A new classification of Cyperaceae (Poales) supported by phylogenomic data

Cyperaceae (sedges) are the third largest monocot family and are of considerable economic and ecological importance. Sedges represent an ideal model family to study evolutionary biology due to their species richness, global distribution, large discrepancies in lineage diversity, broad range of ecological preferences, and adaptations including multiple origins of C4 photosynthesis and holocentric chromosomes. Goetghebeur′s seminal work on Cyperaceae published in 1998 provided the most recent complete classification at tribal and generic level, based on a morphological study of Cyperaceae inflorescence, spikelet, flower, and embryo characters, plus anatomical and other information. Since then, several family‐level molecular phylogenetic studies using Sanger sequence data have been published. Here, more than 20 years after the last comprehensive classification of the family, we present the first family‐wide phylogenomic study of Cyperaceae based on targeted sequencing using the Angiosperms353 probe kit sampling 311 accessions. In addition, 62 accessions available from GenBank were mined for overlapping reads and included in the phylogenomic analyses. Informed by this backbone phylogeny, a new classification for the family at the tribal, subtribal, and generic levels is proposed. The majority of previously recognized suprageneric groups are supported, and for the first time, we establish support for tribe Cryptangieae as a clade including the genus Koyamaea. We provide a taxonomic treatment including identification keys and diagnoses for the 2 subfamilies, 24 tribes, and 10 subtribes, and basic information on the 95 genera. The classification includes five new subtribes in tribe Schoeneae: Anthelepidinae, Caustiinae, Gymnoschoeninae, Lepidospermatinae, and Oreobolinae.


Introduction
Cyperaceae (sedges) are the third largest monocot family (>5600 spp.; Govaerts et al., 2020) and are of considerable economic and ecological importance (Simpson & Inglis, 2001;Spalink et al., 2016aSpalink et al., , 2018. Cyperaceae are an ideal model family to study evolutionary biology due to their species richness, global distribution, large discrepancies in lineage diversity (Escudero & Hipp, 2013), broad range of ecological preferences and diverse phenotypes (Naczi & Ford, 2008), multiple origins of C 4 photosynthesis (Besnard et al., 2009), and the presence of holocentric chromosomes . The family is species-rich in the tropics where it exhibits high generic diversity and a remarkable species richness in the genus Cyperus L. with >960 spp. (Govaerts et al., 2020). High diversity in temperate regions is mostly due to the megadiverse genus Carex L. with >2000 spp. (Govaerts et al., 2020). The history of the family goes back to the early Cenozoic, as supported by a reliable fossil record dating back to the Paleocene (Smith et al., 2009;Spalink et al., 2016b) and evidence of large genera already established by the end of the Eocene (Jiménez-Mejías et al., 2016a), with a probable origin in South America (Spalink et al., 2016b).
Molecular phylogenetic studies on Cyperaceae have relied heavily on relatively few loci, such as a selection of plastid markers and the nuclear markers ITS and ETS (e.g., Semmouri et al., 2019). However, phylogeny estimation is more accurate when conducted with tens to hundreds of nuclear loci, because larger numbers of informative characters help to resolve short branches, and historical processes such as deep coalescence can be taken into account (Johnson et al., 2019). Hence, reducedrepresentation sequencing methods have been developed to sample hundreds of nuclear, orthologous single-copy genes for plant phylogenetic studies (Kadlec et al., 2017;Couvreur et al., 2019;Johnson et al., 2019;Villaverde et al., 2018Villaverde et al., , 2020, allowing users to yield data sets of a larger scale for phylogenetics without the bioinformatic challenges and costs associated with whole-genome sequencing. Larridon et al. (2020) provided an overview of earlier high-throughput sequencing studies on Cyperaceae, whereas more recent studies relying on genomic data already show alternative phylogenetic structure in certain sedge groups not previously recovered using Sanger sequencing (Léveillé-Bourret et al., 2018c;Larridon et al., 2020;Starr et al., 2021;Villaverde et al., 2020Villaverde et al., , 2021. The aim of this study is to resolve the high-level relationships in Cyperaceae and to test the monophyly of the tribes and genera as currently accepted to generate a new classification from subfamily to generic level. We hypothesize that using genome-scale data and an in-depth sampling will provide significantly more phylogenetic information to resolve the topology of the Cyperaceae Tree of Life. Equally, we postulate that the high-throughput technique-targeted sequencing will enable sequencing historical herbarium specimens with poor DNA quality (Brewer et al., 2019), allowing us to place previously unplaced genera in the family phylogeny for the first time.
2 Material and Methods 2.1 Taxon sampling A total of 361 accessions of Cyperaceae were sampled, along with 21 accessions representing other families in order Poales as outgroups (Table S1). The sampling includes nearly all currently accepted genera of Cyperaceae (Govaerts et al., 2020). Three monotypic genera were not sampled, and have never been successfully sequenced using Sanger methods: Nelmesia Van der Veken and Trichoschoenus J.Raynal, which are only known from their type collections, and Rhynchocladium T. Koyama. Costa et al. (2021a) recently changed the generic circumscription in tribe Cryptangieae, re-establishing the monotypic genus Didymiandrum Gilly, whereas Barrett et al. (2021b) recently described a new monotypic genus Ammothryon. These two genera were not sampled. Lab work for samples of three additional monotypic genera, Blysmopsis Oteng-Yeb., Capeobolus Browning, and Khaosokia D.A.Simpson, and the small genus Blysmus Panz. ex Schult. did not provide data of sufficient quality. These genera have been previously successfully placed in the Cyperaceae Tree of Life (e.g., Léveillé-Bourret et al., 2014, 2018cLarridon et al., 2018a;Semmouri et al., 2019). In total, 311 of the 382 accessions were sequenced after enrichment with the Angiosperms353 probes. In addition, 36 accessions enriched with the Angiosperms I kit for Anchored Phylogenomics (Léveillé-Bourret et al., 2018c), including Khaosokia caricoides D.A.Simpson, were mined for reads overlapping with the data generated using the Angiosperms353 probes, as were 6 accessions enriched with Cyperaceae-specific probes , and 20 transcriptomes available on GenBank (Table S1). Angiosperms353 data for most accessions were newly generated for this study, following the protocol established by Baker et al. (2021). In addition, some data were obtained from recent studies (Larridon et al., , 2021cStarr et al., 2021; Table S1).

DNA extraction, library preparation, hybridization, and sequencing
The voucher information and treatment of each sample are provided (Table S1). Molecular work for accessions enriched with the Angiosperms353 probes was carried out at the Sackler Phylogenomics Laboratory, within the Jodrell Laboratory at Royal Botanic Gardens, Kew (Richmond, Surrey, UK). Genomic DNA was extracted from leaf tissue obtained from herbarium specimens or silica-dried samples, using either a modified CTAB approach (Doyle & Doyle, 1987) or a CTAB protocol, based on Beck et al. (2012), modified for optimal simultaneous extraction of 96 to 192 samples (i.e., one or two plates) from degraded (i.e., herbarium) samples (see Supplementary Data Sheet 1 in Larridon et al., 2020). Lastly, 76 accessions were sourced from the Kew DNA Bank (http://dnabank.science.kew.org/) ( Table S1). The samples extracted using a CTAB approach were purified using Agencourt AMPure XP Bead Clean-up (Beckman Coulter, Indianapolis, IN, USA). All DNA extracts were quantified using a Quantus™ Fluorometer (Promega Corporation, Madison, WI, USA) and then run on a 1% agarose gel to assess the average fragment size. Samples with a very low concentration (not visible on a 1% agarose gel) were assessed on an Agilent Technologies 4200 TapeStation System using Genomic DNA ScreenTape (Santa Clara, CA, USA). DNA extracts with average fragment sizes above 350 bp were sonicated using a Covaris M220 Focused-ultrasonicator™ (Covaris, Woburn, MA, USA) following the manufacturer′s protocol to obtain an average fragment size of 350 bp. Dual-indexed libraries for Illumina® sequencing were prepared using the DNA NEBNext® Ultra™ II Library Prep Kit and the NEBNext® Multiplex Oligos for Illumina® (Dual Index Primers Set 1 and 2) from New England BioLabs® (Ipswich, MA, USA) following the manufacturer′s instructions but at half the recommended volumes. The quality of the libraries was evaluated on the TapeStation using High Sensitivity D1000 ScreenTape and the libraries were quantified using a Quantus Fluorometer. The final average library size including the adapters was c. 500 bp. Afterward, the samples were pooled and enriched with the Angiosperms353 probes (Johnson et al., 2018) following the manufacturer′s instructions (myProbes® Manual v4.01, Arbor Biosciences, Ann Arbor, MI, USA). Final products were again run on the TapeStation to assess quality (i.e., average fragment size) so they could be pooled equimolarly for sequencing. After multiplexing library pools, sequencing was performed on an Illumina® MiSeq instrument (San Diego, CA, USA) with v2 (300 cycles at 2 × 150 bp) or v3 (600 cycles at 2 × 300 bp) chemistry at Royal Botanic Gardens, Kew (Richmond, Surrey, UK), or on an Illumina® HiSeq (San Diego, CA, USA) at either Macrogen (Seoul, South Korea)  2.3 Read processing, assembly, and phylogenomic analyses Bioinformatics settings follow Larridon et al. (2021c). Raw reads were trimmed to remove adapter sequences and portions of low quality with Trimmomatic v.0.39 (Bolger et al., 2014) (Johnson et al., 2016) was used to process the qualitychecked, trimmed reads, with default settings except for minimum coverage set to 4×. Paired and unpaired reads from all accessions were mapped to targets with BLASTx (Altschul et al., 1990) using the Angiosperms353 target loci amino acid (AA) sequences (see Supplementary Data Sheet 3 in Larridon et al., 2020). Mapped reads were then assembled into contigs with SPAdes v.3.13.1 (Bankevich et al., 2012). Subsequently, exonerate v.2.2 (Slater & Birney, 2005) was used to align the assembled contigs to their associated target sequence and remove intronic regions (exons data set). HybPiper flags potential paralogs when multiple contigs are discovered mapping well to a single reference sequence. As few random paralog warnings were raised, no sequence was excluded.
Phylogenomic analyses were executed in two rounds (following Zuntini et al., 2021) to improve the inference results. In the first round, all exon sequences with at least 50 bp were recovered and then aligned with MAFFT v.7 (Katoh and Standley, 2013) with the "localpair max iterations 1000" option; sites with more than 30% missing data were removed using Phyutility (Smith & O′Meara, 2012), after which IQ-TREE v.2.1.0 (Minh et al., 2020) was run per gene, followed by TreeShrink (Mai & Mirarab, 2018) with threshold set to 0.05. After this, a quality check was performed to see how many times each accession appeared in each gene tree. Finally, ASTRAL-III v.5.5.11  was run after collapsing branches ≤10% bootstrap (BS) support using Newick Utilities (Junier & Zdobnov, 2010). This round provided the preliminary result. In the second round, again all sequences with at least 50 bp were recovered, those flagged by TreeShrink were removed, and then aligned with MAFFT, after which we generated summary stats in AMAS (Borowiec, 2016). Short alignments (<100 bp) were removed.
For the concatenated IQ-TREE analysis, the individual gene alignments were concatenated in AMAS, and IQ-TREE was run with mode set to "MFP + MERGE" and 10 000 replicates of ultrafast bootstrap replications (Hoang et al., 2018) to generate the final result. We also calculated two measures of genealogical concordance in our data set, the gene concordance factor (gCF) and the site concordance factor (sCF), using the options "-gcf" and "-scf" in IQ-TREE. Trees were plotted in FigTree v.1.4.4 (https://github.com/rambaut/ figtree/releases).

Capture success and data quality
The success of sequence recovery was variable, with an average of 177 genes per sample (above 25% of target size) and 41% of the total potential target (260 802 bp), as indicated in Johnson et al. (2019). The recovery of samples hybridized with other kits was significantly lower: for samples hybridized with the Angiosperms I kit for Anchored Phylogenomics, the recovery was 65 genes, on average, above 25% of target length and 17% of total potential length, whereas samples hybridized with Cyperaceae-specific probes yielded, on average, 45 genes and 18% of potential length (Table S2, Fig. S1).

Phylogenetic relationships
The tree resulting from the coalescent ASTRAL analysis is shown in Fig. 2, and the tree resulting from the concatenated IQ-TREE analysis is shown in Fig. S2. As relationships are very congruent, below we will discuss the relationships as shown in Fig. 2.
Cyperaceae are retrieved as a monophyletic family sister to Juncaceae with strong support (Fig. 2). Within Cyperaceae, 16 main clades are recovered (Fig. 2). Clade 1 represents subfamily Mapanioideae and includes two sister clades representing the tribes Chrysitricheae and Hypolytreae. Clade 2 represents tribe Trilepideae. Clade 3 represents the speciespoor tribe Cladieae. Clade 4 is the Bisboeckelereae-Sclerieae Clade, which includes two subclades representing tribe Bisboeckelereae and tribe Sclerieae, respectively. Clade 5 represents the species-poor tribe Carpheae. Clade 6 is formed of the genus Koyamaea W.W.Thomas & G.Davidse sister to a clade representing tribe Cryptangieae. Clade 7 represents the diverse and species-rich tribe Schoeneae. Tribe Schoeneae includes a range of well-supported clades; however, the nodes in the backbone of the tribe are not all well supported. Also, its position in the backbone of the family is not well supported (LPP = 0.76). Clade 8 consists of tribe Rhynchosporeae. Clade 9 or the Scirpo-Caricoid Clade (SCC Clade) includes a range of species-poor and species-rich lineages: Dulichieae, Khaosokieae, Calliscirpeae, Scirpeae, Trichophoreae, Sumatroscirpeae, and Cariceae. Each tribe is well supported as a monophyletic group as are the backbone nodes. Clade 10 is the Abildgaardieae-Eleocharideae Clade, which falls apart into two sister clades representing the tribes Abildgaardieae and Eleocharideae. Clades 11-14 are often referred to as the Fuireneae s.l. grade, with Clade 11 representing tribe Bolboschoeneae, Clade 12 tribe Fuireneae s.s., Clade 13 tribe Schoenoplecteae, and Clade 14 tribe Pseudoschoeneae. Each tribe is well supported as a monophyletic group and is placed with high support in the backbone of the family. Clades 15 and 16 represent the two main clades of tribe Cypereae, that is, the Ficinia Clade and the Cyperus Clade.

Family Cyperaceae
In our results, Cyperaceae are confirmed as a monophyletic family within the monocot order Poales, sister to Juncaceae (Figs. 2, 3). The relationships inferred within Cyperaceae are mostly congruent with those of previous analyses (Simpson et al., 2007;Muasya et al., 2009a;Escudero & Hipp, 2013;Hinchliff & Roalson, 2013;Jung & Choi, 2013;Spalink et al., 2016b;Semmouri et al., 2019), with some exceptions. Table S4 provides an overview of the main published classifications of the Cyperaceae and the classification proposed in this study, clearly indicating which changes occurred as more data became available. Table 1 provides an overview of the proposed classification.
Most previous molecular studies, which were largely based on chloroplast sequence data, recognized two subfamilies in Cyperaceae (Muasya et al., 2009a;Escudero & Hipp, 2013;Hinchliff & Roalson, 2013;Jung & Choi, 2013;Spalink et al., 2016b;Semmouri et al., 2019). The targeted sequencing data (Fig. 2) confirm the established topology with subfamily Mapanioideae sister to subfamily Cyperoideae. The main morphological arguments to recognize two subfamilies in Cyperaceae relate to the differences in the morphology of the basic units of the inflorescence. In Cyperoideae, inflorescences are composed of one to many spikelets, each consisting of a rachilla bearing few to many glumes that may or may not subtend a flower (e.g., Goetghebeur, 1998). In contrast, the inflorescence units of Mapanioideae are frequently referred to as spicoids (e.g., Kukkonen, 1984;Simpson, 1992;Simpson et al., 2003;Beentje, 2016 and the preferred term here) or pseudospikelets (e.g., Eiten, 1976;Dai et al., 2010), and comprise 1-13(-100) scales. The homology of these units is still unclear. Many authors consider them to be a much-reduced spikelet (the basic inflorescence unit found in most other Cyperaceae; Dahlgren et al., 1985;Simpson, 1992;Vrijdaghs et al., 2006;Prychid & Bruhl, 2013), whereas others view them as a flower in which the regular trimerous structure of the cyperaceous flower has been disturbed (Goetghebeur, 1986(Goetghebeur, , 1998. Most previous studies retrieved tribe Trilepideae as sister to all remaining Cyperoideae (Simpson et al., 2007;Muasya et al., 2009a;Escudero & Hipp, 2013;Hinchliff & Roalson, 2013;Jung & Choi, 2013;Spalink et al., 2016b;Semmouri et al., 2019). This relationship is here confirmed (Figs. 2,3). Otherwise, little congruence can be found concerning the relationships between the early diverging lineages of Cyperoideae in the literature. In Simpson et al. (2007;using only the plastid marker rbcL), a Bisboeckelereae-Sclerieae Clade is the second branching clade in Cyperoideae before Cladium P.Browne. In Muasya et al. (2009a; using the plastid markers rbcL and trnL-F), Cladium branches off before Bisboeckelereae and Sclerieae that form a polytomy with the remainder of Cyperoideae; in Jung & Choi (2013; using the plastid markers rbcL and trnL-F plus one nuclear marker ITS), both clades form a polytomy with the remainder of the Cyperoideae (cf. the Maximum Parsimony results of Semmouri et al., 2019; using five plastid markers and two nuclear markers ETS and ITS). However, in Hinchliff & Roalson (2013; supermatrix approach with scaffold based on two chloroplast markers ndhF and rbcL), Cladium is strongly supported as sister to Schoeneae, with the Bisboeckelereae-Sclerieae Clade retrieved as a separate lineage. In the Maximum Likelihood and Bayesian Inference results of Semmouri et al. (2019), and in the results of Spalink et al. (2016b; using the plastid markers matK, ndhF, rbcL, and trnL-F), Cladieae branches off right after Trilepidae, followed by a Bisboeckelereae-Sclerieae Clade sister to the rest of Cyperoideae. In our results ( Fig. 2A), after subfamily Mapanioideae (Clade 1), tribe Trilepideae (Clade 2), tribe Cladieae (Clade 3), the Bisboeckelereae-Sclerieae Clade (Clade 4), tribe Carpheae (Clade 5), and tribe Cryptangieae (Clade 6), followed by the rest of subfamily Cyperoideae, branch off subsequently.
The topology of the family (Figs. 2, 3) raises interesting evolutionary and developmental questions in that Clades 1-6 are largely composed of tribes that are characterized by species having unisexual flowers (with the exception of Cladieae and Carpheae), in contrast to the remaining Cyperaceae tribes that are largely characterized by having bisexual flowers (with the exception of the tribes Khaosokieae and Cariceae in the Scirpo-Caricoid Clade). These clades are also characterized by having embryo types that were placed close to the ancestral Juncus-type embryo in the semophylesis (evolutionary sequence) of the embryo types according to Goetghebeur (1986; see also fig. 3 of Semmouri et al., 2019). Goetghebeur (1998) placed most of these tribes in two subfamilies: Chrysitricheae and Hypolytreae in subfamily Mapanioideae, and Trilepideae, Bisboeckelereae, Sclerieae, and Cryptangieae in subfamily Sclerioideae. Tribes Carpheae and Cladieae were only recently recognized (Semmouri et al., 2019) and were previously treated as part of tribe Schoeneae (e.g., Goetghebeur, 1998). Simpson et al. (2007) showed that subfamily Sclerioideae was not monophyletic and suggested maintaining only two subfamilies in Cyperaceae, that is, Mapanoideae and Cyperoideae.

Subfamily Mapanioideae
On the basis of pollen data, Simpson et al. (2003) supported the recognition of the two tribes in subfamily Mapanioideae, that is, Chrysitricheae (Fig. 5D) and Hypolytreae (Fig. 4B). Most Cyperaceae, including tribe Chrysitricheae, have thin-walled, pyriform, pseudomonad pollen, whereas Hypolytreae (forest or forest-margin dwellers, where wind pollination is less or not effective) have thick-walled, spheroidal, "Mapania-type" pollen that is coated with lipids, supporting earlier studies, especially Lorougnon (1973), which suggest that Hypolytreae use animal vectors for pollination (Simpson et al., 2003;Nagels et al., 2009). Simpson et al. (2003) indicated that younger, developmental stages of "Mapania-type" pollen were not available for their study and that pollen ontogeny could not be examined. However, Coan et al. (2010) showed that several Hypolytrum species have pseudomonads, suggesting that "Mapaniatype" pollen in general is pseudomonad.
In most molecular studies, the circumscription of Chrysitricheae and Hypolytreae and the relationships between and within these tribes are not well resolved or have been conflicting (Simpson et al., 2003(Simpson et al., , 2007Muasya et al., 2009a;Hinchliff & Roalson, 2013;Semmouri et al., 2019). A case in point is the inconsistent position of Diplasia karatifolia Rich. Simpson et al. (2003) and Muasya et al. (2009a) placed Diplasia Pers. within tribe Hypolytreae, whereas Semmouri et al. (2019) placed it in a nested position within tribe Chrysitricheae, and its relationship was unresolved in Hinchcliff & Roalson (2013). In Spalink et al. (2016b), Diplasia was positioned as a separate lineage sister to the often retrieved Hypolytreae-Chrysitricheae Clade. Our results recover the tribes Chrysitricheae and Hypolytreae as monophyletic, with Diplasia as sister to the rest of tribe Chrysitricheae with moderate support (LPP = 0.84; Fig. 2A).
A recent molecular phylogenetic study showed that the formerly recognized monotypic genus Principina Uittien in nested within Hypolytrum (A. Mesterházy et al., unpublished data). In our results, the sample of Principina is retrieved as sister to the single included accession of Hypolytrum, confirming a close relationship. In Section 5, we follow A. Mesterházy et al. (unpublished data) and relegate Principina to synonymy.

Tribe Cladieae
Tribe Cladieae (Fig. 5B) is monogeneric including only the cosmopolitan genus Cladium. Uncertainty remains concerning the relationship between Cladium and the monotypic genus Rhynchocladium from the Guiana Shield in Guyana and Venezuela. Despite several attempts, Rhynchocladium has never been successfully included in a molecular study.

Tribe Carpheae
The position of the clade including the genera Carpha Banks & Sol. ex R.Br. and Trianoptiles Fenzl ex Endl. is variable in the literature. In some studies, Carpha is positioned within Schoeneae s.l. clade (Zhang et al., 2004(Zhang et al., , 2007Verboom, 2006;Muasya et al., 2009a;Hinchliff & Roalson, 2013), whereas here and in other studies, Carpha (+ Trianoptiles) is placed outside Schoeneae (Simpson et al., 2007;Jung & Choi, 2013;Viljoen et al., 2013;Larridon et al., 2018a;Semmouri et al., 2019). This clade is also set apart by its unique combination of embryo morphology characters, having an embryo that is more or less rhomboid to top-shaped with a tapered scutellum, with a well-differentiated root cap in a lateral position separated from the coleoptile by a notch. This led Semmouri et al. (2019) to erect a new tribe to accommodate the genera Carpha and Trianoptiles. We retrieve tribe Carpheae (Fig. 4C) as a separate speciespoor clade (Figs. 2, 3).

Tribe Cryptangieae
The relationship between tribe Cryptangieae (Fig. 5G) and the other Cyperaceae tribes varied in different studies (cf. Muasya et al., 2009a;Escudero & Hipp, 2013;Hinchliff & Roalson, 2013;Jung & Choi, 2013;Spalink et al., 2016b;Semmouri et al., 2019). In earlier studies, the diversity of the tribe was represented by only two to three species. On the basis of sequence data of an increased sampling (18 spp.), Costa et al. (2018b) recovered Cryptangieae as monophyletic, supporting its recognition as a tribe in combination with its morphological distinctiveness. However, it remained unclear if the tribe is more closely related to tribe Schoeneae, tribe Carpheae, or to a Carpheae-Schoeneae Clade. Also, it did not include sequence data of the genus Koyamaea. Our targeted sequencing results show that a clade of Koyamaea plus Cryptangieae sensu Costa et al. (2018b) branches after tribe Carpheae and before tribe Schoeneae in the coalescent ASTRAL analysis (Figs. 2, 3); however, in the concatenated IQ-TREE analysis, the positions of Cryptangieae and Schoeneae are inverted (Fig. S2).
Koyamaea neblinensis W.W.Thomas & G.Davidse was described as a new genus and species to science by Thomas & Davidse (1989). Due to its bisexual spikelets, each bearing one pistillate flower and many staminate flowers, presence of well-developed perianth bristles in both kinds of flowers, spirally arranged glumes, and regular nutlet without a cupule, Koyamaea was classified as a new genus of the then recognized subfamily Sclerioideae (Goetghebeur, 1998). As the authors believed that their new genus was not closely related to any other genus, they placed it in its own tribe Koyamaeae. Sanger sequence data could not be obtained for this extremely rare species to test its placement in the family and the value of erecting a separate monotypic tribe for it. High-throughput sequencing techniques are better at dealing with fragmented DNA obtained from historical herbarium specimens (e.g., Buerki & Baker, 2015;Hart et al., 2016;Bakker, 2017;Zeng et al., 2018;Brewer et al., 2019). Our targeted sequencing results are the first to place the genus Koyamaea in the Cyperaceae Tree of Life. It is here inferred as   . Genera including species using the C sister to the genera of Cryptangieae ( Fig. 2A). As there are other arguments linking Koyamaea to Cryptangieae, that is, morphological (flowers with spiral glumes, bearing perianth and lacking cupule, beaked fruits) and anatomical (thickened pericarp) shared features, we opt to include Koyamaea in Cryptangieae. Generic delimitation in Cryptangieae has fluctuated over the years, either with the number of genera considered in the strict sense, including just one or few species, or lumped into a broader Lagenocarpus Nees. Recent molecular studies highlighted the need of an updated generic circumscription (Costa et al. 2018b(Costa et al. , 2021a. In the new interpretation, Cephalocarpus Nees includes the species formerly placed in Everardia Ridley and now encompasses the 20 species of Cryptangieae with an elongate caudex and lateral inflorescences (Costa et al., 2021a(Costa et al., , 2021b. In our results, the monophyly of the newly enlarged genus Cephalocarpus is supported by the concatenated IQ-TREE analysis (Fig. S2), but not by the coalescent ASTRAL analysis where Cephalocarpus angustus (N.E.Brown) S.M.Costa (syn. Everardia angusta N.E.Brown) and Cephalocarpus montanus (Ridl.) S.M.Costa (syn. Everardia montana Ridl.) are not retrieved in a single clade ( Fig. 2A). It should be noted that both species were formerly placed in Everadia; we did not sequence the type species of Cephalocarpus (Cephalocarpus dracaenula Nees). Lagenocarpus (sensu Koyama 2005) species have been split in five genera (Costa et al., 2021a): three of them (Cryptangium Schrader ex Nees, Dydimiandrum Gilly, and Exochogyne C.B.Clarke) with 1-2 species and the others with 10 (Krenakia S.M. Costa et al., 2021a) and 15 species (Lagenocarpus). The genera are distinguished mostly by leaf, inflorescence, and fruit characters (Costa et al., 2021a). The results of Costa et al. (2021a) place Krenakia as sister to a clade encompassing three subclades: (i) Didymiandrum + Exochogyne; (ii) Cryptangium sister to Cephalocarpus; and (iii) Lagenocarpus s.s. Our targeted sequencing results show Koyamaea sister to a clade encompassing the Cryptangieae sensu Costa et al. (2018bCosta et al. ( , 2021a.
Relationships within tribe Schoeneae have not been entirely resolved; however, progress has been made in our understanding of its evolution. Morphologically, tribe Schoeneae is a highly variable group. Previous molecular analyses of the group recovered six main clades: the Caustis Clade, Gahnia Clade, Lepidosperma Clade, Oreobolus Clade, Schoenus Clade, and Tricostularia Clade (Viljoen et al., 2013;Larridon et al., 2018a). In the more deeply sampled phylogenetic study of Semmouri et al. (2019), two additional clades became visible, a clade including the genera Reedia F.Muell. and Gymnoschoenus Nees and a separate lineage of Schoenus paludosus (R.Br.) Roem. & Schult. Our targeted sequencing results confirm the presence of eight main clades in Schoeneae (Fig. 2B). To facilitate the morphological characterization of the main clades in this morphologically diverse tribe, they are recognized as subtribes in Section 5.
Anthelepis Clade: In the BI and ML results of Semmouri et al. (2019), Schoenus paludosus formed a polytomy with the Gahnia Clade and the Oreobolus Clade, revealing its isolated position from other Schoenus species. In fact, Schoenus paludosus also differs morphologically from the true Schoenus species in having one or sometimes two lower male flowers and an upper bisexual flower at each spikelet, besides a non-zigzag rachilla (as opposed to the usual states for the genus of bisexual flowers and upper internodes of the rachilla elongated and prominently zigzag; Wilson, 1993). Schoenus paludosus was recently placed in a new genus Anthelepis R.L.Barrett, K.L.Wilson & J.J.Bruhl together with the species previously named Schoenus guillauminii Kük. and Tricostularia undulata (Thwaites) J.Kern . The Anthelepis Clade is here strongly supported as sister to the remainder of Schoeneae (Fig. 2B).
Caustis Clade: This clade includes the genera Caustis R.Br. and Evandra R.Br. (Fig. 2B). The unexpected placement of a lineage previously included in Tetraria, that is, Tetraria borneensis J.Kern, in the Caustis Clade (Larridon et al., 2018a) is being explored further (Barrett RL & Larridon I, unpublished data). This species could not be sequenced for this study.
Gymnoschoenus-Reedia Clade: This is a small clade of just three morphologically distinctive species placed in the genera Gymnoschoenus and Reedia (Fig. 2B), each with restricted distributions in southern Australia whose affinities have been much debated.
Lepidosperma Clade: Another lineage previously included in Tetraria, that is, the Tetraria capillaris (F.Muell.) J.M.Black species complex, native to Australia and New Zealand, was found to be part of the Lepidosperma Clade (Viljoen et al., 2013;Larridon et al., 2018a;Barrett et al., 2019). A recent taxonomic revision of the Tetraria capillaris species complex resulted in the publication of a new genus Netrostylis (Barrett et al., 2021a). The Lepidosperma Clade appears to have originated in Australia (Viljoen et al., 2013). Previous studies (e.g., Verboom, 2006;Muasya et al., 2009a;Hinchliff & Roalson, 2013;Jung & Choi, 2013;Viljoen et al., 2013;Larridon et al., 2018a;Semmouri et al., 2019) indicated that the clade includes (i) the species-rich genus Lepidosperma occurring from China to Australasia; (ii) Machaerina Vahl (including Baumea Gaudich.), which is widespread from Australia to the Americas; and (iii) the monotypic genus Neesenbeckia Levyns endemic from the South African Cape Floristic Region. As retrieved in, for example, Viljoen et al. (2013) and Larridon et al. (2018a), our results confirm that within the Lepidosperma Clade, the genus Machaerina is sister to a clade with two subclades: (i) Lepidosperma and (ii) Netrostylis sister to Neesenbeckia (Fig. 2B). The latter sister relationship between Netrostylis and Neesenbeckia suggests an unusual dispersal event from Australia to southern Africa. Although some taxonomic issues remain in Machaerina (Barrett RL, Wilson KL & Bruhl JJ, unpublished data), more work is required in Lepido-sperma, which has c. 200 undescribed species in southern Australia (Barrett & Wilson, 2012, 2013. Oreobolus Clade: Larridon et al. (2018a) found that Costularia s.l. was composed of four distinct evolutionary lineages with two lineages being part of the Oreobolus Clade: (i) a much-reduced genus Costularia (Larridon et al., 2019a) and (ii) a small New Caledonian endemic genus Chamaedendron. The circumscription of the other genera in this clade (Fig. 2B), that is, Capeobolus, Cyathocoma Nees, and Oreobolus R.Br., remains unchanged.
Schoenus Clade: As some species of Tetraria and Epischoenus had been shown to be nested within Schoenus (Viljoen et al., 2013), Elliott & Muasya (2017) transferred these species to Schoenus. The broader circumscription of Schoenus is supported by our targeted sequencing results (Fig. 2B), and only a single morphologically variable and geographically widespread genus is recognized in this clade.
The Tricostularia Clade also includes the Australian species Tetraria octandra (Nees) Kük., which Larridon et al. (2018a) suggested should be accepted as T. octandra (Nees) C.B.Clarke, as it is not related to Tetraria. The taxonomic changes made to Costularia s.l. and Tetraria by Larridon et al. (2017Larridon et al. ( , 2018a are supported by our targeted sequencing results (Fig. 2B). More recent research has shown that three Australian species until recently placed in Tetraria, T. australiensis C.B.Clarke, T. microcarpa S.T.Blake, and T. octandra are closely related to Morelotia and Xyroschoenus (Barrett et al., 2021b). Therefore, the decision has been taken to expand the circumscription of Morelotia, by including the three Australian Tetraria species and a Pacific Island species (originally described as Machaerina involuta H.St.John) (Barrett et al., 2021b).
Within the Tricostularia Clade, the genus Tricostularia itself has been reduced in morphological circumscription with the removal of species now placed in Anthelepis . In parallel, the number of species was enlarged with the addition of Lepidosperma aphylla R.Br. and L. exsul C.B.Clarke (Barrett & Wilson, 2012) and ongoing taxonomic revision of species boundaries in southern Western Australia (Barrett, 2012;Barrett et al., 2021b).

Scirpo-Caricoid Clade
A Scirpo-Caricoid Clade (SCC Clade), referred to as the Scirpeae-Dulichieae-Cariceae Clade (SDC Clade) in some previous studies, was recognized in all recent molecular phylogenetic studies of Cyperaceae as a sister group to the Abildgaardieae-Eleocharideae-Fuireneae-Cypereae Clade (FAEC Clade) (Muasya et al., 2009a;Hinchliff & Roalson, 2013;Jung & Choi, 2013;Spalink et al., 2016b;Léveillé-Bourret et al., 2014, 2018cSemmouri et al., 2019). The Scirpo-Caricoid Clade contains 41% of all recognized Cyperaceae species Govaerts et al., 2020), comprising a curious assemblage of tribes that illustrates the full breadth of inflorescence and floral diversity of Cyperoideae. This includes bisexual, monoecious, and dioecious species with empty proximal glumes or all glumes fertile, spirally or distichously inserted flowers, sterile or fertile prophylls, as well as setiform, tepaliform, or absent perianth . There are no recognized synapomorphies for this clade, whose only recognizable characteristic is its center of diversity in cold temperate regions of the Northern Hemisphere Martín-Bravo et al., 2019), contrasting with the southern temperate distribution for Schoeneae or mostly tropical diversity of other major Cyperaceae lineages.
The monogeneric tribe Cariceae contains most of the diversity of the clade, with c. 2000 species Villaverde et al., 2020Villaverde et al., , 2021, and is characterized by a highly derived inflorescence morphology formed of perianthless unisexual flowers, with female flowers strictly associated with the production of secondary branches, and pistils contained within or subtended by the first bract of secondary branches (a prophyll called a perigynium or utricle if closed forming a bottle-like structure; Jiménez-Mejías et al., 2016b). Although relationships within this tribe are not highly supported in the present study, they have been already addressed in Villaverde et al. (2020).
The other 13 genera (c. 88 species) of the Scirpo-Caricoid Clade have all been placed at one point in their history within a broadly circumscribed "tribe Scirpeae," which was essentially defined by a lack of derived characters. Unsurprisingly, most recent studies suggested paraphyly of Scirpeae when thus circumscribed (Hinchliff & Roalson, 2013;Jung & Choi, 2013;Spalink et al., 2016b). However, an ancient rapid radiation near the crown of the Scirpo-Caricoid Clade made previous phylogenetic analyses extremely difficult, with different analyses supporting different topologies with consistently low support (Hinchliff & Roalson, 2013;Jung & Choi, 2013;Spalink et al., 2016b;Semmouri et al., 2019).
A series of recent studies combining plastid and nuclear ribosomal markers (Gilmour et al., 2013;Léveillé-Bourret et al. 2014, genomic data (Léveillé-Bourret et al., 2018c;Villaverde et al., 2020Villaverde et al., , 2021, and morphological data  were able to resolve the most recalcitrant backbone branches of the Scirpo-Caricoid phylogeny. Our present results (Fig. 2C) are in agreement with these recent studies and support the taxonomic treatment of the clade as presented in Léveillé-Bourret & Starr (2019).

Tribe
Calliscirpeae. This monogeneric lineage was recently recognized (Léveillé-Bourret & Starr, 2019) based on species formerly placed in Scirpus L. or Eriophorum L., but that differ by having antrorsely barbed perianth bristles and a Carextype embryo (Gilmour et al., 2013). All previous studies have consistently supported the isolated position of this lineage in the Scirpo-Caricoid Clade, but its phylogenetic position as sister to a Scirpeae-Trichophoreae-Sumatroscirpeae-Cariceae Clade has never received strong support (Léveillé-Bourret et al., 2014, 2018cSemmouri et al., 2019). In our results, tribe Calliscirpeae (Fig. 6A) branches after Khaosokieae, sister to the remaining lineages of the Scirpo-Caricoid Clade (Fig. 2C).
4.3.8.4 Tribe Scirpeae. Tribe Scirpeae (Fig. 6C) is a lineage of the Scirpo-Caricoid Clade (e.g., Simpson et al., 2007;Muasya et al., 2009a;Hinchliff & Roalson, 2013;Léveillé-Bourret et al., 2014, 2018cSpalink et al., 2016b;Semmouri et al., 2019). As previously discussed (see Section 4.3.8), tribe Scirpeae sensu Goetghebeur (1998) is not mono-phyletic and consists of three separate lineages. As a result, Scirpeae was recircumscribed by Léveillé-Bourret & Starr (2019) to include only species possessing a (sub-)lateral germ pore in their embryos, corresponding to Schoenustype, Fimbristylis-type, or intermediate embryo types. Under this circumscription, Scirpeae is monophyletic. No visible macromorphological character has been found that can unambiguously diagnose this tribe, which means that identification must be done by means of exclusion. Two major subclades are found within this monophyletic Scirpeae (Fig. 2C): (i) a mostly South American group that has been dubbed "Zameioscirpus Clade" is supported in many studies (Dhooge et al., 2003;Muasya et al., 2009a;Léveillé-Bourret et al., 2015), including Amphiscirpus Oteng-Yeb., Phylloscirpus C.B.Clarke, Rhodoscirpus Léveillé-Bourret, Donadío & J.R. Starr, and Zameioscirpus Dhooge & Goetgh.; and (ii) a mostly circumboreal "Scirpus Clade," well supported in our analyses and consistently found in other studies, with the genus Eriophorum L. forming a wellsupported clade nested within Scirpus, thus making Scirpus paraphyletic (e.g., Gilmour et al., 2013;Hinchliff & Roalson, 2013;Jung & Choi, 2013;Léveillé-Bourret et al., 2014, 2018c. The circumscription of Eriophorum and Scirpus is one of the last adjustments needed to achieve a classification of Cyperaceae where all genera are circumscribed as monophyletic entities. Two options for resolving this issue appear viable: (i) Eriophorum can be merged with Scirpus as proposed by Koyama (1958), or (ii) Eriophorum can be maintained by splitting Scirpus into a series of new genera. Both of these solutions have drawbacks. If Eriophorum is treated within Scirpus, the specific epithets for some wellknown species such as Eriophorum gracile W.D.J.Koch would suddenly be unfamiliar to most in the botanical community (i.e., =Scirpus ardea T.Koyama). However, maintaining Eriophorum would require splitting Scirpus into six to eight genera, each consisting of one to a dozen species. As a taxonomically well-sampled and strongly supported phylogeny for Scirpeae is still lacking, the extent of the taxonomic changes needed to split Scirpus remains unclear. Consequently, a decision on merging or splitting should wait until conclusive phylogenetic data are gathered.
4.3.8.6 Tribe Sumatroscirpeae. The genus Sumatroscirpus Oteng-Yeb. was formerly placed in Dulichieae (Goetghebeur 1998), but unlike other members of this tribe, it possesses tubular fertile prophylls similar to the perigynia of Cariceae. Léveillé-Bourret et al. (2018a) demonstrated that Sumatroscirpus is actually sister to Cariceae and that it corresponds to a morphologically transitional lineage between Cariceae and Scirpeae. This result was confirmed by Semmouri et al. (2019) and is here again confirmed with a completely independent data set (Fig. 2C). We, thus, support its recognition as a monogeneric tribe Sumatroscirpeae (Fig. 6E).
4.3.8.7 Tribe Cariceae. Tribe Cariceae (Fig. 4E) is strongly supported as monophyletic as in previous studies (e.g., Global Carex Group, 2015; Starr et al., 2015;Jiménez-Mejías et al., 2016c;Martín-Bravo et al., 2019;Semmouri et al., 2019;Villaverde et al., 2020). The genus Carex has become monophyletic by the inclusion of the formerly recognized segregate genera Cymophyllus Mack., Kobresia Willd., Schoenoxiphium Nees, and Uncinia Pers. (Global Carex Group, 2015). This taxonomic decision agrees with our results (Fig. 2C). The previously cited genera were the only ones included in the most recent treatments of the tribe (e.g., Kükenthal, 1909;Egorova, 1999;Ball & Reznicek, 2002), whose circumscription remains otherwise unaltered. Although the topology within Carex largely reflects recent studies focused on Carex (e.g., Villaverde et al., 2020), the placement of some species such as Carex ncinate L.f.is not well supported, and for a deeper systematic analysis of the genus Carex, we refer to those studies. Data of Carex species used in this study were generated using three different targeted sequencing probe kits: Angiosperms353 (Johnson et al., 2019), Cyperaceae-specific , and Angiosperms I kit for Anchored Phylogenomics (Léveillé-Bourret et al., 2018c). Lower recovery of the Angiosperms353 genes from data generated with the other probe kits may have contributed to lower resolution within Carex in this study.
In the Ficinia Clade, Erioscirpus Palla is the first genus to diverge, before Scirpoides Ség (Fig. 2F). Erioscirpus was previously thought to be more allied to Scirpus and Eriophorum, but molecular studies (Yano et al., 2012;García-Madrid et al., 2015; (2021) sampled 78% of the Ficinia Clade for a nuclear data set including ETS and ITS and in a chloroplast data set including the genes matK, ndhF, rbcL, and rps16, the trnL intron, and trnL-F spacer with the aim to recircumscribe Ficinia and Isolepis as monophyletic genera. On the basis of the topology obtained with their nuclear data set,  (i) broadened the circumscription of Ficinia to include the annual Isolepis species characterized by cartilaginous glumes and including all Isolepis species retrieved outside the core Isolepis clade, and (ii) narrowed the circumscription of Isolepis to encompass only those species retrieved as part of the core Isolepis clade. Two southern African genera that were recently described, Afroscirpoides García-Madr. & Muasya (García-Madrid et al., 2015) and Dracoscirpoides Muasya (Muasya et al., 2012), form a clade in this study (Fig. 2F). Species segregated into Dracoscirpoides and Hellmuthia are atypical for tribe Cypereae, all bearing perianth, and were originally described as part of Scirpus. Hellmuthia is strongly supported as sister to a clade including Ficinia and Isolepis (e.g., Simpson et al., 2007;Hinchliff & Roalson, 2013;Jung & Choi, 2013;Semmouri et al., 2019). The scale-like perianth of Hellmuthia, interpreted to be analogous to similar structures in Chrysitricheae (subfamily Mapanioideae, here as tribe Hypolytreae s.l.) by Haines & Lye (1976), is now thought to be ontogenetically similar to perianth in other Cyperaceae (Vrijdaghs et al., 2006;Muasya et al., 2009b).

Distribution Cosmopolitan
Key to the subfamilies of Cyperaceae 1a. Basic inflorescence unit (=spicoid) usually comprising 2, strongly keeled and opposite basal bracts (rarely 1 and unkeeled), with a further (0-)1-13(-100) scale-like bracts, the bracts subtending 1 stamen, the whole unit with a terminal pistil ………………………………...Mapanioideae 1b. Basic inflorescence unit (=spikelet) consisting of a rachilla bearing few to many glumes that may or may not subtend a flower (but see Hellmuthia) ……………………Cyperoideae The spicoids each have 4-100 (or more) floral bracts and the lowest two bracts are opposite and keeled or, in Chrysitrix, spirally arranged. The floral bracts may or may not subtend a single stamen and each spicoid is terminated by a single pistil. The spicoids are subtended and usually hidden by glume-like spicoid bracts and these units are aggregated into spikes that are analogous to spikelets in Cyperoideae genera. The pollen is pyriform, with the exception of Diplasia in which it is spheroidal. The style is 2-3-fid and the fruits are 2-3-sided or terete with a hard exocarp.

Subfamily
Accepted . For descriptions and notes on the genera, see Goetghebeur (1998).
Distribution Chrysitricheae mainly have a southern hemisphere (Gondwanan) distribution with the exception of Diplasia, which is present in Trinidad and Central and tropical South America. Diagnosis Hypolytreae are rather delicate to very robust (up to 5 m tall), rhizomatous or stoloniferous perennials. The leaves are linear or sometimes with an expanded, linearoblong to broadly oblong blade and pseudopetiole between the blade and sheath, or rarely reduced to bladeless sheaths. The inflorescence bracts are leaf-like to glume-like. The inflorescences are paniculate, capitate, or reduced to a single spike, rarely anthelate. The basic reproductive units comprise spicoids. The spicoids each have 4-15 floral bracts and the lowest two bracts are opposite and keeled. The lowest two bracts usually subtend a single stamen, whereas the remaining floral bracts may or may not subtend a single stamen and each spicoid is terminated by a single pistil. The spicoids are subtended and usually hidden by glume-like spicoid bracts, and these units are aggregated into spikes that are analogous to spikelets in non-mapiniid genera. The pollen is spheroidal. The style is 2-3-fid and the fruits are 2-3-sided or terete with a hard, succulent or occasionally berry-like exocarp.
Distribution Hypolytreae have a pantropical distribution and occur primarily in forest or forest margins, rarely in savannah.
Distribution West and West Central Africa (Afrotrilepis, Microdracoides), Tropical and southern Africa and Madagascar (Coleochloa), northern South America to Brazil (Trilepis). Occurring in tropical areas mostly on inselbergs, growing on shallow soils; one species epiphytic in submontane tropical rain forest.
Distribution Cladium is subcosmopolitan and occurs in swamps and marshes, often in brackish or calcareous habitats.
Distribution Tropics and subtropics to North America. Schoeneae. Anthers are typically conspicuously greenishyellow in this tribe, whereas they are yellow to red-colored in the morphologically similar Schoeneae. Many species of tribe Schoeneae mainly occur in austral temperate dryland habitats that are only seasonally damp (e.g., woodland and heathland), whereas Carpheae occur typically in wetlands and damp areas.
Distribution Whereas the annual Trianoptiles species are endemic to the wetlands of South Africa (SW Cape), perennial Carpha occurs in swamps and along stream sides in the southern and central African mountains, Madagascar, Mascarenes, New Guinea, southern Japan, southeastern Australia, New Zealand, and Chile.

Tribe Cryptangieae
Cryptangieae Benth. in J. Linn. Soc. London, Bot. 18: 366. (1881). Type Cryptangium Schrad. ex Nees Diagnosis Cryptangieae are mostly characterized by unisexual spikelets (except for Koyamaea, with a more basal single female flower and many male flowers above), spirally arranged glumes (distichously arranged in Exochogyne), fruit usually triangular or trigonous in cross-section with three fimbriate perianth scales opposite the flat sides of the nutlet (biconvex and without hypogynous scales in Exochogyne), and Juncus-or Carex-type embryos, although few species have been studied (Goetghebeur, 1998;Semmouri et al., 2019). It seems that all species present a red-pinkish style and stigma, except for some populations in the "campos rupestres" of Chapada Diamantina localities (Bahia, Brazil). Distribution Tropical America, in forested (Koyamaea, Didymiandrum) and open vegetation, mostly at sandy nutrient-poor soils and/or rocky places, from seashores and sandy temporarily wet plains (such as the Amazonian "campinaranas") to high altitudes (such as the "tepuis" and "campos rupestres"). Also, in some mountains associated with the Andes, but with older and nutrientpoor soils, such as the Cordillera del Condor (Ecuador, Peru).
Includes 8 subtribes, 25 genera. Distribution The tribe has a mostly southern hemisphere distribution, in temperate and subtropical areas, with just a small number of taxa in the northern hemisphere. Diagnosis Tufted, sometimes rhizomatous, perennial or annual graminoids; leaves mostly basal; culms semi-terete; leaves well developed; ligulate; leaf margins scaberulous or glabrous, flat to channeled; inflorescence terminal, paniculate or subracemose; glumes obscurely distichous, usually deciduous; rhachilla non-flexuous, straight; flowers subtended by upper glumes; lower flower(s) functionally male, upper bisexual; upper glumes longer than lower; spikelets ranging from few to many grouped together in spikelet bundles; 3 stamens, stigma 3-fid; nutlets ranging in shape from narrow-ellipsoid to obovoid; perianth bristles (3)6, shorter or longer than the nutlet.
Distribution From Sri Lanka to Hainan, New Caledonia to Australia. Whereas A. undulatus is widespread, the other three species are localized.
Key to the genera of Lepidospermatinae 1a. Two middle glumes larger than others; perianth of 6 bristles equal to or longer than the nutlet, persistent on the rachilla; stigma 6-fid ……………………Neesenbeckia 1b. Glumes of increasing length from the base, upper glumes the largest; perianth of thickened scales persistent at base of nutlet or bristles 0-5, shorter than the nutlet; stigma (2-)3-fid…. flowers subtended by upper glumes; lower flower(s) functionally male (rarely bisexual or absent), upper bisexual (rarely functionally male or female); upper glumes longer than lower; spikelets ranging from few to many in spikelet bundles; usually 3 stamens (6 in Cyathocoma), stigma 3-fid; nutlets ranging in shape from ellipsoid to ovoid or obloid; perianth bristles 6 (sometimes not all developing), shorter or longer than the nutlet.
Distribution Primarily Australasia and South Africa, with a few species in Europe, the Americas, and Caribbean Islands (Kern, 1974;Viljoen et al., 2013).
Embryo top shaped in frontal view, root cap developed in a (sub)basal position, and first leaf primordium developed in a lateral position (Carex-type embryo).

Tribe
Distribution Tropical to subarctic northern hemisphere south to southern South America, and Australia through Malesia.
Key to the genera of Scirpeae 1a. Cauline leaves present, node of the distal leaf visible above the sheath of the leaf below ……………………2 1b. Cauline leaves absent, leaves all basal with node of the distal leaf hidden in the sheath of the leaf below ……4 2a. Inflorescence, a white to red cottony mass at maturity due to the exserted flat and silky perianth bristles >10  Type Trichophorum Pers. Diagnosis Flower bisexual or rarely functionally unisexual with remnant of opposite sex, ligule glabrous, prophyll sterile, basal (0-)1-9 glumes of spikelet sterile, lowest glume often with conspicuously longer awn than following glumes, flowers 1-10+ per spikelet, perianth setiform, squamiform, or absent, embryo with a basal root cap and lateral plumule (Carex-type).
Distribution Western China to West Sumatra.

Tribe Eleocharideae
Type Eleocharis R.Br. Diagnosis (Goetghebeur, 1998): Eleocharideae is characterized by its reduced vegetative morphology, leaves reduced to a sheath (no blade), unispiculate inflorescence, Eleocharis-type embryo, and a helio-and helophilous ecology. Characters shared with its sister tribe Abildgaardieae include a differentiated and thickened style base, and moniliform stigmatic hairs. Characters in common with many Fuireneae include a bristle-like perianth, and an embryo with a broadened cotyledon.
Type Abildgaardia Vahl Diagnosis (Goetghebeur, 1998): Abildgaardieae is characterized by its clearly differentiated style base, which is often thickened and persistent on the nutlet, but it is deciduous in a number of species. Glumes of the spikelet are typically spirally arranged, but distichous glumes are present in some species. Moniliform stigmatic hairs present. Embryos are of the related Abildgaardia-, Bulbostylis-, Carex-, Fimbristylis-, Schoenus-and Tylocarya-type. (Semmouri et al., 2019).
Type Bolboschoenus (Asch.) Palla Diagnosis Differs from all other Cyperaceae tribes by this unique combination of characters: Perennials with long rhizomes often forming hard ovoid tubers at tips. Culms many-noded, 3-sided, thickened at base. Leaves well developed, basal and cauline, eligulate with blade often reduced in lower leaves. Inflorescence terminal (in reduced inflorescences, bract may be erect, but clearly leaf-like), a (compound) corymb-like anthela or capitate with 1 to many spikelets. Inflorescence bracts leaf-like, patent, lowermost often suberect. Spikelets with many spirally arranged, deciduous glumes, each subtending a flower. Glumes puberulent, the apex entire to emarginate or deeply 2-fid, awned or mucronate. Flowers bisexual, perianth present, formed by 3-6 parts, shorter to longer than the nutlet, bristle-like, deciduous with fruit. Stamens 3. Styles 2 or 3. Style base persistent, barely thickened, if at all. Nutlets obovate, dorsiventrally lenticular, or trigonous. Pericarp with the three highly differentiated layers, exocarp cells often enlarged and hollow, surface smooth, epidermal cells roughly isodiametric. Embryo fungiform with three primordial leaves and a notch below the root cap (Bolboschoenus-type).
Accepted genus Bolboschoenus (Asch.) Palla (15 spp. Diagnosis Differs from all other Cyperaceae tribes by this unique combination of characters: Annuals or rhizomatous perennials. Culms many-noded, rarely scapose, 3-5-sided, sometimes thickened at base. Leaves usually well developed, basal and cauline, ligule tubular, membranous, with blade often reduced in lower leaves (rarely all leaf blades reduced).
Inflorescence terminal (in reduced inflorescences, bract may be erect, but clearly leaf-like), paniculate to capitate with few to many spikelets. Inflorescence bracts leaf-like, usually sheathing, lowermost bract sometimes erect. Spikelets with many spirally or rarely pentastichously arranged, deciduous glumes, each subtending a flower. Glumes often pubescent, the apex entire and mucronate to awned. Flowers bisexual, perianth present, as long or shorter than nutlet, formed by 3 parts, or when 6 in 2 whorls, the inner parts scale-like, the outer parts bristle-like, rarely all parts reduced or absent or only 1 scale developed, deciduous with the fruit. Stamens 1 to 3. Styles 3. Style base persistent, barely thickened, if at all. Nutlets obovate, triquetrous to trigonous, frequently stipitate, smooth or variously ornamented. Embryo turbinate to weakly fungiform with a horizontally broadened scutellum, first leaf primordium not strongly outgrown, the second leaf primordium either absent or poorly developed (Fuirena-type).

Tribe Schoenoplecteae
Type Schoenoplectus (Rchb.) Palla Diagnosis Differs from all other Cyperaceae tribes by this unique combination of characters: Perennials with long rhizomes sometimes ending in tubers at tips. Culms nodeless, scapose, trigonous to terete, thickened at base. Leaves usually reduced to a sheath, sometimes developing a ligulate blade, but rarely well developed. Inflorescence pseudolateral, rarely clearly terminal, corymb-like anthela or capitate with (1-)few to many spikelets. Inflorescence bracts often large, erect, stem-like, rarely leaf-like, and patent to reflexed (Actinoscirpus). Spikelets with many spirally arranged, deciduous glumes, each subtending a flower. Glumes puberulent to glabrous, the margins often ciliate or laciniate distally, apex entire to emarginate or deeply 2-fid, awned or mucronate. Flowers bisexual. Perianth present, formed by (-5)6 parts, smooth to retorsely scabrid, bristle-like or sometimes plumose, longer or shorter than nutlet, deciduous with fruit. Stamens 2 or 3. Styles 2 to 3. Style base not thickened, persistent. Nutlets smooth, obovate, trigonous, or dorsiventrally lenticular, yellow to dark brown when mature. Fruit epidermal cells isodiametric to narrowly oblong. Embryo fungiform, scutellum turbinate to rhomboid in shape, root cap lateral, first (well developed) and second embryonic leaves basal (Schoenoplectus-type I).
Distribution Tropical and subtropical Asia from India east to China and south to Northeast Australia (Actinoscirpus), predominantly temperate (Schoenoplectus).

Supplementary Material
The following supplementary material is available online for this article at http://onlinelibrary.wiley.com/doi/10.1111/jse.12757/ suppinfo: Table S1. Voucher information for accessions included in the targeted sequencing study. Table S2. Recovery statistics for the genes targeted by the Angiosperms353 probes for the accessions included in this study. Table S3. AMAS summary statistics generated for the exons data set. Invariable columns were removed. Table S4. Overview of the main published classifications of the family Cyperaceae and the classification proposed in this study, clearly indicating which changes occurred as more data became available. See legend on the right of the table and the included notes. Fig. S1. Heatmap of recovery of the Angiosperms353 probes for the accessions included in this study. Fig. S2. Phylogenetic reconstruction of the relationships in Cyperaceae based on analysis of the exons data set. Concatenated IQ-TREE analysis. Values above branches represent UltraFast Bootstrap support; missing values indicate maximum support. Values below the branches represent gCF/sCF values. Bars on the right indicate subfamilial and tribal classification.