Tempo and Mode of Floristic Exchanges between Hainan Island and Mainland Asia: A Case Study of the Persea Group (Lauraceae)

: The origin of island biodiversity has been a topic of interest in biogeography. Two main hypotheses were suggested to explain the ﬂoristic origin of Hainan Island: ancient vicariance vs. recent dispersal. The Persea group of Lauraceae was used to examine their origin on Hainan Island. A data matrix including ﬁve genera and 49 species was assembled, phylogenetic trees of the Persea group were reconstructed using both maximum likelihood and Bayesian inference methods, and a relaxed molecular clock in BEAST was used to estimate the divergence time of the Persea group. The results showed that the Persea group diverged from its sister clade in the Eocene, and that the endemic and common species of Hainan Island and mainland Asia originated relatively late during the Miocene–Pleistocene. Species of the Persea group on Hainan may have arrived from mainland Asia via dispersal or originated via dispersal–isolation–divergence during the Miocene and Pleistocene. The results favor the dispersal hypothesis for the origin of the ﬂora of Hainan Island and negate the vicariance hypothesis.


Introduction
Hainan Island is the largest tropical island in China. It is located south of Guangdong and Guangxi across the Qiongzhou Strait, east of Vietnam across the Beibu Gulf, and north of the Philippines, Malaysia, and Brunei [1]. The island is mountainous in the middle and slopes to the coast in all directions. It has a characteristic tropical monsoon climate (annual mean temperature: 22-26 • C), adequate light (annual mean hours: 2166 h), and abundant rain (annual mean precipitation: 2000-2400 mm) due to its geographic location [2]. The complex topography and favorable hydrothermal conditions have fundamentally shaped the rich floristic diversity [3].
The flora of Hainan has a strong tropical character [4]. Around 62.8% of the 196 families are tropical in nature. At the genus level, tropical elements account for 80.5%, while temperate elements are relatively low in proportion [3]. Hainan is rich in endemic species, 592 species endemic to China are on Hainan. Most of them are in the four mountainous areas of Wuzhi Shan, Jianfeng Ling, Bawang Ling, and Diaoluo Shan [4][5][6]. Around 2078 species are common to Asia in general.
Two hypotheses were proposed to explain the tempo and mode of the floristic origin of Hainan, although all studies agree that the flora of the island is continental in origin. Ancient vicariance is one hypothesis to explain the floristic origin of Hainan since the island was once connected to Vietnam and Guangxi, China, in the Eocene, then moved and rotated to the southeast to finally reach its present position [1,3]. The other hypothesis contends that the flora of Hainan was due to recent dispersal and frequent floristic exchanges with the Asian mainland [7][8][9][10]. These hypotheses have not been tested by phylogenetic studies based on molecular dating analyses. Lauraceae are one of the early diverged lineages of angiosperms. Fossils of the family can be traced back to the mid-Cretaceous and have also been recorded in the Eocene flora of Hainan [11][12][13]. The family has fleshy fruits that are dispersed by birds, among other factors. It is mainly tropical and pantropical, and contains over 3000 species worldwide [14]. The family has the highest species diversity among characteristic tree families in Hainan [1,3] with 122 species and infraspecific taxa in 16 genera. Many species of Lauraceae are also dominant [15]. Within the family, the Persea group, with five genera, Alseodaphne, Alseodaphnopsis, Dehaasia, Machilus, and Phoebe, is well represented in the flora of Hainan. These genera contain species pairs between Hainan and continental Asia, thus providing an ideal opportunity to examine the hypotheses on the floristic origin of Hainan.
To test these hypotheses, we reconstructed a phylogeny of the Persea group based on extensive sampling of species on Hainan and mainland Asia and estimated the divergence time of the species. We expected the divergence time between Hainan and the adjoining Asian mainland to be earlier than the Eocene under the tectonic drift and vicariance hypothesis, but later than the Eocene under the recent dispersal hypothesis.

Taxon Sampling
We selected 49 species from five genera of the Persea group, with Persea excluded because it is not native to Hainan. Six species from three genera (Lindera, Litsea, Neolitsea) of the core Lauraceae were chosen as outgroups based on recent molecular studies of Lauraceae [16,17]. Two nuclear fragments, including the nuclear ribosomal internal transcribed spacer (nrITS) and the second intron of the LEAFY gene (LEAFY intron II), were selected for phylogenetic reconstruction [18]. GenBank accession numbers for nrITS and LEAFY intron II sequences are listed in Table 1.

Phylogenetic Analyses
The nrITS and LEAFY intron II sequences were aligned with the program MAFFT (Version 7.471, Tokyo, Japan) [19] and edited manually using BioEdit (Version 7.0.9.0, Wooster, OH, USA) [20]. Ambiguously aligned fragments of two alignments were removed in batches using Gblocks (Version 0.91b, Barcelona, Spain) [21]. The sequences were concatenated and analyzed further. The best-fit partition model was chosen for the dataset (nrITS and LEAFY intron II) with ModelFinder [22] based on the Bayesian information criterion (BIC). Phylogenetic analyses were performed using the maximum likelihood (ML) and Bayesian inference (BI) methods. ML phylogenies were inferred using IQ-TREE (Version 1.6.8, Vienna, Austria) [23] under the edge-linked partition model for 5000 ultrafast bootstraps [24], and the Shimodaira-Hasegawa-like approximate likelihood ratio test [25]. BI phylogenies were inferred using MrBayes (Version 3.2.6, Stockholm, Sweden) [26], and the Markov chain Monte Carlo (MCMC) algorithm was run for 500,000 generations with a sampling frequency of every 500 generations. The initial 25% of sampled data were discarded as burn-in. Branch support of the BI tree was determined as Bayesian posterior probabilities (BPs).

Divergence Time Estimation
The combined dataset of the BI tree was used for molecular dating analyses with BEAST (Version 2.6.6, Auckland, New Zealand) [27]. We used BEAUti (Version 2.6.6, Auckland, New Zealand) [27] to import the dataset, set the substitution model as GTR, implemented the relaxed clock log-normal, and applied a Birth-Death Model. We ran the analysis for 40,000,000 Markov Chain Monte Carlo (MCMC) generations with a sampling frequency of every 4000 generations. TRACER (Version 1.7.2, Edinburgh, UK) [28] was used to calculate the log file's stationarity. After removing the first 10% of trees as burn-in, we generated a maximum clade credibility (MCC) tree in TreeAnnotator (Version 2.6.3, Edinburgh, UK) [28] and visualized it in FigTree (Version 1.4.2, Guangzhou, China) [29,30].
We used two macrofossils to calibrate divergence time estimates: Alseodaphne changchangensis JH Jin & JZ Li from the Eocene Changchang Formation of the Changchang Basin of Hainan [12] and Machilus maomingensis JH Jin & B Tang from the Eocene Youganwo Formation of the Maoming Basin of Guangdong, southern China [13]. We followed Li et al. [29] Forests 2022, 13, 1722 4 of 9 in using Alseodaphne changchangensis to calibrate the crown node of the Persea group [A: age 37-49 million years ago (Ma)] and applied parameters including a log-normal prior distribution with an offset of 37 Ma, a mean of 1.8, and a standard deviation of 0.35. We used Machilus maomingensis to calibrate the stem age of Machilus (B: age 33.7-33.9 Ma) following Li et al. [29] and applied parameters including a uniform prior distribution with an offset of 0, a lower of 33.7, and an upper of 33.9 (Table 2).

Sequence Characters
The numbers of variable sites and parsimony-informative (PI) sites of the nrITS dataset were 114 bp (23.1%) and 72 bp (14.5%), respectively ( Table 3). The aligned length of the LEAFY intron II was 625 bp, with 24.2% and 16.7% variable and PI sites, respectively ( Table 3). The aligned length of the combined nrITS and LEAFY intron II was 1088 bp, with 30.3% and 14.4% variable and PI sites, respectively (Table 3).

Phylogenetic Analyses
The ML tree based on the nrITS and LEAFY intron II sequences ( Figure 1) showed that the Persea group was monophyletic and divided into four clades with very high support Alseodaphnopsis constituted a small clade, in which Alseodaphnopsis hainanensis from Hainan was sister to Alseodaphnopsis putaoensis from Southeast Asia (BS: 100, PP: 1). Alseodaphnopsis rugosa from Hainan was sister to Alseodaphnopsis maguanensis from Yunnan. In the Phoebe clade, P. hungmoensis from Hainan was clustered with species from Yunnan and adjacent countries, including P. puwenensis, P. tavoyana, P. megacalyx, and P. macrocarpa, with moderate to high support (BS: 89, PP: 0.99).

Divergence Times
The stem and crown ages of the Persea group were estimated to be 48.

Discussion
Hainan was thought to be separated from the mainland (Beibu Gulf) due to Eocene plate tectonic movement, which fundamentally impacted the origin of the flora of Hainan [3,31]. Based on paleomagnetism and volcanism evidence, Zhu [31] proposed that the flora of Hainan originated via ancient vicariance in the Eocene. Our dating results suggest that species of the Persea group on Hainan originated in the Miocene and later, showing a dispersal-isolation-divergence pattern [32]. The split between mboxemphAlseodaphnopsis hainanensis and its sister species Alseodaphnopsis putaoensis represents the earliest divergence, which was estimated to have occurred around 13.5 Ma. Machilus pomifera (endemic to Hainan) diverged from its sister, M. salicoides, around 3.4 Ma. That divergence was the latest event. However, the estimated divergence time for the endemic species of Hainan from sister species was later than the geographic time of separation of Hainan Island from the mainland. We therefore hypothesize that species of the Persea group on Hainan may have been derived from mainland Asia via multiple dispersal events followed by isolation and speciation. Divergence of the species pairs between Hainan and the mainland occurred at different times in the Neogene. All of these divergence events between species pairs on Hainan and the Asian mainland occurred in the Miocene or later, which negates the ancient vicariance hypothesis but supports the dispersal hypothesis on the origin of the Hainan flora. Additionally, it appears that endemism on Hainan originated via a dispersal-isolation-divergence pattern.
Our  [37]. These studies showed that the endemic species of Hainan evolved multiple times since the early Miocene, thereby supporting the hypothesis of recent dispersal rather than ancient vicariance as the origin of the Hainan flora.
Different mechanisms may have contributed to the origin of the endemic flora of Hainan. Geological events have provided opportunities for floristic exchange between Hainan and the Asian mainland. Periodic land bridges during the Pleistocene connected Hainan and southern China, allowing for frequent exchanges [38,39]. Cycas taiwaniana of Hainan, Guangdong, and Fujian, diverged quite recently [40,41] and may have migrated between the mainland and Hainan when a land bridge was available. Plants with fleshy fruits, e.g., Lauraceae, may have entered Hainan via bird dispersal. A large number of avian fossils have been recorded from Miocene strata in Zhaotong, Yunnan Province [42]. Finally, ocean currents may have provided an additional mechanism for floristic exchange, e.g., Cocos nucifera, between Hainan and mainland Asia, including Southeast Asia and southern China.

Conclusions
The tempo and mode of floristic exchanges between Hainan and mainland Asia are complicated. We conducted phylogeny and molecular dating of the Persea group of Lauraceae to test the competing hypotheses of the origin of the flora of Hainan and concluded that the endemic species originated via multiple recent dispersal events, but not due to ancient vicariance. However, it should be acknowledged that we provided only a study of plants with easily dispersable fleshy fruits. To better understand the origin