Cornus wilsoniana Wangerin

Surprisingly rare in cultivation, Cornus wilsoniana is distinguished principally by its remarkable bark, which exfoliates in patches to give a pale mottled appearance comparable with that of Pinus bungeana (Lacebark Pine) or Pseudocydonia sinensis, and which earned it its Chinese vernacular name, Ghost Dogwood (He et al. 2023). Borne in profusion in early summer, the flat-topped cymes of small white flowers recall C. controversa, and are followed by shining black fruit. The handsome foliage is grey-green, the leaves distinctively pale beneath.

The scientific name memorialises Ernest Wilson, who collected the type (Wilson 764) in western Hubei in 1901. Within a few months of its publication by Wangerin (1908), however, the species had become confused with Cornus walteri Wangerin, creating a nomenclaturial tangle that hampers the attempt to trace the early history of either of these species in cultivation. While preparing his description of Cornus wilsoniana, it became clear to Wangerin that the material assembled under Wilson’s number 764 in the herbarium at St. Petersburg included specimens belonging to another species, referrable to an earlier collection by Augustine Henry from Hubei. Wangerin used Henry’s and Wilson’s material (Henry 6382 and W 764 in part) as the type of the new species C. walteri, separating the two taxa on the basis of C. wilsoniana’s cylindrical (not club-shaped) style, as well as its paniculate (rather than strictly corymbose) inflorescence structure (Wangerin 1908). Not all of the sets of Wilson’s herbarium material (by now distributed around the world) included both species, however, and on the basis of the material available to him at Kew in 1909, Hemsley simply placed Wangerin’s ‘Cornus walteri’ in synonymy with C. wilsoniana, as did Schneider, overlooking the fact that C. walteri was based on extraneous material included under W 764 in other herbaria (Hemsley 1909; Schneider 1912). Schneider corrected the record in the second edition of his account of broadleaves in cultivation in Germany, prepared with Silva-Tarouca and Ambrózy (Silva-Tarouca et al. 1922). Molecular data places C. wilsoniana close to C. macrophylla, as one of the older lineages in the ‘sanguinea’ clade of subgen. Thelycrania (Du et al. 2023) – see the genus entry.

As a consequence of Hemsley’s mistaken synonymy, at least some early introductions of Cornus walteri were mislabelled as C. wilsoniana (or sometimes C. wilsonii). Wilson himself seems to have adopted Hemsley’s nomenclature after his return from his 1910–1911 expedition, which has ramifications for the interpretation of his travel accounts. It is striking that from 1911 onwards, Wilson no longer refers to C. walteri, but uses only Hemsley’s name C. wilsoniana. The ‘small trees of Cornus wilsoniana in flower’ around Yichang described by Wilson in 1913 seem likely to been C. walteri – they correspond with the two trees from Xingshan County, Yichang, both labelled ‘C. walteri’ in photographs taken in 1910 (O-73 and O-17). And in an address to the Horticultural Club of Boston in November 1911, Wilson reported the introduction of ‘Cornus wilsoniana (20 to 40 feet, flower white, in corymbs)’ a description that matches C. walteri (Wilson 1911). Had the examples of ‘Cornus wilsoniana’ mentioned in his 1913 account been the true species, Wilson would surely have remarked on the mottled bark.

Were it not for Wilson’s single collection in 1901, and a superb photograph taken by him of a 20 m tree growing by the side of a path in the Chin-Tsing Shan, western Sichuan (Silva-Tarouca et al. 1922), one might (almost) be tempted wonder whether Wilson was familiar with the true Cornus wilsoniana at all. Its complete absence, bark and all, from his horticultural works such as Aristocrats of the Garden (Wilson 1917, 1938) and its sequel More Aristocrats of the Garden (Wilson 1928) is quite noteworthy, strongly suggesting that the true species cannot then have been in cultivation on either side of the Atlantic. This is also the implication of the species’ absence from successive editions of Bean’s Trees and Shrubs. By 1933, the name C. wilsoniana Hort. was recorded by Bean as a horticultural synonym for C. walteri (itself not then commonly grown), and we can reasonably presume that such plants as the purported ‘Cornus wilsonii’ grown by P.D. Williams at St. Keverne gardens, Cornwall in the late 1920s were really C. walteri, incognito (K[ingdon]-W[ard] 1928).

Nevertheless, it does seem that some introductions, at least, might have been the true Cornus wilsoniana. John ‘Lilacseed’ Dunbar, who played such an influential role in the history of early Wilson introductions to the United States (and who is memorialised in the nothospecies Cornus × dunbarii) grew two specimens together at Highland Park, Rochester, New York, where they flowered for the first time in July 1918. These young plants stood out not for their still immature bark, but for the abaxially glaucescent leaves with adpressed hairs (Dunbar 1918a) characteristic of C. wilsoniana (the leaves of C. walteri are pale green beneath (Rehder 1940)); one of these plants was reported killed and the other injured in the cold winter of 1917–1918 that set back much young material raised from Wilson’s American introductions (Dunbar 1918b). By 1940, Rehder thought that the species was probably not in cultivation, and that plants grown under this name were referrable to C. walteri.

Cornus wilsoniana remains rare in cultivation even now – surprising, perhaps, for a species found in at least twelve Chinese provinces. One reason why the species appears not to have been collected by the many expeditions that visited China since its ‘reopening’ to the West in the 1980s may be the relatively low elevation at which it occurs, up to 1100 metres asl. Expeditions focused on hardier, higher-altitude plants suited to North American or North European conditions may simply have overlooked the species. On the other hand, shrubby collections made by George Forrest in 1919 at far greater altitudes and referred to C. wilsoniana (or C. wilsoniana aff.) are questionable: Forrest 19119 (from southeast Tibet) and 14760 (Mekong-Salween divide) were both collected at 2750–3000 m, outside the species’ accepted range; further investigation of his specimens may shed light on this.

The attractive tree at the Sir Harold Hillier Gardens, Hampshire was grown from seed obtained by Roy Lancaster from Hangzhou Botanic Garden in Zheijang province, China, through its 1985 Index seminum (no. 345). This batch of seed was distributed to a number of gardens and growers, though only two other plants from this batch are known in Britain – one at the Cambridge Botanic Garden, now over 8 m tall, and another at Nymans in Sussex, both thriving. Seed from Hangzhou received by Strasbourg Botanic Garden has produced a much-admired tree (8 m tall by 2002) whose progeny are now growing in a number of French gardens and arboreta, including Le Vastérival, near Dieppe, and Rastel, Tarn-et-Garonne (Demoly & Picard 2005; V. Malecot, pers. comm. 2022). Another fine tree in Nantes may be from the same source (T. Christian, pers. comm. 2026).

Two further plants grow at the Hillier Gardens, both raised from wild-origin seed distributed by Shanghai Botanic Garden, collected on Emei Shan, Sichuan in 2003 (D. Jewell, pers. comm. 2022). Having grown away unchecked, these are now beginning to show the characteristic marbled bark. A tree at White House Farm, Kent, is growing with similar vigour, originating as a collection by Bob Cherry in Yunnan (M. Foster, pers. comm 2023).

The species is now commercially available on both sides of the Atlantic, and in Australia. In the Southern United States, the SFA Gardens, Nacogdoches, have raised and distributed plants grown from seed collected on Zijin Shan (Purple Mountain), Jiangsu Province. In the climate of eastern Texas (USDA Zone 8b) the tree has proven very hardy on alluvial soils, surviving –20°C in the exceptionally severe winter of 2021–2022. This introduction is marcescent in habit, retaining its dead foliage through the winter rather than dropping it; autumnal color is described as disappointing (D. Creech, pers. comm. 2026).

In Britain, Cornus wilsoniana is best characterised as a semi-evergreen species, defoliating at low temperatures, but otherwise elegant and clean in appearance. The character of the bark improves with age, shown to best advantage when grown in good light. In cultivation it is set to be a small tree, often multi-stemmed or low-branching to form an open crown. The Hillier specimen (the British champion tree) planted in 1988 measured a respectable 9 metres tall in 2022, comparable with the 10 m trees that Augustine Henry found growing by rivers and streams, forming a dome-shaped crowns (O’Brien 2011). The remarkable specimens of 40 m reported from Chinese forests seem a distant, probably unattainable prospect in cultivation. The best flowering has been seen following a hot summer (pers. obs.). In its native China, it has been suggested for use as a street tree, while the oil-rich fruits and heavy hard timber are valuable products (Xiang & Boufford 2005). The fruit provides an edible oil that is extremely high in polyunsaturated fatty acids, showing a marked hypolipidemic effect (He et al. 2023).

The sister species Cornus oligophlebeia Merrill has a more southerly distribution, from southeast Yunnan and Bhutan southwards through Myanmar, Thailand and South Vietnam (Du et al. 2023; Xiang & Boufford 2005; POWO 11/12/2025). It is an autumn-flowering tree to 15 m, fruiting in later winter, and with large, leathery, more-or-less glabrous leaves (Merrill 1942). It is not known to be in cultivation, and seems unlikely to be hardy in our study area, but would be interesting to trial in congenial situations.