Malus Mill.

TSO logo


Kindly sponsored by
Francine von Finck: after many informative Tours and Study Days with the I.D.S I feel it only fitting to help and promote such a wonderful organisation.


Julian Sutton (species), Nick Dunn (cultivars)

Recommended citation
Sutton, J. & Dunn, N., 'Malus' from the website Trees and Shrubs Online ( Accessed 2021-06-19.


  • Rosaceae

Common Names

  • Apples
  • Crab Apples
  • Crabs


Reproduction without fertilisation usually by the asexual production of seeds (agamospermy) (as in e.g. Citrus Sorbus). Includes vegetative reproduction (stolons rhizomes suckers etc.) (as in e.g. Ulmus).
(sect.) Subdivision of a genus.
Taxon that reproduces only or regularly by apomixis.
Reproduction without fertilisation usually by the asexual production of seeds (agamospermy) (as in e.g. Citrus Sorbus). Includes vegetative reproduction (stolons rhizomes suckers etc.) (as in e.g. Ulmus).
(in Casuarinaceae) Portion of branchlet between each whorl of leaves.
Made up or consisting of two or more similar parts (e.g. a compound leaf is a leaf with several leaflets).
Innermost layer of the fruit wall. Can be membranous and indistinguishable from the other layers of the fruit wall or may be hard woody and distinctive (see drupe).
IUCN Red List conservation category: ‘there is no reasonable doubt that the last individual [of taxon] has died’.
Plant originating from the cross-fertilisation of genetically distinct individuals (e.g. two species or two subspecies).
Cup-shaped or tubular structure at the base of a flower (‘floral cup’) formed by enlargement of the receptacle and/or the bases of the floral parts.
(botanical) Contained within another part or organ.
Flower-bearing part of a plant; arrangement of flowers on the floral axis.
(of hybrids) Formed by fertilisation between species of different genera.
(of fruit) Vernacular English term for winged samaras (as in e.g. Acer Fraxinus Ulmus)
Compartment of the ovary. loculicidal (of dehiscent fruit) Splitting between the locules. (Cf. septicidal.)
Species distinguished on the basis of minute differences of morphology. Generally used only for species that reproduce via apomixis (e.g. Sorbus).
(of a group of taxa) With a single ancestor; part of a natural lineage believed to reflect evolutionary relationships accurately (n. monophyly). (Cf. paraphyly polyphyly.)
Lowest part of the carpel containing the ovules; later developing into the fruit.
Small grains that contain the male reproductive cells. Produced in the anther.
Act of placing pollen on the stigma. Various agents may initiate pollination including animals and the wind.
Fleshy fruit with leathery core. Typical of Rosaceae subfamily Maloideae (e.g. Malus).
Central axis of an inflorescence cone or pinnate leaf.
Generally an elongated structure arising from the ovary bearing the stigma at its tip.
(var.) Taxonomic rank (varietas) grouping variants of a species with relatively minor differentiation in a few characters but occurring as recognisable populations. Often loosely used for rare minor variants more usefully ranked as forms.



Julian Sutton (species), Nick Dunn (cultivars)

Recommended citation
Sutton, J. & Dunn, N., 'Malus' from the website Trees and Shrubs Online ( Accessed 2021-06-19.

A genus of very approximately 30 species, native to Asia, Europe and North America. Trees or shrubs, deciduous or rarely semi-evergreen, lateral shoots sometimes thorny. Leaves simple, alternate, unlobed or lobed (sometimes both on the same tree), margin toothed; stipules present, usually falling early. Inflorescence a corymb or raceme. Flowers bisexual, stalked, normally in spring; calyx 5-lobed, usually persistent; petals 5, variously rounded, white, pink or red; stamens 15–50; ovary inferior, with (3–)5 locules; styles 2–5, joined at base. Fruit a variably rounded, fleshy pome, sometimes with stone cells; green, yellow or red, with 1 or 2 seeds per locule. (Bean 1981Gu et al. 2003Dickson 2015Cullen et al. 2011Huxley et al. 1992).

In gardens, the genus Malus presents three faces. There are the orchard apples, M. domestica, which we treat only briefly since, although often extremely attractive, they are primarily grown as fruit rather than ornamental or landscape trees, and therefore beyond the scope of Trees and Shrubs Online – for those interested in orchard apples, the M. domestica article offers some suggestions for further reading. Then there are the ornamental crabs of gardens, mostly hybrids developed over the last two centuries in western gardens, although there is a much longer tradition of ornamental Malus in East Asia. We describe these in the pages ‘Malus Cultivars A-B’ etc., although those which are clearly selections of wild species can also be found under the appropriate species. General comments on cultivation, propagation, and diseases later in this article apply as much to ornamental hybrids as to any other apples. Finally there are the wild Malus species, taxonomically difficult, remarkably under-researched and therefore poorly understood. With some important exceptions, they are trees of arboreta and botanic gardens rather than general horticulture. Our aim with these is to avoid giving a false sense of orderly understanding; rather to expose uncertainty and dubious received wisdom, as much as to catalogue what little is certainly known.

Apples belong to a distinct group within the Rosaceae whose fruits are pomes. A pome is a type of accessory fruit (or ‘false fruit’) which develops from more than just the ovary (Esau 1977). The wall of the inferior ovary becomes the endocarp – the leathery core of an apple or the stony layer around a hawthorn (Crataegus) seed – while most of the fleshy outer layers are derived from the hypanthium at the base of the flower (Phipps et al. 1991). These pome-bearing plants have long been recognized as a group, but under different names at various taxonomic levels, from subfamily Maloideae to sub-tribe Malinae by way of tribe Pyreae or Maleae (Robertson et al. 1991; Juniper & Mabberley 2019). Significant genera include Amelanchier, Aronia, Chaenomeles, Cotoneaster, Crataegus, Cydonia, Docynia, Mespilus, Photinia, Pyracantha, Pyrus, all those genera once lumped together as Sorbus, as well as Malus itself. Malus is distinguished by the combined characters of the carpels being fused together and fully enclosed by the hypanthium, the styles being fused together at the base, and having 1–2 ovules (and hence seeds) per locule.

The balance of fossil and molecular evidence points to a great radiation in this group during the late Miocene (11.6–5.3 MYA), following a genome duplication which established the basic chromosome number as 2n=34 (Spengler 2019). Genera within this group are not always easy to delimit, and concepts have varied widely through the history of botany: witness the great diversity of plants once brought together in Sorbus, and the present difficulties in establishing more natural genera. Even the orchard apple was first named Pyrus malus by Linnaeus in 1753; two centuries later Chittenden (1951) was able to say that ‘the species of Malus are still often found in catalogues, &c., under Pyrus’. Even today there are those who would sidestep the problem by lumping all these genera into a vast, probably monophyletic, but quite useless concept of Pyrus (Christenhusz et al. 2018), surely a counsel of despair. Moreover, intergeneric hybrids are unusually frequent, suggesting weak genetic barriers to breeding: examples include × Pyronia, × Sorbopyrus, and × Sorbaronia. Phylogenetic studies of the group have been helpful (Li et al. 2012; Sun et al. 2018) but really thorough sampling of the diversity of each genus and all contentious species is still required.

Some of the trees and shrubs we treat here as Malus have sometimes been placed in the segregate genera Docyniopsis, Eriolobus, Prameles and Sinomalus, but for several of these there has been little consistency in which species are included; in addition, Rushforth (2018) suggests placing several North American species in Chloromeles, a validly published genus but in which no species currently recognized have been named. While molecular evidence can be marshalled in support of any chosen treatment, molecular studies taken together do not yet give a clear or consistent picture of relationships within Malus. They are, however, generally compatible with the idea that Malus in the wide sense (or at least Malus plus Docynia) is monophyletic (for example Li et al. 2012; Sun et al. 2018). Even Rushforth (2018), who would split Malus to an unusual degree, admits that keeping them all in Malus is a ‘logical option’. It is worth noting that apples seem not to have become involved in the free-for-all of intergeneric hybridization.

The wild apple species are in general poorly understood. Estimates of the number vary from 25 to well over 50. There are several reasons for this: paucity of research on wild apples, most of which do not have obvious value in commercial breeding, is a significant problem; the rather subtle and few characters which can be used to distinguish species compound the problem. The role of hybridization has scarcely been explored, but may well be very significant, while polyploidy and apomixis confuse matters considerably in some groups.

The experience of breeding orchard apples and ornamental crabs suggests that there are few if any barriers to breeding between apple species (Fiala 1994). Also, the open, bowl-shaped Malus flower with a mass of stamens pointing in all directions suggests adaptation to pollination by generalist insects. A rare example of research on the pollination of a wild apple, M. komarovii in north eastern China, found 46 species of insect visitor, including bees, flies, beetles and butterflies (Feng et al. 2015). Taken together this implies that hybridization is likely to occur when species meet. For example, modern M. domestica seems to be the result of domesticated M. sieversii, carried west from Central Asia on the Silk Road, meeting and hybridizing with M. sylvestris, while M. asiatica resulted from M. sieversii meeting M. baccata in East Asia (Cornille et al. 2012, 2014; Duan et al. 2017). The role of more ancient hybridization in the origins of the wild species has scarcely been explored, but some molecular studies hint at its importance (Volk et al. 2015).

Some Malus are apomictic. Apomixis is the production of seeds whose embryos develop from maternal tissue instead of through fertilization; offspring are then clones of the parent. Apomictic Malus are apogamous: that is, pollination and the second fertilization event which gives rise to the seed’s endosperm tissue but not to any part of the resulting seedling, are still required for seeds to be produced. Apomixis is common among pome-bearing Rosaceae, where it is associated with polyploidy. Tetraploid apomicts result from doubling of the chromosome complement in an individual of a sexual, diploid species or in a hybrid: apomixis allows that individual to spawn a population of clones – a microspecies. Triploid microspecies originate when a sexual diploid is fertilized by pollen from a tetraploid apomict. Swarms of very similar but distinguishable microspecies have long been known among whitebeams (Aria) in Europe (Rich et al. 2010). More recently, they have been recognized among rowans (Sorbus in the narrow sense) in Asia (McAllister 2005). This has been a horticulturally significant discovery, resulting both in the finding of new, garden-worthy, apomictic microspecies which can easily be clonally propagated by seed (see Sorbus frutescens), and in disruption when a familiar garden plant is found to be one of several horticulturally distinct microspecies in the orbit of a sexual diploid (see S. discolor, S. glabriuscula, S. pseudohupehensis). Studies of the implications of apomixis in Malus are in their infancy, but we can expect similar consequences here in due course. Apomixis is well known in the Chinese M. hupehensis (Dirr 2009) but has also been identified in other East Asian taxa, M. bhutanica, M. rockii, M. sikkimensis and M. toringo var. sargentii (Sax 1959), as well as some in the confusing North American complex around M. coronaria (Dickson 2015). It is still far from clear what any of this means about the status and relationships of these wild apples.

The fleshy pomes of Malus encourage seed dispersal by animals. Beyond anecdotal observations, there has been little study of this. Small-fruited species certainly attract birds in cultivation (Fiala 1994). The evolution of larger fruits is thought to have been driven by a switch to dispersal by mammals, perhaps extinct megafauna (Spengler 2019). Bears and horses are today the principal dispersal agents in wild M. sieversii in Central Asia (Juniper & Mabberley 2019).

This genus has never been given a thorough taxonomic revision with global scope, and desperately needs one. Alfred Rehder came closer than anyone in the early 20th century, a period when the flora of central and western China was first becoming known to western science. His work is scattered in the literature (but notably Rehder 1920 and in Sargent 1916), and reflects his tendencies as a taxonomic splitter. Moreover, he described many cultivated hybrids as nothospecies with latinized binomials; while in keeping with the practice of his times, this leaves an unhelpful legacy, since many of those plants might have been better treated as cultivars. Recent regional floras (Gu et al. 2003; Dickson 2015; Iketani & Ohashi 2001) attempt to make sense of the described species, but their inevitably limited scope and the differing taxonomic outlooks of their compilers do little to help build a unified picture. A modern flora of Central Asia, one important centre of apple diversity, is lacking. In the plethora of published specific names in Malus, many are very rarely seen, even in old literature, and in some cases it is hard to tell which ‘modern’ species they refer to. We list only the more frequently encountered and less ambiguous synonyms. Rather than pick one of the competing (and less than satisfactory) classifications of Malus, we choose to outline the diversity of the genus through a series of informal clans, more-or-less fuzzily defined as befits our collective ignorance.

The first of these clusters of species (‘Clan A’ in the key) includes those most readily recognizable as apples, M. domestica, its primary wild progenitor M. sieversii, and the European M. sylvestris, as well as M. kirghisorum and M. orientalis. Distributed from Europe to northern China, they have relatively large fruits (>1.5 cm diameter), calyces which remain firmly attached to the ripe fruit, and unlobed leaves. We also mention several other more dubiously distinct taxa in brief. The Chinese M. asiatica, M. prunifolia, and M. spectabilis have some features in common with this group and are perhaps of ancient hybrid origin.

A second ‘clan’ (‘C’ in the key) is centred on M. baccata and is native to East Asia and the Himalayas. These species have small fruits with deciduous calyces. Those with unlobed leaves (‘C1’) include M. baccata itself, M. halliana, M. hupehensis, M. rockii, M. sikkimensis, and M. spontanea. Those with at least some of their leaves lobed (‘clan C2’ in the key) are M. bhutanica, M. komarovii, M. toringo, and M. transitoria.

A final East Asia clan (‘D’ in the key, small-fruited with persistent calyces) is centred on M. yunnanensis; M. honanensis, M. kansuensis, M. ombrophila, and M. prattii belong here. Their flowers are superficially more hawthorn-like than others and tend to have a musty rather than sweet smell, while the rachis of the inflorescence is longer than in other groups. M. fusca from the Pacific coast of North America, has much in common with these.

The remaining North American species, which we treat as M. coronaria, M. angustifolia, and M. ioensis, form a very distinct group (‘clan B’ in the key) which is almost universally recognized as Section Chloromeles, although variation within this group is complex and incompletely understood. Several outlying species stand alone: M. tschonoskii from Japan; the subtropical Asiatic M. doumeri; and from the Mediterranean region M. trilobata and M. florentina, the latter sometimes thought to be an intergeneric hybrid.

The horticultural selections known crab apples, widely grown and admired for their abundant flowers and fruits, coupled with hardiness and other useful qualities, are of complex origin and differing characters. Even when a cultivar can be directly attributed to a species or nothospecies, we treat them all alphabetically under the heading Cultivars A-B, etc. Departing from the usual narrative style we have adopted a tabular format for crab apple cultivar entries, which enables rapid comparisons and access to the key points of distinction. These have mostly been prepared by Nick Dunn (some are by Julian Sutton) who is not only a nurseryman of distinction with exceptional knowledge of these cultivars, but a connoisseur of crab apple jelly, and many descriptions contain a useful comment on their suitability for this purpose

The distinction between ‘apples’ and ‘crab apples’ is a vague and purely horticultural one based on fruit size and sometimes flavour, crabs being those with relatively small and perhaps bitter or very sour fruit. On this basis almost all Malus except the majority of M. domestica and M. sieversii may be considered crabs. The terms as we use them are not mutually exclusive, and we are happy also to refer to any Malus as an apple.

Wild apples are primarily trees of open habitats; even when floras claim woodland as their habitats, they tend to be found in sparse areas, along margins and in clearings. It follows that in the garden they are best grown in full sun with good air movement. Although most tolerate dappled shade, fruiting and autumn colour will be best in sun. Most are very cold hardy, some to the extent that they are important ornamentals in the northern prairies of North America. As for so many plants, a fertile, moist, but reasonably drained soil is ideal, but most crabs will tolerate a wide range of conditions. Waterlogging is generally a step too far, as is salinity, although Malus fusca tolerates both in its wild habitats and should be tried much more widely in such places, grown on its own roots. With aesthetics rather than productivity the aim, winter pruning is generally about removing dead or diseased wood and encouraging the desired limb structure.

Self-incompatibility in orchard apples is well studied (OECD 2019 and references therein); fruit set in self-pollinated apples varies between cultivars, but is usually under 10%, and growers understand the need to plant more than one cultivar with compatible flowering times to ensure a good crop. What evidence there is points to a degree of self-incompatibility being the norm in wild apples, although the strength of incompatibility varies between species, and polyploidy can allow full self-fertility. The horticultural literature generally fails to address this as an issue in ornamental crabs. This may be because the desire to maximise yield is less strong, or simply because in practice it is rare to plant an ornamental crab more than a bee’s flight from other Malus. Certainly, ornamental crabs generally fruit well in urban/suburban environments without any care being taken to provide pollinators. However, in cases of poor fruiting, it might be worth considering this as an explanation.

Microbial diseases rather than insect pests are the main problems for the Malus grower: the same diseases affect both crabs and orchard apples (Royal Horticultural Society 2021; Fiala 1994; Baker 1991). Where there is no overriding reason to grow a particular variety, choosing disease resistant species and cultivars is the simplest, most effective response. Apple scab, caused by the ascomycete fungus Venturia inequalis, is found almost wherever apples are grown but is worst in humid areas, spread by airborne spores, and overwintering on fallen leaves. Olive green spots, velvety with spores, appear on the leaves in spring, darkening later, with leaves yellowing and falling early; fruits too are blotched and deformed, and branchlets may crack, allowing in apple canker. Removal of infected twigs and clearing away fallen leaves and fruit may help. Apple canker, caused by another ascomycete Neonectria ditissima, attacks twigs and larger branches; entering through wounds or buds, it causes round or oval indented areas with dead bark, enlarging over the years and sometimes killing the branch. Infected wood should be pruned out, applying wound paint to the cut. The problem may be worse on heavy, wet soils. Fireblight, caused by the bacterium Erwinia amylovora, originated in North America where it is a serious problem in some areas, but was introduced in the mid-20th century to northern Europe (including the UK and Ireland, where it is still quite rare as a problem in ornamental crabs – N. Dunn pers. obs. 2021), and is spreading across the continent. Outbreaks are explosive and not easily predicted, but are associated with warm, moist weather. Bacteria usually enter through the flower, from cells overwintering in bark cankers or carried on the wind. The infection spreads very rapidly under the bark, staining the cambium red-brown and killing shoots, even whole large branches; leaves suddenly wilt and die, leaving the area looking as if it has been scorched by an out-of-control bonfire. Affected branches should be cut out as soon as possible, to at least 60 cm below visibly infected tissue – but only in dry weather to avoid reinfection through the cut surface – and tools sterilized afterwards. In eastern and central North America only, cedar-apple rust is a major problem (Tattar 1989; Fiala 1994). It is caused by the basidiomycete fungus Gymnosporangium juniperi-virginianae, whose life cycle requires both Malus and Juniperus virginiana (Eastern Red Cedar). Galls on the twigs of Juniperus carrying a perennial infection release teliospores in spring. These are carried in the air, infecting newly emerged Malus leaves. Oranges spots up to 2 cm across appear, which in late summer release aeciospores from small cup-shaped structures on the underside; these infect Juniperus. Severe infection causes premature leaf fall, stunted growth and even tree death in the most susceptible apples, such as M. ioensis. To eliminate all Juniperus growing within about 1.5 km of the Malus is rarely possible or desirable, and to plant resistant crabs is the usual response (Fiala 1994). Various powdery mildews affect apples, showing as a powdery white coating on affected leaves (which may be deformed), and sometimes flowers or fruit. The fungi overwinter on infected shoots and fallen leaves, both of which should be removed in winter. Excessive humidity caused by poor air movement is a major risk factor: open, sunny positions suit Malus best.

Most commercial propagation of Malus is by budding or grafting (Humphrey 2019; Fiala 1994). Both T-budding and chip-budding (in summer, typically early August in the UK) are possible. Whip-and-tongue grafts are made in early spring, late in the dormancy period, either as bare root ‘bench’ grafting or ‘field’ grafting directly onto established rootstocks. Medium-vigour MM106 and MM111, along with the phytophthora-resistant M116, are the most popular rootstocks for ornamental crabs (see Malus Rootstock Cultivars for more details). The more vigorous M5 would increase tree size by around 20%, while for very dwarf patio trees M27 would reduce vigour by at least 60% (N. Dunn pers. obs. 2021). Using seed-raised M. domestica or the seed strain M. sylvestris ‘Bittenfelder’ as a rootstock is an old-fashioned practice which will not result in uniform, high quality nursery stock and can cause excessive suckering on weaker cultivars. From time to time cuttings have come into favour, even with some highly commercial North American nurseries (Fiala 1994; Dirr 2009). Softwood cuttings rooted under mist in summer are the norm, although on a domestic scale hardwood cuttings taken in early winter and placed in a cold frame are sometimes successful. For the gardener even more than the nurseryman, the main drawback of cuttings is the lack of consistent root characteristics across cultivars, each variety having its own set of strengths and weaknesses to be understood. Propagation from seed is straightforward, germination requiring prolonged exposure to cold, and is essential for new wild introductions as well as breeding novel cultivars. However, since Malus are mostly outbreeders, even species raised from cultivated seed are likely to be hybridized. In facultative apomicts, most seedlings are likely to be ‘true’. Finally, proven tissue culture techniques for Malus exist, and are becoming more common in commercial settings, but the high cost limits their use (N. Dunn pers. obs. 2021).

In the UK there are good collections of wild Malus at RBG Edinburgh, the Yorkshire Arboretum, and – still mostly of younger trees – at Ness Botanic Gardens, Cheshire (Royal Botanic Garden Edinburgh 2020; Yorkshire Arboretum 2020; University of Liverpool 2020). Two National Plant Collections include a large range of hybrid crabapple cultivars as well as wild species: these are at Barnards Farm, West Horndon, Essex and at Jodrell Bank, Cheshire (Barnards Farm 2020; University of Manchester 2020); the collection of orchard apples at Brogdale, Kent also includes ornamental crabs (Brogdale Collections 2018). In the Netherlands, there is an extensive Malus collection at the Belmonte Arboretum, Wageningen (Belmonte Arboretum 2020). Among Belgian collections, the genus is particularly well represented at Arboretum Hof Ter Saksen, Beveren; Arboretum Provinciaal Domein Bokrijk; and Arboretum Robert Lenoir, Rendeux (Plantcol 2020). Crabapple cultivars are widely planted in North American collections, while the species are particularly well represented at the Arnold Arboretum, MA, and the Morton Arboretum, IL (Arnold Arboretum 2020, Morton Arboretum 2020).

As discussed above, this is a taxonomically difficult, poorly understood genus. Published keys to Malus tend to be incomplete, based on limited specimens and often not to work in practice. To avoid giving a spurious sense of confidence, the following key (to species only) does not attempt to differentiate members of some of the more difficult ‘clans’ outlined above. Species lists for these are as follows:

Clan A: asiatica, domestica, kirghisorum, orientalis, prunifolia, sieversii, spectabilis, sylvestris

Clan B: angustifolia, coronaria, ioensis

Clan C1: baccata, halliana, hupehensis, rockii, sikkimensis, spontanea

Clan C2: bhutanica, komarovii, toringo, transitoria

Clan D: honanensis, kansuensis, ombrophila, prattii, yunnanensis

Identification key

1aAt least some calyces falling before fruit fully ripe, sepals often rather fleshy2
1bCalyx unambiguously persistent6
2aAll leaves unlobedClan C1
2bAt least some leaves lobed on long shoots in summer3
3aAll leaves strongly lobed, flowers hawthorn-like, leaves rough, grey beneath, very late flowering, Europeflorentina
3bWithout this combination of characters, East Asia or W North America4
4aPetals orbicular to broadly obovate, flowers with a yeasty hawthorn-like scent5
4bPetals narrower, ovate to elliptic or obovate, flowers with a sweet, typically apple-blossom scentClan C2
5aPetals 10–15 mm long, fruit 6–10 mm diameter, W North Americafusca
5bPetals 8–10 mm long, fruit 10–15 mm diameter, Chinakansuensis
6aFruit green, waxy, fragrant, core not enclosed at apex, stone cells abundant around core; anthers pink or purple before dehiscence; leaves sometimes shallowly lobed on long shoots only; C and E North America7 (Clan B)
6bWithout this combination of characters, Asia or Europe9
7aSepals and leaves hairy beneathioensis
7bSepals glabrous beneath; leaves glabrous beneath except on the veins8
8aFlowering shoot leaves elliptic or oblong, base usually cuneate, apex more or less roundedangustifolia
8bFlowering shoot leaves proportionately broader, ovate or lanceolate, base usually rounded or cordate, apex more or less pointedcoronaria
9aFruit with endocarp only partially attached to hypanthium, the free portion cone-shaped; calyx persistent; branches without thorns10
9bFruit with endocarp fully attached to hypanthium; calyx persistent or not; branches with or without thorns12
10aStamens about 20; leaves with 3 lobes extending more than 1/3 of the way to the midrib; Europe and W Asiatrilobata
10bStamens 40–55; leaves more shallowly lobed or unlobed; E Asia11
11aLeaves sometimes shallowly lobed, petiole and underside wooly tomentose; Japantschonoskii
11bAll leaves unlobed, petiole and underside wooly tomentose at first, becoming glabrous; China and Indochinadoumeri
12aFruits <2 cm across; rachis of inflorescence 1–2 cm; flowers superficially 'hawthorn-like'Clan D
12bFruits >2 cm across; rachis of inflorescence <1 cm; flowers 'apple-like'Clan A