Generic status of Winitia (Annonaceae, Miliuseae) reaffirmed by molecular phylogenetic analysis, including a new species and a new combination from Thailand

1,3,4 Herbarium, Division of Plant Science and Technology, Faculty of Science, Chiang Mai University, 239 Huay Kaew Rd., Chiang Mai 50200, Thailand. 1 Research Center in Bioresources for Agriculture, Industry, and Medicine, Chiang Mai University, 239 Huay Kaew Rd., Chiang Mai 50200, Thailand. 2 Sichon, Nakhon Si Thammarat 80120, Thailand. 5 Department of Botany and Microbiology, Ohio Wesleyan University, Delaware, Ohio 43015, USA.


Introduction
Annonaceae Juss. are a large pantropical family of flowering plants prominent in lowland rainforests and consisting of ca 2430 species (Couvreur et al. 2019) in 110 genera (Guo et al. 2017b;Chaowasku et al. 2018aChaowasku et al. , 2018bXue et al. 2018). Infrafamilial classifications and generic realignments of the family have been stabilized lately with the help of molecular phylogenetics (e.g., Zhou et al. 2009Zhou et al. , 2010Chatrou et al. 2012;Chaowasku et al. 2012;Xue et al. 2012;Guo et al. 2017aGuo et al. , 2017b. Additionally, in combination with thorough morphological examinations, a number of previously undescribed genera have been established in the past 12 years (Mols et al. 2008;Couvreur et al. 2009Couvreur et al. , 2015Chaowasku et al. 2012Chaowasku et al. , 2013Chaowasku et al. , 2015Chaowasku et al. , 2018aChaowasku et al. , 2018bGuo et al. 2014;Xue et al. 2018). One of these is the genus Winitia Chaowasku (Chaowasku et al. 2013), which is closely allied to the monotypic genus Stelechocarpus Hook.f. & Thomson (Hooker & Thomson 1855) and the genus Sageraea Dalzell (Dalzell 1851) composed of nine species (van Heusden 1997). Despite a rather poor support for the sister relationship of Winitia and Stelechocarpus (Chaowasku et al. 2013), the former was reduced into synonymy of the latter by Turner (2016).
Winitia can be recognized by a suite of morphological traits, e.g., monoecious habit; conspicuously thick, fleshy, and ± rose-colored petals; multicolumellar stigmas, i.e., each stigma with multiple columnar lobes; rather massive, ± blackish brown, and multi-seeded monocarps as well as several palynological features (Chaowasku et al. 2013). The genus is distributed from southern Thailand and southern Vietnam through the Malay Peninsula to Sumatra and Borneo, and currently contains two species: the widespread W. cauliflora (Scheff.) Chaowasku (Chaowasku et al. 2013;basionym: Scheffer 1881) and the type species W. expansa Chaowasku (Chaowasku et al. 2013) endemic to Thailand; so far, only a single accession of each species has been included in phylogenetic analyses (Chaowasku et al. 2013: accession of W. cauliflora was from cultivation in Bogor Botanical Garden, Indonesia).
In Gardner et al. (2015: 174) there are photographs of Winitia sp., which occurs in Surat Thani Province of Thailand and superficially resembles W. expansa, especially in the inner petals spreading at anthesis, but some features are different. Further, W. cauliflora occurring in Thailand has never been included in any phylogenetic inferences. The aims of the present study are, therefore, to (1) elucidate the phylogenetic position of Winitia sp. from Surat Thani Province and W. cauliflora accessions from Thailand and (2) determine their taxonomic status. The additional accessions will also provide a test of the monophyly of the genus.  [Couvreur et al. 2009] and Brieya fasciculata De Wild. [De Wildeman 1914]), which were assigned as outgroups, and Annickieae Couvreur. The Malmeoideae minus Piptostigmateae and Annickieae have been previously shown to constitute a strongly supported clade (e.g., Guo et al. 2017b 3 (matK, ndhF, rbcL, and ycf1), one intron (trnL), and two intergenic spacers (psbA-trnH, trnL-trnF). The ycf1 sequences of the outgroups plus Monocarpia euneura Miq. (Miquel 1865) were not available.

DNA extraction, amplification and sequencing
All methods used for DNA extraction, amplification, and sequencing in the present study were the same as those described in Chaowasku et al. (2018a), with the same primer sequences as used in Chaowasku et al. (2012) except for the psbA-trnH intergenic spacer, which used Fw-GTTATGCATGAACGTAATGCTC (Sang et al. 1997) and Rv-CGCGCATGGTGGATTCACAATCC (Tate & Simpson 2003).

Phylogenetic analyses
Sequences were edited using the Staden package (Staden et al. 2000; available from http://staden.sourceforge.net/) and subsequently aligned by Multiple Sequence Comparison by Log-Expectation (MUSCLE;Edgar 2004) in MEGA7 (Kumar et al. 2016). The alignments were then manually optimized on the basis of homology assessment using the similarity criterion (Simmons 2004). A total of 7026 nucleotide plus nine non-autapomorphic indel characters were included. Indel coding followed the simple method of Simmons & Ochoterena (2000). An inversion of 15 continuous nucleotides in the psbA-trnH intergenic spacer was observed in some sequences and changed to its reverse complement to be homologically alignable to the remaining sequences, following Pirie et al. (2006). Parsimony analysis was carried out in TNT ver. 1.5 (Goloboff & Catalano 2016). All characters were equally weighted and unordered. Incongruence among regions was evaluated by individually analyzing each region to see if there was any significant conflict in clade support (e.g., Wiens 1998). Multiple most parsimonious trees were produced by a heuristic search of the combined data, with 9000 replicates of random sequence addition, saving 10 trees per replicate, and using the tree bisection and reconnection (TBR) branch-swapping algorithm. Clade support was measured by symmetric resampling (SR; Goloboff et al. 2003). A default change probability was used. One hundred thousand replicates were run, each with four replicates of random sequence addition, saving four trees per replicate. A clade with SR ≥ 85%, 70-84% or 50-69% was regarded as strongly, moderately or weakly supported, respectively.
Maximum likelihood analysis was accomplished in IQ-TREE ver. 1.6.10 (Nguyen et al. 2015) using partition models (Chernomor et al. 2016) employed under the "-spp" command, whereas Bayesian Markov chain Monte Carlo (MCMC; Yang & Rannala 1997) phylogenetic analysis was implemented in MrBayes ver. 3.2.6 (Ronquist et al. 2012). Both analytical methods were performed via the CIPRES Science Gateway ver. 3.3 (Miller et al. 2010). The data matrix was divided into seven partitions based on DNA region identity (the trnL intron and the adjacent trnL-trnF intergenic spacer were united as a single partition) plus a binary indel-coded partition. The most suitable model of sequence evolution for each DNA partition was chosen by Akaike Information Criterion (AIC) scores, using FindModel (http://www.hiv.lanl.gov/content/sequence/findmodel/findmodel.html) (Posada & Crandall 1998). The General Time Reversible (GTR; Tavaré 1986) nucleotide substitution model with a gamma distribution for among-site rate variation was chosen for five partitions (matK, ndhF, rbcL, trnLF [= trnL intron + trnL-trnF intergenic spacer] and ycf1) and the Hasegawa-Kishino-Yano (HKY; Hasegawa et al. 1985) substitution model with a gamma distribution for among-site rate variation was chosen for the remaining partition (psbA-trnH).
In the maximum likelihood analysis, the model "JC2+FQ+ASC" was selected by corrected AIC scores for the binary indel partition. Clade support was evaluated by non-parametric bootstrap resampling method (BS; Felsenstein 1985) with 2000 replicates. A clade with BS ≥ 85%, 70-84% or 50-69% was regarded as strongly, moderately or weakly supported, respectively. In the Bayesian analysis, the "coding=variable" setting was chosen for the binary indel partition, which was employed with a simple F81-like model without a gamma distribution for among-site rate variation. Three independent analyses, each using four MCMC chains, were simultaneously run; each run was set for 10 million generations. The European Journal of Taxonomy 659: 1-23 (2020) 4 default prior settings were used except for the prior parameter of rate multiplier ("ratepr" [=variable]). The temperature parameter was set to 0.08. Trees and all parameter values were sampled every 1000 th generation. Convergence was assessed by checking the standard deviation of split frequencies of the runs with values < 0.01 interpreted as indicating a good convergence and by checking for adequate effective sample sizes (ESS > 200) using Tracer ver. 1.6 (Rambaut et al. 2013). The first 25% of all trees sampled were removed as burn-in, and the 50% majority-rule consensus tree was constructed from the remaining trees. A clade with posterior probabilities (PP) ≥ 0.95, 0.9-0.94 or 0.5-0.89 was regarded as strongly supported, weakly supported or unsupported, respectively.

Morphology
The indumentum terminology used followed Hewson (1988). When there was a single observation or measurement, the word 'circa' (ca) was added before. Appendix 2 indicates a list of ten specimens morphologically studied.
Upon a closer observation at part of the multiple sequence alignment of the trnL-trnF intergenic spacer, it is clear that Winitia and Stelechocarpus differ from each other by having dissimilar indel structures: absence of a gap in Winitia vs presence of a gap in Stelechocarpus (Fig. 2). This eight-base-pair indel is one of the nine indel-coded characters included in the binary indel partition.   Fig. 3A); hence, for now we still consider it as a junior synonym of W. cauliflora until more evidence is obtained.
On the basis of personal observations by the second author, W. longipes is generally found near streams. During monsoon seasons, the areas where this species occurs are shortly flooded, and it is possible that the ripe monocarps fallen on the ground are taken away by the water current. Sometimes the seeds are found germinated despite still being partially covered by the pericarp. Winitia longipes is widespread in Nakhon Si Thammarat and neighboring provinces where several protected areas are located, but many individuals do occur outside the protected areas; these unprotected forests diminish nearly every single day due mainly to agricultural expansion (pers. obs.). Therefore, we recommend that its conservation status based on IUCN (2012) be assessed as "Near Threatened (NT)".

Diagnosis
Morphologically similar to Winitia expansa, differs by having a non-glaucous appearance on the petal adaxial surface (glaucous in W. expansa) and shallow pits on the adaxial surface of the inner petals (no pitted structures in W. expansa). Moreover, its inflorescences are mostly clustered on large roots and at the unswollen base of trunks, whereas the inflorescences of W. expansa are mostly clustered at the swollen base of trunks.

Etymology
Named after Thailand, a country where this species is endemic.

Stamens
Tightly packed, not bending inward More loosely packed, slightly bending inward (best observable in outer-whorl ones)

Distribution, habitat, and phenology
Surat Thani Province, Phanom District (Fig. 6); occurs in evergreen forests among rugged limestone outcrops; at an elevation of ca 400 m; flowering material collected in May and June.

Field notes
Flowers orange-pink.

Notes
The new species is so far known only from Phanom District. We observed only three individuals, which occur among limestone outcrops beside a trail surrounded by a rubber tree plantation. In Khlong Phanom National Park where this species has also been reported (Gardner et al. 2015) less than ten individuals were found (pers. comm., S. Gardner). Based on our observations and estimations, we believe at least the category "endangered: EN D" (IUCN 2012) is applicable.

Discussion
The monophyly of Winitia, when more accessions have been added, remains maximally supported ( Fig. 1), as compared with previous studies that included only two accessions (Chaowasku et al. 2013(Chaowasku et al. , 2018bGuo et al. 2017b). The genus, however, was subsumed under Stelechocarpus by Turner (2016), principally based on the arguments that (1) the erection of Winitia has caused Stelechocarpus to become a monotypic genus, which conveys little taxonomic information, and (2) Stelechocarpus inclusive of Winitia is identifiable by a raised leaf midrib adaxially. However, there are also clear morphological differences, including petal color and texture, floral dimorphism (absent vs present), locations of male and female flowers in individuals (mixed vs separate), and stamen and stigma morphology that warrants the recognition of two separate genera (Chaowasku et al. 2013). These differences are comparable to those of two sister monotypic genera, Mwasumbia Couvreur & D.M.Johnson (Couvreur et al. 2009) and Sirdavidia Couvreur & Sauquet (Couvreur et al. 2015), belonging to the African tribe Piptostigmateae of subfamily Malmeoideae and primarily differing from each other in petal color and configuration as well as stamen morphology. Furthermore, support for the sister relationship of Mwasumbia and Sirdavidia was already strong with only three plastid regions (rbcL exon, trnL intron, and trnL-trnF intergenic spacer; Couvreur et al. 2015) included, whereas support for the sister relationship of Winitia and Stelechocarpus is still rather poor and the branch uniting the two genera is very short (Fig. 1) despite using seven plastid regions and adding more accessions. In addition, the branches leading to Winitia and to Stelechocarpus are relatively long (Fig. 1) and comparable to those leading to Monoon Miq. (Miquel 1865) and to Neo-uvaria Airy Shaw (Airy Shaw 1939). Given the same amounts of nucleotide data, the support for the sister relationship of Monoon and Neo-uvaria is even much higher with a longer branch uniting them (Fig. 1).
It is worth noticing that there is an indel of eight continuous base pairs in the trnL-trnF intergeneric spacer potentially diagnostic for generic discrimination in the tribe Miliuseae (Fig. 2), i.e., all members in the same genus possess the same indel structure (absence or presence of an eight-base-pair gap; personal observations based on all GenBank accessions of the trnL-trnF intergeneric spacer indicated in Chaowasku et al. 2014Chaowasku et al. , 2018aChaowasku et al. , 2018bGuo et al. 2017b). This gap is present in Stelechocarpus and absent in Winitia (Fig. 2). Besides Miliuseae, members of other tribes of Malmeoideae and other subfamilies also possess this indel (pers. obs.), but its utility for generic discrimination is still unknown, awaiting detailed observations.
Regarding pollen morphology, the two genera do not possess the same infratectum type, i.e., ± columellate/coarsely granular in Winitia vs finely and densely granular in Stelechocarpus (Chaowasku et al. 2013); in Miliuseae, there have been no reports of different infratectal structures occurring in the same genus (Chaowasku et al. 2014). There are also other palynological traits (exine sculpturing, tectum, and basal layer) that are significantly different in the two genera (Chaowasku et al. 2013).
In recognizing the genus Winitia, we believe that the most important issue to consider is the support for monophyly, then the morphological distinctions as well as relative phylogenetic branch lengths.
On the basis of the above-mentioned morphological and phylogenetic evidence reaffirming a clear-cut differentiation of Winitia and Stelechocarpus, we are therefore strongly convinced that the recognition of the genus Winitia is appropriate. Rather equivalent to the case of Winitia and Stelechocarpus, in Annonaceae acceptance of a non-monotypic genus being the sister group of a monotypic genus is not unprecedented, for example, Fusaea (Baill.) Saff. (Safford 1914; basionym: Baillon 1868) with a monotypic Duckeanthus R.E.Fr. (Fries 1934) as the sister group, Letestudoxa Pellegr. (Pellegrin 1920) with a monotypic Pseudartabotrys Pellegr. (Pellegrin 1920) as the sister group, and Mitrella Miq. (Miquel 1865) with a monotypic Pyramidanthe Miq. (Miquel 1865) as the sister group (see Guo et al. 2017b). Appendix 1 (continued). Voucher specimens for molecular phylogenetic analyses with GenBank accession numbers indicated. -signifies unavailable sequences; * signifies sequences newly produced in this study. Appendix 1 (continued). Voucher specimens for molecular phylogenetic analyses with GenBank accession numbers indicated. -signifies unavailable sequences; * signifies sequences newly produced in this study.