Qian and Ricklefs reply

The estimated distribution of vegetation during the Last Glacial Maximum (18,000 yr bp) and at 6,000 yr bp, on the basis of fossil-pollen data1 and climate, shows that mesic vegetation types were shifted southwards during the Last Glacial Maximum relative to their modern distributions in both eastern Asia and eastern North America2,3. However, the temperate deciduous forest biome, which currently harbours many of the eastern Asian/eastern North American disjunct genera we discussed4, is poorly represented as a vegetation type in fossil-pollen deposits from the Last Glacial Maximum, casting doubt on the accuracy of the reconstruction by Harrison et al.

For example, the palaeovegetation of the Last Glacial Maximum for the vast area between 20° and 30° N and 105° and 120° E was reconstructed from only six pollen localities, including only one from the interior of eastern Asia, which currently harbours the greatest plant diversity in the region.

It is not possible to determine whether temperate forests coalesced or fragmented during the Last Glacial Maximum without more detailed information. Nonetheless, the reconstruction by Harrison et al. confirms that the currently isolated temperate forests of China, Japan, and probably the Korean peninsula, were connected during the Last Glacial Maximum.

Speciation and extinction are processes of populations, not biomes. Climate change causes restructuring of forest communities, probably including the dispersion of temperate deciduous forest taxa among other vegetation types5. Vegetation maps cannot provide a detailed picture of past fragmentation and coalescence of particular species populations. Furthermore, because most eastern Asian/eastern North American disjunctions pre-date the onset of major glacial climate cycles in the Northern Hemisphere6, diversification of disjunct lineages may have began under pre-Pleistocene climate conditions that were quite different from those in the present or in the Last Glacial Maximum. In these pre-Pleistocene conditions, eustatic sea-level changes may have been important in isolating and rejoining populations.

Reconstruction of vegetation history, combined with physiographical and climatic heterogeneity, conveys a general impression of the capacity of a region to promote or retard diversification through allopatric speciation7. However, the biome reconstruction by Harrison et al., including their interpretation of restricted deciduous forest vegetation during glacial maxima, does not contradict the idea that the more complex landforms and climates of eastern Asia provided greater opportunities for allopatric formation of new species compared with eastern North America.