|Home | About | Journals | Submit | Contact Us | Français|
Sisyrinchium (Iridaceae: Iridoideae: Sisyrinchieae) is one of the largest, most widespread and most taxonomically complex genera in Iridaceae, with all species except one native to the American continent. Phylogenetic relationships within the genus were investigated and the evolution of oil-producing structures related to specialized oil-bee pollination examined.
Phylogenetic analyses based on eight molecular markers obtained from 101 Sisyrinchium accessions representing 85 species were conducted in the first extensive phylogenetic analysis of the genus. Total evidence analyses confirmed the monophyly of the genus and retrieved nine major clades weakly connected to the subdivisions previously recognized. The resulting phylogenetic hypothesis was used to reconstruct biogeographical patterns, and to trace the evolutionary origin of glandular trichomes present in the flowers of several species.
Glandular trichomes evolved three times independently in the genus. In two cases, these glandular trichomes are oil-secreting, suggesting that the corresponding flowers might be pollinated by oil-bees. Biogeographical patterns indicate expansions from Central America and the northern Andes to the subandean ranges between Chile and Argentina and to the extended area of the Paraná river basin. The distribution of oil-flower species across the phylogenetic trees suggests that oil-producing trichomes may have played a key role in the diversification of the genus, a hypothesis that requires future testing.
During their evolution, flowering plants have developed a wide variety of strategies to attract and reward pollinators. Plant–pollinator interactions are key components of the dynamics of most terrestrial ecosystems and, in a world where biodiversity is jeopardized by anthropogenic changes, analysing the evolutionary history of species and understanding the mechanisms involved in their evolution, such as plant–pollinator interactions, is crucial (Steffan Dewenter et al., 2006; Waser, 2006). Furthermore, since species interactions are considered to play a central role in many speciation events, studying the evolutionary history of traits closely linked to uncommon interactions could contribute to improve our knowledge of the mechanisms involved (Rieseberg and Willis, 2007).
Insects represent the largest group of animals visiting flowers to collect resources. Most visit flowers to collect pollen and nectar, but some insects seek other resources. Relationships between oil-secreting flowers and oil-collecting bees constitute an example of a functional specialization and uncommon interaction between plants and pollinators (Minckley and Roulston, 2006). Flowers offering an oil resource are found in 11 families according to the APG system (APG III, 2009), distributed across the angiosperms among unrelated orders (Buchmann, 1987; Rasmussen and Olesen, 2000; Steiner and Whitehead, 2002; Neff and Simpson, 2005; Renner and Schaefer, 2010). The latest comprehensive study found that oil-bee pollination has evolved at least 28 times independently, and 1500–1800 species have developed oil-producing organs called elaiophores (Renner and Schaefer, 2010). These structures, located on various floral parts, constitute local glandular fields within the flower and can be anatomically separated into two categories (Vogel, 1969, 1974, 2009). Epithelial elaiophores consist of glandular epithelial or epidermal cells and their oil secretions are stored below a protective cuticle, forming small blisters, whereas trichomatic elaiophores consist of hundreds to thousands of oil-producing glandular trichomes (Buchmann, 1987; Silvera, 2002; Machado, 2004). The latter produce oil continuously and secretion is often unprotected, but oil can in some cases be accumulated in a subcuticular space at the tip of the trichome (Vogel, 1974; Buchmann, 1987; Cocucci and Vogel, 2001). Epithelial and trichomatic elaiophores produce non-volatile oils gathered by the females of specialized oil-collecting bees belonging to two families of Hymenoptera, Mellitidae and Apidae (Michener, 2007), which have developed morphological adaptations on their legs or abdomen to harvest and store lipids (Schlindwein, 1998; Cocucci et al., 2000; Vogel, 2009). Documented uses of floral lipids by oil-collecting bees show that females mix oils with pollen and use this mixture to feed their larvae (Vogel, 1974; Simpson and Neff, 1981). In some species, it has been observed that females also cover the brood cell walls of their nest with a complex set of different substances that contains floral lipids (Buchmann, 1987; Cane et al., 1983; Alves-dos-Santos et al., 2002).
Sisyrinchium (Iridaceae: Iridoideae: Sisyrinchieae) is a large and complex genus with a distribution spanning the American continents from subarctic areas to Tierra del Fuego. Many taxonomic studies have already been conducted, mainly based on morphological traits (Klatt, 1861; Baker, 1878; Bentham and Hooker, 1883; Rudall et al., 1986; Goldblatt et al., 1990; Ravenna, 2000, 2002, 2003b), but the systematics of the genus remain poorly resolved. The number of recognized species varies from approx. 80 (Goldblatt et al., 1989) to approx. 200 (Rudall et al., 1986) and the subgeneric divisions are not well defined and are, as such, unsatisfactory (Goldblatt et al., 1990; Cocucci and Vogel, 2001; Ravenna, 2003b). Moreover, Central and South American representatives of the genus remain largely unknown: 24 % of the 206 taxa accepted by the World Checklist of Iridaceae were described from these areas during the past 10 years (Barker, 2004), suggesting that many species still remain to be described.
Approximately 35 % of Sisyrinchium species have flowers with elaiophores of the glandular trichome type called hereafter nuptial trichomes and located either on the staminal column or on the adaxial side of the tepals, sometimes on both parts. Species with elaiophores are almost exclusively South American. Species diversity in Sisyrinchium is estimated to be highest in South America, mostly around the Paraná river basin and along the subandean ranges, which also corresponds to the range area where oil-collecting bees visiting oil-producing species of Sisyrinchium have been observed (Cocucci and Vogel, 2001). A number of Sisyrinchium species from North America bear stipitate glandular trichomes on their staminal column (McVaugh, 1989; Cholewa and Henderson, 2002). These trichomes are scarce, scattered towards the basal part of the staminal column, slender with a small blister of secretion towards the tip of the trichome head. However, no oil-collecting bee has ever been recorded as visiting North American species with such glandular trichomes. These observations led Cocucci and Vogel (2001) to propose a southern Neotropical origin for Sisyrinchium. The presence of glandular trichomes on several North American species might be a residual condition evolved from elaiophores (Cocucci and Vogel, 2001). Other species have pollen flowers, devoid of trichomes either on the filamental column or on the adaxial side of tepals. Members of this third category are widely distributed throughout the range of the genus.
The goal of this study was to elucidate the phylogenetic relationships among Sisyrinchium species and genera of Sisyrinchieae (Fig. 1), using a total evidence approach, and to test the monophyly of the subgeneric divisions as defined in the existing classifications. The resulting historical framework was used to analyse geographical patterns, optimize the evolutionary history of elaiophores and make hypotheses about potential shifts in the pollination system.
Taxa sampled, voucher information and GenBank accession numbers are listed in Appendix 1. A total of 101 Sisyrinchium accessions from South and North America, representing 85 species covering the different subgeneric arrangements proposed in the literature, were sampled. With the exception of S. jamesonii, it was not possible to obtain plant material from section Segetia (Ravenna, 2003b), which includes approximately four Andean species from Argentina, Bolivia, Peru, Ecuador and Venezuela. Outgroups were selected from the genera Olsynium (five species), Orthrosanthus (one species) and Solenomelus (two species), which are closely related to Sisyrinchium within Sisyrinchieae (Goldblatt et al., 2008). Plant material was mostly sampled from the wild or from cultivated specimens obtained from seeds collected in the wild and held in botanical gardens or national collections. A special effort was put into sampling of S. micranthum, a species that exhibits a high level of morphological plasticity and is closely related to S. laxum and S. rosulatum (Supplementary Data Fig. S1, available online). Previous attempts to generate a taxonomical classification for these variants have failed because plants often present contradictory combinations of character states (Johnston, 1938; Ravenna, 2001b). Since the geographical distribution of S. micranthum, from Canada to South Chile (Ravenna, 2001a, b), is probably the widest of any Sisyrinchium species, specimens were widely sampled to represent morphological variation and cover as much as possible its distribution range.
Total DNA from fresh or silica gel-dried leaves was extracted using the NucleoSpin® Plant II (Macherey-Nagel, Düren, Germany) extraction kit, following the manufacturer's instructions. A combination of quickly and slowly evolving coding and non-coding DNA regions were used to infer phylogenetic relationships among the taxa sampled. Three coding plastid DNA regions (rpoC1, rpoB and matK), two plastid DNA intergenic spacers (trnH-psbA and trnQ-rps16), two mitochondrial DNA introns (nad1-2/3 and nad4-1/2) and the nuclear ribosomal DNA internal transcribed spacer (ITS) region, including ITS1, ITS2 and the 5·8S gene, were used. These loci have shown their potential to complement each other and improve the resolution of the phylogenetic signal at different taxonomic levels (Freudenstein and Chase, 2001; Chat et al., 2004; Kress et al., 2005; Shaw et al., 2005, 2007; Chase et al., 2007; Hollingsworth et al., 2009). Primers used to amplify each DNA region and additional primers used for sequencing are given in Table S1 in the Supplementary Data. The ITS region was first PCR-amplified for ten different samples using primers ITS5 and ITS4 (White et al., 1990). The visualization of PCR products on a 1 % agarose gel revealed two different amplification products (approx. 500 and 700 bp). Both PCR products were size selected on the agarose gel and extracted with the MinElute gel extraction kit (Qiagen, Australia). After sequencing and alignment, these regions were BLASTED against GenBank to verify their identity. The smaller fragment proved to be a contamination from fungal DNA. A semi-nested PCR amplifying only the 700-bp fragment was achieved using a specifically designed forward primer described in Table S1 in the Supplementary Data. PCR amplifications were performed using a PTC-100 MJ-Research thermal cycler in 30 µL total volume reaction with the following reaction components: 1·5 µL of genomic DNA (approx. 15–50 ng), 1 µm of each primer, 250 µm of dNTP, 1× rTaq buffer, 2·5 mm MgCl2 and 0·2 U Taq polymerase (Taq CORE Kit 10; MP Biomedicals, Illkirch, France). The rpoC1 and rpoB loci were successfully amplified with the addition of DMSO (1·2 µL) to the PCR mix. The detailed PCR conditions for each DNA locus used in this study are given in Table S2 in the Supplementary Data. Each DNA region was amplified as a single fragment except in a few cases where internal primers originally designed for sequencing were used to amplify the targeted region in smaller fragments. PCR products were purified and sequenced at the Genoscope (www.genoscope.fr). Raw forward and reverse sequences for each sample were assembled with CodonCode Aligner 3·5·3 (CodonCode Corporation, Dedham, MA, USA); ambiguous bases were corrected after examination of chromatograms, and consensus sequences were edited.
Alignments were first produced using ClustalX (Thompson et al., 1997) and MUSCLE (Edgar, 2004) and further improved manually using MEGA4 (Tamura et al., 2007). All detected polymorphisms [SNPs and insertion–deletions (indels)]were visually checked and further validated using a base quality threshold above 20. The indels of the non-coding regions (trnH-psbA and trnQ-rps16 spacers, nad1-2/3, nad4-1/2, ITS1 and ITS2) shared by two or more taxa were coded as binary characters using GapCoder (Young and Healy, 2003), a program based on simple indel coding sensu Simmons and Ochoterena (2000). All eight DNA regions were first analysed independently (results not shown) and further combined according to the genome. Since the comparison of the resulting topologies of plastid DNA and mitochondrial DNA regions revealed no instances of well-supported topological differences and provided a higher resolution when analysed together than separately, they were combined and compared with the ITS results. Two minor topological incongruences were detected (see ‘Combined plastid DNA, mtDNA and nuclear DNA dataset’ in Results). However, since no major topological conflict was detected, all data were combined into a single supermatrix for subsequent analysis.
Datasets including indels coded with GapCoder were analysed using the parsimony criterion in PAUP* version 4·0b10 (Swofford, 2002). All the MP analyses used heuristic searches with 1000 random addition replicates, tree-bisection-reconnection (TBR) branch swapping and multrees on, with all character states unordered and equally weighted, and indels coded as previously described. Strict and majority-rule consensus trees were calculated from all most-parsimonious trees. The robustness of nodes was evaluated using MP with 1000 bootstrap replicates of new heuristic searches (100 random addition replicates, TBR branch swapping, multrees off). The computer used ran out of memory during heuristic searches for five datasets (psbA-trnH, trnQ-rps16, nad1-2/3, nad4-1/2 and ITS), which meant that analyses for these regions and the combined datasets were conducted using a method developed for this paper and the parsimony ratchet (Nixon, 1999) described in Appendix S1 in the Supplementary Data.
ML (Felsenstein, 1981) and Bayesian MCMC (Yang and Rannala, 1997) analyses were performed. Model parameters listed in Table S3 in the Supplementary Data were set to those calculated by MrModeltest 2·3 (Nylander, 2004). The Akaike information criterion was chosen to select the most appropriate model of DNA substitution for each dataset or data partition used in the analyses. ML analyses were performed using PhyML 3·0 online web server (Guindon et al., 2005) on data matrices excluding the coded indels, since indels cannot be dealt with in PhyML. The reliability of ML topologies was assessed by non-parametric bootstrap tests using 200 pseudo-replicates.
Bayesian analyses were run using MrBayes 3·1·2 (Ronquist and Huelsenbeck, 2003) with the data partitioned and the most appropriate evolutionary model implemented for each partition, as indicated in Table S3 in the Supplementary Data. Coded indels were included in a separate data matrix and treated using a simple model with variable rates. Two independent runs, each comprising four Markov chains (one cold and three heated) and starting with a random tree, were performed simultaneously for 107 generations, sampling trees at every 100th generation. The convergence diagnostic was calculated every 104 generation and its critical value was set to stop the analysis automatically when the standard deviation of the split frequencies had reached the value defined by the stopval command (stoprule = yes stopval = 0·01). In all analyses, the first 25 % trees from each run were discarded as burnin. Resulting trees from the two independent runs were then pooled to produce one 50 % majority-rule consensus tree and Bayesian posterior probabilities were generated for the resulting tree.
Phylogenetic trees resulting from all three analyses (MP, ML and Bayesian) were combined to build a consensus tree manually. For each tree, a given node was kept in the consensus tree only if the bootstrap support for MP or ML was >80 %, or if the posterior probability (PP) was >0·95. Consensus trees were assembled for each genome dataset and for the combined plastid DNA + mitochondrial DNA dataset in order to identify potential topological conflicts among these different combinations. The consensus tree based on the whole dataset analyses was constructed accordingly.
Glandular trichomes are present in the flowers of numerous Sisyrinchium species and for several South American species it has been documented that they secrete oil collected by the pollinating bees (Cocucci and Vogel, 2001). These trichomes are located on the stamens and the adaxial side of tepals but the nature of their secretion has not yet been identified for many species, especially those from North America. Following the definition of nuptial nectaries of Delpino (1874) and Fahn (2000) these trichomes will be called nuptial trichomes in the present study to emphasize the difference between them and glandular trichomes that can be located elsewhere on the flower. Among the 85 species of Sisyrinchium included in this work, 54 bear nuptial trichomes and 50 of them were studied for oil-producing trichomes. Fresh flowers at anthesis were observed under a Zeiss Stemi SV6 stereomicroscope (Carl Zeiss AG, Göttingen, Germany) to record the shape, position and density of glandular trichomes. Detailed observations of the morphological structure of nuptial trichomes were carried out directly on fresh flower buds and fresh flowers at anthesis, with a scanning electron microscope (SEM Hitachi S-3000N; Tokyo, Japan). Dissected organs were mounted on an aluminium specimen holder and observed under low partial vacuum and Peltier cooling stage. SEM pictures were taken using an environmental secondary electron detector.
Nile Red staining, which allows the detection of lipids according to their hydrophobicity (Greenspan et al., 1985; Diaz et al., 2008), was used to obtain information about the oil content of fresh flowers at anthesis. The emission spectrum of Nile Red shifts from red in the presence of polar lipids to yellow when they are combined with non-polar lipids. Observations were made using a Nikon AZ100 macroscope (Nikon France, Champigny-sur-Marne, France). Nile Red yellow emission was observed with 450–500 nm excitation and 535 ± 20 emission filters; red emission was observed with 450–500 excitation and 610 long-pass emission filters.
For character optimization, nuptial trichomes were coded using two discrete characters, respectively, the type (A) and localization (B) of nuptial trichomes. Three states were defined for the first character, namely (0) absent, (1) oil-producing trichomes and (2) non-oil-producing trichomes. Nuptial trichome localization was coded as follows: (0) not applicable (i.e. absence of trichomes); (1) present on the filamental column or filaments (depending on the degree of fusion); (2) present on the filamental column and adaxial side of tepals; and (3) present on the adaxial side of tepals only. We considered that the presence of trichomes on both the filamental column and the adaxial side of tepals should not be treated as a polymorphism but as a separate character state, with potentially different functional implications for the plant–insect interaction, due to the extended surface covered by the trichomes. The consensus trees resulting from the plastid DNA + mitochondrial DNA analysis, the ITS analysis (results not shown) and the combined (supermatrix) analysis were used for character optimization with the MP and ML methods implemented in MESQUITE 2·73 (Maddison and Maddison, 2010). With MP, character states were treated as unordered, allowing any transition among states. ML optimization was conducted using the MK1 model of evolution (Schluter et al., 1997; Pagel, 1999), which gives equal probability for changes between all character states.
Ancestral patterns of geographical distribution were inferred using the same phylogenetic trees as in the above paragraph. The four main geographical areas defined as character states (Fig. 2) represent the overall distribution of Sisyrinchium species (plus the outgroups) on the American continent. A fifth character state was defined for S. acre which is found on the island of Maui in the Hawaii archipelago. The distribution area of each species included in the study was verified (Barker, 2004) and further literature-based investigations were conducted to obtain information on the distribution of species not accepted by the World Checklist of Iridaceae. Detailed descriptions of the way character states were defined are given in Appendix S2 in the Supplementary Data. The distribution area of the genus was optimized similarly to what is described in the above paragraph except that character states were treated alternatively as unordered or ordered, with a step matrix which allowed a penalty for changes between non-adjacent areas. Due to the presence of several multi-state taxa, only MP optimization was performed. Since the tree used was not fully resolved it was not possible to infer the biogeographical history using the parsimony-based dispersal-vicariance method (Ronquist, 1996, 1997).
Alignments of coding regions of both plastid and mitochondrial regions were straightforward because insertion/deletion (indel) events were absent. The recently discovered rps19 gene was present in the spacer region between the trnH and psbA genes in all four genera included in the present study, as in most monocots (Wang et al., 2008). Alignments were manually improved in intron regions. MP, ML and Bayesian analyses were conducted independently for each plastid DNA and mitochondrial DNA regions. Table 1 gives the number of potentially parsimony informative characters, number of most-parsimonious trees, tree lengths, and the consistency and retention indices (CI and RI, respectively) for the strict consensus trees resulting from MP analyses of the separate and combined datasets. There was little homoplasy in four of the plastid DNA datasets (CI ≥ 0·80), except for psbA-trnH (CI = 0·65). In the same way, the level of homoplasy was relatively low for the two mitochondrial DNA datasets (CI ≥ 0·76). Trees were rooted with Orthrosanthus as sister to all other genera, according to the most recent phylogenetic analysis of Iridaceae (Goldblatt et al., 2008). The trees resulting from the MP, ML and Bayesian analyses on separate datasets showed nearly identical topologies for each genome. All plastid-based analyses but one resulted in trees displaying three strongly supported clades [parsimony and likelihood bootstrap supports (PBS and LBS, respectively) >80 % and posterior probability (PP) >0·95), each corresponding to a monophyletic genus, respectively Solenomelus, Olsynium and Sisyrinchium. The ML analysis of the psbA-trnH locus only moderately supported the monophyly of Sisyrinchium (LBS = 57·6). Analyses based on the rpoB, matK, psbA-trnH and trnQ-rps16 loci confirmed the position of Olsynium as sister to Sisyrinchium and Solenomelus as sister to Olsynium + Sisyrinchium with strong support. The only exception was the moderate support obtained for the sister group relationship between Olsynium and Sisyrinchium in the ML analysis of the psbA-trnH locus (LBS = 58·8). The mitochondrial-based analyses gave identical results with strong support for the monophyly and the relationships of the three genera, except for the clade Solenomelus in the MP analysis of the nad4-1/2 locus (PBS = 63·8). Almost all analyses resulted in the monophyly of Solenomelus, Olsynium and Sisyrinchium. Since no incongruence was detected among tree topologies, plastid and mitochondrial markers were combined into a single matrix for each genome.
PCR amplification of the ITS region failed for two samples (Orthrosanthus monadelphus and Olsynium douglasii, which were therefore not available for the subsequent analyses of the nuclear dataset. The alignment included 715 characters (665 from sequences and 50 from coded indels) of which 312 (43·6 %) were potentially phylogenetically informative. The parsimony ratchet yielded 4012 equally most-parsimonious trees (details are given in Table 1) of 916 steps, CI (excluding uninformative characters) = 0·63 and RI = 0·87. Tree topologies resulting from MP, ML and Bayesian analyses showed no major incongruence. The monophyly of Olsynium (PBS = 100 %, LBS = 100 %, PP = 1) and Sisyrinchium (PBS = 88·7 %, LBS = 88·0 %, PP = 0·98) was strongly supported, as was the position of Olsynium as sister to Sisyrinchium (PBS = 98·1 %, LBS = 99·0 %, PP = 1). Detailed results are given in Fig. S2 in the Supplementary Data.
No major incongruence was detected among topologies based on the separate analyses, allowing the datasets to be combined. Plastid markers were combined into a single matrix totalling 4845 characters (including the 245 coded indels), of which 494 (10·2 %) were potentially phylogenetically informative. Most of the potentially parsimony informative characters (66·5 %) belonged to the spacers in the psbA-trnH and trnQ-rps16 regions. There were 3402 characters in the total mitochondrial DNA matrix (including 62 coded indels), of which 144 (3·9 %) were potentially phylogenetically informative. Parsimony informative characters were mainly found in the introns of the mitochondrial DNA regions studied (89·5 %). Results obtained with the plastid and the mitochondrial matrices, respectively, are given in Figs S3 and S4 in the Supplementary Data.
Parsimony ratchet analysis of all eight DNA regions representing all three genomes yielded 3340 equally most-parsimonious trees of 2462 steps, CI (excluding uninformative characters) = 0·68 and RI = 0·90. ML searches produced a best ML tree with –ln L = 25978 (Fig. 3). The 30 901 trees retained after the burn-in from the two runs of the Bayesian analysis were summarized into a 50 % majority-rule consensus tree. Tree topologies resulting from the total evidence analyses with parsimony ratchet, likelihood and Bayesian approaches were largely congruent as shown by the number of shared nodes in the consensus tree (Fig. 4). The monophyly of Solenomelus, Olsynium and Sisyrinchium identified with the separate analyses is strongly supported in the consensus tree (PBS = 100 %, LBS = 100 %, PP = 1 for each genus), and the placement of Olsynium as sister genus to Sisyrinchium was also strongly supported (PBS = 100 %, LBS = 100 %, PP = 1). According to the topology and node supports observed, nine monophyletic groups were identified within Sisyrinchium, clades I to IX, all well supported in all of the analyses (PBS ≥ 90 %, LBS ≥ 90 %, PP = 1), except clade IX which was strongly supported only in the Bayesian analysis (PP = 1), but only moderately supported in the MP and ML analyses (PBS = 75·7 %, LBS = 59 %). All internal nodes concerning the relative placement of the clades appeared strongly supported in the Bayesian analysis, but the results obtained with the parsimony and likelihood analyses were more contrasted. The sister group relationship of clades I and II to the rest of the genus was strongly supported (PBS = 100 %, LBS = 100 %). Strong support was also found for the respective sister relationships of clade III (PBS = 83·8 %, LBS = 93 %), clade IV (PBS = 100 %, LBS = 100 %) and clade V (PBS = 100 %, LBS = 100 %) to the rest of the genus. Moderate to weak support was obtained for the respective sister relationships of clade VI (PBS = 56·1 %, LBS = 74·5 %), clade VII (PBS = 75·8 %, LBS = 71 %) and clade VIII (PBS = 86·8 %, LBS = 64 %) with the remaining species.
The separate plastid DNA + mitochondrial DNA and ITS analyses resulted in only two conflicting nodes which concerned only one species each time (Fig. 5). In the plastid DNA + mitochondrial DNA phylogeny, S. jamesonii was sister to all other Sisyrinchium species, although this relationship appeared strongly supported only in the Bayesian analysis (PBS = 64·1 %, LBS = 74·0 %, PP = 0·97), whereas in the ITS phylogeny the species is found with clade II within a polytomy and clade I is sister to all other Sisyrinchium species with strong support in each analysis (PBS = 88·7 %, LBS = 88·0 %, PP = 0·98). Relationships among clades I, II and S. jamesonii were unresolved in the total evidence analyses, and were consequently represented as a polytomy in the consensus tree of all total evidence analyses. The only noticeable incongruence concerned the position of S. chilense which varied greatly between the topologies obtained after analysis of the combined plastid DNA + mitochondrial DNA dataset, and the ITS data matrix. In the first case S. chilense fell within clade V with relatively strong support (PBS = 81·5 %, LBS = 75·5 %, PP = 1), whereas in the ITS phylogeny S. chilense fell within clade IX, which is only distantly related to clade V, also with strong support (PBS = 97·9 %, LBS = 95 %, PP = 1). This incongruence led to the placing of S. chilense at an intermediate position in the consensus tree from the total evidence analyses.
The character optimizations on the tree obtained from the full molecular dataset are reported in Fig. 6: tree A shows the type of trichomes and tree B indicates their localization. The present observations confirmed that Orthosanthus monadelphus and all Olsynium species included in the present study lack nuptial trichomes. In Solenomelus which comprises only S. segethi and S. pedunculatus, two different types of nuptial structures were identified. The filamental column of S. pedunculatus was entirely covered by unicellular glandular trichomes without any secretion outside the trichomes or any associated secretory cavity, whereas S. segethi had multicellular scales all along the column, covered by secretions towards the lower part of the column. No oil secretion was detected using Nile Red for either species. All nuptial trichomes observed within Sisyrinchium were glandular and unicellular, except in S. minutiflorum (see description below). It was observed that the nuptial glandular trichomes of Sisyrinchium always produced a blister filled with secretions towards the tip (Figs 77–9), except for two species: S. hasslerianum and S. hoehnei. Secretions were stored below the cuticle, which is pulled away from the primary wall to form a subcuticular space. The thickness of the cuticle varies across species.
All species were completely devoid of nuptial trichomes.
Three species included in clade II are devoid of nuptial trichomes (S. convolutum, S. mandonii and S. angustissimum), but the other three species within this clade had nuptial trichomes sparsely located all along the filamental column. These trichomes were scarce, with a slender stalk, a small blister of non-oil secretion with a particularly thin and weak cuticle.
All species except two (S. convallium and S. cf. nervosum) bore nuptial trichomes and all states defined for the localization of nuptial trichomes were present within the clade. No general pattern of trichome distribution on the filamental column could be drawn from our observations. Trichomes were densely distributed along the column and on the free part of the filaments for S. macrocarpum, they densely covered only the lower two-thirds of the filamental column of S. pearcei and S. papillosum, whereas S. graminifolium and S. aff. graminifolium bore trichomes sparsely distributed all along the column. When nuptial trichomes were distributed on the adaxial side of tepals, they were always concentrated at their base.
Whatever their location, all nuptial trichomes in this clade produced oil. Apart from samples of S. graminifolium and S. aff. graminifolium, which had trichomes resembling those of clade II, clade III trichome cells exhibited a short thick stalk and the secretory cavity is larger, enclosed in a thinner cuticle, than in the other clades including species bearing oil-producing trichomes.
All species were completely devoid of nuptial trichomes.
All members of this clade bore nuptial trichomes on the filamental column and oil secretion was detected using Nile Red for all the species tested. The distribution pattern of trichomes was similar in almost all the species observed: trichomes were densely tufted along the lower third of the column and sparsely distributed on the upper two-thirds. The trichome stalk was long and slender and the blister size varied along the column: trichomes on the lower third exhibiting a larger blister than those on the upper two-thirds. Four species within this clade showed different distribution patterns of their trichomes: S. fasciculatum, S. hasslerianum and S. hoehnei had reflexed trichomes densely tufted along the upper two-thirds of the filamental column and sparsely distributed on the lower third, whereas few trichomes were present only at the base of the column of S. rambonis.
Two species were devoid of nuptial trichomes but the other species forming this clade had dense nuptial trichomes along the lower third of the filamental column. All nuptial trichomes observed in this clade produced oil that accumulated in the secretory cavity.
The different morphotypes of S. micranthum and closely related species in this clade had dense oil-producing trichomes on the lower thickened part of the filamental column. This thickened area varied in extent from just the lower half to the entire length of the fused part of the filaments among the different samples. Taxa included in clade VII exhibited an unusual feature in the genus: oil-producing trichomes formed a row along the middle vein of each tepal. This line usually ended at the lower third of the patent disposition of tepals.
The two species forming this clade bore oil-producing trichomes concentrated on the lower quarter of the filamental column. The remainder of the filamental column of S. minus was covered by sparse and scattered trichomes, whereas the middle half part of the column of S. minutiflorum was glabrous and the upper quarter was covered by reflexed multicellular deep-purple trichomes. Located just below the anthers, they exhibited a multicellular stalk and terminated in a glandular cell (as already shown by Cocucci and Vogel, 2001).
Based on geographical distribution and trichome secretion, species of this clade could be separated into two groups. Species from South America had oil-producing trichomes on the filamental column. Trichomes were densely arranged on the lower third part of the column in S. patagonicum and S. cf. macranthum, whereas they extended onto the thickened lower two-thirds of the column in S. platense, with few scattered trichomes on the upper part. Furthermore, this latter species had a row of sparse oil-producing trichomes along the lower third of the middle vein of tepals.
All other species were from North America. Nuptial trichomes with a long and slender stalk were always present on the filamental column but, even if they were topped by a small blister of secretion, they did not produce oil. The trichomes were always scarce and were sparsely distributed from just around the column base to the entire lower two-thirds, depending on the species.
The MP optimization of two aspects of trichome evolution is shown in Fig. 6. Values obtained from the ML optimizations are indicated for the most internal nodes of the mirror trees. These results show that the ancestral state for Solenomelus, Olsynium and Sisyrinchium was the absence of trichomes. They are absent from the early diverging groups of Sisyrinchium. Non-oil-producing trichomes appeared several times on the filamental column within clade II. The ancestral state for clade III is unambiguously the presence of oil-producing trichomes, but their ancestral localization is ambiguous with both MP and ML optimization: trichomes first appearing on both the filamental column and the adaxial side of tepals or only on the tepals is equally parsimonious and more likely than the other states. The ancestral state for clade IV was the absence of trichomes. Oil-producing trichomes evolved in the common ancestor of clade V to clade IX and, except for clade VII, the ancestral localization was on the filamental column. For clade VII, the ancestral localization was on both the filamental column and the adaxial side of tepals. The number of origins of oil-producing trichomes (one or two) is ambiguous with the MP optimization, but the ML optimization suggests that a single origin is more likely.
The optimization of biogeographical patterns on the phylogenetic trees revealed the existence of a strong phylogenetic signal for this character in Sisyrinchium and in the two closely related genera Olsynium and Solenomelus (Fig. 10). Solenomelus is restricted to the central and southern part of Area A, whereas Olsynium is distributed throughout Area A and the southern part of Area C (except O. douglasii which has a range area extending towards the western United States). Within Sisyrinchium, S. jamesonii, clade I and clade II are restricted to Area C, except S. acre which is endemic to the Hawaii archipelago. The distribution area of clade III covers the central and southern parts of Area A (Chile and Argentinean Andes) and corresponds to the red coloured part within this area (Fig. 2). Most of the species within clades IV and V occur only within Area B (southern Brazil, Paraguay, northern Argentina and Uruguay), except S. alatum, S. palmifolium subsp. palmifolium and S. vaginatum subsp. vaginatum which are also distributed in the southern part of Area C (Venezuela, Ecuador, Peru and Bolivia). Sisyrinchium chilense is distributed mainly in Area B (Chile, Argentina and Falkland Islands) but is also present in the southern part of Area C (Ecuador, Peru and Bolivia). Apart from S. cf. valparadiseum which is found only in the central part of Area A, all species in clade VI are distributed within Area B (southern Brazil, Paraguay, northern Argentina and Uruguay). The range recorded in the World Checklist of Iridaceae for S. micranthum (Barker, 2004) was applied similarly to all morphotypes of the complex, because their respective ranges were impossible to determine accurately from the existing data. Most taxa of clade VII are distributed in Areas A, B and C, with S. laxum restricted to Areas A and B (southern Brazil, Paraguay, Argentina and Uruguay), and S. rosulatum restricted to Area D (south-eastern USA to north-eastern Mexico). The two species grouped in clade VIII are restricted to Area B. More than 80 % of clade IX is exclusively North American (Area D), except S. cf. macranthum and S. patagonicum which are native to Area A (Chile and western Argentina) and S. platense to Area B (southern Brazil, Paraguay, northern Argentina and Uruguay). Although ancestral nodes were well resolved for Solenomelus and Sisyrinchium, and for clades I to VIII of Sisyrinchium, deeper nodes and the ancestral state for clade IX were mostly ambiguous, preventing us from formulating solid hypotheses about the evolutionary history of the geographical distribution of the genus Sisyrinchium. Results were identical whether character states were treated as unordered or with a step matrix which allowed a double weight for changes between the non-adjacent areas.
The phylogenetic hypothesis presented here includes the most extensive sampling of Sisyrinchium and its two closest relatives Olsynium and Solenomelus performed so far. A total evidence approach was used to infer phylogenetic relationships within Sisyrinchium, the most species-rich genus of Iridaceae on the American continent. Data from the three genomes (plastid, mitochondrial and nuclear) were combined after checking that there was only minor incongruence among them, which might be related to gene flow or possibly hybrid origin of the taxa concerned. Indeed, limited breeding studies (Henderson, 1976; Cholewa and Henderson, 1984) concerning ten North American species have shown that several species are genetically compatible, producing fertile hybrids. The total evidence approach allowed eight well-supported and one moderately supported clades to be identified in Sisyrinchium. Most of the phylogenetic information came from the non-coding regions of all three genomes, and the information contained in indels had a major impact on tree topology. However, although coding regions were strongly conserved, they were useful for resolving the deeper nodes.
The most comprehensive infrageneric classification performed so far was proposed by Bentham and Hooker (1883) who recognized four sections in Sisyrinchium (Rudall et al., 1986). Recently, two subgenera, Echtronema and Sisyrinchium, were retained as subdivisions of Sisyrinchium, excluding two sections (Eriphilema and Nuno) previously included in the genus by Bentham and Hooker (1883) and now grouped with the genera Phaiophleps and Chamelum under the generic name Olsynium (Goldblatt et al., 1990). The phylogenetic study of Goldblatt et al. (1990), based on the morphological, leaf anatomical and basic chromosome number characters detailed in Table 2, suggested that species classified in subgenus Sisyrinchium were actually more closely related to Solenomelus and Olsynium than the species of subgenus Echtronema. Nevertheless, the trait differences used to discriminate the two subgenera are unclear and subgenus Echtronema is currently regarded as a heterogeneous group of species (Goldblatt and Manning, 2008). The present study shows that neither Echtronema nor Sisyrinchium is monophyletic. Most species of subgenus Echtronema are more closely related to Olsynium and Solenomelus than those of subgenus Sisyrinchium (Fig. 11).
An alternative proposal to subdivide the genus Sisyrinchium s.s. (excluding Olsynium), into eight sections has recently been advanced by Ravenna (2000, 2002, 2003b) on the basis of morphological characters (Table 2). The subdivisions recognized in his classification are partly supported by the present results, with some exceptions (Fig. 11). Species of section Hydastylus are distributed among clades I and IV, the latter also including the species forming section Viperella. This was unanticipated since the latter section is characterized by distinctive morphological characters, such as absent or highly reduced basal leaves, and caulescent plants bearing many short caulinate leaves (Ravenna, 2003b). The present results suggest, however, that section Hydastylus should be restricted to the species of clade I, which corresponds to the initial morphological definition of Hydastylus as a genus distinct from Sisyrinchium (Bicknell, 1900). The most evident morphological differences reside in the ancipital stem (simple and scapose, terminated by a spathe of two conduplicate bracts) and in the yellow flowers. Tepals are obtuse or acute but neither aristate nor emarginate, filaments are more or less adherent below but free and spreading for more than half their length, anthers are versatile, slender style branches are divergent and the ovary is strictly glabrous. This set of morphological characters differentiates species of clade I from all other Sisyrinchium species. Similarly, since clade IV is well supported, its status as a section should be considered. However, further investigations are needed to identify morphological synapomorphies for this group. The section Echtronema corresponds almost entirely to clade II, except for S. graminifolium which belongs to clade III. The species of section Spathirachis, characterized by sessile lateral inflorescences with membranous spathes along the floral stem (Goldblatt et al., 1990; Ravenna, 2003b), are found in clade III. However, this clade also includes S. graminifolium, a species with pedunculate lateral spathes classified in section Echtronema, whereas S. brasiliense and S. avenaceum, with sessile lateral spathes, have been classified in section Hydastylus by Ravenna (2000) on the basis of the herbaceous nature of the spathes, and belong to clades IV and VI, respectively. The present results suggest, therefore, that the presence of sessile lateral spathes, on which section Spathirachis is mainly based, is a homoplasious and non-discriminating trait in the framework of a natural classification. Species classified in sections Lenitium and Scirpeocharis are grouped together in clade VII, with a monophyletic group of four species of section Lenitium (ancipitate, winged stems) and one species of section Sisyrinchium (S. rambonis) sister to a clade which includes all species of section Scirpeocharis (cylindrical stems) and two species of the former section, S. sellowianum and S. soboliferum, nested among species belonging to section Scirpeocharis. These results strongly suggest that the two sections are not two clades but only a single natural group. The presence of S. rambonis in clade VII suggests that the traits used to discriminate this section should also be questioned. Apart from S. rambonis, section Sisyrinchium sensu Ravenna consists of clades VI to IX, which form a monophyletic group on the basis of ITS, with strong support (results not shown): PBS = 97·9 %, LBS = 95·0 %, PP = 1. This group is only moderately supported in the total evidence analyses, the supports obtained from the ML and Bayesian analyses being moderate (LB = 74·5 %, PP = 0·95) and that from the MP analysis being weak (PB = 56·1 %). The phylogenetic pattern corresponding to this section (three well-supported clades, VI to VIII, and clade IX moderately supported) suggests the section needs to be revised, all the more since combined morphological and geographical features discriminate among the four clades.
Section Segetia was represented in the present study by a single accession (S. jamesonii), making an assessment of the monophyly of this section impossible. Furthermore, S. jamesonii is unplaced with regard to clades I and II.
The optimization of nuptial trichomes on a phylogenetic tree (Fig. 6; Tree A) clearly shows that the presence of glandular trichomes on the stamens and/or the adaxial side of tepals is derived in Sisyrinchium (with probably three independent origins). The results show that the presence of nuptial trichomes along the filamental column of Solenomelus pedunculatus results from an independent transition. In the same way, an independent transition led to the presence of nuptial trichomes on the filamental column in species of clade II and, although they were secretory, they produced no oil. The results suggest that these trichomes evolved several times within clade II but since this hypothesis is well supported only in the Bayesian analysis, we considered only one transition. In contrast, it is unclear whether the oil-producing trichomes have a single or two evolutionary origins, both scenarios being equally parsimonious on strongly supported internal nodes. The ML optimization favours a single origin for oil-producing trichomes, since the probability of state 1 (presence) at the corresponding node indicated by a black arrow in Fig. 6 (Tree A) is higher than the probability of state 0 (absence) (P1 = 0·768 vs. P0 = 0·199). However, optimization on trees based on separate datasets (plastid DNA + mitochondrial DNA on the one hand, ITS on the other hand) favours two independent origins. For both, the probabilities associated with absence of oil-producing trichomes at the relevant node are higher than the probability for their presence, respectively P0 = 0·659 and P1 = 0·304 for the plastid + mitochondrial dataset, and P0 = 0·638 and P1 = 0·342 for the ITS dataset. The probability values obtained for each state under ML analyses vary in the same way at the following node indicated by a red arrow in Fig. 6 (Tree A): P0 = 0·304 and P1 = 0·678 on the phylogenetic tree from the full molecular dataset and P0 = 0·633 or 0·638 and P1 = 0·350 or 0·342 when optimization is performed on trees based on the separate datasets.
The species in clade III are distributed from northern-central Chile to Tierra del Fuego, along the southern Andes (Fig. 2). They share this distribution area with only a few other oil-producing Sisyrinchium species, namely S. valparadiseum (clade VI), S. chilense and S. patagonicum (clade IX), which are not phylogenetically related to clade III, and the range area of this clade is totally disconnected from those of clades IV and V. Furthermore, all states for the distribution of nuptial glandular trichomes are represented in clade III (Fig. 6; Tree B) and this combination is unique among the clades identified within the genus. The comparative observation of trichome morphology in almost all species included in the present study revealed several distinctive features for the species of clade III: the stalk of nuptial glandular trichomes is short and thick and the secretory cavity is proportionately larger with a thinner cuticle (Figs 7, ,88 and and9)9) than observed in the other clades. It is interesting to note that the distribution area of species forming clade III is included within the range of oil-collecting bees of the genus Chalepogenus, whereas species of clade V are present within the range area of oil-collecting bees of the genus Lanthanomelissa. It has been shown that several species of bees belonging to these two genera collect oil from Sisyrinchium species (Vogel, 1974; Cocucci and Vogel, 2001). This body of observations leads us to suggest that oil-producing trichomes might have appeared at least twice independently within the genus. The present observations, compared with those previously published (Cocucci and Vogel, 2001), suggest that shifts in the pollination system occurred between the species originating from areas where oil-collecting bees are scarce or absent and species located in areas where oil-collecting bees are known to collect oil on flowers of Sisyrinchium. Thus, the shift from an ancestral state devoid of nuptial trichomes towards the presence of oil-producing trichomes in clades III and IV, or the loss of oil production within clade IX in the North American species, is probably related to the occurrence of tropical oil-collecting bees which are able to forage for oil on Sisyrinchium. These are absent from the central part (Area C) and the northernmost part (Area D) and present in the two other sub-areas of the distribution range of the genus (Cocucci and Vogel, 2001).
It is important to note that our analyses of trichome evolution are preliminary because the content of secretory trichomes was only tested for oil and not for other types of chemicals. As a result, the scenarios for trichome evolution proposed in our study are over-simplified, but they still provide a first insight into general trends for the genus Sisyrinchium.
A South American origin for Sisyrinchium is suggested by the location of the centres of diversity of the genus (Cocucci and Vogel, 2001) and phylogenetic studies of Iridaceae (Goldblatt et al., 2008) and of Sisyrinchieae (Goldblatt et al., 1990). Although the phylogenetic relationships between S. jamesonii and the species grouped in clades I and II is unresolved in our tree, the results suggest that the genus arose somewhere between Bolivia and south-western USA. The polytomy at the base of the tree, due perhaps to the poor sampling in section Segetia, does not allow us to suggest a more precise area for the geographical origin of the genus. Based on our results, two dispersal events might have occurred toward the subandean ranges of Area A (clade III) and the Paraná river basin (Area B) at the base of clades IV to VIII. Unresolved nodes within clade IX prevent us from proposing a geographical origin for this clade.
In this study, molecular, morphological and biogeographical evidences were used to clarify phylogenetic relationships in Sisyrinchium and provide a first insight into the evolutionary history of oil-producing trichomes in this genus. The results confirm the monophyly of the genus and the relative positions of the genera Olsynium and Solenomelus. Several main clades were identified within Sisyrinchium which are only partially congruent with the subdivisions recognized in the most recent infrageneric classification. Nuptial glandular trichomes evolved at least twice, more probably three times, independently during the evolutionary history of the genus. While the first transition, in the basal part of the tree, concerns a limited number of extant species, the two following transitions to oil-producing trichomes were placed at the base of two species-rich clades. These transitions were associated with major biogeographical and ecological changes in the resulting clades. These observations suggest that the acquisition of oil-producing trichomes and the subsequent loss of the trichome ability to produce oil among most of the North American species may well have played a key role in patterns of speciation in Sisyrinchium. This work provides an essential framework for future studies of the relationship between oil-secreting flowers and oil-collecting bees in Iridaceae, but a specific effort concerning the sampling of the species closely related to S. jamesonii and the basal clades I and II remain particularly crucial for improving the resolution at this phylogenetic level. Further phylogenetic studies that incorporate additional molecular markers as well as morphological, cytological and ecological characters are also needed to define more clearly the circumscription of taxonomic subdivisions in the genus and to gain a better understanding of speciation patterns. Studying the evolutionary history of oil-secreting structures in detail within Sisyrinchium, one of the largest genera of angiosperms involved in specialized oil-bee pollination, could help us to understand the extent to which species diversification of a plant lineage has been driven by this particular pollination system.
Supplementary data are available online at www.aob.oxfordjournals.org and consist of the following. Figure S1: Vouchers and flowers illustrating the morphological plasticity observed within S. micranthum and closely related species. Figure S2: Consensus tree based on the strict consensus tree of the parsimony ratchet, the estimated maximum-likelihood tree and the Bayesian 50 % majority-rule consensus tree obtained from the analysis of the nuclear dataset. Figure S3: Consensus tree based on the strict consensus tree of the parsimony ratchet, the estimated maximum-likelihood tree and the Bayesian 50 % majority-rule consensus tree obtained from the analysis of the chloroplast dataset. Figure S4: Consensus tree based on the strict consensus tree of the parsimony ratchet, the estimated maximum-likelihood tree and the Bayesian 50 % majority-rule consensus tree obtained from the analysis of the mitochondrial dataset. Table S1: Primers used for amplifying and sequencing. Table S2: PCR profiles for DNA amplification. Table S3: Data and models used in analyses. Appendix S1: Alternative methods used for the maximum parsimony analyses. Appendix S2: Method used to define geographical areas as character states to infer the ancestral patterns of distribution.
The authors are grateful to R. and E. Heaton (NCCPG Sisyrinchium National Collection, Newton Abbot, UK) who provided an important part of the plant material. We thank them deeply for their continuous help, kindness and availability. We are also indebted to S. Aubert and R. Douzet (Jardin Botanique de la Station Alpine du Lautaret, France) and F. Pautz (Jardin Botanique de Lyon, France) for generously providing seeds, dried specimens and leaf material for DNA extractions, and J. F. Bertrand, A. Dubois, T. Genevet and L. Saunois (Université Paris-Sud, France) for their investment in maintaining the living collection of Sisyrinchium developed in our University. M.-N. Soler (IFR87) generously performed the light microscope studies with competence. This work received funding from the French/Brazilian CAPES/COFECUB cooperation project Sv550/07, from the IFR87 ‘La plante et son environnement’, from the ‘Consortium National de Recherche en Génomique’, and the ‘service de systématique moléculaire’ of the Muséum National d'Histoire Naturelle (IFR 101). It is part of the agreement no. 2005/67 between the Genoscope and the Muséum National d'Histoire Naturelle on the project ‘Macrophylogeny of life’ directed by G. Lecointre.
Taxa (identification number), geographical origin (specimens cultivated in the greenhouse at Université Paris-Sud 11 or elsewhere are noted after the geographical origin information), vouchers [voucher specimens are deposited in the following herbaria: Universidade Federal do Rio Grande do Sul (ICN) and Université Paris-Sud (UPS); thirteen samples are only cultivated in the R. & E. Heaton NCCPG Sisyrinchium collection without voucher deposited], and GenBank accession number (rpoC1, rpoB, matK, trnH-psbA, trnQ-rps16, nad1-2/3, nad4-1/2, ITS). Names in parentheses are names that are not accepted by the World Checklist of Monocots (Barker, 2004). ‘na’ indicates no sequence available for the accession.
Species; Geographical origin; Voucher (Herbarium); GenBank accessions: rpoC1, rpoB, matK, trnH-psbA, trnQ-rps16, nad1-2/3, nad4-1/2, ITS.
Orthrosanthus monadelphus Ravenna (SP174); Mexico: Oaxaca (cultivated); Chauveau & Pautz H09049 (UPS); HQ606558, HQ606668, HQ606778, HQ606888, HQ606998, HQ607216, HQ607326, na. Solenomelus pedunculatus (Gillies ex Hook.) Hoch. (SP221); Chile: Region V (cultivated); Chauveau H09044 (UPS); HQ606679, HQ606789, HQ606899, HQ607009, HQ607227, HQ607337, HQ607117. Solenomelus segethi (Phil.) Kuntze (SP005); Argentina: Neuquen; Chauveau & Aubert H09001 (ICN); HQ606486, HQ606596, HQ606706, HQ606816, HQ606926, HQ607144, HQ607254, HQ607036. Olsynium biflorum (Thunb.) Goldblatt (SP006); Argentina: Santa Cruz (cultivated); R. & E. Heaton OLS100·03 (living collection); HQ606487, HQ606597, HQ606707, HQ606817, HQ606927, HQ607145, HQ607255, HQ607037. Olsynium douglasii (A.Dietr.) E.P.Bicknell (SP007); Canada: British Columbia (cultivated); R. & E. Heaton OLS100·02 (living collection); HQ606488, HQ606598, HQ606708, HQ606818, HQ606928, HQ607146, HQ607256, na. Olsynium junceum (E.Mey. ex C.Presl) Goldblatt (SP191); Argentina: Chubut; Chauveau & Aubert H09026 (ICN); HQ606568, HQ606678, HQ606788, HQ606898, HQ607008, HQ607226, HQ607336, HQ607116. Olsynium junceum (E.Mey. ex C.Presl) Goldblatt (SP837); Argentina: Mendoza; Chauveau & Aubert H09050 (ICN); HQ606593, HQ606703, HQ606813, HQ606923, HQ607033, HQ607251, HQ607361, HQ607141. Olsynium junceum ssp. colchaguense (Phil.) J.M.Watson & A.R.Flores (SP838); Argentina: Rio Negro; Chauveau & Aubert H09051 (ICN); HQ606594, HQ606704, HQ606814, HQ606924, HQ607034, HQ607252, HQ607362, HQ607142. Olsynium scirpoideum (Poepp.) Goldblatt (SP009); Chile: Region Metropolitana (cultivated); R. & E. Heaton OLS104·17 (living collection); HQ606489, HQ606599, HQ606709, HQ606819, HQ606929, HQ607147, HQ607257, HQ607038. Sisyrinchium acre H.Mann (SP010); USA: Hawaii (cultivated); Chauveau & Heaton H09053 (UPS); HQ606490, HQ606600, HQ606710, HQ606820, HQ606930, HQ607148, HQ607258, HQ607039. (Sisyrinchium alatum Hook.) (SP100); Brazil: Santa Catarina; Eggers & Souza-Chies 232 (ICN); HQ606536, HQ606646, HQ606756, HQ606866, HQ606976, HQ607194, HQ607304, HQ607085. Sisyrinchium angustifolium Mill. (SP014); USA (cultivated); Chauveau H09002 (ICN); HQ606491, HQ606601, HQ606711, HQ606821, HQ606931, HQ607149, HQ607259, HQ607040. Sisyrinchium angustissimum (B.L.Rob. & Greenm.) Greenm. & C.H.Thomps. (SP032); Guatemala (cultivated); Chauveau H09008 (UPS); HQ606502, HQ606612, HQ606722, HQ606832, HQ606942, HQ607160, HQ607270, HQ607051. Sisyrinchium arenarium Poepp. (SP819); Chile: Region VII (cultivated); Chauveau H09047 (UPS); HQ606589, HQ606699, HQ606809, HQ606919, HQ607029, HQ607247, HQ607357, HQ607137. Sisyrinchium arizonicum Roth. (SP236); USA (cultivated); Chauveau H09032 (UPS); HQ606577, HQ606687, HQ606797, HQ606907, HQ607017, HQ607235, HQ607345, HQ607125. Sisyrinchium avenaceum Klatt (SP018); Brazil: Rio Grande do Sul; Eggers & Souza-Chies 280 (ICN); HQ606494, HQ606604, HQ606714, HQ606824, HQ606934, HQ607152, HQ607262, HQ607043. Sisyrinchium balansae Baker (SP141); Brazil: Paraná; Eggers & Souza-Chies 364 (ICN); HQ606548, HQ606658, HQ606768, HQ606878, HQ606988, HQ607206, HQ607316, HQ607097. Sisyrinchium bellum S.Watson (SP019); USA: California (cultivated); Eggers, Chauveau & Heaton H09055 (UPS); HQ606495, HQ606605, HQ606715, HQ606825, HQ606935, HQ607153, HQ607263, HQ607044. Sisyrinchium aff. bellum S.Watson (SP020); USA (cultivated); Chauveau H09005 (UPS); HQ606496, HQ606606, HQ606716, HQ606826, HQ606936, HQ607154, HQ607264, HQ607045. Sisyrinchium bermudiana L. (SP021); Bermuda (cultivated); R. & E. Heaton SIS156·00 (living collection); HQ606497, HQ606607, HQ606717, HQ606827, HQ606937, HQ607155, HQ607265, HQ607046. (Sisyrinchium brachypus (E.P.Bicknell) J.K.Henry) (SP023); USA: California (cultivated); R. & E. Heaton SIS102·01 (living collection); HQ606499, HQ606609, HQ606719, HQ606829, HQ606939, HQ607157, HQ607267, HQ607048. Sisyrinchium brasiliense (Ravenna) Ravenna (SP147); Brazil: Paraná; Eggers & Souza-Chies 379 (ICN); HQ606550, HQ606660, HQ606770, HQ606880, HQ606990, HQ607208, HQ607318, HQ607099. Sisyrinchium bromelioides R.C.Foster ssp. bromelioides (SP163); Brazil: Santa Catarina; Eggers & Souza-Chies 410 (ICN); HQ606556, HQ606666, HQ606776, HQ606886, HQ606996, HQ607214, HQ607324, HQ607105. (Sisyrinchium cf. burchellii Baker) (SP138); Brazil: Paraná; Eggers & Souza-Chies 361 (ICN); HQ606547, HQ606657, HQ606767, HQ606877, HQ606987, HQ607205, HQ607315, HQ607096. Sisyrinchium caeteanum Ravenna (SP102); Brazil: Santa Catarina; Eggers & Souza-Chies 224 (ICN); HQ606537, HQ606647, HQ606757, HQ606867, HQ606977, HQ607195, HQ607305, HQ607086. Sisyrinchium californicum (Ker Gawl.) Dryand. (SP022); USA (cultivated); Chauveau H09006 (ICN); HQ606498, HQ606608, HQ606718, HQ606828, HQ606938, HQ607156, HQ607266, HQ607047. Sisyrinchium aff. californicum (Ker Gawl.) Dryand. (SP809); Mexico (cultivated); Eggers, Chauveau & Heaton H09040 (ICN); HQ606584, HQ606694, HQ606804, HQ606914, HQ607024, HQ607242, HQ607352, HQ607132. Sisyrinchium campestre E.P.Bicknell (SP234); USA (cultivated); Chauveau H09030 (UPS); HQ606575, HQ606685, HQ606795, HQ606905, HQ607015, HQ607233, HQ607343, HQ607123. Sisyrinchium chapelcoense Ravenna (SP017); Argentina: Neuquen; Chauveau & Aubert H09004 (ICN); HQ606493, HQ606603, HQ606713, HQ606823, HQ606933, HQ607151, HQ607261, HQ607042. Sisyrinchium chilense Hook. (SP238); Peru: Apurimac (cultivated); Eggers, Chauveau & Heaton H09054 (UPS); HQ606579, HQ606689, HQ606799, HQ606909, HQ607019, HQ607237, HQ607347, HQ607127. Sisyrinchium claritae Herter (SP028); Brazil: Rio Grande do Sul; Eggers & Souza-Chies 267 (ICN); HQ606500, HQ606610, HQ606720, HQ606830, HQ606940, HQ607158, HQ607268, HQ607049. Sisyrinchium commutatum Klatt ssp. commutatum (SP031); Brazil: Paraná; Eggers & Souza-Chies 245 (ICN); HQ606501, HQ606611, HQ606721, HQ606831, HQ606941, HQ607159, HQ607269, HQ607050. (Sisyrinchium commutatum ssp. capillare (Baker) Ravenna) (SP133); Brazil: Paraná; Eggers & Souza-Chies 351 (ICN); HQ606545, HQ606655, HQ606765, HQ606875, HQ606985, HQ607203, HQ607313, HQ607094. Sisyrinchium convallium Ravenna (SP232); Argentina: Chubut; Chauveau & Aubert H09028 (ICN); HQ606573, HQ606683, HQ606793, HQ606903, HQ607013, HQ607231, HQ607341, HQ607121. Sisyrinchium convolutum Nocca (SP108); Mexico: Mexico (cultivated); Eggers, Chauveau & Heaton H09019 (UPS); HQ606538, HQ606648, HQ606758, HQ606868, HQ606978, HQ607196, HQ607306, HQ607087. Sisyrinchium cuspidatum Poepp. (SP038); Chile: Region IV (cultivated); Eggers, Chauveau & Aubert H09011 (ICN); HQ606505, HQ606615, HQ606725, HQ606835, HQ606945, HQ607163, HQ607273, HQ607054. Sisyrinchium demissum Greene (SP034); USA: Arizona; Chauveau H09009 (ICN); HQ606503, HQ606613, HQ606723, HQ606833, HQ606943, HQ607161, HQ607271, HQ607052. Sisyrinchium densiflorum Ravenna (SP119); Brazil: Santa Catarina; Eggers & Souza-Chies 321 (ICN); HQ606540, HQ606650, HQ606760, HQ606870, HQ606980, HQ607198, HQ607308, HQ607089. Sisyrinchium elmeri Greene (SP813); USA: California (cultivated); Chauveau & Heaton H09042 (UPS); HQ606586, HQ606696, HQ606806, HQ606916, HQ607026, HQ607244, HQ607354, HQ607134. (Sisyrinchium fasciculatum Klatt) (SP152); Brazil: Paraná; Eggers & Souza-Chies 391 (ICN); HQ606552, HQ606662, HQ606772, HQ606882, HQ606992, HQ607210, HQ607320, HQ607101. Sisyrinchium fiebrigii I.M.Johnst. (SP121); Brazil: Paraná; Eggers & Souza-Chies 325 (ICN); HQ606541, HQ606651, HQ606761, HQ606871, HQ606981, HQ607199, HQ607309, HQ607090. (Sisyrinchium aff. flaccidum E.P.Bicknell) (SP817); USA (cultivated); Eggers, Chauveau & Heaton H09043 (ICN); HQ606588, HQ606698, HQ606808, HQ606918, HQ607028, HQ607246, HQ607356, HQ607136. Sisyrinchium funereum E.P.Bicknell (SP035); USA: California (cultivated); R. & E. Heaton SIS126·00 (living collection); HQ606504, HQ606614, HQ606724, HQ606834, HQ606944, HQ607162, HQ607272, HQ607053. Sisyrinchium graminifolium Lindl. (SP183); Chile: Region VII (cultivated); Eggers & Chauveau H09024 (UPS); HQ606563, HQ606673, HQ606783, HQ606893, HQ607003, HQ607221, HQ607331, HQ607111. Sisyrinchium aff. graminifolium Lindl. (SP821); Chile: Region V (cultivated); Chauveau H09046 (UPS); HQ606590, HQ606700, HQ606810, HQ606920, HQ607030, HQ607248, HQ607358, HQ607138. Sisyrinchium aff. graminifolium Lindl. (SP826); Chile: Region IV (cultivated); Chauveau H09045 (UPS); HQ606591, HQ606701, HQ606811, HQ606921, HQ607031, HQ607249, HQ607359, HQ607139. Sisyrinchium hasslerianum Baker (SP146); Brazil: Paraná; Eggers & Souza-Chies 377 (ICN); HQ606549, HQ606659, HQ606769, HQ606879, HQ606989, HQ607207, HQ607317, HQ607098. Sisyrinchium hoehnei I.M.Johnst. (SP135); Brazil: Paraná; Eggers & Souza-Chies 355 (ICN); HQ606546, HQ606656, HQ606766, HQ606876, HQ606986, HQ607204, HQ607314, HQ607095. Sisyrinchium idahoense E.P.Bicknell (SP040); USA: California (cultivated); R. & E. Heaton SIS183·00 (living collection); HQ606506, HQ606616, HQ606726, HQ606836, HQ606946, HQ607164, HQ607274, HQ607055. Sisyrinchium idahoense var. macounii (E.P.Bicknell) Douglass M.Hend. (SP041); USA (cultivated); R. & E. Heaton SIS180·01 (living collection); HQ606507, HQ606617, HQ606727, HQ606837, HQ606947, HQ607165, HQ607275, HQ607056. Sisyrinchium jamesonii Baker (SP237); Peru: Apumirac (cultivated); Chauveau & Heaton H09052 (UPS); HQ606578, HQ606688, HQ606798, HQ606908, HQ607018, HQ607236, HQ607346, HQ607126. Sisyrinchium laxum Otto ex Sims (SP049); Argentina: Neuquen; Chauveau & Aubert H09014 (ICN); HQ606510, HQ606620, HQ606730, HQ606840, HQ606950, HQ607168, HQ607278, HQ607059. Sisyrinchium littorale Greene (SP042); Canada (cultivated); Chauveau H09012 (ICN); HQ606508, HQ606618, HQ606728, HQ606838, HQ606948, HQ607166, HQ607276, HQ607057. Sisyrinchium longipes (E.P.Bicknell) Kearney & Peebles (SP181); USA (cultivated); Chauveau H09023 (ICN); HQ606562, HQ606672, HQ606782, HQ606892, HQ607002, HQ607220, HQ607330, HQ607110. Sisyrinchium luzula Klotzsch ex Klatt (SP128); Brazil: Paraná; Eggers & Souza-Chies 341 (ICN); HQ606543, HQ606653, HQ606763, HQ606873, HQ606983, HQ607201, HQ607311, HQ607092. Sisyrinchium macrocarpum Hieron. ssp. macrocarpum (SP046); Argentina (cultivated); Chauveau H09013 (UPS); HQ606509, HQ606619, HQ606729, HQ606839, HQ606949, HQ607167, HQ607277, HQ607058. (Sisyrinchium macrocarpum ssp. laetum Ravenna – unplaced name) (SP235); Argentina (cultivated); Chauveau H09031 (UPS); HQ606576, HQ606686, HQ606796, HQ606906, HQ607016, HQ607234, HQ607344, HQ607124. Sisyrinchium cf. macranthum Griseb. (SP239); Argentina (cultivated); R. & E. Heaton SIS900·166 (living collection); HQ606580, HQ606690, HQ606800, HQ606910, HQ607020, HQ607238, HQ607348, HQ607128. Sisyrinchium macrophyllum Greenm. (SP050); Mexico: Mexico (cultivated); Chauveau & Heaton H09015 (UPS); HQ606511, HQ606621, HQ606731, HQ606841, HQ606951, HQ607169, HQ607279, HQ607060. Sisyrinchium mandonii Baker (SP807); Venezuela (cultivated); Chauveau & Heaton H09039 (ICN); HQ606583, HQ606693, HQ606803, HQ606913, HQ607023, HQ607241, HQ607351, HQ607131. (Sisyrinchium marchio (Vell.) Steud.) (SP161); Brazil: Santa Catarina; Eggers & Souza-Chies 407 (ICN); HQ606555, HQ606665, HQ606775, HQ606885, HQ606995, HQ607213, HQ607323, HQ607104. Sisyrinchium cf. marchioides Ravenna (SP160); Brazil: Santa Catarina; Eggers & Souza-Chies 406 (ICN); HQ606554, HQ606664, HQ606774, HQ606884, HQ606994, HQ607212, HQ607322, HQ607103. Sisyrinchium megapotamicum Malme (SP052); Brazil: Santa Catarina; Eggers & Souza-Chies 236 (ICN); HQ606512, HQ606622, HQ606732, HQ606842, HQ606952, HQ607170, HQ607280, HQ607061. Sisyrinchium miamense E.P.Bicknell (SP053); USA: Florida (cultivated); R. & E. Heaton SIS176·00 (living collection); HQ606513, HQ606623, HQ606733, HQ606843, HQ606953, HQ607171, HQ607281, HQ607062. Sisyrinchium micranthum Cav. (SP839); Uruguay: Colonia; Chauveau & Aubert H09052 (ICN); HQ606595, HQ606705, HQ606815, HQ606925, HQ607035, HQ607253, HQ607363, HQ607143. Sisyrinchium micranthum Cav. Type B (SP190); Brazil: Santa Catarina; Eggers & Souza-Chies 251 (ICN); HQ606567, HQ606677, HQ606787, HQ606897, HQ607007, HQ607225, HQ607335, HQ607115. Sisyrinchium micranthum Cav. Type BY (SP186); Brazil: Rio Grande do Sul; Eggers & Souza-Chies 266-F (ICN); HQ606565, HQ606675, HQ606785, HQ606895, HQ607005, HQ607223, HQ607333, HQ607113. Sisyrinchium micranthum Cav. Type GL (SP060); Brazil: Rio Grande do Sul; Eggers & Souza-Chies 261-K (ICN); HQ606517, HQ606627, HQ606737, HQ606847, HQ606957, HQ607175, HQ607285. Sisyrinchium micranthum Cav. Type LLP (SP062); Brazil: Santa Catarina; Eggers & Souza-Chies 234 (ICN); HQ606518, HQ606628, HQ606738, HQ606848, HQ606958, HQ607176, HQ607286, HQ607067. Sisyrinchium micranthum Cav. Type LLY (SP059); Brazil: Paraná; Eggers & Souza-Chies 244-E (ICN); HQ606516, HQ606626, HQ606736, HQ606846, HQ606956, HQ607174, HQ607284, HQ607065. Sisyrinchium micranthum Cav. Type LLY (SP056); Mexico, Chiapas (cultivated); Chauveau & Heaton H09054 (UPS); HQ606515, HQ606625, HQ606735, HQ606845, HQ606955, HQ607173, HQ607283, HQ607064. Sisyrinchium micranthum Cav. Type S (SP189); Brazil: Santa Catarina; Eggers & Souza-Chies 229-A (ICN); HQ606566, HQ606676, HQ606786, HQ606896, HQ607006, HQ607224, HQ607334, HQ607114. Sisyrinchium micranthum Cav. Type T (SP054); Brazil: Paraná; Eggers & Souza-Chies 242-A (ICN); HQ606514, HQ606624, HQ606734, HQ606844, HQ606954, HQ607172, HQ607282, HQ607063. Sisyrinchium minus Engelm. & A.Gray (SP063); Brazil: Santa Catarina; Eggers & Souza-Chies 230 (ICN); HQ606519, HQ606629, HQ606739, HQ606849, HQ606959, HQ607177, HQ607287, HQ607068. Sisyrinchium minutiflorum Klatt (SP066); Brazil: Rio Grande do Sul; Eggers & Souza-Chies 285 (ICN); HQ606520, HQ606630, HQ606740, HQ606850, HQ606960, HQ607178, HQ607288, HQ607069. Sisyrinchium montanum Greene (SP068); Canada: British Columbia (cultivated); R. & E. Heaton SIS114·02 (living collection); HQ606521, HQ606631, HQ606741, HQ606851, HQ606961, HQ607179, HQ607289, HQ607070. Sisyrinchium nashii E.P.Bicknell (SP233); unknown origin, cultivated in UPS Bot. Gard.; Chauveau H09029 (ICN); HQ606574, HQ606684, HQ606794, HQ606904, HQ607014, HQ607232, HQ607342, HQ607122. (Sisyrinchium cf. nervosum Phil.) (SP016); Argentina: Neuquen; Chauveau & Aubert H09003 (ICN); HQ606492, HQ606602, HQ606712, HQ606822, HQ606932, HQ607150, HQ607260, HQ607041. Sisyrinchium nidulare (Hand.-Mazz.) I.M.Johnst. (SP071); Brazil: Paraná; Eggers & Souza-Chies 240 (ICN); HQ606522, HQ606632, HQ606742, HQ606852, HQ606962, HQ607180, HQ607290, HQ607071. Sisyrinchium ostenianum Beauverd (SP170); Brazil: Rio Grande do Sul; Eggers & Souza-Chies 434 (ICN); HQ606557, HQ606667, HQ606777, HQ606887, HQ606997, HQ607215, HQ607325, HQ607106. Sisyrinchium pachyrhizum Baker (SP073); Brazil: Rio Grande do Sul; Eggers & Souza-Chies 281 (ICN); HQ606523, HQ606633, HQ606743, HQ606853, HQ606963, HQ607181, HQ607291, HQ607072. Sisyrinchium palmifolium L. ssp. palmifolium (SP175); unknown origin, cultivated in UPS Bot. Gard.; Chauveau H09020 (UPS); HQ606559, HQ606669, HQ606779, HQ606889, HQ606999, HQ607217, HQ607327, HQ607107. Sisyrinchium palmifolium ssp. fuscoviride (Ravenna) Ravenna (SP076); Argentina (cultivated); R. & E. Heaton SIS199·00 (living collection); HQ606524, HQ606634, HQ606744, HQ606854, HQ606964, HQ607182, HQ607292, HQ607073. Sisyrinchium papillosum R.C.Foster (SP249); Chile (cultivated); Chauveau H09037 (ICN); HQ606582, HQ606692, HQ606802, HQ606912, HQ607022, HQ607240, HQ607350, HQ607130. Sisyrinchium parvifolium Baker (SP224); Brazil: Santa Catarina; Eggers & Souza-Chies 237-C (ICN); HQ606571, HQ606681, HQ606791, HQ606901, HQ607011, HQ607229, HQ607339, HQ607119. Sisyrinchium patagonicum Phil. ex Baker (SP223); Chile (cultivated); Chauveau H09027 (ICN); HQ606570, HQ606680, HQ606790, HQ606900, HQ607010, HQ607228, HQ607338, HQ607118. Sisyrinchium pearcei Phil. (SP180); Chile: Region IX (cultivated); Chauveau H09022 (ICN); HQ606561, HQ606671, HQ606781, HQ606891, HQ607001, HQ607219, HQ607329, HQ607109. Sisyrinchium platense I.M.Johnst. (SP084); Brazil: Rio Grande do Sul; Eggers & Souza-Chies 187 (ICN); HQ606526, HQ606636, HQ606746, HQ606856, HQ606966, HQ607184, HQ607294, HQ607075. Sisyrinchium pruinosum E.P.Bicknell (SP816); USA: Texas (cultivated); Eggers, Chauveau & Heaton H09048 (UPS); HQ606587, HQ606697, HQ606807, HQ606917, HQ607027, HQ607245, HQ607355, HQ607135. Sisyrinchium purpurellum ssp. trichospathum Ravenna (SP132); Brazil: Paraná; Eggers & Souza-Chies 350 (ICN); HQ606544, HQ606654, HQ606764, HQ606874, HQ606984, HQ607202, HQ607312, HQ607093. Sisyrinchium rambonis R.C.Foster (SP835); Brazil: Rio Grande do Sul; Eggers & Souza-Chies 571 (ICN); HQ606592, HQ606702, HQ606812, HQ606922, HQ607032, HQ607250, HQ607360, HQ607140. Sisyrinchium rectilineum Ravenna (SP123); Brazil: Paraná; Eggers & Souza-Chies 332 (ICN); HQ606542, HQ606652, HQ606762, HQ606872, HQ606982, HQ607200, HQ607310, HQ607091. Sisyrinchium restioides Spreng. (SP097); Brazil: Santa Catarina; Eggers & Souza-Chies 252 (ICN); HQ606534, HQ606644, HQ606754, HQ606864, HQ606974, HQ607192, HQ607302, HQ607083. Sisyrinchium rosulatum E.P.Bicknell (SP085); USA: Georgia (cultivated); ICN 187084 (ICN); HQ606527, HQ606637, HQ606747, HQ606857, HQ606967, HQ607185, HQ607295, HQ607076. Sisyrinchium sarmentosum Suksd. ex Greene (SP086); USA (cultivated); Chauveau H09017 (ICN); HQ606528, HQ606638, HQ606748, HQ606858, HQ606968, HQ607186, HQ607296, HQ607077. Sisyrinchium scabrum Schltdl. & Cham. (SP088); cultivated (material provided by the Royal Botanic Gardens, Kew – 05259); R. & E. Heaton SIS114·02 (living collection); HQ606530, HQ606640, HQ606750, HQ606860, HQ606970, HQ607188, HQ607298, HQ607079. Sisyrinchium scariosum I.M.Johnst. (SP087); Brazil: Rio Grande do Sul; Eggers & Souza-Chies 277 (ICN); HQ606529, HQ606639, HQ606749, HQ606859, HQ606969, HQ607187, HQ607297, HQ607078. Sisyrinchium sellowianum Klatt (SP091); Brazil: Santa Catarina; Eggers & Souza-Chies 253 (ICN); HQ606531, HQ606641, HQ606751, HQ606861, HQ606971, HQ607189, HQ607299, HQ607080. Sisyrinchium setaceum Klatt (SP092); Brazil: Rio Grande do Sul; Eggers & Souza-Chies 214 (ICN); HQ606532, HQ606642, HQ606752, HQ606862, HQ606972, HQ607190, HQ607300, HQ607081. Sisyrinchium soboliferum Ravenna (SP148); Brazil: Paraná; Eggers & Souza-Chies 381 (ICN); HQ606551, HQ606661, HQ606771, HQ606881, HQ606991, HQ607209, HQ607319, HQ607100. Sisyrinchium striatum Sm. (SP095); Chile (cultivated); Chauveau H09018 (ICN); HQ606533, HQ606643, HQ606753, HQ606863, HQ606973, HQ607191, HQ607301, HQ607082. Sisyrinchium aff. strictum E.P.Bicknell (SP811); USA (cultivated); Chauveau & Heaton H09041 (UPS); HQ606585, HQ606695, HQ606805, HQ606915, HQ607025, HQ607243, HQ607353, HQ607133. Sisyrinchium tenuifolium Humb. & Bonpl. ex Willd. (SP184); Mexico: Veracruz (cultivated); Chauveau & Pautz H09025 (ICN); HQ606564, HQ606674, HQ606784, HQ606894, HQ607004, HQ607222, HQ607332, HQ607112. Sisyrinchium tinctorium Kunth (SP240); Ecuador (cultivated); Chauveau & Heaton H09034 (ICN); HQ606581, HQ606691, HQ606801, HQ606911, HQ607021, HQ607239, HQ607349, HQ607129. Sisyrinchium aff. tinctorium Kunth (SP179); cultivated in UPS Bot. Gard. (material provided by MNHN Bot. Gard. 46620); Eggers & Chauveau H09021 (UPS); HQ606560, HQ606670, HQ606780, HQ606890, HQ607000, HQ607218, HQ607328, HQ607108. Sisyrinchium uliginosum Ravenna (SP154); Brazil: Paraná; Eggers & Souza-Chies 393 (ICN); HQ606553, HQ606663, HQ606773, HQ606883, HQ606993, HQ607211, HQ607321, HQ607102. Sisyrinchium vaginatum Spreng. ssp. vaginatum (SP226); Brazil: Rio Grande do Sul; Eggers & Souza-Chies 263 (ICN); HQ606572, HQ606682, HQ606792, HQ606902, HQ607012, HQ607230, HQ607340, HQ607120. Sisyrinchium cf. valparadiseum Ravenna (SP082); Argentina, Rio Negro; Chauveau & Aubert H09016 (ICN); HQ606525, HQ606635, HQ606745, HQ606855, HQ606965, HQ607183, HQ607293, HQ607074. Sisyrinchium weirii Baker (SP099); Brazil: Paraná; Eggers & Souza-Chies 248 (ICN); HQ606535, HQ606645, HQ606755, HQ606865, HQ606975, HQ607193, HQ607303, HQ607084. Sisyrinchium wettsteinii Hand.-Mazz. (SP115); Brazil: Santa Catarina; Eggers & Souza-Chies 250 (ICN); HQ606539, HQ606649, HQ606759, HQ606869, HQ606979, HQ607197, HQ607307, HQ607088.