A new spontaneous hybrid in Gunnera subgenus Panke (Gunneraceae) widespread in the British Isles, with notes on the typification of G. manicata

. Gunnera x cryptica J.M.H.Shaw representing the hybrid G. manicata Linden ex André × G. tinctoria (Molina) Mirb. is described from spontaneous seedlings in cultivation. It is widespread in the British Isles, and appears to have completely replaced G. manicata in cultivation. Details of the earliest publication, authorship and typification of G. manicata are provided, along with the origin and history of the hybrid, and notes on G. tinctoria including a previously overlooked synonym G. thyrsiflora Ruiz ex Barreiro.

For over 150 years, giant Gunnera plants have been widely cultivated for their spectacular sized foliage and architectural value, usually under the names G. tinctoria (Molina) Mirb. and G. manicata Linden ex André. However, distinguishing these two species of Gunnera has long been regarded as challenging (Goldring, 1879;Elwes & Stapf, 1919;Stapf, 1919;Clement, 2003;Grant, 2004). This has become more critical with the wider appreciation of the invasive threat posed by G. tinctoria (Silva et al., 1996;Taranaki Regional Council, 2007;Sheehy Skeffington & Hall, 2011), which is listed in UK Invasive Alien Species legislation as a Species of Special Concern and in the EU as a Species of Union Concern. In order to be able to understand better the distribution of these two taxa and to assist those managing natural areas where Gunnera occurs, as well as to provide guidance to gardeners and the horticultural sector, a study was undertaken to determine the identification of these species in UK and Ireland. This investigation used both morphological and molecular approaches and the latter is reported in another paper (Edwards et al., in prep.). Here we report on the principal conclusion on the identity of plants widely known in cultivation as G. manicata.
Both species originate from South America: G. manicata occurs in the Atlantic rainforest habitat in southern Brazil (Paraná, Rio Grande do Sul, Santa Catarina) (Hassemer, 2017(Hassemer, , 2019, whereas G. tinctoria occurs in Chile and S.W. Argentina where it occurs in a wide range of habitats, including the edges of forest and alongside rivers in the wetter parts of the country, especially the island of Chiloé (Gardner et al., 2015;Gioria & Osborne, 2013). Both species belong in subgenus Panke, and share the same chromosome number (2n = 34) (Dawson, 1983).
Introduction into cultivation in Western Europe in the latter part of the nineteenth century brought these two closely related species together. The results of the molecular analysis of samples from plants in their native ranges and in cultivation demonstrated that G. manicata is not found in cultivation in UK and Ireland, and the existence of a hybrid between the two species (Edwards et al., in prep.) which is here described. Spontaneous hybridisation between species of Gunnera subgenus Panke has been observed in the wild on several occasions (Palkovic, 1978;Mora-Osejo et al., 2011) so that this instance is by no means unique.  56: 97 (1904), (Arnott, 1908a;Nelson & Grills, 1998).
G. scabra var. longiscapa T. Smith ex Burbidge in Flora andSylva 1901: 294 (1901), (Smith, 1904;Nelson & Grills, 1998  The hybrid can attain larger dimensions than either parent due to hybrid vigour (Grant, 2004); however, caution should be exercised when using this to identify an individual because plant and leaf size are also strongly influenced by age and growing conditions. Observations from living plants demonstrate that the hybrid can reach greater height than either of its parents, the Brazilian G. manicata or Chilean G. tinctoria, and have larger leaves. However, it may be difficult to distinguish plants of either parent from the hybrid in publications using leaf measurements. Both in cultivation and in the wild leaf dimensions are considerably influenced by such factors as the age of the plant, depth of soil, water supply and nutrients. Exaggeration and 'guestimates' can also be responsible for large leaf sizes in publications. Container-grown plants produce smaller leaves, while carefully cultivated plants tend to become larger than their wild counterparts do. Just as skilled cultivators who produce prize-winning vegetables can produce plants of great size, so too competitive Victorian gardeners used various techniques to produce enormous plants of Gunnera. Records describe the excavation of large pits filled with rich compost (Jenkins, 1908), regular applications of manure (Burbidge, 1901;Arnott, 1908b) and artificial fertilizers (Bartlett, 1904(Bartlett, , 1908 and removal of inflorescences (Mayne, 1908).
Several well-illustrated identification guides available on the internet, describe the differences between G. tinctoria and G.  . 1) and an explanation on how to distinguish the two species (Goldring, 1879). By 1894 Gunnera seed obtained from Trelissick Gardens, Cornwall, and subsequently raised at Livermere Park, Bury St Edmunds, produced seedlings that varied greatly in colour, style, leaf shape and size (Tallack, 1894). In 1901 (by which time such variable seedlings would be flowering) Burbidge reported on a plant called G. scabra var longiscapa with very robust growth, that carried long tapering flower-spikes unlike the usual club-shape. (Burbidge, 1901). Here 'tapering' is comparable to 'conical' as used to describe the inflorescence of hybrid plants by some authors. It indicates a G. manicata influence. Burbidge also remarked on the seedlings varying in character, size and shape of the leaf, which he attributed to the "two or three named varieties of G. chilensis now found in gardens" (Burbidge, 1901). He elaborated, "There are also forms distinct in leaf: one has the leaves not fully spread but more or less rigidly cupped, the veins and footstalks bright red, and the leaf surface bronze-green; another has flat spreading leaves in which the ribs and veins are pale, and the leaf a purer green. Intermediate forms connect the two." Then in 1905 Tom Smith of Daisy Hill Nursery, Newry, lists several different varieties of giant Gunnera for sale (Nelson & Grills, 1998). These observations indicate that hybrid plants were appearing in cultivation from the late 1870s onwards. This date agrees well with a possible origin of the hybrid around 1873, by which time plants of G. manicata from Linden's first homeproduced seed batch had matured, resulting in greater availability of the new introduction so that large flowering specimens of both species were often grown together for a spectacular display. A circumstance that would have resulted in multiple origins of the hybrid G. × cryptica. Due to the loss of G. manicata seedlings and plants during cold winters (Tallack, 1894), along with garden selection for size and vigour, the hybrid gradually supplanted it. For example, McMillan Browse (2007) reported that at Heligan Gardens, Cornwall, the original plantings of giant rhubarb (Gunnera "manicata") thrived in the bogs of the lower valley. However, a severe late frost destroyed many tender plantings in April 1891; hence, much of what is currently present dates from the very end of the nineteenth century. Consequently, original planting that may have been G. manicata would unknowingly have been replaced by plants of G. × cryptica. While it is not impossible that genuine G. manicata may yet be rediscovered in a sheltered garden, this scenario seems increasingly unlikely. Consequently, all plants from the British Isles and Ireland examined in this study were determined to be either G. tinctoria or G. × cryptica (Edwards et al., in prep.).

Application of the name Gunnera manicata.
Since G. manicata was described from cultivated material grown by Linden, and the type specimen was obtained from a plant cultivated in Britain about 20 years later, the application of the name G. manicata had to be established. The question arises whether the name should apply to the plants found wild in Brazil or to the hybrid from cultivation here named as G. × cryptica? The matter is complicated by the first appearance of the name in horticultural literature and so the following notes on the authorship and first publication of the name, and origin of the type material, are included. These update the accounts provided by Wanntorp (2003) and Shaw (2007). Should it prove necessary, an alternative later name for the Brazilian plants is available as G. brasiliensis Schindler, the application and typification of which is considered in detail by Hassemer (2019). Interestingly the type collection of G. brasiliensis, E.H.G.Ule 1229 (P) appears to consist of material from more than one individual plant, including several immature seedlings with immature foliage, and a detached solitary immature larger leaf. There is a detached partial inflorescence (on P00238859) which is the only extant mature plant part present since the duplicate at B was destroyed. Consequently this inflorescence should be designated the lectotype specimen of G. brasiliensis Schindler. A possibility is that this unusual collection may represent seedlings from a wild seed collection subsequently cultivated, perhaps at the Botanical Garden in Rio de Janeiro. This could account for the locality on the label being Rio de Janeiro followed by Serra do Oratario, which two locations are geographically distant.

The author citation for Gunnera manicata.
When Wanntorp et al. (2002) investigated the history of the name G. manicata they concluded that André (1873) had validated Linden's prior use of the name, which they considered was published as a nomen nudum. Subsequent research by Shaw (2007) drew attention to the previous validating description of Delchevalerie (1867) and Linden ex Delchev. became the recognised authority for G. manicata (Mora-Osejo et al. 2011). However, further investigation has found that prior to this Linden (1865) and André (1866) provided short descriptive phrases, which should be taken into consideration. The first use of the name G. manicata can now be traced to Linden (1865) where plants were offered for sale at 1 franc, with the comment "nouveau et très-beau". As this does not enable comparison with another taxon, it is to be regarded as an earlier nomen nudum.
In André (1866) the description of G. manicata included "…à feuilles plus grandes encore et très-longuement pétiolées." [with even larger leaves and very long petioles]. This enabled comparison with G. tinctoria (as G. scabra), the only other large Gunnera of subgenus Panke then in cultivation. It would appear that this publication of the name has previously been overlooked or dismissed as being a nomen nudum, but on advice from R. Govaerts (Kew) this is accepted as a validating diagnosis.

Typification of Gunnera manicata Linden ex André.
When Baker (1886) described G. manicata Linden, citing Linden's (1867b) publication of the name with its very minimal description and no type specimen citation, his detailed description was based solely on one specimen (F.Ross s.n., K).
In accordance with ICN, Art. 9, Note 1, (Turland et al., 2018) this specimen must be accepted as the holotype of Baker's name, and consequently G. manicata Linden was inadvertently neotypified. Subsequently, Wanntorp et al. (2002) neotypified G. manicata Linden ex André (based on André, 1873) using this same specimen, unaware of the previous typification. This investigation has now shown that G. manicata Linden ex André based on André (1866) is the earliest known valid publication of the name but also was not typified at the time. In order to ensure consistent application of the name G. manicata, the same specimen used by Baker (1886) and Wanntorp et al. (2002)  Origin of the plant from which the neotype specimen was made.
The neotype specimen was collected from Sir George Macleay's garden in 1886. It has been established that this garden was at Pendell Court, which is near Bletchingly, Surrey (Boulger, 2004). It appears that Macleay took up residence at Pendell Court in the late 1860s, initially as a tenant and subsequently purchasing it in 1876 (Anon. 'B', 1886;Driver, 2012Driver, , 2016Wilman, 2021). Indeed, this is mentioned in a letter from J.D. Hooker to Charles Darwin dated 6 or 7 July 1870 (Darwin letter 7267), where he comments, "We spent last Sunday at Mr G. Macleay's who has taken Pendell Court near Bletchyngly [sic]…" In 1876 Macleay appointed the noted plantsman Charles Green as Head Gardener, who supervised the remodelling of the gardens (Driver, 2012). It seems likely that he had previously become acquainted with the work of Charles Green through William Saunders as both held office within the Linnean Society of London. At the time, Green was head gardener for Saunders' collection. After Saunders experienced financial difficulties, Green was able to find employment with Macleay at Pendell Court, after a brief period attempting to run his own plant nursery (Driver, 2012(Driver, , 2016. In 1878, two years after Green started work at Pendell Court, a feature article in Gardeners' Chronicle on the gardens at Pendell Court noted, "At one end of the lake is a noble clump of Gunneras, G. manicata, here far exceeding in the magnitude of its leaves (4 feet across) its near ally G. scabra [G. tinctoria], growing in juxtaposition. A nobler composition than this group of Gunneras on a slightly sloping bank at the end of the lake, with a thriving young Oak as a background, would be difficult to realise." (Anon, 1878). Shortly before this on 4 June 1878 the RHS Floral Committee had awarded Gunnera manicata a First Class Certificate, based on an inflorescence and a young leaf of G. manicata, exhibited by Charles Green (Denny, 1878).
While there is no evidence of the source of the Gunnera manicata planted at Pendell Court at that time, for it to have achieved the size it had by 1878 argues for it being first planted some years before while Macleay was renting the property, or that it arrived at Pendell Court as a well-grown plant (cf. Gumbleton, 1884). A possible explanation is that it had been acquired by Green who, while working for Saunders, had opportunity to purchase stock directly from William Bull. Gunnera manicata is first offered for sale in the UK in Bull's nursery catalogue in 1867, and the description he gives is a direct translation of that provided in Linden's catalogue (Linden, 1867a), which suggests that Bull was able to secure plants directly from that source.
Taking into account the size and therefore the likely age of the Gunnera at Pendell Court it seems most probable that the plant was one of the seedlings from the original introduction of G. manicata being distributed by Linden from 1867. While elsewhere we indicate that the hybrid probably arose sometime around 1873, this was at the nursery of van Houtte, and would seem to be too recent to have produced a plant the size of that evident at Pendell Court in 1878. Unfortunately, there is no longer any trace of Gunnera at the Pendell Court estate, now The Hawthorns School, and so living plants could not be examined. However, the school provided a monochrome print of unknown date, depicting a large plant of G. manicata growing by a lakeside at Pendell Court (Fig. 2). From the leaf features visible, it is identifiable as G. manicata.

Identity of the type
William Goldring (1854Goldring ( -1919 was a gardener at Kew from 1875-1879, before becoming assistant editor of The Garden in 1879. Therein he authored an article on Gunnera, in which he enumerated the differences between the two large species, basing his observations on the G. manicata plant at Pendell Court, which he acknowledged as the largest, and therefore possibly the oldest, he had encountered (Goldring, 1879). The illustrations accompanying his article (Fig. 1) depicted material from this plant, which later became the source of the holotype. As the Pendell Court plant had acquired this reputation, it is understandable that seven years later when faced with a large Gunnera leaf from the garden curator at Kew requiring verification, J. G. Baker of the Kew herbarium, after expressing astonishment at the absence of any herbarium material, applied to Pendell Court for reference material (Baker, 1886). The resulting leaf and inflorescence became the type of G. manicata. Morphological examination of the neotype from this site, comprising five sheets, and the published description (Baker, 1886) shows that it represents the Brazilian species rather than the hybrid. The leaf parts of the plant that were available on the neotype herbarium sheets were photographed and printed out on paper and the original shape was partly reconstructed by matching up the available fragments. The discernible shape and vein branching pattern matched Brazilian G. manicata rather than the hybrid. This also was compared with images and drawings in Fevereiro & Barbosa (1976), herbarium sheets from NY and colour images of wild plants collected in Brazil. In addition, the length of the inflorescence branches on the neotype exceeds that recorded for the hybrid, and matches the Brazilian material, as do the flowers. The complete flowers of G. manicata are only known from drawings made in Brazil by Fritz Muller (Fig. 3) that appear in a letter to C. Darwin dated 12 September 1875 which Darwin passed on to Hooker at Kew. The drawing is reproduced in Gonzalez & Bello (2009), and Fritz Muller's drawings provided the basis for the plates in Schwacke (1890) and Schindler (1905). Muller lived in Blumenau, Santa Catarina state, from where the Serra do Mar was easily accessible. This was also the locality where Libon made his original collection (Stapf, 1919). The flowers in G. manicata are variable with apparently a mixture of functionally pistillate, staminate and hermaphrodite flowers present on most inflorescence branches. Muller stated (folio 1 of the letter) that perfect flowers, meaning those with all parts including petals as he illustrated (Fig. 3), only occurred towards the branch apex. An observation that has been confirmed for 24 other related species mostly in subgenus Panke (Gonzalez & Bello, 2009). However the sepals are polymorphic, sometimes reduced to triangular enations like those illustrated for G. × cryptica (Fig. 4), but often elongated with a stipitate swollen base (Fig. 3) which may contain the basal hydathode reported by Gonzalez & Bello (2009) that is thought to ensure Nostoc travels with the developed fruit to aid the resultant seedling. These unusual sepals do appear to be unique to G. manicata in subgenus Panke, and are present on the neotype specimen of G. manicata at Kew providing a useful confirmation, linking the cultivated neotype with Brazilian material.
Hence, from morphological and to some extent historical evidence the neotype specimen of the name G. manicata applies to the Brazilian species and the hybrid G. manicata × G. tinctoria should be described under a new name, G. × cryptica. Library research has also revealed that it was in André (1866) that the geographical misinformation that G. manicata came from 'Nouvelle-Grenade', now Colombia, was first introduced, rather than the usually cited André (1873) as by Stapf (1919), Wanntorp et al., (2002) and Wanntorp (2003).

A note on Gunnera tinctoria.
During this investigation, it has become apparent that G. tinctoria (Fig. 5) is itself poorly understood. It is a complex and variable taxon that invites further taxonomic investigation and may contain yet unrecognised entities. For instance, plants in cultivation and naturalised in the British Isles and wild in Chile are described as without petals. Those Floras that provide sufficiently detailed descriptions are all unanimous on this absence of petals, including Sell & Murrell (2007), Tebbitt (2011), Muñoz-Schick (1980) and Mora-Osejo et al. (2011). Schindler (1905) thought that flowers were imperfect in European cultivation due to the petals always aborting and falling early. Gonzalez & Bello (2009) did not include material of G. tinctoria or G. manicata in their study of intra-individual variation of Gunnera flowers. However, they concluded that floral unisexuality in subgenus Panke is restricted to the formation of female flowers by incomplete development of the androecium. Such flowers usually lack petals which appear to be associated with protection of the anthers and fall at anthesis. Possibly also relevant here is the intriguing report by Gardner et al. (2015) that G. tinctoria were observed that produced separate male and female inflorescences. In contrast, two publications from Argentina describe the petals, even illustrating them on a hermaphrodite flower, and employing them as key characters to separate G. tinctoria with entire margined, glabrous petals from G. apiculata Schindler with irregular margined, pilose petals. Molina (1978: fig. 3) provides an illustration based on O'Donell 2087 (BAB, LIL), collected from Neuquén, that clearly shows two petals present, though smaller than the anthers. Flora Patagonica (Molina, 1988) comments the flowers are generally perfect, meaning hermaphrodite with all parts present, and describes them thus, "petals 2, 2.5 x 0.5 mm, obovate, somewhat concave or navicular, with a small mucro, readily falling, glabrous." Thus the possibility exists that there is regional variation or an unrecognised taxon may be present.
During the course of this study a previously overlooked name for G. tinctoria was uncovered. The name Gunnera thyrsiflora Ruiz appears in Schultes & Jaramillo-Arango (1998: 212), which is an English translation based on a collation of two unpublished manuscript accounts in the library of the London Natural History Museum in Spanish by Ruiz, providing a narrative of the Ruiz and Pavon expedition to Peru and Chile from 1777 to 1788. An earlier draft manuscript by Ruiz discovered in Spain was edited and published in Spanish by Agustín J. Barreiro in 1931. As this predated the ICBN requirements for a Latin description or type designation, several previously overlooked names for South American plants were validated inadvertently in the Barreiro edition (Shaw, 2022a, b). In turn, Barreiro's edition was translated in to English by Dalhgren (1940). The Colombian scholar and botanist Jaramillo-Arango (1952, 2: 130) identified G. thyrsiflora with G. magellanica Lam., but in view of the vernacular name cited as Panke, the description of uses by natives, and the dimensions in the description it is evidently G. tinctoria that was described by Ruiz in his account. Gunnera thyrsiflora Ruiz ex Barreiro, Relaciόn del Viaje: 192 (1931). Lectotype here proposed: Ruiz & Pavon s.n. (S), which is also the holotype of G. scabra Ruiz & Pavon (1798). Both G. thyrsiflora and G. scabra are synonyms of Gunnera tinctoria (Molina) Mirb., Hist. Nat. Pl. 10: 141 (1805).
Additionally, it is worth noting that in the past several authors have confused G. tinctoria with other Gunnera taxa from western South America. Hence, its distribution has been said to extend into Bolivia and Colombia (Gioria & Osborne 2013). However, the species concepts and taxa delimited by Mora-Osejo et al. (2011) in Flora Neotropica, Gunneraceae are accepted in this study. This is of relevance to identifying and recording material present and previously grown in the British Isles and Ireland as numerous introductions of live material have been made into cultivation, some of which appear to represent G. tinctoria var. valdiviensis (L.E. Mora) Mora, Pabon-Mora & F.Gonzalez. Reports of cultivated plants of G. tinctoria that produce two distinct types of inflorescence from a single individual (such as at Wisley) may represent hybrids between var. tinctoria and var. valdiviensis -a fascinating possibility that requires further investigation.