Tree to 40 m × <1 m dbh. Crown conical or pyrmidal (broadly so in open grown trees), becoming columnar and flat-topped with age. Bark of young trees smooth at first, pinkish- or brownish-grey, developing thin, papery flakes from a young age; becoming rough, scaly, and fissured with age, grey brown in old trees. First order branches horizontally spreading, relatively short, often with the ends upcurved on young or vigorous trees. Branchlets firm, slender, glossy pale pinkish-brown or -grey, grey with pinkish undertones with age, glabrous, weakly ridged behind leaves. Vegetative buds ovoid or broad-conical, c. 5 × 4 mm, largest on leading shoots, slightly resinous, pale chestnut brown. Leaves arranged in multiple ranks each side of shoot, those of the lower ranks more or less pectinate beneath the shoot, those of the upper ranks assurgent, only rarely back-swept (recurved) from the base on vigorous leading shoots especially high in the crown, (1.2–)1.5–3 cm long, 1.9–2.5 mm wide; base twisted or recurved, apex obtuse or acute, glossy green and grooved above, sometimes with a glaucescent bloom, undersides with two dull whitish-green stomatal bands separated by a broad mid-rib, margins very narrow, a few flecks of stomata occasionally present on the upper surface near the apex. Pollen cones 1–1.5 cm long, yellow with reddish microsporophylls. Seed cones often clustered, short-pedunculate, ovoid or ovoid-oblong, apex obtuse or (rarely) papilliform, 5–8(–9) × 2.5–3.5 cm, violet-purple when immature, ripening grey-purple-brown; seed scales broadly flabellate-cyanthiform, 1.5–2 × 2–2.5 cm at mid-cone; bracts spathulate, 1–1.5 cm long, fully included at maturity or with the cusp just exserted; rachis cylindro-conical. (Farjon 2017; Debreczy & Rácz 2011; Fu, Li & Mill 1999).
Distribution China SW Gansu, N Sichuan
Habitat Mixed-coniferous mountain forests, 2300–3600 m asl, often in exposed, cold, and dry situations such as on cliffs and on winter-dry slopes, commonly associating with Juniperus spp., Picea asperata, P. likiangensis subsp. balfouriana, P. wilsonii and Betula albo-sinensis.
USDA Hardiness Zone 5
RHS Hardiness Rating H7
Conservation status Vulnerable (VU)
Taxonomic note Abies recurvata belongs to a group of closely related firs affiliated with A. chensiensis. Their distinctiveness has been amply demonstrated (e.g. Shao et al. 2020; Rushforth 1984) but they continue to be subject to significant divergences in taxonomic treatment. Here we have elected to treat them in the simplest possible way, as distinct but closely related species: A. chensiensis, A. ernestii, A. recurvata, and A. salouenensis. See ‘Abies chensiensis and its relatives’ for further discussion.
Hailing from a much drier and colder region of China than many of its near relatives, Min Fir has proven somewhat more resilient in collections than species such as A. delavayi, A. forrestii, and A. salouenensis from further south. There the climate is influenced to some degree by the monsoons, but in those areas of northern Sichuan and southern Gansu where Abies recurvata is native, the summers are less humid, and most precipitation falls as snow. This combination of cold-hardiness, tolerance of drier sites, and the greater heat tolerance common to most A. chensiensis relatives, renders Min Fir a distinctly useful fir in gardens. Thriving examples may be found in collections of continental and maritime climate zones alike, and on a range of different soil types, from acid podzols to almost pure chalk. It is a frustrating irony, then, that it should remain so rare as it does.
The species was described to science in 1906, based on material Wilson had collected three years earlier during his second visit to China, in the Min valley south of Songpan, Sichuan (W 3021) but there is no record of any seed introduction from this expedition (Rushforth 1999). Seed followed in 1910, again gathered by Wilson on this his fourth expedition – from which came his famous ‘lily limp’, the toll he paid for surviving a landslide during his pursuit of the exquisite Lilium regale – and original trees of W 4051 and 4057 are extant in several collections on both sides of the Atlantic, including Wakehurst in the UK and the Arnold Arboretum in the USA (pers. obs.; Arnold Arboretum 2020). Wilson was clearly a fan, later writing ‘[A. recurvata] is one of the most desirable of the family and was well worth a long journey to introduce it to cultivation’ (Wilson in Bean 1976). This is not to say, though, that every older tree is necesarily derived from Wilson’s gatherings, for Joseph Rock also collected A. recurvata at least twice during his 1925–1927 expedition, and material under Rock 15087 and 15088 is extant in Danish collections. Rock’s gatherings were also distributed by Olsen’s nursery during the late 1920s and 1930s (Wagner 1992).
In optimal conditions A. recurvata grows quickly, and some of the largest examples are known to be Wilson trees. These include a forest plot at Blaize Bailey in the Forest of Dean, UK, from where the Tree Register records several trees in excess of 20 m, the largest being an exceptional 31.2 × 0.7 m dbh (but most 21–24 × 0.45–0.63 m) in 2015 when aged c. 100 years (Tree Register 2020). The largest open-grown trees in arboretum settings include a remarkable tree at Benmore Botanic Garden in Argyll, Scotland, 25 × 0.91 m in 2017; one near Sevenoaks in Kent, England, 22 × 0.43 m in 2014; and at Fota, Co Cork, Ireland, 19.6 × 0.44 m in 2017, when aged c. 80 years (Tree Register 2020). In their survey of firs in the Arnold Arboretum, Warren & Johnson noted that it had been a slow grower there; the largest of three trees from Wilson’s gatherings was only 9 m in 1988, aged c. 78 years (Warren & Johnson 1988). Another ‘slow form’, very likely from Wilson seed, grows at Devonhall in Clackmannanshire, Scotland.
Older trees of A. recurvata (and young grafts taken from old trees) are quite distinctive in several features of their foliage: the somewhat assurgent arrangement of the leaves above the shoot; the shoots glabrous, shining, pale pinkish-brown; the leaves blunt tipped, with entire, obtuse, or acuminate apices, often with a glaucous bloom on the upper surface. This combination of characters is only likely to be confused, on older trees, with its close relatives A. ernestii and A. holophylla, these can be distinguished in the following ways: A. holophylla with longer, sharp-pointed, paler green needles, rarely if ever with a glaucous hue; A. ernestii with longer more pectinate needles, with ermarginate or bifid apices (on sterile shoots), and seed cones with a (usually) dimpled apex (Rushforth 1999). Confusion with A. chensiensis and A. salouenensis is unlikely on account of their longer leaves.
Young seed-raised trees can be difficult to separate from A. ernestii. In youth, A. recurvata foliage can be more pectinate than in old age, especially when specimens are shaded, which they often are in collections as curators place them under existing dappled canopy, logically taking advantage of the species’ shade-tolerance. The best means of separating them in such situations is by the leaves with rounded or pointed apices in A. recurvata cf. emarginate or bifid in A. ernestii; the shoots pink-tinged and glabrous in A. recurvata cf. very pale yellowish-brown (lacking the pink hue) and faintly pubescent (at least when young) in A. ernestii (Debreczy & Rácz 2011; Rushforth 1999).
Whilst the type specimen exhibits the strongly backswept leaves for which the species was named, few trees in cultivation exhibit this trait particularly strongly, even when open-grown. It may be most commonly observed on vigorous leading shoots, especially in the upper crown (Rushforth 1999). The most remarkable example of this feature observed in research for this account was on a tree in a private collection in western Hungary, growing in an area with a strongly continental climate and slightly alkaline soils (pers. obs. 2019); it is illustrated below. This is a relatively young tree of c. 30 years and c. 10 m tall, grown from seed gathered in China by Debreczy and Rácz. The same feature was also very pronounced on a tree growing in the warm, and often dry climate of Gwavas Station, Hawkes Bay, New Zealand (pers. obs. 2017).
Min Fir has not been widely collected since China re-opened to western collecting teams in 1980. Its somewhat inconvenient distribution in northern Sichuan has been just beyond the reach of most expeditions, which have understandably been focussed on the greater diversity that exists further south. The first young plants raised after 1980 are traceable to seed sent out by the Chinese Academy of Forestry, and these are represented in collections including Gothenburg in Sweden, and Bedgebury and the Yorkshire Arboretum in the UK, usually by only one or two trees in each location (pers. obs.). At least one consignment from the Academy, sent in 1979 or 1980, bore the tantalising name ‘Abies squamata’, but all trees from this batch seen in the UK fit A. ernestii. Curiously, a single tree from the same batch at Gothenburg (accession 1980–0001) fits A. recurvata, suggesting a mixed collection (pers. obs.). This grows in dappled shade beneath an Oak canopy in the Rhododendron collection. Nearby, another specimen (accession 1986–0055) is descended from a slightly later Academy distribution, this time correctly named. Another from the same batch grows at the Yorkshire Arboretum (accession 19869027), this was 9.7 m × 22 cm dbh in 2021 (J. Grimshaw pers. comm. 2021).
The SICH expeditions seem only to have collected A. ernestii, with the single exception of SICH 2354 which is may be genuine A. recurvata (it has not been seen in research for this work). The 1994 Alpine Garden Society expedition to China collected this species (ACE 675) but no living trees have been traced. More recently, in 2011, BCHM 96 was collected as A. recurvata. This gathering has been widely distributed among collections both in western Europe and in North America, but while young plants generally look good for A. recurvata (pers. obs.) the size of the parent tree’s cones (seen in photographs) arouse suspicion (K. Rushforth pers. comm. 2020). This gathering may yet turn out to represent A. ernestii, or a blurring of the (artificial) boundaries between these two, very closely related taxa.
This relative lack of new introductions has prompted the frequent repropagation of old trees. At the Arnold Arboretum an original tree of W 4057 was propagated in 1986; the two resulting plants are now 13.2 and 15.6 cm dbh respectively, perhaps suggesting a faster rate of growth compared with the originals (Arnold Arboretum 2020). Young grafts of old trees may be found in several collections in the UK, for example at Crathes in Scotland. A young tree at Pampisford, Cambridgeshire, was sourced from the late Derek Spicer; whilst its provenance is uncertain it is remarkable for its health and relative vigour here, on shallow soils above solid chalk in a very dry part of England (pers. obs. 2020).