Abies in the Sino-Himalaya

Abies chensiensis and its relatives

Abies chensiensis Tiegh. was published in 1892, based on material David had collected in the Qinling mountains in modern day Shaanxi, China (Farjon 2017). This is the oldest name in a group of closely related Chinese firs whose relations have been hotly debated over the years, perhaps best summed up as the A. chensiensis complex. The next name to be published in this complex was A. recurvata Masters in 1906, followed by A. ernestii Rehder in 1939, and A. salouenensis Bordères & Gaussen in 1947. These three names have, over the years, been deployed in just about every conceivable combination, as subspecies and varieties of one another or of A. chensiensis, to the extent that it would be pointless to relay all these combinations here. It comes as no surprise that these taxa are often confused in collections. One symptom of the confusion surrounding them is the suite of inconsistencies that exists in scientific descriptions in major literature. For example, the Flora of China (Fu, Li & Mill 1999) description of A. ernestii suggests that it has greenish-coloured immature seed cones, ripening yellowish-brown, whereas Rushforth (1984), Farjon (1990; 2017) and Debreczy and Rácz (2011) all agree they are violet purple.

These firs may all be distinguished from other species complexes in the region by the combination of: pinkish-grey or yellowish-grey branchlets (cf. brown, red-brown, maroon, purple, or distinctly buff); relatively large, pale chestnut-brown ovoid-conical buds that are usually only slightly resinous (cf. small and globose, or large-ovoid but red-brown and densely covered in resin); and leaves with dull undersides and dull stomatal bands (cf. usually quite striking whitish bands). These firs all belong to Section Momi, according to the infrageneric classification of Farjon and Rushforth (as revised in Farjon (2017)) hence are closely related to the more familiar A. firma and A. homolepis of Japan. Besides the morphological similarities in this group their ecologies are an important link. Particularly in China, most of these species generally occur at lower elevations than those of other Sections that are distributed in the same regions. As such, they tend to occur with a broader range of associates, including a greater proportion of broadleaves, and are well adapted to compete with these by being notably shade tolerant in youth and, on the whole, somewhat more heat tolerant than many of the more familiar Asiatic firs of higher elevations (Debreczy & Rácz 2011).

Rushforth revised this group of Chinese firs in 1984, demonstrating that A. ernestii was referable to A. recurvata at the rank of variety, and that A. salouenensis was referable to A. chensiensis at the rank of subspecies. He also published an additional subspecies, A. chensiensis subsp. yulongxueshanensis based on material from the Lijiang Range in Yunnan (Rushforth 1984). This revision has strongly influenced the ‘western view’ of this group ever since; for example it has been followed by Farjon in all his major works (e.g. 1990; 2017). However, the Flora of China account differs in several respects, chiefly by not recognising any infraspecifics under A. chensiensis or A. recurvata, and by referring A. salouenensis to A. ernestii, which was treated at species rank. The same work synonymises Rushforth’s yulongxueshanensis taxon within their interpretation of salouenensis (Fu, Li & Mill 1999).

A recent paper (Shao & Xiang 2015) sought to resolve this complex through a study combining morphological and molecular data. Their results supported A. chensiensis, A. ernestii, and A. recurvata as good species. They further supported salouenensis as a distinct entity but maintained the Flora of China approach in treating it as a variety of A. ernestii. Notwithstanding the application of modern scientific methods to this conundrum, Shao & Xiang’s findings do not particularly tell us anything new: Clearly, whatever one’s species concept, there is broad agreement now backed up by molecular evidence that the A. chensiensis complex comprises at least four unique taxa: chensiensis; ernestii; recurvata; and salouenensis, but their apparent failure to include material from the type locality of A. recurvata in their sampling could be interpreted as a significant weakness of their study (K. Rushforth pers. comm. 2020).

The status of yulongxueshanensis remains contentious. Rushforth distinguished it from A. chensiensis based on leaves of coning shoots with marginal (cf. medial) resin canals and the larger seed cones (10–14 cm cf. 7–10 cm); from salouenensis he distinguished it based on the shorter leaves of sterile shoots (to 4.5 cm) (Rushforth 1984). Shao & Xiang did not specifically consider it; their sampling clearly infers allignment with the Flora of China view equating this taxon to salouenensis (Shao & Xiang 2015). Debreczy & Rácz (2011) don’t refer to yulongxueshanensis at all, but in discussing the conifer diversity of the Lijiang Massif and Yulongxue Shan in particular, make reference only to A. salouenensis, again inferring agreement with the Flora of China position. The same authors diverge from standard western approaches to the chensiensis complex by recognising all the constituents at species rank, as A. chensiensis, A. recurvata, A. ernestii, and A. salouenensis (Debreczy & Rácz 2011). This alligns closely with the Flora of China approach, differing only in the recognisiton of A. salouenensis at species rank, which is not an unreasonable approach given its distinctiveness.

In light of all that is now known about this group, including the nomenclatural turmoil already aluded to, the simplicity of Debreczy and Rácz’s approach holds considerable appeal, and we have elected to adopt it in this account for Trees and Shrubs Online. Only the status of a fifth taxon, A. chensiensis subsp. yulongxueshanensis Rushforth, remains unclear. In describing this Rushforth recircumscribed material that had previously been assigned to A. (chensiensis subsp.) salouenensis (Rushforth 1984), so until this taxon is further clarified we will follow the Flora of China in referring it to A. salouenensis.

Two other Chinese firs are related to A. chensiensis but are virtually unknown in cultivation and rarely discussed in connection with Shensi Fir. These are A. beshanzuensis and A. ziyuanensis. When considered distinct species (the latter is sometimes considered a variety of the former) each is a critically endangered relict, clinging to existence on a few isolated mountain tops. Farjon & Rushforth (1989) classified both in Section Momi along with A. chensiensis and its relatives discussed above, and each shows affinities with that species and with A. firma of Japan (Farjon 2017). Xiang et al. (2018) uphold this placement, but neither A. beshanzuensis nor A. ziyuanensis was considered in the study of Shao & Xiang (2015).

Abies delavayi and its relatives

Abies delavayi was western botany’s first introduction to the extraordinary diversity of firs in south west China, and as the first of these to be described to science it is a name that has stood the test of time. Unfortunately, it has also become much confused, in gardens, in herbaria, and in literature. Franchet published the name in 1899 based on material Delavay had gathered on the Cangshan, near Dali in Yunnan, in 1884 (Farjon 2017). Over the next few decades further collections of ‘blue-coned’ Abies were introduced from southwest China and it soon became clear many of these represented hitherto unnamed entities. The next to be named was A. forrestii in 1919, but at the time of its discovery in 1910 Forrest himself referred it to A. delavayi, not recognising its distinctiveness.

In 1932 Bruce Jackson reduced A. forrestii and A. faxoniana (now A. fargesii var. faxoniana q.v.) to varieties of A. delavayi (Jackson 1932) and thus began an unfortunate tradition of using the latter as a taxonomic dumping ground. Into it would be sunk so many taxa that most modern taxonomists now consider distinct species (Grimshaw & Bayton 2009). This is one reason that names like A. delavayi, A. fabri, and A. forrestii continue to send shivers down the spines of the uninitiated – their undeniable affinities are another. This taxonomic maelstrom has left an unfortunate legacy in that most early and mid-20th century literature on the subject is abjectly unhelpful, and unless the reader is an ardent specialist they would be well advised to discount anything written prior to 1984.

One reason for proposing such a clean sweep – and dismissing the works of so many august names – is that the reality is far simpler than historic literature would suggest: A. delavayi and its close relatives (treated here as A. fansipanensis and A. nukiangensis, but these are often reduced to infraspecific rank in western treatments) can all instantly and reliably be distinguished from nearly all other firs – including all the other blue-coned firs of the Himalaya and south west China – by their prominently revolute needle margins, giving the needles a distinctive, narrow appearance. In A. delavayi and A. fansipanensis this character can be so extreme that the margins partly obscure the stomatal bands on the needle undersides. Only with A. densa is there potential for confusion, but that species differs in its yellowish-brown or occasionally pale reddish-brown shoots (cf. rich dark reddish-brown or maroon) and in its more prominently ridged and grooved shoots (cf. comparatively smooth) (Rushforth 1987; Fu et al. 1999; Farjon 2017). Some sources suggest A. fabri can approach A. delavayi in this character but cultivated trees of A. fabri have weakly recurved needle margins at best, not revolute (pers. obs.). It is also important to note that poorly preserved herbarium specimens can exhibit recurved needle margins in species that would be quite flat in their living state, and care must be taken when examining such material (this problem may be one reason so many herbarium-based literature sources suggest A. fabri has weakly revolute needle margins, when in fact living plants do not).

Throughout these accounts only collections whose identities have been confirmed will be discussed, and the same principle is applied to trees of unknown origin. This is why there are only scant references to historic trees.

In reality, then, the A. delavayi complex extends only to three well defined entities: A. delavayi itself, A. fansipanensis, and A. nukiangensis. These three entities are all treated at species rank here following the Flora of China, but the alternative approach, of placing the latter two taxa below A. delavayi, is reasonable and probably more familiar in western treatments. The latest contribution to the debate on how best to treat these entities combines morphological and molecular investigative techniques and concludes that they are best regarded as belonging to one widely distributed, variable species, viz. A. delavayi s.l. (Shao et al. 2020). Interestingly Shao et al. (2020) regard A. fabri as belonging to the same affinity, despite its discreet distribution in Sichuan and its morphological distinctiveness: A. fabri differs quite consistently in its paler shoot colour, needles with only slightly recurved margins, and more pectinate foliage (Rushforth 1984). Whilst it may be closely related, it is unhelpful to think of A. fabri as part of the A. delavayi complex, which as defined here has quite a large distribution from the Hengduan mountains in northwest Yunnan, including the locus classicus of the Cangshan (A. delavayi); south through the Mekong and Salween valleys to the Gaoligongshan and northern Myanmar (A. nukiangensis), to a disjunct population in northern Vietnam (A. fansipanensis) (Debreczy & Rácz 2011; Fu et al. 1999); from this great sweep it extends westward into the mountains of Arunachal Pradesh, as A. nukiangensis, but where and how in this region it gives way to A. densa is poorly understood. (A. fordei q.v. is more or less restricted to the northern side of the Yarlung Tsangpo / Brahmaputra watershed and is amply distinct from both A. densa and A. delavayi / nukiangensis (Rushforth 2009) but whether it is the sole fir in this area is unclear. See also discussion under ‘A. delavayi agg. in Arunachal Pradesh’).

Further circumscriptions of the A. delavayi complex may lie in wait in the future, especially as its westernmost enclaves in northern Myanmar and Arunachal Pradesh become better studied. Future research may yet bear out Shao et al’s findings (Shao et al. 2020) and justify the placement of the taxa fansipanensis and nukiangensis at infraspecific rank. Were this to be the case, their discreet distributions suggest the rank of subspecies would be most appropriate, and the late John Silba made the appropriate combinations available in a 2008 paper (ipni.org).

A variety, A. delavayi var. motuoensis, was described from south east Tibet in 1975 and this is often encountered in literature. This is an obscure taxon, barely known outside China, and its true allegiance may lie with another species entirely – it is discussed under ‘A. delavayi agg. in Arunachal Pradesh’.

Abies delavayi and its close relatives can all be distinguished by their conspicuously revolute needle margins, evident here on a tree of P&M 89 growing at Howick, Northumberland. Only A. densa can approach A. delavayi in this regard, but it can be distinguished by shoot characters. Image Tom Christian.

Abies forrestii and its relatives

Just as with Abies delavayi, A. forrestii is surrounded by a suite of closely related taxa that have been variously treated at species and infraspecific rank. Charles Coltman-Rogers described A. forrestii in 1919; it had been introduced some years earlier (as A. delavayi) but it would be several years until its distinctiveness was appreciated. Like A. delavayi it has glabrous shoots, but it differs most obviously in its needles with perfectly flat margins, and in its larger seed cones with bracts only partially exserted at maturity. It also has a discreet distribution (Farjon 2017; Debreczy & Rácz 2011).

In 1929 Viguié & Gaussen described a variety, A. forrestii var. smithii, differing from the type in its hairy shoots and conspicuously exserted bracts with an abrupt, long cusp. This has never been given species rank in any subsequent revisions, although some authors, notably those in the Flora of China, give priority to the pubescence and thus treat it under A. georgei rather than A. forrestii (Farjon 2017). Orr described A. georgei in 1933, again with hairy shoots and prominently exserted bracts, but this time with the bract scale tapering into the cusp (Orr 1933). Next, Bordères-Rey & Gaussen described A. ferreana in 1947, differing in its very dense and long pubescence, medial resin canals, and shorter needles (Bordères-Rey & Gaussen 1947), and the next year, 1948, they named A. rolii and A. yuana (Bordères-Rey & Gaussen 1948), by this point differentiating species based on very minor variations. Some years later, W.C. Cheng & L.K. Fu described A. chayuensis in 1975 ‘similar to A. ferreana’ but differing in shoot colour and in minor vegetative characters such as needle length (Cheng, Fu & Cheng 1975).

These many names, all of them imperfectly known and inconsistently used, were rationalised to some extent by Farjon & Silba in a 1990 Phytologia paper in which A. chayuensis and A. ferreana were reduced to varieties of A. forrestii, and in which A. rolii and A. yuana were synonymised with this new interpretation of A. ferreana. These revisions would form the basis of Farjon’s approach to this group of firs in his 1990 Pinaceae monograph (Farjon 1990) in which he made further revisions by reducing A. georgei to a variety of A. forrestii, and by placing the name A. chayuensis into synonymy with his concept of A. ferreana.

In so doing, Farjon (and Silba) lumped within their interpretation of A. forrestii an extraordinary degree of diversity in what are generally considered key characters, even by the standards of this complex genus. This is why other treatments, most notably those of the Flora of China (Fu, Li & Mill 1999) and Conifers Around The World (Debreczy & Rácz 2011) differ so markedly from the treatments of Farjon. Farjon’s interpretation of the forrestii complex in his 2017 handbook is largely unchanged from his Pinaceae monograph; he continues to recognise A. forrestii var. forrestii, A. forrestii var. ferreana, A. forrestii var. georgei, and A. forrestii var. smithii (Farjon 2017).

The Flora of China account recognises A. forrestii s.s., affords both A. ferreana and A. georgei species rank, treats the entity smithii as a variety of A. georgei, and continues to recognise A. chayuensis, differentiating it from A. ferreana on the basis of shoot colour and a discreet distribution in SE Xizang (Tibet). An additional infraspecific taxon, A. ferreana var. longibracteata L.K. Fu & Nan Li, is also recognised (Fu, Li & Mill 1999).

Debreczy and Rácz err closer to the Flora of China treatment by recognising A. forrestii as a variable species but with no infraspecifics, treat A. ferreana and A. georgei as species, recognise A. ferreana var. longibracteata and A. georgei var. smithii, but do not commit to the validity or otherwise of A. chayuensis, having never seen it in the wild (Debreczy & Rácz 2011).

When he described A. fordei in 2009, Rushforth suggested the Flora of China approach to the forrestii group is ‘reasonable IF [sic] priority is given to the pubescence’ (Rushforth 2009), a position he maintains, and indeed is ‘coming around to’ in 2020 (K. Rushforth pers. comm. 2020). As was explained in the genus article, vegetative characters are afforded priority here because they are easy to assess in cultivation, therefore we have elected to follow an emerging consensus on this group of closely related firs by treating them thus: A. forrestii Coltm-Rog.; A. ferreana Bordères & Gaussen; and A. georgei Orr.

Were Viguié & Gaussen’s var. smithii to be recognised it would sit within this concept of A. georgei (viz. A. georgei var. smithii (Viguié & Gaussen) W.C. Cheng & L.K. Fu), but here we argue that it does not merit taxonomic disctintion and it is reduced to synonymy of A. georgei.

A. ferreana var. longibracteata L.K. Fu & Nan Li was described in 1997 and is here treated as a synonym of A. georgei – see that article for discussion.

The name A. chengii is usually encountered in connection to A. forrestii. This is afforded full treatment here, but in line with an emerging consensus we are amending it to a nothospecies: A. × chengii Rushforth. See that article for a full discussion.

A. chayuensis W.C. Cheng & L.K. Fu is probably a good species, but it is doubtful whether it is in cultivation.

Of those taxa that are in cultivation several remain poorly understood; their relationships and the diversity inherent within them are still shrouded in uncertainty. A. ferreana Bordères & Gaussen provides perhaps the best example: Bordères-Rey & Gaussen (1948) differentiated A. rolii (from the Mekong-Salween divide) and A. yuana (from Chungtien) based on relatively minor characters, and in their publications of all three names (1947; 1948) they provide tabular comparisons between them, and with A. forrestii and A. georgei, but they do not always compare like with like, and many of the vegetative characters described were taken from coning shoots, from whence those characters are known to be atypical and variable. This is at least partly why most modern works consider vegetative characters from sterile shoots only, making comparison between these and Bordères-Rey & Gaussen’s original texts difficult.

Another frustration is that in each case Bordères-Rey & Gaussen based their new species on a single specimen: beyond assigning a holotype to each name no other collections are cited, making it all but impossible to contextualise these ‘species’ with the diversity that exists in wild populations. It is telling that Farjon (1990; 2017), the Flora of China (Fu, Li & Mill 1999), and Debreczy & Rácz (2011) come into rare alignment by considering both A. rolii and A. yuana synonymous with A. ferreana. Nevertheless, several introductions to cultivation, though clearly allied to the A. forrestii aggregate, do not comfortably fit any of the known and widely agreed taxa (A. ferreana, A. forrestii, and A. georgei) nor do they fit A. chayuensis – itself not well understood. These anomalies include Yu 14627 from Muli, and H&M 1447 from Mt Luoji. Until the A. forrestii aggregate is subject to greater scrutiny, they may best be labelled ‘A. forrestii agg.’.

Abies fargesii's flaking bark quickly separates it from A. forrestii, A. ferreana and A. georgei, which it can otherwise resemble in foliage characters. Image Tom Christian.