   Parameter estimates and P-values from multiple regression models on distance matrices of phylogenetic beta diversity of tree assemblages between islands, including the influence of gaps, distance, differences in area and differences in elevation between the islands.
ModelExplanatory variablesRegression coefficientsP-valueR2P (R2)
  1. Phylogenetic beta diversity was calculated using UniFrac. TGapM and KGapM, dummy variables for the matrices of the Tokara and Kerama gaps, respectively; DistM, matrix of distance between islands (ln km); AreaM, matrix of differences in area between islands (ln km2); EleM, matrix of differences in maximum elevation between islands (ln m).

Two variablesIntercept0.4130.8000.2840.001
TGapM0.1490.001  
KGapM0.0150.585  
Three variablesIntercept0.3970.9650.2990.001
TGapM0.1140.001  
KGapM−0.0630.200  
DistM0.0000.062  
Five variablesIntercept0.3750.9990.3170.001
TGapM0.1120.001  
KGapM−0.0570.270  
DistM0.0000.076  
AreaM0.0000.506  
EleM0.0000.136  

Combining the findings of Nakamura et al. (2009) and this study, a substantial discontinuity in the flora exists between the northern Ryukyus and central/southern Ryukyus. During the Neocene period, the Ryukyu Archipelago was connected to the surrounding source islands of Kyushu and Taiwan, but in the middle to late Pliocene, the Tokara Gap isolated the central and southern Ryukyus from Kyushu and the northern Ryukyus; subsequently, the Kerama Gap separated the central Ryukyus from Taiwan and the southern Ryukyus during glaciations. The significant increase in phylogenetic beta diversity with the Tokara Gap and geographical distance suggests a consequence of isolation and rejoining of tree assemblages through the idiosyncratic geohistory. The insignificant role of the Kerama Gap may be attributable to its not having existed as long as the Tokara Gap. The phylogenetic difference between floras in the northern and central/southern Ryukyus demonstrates that a deep imprint exists in the phylogenetic structure of the current flora on the islands caused by the Tokara Gap and separate distance-limited dispersal from the two adjacent source islands of Kyushu and Taiwan. The pattern of phylogenetic beta diversity plays a principal role in identifying historical and contemporary constraints for the assembly of species diversity on continental islands (Graham & Fine, 2008), where partitioning the centre of origin and its descendant areas along the gradient of island ontogeny is more difficult than on oceanic islands (Sanmartín et al., 2008; Whittaker et al., 2008).

Acknowledgements

We thank Matt Lavin and Fredrik Ronquist for their thoughtful comments and suggestions. Financial support was provided by the Japan Society for the Promotion of Science (nos. 21310025 and 22657008).

Editor: Jorge Crisci

Ancillary