Article information
SUMMER
summary
The genus Rehmannia (Ledanidae, Artemisia maprea) consists of about 70 semiparasitic species found in North, Central and South America. Previous phylogenetic studies have not included tropical species and have not taken comprehensive sampling of closely related genera, limiting our systematic understanding of these lineages. We generated newer phylogenetic hypotheses from 179 individuals from 51 species and 5 genera once thought to be the same genus as Foxglove (Candlestick, Foxglove, Quercus rehmannia, Quercus glabra, and Brushed Feather species), with a particular focus on sampling of underrepresented Central and South American taxa. Phylogenetic analyses are based on six cpDNA regions (rbcL, matK, trnT (UGU)-trnF (GAA), rps2, rpoB, and psbA-trnH) and four nuclear regions (ITS, PPR-AT1G09680, PPR-AT3G09060, and PPR-AT5G39980) and are performed using maximum likelihood and Bayesian inference methods. In addition, our data were added to a previously published Liedangidae-wide dataset to reveal the location of these lineages in the comprehensive phylogenetic context using maximum likelihood. Our results reveal a strongly supported lineage of the genus Rehmannia, which is a sister to taxa from southwestern North America and Mexico, and includes distinct Brazilian and Andean clades. The genus B. bristula is a genus of hummingbird pollination in Brazil, which is nested in the branch of the genus Rehmannia brasiliensis and has strong support. We confirm the inconsistencies between chloroplast and nuclear divisions, especially regarding the location of the early divergent lineages in the genus Rehmannia. The monotypic genus Candle Bottle, once included in the genus Rehmannia, is now classified in the family Buchnereae in the family Ledonidae, a position further confirmed by the anther morphology. Ledonaceae is made up of many species from the tropics that have not yet been included in phylogenetic studies, and our study highlights the importance of more comprehensive sampling of the placement of these lineages to clarify taxonomy, biogeography, and trait evolution. Keywords: Candle Bottle Flower, Hemiparasitic, Hummingbird Pollination, Lamiformes, Scrophulariae.
introduction
Intensively sampling, extensive phylogenetic studies provide a framework for studying macroevolutionary processes (e.g., Smith and Donoghue 2008; Edwards and Smith 2010; Tank et al., 2015; Lu et al., 2018); However, without open access sequence data, they may lack representativeness of under-studied taxa. A recent phylogenetic synthesis of the largest parasitic branch of angiosperms (Mortimer et al., 2022) has used such a framework to explore the dynamics of diversity and the transition to holoparasitism, and to highlight key gaps in our taxonomic knowledge. While Mortimer et al. (2022) were able to sample about 40% of the species diversity of the family Leedonidae, the lack of available data for many lineages, particularly minor genera (5 species) and genera from the paleotropics and neotropics, highlighted areas of interest for research. One of the genera in need of taxonomic study is the genus Rehmannia (Lydonidae), the third largest genus of Artemisia marcendae and includes 70 semi-parasitic species, nearly half of which have not yet been included in phylogenetic analysis. Approximately 40 species occur in North America, confined to the eastern part of the continent, and occur mainly in the coastal plains of the southern and eastern United States (Canne-Hilliker 1988), including several endangered taxa (Neel and Cummings 2004; Pettengill and Neel 2008, 2010, 2011). The remaining species are found in Central and South America, with centers of diversity located in the Puna communities of the Andes in Peru and Bolivia, as well as in southeastern Brazil, where many species are confined to the fragmented Campos Rupestris in Minas Gerais (Souza et al. 2001). Within its range, the Agalini species is commonly found in grasslands, savannas, and other open vegetation habitats lacking a closed canopy (Werneck 2011). While most species of Rehmannia are found in well-drained soils, some species occur in moist habitats, salt marshes, or wetlands. Almost all North and Central American species are annuals (two of which are perennial herbaceous plants, A. linifolia (Nutt.) Britton and A. gypsophila B. L. Turner), while South American species differed more in their habits, including slender annuals, suffrutescent and perennials, and large shrubs (Canne-Hilliker 1988). The habits of these species vary widely, and the presence of woodiness is unusual in typical herbaceae, in contrast to North American herbaceous species (Canne-Hilliker and Kampny 1991). North American varieties of the genus Rehmannia typically have pinkish-white to rose-purple bell-shaped corollas containing stamens, while South American varieties have pink-to-rose-purple bell-shaped corollas containing stamens, or narrow, tubular, red/scarlet corollas with exposed stamens (Canne-Hilliker 1988; Figure 1). The relationship between the genus Rehmannia and other genera of the family Ledangidae is not clear. Early studies classified the genus Rehmannia in the genus Rehmannia (1837), along with the genus Candle and Quercus glabra. (1837). Foxglove spp., Quercus rehmannia, Foxglove spp., Cassia nigra, Rehmannia spp., and Rehmannia lanceophyllum spp., as members of the Scrophularia family (Bentham 1846; Pennell1913）。 The family was later disbanded, resulting in the transfer of the genus Rehmannia and other semiparasitic taxa to the Leedonaceae family (Wolfe 1999; Olmstead et al., 2001, Tank et al., 2006). Although the name Lingcong grass is widely used, it is synonymous with the genus Monjouw in the family (Lanjouw 1961), while the name Rehmannia has been used by Britton (1913) and Pennell (e.g., 1913, 1928, 1929). In the genus gerardioid, Quercus glabradiol, Foxglove spp., Quercus rehmannia spp., Rehmannia spp., and Cassia nigra genera have been considered independent genera (sensu Pennell 1928). The genus Candelabra (Figure 2) is a monotypic genus with a range from Panama to Brazil, and its location has been uncertain. Although some taxonomic literature uses the name Rehmannia-hispidula (e.g., Canne 1980; Canne-Hilliker1988； Souza et al., 2001; Souza and Giulietti 2009), but Pennell (Mart.) has been accepted by Flora of the World Online (wfoplantlist.org). Campestris (Mart.) Benth., a synonym of E. splendida J.C. Mikan (Souza and Giulietti 2009), was identified as a sister of a sample of the genus Foxglove (Bennett and Mathews 2006; McNeal et al., 2013; Mortimer et al., 2022). Additional sampling of the genera Brushing and Rehmannia is needed to clarify the location of this species of Brazil pollinated by hummingbirds (Figure 2).
如图1.帚地黄属的花变异，显示钟状花冠（A-B）和漏斗状管状花冠（C-F），包括雄蕊（A-D）和外露雄蕊（E-F），颜色从淡粉色到红色变化/品红。 A.帚地黄属-kingsii；大开曼岛（照片：Maribeth Latvis）。 B.帚地黄属-genistifolia; 巴西（照片：Maribeth Latvis）。 C-D.帚地黄属-lanceolata;； 秘鲁（照片：Maribeth Latvis）。 E.帚地黄属- scarlatina；玻利维亚（照片：Judith Canne-Hilliker）。 F.帚地黄属- angustifolia；巴西（照片：Maribeth Latvis）。
As shown in Figure 2, the flower patterns of the genus Brushed Feather (A) and the genus Candlestick (B), two gerardioid genera that are taxonomically confused with Rehmannia (Photo: Maribeth Latvis). The taxonomy of genera and subgenera within the genus Rehmannia has been extensively revised. Pennell (1929) initially proposed five genera, but later re-limited them to three genera, five subgenera (1935). Work based on nutrition and seed morphology and karyotype led to further revisions and recognition of five genera: A. Heterophyllae Canne-Hilliker, A. Linifoliae Canne-Hilliker, A. purpureae Canne-Hilliker and A. Tenuifoliae There are three subgenera in the genus Canne-Hilliker, A. Pedunulares Canne-Hilliker, A. Purpureae Canne-Hilliker, and A. Setaceae CanneHilliker (Canne 1979, 1980, 1981, 1983, 1984; Canne-Hilliker 1987, Stewart and Canne-Hilliker 1998). Neel and Cummings (2004) examined 15 North American Rehmannia spp. using three chloroplast DNA (cpDNA) regions (rbcL, matK, and ndhF), providing a valuable test for previous morphology-based taxonomy. The monophyletic nature of A.section Erectae is supported, while A.section Purpureae and A.section Heterophyllae are multiphyletic, although limited taxon sampling and low sequence variation hinder conclusions about other families and subfamilies. Pettengill and Neel (2008) provided a more comprehensive phylogenetic framework by supplementing these data with additional cpDNA regions (rpoB, rps2, trnT-trnF, psbA-trnH) and nuclear ITS regions, and by expanding taxon sampling to include 29 species of A. rehmannia species. When multiple germplasms of each species are analyzed, many North American species form monophyletic groups (Pettengill and Neel 2008), while exceptions (e.g., A. acuta Pennell, Federally Protected Species) directly affect conservation status (Pettengill and Neel 2011). Pettengill and Neel (2008) restored 6 main branches, but the order of branches between them lacked sufficient resolution, and few previously restricted sections and subsections were restored. In view of these findings, an updated classification based on extended phylogenetic analysis is warranted. In a follow-up study, Pettengill and Neel (2010) used the genus Rehmannia as a model clade, further demonstrating the efficacy of the proposed barcode region in distinguishing rare congeners. While the studies reviewed above have greatly contributed to the understanding of this taxonomically difficult genus, South and Central American Rehmannia spp. were not sampled, thus missing significant diversity within this group. In order to fully assess the monophyletic nature of this genus, it is necessary to include the genera Brushed Feather and the genus Candeliforma and to perform comprehensive sampling within the geographical range of the genus Rehmannia (Canne 1980; Pettengill and Neel 2008). The aim of this study was to elucidate the phylogenetic relationships within and with closely related genera (Quercus spp., Quercus spp., Quercus rehmannia spp., Quercus rehmannia spp., Quercus spp., and Quercus spp.) and to address one of the major taxonomic gaps in Artemisia spp. (Mortimer et al., 2017). 2022）。 By adding previously unsampled species from the Caribbean and Central and South America, this significantly improved phylogenetic hypothesis will provide a framework for the reclassification of the genus and subsequent studies involving historical biogeography, diversification, and trait evolution. In addition, several species of the genus Rehmannia are geographically restricted in South and Central America, occurring in threatened habitats, such as the Andean Puna community and the Brazilian Cerrado (Ratter et al., 1997), and the phylogeny of well-sampled phylogeny of this genus is an important first study. Steps to guide future conservation efforts and resulting policy decisions, e.g. Pettengill and Neel (2011).
Materials and methods
Taxonomic sampling—Here, we propose a phylogenetic hypothesis based on DNA sequences involving 51 species of Rehmannia and 5 outergroups that were once thought to be the same genus as Rehmannia (gerardioid genera: Candle Bottle, Foxglove Spp., Quercus Radix, Quercus Radix, Quercus Radix, and Brushed Feather Spp.). We used available data from the North American taxa (Neel and Cummings, 2004; Pettengill and Neel 2008) and our own taxonomic data from South America, Central America, and the Caribbean, as well as the recently described species from Florida, A. flexicaulis Hays (2010). All newly sampled specimens for this study were legally collected with proper licensing. In addition to the 7 cpDNA and 1 nuclear region sampled in Pettengill and Neel (2008), we added 3 low-copy nuclear sites from the pentapeptide repeat (PPR) gene family (Yuan et al., 2009:P PRAT1G09680, PPR-AT3G09060, and PPR-AT5G39980, which have been used in the genus Verbena (Verbenaceae) (Yuan et al., 2010) and Bellflower (Platycodonaceae) (Crowl et al., 2014). In addition to centralized phylogenetic analysis of gerarddioid genera, we added data to Mortimer et al.'s densely sampled columnaceae supermatrix. (2022) Understanding the place of these ancestries in the wider family context. We included 179 individuals in this study, representing 51 species of Rehmannia (with a focus on previously unsampled South and Central American taxa), four species of Pedillaryia, Hispidula, and 4 other outergroups (Pedillaryia, Pectinata, Wrightii, and Macrophylla) (Appendix 1). Given the objectives of this study, we endeavoured to include the genus Lycopodium and the genus Candela, which includes multiple species of the genus Lycophyllum as well as the Philodula genus-hispidula germplasm from Brazil and Panama. DNA was obtained from J. Pettengill and M. Neel, as well as the outer groups of Quercus pedillaryia, Quercus glabratum-pectinata, Foxglove spp., and Quercus rehmannia. This study relied on the same germplasm of species of the genus Rehmannia and outergroups used in Pettengill and Neel (2008). Our sampling represented every part and subpart of the genus Rehmannia, spanning the geographic range of the genus, and including as many individuals of each species as possible in an attempt to consider the species as a testable hypothesis (Baum 1998; Pettengill and Neel 2008). However, for geographically restricted species (e.g., Starman, Etanbonsi, Golden Silk Star) and some hard-to-obtain Andean species, one individual is used per species. Notably, the critically endangered Rehmannia spp.-acuta is now considered synonymous with A. decemloba due to its lack of genetic and morphological distinctiveness (Pettengill and Neel 2011). DNA sequencing and alignment—Extract silica-dried material collected in Brazil using the QIAGEN DNeasy Plant Mini Kit (Qiagen Corporation, Valencia, CA) following the manufacturer's instructions, but extend the incubation time in AP1 buffer to overnight. Other silica dried material and plant specimens were extracted using a modified CTAB protocol (Doyle and Doyle 1987). We amplified and sequenced six chloroplast regions (rbcL, matK, trnT (UGU)-trnF (GAA), rps2, rpoB, and psbA-trnH), ribosomal ITS, and three single-copy nuclear site series (AT1G09680, AT3G09060, AT5G39980) from pentapeptide repeat (PPR) genes (Table 1). PCR profiles and primers for cpDNA and ITS amplification and Sanger sequencing mostly follow Pettengill and Neel (2008) and the references cited therein. Amplification and sequencing of putative single-copy nuclear PPR sites were performed according to the protocol outlined in Yuan et al. （2009、2010）。 For some taxa, degraded DNA extraction requires the use of internal primers to amplify regions larger than 800 bp (e.g., rbcL, Table 1). The three nuclear loci from the PPR gene family (AT1G09680, AT3G09060, AT5G39980, Yuan et al., 2009) were selected because they have been shown to have orthologous and intra-genus variability (Yuan et al., 2010; Crowl et al., 2014). Following the protocol outlined by Yuan et al., PCR amplification and Sanger sequencing were performed using a single forward and reverse primer pair (Table 1). （2010）。 After repeated attempts using different primer combinations and PCR reagents, we had limited success in amplifying and sequencing these regions in plant specimens. Thus, sequences obtained for these regions are incorporated into our data matrix, and unsuccessful joins are coded as missing data. All PCR products were purified using ExoSAP-IT (Affymetrix, Santa Clara, CA, USA) and sequenced on a ABI3730XLDNA sequencer (Applied Biosystems, Fullerton, CA, USA) following the manufacturer's protocol at the Florida Interdisciplinary Biotechnology Research Center. Use Sequencher 4.5 (GeneCodes, Ann Arbor, MI, USA) or Geneious 5.3.4 (Biomatters Ltd, Auckland, New Zealand; Drummond et al., 2011). Sequences were aligned using the default settings using MAFFT version 6 (Katoh et al., 2005) and then manually adjusted in Se-Al 2.0a11 (Rambaut 2002) or Geneious 5.3.4. The alignment portions where nucleotide homology is difficult to infer and the beginning and end of genes with incomplete data (close to the gap threshold of 0.2) are manually excluded from subsequent phylogenetic analysis. Gaps were coded as missing data and excluded from the analysis. Table 1.Primers for PCR and Sanger sequencing, primer sequences, and original citations.
Although tandem repeats of different copies in ITS may confound phylogenetic inferences (e.g., NietoFeliner and Rossello 2007), we did this by directly using two primer pairs (ITS1 and ITS4, or ITS4 and ITS5; Table 1) Sanger sequencing was performed consistently to obtain clean ITS sequences. When polymorphic sites are occasionally identified in the chromatogram, they are encoded as IUPAC fuzzy bases and are considered uncertain in the analysis. As Yuan et al. said. (2010), polymorphic loci due to putative intra-individual allele variants in PPR loci, when discovered, are also encoded as ambiguous characters. Sampling within the Ledanaceae family – we obtained nucleotide alignments from Mortimer et al. (2022) For three chloroplast regions, rbcL, matK, and rps2, and one nuclear region, ITS. Our sequences were added to the Lendaceae-wide dataset and realigned using MAFFT version 7.49 (Katoh and St & ley 2013) under the default settings of Geneious Prime 2022.1.1 (http://www.geneious.com, Kearse et al. 2012). The sequence of Paulownia (the same sister of the family Ledanidae) (Refulio-Rodriguez and Olmstead 2014) acts as an out-taxa. Phylogenetic analysis – We constructed a combined chloroplast dataset (rbcL, matK, trnT (UGU)-trnF (GAA), rps2, rpoB and psbA-trnH) and a combined chloroplast 1-core dataset (rbcL, matK, trnT (UGU) trnF (GAA), rps2, rpoB and psbA-trnH, ITS, AT1G09680, AT3G09060 and AT5G39980)。 Each nuclear region (ITS, AT1G09680, AT3G09060, and AT5G39980) was also analysed independently. For each data partition, we analyzed a reduced dataset containing only one individual per species and an extended dataset containing multiple individuals per species. The former has a more complete data matrix, while the latter has a greater proportion of individuals missing data for one or more gene regions. To represent each species as a single tip in a simplified dataset, the same sequence was removed from each alignment, retaining the germplasm with the highest cross-regional coverage. To further reduce missing data, in some cases, chimeric tips are created by combining regions of different germplasms of the same species (e.g., trnT-trnF from Rehmannia spp.-lanceolataJCH2663 and AT1G09680 from Rehmannia spp.-lanceolataMLPE01). Tables 2 and 3 describe the data partitions for shrinking and scaling datasets, respectively. We used ModelFinder (Kalyaanamoorthy et al., 2017), implemented in IQ-TREE version 1.6.12 (Nguyen et al., 2015), to find the best-fitting replacement model for each partition of all datasets using BIC, allowing each partition to have its own evolutionary model velocity. Ultra-fast guidance was used (UFBoot, Minh et al., 2013; Hoang et al., 2018) performed partition maximum likelihood (ML) analysis for all datasets and performed 1000 replicates using IQ-TREE to assess clade support. The model selected for each data partition (for the scaled-down dataset) and Table 3 (for the scaled-out dataset) are reported in Table 2. It should be noted that bootstrap support values from standard nonparametric bootstraps (BS) (implemented in previous phylogenetic analyses of Rehmannia spp.) (Neel and Cummings 2004; Pettengill and Neel 2008) than using UFBoot for ultra-fast bootloaders (in results), and these values cannot be directly compared with our support values (Minh et al., 2013). We consider clades with 95% support using UFBoot to have moderate support, which corresponds to 70-80% BS support using standard nonparametric guidance (Minh et al., 2013). Table 2.Simplified data partition list showing the nucleotide location of each region and the alignment length (no gaps) (in bp) of the tandem partitions analyzed, the number of sequences per partition, and the sequence evolution model inferred using ModelFinder (Kalyaanamoorthy et al., 2017).
Table 3.Extended data partition list showing the alignment length (no gaps) (in bp) of the tandem partitions for each region of nucleotide location and analysis, the number of sequences per partition, and the sequence evolution model inferred using ModelFinder (Kalyaanamoorthy et al., 2017).
The topology of all analyses is visualized in FigTree 1.4.4 (http://tree.bio.ed.ac.uk/software/figtree/). All supplementary plots, fasta files, and input and output files for all phylogenetic analyses are archived in the Dryad digital repository (Latvis et al., 2024). The Ledanidae-wide dataset consisted of sampling by Mortimer et al. (2022) combined with our matK, rbcL, rps2, and ITS data and analyzed in a similar manner to the ML analysis focused on Rehmannia spp. described above. We constructed a combined chloroplast dataset (matK1rbcL1rps2) and a combined chloroplast 1-kernel dataset (matK1rbcL1rps21ITS) and performed ML analysis in IQ-TREE and ITS, respectively. ModelFinder is used to find the best-fitting alternative model for each partition area. Partitions and models are reported in Table 4. Ultrafast bootstrap was used using IQ-TREE (UFBoot, Minh et al., 2013; Hoang et al., 2017) for partitioned ML analysis, 1000 replicates. Bayesian phylogenetic analysis was performed for each data partition described in the simplified Rehmannia dataset (one individual per species, with a more complete matrix) using MrBayes3.2.7a (Ronquist et al., 2012). We used PartitionFinder2 (Lanfear et al., 2017) to find the best-fitting replacement model implemented in MrBayes, using the greedy algorithm (Lanfear et al., 2012) and BIC for model evaluation, assuming link branch length (Supplementary Materials, Latvis et al., 2017). 2024）。 Each assay consisted of two independent MCMC runs for 30,000,000 generations, with four chains (one heated, three cold) at a chain temperature of 0.2. Trees are sampled every 1000 generations, and the first 25% is discarded as aging. The stationarity of the analysis is assessed by examining the mean standard deviation of the splitting frequency of the two runs.
outcome
Sequences and matrices – Table 3 reports the size of the alignment and the replacement model after the gap is removed. The above dataset centered on Rehmannia spp. excludes Hispidula because preliminary analysis shows that its location is outside other outer taxa with extremely long molecular branch lengths. In order to further elucidate the location of Hispidula genus, this taxon was included in the analysis of the family Ledanidae. Table 4 provides detailed information on the Column Dang family range partition and the best fit alternative model. Table 4.List of data partitions within the family showing the length of the analytical alignment (no gaps) (in bp) for nucleotide positions and ligation partitions in each region, the number of sequences per partition, and the sequence evolution model inferred using ModelFinder (Kalyaanamoorthy et al., 2017).
Phylogenetic analysis—The topology examined provided strong support for the monophyletia of the genus Rehmannia, which was well nested within the inner group and closely related to the species of Rehmannia brasiliensis (Figs. 3-6). Further discussion of the Foxglove clade will now also include a generic construct of the genus Brush. In general, we first discuss the results of machine learning analysis and then the results of Bayesian inference. For the scaled-down and expanded versions of our Foxglove-centric dataset, we recovered well-supported differences between cpDNA and nuclear partitioning. We first discuss the results of the cpDNA1nuclear combinatorial partitioning based on a reduced dataset, and then highlight the key differences between cpDNA and ITS reduced partitioning. The reduced Foxglove dataset had the least missing data, and the UFBS values were more likely to reflect the organic relationship. Next, we discuss the extended Foxglove dataset, which has multiple germplasms per species, but lacks more data to assess the monophylethraxis of the described species. We focused on the nuclear ITS topology on the analyzed PPR sites, as the ITS region has better taxon sampling. Individual topologies of PPR sites for the reduced (Figures S1-S3) and extended versions of the Rehmannia dataset (Figures S4-S6) are provided in the supplementary material (Latvis et al., 2024). For the simplified Rehmannia dataset, two clades were recovered by cpDNA+ nuclear partitioning (Figure 3): 1) supporting the poorer North American clades, including A. sectionsEectae, Heterophyllae (represented by A. auriculata), Linifoliae, Purpureae, and Tenuifoliae (UFBS=59%), represented by Pettengill and Neel ( 2008), and 2) a sample from A. A. Heterophyllae (A. heterophylla(Nutt.) Two species consisting of sisters Small ex Britton and A. calycina Pennell branched into the clade (UFBS = 100%) of all South American Rehmannia species (Figure 3). The well-supported South American clade (UFBS=96%) includes the Brazilian clade (UFBS=98%) and the predominantly Andean clade (UFBS=100%) (Figure 3). As previously mentioned, the species of the genus Brushed Feather belongs to the Brazilian clade, forming a moderately supportive clade (UFBS = 94%). Rehmannia genistifolia (Cham. and Schltdl.) D'Arcy is a species that is widely distributed in southern Brazil, Argentina, Chile and Bolivia, and belongs mainly to the Andean clade. The addition of three single-copy nuclear PPR sites, although sampling is incomplete, fails to resolve the order of branching between major lineages in North America, and the support value is still below 75% for UFBS. In contrast to several well-supported relationships between species of the genus Foxglove, we generally observe short molecular clanido lengths and low sequence variation between closely related species in the Brazilian and Andean clades, and most species-level relationships remain unclear in the current dataset (Figure 3). Exceptions include A. stenantha (Diels) D'Arcy and A. lanceolata Ruiz & Pav.) D'Arcy, and the close relationship between the genus Splendida J. C. Mikan and E. caesarea (Cham. and Schltdl.) V.C. Souza (both UFBS = 99%). As for the North American species that were first included in molecular phylogenetic analysis, A. flexicaulis (2010) belongs to the genus A. Eectae and appears to be a sister of A. obtusifolia Raf. （UFBS=100%）。 Rehmannia spp.-kingsii has a highly supportive sister relationship with A. maritima (Raf) in the subgenus A. Purpureae (UFBS=100%). Rehmannia spp. albida Britton and Pennell are another species from the Caribbean that also appear to be related to A. The Purpureae sensu stricto subgenus is closely related, but does not have strong UFBS support. Foxglove-gypsophila was placed in the subgroup A. pedunulares and was associated with A. strictifolia (Benth.) Pennell and A. pedunularis (Benth.) Pennell is in the highly supportive clade (UFBS=99%). The relationship recovered from the cpDNA partition of the simplified Rehmannia dataset (Figure 4) is similar to that recovered from the cpDNA + nuclear partition described above, forming a predominantly North American clade and South American clade + A. heterophylla and A. calycina. However, in the cpDNA topology, the North American clade was highly supported (UFBS 98%) and again showed A. filicaulis (Benth.) Pennell and A. divaricata (Chapm.) Pennell (A. section Tenuifolieae) is a sister of the rest of the branches. North American species (Figure 4). In contrast, the ITS topology (Figure 5) considers A. filicaulis and A. divaricata as sisters of other species of Foxglove (including all North and South American species) with high support (UFBS 99%). The North American clade suggested by the cpDNA dataset is not supported in the ITS dataset, and the North American Species forms a node with a poor support hierarchy, resulting in a highly supported clade containing South American species + A. heterophylla (UFBS 99%). The ITS sequence data for A. calycina is missing, so we are unable to place this taxon in A. heterophylla. In addition, there are some well-documented inconsistencies between cpDNA and ITS simplified datasets in North American species. It is worth noting that A. setacea (J.F. Gmel.) close relationship between them. and A. plukenetii (Elliott) Raf.ITS topology (Figure 5) does not support analyses based on the cpDNA dataset (UFBS100%, Figure 4) and morphology (Canne 1983), where the species is divided into two separate clades: A. plukenetii and A. tenuifolia (Vahl), A. albida, A. maritima, and A. kingsii ( A. subsectionPurpureae), A. laxaPennell (A. subsectionSetaceae) and A. gattingeri (Small) SmallexBritton (A. sectionElectae)) (UFBS 97%), and A. setacea and A. fasciculataElliott), A. purpurea ( L.) Pennell and A. paupercula (A. Gray) Britton (A. subsection Purpureae; UFBS 96%). The trees are based on an extended zoning centered on the genus Rehmannia spp., with multiple germplasms for each species, and each tree shows members of the sister branches of the South American clade, Heterophylla A. (A. calycina and A. heterophylla), which consists of the strongly supported Brazilian clade (including the genus Brushophylla) and the main one. Andean branch. The tree in the extended partition often reveals well-supported instances of inconsistency (e.g., the location of the conflict) between the combinations shown in the simplified Rehmannia dataset (Figure S5), cpDNA (Figure S6), and ITS (Figure S7 A.setacea and A.plukenetii), with poor resolution along the backbone of that genus. However, the cpDNA+ nuclear expansion tree (Figure S7) shows that A. filicaulis and A. divaricata (A. section Tenuifolieae) are sisters of the rest of the genus Rehmannia and are the ranks that lead to the South American clade+, A. calycina and A. heterophylla, similarly narrowing and enlarging ITS partitions (Figure 5 and Figure S9). Conversely, the extended cpDNA topology showed a weakly supported North American clade, which included A. filicaulis and A. divaricata (UFBS59%, Figure S8), and this relationship was more strongly supported in a simplified cpDNA partition with more data (UFBS98%, Figure 4).
As shown in Figure 3. IQ-TREE's maximum likelihood consensus tree is based on a cascaded cpDNA+nuclear partition of the simplified Rehmannia dataset. The number above the branch indicates UFBS support based on 1000 bootstrap replications. Chapters and subsections are color-coded.
As shown in Figure 4, the maximum likelihood consensus tree of IQ-TREE based on cpDNA partition is used for the simplified Foxglove dataset. The number above the branch indicates UFBS support based on 1000 bootstrap replications. Chapters and subsections are color-coded.
Figure 5.The maximum likelihood consensus tree of IQ-TREE based on ITS partitioning is used for the simplified Foxglove dataset. The number above the branch indicates UFBS support based on 1000 bootstrap replications. Chapters and subsections are color-coded. When examining the relationships between multiple sampled individuals of each species in our extended Foxglove division, most of the North American species described form distinctive, well-supported clades, while most South American species are polyphyletic (Figures S7-S9). Pettengill and Neel (2008) describe some of the ill-defined species boundaries in North American species, in particular the critically endangered A. acuta versus the polyphyletic A. decemloba (Greene) Pennell and A. tenella Pennell. In addition, A. harperi germplasm was resolved at different positions in the cpDNA expansion partition (either with A. gattingeri or with A. fasciculata, A. purpurea and A. paupercula, Figure S8), and together in ITS expanded partitioning (Figure S9), which led Pettengill and Neel (2008) to hypothesize a chloroplast capture scenario involving A. harperi. All extended-partition-based trees show a highly supported branch, including the germplasm of A. kingsii and A. maritima (UFBS>99%, Figures S7-S9) on relatively long branches; However, germplasm itself does not form clades corresponding to each species, and there is little molecular differentiation between them.
As shown in Figure 6, the circular illustration tree (bottom left) depicts the maximum likelihood consensus of the IQ-TREE for cpDNA+ nuclear partitioning based on the Ledangidae-range dataset (903 species). The families within the Column family are color-coded. The stars indicate detailed, extended phylogenetic locations: the genera Rehmannia spp., the genera Foxgilla, and other gerardioid genera are nested in the Artemisia family (red in the upper right corner), while the genus Candelabra is nested in the Buchnereae family (bottom right is blue, and Candelabusa is indicated in bold blue text). The number above the branch on the extension tree indicates UFBS support based on 1000 bootstrap replications. The individual gene trees of PPR-AT1G09680, PPRAT3G09060, and PPR-AT5G39980 (Figures S1-S6) each contain a highly supported South American clade (UFBS>95%), which included the genus Brush, which was consistent with other datasets and partitions; However, they often have low UFBS support values and limited taxon sampling, making it impossible to draw conclusions about the location of taxa with conflicting positions in other datasets (e.g., A. filicaulis, A. divaricata, A. harperi, and North American clades and ranks). The results of Bayesian inference analysis using MrBayes on the simplified Rehmannia dataset are consistent with the results reported by ML analysis using IQ-TREE for all partitions. The UFBS support values assessed in the IQ-TREE correlated with the posterior probability assessed in MrBayes. The 50% majority rule topology of MrBayes is shown in Figures 2 and 3. S10-S12。 Using Mortimer et al.'s Liedang University of Science and Technology Comprehensive Phylogeny. (2022), we were able to confidently place the genus Candeliformum in the Buchnereae family and analyze all partitions, while the genus Rehmannia and other closely related genera were placed in the Artemisia family as expected. In Buchnereae, the cpDNA+ nuclear topology places the sisters of the genus Candelabra in a clade composed of Hyobanche, Aeginetia, Christisonia, and Alectraalba (UFBS99%, Figure 6). The ITS partition (Supplementary Materials, Latvis et al., 2024) shows a very similar relationship with poor support (UFBS50%), with the exception of Aeginetia+Christisonia, which belongs to the poorly supported clade (UFBS50%) with Alectra, Escobedia, Melasma, and Centranthera. While cpDNA partitioning (Supplementary Material, Latvis et al., 2024) confirms the position of the genus Isochrysanthemum in Buchnereae (UFBS94%), its location lacks strong support, limiting definitive conclusions about closely related lineages.
discuss
These results have greatly improved our understanding of the phylogeny of Rehmannia and elucidated the relationship with isolated genera. Building on previous work by Pettengill and Neel (2008) and Neel and Cummings (2004), we provide almost twice the taxon samples, of which approximately 80% represent ingroup species targeting the South and Africa (now including the genus Brushed Feathers). Biodiversity of the genus Rehmannia in Central America. This is the first time that these species have been included in molecular phylogenetic studies, providing insights into the monophyletic nature of the genus Rehmannia spp., which have long been taxonomically confused with Rehmannia spp. The phylogenetic hypothesis proposed here provides a solid basis for taxonomic decision-making and future research on historical biogeography, diversification, and trait evolution within the clade. In addition, we used a comprehensive phylogenetic framework of Orobanchaceae within the Buchnereae family, a mysterious, monotypic genus away from the Foxglove and Artemisia equinsis families, within the Buchnereae family. Large, taxa-dense synthetic phylogeny, e.g., Mortimer et al. (2022), which not only shed light on the "dark regions" of trees that require research focus (sensu Hinchliff et al. 2015) and enable detailed macroevolutionary studies, they also provide an important resource for mobilizing taxonomic findings, as we do here. We discuss the important findings further below, with a focus on the newly incorporated South and Central American taxa. Brush Feather Genus-In, Candelabra Genus-Out - Blast Feather Spp.-J.C. Mikan (1821) currently consists of six species: E. caesarea (Cham. and Schltdl.) V.C.Souza、E.eitenorum Barringer、E.macrodonta(Cham.) Several species are now synonymous with the larger E. splendida J. C. R.B.de Mikan species complex (e.g., E. Campestris Spixex Mart., E. petiolata Barringer, E. nervosa Benth.). The nesting of the genus Brushed Feather in the Foxglove clade of Brasilica directly affects future taxonomic decisions for these genera. The name Rehmannia - Sensustricto is suitable for a larger clade covering 60-70 species, while the old name Brushed Feather is only for a few species. In order to maintain the stability of the nomenclature, we recommend retaining the name Rehmannia and transferring species from the genus Rehmannia to Rehmannia (Souza et al. are in preparation). Species currently thought to be of the genus Brushed Feather have a range of floral characteristics, which may be the result of hummingbird pollination syndrome, which distinguishes them from Rehmannia s.s.s., including red to orange tubular or funnel-shaped corollas, exposed stamens, and downy anther tepals. Hummingbird-pollinated species of Rehmannia may also possess one or more of these characteristics, blurring the morphological boundaries between the two genera (Souza et al., 2001). The location of the genus Foxglove relative to other Brazilian hummingbird pollinator species (e.g., A. angustifolia [Mart.]D'Arcy) is unclear, and the possibility of multiple evolutions of hummingbird pollination and related morphology within the Brazilian clade remains, particularly in the Campos rupestres lineage in the Spanish mountains, where the prevalence of hummingbird pollination is much higher than in other open-habitat upland communities (Vasconcelos and Lombardi 2001; Freitas and Sazima 2006). The Andean Foxglove hummingbird species with a red tubular corolla was placed in Virgularia (Von Martius 1829, Bentham 1837; Figure 1E), this distinction was recognized by Pennell (1928), but these species were later transferred to the genus Rehmannia d'Arcy (1978) due to its morphological similarity to the species of Rehmannia spp. in eastern North America. Barringer (1985) hypothesized that Virgularia should be considered a member of the genus Virgularia, although our results show a more distant relationship. The distribution range of the genus Monomorphic Candle Bottle extends from northern Brazil to southern Mexico, and was initially separated from Rehmannia due to its peduncles with small bracts, narrow seeds, and unequal anther sacs (Pennell 1920; Figure 2B). Pennell (1928, 1929) noted that the anther morphology of the genus Hispidula appears to "tend" to the single-celled anthers of Buchnera, Harveya, and Hyobanche in the family Buchnereae of the family Ledanidae, and hypothesized a close relationship with these lineages. Nonetheless, D'Arcy (1978) included the species in the genus Rehmannia, arguing that anther differences may be attributed to flower age and therefore insufficient to distinguish it from the general structure of the genus Rehmannia (D'Arcy 1978). Our results do not support the classification of Hispidula to the family Foxglove or Artemisia spp., but rather confidently place Ichspidula as a member of the Buchnereae family (Figure 6), as Pennell originally postulated (1928, 1929). We included the Candela-hispidula ensemble from Panama and Brazil, representing the southern and northern limits of its range, and the sequence variation between these germplasms was very low for the gene regions we sampled. Anther vesicles are taxonically important in the family Leedonaceae and have been used to divide taxa: the genera Rehmannia and Cassia nigra (Artemisia macula) always have the same film, while the Castillejinae subfamily (Artemisia masenidae) is defined by unequal anther vacuoles (Tank et al., 2009). In the Buchnereae family, Pseudosopubia, Sopubia, and Ghikaea have a significantly reduced follicular membrane and are sterile, while Buchnera, Striga, and Rhamphicarpa have true single-shell anthers with a complete abortion of one follicular membrane (Hoffmann and Fischer 2004; Fischer2004）。 Pennell's observations (1928, 1929) are well-founded regarding the similarities observed in the reduction of anther vesicles observed in Candelabra and Buchnereae. In Buchnereae, our Ledanidae-wide cpDNA1 nuclear partition placed the genus Candelabra within a highly supportive clade (UFBS = 99%, Figure 6), which included Centranthera, Melasma, Escobedia, Alectra, Hyobanche, Christisonia, Aeginetia, Harveya, and Parastriga, It was resolved into a holoparasitic clade composed mainly of Hyobanche, Christisonia, Aeginetia, Harveya, and Parastriga (UFBS=99%). The relationship of this annual semiparasitic taxa (based on green and well-developed vegetative morphology) to the all-parasitic clade of the Ledanaceae family will provide a critical context for understanding the parasitic evolution between these lineages, especially in light of the known changes in rbcL gene expression between Hyobanche and Harveya (Wolfe and R and le2001; R and Le and Wolfe 2005). In addition, most lineages within this broader clade are found in Africa and Asia, and there is evidence of at least two transitions to the New World (Melasma physalodes (D. Don) Melch.+Escobedia, and within Alectra; Morawetz and Wolfe 2009). The genus Candle represents another neotropical lineage embedded in this clade, and its location will help us understand the global biogeographic patterns of these tropical plants. Topological inconsistencies in the base of Rehmannia spp.—Within Rehmannia spp., we found disagreements between genomic partitions regarding the highly supported support of A. divaricata + A. filicaulis (A. Tenuifolieae): sisters of the North American clade with cpDNA partitions (UFBS=98%, Figure 4)) and sisters of other genera Rehmannia (including North and South American species) with ITS partitions (UFBS=99%, Figure 5). Pittengill and Neel (2008) also describe similar topological inconsistencies in these taxa, but their position relative to A. heterophylla + A. calycina (A. section Heterophyllae) and South American taxa is unclear. Both species have G3 endangered status (NatureServe 2022) and are found in the remains of the longleaf pine savanna in the southeastern United States. Morphologically, they are taxonomicly difficult and have features that distinguish them from other species of the genus Rehmannia, such as unusual trichome morphology, seedling morphology (Canne 1983), and stem anatomy (Canne Hilliker and Kampny 1991), as well as flattened, occluded corolla (Pettengill and Neel 2008). This nuclear inconsistency (e.g., Soltis and Kuzoff 1995) may reflect hybridization or chloroplast capture at the base of Rehmannia spp., but thoroughly unraveling the effects of these processes requires a multi-pronged approach to large multi-locus datasets (e.g., species trees). and network analysis, inconsistency quantification, topology testing; Morales-Briones et al., 2021. The South American clade + A. Heterophyllae also differed in position between plastid and nuclear datasets, forming an early divergent clade sister of the major North American clade in the cpDNA partition (UFBS = 100%, Figure 4), but a North American species nested within the ITS partition (Figure 5), despite a lack of statistical support. Relationships between North American species – As mentioned earlier, most of the relationships we recover in the North American lineage and the overall support values for topology are highly similar to those reported in Pettengill and Neel (2008), with the main exception being the A. Heterophyllae section, which we will discuss below. Therefore, we will discuss the differences between our results and those of Pettengill and Neel (2008) and the location of the additional taxa sampled in this study. High UFBS support values support the nomenclature of most species that contain multiple individuals, but shorter internal clades and low UFBS in the topology mask deeper clade patterns on the phylogenetic backbone. In a biogeographic analysis of temperate species in North America, Roy et al. (2020) suggests that rapid radiation between the major clades of the genus Rehmannia occurred during the mid-to-late Miocene, accompanied by a drop in temperature and greater seasonality, which may confound the internal clade pattern. Although Pettengill and Neel (2008) argue that the A. heterophylla group represents all other sisterhoods of the genus Rehmannia, our analysis suggests that this group (represented by A. calycina and A. heterophylla) may have formed a well-supported sisterhood with the new A. group. Species of Rehmannia spp. were sampled from South America. Pennell (1935) had suggested early differentiation locations for heterophyllous A. nodes because of their relatively large, lanceolate leaves, glabrous stems, rectangular capsules, and anther morphology, but Canne (1981) noted that many of these features also occur in South American species of Rehmannia species. Although this taxon was previously described as Tomanthera auriculata (Michx.) Raf., but now it is Foxglove - auriculata (Michx.) S. F. Blake (1918) and is placed in the genus A. Heterophyllae because they share some common stem and leaf anatomical features, such as thickened outer epidermal cell walls, thick-horned tissues, tangentially elongated fibrous bands in the stem, and subepidermal thick-horned tissues in the midrib of leaves (an unusual feature of Rehmannia spp.) (Canne-Hilliker and Kampny, 1991). However, our results suggest that A. auriculata is more closely related to other North American species than to A. section Heterophyllae+ South American clade. Foxglove densiflora, a vulnerable species confined to the calcareous prairies of Kansas, Oklahoma, and Texas (NatureServe 2022), was also separately identified as Tomanthera and is now included in the genus A. Heterophyllae. We are unable to include Densiflora, and its relationship to the previously mentioned species is uncertain, although it shares morphological similarities with auriculata (CanneHilliker and Hays 2020). With the addition of A. albida and A. kingsii, we obtained more representation of Caribbean species, both of which belong to the subgenus A. Purpureae. Foxglove spp.-kingsii is endemic to the Salina wetlands of Grand Cayman and is known only in a small number of small populations in communities dominated by alder (A. chinesis) and huakrasa (sedges) (Diochon et al., 2003). It was listed as critically endangered on the IUCN Red List of Threatened Plants in 2014 and is protected by the Cayman Islands National Trust (Burton and Barrios 2014). Although Diochon et al. (2003) investigated the ecological and habitat requirements of A. kingsii, this species is taxonomically poorly known and was not included in molecular analyses prior to this study. The phylogeny of the inclusion of A. kingsii in the range of Foxglove is an important contribution to the development of effective conservation strategies (reviewed by Soltis and Gitzendanner, 1999). The results of the simplified Rehmannia dataset show that A. kingsii has a highly supportive sister relationship with A. maritima in all partitions (UFBS=100%, Figures 3-6) and that a fairly long molecular clade separates these two species from other North American species. Foxglove-maritima is a halophytic salt marsh species with a wide distribution, extending from the Gulf Coast to Nova Scotia, and is recognized for its fleshy, blunt leaves and sepals, branching structure, and longer peduncle compared to other members of the genus A. Purpureae. Pennell (1929) noted that these morphological differences may represent local adaptation to halophytic habitats. When examining the topology from the extended Foxglove dataset, two germplasms of A. kingsii were mixed with those of A. maritima, and there was no support for the different clades of the two species (Figures S7-S9). It is still possible that A. kingsii nested within A. maritima, representing the Caribbean ancestry of this widely distributed species, which could affect its conservation status. Pennell (1929) classified the genus Foxglove-albida into the genus A. Purpureae and distinguished the species by very pale or white corolla, depressed globular capsules, narrow seeds, and short leaves that appear to rise rather than unfold. Foxglove-albida is found in the pine forests of Jamaica and Cuba, and Pennell believes that the distinctive morphology of this species is due to "a number of isolated genetic lines that have long been characteristic of its particular geographical area" (Pennell 1935, p. 119), representing an early divergent lineage with the rest of the section. Our analysis resolved the relationship between A. albida and members of the subgenus A. Purpureae and Setaceae + A. gattingeri (erect fraction) (UFBS=97%, Figure 3) within the clade, but lacked support for explicit placement. Two species from Mexico, the genus Rehmannia-pedunulousis and A. gypsophila, were also first shown in molecular analysis. Based on our results, both species are apparently members of the subgenus Pedunulouses of the genus A. Purpureae, a clade consisting of several species from southwestern North America and Mexico. Pennell (1929) hypothesized that the relationship between A. pedunularis and A. strictifolia was very close, as they both had strongly upward-curved corollas and relatively long sepal lobes. Foxglove-pedunularis is characterized by its exceptionally large corolla, very long peduncles, and unmodified racemes, which Pennell (1929) interprets as "primitive" within the genus. Based on these morphological characteristics, as well as the distribution throughout Mexico, Pennell (1929) further hypothesized that A. pedunularis represented an ancient species, providing evidence of a Mexican "gateway" to North America from the south. Foxglove-gypsophila is a soil endemic to gypsum sediments, first described by Turner (1986) from Cerro Potosı, Nuevo León, Mexico. This taxon is similar to A. pedunulousis in corolla and fruit morphology, and although the corolla is slightly smaller and the stem is glabrous, it is easily distinguished by its perennial habits and aerial stem bundles that emerge from the tough perennial canopy (Turner 1986). The close relationship between A. gypsophila, A. pedunularis, and A. strictifolia proposed by Pennell (1929) and Turner (1986) is also supported by these analyses, and these species form a highly supported clade across all topologies. It is still possible that A. gypsophila is a lineage nested in A. pedunularis or A. strictifolia, but improved taxonomic sampling of these more widely distributed species is needed to address this issue. Finally, we included in the sample the genus Rehmannia-flexicaulis, which was first described by Hays (2010). This species occurs in the mesoprairie to pine savanna of northeastern Florida and may be endemic to these communities. Compared to its Florida counterpart, foxiculis appears to be less tolerant of drought conditions. This species is distinguished from other long-stemmed species of the genus Rehmannia by its weak main stem ascent, which allows the plant to have a spreading or drooping growth habit and is supported by surrounding vegetation (Hays 2010). Our analysis placed A. flexicaulis within the clade formed by most members of the genus A. Erectae, and as a sister species of A. obtusifolia, it has high support in a tree based on cpDNA+nuclear and cpDNA reduction dataset partitioning (UFBS=100%, Figure 3), 4), but its location in the ITS simplified partition tree is unclear (Figure 5). Flexicaulis shares some morphological features with A. obtusifolia, including strong quadrangular stems and linear key-shaped leaves. Relationships in the South American clade – about 25-30 species of Rehmannia are thought to be present in South America, with centers of diversity located in the Andean Puna community and in the southeastern highlands of Brazil, specifically the Campo srupestres vegetation in Minas Gerais (Cannes-Hilliker 1988, Souza et al. 2001). South American species have received much less attention than their North American close relatives, and it is likely that there are other species waiting to be discovered (J. Canne-Hilliker, personal communication). Members of this South American clade have greater morphological variability than North American species, including annual and perennial taxa (including woody members) and diverse corolla morphology. Rehmannia spp. usually have bract leaves whose size does not decrease along the inflorescence axis, so the flowers are often described as solitary or axillary (Canne-Hilliker 1988). In contrast, North American species have been described as having racemes, with some annual taxa with characteristic modifications (Pennell 1929). Despite these general differences, the basic arrangement of the Agalini flowers is the same in North and South America. In these analyses, all South American species formed a highly supportive clade in all topologies, sisters of the genus A. heterophyllae (represented in this study by A. calycina and A. heterophylla). Branching patterns in the topology suggest that it is likely that the genus Rehmannia entered South America from the north, rather than the other way around, as suggested by Pennell (1929), but this requires a formal biogeographic analysis. This South American clade is further divided into well-supported Brazilian and Andean clades in all topologies; However, the level of sequence variation in both clades is very low, and the shorter molecular branch length hinders the inference of most species-level relationships. We believe that this may be the result of the recent rapid radiation of the genus Rehmannia to South America, a pattern that can also be found in many other flora (Donoghue 2008 review; Antonelli and Sanmart 2011; Hughes et al., 2013). Both the Brazilian and Andean clades include lowland species distributed in open habitats, as well as species confined to high-altitude grassland communities (e.g., Pune, Camposrupestres, Camposed de Campos altitude), which increases the likelihood of biome transitions within South America. In the Brazilian branch, A. communis (Cham. and Schltdl.) D'Arcy is an elongated annual taxon more widely distributed in lowland grassland ecosystems, highly supported as an early divergent lineage and a sister to all other species in Brazil. Reduced cpDNA + nuclear partitioning (UFBS=98%, Figure 3), or in reduced cpDNA partitioning (UFBS=97%, Figure 4) and reduced ITS partitioning (UFBS=87%, Figure 5). bandeirensis Barringer early differentiation of trees. As mentioned earlier, there is very low sequence variation to elucidate the relationship between the remaining Brazilian species, but the associated changes in hummingbird pollination and its corolla morphology are likely derived conditions within the genus Rehmannia. In addition, some plants endemic to Campos Rupest are woody perennials, a condition that also appears to have originated from the Brazilian clade. In the major Andean clades, we find a sisterly relationship (UFBS> between the genus Rehmannia-stenantha and the widespread A. lanceolata95%, Figure 3-5). These taxa are also closely associated with A. pennellii Barringer in a simplified cpDNA partition tree (UFBS=98%, Figure 4). Both A. stenantha and A. lanceolata are shrub-like, and the latter produces a low, stout trunk after several years of growth, which contrasts with the herbaceous A. pennellii (Canne-Hilliker 1988). Genistifolia is an extensive taxa found in the southern regions of the genus Rehmannia (Argentina, Chile, Paraguay, Uruguay, Brazil, Bolivia and Peru) and is apparently nested within the Andean clade. The Andean clade of the genus Foxglove also has hummingbird pollinated species, conditions that appear to occur independently in both the Andes and Brazil. While we were able to sample many species of Rehmannia spp. in South America, one species of particular taxonomic significance was not taken into account in our collection: Rehmannia spp.-mudulosa (GMBarroso) VCSouza, native to the seasonally dry tropical biome of Pará, Brazil. , Mato Grosso and Mato Grosso do Sul. This species was isolated from the genus Rehmannia and was identified by Barrosso (1956) as a genus of Schizocillus based on its unusual zygotic calyx and multipart part of the calyx or lobes. It was later transferred to the genus Rehmannia because the morphological differences were considered too small to be considered a separate genus (Souza and Giulietti 2009). The inclusion of this species is needed to further examine the general limits of the genus Rehmannia. Canne-Hilliker (1988) also provides haploid chromosome counts for 15 populations of 10 species, most of which are from the Andean clade (A. communis from the Brazilian clade is also added), with a "monotonic consistency" of n516. This variation in the number of concordance chromosomes does not provide any useful information in deciphering the relationship between these taxa, and contrasts with the highly morphological variation found within the Andean clade. Although chromosome counts for the rest of the Brazilian species are still required, the base number for all South American species is interpreted as n=16 (Canne-Hilliker 1988). In contrast, the chromosome numbers of North American species with lower morphological diversity have been recorded as n=13 or n=14 and have been incorporated into the taxonomic classification of these taxa (Canne 1981, 1984). In the major North American clade, n=14 for all species, except A. sections Pedunulares and Eectae, which have n=13, and are supported by a minority in Neel and Cummings (2004) and Pettengill and Neel 2008) as sister groups. Heterophyllae was resolved as a sister of the South American clade and also had n=14. Chromosome counts have not been performed on the sampled North American outgroups, but the rest of the genus Rehmannia has n = 14, which appears to be the ancestor haploid chromosome number of this clade (Canne 1981, 1984; Thus, it is inferred that diploid occurs twice, once in the major North American clade (from n=14 to n=13) and once in the South American clade (from n=14 to n=16).
conclusion
The phylogenetic hypothesis of the genus Rehmannia proposed here has greatly improved the understanding of the evolutionary patterns of this different genus in the family Ledanidae. In particular, we provide a significantly expanded taxonomic sampling that excludes representation of the substantial diversity found in South America, Central America, and the Caribbean, compared to previous molecular studies of Rehmannia (Neel and Cummings 2004, Pettengill and Neel 2008). In the process, we also contributed to the necessary taxonomic organization of other small and poorly understood genera in the family Ledanidae (B. spp. and S. candelabra). The family Ledandae consists of several small genera (5 species) that have not yet been included in molecular studies (Mortimer et al., 2022), and they may be nested within larger genera, as exemplified by the genus Brush, or misunderstood morphologically based on taxonomic kinship, as in the case of the genus Candeliforma. Our results will provide a basis for the study of interesting biogeographic patterns, character evolution, and diversification transitions within the genus Rehmannia, as well as provide a background for future research in the Buchnereae family. In addition, the relationships restored in our analysis will be incorporated into future taxonomic and naming decisions, especially with regard to species now included in the genus Brush. Increasing molecular and taxon sampling is an important next step in determining the species level relationships and species definition of Rehmannia species. Hundreds of single-copy nuclear loci can now be rapidly generated to infer species trees using multi-species merging methods and tease out sources of gene tree inconsistencies. These methods are particularly important for unraveling the speciation history of the Agarinis members of the recent and rapidly radiating South America, and to provide a robust topology to explain their diverse morphology and ecology. In addition, by efficiently placing the genus Candeflora in a wider Leedon family synthesis (Mortimer et al., 2022), we demonstrated the utility of a large, species-dense phylogenetic framework in taxonomic discovery.
期刊：Systematic Botany
文章标题：An Evolutionary Framework for Agalinis (Orobanchaceae; The False Foxgloves) Reveals a Rapid South American Radiation that Includes Esterhazya
作者信息：Maribeth Latvis,Vinicius Castro Souza,David C. Tank, Pamela S. Soltis,and Douglas E. Soltis
Original link: https://datadryad.org/stash/share/J5qo1ssQgGbaQNX7kxYZYgyWxXTtMO6_UPsD0l-nEzg
The pictures in the article and the cover picture are from the original article