Aim We investigated the historical biogeography and diversification of Gentiana L. (Gentianaceae). Our study depicts the origin and dispersal routes of this alpine genus, and the role of the uplift of the Qinghai–Tibet Plateau (QTP) and past climate changes as triggers for its diversification. Location Tibeto‐Himalayan region and world‐wide mountain habitats. Methods Our sampling represents more than 50% of the extant Gentiana species, including all sections across their entire geographical ranges. We investigated the evolutionary history of Gentiana using phylogenetic reconstructions (maximum likelihood and Bayesian inference) of ITS, atpB–rbcL and trnL–trnF sequences, as well as molecular dating with beast. We tested two approaches of ancestral area reconstructions (DEC, DIVA) in BioGeoBEARS and investigated diversification rates using BAMM. Results The common ancestor of Gentiana and subtribe Gentianinae lived in the QTP region at around 34 (25–45) million years ago (Ma), and 40 (29–52) Ma respectively. From the surroundings of the QTP, Gentiana lineages dispersed to eastern China, Taiwan, Europe, North and South America, Australia and New Guinea, from mid‐Miocene onward (c. 15 Ma–present), with only one older dispersal event to Europe (c. 37–21 Ma). Diversification rates gradually increased over time, and two switches of diversification rates were identified in Gentianinae (c. 7 Ma, simultaneously in the Pneumonanthe/Cruciata lineage and in Tripterospermum). Main conclusions Gentiana existed in the QTP region throughout most of its uplift history following the India‐Asia collision. This region acted as the primary source area for dispersals to many areas of the world. Because steady increase in diversification rates coincides with the extension of the QTP, we argue that the museum theory rather than the explosive radiation theory prevails for gentians in this region, although rare shifts of diversification rates are associated with niche shifts across the alpine/subalpine ecotone.

Aim We identify the main forest vegetation types in Taiwan, provide their formal definitions and describe their species composition, habitat affinities and distribution. Location Taiwan. Methods A data set of 9822 vegetation plots with environmental characteristics recorded in the field or derived from digital maps in GIS was compiled from historical literature and an extensive field survey. Using expert knowledge, 6574 of these plots were used to build a classification into broad vegetation types. The units of the resulting classification were formally defined using a Cocktail determination key, which can be used for the automatic assignment of new vegetation plots to these vegetation types. Results Twelve vegetation types of zonal forests and nine types of azonal forests were distinguished. Zonal types in the subtropical region, from high mountains to foothills, are Juniperus subalpine coniferous woodland, Abies–Tsuga upper‐montane coniferous forest, Chamaecyparis montane mixed cloud forest, Fagus montane deciduous broad‐leaved cloud forest, Quercus montane evergreen broad‐leaved cloud forest, Machilus–Castanopsis sub‐montane evergreen broad‐leaved forest, Phoebe–Machilus sub‐montane evergreen broad‐leaved forest and Ficus–Machilus semi‐evergreen foothill forest. Zonal types in the tropical region, from high mountains to foothills, are Pasania–Elaeocarpus montane evergreen broad‐leaved cloud forest, Drypetes–Helicia sub‐montane evergreen broad‐leaved forest, Dysoxylum–Machilus foothill evergreen broad‐leaved forest and Aglaia–Ficus foothill evergreen broad‐leaved forest. Azonal types are Illicium–Cyclobalanopsis tropical winter monsoon forest, Pyrenaria–Machilus subtropical winter monsoon forest, Diospyros–Champereia tropical rock‐outcrop forest, Zelkova–Quercus subtropical rock‐outcrop forest, Pinus successional woodland, Alnus successional woodland, Trema–Mallotus successional woodland, Scaevola–Hibiscus seashore woodland and Kandelia mangrove. Conclusions The diversity of forest vegetation in Taiwan is strongly structured by the temperature and moisture gradient. Along the temperature gradient, five altitudinal zones can be recognized. Azonal forest types develop at sites affected by the winter monsoon, on steep slopes, rocky soils, in seashore saline habitats and in places disturbed by fire, landslides and human activities. Zonal vegetation contains a higher ratio of endemic and Pacific species and occurs in wetter habitats, whereas azonal vegetation contains co‐existing species from different regions and usually occurs in drier habitats.

Phylogeography (Avise et al., 1987) is a branch of historical biogeography that involves determining the history of taxa in space and time by integrating the phylogenetic and geographical patterns. This is carried out especially at the level of species complex or intraspecific populations to reveal how the present distribution patterns of taxa have been shaped by geological events or other factors. As the taxon is at species level, the scale in time and space is much shorter and smaller than the study of generic or higher taxonomic levels, and is suitable for revealing the taxon's history from the Quaternary, < 2 Myr.The origin of the island Taiwan can be traced to the Pliocene, about 4-5 Ma, when it began to emerge. Taiwan quickly became the present shape at about 2 Ma through mountain building (Ho, 1982;Shaw, 1996). It is thus suitable to study Taiwan's species based on phylogeography. Taiwan ABSTRACT Aim This paper described current phylogeographical patterns of chloroplastic DNA variation of Trochodendron aralioides, a temperate tree species, and inferred its possible refugium in Taiwan. This information was compared with the known phylogeographical pattern of subtropical tree species.Location A total of 24 populations were sampled including 20 from Taiwan, two each from the Ryukyus and Japan.Methods A haplotype network was constructed by computer program TCS, various parameters of genetic diversity were calculated and neutrality was tested by computer program DnaSP. To examine the similarity of genetic structure among populations, a maximum parsimony tree was reconstructed by computer program PAUP*. The results of isozyme of T. aralioides from a previous publication were incorporated into this study to infer the phylogeographical history.Results Nine haplotypes according to six substitutions, two indels and one inversion of the two cpDNA intergenic spacer fragments (petG-trnP and petApsbJ) of T. aralioides were recognized. Genetic structure of the population of Japan is totally different from those of Taiwan and the Ryukyus. In Taiwan, the genetic structure was differentiated among populations revealed by G st ¼ 0.700 and N st ¼ 0.542, and the population genetics was clearly spatially structured. Two population groups were recognized. The first group was distributed islandwide and extended to the Ryukyus. The second group contained five of the seven known haplotypes, and was restricted to the area between latitude 24°46¢ and 24°06¢ N.Conclusions In Taiwan, north-central area between latitude 24°46¢ and 24°06¢ N is potentially a refugium during the last glaciations. This finding is contradicted to subtropical species as Cyclobalanopsis glauca.

Montane cloud forest is one of the most endangered ecosystems. However, there are few comprehensive studies on the distribution of subtropical montane cloud forest (SMCF). Chamaecyparis forest is one type of SMCF in Taiwan, distributed across the whole island. This study describes eleven types of this forest in Taiwan based on the Braun‐Blanquet approach. Plots were selected from the National Vegetation Database of Taiwan. Two alliances were defined, both of which belong to the order Fagetalia hayatae. Topography and altitude explain the contrasting habitat requirements of these two alliances, whereas seasonality of moisture, soil properties and altitude explain differences in floristic composition at the association level. The alliance of Chamaecyparidion formosanae on slopes and ridges includes coniferous or mixed coniferous and evergreen broad‐leaved forests; it is found at higher altitudes and is more influenced by the summer monsoon than the other alliance. Five associations are defined within this alliance. The alliance of Pasanio kawakamii‐Machilion japonicae growing on slopes and in valleys contains evergreen broad‐leaved forests or forests with a mixture of coniferous and evergreen broad‐leaved species. Six associations can be determined under the alliance of Pasanio kawakamii‐Machilion japonicae. Classification of each syntaxon was formalized using Cocktail Determination Key.

Ecological speciation has long been noted as a central topic in the field of evolutionary biology, and investigation into the relative importance of ecological and geographical factors is becoming increasingly emphasized. We surveyed genetic variation of 277 samples from 25 populations of nine Rhododendron species within Tsutsusi subgenus in Taiwan using simple sequence repeats of expressed sequence tags. Bayesian clustering revealed four genetic lineages: (1) the Rhododendron simsii, Rhododendron kanehirai, and Rhododendron nakaharae lineage (lineage 1); (2) the R h o d o d e n d ro n l o n g i p e r u l a t u m , R h o d o d e n d ro n breviperulatum, and Rhododendron noriakianum lineage (lineage 2); (3) the Rhododendron rubropilosum lineage (lineage 3); and (4) the Rhododendron oldhamii lineage (lineage 4).Asymmetric introgressions were found from lineage 3 into lineages 1 and 2 (introgressed lineages). Genetic admixture of non-R. oldhamii species was also revealed by a neighborjoining tree. Variation partitioning showed that environment explained much larger portions of genetic variation than geography between non-introgressed lineages (i.e., between R. oldhamii and other lineages). However, the Mantel and partial Mantel tests and the multiple matrix regression with randomization found that isolation-by-distance played a more important role than isolation-by-environment (IBE) in contributing to genetic variation in most between lineage comparisons. Nevertheless, strong IBE was found when compared between non-introgressed lineages of R. oldhamii and R. rubropilosum, suggesting post-speciation ecological divergence. Several environmental variables, including annual mean temperature, aspect, isothermality, seasonal precipitation, slope, and soil pH, could be important ecological drivers involved in reproductive isolation between R. oldhamii and non-R. oldhamii species within the Tsutsusi subgenus.