| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Department of Botany, Washington State University, Pullman, Washington 99164-4238 USA
Received for publication November 11, 1999. Accepted for publication June 29, 2000.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
Key Words: Cevallia • Fuertesia • Gronovia • Gronovioideae • ITS • Loasaceae • matK • Petalonyx.
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
) that are distributed primarily in the New World, where they extend from the southern and western United States through much of Mesoamerica and South America. Only two genera are located outside of the New World: Plakothira in the Marquesas Islands of the South Pacific (Florence, 1985
) and Kissenia in the southwest and horn of Africa and the southwestern Arabian Peninsula (Dandy, 1926
). The morphological diversity of the family has led some to question its monophyly (Payer, 1857
; Leins and Winhard, 1973
; Cronquist, 1981
). Both morphological (Hufford, 1990
) and molecular (Hempel et al., 1995
) data, however, have supported the monophyly of Loasaceae as traditionally circumscribed by Urban and Gilg (1900)
and Gilg (1925)
. Recent phylogenetic analyses have resolved the systematic placement of Loasaceae in the Cornales (Downie and Palmer, 1992
; Hufford, 1992
; Albach, 1998
) as the sister clade to Hydrangeaceae (Xiang et al., 1993
; Hempel et al., 1995
; Soltis et al., 1997
; Xiang, Soltis, and Soltis, 1998
). Thus, considerable advances have been made in Loasaceae systematics. However, evolutionary relationships within the family remain unclear. This is a consequence not only of the limitations of previous phylogenetic analyses to discern evolutionary relationships in the family (Hufford, 1988b
; Hempel et al., 1995
), but also of the patterning of diversity in the group.
Discussions of relationships in Loasaceae have been influenced by the subfamilial classification of Urban and Gilg (1900)
, who emphasized floral states in the recognition of three subfamilies: Loasoideae, Mentzelioideae, and Gronovioideae (Table 1). Mentzelioideae and Loasoideae are polystemonous (except Schismocarpus, which is diplostemonous), multicarpellate, multiovulate (having at least two ovules), and have dehiscent fruits (except Kissenia, which has dry, indehiscent fruits). Gronovioideae, in contrast, have simpler flowers that have a haplostemonous androecium, a uniovulate gynoecium, and indehiscent fruits. The relative floral simplicity in this subfamily has limited our ability, using morphological characters (Hufford, 1988b
, and unpublished data), to resolve the relationships of gronovioid genera. Moreover, it has been suggested that the floral characters of Gronovioideae, because they involve simplifications alone, do not provide sufficient support for the monophyly of the subfamily (Davis and Thompson, 1967
).
|
were the first to question the monophyly of Gronovioideae. They suggested that Gronovioideae consisted of three divergent groups (1. Petalonyx, 2. Cevallia, 3. Gronovia-Fuertesia) "whose affinities lie with other genera in the family" (Davis and Thompson, 1967
, p. 11). They considered haplostemony and a uniovulate ovary to be given undue weight in Urban and Gilg's (1900
, also Gilg's, 1925
) circumscription of Gronovioideae. Davis and Thompson (1967)
emphasized that genera of Gronovioideae vary greatly in haploid chromosome number (Petalonyx, n = 23; Cevallia, n = 13; Gronovia, n = 37; Fuertesia, n is unknown) and stamen morphology. Davis and Thompson (1967)
did not provide explicit alternative hypotheses of relationships for gronovioid genera, although they did suggest that Petalonyx was allied with Mentzelioideae. Similarly, Poston and Nowicke (1993)
suggested that pollen and trichome character states do not support the Gronovioideae as a group of closely related species. They indicated that Petalonyx may be more closely related to Mentzelioideae than to other gronovioids because it shares a striate tectum and endexine accumulation under an incurved colpus margin with the pollen of Mentzelioideae. Poston and Nowicke (1993)
emphasized that Cevallia, Fuertesia, and Gronovia differ from Petalonyx in having trichomes that vary from the two common retrorsely barbed types found in Mentzelioideae. Hempel's (1995)
phylogenetic analyses of Loasaceae based on sequences of the cpDNA genes ndhF and rbcL did not support the monophyly of Gronovioideae; however, her sampling was limited to seven taxa.
Despite suggestions that Gronovioideae may not be monophyletic (Davis and Thompson, 1967
; Poston and Nowicke, 1993
; Hempel, 1995
), Cevallia, Fuertesia, Gronovia, and Petalonyx among Loasaceae uniquely share haplostemony and a uniovulate ovary. The novelties found among the gronovioid genera, including pseudosympetaly in some Petalonyx, stigma and anther elaboration in Cevallia, and divergent chromosome numbers in the genera for which counts are available (Davis and Thompson, 1967
), are not sufficient to argue against monophyly. The relative floral simplicity of gronovioid genera is unusual in Loasaceae in which elaboration, particularly in the androecium, has been a key aspect of diversification. The floral morphological disparity between Gronovioideae and other Loasaceae makes them particularly interesting. Notably, the possible monophyly of Gronovioideae raises questions about how modes of floral diversification in a clade distinguished by relative floral simplicity differ from those in clades of the family characterized by more elaborate flowers.
Adequate testing of the monophyly of Gronovioideae requires broad taxon sampling among Loasaceae. This sampling is critical given our limited understanding of relationships in the family. Among the three subfamilies, only the monophyly of Loasoideae has not been questioned. Centrifugal stamen initiation and complex, synorganized staminodes (called nectar scales) support the monophyly of Loasoideae; however, relationships among genera in this group are not well understood. Genera of subfamily Loasoideae, such as Loasa and Cajophora as circumscribed by Urban and Gilg (1900)
, have been suggested to be paraphyletic (Poston and Thompson, 1977
; Poston, 1979
; Weigend, 1997a, b
). Brown (1971)
questioned the monophyly of Gilg's (1925)
Mentzelioideae. He suggested that Eucnide, as well as subfamilies Loasoideae and Gronovioideae, were nested within Mentzelia. Phylogenetic analyses of the family by Hempel (1995)
indicated that Eucnide nests within Mentzelia section Mentzelia.
We test the monophyly of Gronovioideae by implementing a broad sampling of Loasaceae and using DNA sequence data from the chloroplast gene matK to resolve evolutionary relationships among major clades in the family. The utility of matK in resolving relationships among "genera" has been previously demonstrated (e.g., Johnson and Soltis, 1994, 1995
; Steele and Vilgalys, 1994
; Xiang, Soltis, and Soltis, 1998
). To provide additional data to assist in resolving relationships among gronovioids, DNA sequences of a nuclear ribosomal internal transcribed spacer region (ITS) are used. ITS sequences, which evolve more rapidly than those of matK, have been valuable in formulating hypotheses of relationships among closely related genera and among species (e.g., Wojciechowski et al., 1993
; Baldwin et al., 1995
; Eriksson and Donoghue, 1997
). The primary goals in using these molecular data sets were to (1) resolve the evolutionary relationships of major clades of Loasaceae as a means of testing the monophyly of the Gronovioideae and (2) identify sister-group relationships for genera of Gronovioideae.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
). It was designed to provide reasonable sister clade options for genera of Gronovioideae, given the hypotheses of Davis and Thompson (1967)
and Poston and Nowicke (1993)
that the subfamily was not monophyletic, or for the most reasonable placement of a monophyletic Gronovioideae. After the preliminary analyses based on matK sequences found a monophyletic Gronovioideae, all species of the subfamily, except G. grandiflora, were sampled for ITS-1 sequences to provide additional data to help resolve sister-group relationships among the genera.
Outgroup selection was based on earlier studies that showed a sister-group relationship between Hydrangeaceae and Loasaceae and their placement in Cornales (Xiang et al., 1993
; Hempel, 1995
; Soltis, Xiang, and Hufford, 1995
; Xiang, Soltis, and Soltis, 1998
). Based on those results, Hydrangeaceae, core cornoids, and more distantly related members of Ericales s. l. were chosen as outgroups. New sequences of Hydrangeaceae and Hydrostachys multifida were obtained (Appendix), and these were combined with sequences from GenBank for Actinidia chinensis, Cornus chinensis, Alangium platanifolium, Davidia involucrata, Fouquieria splendens, and Sarracenia purpurea to use as outgroups for the broad analysis based on matK sequences. All of these outgroup taxa as well as various combinations of selected subsets of them were used in preliminary phylogenetic analyses, but each analysis resulted in the same ingroup topology. Thus, the final analyses were conducted using only Jamesia, Cornus, and Curtisia as outgroups to reduce the number of equally parsimonious trees that result from alternative placements among the outgroups. For the analysis of species-level relationships in the Gronovioideae using ITS-1 sequences (and a combination of matK and ITS-1), three species (Mentzelia torreyi, M. hispida, and M. nuda) were selected as outgroups based on our matK results.
DNA isolation and sequencing
Total DNAs were extracted from either dried herbarium or fresh specimens using the miniprep procedure of Doyle and Doyle (1987)
as modified by Soltis et al. (1991)
. Double-stranded DNAs were amplified using the polymerase chain reaction (PCR), following Johnson and Soltis (1995)
, with matK-710F and trnK-2R as PCR primers for most matK sequences. Material from some herbarium specimens required two different sets of primers to obtain the same length of DNA: matK-710F/matK-1848R and matK-1470F/trnK-2R. Double-stranded PCR products for cycle sequencing used matK-710F, matK-1470R, matK-1848R, and matK-1713F primers. When greater sequence overlap was sought or the above primers proved ineffective, presumably due to primer divergence, one or more of the following primers were used: matK-1235R, matK-1470F, and trnK-2R. All primers were from Johnson and Soltis (1995)
.
The procedure described above was also followed for ITS-1, with N-nc18S10 and ITS2, ITS4 or C26A (White et al., 1990
; Soltis et al., 1997
) used to amplify double-stranded DNA. The following cycling protocol and an MJ Research PTC-100 thermal cycler were used for ITS amplification: 97°C for 1 min followed by 25 cycles of 95°C for 30 sec, 58°C for 30 sec, and 72°C for 1 min, with a final extension at 72°C for 10 min. The sequencing primers used were ITS2 and N-nc18S10 (White et al., 1990
; Soltis et al., 1997
). Sequences of both ITS and matK were obtained using an ABI model 377 automated DNA sequencer.
Phylogenetic analyses
Data sets
All sequences were manually aligned in SeqApp (Gilbert, 1992
). Data matrices were prepared in MacClade (Maddison and Maddison, 1992
).
Parsimony
Parsimony analyses used heuristic searches with random taxon addition sequences (replicated 100 times), tree bisection-reconnection (TBR) branch-swapping with all character states unordered and equally weighted, and saving all most parsimonious trees. Indels were treated in two ways: (1) as missing data and (2) as binary characters. Multiple most parsimonious trees were combined as a strict consensus tree. Data were analyzed using PAUP 4.0b1 (Swofford, 1998
) on Macintosh Power PCs. Standard measures of homoplasy (consistency index [CI], retention index [RI], and rescaled consistency index [RC]) were calculated. Measures of homoplasy were assessed with invariable characters removed from the data set. Bootstrap (BS) analyses (Felsenstein, 1985
) were conducted using heuristic searches (200 replicates for matK and 500 replicates for ITS-1 and the combined ITS-1/matK data) in PAUP 4.0b1. Decay indices (DI) (Bremer, 1988
; Donoghue et al., 1992
) were estimated using the "converse constraint" method of Baum, Sytsma, and Hoch (1994)
as implemented in PAUP 4.0b1. The congruence of the matK and ITS-1 data sets for Gronovioideae was assessed using incongruence length difference (ILD) (Farris et al., 1995
; Cunningham, 1997
) calculated using the partition homogeneity test in PAUP 4.0b1.
Maximum likelihood
The long branches of Schismocarpus, Cevallia, and Mentzelia torreyi in our parsimony results compelled us to use maximum likelihood analysis (ML) to account for unequal substitution rates (Huelsenbeck, 1995
), which can create problems for the placement of long branches under parsimony (Felsenstein, 1978
). A matK data matrix for the ML analysis was constructed by reducing the number of taxa to 24 from those used for the parsimony analyses to permit a reasonable computation time. The taxon sampling for this data set was established to permit variation among deep nodes, and all genera, except Aosa and Plakothira, were included. The matK sequence for Plakothira was nearly identical to that of Klaprothia fasciculata (uncorrected sequence divergence <1%). All possible characters (variable and invariable) were included in this analysis in which indels were treated as binary characters and sequence regions that had missing data were removed. The combined ITS-1 and matK data set (same taxa as in the parsimony analysis of the combined data), in which ITS-1 indels were treated as binary characters, were also analyzed using ML.
Maximum likelihood can underestimate the amount of change in long branches if among-site rate variation is ignored in the likelihood model (Waddell, 1995
; Yang, 1996
); thus, different model parameters were examined to find the model that best fits our data. An iterative search strategy proposed by Swofford et al. (1996)
and Sullivan, Markert, and Kilpatrick (1997)
was used to evaluate 16 models of molecular evolution. Model parameters were optimized on an initial set of most parsimonious trees (six from the matK data and ten from the combined matK and ITS-1 data) that were found in searches that used parsimony. The best-fit evolutionary model GTR + I + 
had the highest likelihood for the matK (ln L = -5698.578) and the combined matK and ITS-1 (ln L = -2842.218) data sets. This model, however, had only one more free parameter than the GTR + model (ln L = -5698.638 for the matK data; ln L = -2842.262 for the combined data sets). A standard likelihood-ratio test indicated that the improvement of the likelihood score associated with the one additional parameter in the GTR + I + model was not significant for either the matK (
2 = 0.1200, P > 0.05) or combined data sets (2 = 0.0868, P > 0.05). Thus, the simpler GTR + model was selected to reduce computation time in the ML analyses (likelihood estimates of model parameters are presented in Table 2). An ML search (heuristic searches with ten random input orders and TBR branch swapping) was conducted using the chosen GTR + substitution model. Nodal support was estimated using bootstrap analyses (100 replicates) under the best-fit model.
|
study were investigated with the Kishino-Hasegawa test (K-H test; Kishino and Hasegawa, 1989
) as implemented in PAUP 4.0b1. The test uses the standard error of differences in single-site likelihoods between two tree topologies to estimate the significance of an observed difference between them. Alternative trees were reconstructed by implementing a backbone constraint, consisting of equivalent taxa to those used in the Hempel (1995)
analyses in which topologies conflicted with our results. The entire data set was then analyzed using the best-fit model under likelihood with the backbone constraint enforced to find the most likely topology given the constraint backbone. Three backbone constraints were enforced to include (1) both the Eucnide-Mentzelia section Mentzelia and Schismocarpus-Gronovia clades; (2) only an Eucnide-Mentzelia section Mentzelia clade; and (3) only a Schismocarpus-Gronovia clade. |
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
|
|
was paraphyletic. Eucnide was placed as the sister of the rest of the family, and Schismocarpus branched from the next more distal node. A monophyletic Mentzelia (BS = 99, DI = 6) was the sister of Gronovioideae (BS = 97, DI = 3). Several clades of Mentzelia had robust support (BS of 87% or higher for all). Species sampled from sections Bartonia and Bicuspidaria formed a clade whose members share a 6-bp deletion (BS = 100, DI = 7; Fig. 2). They were the sister of section Trachyphytum. The Bartonia-Bicuspidaria-Trachyphytum clade was the sister of sections Mentzelia and Dendromentzelia. Mentzelia torreyi was sister to the rest of the genus.
|
Subfamily Loasoideae was characterized by a basal polytomy consisting of Kissenia, Huidobria chilensis, Klaprothieae (Klaprothia and Plakothira), and a clade of all other Loasoideae in the strict consensus of all trees. In the strict consensus of the island consisting of 33 most parsimonious trees, Huidobria chilensis and Kissenia formed a clade that was sister to all other Loasoideae. In the strict consensus of the island consisting of 66 most parsimonious trees, a clade including Kissenia and Klaprothieae was sister to all other Loasoideae. Klaprothieae were strongly supported, including a 6-bp deletion, as monophyletic (BS = 100, DI = 31; Fig. 2). Loasa s. l. (sensu Urban and Gilg, 1900
) was found to be paraphyletic. The monophyly of Cajophora was well supported (BS = 99, DI = 6), and it forms the sister group to Scyphanthus (BS = 73, DI = 2). The relationships of the Cajophora-Scyphanthus clade to Blumenbachia and particular species of Loasa were not resolved. Aosa and Nasa formed a polytomy with the Loasa-Blumenbachia-Cajophora-Scyphanthus clade.
Maximum likelihood analysis of matK
Likelihood analysis converged on one tree (ln L = -5698.638). The topology (Fig. 3) was consistent with the parsimony analysis using 41 ingroup taxa (Fig. 1). It resolved Kissenia as the sister of Huidobria chilensis and Blumenbachia as the sister of the Cajophora-Scyphanthus clade, although support for both relationships was modest. Nodal support as assessed by bootstrap analyses was generally greater than found for the parsimony analysis (Fig. 3).
Alternative topologies were examined using a K-H test (Kishino and Hasegawa, 1989
) to explore the significance of an observed difference in likelihood between two alternative phylogenetic hypotheses. The likelihood values of the best maximum likelihood trees that were constrained to reflect Hempel's (1995)
results, which placed Schismocarpus as the sister of Gronovia and nested Eucnide within Mentzelia section Mentzelia, were significantly worse than the value for the best maximum likelihood tree (Fig. 3) based on the matK sequence data as determined by the K-H test (P < 0.0001; Table 4).
|
Parsimony analysis of ITS-1
When indels were treated as binary characters, then a single tree of 196 steps (CI = 0.85, RI = 0.73, RC = 0.62; Fig. 4) was found. When indels were treated as missing data, three trees of 186 steps (CI = 0.85, RI = 0.72, RC = 0.62) were found, and these differ only from the trees found when indels were included in not resolving the core gronovioid clade.
|
Phylogenetic analyses of combined matK and ITS-1 data
The ILD test indicated that the matK and ITS-1 data sets were not significantly more incongruent than random partitions (P > 0.38 without indels as binary characters; P > 0.31 with indels as binary characters). After alignments, the combined ITS and matK data matrix of 11 taxa (same as for ITS-1 alone) had 312 variable characters, including 119 parsimony-informative characters (including indels).
Phylogenetic analyses that combined matK and ITS-1 data for Gronovioideae (with indels treated as missing data or as binary characters) resulted consistently in trees that were the same as the ITS-1 tree derived from the analysis in which indels were treated as binary characters (Fig. 4). When indels were included as binary characters, parsimony analysis of the combined data set resulted in three trees of 397 steps (CI = 0.89, RI = 0.83, RC = 0.75; strict consensus in Fig. 4). Combining the matK and ITS-1 data increased the support for most clades (as measured by bootstrap and decay values) relative to the results of either data set alone (Fig. 4). ML analysis of the combined data resulted in one best tree (ln L = -2842.218) that had the same topology as that from the parsimony analyses, although branch robustness varied slightly from the parsimony results (Fig. 4).
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
, and Loasoideae. Key insights from this analysis include support for (1) the sister-group relationship of Mentzelia and Gronovioideae; (2) the monophyly of Loasoideae; (3) the placement of Schismocarpus as the sister of the clade that consists of Mentzelia, Gronovioideae, and Loasoideae; and (4) the placement of Eucnide as the sister to the rest of the family.
Eucnide and Schismocarpus
The placement of Eucnide and Schismocarpus as branches at successive basal nodes reflects Hufford's (1989b)
suggestion that their shared morphological states could be symplesiomorphies. Our results for Schismocarpus and Eucnide differ considerably from those found by Hempel (1995)
, whose cladograms place Eucnide in Mentzelia section Mentzelia and Schismocarpus as the sister of Gronovia. If we model these placements by modifying our most parsimonious cladograms, then the lengths of the parsimony trees increase by 27 and 21 steps, respectively (Table 4). Hempel's (1995)
results have limited robustness as measured by bootstrap values, and there are no distinctive morphological synapomorphies that support her results for the placement of Eucnide and Schismocarpus. We did find, however, that Schismocarpus and Gronovia share a deletion, which was the only homoplastic indel in our matK data (Fig. 2). Discrepancies between our results and those of Hempel (1995)
are possibly due to the limited taxon sampling in her analysis and/or different phylogenetic signals in the data she used.
Given the highly divergent placement of Eucnide and Schismocarpus in our results relative to that of Hempel (1995)
, and the higher sequence divergence of the latter genus (Fig. 2), we applied ML to offer an alternative strategy to discover the relationships of Schismocarpus and Eucnide. The ML results are consistent with those of our parsimony analysis, and the clades are robust as measured by bootstrap values (Fig. 3). In addition, the likelihood values of the best likelihood trees constrained to Hempel's (1995)
placement of Schismocarpus and Eucnide, as determined by the K-H test (P < 0.001; Table 4; Kishino and Hasegawa, 1989
), were significantly worse than those of the best ML tree (Fig. 3).
Loasoideae
The DNA sequence data, including a 6-bp deletion (Table 3), complement a suite of morphological characters, especially androecial states, that support the monophyly of subfamily Loasoideae. This group, in which most species are distributed primarily in Central and South America, is probably the least known in the family and generic circumscriptions in it have been controversial (Brown and Kaul, 1981
; Poston and Thompson, 1977
; Weigend, 1997a
). The delimitation of Loasa (= Loasa s. l., Fig. 1) used in Urban and Gilg's (1900)
monograph has been followed by most workers. Grau (1997)
and Weigend (1997b)
have advocated segregating Huidobria and Presliophytum as well as the new genera Aosa, Chichicaste, and Nasa from Loasa s. l. Chichicaste and Presliophytum were not sampled for our study, although representatives of Aosa, Huidobria, and Nasa were included. As shown in the strict consensus tree (Fig. 1), Huidobria is among the clades that branch from the polytomy at the base of Loasoideae (Fig. 1). The two sampled species of Nasa are a monophyletic group that branches from a polytomy that also includes the one sampled species of Aosa and a clade that consists of Loasa s. s., Blumenbachia, Cajophora, and Scyphanthus.
Circumscription of Cajophora, especially in regard to Blumenbachia, has also been debated (Poston, 1979
; Weigend, 1997b
). An alliance of Cajophora and Blumenbachia has been generally accepted because they share helically twisted fruits (Urban and Gilg, 1900
). Weigend (1997b)
, however, advocated submerging Cajophora and Scyphanthus, but not Blumenbachia, in Loasa s. s. Although the sampling of species in Cajophora for our study has not been sufficient to examine problems of its circumscription in regard to Blumenbachia, our results show that the sampled species are strongly supported as a monophyletic group (Figs. 1 and 3). The hypothesis of monophyly for Cajophora is supported by its unique possession among Loasoideae of a base chromosome number of n = 8, whereas all other sampled Loasoideae vary from n = 12 to n = 24 (Poston and Thompson, 1977
). Additional taxon sampling in Cajophora, especially among all of the recognized sections, will be important before we can be assured of its monophyly. Our results place the poorly known Chilean genus Scyphanthus as the sister clade of Cajophora. The results also show that Cajophora, Scyphanthus, and Blumenbachia are nested among clades of a paraphyletic Loasa s. s. (sensu Weigend, 1997b
).
Hufford (1990)
suggested Klaprothieae (Table 1) may be the most plesiomorphic members of Loasoideae. Our results identify Klaprothieae as a clade that diverges from a polytomy at the base of Loasoideae. Weigend's (1997c
, p. 42) contention that Klaprothieae are "firmly connected to Loasaceae via Loasa [= Aosa fide Weigend, 1997b
] plumieri and Plakothira frutescens" is not supported by our results (Fig. 1).
The placement of Kissenia is equivocal in the parsimony analysis: it can be placed as the sister of either Huidobria chilensis or Klaprothieae. The ML trees place Kissenia as the sister of H. chilensis, although this relationship is weakly supported. Kissenia and H. chilensis notably lack a 6-bp deletion found in all other Loasoideae. Poston (1979)
previously suggested an affinity of H. chilensis with Kissenia based on shared nectar scale morphology. Presliophytum also has relatively simple nectar scales as are found in Huidobria and Kissenia (Weigend, 1997b
), and its inclusion in future phylogenetic studies will be critical for understanding the early diversification of Loasoideae.
Mentzelia
Support provided by the matK data for the monophyly of Mentzelia is consistent with the distribution of iridoid compounds. For example, Jensen, Mikkelsen, and Nielsen (1981)
and Damtoft, Jensen, and Nielsen (1993)
showed that sampled species of Mentzelia were unique among Loasaceae in having iridoids (derived from epoxydecaloside and decaloside) other than loganin and/or secoiridoids; however, all genera have not been sampled and few species of Mentzelia have been sampled.
Broad patterns of relationships among species in Mentzelia remain largely unexplored. Four to six sections have been recognized by recent workers. The different numbers of sections recognized depend on (1) whether M. arborescens is segregated as the monotypic section Dendromentzelia or included in section Mentzelia and (2) whether M. torreyi is segregated as the monotypic section Micromentzelia or included in section Bartonia.
Our results provide important preliminary insights into these problems of sectional delimitation and relationships among clades in Mentzelia. Urban and Gilg (1900)
placed M. torreyi in the monotypic section Micromentzelia, whereas Darlington (1934)
placed it in section Bartonia. Our results place M. torreyi as the sister to the rest of Mentzelia and are not consistent with Darlington's (1934)
inclusion of it in section Bartonia. Thus, the continued recognition of the monotypic section Micromentzelia is supported. Urban and Gilg (1900
; Gilg, 1925
) circumscribed the monotypic section Dendromentzelia for the arborescent M. arborescens, whereas Darlington (1934)
placed M. arborescens in section Mentzelia. Hill (1976)
retained M. arborescens as section Dendromentzelia because its seed coat characteristics suggested unclear affinities with either section Mentzelia or Bartonia. Our results (Fig. 1) provide preliminary support for the monophyly of M. arborescens and section Mentzelia (represented by M. hispida); however, additional species of section Mentzelia must be sampled before we can assess reasonably their relationship to M. arborescens and Darlington's (1934)
circumscription of the section.
Only sections Trachyphytum (Zavortink, 1966
) and Bicuspidaria (Daniels, 1970
) have been revised since Darlington's (1934)
monograph of the genus. Our results indicate that sections Trachyphytum and Bicuspidaria form a monophyletic group with section Bartonia, which is consistent with results from Hempel (1995)
. Hempel's (1995)
results from the phylogenetic analysis of ITS data placed sections Bartonia, Bicuspidaria, and Trachyphytum in a clade that was sister to section Mentzelia. These congruent findings differ from the suggestion of Hill (1976)
, who indicated that section Bartonia was more closely related to section Dendromentzelia than to sections Bicuspidaria and Trachyphytum. Our data show that the monophyly of sections Bicuspidaria and Bartonia is also supported by a derived 6-bp deletion.
Gronovioideae
Our results strongly support the monophyly of Gronovioideae and contrast with suggestions that gronovioid genera are not a natural alliance (e.g., Davis and Thompson, 1967
; Poston and Nowicke, 1993
). Our results lead us to disagree with Weigend's (1997b
, p. 203) proposal "that it seems well justified to remove [Gronovioideae] from Loasaceae and create a family Gronoviaceae in its own right." Our results strongly support the placement of gronovioids within Loasaceae and following Weigend's proposal would make the family paraphyletic. Our results are consistent with the traditional assumption that the haplostemony and uniovulate gynoecium of Gronovioideae are derived features that evolved only once in Loasaceae. Thus, pollen (Poston and Nowicke, 1993
) and karyotype (Davis and Thompson, 1967
) character suites of Gronovioideae need to be seen as especially labile in the evolution of this small clade.
Indeed, pollen morphology is diverse in the Gronovioideae (Poston and Nowicke, 1993
). Petalonyx shares a plesiomorphic striate tectum with Mentzelia, Eucnide, and Schismocarpus. Core gronovioids have diverse tectum forms: Cevallia is spinulose, Gronovia is beaded-echinate, and Fuertesia is rugulose-punctate. In addition, the Gronovioideae have diversity in columellae length, definition of the foot layer, and form of operculum cover (Poston and Nowicke, 1993
).
Our results strongly support a sister-group relationship for Mentzelia and Gronovioideae (BS = 97). The alliance of Gronovioideae and Mentzelia calls attention to suggestions (Davis and Thompson, 1967
; Poston and Nowicke, 1993
) that Petalonyx, in particular, is similar to Mentzelia, although their shared pollen states appear to be symplesiomorphies. Floral ontogenetic data have also provided a possible morphological synapomorphy for Gronovioideae and Mentzelia (Moody and Hufford, 2000
). The corolla of both Mentzelia and Gronovioideae is initiated on the same transectional plane as the sepals, and it is not lifted by zonal growth to form a platelike region on which the stamens are also positioned. In contrast, the other investigated Loasaceae, including Eucnide, Schismocarpus, and members of Loasoideae, have a phase during early corolla development when zonal growth lifts the petals as a synorganized platelike zone above the insertion of the sepals (Leins and Winhard, 1973
; Hufford, 1988a, 1989b, 1990
). The stamens are subsequently initiated in positions that alternate with the petals on the uplifted plate of tissue. The zonal growth associated with this petal-stamen "plate" is lacking in investigated Gronovioideae (including Petalonyx linearis, Hufford, 1989a
) and Mentzelia (Brown and Kaul, 1981
; Hufford, 1990
). Because the petal-stamen "plate" is present in Eucnide and Schismocarpus, which branch from successive nodes at the base of our tree (Fig. 1), as well as in Loasoideae, we hypothesize that it is plesiomorphic for the family. Thus, the absence of the petal-stamen "plate" is a morphological synapomorphy for the Mentzelia-Gronovioideae clade.
Within the Gronovioideae, our results clearly support two main clades: (1) Petalonyx and (2) "core gronovioids," including Cevallia, Fuertesia, and Gronovia. The core gronovioids share morphological synapomorphies that include reduced petal size relative to sepals and largely bifacial filaments. The core gronovioids also share numerous deletions in ITS-1 (Fig. 4). Within the core gronovioids, Fuertesia and Gronovia share morphological synapomorphies that include a vine habit, apically bifid trichomes, and a nectary that is enlarged to form a bowl-shaped disc (Davis and Thompson, 1967
; Poston and Nowicke, 1993
; Moody and Hufford, 2000
). They also share two indels (Fig. 4).
Each gronovioid genus has distinctive autapomorphies. Cevallia has an elongate distal protrusion from its anthers and a dendritic trichome type. Gronovia has leaves with a 3–5 lobed margin and cordate lamina base, petal size reduced relative to sepals, trichomes with laterally positioned, two-hooked barbs on the stamens, and a nectarial disk covered by apically and laterally barbed trichomes. Fuertesia has lobed petals, trichomes with basal barbs, and trichome growth on both abaxial and adaxial surfaces of the anther.
Our results provide insights on interspecific relationships among the five species of Petalonyx. Petalonyx thurberi, P. nitidus, and P. parryi notably form a monophyletic group (Fig. 4). These three species are unusual in having flowers with fused petals and stamens positioned outside of the corolla (Davis and Thompson, 1967
). This clade of three species is the only group in the family outside of Eucnide to express sympetaly. The sympetaly, however, differs from that of Eucnide (Hufford, 1988a
) in arising from postgenital fusion of the petals after the stamens have grown outward from initiation positions in a whorl centripetal to the corolla (L. Hufford, unpublished data). The other two species of Petalonyx (P. linearis and P. crenatus) have free petals. Petalonyx crenatus is unique among Gronovioideae in having staminodes. It has two fertile stamens, whereas the other three stamens consist only of cylindrical filaments.
Conclusions
Our results strongly support a monophyletic Gronovioideae, despite contrary suggestions by Davis and Thompson (1967)
and Hempel (1995)
. Molecular data have been useful in elucidating the sister-clade relationship of Gronovioideae and Mentzelia, a result that had not previously been suggested. These results also provide the first support for the sister-group relationship of Loasoideae and the Gronovioideae-Mentzelia clade. Our results are consistent with the suggestions of Brown (1971)
and phylogenetic results of Hufford (1988b)
that indicated Mentzelioideae as circumscribed by Urban and Gilg (1900)
, Gilg (1925)
, and Ernst and Thompson (1963)
were paraphyletic. We suggest that Mentzelioideae be restricted only to include Mentzelia (i.e., that Eucnide and Schismocarpus be segregated from the subfamily). The results significantly support the placement of Eucnide as the sister of the rest of the family. Preliminary insights have been provided for relationships among genera of Loasoideae and among species of Mentzelia, although both of these topics will require further phylogenetic study.
|
|
FOOTNOTES |
|---|
3 Current address: Department of Ecology and Evolutionary Biology, University of Connecticut, Storrs, Connecticut 06269 USA.
4 Current address: School of Biological Sciences, Washington State University, Pullman, Washington 99164-4236 USA.
|
LITERATURE CITED |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
Baldwin, B. G., M. J. Sanderson, J. M. Porter, M. F. Wojciechowski, C. S. Campbell, and M. J. Donoghue. 1995 The ITS region of nuclear ribosomal DNA: a valuable source of evidence on angiosperm phylogeny. Annals of the Missouri Botanical Garden 82: 237–277
Baum, D. A., K. J. Sytsma, and P. C. Hoch. 1994 A phylogenetic analysis of Epilobium (Onagraceae) based on nuclear ribosomal DNA sequences. Systematic Botany 19: 363–388[CrossRef][ISI]
Bremer, K. 1988 The limits of amino acid sequence data in angiosperm phylogenetic reconstruction. Evolution 42: 795–803[CrossRef][ISI]
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
———, and R. B. Kaul. 1981 Floral structure and mechanism in Loasaceae. American Journal of Botany 68: 361–372[CrossRef][ISI]
Cronquist, A. 1981 An integrated system of classification of flowering plants. Columbia University Press, New York, New York, USA
Cunningham, C. W. 1997 Can three incongruence tests predict when data should be combined? Molecular Biology and Evolution 14: 733–740[Abstract]
Damtoft, S., S. R. Jensen, and B. J. Nielsen. 1993 Schismoside, an iridoid glycoside from Schismocarpus matudai. Phytochemistry 32: 885–889[CrossRef][ISI]
Dandy, J. E. 1926 Notes on Kissenia and the geographical distribution of the Loasaceae. Kew Bulletin 1926: 174–180
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: 103–226[CrossRef]
Davis, W. S., and H. J. Thompson. 1967 A revision of Petalonyx (Loasaceae) with a consideration of affinities in subfamily Gronovioideae. Madroño 19: 1–18
Donoghue, M. J., R. G. Olmstead, J. F. Smith, and J. D. Palmer. 1992 Phylogenetic relationships of Dipsacales based on rbcL sequences. Annals of the Missouri Botanical Garden 79: 333–345
Downie, S. R., and J. D. Palmer. 1992 Restriction site mapping of the chloroplast DNA inverted repeat: a molecular phylogeny of the Asteridae. Annals of the Missouri Botanical Garden 79: 266–283[CrossRef][ISI]
Doyle, J. J., and J. L. Doyle. 1987 A rapid DNA isolation method for small quantities of fresh leaf tissue. Phytochemistry Bulletin 19: 11–15
Eriksson, T., and M. J. Donoghue. 1997 Phylogenetic relationships of Sambucus and Adoxa (Adoxoideae, Adoxaceae) based on nuclear ribosomal ITS sequences and preliminary morphological data. Systematic Botany 22: 555–573[CrossRef][ISI]
Ernst, W. R., and H. J. Thompson. 1963 The Loasaceae in the southeastern United States. Journal of the Arnold Arboreteum 44: 138–142
Farris, J. S., M. Kallersjo, A. G. Kluge, and C. Bult. 1995 Testing significance of incongruence. Cladistics 10: 315–319[CrossRef][ISI]
Felsenstein, J. 1978 Cases in which parsimony and compatibility methods will be positively misleading. Systematic Zoology 27: 401–410[CrossRef][ISI]
———. 1985 Confidence limits on phylogenies: an approach using the bootstrap. Evolution 39: 783–791[CrossRef][ISI]
Florence, J. 1985 Sertum polynesium I. Plakothira Florence (Loasaceae), Genre nouveau des iles Marquises. Bulletin du Muséum National d'Histoire Naturelle. Section G, Adansonia, Botanique, Phytochimie 3: 239–245
Gilbert, D. G. 1992 SeqApp: a biosequence editor and analysis application. Privately published by the author, Bloomington, Indiana, USA
Gilg, E. 1925 Loasaceae. In A. Engler and K. Prantl [eds.], Die Naturlichen Pflanzenfamilien, 2nd ed. 21: 522–543
Grau, J. 1997 Huidobria, eine isolierte Gattung der Loasaceae aus Chile. Sendtnera 4: 77–93
Hempel, A. L. 1995 Molecular systematics of the Loasaceae. Ph.D. dissertation, University of Texas, Austin, Texas, USA
———, P. A. Reeves, R. G. Olmstead, and 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][ISI]
Hill, R. J. 1976 Taxonomic and phylogenetic significance of seed coat microsculpturing in Mentzelia (Loasaceae) in Wyoming and adjacent western states. Brittonia 28: 86–112[CrossRef][ISI]
Huelsenbeck, J. P. 1995 Performance of phylogenetic methods in simulation. Systematic Biology 44: 17–48
Hufford, L. 1988a Roles of early ontogenetic modifications in the evolution of floral morphology of Eucnide (Loasaceae). Botanische Jahrbücher für Systematik 109: 289–333
———. 1988b Systematics of Loasaceae: a consideration of familial circumscription and a cladistic analysis of generic relationships. American Journal of Botany 75(pt. 2): 181
———. 1989a Structure of the inflorescence and flower of Petalonyx linearis (Loasaceae). Plant Systematics and Evolution 163: 211–226[CrossRef][ISI]
———. 1989b The structure and potential loasaceous affinities of Schismocarpus. Nordic Journal of Botany 9: 217–227
———. 1990 Androecial development and the problem of monophyly of Loasaceae. Canadian Journal of Botany 68: 402–419
———. 1992 Rosidae and their relationship to other nonmagnoliid dicotyledons: a phylogenetic analysis using morphological and chemical data. Annals of the Missouri Botanical Garden 79: 219–248
Jensen, S. R., C. B. Mikkelsen, and B. J. Nielsen. 1981 Iridoid mono- and di-glycosides in Mentzelia. Phytochemistry 20: 71–83[CrossRef][ISI]
Johnson, L. A., and D. E. Soltis. 1994 MatK DNA sequences and phylogenetic reconstruction Saxifragaceae s. str. Systematic Botany 19: 143–156
———, and ———. 1995 Phylogenetic inference in Saxifragaceae sensu stricto and Gilia (Polemoniaceae) using matK sequences. Annals of the Missouri Botanical Garden 82: 149–175[CrossRef][ISI]
Kishino, H., and M. Hasegawa. 1989 Evaluation of the maximum likelihood estimate of the evolutionary tree topologies from DNA sequence data, and the branching order of Hominoidea. Journal of Molecular Evolution 29: 170–179[CrossRef][ISI][Medline]
Leins, P., and W. Winhard. 1973 Entwicklungsgeschichtliche studien an Loasaceen-Bluten. Oesterreichische Botanische Zeitschrift 122: 145–165[CrossRef]
Maddison, W. P., and D. R. Maddison. 1992 MacClade: analysis of phylogeny and character evolution, version 3.0. Sinauer, Sunderland, Massachusetts, USA
Moody, M. L., and L. Hufford. 2000 Floral ontogeny and morphology in Cevallia, Fuertesia, and Gronovia (Loasaceae subfamily Gronovioideae). International Journal of Plant Sciences 161: 869–883
Payer, J. B. 1857 Traite d'organogénie comparée de la fleur. (1966 reprint) Stechert-Hafner, New York, New York, USA
Poston, M. 1979 A systematic examination of the Loasoideae (Loasaceae): a realignment of the genera. Ph.D. dissertation, University of California, Los Angeles, USA
———, and J. W. Nowicke, 1990 A reevaluation of Klaprothia and Sclerothrix (Loasaceae: Klaprothieae). Systematic Botany 15: 671–678[CrossRef][ISI]
———, and ———. 1993 Pollen morphology, trichome types and relationships of the Gronovioideae (Loasaceae). American Journal of Botany 80: 689–704[CrossRef][ISI]
——–, and H. J. Thompson. 1977 Cytotaxonomic observations in Loasaceae subfamily Loasoideae. Systematic Botany 2: 28–42[CrossRef][ISI]
Soltis, D. E., P. S. Soltis, T. G. Collier, and M. L. Edgerton. 1991 Chloroplast DNA variation within and among genera of the Heuchera group (Saxifragaceae): evidence for chloroplast transfer and paraphyly. American Journal of Botany 78: 1091–1112[CrossRef][ISI]
———, ———, et al. 1997 Angiosperm phylogeny inferred from 18S ribosomal DNA sequences. Annals of the Missouri Botanical Garden 84: 1–49
———, Q.-Y. Xiang, and L. Hufford. 1995 Relationships and evolution of Hydrangeaceae based on rbcL sequence data. American Journal of Botany 82: 504–514[CrossRef][ISI]
Steele, K. P., and R. Vilgalys. 1994 Phylogenetic analyses of Polemoniaceae using nucleotide sequences of the plastid gene matK. Systematic Botany 19: 126–142
Sullivan, J. J., A. Markert, and C. W. Kilpatrick. 1997 Phylogeography and molecular systematics of the Peromyscus aztecus species group (Rodentia: Muridae) inferred using parsimony and likelihood. Systematic Biology 46: 426–440[CrossRef][ISI][Medline]
Swofford, D. L. 1998 Phylogenetic analysis using parsimony, version 4.0b1. Sinauer, Sunderland, Massachusetts, U.S.A
———, G. J. Olsen, P. J. Waddell, and D. M. Hillis. 1996 Phylogenetic inference. In D. M. Hillis, C. Moritz, and B. K. Mable [eds.], Molecular systematics, 2nd ed., 407–514. Sinauer, Sunderland, Massachusetts, USA
Thompson, H. J., and W. R. Ernst. 1967 Floral biology and systematics of Eucnide (Loasaceae). Journal of the Arnold Arboretum 48: 56–88[ISI]
Urban, I., and E. Gilg. 1900 Monographia Loasacearum. Nova Acta Academiae Caesareae Leopoldino-Carolinae Germanicae Naturae Curiosorum 76: 1370
Waddell, P. 1995 Statistical methods of phylogenetic analysis, including Hadamard conjugations, LogDet transforms, and maximum likelihood. Ph.D. dissertation, Massey University, Palmerston North, New Zealand
Weigend, M. 1997a Names and types in Cajophora K. Presl. s. str. (Loasaceae). Sendtnera 4: 221–242
———. 1997b Nasa and the conquest of South America—systematic rearrangements in Loasaceae Juss. Ph.D. dissertation, Ludwig-Maximilians-Universität, Munich, Germany
———. 1997c Some aspects of the biogeography, morphology and systematics of Loasoideae in northern South America. In R. Valencia and H. Balslev [eds.], Estudios Sobre Diversidad y Ecologia de Plantas, 37–50. Pontificia Universidad Catolica del Ecuador, Quito, Ecuador
White, T. J., T. Bruns, S. Lee, and J. Taylor. 1990 Amplification and direct sequencing of fungal ribosomal RNA genes for phylogenetics. In M. A. Innis, D. H. Gelfand, J. J. Sninsky, and T. J. White [eds.], PCR protocols: a guide to methods and applications, 315–322. Academic Press, San Diego, California, USA
} = observed difference in ln likelihood. {
} = standard deviation of distribution of single-site likelihood differences