Molecular Phylogenetic Evidence and Biogeographic History of Indian Endemic Portulaca L. (Portulacaceae) Species

: The genus Portulaca L. belongs to the monogeneric family Portulacaceae and consists of about 157 species worldwide. In India, it is represented by 11 taxa; among them, Portulaca badamica , Portulaca lakshminarasimhaniana , Portulaca oleracea var. linearifolia , and Portulaca laljii are endemic. So far, the phylogenetic positions of these species have not yet been analyzed. We have reconstructed the Bayesian and maximum likelihood phylogenies based on a combined chloroplast and nuclear DNA sequence dataset to reveal phylogenetic placements of Indian Portulaca . Phylogenetic analyses indicate that all the sampled Indian Portulaca species (except Portulaca wightiana ) are placed in the AL clade, which contains most of the known species of the family Portulacaceae. We used reconstructed phylogeny to study the historical biogeography of Indian endemic species by employing S-DIVA analysis. S-DIVA analysis suggested P. lakshminarasimhaniana has origin in India, it may be the result of in situ speciation in India, and P. badamica was dispersed from Africa to India. We have also discussed the systematic placements of endemic species and their morphological relationships with closely allied species. In addition, this study also provides taxonomic treatment for endemic species.


Introduction
The genus Portulaca L. includes about 157 taxa, which are annual and perennial species [1][2][3][4][5]. Portulaca species are distributed worldwide in the tropical and subtropical areas with centers of diversity in South America and Africa [1]. Ocampo and Columbus [1] performed comprehensive phylogeny of the genus Portulaca; in which they discussed the phylogenetic analysis and historical biogeography of the genus. Their phylogenetic analyses reveal two major lineages: the OL clade (opposite-leaved species) and the AL clade (an alternate to subopposite-leaved species). Species from the OL clade are mainly distributed in Africa, Asia, and Australia, whereas species from the AL clade originated in the New World [1]. AL clade comprises three strongly supported groups: the Oleracea, Umbraticola, and Pilosa clades.
Sivarajan [6] made the first attempt to revise the genus Portulaca from India and recognized four distinct species and three infra-specific taxa. He treated Portulaca grandiflora Hook. as a subspecies of Portulaca pilosa L. and Portulaca tuberosa Roxb. was considered a variety of P. pilosa subsp. pilosa. Singh and Sanjappa [7] reported six species of Portulaca Diversity 2022, 14, 443 2 of 16 from India, viz. P. grandiflora Hook., P. oleracea L., P. pilosa, P. quadrifida L., P. tuberosa, and P. wightiana Wall. ex Wight and Arn. and one variety, i.e., P. oleracea var. linearifolia Sivarajan and Manilal. Dalavi et al. [3] described two new species from India: Portulaca badamica S.R.Yadav and Dalavi and Portulaca lakshminarasimhaniana S.R.Yadav and Dalavi. Recently, Sivaramakrishna and Yungandhar [8] described Portulaca laljii from the Eastern Ghats of India. Portulaca umbraticola Kunth. is a highly polymorphic species cultivated throughout India for ornamental foliage and flowers [9]. It is differentiated from the rest of the Indian species by means of winged pyxis [9]. Therefore, up to date, there are total of 11 taxa of Portulaca reported from India. Among these, P. badamica, P. lakshminarasimhaniana, P. oleracea var. linearifolia, and P. laljii are endemic to India, and their phylogenetic positions have yet not been assessed. Details of species found in India, their geographical distribution, and their representation in molecular studies are listed in Table 1. Ocampo and Columbus [1] provided comprehensive phylogenetic analyses based on a combined dataset of three plastid regions (ndhA intron, ndhF, and trnT-psbD) and the nuclear ITS region. They studied phylogeny based on the historical biogeography of the genus and hypothesized that the ancestral distribution area of Portulaca included southern hemisphere continents and Asia. preparation of all the species of Portulaca was performed by the standard protocol adopted from Jain and Rao [12]. To confirm the identity and distribution of every species of Portulaca from India we explored all the available literature such as Dalavi et al. [3,4], Ocampo [2], and Sivarajan [6]. Details of collected plant species are provided in Table 2. Total genomic DNA was extracted from young and fresh leaves of collected plant species using the modified CTAB method [13] with some modifications mentioned in Tamboli et al. [14]. We amplified and sequenced the ndhA, ndhF, trnT-psbD, and ITS region from them. We constructed a dataset of 63 Portulaca species in 11 infraspecific taxa, which covers sampling from all its nine major distribution areas and is used for further phylogenetic and biogeographic analyses. We created a dataset of 89 accessions of Portulaca, which contains sequencing data of 80 accessions of Portulaca taken from previous studies [1,10,11] and sequencing data of 9 Indian Portulaca accessions generated in this study ( Table 2). Single representative species from three families, namely Talinopsis frutescens A. Gray (Anacampserotaceae), Pereskia aculeata Mill. (Cactaceae), and Talinum paniculatum (Jacq.) Gaertn. (Talinaceae) selected as an outgroup as Anacampserotaceae, Cactaceae, and Talinaceae are the closest relatives to Portulacaceae [1]. Details of the DNA sequence retrieved from GenBank are provided in Supplementary Material S1.

PCR and DNA Sequencing
A list of primers and PCR conditions is listed in Table 3. In this study, we generated 26 sequences out of that 10 were amplified and sequenced at the Department of Biochemistry, Shivaji University, Kolhapur, India. The other 16 were sequenced at Macrogen Corporation (Seoul, Korea). The procedure of PCR amplification and DNA sequencing was done at the Department of Biochemistry, Shivaji University, India, as follows: PCR reactions were performed in 25 µL of medium containing 12.5 µL AmpliTaq Gold ® 360 Master mix (Applied Biosystem, Foster City, CA, USA), 8.5 µL nuclease-free water, 0.5 µL GC enhancer (Applied Biosystem, Foster City, CA, USA), 0.5 µL each forward and reverse primer (15 µM), and 2.5 µL of DNA. PCR reactions were carried out in a thermal cycler (BIO-RAD Laboratories, Hercules, CA, USA). Amplified PCR products were then purified using the GenEluteTM PCR Clean-Up kit (SIGMA-ALDRICH, St. Louis, MO, USA). Sequencing of purified PCR products was done using BigDye ® Terminator v 3.1 cycle sequencing kit (ThermosFisher Scientific, Waltham, MA, USA) on 3500 Genetic Analyzer (Applied Biosystems, Foster City, USA). The protocol for PCR amplifications carried out at Plant Systematics Laboratory, Department of Biology, Kyungpook National University, Daegu, South Korea was as follows: PCR reactions were performed in 30 µL medium containing 15 µL nuclease-free water, 10 µL EmeraldAmp GT PCR Master Mix (TaKaRa, Kusatsu, Shiga, Japan), and 1 µL each forward and reverse primer (10 µM). PCR reactions were carried out in a thermal cycler (TaKaRa, Kusatsu, Shiga, Japan). Amplified PCR product was further purified and sequenced at Macrogen Corporation (Seoul, South Korea). Sequences of ndhA intron, ndhF, trnT-psbD, and ITS region were submitted to the GenBank NCBI database. GenBank accession numbers are provided in Table 2.

Sequence Alignment and Phylogenetic Analyses
Output files containing raw DNA sequences generated after the sequencing run were checked and edited using Sequencher v. 5.1 (Gene Codes Corporation, Ann Arbor, MI, USA) [15]. Multiple sequence alignment of each gene was carried out using MUSCLE [16] implemented in MEGA 7 [17] and further refined using GBlocks v. 0.91b (Information Génomique et Structurale, France) [18]. For checking incongruence among four sequence datasets, we performed the incongruence length difference (ILD) test [19] in PAUP 4.0a152 [20] as well as assessed the combinability of the different four loci by constructing phylogeny based on each gene and comparing topologies. The ILD test was performed with 1000 heuristic searches, by keeping simple taxon addition and TBR as branch swapping algorithm. Results of individual analyses revealed no significant topological conflicts among four different markers (Figures S1-S4) and the ILD test also revealed no significant incongruence between the chloroplast and nrITS dataset (p-value = 0.0100); thus, all subsequent analyses were performed on a combined data set (Supplementary Material S2). The ndhA, ndhF, trnT-psbD, and ITS sequence datasets were concatenated into a single matrix (Supplementary Material S2) using SequenceMatrix [21]. jModelTest 2 [22] on XSEDE on the CIPRES science gateway (http://www.phylo.org/ (accessed on 23 March 2022)) [23] was used to find out the best-fit nucleotide substitution model for the combined dataset under the Akaike Information Criterion (AIC). The best-fit nucleotide model for the combined dataset was GTR + I + G.
The Bayesian and Maximum likelihood methods were performed for phylogenetic analyses. Bayesian phylogenetic (BI) analysis was performed using MrBayes v.3.2.6 [24] on XSEDE via the CIPRES portal [23]. Markov chain Monte Carlo (MCMC) was conducted using the best-fit nucleotide model and consisted of four independent runs of 50,000,000 generations each with one cold chain and three heated chains with a tree sampled every 1000 generations. The first 10% trees were discarded as burn-in, and the remaining were used to obtain a 50% majority-rule consensus Bayesian tree having clades supported by posterior probabilities (PPs). At the end of the run, the average standard deviation of split frequencies was below 0.0100. Tracer 1.7.1 [25] was used to check effective sample size (ESS); the ESS presented values were well above 400 for all statistics, which ensured enough selected settings for the sampling [26]. Maximum likelihood analysis (ML) was performed using RaxML-HPC v.8.0 on XSEDE [27] via the CIPRES portal [23] using the GTR + G model with 1000 rapid bootstrap replicates.

Biogeographic Analysis
The distribution range of Portulaca was divided into nine areas: (A) South America, (B) North America, (C) Caribbean, (D) Central America, (E) Africa, (F) Asia, (G) Australia, (H) Islands of Pacific Ocean, and (I) Europe. We used S-DIVA analysis implemented in RASP v 3.2 (http://mnh.scu.edu.cn/soft/blog/RASP (accessed on 2 May 2022)) [28] to reconstruct the possible ancestral ranges of the genus Portulaca on the Bayesian phylogenetic trees. To account for uncertainties in phylogeny, we used 811 Binary trees and ran S-DIVA on all of them. The maximum number of areas at each node was kept as 4. S-DIVA analyses were performed on the Bayesian tree with all compatible groups ( Figure S5) as it does not accept polytomies in the target tree. The possible ancestral ranges at each node on a selected tree were obtained. The RASP program output of S-DIVA analysis for important nodes discussed in this study are provided in Supplementary Material S3.

Historical Biogeography Reconstructions S-DIVA Analysis
The reconstruction is shown in a maximal S-DIVA value of 5058.717 ( Figure 2). Node 180 indicates there are three possibilities; South America (A), Asia (F), and Australia (G) may have the place of origin for comment ancestor of the OL and AL clades ( Figure 2). The Indian endemic species P. oleracea var. linearifolia, which is placed in the Oleracea clade, was dispersed from either North America (B) or Africa (E) to India (Asia (F)) (evidence from Node 114). Recently described Indian endemic species, P. lakshaminarsimhaniana has its origin in India (Asia) (evidence from Node 162) and Node 136 suggests a dispersal event with 82.50% probability, which indicates that the dispersal of P. badamica from Africa (E) to India (Asia (F)) ( Figure 2).

Discussion
Previous studies on Portulaca demonstrated phylogenetic analysis based on a combined data matrix that retrieved highly supported phylogeny, which can be used as a framework to show phylogenetic placements of species [1,10,11]. Our reconstructed BI and ML phylogenies resulted in the recovery of all major clades observed in previous phylogenies [1,10,11]. Those are the OL clade and AL clade with four subclades: the Cryptopetala clade, Oleracea clade, Umbraticola, and Pilosa clade with substantial Bayesian PP value Maximum likelihood bootstrap support (Figure 1).

Systematic Positions and Morphological Relationships of Indian Endemic Species with Their Closely Related Species
Out of 11 species of Indian Portulaca, P. badamica, P. lakshminarsimhaniana, P. oleracea var. linearifolia, and P. laljii are endemic to India. P. badamica and P. lakshminarsimhaniana were placed in the Pilosa clade ( Figure 1). Species belonging to this clade have alternate leaves with conspicuous hairs in the axils, and flowers are generally arranged in heads; in some cases, the capitulum is reduced to very few flowers [1]. Portulaca badamica (JVD-1261) shows a sister relationship to P. massaica and P. foliosa with Bayesian PP value = 1 and ML BS = 98 ( Figure 1). Portulaca badamica is morphologically closer to P. pilosa and P. oleracea var. linearifolia [3]. Portulaca badamica differs from P. pilosa by its annual, slender, sparsely branched, erect, habit (compared to perennial, robust, much-branched, spreading habit), usually cleistogamous yellow flowers (compared to chasmogamous, pink flowers), 3-4-fid style (compared to 5-7-fid style), 8-12 stamens (compared to 20-25 stamens), and bluish seeds with stellulate flat cells without central elevations (compared to bluish seeds with stellulate flat cells with central elevations) [3]. It is also similar to Portulaca oleracea var. linearifolia but differs in having hairs in axils of leaves (compared to completely glabrous habit), 1-6-flowered capitulum (compared to 1-10-flowered capitulum), 3-4-fid style (compared to 5 fid style), very delicate, transparent sepals (compared to thick green sepals covering capsules), and bluish seeds having stellulate flat cells without central elevations (compared to black seeds with circular central elevations in cells) [3].
P. lakshminarsimhaniana (JVD-1260) is grouped with our added Indian accession of P. pliosa (JVD-1259) and P. grandiflora (JVD-1264) with PP = 1 and ML BS = 68 ( Figure 1). P. lakshminarsimhaniana is morphologically closer to P. suffrutescens [3], and both of these species are placed in the Pilosa clade ( Figure 1). Portulaca lakshminarasimhaniana differs from P. suffrutescens by its non-tuberous root (compared to tuberous root), stem scabrous with scaly bark (compared to a smooth stem without scaly bark), dark pink to red flowers (compared to magenta to orange flowers), 8-15 stamens (compared to more than 40 stamens), strictly 4 fid style (compared to 5-6-fid style) and oblong capsules (compared to sub-globose capsules) [3]. P. oleracea var. linearifolia (JVD-1263) was placed in the Oleracea clade and grouped with the other members of the Olereacea group (Figure 1). Unfortunately, we could not sample recently described P. laljii in this study, which is flagged for future molecular studies. P. laljii is closely related to P. lakshminarsimhaniana but differs in its tuberous root (compared to non-tuberous root), stem herbaceous smooth without scaly bark (compared to woody, scabrous with scaly bark), pinkish-orange flowers with the deep pinkish-orange throat (compared to pink to red flowers with dark red throat), 6-7 stamens (compared to 8-15 stamens), strictly 3 fid style (compared to 4 fid style), and prolate capsule (compared to oblong capsule) [8]. If the phylogenetic placement of P. laljii is studied in the future, we hypothesize that it would fall in the Pilosa clade, based on morphological similarities with P. lakshminarsimhaniana.

Biogeographic History of Indian Endemic Species
Phylogeny-based historical biogeographical reconstructions are now essential to illuminate the organism's evolutionary history [29]. A previous historical biogeography of the genus was studied using the S-DIVA method [1] and in this study, we used that method to determine the historical biogeography of Indian endemic species. We have enriched Diversity 2022, 14, 443 9 of 16 the sampling of Ocampo and Columbus [1] by the addition of nine Indian Portulaca taxa (among which three are endemic to the region) and recently described two new species, namely P. juliomartinezii G. Ocampo [10] and P. almeviae G. Ocampo [11]. Phylogeny-based historical biogeographic reconstruction of Ocampo and Columbus [1] proposed three of the four possibilities in the southern hemisphere (Africa, Asia, Australia, and South America) as the place of origin of Portulaca. Our S-DIVA analysis also revealed that South America (A), Asia (F), and Australia (G) might have the place of origin for the genus (Figure 2). The place of origin of Portulaca is uncertain, and several hypotheses about an ancestral area of the genus were previously proposed [30][31][32]. The greatest biodiversity of Portulaca species is in the southern hemisphere, and it would be originated there [30,32]. This is the possible reason behind having the origin of the genus in any of the continents of the southern hemisphere.
S-DIVA suggested two possibilities for the dispersal of P. oleracea var. linearifolia, from either North America (B) or Africa (E) to India. The molecular phylogeny of Ocampo and Columbus [1] and phylogeny presented in this study show that P. oleracea is not monophyletic and subspecies of oleracea is now distributed worldwide as a weed. A previous study [1] and this study show that there are a number of subspecific species of P. oleracea from the Oleracea clade inferred with the highest probability to have originated in North America (Node 111, Figure 2). The addition of more sampling from the P. oleracea clade complex will be helpful to gain more knowledge about the geographic origin, although it is very difficult to estimate ancestral area for taxa having widespread distribution as shown in other groups within Cactineae [33,34]. P. badamica was dispersed from Africa to India (evident from Node 136, Figure 2). It could be very well that there were African populations of Portulaca that were the source of dispersal to India. Although, there is no obvious mechanism of species dispersal to such great distances. Ridley [35] suggested birds can disperse the seeds up to a long distance, water also can act as a dispersal agent; including intercontinental (by sea) and intracontinental dispersal, it is demonstrated that seeds can remain viable after floating for a number of weeks in sea water [35,36]. Another dispersal agent is tropical storms (a wind dispersal mechanism), it has been proposed to distribute seeds from the Caribbean to the southeastern USA [37]. Humans are also effective dispersal agents [35] and have widely propagated species (intentionally or inadvertently), especially in urban areas, agricultural fields, along roads, and at ports. P. lakshminarasimhaniana is a sister to the widespread cosmopolitan P. pilosa species and it may have originated in India. It could very well be that P. lakshminarasimhaniana diverged from the P. Pilosa Indian populations within India, i.e., the result of in situ speciation in India. There are more population sampling and population genetic approaches that are necessary to investigate this further. Morphological description: seasonal, slightly branched, erect, slender herbs, 5-15 cm in height with fibrous roots. Stem: erect, slender, sparsely pilose in earlier stages, glabrous when old. Leaves: 0.8-1.5 × 0.1-0.3 cm, simple, alternate, linear terete, subsessile, glaucous, veins not distinct; petiole with a small tuft of hairs at nodal portion. Flowers: 1-6, sessile in terminal capitulum surrounded by 4-7 involucral bracts and a ring of white hairs. Flowers 1 cm across, usually cleistogamous, rarely chasmogamous, sessile, yellow. Sepals: 2, 4-6 × 2-3 mm, ovate, connate at base, membranous, 4-6-veined, pale yellow in color, glabrous. Petals: 4-5, 5-6 × 2-3 mm, connate at the base, ovate to obovate, glabrous, entire, pale yellow to bright yellow. Stamens: 8-12, free, filaments connate at the base forming ring and adnate to petals, unequal in length; anthers 0.7-0.9 mm, bithecous, yellowish in color, dehiscing via longitudinal slits; filaments 1.0-2.5 mm long, glabrous. Gynoecium: 4-5 carpellary, syncarpous; ovary 2-3 × 1-2 mm, sub-globose, lower portion sunk into the base of the calyx tube, glabrous. Style: 3-4 mm long, glabrous, 3-4 fid, pale yellow; stigma 1-2 mm, papillate, yellow. Fruits: circum-scissile pyxis, 4-6 × 2-3 mm, sub-globose, basal disc, and upper operculum nearly equal in length. Seeds: 40-60 per capsule, sub-reniform, 0.6-0.7 mm in diameter, bluish black with stellulate flat cells arranged in 3-4 circular rings without central elevations.   Habitat: Plant shows a very discrete population in moist sandy plains. Flowering and fruiting: May to August. Distribution: India (Only known from Badami hills of Bagalkot District, Karnataka) [3,4]. Status: The species is not yet accessed for IUCN categorization and is kept as DD. It is endemic to south peninsular India. Distribution: India (Only known from Badami hills of Bagalkot District, Karnatak [3,4].
Status: The species is not yet accessed for IUCN categorization and is kept as DD. is endemic to south peninsular India.    Portulaca laljii P.Sivaramakrishna and P. Yugandhar.
We were unable to sample this species in this study, the full morphological description of this species is provided in Sivaramakrishna and Yugandhar [8].  Figure 5).

Conclusions
The present study confirms phylogenetic placements of Indian endemic Portulaca species, namely P. badamica, P. lakshminarsimhanina, and P. oleracea var. linearifolia. Morphological similarities and differences between endemic and closely related species were also addressed. Reconstructed phylogeny reveals that all the sampled Indian Portulaca species (except P. wightiana) are phylogenetically placed in the AL clade. This clade includes the best-known species of the family Portulacaceae and is characterized by alternating to sub-opposite leaves [1]. In addition to this, the present study provides taxonomic treatment for endemic species and a distribution map ( Figure 6). S-DIVA analysis suggested that Indian endemic species P. badamica were dispersed from Africa to India and P. lakshminarasimhaniana may be diverged from the Indian P. Pilosa populations and the result of in situ speciation in India. Supplementary Materials: The following supporting information can be downloaded at: www.mdpi.com/article/10.3390/d14060443/s1, Supplementary Material S1: Details of Portulaca accessions included in phylogenetic analyses: herbarium number and GenBank accession numbers for ITS, ndhF, trnT-psbD and ndhA intron. Supplementary Material S2: Aligned combined sequence data (ITS + ndhA + ndhF + trnT-psbD) matrix. Supplementary Material S3: RASP program results of S-DIVA analysis for important nodes discussed in this study. Figure S1: Maximum likelihood tree based on ITS data. Figure S2: Maximum likelihood tree based on ndhA data. Figure S3: Maximum Supplementary Materials: The following supporting information can be downloaded at: https:// www.mdpi.com/article/10.3390/d14060443/s1, Supplementary Material S1: Details of Portulaca accessions included in phylogenetic analyses: herbarium number and GenBank accession numbers for ITS, ndhF, trnT-psbD and ndhA intron. Supplementary Material S2: Aligned combined sequence data (ITS + ndhA + ndhF + trnT-psbD) matrix. Supplementary Material S3: RASP program results of S-DIVA analysis for important nodes discussed in this study. Figure S1: Maximum likelihood tree based on ITS data. Figure S2: Maximum likelihood tree based on ndhA data. Figure S3: Maximum likelihood tree based on ndhF data. Figure S4: Maximum likelihood tree based on trnT-psbD data. Figure S5: Bayesian phylogenetic analysis on the combined dataset with all compatible groups. Figure  S6: Maximum likelihood tree based on combined (ITS + ndhA + ndhF + trnT-psbD) dataset with the highest log-likelihood value (−22,064.18).