Catalpa bungei C.A.Mey.

Catalpa bungei occupies a basal position in sect. Catalpa, its lineage having diverged at an early date from the clade containing the two North American catalpas and the other Chinese species, C. ovata (Li 2008). From all of these, C. bungei is distinguished by its pink flowers, racemose or corymbose (not paniculate or thyrsoid) inflorescences. The large pink to rosy-purple flowers and elegantly tapering leaves make it a tree of great distinction, accounting for its widespread cultivation in China, particularly in temple grounds (Elwes & Henry 1906–1913). Valder (1999) includes a photograph of a fine specimen in the Imperial Garden of the Forbidden City, and it was probably in this vicinity that the tree was first seen and collected in 1831 by Alexander von Bunge (1803–90), who noted that it was frequently found in gardens under the local name ‘stinking tree’ (arbor foetens – for the foetid leaves (von Bunge 1835)). Having apparently seen only cultivated plants, von Bunge queried whether the tree was indigenous to China, misidentifying it as the North American C. syringifolia Sims (= C. bignonioides, the only temperate American species then known). It was C.A. Meyer (1837) who noticed the distinctive unbranched inflorescence structure of von Bunge’s specimens and named the new species for its original collector.

The early-twentieth-century collector Frank Meyer hailed C. bungei as ‘one of the finest flowering trees in the world’, but the species is variable, and has not always elicited such admiration. Part of the problem lies in the fact that the species – in all its diverse phenotypes – is simply not well enough known in Western horticulture, being both under-represented in collections and persistently misidentified. It is salutary to realise that, as late as 1894, when Louis Édouard Bureau prepared his revision of Catalpa, the species had never flowered in France – his observations of the flowers were based on notes by the Abbé David, and a Chinese painting of a flowering branch sent to him by a correspondent (Bureau 1894). There is no question that the species deserves to be much more widely planted.

As currently circumscribed, C. bungei encompasses a number of pink(ish)-flowered taxa formerly treated as distinct, of which three – C. bungei sensu stricto, C. fargesii and C. duclouxii – have at various times enjoyed broad acceptance as species. The extensive and historically rather confused synonymy reflects the species’ wide geographical range – a broad swathe of the Yellow River basin across central and eastern China – as well as the degree of natural variation occurring within it. Taxonomic distinctions within this broad species were formerly made on the basis of the type and density of pubescence, or (in Dode’s case) inflorescence architecture, though in both cases the variation in question has been found to be a continuous character, and not a sound basis for taxonomic distinctions (Rehder in Sargent 1913, Olsen & Kirkbride 2017). Catalpa duclouxii was demoted to a form of C. fargesii by Gilmour (1936); Olsen and Kirkbride (2017) finally reduced all of these taxa to synonymy with C. bungei following extensive examination of specimens at three Chinese herbaria (in Beijing, Nanjing and Kunming).

Though all recent studies agree on the appropriateness of treating this group as C. bungei sensu lato, a significant degree of genetic differentiation has been detected within the broadly conceived species. This (often silent) variation may not justify the erection of infraspecific divisions, but reflects the continued playing-out of large-scale evolutionary processes. Li (2008) found C. bungei s.s. and C. fargesii to be most closely related, with C. duclouxii as a sister group, suggesting a distinct genetic division between them. This interpretation was broadly confirmed by the higher-powered analysis of Dong et al. (2022), though that study also detected a high degree of confusion in the inheritance patterns of the approximately 5000 genes analysed. While upholding the treatment of the three taxa as C. bungei s.l., Dong et al. raised the possibility that C. bungei s.s. may be of hybrid origin, with the other two taxa implicated in the parentage. Further molecular data will doubtless shed light on these complex relationships.

As a valuable native timber tree, C. bungei has been the object of sustained attention on the part of Chinese researchers, from the perspectives of both agroforestry and ecological restoration and management (Jian et al. 2022). It is notable that the systematic treatment of the species in Chinese horticultural science differs from both the broad approach adopted by Olsen & Kirkbride (2017) and the finer resolution of the Flora of China (Zhang & Santisuk 1998) in continuing to recognise discrete infraspecific taxa. For instance, Xiao et al. (2019) treat C. fargesii f. duclouxii as a distinct (northern) population with a distribution on the Yunnan-Guizhou plateau, in contrast to the southern C. bungei s.s. centred on the Yellow and Yangtze River region. Zhang et al. (2023) likewise distinguish between C. bungei and C. duclouxii. The evident practical utility of such population-based infraspecific concepts indicates that there is much still to be learned by westerners about C. bungei’s distribution, biogeography and variation: this remains a poorly understood species outside China.

Not only does the degree of vestiture in C. bungei represent a complete continuum from glabrous to densely pubescent, but leaves across the species are highly variable in size and shape, even varying on the same tree. Indeed, the author of C. bungei, C. A. Meyer, described C. bungei var. heterophylla simultaneously with the type variety, now synonymised (Meyer 1837). Having observed natural populations, Augustine Henry reported the leaves of old trees as being ‘entire or two- to three-lobed’, but dentate-margined on branches terminating in an inflorescence, the latter form being predominant on young trees (Elwes & Henry 1906–1913). Zhang et al. (2023) obtained a remarkable variety of leaf shapes – from entire to thirteen-lobed, and in varying aspect ratios – in the F₁ progeny of crosses between two individuals from different C. bungei cultivars (one of which was referred to C. duclouxii), a phenomenon that appears consistent with the hybrid origin posited by Dong et al. (2022).

As a wild tree, Catalpa bungei has a wide distribution in upland areas of China and the Yellow River basin (Jian et al. 2022). Henry observed it growing in Hubei, in ‘mixed forests of deciduous trees’ (Elwes & Henry 1906–1913), and Wilson collected it (under the names C. fargesii and C. duclouxii) in western Hubei and Sichuan (Sargent 1913). Joseph Rock gathered cuttings at 2865 m asl in the Likiang valley, western Yunnan in 1922 (as C. duclouxii 3217, PI 55931), noting that ‘at this altitude the plants do not bear seed’, though he also gathered seed from trees growing on the Likiang plain at 2680 m asl (Office of Foreign Seed and Plant Introduction 1923).

To this day not enough is known of the natural history and distribution of this species, partly on account of the linguistic barrier that limits Western botanists’ ability to profit from the extensive literature on Catalpa in Chinese journals. One important question that needs to be better understood is the relationship between cultivated plants and wild populations. It is significant that, beginning with von Bunge’s first sighting of the tree in the temple yards of Beijing, westerners have encountered the species as cultivated or (in the case of field pollards, coppice or secondary-growth forest) as semi-cultivated trees, gaining perhaps only a partial picture of the tree’s range and characteristics. Perhaps this is one of the reasons why the synonymous taxa were maintained for so long as species, as Grimshaw and Olsen (2011) hinted with respect to C. duclouxii and C. bungei. The tree’s famed shyness to produce seed (in contrast to C. ovata) is also relevant in this respect, since the tree appears to be largely propagated by clonal propagation from cuttings or suckers – see the instructive first-hand account of the domestic cultivation of C. bungei on a farmstead in Yunnan in Olsen and Kirkbride (2017). When fruiting does occur, the seed is a valuable commodity for commercial distribution, the tree being a source of valuable timber: Rock records that seed was collected from trees in western Yunnan and centrally distributed (from Tengyueh, modern Tengchong), suggesting that the distributions and phenotypes encountered by Western botanists may not reflect natural ranges (Office of Foreign Seed and Plant Introduction 1923).

The circumstances of C. bungei’s introduction to Europe pose a number of historical problems. Writing in 1885, Lavallée (whose testimony is to be taken seriously) considered that the true species had still never been introduced to Europe (Lavallée 1885). Nevertheless, on several occasions in the late 1840s and 1850s, Parisian gardeners reported having received C. bungei from consignments of seed from Eastern China, often provided through diplomatic or governmental channels (Masson 1854). The thriving young, as yet unflowered plants that resulted gave rise to reports that the species had been successfully propagated (Jacques 1855), but on coming into flower these ‘C. bungei’ (or ‘C. bungeana’) specimens invariably turned out to be the yellow-flowered C. ovata (Grimshaw & Olsen 2011). Nevertheless the misconception that C. bungei was in cultivation in France took hold, and a date of introduction of 1848 is often found in the literature; an even earlier date of 1838 given by the usually well-informed Charles Baltet (1895) can also not be substantiated. The suspicious remark of one German visitor to the Jardin d’acclimatation during the 1867 Paris Exposition, that an abundantly shooting specimen of ‘Catalpa bungeana’ looked remarkably similar to Catalpa kaempferi (= C. ovata) (Beer 1869) demonstrates that the confusion had now become ingrained; before long it had extended into Europe more widely and to America (Hooker 1882, Elwes & Henry 1906–1913, Grimshaw & Olsen 2011).

The consequences of this misidentification continue to reverberate, part of a nomenclatural tangle that eventually implicated not two, but three Catalpa taxa (see the entry for C. bignonioides ‘Nana’, the ‘Catalpa bungei’ of the horticultural trade). Nevertheless, the mid-nineteenth-century Parisian gardeners may be forgiven for their certainty that they had raised two separate Catalpa species. Given the morphological variability within the genus, particularly with respect to leaf shape, it is not hard to imagine how contrasting phenotypes of C. ovata might have been attributed to two different species, particularly if the original seed had been mislabelled. Even for modern horticulturists it is not easy to tell young catalpa seedlings from one another by vegetative characters alone. The foetid (‘datura-like’) leaf odour of one of the spurious ‘C. bungei’ seedlings (Jacques 1855, Quihou 1865) also proved an unfortunate red herring: C. bungei’s malodorous reputation was well known, but what was not then realised was that some phenotypes of C. ovata also have foetid leaves (Dode 1907, J. Grimshaw, pers. obs. 2025).

Still, the failure of this most beautiful of ornamental trees to reach Europe at a time of great botanical curiosity and energetic plant importation does seem remarkable. The reasons are no doubt complex, but it is worth noting the poor condition in which much Chinese seed arrived in France, most of which failed to germinate, as Jacques (1851) complained at the time. And, of course, C. bungei seed was simply scarcer than that of the copiously fruiting C. ovata (which – a temple tree like C. bungei – would be easy to harvest): the reluctance of C. bungei to set seed is now recognised, and this characteristic is likely to have been still more marked in cultivation, when trees might well have been pollarded (Grimshaw & Olsen 2011).

In fact, the earliest documented introduction of C. bungei from eastern China – astonishingly – was not until 1904, when seeds from the Beijing district were sent by the diplomat E. T. Williams to the Arnold Arboretum, where a tree from this consignment (Acc: 12927) still flowers prolifically (Olsen & Kirkbride 2010). Frank Meyer collected it in the same area in 1905, from the Yellow Temple, North of Peking (PI 16914) (Cunningham 1984). Material deriving from collections from this area remains extremely scarce.

In Western cultivation the tree is most commonly encountered as plants deriving from collections from western China during the decades either side of 1900, and introduced under the names C. fargesii or C. duclouxii. Trees from these western populations were identified by early collectors as a showier form of pink-flowered catalpa with inflorescences that were distinctly fuller than those of C. bungei, and with larger flowers. The first herbarium specimens from this region were collected by Augustine Henry in Sichuan in 1888 (Henry 5856a) (O’Brien 2025), and further examples were collected by Paul Farges in Sichuan (Farges 495).

Bureau based the new species Catalpa fargesii on Farges’ material, distinguishing it from C. bungei not only by the fuller inflorescence, but also by the branched or stellate hairs on most parts of the plant, though Bureau’s emphasis on vestiture proved to be misplaced – Olsen and Kirkbride (2017) demonstrated that pubescence must be seen as a continuous character in this broad species. The glabrous Catalpa duclouxii, recognised as distinct by Dode on the basis of its more complex, branched inflorescences, has for more than a century been generally held to be inseparable from C. fargesii, from which it differs in its vestiture alone (Rehder in Sargent 1913); it was reduced to a forma of that species by Gilmour (1936). At the other end of the scale, particularly densely pubescent individuals may have been ascribed by collectors in the field to C. vestita Diels on the strength of the epithet alone, though in fact Diels’ species was explicitly based on a fewer-flowered inflorescence than typical C. bungei (Diels 1901). The group of ‘Catalpa vestita in full flower’ encountered by Ernest Wilson in northwestern Sichuan (near Songpan) in early June 1910, sounds closer to Bureau’s C. fargesii, with their inflorescences ‘cylindrically branched, with flowers equal in size to C. bignonioides’ (Wilson 1910). Rock’s 1922 collections 3097 and 3217, from Yunnan, were described as having flowers ‘borne in full panicles’ (Office of Foreign Seed and Plant Introduction 1923).

Nevertheless, although there evidently exist showy forms of C. bungei of sufficient horticultural distinction to have attracted the attention of early collectors, the species’ almost constant variation in a broad suite of characters makes it untenable to recognise these forms as taxonomic groups, though the concept of both a Duclouxii Group and a Fargesii Group is sometimes encountered (e.g. Lancaster et al. 2019). Rather than attempting to circumscribe arbitrary groups of cultivated trees, it is probably more informative simply to be aware of trees’ diverse provenances. In British and American contexts, the forms of C. bungei most likely to be encountered derive from material distributed under Wilson 640 – in fact collected from two different trees more than 700 km apart, one from Qing-shen (Ching-chi Hsien) in western Sichuan (12–16 m, flowers ‘very light purple with yellow throat’), the other from Fang-xian (Fang Hsien) in western Hubei (13–16 m, flowers ‘rose-pink with orange markings in throat’). Rehder (in Sargent 1913) distinguished the Sichuan W 640 by its fuller infloresences (the lower branches bearing 2–3 flowers), in contrast to the simple lateral axes of the Hubei W 640. This material was collected as C. duclouxii, but Rehder noted that it had been distributed under both synonyms, C. duclouxii and C. fargesii. Material from one of these W 640 collections growing at the Royal Botanic Garden, Edinburgh (Acc.: 20071778, as C. fargesii f. duclouxii) has glabrous leaves of variable shape and lobing, flushed red on first emergence. Plants grown under this collection number elsewhere (for instance at Valley Gardens, Surrey) have entire leaves, with pubescent undersides. Flowers on these plants are generally pale pink, with a lightly red-dotted corolla and hairy pedicels. Veitch’s catalogues from 1911 and 1913 advertise a form of C. fargesii (five-lobed, glabrous, deep-green leaves, purportedly with pale purple flowers); C. vestita (purportedly rosy-pink) is listed in 1913.

Other Wilson collections under C. duclouxii include Wilson for Veitch 976, collected in 1900 from eastern Sichuan, which is probably the material introduced by Veitch under the name C. fargesii before 1904 (Veitch 1904), the names interchanged as would happen subsequently with W 640. Wilson for Veitch 4289, collected in 1904, is said by Rehder to have ‘flowers white, suffused and spotted with pink’ (Rehder in Sargent 1913), and the fuller, more complex inflorescence-type. Wilson’s vast archive of images at Harvard University includes a photograph of C. duclouxii growing as a field tree at approx. 900 m asl in a cultivated terrace near Yunyang Hsien, Eastern Sichuan. The tree has an ascending, narrow, rather open crown to approx. 21 m, with two codominant stems forking at about 2 m above ground (girth 1.8 m) (photograph O-192; 2 July 1910); it is not in flower, and does not correspond to any of Wilson’s known collections A mass of juvenile suckers beneath the tree bears leaves with 5–7 lobes leaves, though the mature foliage on the tree itself is 3-lobed or entire.

Wilson’s collections under C. fargesii, all from western Hubei, are W 636 (‘rosy-purple’ or ‘pale purple’, 1907) and W 748 (‘bright rosy-purple’ flowers, 1907) and W 4556 (‘rosy-pink’ flowers, 1910) (Rehder, in Sargent 1913). A photograph of W 636 taken in May 1907 shows a shapely ascending tree (21 m, girth 1.8 m) with a slender crown, growing at 1100 m asl at the upslope edge of woodland bordering a cultivated field near Fang Hsien; there is what appears to be at least one smaller suckering trunk growing just outside the canopy (photograph Z-703).

A number of botanical institutions grow material originating from Kunming Botanic Garden, Yunnan, among them the United States National Arboretum, whose plants – received as seed, under C. duclouxii – are described as bearing ‘masses of showy pink flowers in spring’ (Acc: 53522). The leaves are heart-shaped, hand-sized, with the surface almost glossy. Two other specimens from the same source have reached 8 m and 10 m (Grimshaw & Olsen 2011). Curiously, in Washington, D.C. (USDA zone 7a) these trees exhibit a semi-evergreen habit, losing older leaves as winter progresses. At Strasbourg Botanic Garden a tree with this provenance has striking corymbose inflorescences in which the corolla lobes are densely purple-dappled to the margins, quite different in effect from the more pallid flowers of material of W 640 origin, or of the E. T. Williams collection at the Arnold Arboretum (Acc. 12927).

Collections from northern China are scarcer. This region seems to have been the preserve of Frank Meyer, who introduced material from Wangyuko, Shanxi in 1914. Interestingly, this collection (PI 38254) was imported into the United States as ‘young rooted trees’ (United States Department of Agriculture 1916) – consistent with the customary means of propagation from root suckers (as Meyer noted), but also supplying interesting context for the previously noted absence of true C. bungei seed in Paris in the mid nineteenth century.

Opinions are mixed on the merits of C. bungei as specimen trees, at least insofar as concerns the forms generally grown in British gardens, whose beautiful pale pink flowers are exceptionally attractive and can be borne in large numbers, but are apt to be ‘rather gaunt’, and ‘narrow-crowned’ in habit (Bean 1976). Describing C. bungei as ‘the finest June-flowering tree for temperate regions’, the late Nigel Muir summed up the situation quite well: ‘Forty to fifty years ago there was an exceptionally beautiful specimen at Kew which was much superior to any other trees of the species growing there. […] Like so many splendid trees it has been little used for any purpose, especially urban planting. It was rare half a century ago and it is still about as rare today. Catalpa fargesii could make a very good and interesting street tree. It has a fine upright habit for this use and the leaves are not too large to be at all oppressive’ (quoted in Grimshaw & Olsen 2011).

Together with C. speciosa, C. bungei represents the northern limit of the Bignoniaceae, a family that has overwhelmingly diversified in the tropics and subtropics, having made few inroads into higher latitudes. As one would expect, it is a hardy species, though one might expect a degree of variation given the species’ wide latitudinal range. Trees grown in Belgium under the name C. duclouxii have fallen victim to killing frosts in late spring, though the tree appears to be winter hardy when the wood is ripened (Grimshaw & Olsen 2011). Frank Meyer emphasised the need for ‘quite hot’ summers and winters that are ‘not too severe’, noting that the tree ‘prefer[s] a porous soil [and] thrives best when planted close to irrigation canals and on sheltered places’ (United States Department of Agriculture 1916).