sign in sign up
<<
Home , Altingiaceae, Aphanopetalum, Boykinia, Cercidiphyllum, Chrysosplenium, Corylopsis

TAXON 56 (1) • February 2007: 65–73 Soltis & al. • and relationships of

Monophyly and relationships of the enigmatic Peridiscaceae

Douglas E. Soltis1, Joshua W. Clayton1, Charles C. Davis2, Matthew A. Gitzendanner1, Martin Cheek3, Vincent Savolainen3, André M. Amorim4 & Pamela S. Soltis5

1 Department of Botany, University of Florida, Gainesville, Florida 32611, U.S.A. [email protected] (author for correspondence) 2 Harvard University Herbaria, Department of Organismic and Evolutionary Biology, Cambridge, Massachusetts 02138, U.S.A. 3 Royal Botanic Gardens, Kew, Richmond TW9 3DS, U.K. 4 Departamento de Ciências Biológicas, Universidade Estadual de Santa Cruz, Illhéus, 46.650-000, Bahia, Brazil 5 Florida Museum of Natural History, University of Florida, Gainesville, Florida 32611, U.S.A.

Peridiscaceae, comprising Peridiscus, , and Whittonia, are an enigmatic angiosperm family of uncertain composition and placement. Although some have placed Soyauxia in other families (e.g., , Medusandraceae), rather than in Peridiscaceae, sequence data for five (material of Whittonia could not be obtained) provide strong support for a of Soyauxia and Peridiscus. This evidence, combined with the strong morphological similarity of Peridiscus and Whittonia, support a monophyletic Peridiscaceae of three genera. Molecular analyses of a three- (rbcL, atpB, 18S rDNA) dataset for 569 taxa indicate that Peridiscus + Soyauxia together with Daphniphyllaceae form a clade that is sister to the rest of . Maximum likelihood and Bayesian analyses of Saxifragales using a five-gene (rbcL, atpB, matK, 18S rDNA, 26S rDNA) dataset place Peridiscaceae (posterior probability of 1.00) Peridiscaceae as sister to the remainder of Saxifragales, albeit without high posterior probability (pp = 0.78). Parsimony places a well-supported Peridiscaceae (100% bootstrap) as sister to Paeoniaceae within a paraphyletic , a placement that may be due to long-branch attraction. Following removal of Paeoniaceae from the dataset, parsimony place Peridiscaceae as sister to the remainder of Saxifragales. Although the placement of Peridiscaceae is not well supported in any analysis, molecular data suggest that Peridiscaceae do not have as their closest relatives , , Pterostemonaceae, , or , but instead are more closely related to woody members of Saxifragales (, Cercidiphyllaceae, Hamamelidaceae, and Daphniphyllaceae); several morphological features similarly suggest a relationship of Peridiscaceae to these woody families. The low support for the placement of Peridiscaceae is not surprising; previous analyses indicate that Saxifragales underwent a rapid, ancient radiation, and resolving relationships among members of the clade, particularly the grade of woody taxa, has been extremely difficult.

KEYWORDS: molecular systematics, Peridiscaceae, rapid radiation, Saxifragales

fragaceae. The circumscription of Saxifragales based on INTRODUCTION molecular data departs markedly from previous morphol- A brief history of Saxifragales. — The composition ogy-based classifications, which placed these families in of Saxifragales as revealed in recent analyses is one of the three different subclasses (e.g., Cronquist, 1981; Takhtajan major surprises of molecular phylogenetic studies of an- 1997; Morgan & Soltis, 1993; Qiu & al., 1998; reviewed giosperms (e.g., Chase & al., 1993; Morgan & Soltis, 1993; in Soltis & al., 2005): Altingiaceae, Hamamelidaceae, D. Soltis & al., 1997a, 2000; Qiu & al., 1998; Hoot & al., Cercidiphyllaceae, and Daphniphyllaceae were placed in 1999; P. Soltis & al., 1999; Savolainen & al., 2000a, b; re- Hamamelidae; Saxifragaceae, Iteaceae, Pterostemonace- viewed in D. Soltis & al., 2005). The (APG II, 2003) ae, Grossulariaceae, Crassulaceae, and Haloragaceae s.l. now includes Altingiaceae, Cercidiphyllaceae, - were treated in ; and Paeoniaceae were placed in ceae, Daphniphyllaceae, Grossulariaceae, Haloragaceae Magnoliidae or Dilleniidae (reviewed in Cronquist, 1981). s.l. (expanded to include Tetracarpaeaceae, Penthoraceae, Several members of Saxifragales were considered and [formerly of ]), Hama- closely related in some previous classifications: Saxi- melidaceae, Iteaceae, Paeoniaceae, Pterostemonaceae fragaceae, Grossulariaceae, Iteaceae, Pterostemonaceae, (sometimes included in Iteaceae; APG II, 2003), and Saxi- Penthoraceae, and Tetracarpaeaceae were considered

65 Soltis & al. • Monophyly and relationships of Peridiscaceae 56 (1) • February 2007: 65–73

part of a much more broadly defined Saxifragaceae s.l. (Brenan, 1953). Cronquist (1981) and Takhtajan (1997) (Engler, 1930; Morgan & Soltis, 1993), and a close rela- placed Whittonia and Peridiscus together in Peridisca- tionship of these families to Crassulaceae was also pro- ceae, and Soyauxia in Flacourtiaceae; Peridiscaceae and posed by Cronquist (1981) and Takhtajan (1987, 1997). Flacourtiaceae were considered closely related members Though not linked with these families, Hamamelida- of by both authors. ceae, Altingiaceae, Cercidiphyllaceae, and Daphniphyl- Based on rbcL, atpB, and 18S rDNA sequence data laceae have also been considered closely related (e.g., for Peridiscus, Davis & Chase (2004) suggested that the “lower hamamelids” Walker & Doyle, 1975). Peridiscaceae should also be included in Saxifragales. Haloragaceae and Paeoniaceae, however, have never However, the phylogenetic position of Peridiscus within been placed with any members of Saxifragales in any Saxifragales was unclear. Because only one of three gen- classifications prior to APG (1998) and APG II (2003). era of Peridiscaceae was sampled (Peridiscus) by Davis Haloragaceae had been placed in or near the rosid order & Chase (2004), both the monophyly of Peridiscaceae as Myrtales (Cronquist, 1981), whereas Paeoniaceae had well as the placement of the family within Saxifragales been considered closely related to Magnoliaceae (Sawa- require additional investigation. da, 1971), (Takhtajan, 1997), or Dillenia- In an effort to clarify the composition of Peridisca- ceae (Cronquist, 1981). The inclusion of Aphanopetalum ceae and relationships of the family within Saxifragales, in Saxifragales as part of an expanded Haloragaceae was we conducted a broad molecular phylogenetic analysis unexpected. Based on anatomical data Dickison & al. of the angiosperms, as well as focused analyses of Saxi- (1994) noted similarities between Aphanopetalum and fragales, adding sequences of four genes for two genera some of the traditionally recognized Saxifragaceae s.l. of Peridiscaceae (Peridiscus and Soyauxia; material of (e.g., Cronquist, 1981), including members of Saxifragoi- Whittonia could not be obtained) to existing multigene deae, a group that is now part of Saxifragales. datasets. Although early, broad analyses of rbcL sequences first revealed the Saxifragales clade (Chase & al., 1993; Morgan & Soltis, 1993; Chase & Albert, 1998), they did not provide bootstrap support (BS) greater than 50% for MATERIALS AND METHODS the clade. Investigation of atpB alone, as well as atpB DNA amplification and sequencing. — We iso- + rbcL, also indicated a monophyletic Saxifragales, but lated DNA following the general method of Doyle & neither of these studies provided BS > 50% (Savolainen Doyle (1987). A 750-base pair (bp) section of matK for & al., 2000a, b). 18S rDNA alone and 18S rDNA + rbcL Soyauxia was amplified using trnK-3914F and 1470R ultimately revealed a Saxifragales clade with moderate (Johnson & Soltis, 1994); the sequencing primers used jackknife support (e.g., D. Soltis & al., 1997b; D. Soltis were 710F, 934F, and 1470R (Johnson & Soltis, 1994, & Soltis, 1997), and the clade received strong support 1995). PCR products were sequenced in both directions in analyses of a dataset based on plastid rbcL and atpB using Beckman-Coulter Dye Terminator Cycle Sequenc- and nuclear 18S rDNA (Hoot & al., 1999; D. Soltis & al., ing Quick-Start kits and a Beckman-Coulter automat- 2000). ed sequencer following the manufacturer’s protocols Enigmatic Peridiscaceae. — Based on anatomical (Beckman Coulter, Inc., Fullerton, CA, U.S.A.). A 1300- characters, some (Sandwith, 1962) considered Peridis- bp region of matK was sequenced for Peridiscus using caceae to comprise three genera: Peridiscus (1 ), the primers matK-400F, matK-842F, and matK-1390R of Soyauxia (9 species, including 3 undescribed species Cameron & al. (2001), which approximately corresponds – Cheek, unpubl. data), and Whittonia (1 species). The to the 3’ region used by Hilu & al. (2003). rbcL and atpB three genera have a highly disjunct distribution; Peri- were amplified and sequenced following Savolainen & discus and Whittonia occur in northern , al. (2000a) except that sequencing reactions were run on whereas Souyaxia is native to Africa. However, other an Applied Biosystems 3730 DNA Analyzer. The 18S authors have not considered these three genera to be rRNA gene was amplified with primers 20F and 1324R members of the same family. Whereas Whittonia and following the general methods of D. Soltis & al. (1997a). Peridiscus have been considered closely related based Primers 20F, 1324R, and 461R (5’ TTT GCG CGC CTG on numerous shared vegetative, anatomical, and re- CTG CCT TCC 3’; Liz Caddick, pers. comm.) were used productive characters (e.g., Sandwith, 1962; Metcalfe, for sequencing. A portion (~750 bp) of 26S rDNA was 1962; Takhtajan, 1997), the relationships of Soyauxia amplified for Peridiscus in four separate pieces using have been problematic. Souyaxia has variously been primers N-nc26S1 to 950rev, N-nc26S5 to 1839rev, N- considered part of (Oliver in Bentham nc26S9 to 2782rev, and N-nc26S11 to 3331rev following & Hooker, 1862–1883; reviewed in Brenan, 1953), Fla- Kuzoff & al. (1998); these primers were also used for courtiaceae (e.g., Sandwith, 1962), or Medusandraceae sequencing. A small portion of Soyauxia was amplified

66 TAXON 56 (1) • February 2007: 65–73 Soltis & al. • Monophyly and relationships of Peridiscaceae

and sequenced using N-nc26S9 and 2426rev primers. separately. Two independent analyses ran for 5 million 18S rDNA and 26S rDNA sequencing reactions were run generations using four Markov chains sampled every on an Applied Biosystems 3730 DNA Analyzer. 100th generation. The first 100,000 generations were dis- Alignment and phylogenetic analyses. — Se- carded as burn-in; other parameters were left at default quences were easily aligned visually by adding them to values. Posterior probabilities were calculated using the the existing three-gene, 567-taxon matrix (P. Soltis & al., resulting trees from both runs. 1999; D. Soltis & al., 2000) and to the five-gene, 40-tax- ML analyses were implemented in PAUP* using the on Saxifragales matrix (Fishbein & al., 2001; Fishbein GTR+I+G model block generated by MODELTEST. The & Soltis, 2004). Short regions at the beginning and ends analysis used a heuristic search with 10 random addition of genes for which sequences were incomplete, as well replicates, TBR branch swapping, and the MulTrees op- as several short regions of 26S rDNA that were difficult tion in effect. to align (see Kuzoff & al., 1998) were excluded from the Trees and datasets have been deposited in TreeBase. analysis. Maximum parsimony (MP) analyses were conduct- ed using PAUP* 4.0 (Swofford, 2000). MP analyses of the 569-taxon dataset began with 10,000 random taxon RESULTS addition replicates with no branch swapping, with one We obtained 750 bp and 1300 bp of matK for Soy- saved per replicate. These 10,000 starting trees were auxia talbotii and Peridiscus lucidus, respectively, and then used in a second round of heuristic searches with 548 bp and 2930 bp of 26S rDNA for Soyauxia and Peri- tree-bisection-reconnection (TBR) branch swapping, discus, respectively. Nearly complete sequences of rbcL, MulTrees in effect, holding no more than 10,000 most atpB, and 18S rDNA were obtained for Soyauxia. Se- parsimonious trees. Bootstrap analysis was not conduct- quences are deposited in GenBank (Appendix). ed given the large size of this dataset. Our three-gene analysis using the 567-taxon data- Parsimony analysis of the five-gene Saxifragales da- set (P. Soltis & al., 1999: D. Soltis & al., 2000) with se- taset used heuristic searches of 1,000 random addition quences added for Peridiscus and Soyauxia confirmed replicates with no more than 10 trees saved per replicate, the placement of Peridiscaceae within Saxifragales (tree TBR swapping, and MulTrees in effect. Bootstrap sup- not shown). In this analysis, Peridiscus and Soyauxia port (BS; Felsenstein, 1985) was estimated from 1,000 formed a clade and appeared as sister to Daphniphyl- bootstrap replicates using 10 random taxon additions per lum. This clade of Peridiscaceae + Daphniphyllaceae ap- replicate, TBR branch swapping, MulTrees in effect, and peared as sister to the remainder of Saxifragales. saving all trees. Several different analyses of the Saxi- The parsimony analysis of the Saxifragales dataset fragales dataset were conducted. In our primary analy- based on all five genes resulted in a single shortest tree sis, we included all five genes. However, when initial and revealed a strongly supported Saxifragales clade results (below) suggested the possibility of long-branch (99% BS) consisting of two main subclades (Fig. 1). attraction with the addition of 26S rDNA data, Peridis- In the shortest tree obtained, a clade of Peridiscus and caceae and Paeoniaceae were alternately excluded from Soyauxia is well supported (100% BS); Peridiscaceae, the five-gene dataset. We also conducted a parsimony in turn, receive support (84% BS) as sister to a well- analysis in which we excluded 26S rDNA. supported (100% BS) Paeoniaceae. Peridiscaceae plus We also conducted Bayesian and maximum likeli- Paeoniaceae are nested within a paraphyletic Hamameli- hood (ML) analyses using the five-gene dataset for Saxi- daceae, albeit without support greater than 50%. Peridis- fragales: one analysis was conducted with all five genes, caceae, Paeoniaceae, and Hamamelidaceae appear in a and a second analysis was conducted with 26S rDNA weakly supported (55% BS) clade with Altingiaceae and sequences excluded (as above, for parsimony). MODEL- Daphniphyllaceae (Fig. 1). Cercidiphyllaceae are sister TEST v3.7 (Posada & Crandall, 1998) was used to select to this clade, with Altingiaceae plus Daphniphyllaceae the best evolutionary model for the Bayesian and ML sister to the clade of Hamamelidaceae, Peridiscaceae, analyses of Saxifragales. Both the hierarchical likelihood and Paeoniaceae. ratio tests (hLRTs) and Akaike information criterion The second main clade is weakly supported (68% (AIC) selected GTR+I+G as the best-fit model. Bayesian BS) and comprises Crassulaceae, Grossulariaceae, analyses were conducted with the MPI-enabled version Haloragaceae, Iteaceae, Pterostemonaceae, and Saxi- of MrBayes 3.1.1 (Huelsenbeck & Ronquist, 2001; Ron- fragaceae. Within this clade, each family is strongly quist & Huelsenbeck, 2003; Altekar & al., 2004). The supported as monophyletic. A well-supported (99% BS) and nuclear genes were designated as sepa- clade of Crassulaceae + Haloragaceae is sister to a well- rate partitions, and all model parameters were unlinked, supported (99% BS) “Saxifragaceae alliance” (Fishbein allowing each partition to optimize parameter values & al., 2001; D. Soltis & al., 2005) of Grossulariaceae,

67 Soltis & al. • Monophyly and relationships of Peridiscaceae TAXON 56 (1) • February 2007: 65–73

Iteaceae, Pterostemonaceae, and Saxifragaceae. Within discaceae are sister to the remainder of Saxifragales (tree the Saxifragaceae alliance, Pterostemonaceae + Iteaceae not shown), albeit without support greater than 50%. In (100% BS) are sister to a clade (99% BS) of (the a second analysis, Paeoniaceae remained in the dataset, sole member of Grossulariaceae) + Saxifragaceae. These and we excluded Peridiscaceae. In this analysis, we re- results are consistent with those of the five-gene analysis covered a single tree that resembled the five-gene tree of of Fishbein & al. (2001), without the inclusion of Peri- Fishbein & al. (2001), with a clade of Cercidiphyllaceae, discaceae. Altingiaceae, Daphniphyllaceae, Hamamelidaceae, and The branches to Paeoniaceae and Peridiscaceae are Paeoniaceae. each very long relative to most other branches in the par- Given that only partial 26S rDNA sequences were simony tree (Fig. 1). We therefore conducted additional obtained for Peridiscaceae (particularly Soyauxia), we analyses to explore the possibility that long-branch attrac- also explored the impact of the 26S rDNA sequence data. tion (Felsenstein, 1978) may have affected the parsimony Parsimony analysis of a dataset from which 26S rDNA topology. In one analysis, we removed Paeoniaceae from sequences were excluded resulted in a topology that dif- the five-gene dataset; in the shortest tree obtained, Peri- fered from that described above. Saxifragales are again

120 11 Altingia ALTINGIACEAE 9 113 143 DAPHNIPHYLLACEAE 100 38 29 33 25 20 100 Hamamelis 55 95 61 16 78 52 HAMAMELIDACEAE 38 21 17 57 98 17 47 11 12 100 16 288 26 PAEONIACEAE 100 8 Paeonia suffruticosa 65 23 51 Paeonia tenuifolia 84 27 Peridiscus 209 PERIDISCACEAE 100 43 Soyauxia 69 CERCIDIPHYLLACEAE 128 Aphanopetalum 79 73 177 5 68 100 100 55 100 HALORAGACEAE 95 89 53 89 85 178 99 Tetracarpaea 258 Crassula 205 100 87 100 CRASSULACEAE 78 100 136 100 92 50 51 68 100 41 68 165 76 46 91 77 100 43 35 157 SAXIFRAGACEAE 28 56 29 207 99 76 45 36 149 Ribes GROSSULARIACEAE 99 116 Choristylis 61 ITEACEAE 46 100 70 100 88 Pterostemon PTEROSTEMONACEAE 235 75 Lambertia 99 140 100 Platanus 100 52 110 Tetracentron 55 OUTGROUPS 100 Trochodendron 85 107 Leea 100 92 Vitis 50 changes Fig. 1. Shortest tree resulting from parsimony analysis of a five-gene dataset for Saxifragales (length = 6157; consistency index [CI] = 0.521; retention index [RI] = 0.616). Numbers above branches indicate branch lengths; numbers below branches are bootstrap values. Arrows place bootstrap values on short branches.

68 TAXON 56 (1) • February 2007: 65–73 Soltis & al. • Monophyly and relationships of Peridiscaceae

well supported (BS 100%) as a clade, but a well-sup- two . One of these clades (pp = 0.82) consists of ported Peridiscaceae (BS 100%) appear as sister to the a trichotomy of Cercidiphyllaceae + Daphniphyllaceae remainder of Saxifragales, albeit without BS greater than (pp = 0.75), Altingiaceae (pp = 1.0), and Hamamelida- 50% (tree not shown). Daphniphyllaceae now appear as ceae (pp = 1.0). The second clade (pp = 0.88) consists of part of a pentachotomy that also includes: (1) Hamame- Paeoniaceae as sister to a clade of the remaining Saxi- lidaceae, (2) Cercidiphyllaceae, (3) Altingiaceae, and (4) fragales (pp = 0.92), which comprise a clade (pp = 1.0) a large clade of Grossulariaceae, Saxifragaceae, Iteace- of Crassulaceae + Haloragaceae as sister to a clade (pp = ae, Pterostemonaceae, Paeoniaceae, Crassulaceae, and 1.0) of (Iteaceae + Pterostemonaceae; pp = 1.0) + (Gros- Haloragaceae. sulariaceae + Saxifragaceae; pp = 1.0) (Fig. 2). In the Bayesian analysis of the five-gene dataset Maximum likelihood resulted in a topology (tree (Fig. 2), Peridiscaceae, represented by Peridiscus and not shown; –lnL = 46,746.02229) identical to the Bayes- Soyauxia, are again sister taxa (posterior probability ian tree. [pp] = 1.0). Peridiscaceae are sister to the remainder of When 26S rDNA sequence data were removed, Saxifragales. The remaining members of the order form Bayesian and ML analyses placed a clade (pp = 0.68) of

1.00 Altingia ALTINGIACEAE 0.75 Liquidambar Cercidiphyllum CERCIDIPHYLLACEAE Daphniphyllum DAPHNIPHYLLACEAE 0.82 1.00 Corylopsis 1.00 Hamamelis 1.00 Disanthus HAMAMELIDACEAE Mytilaria Exbucklandia 1.00 Rhodoleia 1.00 Aphanopetalum 1.00 1.00 Haloragis 1.00 Myriophyllum HALORAGACEAE Penthorum 1.00 1.00 Tetracarpaea 0.78 Crassula 1.00 1.00 Dudleya CRASSULACEAE 1.00 Sedum Kalanchoe 0.92 1.00 Boykinia Sullivantia

1.00 0.66 Chrysosplenium 1.00 Peltoboykinia SAXIFRAGACEAE 1.00 Saxifraga 1.00 1.00 Heuchera 1.00 Micranthes Ribes 1.00 GROSSULARIACEAE 0.88 Choristylis 1.00 ITEACEAE Itea 1.00 Pterostemon PTEROSTEMONACEAE Paeonia brownii 1.00 Paeonia californica PAEONIACEAE 1.00 Paeonia tenuifolia Paeonia suffruticosa 1.00 Peridiscus Soyauxia PERIDISCACEAE 1.00 Lambertia 1.00 Platanus 1.00 Tetracentron Trochodendron OUTGROUPS Leea Vitis 0.005 substitutions/site Fig. 2. Majority rule consensus showing the posterior probabilities (above branches) based on Bayesian analysis of a five- gene dataset for Saxifragales (see Methods for details). Arrows place pp values on short branches.

69 Soltis & al. • Monophyly and relationships of Peridiscaceae TAXON 56 (1) • February 2007: 65–73

Peridiscaceae (pp = 1.0) + Daphniphyllaceae as sister to niaceae from our five-gene dataset, parsimony places the remainder of the order. Peridiscaceae + Daphniphyl- Peridiscaceae as sister to the remainder of Saxifragales. laceae were subsequently followed by a grade of Hama- When Peridiscaceae are removed, Paeoniaceae are sister melidaceae, Altingiaceae, and Cercidiphyllaceae as sister to Daphniphyllaceae (and part of a clade that includes to a clade of Paeoniaceae plus the remaining Saxifragales Altingiaceae, Hamamelidaceae, and Cercidiphyllaceae). (Crassulaceae, Haloragaceae, Iteaceae, Pterostemonace- This is the same topology recovered in the five-gene ae, Grossulariaceae, and Saxifragaceae). Relationships analysis of Fishbein & al. (2001); these results further among members of the Hamamelidaceae, Altingiaceae, support our hypothesis that the parsimony placement of and Cercidiphyllaceae grade have pp values below 0.80. Peridiscaceae and Paeoniaceae in our five-gene analysis In contrast, relationships among Crassulaceae, Halor- is the result of long-branch attraction. Removal of 26S agaceae, Iteaceae, Pterostemonaceae, Grossulariaceae, rDNA sequence data provides additional support for a and Saxifragaceae were again well supported (pp = 1.0) basal placement of Peridiscaceae; when these data are and identical to those described above. removed, the parsimony tree is like the Bayesian and ML trees with Peridiscaceae sister to the rest of Saxifragales. However, none of the interfamilial relationships noted receives BS greater than 50% or a high posterior DISCUSSION probability. Although our analyses clearly indicate that Our phylogenetic analyses support the hypothesis a monophyletic Peridiscaceae are part of Saxifragales, (Sandwith, 1962) that Peridiscaceae comprise Whitto- and typically place Peridiscaceae as sister (either alone nia, Peridiscus, and Souyaxia. Whereas Whittonia and or with Daphniphyllaceae) to the rest of Saxifragales, Peridiscus are morphologically similar and have been molecular data have so far not identified a well-sup- considered closely related (e.g., Metcalfe, 1962), the rela- ported placement of the family. Perhaps one notewor- tionships of Soyauxia have been problematic. For exam- thy result is that Peridiscaceae are not part of the large ple, Cronquist (1981) and Takhtajan (1997) both placed clade of Crassulaceae, Haloragaceae, Iteaceae, Pteroste- Whittonia and Peridiscus together in Peridiscaceae, but monaceae, Grossulariaceae, Saxifragaceae. Instead our Soyauxia was placed in Flacourtiaceae. Molecular data data suggest that Peridiscaceae may be part of a clade firmly place Souyaxia with Peridiscus and, combined or grade of woody taxa that includes Cercidiphyllaceae, with the strong morphological similarity of Peridiscus Daphniphyllaceae, Hamamelidaceae, and Altingiaceae. and Whittonia, support a monophyletic Peridiscaceae of Relationships among the woody families of Saxifragales, three genera. Several distinctive morphological features including Peridiscaceae, remain largely unresolved, a are also shared by all three genera of Peridiscaceae. For result that may be attributable to an early, rapid radia- example, the endosperm walls are massively thickened tion in the clade (Fishbein & al., 2001). As also noted by in genera of Peridiscaceae (Sandwith, 1962; Stevens, Fishbein & al. (2001), it may require a large number of 2004). Genera of the family also have a distinctive peti- additional base pairs to resolve deep-level phylogenetic olar anatomy (Metcalfe, 1962; Stevens, 2004). relationships in Saxifragales. Broad analysis of a three-gene, 569-taxon dataset Are there morphological characters that unite mem- for angiosperms (D. Soltis & al., 2000) confirms the bers of Peridiscaceae with other Saxifragales? As re- placement of Peridiscaceae within Saxifragales (Davis viewed elsewhere (e.g., D. Soltis & al., 2005), the mor- & Chase, 2004); Peridiscaceae appear as sister to Daph- phological diversity encompassed by Saxifragales is niphyllaceae; this clade is, in turn, sister to the remainder large, as seen in the comparison of features in Table 1. of Saxifragales. As a result, it is hard to identify features that are syn- Five-gene analyses of Saxifragales suggest that Peri- apomorphies for the clade. However, the woody mem- discaceae may be sister to the remainder of Saxifragales. bers do share a suite of anatomical characters (Table 1). Bayesian and ML searches of the five-gene dataset place Daphniphyllaceae, Hamamelidaceae, Altingiaceae, and Peridiscaceae as sister to the remainder of the clade (Fig. Cercidiphyllaceae all possess vessels with scalariform 2). Parsimony analyses appear to be affected by long- perforation plates, imperforate tracheary elements with branch attraction (Fig. 1), with Peridiscaceae and Paeoni- distinctly bordered pits, and parenchyma that is aceae (both on very long branches) forming a clade nest- apotracheal and diffuse (Baas & al., 2000) (Table 1). ed within Hamamelidaceae. Long-branch attraction may Peridiscaceae, although poorly studied, also exhibit have similarly affected the parsimony analysis of Saxi- these same anatomical features (Table 1) (Metcalfe, fragales by Davis & Chase (2004). Their analysis, based 1962; see Cronquist, 1981; Stevens, 2004). In contrast, on rbcL, atpB, and 18S rDNA, also placed Peridiscaceae these features either do not occur, or occur rarely in oth- (represented in that study only by Peridiscus) as sister er Saxifragales (represented by Crassulaceae in Table to Paeoniaceae. Significantly, when we removed Paeo- 1), which are primarily herbaceous (e.g., Crassulaceae,

70 TAXON 56 (1) • February 2007: 65–73 Soltis & al. • Monophyly and relationships of Peridiscaceae

Saxifragaceae), or (e.g., Iteaceae, Grossulariace- Saxifragales, including Cercidiphyllaceae, Daphniphyl- ae). For example, whereas Daphniphyllaceae, Hamame- laceae, and Hamamelidaceae (Table 1). lidaceae, Altingiaceae, and Cercidiphyllaceae have ves- A possible feature uniting Peridiscaceae with other sels with scalariform perforation plates, other families Saxifragales may be position and gynoecial devel- of Saxifragales typically have simple perforation plates opment. Typically, ovary positions have been described (Aphanopetalum also has scalariform perforation plates; as superior versus inferior, and decisions concerning both simple and scalariform perforation plates have been structural homology made exclusively on the basis of reported in Paeoniaceae). anthetic (i.e., mature) floral structure. Ovary position Other morphological features are also shared by varies in Peridiscaceae. The of Peridiscus woody Saxifragales. Peridiscaceae, Altingiaceae, Cer- is half-inferior, described as half-sunken into the disk cidiphyllaceae, Daphniphyllaceae, and Hamamelidaceae (Sandwith, 1962), whereas of Whittonia and all possess pendulous (Table 1), a feature found Soyauxia are described as hypogynous (Brenan, 1953). infrequently elsewhere in Saxifragales (e.g., Haloraga- This degree of variation in ovary position among related ceae). Peridiscaceae also exhibit a tanniniferous layer of genera is also seen in other Saxifragales. For example, cells in the tegmen, a feature also found in other woody in Saxifragaceae, the complete range of ovary positions

Table 1. Morphological comparison of Peridiscaceae with some representative families of Saxifragales (modified from Davis & Chase, 2004; characters taken from Cronquist, 1981; Baas & al., 2000; D. Soltis & al., 2005; Stevens, 2004). Character Peridiscaceae Daphniphyl- Hamamelida- Altingiaceae Cercidiphyl- Crassulaceae laceae ceae laceae Primary veins palmate pinnate palmate palmate palmate or pin- palmate nipalmate Stomates anomocytic paracytic various paracytic anomocytic anisocytic Stipules present absent present present present present Scalariform per- present present mostly present present present absent foration plates Tracheary ele- imperforate with imperforate with imperforate with imperforate with imperforate with imperforate with ments bordered pits bordered pits bordered pits bordered pits bordered pits simple pits Wood paren- apotracheal, apotracheal, apotracheal, apotracheal, apotracheal, absent; chyma diffuse diffuse diffuse diffuse diffuse herbaceous axillary fascicles/ axillary spikes (seldom ; globose terminal on cymose racemes racemose or head sympodial short paniculate) shoots absent absent usually present absent absent present Floral subtending present present absent present (carpel- present absent late flowers) 4–7, imbricate 2–6 (rarely ab- 4–5(–10), imbri- absent absent usually 5 sent), imbricate cate numerous, anther 5–6, 4–5(–10), anther 4–10, dehiscence 4–5(–10), anther usually 10, dehis- dehiscence by by slits dehiscence by by slits dehiscence by cence by slits slits slits/valves slits/valves Carpel number 3–4 2(–4) 2 (3) 2 1 3–10

Ovary position half inferior superior half to fully half inferior does not apply in superior inferior that there is no Style length short short long short long long

Ovules/placenta- pendulous from pendulous from pendulous from pendulous from in two rows parietal tion top of the ovary top of the ovary top of the ovary top of the ovary one seeded one seeded drupe capsular capsular follicle

Tanniniferous present present present absent present absent layer of cells in the tegmen Endosperm abundant, walls abundant, oily scanty, oily and scanty scanty, oily copious to scanty, massively thick- and protein- proteinaceous oily and protein- ened aceous aceous Embryo small small, straight large, straight long large, spatulate long

71 Soltis & al. • Monophyly and relationships of Peridiscaceae TAXON 56 (1) • February 2007: 65–73

from what has been described as superior to completely Baas, P., Wheeler, E. & Chase, M.W. 2000. Dicotyledonous inferior, as well as a range of intermediates, has been wood anatomy and the APG system of angiosperm and described within and Saxifraga (Kuzoff classification. Bot. J. Linn. Soc. 134: 3–17. & al., 2001; D. Soltis & Hufford, 2002). Studies of flo- Bentham, G. & Hooker, J.D. 1862–1883. Genera Plantarum. Reeve, London. ral development in Saxifragaceae and other Saxifragales Brenan, J.P.M. 1953. Soyauxia, a second of Medusan- indicated that species reported to have superior ovaries draceae. Kew Bull. 4: 507–511. actually have an appendicular epigynous ground plan, Cameron, K.M., Chase, M.W., Anderson, W.R. & Hills, a developmental program that leads to the formation of H.G. 2001. Molecular systematics of : inferior ovaries (e.g., Kuzoff & al., 2001: D. Soltis & Huf- evidence from plastid rbcL and matK sequences. Amer. J. ford, 2002). Thus, these ovaries are not truly superior but Bot. 88: 1847–1862. Chase, M.W. & Albert, V.A. 1998. A perspective on the con- represent “superior mimics” or “pseudosuperior” ovaries. tribution of plastid rbcL DNA sequences to angiosperm Appendicular epigyny typifies nearly all Saxifragales . Pp. 488–507 in: Soltis, D.E., Soltis, P.S. and is also plesiomorphic for the clade. Ovaries termed & Doyle, J.J. (eds.), Molecular Systematics of Plants II: “superior” in Iteaceae and Paeoniaceae are, in fact, pseu- DNA Sequencing. Kluwer, Boston. dosuperior (reviewed in D. Soltis & al., 2005; for floral Chase, M.W., Soltis, D.E., Olmstead, R., Morgan, G.D., Les, development in Paeoniaceae see also Hiepko, 1965), de- D.H., Mishler, B.D., Duvall, M.R., Price, R.A., Hills, rived via an appendicular epigynous ground plan and are H.G., Qiu, Y.-L., Kron, K.A., Rettig, J.H., Conti, E., Palmer, J.D., Manhart, J.R., Sytsma, K.J., Michaels, comparable to those observed in Saxifragaceae (D. Soltis H.J., Kress, W.J., Karol, K.G., Clark, W.D., Hedron, & al., 2005). In Saxifragaceae and other Saxifragales, M., Gaut, B.S., Jansen, R.K., Kim, K.-J., Wimpee, ovary position at anthesis is a consequence of the amount C.F., Smith, J.F., Furnier, G.R., Strauss, S.H., Xiang, of vertical extension of the inferior vs. superior region Q.-Y., Plunkett, G.M., Soltis, P.S., Swensen, S.M., of the ovary (Kuzoff & al., 2001; D. Soltis & al., 2003). Williams, S.E., Gadek, P.A., Quinn, C.J., Egguiarte, This same flexibility or lability in ovary position devel- L.E., Golenberg, E., Learn, G.H., Jr., Graham, S.W., opment, shown to be responsible for generating variation Barrett, S.C.H., Dayanandan, S. & Albert, V.A. 1993. Phylogenetics of seed plants: an analysis of nucleotide se- in ovary position in Saxifragaceae, Iteaceae, and other quences from the plastid gene rbcL. Ann. Missouri Bot. Saxifragales may also be operating in Peridiscaceae. Gard. 80: 526–580. In conclusion, our data suggest a monophyletic Peri- Cronquist, A. 1981. An Integrated System of Classification of discaceae that includes the problematic Soyauxia; molec- Flowering Plants. Columbia University Press, New York. ular data suggest a placement of the enigmatic Peridisca- Davis, C.C. & Chase, M.W. 2004. are sister to ceae as sister to the remainder of Saxifragales. Resolving Malpighiaceae; Peridiscaceae belong to Saxifragales. the precise placement of the family with any measure of Amer. J. Bot. 91: 262–273. Dickison, W.C., Hils, M.H., Lucansky, T.W. & Stern, W.L. internal support will be a challenge, however, given that 1994. Comparative anatomy and systematics of woody all available data suggest that Saxifragales represent an Saxifragaceae Endl. Bot. J. Linn. Soc. 114: 167–182. ancient, rapid radiation (Fishbein & al., 2001). Doyle, J.J. & Doyle, J.L. 1987. A rapid DNA isolation pro- cedure from small quantities of fresh tissue. Phyto- chem. Bull. 19: 11–15. Engler, A. 1930. Saxifragaceae. Pp. 74–226 in: Engler, A. & ACKNOWLEDGMENTS Prantl, K. (eds.), Die natürlichen Pflanzenfamilien, 2nd ed., 18a. Engelmann, Leipzig. This research was supported in part by an Assembling the Felsenstein, J. 1978. Cases in which parsimony and compat- Tree of Life (AToL) grant EF-0431266 (NSF). We thank Martyn ibility methods will be positively misleading. Syst. Zool. Powell for his help with DNA sequencing. We thank two anony- 27: 401–410. mous reviewers for their helpful comments. Felsenstein, J. 1985. Confidence limits on phylogeny: an ap- proach using the bootstrap. Evolution 39: 783–791. Fishbein, M. & Soltis, D.E. 2004. Further resolution of the rapid radiation of Saxifragales (angiosperms, ) LITERATURE CITED supported by mixed-model Bayesian analysis. Syst. Bot. 29: 883–891. Altekar, G., Dwarkadas, S., Huelsenbeck, J.P., & Ron- Fishbein, M., Hibsch-Jetter, C., Soltis, D.E. & Hufford, L. quist, F. 2004. Parallel Metropolis-coupled Markov chain 2001. Phylogeny of Saxifragales (angiosperms, eudicots): Monte Carlo for Bayesian phylogenetic inference. Bioin- analysis of a rapid, ancient radiation. Syst. Biol. 50: 817–847. formatics 20: 407–415. Hiepko, P. 1965. Das zentrifugale Androeceum der Paeo- APG. 1998. An ordinal classification for the families of flow- niaceae. Ber. Deutsch. Bot. Ges. 77: 427–435. ering plants. Ann. Missouri Bot. Gard. 85: 531–553. Hilu, K.W., Borsch, T., Müller, K., Soltis, D.E., Soltis, P.S., APG II. 2003. An update of the Angiosperm Phylogeny Group Savolainen, V., Chase, M.W., Powell, M., Alice, L.A., classification for the orders and families of flowering Evans, R.C., Sauquet, H., Neinhuis, C., Slotta, T.A., plants: APG II. Bot. J. Linn. Soc. 141: 399–436. Rohwer, J.G. , Campbell, C.S. & Chatrou, L. 2003. An-

72 TAXON 56 (1) • February 2007: 65–73 Soltis & al. • Monophyly and relationships of Peridiscaceae

giosperm phylogeny based on matK sequence informa- Sawada, M. 1971. Floral vascularization of Paeonia japonica tion. Amer. J. Bot. 90: 1758–1776. with some consideration on systematic position of the Hoot, S.B., Magallón, S. & Crane, P.R. 1999. Phylogeny of Paeoniaceae. Bot. Mag. (Tokyo) 84: 51–60. basal eudicots based on three molecular data sets: atpB, Soltis, D.E., Fishbein, M. & Kuzoff, R.K. 2003. Re-evaluat- rbcL, and 18S nuclear ribosomal DNA sequences. Ann. ing the evolution of epigyny: data from phylogenetics and Missouri Bot. Gard. 86: 1–32. floral ontogeny. Int. J. Pl. Sci. 164: S251–S264. Huelsenbeck, J.P. & Ronquist, F. 2001. MRBAYES: Bayes- Soltis, D.E., Hibsch-Jetter, C., Soltis, P.S., Chase, M. & ian inference of phylogeny. Bioinformatics 17: 754–755. Farris, J.S. 1997b. Molecular phylogenetic relationships Johnson, L.A. & Soltis, D.E. 1994. matK DNA sequences among angiosperms: an overview based on rbcL and 18S and phylogenetic reconstruction in Saxifragaceae sensu rDNA sequences. Pp. 157–178 in: Iwatsuki, K. & Raven, stricto. Syst. Bot. 19: 143–156. P.H. (eds.), Evolution and Diversification of Land Plants. Johnson, L.A. & Soltis, D.E. 1995. Phylogenetic inference Springer, Tokyo. using matK sequences. Ann. Missouri Bot. Gard. 82: Soltis, D.E. & Hufford, L. 2002. Ovary development and 149–175. diversification in Saxifragaceae. Int. J. Sci. 163: Kuzoff, R.K., Hufford, L. & Soltis, D.E. 2001. Structural 277–293. homology and developmental transformations associated Soltis, D.E. & Soltis, P.S. 1997. Phylogenetic relationships with ovary diversification in Lithophragma (Saxifraga- among Saxifragaceae sensu lato: A comparison of topolo- ceae). Amer. J. Bot. 88: 196–205. gies based on 18S and rbcL sequences. Amer. J. Bot. 84: Kuzoff, R.K., Sweere, J.A., Soltis, D.E., Soltis, P.S. & Zim- 504–522. mer, E.A. 1998. The phylogenetic potential of entire 26S Soltis, D.E., Soltis, P.S., Chase, M.W., Mort, M., Albach, rDNA sequences in plants. Molec. Biol. Evol. 15: 251–263. D., Zanis, M., Savolainen, V., Hahn, W., Hoot, S., Fay, Metcalfe, C.R. 1962. Notes on the systematic anatomy of M., Axtell, M., Swensen, S., Nixon K. & Farris, J. Whittonia and Peridiscus. Kew Bull. 15: 472–474. 2000. Angiosperm phylogeny inferred from a combined Morgan, D.R. & Soltis, D.E. 1993. Phylogenetic relationships data set of 18S rDNA, rbcL and atpB sequences. Bot. J. among Saxifragaceae sensu lato based on rbcL sequence Linn. Soc. 133: 381–461. data. Ann. Missouri Bot. Gard. 80: 631–660. Soltis, D.E., Soltis, P.S., Endress, P.K. & Chase, M.W. Posada, D. & Crandall, K.A. 1998. Modeltest: testing the 2005. Phylogeny and Evolution of Angiosperms. Sinauer model of DNA substitution. Bioinformatics 14: 817–818. Associates, Sunderland. Qiu, Y.-L., Chase, M.W., Hoot, S.B., Conti, E., Crane, P.R., Soltis, D.E., Soltis, P.S., Nickrent, D.L., Johnson, L.A., Sytsma, K.J. & Parks, C.R. 1998. Phylogenetics of the Hahn, W.J., Hoot, S.B., Sweere, J.A., Kuzoff, R.K., Hamamelidae and their allies: parsimony analyses of nu- Kron, K.A., Chase, M., Swenson, S.M., Zimmer, E., cleotide sequences of the plastid gene rbcL. Int. J. Pl. Sci. Chaw, S.M., Gillespie, L.J., Kress, W.J. & Sytsma K. 159: 891–905. 1997a. Angiosperm phylogeny inferred from 18S ribosom- Ronquist, F. & Huelsenbeck, J.P. 2003. MRBAYES 3: al DNA sequences. Ann. Missouri Bot. Gard. 84: 1–49. Bayesian phylogenetic inference under mixed models. Soltis, P.S., Soltis, D.E. & Chase, M.W. 1999. Angiosperm Bioinformatics 19: 1572–1574. phylogeny inferred from multiple genes as a research tool Sandwith, N.Y. 1962. A new genus of Peridiscaceae. Kew for comparative biology. Nature 402: 402–404. Bull. 15: 467–471. Stevens, P.F. 2004. Angiosperm Phylogeny Website. Version Savolainen, V., Chase, M., Morton, C.M., Soltis, D.E., 5, May 2004. http://www.mobot.org/MOBOT/research/ Bayer, C., Fay, M.F., De Bruijn, A., Sullivan, S. & APweb/ Qiu,Y.-L. 2000a. Phylogenetics of flowering plants based Swofford, D.L. 2000. PAUP*. Phylogenetic Analysis Using upon a combined analysis of plastid atpB and rbcL gene Parsimony (*and Other Methods), version 4. Sinauer As- sequences. Syst. Biol. 49: 306–362. sociates, Sunderland. Savolainen, V., Fay, M.F., Albach, D.C., Backlund, A., Van Takhtajan, A. 1987. System of Magnoliophyta. Academy of der Bank, M., Cameron, K.M., Johnson, S.A., Lledo, Sciences USSR, Leningrad. M.D., Pintaud, J.-C., Powell, M., Sheahan, M.C., Takhtajan, A. 1997. Diversity and Classification of - Soltis, D.E., Soltis, P.S., Weston, P., Whitten, W.M., ing Plants. Columbia University Press, New York. Wurdack, J. & Chase, M.W. 2000b. Phylogeny of the Walker, J.W. & Doyle, J.A. 1975. The basis of angiosperm eudicots: a nearly complete familial analysis based on phylogeny: palynology. Ann. Missouri Bot. Gard. 62: rbcL gene sequences. Kew Bull. 55: 257–309. 664–723.

Appendix. Species of Peridiscaceae analyzed, collection and voucher data. Species; collection; location of voucher; GenBank accession number. Peridiscus lucidus Benth.; Soares 205 (CEPEC); matK: DQ411570; 26S rDNA: DQ400571. Soyauxia talbotii Oliv.; Cheek 10617 (K); matK, 26S rDNA: DQ400572, DQ241372; 18S rDNA: AM111355; rbcL: AM111356; atpB: AM111257

73