Proposal to recognise the tribes Adinobotryeae and Glycyrrhizeae (Leguminosae subfamily Papilionoideae) based on chloroplast phylogenomic evidence

Abstract Within the legume family, the taxonomic status of subtribe Glycyrrhizinae of tribe Galegeae and of the genus Adinobotrys has been re-assessed. Based on genome skimming data, we conducted phylogenomic analyses of the inverted repeat-lacking clade within subfamily Papilionoideae. The results support the sister relationship between Glycyrrhizeae and Adinobotrys. Glycyrrhizeae is resurrected based on Glycyrrhiza and Glycyrrhizopsis, and a new tribe, Adinobotryeae, is proposed to accommodate Adinobotrys.

Within the GAW clade, the temperate herbaceous genera Glycyrrhiza L., with Meristotropis Fisch. & C.A.Mey. being treated within Glycyrrhiza, and Glycyrrhizopsis Boiss. are sister taxa (Duan et al. 2020(Duan et al. , 2021; also see Fig. 1). Tribe Glycyrrhizeae was proposed by Rydberg (1917) to accommodate Glycyrrhiza, which was degraded as subtribe Glycyrrhizinae of tribe Galegeae by Rydberg (1923) and was then slightly emended by Polhill (1981) with Glycyrrhizopsis species treated as a synonym of Glycyrrhiza. Although Glycyrhizeae was not widely accepted by taxonomists, the taxonomic position of Glycyrrhizinae is also questionable. Galegeae is a polyphyletic and morphologically diverse tribe (Sanderson and Liston 1995;Lock and Schrire 2005;Duan et al. 2015), and our prior phylogenetic studies have shown that Glycyrrhizinae is only distantly related to core Galegeae (Duan et al. 2015(Duan et al. , 2021. In consequence, the taxonomic rank of Glycyrrhizeae/Glycyrrhizinae needs to be reconsidered. Apart from Glycyrrhizeae, the remaining taxa of the GAW clade included some woody genera, formerly placed in tribe Millettieae s.l., which were subsequently assigned into two clades: the liana (rarely climbing shrubs) tribe Wisterieae and the tree genus Adinobotrys Dunn (Compton et al. 2019;Duan et al. 2021; also see Fig. 1). Previous phylogenetic results excluded Adinobotrys from Wisterieae or Millettieae (Hu et al. 2000;Duan et al. 2021), but further taxonomic study is required to re-assess the placement of this genus.
In the present study, we used the genome skimming method (Straub et al. 2012;Zhang et al. 2015) to obtain 75 chloroplast (cp) coding sequences (CDSs), and constructed a phylogenomic framework of the GAW clade to test the taxonomic position and status of genus Adinobotrys and tribe Glycyrrhizeae. Our study provides a taxonomic foundation for future evolutionary, systematic and biogeographical studies of the IRLC legumes.

Taxon sampling
Sampling for molecular phylogenetic analyses included the genus Adinobotrys (one species sampled), both of the genera of tribe Glycyrrhizeae (eight species sampled) and 14 of the 15 genera (Serawaia J. Compton   2020, 2021). Following Wojciechowski et al. (2004) and Duan et al. (2021), ten other genera of the IRLC (10 spp.) and three genera of Robinioids were selected as outgroups. Most samples for the study were obtained from fresh field collections or from preserved herbarium specimens (see Table S1 in the supplementary file for details), except that DNA samples of Austrocallerya megasperma (F.Muell.) J.Compton & Schrire, Whitfordiodendron nieuwenhuisii (J.J.Sm.) Dunn, Wisteria brachybotrys Siebold & Zucc. and Wisteria floribunda (Willd.) DC. were accessed from the DNA and Tissue Bank, Royal Botanic Gardens, Kew (https://dnabank.science.kew.org), and the chloroplast genome of Lotus japonicus (Regel) K.Larsen was downloaded from GenBank.

DNA Extraction, Genome Assembly, Annotation and Alignment
We extracted the total genomic DNA following a modified CTAB protocol (Doyle and Doyle 1987). Yield and integrity (size distribution) of genomic DNA extracts were quantified by fluorometric quantification on a Qubit (Invitrogen, Carlsbad, California, USA) using a dsDNA HS kit, as well as by visual assessment on 1% agarose gels. Subsequently, we used all samples to build blunt-end DNA libraries using the NEBNext Ultra II DNA library Prep kit for Illumina (New England Bio-labs) following the protocol of the manufacturer. We pooled the final indexed libraries in equimolar ratios and sequenced them in a single lane of an Illumina XTen sequencing system (Illumina Inc.).
From the raw reads, we filtered out adaptors and low-quality reads in Trimmomatic v.0.33 (Bolger et al. 2014). We checked the quality of the remaining reads using FastQC (www.bioinformatics.babraham.ac.uk/projects/fastqc/) and performed de novo assembly in SPAdes 3.11 (Bankevich et al. 2012) with the k-mer of 75, 85, 95 and 105. A customised python script was employed (Jin et al. 2018) with its default parameters to apply BLAST and a built-in library to connect verified contigs into plastomes in SPAdes. We annotated the assembly of the resulting complete cp genomes using the Dual Organellar GenoMe Annotator (DOGMA) (Wyman et al. 2004) with Glycyrrhiza glabra L. [GenBank Accession #: NC_024038; Sabir et al. (2014)] as a reference (see Suppl. material 1 for details of the annotated cp genomes). Start and stop codons and intron/exon boundaries for protein-coding genes were checked manually. As our samples covered a vast phylogenetic range of clades within the IRLC legumes, the character of the cp genomes, such as genomic structure and gene order, varied dramatically (Sabir et al. 2014;Sveinsson and Cronk 2016). To avoid introducing potential error from the various genomic characters, we extracted 75 protein coding sequences (CDSs; as in Kang et al. 2018, except for rps12) from the annotated genomes and concatenated them with Geneious (Kearse et al. 2012) (see Table S2 in the supplementary file for details of the CDSs).

Phylogenetic Analyses
We aligned the cp CDS sequences independently with MAFFT v.7 (Katoh and Standley 2013). The alignment was partitioned (Thode et al. 2020) and the best nucleotide substitution models for each CDS were inferred using PartitionFinder 2 (Lanfear et al. 2016) under the default settings (for the best models see Suppl. material 2). Accordingly, we carried out separate phylogenetic analyses using Bayesian Inference (BI; Rannala and Yang 1996; Mau et al. 1999) implemented in the programme MrBayes 3.2.5 (Ronquist and Huelsenbeck 2003;Ronquist et al. 2012) by applying default prior settings. Each BI was performed by applying two independent runs of the Markov Chain Monte Carlo (MCMC) for 10 million generations with sampling every 1,000 generations. We discarded the first 2,500 trees as burn-in and summarised the remaining posterior topologies as a maximum clade credibility (MCC) tree. The analysis stationarities were verified with Tracer v.1.6 (http://tree.bio.ed.ac.uk/software/tracer) by ensuring that all ESS values exceeded 200 and the convergence was confirmed between independent runs. In addition to BI, we also performed Maximum Likelihood (ML) analyses for cp CDSs, respectively, using IQ-TREE v.1.6 (Nguyen et al. 2015) with the following settings: rapid bootstrap analysis with 1,000 replicates, followed by a search for best-scoring ML tree starting with a random seed.

Results
All the sequenced plastomes were successfully assembled into complete circular configurations. The sizes of the cp genomes ranged from 122,310 to 156,702 bp and the GC contents were between 33.7% and 35.9% (see Suppl. material 1 for details). The alignment length of concatenated cp CDSs for the analyses in MrBayes and IQ-TREE was 72,106 bp. Our ML results were congruent in topology with the corresponding BI trees, and the support values of the former (as "LBS" hereafter) were thus labelled on the corresponding branches of the latter (as "PP" hereafter). Phylogenetic reconstructions ( Fig. 2) supported the monophyly of the IRLC (PP = 1, LBS = 100%), and the GAW clade (PP = 1, LBS = 71%) was sister to the rest of the IRLC taxa (PP = 1, LBS = 100%). Within the GAW clade, Adinobotrys and the well-supported Glycyrrhizeae (PP = 1, LBS = 100%) formed a clade (PP = 1, LBS = 80%), which was sister to tribe Wisterieae (PP = 1, LBS = 100%).

Discussion
Within the GAW clade, some tropical/subtropical (rarely temperate) woody liana/tree species, formerly belonging to Millettieae s.l., clustered with the temperate herbaceous Glycyrrhiza-Glycyrrhizopsis clade (Compton et al. 2019;Duan et al. 2020Duan et al. , 2021; also see Fig. 2). Recently, Compton et al. (2019) and Duan et al. (2021) assigned the above-mentioned liana/tree group into two non-sister clades: genus Adinobotrys and tribe Wisterieae (14 genera), corroborated by our cp CDSs trees (Fig. 2). Herein, we propose a monogeneric tribe Adinobotryeae based on Adinobotrys, and resurrect the tribe Glycyrrhizeae corresponding to the Glycyrrhiza-Glycyrrhizopsis clade (see Taxonomic Treatment) for the following reasons: A. Our phylogenomic analyses (Fig. 2) validated that Adinobotrys does not belong to Wisterieae; and the Glycyrrhiza-Glycyrrhizopsis clade is phylogenetically distant from both of the core groups of Galegeae [i.e. the Erophaca-Astragalean clade; see Lock and Schrire (2005)] and the type genus (i.e. Galega L.) (also see Wojciechowski et al. 2004;Duan et al. 2015Duan et al. , 2021Compton et al. 2019). It is thus reasonable to divide the GAW clade into three independent groups to ensure monophyly of Adinobotrys and the Glycyrrhiza-Glycyrrhizopsis clade with respect to tribe Wisterieae.
B. Our recent analyses indicated that a chloroplast capture event may have occurred in the GAW clade, and the common ancestors of the Adinobotrys and Wisterieae were hypothesised to be the putative paternal and maternal parents of Glycyrrhiza-Glycyrrhizopsis clade, respectively (Duan et al. 2021). Since Wisterieae was already known as a welldefined tribe in the GAW clade, its counterpart groups (Adinobotrys and the Glycyrrhiza-Glycyrrhizopsis clade) should also be recognised at the tribal rank. Such an evolutionary event-linked taxonomy was also noted in previous studies (e.g. Degtjareva et al. 2012).
C. As far as morphological differentiation is concerned, Glycyrrhizeae, containing the widely distributed Glycyrrhiza and the Anatolian endemic Glycyrrhizopsis Boiss. (Meng 2005;Duan et al. 2020; see Fig. 1C-I), is easily distinguished from Wisterieae and Adinobotrys by its herbaceous habit and temperate distribution. Adinobotrys, which has often been treated as part of Callerya Endl. within Wisterieae (Polhill 1981;Schot 1994;Schrire 2005), has a unique evergreen tree habit in contrast to the predominantly liana habit of Wisterieae (Fig. 1A, B; also see Compton et al. 2019;Compton and Schrire 2020).
Despite the tribal revision above, taxonomic questions still remain. Since Glycyrrhizeae, and Astragaleae sensu Duan et al. (2021), corresponding to the aforementioned Erophaca-Astragalean clade, were removed from tribe Galegeae, the remaining genus Galega was found to be nested within the Vicioid clade (Steele and Wojciechowski 2003;Wojciechowski et al. 2004;Duan et al. 2021). Therefore, tribe Galegeae needs further taxonomic consideration before either being merged into a larger tribe corresponding to the Vicioid clade, or re-instated in its earlier, narrower sense, i.e. as the monogeneric tribe circumscribed by Hutchinson (1964). In addition, the recent work of Compton and Schrire (2020) expanded the genus Adinobotrys sensu Compton et al. (2019) to contain four species, although its infra-generic relationships need to be studied in more detail.
Distribution and habitat. Evergreen forests in Brunei, Cambodia, India, Indonesia (Java, Kalimantan and Sumatra), Laos, Malaysia (Peninsula and Sabah), Myanmar, Thailand and Vietnam, below 1200 m in altitude. Note. The tribe includes two genera, Glycyrrhiza and Glycyrrhizopsis, with the latter confined in S. Anatolia. The root of Glycyrrhiza is widely used as medicine and in the food industry (see Duan et al. 2020