Cornus scabrida Franch.

Subgenus Thelycrania includes a group of very similar Chinese dogwoods described during the first two decades of the twentieth century, including Cornus bretschneideri L.Henry, C. alsophila W.W.Sm., C. poliophylla C.K.Schneid. & Wangerin, C. schindleri Wangerin and C. monbeigii Hemsl. This complex of species, which forms a well-supported sub-clade, made a late appearance in Bean’s Trees and Shrubs: none of them was treated in the early two-volume editions (Bean 1914, 1916), but two species (Cornus hemsleyi and C. monbeigii) were included in the supplementary third volume produced in response to the huge number of early-twentieth-century introductions from the region (Bean 1933). In Flora of China (Xiang & Boufford 2005), which we broadly follow, several of these species are placed in synonymy or at inferior ranks, though the picture given in that work is complicated by subsequent molecular data.

By and large, the horticultural value of these plants is inversely proportional to the taxonomic trouble they cause. Bean wrote no more than the truth when he described them as ‘of no particular merit from a garden point of view’ (Bean 1933), but their circumscription and relationship has long been a problem, as the extensive synonymy of Cornus scabrida hints. Xiang & Boufford (2005) treated C. scabrida under the name C. schindleri, and included within it the subspecies poliophylla (C.K.Schneid. & Wangerin) Q.Y.Xiang. No combination for this taxon at subspecific rank currently exists under C. scabrida, so we treat it as a full species (Cornus poliophylla), an approach that finds some support from a recent near-comprehensive phylogenetic study, in which ‘C. schindleri’ and C. poliophylla were recovered as distinct taxa, albeit samples under another synonym – Cornus monbeigii, described from Yunnanese material by Hemsley (1909) – were placed as sister to C. hemsleyi, at a remove from C. scabrida (Du et al. 2023). Given the small sample sizes involved, and the difficulty (notorious in Cornus) in validating voucher specimens, those findings ought perhaps to be treated as provisional, pending a much-needed monographic revision that would integrate morphology with molecular and other data. For the present, we continue to include C. monbeigii within C. scabrida.

Cornus scabrida is distributed widely throughout central and southern China, from Tibet westwards into Hubei and Henan. George Forrest collected flowering specimens on several occasions during the early years of the twentieth century, Forrest 389 from the Mekong-Salween divide in 1904, Forrest 2423 in mixed forests on the eastern flank of the Lichiang Range in June 1906 (Diels 1912), and Forrest 5837 from the same area in 1910 (Herbier Muséum Paris 1917). We learn from Bean (1933) that the species’ first introduction into cultivation, as C. monbeigii, was from a 1917 Forrest collection from the Mekong-Salween watershed. This must have been Forrest 14797, collected there in September 1917 at 3000 m asl., found in open thickets. Circumstantial evidence strongly suggests that a veteran tree (Acc. 19744311) growing at RBG Edinburgh represents Forrest’s original introduction, the tree described on a 1974 herbarium sheet as having grown in the collection ‘without a label for many years’, and referred (with scrupulous circumspection) to the fruiting herbarium specimen of Forrest 14797 in the Garden’s herbarium (RBG Edinburgh 1917, 1974). Such collections may be of conservation significance, since Cornus monbeigii – that is, northwest Yunnanese populations of C. scabrida from Weixi, Lijiang and Degin – like many forest species from this area, was listed as Vulnerable on the IUCN Red List (Sun 1998), owing to loss and degradation of habitat (Oldfield, Lusty & MacKinven 1998).

It seems far from unlikely that other instances of Forrest’s material might persist, incognito, in historical collections. However, most plants in present-day cultivation appear to derive from more recent introductions under C. schindleri, the product of expeditions to Sichuan in the early 1990s (SICH 598 in 1991, SICH 693 in 1992). This material was distributed to several gardens in the United Kingdom and the United States, and accounts for all of the material grown at RBG Kew.

Although more likely to be encountered as a large shrub, Cornus scabrida can form a small to medium-sized tree, as in the case of the Edinburgh specimen, which is multi-stemmed (perhaps attributable to early coppicing) with a substantial low bole and a spreading, rounded crown: its tendency to produce annual suckering stems (bright-green in this case) betrays its close kinship with the rest of the shrubby dogwoods in subgenus Thelycrania. The tree’s evident success in Edinburgh offers a counterpoint to the report by Grimshaw and Bayton (2009) that plants of Sichuanese origin at the Hillier Garden had been killed to the base by frost, and that the species might require hotter summers to ripen the wood. Given the degree of variation across the species’ broad range (Xiang & Boufford 2005), it would hardly be surprising if different provenances corresponded to variations in hardiness.