Boronia (Wilson, 1998)

Duretto, Marco F., Heslewood, Margaret M. & Bayly, Michael J., 2023, A molecular phylogeny of Boronia (Rutaceae): placement of enigmatic taxa and a revised infrageneric classification, Australian Systematic Botany 36 (2), pp. 81-106 : 98-99

Section Boronia

The circumscription of Boronia section Boronia (only B. citrata and B. rozefeldsii not sampled for this molecular study) is similar to that outlined by Bayly et al. (2015), apart from the removal of B. coriacea (now placed in section Inornatae ), which they had provisionally placed in section Boronia . Both unsampled species can be placed in section Boronia on the basis of morphology (see Albrecht and Walsh 1993; Duretto 2003). The section has a unique seed type: the testa is smooth and the adaxial hilum is linear and sunken in a groove that is surrounded by glossy labiose margins ( Wilson 1998; Choi et al. 2012). Within the section, there are two well-supported clades, namely, clade A containing species found in south-eastern Australia and Tasmania ( B. citrata and B. rozefeldsii not sampled) and including the type species of Boronia , B. pinnata Sm. , and clade B with all south-western Australian species. There are no obvious morphological characters, or combinations of characters, supporting either of these clades. However, most species in clade A have pedunculate, open and multi-flowered inflorescences (exceptions are the simple leaved B. deanei , B. rhomboidea , B. serrulata and some Tasmanian species), whereas those in clade B usually have flowers that are solitary or paired, although small cymes are also present in all species of clade B 2, and three-flowered cymes are also present in B. stricta and B. virgata Paul G.Wilson of clade B 3.

The internal structure within clade A is not well supported and there is significant conflict between analyses. In the combined and plastid analyses, B. filifolia (SA, W Vic.) is resolved as sister to a clade containing the remainder of the species. This species is unusual in the section in being glabrous apart from the flowers and having filiform leaves or leaflets. Very narrow leaves are also found in B. deanei (NSW) , which is also glabrous apart from a raised ring of tuberculae on the glands on the stems and leaves. Structure within the remainder of clade A is not well supported, although B. falcifolia (coastal NSW, SE Qld) is sometimes sister to a clade of the remaining species. In the combined (1.00 PP, 58 JK) and nuclear (1.00 PP) analyses, there is also a large polytomy (clade A 1) with all Tasmanian species (excepting B. rhomboidea ) as well as most species found in western Victoria and South Australia (apart from B. filifolia ).

The south-western Australian clade (clade B), in contrast to the south-eastern Australian clade, has well-supported structure and contains three strongly to robustly supported clades (clades B 1, B 2, B 3). Clade B 1 is characterised by persistent or tardily caducous petioles (falling after the lamina has fallen) and the species within it show considerable morphological variation. Clade B 2 is characterised by woolly staminal filaments (unique in section Boronia ) and simple leaves (widespread in Boronia ). The names Boronia series Variabiles Benth. (type: B. crenulata ) and series Terminales Benth. (type: B. capitata ) are both available for clade B 2 and were published in the same publication, and the former is chosen here. Clade B 3, as with clade B 1, contains significant morphological variation, and like clades A and B, lacks clearly discernible morphological apomorphies. The name Boronia series Heterandrae Benth. (type: B. megastigma ) is available for this clade.

Most species of Boronia have simple flowers without much modification, typical of most members of the family Rutaceae , that is, a fully open, pink or white corolla, erect stamens that are longer than the ovary, anthers that are approximately equal in size and fertile, and a cylindrical style topped with a minute stigma. Within both clades B 1 and B 3, there are suites of species with highly modified flowers that appear to be driven by specialised pollinator associations with day moths of the family Heliozelidae ( Milla 2019; Wild 2022; L. Milla et al., in prep.). Clades B 1 and B 3 resolve together in the nuclear-only analyses, hinting at a possible shared origin of some of these features. For example, sterile antesepalous anthers that are significantly different in size from the antepetalous anthers occur in both clades B 1 and B 3. In clade B 1, B. purdieana and B. tetrandra are sister species that have minute sterile antisepalous anthers. By contrast, in clade B 3, B. megastigma (sister species of B. clavata ), and the sister taxa B. heterophylla and B. molloyae , have very large, dark-coloured, sterile, antesepalous anthers. All these species, with B. clavata , have very large stigmas and, in both clades, some species have stigmas that have antesepalous lobes: in B 1, B. purdieana and B. tetrandra , and in B 3, B. megastigma (see Duretto et al. 2013, fig. 24). There are several other floral features that appear to be associated with specialised pollinator–host associations, including pendulous flowers, cup-shaped flowers, unusual petal colours (e.g. green, yellow, red, brown), contrasting petal colours ( B. megastigma where the adaxial surface is yellow and abaxial surface is brown), variously shaped stamens, lobed or hairy discs, sunken ovaries, absent styles and large stigmas (see descriptions in Duretto et al. 2013). There is a diversity of floral forms that interestingly does not lend itself to a formal classification as there are many other species with more typical flowers related to these clades. Most species in south-eastern Australia and Tasmania have simple unmodified flowers. However, two species of clade A, namely, B. serrulata and B. floribunda Sieb. ex Rchb. (both confined to the Sydney region sandstones, NSW), also have modified flowers with large spherical stigmas and filaments with a dense tuft of hairs at the apex. Curiously these species are not sister taxa and the features seem to be a case of parallel evolution. This association and apparent co-evolution of certain moths of the family Heliozelidae with their host species are part of another study ( Milla 2019; L. Milla et al., in prep.) and will not be dealt with further here.

A taxonomic challenge in Boronia section Boronia is that there are several clades that are strongly supported by both molecular and morphological data, including B 1 (persistent or tardily caducous petioles), B 2 (woolly filaments) and the species pairs in clades B 1 and B 3 discussed above. By contrast, clades A, B and B 3 have good to strong support in the analyses presented here but do not have readily identified morphological apomorphies. This is similar to the situation seen in other Australian genera in Rutaceae , for example, Asterolasia F.Muell. and Phebalium Vent. , where there is strong molecular support for clades that are confined to either eastern or south-western Australia, but no obvious morphological characters to support the clades ( Duretto et al. 2023). One of the roles of a cladistic analysis is to identify well-supported groups that then can be recognised in a preferably robust formal taxonomy, which is very useful in larger taxonomic groups such as Boronia . Robust formal classification helps with the development of identification tools, identification of conservation priorities, as well as the placement of taxa new to science, which is continuing in Boronia (e.g. Barrett et al. 2015; Duretto 2019).

There are a number of options available on how to proceed with the classification of the species in section Boronia . One would be to retain the current classification with no further grouping of the species. A second would be to formally recognise those clades with morphological synapomorphies as series (clades B 1, B 2, B. megastigma + B. clavata etc.) and leave the remainder as incertae sedis. Unfortunately, the type species, B. pinnata , is one of the many species not part of a morphologically well-defined group. A third option is to formally recognise the four major clades with strong to robust molecular support (clades A, B 1, B 2, B 3) as series, acknowledging that two of these are difficult to identify on morphological grounds but that all have support on the basis of molecular data. Option three does not create significant issues nomenclaturally: clade A retains the name Boronia as it contains the type species; clade B 1 does not have an available name but is easily defined and thus can be formally described; clade B 2 has two available names at the appropriate rank (of equal priority as they are described in the same publication and have never been considered synonymous) and is easily defined on morphological grounds; and clade B 3 has a name available at the appropriate rank. The issue here is that both clades A and B 3 do not have morphological apomorphies and are diverse morphologically. The result would be a complicated key to series, which is not an insurmountable issue, just not ideal. We are applying the third option (see Taxonomy).