Understanding Diversity and Systematics in Australian Fabaceae Tribe Mirbelieae

: Australia has a very diverse pea-ﬂowered legume ﬂora with 1715 native and naturalised species currently recognised. Tribe Mirbelieae s.l. includes 44% of Australia’s peas in 24 genera with 756 recognised species. However, several genera within the Pultenaea alliance in tribe Mirbelieae are considered to be non-monophyletic and two main options have been proposed: option one is to merge ca. 18 genera containing ca. 540 species (the largest genus, Pultenaea has nomenclatural priority); and option two is to re-circumscribe some genera and describe new genera as required to form monophyletic groups. At the species level, option one would require 76% of names to be changed; whereas based on available data, option two is likely to require, at most, 8.3% of names to change. Option two therefore provides the least nomenclatural disruption but cannot be implemented without a robust phylogenetic framework to deﬁne new generic limits. Here we present novel analyses of available plastid DNA data ( trn L-F) which suggest that option two would be feasible once sufﬁcient data are generated to resolve relationships. However, the reticulate evolutionary histories or past rapid speciation suggested for this group may prevent the resolution of all nodes. We propose targeted use of Next-Generation Sequencing technology as the best way to resolve relationships between the key clades in the tribe and present a framework for such a study. An overview of current taxonomy in the tribe is presented, along with the state of taxonomic knowledge and availability of published descriptions for electronic ﬂora treatments. Several new combinations and typiﬁcations are published in an appendix.


Introduction
Legumes (Family Fabaceae/Leguminosae) are the third largest plant family behind daisies (Asteraceae) and orchids (Orchidaceae) and the family contains an incredible range of morphological and ecological diversity [1]. Legumes date back to the Maastrichtian or early Paleocene, diversifying after the Cretaceous-Paleogene boundary mass extinction event [2]. Estimating the phylogeny and establishing a useful classification system is naturally a challenging task in such a diverse lineage, however significant advances have been achieved in recent decades, particularly through contributions to the series Advances in Legume Systematics [3][4][5][6][7][8][9][10][11][12][13][14]. Hundreds of publications have addressed the relationships between particular clades of Fabaceae using phylogenetic data, and a number of important

Tribal Limits
Tribe Mirbelieae has been recognised as a natural group for four decades [37], sharing with the closely related tribe Bossiaeeae an endemic distribution in Australia and distinctive yellow and red corolla markings, hence the common name "egg-and-bacon" peas. Earlier classifications relied on morphology-based analyses to assess the relationships of genera in the Mirbelieae (Table 1; [37][38][39][40][41]). Subsequent studies used DNA sequences to estimate the phylogeny and interpreted the morphology from its fit to the molecular trees. A molecular phylogenetic framework for understanding taxonomic limits of the tribes has only partially been developed [34,36,[42][43][44][45][46][47][48]. Both chloroplast DNA sequences (especially trnL-F) and nuclear ribosomal DNA (mainly ITS) have been used and are often combined for analysisexcept when conflicting (e.g., [34,48]). Molecular studies have consistently found both Bossiaeeae and core Mirbelieae to be monophyletic. These groups differ in embryology [49]. A clade comprising core Mirbelieae plus Isotropis Benth. shares reduced, 5-nucleate female gametophytes as a synapomorphy, while Bossiaeeae share normal 8-nucleate megametophytes but with giant antipodals. The other genera of Mirbelieae s.l., including Daviesia Sm., Gompholobium Sm., Sphaerolobium Sm., Erichsenia Hemsl. and Viminaria Sm., have giant antipodals. Relationships among the genera of the giant antipodals group (including Bossiaeeae) have varied among analyses, and also between the genomes [46,48]. This group has usually been found to be non-monophyletic with core Mirbelieae nested inside. However, in the ITS partition, a base-composition bias (non-stationarity) between groups with the two embryological traits may explain the non-monophyly [46]. When stationarity was partially corrected by using a Logdet model in a neighbour-joining analysis, the giant antipodals group was found to be monophyletic, though without branch support [46]. That is, the two embryological groups were reciprocally monophyletic. As monophyly of core Table 1. Genera of Mirbelieae s.l. (column 1); previously published suprageneric classifications (columns 2 and 3); embryosac groups (column 4); usage in the present review; columns 5 and 6 indicate whether genera are monophyletic (yes/no/-) in published phylogenies with separate cpDNA and ITS trees [36,[46][47][48].

Genus-Level Questions
Most recognised genera in Mirbelieae were established in the first half of the 19th Century [50][51][52][53][54][55][56][57][58][59][60][61][62][63][64][65][66][67]. Despite most genera being established for so long, generic boundaries within core Mirbelieae remain contentious because published phylogenies do not resolve a number of these genera as monophyletic [34,36,68] (Table 1). Orthia et al. [34,36] suggested that the evolutionary history of the core Mirbelieae may be complex, with potential for ancestral polymorphism, hybridisation, and incomplete speciation/incomplete lineage sorting. Data also suggest a recent radiation of extant taxa and all of these factors are potentially resulting in a lack of resolution and statistical support for the major clade relationships in the tribe. A rapid radiation during the tertiary has been suggested for Diversity 2021, 13, 391 4 of 38 many legume groups [22]. Interpretation of phylogenetic histories should take these factors into account. However, little is known of the presence or frequency of polyploidization or apomixis in the tribe. Likewise, there is insufficient data to infer any consistency of maternal inheritance of the chloroplast genome among these genera.
Phylogenetic analyses from the Cook and Crisp labs [69][70][71] used both nuclear ITS data and chloroplast data, including but not necessarily restricted to trnL-F. Trees estimated from the separate genomes were generally congruent and therefore results from analyses of combined data were used. However, in an analysis of Pultenaea s.l. [34], the chloroplast and nuclear trees were in conflict and the results were presented separately. Species-level sampling is still insufficient to resolve a number of generic boundaries, and several key nodes in the backbone of the trnL-F phylogeny remain unsupported. Some additional data are available for other markers, including ITS, ETS, trnK, matK, psbA-trnH and trnL, but sampling density is relatively low. However, these results do provide support for many clades, including Gastrolobium R.Br., Jacksonia R.Br. ex Sm., Mirbelia Sm., core Oxylobium Andrews, and core Pultenaea, which contain most species in the core tribe. The resolution of these key clades as monophyletic provides a high degree of confidence that alternate classifications can be devised to minimise nomenclatural change once data are generated that can provide sufficient phylogenetic resolution in order to flesh out the backbone of the phylogeny. The current lack of resolution means that the best outcome for taxonomic stability (lumping or dividing existing genera) is uncertain.
One proposed solution is to include all genera in the Pultenaea alliance in a greatly enlarged mega-genus Pultenaea [34], but this has been met with little enthusiasm from the botanical community and general public. However, an alternative solution is not available currently, and cannot be obtained without significant new molecular data. If additional data are generated that resolve relationships, then it is likely that Aotus Sm., Eutaxia R.Br., Podolobium R.Br. and Pultenaea will require significant re-circumscription. Minor adjustments are likely to be required in Callistachys Vent., Chorizema Labill. (potentially to include Podolobium s.s.), Oxylobium and Phyllota (DC.) DC. ex Benth. Several new genera are likely to be required in order to define monophyletic genera within the tribe, however this option may require less nomenclatural change than the creation of a giant Pultenaea including ca. 540 species.
Despite relatively recent revisions [93][94][95][96][97], morphological assessments of Pultenaea s.s. have identified many species circumscriptions that remain uncertain (R.L. Barrett, J.A.R. Clugston & M.A.M. Renner, unpubl. data). There are also unresolved issues of typification, as not all relevant material was located during these revisions (R.L. Barrett, unpubl. data) We estimate that around a third of the 148 species currently included in Pultenaea should have their taxonomic circumscriptions revised. Many changes differing from these treatments have already been adopted by the Flora of South Australia [124] and VicFlora [139]. Detailed assessment of the Pultenaea glabra Benth. complex by M.A.M. Renner et al. (in prep.), has identified six new species. As Pultenaea glabra is already listed as rare in its broadest circumscription, all of the segregate species are likely to be listed as threatened taxa within New South Wales. It is likely that many additional taxa that are of conservation priority should be recognised in Pultenaea.

Biogeographic Patterns
Tribe Mirbelieae is of significant interest for understanding the evolution of the Australian flora due to its distribution across Australia, high species diversity [68,140], high levels of local endemism [71,[141][142][143], morphological diversity [46], breeding systems [144], and specialised pollination syndromes [48,145]. There are contrasting centres of endemism in both southwestern (particularly Daviesia and Gastrolobium) and southeastern Australia (particularly Bossiaea, Dillwynia and Pultenaea), a pattern not seen in many species-rich groups in Australia [40,146,147]. These parallel patterns offer unique opportunities for understanding broader patterns and processes of speciation in the two regions. These strong biogeographic trends are reflected in phylogenetic reconstructions of the tribe (Figure 1). A comparison of south-west and south-east diversity patterns in Daviesia and Bossiaea [70] revealed that the geographic overlap of clades was significantly greater for Daviesia in the south-west than in the south-east but that this was reversed for Bossiaea. Despite this, diversification rates did not differ between the regions in either genus over the last 10 Myr [70]. Rather, the interaction of multiple factors likely explains the diversity differences between the two regions. The smaller south-western geographic ranges of species in both genera are probably explained by the steeper climatic gradients in that region. Daviesia is far Diversity 2021, 13, x FOR PEER REVIEW 9 of 39 (D)

Phylogenetic Analyses
A phylogeny of core Mirbelieae has been inferred here using the trnL intron and trnL-F spacer of cpDNA. Previous studies have only included selected data, so a novel analysis is justified in order to assess all of the available data for this marker. All available sequences for tribe Mirbelieae were downloaded from NCBI Genbank ( Table 2). As genera outside core Mirbelieae are relatively distantly related [46], and the key questions we wish to address relate to core Mirbelieae, only six species outside this core were included as outgroups (Table 2). Randomly selected sequences were checked by using BLAST (http: //blast.ncbi.nlm.nih.gov/Blast.cgi (accessed on 25 April 2021)) to ensure that they were not from a contaminated source. A preliminary alignment was performed using MAFFT ver. 7.450 [148,149] and minor corrections were made manually with Geneious Prime (ver. 2021.1.1, see https://www.geneious.com, accessed on 25 April 2021). Phylogenetic analyses were performed using Bayesian inference implemented in MrBayes (ver. 3.2.6, see https://github.com/NBISweden/MrBayes/releases/tag/v3.2.6 (accessed on 25 April 2021)) [147,148]. Daviesia longifolia Benth. was designated as the outgroup for analyses. Four Markov-Chain Monte Carlo (MCMC) chains were run for 20 million generations, with one tree sampled every 5000 generations at a temperature of 0.2 with default priors (gamma), and GTR substitution model, GTR+I+G (identified by the Akaike Information Criterion corrected for small sample sizes in jModelTest2; https://github.com/ddarriba/jmodeltest2 (accessed on 30 June 2021)). The first 2,000,000 trees recovered were discarded as burn-in (trees produced before convergence). Stationarity and mixing were assessed using Tracer (ver. 1.7.1) [150].

Results
The sequence data comprised 200 sequences for trnL-F including the outgroup samples ( Table 2). Alignment and character statistics are shown in Table 3. Stationarity and mixing were confirmed with an effective sample size of 79,932; a mean standard deviation of the split allele frequencies of 4.931 × 10 −3 ; and consistency in the tracer run. Sphaerolobium is strongly supported as a monophyletic clade sister to core Mirbelieae. The posterior tree is dominated by a hard polytomy from which all major clades of core Mirbelieae are derived ( Figure 1). This limits the degree to which relationships between clades can be discussed, so instead, we will focus on the major clades that are strongly supported within polytomy and their correlation with currently recognised genera. Significantly, we recover a moderately supported clade that contains most Pultenaea species, with four strongly supported subclades within this clade. Support for this clade, probably the first to diverge in core Mirbelieae, appears to be influenced by outgroup selection, as analyses including more outgroup taxa recover the same clade with strong support (>0.95) (not shown). Three of these subclades are endemic to southeastern Australia, while one also has numerous species in southwestern Australia. Core Oxylobium is strongly supported as monophyletic, with the inclusion of Mirbelia oxylobiodes F.Muell., which is formally transferred to Oxylobium below. Dillwynia, Eutaxia, Jacksonia, Leptosema and Phyllota are all strongly supported as monophyletic based on only limited sampling. Aotus is separated into two lineages based on just four included species, reflecting its diverse morphology. Euchilopsis and Urodon are together weakly supported as sister to Phyllota. A clade of Western Australian species currently included in Pultenaea s.l. is strongly supported as sister to Latrobea, which is strongly supported as monophyletic based on two included species. The position of 'Otion microphyllum' is not supported but it has morphological affinities to Aotus.
Mirbelia is strongly supported as monophyletic, with the exception of M. oxylobioides F.Muell. which clearly belongs in Oxylobium. The position of Pultenaea brachytropis Benth. is not supported. It appears, on morphology, to be allied to P. craigiana C.F.Wilkins, Orthia and Crisp (not sampled), and the two species may form an independent lineage. A second clade of Western Australian species currently included in Pultenaea s.l. is strongly supported as sister to a clade that includes Stonesiella, Almaleea and Pultenaea adunca Turcz., making this a morphologically heterogenous clade. The relationship between the Chorizema species and the type clade of Podolobium remains unclear, as these nodes are not supported in our analyses, but it is possible that these two genera should be united. Three species of Podolobium are strongly supported as sister to Callistachys lanceolata, and they are formally recognised as Callistachys species below. Finally, there is strong support for the broad concept of Gastrolobium adopted by Chandler et al. [50]. This definition of Gastrolobium includes significant morphological diversity, reflecting the fact that it now includes the former genera Brachysema R.Br., Jansonia Kippist and Nemcia Domin and Oxylobium lineare (Benth.) Benth. (= Gastrolobium ebracteolatum G.Chandler and Crisp).

Discussion
As highlighted by Orthia et al. [34,36], the resolution of monophyletic genera is impossible from these data unless the entirety of core Mirbelieae is incorporated in a single mega-genus, to which the name Pultenaea would apply. Such a proposal has met with significant resistance from both the general public and botanical community, so it is our hope that recent advances in sequencing technology can generate novel data to resolve relationships among the majority or all of the lineages that currently form the hard polytomy recovered in our analyses. If these relationships can be resolved, then a classification can be proposed that minimises taxonomic disruption and maintains as many traditionally recognised genera as possible, though significant re-circumscription of some genera appears inevitable [34,36].
Careful assessment and reanalysis of molecular data (from the trnL-F marker) available on GenBank, including a larger number of taxa, does recover a higher proportion of supported monophyletic clades than found by Orthia et al. [34,36], including a clade containing most Pultenaea species (130 of 148). Pultenaea is the largest genus in the tribe, and it contains a high degree of morphological and phenotypic variation. The Pultenaea species excluded from the core clade are all endemic to south-western Australia, and their inclusion in Pultenaea has been questioned previously [96].

How to Resolve Generic Relationships within Mirbelieae?
The close evolutionary history of the Mirbelieae, which is potentially due to hybridisation, incomplete lineage sorting and recent radiation [34,36], can make genetic identification of closely related species and individuals very difficult. This is often due to little genetic sequence divergence, which is especially problematic when using traditional approaches such as DNA barcoding [151]. However, for groups that may have recently diversified, population genetic markers, such as microsatellites, are an excellent way to determine genetic differences between closely related species [152][153][154]. In many cases, these closely related species share a close geographic range, which can lead to interspecific hybridisation and outbreeding depression, which can affect species delimitation [155].
Next-Generation Sequencing has revolutionised the way that we handle and think about molecular data, due to its ability to sequence whole genomes. However, the costs and bioinformatics processing associated with full-genome studies are often intractable under most research budgets, and for answering most evolutionary and phylogenetic questions, this quantity of data is not required [156]. Techniques, such as restriction associated DNA sequencing (RADseq), offer an excellent and cost-effective approach that can be used to find informative genome-wide markers and provide unprecedented phylogenetic resolution [157]. However, markers developed by using RADseq are taxon-specific and not cross-applicable between genera or families [158]. In recent years, huge advances have been made in targeted sequence capture that allow for the enrichment of hundreds of informative markers throughout the genome that are cross-applicable throughout angiosperms and gymnosperms [159][160][161][162][163].
Johnson et al. [164] have recently developed a commercially available targeted baits probe set (myBaits Angiosperms 353) that can generate a range of informative low-copy exons from the nuclear genome, which are cross-applicable across angiosperms. This baits set has particular utility for herbarium specimens that are up to 100 years old or sometimes older [162]. To date, the Angiosperms 353 probe set has provided impressive phylogenetic resolution in Schefflera J.R. Forst. and G. Forst. [165], Cyperaceae [160,161,166] and Gesneriaceae [167], providing unprecedented support at both the backbone and species levels. Additionally, the myBaits Angiosperms 353 kit, currently being used as part of the Plant and Fungal Trees of Life (PAFTOL) [163,164,168], is a method that will form part of the Phylogenomics Working Group (PWG) of the Genomics for Australian Plants Framework Data Initiative (GAP; https://www.genomicsforaustralianplants.com (accessed on 25 April 2021)), setting a new standard for angiosperm genomics.
The resolution of tribal, generic and species relationships within Fabaceae tribe Mirbelieae can realistically only be obtained by the targeted sampling of key species for the generation of significant new molecular data [34,36]. The aim of such sampling is to develop a new classification that minimises nomenclatural changes. Considering the complexity and possible recent radiation of Mirbelieae, construction of a well-supported phylogeny will require the use of rich and phylogenetically informative low-copy markers to allow for competing phylogenetic signals in the data to be identified and resolved [159]. Using targeted capture for Mirbelieae should not only provide good phylogenetic resolution of genera but also at the species level [169]. Therefore, the use of target capture sequencing should directly aid in species circumscription. A significant number of species complexes still require resolution (at least 28 in Pultenaea s. s.; unpubl. data by the authors), so this is an important consideration when selecting a technique.
To accomplish this, we will utilise Next-Generation Sequencing in the form of targeted baits capture using the myBaits Angiosperms 353 universal probe set in order to obtain informative markers from the nuclear genome [163,164]. Enabling the sequencing of a large number of informative low-copy markers throughout the nuclear genome at an affordable cost should allow us to resolve the relationships between major clades and therefore redefine genera. We will sample multiple individuals within the example species complexes in order to test the utility of the PAFTOL markers [163] for resolution of such taxonomic problems, guiding future studies. Where we are able to resolve species limits within this project, we will revise or create new descriptions for the Australian eFlora and publish new species as required.
A wide range of DNA samples have already been collected by the authors of this paper and continued targeted fieldwork will ensure that the most important taxa and populations can be included in order to address our key questions. Currently, silica-dried DNA samples are held for~95% of key taxa required for molecular work (subset from >1300 samples of Faboideae held by project collaborators, with additional samples in the NSW DNA bank). Further samples will be obtained through fieldwork for the remaining species to represent all the key clades of Mirbelieae in order to provide the maximum phylogenetic resolution and support from the Angiosperms 353 bait set [162,163]. Sampling from herbarium specimens will also be undertaken where no field-collected material is available. It is critical that all key taxa (including all type species) are included in a phylogeny to help fully resolve generic level phylogenetic relationships in Mirbelieae.  .). A few anomalous species may be placed in the new genera, or in expanded definitions of existing genera. Two lineages currently included in Pultenaea s.l., both endemic to Western Australia, appear to be well-supported clades sister to other Western Australian genera, and new generic names may be required for these clades if they continue to be supported by additional data and sampling. 'Otion' remains an informal name that also requires formal circumscription and validation. In summary, current taxonomic descriptions are available in the published literature for 595 species, while 160 species need revised or new descriptions (at least 32 of these species are unnamed).

Current Availability of Descriptive Information and Interactive Identification Tools
The release of profiles on the Australian eFlora under current generic circumscriptions is not seen to be a viable option, as many of the generic and species boundaries are known to be problematic. An example is the generic name 'Otion', first proposed in 1982, but never formalised, leaving its constituent species orphaned. Combining these into Pultenaea s.l. until generic boundaries are resolved would only further distort the circumscription of that genus and be problematic. The definition of Pultenaea s.s. is also especially challenging at present, but the transfer of just 18 species from Pultenaea will likely resolve this as the remainder of Pultenaea species are resolved as a monophyletic clade.
Published revisions are available for 595 species within Mirbelieae, leaving 160 species that require revision, or new descriptions (Table 4). We aim for the completion of eFlora treatments for all taxa in tribe Mirbelieae (24-33 genera and ca. 755 species). Treatments have already been provided to ABRS for an additional 57 pea genera (including 483 species) from other tribes. An ongoing project by R. Butcher, I.D. Cowie & T.D. Macfarlane et al.
( [170][171][172][173][174][175][176]; Butcher, unpublished data) to revise Australian Tephrosia Pers. will add ca. 111 species. At the completion of these projects, ca. 1370 species will be available for eFlora, leaving a gap of ca. 393 species for completion of Fabaceae for the Flora of Australia. Table 4. Summary of genera in tribe Mirbelieae reflecting putative phylogenetic relationships, number with molecular data, and progress towards eFlora treatments. Genera or clades marked with an * are yet to be adequately circumscribed and require additional genetic data.

Interactive Keys
The Pea Key provides a user-friendly resource for the identification of Australian Faboideae [138]. However, technological changes since its first edition, released in 2007, require the key to be updated with improved accessibility. Less than 50 species of Mirbelieae need to be added and the functionality of the key must be improved in order to make it more attractive and to increase its use by a wider audience. An expansion of the character sets for Mirbelieae species would enable a more reliable tool in the identification of taxa to species-level.
The Pea Key currently includes ca. 1500 species among 136 genera, based on 67 descriptive characters (hosted by ANBG, last updated 2007). A new interactive key has recently been published on FloraBase for Western Australian peas [177]. This key is based on 70 characters, is well-illustrated, and is user-friendly. The WA key provides an excellent tool to aid in the identification of 507 Mirbelieae species, and we aim to update existing data, images and illustrations in The Pea Key for the 247 Mirbelieae species that occur outside of Western Australia in order to align with, and complement, the Western Australian pea key. A new interactive key has also been completed for Victoria, covering 309 species based on 50 characters [139].
Images and descriptive data were mostly absent in the first edition of The Pea Key. Rather than creating a stand-alone dataset of descriptive information and images, it is proposed that the second edition links to the newly created eFlora platform through unique identifiers provided by the Atlas of Living Australia (ALA). Linking to the eFlora will automatically utilise images in the Australian Plant Image Index (APII). Images of key morphological features provide powerful identification aids and representative images are provided to demonstrate the large range of morphological variation found in the tribe (Figures 2-6).

Conclusions for Future Taxonomic Classification
Core Tribe Mirbelieae, as defined here, currently includes 18 genera, but our findings show that at least some of these genera cannot be maintained in their current circumscriptions. Options that create monophyletic genera include the recognition of a single large genus, or up to 27 genera (in which case up to six genera would be new). We propose to use targeted sampling of all recognised and possible novel genera, representing known diversity in the group, in order to determine a novel generic level classification that minimises taxonomic changes. We anticipate that with new data to confirm relationships between the major lineages identified in our study, the majority of Pultenaea species can be retained in that genus. The large genera Gastrolobium, Jacksonia and Mirbelia are likely to be maintained in the current circumscriptions. Most of the other genera in core Mirbelieae require further sampling to determine revised monophyletic units that have utility as genera. This resolution is critical to the appropriate placement of species in the Australian eFlora.
require further sampling to determine revised monophyletic units that have utility as genera. This resolution is critical to the appropriate placement of species in the Australian eFlora.

Conflicts of Interest:
The authors declare no conflict of interest.
While many changes in generic circumscription are expected, most require more data and better sampling before the best nomenclatural solutions can be determined [34,36]. However, there is sufficient support to justify a few taxonomic changes and these are validated below. The opportunity is taken to select types for genera where type species have never been designated, as far as we can determine. Specific lectotypes for species and infraspecific taxa are also designated in order to place names correctly in advance of a new Notes: Callistachys has been rejected against the conserved name Oxylobium Andrews (1807) [178] if the two genera are united, but this has no effect when they are considered distinct genera. We anticipate that the two genera will both remain accepted. Three eastern Australian species previously included in Podolobium are clearly closely related to the Western Australian species Callistachys lanceolata ( Figure 1C; [179]), and they are here included in Callistachys where names are already available [40]. Oxylobium alpestre F. Muell., Victoria-Parliamentary Papers-Votes and Proceedings of the Legislative Assembly 3, 12 (1853), nom. inval., nom. nud. Notes: Mueller [180] described the fruit of this species, with no details of the flowers. Numerous syntypes have been located, many without collection dates. It is unknown whether he did have flowering material, so both fruiting and flowering sheets are considered syntypes. We select a fruiting sheet at MEL as the lectotype as this is one of the the largest specimens at MEL, it has a collection date firmly establishing that it was collected before the species was named, and it is also a good match for the protologue. There is also a duplicate at K.   [57]. All four species were recognised by Bentham [59] when he raised the section to genus rank. However, Bentham later reduced three of these to synonymy under Phyllota phylicoides (Sieber ex DC.) Benth. [62] Phyllota squarrosa (Sieber ex DC.) Benth. has since been reinstated. We note that Bentham [59] specifically attributed the genus name to de Candolle (probably based on de Candolle ([57]: 113) stating 'An genus proprium?' [59] We therefore accept the authorship of the genus name as '(DC.) DC. ex Benth.' Jancey [92] undertook a detailed study of Phyllota in New South Wales, providing detailed notes on the type species for the species, but did not select a type for the genus. We here select Pultenaea phylicoides as the type of Pultenaea sect. Phyllota (and therefore of Phyllota), as it is only one of the four original species later recognised by Bentham [62]. Walpersia burtonioides was described as a new genus and species from South Africa based on erroneous label data and it is actually an Australian taxon [183]. Typification: There are three sheets of Phyllota barbata at W collected by von Hügel, each with different collection numbers, so they are here regarded as syntypes. W 0046869 is designated as the lectotype as it is the only sheet that bears the name 'Phyllota barbata' on the original labels.

Phyllota barbata
Two sheets of the type collection of Phyllota villosa have been located. We here designate the sheet at KW as lectotype as this is the sheet used by Turczaninow when describing the species. Notes: Orthia et al. [96] noted the morphological similarity of Phyllota barbata and Pultenaea barbata and even suggested they were possibly conspecific. We conclude that the two names might best be treated as a single species based on study of the type specimens, however the names are currently both in use for two quite distinct taxa in Western Australia, so further studies are advised on the application of both names. The two taxa were named in separate genera, but independently given the same species epithet. If the two taxa are united, the younger name has priority if the taxon is included in Pultenaea (as Orthia et al. [96]  Benth., as this is the best described of the two species included by Bentham [60], and the only name that is legitimate. Pultenaea barbata C.R.P. Andrews, Journal of the West Australian Natural History Society 2(1): 38-39 (1904 [93] attempted to raise Pultenaea parrisiae subsp. elusa to the species rank, however they unfortunately cited the entire page range of the publication, so the combination was not validly published (ICN Article 41.5 [181] [62], with two sheets now located at K and two other sheets at BM and MEL. Only the sheet at MEL bears the name 'var. recurvifolia', but this does not appear to be in Bentham's script, though, unfortunately, the bottom of the label has been cut off at some point. The sheets at BM and K bear the name P. recurvifolia, indicating that they were distributed much later. For this reason, we choose the MEL sheet as lectotype. A second collection (Mouth of the Glenelg [River], W. Allitt s.n. (K 000118045)) was annotated as the 'holotype' by M.D. Crisp in 1982, but it was not cited by Bentham [62], so it has no type status. Notes: de Kok & West [94] erroneously treated this taxon as a synonym of P. daltonii, a later name. Further investigation has shown that P. recurvifolia is more closely allied to P. hispidula R.Br. ex Benth., as noted in VicFlora ( [133], updated by Stajsic 2019). Its taxonomic status requires further investigation, and it is uncertain whether the taxon should be reinstated or included as a synonym of P. hispidula. Notes: Bentham [59] described two species in his new genus Isotropis, I. biloba Benth. and I. striata Benth. but did not designate either as the type as this was not a practice at the time.
Bentham [62] recognised I. striata as a good species, but included I. biloba as a synonym, having examined a broader range of specimens since his original treatment. On that basis, we here designate I. striata as the type of the genus. The application of both names has been considered uncertain in Australian literature, but one of us (MDC) examined the type material of each taxon at W in 1982, and these specimens have recently become available online. They are both confirmed as synonyms of I. cuneifolia, as indicated by Bentham ([62]; as Callistachys cuneifolia Sm.). Both names are applicable to I. cuneata subsp. cuneata based on the characters defined by Keighery [113], though this taxon is variable and additional taxa are likely to be recognised.