Hydrangea serrata (Thunb.) Ser.

A small to medium deciduous shrub for lightly shaded situations, Hydrangea serrata is closely related to H. macrophylla but is smaller in all its parts, much hardier and earlier flowering. Long cultivated in Japan, variants in flower colour, inflorescence form and habit have all been selected. European and North American gardeners have rather neglected it in favour of H. macrophylla-based cultivars, but this is changing.

H. serrata is an altogether finer-textured shrub than H. macrophylla. It has much thinner branchlets, making an erect to spreading clump shooting less readily from the base, and rarely much more than 1 m tall in gardens, much shorter in some clones. Its leaves are thinner textured, finely or coarsely toothed, with a matt surface rather rough to the touch and often hairier on the veins, at least at first (Ohba 2001; van Gelderen & van Gelderen 2004). Inflorescences (typically lacecap form) are terminal, but sometimes also on short lateral branches. Sterile flowers have 3–5 sepals, sometimes unequal, with toothed or entire margins. At least in gardens, H. serrata begins flowering in early summer, significantly before H. macrophylla. As in H. macrophylla, flower colour is susceptible to aluminium availability, higher levels giving bluer colours, but some clones turn blue far less readily, if ever. White flowered forms are frequently grown, as well as ones in which red/pink pigmentation starts at the sepal margin, spreading only as the flower ages (the beni cultivars). Mophead (temari) and numerous double (yae) forms have been selected (Foster 2023). Leaf colour is generally more interesting than in H. macrophylla, with red or purple tints common on young growth and again in autumn.

Although it reaches sea level in the wild, growing alongside H. macrophylla on the Izu Peninsula, Honshu, and sometimes hybridizing with it (Hirasawa 2017) this is primarily an understorey shrub of montane forests, and is consequently much hardier. The terminal inflorescences are within winter buds formed the previous autumn, so winter-hardiness of the woody shoots allows flowering in cold areas (USDA Zone 6, or even below) where most H. macrophylla-based cultivars die back and barely flower without heroic efforts by the gardener. Dirr (2021) cites a plant in Iowa flowering on old wood after experiencing –34°C the previous winter, although this is quite exceptional. As a woodland plant, light shade and shelter from strong winds are strongly recommended – in some cold areas it is virtually ungrowable in exposed positions (Boebel 2011). A well-drained but not dry soil is (as so often) ideal (Foster 2023). Minimal pruning is advisable, as with H. macrophylla, but untidy, overmature bushes can be cut back to near the base to encourage strong young shoots, although there will be little or no flowering in the first year.

Forms of what we now know as H. serrata first came to Europe in the 19th century from Philipp von Siebold’s living collection; some must have been among the earlier arrivals since the first grown in Britain was imported (perhaps from Leiden) before 1844 by Hugh Low’s Clapton Nursery, London (Lindley 1844). By the end of the century several names which seem to relate to H. serrata forms, as well as probable hybrids, were current in Britain (Bean 1898). The great bulk of introductions have been in the 20th century, mostly established Japanese cultivars under correct, recognizable or garbled names, with a few wild collections by visiting Europeans and Americans thrown in, Ernest Wilson’s W 7820 (Shikoku, 1915) having a wider circulation than most of those and proving one of the hardiest in North America (Dirr 2021).

Since Thunberg (1784) first described the species (as Viburnum serratum), the complex patterns of variation within H. serrata have defied taxonomists’ efforts to catalogue them in a stable, comprehensible way. Numerous taxa have been described, particularly as varieties, but there is currently no single published account of the whole group which makes complete sense to outsiders. Hideaki Ohba’s work for Flora of Japan (Ohba 2001) has already been overtaken by his own revisiting of Siebold’s specimens from which many taxa were first described (Ohba & Akiyama 2013), while others have identified entirely new taxa (e.g. Yamazaki 2001). One recent molecular study suggests that this is a paraphyletic species, with H. macrophylla sister to only a subset of H. serrata (Hirota et al. 2022); some southern Japanese forms are then hived off into H. acuminata Siebold & Zucc., but this does not yet leave us with a usable scheme for the whole group. Modern Japanese work seems never to engage with Korean material; conversely a Korean molecular study apparently contains no Japanese H. serrata forms (Kim, Park & Park 2016). Rather than try to describe varieties (or further species!) in these shifting sands, we briefly mention those names most often seen in cultivation, with key features following Ohba (2001). H. serrata var. angustata (Franch. & Sav.) H.Ohba has lanceolate, sweet-tasting leaves under 10 cm long – var. acuminata is subsumed here on this scheme; var. thunbergii (Siebold) H.Ohba has smaller sterile flowers, with sepals under 2 cm long, with less attenuated bases and leaves sometimes sweet-tasting; var. yesoensis (Koidz.) H.Ohba has larger leaves and flowers than the type, leaves widely elliptic,10–17 × 5–10 cm, inflorescences 10–17 cm across, sterile flowers to 4 cm across. Plants in gardens under variants of these names often do not match the descriptions, or indeed one another – van Gelderen & van Gelderen (2004) nicely lay bare the confusion and uncertainty.​

Some H. serrata populations (including cultivars with amacha in their name – Foster 2023) accumulate phyllodulcin, a very sweet dihydroisocoumarin, in their leaves. The substance seems to be released enzymatically from a glycoside on leaf damage or drying (Suzuki et al.1977). They are cultivated commercially in Japan, harvested just before flowering, and fermented to produce the sweet tea amacha, widely drunk in Japan and beyond, as well as being poured ceremonially over statues of the Buddha (Kimberley’s Kyushu 2022; Wikipedia 2024). In a study of wild plants in a western Japanese beech forest (Ujihara, Shinozaki & Kato 1995), sweet plants were found to occur in discrete, local, subpopulations. Phyllodulcin levels peaked around flowering time and increased on partial defoliation. Transplantation experiments suggested that habitat did not affect the plant’s ability to make phyllodulcin; sweet and non-sweet plants were affected equally by leaf mining caterpillars, arguing against a defensive function.

H. serrata and H. macrophylla share the same chromosome number (2n=36, Cerbah et al. 2001) and can hybridize. Some of the older cultivars once attributed to H. serrata, such as ‘Grayswood’ and ‘Mariesii’ are probably hybrids (Association Shamrock 2022). Some are known to be the result of deliberate crosses (e.g. ‘Garden House Glory’, ‘Preziosa’, ‘Mak20’). However, there is real uncertainity over to what extent H. serrata genes are present in the great bulk of cultivars which have the general look of H. macrophylla. The alphabetical articles, ‘Hydrangea Section Macrophyllae Cultivars A’ etc, include pure H. serrata, pure H. macrophylla, known hybrids between them, and the great mass whose exact genetic heritage is unclear; anything else would introduce guesswork and controversy. We make clear the status of each (“near macrophylla”, macrophylla × serrata” etc.). The great majority of pure H. serrata cultivars we include are of Japanese origin. There seems to have been a perception in the European and North American nursery trades that there is not a large market for H. serrata selections (M. Foster pers. comm. 2023). In Britain, Maurice Foster has spent many years raising and selecting both H. serrata cultivars and macrophylla × serrata primary hybrids, with exceptional results: a few are illustrated below. They remain unreleased and without formal cultivar names: it is hoped that this situation will change.