| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Systematics |
2School of Biological Sciences, Washington State University, Pullman, Washington 99164-4236 USA;and 3Jodrell Laboratory, Royal Botanic Gardens, Kew, Richmond, Surrey, TW9 3DS, UK
Received for publication December 3, 2002. Accepted for publication March 18, 2003.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
Key Words: Loasaceae • maximum likelihood • parsimony • phylogeny • Shimodaira-Hasegawa (SH) test • systematics
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
. Flowers of these two genera are choripetalous, polystemonous, and syncarpous, usually with several ovules. The morphological disparity of Loasaceae was heightened when Bartling (1825) included in the family Gronovia, in which flowers are haplostemonous and have only a single ovule. Reichenbach (1837)
was the first to circumscribe genera in the three groups Gronovieae, Mentzelieae, and Blumenbachieae that would ultimately correspond to the modern subfamilies. Not all 19th-century systematists accepted Reichenbach's groups. For example, Baillon (1888)
recognized only two series in the family, including Gronovieae, corresponding to Reichenbach's group of that name, and Loaseae, which combined Mentzelieae and Blumenbachieae of Reichenbach. In contrast to workers who have grouped gronovioid, mentzelioid, and loasoid taxa as Loasaceae, Endlicher (1841)
and Weigend (1997
; Weigend et al. 2000
) segregated the gronovioids from Loasaceae as the family Gronoviaceae.
Gilg (1895
, 1925
) and especially Urban and Gilg (1900)
provided influential systematic treatments of Loasaceae in which the subfamilies Gronovioideae, Mentzelioideae, and Loasoideae were circumscribed in a manner that has been largely followed since those publications (including Weigend [1997
] despite his elevation of Gronovioideae to the familial level). The monograph of Urban and Gilg (1900)
, a comprehensive taxonomic treatment of Loasaceae, presented revisions for genera and provided several new subgeneric groups. Since Urban and Gilg (1900)
, generic circumscriptions have been debated. Thompson and Ernst (1967)
revised Eucnide and reduced Sympetaleia to synonymy within Eucnide. Various workers have questioned broad circumscriptions of the larger genera Cajophora, Loasa, and Mentzelia. For example, Poston and Thompson (1977)
suggested that Cajophora sensu lato (s.l.) was polyphyletic and hypothesized that Cajophora section Bialatae was more closely related to Blumenbachia than to other Cajophora. Weigend (1997)
excluded sections Angulatae and Bialatae from Cajophora and placed them in Blumenbachia. Grau (1997)
resurrected Huidobria, which had been included as a section of Loasa s.l. by Urban and Gilg (1900)
. Weigend (1997)
segregated the new genera Aosa, Chichicaste, Nasa, and Presliophytum from Loasa s.l. Several authors have advocated segregating the genera Acrolasia and Nuttallia from Mentzelia (Rydberg, 1903
; Davidson, 1916
; Weber and Wittman, 2001
). Brown (1971)
, Hempel (1995
; Hempel and Jansen, 1996
), and Weigend (1997)
considered Mentzelia to be paraphyletic. Aside from the several segregate genera, only three new genera lying clearly outside of those circumscribed by Urban and Gilg (1900)
have been described. The first of these, Schismocarpus, was described by Blake (1918)
and allied with Mentzelioideae. The two other novel genera, Plakothira of Florence (1985)
and Xylopodia of Weigend (1997)
, were relegated to Loasoideae.
These taxonomic studies have provided considerable insight into the distribution of character diversity among Loasaceae and clarified key issues in the systematics of the family. Circumscription problems and questions of evolutionary diversification, however, require an approach centered primarily on hypotheses of monophyletic groups and their relationships. Hufford (1988)
first applied phylogenetic systematics to problems of monophyly in the Loasaceae; however, his analysis of morphological characters derived strictly from the literature resolved few relationships. Subsequent phylogenetic studies by Poston and Nowicke (1993)
and Hempel (1995
; Hempel and Jansen, 1996
) were hampered by limited taxon sampling. Moody et al. (2001)
considerably advanced our understanding of relationships in Loasaceae in a phylogenetic study that sampled taxa broadly within the family and applied analyses of DNA sequences of the plastid gene matK to examine the placement of gronovioids. The results of Moody et al. (2001)
identified several major clades in Loasaceae, including the Gronovioideae-Mentzelia clade, and provided good support for many monophyletic groups.
Our phylogenetic analyses extend those of Moody et al. (2001)
by adding DNA sequence data from the plastid trnL-trnF region (which includes the trnL intron and trnL-trnF intergenic spacer; Taberlet et al., 1991
) to data from the matK gene and increasing taxon sampling within Loasaceae. Both maximum-parsimony and maximum-likelihood analyses are applied as optimality criteria for the selection of phylogenetic trees. We examine further the support for major clades of Loasaceae, considering especially Loasoideae and Mentzelia, and clarify where additional data are needed to resolve clades. We explicitly evaluate whether prior circumscriptions at the familial, subfamilial, generic, and subgeneric levels correspond to monophyletic groups and, if so, how well they are supported by our character data. Our approach is to model existing circumscriptions and hypotheses of evolutionary relationship as topologies, then search for the most parsimonious and most likely trees using these topologies as constraints. We compare the tree lengths and likelihoods, respectively, from constrained and unconstrained analyses as a means to evaluate the relative support, using our data, for various hypotheses. We use the Shimodaira-Hasegawa test (Shimodaira and Hasegawa, 1999
) to establish the significance of the differences we find in constrained and unconstrained topologies from maximum-likelihood analyses.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
, 1998
, 2002
; Hempel et al., 1995
; Olmstead et al., 2000
; Hufford et al., 2001
). Nine outgroups were used for the analysis of matK; based on those results, three outgroups were applied for the analysis of the trnL-trnF and the combination of matK and trnL-trnF. Our sampling in Loasaceae aimed to encompass groups delimited in or suggested to be problematic by earlier taxonomic studies (Urban and Gilg, 1900
; Gilg, 1925
; Darlington, 1934
; Daniels, 1970
; Brown, 1971
; Poston and Thompson, 1977
; Poston and Nowicke, 1993
; Hempel, 1995
; Weigend, 1997
). We intensified taxon sampling when possible in clades of the family found by Moody et al. (2001)
to examine the composition of monophyletic groups. Our sampling includes representatives from most suprageneric entities recognized in Urban and Gilg's (1900)
comprehensive monograph of Loasaceae. Among the genera recognized in modern treatments of the family, all were sampled except for those described only recently, including Xylopodia, which is known only from the type specimens, and the segregate genus Chichicaste (Weigend, 1997
). We sampled broadly among the species recognized in the sections of Mentzelia. For a list of taxa and accession sampled see the Supplementary Data accompanying the online version of this article.
DNA sequences
The matK sequences for several outgroups and Loasaceae were obtained from Moody et al. (2001)
. New matK sequences were obtained for an additional 32 Loasaceae. All trnL-trnF sequences used in the study were new. Total DNA was extracted from either herbarium or silica-dried specimens of leaves using a standard cetyltrimethylammonium bromide (CTAB) procedure (Doyle and Doyle, 1987
). The polymerase chain reaction (PCR) mixes varied somewhat, although most used 20 mmol/L Tris-HCl pH 8.3, 50 mmol/L KCl, 1.5 mmol/L MgCl2, 0.01% Tween-20, 150 µmol/L dNTPs, 0.5 µmol/L forward- and reverse-amplification primers, diluted DNA template (1 : 10–1 : 100, depending on extraction quality), and water to a total volume of 25 µL. The PCR primers for matK were matK-710F and trnK-2R (Johnson and Soltis, 1995
) and for trnL-trnF were c and f (Taberlet et al., 1991
). Sequencing primers were matK-710F, trnK-2R, matK-1713F, and matK-1848R (citations in Moody et al., 2001
) and trnL c and f. Sequences were aligned manually in Se-Al (Rambaut, 1996
). In selected regions of the aligned sequences, character homology was equivocal; hence, for the analysis one 6-bp region of matK and 20 short regions of trnL-trnF were deleted.
Phylogenetic analyses
All phylogenetic analyses used PAUP* 4.0 (Swofford, 2002
). Parsimony analyses were conducted independently on the matK and trnL-trnF data sets and on a data set in which matK and trnL-trnF sequences were combined. Heuristic searches included 1000 starting trees built by random taxon addition followed by tree bisection-reconnection (TBR) branch-swapping. All character state transitions were equally weighted and unordered. Indels were treated as missing data. Tree statistics and measures of homoplasy were calculated using PAUP* with uninformative characters removed. Multiple most parsimonious trees were combined in PAUP* to construct strict consensus cladograms. The robustness of clades was assessed using branch decay (Bremer, 1988
; Donoghue et al., 1992
) and bootstrap (Felsenstein, 1985
) analyses. Decay analyses were implemented using AutoDecay (Eriksson, 1999
) and PAUP* (Swofford, 2002
). Bootstrap analyses were implemented in PAUP* using the heuristic search procedures described, including 500 pseudoreplicates for the analysis of the trnL-trnF data and 1000 pseudoreplicates for the analyses of the matK and combined data sets (maxtrees was set at 1000 for the independent data sets and 2000 for the combined data set).
Maximum-likelihood (ML) analyses were conducted only on the combined matK and trnL-trnF data. To reduce computational times, taxon sampling was reduced to two outgroup and 27 ingroup taxa. Taxa were sampled to represent the major clades found in the parsimony searches and to facilitate evaluating prior taxonomic and phylogenetic hypotheses. Modeltest (Posada and Crandall, 1998
), which uses likelihood ratio tests to compare 56 alternative models for DNA substitution, was used to select a substitution model for the reduced taxon data set. Two tree topologies, one inferred using maximum parsimony (as specified above for the combined data set) and the other using neighbor joining, were used in Modeltest to determine the model that best fit the combined matK and trnL-trnF data. The model of DNA substitution selected by Modeltest, a submodel of the GTR + 
model in which only one parameter was needed for transitions, was the same for both of the preliminary topologies. The ML analysis used was a heuristic search procedure that included a single starting tree built by random taxon addition followed by TBR branch-swapping. The robustness of the ML topology was assessed using bootstrap analysis, which was limited to 50 pseudoreplicates because of computational time. The bootstrap analysis used a heuristic procedure in PAUP* that included random taxon addition and TBR branch swapping.
Alternative topologies
Hypotheses of taxonomic groups and their interrelationships can be modelled as cladogram topologies. We designed five constraint topologies based on existing hypotheses of relationships; each constraint specified a few nodes to force the monophyly of selected groups of taxa. We applied these constraints to searches using the combined matK and trnL-trnF data under both parsimony and likelihood criteria. The parsimony analyses used the same 70 taxa as in the unconstrained parsimony analysis of the combined data. We conducted full heuristic searches for the most parsimonious cladograms under each of the topology constraints (all analyses swapped to completion), permitting us to compare the lengths of constrained topologies to that of the most parsimonious topologies from the unconstrained analyses. Constrained ML searches used the same 29 taxa, model of DNA substitution, and search procedure as the unconstrained ML search described earlier.
We used the Shimodaira-Hasegawa (SH) test (Shimodaira and Hasegawa, 1999
) to compare the results of our analyses to prior hypotheses. The SH test provides a statistical evaluation of the differences in likelihoods between trees of interest; we compared those trees that resulted from the constrained and unconstrained ML analyses. Some likelihood-based tests are compromised by the inclusion of both a priori and a posteriori hypotheses (e.g., the Kishino-Hasegawa test; Shimodaira and Hasegawa, 1999
; Goldman et al., 2000
). To address this problem, investigators have advocated as viable alternatives either parametric bootstrapping (Huelsenbeck et al., 1996
) or the SH test (Shimodaira and Hasegawa, 1999
; Goldman et al., 2000
), which is less computationally intensive and more conservative (Buckley, 2002
). The SH test adjusts the significance values as needed for multiple comparisons and conducts one-tailed tests that are appropriate when some of the trees are determined a posteriori (Shimodaira and Hasagawa, 1999
; Goldman et al., 2000
). The SH test is designed for multiple, simultaneous comparisons among trees and assumes that the true tree is included. The set of topologies that we compared included the ML tree from the unconstrained search, the most likely tree from each constrained ML search, and a set of trees based on a parsimony analysis of the 29 taxon data set. Trees from a parsimony analysis were included because of the possibility that the true tree might not be the ML tree from the unconstrained analysis. From the parsimony analysis, we included the most parsimonious tree (same topology as the ML tree) and all trees up to five steps longer (a total of 887 trees) in the SH test. It was logistically impractical to consider all possible topologies for the 29 taxon data set in the SH test; topologies up to five steps longer than the most parsimonious tree provided a broad range of trees that shared the well-supported branches found in our analyses, and we were able to include all trees up to five steps longer that were found. Shimodaira and Hasegawa (1999
, p. 1115) suggested that "extremely unlikely topologies" not be included in the set of topologies compared in the SH test, and topologies more than five steps longer than the most parsimonious tree tended to have multiple branches that were inconsistent with existing ideas about relationships in Loasaceae. Earlier studies that applied the SH test have not included as large a set of topologies for simultaneous comparison (e.g., Buckley et al., 2001
; Hahn, 2002
; Leaché and Reeder, 2002
; Silberman et al., 2002
; Turmel et al., 2002
). The SH test was implemented in PAUP*, using 1000 bootstrap replicates that applied the RELL method for resampling (Kishino et al., 1990
) to establish a null distribution for the test statistic.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
95%) for the monophyly of several major clades, including Loasoideae-Gronovioideae-Mentzelia, Loasoideae, Presliophytum-Loasa-Blumenbachia-Cajophora-Scyphanthus, Presliophytum-Loasa malesherbioides, Klaprothieae, Gronovioideae-Mentzelia, Gronovioideae, and Cevallia-Fuertesia-Gronovia. Several genera also received a similar level of very strong support, including Blumenbachia, Cajophora, Presliophytum, Nasa, Petalonyx, Mentzelia, and Eucnide. Support was particularly strong for "backbone" nodes of Mentzelia (BS 94%). The basal nodes of Loasaceae, including the Eucnide and Schismocarpus clades, received more modest support (BS = 78–82%).
|
94%).
|
|
|
|
resulted in trees that were 15 state changes longer than the most parsimonious trees (Fig. 8). The constrained analysis that forced the monophyly of each of the three subfamilies Mentzelioideae, Gronovioideae, and Loasoideae as circumscribed by Urban and Gilg (1900)
resulted in trees that were 16 state changes longer than the most parsimonious trees (Schismocarpus had not been described at the time of Urban and Gilg's [1900
] monograph and was included in Mentzelioideae as suggested by Blake [1918
]) (Fig. 7). The constrained analyses that forced the gronovioids (Cevallia, Fuertesia, Gronovia, and Petalonyx) to be the sister to the rest of the Loasaceae, modeling the segregation of Gronoviaceae from Loasaceae as advocated by Endlicher (1841)
and Weigend (1997
; Weigend et al., 2000
), resulted in trees that were 18 state changes longer than the most parsimonious trees (Fig. 6).
|
|
, (3) the monophyly of Loasa s.l. as circumscribed by Urban and Gilg (1900)
, and (4) the monophyly of Mentzelia torreyi and Mentzelia sect. Bartonia. The SH test did not reject either the trees that resulted when M. reflexa was constrained to form a monophyletic group with Mentzelia sect. Bartonia or any of the trees derived from the unconstrained parsimony analysis. |
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
The best topologies found for the combined matK and trnL-trnF data sets using parsimony and likelihood analyses were entirely consistent. Many clades recovered in these analyses were very strongly supported. Less support was found for the most basal nodes of the family. Additional data need to be acquired to evaluate and presumably provide enhanced support for the placements of Eucnide and Schismocarpus. The poorest support for clades and, indeed, a lack of clade resolution, was found in some of the more terminal clades, especially in Mentzelia, in which we found little sequence divergence. The unresolved relationships among species in the sections of Mentzelia contrasts greatly to the very strong support for relationships among most of the sections. This pattern of support may reflect the antiquity of the major clades of Mentzelia in contrast to what may have been relatively recent taxon radiations in the sections, each of which is associated with arid provinces in western North America and Mexico that originated during the latter part of the Tertiary (Axelrod, 1950
; Cronquist, 1978
; Van Devender et al., 1987
; Graham, 1993
).
Splitting Loasaceae
Several authors have questioned the monophyly of Loasaceae. Most of these authors (Payer, 1857
; Hirmer, 1916
; Leins and Winhard, 1973
; Ronse Decraene and Smets, 1987
) suggested splitting taxa with centripetal androecial development (e.g., Mentzelia) from those that have centrifugal androecial development (e.g., Cajophora and Nasa). Such suggestions proved problematic when Hufford (1990)
demonstrated that some Loasoideae, including Cajophora and Nasa, purported to have centrifugal androecia actually begin androecial development with a phase of centripetal initiation of stamen primordia. Dickson (1866)
and Hufford (1990)
demonstrated that androecia throughout Loasaceae s.l. (as circumscribed in Urban and Gilg, 1900
) share common developmental attributes that could be expected if the family were monophyletic. Our results indicate that either diplostemony (characteristic of Schismocarpus) or polystemony in which stamen initiation is strictly centripetal (as in Eucnide) was plesiomorphic in Loasaceae; polystemonous androecia that have phases of centripetal and centrifugal initiation or strictly centrifugal initiation patterns are novelties within a monophyletic Loasaceae in Loasoideae. Although neglected by Payer (1857)
, Hirmer (1916)
, Leins and Winhard (1973)
, and Ronse Decraene and Smets (1987)
, who focused on differences in polystemonous androecia when suggesting that Loasaceae were not monophyletic, Gronovioideae have haplostemonous androecia, a derived simplification of androecial development within Loasaceae (Moody and Hufford, 2000
).
Gronovioids have also been segregated from Loasaceae as their own family, Gronoviaceae (Endlicher, 1841
; Weigend, 1997
). Weigend (1997)
placed Gronoviaceae as the sister of Loasaceae in a phylogenetic diagram. This placement is inconsistent with our results (e.g., their placement within Loasaceae is supported by two nodes that have bootstrap values greater than 95%). A constrained parsimony analysis that forced a sister group relationship between gronovioids and the rest of Loasaceae resulted in topologies that were 18 steps longer than the most parsimonious cladograms in the unconstrained analysis of matK and trnL-trnF sequences (Table 1). The ML topology constrained to place gronovioids as the sister to the rest of Loasaceae was rejected using the SH test (P = 0.036). Our results provide very strong support for the sister group relationship of Gronovioideae and Mentzelia.
Subfamilies
The three subfamilies of Loasaceae—Loasoideae, Mentzelioideae, and Gronovioideae—as circumscribed by Urban and Gilg (1900)
, have been modified little to date, and no major realignments or new subfamilies have been proposed. Mentzelioideae were expanded to incorporate Schismocarpus (Gilg, 1925
), following the recommendation of Blake (1918)
. Loasoideae were expanded to include Plakothira (Florence, 1985
) and Xylopodia (Weigend, 1997
). Circumscriptions of the subfamilies have received only modest challenges (Ernst and Thompson, 1963
), and most of these have focused on Gronovioideae (Davis and Thompson, 1967
; Poston and Nowicke, 1993
). Hempel's (1995)
phylogenetic analysis of Loasaceae did not recover a monophyletic Gronovioideae; however, she sampled only seven species in the family for the plastid genes ndhF and rbcL. In contrast to challenges to the circumscription of Gronovioideae, Weigend (1997)
emphasized that gronovioids had several synapomorphies and argued for their monophyly. Moody et al. (2001)
used matK sequences from a broad sampling of Loasaceae in phylogenetic analyses that provided strong support for the monophyly of Gronovioideae. Although our trnL-trnF data alone provided little information on Gronovioideae aside from strong support for the monophyly of Petalonyx (Fig. 2), the addition of these data to the matK sequences have provided increased support for the Moody et al. (2001)
results as assessed by decay and bootstrap analyses.
Although our results provide support for the monophyly of Loasoideae and Gronovioideae of Urban and Gilg (1900)
and Gilg (1925)
, they are clearly incongruent with their concept of Mentzelioideae. No morphological synapomorphies have been suggested previously for Mentzelioideae, and some authors have hypothesized that they are paraphyletic to Loasoideae and/or Gronovioideae (Brown, 1971
; Hufford, 1988
). Scant attention has been given to the circumscription of Mentzelioideae, perhaps because it lacks the notable morphological elaborations of Loasoideae flowers and simplifications of Gronovioideae flowers. Urban and Gilg (1900)
considered the subfamily to consist of the genera Eucnide, Mentzelia, and Sympetaleia. Thompson and Ernst (1967)
reduced Sympetaleia, a group of three species centered largely in Baja California, to a section of Eucnide. In our results, Eucnide sensu Thompson and Ernst (1967)
is strongly supported as monophyletic for the species sampled, which includes E. aurea, a representative of sect. Sympetaleia, placed as the sister to the rest of the genus. Blake's (1918)
alliance of Schismocarpus with Mentzelioideae was challenged by Ernst and Thompson (1963
, p. 141), who suggested that the genus was "discordant in the Mentzelioideae." The Moody et al. (2001)
results failed to recover a monophyletic Mentzelioideae. Their results, which are replicated in our analyses of the combined matK and trnL-trnF data set, placed Eucnide as the sister to the rest of Loasaceae and Schismocarpus at the next internal node. As noted earlier, Mentzelia was placed as the sister of Gronovioideae.
We examined the Urban and Gilg (1900)
and Gilg (1925)
approach to the circumscription of three subfamilies in Loasaceae using topology constraints. Our constraint topologies forced not only the monophyly of Loasoideae (including Plakothira) and Gronovioideae, which had been recovered in our best trees from the parsimony and ML analyses, but also Mentzelioideae (including Schismocarpus). Our constrained parsimony cladograms were 16 steps longer than the most parsimonious trees from unconstrained analysis of matK and trnL-trnF sequences. The ML topology constrained to force the monophyly of the three subfamilies was rejected under the SH test (P = 0.043).
To achieve a subfamilial classification of Loasaceae based on monophyletic groups, we suggest that Loasoideae and Gronovioideae be maintained as circumscribed by Gilg (1925
, except with the addition of Xylopodia and Plakothira to Loasoideae) but that the circumscription of Mentzelioideae should be revised. We recommend that Mentzelioideae be restricted only to Mentzelia. To accommodate Eucnide and Schismocarpus in a revised subfamilial classification, we recommend that each be included in its own subfamily.
Loasoideae
The monophyly of Loasoideae, which are characterized by complex staminodes, has been unchallenged since Urban and Gilg's (1900)
monograph of the family. Our results provide very strong support for the monophyly of Loasoideae. Phylogenetic relationships found within Loasoideae are consistent with those of Moody et al. (2001)
, although our addition of trnL-trnF characters to the matK sequence data has resulted in greater resolution and better support for clades of the subfamily. Our greater taxon sampling permitted us to evaluate earlier systematic treatments and evolutionary hypotheses that were not examined by Moody et al. (2001)
.
Urban and Gilg (1900)
and Gilg (1925)
recognized in Loasoideae the three tribes Klaprothieae, Kissenieae, and Loaseae. Weigend (1997)
retained Klaprothieae and Loaseae but argued that Kissenia, the only genus of Kissenieae, evolved among Loaseae; thus, he reduced Kissenieae to synonymy in Loaseae. Our results support the monophyly of Loaseae and Klaprothieae; however, we found strong support for a sister group relationship of Klaprothieae and Kissenia, which has not been previously suggested. Our results place the Klaprothieae-Kissenia clade as the sister to the rest of Loasoideae, a group equivalent to the Loaseae of Urban and Gilg (1900)
and Gilg (1925)
. We suggest that the three tribes Klaprothieae, Kissenieae, and Loaseae be maintained in Loasoideae. We emphasize the caveat, however, that little support was found for the two most basal nodes of Loaseae (Fig. 3). The weak nodes at the base of the Loaseae raise the possibility that the tribe could be paraphyletic to the Klaprothieae-Kissenia clade. Such arrangements should be tested as additional phylogenetic data become available.
Within Loaseae, Weigend (1997)
distinguished "lower Loaseae" from "higher Loaseae." In his summary phylogenetic diagram, his "lower Loaseae" was a grade that consisted of Huidobria, Presliophytum, Kissenia, Chichicaste (= Loasa grandis), and Loasa malesherbioides. Weigend's putatively monophyletic "higher Loaseae" included Aosa, Nasa, Loasa sensu stricto (s.str.), Scyphanthus, Cajophora s.str., Cajophora sect. Angulatae, Cajophora sect. Bialatae, and Blumenbachia. Not all of these groups were sampled for our analyses (either unavailable or the specimens sampled had degraded DNA), but our results did not recover either Weigend's "lower Loaseae" grade or "higher Loaseae" clade. For example, our results placed both Presliophytum and Loasa malesherbioides in what would correspond to Weigend's "higher Loaseae."
The monophyly of Loasa as circumscribed broadly by Urban and Gilg (1900)
has been recognized as problematic (Grau, 1997
; Weigend, 1997
) and was not supported by our data. When we forced the monophyly of Loasa s.l., resulting cladograms were 15 steps longer than the unconstrained most parsimonious cladograms, and the constrained ML tree was rejected by the SH test (P = 0.009; Table 1). Recently, Weigend (1997)
argued for a circumscription of Loasa s.str. that would include only Urban and Gilg's (1900)
Loasa sect. Loasa series Acanthifoliae, Macrospermae, Floribundae, Pinnatae, Volubiles, Acaules, and Deserticolae. Weigend (1997)
segregated various portions of Urban and Gilg's Loasa s.l. as new genera, including: Loasa sect. Presliophytum as Presliophytum; Loasa sect. Loasa series Grandiflorae, Alatae, Saccatae, Carunculatae, and the L. venezuelensis group as Nasa; Loasa sect. Loasa series Pusillae, Corymbosae, and Parviflorae as Aosa; and Loasa grandis as Chichicaste. Grau (1997)
resurrected Huidobria for Urban and Gilg's (1900)
Loasa sect. Huidobria. We did not find support for the monophyly of Huidobria; however, the nodes at which these two species diverged had little support. Two characters may support the monophyly of the genus: chromosome numbers of 2n = 36 and the morphology of the staminodial scales of flowers (Grau, 1997
). Our results provide very strong support for the monophyly of Nasa and Presliophytum and are consistent with the recognition of Aosa. Chichicaste was not available to sample for this study.
Weigend (1997)
argued that L. malesherbioides should also be segregated from Loasa, although he made no formal taxonomic change. Our results, which strongly support the placement of L. malesherbioides as the sister of Presliophytum, are consistent with his suggestion. Given the robust placement of L. malesherbioides in our results, a sensible option might be to transfer the species to Presliophytum, expanding slightly the circumscription of this genus.
Loasa acanthifolia, the type species for Loasa, was included in our parsimony analysis of the matK data (this DNA did not amplify for trnL-trnF) and placed with L. pallida as the sister of Cajophora and Scyphanthus. A similar placement for L. pallida was found in the parsimony and ML analyses of the combined matK and trnL-trnF data set. The only other member of Weigend's (1997)
Loasa s.str. sampled for our analyses was L. heterophylla, which was one of the 10 species placed by Urban and Gilg (1900)
in series Macrospermae. Although the placement of L. heterophylla was shown under bootstrap and decay analyses to have only weak support in the parsimony results, its separation from L. pallida indicates that Weigend's Loasa s.str. warrants further revisionary and phylogenetic study.
Urban and Gilg's (1900)
broad circumscription of Cajophora has also been of concern to systematists. Poston and Thompson (1977)
suggested that Cajophora sect. Bialatae was more closely related to Blumenbachia than to other Cajophora. Weigend (1997)
excluded both sects. Angulatae and Bialatae from Cajophora and placed them in Blumenbachia. Material of Cajophora sects. Angulatae and Bialatae was not available to sample for this investigation. Our results provided very strong support for the monophyly of the sampled Cajophora, a set of species that corresponds to the limited boundaries of Weigend's (1997)
Cajophora s.str. Within Cajophora s.str. our sampling was limited, but our results indicated that Urban and Gilg's (1900)
sect. Orthocarpae (including the sampled species C. carduifolia, C. cirsiifolia, and C. chuquitensis) was paraphyletic to both sects. Dolichocarpae (represented by C. clavata) and Platypetalae (represented by C. canarinoides). A sister group relationship of the sampled Cajophora to the monotypic genus Scyphanthus was very strongly supported.
Mentzelia
Our results contributed substantially toward understanding the monophyly of Mentzelia as well as its major clades and their relationships. Brown (1971)
considered Mentzelia to be paraphyletic to the rest of Loasaceae. Hempel (1995
; Hempel and Jansen, 1996
), and Weigend (1997)
considered Mentzelia to be paraphyletic to Eucnide. This phylogenetic concept was not supported by our results, which showed the sister clade relationship of Mentzelia to Gronovioideae to be very strongly supported. We found very strong support for the monophyly of Mentzelia as traditionally circumscribed (e.g., by Urban and Gilg [1900
] and Gilg [1925
]). The monophyly of the traditional Mentzelia and the pattern of relationships displayed among its clades weigh against the recognition of the segregates Acrolasia Presl and Nuttallia Raf. (e.g., Rydberg, 1903
; Davidson, 1916
; Weber and Wittmann, 2001
), which would render paraphyletic a more narrowly circumscribed Mentzelia.
The most recent comprehensive revision of Mentzelia by Darlington (1934)
divided the genus in the four sects. Mentzelia, Bartonia, Trachyphytum, and Bicuspidaria; this contrasts with Urban and Gilg's (1900)
earlier recognition of seven sections. The difference between these two treatments lies particularly in the placements of M. arborescens, M. reflexa, and M. torreyi. Urban and Gilg (1900)
placed each of these species in its own monotypic section: M. arborescens in sect. Dendromentzelia, M. reflexa in sect. Octopetaleia, and M. torreyi in sect. Micromentzelia.
Darlington (1934)
combined sect. Dendromentzelia with sect. Mentzelia, placing M. arborescens in the latter. Thompson and Lewis (1955)
followed the Darlington treatment of M. arborescens, including it in sect. Mentzelia; however, Ernst and Thompson (1963)
treated it as sect. Dendromentzelia. Our results provided modest support for the placement of M. arborescens outside of sect. Mentzelia s.str., which is consistent with Urban and Gilg's (1900)
placement of the species in its own section. Our parsimony analyses of the combined matK and trnL-trnF data found some trees in which M. arborescens was placed as the sister of sect. Mentzelia s.str. and others in which it was placed as the sister of the sects. Bartonia-Bicupidaria-Trachyphytum clade. The ML analysis was consistent with the latter set of parsimony trees, but as with the parsimony results the M. arborescens node had weak support. Thus, although the placement of M. arborescens remains equivocal, our results are consistent with Urban and Gilg's distinction between sects. Dendromentzelia and Mentzelia.
Darlington (1934)
also collapsed Urban and Gilg's (1900)
sects. Octopetaleia and Micromentzelia, placing both M. reflexa and M. torreyi in sect. Bartonia. Thompson (1963)
suggested that Darlington's treatment of these two species might have been "inadvertent" and noted that both "are clearly more similar to species of other sections" (p. 17). Contrary to Darlington's treatment, our results provided very strong support for the exclusion of M. torreyi from sect. Bartonia. When we used constraint topologies to force M. torreyi to form a monophyletic group with the species of sect. Bartonia the resulting topologies were 36 steps longer than the most parsimonious constrained trees, and the SH test rejected this placement as inconsistent with our data (P < 0.0005). In contrast, the constrained analyses were less conclusive in regard to the placement of M. reflexa. Constrained analyses in which M. reflexa was forced to be monophyletic with sect. Bartonia resulted in trees that were only two steps longer than the most parsimonious cladograms from unconstrained analyses, and the SH test did not reject similarly constrained ML topologies (P = 0.991). The results of the constrained analyses for M. reflexa may be more a consequence of the low sequence divergence (short branch length and decay value = 2 in analyses using the combined matK and trnL-trnF data) for sect. Bicuspidaria than a problem inherent solely to M. reflexa. Our results found strong support for the monophyly sect. Bicuspidaria, including M. reflexa, which is consistent with Daniels's (1970)
revision of the section.
Conclusions
Our results provide additional support for the major clades of Loasaceae found by Moody et al. (2001)
. We test key prior hypotheses using the approach of Shimodaira and Hasegawa (1999)
. These tests reject the placement of gronovioids as the sister of the rest of Loasaceae; in contrast to the strongly supported placement of gronovioids as the sister of Mentzelia in our results. We are able to reject the best topologies that modeled the subfamilial taxonomy of Urban and Gilg (1900)
and Gilg (1925)
, making Mentzelioideae as well as Gronovioideae and Loasoideae monophyletic. Our results provide very strong support for the monophyly of Gronovioideae and Loasoideae and demonstrate the paraphyly of Mentzelioideae. We recommend that Mentzelioideae be restricted only to Mentzelia and new subfamilies described to accommodate Eucnide and Schismocarpus. The SH test rejects the best topologies in which we force the monophyly of Loasa s.l. as circumscribed by Urban and Gilg (1900)
and Gilg (1925)
. Our results demonstrate the paraphyly of Loasa s.l. and provide support for several new genera proposed by Weigend (1997)
. Additional taxon sampling is needed to examine support for Loasa s.str. as circumscribed by Weigend (1997)
. Among our novel results is the placement of Kissenia as the sister of Klaprothieae. The Kissenia-Klaprothieae clade forms the sister of Loaseae, providing support for the three tribes recognized by Urban and Gilg (1900)
in Loasoideae. Additional phylogenetically informative data are needed to examine support for the basal nodes of Loasoideae and to resolve relationships among the more terminal branches (e.g., within Cajophora). Similarly, more data are needed to resolve relationships among species within sections of Mentzelia. The lack of resolution within sections of Mentzelia contrasts with the very well supported clades that correspond to the sections of Urban and Gilg (1900)
.
|
FOOTNOTES |
|---|
4 Present address: Section of Ecology and Evolutionary Biology, Division of Biological Sciences, University of California, Davis, California 95616 USA
5 Present address: Compton Herbarium, Kirstenbosch Research Centre, Private Bag X7, Claremont 7735, Cape Town, Republic of South Africa
6 Author for reprint requests (e-mail: hufford{at}mail.wsu.edu
)
|
LITERATURE CITED |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
Baillon H. E. 1888 The natural history of plants, vol. 3. Reeve and Company, London, UK
Bartling F. T. 1825 Ordines Naturales Plantarum. Diettrich, Göttingen, Germany
Blake S. F. 1918 New plants from Oaxaca. Contributions of the Gray Herbarium, New Series 53: 55-65
Bremer K. 1988 The limits of amino acid sequence data in angiosperm phylogenetic reconstruction. Evolution 42: 795-803[CrossRef][Web of Science]
Brown D. K. 1971 A study of floral morphology in the Loasaceae with emphasis on relationships among the subfamilies. Ph.D. dissertation, University of Nebraska, Lincoln, Nebraska, USA
Buckley T. R. 2002 Model misspecification and probabilistic tests of topology; evidence from empirical data sets. Systematic Biology 51: 509-523[CrossRef][Web of Science][Medline]
Buckley T. R. C. Simon H. Shimodaira G. K. Chambers 2001 Evaluating hypotheses on the origin and evolution of the New Zealand alpine cicadas (Maoricicada) using multiple-comparison tests of tree topology. Molecular Biology and Evolution 18: 223-234
Cronquist A. 1978 The biota of the intermountain region in a geohistorical context. Great Basin Naturalist Memoirs 2: 3-15
Daniels G. S. 1970 The floral biology and taxonomy of Mentzelia section Bicuspidaria (Loasaceae). Ph.D. dissertation, University of California, Los Angeles, California, USA
Darlington J. 1934 A monograph of the genus Mentzelia. Annals of the Missouri Botanical Garden 21: 103-227[CrossRef]
Davidson A. 1916 A revision of the western mentzelias. Southern California Academy of Sciences 5: 13-18
Davis W. S. H. J. Thompson 1967 A revision of Petalonyx (Loasaceae) with a consideration of affinities of subfamily Gronovioideae. Madroño 19: 1-18
de Jussieu A. L. 1804 Memoire sur les Loasa. Annales du Muséum National d'Histoire Naturelle 5: 19-27
Dickson A. 1866 On the morphological constitution of the androecium of Mentzelia, and its analogy with that of certain Rosaceae. Transactions of the Botanical Society of Edinburgh 8: 288-298
Donoghue M. J. R. G. Olmstead J. F. Smith J. D. Palmer 1992 Phylogenetic relationships of Dipsacales based on rbcL sequences. Annals of the Missouri Botanical Garden 79: 333-345
Doyle J. J. J. L. Doyle 1987 A rapid DNA isolation procedure for small quantities of fresh leaf tissue. Phytochemical Bulletin 19: 11-15
Endlicher S. L. 1841 Genera Plantarum, Supplement 1. Beck, Vienna, Austria
Eriksson T. 1999 AutoDecay ver. 4.0 (program distributed by the author). Bergius Foundation, Royal Swedish Academy of Sciences, Stockholm, Sweden
Ernst W. R. H. J. Thompson 1963 The Loasaceae in the southeastern United States. Journal of the Arnold Arboretum 44: 138-142
Felsenstein J. 1985 Confidence limits on phylogenies: an approach using the bootstrap. Evolution 39: 783-791[CrossRef][Web of Science]
Florence J. 1985 Sertum polynesicum. I. Plakothira Florence (Loasaceae), genre nouveau des îles Marquises. Adansonia 7: 239-245
Gilg E. 1895 Loasaceae. In A. Engler and K. Prantl [eds.], Die natürlichen Pflanzenfamilien, 3 (6a), 100–121. Engelmann, Leipzig, Germany
Gilg E. 1925 Loasaceae. In A. Engler and K. Prantl [eds.], Die natürlichen Pflanzenfamilien, 2nd ed., 21, 522–543. Engelmann, Leipzig, Germany
Goldman N. J. P. Anderson A. G. Rodrigo 2000 Likelihood-based tests of topologies in phylogenetics. Systematic Biology 49: 652-670[CrossRef][Web of Science][Medline]
Graham A. 1993 History of the vegetation: Cretaceous (Maastichtian)-Tertiary. In Flora of North America Editorial Committee [eds.], Flora of North America, vol. 1, 57–70. Oxford University Press, Oxford, UK
Grau J. 1997 Huidobria, eine isolierte Gattung der Loasaceae aus Chile. Sendtnera 4: 77-93
Hahn 2002 A molecular phylogenetic study of the Palmae (Arecaceae) based on atpB, rbcL and 18S nr DNA sequences. Systematic Biology 51: 92-112[CrossRef][Web of Science][Medline]
Hempel A. L. 1995 Molecular systematics of the Loasaceae. Ph.D. dissertation, University of Texas, Austin, Texas, USA
Hempel A. L. R. K. Jansen 1996 Derivation of a florally diverse genus, Eucnide, from within Mentzelia. American Journal of Botany 83: (suppl.) 162 (Abstract) .[CrossRef][Web of Science]
Hempel A. P. A. Reeves R. G. Olmstead R. K. Jansen 1995 Implications of rbcL sequence data for higher order relationships of the Loasaceae and the anomalous aquatic plant Hydrostachys (Hydrostachyaceae). Plant Systematics and Evolution 194: 25-37[CrossRef][Web of Science]
Hirmer M. 1916 Beiträge zur Morphologie polyandrischer Blüten. Flora 110: 140-192
Huelsenbeck J. P. D. M. Hillis R. Nielsen 1996 A likelihood ratio test of monophyly. Systematic Biology 45: 546-558[CrossRef]
Hufford L. 1988 Systematics of Loasaceae: a consideration of familial circumscription and a cladistic analysis of generic relationships. American Journal of Botany 75: (suppl.) 181.
Hufford L. 1990 Androecial development and the problem of monophyly of Loasaceae. Canadian Journal of Botany 68: 402-419[Web of Science]
Hufford L. M. L. Moody D. E. Soltis 2001 A phylogenetic analysis of Hydrangeaceae based on sequences of the plastid gene matK and their combination with rbcL and morphological data. International Journal of Plant Sciences 162: 835-846[CrossRef]
Johnson L. A. D. E. Soltis 1995 Phylogenetic inference in Saxifragaceae sensu sticto and Gilia (Polemoniaceae) using matK sequences. Annals of the Missouri Botanical Garden 82: 149-175[CrossRef][Web of Science]
Kishino H. T. Miyata M. Hasegawa 1990 Maximum likelihood inference of protein phylogeny and the origin of chloroplasts. Journal of Molecular Evolution 30: 151-160
Leaché A. D. T. W. Reeder 2002 Molecular systematics of the eastern fence lizard (Soeloporus undulatus): a comparison of parsimony, likelihood, and Bayesian approaches. Systematic Biology 51: 44-68[CrossRef][Web of Science][Medline]
Leins P. W. Winhard 1973 Entwicklungsgeschichtliche Studien an Loasaceen Blüten. Oesterreichische Botanische Zeitschrift 122: 145-165[CrossRef]
Moody M. L. L. Hufford 2000 Floral ontogeny and morphology of Cevallia, Fuertesia, and Gronovia (Loasaceae subfamily Gronovioideae). International Journal of Plant Sciences 161: 869-883[CrossRef]
Moody M. L. L. Hufford D. E. Soltis P. S. Soltis 2001 Phylogenetic relationships of Loasaceae subfamily Gronovioideae inferred from matK and ITS sequence data. American Journal of Botany 88: 326-336
Olmstead R. G. K.-J. Kim R. K. Jansen S. J. Wagstaff 2000 The phylogeny of Asteridae sensu lato based on chloroplast ndhF gene sequences. Molecular Phylogenetics and Evolution 16: 96-112[CrossRef][Web of Science][Medline]
Payer J.-B. 1857 Traité de d'organogénie comparée de la fleur. Stechert-Hafner, New York, New York, USA (1966 reprint)
Posada D. K. A. Crandall 1998 Modeltest: testing the model of DNA substitution. Bioinformatics 14: 817-818
Poston M. E. J. W. Nowicke 1993 Pollen morphology, trichome types, and relationships of the Gronovioideae (Loasaceae). American Journal of Botany 80: 689-704[CrossRef][Web of Science]
Poston M. E. H. J. Thompson 1977 Cytotaxonomic observations in Loasaceae subfamily Loasoideae. Systematic Botany 2: 28-35[CrossRef][Web of Science]
Rambaut A. 1996 Se-Al: sequence alignment editor, version 1. http://evolve.zoo.ox.ac.uk/Se-Al/Se-Al.html. Oxford, UK
Reichenbach L. 1837 Handbuch des natürlichen Pflanzensystems. Arnold, Dresden, Germany
Ronse-Decraene L.-P. E. Smets 1987 The distribution and the systematic relevance of the androecial characters oligomery and polymery in the Magnoliophytina. Nordic Journal of Botany 7: 239-253
Rydberg P. A. 1903 Some generic segregations. Bulletin of the Torrey Botanical Club 30: 271-281[CrossRef]
Shimodaira H. M. Hasegawa 1999 Multiple comparisons of log-likelihoods with applications to phylogenetic inference. Molecular Biology and Evolution 16: 1114-1116[Web of Science]
Silberman J. D. A. G. B. Simpson J. Kulda I. Cepicka V. Hampl P. J. Johnson A. J. Roger 2002 Retortamonad flagellates are closely related to diplomonads—implications for the history of mitochondrial function in eukaryote evolution. Molecular Biology and Evolution 19: 777-786
Swofford D. L. 2002 PAUP*: phylogenetic analysis using parsimony (*and other methods). Version 4.0. Sinauer Associates, Sunderland, Massachusetts, USA
Taberlet P. L. Gielly G. Pautou J. Bouvet 1991 Universal primers for the amplification of three non-coding regions of chloroplast DNA. Plant Molecular Biology 17: 1105-1109[CrossRef][Web of Science][Medline]
Thompson H. J. 1963 Cytotaxonomic observations on Mentzelia, sect. Bartonia (Loasaceae). Madroño 17: 16-22
Thompson H. J. W. R. Ernst 1967 Floral biology and systematics of Eucnide (Loasaceae). Journal of the Arnold Arboretum 48: 56-88[Web of Science]
Thompson H. J. H. Lewis 1955 Chromosome numbers in Mentzelia (Loasaceae). Madroño 13: 102-107
Turmel M. M. Ehara C. Otis C. Lemieux 2002 Phylogenetic relationships among streptophytes as inferred from chloroplast small and large subunit rRNA gene sequences. Journal of Phycology 38: 364-375[CrossRef][Web of Science]
Urban I. E. Gilg 1900 Monographia Loasacearum. Nova Acta Academiae Caesareae Leopoldino-Carolinae Germanicae Naturae Curiosorum 76: 1-370
Van Devender T. R. R. S. Thompson J. L. Betancourt 1987 Vegetation history of the deserts of southwestern North America; the nature and timing of the Late Wisconsin-Holocene transition. In W. F. Ruddiman, and H. E. Wright, Jr. [eds.], North America and adjacent oceans during the last glaciation, 323–352. Geological Society of America, Boulder, Colorado, USA
Weber W. A. R. G. Wittmann 2001 Colorado flora: eastern slope, 3rd ed. University Press of Colorado, Boulder, Colorado, USA
Weigend M. 1997 Nasa and the conquest of South America. Ph.D. dissertation, Ludwig-Maximilians-University, Munich, Germany
Weigend M. J. Kufer A. A. Müller 2000 Phytochemistry and the systematics and ecology of Loasaceae and Gronoviaceae (Loasales). American Journal of Botany 87: 1202-1210
Xiang Q.-Y. M. L. Moody D. E. Soltis C.-Z. Fan P. S. Soltis 2002 Relationships within Cornales and circumscriptions of Cornaceae—matK and rbcL sequence data and effects of outgroups and long branches. Molecular Phylogenetics and Evolution 24: 35-57[CrossRef][Web of Science][Medline]
Xiang Q.-Y. D. E. Soltis D. R. Morgan P. S. Soltis 1993 Phylogenetic relationships of Cornus sensu lato and putative relatives inferred from rbcL sequence data. Annals of the Missouri Botanical Garden 80: 723-734[CrossRef][Web of Science]
Xiang Q.-Y. D. E. Soltis P. S. Soltis 1998 Phylogenetic relationships of Cornaceae and close relatives inferred form matK and rbcL sequences. American Journal of Botany 85: 285-297[Abstract]
This article has been cited by other articles:
|
|
 |
|
  M. Ackermann, M. Achatz, and M. Weigend Hybridization and crossability in Caiophora (Loasaceae subfam. Loasoideae): Are interfertile species and inbred populations results of a recent radiation? Am. J. Botany, September 1, 2008; 95(9): 1109 - 1121. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. ACKERMANN and M. WEIGEND Nectar, Floral Morphology and Pollination Syndrome in Loasaceae subfam. Loasoideae (Cornales) Ann. Bot., September 1, 2006; 98(3): 503 - 514. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. Shaw, E. B. Lickey, J. T. Beck, S. B. Farmer, W. Liu, J. Miller, K. C. Siripun, C. T. Winder, E. E. Schilling, and R. L. Small The tortoise and the hare II: relative utility of 21 noncoding chloroplast DNA sequences for phylogenetic analysis Am. J. Botany, January 1, 2005; 92(1): 142 - 166. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
