In this short introduction to this issue, we deliberately use common names in the title, reflecting the complicated inter-relationships and taxonomic history of this group of native trees and their relatives. There are numerous botanical names available for many of the species involved (as well as for their relatives); across the community of researchers there is no clear consensus on which genera should be recognised, and widely differing systems have been proposed. In this issue we have decided to use the names for each taxon in Sorbus L., as this is the system used in the most recent treatment of the British and Irish taxa (Rich et al., 2010) and in the most recent flora (Stace, 2019), but noting that Sorbus in this sense (Sorbus sensu lato) is certainly not a natural (monophyletic) group. We point the reader in the direction of some of the important literature in which the complications and the reasons for them are discussed, beginning with a short account of the group of Rosaceae to which these species belong, Rosaceae subtribe Malinae. Then, we discuss the inter-relationships between the taxa treated here as Sorbus in more detail. Previously treated as Rosaceae subfamily Maloideae (for a discussion of this, see, e.g., Rich et al., 2010, and references therein), Rosaceae subtribe Malinae are now generally considered part of subfamily Amygdaloideae (see, e.g., Sun et al., 2018), sometimes incorrectly called subfamily Spiraeoideae. As with subfamilies, different names have been used at the subtribal level: Rosaceae subtribe Malinae have also been called subtribe Pyrinae, but the name Malinae has been conserved (Reveal, 2012); this is the name that should be used, following the International Code of Botanical Nomenclature (see McNeill & Turland, 2011). Rosaceae subtribe Malinae are widely distributed in the temperate parts of the Northern Hemisphere, and some taxa are widely planted elsewhere as fruit trees or ornamentals. They have been hypothesised to be of allopolyploid origin, resulting in the base chromosome number (n = 17) being higher than that found in other groups of Rosaceae (e.g. Evans & Campbell, 2002). Morphologically, they are characterised by the pome (the apple-like fruit), unlike related genera with dry dehiscent fruits that have been included in tribe Maleae by some authors; see, e.g., Sun et al., 2018, who recommended that these related genera (Gillenia Moench, Kageneckia Ruiz & Pav. + Lindleya Kunth, Vauquelinia Corrêa ex Bonpl.) should be treated as three small subtribes of tribe Maleae. As at higher taxonomic levels, taxonomy within Rosaceae subtribe Malinae has also been controversial, with widely differing numbers of genera being recognised by different authors. Linnaeus (1753) included species now regarded as belonging to Malinae in four genera (Crataegus L., Mespilus L., Pyrus L. and Sorbus L.), based largely on style number, but this classification was soon altered by other authors. Sorbus was included in Linnaeus' group characterised by three styles (“Icosandria Trigynia”), but the species he included in it differ in predominant style number: S. domestica L., TRUE SERVICE TREE, has five (seven) styles, whereas S. aucuparia, ROWAN or MOUNTAIN ASH, generally has three or four styles. As S. domestica is not a match for the definition of Sorbus sensu Linnaeus, S. aucuparia has been accepted as the type species of the genus. Readers who wish to know more about the details concerning this decision and taxonomic consequences of it are referred to Sennikov (2014). Three examples are given here to show that taxa in this group had been treated as members of several or many genera. Even before the end of the 18th Century, COMMON WHITEBEAM, Linnaeus' Crataegus aria L., had been transferred to Sorbus [as S. aria (L.) Crantz, in 1763], Mespilus [as M. aria (L.) Scop., in 1771], Pyrus [as P. aria (L.) Ehrh., 1784], Hahnia [as H. aria (L.) Medik., in 1793] and Lazarolus [as L. aria (L.) Borkh., in 1798], and names now exist for it in at least 11 genera and two nothogenera (‘genera’ resulting from hybridisation between members of two or more genera). See Fay et al. (2022a, this issue) for a partial list of synonyms of Sorbus aria. Likewise, APPLE (included in Pyrus by Linnaeus as P. malus L.) has synonyms in Malus Mill. [as M. domestica (Suckow) Borkh.], Pyrenia Clairv. and Sorbus. Linnaeus' Mespilus included taxa now often treated as members of Amelanchier Medik., Cotoneaster Medik., Pyracantha M.Roem. and Sorbus sensu lato (Chamaemespilus Medik.). These examples illustrate the dramatic changes in generic delimitation that have occurred in Rosaceae subtribe Malinae. Robertson et al. (1991) provided a useful historical summary of generic concepts used by different authors. Also in Malinae, many taxa in nature and in horticulture are the result of hybridisation between two (or more) genera; see fig. 1 in Robertson et al. (1991) for a summary of the intergeneric crosses known to occur. This situation results in the use of nothogenera, e.g. ×Sorbaronia C.K.Schneid. (= Sorbus × Aronia Medik.), ×Pyraria A.Chev. (Aria × Pyrus). ×Sorbocotoneaster Pojark. (= Sorbus × Cotoneaster) or ×Amelasorbus Rehder (Amelanchier × Sorbus). If narrower genera are used, notably in Sorbus sensu lato, some authors (e.g. Sennikov & Kurtto, 2017; Kurtto et al., 2018; Sennikov, 2018) have treated some of the resulting nothogenera as genera in their own right, and this can lead to the use of other names, including Hedlundia Sennikov & Kurtto [Aria (Pers.) J.Jacq. ex Host × Sorbus sensu stricto] and Scandosorbus Sennikov [= Aria × Sorbus sensu stricto × Torminalis Medik., replacing the illegitimate Borkhausenia Sennikov & Kurtto; see Sennikov, 2018]. The use of these genera/nothogenera does not, however, fully reflect the complexity of the situation (see below). In contrast with other widely recognised genera of Malinae (which are largely monophyletic, i.e. they form a natural group of species that are more closely related to each other than they are to species in other genera), Sorbus sensu lato is problematic, as it is certainly polyphyletic, i.e. it consists of lineages that are less closely related to each other than they are to other genera. Rushforth (2016) stated that ‘I clearly consider that the broad concept of Sorbus is pathologically flawed’, and he is not alone in thinking like this! Many papers have been published in which relationships between members of Sorbus sensu lato (and other Malinae) have been investigated on the basis of morphological characters and a range of molecular markers. However, these fail to reveal consistent patterns of relationships, in part due to differences in sampling and in part due to lack of support for critical branches in the phylogenetic trees. In addition, evidence from nuclear DNA suggests various patterns of relationships in the subtribe that are incongruent with those suggested by evidence from plastid DNA (probably largely due to the frequent occurrence of hybridisation). Campbell et al. (2007), for example, compared phylogenetic trees based on five nuclear and six plastid DNA regions. With the combined plastid regions, members of Sorbus fell in three places in the tree: Sorbus subgenus Aria Pers. [represented by S. alnifolia (Siebold & Zucc.) K.Koch 1] was related to Heteromeles M.Roem. and Cotoneaster; Sorbus subgenus Cormus (Spach) Duch. (S. domestica) was sister to Pyrus + Sorbus sensu stricto (represented by S. aucuparia and S. americana); and Sorbus subgenus Chamaemespilus (Medik.) K.Koch [represented by S. chamaemespilus (L.) Crantz] was sister to Sorbus subgenus Torminaria (DC.) K.Koch [represented by S. torminalis (L.) Crantz]. In other similar studies (e.g. Evans & Campbell, 2002; Creissen, 2009; Lo & Donoghue, 2012; Sun et al., 2018; Ulaszewski et al., 2021), the authors also failed to recover a monophyletic Sorbus sensu lato. In Europe, the recognition of narrower genera for the groups within Sorbus sensu lato to reflect the lack of monophyly has led to the use of ten genera (five non-hybrid, five hybridogenous) by, e.g., Kurtto et al. (2018) and Sennikov & Kurtto (2017), who proposed the many new binomial combinations necessary to reflect this generic treatment. [McAllister, 2005, in his monograph Mountain ash and other rowans (Sorbus sensu stricto), referring to the splitting of Sorbus sensu lato into separate genera stated that ‘the adoption of this classification will require the creation of many new combinations for the apomictic species of hybrid origin’]. Other narrower genera have been recognised for taxa in Asia. For example, Rushforth (2018) recognised seven additional genera (five newly described in that paper) for members of Sorbus sensu lato, bringing the number of genera to 17. Also in Malinae, he described another genus (Prameles Rushforth), bringing the number of genera in the subtribe recognised in his treatment to 45. Even in Britain, the use of separate generic names for the hybridogenous taxa by Kurtto et al. (2018) and Sennikov & Kurtto (2017), however, fails to address the full complexity of the situation, as some of the species result from more than one hybridisation event. For example, Robertson et al. (2004) showed that Sorbus arranensis Hedl. (ARRAN WHITEBEAM) was the result of hybridisation between S. aucuparia and S. rupicola (Syme) Hedl. (ROCK WHITEBEAM); in the system of Sennikov & Kurtto (2017), this was treated as Hedlundia arranensis (Hedl.) Sennikov & Kurtto. Sorbus pseudofennica E.F.Warb. (ARRAN SERVICE TREE), treated as H. pseudofennica (E.F.Warb.) Sennikov & Kurtto, is in turn the result of back-crossing of Hedlundia arranensis onto Sorbus aucuparia (Robertson et al., 2004). While including this species in Hedlundia reflected the broad contributions of the parental genomes, extending the splitting approach to its logical conclusion, the Arran service tree could be treated in another nothogenus with a name like “×Sorbohedlundia”. To make matters yet more complex again, Sorbus pseudomeinichii Ashley Robertson (FALSE ROWAN) is the result of a further round of back-crossing of the ARRAN SERVICE TREE onto S. aucuparia (Robertson & Sydes, 2006). Should this be in “×Sorbosorbohedlundia”? Including taxa like S. pseudofennica and S. pseudomeinichii in Hedlundia, as done by Sennikov & Kurtto (2017), means that Hedlundia becomes a name applied to taxa with distinct evolutionary histories. As an alternative to this, a system advocating a broad concept of Pyrus for the taxa in Sorbus sensu lato and some related genera of Malinae was proposed in Christenhusz et al. (2018). Advantages of this system were that the use of nothogenera could (largely) be avoided. This is line with treatments of other groups in which hybridisation causes taxonomic complexity. Three examples of such treatments are given here. In Orchidaceae subtribe Oncidiinae, a number of previously widely recognised genera have been subsumed into Oncidium Sw. (e.g. Chase et al., 2008), greatly simplifying the taxonomy of the many important horticultural hybrids in this group. In Asparagaceae, Chionodoxa Boiss. has been subsumed into Scilla L. (e.g. David, 2018), removing the need for the use of ×Chionoscilla J.Allen ex G.Nicholson. Likewise, in ferns, the inclusion of species of Ceterach Willd. and Phyllitis Hill in Asplenium L. (e.g. Schneider et al., 2004; Smith et al., 2006) has removed the need for the use of the nothogenera ×Asplenoceterach D.E.Mey. and ×Asplenophyllitis Alston. Other points in favour of a broad Pyrus were that many species of Sorbus sensu lato already had available names in Pyrus and that Pyrus sensu lato overcame the problem of the polyphyly of Sorbus. However, some members of the community regarded the recognition of such a broad concept of Pyrus (despite the long history) to be a step too far, with some of the groups included in this concept of Pyrus being separately recognisable (apples, pears, cotoneaster etc); as such, some botanists saw this as being a case of solving one problem but creating another. Yet another treatment of Sorbus sensu lato, suggesting the treatment of the taxa as two genera, was put forward on the basis of a combined analysis of plastid DNA by Sun et al. (2018), one composed of Sorbus subgenera Aria, Chamaemespilus and Torminalis, the other of subgenera Cormus and Sorbus and Micromeles folgneri C.K.Schneid. [Micromeles is a genus with a particularly problematic taxonomic history (see, e.g., Rushforth, 2016), and M. folgneri has synonyms in Sorbus, Aria and Alniaria Rushforth]. However, even in their analyses, Sun et al. (2018) were unable to exclude the possibility that Sorbus sensu stricto + M. folgneri are more closely related to Pyrus sensu stricto than they are to Sorbus subgenus Cormus, and the topologies for relationships in Malinae recovered in their analyses do not agree fully with those recovered in other analyses. In addition to the question of which genera to recognise, there has also been much discussion about what constitutes a species in Sorbus sensu lato. Species concepts vary widely in Sorbus, reflecting the combination of sexual reproduction, hybridisation and apomixis (see, e.g., McAllister, 2005; Rich et al., 2010) and the widespread of occurrence of allopolyploidy (see Pellicer et al., 2012, and references therein). Aldosoro et al. (2004), for example, applied a much broader species concept in their monograph of Sorbus subgenera Aria and Torminaria than, for example, Rich et al. (2010). As examples of the effect of the broader species concept, Aldosoro et al. (2004) included S. subcuneata Wilmott (SOMERSET WHITEBEAM; Fay et al., 2022b, this issue) and several other taxa in a broad S. latifolia (Lam.) Pers. (FONTAINEBLEAU WHITEBEAM), and other species were similarly sunk into a broad S. aria. This treatment has not, however, been widely accepted, and we maintain S. subcuneata and the other species subsumed into S. latifolia by Aldosoro et al. (2004) as distinct species here. For one definition of what constitutes a species in Sorbus, see Rich et al. (2010), pp. 3–6. In summary, several taxonomic frameworks are available for the rowans, whitebeams and service trees. None of these has gained universal recognition among the community of people working on these plants, and each system has its own advantages and disadvantages. Many of the problems arise from trying to fit to complex evolutionary histories into simple hierarchical frameworks. For this reason, we have decided to go with the status quo, using the botanical names used in Rich et al. (2010), the most comprehensive treatment of the British and Irish species, and in Stace (2019), the most recent flora of Britain and Ireland. The names in Sorbus (despite the polyphyly of the genus in the broad sense) are still widely used by botanists in Britain and Ireland. In time, one of the systems outlined above as alternatives may gain traction and thus wider acceptance, but this point has not yet been reached. For completeness, we provide the synonyms that apply in the various systems in the individual species accounts, but we ask the reader to enjoy these wonderful plants and not to worry too much about the taxonomic complexities. We thank the many colleagues and friends who have worked with us on these plants and who have discussed their complexities with us over many years – it's been challenging, but also great fun!
Search from 250 M+ research papersSearch from 250 M+ research papers
Open Access
Feb 29, 2024
Tom W. May 