﻿Malus includes Docynia (Maleae, Rosaceae): evidence from phylogenomics and morphology

﻿Abstract Docynia has been treated as a separate genus or merged into Cydonia or Docyniopsis. Our phylogenomic evidence from 797 single-copy nuclear genes and plastomes confirmed the sister relationship between Docynia and Docyniopsis. By integrating the phylogenomic and morphological evidence, we propose to accept a broad generic concept of Malus and merge Docynia into Malus. Three new combinations are also made here: Malusdelavayi (Franch.) B.B.Liu, M.indica (Wall.) B.B.Liu and M.longiunguis (Q.Luo & J.L.Liu) B.B.Liu.

Recent phylogenetic and phylogenomic studies presented strong topological discordance amongst nuclear/plastid genes and showed cytonuclear conflicts (referring to fig. 1 in Liu et al. (2022)). Docynia is closely related to Docyniopsis (C.K.Schneid.) Koidz. (= Malus sect. Docyniopsis C.K.Schneid.), based on the plastomes and the nuclear sequences (Lo and Donoghue 2012;Liu et al. 2019Liu et al. , 2020aLiu et al. , 2020bLiu et al. , 2022Jin et al. 2023). Several shared morphological characteristics have also supported their close relationship, i.e. cone-shaped non-adnate part of the ovaries (Fig. 1B1, C1), fully connate carpels (Fig. 1B1, C1), incurved and persistent calyx, numerous scattered sclereids throughout the flesh, juvenile leaves deeply lobed and similar flavonoid chemistry (Williams 1982;Robertson et al. 1991;Kalkman 2004). However, Jin (2014) proposed an alternative phylogenetic inference, based on the whole plastome, the sister relationship between Docynia and Cydonia. Additionally, Xiang et al. (2017) inferred a close relationship between Docynia and Eriolobus M.Roem., based on the transcriptomic data and this result provided another line of evidence for Schneider's (1906) taxonomic transfer. However, Xiang et al. (2017) sampled only four apple-related species, Malus baccata (L.) Borkh., M. domestica (Suckow) Borkh., Docynia delavayi (Franch.) C.K.Schneid. and Eriolobus trilobatus M.Roem., the inferred phylogenomic topology based on this limited taxon sampling; thus, an accurate species relationship was not presented. Therefore, the argument that Docynia should be transferred to the genus Eriolobus, based on a strongly-supported sister relationship between the two taxa is untenable, as this evidence with limited taxon sampling is insufficient to justify a taxonomic reclassification proposed by Schneider (1906). Liu et al. (2022) sampled 39 individuals representing 18 wild species and provided a robust backbone of the apple and its allies in the framework of the tribe Maleae integrating 797 single-copy nuclear genes (SCN genes) and whole plastome data (Fig. 2). This phylogenomic analyses resolved the phylogenetic position of Docynia, placing it within Malus sensu lato (Liu et al. 2022).
In this study, we aim to transfer three currently-recognised species of Docynia to Malus.

Materials and methods
We sampled 77 individuals in the framework of Maleae, of which 39 were apple-related species and the other 38 were outgroup species. All these 77 samples were performed for deep genome skimming (DGS) sequencing with 5-10G data for each sample. We assembled the whole plastome using NOVOPlasty v. 4.3.1 (Dierckxsens et al. 2016) and a successive assembly approach (Liu et al. 2021).   Given the rich genomic resources in various lineages of Rosaceae, we screened 797 nuclear SCN genes from six genomes, Malus baccata, M. domestica, Pyrus betulifolia Bunge, P. bretschneideri Rehder, P. ussuriensis Maxim. × P. communis L. and P. pyrifolia (Burm.f.) Nakai. We assembled these 797 nuclear SCN genes for these 77 samples using HybPiper pipeline v. 1.3.1 (Johnson et al. 2016). The assembled sequences were then cleaned with a series of procedures, such as trimAL v. 1.2 (Capella-Gutiérrez et al. 2009), AMAS v. 1.0 (Borowiec 2016), TreeShrink v. 1.3.9 (Mai and Mirarab 2018) and Spruceup (Borowiec 2019). We combined the concatenated and coalescent-based methods for accurate phylogenetic inference. As for the concatenated-based method, we performed Maximum Likelihood (ML) tree inference using RAxML 8.2.12 (Stamatakis 2014) and IQ-TREE2 v. 2.1.3 (Minh et al. 2020). The shrunken trees from TreeShrink (Mai and Mirarab 2018) were used as input to estimate a coalescent-based species tree with ASTRAL-III (Zhang et al. 2018). The detailed parameters refer to the materials and methods in Liu et al. (2022).

Results and discussion
The phylogenetic relationship between Docynia and Malus has been controversial for two centuries. Our results revealed that all these nine nuclear and plastid trees in our study (Liu et al. 2022) demonstrated the paraphyly of Malus s.s., with Docynia nested within it (Fig. 2) and this was also confirmed in several recent molecular studies (Lo and Donoghue 2012;Xiang et al. 2017;Liu et al. 2020a). The sister relationship between Docynia and Malus doumeri A.Chev. (= Docyniopsis) was confirmed either in the nuclear or plastid trees, except for the conflicting phylogenetic placement of the Docynia-Docyniopsis clade. As indicated in our previous study (Liu et al. 2022), the close relationship between Docynia and Docyniopsis (Fig. 3) was also supported by the morphological evidence, such as cone-shaped non-adnate part of the ovaries (Figs 1B1, C1, 3I), fully connate carpels (Figs 1B1, C1, 3K), incurved and persistent calyx (Fig.  3A, F, I, J), numerous scattered sclereids throughout the flesh, juvenile leaves deeply lobed and similar flavonoid chemistry (Williams 1982;Robertson et al. 1991;Kalkman 2004).
Despite the monophyly of narrowly-circumscribed small genera in Malus s.l., including Chloromeles (Decne.) Decne., Docynia, Docyniopsis, Eriolobus M.Roem. and Malus sensu stricto, we believe that such narrow generic concepts may be impractical for use by botanists, ecologists, conservation biologists and horticulturalists. Given the prevalence of reticulations in angiosperms, we recommend integrating multiple lines of evidence for accurate taxonomic treatments, including morphology, phylogenomics, cytology, biogeography and ecology, as proposed by integrative systematics ). Traditionally, taxonomic circumscription was often focused solely on the taxonomic community, with little consideration given to its broader implications. However, today there is a growing recognition that taxonomic circumscription can have far-reaching effects on many aspects of biology, including conservation, ecology and evolution. By considering the needs of the broader biological community, taxonomic circumscription can help to ensure that taxonomic classifications are more valuable and relevant to a wider range of researchers and practitioners. An excessive inclination towards separating genera can hinder the advancement of research programmes for understanding evolution across all descendants stemming from a common ancestor. Additionally, by educating the general public about the importance of taxonomic circumscription, we can help foster a greater appreciation for biodiversity and its role in understanding and conserving it. In summary, taxonomic circumscription today should be viewed as a tool for serving the needs of both the taxonomic community and the broader biological community, as well as educating the general public about the importance of biodiversity and taxonomy (Wen et al. 2015Funk 2018).
With all these considerations, we propose using the broad generic concept of Malus, which includes all members of Malus sensu Gu and Spongberg (2003) and the species in Docynia. We here formally transferred the three currently-recognised species of Docynia to Malus in the following text.   = Malus docynioides C.K.Schneid., Bot. Gaz. 63: 400 (1917 (1831) did not designate a specimen as the holotype, but only provided an illustration, which is considered to be the holotype (Fig. 4). However, the accurate identification of this species will be significantly impeded due to the limited morphological details in the illustration compared to the specimens (Turland et al. 2018). Consequently, it becomes necessary to select a single specimen as the epitype in order to distinguish it from its closest relatives, such as Malus delavayi and M. longiunguis. Decaisne (1874)   Distribution. China (Sichuan). Note 2. In the protologue, the holotype is indicated as being deposited in the herbarium of Xichang College (HXCH, Luo et al. 2011); however, this holotype specimen was then sent to the China National Herbarium (PE).