| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Systematics and Phytogeography |
2Department of Plant Biology, University of Georgia, 2502 Plant Sciences, Athens, Georgia 30602-7271 USA; 3Department of Natural History, Florida Museum of Natural History, University of Florida, P.O. Box 117800, Gainesville, Florida 32611-7800 USA; 4Department of Botany, University of Florida, P.O. Box 118526, Gainesville, Florida 32611-8526 USA
Received for publication December 13, 2005. Accepted for publication May 15, 2006.
ABSTRACT
As currently defined, the 24 species of Schoenocaulon occur in three disjunct areas: north central Florida (one species, S. dubium), southern Peru (portion of the range of S. officinale), and the region from southeastern New Mexico–Texas south to Venezuela; the 20 species endemic to Mexico are geographically restricted. Species delimitations, often based on tepal morphology, have been problematic. Our analyses of ITS sequence data for all 27 species and infraspecific taxa support recognition of two new species and recircumscription and placement of elements of the polyphyletic S. ghiesbreghtii and S. mortonii complexes. For taxa with adequate sampling, our data also indicate 11–12 cladospecies and 3–6 metaspecies according to the apomorphic species concept. The resolved phylogeny, correlated with geography and morphology, allows insight into biogeographical diversification and the evolution of some unusual morphological characters within the genus, such as nectary differentiation and tepal margin type.
Key Words: ITS • Liliales • Melanthiaceae • Mexico • molecular phylogeny • Schoenocaulon
The current definitive work on Schoenocaulon A. Gray comprises studies by Frame (1989
, 1990
, 2001
, 2002
; Frame et al., 1999
; López-Ferrari et al., 2000
); the 27 taxa (24 species and three varieties), listed in Table 1, form the basis of this study. These species are narrowly endemic, generally inhabiting xeric ecosystems (barrens, prairies, alpine grasslands, and pine–oak forests) in three disjunct geographic areas (Fig. 1). Twenty-three species occur in southeastern New Mexico–Texas south to Venezuela, and S. officinale also occurs as a disjunct in southern Peru. The 20 Mexican endemics (Table 1) are geographically restricted, often to a single mountain range, and the three widespread species in that area (at least S. officinale and some elements of the S. ghiesbreghtii complex) may have been distributed by Native Americans for medicinal purposes (Frame, 1990
; Zomlefer, 1997
). The remaining disjunct, S. dubium, is restricted to north central to southeastern peninsular Florida (Wunderlin and Hansen, 2003
, 2005
). Reports of this species as possibly occurring in Georgia (e.g., Gray, 1837
; Gates, 1918
; de Zerpa, 1951
; Duncan and Kartesz, 1981
), as originally cited by Michaux (1803)
, have not been verified by herbarium voucher specimens (Jones and Coile, 1988
; Frame, 1990
; W. B. Zomlefer, personal observation).
|
|
), tribe Melanthieae (Amianthium, Anticlea, Stenanthium, Toxicoscordion, Veratrum s.l., Zigadenus)—a placement well established with molecular and morphological data (see summaries in Zomlefer, 1997
, 1999
; Zomlefer et al., 2001
, 2003
, 2006
; Zomlefer and Judd, 2002
). The monophyly of Schoenocaulon is well supported by several morphological synapomorphies: a distinctive fibrous-tunicate bulb terminating a reduced/rudimentary (to absent) rhizome covered by dark brown to black scales and fibers (Fig. 2A); a relatively long, bottlebrush-like, spicate (to somewhat racemose) inflorescence of tiny, sessile to subsessile flowers with conspicuously exserted stamens (Fig. 2B and C); and appendaged seeds (Fig. 2I; Zomlefer, 1997
; Zomlefer et al., 2001
, 2004
, 2006
). Schoenocaulon is also well defined karyologically and chemically. The homogeneity of Schoenocaulon chromosomes (e.g., Fig. 2K) in terms of number (x = 8), ploidy (2n), size (1.5–3.0 µm), and morphology (e.g., distinct kinetochores; interphase heteropycnotic segments) has been confirmed for 11 species (de Zerpa, 1951
; Preece, 1956
; Cave, 1967
; Tamura, 1995
; Frame, 1990
, 2001
).
|
; Frame, 2001
). The many highly endemic Schoenocaulon species (mainly in remote areas of Mexico) have been rarely collected, and then predominantly by collectors circa 1870–1910 under very unfavorable conditions (Iltis, 1998
). In addition, the vegetative habit is relatively uniform, and variation in floral and fruit morphology is generally subtle (Brinker, 1942
; Frame, 1990
). Perianth morphology has typically been used to distinguish species. Tepals vary from elliptic-ovate with erose margins (e.g., S. yucatanense, Fig. 2F) to ligulate with entire margins, as in S. officinale (Fig. 2E). In addition, several species (such as S. texanum, Fig. 2G) have one or two conspicuous hyaline auricles (flanges or teeth) along each side at the tepal base. The melanthioid perigonal gland is typically represented by nectariferous tissue lining a shallow to conspicuous concavity at the tepal base as in Fig. 2D and F; however, a smooth nectariferous zone on each tepal characterizes several species (e.g., S. texanum, Fig. 2G), and a pad-like gland occurs in S. officinale (Fig. 2E; Zomlefer, 1997
). Depending on the species, the flowers may be sessile (Fig. 2C) to subpedicellate when the pollen is shed. Characters of the inflorescence, particularly length and number of flowers/flower ranks, have also been used to define species (Frame et al., 1999
). These floral (and other) character states, however, identified as autapomorphies for particular species, have not been examined in the context as possible synapomorphies to support relationships between species/species groups.
Current investigation
Prior to our investigations, Schoenocaulon had not been the subject of phylogenetic analyses at any taxonomic level. In our previous study of Melanthieae (Zomlefer et al., 2001
), combined ITS and trnL-F data strongly support the position of Schoenocaulon as sister to most of the tribe, i.e., to Toxicoscordion (Amianthium/Veratrum–Stenanthium/Anticlea) with Zigadenus as sister to the entire tribe. That study comprised 29 representatives of Melanthieae, including two of Schoenocaulon (one sample each of S. dubium and S. texanum).
Our current investigation, focused on Schoenocaulon, involved increased sampling to represent all 24 species and the three varieties (Table 1). This was possible only with permission from various herbarium personnel (see Acknowledgments) to sample specimens of these many rarely collected species for molecular analyses. We performed parsimony and Bayesian analyses of sequence data from the internal transcribed spacer region ITS-1, 5.8S, and ITS-2 (together referred to as ITS nuclear ribosomal DNA or nrDNA), a region often used for determining generic and species relationships (summaries in Baldwin et al. [1995
] and Soltis and Soltis [1998
]). From our analyses of molecular data correlated with geographical distribution, we (1) produce an infrageneric phylogeny of Schoenocaulon, (2) evaluate the monophyly of several species complexes, (3) determine the placement of new species and other problematic ones, (4) develop hypotheses concerning biogeographical diversification and morphological evolution, and (5) identify clades requiring further study.
MATERIALS AND METHODS
Plant material
Table 1 is a list of taxa examined and voucher information. The 68 samples of Schoenocaulon encompass all 24 species and the three varieties. Widespread and/or especially variable species (e.g., when infraspecific taxa have been recognized) were represented by several accessions (i.e., [Operational Taxonomic Units, OTUs]) whenever possible (see Kron and Judd, 1997
). Herbarium specimens, ranging in age from 12 to 128 years (indicated in Table 1), provided most material, with the exception of one live plant of S. dubium (Zomlefer 673) and three silica-preserved samples (Chase and Hills, 1991
) of S. texanum (Sivinski 4712, Wendt 7016, Worthington 28255). Successful extractions involved leaf blades and/or tepals. We postulate that the success in amplifying old Schoenocaulon specimens (27 sheets over 50 years old; 11 of these at least 100 years old) is due to their extreme xerophytic habit: their natural drought tolerance may protect the DNA from degradation during the drying process (W. M. Whitten, personal observation; Zomlefer et al., 2004
).
Molecular techniques
DNA was extracted from plant material according to the methods of Doyle and Doyle (1987)
, scaled down to 1.0-mL extraction volumes. DNA was precipitated overnight at –20°C with 0.65 volumes of isopropanol, centrifuged, washed twice with 70% ethanol, and dried. The pellet was resuspended in 75 µL of Tris-EDTA buffer and stored at –20°C. Total DNA from herbarium specimens was purified using QIAquick columns (Qiagen, Santa Clarita, California, USA). Amplification of DNA was generally performed using 50-µL reactions with 35 cycles, 2.5 mmol/L MgCl2, 1.0 mol/L betaine, and a hot start, using Promega (Promega, Madison, Wisconsin, USA) or Epicentre (Epicentre Technologies, Madison, Wisconsin, USA) buffers and Taq polymerase. A touchdown thermal cycling program was used; the initial annealing temperature was 76°C, decreasing 1°C per cycle for 15 cycles, followed by 21 cycles at 59°C. The ITS amplification and sequencing primers depended on the plant material: Sun et al. (1994)
for fresh and silica-dried samples and Blattner (1999)
for herbarium specimens. The PCR products were purified using QIAquick columns and underwent dye terminator cycle sequencing with ABI (Applied Biosystems, Foster City, California, USA) reagents (5-µL reactions).
The final sequencing gel runs were completed by the Interdisciplinary Center for Biotechnology Research (ICBR, University of Florida) with ABI 377 and ABI 373A automated sequencers (Applied Biosystems). Both strands were sequenced to assure accurate base calling. Individual sequences were edited and assembled using the software package Sequencher (GeneCodes, Ann Arbor, Michigan, USA) and aligned manually using Se-Al 2.0 (GeneCodes) on an Apple PowerMac G5 computer. The ends of matrices were trimmed to exclude sequencing artifacts. Sequences are deposited in GenBank (AF303730, AF303731, AY738266–AY738331; see Table 1). The aligned data matrix is also available from the authors (W.B.Z. and W.M.W.; wendyz@plantbio.uga.edu; whitten@flmnh.ufl.edu).
Search strategies
Three sets of cladistic analyses were performed using PAUP* version 4.0b10 (Swofford, 2002
) with all characters weighted equally: (1) ITS data with gaps coded as missing values; (2) "gap analysis" comprising only gaps as characters; and (3) combined analyses of 1 and 2. For analysis 2, a "gap file" (binary matrix) and "gaplist" were generated with PAUPGAP, version 1.12 (Cox, 1997
). We analyzed gap data because the original ITS matrix (analysis 1) indicated several possibly informative indels corresponding to certain clades. The outgroups for Schoenocaulon (Amianthium muscitoxicum [AF303702], Anticlea elegans [AF303725], Stenanthium densum [AF303722], Toxicoscordion nuttallii [AF303718], Veratrum viride [AF303706], Zigadenus glaberrimus [AF303712]) were based on Zomlefer et al. (2001)
. Analyses used the heuristic search option (MULTREES, SPR [subtree pruning-regrafting], 1000 random replicates, holding five trees per replicate). All trees and their statistics (tree length, consistency index [CI], and retention index [RI]) were saved to a log file. A bootstrap consensus tree was generated (1000 replicates, saving five trees per replicate; Felsenstein, 1985
), and we here designate the following categories of bootstrap percentage (BP) support: unsupported (<50%), weak (50–74%), moderate (75–84%), and strong (85–100%). In addition, decay indices (Bremer values) were generated with AutoDecay 4.0.1 (1000 random addition sequence replicates; Eriksson, 1998
).
We performed Bayesian analyses using MrBayes 3.0B (Huelsenbeck and Ronquist, 2001
, 2002
) on the combined data set. The parameters for the Bayesian analyses were as follows: 1set nst = 2 rates = equal; set autoclose = yes; mcmcp ngnen = 2000000; prinfreq = 100; samplefreq = 10; nchains = 4; savebrlens = yes; mcmc; sumt; burnin = 200000; contype = halfcompat. The analysis was repeated with rates = gamma. In each case, the first 10 000 trees were omitted, and the majority rule consensus tree was obtained in PAUP* from the remaining trees.
RESULTS
Table 2 lists tree statistics for the three data sets: ITS data with gaps treated as missing values (hereafter referred to as the "ITS analysis"), "gap matrix" (gaps as binary characters), and the combined (ITS + gap) analysis. The relatively high number of trees is not due to incongruence but to the multiple samples of the same taxon coupled with little ITS variation within certain species complexes. We justify performing the combined analysis by comparison of the bootstrap consensus trees for the three matrices. The combined analysis produced fewer trees (1217) compared to the ITS data alone (2888 trees), with equivalent bootstrap support (11 and eight major clades, respectively, with >85% support) and comparable retention indices (1.66 and 1.67, respectively). These trees do not exhibit hard incongruence, and the ITS (not shown) and combined-data consensus trees (Fig. 3) have similar topologies. Further support for this topology is provided by Bayesian analyses of the combined data set: the resulting tree (not shown) has the same topology with Bayesian posterior probabilities of 100% for all clades having strong bootstrap support in the parsimony analyses.
|
|
Clade A has weak support (BP 73) with combined data, but moderate support (BP 84) with gap analysis. This difference is due to the position of S. comatum sensu stricto (s.s.) (from the type locality, San Luis Potosí, Mexico): a member of clade A and sister to the rest of the clade in these two analyses (Fig. 3), but with ITS data alone, weakly supported (BP 51) as sister to all of Schoenocaulon, a relationship that collapses in the strict consensus (not shown). An element of S. comatum (from Puebla, Mexico) falls in clade B. Within clade A, the monophyletic S. dubium/S. texanum group is strongly supported (combined data BP 100; gap data BP 95) and is sister to the rest of the clade (subclade C: BP 99 and 77, respectively).
Within clade A, subclade C is strongly supported as noted before. In this subclade, the wide-ranging S. ghiesbreghtii element (eight samples) from Texas to Quintana Roo, Mexico, is monophyletic (BP 100 and 96, respectively) and sister to the rest of the subclade (clade H). However, the type entity of this species, represented by five samples restricted to Chiapas, Mexico, is in clade B (see orange blocks in Fig. 3). The rest of subclade C (combined data BP 74) comprises a polytomy of the closely related species S. ignigenum, S. intermedium, S. macrocarpum, and S. plumosum (clade H).
Clade B is strongly supported in the combined analysis (BP 96). Clade B is divided into two strongly supported subclades, D and E, with combined data BPs of 98 and 100, respectively. Within subclade D, the non-type element of S. comatum (from Puebla, Mexico) is sister to the strongly supported clade (BP 93) comprising S. tenuifolium/S. calcicola and S. caricifolium var. oaxacense/S. tenorioi/S. tenue. Schoenocaulon tenuifolium is strongly supported as sister to S. calcicola (BP 93). However, S. caricifolium s.s. (S. caricifolium var. caricifolium) is in subclade F, and the type element of S. comatum is in clade A.
The internal structure of subclade E varies with different analyses: its two main subclades, F and G, have weak bootstrap support (e.g., 60% and 69% with combined data), and relationships within the subclades are generally weakly to moderately resolved in all analyses. Although S. caricifolium var. caricifolium is weakly supported (combined data BP 69) as most closely related to S. rzedowskii/S. obtusum/S. madidorum, the gap data (BP 86) is a species-level synapomorphy, providing further support for its separation from S. caricifolium var. oaxacense in subclade D (see pink shading in Fig. 3). Schoenocaulon mortonii comprises two elements placed in different subclades (yellow blocks in Fig. 3). Schoeonocaulon mortonii s.s. (from the type locality, Michoacán, Mexico) is moderately supported (BP 75) as more closely related to S. pellucidum, S. jaliscence, and S. tigrense than to S. mortonii from México, Mexico, and is sister to S. pellucidum (combined data BP 76; ITS BP 82). The non-type element of S. mortonii is most closely related to (or perhaps conspecific with) S. conzattii (BP 83). The two varieties of S. megarrhizum, var. megarrhizum and var. deminutum, do not form a clade (green blocks in Fig. 3): S. megarrhizum var. deminutum is more closely related to S. mortonii s.s. and S. pellucidum than to S. megarrhizum s.s. Also in clade E is S. ghiesbreghtii s.s. (from the type locality, Chiapas, Mexico), sister to S. pringlei (BP 73), whereas another element of S. ghiesbreghtii (from other Mexican states and Texas; see Fig. 4) has strongly supported placement in clade A.
|
Species recognition in Schoenocaulon
The "phylogenetic species concept" (PSC) is one of several approaches regularly utilized by systematists to define species and has proven to be effective in resolving certain problematic species complexes (e.g., Huck et al., 1989
; Mishler, 1990
; Judd and Karpook, 1993
; Judd et al., 1995
; Guerrero et al., 2004
; Penneys and Judd, 2004
). However, this term, now applied to at least three different criteria, has become ambiguous (see summaries in Baum and Donoghue, 1995
; Wheeler and Meier, 2000
; Judd et al., 2002
). We here employ two specific PSCs to define species of Schoenocaulon: the apomorphic species concept (Donoghue, 1985
; Mishler, 1985
; Mishler and Brandon, 1987
; deQueiroz and Donoghue, 1988;
Mishler and Theriot, 2000
) and the diagnosable species concept (Nixon and Wheeler, 1990
; Davis and Nixon, 1992
; Wheeler and Platnick, 2000
).
Table 3 summarizes the designations of Schoenocaulon species (cladospecies, metaspecies, or paraphyletic) based on the apomorphic species concept using bootstrap support for our three data sets. In our study, we apply the term metaspecies to groupings of populations that are phenetically diagnosable but are not united by molecular apomorphies. These putative metaspecies are here based only on ITS sequence data, and survey of additional DNA regions and/or morphology may reveal synapomorphies. Species indicated as "paraphyletic" are those for which we have molecular evidence that that they are "positively paraphyletic," i.e., certain of their populations are more closely related to other species than they are to other conspecific ones.
|
The ITS analyses provide strong evidence for pairs of cladospecies that are not sisters to one another (Table 3): Schoenocaulon caricifolium (pink blocks in Fig. 3), S. comatum (blue), S. ghiesbreghtii (orange), and S. megarrhizum (green). In addition, the two elements of S. mortonii (yellow) comprise paired, non-sister metaspecies. The recent recognition of several new species of Schoenocaulon—S. ignigenum (Frame et al., 1999
), S. pellucidum (Frame, 1989
), and S. plumosum (Frame et al., 1999
)—is not resolved by our analyses. Schoenocaulon ignigenum (one sample) and S. intermedium (two samples) comprise a clade (i.e., possible cladospecies) with ITS alone and combined data bootstrap support of 100% (see Fig. 3). The clade (ITS BP 75; gap data BP 88; combined data BP 93) with one sample of S. plumosum embedded within paraphyletic S. macrocarpum calls into question the recognition of S. plumosum. Likewise, the separation of S. pellucidum and S. mortonii s.s. (Jalisco, Michoacán) may be problematic; S. conzattii (one sample) with the other element of S. mortonii (México) would be a strongly supported cladospecies (S. conzattii s.l.; ITS BP 81, combined data BP 83).
Examination of morphology, ecology, and biogeography (see following sections, Biogeographical diversification and Morphological evolution) supports the polyphyly of the five species indicated in color in Fig. 3; our strongest evidence, thus far, is for reinstatement of a segregate of S. ghiesbreghtii s.l. and recognition of a geographical disjunct of S. comatum s.l. and a variety of S. caricifolium s.l. as new species. However, the evidence of polyphyly in S. megarrhizum s.l. (i.e., var. megarrhizum not sister to var. deminutum) and in the disjunct populations of S. mortonii s.l. is also compelling, but these groups are more taxonomically difficult. Our work is somewhat hampered by the lack of available material: crucial specimens collected by D. Frame and collaborators, cited by Frame (1989
, 1990
; Frame et al., 1999
), have not yet been deposited at the herbaria (e.g., GH, MICH, NY, UC, US; E. Wood, R. Rabeler, T. Zanoni, B. Ertter, G. Russell, respectively, personal communications). Nonetheless, our conclusion—that these five "species," as recognized by Frame (1989
, 1990
, 2002
; Frame et al., 1999
; López-Ferrari et al., 2000
), are not monophyletic—is well supported. We here summarize our conclusions concerning these five problematic "species." Nomenclatural consequences of our work (including recombinations and new species descriptions) will be addressed in a companion paper (W. B. Zomlefer and W. S. Judd, unpublished manuscript).
Schoenocaulon ghiesbreghtii complex
This species has a convoluted taxonomic history (see Frame [1990]
for summary). Broadly defined, S. ghiesbreghtii has included S. yucatanense (a segregate name sometimes applied to the Mexican Gulf coastal populations; e.g., Brinker, 1942
; Mullin 1994
) and "S. drummondii," an illegitimate name (W. Zomlefer and W. Judd, unpublished manuscript), typically used for Texan populations of this complex (e.g., Correll and Johnston, 1979
; Hatch et al., 1990
; Kartesz, 1994
; Jones et al., 1997
; Diggs et al., 1999
; Turner et al., 2003
)—or even for the entire complex (e.g., Frame, 1989
). Frame (1990
, 2002
; López-Ferrari et al., 2000
) ultimately considered the complex as one highly variable species with two (or possibly three) distinct entities that should be recognized at the intraspecific rank after more detailed analysis (particularly of floral variation). She noted (Frame, 1990
) that collections from Chiapas, Mexico (the type locality; S. ghiesbreghtii s.s.), were particularly robust, growing in rich mountain soil of riparian habitats (elevation 1500–2700 m), and that the remaining populations along the Gulf coast of Texas–Mexico were relatively uniform in morphology (details not given); the Texan plants occur on the Reyosa gravel formation of the Texas Gulf prairie (at or near sea level to 100 m), and the Mexican Gulf populations occur in eastern deciduous tropical forest (Leopold, 1950
; Rzedowski, 1978
), also at relatively lower elevations (70–1500 m).
Besides ecology and distribution, S. ghiesbreghtii s.s. significantly differs morphologically from the rest of the complex; the disjunct Gulf populations (see Fig. 4) of Texas and central Mexico (S. yucatanense s.l.), however, are similar to one another. In addition to larger leaves, flowers, inflorescences, and fruits, S. ghiesbreghtii s.s. can be distinguished from S. yucatanense by the somewhat entire to slightly irregular tepal margins that may comprise double teeth or shallow auricles (vs. erose throughout in S. yucatanense, Fig. 2F); sessile flowers (vs. subsessile); a gynoecium on a short gynophore (vs. sessile); and slightly rugose fruits (vs. smooth). The segregation of S. yucatanense is also supported by our ITS gap data: our eight samples comprise a clade (gap data BP 96) within clade C in Fig. 3, whereas S. ghiesbreghtii s.s. (five samples in clade E) lacks these indels. The morphological divergence between these two DNA-supported clades supports our contention that S. ghiesbreghtii, as circumscribed by Frame (1990
, 2002
; López-Ferrari et al., 2000
), actually represents two cladospecies (Table 3) that are not closely related (Fig. 3).
Two varieties of Schoenocaulon caricifolium
Frame (1990; Frame et al., 1999
) designated the new variety S. caricifolium var. oaxacense for the much less robust plants comprising the southernmost populations of the species, characterized by ecology (scrub oak forests to 2300 m a.s.l. vs. open oak forests to 3300 m a.s.l. for S. caricifolium var. caricifolium); growth type (small colonies of a few plants or growing singly vs. large colonies); and sessile flowers with falcate tepals (vs. short-pedicellate flowers with ligulate tepals). Our examination of the limited material available for this species indicates that the tepals are also shorter and slightly broader in S. caricifolium var. oaxacense. In addition, the nectary of S. caricifolium var. caricifolium consists of an indistinct, smooth, and shiny zone near the tepal base, whereas in S. caricifolium var. oaxacense, the nectariferous tissue lines a conspicuous deep depression. Our ITS gap data (BP 86), a species-level synapomorphy for S. caricifolium s.s. (clade E, Fig. 3), provides further support for its separation from S. caricifolium var. oaxacense (clade D), which lacks these indels. When considered at the specific rank, these two varieties are each likely monophyletic (cladospecies in Table 3).
Two elements of Schoenocaulon comatum
Defined broadly, this species comprises two geographically disjunct entities (see Fig. 4) with the southern populations (Puebla/Oaxaca) comprising more robust plants with larger flowers (Frame 1990
). Our studies confirm that flowers of the southern disjuncts are considerably larger with the tepals ca. 1.5–2.0 times longer than those of plants of San Luis Potosí (S. comatum s.s.). In addition, their floral bracts are more deeply erose with very pronounced (puckered or bullate) abaxial ridges (vs. slightly puckered in S. comatum s.s.), the perigonal nectaries differ (smooth and thickened area vs. nectariferous concavity), and the fruit is broader. The two geographically isolated entities also have different intercostal cell patterns on the leaf (visible on the abaxial surface): rectangular cells running longitudinally in S. comatum s.s. but positioned transversely in the populations from Puebla/Oaxaca. Schoenocaulon comatum s.s. (three samples) is weakly supported as a clade by our ITS gap data (69 BP; not shown); the southern populations are also likely monophyletic (see Table 3: both are cladospecies). Thus, our phylogenetic results strongly support the recognition of the San Luis Potosí populations (S. comatum s.s.) as specifically distinct from the Puebla/Oaxaca populations of S. comatum s.l.
Two elements of Schoenocaulon mortonii, plus S. conzattii and S. pellucidum
Both entities of S. mortonii s.l. are probably metaspecies based on analyses of our ITS data sets (Table 3), but each is well defined by distinctive morphology and distribution. Frame (1990)
mentions that the southern populations (S. mortonii s.s.; Jalisco, Michoacán) are similar to S. pellucidum (discussed below), and therefore, differ (characters not specified) from the northern elements of S. mortonii s.l. (from México). Our examination of specimens reveals that the northernmost populations have conspicuously papillose leaf margins comprising several rows of papillae (vs. a few scattered marginal papillae in S. mortonii s.s.); a stouter pistil with much longer styles, and a larger floral bract with more deeply erose margins and more conspicuously puckered (bullate) abaxial surface. In addition, the nectary is represented by a nectariferous concavity at the tepal base (vs. a smooth zone in S. mortonii s.s.), and the scapes are generally shorter, 65–70 cm (vs. greater than 75 cm). These distinctive features are shared by S. conzattii, known only from the two type specimens (Conzatti & Gonzáles 449; US, GH) collected in 1897. As noted earlier, S. conzattii comprises a cladospecies with this element of S. mortonii s.l.—further supporting that these populations of S. mortonii are likely rediscovered S. conzattii. Thus, we consider the identification by Frame (1990)
of these specimens as incorrect: they actually represent S. conzattii. The inclusion of these specimens within S. conzattii renders it monophyletic and also well characterized morphologically.
Frame (1990)
noted that S. mortonii s.s. is very similar morphologically to S. pellucidum—specifically the leaf width and distance between floral ranks. She (Frame et al., 1989) separates her new species, represented by limited floral material, from S. mortonii s.s. on the basis of bulb scale texture (papery vs. thick, respectively), plus S. mortonii s.s. has a more fibrous tunic on the bulb, probably an adaptation to frequent fires. As discussed previously, our ITS analyses do not support S. pellucidum as a segregate species, and our examination of limited loan material confirms that leaf and floral features support sinking S. pellucidum into S. mortonii s.s., as the morphological features supposedly distinguishing them are inconsistent.
Two varieties of Schoenocaulon megarrhizum
Schoenocaulon megarrhizum var. megarrhizum is weakly supported as a cladospecies (Table 3) in our analyses (ITS BP 64; combined data BP 63) and is positioned well outside the clade embedded with S. megarrhizum var. deminutum (Figs. 3 and 4). As the name of the variety suggests, Frame et al. (1999)
separated S. megarrhizum var. deminutum, comprising the southeastern members of this species, on the basis of its small stature: the delicate plants with relatively small bulbs and short inflorescences (to 18 cm) grow in clumps, compared to the large and robust, solitary plants of S. megarrhizum var. megarrhizum that have bulbs twice as large and inflorescences greater than 20 cm long (Frame, 1990
). In addition, S. megarrhizum var. deminutum has different perianth and filament coloration (green-purple and purple, respectively) than S. megarrhizum s.s. (green tepals, white filaments; Frame 1989
). The new variety is based on two collections: six type specimens (Palmer 419; 1896) and two sheets of one additional collection (Rose 2227; 1897). Our analyses (Figs. 3 and 4) resolve this distinct variety (represented by one sample) in a clade with S. mortonii s.s./S. pellucidum (ITS BP 82; combined data BP 76), with the relationships between these three taxa themselves unresolved. This variety is either a new species or conspecific with the S. mortonii s.s. complex. Our morphological comparisons of S. megarrhizum var. deminutum with S. mortonii s.s./S. pellucidum are inconclusive at this time, however, due to paucity of material available for this variety.
Biogeographical diversification
Figure 4 is the combined (ITS + gap) strict consensus tree for Schoenocaulon with distribution maps for major subclades. Most species are endemics with very restricted distributions, and our analyses support the hypothesis that each clade (comprising 1–6 species) diversified in a particular geographical area. The two notable exceptions (see subsequent discussion), the relatively widespread "S. ghiesbreghtii" (specifically, the segregate S. yucatanense) and S. officinale, may have been spread far from their original localities by Native Americans who traditionally used the seeds as a source for potent pesticides (Frame, 2001
; see also Zomlefer, 1997
).
The position of S. comatum s.s. (restricted to San Luis Potosí, Mexico) as sister to the rest of clade A indicates this clade as a possible central Mexican group. Within clade A (see Fig. 3), the S. dubium–S. texanum clade has a disjunct distribution: S. dubium is endemic to peninsular Florida (sandhill and scrub communities), and sister group S. texanum is much more widespread (west central Texas–southeastern New Mexico to north central Mexico in the low elevation, arid to semi-arid zones of the Central Plateau). This Florida–western North America disjunct pattern is common for many extant xerophytic plants (and animals), due to an extensive, semiarid, circum-Gulf Coast corridor established during low sea levels of the late Miocene–Pliocene (ca. 15–5 million years ago [mya]) and early Pleistocene (2 mya) epochs (Webb, 1990
). By the mid-Pleistocene (1 mya), the biotic affiliation of Florida with the semiarid west was severed: many of the biota in the peninsula became extant relicts when climatic changes produced a vast reduction in xeric and open habitats between Florida and Texas (Webb, 1990
). Based on approximate rates of ITS mutation rates published in the literature (e.g., Ronsted et al. [2002
]: 4.7 x 10–9 to 2.4 x 10–9 substitutions per site per year), the approximate time of divergence of S. dubium and S. texanum (13 substitutions [absolute distance]/726 sites sequenced; 7.5 to 3.8 mya) is compatible with the age of these climate changes. Other examples of the many plant species (or closely related species) endemic to xeric habitats of the Florida peninsula that exhibit this disjunction distribution pattern include: Callirhoe papaver; Ceanothus microphyllus and its western relatives; Eriogonum longifolium var. gnaphalifolium and var. longifolium; Krameria lanceolata and its western relatives; Lyonia ferruginea, L. fruticosa, and L. squamulosa; Nolina brittoniana, N. atopocarpa, and their western relatives; Ulmus crassifolia; and Ziziphus celata and Z. parryi (Ward, 1979
; Judd, 1981
; Judd and Hall, 1984
; Christman and Judd, 1990
).
Subclade C comprises two main geographical elements: the widespread Gulf-Coastal element of S. ghiesbreghtii s.l. (= S. yucatanense) and clade H, mainly restricted to the Sierra Madre Oriental. The exception to the latter distribution involves S. macrocarpum with disjunct populations in the Sierra Madre Occidental (Fig. 4). Frame (1990)
postulated that the western disjuncts of this species may have become separated from a broader distribution in the mid-Tertiary (Oligocene, ca. 30–25 mya) when the formation of the eastern mountain ranges produced very arid conditions in the Central Plateau of Mexico (Rzedowski, 1978
). The distribution of S. yucatanense (S. "ghiesbreghtii" in Figs. 3 and 4) may have been affected by cultivation by indigenous peoples along the eastern coast (at least in the Yucatán Peninsula); for example, the type specimen of this segregate species (Steere 2093, NY, MICH) was found near a pyramid (Uxmal, Yucatán), as have several other collections (e.g., Bunting 4, WIS; Calzeda et al. 6653, Navaráez et al. 625, Puch et al. 567, all CICY).
Clade B overall comprises more southerly distributed elements than clade A. Subclade D, with six species, has a relatively discrete distribution just south of the Transverse Neovolcanic Belt and north of the Isthmus of Tehuantepec (Fig. 4). Within the large clade E, subclade F (S. rzedowskii/S. caricifolium s.s./S. madidorum/S. obtusum) has a more widespread and northern distribution, extending to Aguascalientes (one specimen of S. caricifolium var. caricifolium, Rzedowski & McVaugh 871, MICH, NY). In subclade G, the rarely collected S. mortonii/S. conzattii clade has a similar distribution. Subclade S. pringlei/S. ghiesbreghtii s.s., however, occupies a different area (southern Mexico), with the former species distributed in Morelos, México, Hidalgo, and Veracruz, and the latter, occurring just to the south, restricted to several localities in Chiapas. The regions into which these clades have diversified (the Sierra Madre Oriental, the Transverse Neovolcanic Belt mountains, and the mountains of Chiapas) are considered distinct floristic provinces by Rzedowski (1978)
and are each characterized by numerous endemic species.
The distribution of S. officinale extends from Colima, Mexico, down the western coast of Central America to Costa Rica, plus disjunct populations in south central Peru and northwestern Venezuela (see Fig. 4). This species is the primary source of the medicinal "sabadilla" or "cebadilla" (alkaloids extracted from the seeds) used mainly as a disinfectant and topical pesticide (summary in Zomlefer, 1997
). Sabadilla was an important drug plant for Native Americans of Mexico to northern South America and was later (16th century) adopted by European colonists as a topical powder for eradicating lice. Plants were even cultivated in the mountains of Veracruz, Mexico, and Caracas, Venezuela, for export to Europe during World War I for pediculosis (Frame, 1989
, 1990
). This species, characterized by the largest plants in the genus, undoubtedly has been cultivated and spread by indigenous peoples. Frame (1990)
noted local races with double flowers (7–12 tepals, 6–12 stamens, 3–9 carpels) indicating probable cultivars. Using certain specimens to plot distributions may be inaccurate as collections from markets (e.g., Palmer 1410, GH, US) may have been gathered far from the town in which they were sold (Frame, 1989
, 1990
). The species is likely native to southern Mexico, where it is most diverse and where its closest relatives occur, and was transported southward into Central and South America in connection with its medicinal uses.
The remaining two clades, S. megarrhizum s.s. alone and subclade I (S. tigrense/S. jaliscense/S. megarrhizum var. deminutum/S. pellucidum/S. mortonii), are restricted to western Mexico. Schoenocaulon megarrhizum var. megarrhizum occupies a unique area in the northwestern corner of Mexico (Chihuahua, Sonora, Sinaloa) bordering the Sierra Madre Occidental, whereas clade I has a west central distribution (Durango, Jalisco, Michoacán, Nayarit, San Luis Potosí), another arid-semiarid ecological zone bordered by mountain ranges to the east, west, and south. Members of clade E have, thus, diversified within the Sierra Madre Occidental and the adjacent Pacific Coastal regions, an area constituting distinctive floristic provinces that are diverse and contain numerous endemics (Rzedowski, 1978
).
The current distributions of the outgroups for Schoenocaulon are various: the coastal plain of the southeastern United States (Zigadenus, Amianthium, and Stenanthium), midwestern–western United States (Toxicoscordion), and the much wider ranging Anticlea (Asia, North America south to Guatemala) and Veratrum (North America–Eurasia). The Melanthieae are most diverse in North America: thus, outgroup distributions support the stem-species of Schoenocaulon likely occurring there. Our phylogeny suggests that Schoenocaulon originated in central to southern Mexico, as indicated by the geographical range of S. comatum s.s. (strongly supported sister to the remaining species of clade A [Fig. 3]) and also by the distribution of most species of clade B. Dispersal northward associated with divergence and speciation may have occurred within clade A, especially as indicated by the distributions of S. yucatanense (considered part of S. ghiesbreghtii s.l. by Frame [1990
, 2002
; López-Ferrari et al., 2000
]), S. texanum, and S. dubium (Fig. 4). Within clade B, speciation was likely largely allopatric through range fragmentation, followed by evolutionary divergence, with each of the major clades diversifying in a particular geographical region (see Table 1 and color-coded units and maps of Fig. 4). For example, the species of clade D occur in the mountains just south of the Transverse Neovolcanic Belt, whereas members of clade F (sister to the remaining members of clade E) diversified mainly in the eastern Transverse Neovolcanic Belt and southern mountains of the Sierra Madre Oriental. In contrast, members of clade I likely evolved in the Sierra Madre Occidental and adjacent Pacific Coastal floristic province, while the perhaps phylogenetically isolated S. megarrhizum var. megarrhizum occurs in a unique area in the extreme northwestern portion of the Sierra Madre Occidental/Pacific Coastal region. Superimposed on these natural patterns is the impact of humans on range expansion of the economically important plants, S. officinale (transported into Central and South America) and S. yucatanense (spread to the Yucatán peninsula).
Morphological evolution
Using the molecular cladogram (Fig. 3) as a framework, we traced the pattern of variation (and some general tendencies) in morphological characters by manually mapping character states so that transitions were minimized (i.e., application of parsimony). Below we summarize our investigation of some exemplar character state changes inferred by hypothesis of phylogeny; three examples (pedicel development, nectary type, and maximum scape length) are depicted in Fig. 5. Although homoplasious, several characters are potential synapomorphies for certain clades. In addition, we acknowledge that the limited number of specimens available for several species may present a source of error in evaluating gaps in quantitative characters and additional quantitative study may provide additional support for their value in discerning phylogenetic patterns.
|
; Zomlefer et al., 2001
, 2004
, 2006
) with the flowers varying from subsessile (barely developed pedicels, 0.5–2.0 mm) to sessile (no perceptible pedicel; Fig. 2C). As shown in Fig. 5A, sessile flowers have evolved several times from the ancestral subsessile condition in S. "ghiesbreghtii" (= S. yucatanense) and S. texanum/S. dubium (all in clade A; DELTRAN optimization); S. jaliscence var. regulare and S. pringlei (both in clade G); and the entirely sessile-flowered clade D. Although homoplasious, sessile flowers are likely derived for the latter clade, and a spicate inflorescence with subsessile flowers is synapomorphic for Schoenocaulon.
Unlike Veratrum, in which tepal margin modifications are synapomorphic for particular subgeneric groupings (Zomlefer et al., 2003
), the erose margin in Schoenocaulon (Fig. 2F) is homoplasious, arising independently several times (two species in clade A; three, in clade B [subclades D and F]). The lack of tepal auricles (Fig. 2D–F) is ancestral (outgroup comparison); tepal auricles (Fig. 2G) occur in most of clade B (except for the embedded S. tigrense and S. officinale) and within clade A (S. texanum and subclade H). Therefore, this character is a possible synapomorphy for clade B, which has also evolved independently within clade A.
Tepal color also exhibits some phylogenetic signal. Tepals in Schoenocaulon species are usually greenish (sometimes with a maroon tinge) but are entirely maroon in four of the five members of clade H. This striking coloration is a potential synapomorphy for that clade with a reversal to green in S. macrocarpum. In addition, cream-colored tepals have evolved independently three times, within clade G (S. officinale, S. tigrense, S. pellucidum), although the tepals of S. pellucidum vary from greenish to cream-colored. The bright red color of the flowers of S. tenuifolium is an autapomorphy for that species.
Our analyses confirm that nectariferous tissue lining a concavity at the tepal base (e.g., Fig. 2D and F) is the synapomorphic nectary type for Schoenocaulon (Fig. 5B). A likely synapomorphy for clade I is the reduction of this nectariferous depression to a poorly defined, shiny, smooth zone (e.g., Fig. 2G)—a homoplasious feature also occurring in S. megarrhizum var. megarrhizum, S. caricifolium var. caricifolium/S. rzedowskii, and S. "comatum" (all clade B), and S. texanum (clade A). The thick pad-like gland of S. officinale (Fig. 2E) may represent an autapomorphic reversal to the outgroup condition for that species.
Ovule number per locule provides the strongest morphological marker distinguishing the major clades A and B. The ancestral condition (outgroup comparison) is 5–14+ ovules per locule. (An exception is Amianthium, with 2–4 ovules per locule, a synapomorphy for that genus.) In Schoenocaulon, the species of clade B tend to have few (2–9) ovules per locule in most species, with an increase to 12 in two species of clade G (S. officinale and S. jaliscence). This contrasts somewhat with the overlapping character of 5–16 ovules per locule for clade A. Therefore, the lower ovule number per locule may be a synapomorphy for clade B, with an increased number evolving in two members of that clade.
Plant robustness
The plants of clade E are generally the largest in the genus. Among quantifiable characters relating to plant robustness, scape length is one of the best indicators (see Fig. 5C), as noted by Frame (1990)
for members of subclade I. The entire clade G has maximum scape lengths of 65 cm or longer (therefore, a possible synapomorphy for these species). Lengths are generally shorter in other clades; however, homoplasy occurs (scattered species with longer scapes elsewhere in cladogram: S. dubium, S. plumosum, S. tenuifolium, and S. madidorum). Inflorescence congestion, defined here as at least eight flowers per centimeter of scape at anthesis (e.g., Fig. 2B), is also useful, although very homoplasious: members of clade F all have relatively compact inflorescences (a likely synapomorphy; Fig. 2B), which also characterizes S. ghiesbreghtii/S. pringlei and six other species scattered throughout the cladogram. The number of floral ranks, commonly 4–8 in Schoenocaulon, is reduced to three for the S. tenorioi/S. caricifolium var. oaxacense/S. tenue clade (within clade D). Maximum leaf widths (10–15 mm) occur in clade B (five species: S. pellucidum, S. officinale [Fig. 2B], S. ghiesbreghtii s.s., S. madidorum, and S. tenuifolium), and the maximum number of leaves per plant (nine or more) occurs in clade E (all of clade F; also S. mortonii s.s., S. pellucidum, S. officinale, and S. ghiesbreghtii s.s.) and also in clades D (S. tenuifolium) and H (S. macrocarpum/S. plumosum).
General conclusions
With analyses of ITS sequence data, we were able to discern several cryptic species of Schoenocaulon, which are, nonetheless, clearly diagnosable on the basis of morphology. The resolution of these "split species" (color bars in Fig. 3) supports redefinition of five species complexes, including recognition of two new species, resurrection of a segregate, and reassignment of discordant elements to other species. The molecular cladogram also resolved several well-supported subgeneric clades, each indigenous to certain geographical areas. Indels in ITS (deletions) also provided significant support to some clades. We have also identified additional morphological synapomorphies for Schoenocaulon and several potential derived features (albeit homoplasious) that define most subgeneric clades.
For our investigations of taxa in tribe Melanthieae (e.g., Zomlefer et al., 2001
, 2003
), we have recognized taxa above the rank of species based on principles outlined by Backlund and Bremer (1998)
: first and foremost, taxa are monophyletic but secondarily, they should have strong statistical support and also be more or less recognizable based on morphological characters. Corollary criteria, including the size of the clade, nomenclatural stability, and issues relating to minimizing redundancy in classification, are addressed in Kellogg and Judd (2002)
and APG II (2003)
. Based on these criteria, at this time we do not recommend formal taxonomic recognition of the major subgeneric clades of Schoenocaulon (i.e., clades A through I in Figs. 3, 4, 5) because they lack unambiguous diagnostic morphological markers.
Our study exemplifies the benefit of technological improvements allowing extraction and sequencing of DNA from herbarium specimens of difficult taxa. We were successful in elucidating a well-supported phylogeny for this interesting group of species only by acquiring DNA sequence data from limited, often old, material. Many Schoenocaulon species are now rare, occurring only in a few extant populations distributed within limited geographical areas (Frame et al., 1999
). Their montane habitats are rapidly being overtaken by the growth of cities and the conversion of pine–oak forests for agricultural use (Leopold, 1950
; Iltis, 1998
). Several species face extinction within the next few years unless sufficient natural habitats are protected (Frame, 2001
). We therefore appreciate the opportunity to sample these precious specimens and look forward to improved protocols facilitating use of herbarium material in molecular investigations.
FOOTNOTES
1 The authors thank the curators/collections managers of the following herbaria for loaning crucial material: ARIZ, BRIT, CAS, CICY, F, GH, IEB, K, MEXU, MICH, MO, NMC, NY, PH, RSA, TCD, TEX, UC, UNM, US, WIS, and XAL. George "Rusty" Russell and Debbie Bell (US), Mark Wetter and Hugh H. Iltis (WIS), Tom Zanoni (NY), Mona Bourell (CAS), Barbara Ertter (UC/JEPS), Kent D. Perkins (FLAS), James Solomon (MO), and Emily Wood (GH) graciously assisted W.B.Z. at their respective herbaria. Judy Warnement, Gretchen Wade, and Lisa Decesare (all Harvard University Botanical Libraries), Andrew W. Colligan (archivist, Missouri Botanical Garden Library), and Kanchi Gandhi helped with problematic references. They are also grateful to Robert Sivinski, Tom Wendt (TEX), and Richard D. Worthington (UTEP) for silica samples and voucher specimens; Kent D. Perkins for permission to sample specimens at FLAS; Lucinda McDade (PH) for permission to extract a rare specimen of Schoenocaulon comatum collected in 1876; Kanchi N. Gandhi for assistance on the nomenclature for the S. ghiesbreghtii complex; Hugh H. Iltis for providing various obscure publications on Schoenocaulon; and two anonymous reviewers for constructive criticisms of the text. The Department of Plant Biology, University of Georgia, generously provided funds to W.B.Z. for travel to CAS, FLAS, GH, MO, NY, UC/JEPS, and WIS. 
5 4Author for correspondence (wendyz{at}plantbio.uga.edu
)
LITERATURE CITED
Angiosperm Phylogeny Group II [APG II].. 2003. An update of the Angiosperm Phylogeny Group classification for the orders and families of flowering plants: APG II. Botanical Journal of the Linnean Society 141: 399-436.[CrossRef]
Backlund A. Bremer K.. 1998. To be or not to be—principles of classification and monophyletic plant families. Taxon 47: 391-400.[CrossRef][ISI]
Baldwin B. G. Sanderson M. J. Porter J. M. Wojciechowski M. F. Campbell C. S. Donoghue M. J.. 1995. The ITS region of nuclear ribosomal DNA: a valuable source of evidence on angiosperm phylogeny. Annals of the Missouri Botanical Garden 82: 247-277.[CrossRef][ISI]
Baum D. A. Donoghue M. J.. 1995. Choosing among alternative "phylogenetic" species concepts. Systematic Botany 20: 560-573.[CrossRef][ISI]
Blattner F. R.. 1999. Direct amplification of the entire ITS region from poorly preserved plant material using recombinant PCR. BioTechniques 27: 1180-1185.[ISI][Medline]
Brinker R. R.. 1942. Monograph of Schoenocaulon. Annals of the Missouri Botanical Garden 29: 287-315.[CrossRef]
Cave M. S.. 1967. Documented chromosome numbers of plants. Madroño 19: 134-136.
Chase M. W. Hills H. G.. 1991. Silica gel: an ideal material for field preservation of leaf samples for DNA studies. Taxon 40: 215-220.[CrossRef][ISI]
Christman S. P. Judd W. S.. 1990. Notes on plants endemic to Florida scrub. Florida Scientist 53: 52-73.
Correll D. S. Johnston M. C.. 1979. Manual of the vascular plants of Texas University of Texas at Dallas, Richardson, Texas, USA.
Cox A. V.. 1997. PAUPGAP, version 1.12 Program distributed by author, Royal Botanic Gardens, Kew, UK.
Davis J. I. Nixon K. C.. 1992. Populations, genetic variation, and the delimitation of phylogenetic species. Systematic Biology 41: 421-435.[CrossRef]
deQueiroz K. Donoghue M. J.. 1988. Phylogenetic systematics and the species problem. Cladistics 4: 317-338.
de Zerpa S. M.. 1951. Los cromosomas de Schoenocaulon officinale (cebadilla). Agronomía Tropical 1: 83-84.
Diggs G. M. Lipsomb B. L. O'Kennon R. J.. 1999. Shinners and Mahler's illustrated flora of north central Texas Botanical Research Institute of Texas, Fort Worth, Texas, USA.
Donoghue M. J.. 1985. A critique of the biological species concept and recommendations for a phylogenetic alternative. Bryologist 88: 172-181.[CrossRef][ISI]
Doyle J. J. Doyle J. L.. 1987. A rapid DNA isolation procedure for small quantities of fresh leaf tissue. Phytochemical Bulletin 19: 11-15.
Duncan W. H. Kartesz J. T.. 1981. Vascular flora of Georgia: an annotated checklist University of Georgia Press, Athens, Georgia, USA.
Eriksson T.. 1998. AutoDecay version 4.0.1 Computer program distributed by author, Stockholm University, Stockholm, Sweden. Available at http://www.bergianska.se/index_kontaktaoss_torsten.html.
Felsenstein J.. 1985. Confidence limits on phylogenies: an approach using the bootstrap. Evolution 46: 59-173.
Frame D. M.. 1989. Schoenocaulon A. Gray. In R. McVaugh [ed.] Novo-Galiciana, vol. 15 263-275 University of Michigan Herbarium, Ann Arbor, Michigan, USA.
Frame D. M.. 1990. A revision of Schoenocaulon (Liliaceae–Melanthieae) Ph.D. dissertation, City University of New York, New York, New York, USA.
Frame D. M.. 2001. Chromosome studies in Schoenocaulon (Liliaceae: Melanthieae) a relict genus. Anales del Instituto de Biología, Universidad Nacional Autóma de México, Serie Botánica 72: 123-129.
Frame D. M.. 2002. Schoenocaulon A. Gray. In Flora of North America Editorial Committee [eds.] Flora of North America north of Mexico, vol. 26 79-81 Oxford University Press, New York, New York, USA.
Frame D. M. Espejo A. Lopez-Ferrari A. R.. 1999. A conspectus of Mexican Melanthiaceae including a description of new taxa of Schoenocaulon and Zigadenus. Acta Botánica Mexicana 48: 27-50.
Gates R. R.. 1918. A systematic study of the North American Melanthaceae [sic] from a genetic standpoint. Journal of the Linnean Society, Botany 44: 131-172.
Gray A.. 1837. Melanthacearum Americae septentrionalis revisio. Annals of the Lyceum of Natural History of New York 4: 105-140.
Guerrero A. Judd W. S. Morris A. B.. 2004. A new species of Illicium subsection Parviflora (Illiciaceae) from the Massif de la Hotte, Haiti. Brittonia 56: 346-352.[CrossRef][ISI]
Hatch S. L. Gandhi K. N. Brown L. E.. 1990. Checklist of the vascular plants of Texas Texas A & M University, College Station, Texas, USA.
Huck R. B. Judd W. S. Whitten W. M. Skean J. D. Wunderlin R. P. Delaney K. R.. 1989. A new Dicerandra (Labiatae) from the Lake Wales Ridge of Florida, with a cladistic analysis and discussion of endemism. Systematic Botany 14: 197-213.[CrossRef][ISI]
Huelsenbeck J. P. Ronquist F.. 2001. MrBayes: Bayesian inference of phylogenetic trees. Bioinformatics 17: 754-755.
Huelsenbeck J. P. Ronquist F.. 2002. MrBayes Computer program distributed by the authors, University of California, San Diego, California, USA. Available at http://mrbayes.csit.fsu.edu/.
Itlis H. H.. 1998. Cleome chapalaensis, n. sp., a South American element on the Mexican plateau. Boletín del Instituto de Botánica (IBUG) 5: 413-443.
Jones S. B. Coile N. C.. 1988. The distribution of the vascular flora of Georgia Published by the authors, Athens, Georgia, USA.
Jones S. D. Wipff J. K. Montgomery P. M.. 1997. Vascular plants of Texas University of Texas Press, Austin, Texas, USA.
