New insights into the phylogenetic and biogeographic analysis of Elaeocarpus (Elaeocarpaceae) in China, and further consolidated ‘Acronodia’ as a distinct group

Abstract

Background

Elaeocarpus is the most species-rich genus in Elaeocarpaceae (Oxalidales), comprising 39 species of trees that grow in tropical and subtropical forests in China, 14 of which are endemic. Few studies to focus on the phylogeny of Elaeocarpus in China. Limited available evidence indicates a close phylogenetic relationship between Sect. Ganitrus and Sect. Dicera, while the question of whether the 'Acronodia' group warrants taxonomic separation from Sect. Monocera remains unresolved. The status of group ‘Acronodia’ in Elaeocarpus is uncertain because the combination of molecular fragments available to construct the phylogenetic tree has not been evaluated.

Results

In this study, we compared chloroplast genome sequences on the basis of the alignment of 4 chloroplast genome sequences with the mVISTA and KaKs_Calculator tools. The results revealed that the phylogeny has good bootstrap value for ycf1, ITS and trnS-atpA and that 27 Elaeocarpus species (40 samples), including 3 species that are distributed naturally outside China, are grouped into 2 major clades: one comprising Sect. Ganitrus and Sect. Dicera, and the other consisting of Sect. Monocera and the ‘Acronodia’ group. Furthermore, the results of the phylogenetic analysis with the BEAST tool suggest that Elaeocarpus originated from Southwest China during the early Eocene (40 Ma) and started to diversify south of the Yangtze River during the early Miocene (15 Ma).

Conclusion

Overall, this study highlights the taxonomic utility of chloroplast genomes in Elaeocarpus, and the time and regions of origin will facilitate future studies on conservation.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12870-025-07526-2.

Introduction

Since 2000, the rapid development of molecular marker technologies has driven the application of DNA barcodes to plant systematics in animals [29]. DNA barcodes are easy to amplify in plant chloroplast genomes because they are relatively conserved [6, 15, 40]. Thus, the use of DNA barcoding as a classification tool has been critical in the study of phylogenetic materials, such as medicinal plants, in recent years [.2, 9, 78]. A phylogenetic tree produced from DNA barcodes combined with fossil materials effectively estimates the divergence time of plants [19]. Further studies have shown that a relaxed molecular clock is more accurate [21] and is closely related to historical geological events [1, 33].

Elaeocarpus is the largest genus in Elaeocarpaceae and encompasses approximately 350–400 species that inhabit tropical and subtropical forests, of which 39 species grow in China [14, 55, 76]. Many species within the genus possess considerable practical importance. Several species have attractive evergreen foliage and widely cultivated as ornamental trees, such as E. glabripetalus [55]. Additionally, some species like E. braceanus produce edible fruits and certain parts have been steeped in liquor [71]. Morphological studies revealed that Elaeocarpus in the Flora Reipublicae Popularis Sinicae (FRPS) divided into 3 groups: Sect. Ganitrus, Sect. Dicera, Sect. Monocera [82], while the Sect. Acronodia is admitted to independent from Sect. Dicera in Flora Yunnanica [75] and Flora of China [69]. The sect. Acronodia was established by Master [46] and used in the classification of Elaeocarpus plants in Malaysia [13].

The infrageneric classifications of Elaeocarpus from China are based on intuitive approaches and have not been examined via a phylogenetic approach to test the monophyly of the groups and to determine the morphological synapomorphies. Among the fossil records of Elaeocarpus, the first record of fossil leaves of Elaeocarpus was discovered in Miocene sediments in Australia and India [3, 38]. Moreover, following the rapid progression of molecular genetic techniques and fossil research materials, a phylogenetic tree of Elaeocarpaceae constructed based sequence data from the internal transcribed spacer of nuclear ribosomal DNA [47]. Combine multiple fragments analysis revealed the position of Elaeocarpus in Elaeocarpus alliance (Sericolea, Aceratium and Elaeocarpus) and the main lineages within infragenus [17, 55]. The estimated age of the most recent common ancestor of Elaeocarpus at 46 Mya using only 13 species in Elaeocarpaceae and performed a penalized likelihood analysis of trnL‒trnF region and ITS sequence dataset [17]. The origin and evolution of the Elaeocarpus genus were discussed by Phoon [55], who noted that this group originated in Australia in the Eocene and migrated to surrounding regions, especially Indonesians, mostly in the Oligocene and Miocene.

However, the samples of previous studies focused on mainly from Australian and Indonesian, and lacking the samples from China and Indochina Peninsula, which will reduce the credibility of the assessment of the origin time of Elaeocarpus. In addition, the bootstrap values of the phylogeny tree in some lineages of Elaeocarpus was relatively low. Which indicating that the phylogeny tree constructed of trnL‒trnF region, trnV‒ndhC region and ITS sequence are not suitable for species identification and interspecific relationship analysis in Elaeocarpus. Furthermore, in combination with morphological research, the status of Sect. Acronodia is difficult to define. Here, we selected another divergence hotspot regions of Elaeocarpus to explore the phylogenetic relationships between Elaeocarpus and other genera in detail.

This study aims to (1) evaluate effective molecular markers for phylogeny reconstruction; (2) clarify the phylogenetic position of Sect. Acronodia, and elucidate the intersectional relationships within Elaeocarpus in China; (3) To infer the divergence time and biogeographic history between the genera and genera of the genus Elaeocarpus in China.

Materials and methods

Plant sampling and DNA extraction

The leaf materials were sampled from 27 species of Elaeocarpus (40 samples), including 3 species that are distributed naturally outside China, and were collected from the field in China and from the Royal Botanic Gardens (Table 1). The voucher specimen of the collection was deposited at the Nanling Herbarium of Gannan Normal University (GNNU), the Museum of Beijing Forestry University (BJFC) and the Royal Botanic Gardens (K). Total genomic DNA was extracted via the CTAB method [10] and then sent 4 species (E. japonicus, E. angustifolius, E. hainanensis and E. japonicus var. yunnanensis) to the Sino Geno Max Company for next-generation sequencing via the Illumina HiSeq (TM) 2000 platform [28, 41] in Beijing, China.

Table 1.

Sampled species of Elaeocarpaceae and their voucher specimens

NO	Species	Herbarium code	Voucher specimems number	Geographic Origin	Identifier
1	Elaeocarpus poilanei	GNNU	XYF010165	Diaoluo mountain, Hainan	Yihui Wang
2	Elaeocarpus decipiens	GNNU	PVHJX012837	Sanqing mountain, Jiangxi	Yihui Wang
3	Elaeocarpus duclouxii	BJFC	XZ622	Xichou county, Yunnan	Yifei Xie
4	Elaeocarpus howii	BJFC	XW1869	Malipo county, Yunnan	Yifei Xie
5	Elaeocarpus lacunosus	BJFC	XZ587, XZ592, XZ596	Maguan county, Yunnan	Yifei Xie
6	Elaeocarpus sylvestris	BJFC	XZ2058, XZ632	Fangchenggang City, Guangxi	Yifei Xie
7	Elaeocarpus braceanus	BJFC	XZ541	Malipo county, Yunnan	Yifei Xie
8	Elaeocarpus dongnaiensis	Kew	S-566	The Royal Botanic Gardens	Nguyen. V..D
9	Elaeocarpus serratus	GNNU	PVHJX014292	XiShuangBanNa Tropical Botanical Garden	Yihui Wang
10	Elaeocarpus glabripetalus	BJFC	20160201	Xichou county, Yunnan	Yifei Xie
11	Elaeocarpus austroyunnanensis	BJFC	XZ534, XZ545, XZ549	Nannuo mountain, Yunnan	Yifei Xie
12	Elaeocarpus sikkimensis	BJFC	XZ536	Lancang county, Yunnan	Yifei Xie
13	Elaeocarpus angustifolius	BJFC	140942	Guangxi Academy of forestry	Yifei Xie
14	Elaeocarpus japonicus	BJFC	XZ616, XZ620	Wugong mountain, Jiangxi	Yifei Xie
15	Elaeocarpus japonicus var. yunnanensis	BJFC	XW1746, XZ601	Xichou county, Yunnan	Yifei Xie
16	Elaeocarpus chinensis	BJFC	XW1653	Xichou county, Yunnan	Yifei Xie
17	Elaeocarpus nitentifolius	GNNU	XYF010170	Jianfeng mountain, Hainan	Yihui Wang
18	Elaeocarpus petiolatus	GNNU	XYF010183	Diaoluo mountain, Hainan	Yihui Wang
19	Elaeocarpus bachmaensis	BJFC	G100011, XZ16_558	Baihua mountain, Yunnan	Yifei Xie
20	Elaeocarpus prunifolioides	BJFC	XZ16 558	XiShuangBanNa Tropical Botanical Garden	Yifei Xie
21	Elaeocarpus varunua	BJFC	XW1532	Baizse county, Guangxi	Yifei Xie
22	Elaeocarpus dubius	BJFC	XZ530	Yingge mountain, Hainan	Yifei Xie
23	Elaeocarpus branderhorstii	Kew	24004	The Royal Botanic Gardens	Coode, M.J.E
24	Elaeocarpus pycnanthus	Kew	24006	The Royal Botanic Gardens	Coode, M.J.E
25	Elaeocarpus hainanensis	GNNU	PVHJX014291	Diaoluo mountain, Hainan	Yihui Wang
26	Elaeocarpus rugosus	GNNU	XYF010102	XiShuangBanNa Tropical Botanical Garden	Yihui Wang
27	Elaeocarpus grandiflorus	GNNU	XZ538	Yingjiang county, Yunnan	Yihui Wang
28	Sloanea sinensis	BJFC	XW1956	Wenshan, Yunnan	Yifei Xie
29	Aristotelia fruticosa	Kew	781	The Royal Botanic Gardens	Chase, M.W
30	Crinodendron patagua	Kew	19978	The Royal Botanic Gardens	Chase, M.W
31	Vallea stipularis	Kew	654	The Royal Botanic Gardens	Chase, M.W

Herbarium codes refer to the following institutions

GNNU Nanling Herbarium of Gannan Normal University, China; BJFC Museum of Beijing Forestry University, China; Kew, Royal Botanic Gardens, UK

Genome comparison and phylogenetic analysis

Given that the important lineages within Elaeocarpus in China consist of Sect. Ganitrus, Sect. Dicera, Sect. Monocera, and group ‘Acronodia’, The main chloroplast genome comparisons among the 4 previously published chloroplast genome sequences of Elaeocarpus (E. japonicus MT683335, E. angustifolius MW242787, E. hainanensis MW602804 and E. japonicus var. yunnanensis MW242788) were carried out under the Shuffle‐LAGAN mode via the mVISTA program [4, 48] to elucidate the level of sequence divergence. To estimate the nucleotide substitution rates, we computed the ratio of nonsynonymous substitutions per nonsynonymous site (Ka) to synonymous substitutions per synonymous site (Ks) for genes using the YN model implemented in KaKs_Calculator Version 3.0 [83]. For non-coding sequences, the ratio of the non-coding substitution rate (Kn) to the neutral substitution rate (Ks) was estimated by aligning and comparing adjacent regions. we identified hotspots characterised by high numbers of nucleotide substitutions and highly divergent regions. The primer pairs used for the amplification of genes were designed via Primer Premier 6 (Table 2). The PCR products were verified by gel electrophoresis on a 1.5% agarose gel. To develop specific DNA barcodes, amplification was performed according to the following conditions: 94 °C for 4 min, 35 cycles at 94 °C for 30 s, 54 °C for 30 s and 72 °C for 1 min; and a final extension at 72 °C for 6 min. The PCR products were sequenced directly via the Sanger method [63]. The sequences number from NCBI amplified with high efficiency are presented in Table S1.

Table 2.

List of primers and adapters used for amplification and sequencing of DNA regions

DNA regions	Primer	Sequence (5’to 3’)	Reference	amplification efficiency
trnS-atpA spacer	trnS-2F trnS-2R	TACGGATTCTTGCACAATTC ATTCAAAACACCTCGCTTTC	This study	100%
ycf1	ycf1-2F ycf1-2R	GAATCATCCCATCTTAATTG TTCATTTTAAAGCAGACTCA	This study	100%
ITS	ITS4 ITS5	TCCTCCGCTTATTGATATGC GGAAGTAAAAGTCGTAACAAGG	White et al.,(1990)	100%
ndhF-trnL spacer	ndhF-3F ndhF-3R	TTGCGAAAATTGGTGAATAC CATACCTTTTGTTACACTTC	This study	60%
psbE-petL spacer	psbE-1F psbE-1R	AGCTCCGCATATTCTTGTAC TGAAGGAGCTAAATGAAATA	This study	50%
psbZ-trnG spacer	psbZ-1F psbZ-1R	AACCTATTCGTTCCAGATAC TGATACGCAATGTCTGGATC	This study	60%
rbcL-psaI spacer	rbcL-2F rbcL-2R	CATCTATTGTATTTTCATGT CCATAACCCCAATTGCTACG	This study	80%
ycf3-trnS spacer	ycf3-1F ycf3-1R	GGTCGGAATAGGCAGGTAAT ATAGGCCTTATCCACACAAG	This study	80%
rpl16 spacer	rpl16-1F rpl16-1R	CTATCTCGAAATAATGAATT ATCCATTTATTCTGAGAGGC	This study	60%

Phylogenetic analysis were conducted on selected regions of the above 27 species and outgroups (Sloanea sinensis, Vallea stipularis, Crinodendron patagua, and Aristotelia fruticosa) based on ITS, trnS-atpA and ycf1 sequence regions. Bayesian inference analysis were performed via MrBayes version 3.2.6 [59] on a dataset of three sequence fragments (ITS, trnS-atpA and ycf1) from 31 species. The whole-genome matrix was aligned via MAFFT version 3.73 [35] followed by manual editing in Geneious version 9.1.7 [36]. The optimal DNA substitution model, determined through Bayesian information criterion analysis with jModeltest version 2.1.10 [20, 24], was GTR + F + I. Markov chain Monte Carlo (MCMC) analysis was conducted in MrBayes over 10,000,000 generations, starting from a random tree and sampling every 1,000 generations. The initial 25% of the trees were discarded as burn-in, while the remaining trees were utilized to produce a majority rule consensus tree. Additionally, we estimated a maximum likelihood phylogeny for the genera using RAxML v8.0.0 [66] on the CIPRES web server (www.phylo.org), employing default settings and the GTR + F + I model of sequence evolution.

Molecular clock dating and diversification rates

Using BEAST 1.10.4, we implemented a lognormal relaxed clock model with the GTR + F + I site model, which included four gamma categories. The analysis began with a random starting tree and applied a Yule process tree prior [67]. We conducted Markov chain Monte Carlo (MCMC) simulations for 500 million generations, sampling every 50,000 generations, and confirmed that the effective sample size (ESS) values for all parameters exceeded 200. Subsequently, we employed phyutility software to create an all-compatible consensus tree [64]. Node ages were optimized onto this consensus phylogeny as the median value for a given node across all trees in the posterior distribution that contained the node via TreeAnnotator software [22]. Additionally, the phylogeny was calibrated using 2 fossils and 1 prediction node, For the prediction calibration node, we used the 40 ± 30 Ma split between Vallea and Aristotelia as the calibration point [30]. The fossil specimen of E. angustifolius is from Tasmania in Australia, which is approximately 25 ± 2 Ma old [31]. Another fossil, Sloanea artocarpites is from Baden-Württemberg in Germany was used to constrain the divergence between Sloanea and Vallea to 13 ± 3 Ma [49]. The tree was viewed and edited with Fig Tree version 1.4.0 software (http://tree.bio.ed.ac.uk/software/figtree/). Diversification rates at crown nodes of Elaeocarpus were assessed via the GEIGER package version 2.0.3 [27] in R version 4.1.1.

Biogeography analysis

The maximum clade credibility (MCC) tree was calculated in BEAST to estimate the ancestral geographic ranges in Elaeocarpus via reconstruction of the ancestral state in phylogenies (RASP; [79]). The analysis was based on the dispersal extinction-cladogenesis model (DEC; [60]).

Due to the fact that the occurrence records and actual distribution characteristics of East Asian Elaeocarpus are not strictly confined to artificially defined geographical boundaries, such as species dispersal occur in transitional zones between regions. Eight areas were defined to represent the entire distribution range of Elaeocarpaceae, with areas a to d corresponding to the distributions of the four sections within the Asian Elaeocarpus. (a) southern Yunnan, Hainan, India, Indochina Peninsula; (b) Yunnan, Guangxi, Hainan, Indochina Peninsula; (c) southern Yangtze River, Indochina Peninsula; (d) Japan, southern Yangtze River, northern Vietnam; (e) South America; (f) Australia; (g) Indonesia; and (h) India.

Results

Comparative chloroplast genomic analysis

To identify divergent regions among species, we calculated the common protein-coding genes to assess heterogeneity in evolutionary rates. Among the analyzed genes, large ribosomal protein genes (rpl16) and hypothetical chloroplast reading frames (ycf1) showed relatively high Ka/Ks values (Ka/Ks > 0.7), suggesting that they may have undergone positive selection. The remaining genes exhibited lower Ka/Ks values (Ka/Ks < 0.7), indicating that they are primarily under purifying selection. Furthermore, the Kn/Ks ratios were compared across non-coding regions to identify divergence in evolutionary rates. Among the non-coding regions, those with high Kn/Ks values across 4 species were selected. Elevated Kn/Ks values (Kn/Ks > 6) were detected in the following intergenic regions: ndhF-trnL, psbE-petL, psbZ-trnG, rbcL-psaI, trnS-atpA and ycf3-trnS (Supplementary Fig. S2).

Sequence identity plots of the chloroplast genomes of four species were generated with the annotation of the E. japonicus chloroplast genome as a reference. The results revealed that the four chloroplast genomes presented highly conserved identities (Supplementary Fig. S1). In the chloroplast genomes, the LSC and SSC regions were more divergent than the IR regions were. Furthermore, non-coding regions were more divergent than coding regions, and the highly divergent noncoding regions among the seven chloroplast genomes appeared in the intergenic sequence (IGS), such as ndhF-trnL, psbE-petL, psbZ-trnG, rbcL-psaI, trnS-atpA and ycf3-trnS, and in introns of the genes rpl16 and ycf1. All the above divergent regions were also found to be under positive selection, making them potential candidates for the development of molecular markers and barcodes for future phylogenetic analyses of Elaeocarpus species.

Molecular phylogeny of Elaeocarpus

To determine whether the hotspot regions could be used as DNA barcodes to distinguish Elaeocapus, we selected 8 chloroplast fragments and an rRNA gene fragment to construct phylogenetic trees (Supplementary Fig. S1; Supplementary Fig. S2). The results showed that the phylogenetic tree has good bootstrap value for ycf1, ITS and trnS-atpA, along with a high amplification rate for these regions (Table 2). The phylogenetic tree confirmed that Vallea and Aristotelia are monophyletic and form a clade with Sloanea, and this clade is sister to Crinodendron. Within the Elaeocapus clade, the phylogenetic tree delineates these four groups into two major branches: one comprising Sect. Ganitrus and Sect. Dicera, and the other consisting of Sect. Monocera and the ‘Acronodia’ group. 4 lineages including Sect. Ganitrus, Sect. Dicera, Sect. Monocera, group ‘Acronodia’, are resolved by Bayesian analysis with strong support. The basal node within the Elaeocapus clade is Sect. Ganitrus, which consists of E. angustifolius (posterior probability = 1) and is a sister to Sect. Dicera (posterior probability = 1) consisting of 12 species. Group ‘Acronodia’ (posterior probability = 0.882) within 5 species is part of Sect. Monocera (posterior probability = 0.835) within 9 species (Fig. 1).

Fig. 1.

Strict consensus of Elaeocarpus based on ycf1, ITS and trnS-atpA. The numbers above and below branches are Bayesian posterior probabilities. Morphological character state changes within Elaeocarpus are indicated by braces after the species name with character numbers and sections. The blue bars indicate the 95% highest posterior density (HPD) intervals of the age estimate. The divergence times of the clades and subclades are displayed at the bottom of the picture. The pentagram represents calibration points

Age estimation, ancestral ranges, and diversification rates

The first diversification event via molecular clock analysis was estimated within Elaeocarpaceae at 31.5 Ma (95% highest posterior probability density (HPD): 22.4–43.6 Ma), indicating that Crinodendron began to diversify in the late Oligocene to early Miocene (Fig. 1). The second diversification event included Vallea, Aristotelia and Sloanea. The Aristotelia alliance and Sloanea diverged at 11.2 Ma (HPD: 5.5–19.8 Ma), and Vallea diverged from Aristotelia at 8.5 Ma (HPD: 4.5–12.2 Ma). The Elaeocapus clade diverged approximately 28.8 Ma (HPD: 22.5–36.5 Ma) from the Indochina Peninsula and Southwest China, of which 4 lineages split at different times: Sect. Ganitrus and Sect. Dicera diverged at 24.8 Ma (HPD: 21.2–28.8 Ma). Sect. Monocera alliances, including the ‘Acronodia’ group diverged at 15.9 Ma (HPD: 8.1–23.2 Ma). Sect. Ganitrus and Sect. Monocera evolved into Sect. Dicera and group ‘Acronodia’, with an estimated ancestral range extending from the Indochina Peninsula and Southwest China to southern Yangtze River and Japan. The diversification rates estimated in GEIGER were highest at 4–20 Ma during the middle and late Miocene (Fig. 2).

Fig. 2.

Ancestral state reconstructions and fossil-calibrated phylogeny generated by BEAST and RASP. The divergence times of the clades and subclades are displayed on the bottom left of the picture. Geological time abbreviations: Pli = Pliocene; Q = Quaternary; Ple = Pleistocene; H = Holocene. Distribution areas are shown on the top right of the picture A southern Yunnan, Hainan, India, Indochina Peninsula; B Yunnan, Guangxi, Hainan, Indochina Peninsula; C southern Yangtze River, Indochina Peninsula; D Japan, southern Yangtze River, northern Vietnam; E South America; F Australia; G Indonesia; H India). Semilogarithmic lineage-through-time (LTT) plots are shown on the bottom right of the picture

Discussion

DNA markers and molecular phylogeny

Numerous studies indicate that the detection of widespread adaptive evolution, often inferred from elevated genetic diversity, is fundamentally linked to positive selection acting on specific molecular regions rather than the whole genome sequences [62, 80]. Furthermore, signatures of positive selection serve as reliable indicators for estimating divergence times among species [8, 23, 85]. Traditional DNA barcodes derived from chloroplast loci, such as ITS, rbcL, matK and trnL-trnF, are not suitable for all plant taxonomic groups [44, 53, 74]. In most angiosperms, ycf1 is the most variable plastid genome region and can serve as a core barcode, which is congruent with the phylogenetic tree generated from the DNA fragment in this study [73]. TrnS-rps4 and atpA were proposed for use as DNA barcodes to identify species in ferns [7, 52, 61], which was similar to the findings of our study. However, except for ycf1 and trnS-atpA, all the genes were unavailable for the development of DNA barcodes because of their low PCR amplification efficiency, low sequencing success rate, or poor bootstrap value (Supplementary Fig. S1, Table 2).

In this study, we used the sequence cpDNA and ITS phylogenies (trnS-atpA, ycf1 and ITS) of Elaeocapus to date rather than the trnL-trnF region, trnV-ndhC region and ITS [3, 17, 47, 55]. The constructed phylogenetic tree was broadly congruent with Coode’s system [12]. Unlike previous studies, our study strongly supported the relationships among the four sections and robust species-level support in Sect. Dicera, Sect. Monocera, and the ‘Acronodia’ group. Our phylogenetic tree provides a valuable framework and the best-resolved topology to date for understanding the phylogeny of East Asian Elaeocarpus.

Major clades within Elaeocarpus

The split of Elaeocarpus into two major clades, clade I and clade II, disagreed with previous results [68, 82] (Fig. 1). In addition, according to all previous studies, clade I was divided into Sect. Ganitrus and Sect. Dicera. Clade II was divided into Sect. Monocera and Sect. Acronodia. The morphological differences between the two branches are as that clade I’s petals incised or entire, exocarp shiny; and clade II’s petals laciniate and exocarp inconspicuous.

The three sections and a group within two major clades

Based on the morphological and molecular data, Sect. Dicera is close to Sect. Ganitrus. The majority of the species are described in Sect. Dicera in previous phylogeny studies [3, 14, 46, 55, 68, 82]. The present study included 12 taxa of Sect. Dicera, which formed a well-supported clade. All taxa can be identified not only in morphology but also in molecules [11, 72].

This study examined 11 taxa of Sect. Monocera, and two well-supported subclades were established. The first subclade comprises species with drupe spindle shapes, namely, E. grandiflorus, E. rugosus and E. hainanensis. The second subclade comprises species with drupe ellipsoids, namely, E. prunifolioides, E. bachmaensis, E. varunua, E. bachmaensis, E. dubius and E. petiolatus. A similar result was obtained by Phoon [55], in which the Monocera clade was also divided into 2 groups, but the species studied were distributed only in Malaysia.

For the ‘Acronodia’ group, according to Zhang [82] and Tang [68], the ‘Acronodia’ group is close to Sect. Dicera based morphological classification with flowers polygamous, petals incised or entire, not laciniate, exocarp shiny. This was different from our phylogeny result. Previous studies on Sect. Acronodia, excluding species in East Asia and Indochina, revealed a distribution in western Malesia, with flowers polygamous, 8–12 stamens without awns or significant bristles, 2-merous ovaries, petals scarcely divided or with up to 18 small teeth [13]. E. japonicus and E. nitentifolius were excluded for some of these characteristics, which was inconsistent with our results. Therefore, we redefined the group ‘Acronodia’, which revealed it is close to Sect. Monocera. In other words, the status of the ‘Acronodia’ group in Elaeocapus cannot be clearly determined by morphological and anatomical methods alone.

Paleogene global climate change and its impact on the diversification of Elaeocarpus in Oligocene

Ancestral area reconstructions suggest a widespread distribution of Elaeocarpus in China and the Indochina Peninsula during the Oligocene 31.5 Ma (HPD:18.2–41.5 Ma). This scenario was supported by the wide distribution of fossils with multiple fossils found during the Quaternary in Australia [37, 39, 65] and during the Oligocene in Japan [45, 54] and Asia. Our study inferred an origin from the Indochina Peninsula around the Oligocene, which was induced by the warm and humid peripheral tropical environment from the Oligocene [32]. This long-maintained humid environment provided a source for populating East Asian flora [58]. Climate and vegetation modeling has shown that the rise of the northern part of the Tibetan Plateau and the rise of temperature from the Oligocene had the capacity to alter the Asian monsoon system and induce a precipitation increase across eastern Asia [42], such as Elaeocarpus.

The impacts of animal activities and climate change in the Neogene and Quaternary on the spatial‒temporal patterns of Elaeocarpus in China

The intensification of the East Asian summer monsoon around the Oligocene–Miocene boundary because of the uplift of the northern Tibetan Plateau at 20–25 Ma [70], which induced the spread of Elaeocarpus from Southwest China to the south of the Yangtze River. Recently, During the Quaternary period, tectonic movements and changes in river systems broke up the habitats of Elaeocarpus plants. This created isolated refugees, which compressed the genus's originally continuous distribution ranges. This geographic isolation prompted genetic drift and ultimately led to the emergence of endemic Elaeocarpus species in southwestern China [26, 57, 77]. The profound impact of Quaternary glacial-interglacial cycles on the distribution and genetic structure of endemic plant species in China is further exemplified by comparative phylogeographic studies of co-distributed species such as Taxus wallichiana and Taiwania cryptomerioides. Taxus wallichiana exhibits a marked east–west genetic split, with proposed refugia in the Wuyi Mountains and the Yunnan-Guizhou Plateau. This suggests independent evolutionary trajectories in separate refugia during the Last Glacial Maximum [43]. Similarly, a study on the conifer Taiwania cryptomerioides identified at least four glacial refugia: in the Wuyi Mountains, the Nanling Mountains, the Yungui Plateau, and a larger contiguous refugium in the border region of Vietnam and Yunnan. The study revealed significant genetic divergence among these refugial populations and detailed postglacial expansion routes into surrounding mountain ranges [56]. Thus, the climatic fluctuations of the Quaternary glacial-interglacial cycles likely further triggered population-level diversification in tropical and subtropical regions. The widespread small fruits of Elaeocarpus resulted from a drastic transition from a zonal distribution arid climate to a warm and wet climate in central and eastern China since the early Miocene [25]. In addition, our study indicates that the diversification rates of Elaeocarpus in China were estimated to have been highest at 4–20 Ma during the Miocene, which is consistent with the findings reported for Theaceae [81] and evergreen oak [34].

In addition, the dispersal of fruits by frugivorous birds and hominid apes with rapid radiation evolution that occurred in the early Miocene can quickly expand the distribution range of Elaeocarpus in China [5, 18, 50, 51]. Furthermore, species with small, shiny and ellipsoid drupes, such as Elaeocarpus japonicus and E. bachmaensis, in the terminus of the phylogeny tree evolved from species with large and subglobose drupes. The alteration of the morphology of the drupes and the zoochory mode of dispersal has favored the dispersal of viable propagules across water barriers from China to Japan [16]. Therefore, we infer that the possible migration route of Elaeocarpus in China is from the Indochina Peninsula to China and Japan via birds or some mammals.

Authors’ contributions

Yihui Wang: Formal analysis (Lead), Data curation (Lead), Writing original draft (Lead), Methodology (Lead), Resources (Equal), Software (Lead), Visualization (Lead). Yifei Xie: Conceptualization (Lead), Project administration (Lead), Writing - review & editing (Lead), Funding acquisition (Lead). Jiayi Jin: Methodology (Equal). Jinyue Li: Software (Equal). Xueping Lai: Software (Equal), Data curation (Equal). Xiangdong Qiu: Resources (Supporting). Yang Tong: Resources (Supporting). Zhixiang Zhang: Resources (Supporting), Funding acquisition (Equal).

Funding

This work was supported by the National Natural Science Foundation of China (grant numbers 31110103911, J1310002) and the Science and Technology Project in Jiangxi Province Department of Education (190781)

Data availability

The original contributions presented in the study are included in the article/supplementary material, and the GeneBank accession numbers of each sequence are shown in Table S1. Further inquiries can be directed to the corresponding authors.

Declarations

Ethics approval and consent to participate

This study is plant-based material, does not involve animal or human testing, and does not require ethical review.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.