HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

What's this?

This Article Abstract Full Text (PDF) Supplementary Data Submit a response Alert me when this article is cited Alert me when eLetters are posted Alert me if a correction is posted Services Email this article to a friend Similar articles in this journal Similar articles in Web of Science Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via Google Scholar Google Scholar Articles by Moody, M. L. Articles by Les, D. H. Search for Related Content PubMed Articles by Moody, M. L. Articles by Les, D. H. Agricola Articles by Moody, M. L. Articles by Les, D. H. Social Bookmarking                            What's this?

Phylogenetic systematics and character evolution in the angiosperm family Haloragaceae¹

University of Connecticut, Department of Ecology and Evolutionary Biology, Storrs, Connecticut 06269 USA

Received for publication December 22, 2006. Accepted for publication October 15, 2007.

ABSTRACT

The poorly known Haloragaceae R. Br. (Saxifragales) are highly diverse in habit (small trees to submerged aquatics) and labile in floral merosity (2–4), both uncommon among the core eudicots. This family has a cosmopolitan distribution, but taxonomic diversity is concentrated in Australia. An explicit phylogenetic approach has not previously been utilized to examine relationships or character evolution in this family. We used molecular evidence from nrDNA ITS and cpDNA trnK and matK regions under both Bayesian and parsimony analyses to address phylogenetic relationships. Combined molecular analyses defined a monophyletic Haloragaceae with the woody genera (Haloragodendron, Glischrocaryon) sister to the rest. Relationships among many genera were well resolved, with genera as currently delimited generally well supported, although there were notable exceptions; a new genus (Trihaloragis) is recognized, and the aquatic genus Meionectes is again distinct from Haloragis. Three new species combinations are also recognized. There are multiple (two or three) origins of the submerged aquatic habit in the family and potentially an intermediate reversal to the terrestrial habit, neither previously demonstrated in a core eudicot family using an explicit phylogenetic hypothesis. Ancestral character analyses suggest two origins of trimerous flowers and multiple reductions to dimerous flowers throughout Haloragaceae.

Key Words: aquatic • Bayes • floral merosity • Haloragaceae • ITS • matK • phylogenetics

Haloragaceae are a cosmopolitan family currently with eight genera and about 120 species (Table 1). They are extremely diverse in habit, ranging from small trees to submerged aquatics. Four genera (Glischrocaryon, Gonocarpus, Haloragis, Haloragodendron) are primarily terrestrial, whereas four (Laurembergia, Meziella, Myriophyllum, Proserpinaca) are aquatic/semiaquatic. The habitats of these taxa range from arid deserts to freshwater lakes exceeding 10 m in depth. Haloragaceae also are highly labile in floral merosity (2–4), uncommon among core eudicots. The center of diversity for the family is Australia, where four endemic genera and 70% of the species occur.

Prior to the advent of molecular-based phylogenetic studies, many authors suggested that Haloragaceae had a close relationship to Onagraceae (Schindler, 1905 ; Hutchinson, 1959 ; Melchior, 1964 ), Cornaceae (Thorne, 1968 ; Orchard, 1975 ), or the aquatic Podostemaceae (Cronquist, 1968 ; Takhtajan, 1969 ). These suggestions were based on features from embryology, pollen morphology, and floral vasculature. However, inclusive studies of angiosperms using molecular phylogenetic approaches have supported the placement of Haloragaceae within Saxifragales (Morgan and Soltis, 1993 ; Soltis et al., 1997 ). Fishbein et al. (2001) sampled two Haloragaceae taxa (Myriophyllum sp. and Haloragis sp.) in a phylogenetic treatment of the Saxifragales using sequence data from five genes and found Haloragaceae to form a clade with three genera (Aphanopetalum, Penthorum, Tetracarpaea) not considered previously as being closely allied to Haloragaceae. In turn, Crassulaceae resolved as a well-supported sister group to this clade (Fishbein et al., 2001 ). Consequently, inclusion of both Tetracarpaea and Penthorum in Haloragaceae has been proposed as an option (APG II 2003 ).

Intergeneric relationships
Disparate geographic distributions and highly divergent morphology (Orchard, 1975 ; Schindler, 1905 ) among the four aquatic/semiaquatic genera (Table 1) have made it difficult to hypothesize their evolutionary relationships within Haloragaceae. The aquatic genera have different distributions; neither Proserpinaca (New World) nor Laurembergia (Africa, Asia, and South America) is found in Australia. Meziella is an Australian endemic, while Myriophyllum has a global distribution with its center of diversity (60% of species) in Australia. In contrast, the four terrestrial genera (Table 1) occur almost exclusively in Australia and New Zealand. Three Gonocarpus species extend into Southeast Asia, and a group of four Haloragis species extends east from Australia into islands of the New Caledonia complex (H. prostrata J.R.Forst. & G.Forst.), Rapa Island (H. stokesii R.Br.), and the Juan Fernandez Islands off the west coast of Chile (H. masatierrana Skottsb., H. masafuerrana Skottsb.).

Though many morphological features have been examined in Haloragaceae, intergeneric relationships remain elusive. The circumscription of the aquatic genera (excluding Meionectes) has not been questioned and is supported by morphology (Orchard, 1975 ). However, relationships among these genera are unclear. Unisexual flowers (except in Proserpinaca) and pinnately dissected leaves (except in Laurembergia) are common among the aquatic/semiaquatic genera, but these characteristics also are convergent among many aquatic angiosperms (Sculthorpe, 1967 ; Cook, 1996 ). Other morphological characters less likely to be adaptive to the aquatic habit indicate that some aquatic taxa may not be as closely related to each other as they are to Haloragis or Gonocarpus. For example, pollen characters (Praglowski, 1970 ) associate Haloragis, Gonocarpus, Laurembergia, and Myriophyllum, whereas embryological features (Bawa 1969a , b ) link only Haloragis, Myriophyllum, and Laurembergia. Furthermore, several reproductive and vegetative features (Orchard, 1975 ) are shared more closely among Haloragis, Gonocarpus, and Laurembergia but not Myriophyllum.

Aquatic taxa
Myriophyllum (60+ spp.) is among the most speciose genera of submerged aquatic "dicots" (Cook, 1996 ) and is unique within Haloragaceae in having fruits that separate into individual nutlets at maturity and carpellate flowers that lack petals and often sepals. Meziella is a genus thought to be extinct in Australia until its recent rediscovery by Orchard and Keighery (1993) , who suggested a close affiliation of Meziella with Myriophyllum. Laurembergia may best be characterized as a helophyte (i.e., generally terrestrial but tolerant of prolonged inundation; Cook, 1996 ). Orchard (1975) considered the placement of Proserpinaca within Haloragaceae particularly problematic. He suggested with reservation a "transitional placement" of Proserpinaca between Haloragis and Myriophyllum. The deeply dissected submerged leaves of Proserpinaca are similar to those of Myriophyllum but also occur in the aquatic/semiaquatic members of Haloragis (H. brownii (Hook.f.) Schindl., H. heterophylla Brongn., and H. tenuifolia Benth.). The trimerous, perfect flowers of Proserpinaca rarely occur elsewhere in the family and are found otherwise only in Gonocarpus hexandrus (F.Muell.) Orchard, Haloragis gossei F.Muell., H. tenuifolia, H. trigonocarpa F.Muell., and occasionally in H. digyna Labill. The trimerous flower has been emphasized as an important feature in the generic circumscription of Proserpinaca.

Terrestrial taxa
Prior to Orchard's (1975) revision of Haloragaceae, all of the terrestrial taxa except Glischrocaryon (= Loudonia) were usually included in Haloragis (Brown, 1814 ; Schindler, 1905 ). Orchard (1975) split Gonocarpus (the most speciose terrestrial genus; 38 spp.) from Haloragis based primarily on discordant fruit characters. He suggested (p. 274), "the relationship between [Haloragis and Gonocarpus] is probably not close" and instead proposed a closer relationship between Gonocarpus and the semiaquatic Laurembergia. He also split Haloragodendron from Haloragis based on its woody habit, large flowers, and winged fruits, all characters that were shared with Glischrocaryon spp. Orchard (1975) based the distinction between Glischrocaryon and Haloragodendron primarily on vegetative characters. Conspicuously, Glischrocaryon has adaptations common to arid environments, such as green stems and leaves reduced in size and number along the stem, with most located basally, characteristics not found in Haloragodendron.

Character evolution
Aquatic habit
Multiple origins of the aquatic habit are now generally regarded as having occurred across the angiosperms (Cook, 1996 ). However, multiple origins of submerged aquatic genera in a single family as defined by APG II (2003) are rare among the core eudicots. Character states often associated with aquatic taxa include those related to anemophily (i.e., reduced flowers, dioecy, monoecy, etc.) as well as highly divided or ribbonlike leaves (Sculthorpe, 1967 ). Submerged leaves of aquatic angiosperms undergo extreme structural modification. In contrast to their terrestrial ancestors, the leaves of submerged aquatics characteristically have a reduced/absent cuticle, are astomatal, and often are highly divided and/or ribbonlike to compensate for low light levels, limited CO₂ uptake, and water resistance (Cronk and Fennesy, 2001 ; Sculthorpe, 1967 ). All of these characteristics can be found among the aquatic Haloragaceae.

The aquatic habit manifests itself in various forms within Haloragaceae. Myriophyllum, Meziella, Proserpinaca, Haloragis brownii, and H. tenuifolium all spend part of their life cycle underwater and are considered primarily submerged taxa. However, as in most aquatic plants, their reproductive structures are produced primarily on emergent stems that have variable leaf forms resembling those of their terrestrial relatives. The predominant vegetative form of aquatic Haloragaceae is a submerged stem with lacunal passages and pinnately laciniate leaves, although several Myriophyllum species have only linear, minute leaves. All the submerged aquatic genera have species that display some plasticity in form and/or habit, with some taxa adapted to survival on mudflats through thickened leaves that are reduced in size and segmentation, traits also found in the aquatic Haloragis (H. brownii, H. tenuifolia). In contrast, the helophytic Laurembergia does not include any species with a distinctive submerged form. Given this diversity, Haloragaceae are an ideal group for investigating the evolutionary transition from terrestrial to submerged aquatic habitats.

Floral merosity
Floral merosity often deviates from the basic tetramerous plan that occurs throughout Haloragaceae. Perfect, trimerous flowers are uncommon among the core eudicots (Soltis et al., 2003 ), yet occur in four different Haloragaceae genera (Glischrocaryon, Gonocarpus, Haloragis, Proserpinaca). Moreover, dimerous flowers, another unusual condition among core eudicots, are present in Glischrocaryon behrii (Schindl.) Orchard, Haloragis brownii, Haloragis digyna (usually), and some Myriophyllum spp. Reductions in locule or carpel number have also occurred in some species. Orchard (1975) considered these differences to represent derived reductions within genera, an opinion also shared by Schindler (1905) . Assuming multiple reductions, neither author suggested close relationships between the pinnate-leaved, aquatic, trimerous- and dimerous-flowered species. This lability of floral merosity makes Haloragaceae an ideal group for investigating these evolutionary trends.

Data selection and analyses
Given the extreme diversity and contrasting systematic hypotheses that have been suggested for this family, a phylogenetic approach using molecular markers should help clarify relationships within Haloragaceae as it has for other families (Judd et al., 2002 ). It is important to use multiple data sets to address phylogenetic questions at different taxonomic levels (Swofford et al., 1996 ) and to account for hybridization and/or lineage sorting through incongruent phylogenetic hypotheses based on plastid vs. nuclear genes (Rieseberg and Wendel, 1993 ; Avise, 1994 ; Wendel and Doyle, 1998 ; Sang and Zhong, 2000 ). Hybridization can be restricted taxonomically, being more prominent in some groups than in others (Ellstrand et al., 1996 ; Rieseberg, 1997 ). Because hybridization is suspected or known in Haloragaceae (Orchard, 1975 ; Moody and Les, 2002 ), it is important to consider hybridization when interpreting phylogenetic hypotheses in this family. Although ancient hybridizations are difficult to determine using current phylogenetic methodologies (Linder and Rieseberg, 2004 ), more recent hybridization events can be assessed by comparing phylogenetic hypotheses derived from nuclear and plastid genomes and observing marked incongruence between the resulting tree topologies (Rieseberg, 1991 ; Wendel et al., 1991 , 1995 ). We have chosen to use the nrDNA ITS and the cpDNA trnK introns + matK coding region for their combined effectiveness at elucidating phylogenetic relationships at the generic and subgeneric level (e.g., Moody et al., 2001 ; Les et al., 2002a , b ).

Conflict among phylogenetic hypotheses can also arise because of unequal rates of evolution among data sets. This problem has been addressed by recent advances in Bayesian phylogenetic analyses (Ronquist and Huelsenbeck, 2003 ; Nylander et al., 2004 ), and we use multiple maximum likelihood (ML) models for separate, defined partitions in our combined analyses. Maximum parsimony was also used as an alternative source of reference to compare to Bayesian analysis results. Thus, both parsimony and Bayesian analyses were implemented for comparative purposes.

The major goals of this research were to (1) elucidate phylogenetic relationships among Haloragaceae genera and within some (2) evaluate generic limits as currently proposed and redefine them in the context of the phylogeny where necessary, and (3) examine the evolution of the aquatic habit and floral merosity among Haloragaceae genera using explicit phylogenetic hypotheses.

MATERIALS AND METHODS

Taxon sampling
Ninety-three taxa representing all genera of Haloragaceae and the hypothesized outgroups Aphanopetalaceae, Penthoraceae, and Tetracarpaeaceae (Fishbein et al., 2001 ) were sampled. All known species of Glischrocaryon, Haloragodendron, and Meziella were sampled. Other generic coverage included 23 of 36 (64%) Gonocarpus species and 16 of 26 (62%) Haloragis species, representing all major groupings recognized by Orchard (1975) and all the subsections of Schindler (1905) . Also one of four (25%) Laurembergia species, two of three (67%) Proserpinaca species, and 31 of 65 (48%) of Myriophyllum species (including the major alliances of Orchard [1986] and most subsections of Schindler [1905] ) were sampled. Most taxa were collected in the field and preserved with cetyl trimethylammonium bromide (CTAB) (Rogstad, 1992 ). Several taxa also were sampled from herbarium specimens, and multiple accessions of each taxon were sampled when possible (Appendix 1).

DNA extraction, PCR, and sequencing
Total genomic DNAs were extracted from fresh, CTAB-preserved, and herbarium specimen leaf material using a modified CTAB miniprep procedure (Doyle and Doyle, 1987 ). PCR was used to amplify the ITS-1, ITS-2, and 5.8S region of nuclear ribosomal DNA using the ITS4 and ITS5 primers or, in the case of several herbarium specimens, using ITS3 and ITS4 to amplify the ITS-2 and partial 5.8S region and ITS2 and ITS5 to amplify the ITS-1 and partial 5.8S region (White et al., 1990 ). The cpDNA trnK introns and matK coding-region were amplified using the primers trnK-3914F and trnK-2R (Johnson and Soltis, 1994 ). Several additional primers were used to amplify trnK and matK from DNA of herbarium leaf material including matK68F, matK1872R (Johnson and Soltis, 1994 ), matK 900F (Moody and Les, 2002 ), and newly developed trnK5R (5'-TCTTGGGTTATCAAATGATA) and matK70R (5'-GTTGTGTTGACGAAAT). PCR protocols and conditions were the same as described in Moody et al. (2001) . Cycle-sequencing of ITS used combinations of the ITS2, ITS3, ITS4, and ITS5 primers (White et al., 1990 ). The ITS region was cloned for some taxa using the TOPO TA cloning kit (Invitrogen, Carlsbadm California, USA) to identify and exclude fungal contaminants. The trnK introns and matK region were sequenced using trnK-3914F, matK68F, matK1872R, matK900F, trnK360F, trnK2R, matK70R, trnKR, and a newly developed Haloragaceae specific primer trnK3F (5'-CGTCGATTGTGCGTA). Sequences were obtained using Big Dye terminator technology on an ABI 3100 automated sequencer (Applied Biosystems, Foster City, California, USA).

Phylogenetic analyses
Sequences were edited using the program Sequencher 4.1.2 (Gene Code Corp., Ann Arbor, Michigan, USA) and manually aligned using the program MacClade 4.01 (Maddison and Maddison, 2001 ). Indel alignments were carefully assessed as in Graham et al. (2000) . Parsimony analyses were performed with the program PAUP* version 4.0b8 (Swofford, 2000 ) using heuristic searches with random taxon-addition sequences, max trees undefined, and tree bisection-reconnection with unordered, equally weighted characters with 100 analysis replicates. Indels were treated as missing data, and indel regions lacking variable data were removed. ITS data were easily aligned in conserved regions, but ITS regions with highly ambiguous alignment (84 bp from nine regions) were removed from the data set when necessary (Appendix S2; see Supplemental Data accompanying online version of this article). Standard measures of homoplasy, such as consistency index (CI), retention index (RI), and rescaled consistency index (RC) excluding uninformative characters, and the level of internal support (bootstrap values) were calculated using PAUP* 4.0b8. Bootstrap analyses were conducted using 200 replicate heuristic searches as indicated earlier with max trees set at 10 000.

Bayesian analyses were performed using the program MrBayes version 3.0B4 (Huelsenbeck and Ronquist, 2001 ). The nrDNA ITS, trnK, and matK data sets were examined initially to determine the best-fit DNA substitution model using the program MrModeltest version 1.1b (Nylander, 2003 ). The matK data set was analyzed with each codon position representing a separate data set and each having an individual best-fit model. Initially, two separate analyses were performed: (1) nrDNA ITS and (2) cpDNA trnK and matK. Subsequently, the nrDNA and cpDNA data sets were combined. Bayesian analyses were performed twice for 3.0 x 10⁶ generations using the best-fit model for each character partition. Trees were sampled every 1000 generations. Stability of the process was assessed by plotting model –ln likelihood scores against generations to determine equilibrium. Trees sampled before reaching equilibrium were discarded as burn in, with the remaining trees used to generate a 50% majority rule consensus tree where the percentage of the nodes recovered represented each node's posterior probability. Trees recovered from each individual run then were compared for topology and posterior probability to further determine consensus among analyses of each of the three data sets analyzed. Nodal support was determined using Bayesian posterior probabilities (PP) > 0.95 as the threshold criterion for strong support.

Incongruence
Congruence of the ITS and cpDNA data sets was tested using the incongruence length difference (ILD) test as implemented in PAUP*. The nrDNA ITS and cpDNA data sets were analyzed using 1000 homogeneity replicates with heuristic searches as described earlier under parsimony analysis. Incongruence also was determined visually for trees with incongruent topologies between different data sets (nrDNA vs. cpDNA). If incongruence was detected, the conflicting branches were evaluated individually for relative support given parsimony bootstrap and Bayesian posterior probabilities. Eventually, the data were combined regardless of the outcome of the ILD test (see Discussion).

Analysis of character evolution
The habit and merosity data were compiled from several literature sources (Schindler, 1905 ; Orchard, 1975 , 1979 , 1981 , 1985 , 1990 ). Floral merosity was scored as 2, 3, 4, or 5 and habit as terrestrial, aquatic, or semiaquatic. The five most likely trees recovered from the combined data Bayesian analysis were evaluated for topology most closely matching that of the majority rule consensus tree. Character states were then optimized on the best-fit phylogenetic hypothesis so that relative branch lengths, as determined by Bayesian analysis, could be incorporated for ML analysis of ancestral states. Ancestral state optimization was performed using both parsimony and likelihood methodologies implemented in the program Mesquite version 1.04 (Maddison and Maddison, 2003 ). We choose to use a one-rate model following the observations of Mooers and Schluter (1999) . The maximum likelihood model used for the analysis of the morphological data was Mk1 (Lewis, 2001 ).

RESULTS

The nrDNA ITS data set consisted of 681 bp of aligned sequence, with 84 sites removed because of ambiguous alignment. There were 274 variable characters, with 229 being potentially parsimony informative (including outgroups). The cpDNA trnK 5' intron comprised 665 bp of aligned sequence data of which 290 sites were variable and 173 potentially parsimony informative; the trnK 5' intron was not recovered for Tetracarpaea tasmanica Hook.f. The matK data set included 1506 bp of aligned sequence data of which 694 sites were variable and 425 parsimony informative. The trnK 3' data set had 127 bp of aligned sequence data of which 72 sites were variable and 53 potentially parsimony informative; the trnK 3' intron was not recovered for Haloragis acutangula F.Muell., H. serra Brongn., Myriophyllum sp., or Tetracarpaea tasmanica.

Parsimony results
Parsimony analysis of ITS recovered five islands with a total of 12 182 most parsimonious trees with a score of 1746 steps (CI = 0.37, RI = 0.76, RC = 0.28). Parsimony analysis of cpDNA data (trnK + matK) resulted in 87 255 equally parsimonious trees of 1860 steps (CI = 0.61, RI = 0.86, RC = 0.52). Parsimony analysis of the combined data resulted in 28 812 equally parsimonious trees of 3706 steps (CI = 0.48, RI = 0.80, RC = 0.36). Parsimony bootstrap scores are displayed on the results of Bayesian analyses of the respective data sets (Figs. 1–3).

View larger version (22K): [in this window] [in a new window]   Fig. 1. Phylogenetic relationships in Haloragaceae as indicated by a majority rule consensus tree of 28 000 trees (after discarding burn-in) from Bayesian analysis of ITS sequence data analyzed under a GTR+I+ model. Numbers above branches refer to posterior probabilities, and numbers below branches are bootstrap support from parsimony analyses. Branches represented by dashed lines are not found in the strict consensus trees from the parsimony analysis of the same data set. Numbers in bold above branches have the same values in both parsimony and Bayesian analyses. When numbers are lacking below solid branches, parsimony had bootstrap values <50 supporting that node.

View larger version (24K): [in this window] [in a new window]   Fig. 3. Phylogenetic relationships in Haloragaceae as indicated by a majority rule consensus tree of 28 000 trees (after discarding burn-in) from Bayesian analysis based on ITS and cpDNA combined data analyzed using individual DNA substitution models appropriate to each data partition (Table 2). Numbers above branches refer to posterior probabilities, and numbers below branches are bootstrap support from parsimony analyses. Branches represented by dashed lines are not found in the strict consensus trees from the parsimony analysis of the same data set. Numbers in bold above branches have the same values in both parsimony and Bayesian analyses. When numbers are lacking below solid branches, parsimony had bootstrap values <50 supporting that node.

View larger version (19K): [in this window] [in a new window]   Fig. 4. Phylogenetic relationships in Haloragaceae as indicated by phylograms of best likelihood trees selected from 28 000 trees recovered from Bayesian analyses of (A) combined ITS and cpDNA data (second most likely tree also used for ancestral state analysis); (B) cpDNA data; (C) ITS data. Branch lengths are proportional.

View larger version (22K): [in this window] [in a new window]   Fig. 2. Phylogenetic relationships in Haloragaceae as indicated by a majority rule consensus tree of 28 000 trees (after discarding burn-in) from Bayesian analysis based on cpDNA trnK and matK data analyzed using individual DNA substitution models appropriate to each data partition (Table 2). Numbers above branches refer to posterior probabilities, and numbers below branches are bootstrap support from parsimony analyses. Branches represented by dashed lines are not found in the strict consensus trees from the parsimony analysis of the same data set. Numbers in bold above branches have the same values in both parsimony and Bayesian analyses. When numbers are lacking below solid branches, parsimony had bootstrap values <50 supporting that node.

Haloragis (here forward excluding H. brownii-H. tenuifolia) and Gonocarpus (here forward excluding G. hexandrus) formed a well-supported clade in all analyses and were sister to a Laurembergia-Meziella-Myriophyllum clade in the cpDNA and combined analyses. Analyses of ITS resolved these same relationships but weakly supported G. hexandrus (Fig. 1) sister to Laurembergia-Meziella-Myriophyllum. Haloragis was monophyletic in all analyses (Figs. 1–3). Gonocarpus was monophyletic in cpDNA and combined analyses. Analyses of ITS identified a paraphyletic Gonocarpus. Gonocarpus eremophilus Orchard, G. micranthus Thunb., and G. montanus (Hook.f.) Orchard formed a poorly supported basal grade to Haloragis. A clade including Laurembergia, Meziella, and Myriophyllum was present in all analyses, with Laurembergia sister to Myriophyllum-Meziella. ITS resolved a weakly supported sister group relationship of Meziella to Myriophyllum in the Bayesian analysis (PP = 0.89), a hypothesis not resolved by parsimony (Fig. 1). Analyses of combined data and cpDNA supported Meziella as part of a monophyletic Myriophyllum. Within Myriophyllum, two major clades were supported in the combined data analyses: (1) mostly Australian and North American endemics including Meziella and (2) South American taxa and a geographically disparate alliance of Myriophyllum (Fig. 3).

Within Haloragis, H. trigonocarpa was sister to the rest of the genus as supported by cpDNA and combined data, whereas ITS results were equivocal in the Bayesian analysis. Short branches and weak support were resolved for other relationships within this genus, although combined analyses did provide phylogenetic support for some groupings (Fig. 3). In all analyses, H. erecta (Banks ex. Murray) Oken and H. masatierrana resolved as a well-supported clade nested within Haloragis. Several clades within Gonocarpus were recovered consistently using different data sets. A sister group relationship between G. montanus and G. micranthus was well supported, and they were sister to the rest of Gonocarpus (less G. eremophilus) using the cpDNA and combined data. Bayesian analysis of ITS depicted these taxa basal to Gonocarpus-Haloragis with weak support (Fig. 1). Within Gonocarpus, branching after G. micranthus-G. montanus, was the "benthamii clade" (see Fig. 3) containing five taxa endemic to Western Australia. Branching next was the "acanthocarpus clade" (G. acanthocarpus (Brongn.) Orchard, G. ephemerus Orchard, G. leptothecus (F.Muell.) Orchard; Fig. 3) using the cpDNA and combined data. Gonocarpus acanthocarpus and G. leptothecus were well supported as sister taxa in all data sets, but were placed basal to the "benthamii clade" with weak support using the Bayesian analyses of ITS (Fig. 1). Gonocarpus ephemerus was part of the "acanthocarpus clade" using the cpDNA and combined data, but this relationship was not supported using ITS alone. The "tetragynus clade" (see Fig. 3) was resolved in all analyss, although relationships within this clade were inconsistent.

Character analyses
The parsimony reconstruction of ancestral character states was equivocal for aquatic habit (Fig. 5). There were either three independent origins of the aquatic habit at nodes 2 (Haloragis brownii-H. tenuifolia [= Meionectes]), 3 (Proserpinaca), and 6 (Laurembergia-Meziella-Myriophyllum) or two independent origins of the aquatic habit at nodes 1 (Haloragis brownii-H. tenuifolia) and 6 (Laurembergia-Myriophyllum) with an intermediate reversal back to the terrestrial habit beginning with node 3 (Gonocarpus hexandrus [= Trihaloragis]). The first hypothesis was favored by the ML reconstruction of ancestral characters, but not significantly (Fig. 5).

View larger version (31K): [in this window] [in a new window]   Fig. 5. Maximum likelihood reconstructions of ancestral habit character states in Haloragaceae. Semiaquatic (S), aquatic (A), and terrestrial (T) habits were treated separately on the best-fit (second best likelihood score) tree from the combined ITS and cpDNA data analyses to account for branch lengths. Branches are: solid black (terrestrial), black outlined (aquatic), dashed (semiaquatic). Ambiguous ancestral character states at the nodes of interest are indicated by a circle. Aquatic, semiaquatic, and terrestrial habit transitions are indicated in the boxes by each numbered node according to the maximum likelihood estimations of ancestral states.

View larger version (29K): [in this window] [in a new window]   Fig. 6. Maximum likelihood reconstructions of ancestral floral character states in Haloragaceae. Dimerous, trimerous, tetramerous, and pentamerous flowers were treated separately on the best-fit (second best likelihood score) tree from the combined ITS and cpDNA data analyses to account for branch lengths. Branches are represented as black dashed-outline (dimerous), long solid-dashed (trimerous), solid (tetramerous), and short-dashed (pentamerous). Ancestral character states are indicated in the boxes by each pertinent numbered node along with the maximum likelihood probabilities for ancestral states denoted as trimerous (3) or tetramerous (4).

Data and phylogenetic analyses
A primary concern in choosing data for these analyses was to incorporate genes that were informative at multiple taxonomic levels. In this case, relationships of the family Haloragaceae at the intergeneric and intrageneric level required genes with different evolutionary rates. The nrDNA ITS region is highly variable, and its use has been recommended when examining interspecific relationships while its utility for examining intergeneric relationships has varied widely depending on lineage (Baldwin et al., 1995 ). Its high variability may be problematic when the data are applied singly at higher phylogenetic levels (e.g., Baldwin et al., 1995 ; Soltis and Kuzoff, 1995 ; Kim and Jansen, 1996 ), especially when there has been little divergence time between multiple lineages at deep nodes (Donaghue and Sanderson, 1995 ; Seelanen et al., 1997 ). Because the cpDNA matK-coding region evolves at a slower rate, this locus was expected to provide a more accurate picture of higher-level relationships, especially intrageneric relationships, although we also expected that some informative characters would facilitate elucidation of lower-level relationships, especially when including the trnK introns. Not surprisingly, the ITS data set and cpDNA data set provided some incongruent results when treated separately at levels less appropriate for each given data set, and they have very different likelihood model parameters (Table 2). At the family level, ITS displayed high homoplasy (CI = 0.37), a result that proved, as would be expected, to be much more moderate at lower phylogenetic levels (see Table 3). Also, branches along the backbone of the phylogeny were relatively short and had little support, especially under parsimony bootstrap (Figs. 1 and 4c). The cpDNA analyses also resolved short branches along the backbone of the phylogeny, but homoplasy was much lower (CI = 0.67), and nodal support was strong under both parsimony and Bayesian analyses for the resolved relationships.

As discussed earlier, incongruence may also exist between the nuclear and plastid genomes as a consequence of hybridization or lineage sorting. In the case of incongruence between data sets, each case was examined individually, and interpretation of the combined data analyses includes discussion of incongruence where appropriate.

Phylogenetic hypotheses
Penthorum and Tetracarpaea
The position of Penthorum and Tetracarpaea as part of Haloragaceae s.l. has been suggested based solely on results of recent large-scale phylogenetic analyses in which these disparate genera were found to ally as a grade in a clade including Haloragaceae and Aphanopetalaceae (Fishbein et al., 2001 ; APG II, 2003 ). Morphological similarities were not considered in this disposition. Until recently, Penthorum had been included in Saxifragaceae, whereas Tetracarpaea comprised a monotypic Tetracarpaeaceae. Based on the phylogenetic results of Fishbein et al. (2001) , we used Aphanopetalum as an outgroup in these analyses to consider the position of Penthorum and Tetracarpaea within a Haloragaceae s.l. Examination of the phylograms (Fig. 4a–c) and comparisons of pairwise distances of combined data (Table 4) indicated that Penthorum and Tetracarpaea are much more distinct from all Haloragaceae taxa than any Haloragaceae are to each other at the molecular level.

Basal nodes of Haloragaceae
The combined and cpDNA analyses resolved Glischrocaryon-Haloragodendron as sister to the rest of the family (PP = 1.0; BS = 52), a result that conflicted with results from ITS analyses (Figs. 1–3). There was a high level of homoplasy in ITS (CI = 0.37; Table 3) at this deep phylogenetic level along with the weak parsimony bootstrap support (<0.50) along the backbone of the phylogenetic hypothesis (Fig. 1). Additionally, strict consensus of parsimony trees one step longer than the best trees (1747 steps) results in a polytomy of four clades (Haloragodendron-Glischrocaryon, Haloragis brownii-H. tenuifolia, Proserpinaca, and all other Haloragaceae). In contrast, the cpDNA provided strong support for Haloragodendron-Glischrocaryon as sister to all other Haloragaceae (BS = 93; PP = 1.0) with much lower homoplasy (CI = 0.67).

It is noteworthy that when the data sets are combined, Bayesian analysis resolves Glischrocaryon-Haloragodendron sister to all other Haloragaceae with a strong posterior probability (PP = 1.0) but weak parsimony bootstrap (BS = 52; Fig. 3), thus resolving the same phylogenetic hypothesis as the cpDNA taken alone (Fig. 2). Posterior probability has been demonstrated to represent a less biased estimator of confidence than BS (Wilcox et al., 2002 ; Erixon et al., 2003 ), even if the support provided by PPs may represent a more liberal estimate of branch support, leading to the potential for disproportionately high support for some branches (Suzuki, 2002 ). Recent evidence also has shown that PP scan assign high confidence to extremely short, incorrectly inferred branches (Alfaro et al., 2003 ); thus, branch length is important to consider in interpreting the high PP support provided for the contrasting results between the combined and ITS analyses for the basal nodes (Fig. 4a, c). Given the higher quality of the cpDNA data for this level of phylogeny (based on measures of homoplasy; see Table 3), we consider that the resolution and branch support provided by cpDNA in the parsimony and Bayesian analyses and the strongly supported resolution of Haloragodendron-Glischrocaryon as sister to all Haloragaceae in the Bayesian analysis of the combined data analyses is the most likely (Figs. 2, 3). Perhaps the incorporation of additional, more slowly evolving, single-copy nuclear markers would provide a better understanding of higher-level relationships in Haloragaceae.

Glischrocaryon-Haloragodendron
The monophyly of Haloragodendron-Glischrocaryon is well supported, but monophyly of two distinct genera is not supported in all data sets (Figs. 1–3). ITS and cpDNA analyses yield conflicting phylogenetic hypotheses. ITS strongly supports both a monophyletic Glischrocaryon and Haloragodendron, whereas the cpDNA supports paraphyly of Haloragodendron in respect to Glischrocaryon. The incongruence between data sets cannot be easily reconciled. Shallow basal branches are resolved for this clade (Fig. 4a, b) by both data sets, perhaps suggesting a rapid radiation (Fishbein et al., 2001 ; Donoghue and Sanderson, 1992). The combined data also suggest that Haloragodendron is paraphyletic in relation to Glischrocaryon (Fig. 3), but nodal support, especially parsimony bootstrap values, is reduced for basal relationships in this clade indicating the incongruence may have a biological basis (i.e., ancient hybridization, lineage sorting) and needs to be explored further.

Morphology also supports the close relationship of Haloragodendron and Glischrocaryon but does not provide a clear view of generic limits. Orchard (1975) pointed out the strong resemblance of fruit characters between the two, both having a single seed develop (from four ovules) and occupy the entire fruit. Praglowski (1970) emphasized that the pollen morphology of Haloragodendron (= Haloragis sect. Pleianthus subsect. Spongiocarpus in part) and Glischrocaryon was nearly identical and differed from that of all other Haloragaceae, leading him to suggest their merger. Glischrocaryon species are distinct from Haloragodendron based on vegetative morphology. They have annual green, round stems and often highly reduced, alternate leaves (i.e., characteristics associated with the arid habitats [excluding alternate leaves] in which these taxa are found). In contrast, species assigned to Haloragodendron have perennial brown-red, four-angled stems with relatively large opposite leaves. Other Haloragaceae genera (i.e., Haloragis and Gonocarpus) have both alternate- and opposite-leaved members, suggesting the ambiguity of this characteristic in defining generic limits in the family. Given the conflict between molecular analyses, further sequence data are needed to evaluate these inconsistencies before these generic limits can be confidently resolved; thus, the circumscription of these genera remains tentative.

Incongruence between cpDNA and ITS phylogenetic hypotheses is also found regarding the dimerous Glischrocaryon behrii. ITS supports its position as part of a clade including G. angustifolium-G. roei, whereas cpDNA places it as paraphyletic to other Glischrocaryon and sister to the "eastern Haloragodendron" (Figs. 1–3). Glischrocaryon behrii is the only member of Glischrocaryon that ranges into eastern Australia. Its sister group relationship to the "eastern Haloragodendron" (Fig. 2) may reflect an evolutionary split between this clade and Glischrocaryon. The combined analysis supports placement of G. behrii with Glischrocaryon, perhaps reflecting the strong support given by ITS and the relatively short branches resolved for the cpDNA hypothesis (Fig. 4b).

Given our inclusive sampling of Glischrocaryon and Haloragodendron, some discussion of intergeneric relationships is warranted. All data sets resolve a strongly supported "eastern Haloragodendron" (Figs. 1–3), which are geographically disjunct from the western species (H. glandulosum and H. racemosum). A close relationship between Glischrocaryon flavescens (Drumm.) Orchard and G. aureum is also well supported by all data sets, a relationship predicted by Orchard (1975) . We collected multiple samples of the two varieties of Glischrocaryon aureum (Lindl.) Orchard as part of this study. The more widespread G. aureum var. angustifolium (Nees) Orchard is distinct from G. aureum var. aureum at the molecular level (Figs. 1–4) and is strongly supported as the sister to G. roei (cpDNA and combined data) or G. roei + G. behrii (ITS). Orchard (1975) had considered G. aureum var. angustifolium as a link between G. aureum var. aureum and G. roei Endl. It shares vegetative and habitat features with G. roei but lacks its highly inflated fruits. Conversely, G. aureum var. angustifolium possesses the winged fruits found in G. aureum var. aureum. Orchard suggested G. aureum var. angustifolium may have arisen through hybridization between G. aureum and G. roei, and our data do not preclude this possibility. Given the morphological and phylogenetic evidence, we propose that Glischrocaryon angustifolium (= G. aureum var. angustifolium: Table 5) should be recognized as a distinct species.

Proserpinaca and Meionectes
Proserpinaca and Haloragis brownii-H. tenuifolia are closely related yet distinct from each other and retain a basal branching position relative to most other Haloragaceae in all phylogenetic hypotheses (Figs. 1–3). These taxa are disjunct; Proserpinaca is found only in the New World (South America to Canada), whereas Haloragis brownii and H. tenuifolia are Australian endemics. The fossil record shows that Proserpinaca once had a much wider range, possibly having a circumboreal distribution as early as the Pliocene (Katz et al., 1965 ; Praglowski, 1970 ; Huckerby and Oldfield, 1976 ). Although the precise placement of Haloragis brownii-H. tenuifolia in relation to Proserpinaca and Glischrocaryon-Haloragodendron remains uncertain, the monophyly of Haloragis brownii-H. tenuifolia and the placement of this clade outside of Haloragis is clearly supported by all analyses (Figs. 1–3).

The aquatic habit of H. brownii and H. tenuifolia is uncommon in Haloragis (except for the helophyte H. heterophylla) as are the dimerous flowers of H. brownii and the trimerous flowers of H. tenuifolia. The trimerous (sometimes dimerous) perfect flowers and pinnatifid, submerged leaves of H. tenuifolia are similar to those in Proserpinaca. The most evident morphological differences between Haloragis brownii-H. tenuifolia and Proserpinaca are found in other floral characters. Whereas Haloragis brownii and H. tenuifolia have conspicuous, hooded petals and two whorls of stamens, Proserpinaca has only rudimentary petals and a single whorl of stamens, reductions that are also common elsewhere in the family. Brown (1814) recognized a monotypic Meionectes (= M. brownii), but most subsequent authors have synonymized Meionectes with Haloragis (Candolle, 1828 ; Schindler, 1905 ; Orchard, 1975 , 1990 ). Based on the molecular evidence (Figs. 1–3) and the unique morphology of Haloragis brownii-H. tenuifolia in relation to other Haloragaceae, we reinstate the genus Meionectes H.Br., which we circumscribe as comprising solely these two species (see Table 5).

Gonocarpus
Various discrepancies exist in respect to the relationships proposed for Gonocarpus species in the treatments of Orchard (1975) and Schindler (1905) . Notably, Schindler (1905) placed Gonocarpus in synonymy with Haloragis, yet all known species at the time (now included in Gonocarpus) were in section Monanthus of Haloragis, except G. hexandrus, G. nodulosus Nees, and G. paniculatus (R.Br. ex Benth.) Orchard (Appendix S1, see Supplemental Data in online version of this article). Gonocarpus nodulosus was placed within its own subgenus (Pseudohaloragis), and G. hexandrus was included in subsection Trihalorraghis along with the aquatic H. tenuifolia, which has already been discussed. Orchard (1975) split Gonocarpus from Haloragis but did not interpret the genus in a phylogenetic context. Our combined data support a monophyletic Gonocarpus (excluding G. hexandrus) sister to Haloragis.

Although Orchard (1975 , 1990 ) placed G. hexandrus within Gonocarpus, this was the only species for which he found no affiliation with other taxa in the genus. This species is unique among Gonocarpus with its trimerous flowers and racemose inflorescences (the last found otherwise only in G. paniculatus). In the final combined data analysis (Fig. 3), this taxon is sister to a clade of Haloragis-Gonocarpus + Laurembergia-Meziella-Myriophyllum. This relationship is inconsistent between cpDNA and ITS. Bayesian analysis of ITS weakly supported G. hexandrus as sister to Laurembergia-Meziella-Myriophyllum (PP < 0.95), whereas parsimony did not resolve this relationship (Fig. 1). Alternatively, cpDNA weakly resolved G. hexandrus as sister to Haloragis-Gonocarpus + Laurembergia-Meziella-Myriophyllum (also representative of the combined data). This taxon is clearly distinct phylogenetically from Gonocarpus and all other Haloragaceae genera (Figs. 1–3) and unique from all Gonocarpus taxa with trimerous flowers, thus we recognize it as the distinct, monotypic genus Trihaloragis (see Table 5). Three subspecies of G. hexandrus have been recognized (Orchard, 1975 , 1990 ), and further studies of this poorly known taxon are warranted.

Although incomplete, our representative sampling of Gonocarpus allows for some discussion of relationships in the genus. All analyses resolve the sister relationship of G. montanus (Hook.f.) Orchard (= H. depressa var. montana) and G. micranthus (PP = 1.0). Although Orchard (1975) suggested that these taxa were only distantly related, both were included in subsection Lamprocalyx by Schindler (1905) because they share similar fruit types and large, deltoid calyx lobes. The "benthamii clade" is well supported (Figs. 1–3) and corresponds with Orchard's (1975) hypothesis of a closely related Western Australia group (Appendix S1: node 1, group D). This clade includes G. paniculatus (the only racemose Gonocarpus), which Schindler (1905) placed in subsection Spongiocarpus among the taxa now included in Haloragodendron (Appendix S1). A sister relationship of G. acanthocarpus and G. leptothecus is well supported in all analyses, an expected result given their synonymous treatment as G. acanthocarpus by most authors prior to Orchard (1975) .

A clade of several closely related taxa (here referred to as the tetragynus clade; Fig. 3) is also well supported in all analyses, but relationships among the species remain dubious. Before the description of G. humilis and G. oreophilus by Orchard (1975) , herbarium specimens of the taxa from this clade typically were identified either as G. tetragynus or G. teucrioides. Although these taxa appear to be closely related (as would be expected from morphology), they also are divergent at the molecular level (Fig. 4). However, incongruence between phylogenetic hypotheses based on nuclear and chloroplast markers among these taxa may indicate a history of hybridization (Figs. 1–2). Forms intermediate between G. elatus and either G. oreophilus or G. mezianus (see Fig. 3) have been described (Orchard, 1975 ). For this study, extreme morphological forms of members of the tetragynus clade were used for molecular analyses. Forms with morphological intergradations among these taxa also were collected in the field and possessed various degrees of polymorphism at ITS nucleotide sites (Appendix S1). The polymorphic states reflected differences at sites that were clearly associated with the well-defined morphological taxa, which we interpreted as a consequence of occasional, recent hybridization among these taxa. To better understand the relationships among members of the tetragynus clade and their propensity for hybridization, additional studies with more intensive population-level sampling are needed.

Haloragis
Although species richness in Haloragis was reduced by splitting Gonocarpus and HaloragodendronOrchard (1975) , the genus still displays wide morphological diversity among the >26 species remaining. Characters within Haloragis include alternate or opposite leaves, terrestrial or semiaquatic habit, two-, three- or four-merous flowers, and 2–4-loculed ovaries. All our data strongly support the monophyly of Haloragis (excluding H. brownii-H. tenuifolia). Haloragis can be distinguished from Gonocarpus by a host of morphological characters involving carpel septation, locule number, and many-flowered dichasia (Orchard, 1975 ).

There is little resolution and/or weak support for most intrageneric relationships in Haloragis, yet some are notably well supported and congruent among data sets. Combined data and cpDNA analyses agreed in resolving H. trigonocarpa as sister to all other Haloragis, whereas ITS analyses resulted in a basal polytomy (Fig. 1). Haloragis trigonocarpa is one of two trimerous species in Haloragis. The other species, H. gossei, was not sampled but recently has been considered closely related to H. trigonocarpa, the two differing only in minor fruit characteristics (Orchard et al., 2005 ). Also notable is the position of the helophytic H. heterophylla as nested within Haloragis, clearly demonstrating its independent origin from the aquatic Meionectes (= H. brownii-H. tenuifolia; Fig. 3).

Laurembergia-Meziella-Myriophyllum
Laurembergia is sister to Meziella-Myriophyllum in all analyses, and this relationship is well supported (Figs. 1–3). The helophytic Laurembergia differs from other aquatic taxa in the family by lacking a submerged form that is distinct from the terrestrial form. In its fruit and ovary structure, Laurembergia is strikingly similar to Gonocarpus, which was the affiliation favored by Orchard (1975) . However, it is the only genus in the family besides Myriophyllum in which truly unisexual flowers are found, although functional monoecy has been described for several Gonocarpus.

A nested position of Meziella within Myriophyllum is well supported in the cpDNA and combined analyses. Its position is resolved as sister to all Myriophyllum in ITS Bayesian analyses with weak support (PP = 0.89), whereas parsimony did not resolve the relationship but included Meziella instead as part of a polytomy with two Myriophyllum clades. The strong support in the combined and cpDNA analysis for the inclusion of Meziella within Myriophyllum and lack of support for the ITS results brings into question its generic status. Meziella has been regarded as distinct from Myriophyllum based on a host of characters, but most conspicuously its four nutlets do not separate because of a persistent spiny calyx, whereas in Myriophyllum the nutlets lack the spiny calyx and separate at maturity (Orchard and Keighery, 1993 ). Within Haloragaceae, four woody nutlets form only within Myriophyllum and Meziella. Other characters used as evidence to support the unique generic status of Meziella (e.g., single whorl of stamens and apiculate stamens [Orchard and Keighery, 1993 ]) also occur in some Myriophyllum species (e.g., M. mattogrossense Hoehne and M. decussatum Orchard [not included in these analyses]). Molecular evidence indicates that its status as a distinct genus deserves further consideration. Given that the nrDNA data resolved Meziella sister to Myriophyllum (although with weak support), we suggest a conservative approach, to retain Meziella, pending the outcome of further analyses using more informative genomic DNA data.

Two well-supported clades are resolved within Myriophyllum, one comprising most of the endemic Australian and North American species and another comprising the South American species, circumboreal taxa (e.g., M. sibiricum Kom., M. verticillatum L.), and a group of several predominately southern hemisphere species once placed in synonymy with M. elatinoides (now part of the "elatinoides clade"; Fig. 3). Although our sampling of Myriophyllum provided a thorough coverage of the major lineages, which was sufficient to address the questions explored in this study, a more comprehensive survey of taxa has been conducted specifically to examine the systematics of the genus (Moody, 2004 ; M. Moody and D. Les, unpublished data) and will be discussed in detail in a subsequent treatment.

Character evolution
Inherent to the accuracy of our ancestral state reconstruction is the accuracy of our combined phylogenetic analyses. Some of the relationships resolved by the combined data are incongruent with our phylogenetic hypothesis based on ITS data alone, most importantly regarding the phylogenetic position of Meionectes, Proserpinaca and Trihaloragis. The well-supported resolution and agreement between cpDNA and combined phylogenetic hypotheses regarding deep nodes suggests that the hypotheses based on these data sets are the most reliable regarding ancestral character-state reconstruction for Haloragaceae. In some cases, relatively short branches were recovered regarding crucial relationships (i.e., Trihaloragis) for our ancestral character-state analyses (Fig. 4a) because of this ML ancestral character state reconstructions, which take into account branch length, are particularly insightful (Figs. 5, 6).

Aquatic habit
The evolution of the aquatic habit is suspected to have at least 50 and perhaps upward of 100 separate origins among angiosperms (Cook, 1996 ). Although the advent of the aquatic habit is suspected to have evolved multiple times within core eudicot families (Cook, 1996 ), evidence supporting multiple origins of the submerged aquatic habit within a single core eudicot family has not yet been presented to the best of our knowledge. The transition from a terrestrial to aquatic habit, followed by a reversal to a terrestrial habit within angiosperms also appears not to have been addressed previously. A recent, extensive literature search uncovered only a single reference in which such reversals were proposed (Cook, 1999 ); however, no such hypothesis has yet been tested using a comparative phylogenetic approach.

Unfortunately, our results are equivocal as to whether a reversal from an aquatic to a terrestrial habit has occurred in Haloragaceae (Fig. 5). The ML analyses of ancestral characters yielded a higher probability in support of three origins of the aquatic habit (rather than two independent origins to aquatic habit with an intermediate reversal; Fig. 5), but the level of probability was not significant. Notably, species of Proserpinaca and Meionectes share the ability to survive terrestrial conditions for prolonged periods of time. Their retention of terrestrial characteristics such as rigid stems and leaves with cuticles and stomata could facilitate a transition back to land. It must be noted that a transition to an aquatic habit directly from woody, perennial Haloragodendron-Glischrocaryon ancestors (Fig. 5) may represent a more extreme shift, but whether transitional taxa may have existed but become extinct is unknown.

Our phylogenetic evidence does suggest at least two transitions to an aquatic habit in Haloragaceae (Fig. 5). One of the notable aspects concerning multiple aquatic origins in the family is the convergence of similar vegetative characteristics (adaptations to submersed conditions) among the aquatic genera. Most notable are the pinnate leaves in Myriophyllum and Proserpinaca. Although highly dissected leaves are common among aquatic taxa (Sculthorpe, 1967 ; Cronk and Fennessy, 2002), the morphology of a linear, central rachis with linear, parallel, lateral pinnae borne from the rachis is uncommon, yet it is a feature that appears to have evolved independently within Haloragaceae. Meionectes brownii and M. tenuifolia have a similar pinnate pattern in their submerged leaves but differ in having either multifid pinnae (M. brownii) or a flattened laminate rachis (M. tenuifolia). The convergent characteristic of an amphibious habit among the aquatic Haloragaceae genera also is noteworthy; an amphibious habit may well have facilitated the adaptation of Myriophyllum and Meionectes to Australia's severe climatic fluctuations.

Merosity
Recent phylogenetic analyses of the angiosperms have defined a well-resolved core eudicot clade (APG II, 2003 ). The early diverging eudicots have been recognized as having a high level of lability in floral merosity and floral form (Endress, 1994 ; Drinnan et al., 1994 ), a characterization largely supported by ancestral character state reconstructions on phylogenetic hypotheses (Albert et al., 1998 ). In turn, the core eudicots appear to have a much less labile merosity with pentamery prominent (Endress, 1990 ; Drinnan et al., 1994 ; Albert et al., 1998 ). Recent phylogenetic analyses of core eudicots suggest a sister-group relationship of Gunnerales to the core eudicots, leading to the hypothesis that the dimerous flowers found among Gunnerales and other eudicots represent a transitional merosity leading to pentamery in the core eudicots (Soltis et al., 2003 ).

Although tetramerous flowers are common among core eudicot families (e.g., Brassicaceae, Crassulaceae, Haloragaceae, etc.), dimerous and trimerous flowers are unusual. Haloragaceae are nested well within Saxifragales in the core eudicots, yet vary extremely in merosity. Dimerous flowers are found in Glischrocaryon behrii, Meionectes brownii, Haloragis digyna, and Myriophyllum coronatum Meijden. Soltis et al. (2003 , p. 466) state, "In core eudicots there is sometimes variation between a pentamerous and tetramerous perianth merosity, but there is not a dimerous perianth (in contrast to the early-branching eudicots)" [emphasis ours], a statement that needs modifying. Although dimerous flowers may be uncommon in the core eudicots, evidence suggests that they have evolved by way of reductions multiple times within Haloragaceae (Fig. 6).

Trimerous flowers are uncommon among the core eudicots, although they are found among the unisexual flowers of Viscaceae and Fagaceae. Perfect trimerous flowers such as those found in Haloragaceae are even more uncommon (Judd et al., 2002 ). Both parsimony and ML ancestral character state reconstructions support a transitional pathway to trimerous flowers from tetramerous flowers with a reversal to tetramery (Fig. 6) rather than multiple reductions within Haloragis and Gonocarpus (Schindler, 1905 ; Orchard, 1975 ). Although it is indisputable that labile merosity characterizes the early-branching eudicots (Endress, 1994 ; Drinnan et al., 1994 ), lability of merosity among core eudicots may deserve further evaluation at the family level. Our results have shown Haloragaceae to be a clear example of highly labile merosity within the core eudicots with multiple origins of dimerous and trimerous flowers (Fig. 6).

Conclusions
A molecular phylogenetic approach toward the systematics of Haloragaceae has helped to resolve relationships within the family that previously were difficult to elucidate. Reinstatement of Meionectes and recognition of Trihaloragis provide new perspectives in the taxonomy and evolution of Haloragaceae. Genetic distances based on DNA sequence data and patterns of morphological character distributions combine to show a distant sister group relationship of Haloragaceae to Penthorum and Tetracarpaea and support a reevaluation of the option of merger of these genera with the family. Phylogenetic analysis of DNA sequence data indicates that the aquatic habit has evolved multiple times within Haloragaceae; however, a possible reversal from an aquatic to a terrestrial habit within the family cannot be rejected. A further assessment of the evolution of aquatic habit that couples phylogenetic information with developmental studies would be beneficial and might help to uncover patterns in the evolution of characteristics adaptive to the aquatic habit. Merosity is labile among Haloragaceae with dimerous and trimerous flowers, apparently having evolved multiple times within this angiosperm family, a pattern not predicted by the distribution of floral characteristics within other core eudicots. Comprehensive phylogenetic studies continue to be a vital tool in assessing character evolution.

Taxonomy
Trihaloragis
M. L. Moody & D. H. Les, gen. nov. TYPE: Haloragis hexandra F. Muell., Fragm. 3. 1862. Trihaloragis hexandra (F. Muell.) M. L. Moody & D. H. Les.

Herba perennis vel suffrutex, caules 4-costati. Folia in partibus inferioribus opposita, superis alterna, lanceolata ad oblanceolata. Flores 3-meri, perfecti; sepala 3, deltoidea, viridia; petala 3, cucullata, viridia ad cremea; stamina 6; styli 3; ovarium viride, ovoideum, 3-costatum, incomplete 3-loculatum, septis solum in partibus inferioribus, loculis petalis oppositis, uniovulatis.

Perennial herb or subshrub, stems 4-ribbed. Leaves opposite basally becoming alternate in upper parts, lanceolate to oblanceolate. Flowers trimerous, perfect; sepals 3, deltoid, green; petals 3, hooded, green to cream; stamens 6; styles 3; ovary green, ovoid, 3-ribbed, incompletely 3-locular, septae present only in lower part, locules opposite the petals, one ovule per locule.

Trihaloragis hexandra
(F. Muell.) M. L. Moody & D. H. Les, comb. nov. Basionym: Haloragis hexandra F. Muell., Fragm. 3, p. 31. 1862. Haloragis lanceolata R. Br. ex Benth., Fl. Aust 2. 1864. Gonocarpus hexandrus (F. Muell.) Orchard, Tax. Rev. Fam. Halorag. 1975. TYPE: Australia: "In locis uliginosis ad sinum Wilson's Inlet Novae Hollandiae austro-occidentalis." Oldfield, s.n. (holotype MEL!; isotype MEL!).

Glischrocaryon angustifolium
(Nees) M. L. Moody & D. H. Les, comb. et stat. nov. Basionym: Loudonia aurea var. angustifolia Nees in Lehm., Pl. Preiss 1. p. 159. 1844. Glischrocaryon aureum var. angustifolium (Nees) Orchard, Tax. Rev. Fam. Halorag. 1975. TYPE: "In glareosis sterilibus districtus Hay m. Novembri a. 1840. Herb." Preiss 2079 (holotype B?: not located; isotype P!).

Meionectes tenuifolia
(Benth.) M. L. Moody & D. H. Les, comb. nov. Basionym: Haloragis tenuifolia Benth., Fl. Aust. 2, p. 477. 1864. Type: Australia: "W. Australia, Drummond, 4th Coll. n. 86" (isotype MEL!).

APPENDIX 1.

Taxa used for molecular analyses. When multiple taxa are listed, these multiple specimens were sampled and sequenced for ITS. GenBank numbers follow taxonomic authority. An em dash (—) in place of data for a gene indicates missing data. Taxa for which DNA was extracted directly from herbarium specimens are represented by an asterisk (*) before the collector's name. ANBG –Australian National Botanic Gardens. HPBG – Harold Porter Botanical Garden. Herbarium abbreviation as per Index Herbariorum.

Taxon—GenBank accession nos.: ITS, trnK 5' intron, matK, trnK 3'; Voucher specimen Herbarium, Source.

Glischrocaryon angustifolium (Nees) Moody & Les—EF178776, EF178958, EF179050, EF178866. Moody 393 CONN, WA, Australia; Moody 416 CONN, WA, Australia. G. aureum (Lindl.) Orchard—F178779, EF178961, EF179053, EF178869. Moody 386 CONN, WA, Australia; Moody 422 CONN, WA, Australia. G. behrii (Schindl.) Orchard—EF178777, EF178959, EF179051, EF178867. Moody 436 CONN, Vic., Australia; ANBG, ACT, Australia, G. flavescens (Drumm.) Orchard—EF178778, EF178960, EF179052, EF178868. Moody 394 CONN, WA, Australia. G. roei Endl.—EF178775, EF178957, EF179049, EF178865. Moody 397 CONN, WA, Australia. G. roei x G. angustifolium, Moody 391 CONN, WA, Australia.

Gonocarpus acanthocarpus (Brongn.) Orchard—EF178757, EF178939, EF179031, EF178847. *Thompson 187 NSW, QLD, Australia. G. benthamii Orchard—EF178756, EF178938, EF179030, EF178846. Moody 399 CONN, WA, Australia. G. elatus (A.Cunn. ex. Fenzl) Orchard—EF178752, EF178934, EF179026, EF178842. Moody 489 CONN (cult.) NSW, Australia. G. ephemerus Orchard—EF178754, EF178936, EF179028, EF178844. *Patrick 3463 PERTH, WA, Australia. G. eremophilus Orchard—EF178769, EF178951, EF179043, EF178859. *Latz 8808 NSW, NT, Australia. G. hexandrus (F. Muell.) Orchard—EF178759, EF178941, EF179033, EF178849. *Bright 93 PERTH, WA, Australia; Lepschi 3360 PERTH, WA, Australia. G. humilis Orchard—EF178763, EF178945, EF179037, EF178853. Moody 429 CONN, NSW, Australia; Moody 487 CONN, NSW, Australia. G. intricatus (Benth.) Orchard—EF178753, EF178935, EF179027, EF178843. Moody 401 CONN, WA, Australia. G. leptothecus (F. Muell.) Orchard—EF178758, EF178940, EF179032, EF178848. Martine 803 CONN, NT, Australia. G. longifolius (Schindl.) Orchard—EF178772, EF178954, EF179046, EF178862. cult. ANBG, ACT, Australia; *James s.n. NSW, Australia. G. mezianus (Schindl.) Orchard—EF178760, EF178942, EF179034, EF178850. Moody 442 CONN, NSW, Australia. G. micranthus, Thunb. —EF178771, EF178953, EF179045, EF178861. Moody 479 CONN, NSW, Australia; Moody 463 CONN, NSW, Australia. G. montanus (Hook. f.) Orchard—EF178770, EF178952, EF179044, EF178860. Moody 448 CONN, NSW, Australia; Moody 449b CONN, NSW, Australia. G. nodulosus Nees—EF178762, EF178944, EF179036, EF178852. Moody 396 CONN, WA, Australia. G. oreophilus Orchard—EF178751, EF178933, EF179025, EF178841. Moody 490 CONN, cult. ACT, Australia. G. paniculatus (R. Br. ex Benth.) Orchard—EF178765, EF178947, EF179039, EF178855. Moody 414 CONN, WA, Australia; Moody 406 CONN, WA, Australia. G. pusillus (R. Br. ex Benth.) Orchard—EF178766, EF178948, EF179040, EF178856. Moody 401 CONN, WA, Australia; *Middleton 101 PERTH, WA, Australia. G. salsoloides Rchb. ex Spreng. —EF178755, EF178937, EF179029, EF178845. *Coveny 17342 CANB, NSW, Australia. G. scordioides (Benth.) Orchard—EF178750, EF178932, EF179024, EF178840. Moody 392 CONN, WA, Australia; *Hislop 1233 PERTH, WA, Australia. G. tetragynus Labill. —EF178764, EF178946, EF179038, EF178854. Moody 468 CONN, NSW, Australia; Moody 474 CONN, NSW, Australia. G. teucrioides DC. —EF178761, EF178943, EF179035, EF178851. Moody 462 CONN, NSW, Australia; cult. ANBG, ACT, Australia. G. trichostachyus (Benth.) Orchard—EF178767, EF178949, EF179041, EF178857. Moody 398 CONN, WA, Australia. G. urceolatus Orchard—EF178768, EF178950, EF179042, EF178858. *Bean 14450 NSW, QLD, Australia. Gonocarpus(hybrid?). Moody 459 CONN, Vic., Australia; Moody 472 CONN, NSW, Australia; Moody 478 CONN, NSW, Australia.

Haloragis acutangula F. Muell.—EF178749, EF178931, EF179023, —. *Bates 33500 PERTH, SA, Australia. H. digyna Labill.—EF178747, EF178929, EF179021, EF178838. Moody 411 CONN, WA, Australia. H. dura Orchard—EF178745, EF178927, EF179019, EF178837. Moody 385 CONN, WA, Australia; Moody 390 CONN, WA, Australia. H. erecta (Banks ex. Murray) Oken—EF178736, EF178918, EF179010, EF178828. *Rixon 20 NSW, Auckland, NZ. H. exalata F. Muell. —EF178741, EF178923, EF179015, EF178833. *Miles s.n. PERTH, NSW, Australia; Miles s.n. PERTH, NSW, Australia. H. eyreana Orchard—EF178740, EF178922, EF179014, EF178832. *Jusaitis A78 PERTH, SA, Australia. H. foliosa, Benth. —EF178748, EF178930, EF179022, EF178839. Moody 420 CONN, WA, Australia. H. glauca Lindl. EF178744, EF178926, EF179018, EF178836. Moody 426 CONN, NSW, Australia. H. hamata Orchard—EF178738, EF178920, EF179012, EF178830. *Cranfield 10585 PERTH, WA, Australia. H. heterophylla Brongn.—EF178743, EF178925, EF179017, EF178835. Moody 458 CONN, NSW, Australia; Moody 471 CONN, WA, Australia. H. masatierrana Skottsb.—EF178737, EF178919, EF179011, EF178829. (cult.) CONN. H. serra Brongn. —EF178746, EF178928, EF179020, —. Moody 488 CONN (cult.) NSW, Australia. H. stricta R. Br.—EF178742, EF178924, EF179016, EF178834. *Williams 88074 PERTH, QLD, Australia. H. trigonocarpa F. Muell.—EF178739, EF178921, EF179013, EF178831. *DJE 3100 PERTH, WA, Australia.

Haloragodendron baeuerlenii (F. Muell.) Orchard—EF178783, EF178965, EF179057, EF178873. Moody 483 CONN (cult.) NSW, Australia. H. gibsonii Wilson & Moody—EF178785, EF178967, EF179059, EF178875. Moody 480 CONN (cult.) NSW, Australia. H. glandulosum Orchard—EF178780, EF178962, EF179054, EF178870. (cult.) ANBG 2000, ACT, Australia. H. lucasii (Maiden & Betche) Orchard—EF178784, EF178966, EF179058, EF178874. Moody 481 CONN (cult.) NSW, Australia; EF178786, EF178968, EF179060, EF178876. Moody 482 CONN (cult.) NSW, Australia; Moody 485 CONN, NSW, Australia; Moody 486 CONN, NSW, Australia. H. monospermum (F. Muell.) Orchard—EF178781, EF178963, EF179055, EF178871. Moody 475 CONN, NSW, Australia; Moody 491 CONN (cult.) ANBG, ACT, Australia. H. racemosum (Labill.) Orchard—EF178782, EF178964, EF179056, EF178872. Moody 400 CONN, WA, Australia; Moody 402 CONN, WA, Australia; Moody 403 CONN, WA, Australia.

Laurembergia repens (L.) P. J. Berguis—EF178735, EF178917, EF179009, EF178827. J. P. Rourke (cult.) HPBG, Cape Town, South Africa; Williams 113 C, South Africa.

Meionectes brownii Hook. f. —EF178773, EF178955, EF179047, EF178863. Moody 438 CONN, Vic, Australia; Moody 408 CONN, WA, Australia. M. tenuifolia (Benth.) Moody & Les—EF178774, EF178956, EF179048, EF178864. *Wright s.n. PERTH, WA, Australia; Cranfield 15976 PERTH, WA, Australia.

Meziella trifida (Nees) Schindl. —EF178734, EF178916, EF179008, EF178826. Moody 410 CONN, WA, Australia; Moody 405 CONN, WA, Australia; Moody 404 CONN, WA, Australia.

Myriophyllum alpinum Orchard—EF178720, EF178902, EF178994, EF178812. Moody 449 CONN, NSW, Australia; Moody 453 CONN, NSW, Australia. M. alterniflorum DC.—EF178704, EF178886, EF178978, EF178797. Moody 109a CONN, WI, USA; Moody 111 CONN, WI, USA. M. aquaticum (Vellozo) Verd.—EF178727, EF178909, EF179001, EF178819. Moody 51 CONN, CA, USA; Moody 56 CONN, FL, USA. M. balladoniense Orchard—EF178708, EF178890, EF178982, EF178801. Moody 389 CONN, WA, Australia. M. caput-medusae Orchard—EF178703, EF178885, EF178977, EF178796. Moody 443 CONN, NSW, Australia; Moody 446 CONN, NSW, Australia. M. coronatum Meijden—EF178717, EF178899, EF178991, EF178809. *Paijmans 3039 CANB, QLD, Australia. M. crispatum Orchard—EF178721, EF178903, EF178995, EF178813. Moody 445 CONN, NSW, Australia; Moody 437 CONN, NSW, Australia. M. drummondii Orchard—EF178725, EF178907, EF178999, EF178817. Moody 409 CONN, WA, Australia; Moody 417 CONN, WA, Australia. M. farwellii Morong—EF178731, EF178913, EF179005, EF178823. Moody 97 CONN, MN, USA; Moody 106 CONN, WI, USA. M. filiforme Benth.—EF178716, EF178898, EF178990, EF178808. *Wilson 1810 NSW, NT, Australia. M. heterophyllum Michx. —EF178733, EF178915, EF179007, EF178825. Moody 101 CONN, MN, USA; Moody 143 CONN, MA, USA. M. latifolium F.Muell.—EF178729, EF178911, EF179003, EF178821. *Orchard 4793 NSW, NSW, Australia; Jacobs 6706 NSW, NSW, Australia. M. laxum Shuttlew.—EF178732, EF178914, EF179006, EF178824. Moody 170 CONN, SC, USA; Moody 77 CONN, FL, USA. M. lophatum Orchard—EF178718, EF178900, EF178992, EF178810. Moody 456 CONN, NSW, Australia; Moody 455 CONN, NSW, Australia. M. mattogrossense Hoehne—EF178728, EF178910, EF179002, EF178820. Ritter 2314 LPB, Carrasco, Bolivia. M. muelleri Sond.—EF178719, EF178901, EF178993, EF178811. *Jacobs 6597 NSW, NSW, Australia. M. oguraense Miki—EF178705, EF178887, EF178979, EF178798. Kadono s.n. HYO, Hyogo, Japan. M. papillosum Orchard—EF178724, EF178906, EF178998, EF178816. Moody 424 CONN, NSW, Australia; Les 614 CONN, NSW, Australia. M. pedunculatum Hook. f.—EF178711, EF178893, EF178985, EF178804. Moody 452 CONN, NSW, Australia; Moody 467 CONN, NSW, Australia. M. petraeum Orchard—EF178712, EF178894, EF178986, EF178805. *Archer 1564 NSW, WA, Australia; *Brown 1123 PERTH, WA, Australia. M. quitense Kunth.—EF178700, EF178882, EF178974, EF178793. Moody 183 CONN, OR, USA. M. salsugineum Orchard—EF178701, EF178883, EF178975, EF178794, Moody 412 CONN, Vic, Australia; Les s.n. (cult.) U of Tasmania. M. sibiricum Kom.—EF178706, EF178888, EF178980, EF178799. Moody 82 CONN, CA, USA; Moody 99 CONN, MN, USA. M. spicatum L.—EF178702, EF178884, EF178976, EF178795. Moody 86 CONN, CA, USA; Moody 117 CONN, WI, USA. M. tenellum Bigelow—EF178730, EF178912, EF179004, EF178822. Moody 110 CONN, WI, USA; Moody 93 CONN, MN, USA. M. tillaeoides Diels—EF178710, EF178892, EF178984, EF178803. Moody 415 CONN, WA, Australia. M. trachycarpum F. Muell.—EF178715, EF178897, EF178989, EF178807. Jacobs 8843 NSW, NT, Australia; Martine 863 CONN, WA, Australia. M. ussuriense Maxim.—EF178726, EF178908, EF179000, EF178818. Kadono "Kasai-1" HYO, Hyogo, Japan; Kadono "Kasai-2" HYO, Hyogo, Japan. M. verrucosum Lindl.—EF178707, EF178889, EF178981, EF178800. Moody 427 CONN, NSW, Australia; Les 601 CONN, NSW, Australia. M. verticillatum L.—EF178709, EF178891, EF178983, EF178802. Moody 177 CONN, OR, USA; Moody 147 CONN, ME, USA. M. votschii Schindl.—EF178714, EF178896, EF178988, EF178806. *Rixon 31 NSW, Australia. M.sp. EF178713, EF178895, EF178987, —. Les 542 CONN, NSW, Australia. M.sp. nov. a, EF178722, EF178904, EF178996, EF178814. Les 622 CONN, NSW, Australia. M.sp. nov. b, EF178723, EF178905, EF178997, EF178815. Moody 476 CONN, NSW, Australia.

Proserpinaca palustris L.—EF178787, EF178969, EF179061, EF178877. Moody 59 CONN, FL, USA; Moody s.n. CONN, CT, USA. P. pectinata Lam.—EF178788, EF178970, EF179062, EF178878. *Nelson 19432 USCH, SC, USA; *Horn 4610 USCH, SC, USA.

Outgroups:

Penthorum sedoides L.—EF178789, EF178971, EF179063, EF178879. Moody 515 CONN, CT, USA. Tetracarpaea tasmanica Hook. f.—EF178790, — , EF179064, — *Wiecek 577 NSW, TAS, Australia. Aphanopetalum clematideum (Harv.) Domin—EF178792, EF178972, EF179065, EF178880. Moody 421 CONN, WA, Australia. A. resinosum (Harv.) Domin—EF178791, EF178973, EF179066, EF178881. Moody 484 CONN (cult.) NSW, Australia.

View larger version (16K): [in this window] [in a new window]   Fig. 4. Continued.

¹ The authors are especially grateful to S. Jacobs and P. Wilson (NSW) for extensive field assistance in Australia; also to R. Bayer, R. Cranfield, G. Crow, C. B. Hellquist, Y. Kadono, C. Martine, N. Ritter, J. Rourke, and G. Towler for assistance and/or collections in the field. They thank N. Tippery for assistance with Latin descriptions; P. Lewis, L. Lewis, C. Jones, and three anonymous reviewers for helpful discussion and/or comments on the manuscript; and the staff of the following herbaria for their generosity in loaning specimens: C, CAL, CANB, CBG, HBG, MEL, NSW, PERTH, R, UBC, UPS, USCH. This research was funded by NSF DDIG 0309123, NEBC Graduate Student Research Award Program, Karling Graduate Student Research award (BSA), William R. Anderson Student Research Grant (ASPT), and the University of Connecticut Bamford Endowment Fund.

² Author for correspondence (e-mail: miclmood{at}cyllene.uwa.edu.au ; current address: University of Western Australia, Department of Plant Biology, 35 Stirling Highway, Crawley, WA 6009, Australia

LITERATURE CITED

Albert V. A. Gustaffson M. H. G. Dilaurenzio L.. 1998. Ontogenetic systematics, molecular developmental genetics, and the angiosperm petal. In D. E. Soltis, P. S. Soltis, J. J. Doyle [eds.], Molecular systematics of plants II, 349-374. Kluwer, New York, New York, USA..

Alfaro M. E. Zoller S. Lutzoni F.. 2003. Bayes or bootstrap? A simulation study comparing the performance of Bayesian Markov chain Monte Carlo sampling and bootstrapping in assessing phylogenetic confidence. Molecular Biology and Evolution 20: 255-266..[Abstract/Free Full Text]

Angiosperm Phylogeny Group [APG].. 1998. An ordinal classification for the families of flowering plants. Annals of the Missouri Botanical Garden 85: 531-553..[CrossRef][Web of Science]

Angiosperm Phylogeny Group [APG II].. 2003. An updated classification of the angiosperms. Botanical Journal of the Linnean Society 141: 399-436..[CrossRef][Web of Science]

Avise J. C.. 1994. Molecular markers, natural history and evolution. Chapman and Hall, New York, New York, USA..

Baldwin B. G. Sanderson M. J. Porter J. M. Wojciechowski M. F. Campbell C. S. Donoghue M. J.. 1995. The ITS region of nuclear ribosomal DNA: a valuable source of evidence on angiosperm phylogeny. Annals of the Missouri Botanical Garden 82: 247-277..[CrossRef][Web of Science]

Barker F. K. Lutzoni F. M.. 2002. The utility of the incongruence length difference test. Systematic Biology 51: 625-637..[CrossRef][Web of Science][Medline]

Bawa B. S.. 1969a. Embryological studies on the Haloragidaceae. II. Laurembergia brevipes Schindl. and a discussion of systematic considerations. Proceedings of the National Academy of Sciences, India 35: 273-290..

Bawa B. S.. 1969b. Embryological studies on the Haloragidaceae. III. Myriophyllum intermedium DC. Beiträge zur Biologie der Pflanzen 45: 447-464..

Brown R.. 1814. General remarks, geographical and systematical on the botany of Terra Australis. In M. Flinders [ed.], A voyage to Terra Australis, (533–613). G. & W. Nichol, London, UK..

Bull J. J. Huelsenbeck J. P. Cunningham C. W. Swofford D. L. Waddell P. J.. 1993. Partitioning and combining data in phylogenetic analyses. Systematic Biology 42: 384-397..[Free Full Text]

Candolle A. P. de.. 1828. Prodromus systematis naturalis regni vegetabilis, sive enumeratio contracta ordinum generum specierumque plantarum huc usque cognitarum, juxta methodi naturalis normas digesta, vol. 3. Sumptibus Sociorum Treuttel et Wurtz, Paris, France..

Chippindale P. T. Wiens J. J.. 1994. Weighting, partitioning, and combining characters in phylogenetic analysis. Systematic Biology 43: 278-287..[Free Full Text]

Cook C. D. K.. 1996. Aquatic plant book. Academic Publishing, New York, New York, USA..

Cook C. D. K.. 1999. The numbers and kinds of embryo-bearing plants which have become aquatic: a survey. Perspectives in Plant Ecology, Evolution and Systematics 2: 79-102..[CrossRef]

Cronk J. K. Fennessy M. S.. 2001. Wetland plants: biology and ecology. Lewis Publishers, Boca Raton, Florida, USA..

Cronquist A.. 1968. The evolution and classification of flowering plants. Houghton Mifflin, Boston, Massachusetts, USA..

Darlu P. Lecointre G.. 2002. When does the incongruence length difference test fail?. Molecular Biology and Evolution 19: 432..[Abstract/Free Full Text]

Dolphin K. Belshaw R. Orme D. L. C. Quicke D. L. J.. 2000. Noise and incongruence: interpreting results of the incongruence length difference test. Molecular Phylogenetics and Evolution 17: 401-406..[CrossRef][Web of Science][Medline]

Donaghue M. J. Sanderson M. J.. 1992. The suitability of molecular and morphological evidence in reconstructing plant phylogeny. In P. S. Soltis, D. E. Soltis, J. J. Doyle [eds.], Molecular systematics of plants, 340-368. Chapman and Hall, New York, New York, USA..

Doyle J. J. Doyle J. L.. 1987. A rapid DNA isolation procedure for small quantities of fresh leaf tissue. Phytochemical Bulletin 19: 11-15..

Drinnan A. N. Crane P. R. Hoot S. B.. 1994. Patterns of floral evolution in the early diversification of non-magnoliid dicotyledons (eudicots). Plant Systematics and Evolution 8: 93-122..[CrossRef]

Ellstrand N. C. Whitkus R. Rieseberg L. H.. 1996. Distribution of spontaneous plant hybrids. Proceedings of the National Academy of Sciences, USA 93: 5090-5093..[Abstract/Free Full Text]

Endress P. K.. 1990. Patterns of floral construction in ontogeny and phylogeny. Biological Journal of the Linnean Society 39: 153-175..[CrossRef][Web of Science]

Endress P. K.. 1994. Floral structure and evolution of primitive angiosperms: recent advances. Plant Systematics and Evolution 192: 79-97..[CrossRef][Web of Science]

Erixon P. Svennblad B. Britton T. Oxelman B.. 2003. Reliability of Bayesian posterior probabilities and bootstrap frequencies in phylogenetics. Systematic Biology 52: 665-673..[Abstract/Free Full Text]

Fishbein M. Hibsch-Jetter C. Soltis D. E. Hufford L.. 2001. Phylogeny of Saxifragales (angiosperms, eudicots): analysis of a rapid, ancient radiation. Systematic Biology 50: 817-847..[Abstract/Free Full Text]

Graham S. W. Reeves P. A. Burns A. C. E. Olmstead R. G.. 2000. Microstructural changes in noncoding chloroplast DNA: interpretation, evolution, and utility of indels and inversions in basal angiosperm phylogenetic inference. International Journal of Plant Sciences 161: S83-S96..[CrossRef][Web of Science]

Huckerby E. Oldfield F.. 1976. The quaternary vegetational history of the French pays basque, II. Plant macrofossils, and additional pollen-analytical data. New Phytologist 77: 499-526..[CrossRef][Web of Science]

Huelsenbeck J. P. Ronquist F.. 2001. MrBayes: Bayesian inference of phylogenetic trees. Bioinformatics 17: 754-755..[Abstract/Free Full Text]

Hutchinson J.. 1959. The families of flowering plants, vol. 1, 2nd ed. Clarendon Press, Oxford, UK..

Johnson L. A. Soltis D. E.. 1994. matK DNA sequences and phylogenetic reconstruction in Saxifragaceae s. str. Systematic Botany 19: 143-156..[CrossRef][Web of Science]

Judd W. S. Campbell C. S. Kellogg E. A. Stevens P. F. Donoghue M. J.. 2002. Plant systematics: a phylogenetic approach, 2nd ed. Sinauer, Sunderland, Massachusetts, USA..

Katz J. A. Katz S. V. Kipianoi M. G.. 1965. Atlas and keys of fruits and seeds occurring in the quaternary deposits of the USSR Moscow. Academy of Sciences of the U.S.S.R., Commission for Investigations of the Quaternary Period, Nauka, Moscow, U.S.S.R..

Kim Y-D. Jansen R. K.. 1996. Phylogenetic implication of rbcL and ITS sequence variation in the Berberidaceae. Systematic Botany 21: 381-396..[CrossRef][Web of Science]

Les D. H. Crawford D. J. Landolt E. Gabel J. D. Kimball R. T.. 2002a. Phylogeny and systematics of Lemnaceae, the duckweed family. Systematic Botany 27: 221-240..[Web of Science]

Les D. H. Moody M. L. Jacobs S. W. L. Bayer R. J.. 2002b. Systematics of seagrasses (Zosteraceae) in Australia and New Zealand. Systematic Botany 27: 468-484..[Web of Science]

Lewis P. O.. 2001. A likelihood approach to estimating phylogeny from discrete morphological character data. Systematic Biology 50: 913-925..[Abstract/Free Full Text]

Linder C. R. Rieseberg L. H.. 2004. Reconstructing patterns of reticulate evolution in plants. American Journal of Botany 91: 1700-1708..[Abstract/Free Full Text]

Maddison D. R. Maddison W. P.. 2001. MacClade. Analysis of phylogeny and character evolution. 4.01. Sinauer, Sunderland, Massachusetts, USA..

Maddison W. P. Maddison D. R.. 2003. Mesquite: a modular system for evolutionary analyses, version 1.04. Website http://mesquiteproject.org [accessed October 2006]..

Melchior H.. 1964. Engler's syllabus der Pflanzenfamilien, vol. 2, 12th ed. Gebrüder Bornträger, Berlin, Germany..

Moody M. L.. 2004. Systematics of the angiosperm family Haloragaceae R. Br. emphasizing the aquatic genus Myriophyllum: phylogeny, hybridization and character evolution. Ph.D. dissertation, University of Connecticut, Storrs, Connecticut, USA..

Moody M. L. Hufford L. Soltis D. E. Soltis P. S.. 2001. Phylogenetic relationships of Loasaceae subfamily Gronovioideae inferred from matK and ITS sequence data. American Journal of Botany 88: 326-336..[Abstract/Free Full Text]

Moody M. L. Les D. H.. 2002. Evidence of hybridity in invasive watermilfoil (Myriophyllum) populations. Proceedings of the National Academy of Sciences, USA 99: 14867-14871..[Abstract/Free Full Text]

Mooers A. O. Schluter D.. 1999. Reconstructing ancestor states with maximum likelihood: support for one- and two-rate models. Systematic Biology 48: 623-633..[Free Full Text]

Morgan D. R. Soltis D. E.. 1993. Phylogenetic relationships among Saxifragaceae sensu lato based on rbcL sequence data. Annals of the Missouri Botanical Garden 80: 631-660..[CrossRef][Web of Science]

Nylander J. A. A.. 2003. MrModeltest 1.1b. Department of Systematic Zoology, EBC, Uppsala University, Sweden. Available at website http://www.abc.se/nylander/..

Nylander J. A. A. Ronquist F. Huelsenbeck J. P. Nieves-Aldrey J. L.. 2004. Bayesian phylogenetic analysis of combined data. Systematic Biology 53: 47-67..[Abstract/Free Full Text]

Orchard A. E.. 1975. Taxonomic revisions in the family Haloragaceae. 1. The genera Haloragis, Haloragodendron, Glischrocaryon, Meziella and Gonocarpus. Bulletin of the Auckland Institute and Museum 10: 1-299..[Medline]

Orchard A. E.. 1979. Haloragaceae in Australasia. 1. New Zealand: a revision of the genus and a synopsis of the family. Brunonia 2: 247-287..[CrossRef]

Orchard A. E.. 1981. A revision of South American Myriophyllum (Haloragaceae) and its repercussions on some Australian and North American species. Brunonia 4: 27-65..[CrossRef]

Orchard A. E.. 1985. Myriophyllum (Haloragaceae) in Australasia. II. The Australian species. Brunonia 8: 173-291..[CrossRef]

Orchard A. E.. 1990. Flora of Australia: Haloragaceae, vol. 18. AGPS Press, Canberra, Australia..

Orchard A. E. Keighery G. J.. 1993. The status, ecology and relationships of Meziella (Haloragaceae). Nuytsia 9: 111-117..

Orchard A. E. Lepschi B. J. Hislop M.. 2005. New taxa, a new record and a rediscovery in Western Australian Haloragis (Haloragaceae). Nuytsia 15: 431-443..

Praglowski J.. 1970. The pollen morphology of the Haloragaceae with reference to taxonomy. Grana 10: 159-239..

Rieseberg L. H.. 1991. Homoploid reticulate evolution in Helianthus: evidence from ribosomal genes. American Journal of Botany 78: 1218-1237..[CrossRef][Web of Science]

Rieseberg L. H.. 1997. Hybrid origins of plant species. Annual Review of Ecology and Systematics 28: 359-389..[CrossRef][Web of Science]

Rieseberg L. H. Wendel J. F.. 1993. Introgression and its consequences in plants. In R. G. Harrison [ed.], Hybrid zones and the evolutionary process, 70-109. Oxford University Press, New York, New York, USA..

Rogstad S. H.. 1992. Saturated NaCl-CTAB solution as a means of field preservation of leaves for DNA analyses. Taxon 41: 701-708..[CrossRef][Web of Science]

Ronquist F. Huelsenbeck J. P.. 2003. MrBayes 3: Bayesian phylogenetic inference under mixed models. Bioinformatics Applications Note 19: 1572-1574..[Abstract/Free Full Text]

Sang T. Zhong Y.. 2000. Testing hybridization hypotheses based on incongruent gene trees. Systematic Biology 49: 422-434..[Abstract/Free Full Text]

Schindler A. K.. 1905. Halorrhagaceae. In H. G. A. Engler [ed.], Das Pflanzenreich, IV, 1-133. Wilhelm Engelmann, Leipzig, Germany..

Sculthorpe C. D.. 1967. The biology of aquatic vascular plants. Edward Arnold, London, UK..

Seelanen T. A. Schnabel A. Wendel J. F.. 1997. Congruence and consensus in the cotton tribe (Malvaceae). Systematic Botany 22: 259-290..[CrossRef][Web of Science]

Shaw H. K. A.. 1966. A dictionary of the flowering plants and ferns, 7th ed. Harvard University Press, Cambridge, Massachusetts, USA..

Soltis D. E. Hibsch-Jetter C. Soltis P. S. Chase M. Farris J. S.. 1997. Molecular phylogenetic relationships among angiosperms: an overview based on rbcL and 18S rDNA sequences. In K. Iwatsuki, P. H. Raven [eds.], Evolution and diversification of land plants. Springer, Tokyo, Japan..

Soltis D. E. Kuzoff R. K.. 1995. Discordance between nuclear and chloroplast phylogenies in the Heuchera group (Saxifragaceae). Evolution 49: 727-742..[CrossRef][Web of Science]

Soltis D. E. Senters A. E. Zanis M. J. Kim S. Thompson J. D. Soltis P. S. Ronse De Craene L. P. Endress P. K. Farris J. S.. 2003. Gunnerales are sister to other core eudicots: implications for the evolution of pentamery. American Journal of Botany 90: 461-470..[Abstract/Free Full Text]

Stebbins G. L.. 1974. Flowering plants: evolution above the species level. Harvard University Press, Cambridge, Massachusetts, USA..

Suzuki Y. Glazko G. V. Nei M.. 2002. Overcredibility of molecular phylogenies obtained by Bayesian phylogenetics. Proceedings of the National Academy of Sciences, USA 99: 16138-16143..[Abstract/Free Full Text]

Swofford D. L.. 2000. Phylogenetic analysis using parsimony, version 4.0. Sinauer, Sunderland, Massachusetts, USA..

Swofford D. L. Olsen G. J. Waddell P. J. Hillis D. M.. 1996. Phylogenetic inference. In D. M. Hillis, C. Moritz, B. K. Mable [eds.], Molecular systematics, 407-514. Sinauer, Sunderland, Massachusetts, USA..

Takhtajan A.. 1969. Flowering plants: origin and dispersal. Smithsonian Institution Press, Washington, D.C., USA..

Thorne R. F.. 1968. Synopsis of a putatively phylogenetic classification of the flowering plants. Aliso 6: 57-66..

Wendel J. F. Doyle J. J.. 1998. Phylogenetic incongruence: window into genome history and molecular evolution. In P. Soltis, D. Soltis, J. Doyle [eds.], Molecular systematics of plants II, 265-296. Kluwer, Dordrecht, Netherlands..

Wendel J. F. Schnabel A. Seelanan T.. 1995. Bidirectional interlocus concerted evolution following allopolyploid speciation in cotton (Gossypium). Proceedings of the National Academy of Sciences, USA 92: 280-284..[Abstract/Free Full Text]

Wendel J. F. Stewart J. Mc D. Rettig J. H.. 1991. Molecular evidence for homoploid reticulate evolution among Australian species of Gossypium. Evolution 45: 694-711..[CrossRef][Web of Science]

White T. J. Bruns S. L. Taylor J.. 1990. Amplification and direct sequencing of fungal ribosomal RNA genes for phylogenetics. In M. A. Innis, D. H. Gelfand, J. J. Sninsky, T. J. White [eds.], PCR protocols: a guide to methods and applications, 315-322. Academic Press, San Diego, California, USA..

Wilcox T. P. Zwickl D. J. Heath T. A. Hillis D. M.. 2002. Phylogenetic relationships of the dwarf boas and a comparison of Bayesian and bootstrap measures of phylogenetic support. Molecular Phylogenetics and Evolution 25: 361-371..[CrossRef][Web of Science][Medline]

Yoder A. D. Irwin J. A. Payseur B. A.. 2001. Failure of the ILD to determine data combinability for slow loris phylogeny. Systematic Biology 50: 408-424..[Web of Science][Medline]

CiteULike    Complore    Connotea    Del.icio.us    Digg    Facebook    Reddit    Technorati    Twitter    What's this?

This Article Abstract Full Text (PDF) Supplementary Data Submit a response Alert me when this article is cited Alert me when eLetters are posted Alert me if a correction is posted Services Email this article to a friend Similar articles in this journal Similar articles in Web of Science Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via Google Scholar Google Scholar Articles by Moody, M. L. Articles by Les, D. H. Search for Related Content PubMed Articles by Moody, M. L. Articles by Les, D. H. Agricola Articles by Moody, M. L. Articles by Les, D. H. Social Bookmarking                            What's this?