Paeonia L.

TSO logo


Kindly sponsored by
Sponsored by David Sayers in memory of Hedvika Fraser.


Julian Sutton (2020)

Recommended citation
Sutton, J. (2020), 'Paeonia' from the website Trees and Shrubs Online ( Accessed 2024-07-12.


  • Paeoniaceae

Common Names

  • Peonies
  • Paeonies
  • Woody Peonies
  • Tree Peonies


(sect.) Subdivision of a genus.
Pollen-producing structure of flower at the tip of the filament; part of a stamen.
Situated in an axil.
Immature shoot protected by scales that develops into leaves and/or flowers.
(of a plant or an animal) Found in a native state only within a defined region or country.
Outside the flower. Used to describe nectaries situated on vegetative parts of a plant.
A group of genera more closely related to each other than to genera in other families. Names of families are identified by the suffix ‘-aceae’ (e.g. Myrtaceae) with a few traditional exceptions (e.g. Leguminosae).
Plant originating from the cross-fertilisation of genetically distinct individuals (e.g. two species or two subspecies).
(botanical) Contained within another part or organ.
Taxonomic account of a single genus or family.
The visible form of an organism.
keel petal
(in the flowers of some legumes) The two front petals fused together to form a keel-like structure.
Petal-like. May refer to sepals or stamens modified into a petal-like form.
Act of placing pollen on the stigma. Various agents may initiate pollination including animals and the wind.
Arranged in a net-like manner.
(sect.) Subdivision of a genus.
(pl. taxa) Group of organisms sharing the same taxonomic rank (family genus species infraspecific variety).
Arrangement of three or more organs (leaves flowers) around a central axis. whorled Arranged in a whorl.



Julian Sutton (2020)

Recommended citation
Sutton, J. (2020), 'Paeonia' from the website Trees and Shrubs Online ( Accessed 2024-07-12.

A genus of about 33 species (about 9 of them shrubby, the remainder herbaceous), distributed from East Asia to the Mediterranean Basin, with two outliers in western North America. The following relates to the woody species, all native to China.

Shrubs to 3.5 m; stems sympodial, the apical part of each year’s growth dying back. Roots thick and fleshy, sometimes tuberous. Leaves alternate and compound, either ternate or pinnate and sometimes further divided or lobed. Flowers terminal on the current year’s growth, and sometimes also in upper axils. Bracts leaflike, varying in shape and size, grading into sepals. Disc extended upwards to form a sheath partly or entirely enclosing the carpels. Sepals 2–9, apex usually caudate; petals 5–13, many more in some garden hybrids; stamens very numerous, the innermost dehiscing first; carpels about 5, free; styles present or absent; stigmas laterally flattened, recurved, crested. Fruit a follicle containing a few seeds to 1.3 cm diameter. (Bean 1976; Hong et al. 2001; Hong 2010)

Woody peonies, mostly complex hybrids, have been prized as garden plants in East Asia for more than 1000 years. Despite steady introductions over more than 200 years – not always of the most suitable plants – most have failed to enter the garden mainstream in our area, and have largely been the province of enthusiastic specialists. Ignorance of the wild species and, outside China, of the diverse nature and multiple origins of the hybrids, has not helped. Chosen well, however, woody peonies can be impressive, easily grown garden shrubs (never trees, despite the ‘tree peony’ moniker) in much of our area.

The genus Paeonia is taxonomically isolated (Sang et al. 1997). Once grouped in the Ranunculaceae, it is now generally placed in a family of its own, Paeoniaceae, within the order Saxifragales (Stevens 2020). Identification of its closest relatives by reconstructing the rapid, early diversification of the Saxifragales using molecular data has proved problematic. The leading suggestion, that Paeoniaceae might be sister to a clade comprising Altingiaceae, Hamamelidaceae, Cercidiphyllaceae and Daphniphyllaceae, a group rich in familiar woody plants, is not unambiguously supported by molecular data and has never been raised on morphological grounds (Jian et al. 2008).

Molecular data have not altered the long-term consensus that Paeonia divides into three natural groups, treated as sections in Hong’s (2010) revision of the genus. Sections Onaepia Lindl. (W. North America) and Paeonia (Europe and Asia) are exclusively herbaceous. Section Moutan D.C. is exclusively woody, and endemic to China. While hybrids within sections are common, there are strong (but not insurmountable) barriers to breeding between woody and herbaceous peonies. Herbaceous species, especially in the Mediterranean Basin, are taxonomically troublesome due to a history of repeated divergence and hybridization – reticulate evolution – with polyploidy involved (Sang et al. 1997). However, taxonomic problems around woody species (all diploids, 2n=10 – Yang et al. 2020), are largely the result of human influence.

People and peonies have lived together in China for thousands of years, linked by both beauty and utility. Herbaceous peonies were domesticated more than 3,900 years ago, and woody peonies have been cultivated for at least 2,000 years, initially for medicine (Cheng 2007). The outer layers of the roots of all woody peonies are used as a traditional medicine, danpi, a treatment for diverse various ailments, mostly connected in some way to the blood. Its importance has led to movements of plants back and forth between the wild and cultivation, with an inevitable mix of selection, hybridisation and blurring of the distinction between natural and garden populations (McLewin & Chen 2006; Cheng 2007). Serious interest in woody peonies as ornamentals began in China well over 1,000 years ago (Li 2005). Active selection for ornamental traits in garden hybrids followed rather different lines in several climatically diverse regions of China, as well as in Japan (Li 2005). All this, combined with a long history of habitat change in the more densely populated parts of China has led to a rarity of truly wild material, the basis of any understanding of the wild species underlying woody peony diversity.

Until well into the 20th century, scientific names were applied to woody peonies only by westerners ignorant both of these issues and of the full diversity of plants in East Asian cultivation. The first plants to reach Europe were garden hybrids, but were named as if they were species – Miller’s P. suffruticosa of 1804 is a prime example, still sadly in use. The period of large scale collecting by westerners in central and western China, in the late 19th and early 20th centuries, brought potentially wild material to attention. Two species described in this period, P. delavayi and P. ludlowii, are found well to the west of other species, and had experienced much less human interference. They are still recognised today, grouped as subsection Delavayanae (Hong 2010). Understanding of the remaining species, grouped in subsection Vaginatae, was poorly served during this period. It has taken thorough fieldwork by Chinese researchers over at least 30 years to build a coherent picture of these, the wild ancestors of most garden hybrids, culminating in Hong’s (2010, 2011) monograph. Far from the last word, this might best be considered the end of the beginning of this work, which has involved repeated taxonomic change and sometimes vitriolic disputes. Most of these species are rare, some vanishingly so, and the existence of one (P. ostii) has in effect been inferred from domesticated plants.

Our approach here has been to explain the diversity of cultivated plants rather than formally to classify them. We describe all species currently recognised and discuss their status in cultivation – few are at all widely grown. We treat the hybrids as informal assemblages – groups rather than Cultivar Groups – reflecting their genetic and geographical origins, an approach widely used in the English language Chinese literature (Wang 1998; Li 2005). The benefit of this is that it brings together bodies or more-or-less shared features, including ancestry, morphology, climatic adaptation and breeding potential. The drawback is that it is almost impossible rigorously to define these groupings and it is not always possible easily to place an unidentified cultivar; they are ‘blurred at the edges’, reflecting the realities of gene flow between breeding lines. There are four major Chinese groupings, the Central Plains, Gansu, Southern Yangtze and Southwest cultivars. Derived from the Central Plains group, the Japanese cultivars have distinctive qualities. We treat cultivars raised in Europe or North America without other genetic input as belonging to these groupings. More precisely defined, the Lutea Hybrids comprise all woody hybrids with P. delavayi in their ancestry (American Peony Society 2020). The Itoh cultivars are hybrids between woody peonies and the herbaceous P. lactiflora. In both these groups, however, distinctions are becoming less clear with successive generations of backcrossing Lutea Hybrids to Japanese cultivars in particular, buy also some Itohs. As low levels of fertility are found in some of the usually sterile Itoh cultivars it seems likely that even this distinction will become blurred eventually.

Formal attempts to classify hybrids have struggled to reconcile these meaningful but loosely delineated assemblages with the need for precise definition and engagement with existing latinized names. Further, a clear summary of breeding lines (Yang et al. 2020) shows that development of hybrids across the genus has been reticulate, and so not easily forced into a hierarchical scheme. All make use of the familiar but ambiguous name ‘P. suffruticosa’ (q.v.) in one way or another. Applied at some time or another to pretty much any species or hybrid within subsection Vaginatae, many Chinese botanists continue to use this name for hybrids within the section, without further categorization (for example Hong 2011; Zhou et al. 2014). The American Peony Society (2020) recognises three groups of woody hybrids plus the Itoh Group; Lutea Hybrids are defined above; Rockii Hybrids are those with flares on the petal bases, evidence of P. rockii in their ancestry; and Suffruticosa Hybrids are the remaining hybrids within subsection Vaginatae. This cuts across the distinction between Gansu and Central Plains cultivars, raising a generalisation (Gansu cultivars have much rockii influence and usually have flares, Central Plains cultivars have less rockii involvement and more rarely have flares) to the status of definition. Cheng (2007) attempts to handle the full diversity of hybrids within a taxonomic framework. Central Plains, Southern Yangtze and Southwest cultivars are treated as Cultivar Groups (under other names) within P. × suffruticosa, as are Japanese, European and American cultivars descended from them. He places all Gansu cultivars within P. rockii, distinguishing domesticated P. rockii as Rock Group and hybrids as Gansu Group, surely a taxonomic nonsense whether or not one could ever be certain which plants are hybrids. P. × lemoinei rather than Lutea Hybrids is used, a narrower grouping which leaves many modern hybrids unplaced. Other systems fall somewhere between these poles. The major differences in practice are whether or not Gansu cultivars are included within P. × suffruticosa, and whether to use Lutea Hybrids or P. × lemoinei. In all cases, parentage (known or presumed) plays a part in category definitions. It seems likely, however, that within a few years breeding will blur these categories so much that only a classification based entirely on observable features will suffice. Anyone making cultivar listings will feel the temptation to reach for ‘suffruticosa’ in some form as a catch-all. In our view this concept is not helpful and we strongly urge its discontinuation. In trying to avoid this dangerously convenient name, the use of a suffix such as [S] to denote shrubby species, whether or not they are grouped within some other taxon, is available: the Royal Horticultural Society’s Plant Finder (Cubey 2019) takes this approach, albeit in a limited way.

Woody peonies produce long annual shoots bearing the leaves; if the shoot produces flowers, one is terminal and any others are in upper leaf axils. A substantial portion of the new shoot dies over winter (generally more in colder areas), so that only the basal part contributes to the lengthening woody stem. The following year’s shoot grows out from an axillary bud; the woody stem is therefore a sympodium (Hong 2010). New shoots sometimes arise from the woody rootstock, and some spread laterally to the extent that they can be called stoloniferous; in this way woody peonies display anything from one or two ‘trunks’ to a veritable thicket.

The open, bowl shaped flowers with many stamens suggest a generalist pollination strategy. In a nectar-producing population of P. delavayi in Yunnan, various bees, butterflies, beetles and ants visited, bees being the most effective pollinators (Li et al. 2014). P. jishanensis, which produces no nectar, was visited by bees and beetles (Luo et al. 1998), while He et al. (2020) found that bumblebees and honeybees were effective pollinators of domesticated P. ostii grown for seed oil. As a rule, woody peonies are self-fertile, but only at a low level, favouring outcrossing (Yang et al. 2020); in the garden context, this encourages uncontrolled hybridisation. The large, non-fleshy seeds of peonies seem not to be dispersed by animals, and move only short distances (Wang 2019).

Two flower features are particularly interesting. Firstly, the disc below the carpels is extended upwards, partly or wholly enclosing them; the size and colour of this sheath is an important identification feature. In P. delavayi, the disc produces nectar in some populations (Hong et al. 1998); in woody peonies of subsection Vaginatae only extrafloral nectar is produced, at the bases of some bracts, which explains the attraction of ants to peony buds, but has no role in pollination. Secondly, the flowers of peonies, both herbaceous and woody, seem especially susceptible to doubling.

As in other flowers, doubles are the result of floral organ primordia, which would otherwise become stamens or carpels developing as petal-like structures. The number of primordia doing this, and the extent to which they resemble petals, determines the flower form, and is horticulturally very significant. The American Peony Society (2020) recognises six categories, developed for herbaceous peonies but mostly also relevant to woody cultivars. Single flowers are defined as having typically 5, but up to 15 petals, but if more than 5 opening flat to show the stamens and carpels. The Japanese form (not seen in woody cultivars) has stamens replaced by more-or-less sterile staminodes, which still somewhat resemble stamens. The distinction between Anemone and Bomb forms is really relevant to herbaceous peonies: both have petaloid stamens, which are distinctly shorter and/or narrower than the true petals; in Anemone flowers they are small and yellow (i.e. anther-coloured) while in Bomb flowers they are the same colour as the petals, long, but narrower than true petals. Flowers looking very much like bomb forms are known among Central Plains cultivars but, confusingly, are classified as Anemone form in China (Wang 1998). Semi-Double flowers have more full-size petals than singles, or they do not open flat, giving a ‘fuller’ look, although not entirely obscuring the stamens and carpels. Semi-Doubles cover three Chinese categories, Lotus, Chrysanthemum and Rose, with increasing number and size of petals, sometimes grouped together as Hundred Petals forms (Wang 1998). In Full Double flowers all primordia develop as petals, except sometimes for a few stamens among the petals. These residual stamens are usually in a whorl, giving rise to a ‘flower in flower’ appearance: this is known as proliferation, and is given much prominence in China. Full Doubles are called Crown or Globular form in China, the latter shows little variation in petal size throughout the flower (Wang 1998). However carefully categories are defined, some cultivars will fall on the boundaries. American and Chinese categories do not neatly coincide; Chinese Crown varieties seem to fall between Bomb and Full Double, for example. This is not an exact science.

Some general points can be made about cultivation (Bean 1976; McLewin & Chen 2006; American Peony Society 2020). While most wild peonies are found in more or less wooded places, crowding and low branches overhead are their enemies. Open positions (not necessarily sunny) with good air movement, on fertile but well-drained soil, are basic recommendations. Sir Peter Smithers, who grew and bred many woody peonies, planted them on a sunny, south-facing slope in his Swiss garden, destined to become a woodland with many magnolias (Smithers 1995): initially they ‘flourished exceedingly’ but as open space diminished through time many declined and were lost. Gansu cultivars unambiguously perform best in full sun (McLewin & Chen 2006). Planting or potting is best done in September, before root growth begins in earnest. Grafted specimens in particular should be planted deeply, with the graft union around 10 cm deep: this discourages shooting from the understock and encourages the scion to produce its own root system (Bean 1976; Smithers 1995). Itoh cultivars should, like herbaceous peonies, be planted with the roots covered but with buds showing at the surface.

Woody peonies are relatively free of pests and diseases, but peony blight caused by the fungus Botrytis paeoniae can be serious, causing young shoots to wilt and die. A grey mould may be seen, producing brownish or black spores, which can spread to young leaves and flower buds, which then rot. Affected shoots and all snags of dead wood should be cut out and destroyed. Fungicides can be used as growth emerges, and thereafter, but good air movement and garden hygiene, removing dead leaves as they fall, are the best prevention (Bean 1976; Missouri Botanical Garden 2020). Care should be taken in propagation and pruning to avoid spreading viruses.

Beyond tidying, there should be no need for pruning, although McLewin & Chen (2006) find that strongly growing Gansu cultivars will tolerate almost any pruning inflicted upon them. Winter pruning to keep the shrub artificially short might sometimes be useful, but it removes the large axillary buds which will become next season’s flowering shoots. Smithers (1995) describes a labour-intensive Japanese technique, which gets around this problem. After flowering, as soon as buds are visible in the leaf axils (probably in June) they are removed from the upper axils with a sharp knife; it is then the lower axillary buds which fatten up. Just before leaf fall the stems are pruned back to these buds.

Some groups are better suited to our area than others. Apart from the widely growable P. delavayi and the less hardy P. ludlowii, which is a good garden plant in Britain but unsuited to the peony-growing heartland of the central United States, genuine wild species are very rarely seen in gardens. The Central Plains cultivars, hugely popular in parts of China, are adapted to hot, rainy summers, and prove difficult to grow well in most of our area. Japanese cultivars derived from them have been more satisfactory. Gansu cultivars seem more adaptable and can make excellent garden plants in much of our area, as can Lutea Hybrids and Itoh cultivars.

Division (in September, but winter is sometimes suggested) and grafting are the usual clonal propagation methods (Bean 1976; McLewin & Chen 2006). Grafting, in September or October, is usually onto a herbaceous understock, generally a form of P. lactiflora or P. mascula. Side wedge grafts are traditional, due to the size mismatch between the herbaceous peony’s tuberous root and the scion. The understock should decline and die once the scion’s root system is well established, although it is sometimes removed. On a domestic scale, layering is possible, if slow. With Gansu cultivars, cuttings of midsummer new growth, cut just into old wood, with some leaf reduction, bottom heat and fairly high humidity can be successful, but are rarely tried (McLewin & Chen 2006). Research on tissue culture of woody peonies is ongoing but apparently not yet commercial. However, it is the leading method for the bringing to market and mass production of new Itoh cultivars. Otherwise, these can be divided in September or October as with herbaceous peonies. Bearing in mind that garden seed is often outcrossed and that seedlings of hybrids and cultivars will not be ‘true’, seed should be sown when ripe and exposed to cold overwinter. Germination follows in spring, but ungerminated seed should be retained for another year (Smithers 1995).

The following routes into the vast literature of peonies might prove helpful. For wild species, a modern monograph of the entire genus exists (Hong 2010; Hong 2011). Stern’s (1946) monograph is taxonomically outdated, but still provides much useful information on early western experience with woody peonies. Cheng (2007) and Yang et al. (2020) give clear overviews of international breeding efforts. For detailed, illustrated treatments of Chinese cultivars see Wang (1998) and Li (2005), both available in English; Hashida (1990) does the same for contemporary Japanese cultivars, in Japanese. The Gansu cultivars and P. rockii are covered thoroughly and engagingly, from an integrated Chinese/Western viewpoint, by McLewin & Chen (2006); detailed cultivar descriptions are given by Li (2005). For the international register of cultivars (but not all are registered) and snappy articles from a North American standpoint, see American Peony Society (2020). Other online resources include the websites of specialist nurseries (for example Cricket Hill Garden 2020; Pivoines Rivière 2020; Solaris Farms 2020; Tetzlaff 2020; Tree Peony Company 2020) and the extraordinary, if labyrinthine, hoard of information and literature which is Carsten Burkhardt’s Web Project Paeonia (Burkhardt 2020).

The following key applies only to wild species (Hong et al. 2001; Hong & Pan 2007; Hong 2011).

Identification key

1aDisc fleshy, surrounding only the base of the carpels at flowering time; carpels glabrous.2 (subsection Delavayanae)
1bDisc leathery, surrounding the carpels to at least halfway; carpels hairy or glabrous.3 (subsection Vaginatae)
2aCarpel usually 1; follicle >4.5 cm long; petals always yellow; plant 1.5–3.5 m tall.P. ludlowii
2bCarpels usually 2–5; follicle <4.5 cm long; petals not always yellow; plant <2 m tall.P. delavayi
3aDisc reaching halfway up the carpels at flowering time.P. decomposita
3bDisc reaching more than halfway up carpels at flowering time4
4aCarpels glabrous, usually 3 or 4, leaflets usually more than 20, all lobed.P. rotundiloba
4bCarpels densely hairy, usually 5; leaflets usually less than 20, if not, at least some entire.5
5aLower leaves pinnate, leaflets >9.6
5bLower leaves biternate, leaflets usually 9.7
6aLower leaves 2-pinnate, leaflets <16, petals without basal flare.P. ostii
6bLower leaves (2–)3-pinnate, leaflets >16, petals with basal flare.P. rockii
7aLeaflets mostly entire, often reddish above, densely hairy in axils of main veins beneath; petals often with red basal spot.P. qiui
7bLeaflets mostly entire, green above, beneath glabrous, hairy on veins or hairy throughout; petals without basal spot.8
8aLeaflets glabrous; sepals with caudate apex; petals pink.P. cathayana
8bLeaflets hairy beneath, at least on the veins; sepals with rounded apex; petals white, sometimes edged pink.P. jishanensis