Confirmation of species independence and affinity of Musa huangbaioa (Musaceae) – rare endemic banana of China – according to the molecular phylogenetic data

1 Komarov Botanical Institute of the Russian Academy of Sciences, Prof. Popova St., 2, St. Petersburg, 199376, Russia 2 Altai State University, Lenina Pr., 61, Barnaul, Russia 3 Institute of Botany of Chinese Academy of Sciences, Nanxincun, No.20, Xiangshan, Beijing, China 4 St.-Petersburg State University, Universitetskaya quay, 7−9, St.-Petersburg, Russia 5 ORCID iD: https://orcid.org/0000-0003-2859-9820 6 E-mail: nnosov2004@mail.ru; ORCID iD: https://orcid.org/0000-0002-7096-9242 7 ORCID iD: https://orcid.org/0000-0002-1052-4575 8 ORCID iD: https://orcid.org/0000-0003-3425-1011 9 ORCID iD: https://orcid.org/0000-0003-1146-1622 * Corresponding author


Introduction
Musaceae is a small paleotropical family which is one of the most important food crops in the world. Its morphological taxonomy is rather well established, nevertheless, there are some difficulties in establishing more precise phylogenetic system basing on molecular genetic data.
As a genus, Musa was established by Linnaeus (1753), but the first infrageneric classification was performed only more than a century later by Sagot (1887). He divided bananas into three groups: giant bananas, fleshy (edible) bananas and ornamental bananas with upright inflorescences and brightly colored bracts. Subsequently, Baker following Sagot's classification formally divided bananas into 3 subgenera: Physocaulis Baker, Eumusa Baker and Rhodochlamys Baker (Baker, 1893). Then Cheesman elevated the first group, giant bananas, to the generic level as the genus Ensete Bruce ex Horan. (Cheesman, 1947). The genus Musa, he divided into four sections basing on morphology and chromosome numbers: Eumusa (x = 11), Rhodochlamys (x = 11), Australimusa Cheesman (x = 10) and Callimusa Cheesman (x = 9, 10). This classification was widely accepted and used in many treatments of the genus. Later Argent (1976) proposed a new section, Ingentimusa Argent, with the lowest chromosome number, x = 7. The section is monotypic and distributed only in New Guinea.
New methods of analysis become actual due to the rather morphological unity of the genus . After the development of molecular phylogenetic methods, many phylogenetic schemes basing on molecular data were created. RFLP and AFLP revealed some inconsistencies with the traditional system of the genus Musa, e. g., sect.
Rhodochlamys grouped in the sect. Musa and sect. Australimusa occurred in the sect. Callimusa . In addition, some phylogenetic analyses were performed according to the nuclear and chloroplast gene data. Two main clades were found in the whole genus. The first clade comprised sections Musa and Rhodochlamys and the second species of sections Callimusa, Australimusa and Ingentimusa . Within the whole family genera Musa, Musella (Franch.) H. W. Li and Ensete were monophyletic, but none of the sections previously defined by morphological features. For example, authors suggested the sections Musa and Rhodochlamys combining into the sect. Musa because of the common basic chromosome number, x = 11, and also some morphological characters Liu et al., 2010). The species from the second clade also have common basic chromosome number x = 7, 9, 10. Previous studies revealed that infrageneric classification of bananas is closely correlated with basic chromosome numbers, which lead to the reproductive isolation (Cheeseman, 1947). Of course, new methods of molecular phylogeny will help us clarifying the affinity of the rare and new Musa species.
The genus Musa comprises about 70 species (Simmonds, Weatherup, 1990;De Langhe, 2000;Wu, Kress, 2000;Wong et al., 2002;Häkkinen, Väre, 2008;Häkkinen, 2009Häkkinen, , 2013Lý et al., 2012). Despite rather knowledge of the genus, some new species have been recently discovered (Lý et al., 2012, Singh, 2014. In 1987 banana sample with clear morphological distinction from the other species was found in China (Zhu, 1987). Its relationships are not very clear. In addition, this species was described only in Chinese local journal and was very little known for the scientists.

Confirmation of species independence and affinity of Musa huangbaioa
A series of morphological peculiarities of this newly discovered banana and its probable species status raise a question of molecular phylogenetic analysis of this sample and some other bananas from its possible affinity. Although quite a few bananas were studied by different molecular phylogenetic approaches, such analysis can be rather useful in clarifying the relationships of the newly discovered species and especially in possible hybridization cases. Because of the unusual morphological characters of M. huangbaioa we took into analysis species from different banana sections and clades. For our analysis we used marker sequences from two genome regions: nuclear (ITS) and chloroplast (trnL-trnF). These sequences are widely applied for phylogenetic reconstruction and species identifying (DNA-barcoding) (Hollingsworth et al., 2011). In bananas, as in other families, the usage of different marker sequences is necessary, because Musa is prone to hybridization in some cases (Swangpol et al., 2007).

Materials and methods
Plant samples of the studied species were grown from the seeds collected in December 2016 in Sichuan Province in the territory of the Sichuan Academy of Natural Resource Science Emei Mountain Biology Resource Experiment Station and kept in Botanical Garden of Peter the Great, St.-Petersburg, Russia. Other Musa samples were obtained from greenhouses of Komarov Botanical Institute of the RAS (BIN RAS) and collected during the expedition of the Department Saint-Petersburg Botanical Garden of Peter the Great to Japan in 2019.
We added other sequences from the international GenBank database https://www.ncbi.nlm.nih.gov/ nuccore/?term=). All the sequences used for our analysis are presented in the Table. Totally we analyzed 28 ITS and 26 trnL-trnF sequences using Musella lasiocarpa, a species of closely related genus, as outgroup.   (Edgar, 2004) included in MEGA ver. 7.0 (Kumar et al., 2016) and adjusted manually. Molecular phylogenetic analysis was performed by the Bayesian method and Maximum Likelihood. Models for phylogenetic tree estimation were selected using Akaike Information Criterium (AIC) with MEGA 7.0 (Kumar et al., 2016) and jModelTest 2.1.6 (Darriba et al., 2012). Unambiguous indels were coded with SeqState 1.4.1 (Müller, 2005) and then used in Bayesian analysis as binary characters ("restriction" option). We used Musella lasiocarpa (Franch.) H.W. Li as outgroup, because according to the previous research it occupies rather distant position in the family Musaceae (Simmonds, 1962;. Bayesian analysis was carried out with 1-1.5 million generations until the standard deviation of split frequencies was lower than 0.01. The first 25 % of trees were discarded, option "burnin". Maximum Likelihood analysis was conducted by MEGA 7.0; 1000 bootstrap replications. In the resulting trees, posterior probability is shown at the nodes as the first value, bootstrap value as the second. Clades with 100-90 % of posterior probability and bootstrap index we consider as strongly supported, 89-70 % -as moderately supported and 50-69 %as weakly supported. Indexes below 50 % we regard as no support for the clade.

Taxonomic treatment
M. huangbaioa is an endemic plant growing in subtropical conditions of China. We studied specimen growing in culture at Sichuan Academy of Natural Resource Science Emei Mountain Biology Resource Experiment Station. This species was described by Z. Y. Zhu in Chinese local journal, and thus was forgotten for the majority of researchers.
Here we need to remind of its existence and to provide a correct link to the description of this species. Z. Y. Zhu, 1987,  The plant appearance and some morphological features are presented in Fig. 1.

Results
Musa huangbaioa, a rare endemic banana, forms a clade with various M. basjoo Siebold et Zuccarini ex Iinuma samples according to the nuclear and chloroplast datasets (PP = 99 %, BS = 98 % -ITS data, PP = 87, BS unsupported -trnL-trnF) (Fig. 2,9). It falls into the Clade I that was defined by previous analyses (see Liu et al., 2010;Feng et al., 2016). This species with M. basjoo forms a separate subclade in the Clade I representing the section Musa. It is notable that M. huangbaioa has the same ITS sequences as one sample of M. basjoo cultivated in Mexica (Fig. 2).
According to the ITS data (PP = 99 %, BS = 98 %), the sister subclade in the Clade I corresponds to the sections Musa and Rhodochlamys (Fig. 2).  (Fig. 3). The second subclade of the sections Musa and Rhodochlamys is present as well (PP = 98 %, BS unsupported) but M. acuminata forms a polytomy rather than any of the monophyletic groups (Fig. 3).

Discussion
Species of the most part of banana family (Musaceae) mainly have wide natural range, but M. huangbaioa is described from the foot of Emei mountain and thus is narrow endemic. This is rather unusual feature for the bananas. The morphological characters of Musa huangbaioa show that it belongs to the sect. Musa: its pseudostems are tall, bracts are yellow and sometimes brownish, leaves are with glaucous tinge. This species also differs from all other banana species by unusual ribbed shape of the fruits and undulated petiole margins. The floral characters of M. huangbaioa, as well as the plant height, mostly resemble M. itinerans. Seeds also resemble that of the sect. Musa; they are angular, Confirmation of species independence and affinity of Musa huangbaioa more or less tetrahedral. According to the ITS and trnL-trnF data, this species is closely related to M. basjoo. It is taller than M. basjoo, its height is almost like that of M. acuminata. The floral characters resemble M. itinerans Cheesman, but the stipule is longer than in M. itinerans. Moreover, according to the ITS data, M. huangbaioa is identical to the sample of M. basjoo cultivated in Central America and has the same trnL-trnF sequences as M. basjoo. Speaking of this group as a whole, Chinese group of M. basjoo and M. itinerans is rather uniform and well separated within the sect. Musa (see Simmonds, 1962;Simmonds, Weatherup, 1990;. It could originate in Oligocene (Christelová et al., 2011) and, according to the morphological data, form after the separation of M. balbisiana, M. acuminata and M. schizocarpa lineages (Simmonds, 1962). As shown in our previous article, M. basjoo affinity contains complex hybrid species, probably, with the different maternal genomes. We can assume that this banana, M. huangbaioa, can be modern hybrid with the maternal genome inherited from M. basjoo. Unusual fruit shape can be adaptation to the mountain conditions. According to our data, Musa sect. Rhodochlamys is nested with the sect. Musa in the clade that also comprises M. acuminata affinity group. This well corresponds with the hypothesis by N. Simmonds (1962), that M. acuminata "stock" could be ancestral for the whole section Rhodochlamys. Nevertheless, we do not combine this section with the section Musa due to morphological distinction between these sections. Bananas, as well as many other flowering plants, are subject of reticulate evolution and, thus, their natural system can be the network rather than dichotomous picture. This may really complicate their taxonomy.
The second clade that contains sections Callimusa and Australimusa also presents the reticulation evidence. Musa exotica from the sect. Callimusa, which is rather morphologically distinct from the other species of the sections Callimusa and Australimusa in seed characters , falls into the clade with its relatives with similar seeds, M. coccinea. This fits well with the previous studies Liu et al., 2010;Christelová et al., 2011;Feng et al., 2016). But maternal genome of M. exotica, as it appears, is related to another species of the sect. Callimusa, M. violascens. M. exotica can be the introgressant that originated from the hybridization between members of two clades/lines from the sect. Callimusa retaining the seed traits of M. coccinea group. As in previous cases, this possible hybrid species retains the initial chromosome number of the section, 2n = 20. We can assume that bananas in natural conditions form homoploid hybrids without chromosome number duplication (Feliner et al., 2017).

Conclusions
Molecular phylogenetic methods allowed us to establish the clear relationship of the rare endemic banana species, Musa huangbaioa, which has peculiar and unusual morphological features. M. huangbaioa appeared to be related to the separate Chinese banana group, M. basjoo-M. itinerans affinity; this placement was supported by the seed shape of M. huangbaioa along with the color and shape of male bracts and male flowers. At the same time, our studies showed interesting events of the possible reticulation within this group. This new species, M. huangbaioa, can be hybridogenous and, possibly, its morphological characters formed because of the gene combination. We also confirmed convenience and suitability of the marker sequences from the different genomes, nrITS and trnL-trnF for the identification of affinity in plants, "molecular barcoding" (see Hollingsworth et al., 2011).