The diversification times (crown ages) of the endemic genera were obtained through dated phylogeny in polytypic genera or phylogeography (coalescence time of haplotypes across distributions) in monotypic genera ( Supplementary Table S1). The median and 95% highest-probability density interval of the divergence time for the endemic seed plant genera were collected. The mean and median values of the diversification times were calculated. The number and accumulation of genera diversified in every five million years were calculated and graphed. Diversification times between monotypic and polytypic genera was compared. The first appearance of fossils of endemic relict genera outside or inside of EA were compiled mainly from the studies of Manchester et al. (2009), Chen et al. (2018), and Tang et al. (2018).

To illustrate the spatial distribution of diversification times, we divide EA into different regions. Because it is hard to collect distribution data based on biogeographical regions, level-3 regions defined in the Taxonomic Databases for Plant Sciences (TDWG, Brummitt, 2001) based on botanical countries or aggregations of constituent provinces in a large country fit well to the monsoon climate regions, and the Plants of the World Online (POWO, https://powo.science.kew.org/) database provides well-collected distribution data according to TDWG codes, so 29 regions, namely, Amur, Assam, Bangladesh, Cambodia, China North-Central, China South-Central, China Southeast, East Himalaya, West Himalaya, Hainan, India, Inner Mongolia, Japan, Khabarovsk, Korea, Laos, Manchuria, Mongolia, Myanmar, Nansei-shoto, Nepal, Primorye, Qinghai, Sakhalin, Taiwan, Thailand, Tibet, Vietnam, and Xinjiang (Brummitt, 2001), were identified. The weighted mean diversification times (MDTs, Lu et al., 2018) and average divergence times (ADTs) in each region were calculated by integrating spatial distribution and radiation dates. AGEi represents the diversification times of a genus i (i = 1, …, n) in a region, and Si is the species number in the genus. From this, MDT and ADT were calculated as:

Spatial distribution and species number data of all genera were obtained from the POWO database.

Overall, we obtained divergence times for 96 taxa (101 genera), including seven complexes (Ombrocharis/Perillula, Soroseris/Stebbinsia, Oresitrophe/Mukdenia, Changium/Chuanminshen, Physochlaina/Przewalski, Perilla/Keiskea, and Akebia/Archakebia) and four lineages (Phtheirospermum-1 and Phtheirospermum-2 and Syncalathium-1 and Syncalathium-2), and the taxa comprised 26 trees, 20 shrubs, 48 herbaceous plants, and 3 lianas in 76 polytypic and 20 monotypic taxa ( Supplementary Table S1 ).

The oldest divergence was dated to the late Eocene (35.91 Ma, Hovenia) and the youngest to the late Pleistocene (0.04 Ma, Gaoligongshania). The mean divergence time was 6.94 Ma with a 95% confidential interval (CI) of 5.54–8.34 Ma (median = 4.52 Ma). Diversification events slowly accumulated before a bloom after the late Miocene, and 80% (n = 77) of the sampled genera diversified after 10 Ma ( Figure 2 ). However, radiations of monotypic taxa were more recent (median = 4.05 Ma) than those of polytypic taxa (median = 7.84 Ma) ( Supplementary Figure S1 ). Among monotypic taxa, 80% (16/20) had population divergence after the Miocene (<5.3 Ma) and 55% (11/20) diverged during the Pleistocene ( Figure 2 and Supplementary Table S1 ). Accelerated accumulation of radiation events after the late Miocene (77%, 59/77) also occurred in the polytypic taxa ( Supplementary Figure S2 ).

We found that 23 of 101 sampled genera are represented in fossils ( Table 1 ). The earliest fossils known are found mostly (22/23) before the Miocene with the oldest dating to Triassic (Ginkgo), while the relictualism is mainly (83%, 19/23) diverged after late Miocene ( Supplementary Table S1 ), and about half (12/23) of the earliest fossils in the EA region are not older than Miocene with ancient genera that could be traced back to Cretaceous or even the Jurassic period ( Table 1 ).

We obtain ca. 600 distribution information of the sampled 96 taxa, and the number of taxa distributed in the 29 regions ranges from one to 83 with the highest density located in south and central China (China North-Central, China South-Central, and China Southeast region) ( Figure 2 and Supplementary Table S2 ). The MDTs range from 2.4 Ma (in Khabarovsk) to 16.50 Ma (in Nansei-shoto), and a gradual decrease from coastal regions to inner mainland regions is shown. The MDTs of plants located at the ISM and the transitional region are mostly in the mid-Miocene, and those located at the EASM region are in the middle to late Miocene ( Figure 3 and Supplementary Table S2 ). Moreover, the ADTs have the youngest time in Khabarovsk (ADT = 3.22 Ma) and the oldest in India (ADT = 15.18 Ma), and the MDTs of plants located in the monsoon region are mostly in the late Miocene ( Figure 3 and Supplementary Table S2 ).

The diversification events of over 100 (approximately 17%) endemic seed plant genera were compiled to investigate the evolution of EA plant endemism. A similar ratio of different life forms within the sampled genera compared to that of all endemic genera (Manchester et al., 2009) ensured representativeness. Diversifications (crown ages) were much more recent than origins (stem ages) in both the oldest lineage (35.91 versus 68.25 Ma) and the median values (4.52 versus 13.60 Ma) (Chen et al., 2018). The majority of the diversifications, which occurred after the late Miocene (ca. 10 Ma, Figure 2 ), also lagged behind the formation of the main body of flora, which occurred after the Miocene (ca. 23 Ma) (Chen et al., 2018; Lu et al., 2018). This time lag is similar to that found by Ramírez-Barahona et al. (2020), who shows that the origins of families were 37–56 Ma older than their diversifications across the entire angiosperm tree.

The diversification events began in the late Eocene and its accumulation did not obviously increase until after the late Miocene ( Figure 2 and Supplementary Figure S2 ), which may correlate with the uplift of mountains in southwest China (Harrison et al., 1992; Xing and Ree, 2017) or, more likely, the intensification of the Asian Monsoon (Yu et al., 2017; Wang et al., 2020). The inference receives further support from the spatial pattern. First, the distributions of sampled genera are centered in south and central China, similar to species or endemism diversity found in previous studies (Huang et al., 2016; Xu et al., 2016; Tang et al., 2018), indicating a good representativeness. Then, the MDTs and ADTs in the EASM, ISM, and transitional region fall in the mid-late Miocene ( Figures 1 , 3 ). The fossils of Tertiary relict genera are mostly and widely distributed outside EA and only half of them are firstly found in EA during the Miocene, so relictualism experiences extinctions not only outside but also inside the EA, and the present species/populations are only prosperous after late Miocene when monsoon brings humid climate (Ye and Li, 2022). Moreover, it is shown that the sampled monotypic endemic taxa are mainly diversified after Pliocene, and the number of diversified genera in this period also exceeds that in the late Miocene in polytypic taxa ( Supplementary Table S1 and Figure S2 ). We suggest that the reinforcement of the EASM/ISM during the late Pliocene (An et al., 2001; Wan et al., 2006; Clift et al., 2015) and the glacial–interglacial periodic alternations in Pleistocene (Ye et al., 2017a) may also exert a greater influence on species and population divergence.

Previous studies have demonstrated the influence of monsoon development on the floral evolution in EA at various taxonomic levels. Floral-level investigations of the Sino-Japanese floristic region (Chen et al., 2018), Chinese angiosperms (Lu et al., 2018), and dominant taxa in EBLFs (Xiang et al., 2016; Deng et al., 2017; Yu et al., 2017) have emphasized the climatic changes during the late Oligocene and early Miocene. Li et al. (2021) provided additional evidence by integrating modeling (SDGVM and JeDi-DGVM) and fossil data and found that the orographic evolution in northern Tibet affected vegetation and, especially, plant diversity in EA during the Neogene by altering the monsoon (both summer and winter) system. Support was gained by other studies, such as Zhang et al. (2021) who found that TP growth increased annual precipitation in East China by comparing the effects of TP growth, Paratethys Sea retreat, and global cooling on the East Asian climate change that occurred by the early Miocene. Chen and Su (2020) also found that patterns of woody dicotyledon richness in humid regions of China were mainly affected by the Asian winter monsoon by exploring the relationship between these patterns and climatic variations, especially in the Asian monsoonal climate.

Studies at the regional scale have provided further support for this idea. For example, in the Hengduan Mountains, which has the richest temperate alpine flora worldwide, Ding et al. (2020) recently found that during the early to middle Miocene and late Miocene, accelerated diversification and colonization of adjacent regions were likely driven not only by mountain development but also by Asian summer monsoon intensification. In EBLFs, Ye and Li (2022) compiled 10 radiation dates for relict evergreen genera, four diverged around the late Miocene–Pliocene boundary and four diverged during the late Pliocene, this is resulted from an increasingly humid climate after the late Miocene (An et al., 2001; Wan et al., 2006; Clift et al., 2015) and the absence of large ice sheets in the Pleistocene (Qiu et al., 2017), which allowed these genera to prosper in EBLFs. This is further supported by the origins of the dominant species in EBLFs (Ye and Li, 2022), they collected the origin time of 92 out of 291 (31.6%) dominant species and found that 70 (76.1%) species originated after ca. 8 Ma and accelerated accumulations began after 4 Ma. In Taiwan, an orogenic continental island that has high levels of endemic biological richness (Kier et al., 2009), we found that the late Pliocene–Pleistocene (3.5–2.0 Ma) was a key period for the origin of its floral endemism, as 28% (30/107) of the investigated endemic seed plants originated during this stage (Ye and Tian, unpublished data). Additionally, palynological analyses in East China have shown that well-developed mixed evergreen and deciduous broadleaved forests were strongly influenced by the EASM during the Holocene (Chen et al., 2020).

Evolution at the genus level showed close correlations with the monsoon intensification periods. Previous biogeographic studies have revealed that the intensified EASM during the late Miocene significantly increased the diversification rate in EBLFs (Deng et al., 2017; Yu et al., 2017; Wang et al., 2020). Kong et al. (2017) used environmental correlation analysis to show that the speciation rate of Primulina is tightly linked to the EASM strength, which provides suitable habitats for calciphiles by facilitating the dissolution and weathering of karsts. Furthermore, Kong et al. (2021) indicated that monsoon development, rather than mountain development, was probably the primary driver of the rapid diversification in Oreocharis. The Asian monsoon (EAM and/or IM) intensification had a similar influence on Rhododendron (Xia et al., 2021), Pinus (Liu et al., 2022), Hedychium (Ashokan et al., 2022), and Typhonium (Low et al., 2021). The transition to a monsoon climate system during the late Oligocene may have also resulted in divergence among species, such as Prinsepia (Ma et al., 2019).

Furthermore, monsoon intensification has promoted intraspecific population divergence (Ye et al., 2017a). Phylogeographic studies have suggested that periodic glacial–interglacial alternations are responsible for genetic divergences (Qiu et al., 2011). However, Zhang et al. (2013) found two sympatric Chinese beeches (Fagus lucida and F. longipetiolata) in subtropical China diverged during the late Miocene. These species likely migrated southward due to unsuitable environments in high latitudes; the significantly intensified EASM would have provided a major shelter. Another deciduous tree, Cyclocarya paliurus, originated during the early to middle Miocene and shows two peaks of lineage diversifications during the late Miocene (ca. 9.6 Ma) and late Pliocene (ca. 3.6 Ma), which coincide with the EASM intensifications (Kou et al., 2016). In evergreen plants, such as Cephalotaxus oliveri (Wang et al., 2014), Quercus glauca (Xu et al., 2015), Tetrastigma hemsleyanum (Wang et al., 2015), Phoebe zhennan (Xiao et al., 2020), Engelhardia roxburghiana, and E. fenzelii (Meng et al., 2022), their divergences and demographic expansions have also been correlated with the development of summer monsoons. In addition, the monsoon may act as a vector that facilitates long-distance dispersal (Li et al., 2012) or cause different environmental conditions owing to different monsoons (EAM or IM), which may cause/reinforce ecological adaptations (Liu et al., 2013).

The first concerns with sampling bias. Although a similar ratio of life forms in the sampled genera compare to that of all EA endemic genera is shown, limited sampling (<20%) would result in bias. Besides, an insufficient sample of species in polytypic genera or populations in monotypic genera may cause an underestimation of radiation times (Qiu et al., 2011). In the present study, 73/76 polytypic taxa contain almost all species, 15/20 monophyletic genera cover samples from its entire geographic range, and three have restricted distributions (e.g., Gaoligongshania megalothyrsa) ( Supplementary Table S1 ), minimizing the possibility of underestimations. Finally, the spatial distribution of divergence time is based on level-3 regions of the TDWG, which are mainly divided by political boundaries in EA ( Figure 3 ), and the optimal evaluations of monsoon influence are probably derived from divisions based on monsoon regions ( Figure 1 ), while similar conclusions to this study are likely deduced.

The second would be caused by compiling methods. Following Chen et al. (2018) and Ye and Li (2022), we directly adopt the estimated times in previous studies. In the compiled 96 taxa, the divergence times of 19 taxa are calibrated using substitution rates that are varied among the different taxa (Drouin et al., 2008), and 12 taxa are calibrated only by one fossil or secondary point ( Supplementary Table S1 ). Moreover, sequences from only one or two fragments are obtained to reconstruct the phylogeny ( Supplementary Table S1 ). Although diversification rates or events are assembled after time estimations based on published or self-generated sequence data in Xing and Ree (2017) and Li et al. (2022), bias is inevitable in different dates–phylogenies using different calibrations and genetic markers. Dating a tree of life using the same calibrations and markers may exert a low resolution among genera/species (Lu et al., 2018).

The third is related to the combined influence of different geoclimate changes. The evolution of flora in EA could not solely be explained by the strengthening of the monsoons. The most important contributors to monsoon development, which are likely TP uplift and global cooling (Lu and Guo, 2014; Li et al., 2021), also exert great influence on floral evolution. Mountain uplift could lead to in-situ speciation, preserving dispersed species, preventing extinctions, or creating geographic isolations and so on (Favre et al., 2015; Muellner-Riehl, 2019); such as in T. hemsleyanum, the uplift of eastern TP may have split the species’ ancestral distribution into isolated southwestern and southern populations. Global cooling in Neogene and Pleistocene has driven the extinction of species/populations in the high-latitude region and results in southward retraction (Ye and Li, 2022). For example, the colonization to subtropical China of C. paliurus during the mid-Miocene might be partly explained by its sensitivity to global cooling after the Miocene optimum (Kou et al., 2016). Other geoclimate changes, such as increased precipitation during the dry (winter) season (Li et al., 2021), aridity in the Asian interior (Zhang et al., 2013), and the Holocene Climatic Optimum (Chen et al., 2020), also influence genetic divergence. Additionally, the monsoon not only increased species and genetic diversity but also decreased it by causing the disappearance of some gymnosperm genera (Liu et al., 2017).