The identity of Prunusdielsiana (Rosaceae)

Abstract The valid publication date of Prunusdielsiana was found to be later than that of P.rufoides, which has been considered a synonym of P.dielsiana. Prunusdielsiana is therefore reduced to a synonym of P.rufoides, instead of the reverse. In addition, all previously named varieties of Prunusdielsiana, including var. abbreviata, var. conferta, and var. laxa, as well as P.carcharias are also listed as synonyms of P.rufoides in the present paper.


Introduction
Prunus L. subg. Cerasus A. Gray, a group commonly known as cherries, is historically controversial in its taxonomy. As concluded by Wu et al. (2018), the taxonomy of this clade needs extensive study.
Prunus dielsiana C. K. Schneid. is a species widely distributed around central China and east China (Li and Bartholomew 2003). While reviewing the protologue of this species of P. subg. Cerasus, P. dielsiana C. K. Schneid. was found to be an invalid name. This name first appeared in Schneider's account in 1905. Schneider (1905 proposed a description of a cherry collection, Wilson 308, which he determined as "Prunus szechuanica var.?," indicating his uncertainty about its identification. Schneider stated that should this collection be a new species or a distinct variety of Prunus szechuanica, he would have proposed to name it as "P. dielsiana" or rather "var. dielsiana". Schneider set the varietal name in bold and also indicated acceptance of the varietal status by citing only "Prunus szechuanica var. dielsiana" in the index. Prunus szechuanica var. dielsiana C. K. Schneid. was therefore validly published while P. dielsiana C. K. Schneid. was invalid. Prunus dielsiana was validly published seven years later by Koehne (1912), who provided an entirely new description of the taxon. The name was ascribed to "Schneider in Fedde, Rep. Nov. Sp. I. 68 (1905)", which should be treated as a reference to the basionym, and Koehne's name should be considered as a new combination based on Prunus szechuanica var. dielsiana C. K. Schneid..
Unaware of the fact mentioned above, Yü and Li (1986) Li and Bartholomew (2003). This is incorrect since P. rufoides C. K. Schneid. was validly published in 1905, while the earliest homotypic species-level synonym for C. dielsiana was published in 1912. Thus, Li and Bartholomew (2003) should have listed C. dielsiana (C. K. Schneid.) Yü et Li as a synonym of P. rufoides C. K. Schneid.
Three varieties have been published under Prunus dielsiana. Prunus dielsiana var. conferta and P. dielsiana var. laxa were described by Koehne (1912), based on specimens from western Hubei. They were thought to be different in their involucres (bracts subtending the inflorescence), with the involucres of P. dielsiana var. conferta described as erect and closed and those of P. dielsiana var. laxa described as open or sub-reflected. However, neither of these varieties was accepted as distinct in "Flora Reipublicae Popularis Sinicae" (Yü and Li 1986) and "Flora of China" (Li and Bartholomew 2003), where both were listed as synonyms of Cerasus dielsiana. In addition, P. dielsiana var. conferta was based on the same type as P. szechuanica var. dielsiana, making it a later homotypic synonym that cannot be validly published under Art. 22.2 of ICN (Turland et al. 2018).
Prunus dielsiana var. abbreviata Cardot was described by Cardot (1920) based on Cavalerie et Fortunat 2276 collected from Guizhou (Kouy-Tcheou). He stated that this variety was similar to P. dielsiana var. conferta Koehne, but different in its narrow involucres and very short peduncles hidden in the involucres. This variety was accepted as distinct in "Flora Reipublicae Popularis Sinicae" (Yü and Li 1986) and "Flora of China" (Li and Bartholomew 2003), but we disagree with this interpretation. The distinguishing trait is rather unstable, and even in the type specimen, not all the peduncles are hidden in the involucre. As mentioned by Wang (2014), P. dielsiana is a highly variable species. In our field and herbarium specimen observations, leaf and floral characters, including those of the involucres and peduncles that were used to distinguish these three varieties, exhibited extensive variation (Figure 1). Therefore, we deem that none of the three previously named varieties is worthy of taxonomic recognition.
Prunus carcharias was described based on a leafy branch without flowers and fruit collected from China, Chongqing, Nanchuan County (Koehne 1912), and the name is still unresolved today (Li and Bartholomew 2003). This species was considered as a member of P. sect. Microcerasus (Spach) C. K. Schneid. by Koehne (1912) because its leaf shape and serration were very similar to P. nakaii H. Lév. (P. japonica var. nakaii (H. Lév.) Rehder). However, we believe that this branch must represent an adventitious shoot of P. rufoides (Figure 2), a phenomenon commonly observed in Nanchuan. Leaf dimorphism often occurs in P. subg. Cerasus, such that the leaves on the adventitious shoots and summer shoots may be obviously different from typical leaves of the species. Therefore it is not advisable to describe new species of this clade based solely on differences in leaf morphology, and we consider P. carcharias Koehne to be a synonym of P. rufoides.
Koehne cited three gatherings from Hubei, Wilson 37, Wilson 37a and Wilson 68, when describing P. dielsiana var. laxa. There are 12 sheets of these three gatherings that can be found in the Global Plant Database (JSTOR 2019), and we found that two  sheets of Wilson 37a, both with identification slips from Koehne, were mixed species collections. One of them, HBG511107, contains a flowering branch of P. conradinae Koehne, while another one, US00623845, contains a leafy branch of P. tomentosa Thunb. Therefore, a lectotype was selected in agreement with Art. 9.3 and Art. 9.14 of ICN (Turland et al. 2018).