Shrubs about 0.8 m tall. Leaves glabrous; lower leaves biternate, with 9 leaflets; terminal leaflets obovate—deltoid, 8—10 cm long, 7-9 cm broad, 3- or 5-cleft to the middle or even beyond, lateral leaflets ovate or ovate-lanceolate, 4-7 cm long, 2-4.5 cm broad, entire or shallowly lobed. Flowers solitary, terminal, single; involucrate bracts 2—6 in number, glabrous; sepals 4—5 in number, all caudate at the apex, 3-3.5 cm long, 2-3 cm broad, glabrous; petals 9 or 10 in number, rose, broadly obovate, rounded at the apex, 5-6 cm long, 4-6 cm broad; filaments purple, anthers yellow; disk entirely enveloping carpels at anthesis, purple; stigmas purple.
Chromosome number: 2n=10.
Native to W Henan (Songxian) and W Hubei (Baokang).
When we (Hong et al., 1998) described Paeonia suffruticosa subsp. yinpingmudan as new, we cited two specimens, one from a cliff in the Yinping Hill, Chaohu, Anhui Province (K. Y. Pan & Z.W Xie 9701) and the other from a cultivated shrub next to Mr YANG Hui-Fang’s house in Secaogou Village, Shigunping, Muzhijie Township, Songxian County, Henan Province (D. Y. Hang, Y. Z. Ye & Y. X. Feng H97010). The former has only one leaf and several white petals, whereas the latter, with rose petals, was introduced from a nearby mountain around 1961, as Mr Yang told us. These two specimens were considered to form a single entity, and possibly a wild form of the cultivated tree peony on the basis of similarity of the leaves, although they had different petal colours. It is regretful that we designated 9701, instead ofH97010, as the type of P. suffruticosa subsp. yinpingmudan, considering that the former came from an individual of wild origin with considerable certainty, but ignoring its incompleteness as a specimen.
Our recent DNA sequencing data do not support this taxonomic treatment. On all the molecular trees of the nuclear GPAT gene, the nuclear gene family Adh1A, Adh1B and Adh2, and cpDNA, K. Y. Pan & Z. W. Xie 9701 and P. ostii formed a clade, whereas H97010 either formed an independent clade or formed a clade with P. qiui (Zhao et al., 2004; Lin et al., 2004; Zhou et al., unpub.).
Haw (2006) strongly argued that Paeonia suffruticosa subsp. yinpingmudan from Anhui is an element of P. ostii, but not the direct progenitor of P. suffruticosa. According to his observation, the peony on the cliff from Yinping Hill in Anhui Province has lower leaves with 11 instead of 9 leaﬂets. Mr LI Min and Mr MA Xin-Tang made a trip to Yinping Hill in April of 2006 to further observe the peony on the cliff. They collected petals falling down the cliff and took a large number of photos using a telelens. According to their photos, the lower leaves of this tree peony actually have 13 (not nine) leaﬂets, which are ovate-lanceolate or ovate, and mostly entire (Hong & Pan, 2007). The leaf of the type specimen of P. suffruticosa subsp. yinpingmudan (Pan & Xie 9701, with nine leaﬂets) (Hong et al., 1998) might be a middle leaf but not a lower one. The flowers of this type specimen are purely white, with purple filaments and disc. Therefore, there are no signiﬁcant differences between the tree peony on the cliff (Pan & Xie 9701) and the specimens of P. ostii that we have examined to date. The tree peony in Songxian of Henan Province (Hang et al. H97010), however, differs from P. suffruticosa subsp. yinpingmudan in having the lower leaves strictly with nine leaﬂets, mostly lobed, and rose petals. Therefore, the two collections of tree peonies cited when P. suffruticosa subsp. yinpingmudan was described as new actually belong to two different entities, Pan & Xie 9701 (Anhui) being an element of P. ostii as shown by DNA data and as stated by Haw (2006), whereas Hong at al. H97010 (Henan) might be a real wild form of the cultivated tree peony. Thus, Paeonia suffruticosa subsp. yinpingmudan from Anhui is reduced herein to a synonym of P. ostii, whereas the tree peony from Henan has been described as a new species, P. cathayana (Hong & Pan, 2007).
Mr B. A. Shen’s Paeonia yinpingmudan subsp. henanensis (Shen, 2001) is an illegitimate name because D. Y. Hong and his co-workers never gave the Latin name ‘henanensis’, and neither did Mr Shen designate a type for his name ‘henanensis’. However, from Mr Shen’s short note in Chinese, his ‘henanensis’ is clearly connected with Hong, Ye & Feng H97010.