Hydrangea Gronov. ex L.

TSO logo

Sponsor

Kindly sponsored by a generous supporter who loves hydrangeas.

Credits

Julian Sutton (2025)

Recommended citation
Sutton, J. (2025), 'Hydrangea' from the website Trees and Shrubs Online (treesandshrubsonline.org/articles/hydrangea/). Accessed 2025-07-08.

Family

  • Hydrangeaceae

Synonyms

  • Broussaisia Gaudich.
  • Calyptranthe (Maxim.) Nakai
  • Cardiandra Siebold & Zucc.
  • Cornidia Ruiz & Pav.
  • Deinanthe Maxim.
  • Decumaria L.
  • Dichroa Lour.
  • Heteromalla (Rehder) H.Ohba & S.Akiyama
  • Hortensia Comm.
  • Pileostegia Turcz.
  • Platycrater Siebold & Zucc.
  • Schizophragma Siebold & Zucc.
  • ×Didrangea J.M.H.Shaw

Species in genus

Glossary

capsule
Dry dehiscent fruit; formed from syncarpous ovary.
IDS
International Dendrology Society sponsors of this book.
ovary
Lowest part of the carpel containing the ovules; later developing into the fruit.
section
(sect.) Subdivision of a genus.
taxonomy
Classification usually in a biological sense.
axillary
Situated in an axil.
capsule
Dry dehiscent fruit; formed from syncarpous ovary.
disjunct
Discontinuous; (of a distribution pattern) the range is split into two or more distinct areas.
family
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).
glabrous
Lacking hairs smooth. glabrescent Becoming hairless.
included
(botanical) Contained within another part or organ.
monograph
Taxonomic account of a single genus or family.
monophyletic
(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.)
morphology
The visible form of an organism.
ovary
Lowest part of the carpel containing the ovules; later developing into the fruit.
polyphyletic
(of a taxon) Including entities descended from different ancestors and evolutionary lineages grouped usually as a result of superficial similarities (n. polyphyly). The grouping is therefore not ‘natural’. (Cf. monophyly paraphyly.)
section
(sect.) Subdivision of a genus.
umbel
Inflorescence in which pedicels all arise from same point on peduncle. May be flat-topped (as in e.g. Umbelliferae) to spherical (as in e.g. Araliaceae). umbellate In form of umbel.

References

Credits

Julian Sutton (2025)

Recommended citation
Sutton, J. (2025), 'Hydrangea' from the website Trees and Shrubs Online (treesandshrubsonline.org/articles/hydrangea/). Accessed 2025-07-08.

A genus of 65–100 or more species of deciduous or evergreen shrubs (rarely approaching small trees), woody climbers with aerial roots, and herbaceous perennials, native to Asia, the Americas, New Guinea and Hawaii, heavily concentrated in warm temperate areas.

Branchlets, leaves and inflorescences variously hairy or glabrous. Leaves usually opposite, less often alternate or fasciculated; stipules present or not. Leaf blade simple, margin entire, serrate or dentate (rarely pinnately lobed). Inflorescence terminal (rarely axillary) on current year’s growth, a complex branching system approximating to a corymbose or umbellate cyme, less often to a panicle, usually many-flowered. Flowers of two types, ‘fertile’ and ‘sterile’; wild-type plants of most species have inflorescences largely comprising fertile flowers, with a smaller number of sterile flowers usually around the inflorescence margin, a few species having only fertile flowers (but see note on the entirely herbaceous Section Deinanthe below); mutant forms with sterile flowers predominating are widely cultivated. Fertile flowers small, hermaphrodite; calyx tube adnate to ovary, usually toothed or lobed; petals 4–8, valvate (in all woody species) or imbricate in bud, falling at anthesis either individually or in a few species joined as a ‘cap’, or (less often) persistent; stamens 4–30 (rarely much more numerous); ovary inferior or partially so; styles 1–6, sometimes connate; stigmas terminal or decurrent, occasionally lobed. Sterile flowers consisting of 1–5 large, petaloid sepals, with other floral organs usually absent, occasionally present or reduced. (In the wholly herbaceous Section Deinanthe the situation is reversed, the few-flowered inflorescences consisting of large, showy fertile flowers with a few small sterile flowers at the margin.) Fruit a dehiscent capsule or a fleshy berry. Seeds small (sometimes minute) and numerous, sometimes winged. (McClintock 1957; Pan et al. 2001).

Preface and Acknowledgements

The account of Hydrangea presented here gives a completely new perspective on this familiar and yet very unfamiliar genus. It has been made possible by a very generous donation from a ‘lover of hydrangeas’ who prefers to remain anonymous, but anonymity does not preclude recording our sincere and great gratitude for this support.

Hydrangea is indeed both familiar and unfamiliar. We are all acquainted with a handful of ‘core’ species, but beyond these there is a morass of uncertainty. In this magisterial account of the genus in horticulture, Julian Sutton steers a cautious path through the difficulties, hedging it about with caveats and repeated warnings of the difficulties, attempting to reconcile the patchy science with the plants grown in our gardens. The principles of his approach are clearly discussed in the introduction to the genus, which we urge you to read before approaching the species.

This account of the genus contains over 160,000 words, which would fill an 800 page book containing a fraction of the 1300 images we can provide here. Bean covered 15 species plus 53 varieties and cultivars: here, at publication in May 2025, we present accounts of 81 species and 587 infraspecific taxa. This is​ a consequence of changing taxonomic concepts and the inclusion of formerly segregate genera, as well as an indication of the vastly increased popularity of the genus in recent years that has seen a proliferation of cultivar selections (only sampled here). The work has taken two and a half years to prepare, and has been a monumental effort by Julian Sutton. We thank him most profoundly for this extraordinary piece of work.

Julian and the TSO team are also collectively grateful for the very many people who have helped with information or supplied images, among whom are Jack Aldridge, John Anderson, Ross Bayton, Koen Camelbeke, Glyn Church, Barry Clarke, John David, Michael Dirr, Dave Doggett, Dawn Edwards, Mark Fillan, Maurice Foster, Jess Francis, Barry Grain, Anne Greenall, Rowan Griffiths, Sam Hoadley (Mt Cuba Center), Dan Hinkley, Marco ten Hoope (Kolster BV), Tom Hudson, Kevin Hughes, Hayes Jackson, Ozzie Johnson, Jan De Langhe, Adriana Lopez Villalobos (University of British Columbia), Nick Macer, Guillaume Mamdy, Susan Miller, Irene Palmer (North Carolina State University), Pépinières Minier, Richard Pitts, Charlie Pridham, Alan Shamash, Joe Sime, Mark Weathington (JC Raulston Arboretum), Rip Weaver, Bleddyn & Sue Wynn-Jones, and in the most important supportive role of all, Sarah Sutton.

As always with TSO we will welcome your feedback.

John Grimshaw & Tom Christian

Editors, IDS Trees and Shrubs Online, 15 May 2025

More than Mopheads

Almost every gardener in the temperate zone will be familiar with at least one hydrangea. For many, especially in Europe and New Zealand, their first and defining experience will be of classic ‘mophead’ and ‘lacecap’ cultivars derived primarily from Hydrangea macrophylla, growable with protection or heroic measures even in colder areas but coming to the fore in mild, maritime climates. A younger generation may be equally familiar with the hardier H. paniculata, whose popularity and cultivar diversity has burgeoned. For many North Americans, either H. paniculata ‘Grandiflora’ – the ‘Peegee’ hydrangea – or the native H. arborescens will be the archetype. These summer-flowering deciduous shrubs may be the head of the Hydrangea dragon, but it has a long and diverse tail. Other shrubby species, tall or dwarf, glossy and glabrous or delightfully hairy, bold-leaved or finely textured, have quite different garden uses and effects both in leaf and flower. There are climbers too, deciduous or evergreen, some able to reach the top of a large tree, others manageable on a modest wall. Shrubs and climbers still known to many as Broussaisia, Decumaria, Dichroa, Pileostegia, Platycrater and Schizophragma are now included in Hydrangea, some bringing unexpected features such as colourful, fleshy berries. There are even a few herbaceous hydrangeas, species formerly placed in Cardiandra and Deinanthe.

Taxonomy: the Trouble with Hydrangea

Hydrangea is a taxonomically challenging genus. The great broadening of its scope in recent years (see ‘Hydrangea in the Hydrangeaceae’, below) may be revolutionary, but delineation of species is a bigger, long standing problem. Perpetual anarchy in naming has been the result.

A body of ‘traditional’ taxonomic work based on morphology was able to recognize groupings of apparently related forms without large scale controversy. Elizabeth McClintock’s (1957) monograph is perhaps the best overview of these. Many have proved to be – or at least approximate to – natural groupings, in the light of molecular evidence (e.g. De Smet et al. 2015; Yang et al. 2024). We characterize these as species complexes. What has proved impossible is consensus on how to recognize species within most such complexes. The usual pattern is as follows. First comes a flurry of species descriptions in the 19th and early 20th centuries, each often from one area and sometimes based on a single specimen, without reference to quite similar forms from elsewhere. Next is a period of consolidation, where workers suggest synonomy, often in conflicting and confusing ways. This is followed by McClintock’s monograph, which usually lumps the bulk of each complex into a very broadly drawn, highly variable species. The next period, our own, is one in which everyone who needs to name hydrangeas, whether carrying out field or laboratory research, compiling a regional flora, writing a horticultural reference book or curating any sort of collection, has to choose what is for them the most workable path through the web of names. To immerse oneself in recent hydrangea literature is to realise that in most areas there is absolutely no consensus.

Our task for TSO is to present the reality of plants seen in Western gardens and the various names they carry. Our approach to species complexes has been to ride both horses; we first describe the complex as a whole under its central species (e.g. H. aspera), but then devote separate pages at species level to each of those arguably distinct segregate species which have any real currency in botany or horticulture. This messy situation reflects a messy reality.

Hydrangea in the Hydrangeaceae

Until the late 20th century Hydrangea was placed in a wide-sense Saxifragaceae, a family which proved to encompass some very distantly related plants. The modern consensus is to treat it as the type genus of a much smaller Hydrangeaceae (Cronquist 1981). A body of molecular evidence places the family not in the order Saxifragales but in the Cornales, along with most notably Cornaceae, Nyssaceae and the mainly herbaceous Loasaceae (Stevens 2025). Phylogenetic work (e.g. Kim et al. 2015) amply supports three distinct groups in the family, usually placed in two subfamilies (Stevens 2025). Subfamily Jamesioideae covers Fendlera and Jamesia, shrubs from the American West. Subfamily Hydrangeoideae has two tribes. Philadelpheae comprises Deutzia, Philadelphus, the North American shrubs Carpenteria, Fendlerella and Whipplea, plus the East Asian herbaceous perennial Kirengeshoma. Tribe Hydrangeae consists of only Hydrangea in its widest sense, as used here.

The concept of Hydrangea has expanded over time. Linnaeus (1753) first used the name for H. arborescens, the type species. It derives from the Greek hydor (water) and aggeion (vessel, container), presumably a reference to fruit shape. Further North American species quickly followed. The first Asiatic species became known to Western science in the late 18th century. H. macrophylla was initially described as a Viburnum by Thunberg (1784), understandably given the opposite leaves, and umbel-like inflorescences ringed by showy sterile flowers; Hortensia opuloides (another Viburnum reference!) and Hydrangea hortensis were other early names for it. The 19th and early 20th century saw the great bulk of species being described as hydrangeas from the start. Although the first-described of the Mexican and South American climbers was placed in the genus Cornidia (Ruiz & Pavón 1798), that name was long overlooked, and most were described in Hydrangea. Eight other clearly related genera, Broussaisia, Cardiandra, Decumaria, Deinanthe, Dichroa, Pileostegia, Platycrater and Schizophragma, grew up independently. Apart from debates around the distinctness of Schizophragma from Hydrangea, and of Decumaria and Pileostegia from one another, they were accepted by most.

Only in the molecular era has evidence accumulated that Hydrangea in this narrow sense is not monophyletic, that these other eight genera are actually embedded within it, the result of distinctive directions of Hydrangea evolution, rather than merely its cousins. Two studies have a particularly broad scope (De Smet et al. 2015; Yang et al. 2024), but many others point in the same general direction (e.g. Raman et al. 2023). De Smet et al. (2015) were the first fully to realise an enlarged Hydrangea which included all these plants, complete with a classification of the genus into sections, plus some necessary name changes. Yang et al. (2024) further refined the classification.

This is a recent development, but is gradually becoming adopted in both botany (e.g. Royal Botanic Gardens, Kew 2025) and horticulture (Royal Horticultural Society 2025). An alternative route to natural, monophyletic genera would be to split Hydrangea, giving generic status to sections or clusters of sections traditionally part of Hydrangea, alongside the other eight genera. An attempt to do just this, a purely nomenclatural work focussing on Japanese species (Ohba & Akiyama 2016), has made little or no impact. A few examples of the highly disruptive effect this would have on many established names include Hortensia macrophylla (Thunb.) H.Ohba & S.Akiyama (Hydrangea macrophylla), Heteromalla doniana H.Ohba & S.Akiyama (Hydrangea heteromalla), Heteromalla paniculata (Siebold) H.Ohba & S.Akiyama (Hydrangea paniculata), Calyptranthe petiolaris (Siebold & Zucc.) H.Ohba & S.Akiyama (Hydrangea petiolaris) and Platycrater involucrata (Siebold) H.Ohba & S.Akiyama (Hydrangea involucrata). If then we follow the broad-genus route, some gardeners will inevitably resent the loss of a few names for recognizable groups, but this need not be the case. It is perfectly acceptable to retain the old generic names as common names for these groups – dichroas, schizophragmas and so forth – just as we already speak of cornidias. The world has not ended because we chose collectively to treat Michelia and Manglietia as sections of Magnolia. On the contrary, this emphasis on their true relationships can only have encouraged the wider, intersectional hybridization which is such an important feature of modern Magnolia breeding, and classifying them in a single genus avoids an unhelpful proliferation of nothogeneric names. Serious intersectional hybridization is only now coming to the fore in Hydrangea, but the results are already benefitting gardeners, and the same nomenclatural issues apply: already we have been able to dispense with the nothogenus ×Didrangea (see Hydrangea × versicolor).

Origins

The wide but disjunct range of Hydrangea needs explaining. Like Magnolia, an even larger and better-studied genus, it has a major centre of diversity in warm-temperate East Asia, with some regional disjunctions (between Japan, Korea and Eastern China for example), and secondary centres in the southeastern United States, Mexico and South America. Neither genus has living species in West Asia, Europe, Africa or Western North America. Disjunctions in Magnolia result mainly from contraction during the Ice Ages of a once near-continuous range in the warm temperate Northern Hemisphere forests of the Tertiary (see references under Magnolia). Hydrangea is a younger genus, diversifying only from the mid-Tertiary around 23 MYA (Yang et al. 2023), but even so fossils matching wide-sense Hydrangea are known from Europe and Western North America (summarized in McClintock 1957; Yang et al. 2023). This makes spread by land from an origin somewhere in Asia a viable explanation, the genus reaching the American Southeast via Beringia and Western North America. A North American origin with migration in the opposite direction is equally feasible, and Raman et al. (2023) could reach no conclusion on which was more likely. Matching distribution to phylogeny suggests that species from Mexico, Central and South America are from a different lineage, and long distance dispersal between East Asia and the Americas (perhaps in both directions) was involved here (Yang et al. 2023). Hydrangea seeds are minute, typically about 1 mm long, and usually winged, making this far more plausible than in the larger-seeded Magnolia. This may well explain why Hydrangea has colonized oceanic islands (Hawaii) but Magnolia has not.

Classifying Hydrangea – a Work in Progress

Anyone trying to explore the diversity of a large genus like Hydrangea should find helpful a basic understanding of how taxonomists have subdivided it. Two sectional classifications of modern wide-sense Hydrangea have been published to date, both informed by phylogenetic work as well as morphology (De Smet et al. 2015; Yang et al. 2024). The De Smet classification is a straightforward one with a single tier, splitting the genus into 16 sections. The Yang classification was based on phylogenomic work taking into account a wider range of East Asian species, especially in more doubtful groupings. This expands the classification to 19 sections nested in 5 subgenera. However, the differences are not extreme.

The two underlying phylogenies both indicate that a vast majority of species belong to two major clades, Hydrangea I (widespread including East Asia) and Hydrangea II (East Asia only). Although they diverged early in the history of the genus, they diversified at different times (Yang et al. 2023). Hydrangea I radiated mainly between 20 and 8 MYA, while the diversity in Hydrangea II we see today dates only from about 7 MYA. Relationships within Hydrangea I seem well established, and the two classifications hardly differ here. The main differences lie in interpretation of relationships within Hydrangea II, which includes important species such as H. macrophylla, H. serrata, H. luteovenosa and H. pottingeri (still better known as H. chinensis), as well as the dichroas. Even before the molecular era this was a hotbed of taxonomic angst.

The main practical difference between the classifications is that the Yang group subdivide De Smet’s Section Chinenses, which they consider polyphyletic. While we agree that in the light of current data Chinenses is probably not a natural group, we are not convinced that understanding of the true relationships here is yet settled. A second difference is that confusingly named Section Sinenses (not Chinenses!) is split from Section Decumaria to reflect improved understanding of the relationship between former Decumaria and Pileostegia species. On balance, we choose to follow the simpler De Smet classification here, highlighting uncertainties and the few places where the Yang system diverges. A brief description of each section follows; we include for completeness two entirely herbaceous and one tropical section, all of whose species are beyond the scope of Trees and Shrubs Online. Then come two tables, one comparing the De Smet and Yang classifications, the other relating the De Smet system to the pre-molecular consensus, as used in most or all garden literature to date (classification of Hydrangea sensu trad. from McClintock (1957), plus eight satellite genera).

Unless stated otherwise, plants of each section have opposite leaves, a partially inferior ovary, seeds with wings at one or both ends, and the fruit a dry capsule.

Section Asperae

Deciduous (rarely evergreen) shrubs from Asia. Inflorescences terminal, sometimes also axillary; sterile flowers usually present. Ovary inferior; styles ~2, free; stamens ~10 (but very numerous in H. platyarguta). Capsule apex truncate. This section consists of the tight-knit H. aspera complex plus four morphological outliers, one quite extreme; molecular studies agree that this is a natural grouping, however. The H. aspera complex (medium to large shrubs, sometimes with notably large leaves, with hairy – sometimes strikingly hairy – young shoots and leaf undersides) varies considerably, but defining segregate species defies consensus. We describe the grouping under H. aspera, then give shorter treatments to the arguably distinct HH. kawakamii, longipes, robusta, sargentiana, strigosa and villosa. Of the remaining species, H. sikokiana is a narrow endemic of southern Japan, with unusual lobed leaves. H. involucrata and the rare evergreen H. longifolia have extraordinary globular inflorescence buds, tightly surrounded by broad bracts, much like those of Section Cornidia. Finally H. platyarguta is so distinct that it was formerly given its own genus, as Platycrater arguta. Its few-flowered inflorescences have showy fertile flowers with enlarged white or pink sepals, many yellow stamens, and unusually long styles. (Clade: Hydrangea I).

Section Broussaisia

Evergreen shrub or small tree with large, thick, opposite leaves (more rarely in whorls of three). Dioecious, with male or female flowers in terminal inflorescences; sterile flowers absent. Ovary almost fully inferior; styles more or less fused into a column, with a 4–5-lobed stigma; stamens 10. Fruit a red, fleshy or mealy berry (Fosberg 1939). This is the former genus Broussaisia, a single species Hydrangea arguta (Gaudich.) Y.De Smet & Granados, from the Hawaiian Islands. This outlandish plant is rarely seen outside Hawaii, would almost certainly need tropical glasshouse protection anywhere in our area, and so is beyond our scope. (Phylogenetic position: basal branch of either Hydrangea I or II, or indeed of the genus as a whole – De Smet et al. 2015; Raman et al. 2023; Yang et al. 2024)

Section Calyptranthe

Deciduous woody climbers. Inflorescence with white flowers; sterile flowers usually present. Petals joined together at their apices to form a cap (calyptra) which falls at anthesis. Ovary inferior; styles 2–3, free; stamens 10–20. Seed encircled by a wing. 2 species in East Asia: we describe both H. anomala and H. petiolaris. (Clade: Hydrangea I).

Section Cardiandra

Herbaceous perennials with alternate leaves. Inflorescence with white to pink or purple fertile and sterile flowers, sterile ones with 2–3 enlarged sepals. Styles (2–)3(–4), stamens very numerous. These are the former Cardiandra species. Sometimes described as shrubs or subshrubs (e.g. Pan et al. 2001), the most recent botanical revision (Ohba 1985a, b) unambiguously refers to them all as herbaceous, and this is how they behave in cultivation. There are 2–4 species in East Asia, all in Western cultivation: H. alternifolia Siebold (Japan), H. amamiohsimensis (Koidz.) Y.De Smet & Granados (Ryuku Islands), H. densifolia (C.F.Wei) Y.De Smet & Granados (= Cardiandra formosana Hayata; China (Zhejiang), Taiwan), and H. moellendorffii Hance (China, Ryuku Islands). H. densifolia and H. moellendorffii are sometimes included in H. alternifolia. There are also garden hybrids from deliberate crosses between between H. alternifolia and H. densifolia, H. × agricola (J.M.H.Shaw) J.M.H.Shaw (Shaw 2008), and a number of cultivars. All are elegant plants for the woodland garden, but beyond our scope being herbaceous. (Clade: Hydrangea I).

Section Chinenses

Deciduous (rarely evergreen) shrubs from Asia. Inflorescences terminal, sometimes also axillary; sterile flowers usually present. Stamens ~10. Capsule apex conical. Weakly defined morphologically, as usual in the Hydrangea II clade, and perhaps always a taxonomic bin for awkward cases, this grouping nevertheless has a recognisable ‘look’ with rather narrow leaves and relatively small ‘lacecap’ inflorescences. Species definitions are in long-term flux, and for horticultural convenience we treat them as the H. scandens/pottingeri complex, describing HH. angustipetala, pottingeri (= chinensis), liukiuensis, lobbii, luteovenosa, and scandens itself. Others such as H. kawagoeana, probably not grown in the West, are not described here. On morphological grounds at least, the shadowy H. davidii also appears to belong here. However, there is increasing molecular evidence that this is not a monophyletic grouping. Yang et al. (2024) separated some taxa into new Sections Kawagoeanae and Kwangsienses in an attempt to remedy this, but the whole story of relationships among these hydrangeas is surely still to be told.

Section Cornidia

Evergreen woody climbers with leathery leaves, sometimes extremely vigorous. Plants usually functionally dioecious. Inflorescence terminal, sometimes also axillary, tightly surrounded in bud by a few large bracts which fall at anthesis; sterile flowers often but not always present. Ovary inferior; styles 2–4, stamens 10 or fewer. A large (and growing!) number of species, mostly from the mountains of tropical South and Central America, with three temperate outliers. We describe the temperate species H. integrifolia (Taiwan, the Philippines), H. seemannii (N Mexico), and H. serratifolia (Argentina, Chile). Since some of the tropical species are occasionally grown and may prove surprisingly hardy, we give an overview of these under Hydrangea Species of Section Cornidia. (Clade: Hydrangea I).

Section Decumaria

Evergreen and deciduous woody climbers. Inflorescence terminal, with white flowers; sterile flowers absent. Petals free, persistent at anthesis. Ovary inferior; style short and stout, lobed or with stigmatic lines; stamens 20–30. De Smet et al. (2015) place the two former Decumaria species here; we describe H. barbara (southeastern North America) and H. obtusifolia (= D. sinensis, China). Their classification was based on a phylogeny which omitted H. obtusifolia. Other phylogenies which include it are inconsistent (Raman et al. 2023; Yang et al. 2024). The Yang classification places H. obtusifolia in its own Section Sinenses, but there is no doubt that these plants are closely related to one another and to Sections Pileostegia and Schizophragma. (Clade: Hydrangea I).

Section Deinanthe

Herbaceous, rhizomatous perennials. Stems unbranched with 2–4 large leaves crowded near the apex. Inflorescence terminal, few flowered; fertile and (usually) sterile flowers present, but most unusually the somewhat nodding fertile flowers tend to be larger and showier than the sterile ones. Style with 5 lobes, stamens very numerous. These are the two former Deinanthe species, H. bifida (Maxim.) Y.De Smet & Granados (scattered in S Japan; white flowers, leaf apex usually bifid), and H. caerulea (Stapf) Y.De Smet & Granados (a narrow endemic, W Hubei, China; blue flowers). Both are woodland plants of moist valleys, with relict distributions in the wild (Sakaguchi et al. 2021); H. caerulea is considered endangered (Xu & Xiao 2022). They make beautiful garden plants in cool, woodland settings with abundant moisture during the growing season coupled with good drainage, but are beyond our scope being herbaceous. (Clade: Hydrangea I).

Section Dichroa

Evergreen to semi-evergreen shrubs. Inflorescences terminal and often also axillary, without sterile flowers. Fertile flowers with narrow but showy, coloured petals which often recurve, opening from conspicuous obovoid to globular buds; styles 2–6, free or joined only basally; stamens usually 10 or fewer. Fruit a fleshy blue to purplish berry. This is a taxonomically difficult and under-researched section, with very roughly 12 species in Asia and New Guinea, one highly variable and widespread, the rest much more local and tightly defined. We first describe widespread H. febrifuga, discussing general issues with the Section here, then go on to the more local H. daimingshanensis, H. hirsuta and H. hwangii. (Clade: Hydrangea II).

Section Heteromallae

Deciduous shrubs, sometimes approaching small trees; leaves opposite or in whorls of three. Inflorescences terminal, with fertile and sterile flowers, usually white, often ageing pink or reddish. Styles 3–4, joined basally; stamens ~10. Capsule with conical projecting apex. Seeds winged at both ends. Asiatic plants; we describe both H. paniculata and the H. heteromalla complex, a thorough discussion of the complex under H. heteromalla, then shorter treatments of the more widely accepted segregates H. bretschneideri and H. xanthoneura. (Clade: Hydrangea I).

Section Hirtae.

Deciduous shrubs from East Asia. Inflorescences terminal, sterile flowers absent (if H. hirta is the only species included!) Styles 2–4; stamens ~10. The De Smet classification based this Section on H. hirta alone. The Yang classification includes the mysterious H. davidii here (not sampled by De Smet et al. 2015) as well as H. shaochingii Chun (a rarely recognized Chinese species not described here, previously considered to be in the orbit of H. pottingeri). The brief morphological notes and uncertainties here reflect the ongoing difficulty of classifying species in the Hydrangea II clade.

Section Hydrangea

Deciduous shrubs. Inflorescences terminal, sterile flowers usually present. Flowers white; ovary inferior; styles 2–4; stamens 10. Capsule with truncate apex; seeds not winged. Four species in Eastern North America; we describe the H. arborescens complex (HH. arborescens, cinerea and radiata), and H. quercifolia with its unusual lobed leaves. (Clade: Hydrangea I).

Section Macrophyllae

Deciduous to semi-evergreen Japanese and Korean shrubs. Inforescence terminal, sometimes also axillary; fertile and sterile flowers present in wild-type plants. Styles ~3; stamens 10. Despite the weak morphological description typical of sections in the Hydrangea II clade, the two species (H. macrophylla and H. serrata) are familiar to gardeners and share much in their general ‘look’.

Section Pileostegia

Evergreen woody climbers. Inflorescence terminal, with white flowers; sterile flowers absent. Petals joined together at their apices to form a cap (calyptra) which falls at anthesis. Ovary inferior; style short and stout, ~5-lobed; stamens 8–10. Fruit dehiscing irregularly along ribs; seeds winged at both ends. Most closely related to the climbing Sections Decumaria and Schizophragma. 2 species in China and Japan; we describe more widely cultivated H. viburnoides, and briefly discuss H. tomentella which might perhaps be grown. (Clade: Hydrangea I).

Section Schizophragma

Deciduous woody climbers. Inflorescence with white flowers; sterile flowers usually present, each normally with only one enlarged sepal. Fertile flowers with a single short style and 10 stamens. Seeds winged at both ends. These are the former Schizophragma species, about 7 in Asia, still better known under their Schizophragma names in gardens. They are remarkably similar to one another, and taxonomically troublesome (Liu & Zhu 2011). We describe H. ampla (= S. integrifolium), H. faurei, H. glaucescens, H. hydrangeoides and H. schizomollis (= S. molle). (Clade: Hydrangea I).

Section Stylosae

Deciduous East Asian shrubs. Inforescence with fertile and sterile flowers. Styles ~4; stamens 10. Weakly defined morphologically, and included in the Macrophyllae by McClintock (1957), it was set up by De Smet et al. (2015) to cover H. stylosa and the poorly understood H. indochinensis, both described here. (Clade: Hydrangea II).

Sections following De Smet et al. (2015)

Equivalent following Yang et al. (2024)

Section AsperaeSubgenus Hydrangea, Section Asperae
Section BroussaisiaSubgenus Dichroa, Section Broussaisia
Section CalyptrantheSubgenus Hydrangea, Section Calyptranthe
Section CardiandraSubgenus Cardiandra, Section Cardiandra
Section ChinensesSubgenus Dichroa, Sections Chinenses, Kawagoeanae, Kwangsienses
Section CornidiaSubgenus Hydrangea, Section Cornidia
Section DecumariaSubgenus Decumaria, Sections Decumaria, Sinenses
Section DeinantheSubgenus Cardiandra, Section Deinanthe
Section DichroaSubgenus Dichroa, Section Dichroa
Section HeteromallaeSubgenus Heteromallae, Section Heteromallae
Section HirtaeSubgenus Dichroa, Section Hirtae
Section HydrangeaSubgenus Hydrangea, Section Hydrangea
Section MacrophyllaeSubgenus Dichroa, Section Macrophyllae
Section PileostegiaSubgenus Decumaria, Section Pileostegia
Section SchizophragmaSubgenus Decumaria, Section Schizophragma
Section StylosaeSubgenus Dichroa, Section Stylosae

Sections following De Smet et al. (2015)

Equivalent following McClintock (1957) with satellite genera

Section AsperaeSection Hydrangea, Subsection Asperae, & Genus Platycrater
Section BroussaisiaGenus Broussaisia
Section CalyptrantheSection Hydrangea, Subsection Calyptranthe
Section CardiandraGenus Cardiandra
Section ChinensesSection Hydrangea, Subsection Petalanthae in part
Section CornidiaSection Cornidia, Subsections Monosegia & Polysegia
Section DecumariaGenus Decumaria
Section DeinantheGenus Deinanthe
Section DichroaGenus Dichroa
Section HeteromallaeSection Hydrangea, Subsection Heteromallae
Section HirtaeSection Hydrangea, Subsection Petalanthae in part
Section HydrangeaSection Hydrangea, Subsection Americanae
Section MacrophyllaeSection Hydrangea, Subsection Macrophyllae in part
Section PileostegiaGenus Pileostegia
Section SchizophragmaGenus Schizophragma
Section StylosaeSection Hydrangea, Subsection Macrophyllae in part

Flowers, Flowering & Fruit

Hydrangea foliage is often attractive, but flowers are surely the primary reason why most gardeners choose them. They are never solitary, and usually in many-flowered inflorescences. These are complex branching structures, some with a single main axis (e.g. H. paniculata), but more often with several branches from the base. The terminology of inflorescence form is full of ambiguity and conflicting definitions (Prenner, Vergara-Silva & Rudall 2009), but it seems fair to call these panicle-like and umbel-like respectively. Inflorescences are usually at the ends of shoots, but sometimes also on short axillary shoots.

In most species there are two quite different types of flower within the same inflorescence. The small ‘fertile’ flowers have both male and female organs, petals – usually quite small, but white or coloured, and only very small, greenish sepals. The ‘sterile’ flowers tend to be around the margin of flat or domed inflorescences, but this is not always the case. Such flowers have enlarged, petaloid sepals, making each far larger and showier than the fertile flowers. Sometimes they have a few reduced or even full-sized sexual organs and / or petals, so may not always be strictly fertile, making the name seem inappropriate. Several alternatives have been suggested, including ‘decorative flowers’ (Wong Sato & Kato 2019), ‘marginal flowers’ (Yang et al. 2024), ‘flag flowers’, and ‘ray’ flowers or florets (Foster 2023), but none seem to have caught on, and like the majority we stick with ‘sterile’.

It is often suggested that the sterile flowers have a role in attracting insect pollinators (Darwin 1877); this would certainly make sense, increasing the diameter of the inflorescence’s outline at little extra cost to the plant. Other hypotheses include the enlarged sepals protecting the fertile flowers from rain (Kerner von Marilaun 1895), or acting as landing pads for insects (Wong Sato & Kato 2019). Experimental support for the ‘attraction’ hypothesis comes from a comparison of insect visits to intact inflorescences and those with the sterile flowers cut off, in nine Japanese Hydrangea species: intact inflorescences received significantly more visits; even visits per unit area increased, though less so (Wong Sato & Kato 2019). Lacking sterile flowers is not necessarily a problem. Most plants of H. serratifolia have none; their dense inflorescences dominated in male plants by masses of creamy stamens are very conspicuous, at least to the human eye. The species of Section Dichroa have only fertile flowers; their petals are larger than in most other hydrangeas.

In most species with sterile as well as fertile flowers, the sterile flowers tend to encircle the inflorescence – the ‘lacecap’ form (Haworth-Booth 1984). Mutants with far more numerous sterile flowers covering the head, obscuring the fertile flowers, are known as ‘hortensias’ (usual in continental Europe) or ‘mopheads’ in English-speaking countries. They have been found in wild populations of several species, but are best known in H. macrophylla-based cultivars. In these, both mophead inflorescence and double flower (with extra whorls of petaloid organs) are single gene recessive characters (Waki et al. 2018).

There have been few pollination studies in natural habitats, but it is generally assumed that hydrangeas have a generalist pollination strategy. This is certainly true of H. arborescens, whose flowers attract diverse insects, especially bees and flies (Robertson 1892; Pilatowski 1982). Garden experience of various species suggests the same (pers.obs., UK). With very few exceptions, mophead cultivars of the H. arborescens complex were less visited by insects than wild type (lacecap) forms in a Delaware, USA, garden setting (Hoadley 2021): perhaps insects quickly learn that there are fewer rewards to be had.

The mutability of flower colour in hydrangeas (pink or blue?) is well known far beyond gardening circles. This is most pronounced in H. macrophylla; many other (but not all) species are affected to a lesser extent. Aluminium availability is central (Chalker-Scott 2018 and references therein). Flowers tend toward blue when available aluminium is abundant (typically on acidic soils, or with supplements), pink when at low levels (as on alkaline soil, where aluminium becomes bound to clay particles). Soil pH itself is not the cause: acidic but aluminium deficient soils grow pink hydrangeas. Nor is iron involved as many gardeners believe: pH certainly affects iron availability, but this is not the mechanism here, and iron supplements will not affect colour. For a more detailed discussion see under Hydrangea macrophylla.

In some hydrangeas, flowers are initiated during the season they open, i.e. ‘flowering on new wood’. H. paniculata and the H. arborescens complex are good examples, and will flower well even if heavily pruned in winter or early spring. In far more species, though, they are initiated the previous year and spend the winter inside dormant buds. Most H. macrophylla cultivars fall into this category; loss of old wood to winter cold or heavy pruning, or even simply removing a short section of each old shoot with the terminal bud will drastically reduce flowering the following year. Cultivars which will also flower on axillary shoots, or terminally on the current year’s growth are a major breeding aim (see H. macrophylla).

Most hydrangea fruits are not attractive in themselves, but in Section Dichroa the fruit is a fleshy berry rather than a dry capsule, usually blue but purplish on alkaline soils. This transforms the appearance of the plant, although in structural terms the difference is small. If a dichroa berry is compared with a capsule from a relative such as H. pottingeri or H. macrophylla, they look much the same, the fruit wall of the dichroa berry having simply swollen up. Dichroas which fruit well in the open garden (e.g. H. febrifuga Long March Group) may hold their berries through the winter, apparently ignored by birds. Even in other groups the ageing inflorescence is not necessarily a withered brown mess to be tidied away immediately. Sepals of sterile flowers can be robust and long lasting, sometimes changing colour gradually through subtle ‘antique’ shades over several months; in mopheads where sepals dominate the head, this is especially ornamental. In colder areas old inflorescences may even give the terminal buds a modicum of frost protection.

The Climbing Habit

The climbing way of life probably evolved twice in the history of Hydrangea, once in an ancestor of Sections Calyptranthe and Cornidia, once in an ancestor of Sections Decumaria, Pileostegia and Schizophragma (Yang et al. 2024). Like ivies (Hedera spp.) they are root climbers, with non-twining stems held tightly against tree bark or rock surfaces by vast numbers of short, slender aerial roots. Their mode of growth is similar, whichever section they belong to, although some are evergreen, others deciduous.

Seedlings on the forest floor develop branching horizontal stems growing across the ground. At least where studied, erect side branches grow from buds at the nodes: these are able to colonize any vertical surface they meet (Kato et al. 2014). In some species at least a proportion of these shoots are negatively phototropic, tending to grow towards the shadowy bases of tree trunks. Once onto a trunk, boulder or cliff, extension shoots attached by aerial roots rapidly colonize the tree or rock face. In subsequent years much shorter, non-rooting side shoots grow out from this framework, away from the branch. In many species – certainly in H. hydrangeoides (Ichihashi & Tateno 2011) – these probably carry the bulk of the leaf area, and only side shoots flower. Leaves on extension and side shoots may look confusingly different (Shaw 2012). Like most other root climbers they cannot colonize fine branchlets and so remain in the inner part of the tree canopy, tend not to overpower it, and do not normally cross to neighbouring trees as some twiners may (Ichihashi & Tateno 2011). However, some cornidias (often immensely vigorous plants) are able to produce small tree-like canopies themselves, high in the host canopy and potentially overtopping it (Samain & Martínez Salas 2015).

Understanding their growth pattern suggests methods for propagating vegetatively and establishing climbing hydrangeas, as practised by Crûg Farm Plants, UK (B. Wynn-Jones pers. comm. 2023) but deserving much wider use. For propagation, creeping juvenile shoots rather than aerial side shoots are used. Stock plants are grown in raised nursery beds without vertical surfaces nearby. Mats of branched, creeping shoots develop, rooting in the soil. Chunks are cut out with a spade each year from the mat’s outer reaches, and potted up for sale. On planting, stems should be pegged down along the base of a wall or tree trunk; side shoots which break from lateral buds will readily attach themselves. Too often the nursery trade seems to supply rooted cuttings taken from aerial side shoots, which can be very slow to attach. Many years ago the author attempted in ignorance to establish H. obtusifolia from just such a plant, pressing the shoots back against a wall – rather than the soil at its base – with sticks. This was a spectacular failure: after 5 years it had formed a whippy, much branched shrublet no more than 30 cm tall without once attaching itself.

Economic Uses

It is hard to find evidence that people have used hydrangeas for anything other than ornament and medicine. Hydrangea febrifuga in particular is well known for its use against malaria and other parasites in traditional East Asian medicine (Wang et al. 2024), and members of the H. scandens complex have also been used in this way (Khalil et al. 2003). North American H. arborescens root is said to be diuretic, used by the Cherokee and early European settlers to treat kidney stones (Henkel 1907) while the peeling bark was used to treat diverse ailments by the Cherokee and others (Moerman 2003). Mexican H. seemannii has been used traditionally to treat diabetes; extracts have been shown to reduce blood sugar levels in mice (Pérez-Vásquez et al. 2013). However, these are toxic plants containing a range of cyanogenic glycosides, coumarin glycosides and iridoid glycosides. Any brief search of research literature will reveal a great number of studies on the possible effects of these substances on diverse disease processes. While there seems to be little scientific evidence that hydrangeas contain hallucinogens, there have sometimes been fashions for smoking dried flowers or leaves in the hope that they do (Samuel 2014), probably misguided given their toxicity. A report of H. anomala being used for animal fodder in India (Pant & Samant 2010) is surprising. It undoubtedly as ornamental and cut-flower plants that hydrangeas have their biggest impact on human activity.

Hybridization & Breeding

Interspecific hybrids in Hydrangea are known both in the wild and in gardens, but they are not commonplace. Most are between closely related species; the few known intersectional hybrids are between members of closely related sections. Garden experience is that within tight-knit groupings such as the H. aspera or H. arborescens complexes, apparently hybrid seedlings between different segregate species turn up quite often. H. macrophylla and H. serrata, a closely related pair, apparently hybridize in nature where their ecological ranges overlap (Hirasawa 2017), as well as crossing freely under both garden (Foster 2023; Reed & Rinehart 2007) and experimental (Reed 2000) conditions. Known intersectional hybrids are all artificial, except for some involving Section Dichroa (see especially Hydrangea × versicolor). Even then, the few intersectional crosses resulting in viable seed without resorting to embryo rescue techniques are between members of more closely related sections (Bak & Han 2024). For example climbing Hydrangea ‘Inovalaur’ (SEMIOLA) came from a deliberate cross between H. petiolaris (Section Calyptranthe) and H. seemannii (in sister Section Cornidia). See ‘Hydrangea Intersectional Hybrids’ for more details. Even some species pairings within sections require ovary culture / embryo rescue for success, for example crosses between H. arborescens and H. quercifolia in Section Americanae (Reed 2000; Bak & Han 2024).

Historically, deliberate breeding has been focused within a few species, especially H. macrophylla. In the West, this began around 1903 with French nurseryman Victor Lemoine raising outcrossed seedlings from some of the founding cultivars of European H. macrophylla stocks, which had originated in East Asia where garden races were long established. The extent to which these founders (and hence the European cultivars) carried genes from H. serrata is still debatable. Lemoine’s lacecap cultivars ‘Mariesii Grandiflora’, ‘Mariesii Lilacina’ and ‘Mariesii Perfecta’ are still grown. Mophead breeding soon followed, with Émile Mouillere introducing the enduring cultivars ‘Générale Vicomtesse de Vibraye’ and ‘Madame Émile Mouillère’ in 1908. Breeding work spread across Western Europe, the market driven by the potted plant trade in areas too cold for these to be reliable open-garden plants. Breeders favoured showy mopheads in strong colours, which suited pot culture and could easily be forced into flower for a long selling season. Quite incidentally, some of the myriad cultivars bred for this trade or for cut flowers proved gardenworthy. The Swiss-bred Teller Series, introduced from the 1960s, brought lacecaps back into serious commerce; these were widely tested in the open garden before release, and many (e.g. ‘Blaumeise’) have enduring quality. British, North American and New Zealand markets in the early to mid 20th century were largely fed by continental European breeding (Foster 2023; Haworth-Booth 1984).

From the late 20th century there have been some new directions in H. macrophylla-based breeding, with North American breeders and some in Japan with a global focus joining the Europeans. Extended flowering period (‘remontancy’ to hydrangea afficionados) has been a major concern. One element in this is the ability to flower on new wood, which is essential for flowering in cold areas where the aerial shoots die back in winter. Other ornamental developments include ‘picotee’ flowers with a darker pigmented edge to the sepals, double flowers, and black-stemmed varieties which improve on the old ‘Nigra’ (Boebel 2011, 2018; Dirr 2021; Foster 2023). More deliberate attempts to bring in genes from H. serrata for hardiness have been significant. Intersectional hybrids between H. macrophylla and members of Section Chinenses are now coming over the horizon, with more restrained habit and profuse axillary flowering important aims: the success of H. ‘Ushyd0405’ (RUNAWAY BRIDE®, FAIRYTRAIL BRIDE® etc.) is a taste of what is to come.

Since the mid-20th century, hardier H. paniculata has been a second focus for serious commercial breeding, from a very few much earlier introductions. Jelena and Robert de Belder of Arboretum Kalmthout, Belgium, began the process by raising seedlings from ‘Floribunda’ in 1958 (Hoffman & Houtman 2007). Three generations of seed raising produced an influential set of cultivars including ‘Unique’, ‘Brussels Lace’ and ‘Interhydia’ (PINK DIAMOND). European and North American breeders have continued the process in earnest. Inflorescence size, compact form, the stem strength needed to support huge flower heads, early flowering and precocious pinking of the sepals have all been breeding aims (Dirr 2021). The ratio of sterile to fertile flowers varies more continously than in H. macrophylla, and the term mophead is rarely used, but the equivalent ‘dense’ and ‘very dense’ inflorescences have generally been favoured, sometimes to the horror of those with ‘refined’ taste.

A third modern focus has been the North American species. Some cultivars in the H. arborescens complex are chance seedlings or wild finds, but there has been concerted effort to breed mopheads with stronger stems than the classic, floppy ‘Grandiflora’ and ‘Annabelle’; pink mopheads have been another preoccupation (see under Hydrangea arborescens). However, the great majority of cultivars in H. quercifolia turn out to have been wild finds, particularly in Alabama.

The Cultivar Conundrum

There are – or have been – immense numbers of Hydrangea cultivars, thousands rather than hundreds. Cultivar names are indispensible when discussing garden plant variation, but they present their own problems. Some cultivar names distinguish horticulturally significant points within the diversity of a single species; these might be extreme, rare, or especially desirable variants, but equally might simply represent a commonly grown if unexceptional form of a variable species. They might originally have been vegetatively propagated from a wild plant, selected from a batch of wild seed, appeared by chance in a garden as a seedling or branch sport, or resulted from deliberate crossing of quite different variants of the species. Such cultivars do not stand apart from the ‘true’ species, they belong to it.

Cultivar names also allow us to name garden hybrids in a way compatible with the nomenclature of wild plants. Botanists have described very few nothospecies in Hydrangea, latinized names for particular hybrid combinations (e.g. Hydrangea × versicolor for all hybrids involving only H. macrophylla and H. febrifuga). Cultivar names can be given to distinctive clones within such hybrid species, for example H. × versicolor ‘Cambridge Blue’. In all other cases cultivars are correctly listed under the generic name only, hence Hydrangea ‘Inovalaur’. In hydrangeas a degree of hybridity is often suspected, but nobody is sure. Some of the H. macrophylla-like cultivars definitely have genes from H. serrata in their makeup, but perhaps they are present in many more. We might reflect this by listing a classic mophead as simply Hydrangea ‘Altona’, without specifying a species.

Just as with specific and generic names, a code exists to bring order to cultivar names (Brickell et al. 2016). One important principle is that each cultivar has only one valid cultivar name, usually the first to be published. In recent decades sections of the nursery trade have increasingly used trade designations – little-regulated marketing names for plants which normally also have a valid cultivar name. Some are registered trademarks, some not; many apply to cultivars subject to plant patents or other legal protections, others do not. They are not written in quotes and are distinguished by a different typeface, often – as here – appearing in block capitals, viz: Hydrangea BLUE WAVE is a trade designation of the cultivar Hydrangea ‘Mariesii Perfecta’. A single cultivar may carry more than one trade designation, often in different parts of the world for language or other cultural reasons. More confusing still, protected varieties are often given formal cultivar names which appear meaningless, do not always make pronounceable words and in practice force growers and gardeners to use the trade designation, with any legal restrictions it may carry. For example, Hydrangea BLUEBERRY CHEESECAKE, H. COTTON CANDY and H. TUFF STUFF™ are all trade designations of Hydrangea ‘Mak20’. Sometimes it is very hard to be sure which of several options is the formal cultivar name. Occasionally an extra layer of confusion is added when the same attractive name is picked as the cultivar name of one plant and as a trade designation for another. For example, Hydrangea ‘After Midnight’ and H. ‘Jong02’ (AFTER MIDNIGHT) are entirely different mophead cultivars.

Cultivar series add further complications. Common in today’s plant marketing world, they are usually a range of complementary varieties (a colour range perhaps) from a single breeder, often with a great deal of common genetic heritage. All this is true of the Teller Series of Swiss-bred lacecap hydrangeas introduced from the 1960s onwards. Increasingly though, at least in Hydrangea, there is a tendency for marketing companies to put together series or ‘collections’ of various breeders’ cultivars, often through licensing agreements but sometimes including older, unprotected cultivars (e.g. Royalty® Collection). The series concept then blurs into the idea of a plant brand, very much a feature of the contemporary garden centre trade, especially in North America (McClellan 2024). Some brands include plants from diverse genera, but marketing information and labels may include the brand name as if it were part of a trade designation or cultivar name (e.g. “Hydrangea FLAIR & FLAVORS® BLUEBERRY CHEESECAKE” (Proven Winners 2025), this mouthful containing almost everything but the cultivar name).

In Trees and Shrubs Online we try to use the accepted cultivar name, however ‘silly’ the name, as the primary listing, giving more significant trade designations as synonyms. This is not a happy situation, but every alternative introduces worse problems. The Search facility will locate a cultivar if only a trade designation is known. In the more taxonomically challenging areas it is usually easy to place a cultivar within a species complex, but often impossible or meaningless to assign it to a segregate species. Hence we have separate articles for the following: H. arborescens Species Complex Cultivars, H. aspera Species Complex Cultivars, H. heteromalla Species Complex Cultivars, H. Section Macrophyllae Cultivars (i.e. H. macrophylla, H. serrata and hybrids between them), and H. Intersectional Hybrids. Where there are large numbers, we divide them between alphabetical pages (e.g. H. Section Macrophyllae Cultivars E–F); the many H. paniculata cultivars are also listed across several alphabetical pages.

Our aim is to be thorough but not comprehensive in cultivar coverage. Especially in H. paniculata and Section Macrophyllae there are simply too many cultivars – often here today and gone from the nursery trade tomorrow – for it to be sensible to list everything. That is the role of a cultivar register (Association Shamrock 2022). We have tried to include what is likely to be encountered as established plants in today’s gardens and collections, emphasizing important modern cultivars, older varieties which have persisted well, historically significant plants, and anything else which is particularly distinct and interesting. Some cultivars have picked up multiple names over time; we follow the Register where it is clear on which is the correct name, unless we have reason to disagree. Sometimes there is real uncertainty.

In the Garden, on the Nursery

Climbers apart, the great majority of hydrangeas are woodland floor plants from latitudes 20–40° N, in regions with significant summer rainfall. This goes a long way towards explaining their garden requirements. Hydrangea specialist Maurice Foster sums up the ultimate conditions for growing a range of species as follows: moisture retentive but well drained, fertile loam with plenty of organic matter, and a pH of 6.5 or lower (if blue flowers are to one’s taste), with good light levels but shaded during the hottest, brightest part of the day (Foster 2023). This reflects his own experience in Southern England, with its unpredictable rainfall pattern and (by the standards of wider temperate regions) lack of temperature extremes, high or low, but is a good starting point anywhere.

Waterlogging in winter is almost always to be avoided, and heavy clay should be improved before planting. Summer drought is a problem to most hydrangeas, but a few species are more tolerant of dry soils: H. villosa and H. involucrata ‘Hortensis’ succeeded on dry chalk soil in Frederick Stern’s Highdown garden in Sussex, UK (Bean 1981). A good annual mulching, especially around newly planted hydrangeas is immensely valuable whatever the soil, working against both these harmful extremes.

When it comes to shade, there is a trade-off to be made between coolness and light levels. In most hydrangeas more light gives better flowering, however shade tolerant the plant may be. This is readily seen in many of the climbers: on a north-facing wall (in the Northern Hemisphere) they may grow well, but most flowers will be in the brighter light at the walltop, or where shoots hang over on the sunny side. At least in southern Britain, H. involucrata does not fit this pattern, flowering well even in quite deep shade (Foster 2023; pers. obs.). In full sun, light intensity itself can be a problem in variegated cultivars, the yellow or white zones tending to scorch, and white flowered hydrangeas in particular may brown prematurely, but heat and the drying effect are probably the more general issues (Foster 2023). Species growing well in full sun as well as shade include H. paniculata and H. arborescens. The trend towards more intense summer heatwaves in cool maritime Europe and the Pacific Northwest should focus gardeners’ minds when making new plantings. At lower latitudes with the sun higher in the sky, the trade-off shifts, flowering in shade becoming less of a problem. Many H. macrophylla-based cultivars are recommended for much shadier conditions in New Zealand than they would be in Britain or Belgium, for example (Church 2007).

Few hydrangeas tolerate the cold winters of Central Europe or the American Midwest. H. paniculata and H. serrata are among the hardier species. For those intent on growing H. macrophylla-based cultivars in cold (approximately USDA Zone 6) areas, Tim Boebel (2011; 2018) has developed precise pruning and mulching techniques which have proved very successful for some cultivars (see under H. macrophylla for details). In the end though, microsites matter, whichever environmental variable is in question. The better one knows one’s own garden, the more effectively one can identify the ideal spots for hydrangeas.

The light / shade balance makes open woodland settings ideal for many hydrangeas, edges of clearings and pathsides being especially appropriate. In smaller gardens the mixed border is a natural home for them. Shade in the middle of the day might be provided by a single specimen tree, or taller shrubs ‘pruned up’ (‘one trick wonders’ such as Chimonanthus praecox or larger Philadelphus could be put to use in this way). Tall H. heteromalla, broad mounds of H. aspera, or H. paniculata pruned annually to a 1–1.5 m tall framework can provide an attractive backdrop to herbaceous plants. Less vigorous H. arborescens cultivars pruned to ground level each winter take their place among such perennials. Shrubby hydrangeas are a popular choice alongside buildings, walls giving both some shade and protection from winter cold. The author’s earliest childhood memories of hydrangeas are of pink mopheads invariably growing against walls in East Sussex villages of the late 1960s, not long after Britain’s severe winters of 1946–7 and 1962–3. A less common use is in loose, informal hedges; the prunability of H. paniculata makes it suitable (Foster 2023).

As with any other woody ornamental, hydrangea pruning is about aesthetics rather than plant survival. Almost any woody hydrangea will shoot readily from the base if cut back hard. This is useful when removing old, weak growth or renewing an over-large or poorly formed shrub, but in most cases there is a cost in flowering terms. Hydrangeas fall into four pruning categories (Brown & Kirkham 2004); if in doubt, do less rather than more.

Firstly, H. paniculata and the H. arborescens complex flower terminally on the current year’s growth. Annual pruning is thus possible without loss of flowers, but is far from essential. Annual pruning of H. paniculata brings larger flower heads and more compact – or at least more controlled – form. Initially, stems are cut back only to form a permanent skeleton, short or tall, broad or narrow as desired; thereafter, annual growth is cut back to two buds in late winter. Thinning out the shoots as they grow can further increase panicle size (Bean 1981). Pruning to four buds gives intermediate size and vigour, while postponing pruning until as late as the end of April (southern England) can delay flowering if desired (Lancaster & Wesley 2008). H. arborescens spreads by underground stems to form dense, slowly spreading clumps. It is possible to cut stems back to ground level in late winter, the plant then behaving almost as a herbaceous perennial. Cut-back plants tend to grow less tall, and the vigorous first-year stems have larger inflorescences, potentially eye-catching but prone to flopping in weaker stemmed cultivars (Hoadley 2021), just as in H. paniculata. Otherwise, old inflorescences can be tidied away in winter, leaving the framework of stems intact.

The second group includes the H. macrophylla- and H. serrata-based cultivars, which form clumps shooting readily from the base. These flower terminally, from buds formed the previous year. They can be left entirely alone, but will flower better if in early spring, before bud-break, the older and weaker stems are cut out to the base, and the old flower heads cut out only as far back as large, strong buds. For specialized pruning techniques in cold areas see Boebel (2011; 2018) and under H. macrophylla.

Thirdly, the less bushy H. aspera and heteromalla complexes, as well as H. quercifolia, are best left to form a permanent framework, with only conspicuously weak growth cut out in spring or after flowering.

The final group is the climbers. In the crown of a large tree there is little that can or needs to be done, but on a wall it is often necessary to limit the plant’s ambition. Once a permanent skeleton of rooted stems is established, new extension growth should be cut out as soon as is feasible, ideally as it appears through the summer. If over time the mass of flowering side shoots hangs out too far, these can be cut back in spring to a suitable bud close to the shoot base; to avoid a year without flowers it is wise to spread this over a few years, cutting back a proportion each spring.

When propagating hydrangeas, seed is rarely the first choice, but is necessary with wild-collections, when selecting new cultivars or otherwise looking for a genetically diverse batch of plants. Seeds are minute, best collected by cutting off an entire inflorescence just as capsules begin to dehisce (late autumn to early winter for many species in southern England), and placing it upside down on a sheet of paper indoors, away from draughts. Fresh seed surface-sown in winter usually germinates rapidly (Foster 2023). Most work on germination in Hydrangea broadly suggests that light stimulates germination. Various studies (mainly in H. macrophylla and H. paniculata) show that vernalization is not required, although exposure to cold may increase germination rates (Greer & Rinehart 2010).

Most species are easily increased vegetatively from cuttings. Semi-ripe wood in early summer usually roots quickly in mist, a domestic propagator or a pot enclosed in a clear plastic bag; cuttings with two or three pairs of buds are ideal (Foster 2023; van Gelderen & van Gelderen 2004). Hairy-leaved species may require more care to avoid fungal infections. Hardwood cuttings in late autumn are feasible, but usually slower to root and require care to bring them through the winter (pers. obs.). A tiny amount of material may still make a viable cutting. Foster (2023) describes rooting a variegated sport of H. ‘Mariesii’ from a single node and leaf. Mark Fillan (pers. comm. 2024) demonstrated that single winter buds of H. ‘Garden House Glory’ removed entirely from their parent stems in winter could be rooted in mist and be grown on to produce liner-sized plants by summer. In nursery practice, rooted cuttings were grown in the field until the later 20th century, when pot growing in highly regulated environments became the norm (van Gelderen & van Gelderen 2004).

Layering too is feasible and sometimes happens spontaneously in large, old clumps of some species left to their own devices. It is most often used in large-leaved H. quercifolia and the H. aspera complex (Bean 1981), as well as in climbers (see above). Grafting is very rarely used in Hydrangea. Tissue culture protocols exist for several Hydrangea species (e.g. Ruffoni, Sacco & Savona 2013). Ease of cuttings means that the main use of such techniques is always likely to be rapid build-up of propagating stock in newly developed cultivars. They are essential, though, in establishing novel intersectional hybrids where embryo rescue is required.

In most areas Hydrangea is a mercifully trouble-free genus in the garden. Grazing mammals – whether deer or rabbits – rarely touch them, perhaps thanks to their toxicity, but slugs and snails can be damaging. Among sap-suckers, Hydrangea Scale (Pulvinaria hydrangeae) is an emerging problem, introduced to Europe and North America, probably from East Asia (Salisbury & Malumphy 2017); in parts of Europe it has become a serious pest in urban environments, although Foster (2023) finds it an uncommon one in Britain. If one has to resort to insecticides, midsummer might be the best time when the more susceptible nymphs are present (Salisbury & Malumphy 2017). Mildews can affect hydrangeas, but are usually a symptom of suboptimal growing conditions (Foster 2023; van Gelderen & van Gelderen 2004); plants in the garden may be disfigured, but not killed, and the best solution is to rectify underlying cultural problems. In the American Southeast a fungal leaf spot caused by Cercospora hydrangea disfigures garden plants and nursery stock (Smith 2025) but is much less common or problematic further north and in Europe. Fungicides are used in the nursery trade, but where it is a serious issue in the open garden it seems wise to choose less susceptible hydrangeas: there is considerble variation even between cultivars (Dirr 2021).

Seeing Hydrangeas, Learning More

So many hydrangeas are hardy only in the milder parts of our area, that the better collections are inevitably towards the maritime margins of Western Europe and North America. Many gardens have fine displays, few have comprehensive collections which allow thorough comparisons. One of the Western world’s best and most complete is at Jardin Shamrock, Varengeville-sur-Mer, on the northern French coast (Collection Mondiale d’Hydrangea Shamrock 2024). In Britain, hydrangeas across a wide range of groups are prominent at both RHS Garden Wisley and the Savill Garden, Surrey. Knowledgeable volunteers maintain and continue to develop an impressive collection at Darley Park, Derby; its centrepiece is the near-comprehensive National Collection of H. paniculata cultivars filling a large walled garden (Hydrangea Derby 2025). Susan Miller’s National Collection of H. serrata in the West of Scotland is open seasonally, and is particularly interesting as an example of how to integrate large numbers of hydrangeas of different types into botanically diverse, attractive plantings in a medium-sized domestic garden (Plant Heritage 2025). Trebah Garden in coastal Cornwall is famous for spectacular massed plantings of mophead cultivars dating from the 1950s, predominantly in blue (Trebah Garden 2025). In Belgium there are large collections at among others Arboretum Wespelaar, Merksplas Botanic Garden, Meise Botanic Garden and Arboretum Kalmthout. Many of the best collections however are in private gardens not routinely open to the public. Joining the International Dendrology Society and other specialist plant groups is a fine route into visiting the widest range of woody plant collections and meeting knowledgeable enthusiasts.

Important North American collections are mainly on the Pacific and Atlantic seaboards. University of British Columbia Botanical Garden has a large, botanically diverse collection (University of British Columbia 2025). Other Northwestern collections with many hydrangeas include Washington Park, Seattle (University of Washington Botanic Gardens 2025) and Heronswood Garden, Washington State (Heronswood Garden 2025). On the East Coast, the Cape Cod Hydrangea Display Garden, Massachusetts, has impressive displays, particularly of H. macrophylla-based cultivars (Heritage Museums and Gardens 2025), and Mt Cuba Center, Delaware recently completed a thorough trial of H. arborescens complex cultivars (Mt. Cuba Center 2021). Further south in our area, the JC Raulston Arboretum, North Carolina has a diverse collection (JC Raulston Arboretum 2025), but many of the Southeast’s great hydrangea gardens lie beyond our scope, in Georgia and Alabama for example.

A number of good horticultural books complement our account of Hydrangea, though none to date properly cover the enlarged genus. Foster (2023) is an excellent general account from a British standpoint, the author having particular expertise in the H. aspera group and H. serrata. French authorities Corinne & Robert Mallet have written several useful books pitched at different levels; Mallet, Mallet & van Trier (1992) is available in English and German translations as well as the original French. They have also been instrumental in setting up the International Cultivar Register (Association Shamrock 2022). Coming from a Dutch nursery background, van Gelderen & van Gelderen (2004) illustrate and describe a very large range of cultivars. Once a standard work, Haworth-Booth (19501984) feels increasingly outmoded, mainly of historical interest. Glyn Church (2007) writes directly and with great enthusiasm from a New Zealand standpoint. North American plantsman and hydrangea breeder Michael Dirr’s unique perspective on the genus is presented in his (2004) book, along with an updated version (Dirr 2021) with much on recent breeding work which cannot be ignored.

Turning to scientific literature, there is no botanical revision of broad-sense Hydrangea, and treatments of the old narrow-sense genera are patchy. The most recent revision of narrow-sense Hydrangea is McClintock (1957), whose very broad species concepts are out of step with most contemporary thinking. Sections Pileostegia and Schizophragma are well served by a recent revision (Liu & Zhu 2011), while the evolving taxonomy of Section Cornidia is covered by an important series of research papers (Samain, Hernández Najarro & Martínez Salas 2014; 2019; Samain, Granados Mendoza & Martínez Salas 2021; Granados Mendoza et al. 2021). Beyond these recently revised groups, the best overviews are in regional floras. The vast majority of species (apart from the New World cornidias) fall under the remit of Flora of China (Pan et al. 2001), Flora of Japan (Ohba 2001), and Flora of North America (Freeman 2016). These floras include some contentious taxonomic views. This is Hydrangea, after all, and anyone serious about the genus must get used to that.

Identification key (species only)

1aPlant herbaceous2
1bPlant woody3
2aClump forming perennials with alternate leaves along stemsSection Cardiandra (not treated further here)
2bRhizomatous perennials with 2–4 large leaves clustered near the stem apexSection Deinanthe (not treated further here)
3aWoody climber
3bNon-climbing shrub, rarely a small tree17
4aInflorescence buds large, globular, enclosed by broad involucral bracts; evergreen5 (Section Cornidia)
4bInflorescence buds not as above; deciduous or evergreen7
5aAll flowers fertile, in thyrse-like inflorescences; domatia clearly visible on lower leaf surface; Chile and ArgentinaH. serratifolia (see also Hydrangea Species of Section Cornidia)
5bSterile and fertile flowers present, in umbel-like inflorescences6
6aSterile flowers ringing inflorescence, sepals 1–2 cm; MexicoH. seemannii (see also Hydrangea Species of Section Cornidia)
6bSterile flowers scattered through inflorescence, sepals 1–1.5 cm; TaiwanH. integrifolia (see also Hydrangea Species of Section Cornidia)
7aAll flowers fertile8
7bSterile and fertile flowers present11
8aPetals joined apically to form a cap which falls as flowers open; stamens 8–109 (Section Pileostegia)
8bPetals free; stamens 10–2010 (Section Decumaria)
9aBranchlets, leaves and inflorescences glabrous; widespread in E Asia, commonly cultivatedH. viburnoides
9bBranchlets, leaves and inflorescences with dense, rufous stellate hairs; SE China; at best very rare, probably absent from Western gardensH. tomentella
10aLeaves deciduous to semi-evergreen, 10–12 × 6–8 cm; N AmericaH. barbara
10bLeaves fully evergreen, 3.5–7 × 2–3.5 cm; ChinaH. obtusifolia
11aSterile flowers with >1 enlarged sepal; fertile flowers with petals joined apically to form a cap which falls as flowers open, stamens 10–2012 (Section Calyptranthe)
11bSterile flowers usually with a single enlarged sepal; fertile flowers with petals free, stamens ~1013 (Section Schizophragma)
12aStamens <15H. anomala
12bStamens >15H. petiolaris
13aLeaf margin roughly serrate; undersurface whitish, densely hairy on midvein and in vein axils; Japanese, Korean and Russian Islands, very commonly grownH. hydrangeoides
13bLeaf margin entire, finely serrate or serrulate; mainland E Asia and Taiwan14
14aLeaves glaucous beneath, with prominent hair tufts in vain axils; S China, rare or absent from Western gardensH. glaucescens
14bNot as above15
15aLeaves glabrous beneath; mainland ChinaH. ampla
15bLeaves tomentose beneath16
16aLeaves densely brown-tomentose beneath; mainland China, VietnamH. schizomollis
16bLeaves brown-tomentose on main veins and vein axils beneath, at least at first; TaiwanH. fauriei
17aPetals of sterile flowers at least 5 mm long18
17bPetals of sterile flowers less than 5 mm long, usually ~2 mm19
18aLeaves deciduous. Sterile flowers sometimes present, with broadly ovate sepals. Fertile flowers with very numerous and conspicuous yellow stamens; petals ovate, typically broadly so, white (rarely pink). Fruit a dry capsuleH. platyarguta (Section Asperae)
18bMore or less evergreen. Sterile flowers absent. Fertile flowers with 4-10(-20) stamens; petals more narrowly ovate, elliptic or oblong to lanceolate, blue to pink (rarely white). Fruit a fleshy berry.21 (Section Dichroa) – but see H. febrifuga regardless
19aOvary inferior, both at anthesis and in fruit; capsule apex truncate20
19bOvary only partially inferior, at least in fruit; capsule conical at apex36
20aSeeds not winged; E United States24 (Section Hydrangea)
20bSeeds winged; Asia27 (Section Asperae)
21aBranchlets, petioles, veins (on both leaf surfaces) and inflorescences conspicuously hairy with both short, crisped and long, straight hairs; calyx lobes 4–6 mm, hairy; S China (Guangxi, Yunnan), VietnamH. hirsuta
21bNot as above20
22aBranchlets, petioles and veins with both short, crisped and long, straight hairs; calyx lobes ~1 mm, glabrous; S China (Guangxi, Guizhou), ?VietnamH. daimingshanensis
22bNot as above21
23aLower leaf surface conspicuously hairy with crisped hairs on and between the veins; styles usually only 3; Yunnan, ?VietnamH. hwangii
23bNot as above (a widespread and variable species; Himalaya, through China to SE Asia)H. febrifuga
24aInflorescence pyramidal; leaves pinnately lobedH. quercifolia
24bInflorescence flat to domed; leaves not lobed25 (H. arborescens species complex)
25aLeaf undersurface green, glabrous or with sparse hairs on the veins; sterile flowers rarely presentH. arborescens
25bLeaf undersurface greyish to silver or white, pubescent; sterile flowers sometimes present26
26aLeaf silvery or white beneath, with a matted, felt-like indumentumH. radiata
26bLeaf greyish beneath, hairs not dense enough to appear mattedH. cinerea
27aLeaves shallowly but sharply lobedH. sikokiana
27bLeaves not lobed28
28aInflorescence buds 'drumstick-like', tightly enclosed in large bracts29
28bNot as above30 (H. aspera species complex) – also see H. aspera regardless
29aLeaves lanceolate, rough with simple and 2-branched hairs; endemic to Taiwan; very rare in gardensH. longifolia
29bLeaves oblong to obovate, rough with only simple hairs; endemic to Honshu; frequent in gardensH. involucrata
30aStems, petioles and leaf midribs densely covered in long, fleshy, transparent, tapering trichomesH. sargentiana
30bNot as above31
31aStems, leaf undersides and inflorescence branches rough with short, stiff, appressed, white hairsH. strigosa
31bNot as above32
32aLeaf coarsely velvety beneath, with short but not stiff, spreading white hairs thickly covering every veinH. villosa
32bNot as above33
33aPetioles mostly more than half the length of leaf blade34
33bPetioles less than half the length of leaf blade35
34aLeaves thin, with sparse, appressed, white hairs; fertile flowers with white petals and stamensH. longipes
34bLeaves thicker textured, usually more densely hairy; petals and stamens pigmentedH. robusta
35aNative to TaiwanH. kawakamii
35bNative to mainland AsiaH. aspera
36aInflorescence paniculate, with elongated central axisH. paniculata
36bNot as above37
From this point in the key, be aware of the very poorly understood H. davidii and H. indochinensis, which are not included
37aLower leaf surface glabrous, or with a few short, crisped hairs in the vein axils or veins38
37bLower leaf surface significantly pubescent, at least on the major veins40
38aInflorescence 10–20 cm across, leaves thick-texturedH. macrophylla
38bInflorescence 4–10 cm across, leaves thick or thin39
39aStyles 2.5–3 mm, petals reflexed, especially after floweringH. stylosa
39bStyles 2–2.5 mm, petals not notably reflexedH. serrata
40aSterile flowers absent41
40bSterile flowers present42
41aLeaves longer than 4.5 cm, with coarse, regular serrationsH. hirta
41bLeaves shorter than 4.5 cm, only weakly serrate towards the apex; endemic to Okinawa, probably not in cultivationH. liukiuensis (Section Chinenses)
42aPetioles 2–4 cm long43 (H. heteromalla species complex) – also see H. heteromalla regardless
42bPetioles not more than 2 cm long45 (Section Chinenses) – also see H. pottingeri regardless
43aLower leaf surface pale green, glabrous apart from some greyish hairs on the veins; S ChinaH. xanthoneura
43bLower leaf surface grey to cream or white, more conspicuously hairy44
44aLower leaf surface grey hairy, at least when young, but not so dense as to be velvety; a shrub to 3 m; C to NE ChinaH. bretschneideri
44bNot as above; typically with lower leaf surface with velvety white indumentum; shrub or small tree to 5 m or moreH. heteromalla
45aLeaves under 3.5 cm longH. luteovenosa
45bLeaves more than 3.5 cm long46
46aMore or less evergreen shrub, leaves 8–12 cm long; Taiwan, at lower altitudes; scarcely hardy in our areaH. lobbii
46bDeciduous shrub47
47aSpreading shrub with thin straggling branches; leaves 4–7 cm long, thin textured; sterile flowers usually with 3 sepals; JapanH. scandens
47bErect shrub; leaves 8–12 cm long, thin textured; sterile flowers with ~4 sepals; N TaiwanH. angustipetala
47cErect shrub; leaves 5–12 cm long, thick textured; sterile flowers with 3–4 sepals; NE India, Myanmar, mainland China, S & C TaiwanH. pottingeri