EXTANT SEED PLANTS

Plant woody, evergreen; nicotinic acid metabolised to trigonelline, (cyanogenesis via tyrosine pathway); primary cell walls rich in xyloglucans and/or glucomannans, 25-30% pectin [Type I walls]; lignins rich in guaiacyl units; true roots present, apex multicellular, xylem exarch, branching endogenous; arbuscular mycorrhizae +; shoot apical meristem multicellular, interface specific plasmodesmatal network; stem with ectophloic eustele, endodermis 0, xylem endarch, branching exogenous; vascular tissue in t.s. discontinuous by interfascicular regions; vascular cambium + [xylem ("wood") differentiating internally, phloem externally]; wood homoxylous, tracheids +; tracheid/tracheid pits circular, bordered; sieve tube/cell plastids with starch grains; phloem fibers +; stem cork cambium superficial, root cork cambium deep seated; nodes ?; stomata ?; leaf vascular bundles collateral; leaves megaphyllous [determinancy evolved first, then ad/abaxial symmetry], spiral, simple, axillary buds +[?], prophylls [including bracteoles] two, lateral, veins -5 mm/mm² [mean for all non-angiosperms 1.8]; plant heterosporous, sporangia eusporangiate, on sporophylls, sporophylls aggregated in indeterminate cones/strobili; true pollen [microspores, i.e. no distal pore for release of gametes] +, grains mono[ana]sulcate, exine and intine homogeneous; ovules unitegmic, crassinucellate, megaspore tetrad tetrahedral, only one megaspore develops, megasporangium indehiscent; male gametophyte development first endo- then exosporic, tube developing from distal end of grain, to ca 2 mm from receptive surface to egg, gametes two, with cell walls, with many flagellae; female gametophyte endosporic, initially syncytial, walls then surrounding individual nuclei; seeds "large", first cell wall of zygote transverse, embryo straight, endoscopic [suspensor +], short-minute, with morphological dormancy, white, cotyledons 2; plastid transmission maternal; two copies of LEAFY gene, PHY gene duplication [N/O//A/C and P//BE lines], mitochondrial nad1 intron 2 and coxIIi3 intron present.

MAGNOLIOPHYTA

Plant woody, evergreen; lignans, O-methyl flavonols, dihydroflavonols, triterpenoid oleanane, non-hydrolysable tannins, quercetin and/or kaempferol +, apigenin and/or luteolin scattered, [cyanogenesis in ANITA grade?], lignins derived from both coniferyl and sinapyl alcohols, containing syringaldehyde [in positive Maüle reaction, syringyl:guaiacyl ratio less than 2-2.5:1], and hemicelluloses as xyloglucans; root apical meristem intermediate-open; root vascular tissue oligarch [di- to pentarch], lateral roots arise opposite or immediately to the side of [when diarch] xylem poles; origin of epidermis with no clear pattern [probably from inner layer of root cap], trichoblasts [differentiated root hair-forming cells] 0; stem with 2-layered tunica-corpus construction; wood fibers and wood parenchyma +; reaction wood ?, with gelatinous fibres; starch grains simple; primary cell wall mostly with pectic polysaccharides; tracheids +; sieve tubes eunucleate, with a sieve plate and cytoplasm with P-proteins, companion cells from same mother cell that gave rise to the sieve tube; nodes unilacunar [1:?]; stomata with ends of guard cells level with pore, paracytic, outer stomatal ledges producing vestibule; leaves with petiole and lamina [the latter formed from the primordial leaf apex], development of venation acropetal, 2ndary veins pinnate, fine venation reticulate, veins (1.7-)4.1(-5.7) mm/mm², endings free; flowers perfect, polysymmetric, parts spiral [esp. the A], free, development in general centripetal, numbers unstable; P not sharply differentiated, outer members not enclosing the rest of the bud, smaller than inner members; A many, with a single trace, introrse, filaments stout, anther ± embedded in the filament, tetrasporangiate, dithecal, with at least outer secondary parietal cells dividing, each theca dehiscing longitudinally by action of hypodermal endothecium, endothecial cells elongated at right angles to long axis of anther; tapetum glandular, binucleate; microspore mother cells in a block, microsporogenesis successive, walls developing by centripetal furrowing; pollen subspherical, binucleate at dispersal, trinucleate eventually, tectum continuous or microperforate, ektexine columellar, endexine thin, compact, lamellate only in the apertural regions; nectary 0; G free, several, ascidiate, with postgenital occlusion by secretion, few [?1] ovules/carpel, ovules marginal, anatropous, bitegmic, [outer integument often largely subdermal in origin, inner integument dermal], micropyle endostomal, integuments 2-3 cells thick, nucellus at apex of ovule 1-3 cells thick, megasporocyte single, megaspore lacking sporopollenin and cuticle, chalazal, female gametophyte four-celled [one-modular, nucleus of egg cell sister to one of the polar nuclei], stylulus short, hollow, cavity not lined by distinct epidermal layer, stigma ± decurrent, dry [not secretory]; P deciduous in fruit; seed exotestal; pollen germinating in less than 3 hours, siphonogamy, tube elongated, growing at 80-600 µm/hour, with callose plugs and callose-based walls, penetrating between cells, penetration of ovules within ca 18 hours, distance to first ovule 1.1.-2.1 mm; tube moves between nucellar cells, double fertilisation +, endosperm diploid, cellular [first division oblique, micropylar end initially with a single large cell, chalazal end more actively dividing], copious, oily and/or proteinaceous, embryo cellular ab initio, minute; germination hypogeal, seedlings/young plants sympodial; Arabidopsis-type telomeres [(TTTAGGG)_n]; whole genome duplication, single copy of LEAFY and RPB2 gene, knox genes extensively duplicated [A1-A4], AP1/FUL gene, paleo AP3 and PI genes [paralogous B-class genes] +, with "DEAER" motif, SEP3/LOFSEP and PHYA + C/PHYB + E gene pairs.

Evolution. Possible apomorphies for flowering plants are in bold. Note that the actual level to which many of these features, particularly the more cryptic ones, should be assigned is unclear, because some taxa basal to the [magnoliid + monocot + eudicot] group have been surprisingly little studied, there is considerable variation between families in particular for several of these characters, and also because details of relationships among gymnosperms will affect the level at which some of these characters are pegged. For example, if reticulate-perforate pollen is optimized to the next node on the tree (see Friis et al. 2009 for a discussion), it effectively makes the pollen morphology of the common ancestor of all angiosperms ambiguous... For other features such a a nucellus only one (Nymphaeales) to three cells thick above the embryo sac and a stylar canal lacking an epidermal layer, although plesiomorphous for basal grade angiosperms (Williams 2009), where on the tree a thicker nucellus and a stylar epidermal layer are acquired has not yet been indicated.

NYMPHAEALES [AUSTROBAILEYALES [[CHLORANTHALES + MAGNOLIIDS] [MONOCOTS [CERATOPHYLLALES + EUDICOTS]]]]: vessels + [one position], elements with elongated scalariform perforation plates; axial parenchyma diffuse or diffuse-in-aggregate; tectum reticulate-perforate [here?]; ?genome duplication; "DEAER" motif in AP3 and PI genes lost, gaps in these genes.

SAXIFRAGALES [VITALES + ROSIDS] = ROSANAE Takhtajan: ?  Back to Main Tree

Evolution. The age of the rosid clade has been estimated at 117-108 million years before present (Wikstr&m et al. 2003); fossils assignable to rosids are ca 94 million years (Crepet et al. 2004). The age of the rosids in Anderson et al. (2005) should be treated with caution, since members of both Vitales and Crossosomatales were included in their rosids - and there was also a separate Crossosomatales. However, Wang et al. (2009) and Magallón and Castillo (2009) provide estimates largely in agreement with those of (Wikstr&m et al. 2003) and Davies et al. (2004). The stem group may have diverged 114.4-115.1 million yeas ago (Magallón & Castillo 2009).

Much crown group diversification probably occured within a narrow time interval of 5-15 million years some time around 117-93 million years ago in the late Aptian early Turonian (Wang et al. 2009).

Butterfly caterpillars are common on members of the group, occuring about twice as frequently as might be expected going by species number alone, but the tree habit is also common in the group, and trees perhaps support a correspondingly disproportionate number of larvae... (Janz & Nylin 1998).

Chemistry, Morphology, etc. Smets et al. (2003) characterise the rosids as having receptacular nectaries; note that the Vitales have gynoecial nectaries and Proteaceae have receptacular nectaries (Smets 1988). A relatively large embryo over half the length of the seed may be a feature of this clade (Forbis et al. 2002), although there is considerable variation in Saxifragales and Vitales have small embryos. This character has been pegged at a higher level in the tree (core rosids).

The feature of cuticle wax platelets as rosettes is scattered through this group, but is especially common in Fabaceae and is known from several Malpighiales (see Ditsch & Barthlott 1997 for details). Distinctive mucilage cells in flowers with a much thickened mucilaginous inner periclinal wall and distinct cytoplasm are found especially in this broader group (note: not yet reported from Geraniales, see Matthews & Endress 2006b). The sepals characteristically have three traces from three gaps; in several more basal eudicot clades the outer perianth/sepals have only a single gap (von Balthazar and Endress 2002 - see also Ranunculales). Petal development is often retarded relative to that of other parts of the flower (also in Cabomba and Saruma!). For the genome duplication, see Tuskan et al. (2006); exactly where it is to be placed on the tree is unclear. Similarly, where the loss of the chloroplast infA gene is to be placed is unclear (see Millen et al. 2001).

Phylogeny. Thinking of major branchings within rosids, Hilu et al. (2003: matK analysis alone, although Schumacheria [one member of Dilleniaceae included in the study] was firmly associated with Ericales...) found a possible set of relationships [Rosids [[Dilleniacaeae + Vitaceae] [Saxifragales [Santalales [Berberidopsidales [Caryophyllales + Asterids]]]]]]. Nickrent et al. (2005) also found the position of Saxifragales to be particularly uncertain, although Vitales again tended to go with rosids. Recently, Soltis et al. (2007a) found a grouping [Saxifragales [Vitales + rosids]], both groupings with 1.0 p.p. (see also Zhu et al 2007, but little support). Earlier work had suggested similar relationships, thus Saxifragales were sister to the rest of the group (e.g. P. Soltis et al. 1999: Vitales not included), albeit with weak support. Studies on the duplication of the RPB2 gene and subsequent loss of one of the copies (Oxelman et al. 2004; Luo et al. 2007) suggest that Saxifragales and Rosids are linked by loss of the -I copy, which also has occured in Santalales and Caryophyllales, but not Vitales, Berberidopsidales, Gunnerales or Dilleniales, which have all lost the RPB2-D copy. The variation pattern is complex - and this is without worrying about what is going on within the asterids. Saxifragales and Vitales were successively sister to all eudicots minus Gunnerales (mitochondrial gene only), or to all rosids, but with little support (Zhu et al. 2007). Support for these or very similar hypotheses of relationships seems to becoming stronger. Saxifragales are well supported as sister to [Vitales + rosids] in the 12-gene plus plastid inverted repeat analyses of Wang et al. (2009). It does seem that Vitales may well be sister to the other rosids s. str. (e.g. Jansen et al. 2007), although support for this poasition was only weak in Wang et al. (2009). Palaeohexaploidy also seems to link Vitales with rosids, but the sampling in major core eudicot clades is very poor (Jaillon, Eury et al. 2007), and the implications of this paleohexaploidy in groups suspected of undergoing subsequent bouts of genome duplications in terms of gene copy numbers is unexplored (cf. Velasco et al. 2007, who interpret the data rather differently).

D. Soltis et al. (2003a) found 79% support for rosids sensu stricto, i.e., lacking Vitales and Saxifragales. Within rosids s. str., relationships have also been somewhat unclear (e.g. Soltis et al. 2005b; Jansen et al. 2006a; Bausher at al. 2006; Zhu et al. 2007; versions of this site up to March 2009), but the topology of the tree has recently been much clarified (Wang et al. 2009). The composition of the rosid I clade (= fabids) has been perhaps particularly problematical. In an analysis including the mitochondrial matR and two chloroplast genes, [Celastrales + Oxalidales + Malpighiales] were sister to the N-fixing clade, with weak to moderate support; Crossosomatales were weakly supported as sister to the rosid II clade, while Myrtales and Geraniales appeared to be successively sister to all other rosids - but with little support (Zhu et al. 2007). S.-B. Lee et al. (2006) found some support for the clade [Geraniales + Myrtales] sister to the rosid I clade, although sampling was poor. This [Myrtales + Geraniales] clade was sister to rosid II clade, albeit with weak support, but the combined clade had strong support (weaker using maximum parsimony) and in turn being strongly supported as sister to the [N-fixing clade + [Celastrales, Oxalidales, Malpighiales]] clade, albeit with sketchy sampling (Jansen et al. 2007) - i.e. the same basic set of relationships found in the comprehensive analysis of Wang et al. (2009). See the Dilleniales page for further discussion on the relationships of the rosids.

All things being considered, the circumscription of the rosids can usefully be expanded to include both Saxifragales and Vitales; they link with them phylogenetically (see above) and are morphologically quite similar. See the Dilleniales page for further discussion.

SAXIFRAGALES Dumortier  Main Tree, Synapomorphies.

Ellagic acid, myricetin, flavonols +; (tension wood +); branching from the previous flush [woody members]; cuticle waxes as clustered tubules; leaf teeth with gland broadening distally and with apical foramen, higher order lateral veins joining it; anthers basifixed, with basal pit, sagittate, ?pollen morphology, carpels free, at least apically, micropyle bistomal, (outer integument largely dermal in origin), styluli short, stigmas decurrent, at most slightly wet; fruit dry; seeds ± exotestal; endosperm type?, embryo size?; unique 1 BP [adenosine] insertion in 18S rDNA. - 15 families, 112 genera, 2470 species.

Evolution. Magallón and Castillo (2009) estimated ages of ca 114.2 and 114.8 million years (relaxed and constrained penalized likelihood datings respectively) for the stem group, while the crown group was 106.4 to 107 million years old. Wikström et al. (2001) had suggested that the age of the clade is some 121-111 my, with diversification within it starting 110-92 million years before present; comparable figures given by Anderson et al. (2005) are ca 108 and 102 million years before present respectively. Hermsen et al. (2006b: the topology quite resolved, but with little support and different in detail from that adopted here; see also below) suggested that much diversification occured rather later, 90-84 million years before present in the Late Cretaceous.

Saxifragales contain ca 1.3% of eudicot diversity. They have a very poor representation in the tropics in general and neotropics in particular, which makes the recent inclusion of the tropical Peridiscaceae (see below) the more notable. Apart from the Crassulaceae/Saxifragaceae clade, relationships within Saxifragales remained largely unresolved for some time, and it was suggested that they represented an ancient and rapid radiation (Fishbein et al. 2001; Fishbein & Soltis 2004). More recently Jian et al. (2006, esp. 2008) also estimated the age of the crown group Saxifragales as being rather younger, some 103-83 million years million years before present, with subsequent early diversification perhaps occuring over a period as short as 3-6 million years.

D. Soltis et al. (2007b) suggest possible apomorphies for the clade other than those given above.

Chemistry, Morphology, etc. Roots are diarch (Van Tieghem & Douliot 1888). Saxifragales commonly have scalariform perforation plates, lateral pitting that is mostly scalariform or opposite, bordered pits, etc., but whether these are synapomorphies is unclear (Jian et al. [2006] characterise the largely unresolved woody members and the Saxifraceae + Crassulaceae clades in terms of their wood anatomy). Leaf teeth are basically rosid, although those of Cercidiphyllum are described as being more or less chloranthoid (not very different), while Hamamelidaceae can have teeth with a clear, glandular apex (fothergilloid) and those of Altingiaceae are platanoid (basically, the higher order lateral veins do not quite make it to the tooth - Hickey & Wolfe 1975; Tetracarpaea is similar, also lacking such veins - Hils et al. 1988). Despite appearances, the floral apex in nearly all taxa studied is flat or concave (Fishbein et al. 2000; Soltis & Hufford 2002; D. Soltis et al. 2003b; Soltis et al. 2005b), although Wurdack and Davis (2009) suggest that this is not the case for Peridiscaceae.

For information on the hamamelids as it was beginning to be realised that they might have to be split, see Crane and Blackmore (1989). For pollen, see Hideux and Ferguson (1976), Zavada and Dilcher (1986); floral anatomy and morphology, Gaümann (1919), Bensel and Palser (1975a-d), Hufford and Endress (1989), Drinnan et al. (1995), Fishbein et al. (2000); for chemistry, Giannasi (1986), Jay (1971); for anatomy, Watari (1939), Ramamonjiarisoa (1980), Cutler and Gregory (1998); for seed coat, Krach (1976); and for general morphology, see Hermsen et al. (2006b).

Phylogeny. Saxifragales as circumscribed here are suggested in molecular phylogenies (e.g. D. Soltis et al. 1997; D. Soltis & P. Soltis 1997). In addition to sequence similarity, members have i.a. an insertion in 18S rDNA in common. Peridiscaceae are an unexpected addition to the clade. They were placed in Malpighiales by Savolainen et al. (2000a: see A.P.G. II 2003). However, Davis and Chase (2004; see also Soltis et al. 2007a) rather surprisingly found that the family was to be placed here, also adding Soyauxia, previously placed in Medusandraceae. More recently Wurdack and Davis (2009) have added Medusandra itself to Peridiscaceae.

Within Saxifragales relationships other than the Saxifragaceae/Crassulaceae clade were unclear for some time. Support for the Crassulaceae et al. and Saxifragaceae et al. clades, and also the relationships within them, is generally strong; support for the clade [Pterostemonaceae + Iteaceae] in the latter is also strong (e.g. Soltis et al. 2007a: see below). Although Hilu et al. (2003: matK) did not recover the Saxifragaceae/Crassulaceae clade, there was no strong support for an alternative placement; Cercidiphyllaceae and Daphniphyllaceae were sister taxa, with moderate jacknife support. Hermsen et al. (2006b), who included both molecular and morphological (the latter also from selected fossils) data, also recovered a [Crassulaceae et al. + Saxifragaceae et al.] clade, while all other families were in a clade sister to this. [Paeonia + Daphniphyllum] and [Cercidiphyllaceae + Altingiaceae] were clades in this latter, but with very weak support (<50% bootstrap); Peridiscaceae were not included. Paeonia was linked with moderate support to the Crassulaceae clade, or, more weakly, with the Crassulaceae + Saxifragaceae clades in some analyses in Fishbein et al. (2001); the latter relationship also appeared in a recent study (Fishbein & Soltis 2004: Peridiscaceae not included), but it was linked with low support to Peridiscus itself by Davis and Chase (2004). D. Soltis et al. (2007b) were also unable to recover stable relationships among the woody Saxifragales, long branch attraction (Paeoniaceae, Peridiscaceae) possibly occuring; depending on the analysis, a clade [Peridiscaceae + Paeoniaceae] made Hamamelidaceae paraphyletic, or Peridiscaceae were sister to all other Saxifragales in the study. Despite the addition of more data, Jian et al. (2006) still found it difficult to resolve relationships between the woody members, although it appeared that Peridiscaceae might be sister to the rest of the order, and Paeoniaceae sister to the Crassulaceae and Saxifragaceae clades.

However, Jian et al. (2008), using a variety of large data sets (some with over 50,000 bp) and analyeses, have been able to find strong maximum likelihood and Bayesiam support for the topology used here, although Paeonia in particular moved around the tree in some parsimony analyses. There is little morphological support for the basal branches of this topology, but that seems pretty much par for the course, although characters can be optimised to positions on a number of the shallower branches (cf. Hermsen et al. 2006b).

Some molecular analyses have placed the hitherto unplaced Cynomoriaceae in Saxifragales, perhaps in the Crassulaceae area, although with little support (Nickrent 2002; Nickrent et al. 2005), however, Barkman et al. (2007) found no support for a position in this order - or for any particular position at all. A position in Saxifragales was rejected by Jian et al. (2008), who prefered to place them in Santalales; Balanophoraceae, with which Cynomoriaceae were linked in the past, are also to be included in Santalales (see also Nickrent et al. 2005). Recently Cynomoriaceae have been placed in Rosales sister to Rosaceae based on analysis of chloroplast inverted repeat sequences (Moraceae were the only other family in the order examined: see also the Rosales page), and with strong support; Cynomoriaceae certainly were to be excluded from Saxifragales (good sampling) and several other rosid orders (Zhang et al. 2009). Further studies are needed and the inclusion of Cynomoriaceae here needs confirmation, however, they and a number of other Saxifragales have cellular endosperm while there seems to be little morphology linking Cynomoriaceae and Rosales.

Previous Relationships. Saxifragales includes Hamamelidaceae, a key group classically linking the Englerian Amentiferae (usually dioecious or monoecious woody plants with an ament, or catkin, with small flowers, and sometimes believed to be primitive), to "dicots" with more conventional flowers. However, the old Amentiferae, included in Cronquist's (1981) Hamamelidae, are now in several bits, mostly in the rosids, of which one is here - see also Fagales, which constitutes the major part of Amentiferae, Malpighiales (Salicaceae), Rosales ("Urticales"), etc. (Qiu et al. 1998a) - but also including Eupteleaceae (eudicot), Myrothamnaceae (core eudicot), Eucommiaceae (asterid I/lamiid), etc. A link between woody Saxifragales and Fagales has often been recognised (e.g. Frohne & Jensen 1992), although there seems little point in continuing to do this. The woody Saxifragaceae, which have been included in Saxifragaceae s.l., Hydrangeaceae, or Grossulariaceae, are spread widely through both rosids and asterids (e.g. Morgan & Soltis 1993), examples being Brexiaceae (Celastrales), Hydrangeaceae (Cornales), and Escalloniaceae (Escalloniales - asterid II/campanulid). Most iridoid-negative, herbaceous and/or crassinucellate members remain here. Ironically, three families of Saxifragales s. str. are indeed reliably reported to have iridoids (how many origins?), and are the only families outside asterids known to have these compounds. Daphniphyllanae, Saxifraganae and Hamamelidanae, in which Takhtajan (1997) placed most of the families that are in Saxifragales here, are all in his Hamamelididae.

Includes Altingiaceae, Aphanopetalaceae, Cercidiphyllaceae, Crassulaceae, Cynomoriaceae, Daphniphyllaceae, Grossulariaceae, Haloragaceae, Hamamelidaceae, Iteaceae, Paeoniaceae, Penthoraceae, Peridiscaceae, Saxifragaceae, Tetracarpaeaceae.

Synonymy: Hamamelidineae Thorne & Reveal - Altingiales Doweld, Cercidiphyllales Reveal, Crassulales Lindley, Cynomoriales Burnett, Daphniphyllales Cronquist, Fothergillales Link, Grossulariales Lindley, Haloragales Bromhead, Hamamelidales Grisebach, Iteales Doweld, Medusandrales Brenan

PERIDISCACEAE Kuhlmann, nom. cons.   Back to Saxifragales

Trees; plants Al accumulators; vessel elements with scalariform perforation plates; apotracheal (paratracheal, diffuse) parenchya +; secretory canals + [Medusandra]; petiole bundles annular with wing bundles [Soyauxia], also an adaxial plate [Peridiscus] or an adaxial [Whittonia] or medullary [Medusandra] annular bundle; crystals +; hairs unicellular, lignified [Medusandra]; epidermal wax crystals in rosettes; leaves two-ranked, (spiral, margin serrate), 2ndary veins palmate, petiole pulvinate [Peridiscus, Medusandra], stipule single, adaxial, or paired, lateral; inflorescences axillary, racemose(-spicate) or fasciculate, flowers small; P 4-7; A many, at most slightly connate basally, (anthers monothecal), disc large, annular, hairy or not, or K open, C +; A 5, alternating with K, 5 long hairy staminodes opposite C, disc 0 (Medusandra), G [3-4], ± sunken in disc or not, 1-locular, (placentation free central), with 6-8 pendulous apical epitropous ovules, stigmas punctate; fruit a drupe or capsule; P deciduous or K much enlarged, accrescent, recurved (Medusandra); seed 1, large, coat tanniniferous, walls thin, ± collapsed; endosperm ?development, copious, cell walls thick, pitted, embryo very small, cotyledons foliaceous; n = ?

4[list]/11. South America, tropical W. Africa (map: from Heywood 2007).

• Peridiscaceae are tropical trees with entire, stipulate leaves. The racemose inflorescences have small flowers, the gynoecium has apical ovules, and the styluli are short. The capsular fruit has a single, large seed with a dark-colored but not very hard coat; the embryo is very small and embedded in cartilagineous endosperm.

Chemistry, Morphology, etc. Peridiscaceae are a very poorly known and superficially heterogeneous group. The vascular pitting of Medusandra is scalariform, and the pits are bordered (Metcalfe 1952). Note that the leaves are almost certainly simple, not unifoliate (cf. Brenan 1952; Hutchinson 1973); the rather swollen apex of the petiole is like that of many Euphorbiaceae, Hydnocarpus, Octoknema, etc., which are not usually described as being possibly unifoliolate. Petiole anatomy in the region of the pulvinus is complex. For epidermal wax crystals, see Ditsch and Barthlott (1997). The bracteoles are often inconspicuous (cf. Cronquist 1981). I have not seen Whittonia, but Metcalfe (1962) gives some details of its anatomy.

Peridiscus and Whittonia have monothecal anthers, probably derived within the family. The flowers of Medusandra have long and conspicuous staminodes borne opposite the sepals, hence the generic name. The basic seed morphology/anatomy of Soyauxia and Peridiscus, from either side of the Atlantic, are almost identical, although the two are vegetatively very different (Peridiscus is sometimes even identified as Menispermaceae!).

See Metcalfe (1952b, 1962) and Miller (1975) for anatomy and Soltis et al. (2007b) for general information.

Previous Relationships. Peridiscaceae were included in Violales-Flacourtiaceae by Takhtajan (1997), and a similar position was suggested by A.P.G. III (2003); Soyauxia in particular has been associated with Medusandraceae or Flacourtiaceae (Malpighiales). Medusandra itself was tentatively included in Malpighiales by Soltis et al. (2005b), certainly, the serrate leaf blades and the cauline stipules suggest relationships other than to Santalales or Santalanae (cf. Cronquist 1981; Takhtajan 1997; Thorne 2007).

Synonym: Medusandraceae Brenan, nom. cons.

[Paeoniaceae [Altingiaceae [Hamamelidaceae [Cercidiphyllaceae + Daphniphyllaceae]]]] [[Crassulaceae [Aphanopetalaceae [Tetracarpaeaceae [Haloragaceae + Penthoraceae]]]] [[Iteaceae + Pterostemonaceae] [Grossulariaceae + Saxifragaceae]]]: floral apex flat-concave early in development [G often (semi-)inferior].

[Paeoniaceae [Altingiaceae [Hamamelidaceae [Cercidiphyllaceae + Daphniphyllaceae]]]]: mitochondrial coxII.i3 intron 0; leaves with basically palmate venation.

PAEONIACEAE Rafinesque, nom. cons.   Back to Saxifragales

Perennial herbs to shrublets; iridoids, ethereal oils, flavones +, myricetin and tannins 0; cork "subcortical" [Tiagi 1970]; stem with cortical vascular bundles; vessel elements with simple or scalariform perforation plates; nodes also 5:5; calcium oxalate as crystals; petiole bundles annular; wax tubules with palmitone predominating; indumentum 0 (hairs unicellular); leaves spiral, compound, ultimately ternate, ptyxis variable, margin toothed, base broad, stipules 0; flowers large, terminal, with cortical vascular system; P spiral, all 3:3, K (3-)5(-7), tough, C 5-8(-13), not sharply distinguished; A many, from 5 trunk bundles continuing spiral of C, centrifugal, basal pits?, pollen colporoid, nectary a lobed disc; G (2) 3-5(-15), usu. many ovules/carpel, outer integument 10-20 cells across, massive nucellar cap +, archesporium multicellular, nucellus absorbed before anthesis, stigma expanded, rather oblique, sessile, wet; fruit a follicle, K persistent; funicle fleshy, with apical rim-aril, testa fleshy, vascularised, exotestal cells palisade, variously thickened, the hypodermis palisade, ± lignified, (some mesotesta thickened); endosperm nuclear, with amyloid, chalazal endosperm haustorium +, embryo initially coenocytic, minute; n = 5, chromosomes 10-15 µm long; germination hypogeal; mitochondrial coxII.i3 intron 0.

1[list]/33. N. Temperate, especially East Asia (map: from Stern 1946; Hultén & Fries 1986). [Photo - Fruit]

• Paeoniaceae are perennial herbs or shrublets that are recognisable by their rather large, ± ternately compound, exstipulate leaves, and large flowers. The latter have sepals, petals, numerous centrifugal stamens, a prominent disc, and separate carpels. The fruit is a follicle and the seeds are fleshy.

Evolution. The testa is thick, fleshy and coloured, and in some species the colour contrasts with that of the testa of partly developed and unfertilised seeds (blackish/red) when the follicle opens. The funicle is also fleshy. For the evolution and biogeography of the genus, see Sang et al. (1997).

Chemistry, Morphology, etc. The disc apparently does not secrete nectar (Tamura 2006). Johri et al. (1992) called the micropyle exostomal, however, the inner integument, too, partly forms the micropyle. The funicle is also fleshy, and there may be a small rim aril at the apex. There is no tegmen.

For the perianth, see Hiepko (1964); for general information, see Tamura (2006); Tiagi (1970) and Takhtajan (1988) provide much information on ovules and seeds.

Previous Relationships. Paeoniales were included in Ranunculidae (Takhtajan 1997), and a relationship between Paeoniaceae and Ranunculaceae in particular has often been suggested (Takhtajan 1997; Mabberley [1997] included Glaucidium [see Ranunculaceae here] in Paeoniaceae) because of gross floral similarities between the two. However, they differ in the nature of the petals and nectaries, the development of the androecium, numerous embryological features, etc. (e.g. Tiagi 1970). Dilleniales, in which Paeoniaceae were placed by Cronquist (1981; see Corner 1946), have multistaminate and centrifugal androecia, but differ in gynoecial development, nectary morphology, etc.

[Altingiaceae [Hamamelidaceae [Cercidiphyllaceae + Daphniphyllaceae]]]: cuticle waxes as tubules, nonacosan-10-ol the main wax; inflorescence racemose, flowers sessile; anthers opening ± valvate, connectives apically protruding; embryo long.

Evolution. There are a number of reports of delayed fertilisation from members of all four families, although not in Paeonia (Sogo & Tobe 2006d for references); fertilization in Peridiscaceae is not known.

Chemistry, Morphology, etc. Raffinose and stachyose are common oligosaccharides in phloem exudate in this clade (Daphniphyllaceae not studied: Zimmermann & Ziegler 1975). Hufford and Endress (1989; see also Hersen et al. 2006b) discuss anather morphology and anatomy in detail; members of this clade can have obviously valvate anthers, as in Hamamelidoideae, or the stomium may simply divide at the two ends of the theca, or at least at the base of the theca.

ALTINGIACAEAE Horaninow, nom. cons.   Back to Saxifragales

Trees; resins, route I iridoids +; secretory canals [with resins] in stem; petiole bundle annular, with medullary bundles; stomata paracytic; bud scales +; leaves spiral, flat, lobes conduplicate, stipules on leaf base; plant monoecious, inflorescence ± capitate; P 0; staminate flowers: A 4-10, (anthers with longitudinal slits), pollen spherical, polyporate, fine-reticulate, pistillode +; carpellate flowers: intercarpellary protrusions + [= sterile flowers]; G [2], unsealed, (semi)inferior, (transverse), 20< straight apotropous ovules/carpel [only the lower ones fertile], (micropyle endostomal), stigmas with multicellular protrusions, but no papillae; fruit a septicidal (and loculicidal) capsule; seeds winged, exotesta lignified or not, mesotesta ± sclerotic, endotestal cells oblong, lignified, [or seed exotegmic]; endosperm slight, nuclear; n = 15, 16.

1[list]/13 (for generic limits, see Ickert-Bond & Wen 2006). E. Mediterranean, East Asia to Malesia, Central America (map: see Vink 1957; Wood 1972; Rzedowski 1978; esp. Ickert-Bond et al. 2005). [Photos - Collection]

• Altingiaceae are monoecious, deciduous trees that may be recognised by their spirally-arranged, usually palmately-lobed leaves with petiolar stipules and by their capitate inflorescences. The buds have scales, and the plants often produce resins.

Evolution. The fossil Microaltingia (ca 90 million years before present) has prolate, tricolpate pollen grains with a coarsely reticulate exine, a more or less superior ovary, ovaries with 8 or more ovules per carpel, and perhaps unwinged seeds; it may have been pollinated by insects (Zhou et al. 2001). If correctly assigned here, it is yet another fossil with interestingly plesiomorphous features (see also Calycanthaceae, Platanaceae, Fagaceae, etc.). Ickert-Bond and Wen (2006) give dates for divergence of clades within Altingiaceae; the basal split in the family is between the European + American and East Asian clades, dated somewhere between 19.5 and 54 million years before present.

Chemistry, Morphology, etc. Secretory canals are also reported from Mytilaria (Hamamelidaceae s. str.). There are strongly vascularised phyllomes interior to the staminal whorl; these may be staminodia, nectaries or organs sui generis. The orientation of the carpels varies (Bogle 1986). For the interpretation of the knobs, etc., surrounding the carpellate flowers, see Ickert-Bond et al. (2005). Endocarpial cells are thickened and elongated transverse to the long axis of the fruit, and they look almost palisade in tranverse section of the cells.

The testa is notably thinner than that of most Hamamelidaceae. Ickert-Bond et al. (2005) suggest that in Liquidambar the exotegmen "constitutes most of the seed coat", although it is absent in most Hamamelidaceae, a point also made by others (e.g. Rao 1974). This is not immediately evident in the sections presented (e.g. Fig. 9, G-J) nor in Melikian (1973), but if confirmed (see e.g. Ickert-Bond et al. 2007) it will indeed be a sharp difference to the mesotestal seeds of most Hamamelidaceae.

For information about Hamamelidaceae s.l., see Bogle (1986: floral morphology, etc.), Ferguson (1989: general, esp. fossils), Melikian (1973: seed coat anatomy), Zavada and Dilcher (1986: pollen), and Endress (1993: general). This and Hamamelidaceae - "micropyle faces upwards"?

Phylogeny. Shi et al. (2001) present a molecular phylogeny of Altingiaceae, and this suggests that there is only one genus, in contrast to a morphological phylogenies (Ickert-Bond et al. 2005, 2007), which also places the fossil Microaltingia as sister to the extant representatives of the family.

[Hamamelidaceae [Cercidiphyllaceae + Daphniphyllaceae]]]: ?

HAMAMELIDACEAE R. Brown, nom. cons.   Back to Saxifragales

Trees or shrubs; (C-glycosylflavones +); (vestured pits +; true tracheids +; nodes 5:5); sclereids common; petiole bundles annular (with adaxial bundle; arcuate); stomata variable; hairs usu. stellate; bud scales + (0); leaves two-ranked (opposite, spiral), ± conduplicate-flat or -plicate, (margins entire), (2ndary veins pinnate), stipules cauline; flowers (2-)4-5(-7)-merous; K free to connate, (0), C ribbon-like, adaxially circinate; A = and opposite sepals (3-many), staminodia opposite petals, anthers with conspicuous flaps, (with longitudinal slits), often basifixed, (connective not prolonged), (pollen 6-rugate), nectary a disc, staminodial, or on base of C [last two the same?]; G [2], usu. ± inferior, several-many often epitropous ovules/carpel, outer integument 6-7 cells across, (micropyle zig-zag), hypostase +, styles ± long, stigmas with multicellular protrusions, but no papillae; fruit a loculicidal and septicidal capsule, K often persistent; hilum large, coat often with often discoloration near the hilum, testa thick, hard, multiplicative, exotestal cells thickened (not), mesotesta massive, usually of fibrous sclerotic cells, tegmen tanniniferous; endosperm slight, nuclear (cellular - Parrotiopsis), perisperm +, (polyembryony +).

27[list]/82 - three groups below. Tropical to temperate, esp. East Asia to Australia, not South America (map: from Vester 1940; Vink 1957; Ying et al. 1993; Fl. N. Am. III 1997; Palgrave 2002). [Photos - Collection] [Photo - Flower]

1. Exbucklandoideae Harms

Inflorescence (elongated) capitate; (C 0); G ± inferior, (styluli long); seeds winged (not Mytilaria); exotestal cells alone thickened; n = 8, 12.

3/4-14. Exbucklandia, Rhodoleia, Mytilaria. East Asia, to Assam, East Malesia, to Sumatra.

Synonymy: Exbucklandiaceae Reveal & Doweld, Rhodoleiaceae Nakai

2. Disanthoideae Harms

Nectary at base of C, anthers with longitudinal slits, G; n = 8.

1/1: Disanthus cercidifolius. Japan.

Synonymy: Disanthaceae Nakai

3. Hamamelidoideae Burnett

Leaves with craspedodromous venation; (C 0), 1(-3, 2 sterile) ovules/carpel, (styluli long); fruit with ballistic dispersal of seeds; n = 12.

23/78. Tropical to temperate, esp. East Asia to Australia, not South America.

Synonymy: Fothergillaceae Nuttall, Parrotiaceae Horaninow

• Hamamelidaceae are trees or shrubs that may be recognised by their often stellate or tufted hairs, their strongly stipulate leaves which often have ± palmate venation, their smallish flowers borne in rather dense inflorescences, and their bivalved, loculicidal and septicidal capsules with rather large seeds that have a hard, shiny, bicolored testa. The seed are sometimes forcibly ejected. The bases of the branches are often swollen.

Evolution. The fossil Microaltingia (Altingiaceae) is ca 90 million years old (Zhou et al. 2001), and if Hamamelidaceae are close to Altingiaceae, this also gives a minimum age for Hamamelidaceae. Allonia decandra, a fossil probably to be placed in crown-group Loropetalineae, was collected from the Campanian in the eastern U.S.A., it has twice as many stamens as petals and a lobed disc adaxial to them (Magallón-Puebla et al. 1996); the seeds are rather angled, so there may have been more than one per loculus. For the early Tertiary fossil history of what are now East Asian endemic members of the family, see Manchester et al. (2009).

For the evolution of the flower in Hamamelidoideae, see Magallón (2007; fossils included, optimisation on to more than one topology). There is considerable variation in floral morphology. Corylopsis has rather ordinary-looking flowers with obovate petals, although they are probably derived; the basic condition for the family is to have narrow petals that are adaxially coiled in bud. In Parrotiopsis there are showy inflorescence bracts, and these are bright red in Rhodoleia, where the whole inflorescence is very like the flower of, say, Calycanthaceae. Petals may be lost, and in Fothergilla the inflorescence is made conspicuous by the plump and showy white filaments. Eustigma has quite long styluli and massive, purplish stigmas that are the most conspicuous parts of the flower. Fertilisation in Hamamelidaceae is often much delayed.

Chemistry, Morphology, etc. What is going on with the growth of Exbucklandia? Buds with one bud scale and branches with one or sometimes two prophylls at the very base are common in temperate genera of Hamamelidoideae. There is variation in the direction of initiation of the stamens in multistaminate androecia (Endress 1976); it can be centripetal (Matudaea) or centrifugal (Fothergilla).

Some information is taken from Endress (1970, 1993). For seed anatomy, see Rao (1974), Benedict et al. (2008: also fruits and fossils), and Magállon et al. (2001 and references) and Zhao and Li (2008) for fossils.

Phylogeny. Exbucklandioideae in the circumscription adopted here were apparent only in the analysis of ITS data and good sampling (75% bootstrap, better if gaps scored as a fifth character state: Li et al. 1999b; cf. Shi et al. 1998); the three genera have in the past all been placed in separate subfamilies. With rbcL data, Mytilaria alone was rather weakly supported as sister to [Disanthoideae + Hamamelidoideae] (Li et al. 1999a). Within Hamamelidoideae [Corylopsideae (monotypic) + Loropetaleae (weak support)] were sister to the rest, but tribal interrelationships had for the most part only weak support (Li & Bogle 2001). For a classification of Hamamelidoideae, see Li and Bogle (2001).

Cercidiphyllaceae + Daphniphyllaceae: plant dioecious; flowers small, C 0; pistillate flowers: ovary superior.

CERCIDIPHYLLACEAE Engler, nom. cons.   Back to Saxifragales

Deciduous trees; chalcones, dihydrochalcones +; cork in outer cortex; primary stem with continuous cylinder; prophyll adaxial; leaves usu. opposite, involute, (margins entire), stipule adaxial-petiolar; inflorescence capitate; P 0, floral apex?; staminate "flower": A 16-34 [= several flowers]; carpellate "flower": G 2-8 [= flowers], each subtended by a bract, suture abaxial, many ovules/carpel, styluli long, stigmas decurrent the entire length; fruit a follicle; seeds winged, chalazal appendage with hair-pin loop vascular bundle, testa undistinguished, exotestal cells enlarged, slightly thickened, tegmen tanniniferous; endosperm development?, slight, single suspensor cell notably enlarged; n = 19.

1[list]/2. China and Japan (map: from Heywood 1978; Fu & Hong 2000). [Photos - Collection]

• Cercidiphyllaceae are deciduous trees that may be recognised by the sharp distinction between the very long-lived short shoots and the long shoots; the leaves are more or less opposite, stipulate, and with palmate venation. The plant is dioecious, and what appear to be flowers lack a perianth; staminate "flowers" have several stamens and carpellate "flowers" have 2-8 carpels with long, decurrent stigmas. The fruits are follicles.

Evolution. For the early Tertiary fossil history of Cercidiphyllum, see Manchester et al. (2009). Palaeocene fossils (Joffrea) have 2-carpellate flowers borne on an elongated axis with the adaxial sutures of the carpels facing each other (Crane & Stockey 1985); the "flowers" of today's species are best interpreted as pseudanthia, indeed, the carpels are sometimes slightly separated from one another on the stout green axis, that of the "pedicel"=inflorescence axis. Both individual carpels and groups of stamens are subtended by bracts and are more or less decussately arranged. Yan et al. (2007) suggest that the bract is really a tepal because it is developmentally so different from the vegetative leaves, although it may have teeth or be almost bilobed; on balance, however, it still seems most likely that the structure is a bract.

Chemistry, Morphology, etc. Soltis et al. (2005: fig. 6:11) suggest perhaps inadvertently that the gynoecium is partly inferior. Takhtajan (1997) describes the venation of leaves on the long shoots as being pinnate, but the main secondary veins arise within 5(-10) mm of the base. Endosperm type?

For reported variation in nodal anatomy, see Howard (1979), and for general information, see Endress (1993).

DAPHNIPHYLLACEAE Müller Argoviensis, nom. cons.   Back to Saxifragales

Evergreen trees or shrubs; plants Al-accumulators, route I iridoids, triterpene [squalene] alkaloids +, myricetin, hydrolysable tannins, ellagic acid 0; pericyclic fibers 0; pits bordered; true tracheids +; stomata paracytic (laterocytic, anomocytic); plant glabrous; leaves spiral, ± flat, secondary veins pinnate, stipules 0; flowers pedicellate; P (0) 2-6; staminate flowers: A 5-12(-24), filaments with 3 traces, basal pit indistinct, pistillode 0 (+); carpellate flowers: staminodes 0; G [2(-4)], placentation apical-axile to parietal, (1)2 ± apical ovules/carpel, outer integument 5-6 cells and inner integument ca 4 cells across, micropyle zig-zag, hypostase +, styluli short, recurved, stigmas rather massive, with multicellular protrusions but no papillae; fruit a 1-seeded drupe; seed coat persistent but thin-walled and crushed, or endotegmen tanniniferous, walls thickened; endosperm cellular, perisperm slight, embryo short, cotyledons same width as radicle, (polyembryony +); n = 16; germination epigeal and phanerocotylar, hypocotyl emerges.

1[list]/10. East Asia to Malesia (map: from Huang 1997). [Photo - Fruit]

• Daphniphyllaceae are evergreen trees or shrubs that may be recognised by their rather weakly pseudoverticillate, exstipulate, entire leaves which are often glaucous beneath; the petioles of the leaves of a single innovation often vary considerably in length. The subcorymbose, racemose inflorescence with its small flowers with at most inconspicuous perianth is also distinctive.

Evolution. Some epiplemine Uraniidae (moths) have caterpillars that eat Daphniphyllaceae - and assorted asterids - probably because of the iridoids they have in common (Lees & Smith 1991).

Chemistry, Morphology, etc. The pith is at least sometimes septate. The flowers may be secondarily superior (D. Soltis et al. 2003b); staminodes can be found in both staminate and carpellate flowers. Bhatnagar and Kapil (1982) describe the endotegmic cells as being thickened and variously cutinised or sclerotic.

For information on flower, fruit and embryology, see Bhatnagar and Kapil (1982), see Tang et al. (2009) for stomata, etc., and for general information, see Zhang and Lu (1989) and especially Kubitzki (2006b); for a monograph, see Huang (1965).

Previous Relationships. Daphniphyllaceae have been difficult to place, sometimes being associated with Euphorbiaceae, etc., or placed in a separate order in the Hamamelidae (Cronquist 1981). Balanopaceae (see Malpighiales) were included in a bigeneric Daphniphyllanae (Takhtajan 1997).

[Crassulaceae [Aphanopetalaceae [Tetracarpaeaceae [Haloragaceae + Penthoraceae]]]] [[Iteaceae + Pterostemonaceae] [Grossulariaceae + Saxifragaceae]]: cork superficial; vessel elements with simple perforation plates; petiole bundle(s) arcuate; cuticle waxes not tubular; ovules apotropous [all?]; K persistent, withered; endosperm cellular.

Chemistry, Morphology, etc. Mauritzon (1933) provides information on the ovules and endosperm development for many taxa in this clade.

Phylogeny. The relationships [Crassulaceae [Tetracarpaeaceae [Haloragaceae + Penthoraceae]] [[Saxifragaceae [Iteaceae + Pterostemonaceae]] Grossulariaceae]] were found by Morgan & Soltis (1993); Aphanopetalum (ex Cunoniaceae) is now to be placed in the first major clade.

Crassulaceae [Aphanopetalaceae [Tetracarpaeaceae [Haloragaceae + Penthoraceae]]]: stem with endodermis; nodes 1:1; stipules 0; pollen 3-colporate.

CRASSULACEAE Jaume Saint-Hilaire, nom. cons.   Back to Saxifragales

Succulent herbs to soft-stemmed shrubs; mycorrhizae 0; crassulacean acid metabolism common; flavones, acylated flavonol glycosides, sedoheptulose [sugar reserve], non-hydrolysable tannins +, hydrolyzable tannins 0; red pigment common, even in roots; (cork cortical); young stem with separate bundles; (medullary bundles +); sieve tube plastids lacking starch grains [and protein inclusions]; rays 0; nodes also 1:1-3, 3:3, etc.; cuticle waxes very variable; stomata usu. anisocytic; leaves succulent, ptyxis flat to curved, margins entire (serrate), hydathodes +; inflorescence cymose; A obdiplostemonous, anthers median sagittate, latrorse; G 3-10(+), ± free, (placentation parietal), opposite petals, nectaries [ougrowths of carpels] at base, 1-many ovules/carpel, micropyle bistomal, outer and inner integuments 2 cells across, (styluli long), stigmas punctate to moderately capitate; fruit a follicle; exotestal cells with outer wall ± thickened, inner pigmented layer +; endosperm mostly cellular and variants, chalazal haustorium +, embryo long, suspensor uniseriate, basal cell with mycelium-like haustorial branches; x = 8; germination epigeal and phanerocotylar.

34[list]/1370 - three subfamilies below. Cosmopolitan, esp. the Cape region and Mexico, but few in S. South America and Australia, not in Polynesia, frequently in drier regions. (map: incomplete - see Hultén 1958; Bywater & Wickens 1983; Jürgens 1995; Thiede 1995). [Photo - Flower]

1. Crassuloideae Burnett

Leaves opposite, with several marginal hydathodes only; stamens = K, slightly introrse, (odd carpel abaxial - Tillaea), ovules tenuinucellate, (archesporium multicellular); follicles releasing seeds through apical pore; testa cells with sinuate anticlinal walls, unipapillate; first division of micropylar endosperm cell in horizontal plane; n = 7, 8.

2/196: Crassula (195). Southern Africa, to S.W. Arabia, "Tillaea" more or less world-wide, the only representative of the family in Australia.

Kalanchoideae + Sempervivoideae: lamina with single (sub)apical hydathode; (placentae lobed); seeds costate; first division of micropylar endosperm cell in vertical plane.

2. Kalanchoideae A. Berger

Plant ± woody; bufadienolides +; crystal sand +; C connate; A with spherical connective prolongation, (style +); seeds 4-6-costate, with a micropylar corona; x = 9 [n = 9, 17 (18)].

4/200: Kalanchoe (145), Tylecodon (46). Old World, especially the Karroo in southern Africa, but extending to South East Asia and Malesia, not Australasia.

3. Sempervivoideae Arnott

(Pyrrolidine and piperidine alkaloids +, nonhydrolyzable tannins 0 - esp. Sedum acre group); leaves (opposite); flower 4-32-merous, (C connate), nucellus elongated; (follicle with abaxial dehiscence - Diamorpha); seeds ³6-costate; suspensor with vertical division in micropylar cell; n = ³5, much variation.

28/975: Sedum (420), Echeveria (140), Sempervivum (65), Rhodiola (60), Dudleya (47). Largely N. hemisphere.

• Crassulaceae are recognisable by their succulent herbaceous or soft-stemmed habit, and their flowers, which have the same number of sepals, petals and carpels; the latter are more or less free and have nectariferous scales at their bases. The fruits are usually follicles.

Evolution. Crassulaceae may be some 77-69 million years old, with diversification beginning 41-39 million years before present (Wikstöm et al. 2001).

The distinctive wood, which lacks rays and has very short vessel elements with annular and helical thickening, is probably paedomorphic (t'Hart & Koek-Noorman 1989); plant chemistry, in particular the presence of hydrolyzable tannins and the absence of non-hydrolyzable tannins, as in other woody Saxifragales, is consistent with this idea (Thiede & Eggli 2006). Aeonium, a largely Macaronesian genus with the most endemic species on the archipelago, has striking growth forms each of which seems to have evolved just once (Mes & t'Hart 1996). Crassulacean acid metabolism (CAM) is common throughout the family (e.g. references in Winter & Smith 1996a).

Some species of Kalanchoe produce plantlets, foliar embryos (Yarborough 1932), in notches at the margin of the leaf blade. Both embryogenetic and organogenetic pathways have been coopted, and the young plantlets have cotyledon-like first leaves; species in which development of plantlets is constitutive, i.e. plantlets are produced without the plant being damaged, do not produce viable seed (Garcês et al. 2007).

All ca 200 species of the Echeveria group (Sempervivoideae) appear to be interfertile, a remarkable situation apparently without parallel in flowering plants (Uhl 1992). There have been several origins of sympetaly in Sedoideae s.l. ('t Hart et al. 1999; Carillo-Reyes et al. 2009); both it and epipetaly tend to be weak. The increase in numbers of flower parts in some Sedoideae - some have a multistaminate androecium - is in the context of an increased merousness of the whole flower; the relation between the number of parts of each whorl is unchanged from that of a basic core eudicot flower (see also the asterid I + II clade), i.e. K = C = G; A = 2x C.

Chemistry, Morphology, etc. Anthocyanin is also found in the roots of Saxifragaceae, as well as Melastomataceae, Balsaminaceae, Asteraceae, Droseraceae, and Francoaceae (Krach 1976; Molisch 1928). Sedoheptulose is the most abundant sugar in Crassulaceae; isocitrate is common, unlike other succulents (Thiede & Eggli 2006). The sieve tube plastids are a distinctive variant, lacking starch, the S₀ type (Behnke 1988a). For (mistaken) reports of cortical vascular bundles, see Thiede and Eggli (2006). The leaf blade usually lacks palisade tissue, and the stomata are often on both sides. The stomata may also be heliocytic, with an additional ring of distinct cells outside a basically anisocytic configuration.

There is sometimes only a single vascular trace to the calyx (t'Hart & Koek-Noorman 1989). The ovary is secondarily superior (D. Soltis et al. 2003b). The nucellar epidermal cells seem to be large and more or less radially elongated in Crassuloideae and Kalanchoideae (Mauritzon 1933). Haustoria from places other than the suspensor are reported for Crassulaceae (Mickesell 1990). The testal cells of Crassula often have a single papilla and sinuous anticlinal walls (e.g. Bywater & Wickens 1983). There is no tegmen. Spongberg (1977) notes that the endosperm is usually scanty, while Mabberley (1997) describes it as being copious; the former is correct. For the base chromosome number of the family and of its major clades, see Mort et al. (2001).

Some general information is taken from Spongberg (1977); for embryology, etc., see Mauritzon (1933) and Subramanyam (1970), for chemistry, see Stevens (1995), for general recent accounts of the family, see Eggli (2003: enumeration of all species) and in particular Thiede and Eggli (2006).

Phylogeny. The basic phylogenetic structure of the family seems fairly well established (e.g. van Ham 1995; van Ham & 't Hart 1998; Mort et al. 2001; Mayuzumi & Ohba 2004). The rather highly derived Crassuloideae are sister to the rest of the family; Kalanchoe and relatives are sister to the rest of Sedoideae, and this is in some respects a rather plesiomorphous group (e.g. Mort et al. 2001; Mort 2002; Thiede & Eggli 2006). Tillaea appears to be polyphyletic from within Crassula (Mort et al. 2009). For the phylogeny of Kalanchoe and variation of CAM within it, see Gehrig et al. (2001) and Kluge and Brulfert (1996). Sedum itself occurs in five of the seven main clades apparent in phylogenetic analyses of Sempervivoideae (van Ham 1995; van Ham & 't Hart 1998; Mayuzumi & Ohba 2004); Graptopetalum is also very difficult (Acevedo-Rosas et al. 2004). Within the Acre clade of Sedum, most New World Sedoideae as well as all the old Echeverioideae in the study formed a single clade, although it was only poorly supported (Carillo-Reyes et al. 2009).

Classification. Within Sempervivoideae, generic limits are unclear, and many genera, some previously placed in what were considered to be different subfamilies, e.g. Sedoideae and Echeverioideae, hybridise (e.g. Uhl 1976; 't Hart et al. 1999). Thiede and Eggli (2006) provide a guide through the chaos; note that they prefer to retain a paraphyletic Sedum, although this wil have to be dismembered or its cirumscription dramatically increased.

Synonymy: Cotyledonaceae Martynov, Rhodiolaceae Martynov, Sedaceae Adanson, Sempervivaceae Durande, Tillaeaceae Martynov

Aphanopetalaceae [Tetracarpaeaceae [Haloragaceae + Penthoraceae]]: ?

APHANOPETALACEAE Doweld   Back to Saxifragales

Viny shrub; chemistry?; vessel elements with scalariform perforation plates; pericyclic fibers 0; petiole with 3 (1) bundles; leaves opposite, leaf teeth with a single vein, stipules +, looking like teeth; inflorescence cymose or flowers solitary; flowers 4-merous, hypanthium short, C 0 [rudiments visible only when very young], pollen with rugulate-stellate surface; G seminferior, opposite petals, 1 apical epitropous ovule/carpel, micropyle bistomal, style single, with four canals, branches short; fruit a nut, K enlarging; seed coat?; endosperm development?, embryo curved, size?; n = ?

1/2. W. and E. Australia (map: partly from FloraBase 2004). [Photo - Flower.]

• Aphanopetalaceae are scrambling shrubs with prominently lenticellate stems, opposite and more or less serrate leaves with minute "stipules", and axillary inflorescences. the flowers are apetalous and have a spreading calyx that enlarges in fruit; the fruit itself has a single seed.

Chemistry, Morphology, etc. The ray parenchyma stores starch. Two small bundles soon diverge from the main leaf trace. It is unclear if the stipules "are" stipules or colleters (Kubitzki 2006b). As in Saxifragaceae and Iteaceae, the vascular trace in the petal plane gives a branch to the lateral sepal position, also carpel wall and lateral carpel traces and a single stamen trace; the trace in the sepal plane supplies the carpel wall and median carpel bundle and provides a stamen bundle.

Some information is taken from Jensen (1968: vascular system), Bensel and Palser (1975b: floral anatomy), Dickison (1980b: nodal anatomy), Dickison et al. (1994: anatomy) and Kubitzki (2006b: general).

Tetracarpaeaceae [Haloragaceae + Penthoraceae]: ?

Classification. Although combination of these three rather small families is an option in A.P.G. II (2003), there seems to be little or nothing holding them together morphologically (see also Moody & Les 2007) and they are kept separate in A.P.G. III (2009).

TETRACARPAEACEAE Nakai   Back to Saxifragales

Evergreen shrub; chemistry?; plant glabrous; leaves spiral, petiole short; inflorescence racemose; floral apex convex; flowers 4(-5)-merous, C spatulate, A 4-8, basal pits?, fibrous endothecium 0, nectary 0; G 4 (5), opposite petals, many ovules/carpel, micropyle exostomal, stigmas sessile, not expanded; fruit a follicle; exotestal cells ± elongated longitudinally, outer walls thickened, no mechanical layer; endosperm development?, embryo size?; n = ?

1/1: Tetracarpaea tasmannica. Tasmania. [Photo - Habit]

Chemistry, Morphology, etc. The teeth are perhaps hydathodal, but there are no water pores. The ovary is apparently secondarily superior (D. Soltis et al. 2003b).

See Hils et al. (1988: anatomy) and Kubitzki (2006b: general) for information.

Penthoraceae + Haloragaceae: ?

PENTHORACEAE Britton, nom. cons.   Back to Saxifragales

Rhizomatous herbs; flavonoids +, flavones, myricetin, non-hydrolysable tannins 0; cork ?; young stem with pseudosiphonostele; endodermoid layer?; pericyclic fibers 0; leaves spiral, supervolute, amphistomatic, teeth hydathodal, colleters +; inflorescence cymose, flowers 5-7-merous; hypanthium +, C 0(-7); A 2x K, lacking a basal pit, latrorse; G [5-8], half inferior, apical parts free, opposite sepals, becoming superior, many ovules/carpel, placentae intrusive, adaxial, micropyle?, funicles long, styluli submarginal, stigmas capitate; free part of each carpel basally circumscissile in fruit; exotestal cells with outer wall ± thickened, micropylar operculum endostomal, tegmen otherwise crushed; cell at end of suspensor large, embryo large; n = 8 (9).

1[list]/2. East and South East Asia, E. North America (map: from Hong 1993). [Photo - Penthorum Inflorescence]

• Penthoraceae are rather fleshy herbs with spiral, entire leaves. They may be recognised by their few-branched, monochasial cymose inflorescence, and often apetalous flowers with partly connate carpels with submarginal styles. The fruits are distinctive since the free part of each carpel splits circumscissilely at the base.

Evolution. Although stem group Penthoraceae have been dated to 77-69 million years before present (Wikström et al. 2001), the E. North American/East Asian disjunction is dated to 6.5-2.4 million years before present (Thiede 2006 for references).

Chemistry, Morphology, etc. Cork in the root is initiated in a superficial position (van Tieghem 1899). The sepals are unequal in size and the bracts are lateral to the pedicels. There is a much-enlarged but non-dividing micropylar cell in the embryo suspensor - cf. Haloragaceae and their haustorial suspensor. Danilova (1996) shows the carpels opposite the calyx. The first two pair of seedling leaves are opposite.

See Spongberg (1972, in Saxifragaceae: general), Haskins and Hayden (1987: anatomy), Gornall (1998: as Saxifragaceae) and Thiede (2006: general) for information.

HALORAGACEAE R. Brown, nom. cons.   Back to Saxifragales

Aquatic or amphibious herbs to shrublets (small trees); flavones +, flavonols 0; cork ?; often calcium oxalate crystals in hair-like cortical cells; cuticle waxes 0 (parallel platelets); leaves opposite (spiral), conduplicate-flat, (pinnately compound), colleters +?; plant monoecious (dioecious; flowers perfect), inflorescence dichasial, cymose or fasciculate, or flowers solitary; flowers small, (2-)4-merous, K valvate, C deciduous (0); A 8 (= and opposite sepals), (anthers apiculate), basal pits?, pollen trinucleate, 4-6(-20)-colpate or -porate, ovary inferior, (sub 1-locular), opposite petals or odd member adaxial, 1 (2) pendulous ovules/carpel, nucellar cap, hypostase and poorly developed funicular obturator +, styluli with swollen bases, stigmas (sessile), capitate or not, penicillate, dry; fruit nut-like, (schizocarpic), exocarp often ornamented, stones 1- or 4-seeded; exotesta (and hypodermal layer) persistent, thin-walled, rest obliterated; endosperm +, (nuclear), starchy, haustorial suspensor +, embryo long (short); n = 6, 7 (8).

8[list]/145: Myriophyllum (60). World-wide, but especially Australia (map: from van Steenis 1962; Hultén 1958, 1971; van der Meijden 1971; Wood 1972; Orchard 1981). [Photo - Collection.]

• Haloragaceae are rather small, usually monoecious, often more or less herbaceous plants. Their leaves are serrate or deeply lobed, small, and often opposite or whorled, and their flowers are small (Haloragodendron is an exception) and have penicillate stigmas.

Evolution. Hernández-Castillo and Cevallos-Ferriz (1999) suggest that the fossil Tarahumara sophiae (found in Mexico in deposits laid down ca 70 million years before present) had carpels free from one another but adnate to a hypanthial wall, while its fruit is described as being drupe-like, so its morphology is unlike that of any extant member of the family. Proserpinaca is a common fossil in Europe and Asia, but the genus is no longer found there.

In some species of Haloragodendron the whole inflorescence is coloured. Trihaloragis has flowers in which all whorls are trimetous, very unusual in eudicots, and Moody and Les (2007) point out the extensive variation in floral merism in the family.

Chemistry, Morphology, etc. Nodal anatomy was observed in Haloragis erecta and Laurembergia, ptyxis in the first. Pelargonidin occurs in leaves, as in Saxifragaceae (Doyle & Sogin 1988). Adventitious roots arise between the leaves in Haloragis. Myriophyllum appears to have endostomal ovules (Batygina et al. 1985), while Bawa (1970) noted that the archesporial cell is hypodermal, i.e. the ovule is tenuinucellate. Corner (1976) describes the endosperm as being starchy.

Some information is taken from Orchard and Keighery (1993: Meziella), Orchard (1975) provided a partial monograph of the family (Antipodean taxa), with much detail on floral anatomy, etc., and Kubitzki (2006b) summarized what is known about it; see Praglowski (1970) for pollen.

Phylogeny. See Moody and Les (2001, 2004 and especially 2007, note that in the latter study the nuclear ITS was in some conflict with chloroplast genes) for relationships within the family, which may have an Australian origin. The woody [Glischrocaryon + Haloragodendron] clade is sister to other Haloragaceae, although the monophyly of the two genera themselves is not certain. Much of the rest of the family forms a single clade, but relationships within it are uncertain, Meionectes and Proserpinaca in particular not having a definite position (Moody & Les 2007). The trimerous Trihaloragis is sister to all other members of this clade (Moody & Les 2007).

Previous Relationships. The monotypic Haloragales were placed near Saxifragales by Takhtajan (1997) or linked with Gunneraceae and placed next to Myrtales, as by Cronquist (1981). Historically Gunneraceae and Haloragaceae have been associated, although their pollen is quite different (e.g. Praglowski 1970), the perianth of Gunneraceae is not differentiated into two whorls of sepals and petals, etc.; for the former, see Gunnerales.

Synonymy: Cercodiaceae A.-L. de Jussieu, Myriophyllaceae Schultz-Schultzenstein

[Iteaceae + Pterostemonaceae] [Grossulariaceae + Saxifragaceae]: (vessel elements with scalariform perforation plates); leaves spiral; hypanthium +, stamens = and opposite K; fruit a septicidal capsule.

ITEACEAE J. Agardh, nom. cons.   Back to Saxifragales

C-glycosylflavones +; stipules +; placentation axile, micropyle?, style well developed; endosperm sparse.

2/21. Rather scattered, warm temperate to tropical.

Pterostemon

Shrubs; conical to peltate glandular hairs +; stipules cauline, minute; inflorescence a corymbose cyme; hypanthium?, K valvate, A 5, opposite sepals, filaments flattened, toothed, anthers with basal pits?, pollen 3-colporate, 5 staminodes opposite petals, disc 0; G [5], largely inferior, orientation?, 4-6 ascending apotropous ovules/carpel, (styluli ± radiating), stigma capitate, ?type; C also persistent in fruit; seed coat "cartilaginous"; n = ?

1[list]/3. Mexico (map: from Smith et al. 2004).

Synonymy: Pterostemonaceae Small, nom. cons.

Itea L.

Trees to shrubs; allitol +, flavonols, ellagic acid 0; hairs unicellular; young stem with separate bundles; pith chambered; leaves conduplicate, margins spiny- or gland-toothed, stipules small, on petiole base or adjacent stem; inflorescence axillary, (branched) racemose, flowers rather small; C valvate, anthers lacking basal pits, connective forming apical protrusion, pollen bilateral, 2-porate, ektexine homogeneous, disc or nectary lining hypanthium; G [2] to subinferior, many ovules/carpel, stigma punctate-lobed, wet; fruit valves often attached by the stigma; exotestal cells with outer walls thickened; (endosperm moderate - Itea rhamnoides); n = 11.

1[list]/18. South East Asia to W. Malesia, E. North America, E. and S. Africa (map: from Mai 1985; Aubréville 1974a; Palgrave 2002; Heywood 2007). [Photo - Itea Flower.]

• Pterostemon looks rather like Rosaceae, both being woody and having spiral, serrate, and stipulate leaves, although the leaf blades of Pterostemon tend to be notably shiny. However, the distinctive conical to peltate glandular hairs, the exudate of which produces the shiny leaf surface, and the flowers, each with a whorl of five stamens that have winged and toothed filaments and a whorl of five staminodes, allow it to be easily recognised.

• Itea may be recognised by its stems with chambered pith, spirally-arranged serrate leaves with superposed axillary buds, paniculate to racemose inflorescences with rather small flowers, and septicidal fruits the valves of which often remain attached by the stigma. The flowers have a hypanthium, the corolla is valvate, and there are only two carpels.

Evolution. The fossil Divisestylus, from the late Cretaceous some 90 million years before present and perhaps part of the Itea clade, has five stamens opposite the sepals and ovaries and stigmas fused, but there are separate styles - just like Itea. However, the pollen is tricolpate and striate, suggesting that the 2-porate pollen mentioned above is not a synapomorphy for the whole clade (Hermsen et al. 2003), merely for extant Itea. However, the distinctive pollen of Itea is known fossil in Europe in the Eocene and somewhat later in North America (Hermsen et al. 2003).

Chemistry, Morphology, etc. For the chemistry of Itea, see Bohm et al. (1999). Pterostemon also has flavones, in this being like other Saxifragales. Choristylis (= Itea) lacks axial parenchyma. The ovule of Itea is sometimes described as being unitegmic, with the integument 6-7 cells across, but this appears to be incorrect (Kubitzki 2006b). Although the carpels are free initially, they become connate, even along the style (Ge et al. 2002), so perhaps the styles are more accurately described as being postgenitally connate styluli.

For additional information is taken from Spongberg (1972: general), Bensel and Palser (1975b: floral anatomy), Ramamonjiarisoa (1980: anatomy), Gornall et al. (1998: general), Ge et al. (2002: floral development) and Kubitzki (2006b: general).

The pericyclic fibers of Pterostemon seem to be weakly developed and the androecium is obdiplostemonous. Similar peltate glandular hairs are known from Grossulariaceae. For more information, see Goldberg (1986), Wilkinson (1994, 1998) and especially Kubitzki (2006b: as Pterostemonaceae), but the genus is not well known.

Classification. Combination of Iteaceae and Pterostemonaceae was optional (as Iteaceae s.l.), see A.P.G. II (2003); this broadened circumscription was formally adopted by A.P.G. III (2009).

Grossulariaceae + Saxifragaceae: G [2-3]; (endosperm helobial).

GROSSULARIACEAE de Candolle, nom. cons.   Back to Saxifragales

Shrubs; cork cambium outer cortical/pericyclic; underground stems with endodermis; pericyclic fibers 0; leaves conduplicate-plicate, margins also lobed, teeth hyadthodal, 2ndary veins palmate, leaf bases broad and with thin margins (paired prickles at the nodes); inflorescence racemose, pedicels articulated; (flowers 4-merous), nectary at base of well-developed hypanthium, C small, open, (staminodes +; A 10), pollen 5-15-porate, with distinctively rugose ectoapertures, tectum complete, G inferior, usu. median, placentation parietal, many ovules/carpel, outer integument 3-5 layers across, hypostase +, style single, long, stigma capitate, wet; fruit baccate; seeds hard, arillate, exotestal cells palisade, mucilaginous [all told this layer 3-6 cells thick], endotestal cells crystalliferous, radial and inner walls lignified, ?tegmen cells elongated, tanniniferous; endosperm hemicellulosic, embryo short; n = 8, chromosomes 1.5-2.5 µm long; germination epigeal.

1[list]/150: Ribes. Temperate N. hemisphere, also along the Andes (map: see Hultén & Fries 1986; Jalas et al. 1999; Malyschev & Peschkova 2004 - incomplete). [Photos - Collection.]

• Grossulariaceae may be recognised by their lobed, spiral leaves with broad bases, and flowers with a well-developed hypanthium, relatively large and colored sepals and smaller petals, a usually bicarpellate gynoecium, and parietal placentation; the fruit is baccate.

Evolution. The fruits of Ribes are an important food for Andean frugivorous birds. Several species of insects have been recorded as eating species of Ribes (Weigend 2006). A number of fungi, including the white pine blister rust (the basidiomycete Cronartium ribicola), spend part of their life cycles on the plants of this genus; in parts of North America largely unsucessful attempts have been made to eradicate Ribes so as to disrupt the life cycle of this damaging fungus, Ribes harbouring the telial stage.

Chemistry, Morphology, etc. Stem collenchyma is well developed. Nectar glands on the anthers are reported from some species. There is considerable variation in pollen morphology, and Ribes divaricatum has pentacolpo-di-orate pollen (Weigend 2006). Some information is taken from Cutler and Gregory (1998: anatomy); see Weigend (2006) for a general account.

Phylogeny. Weigend et al. (2002) and Senters and Soltis (2003) suggest phylogenies for Ribes.

Previous Relationships. Grossulariaceae as circumscribed by Cronquist (1981) are very heterogeneous, and include genera now placed in Phyllonomaceae (inc. Dulongiaceae), Escalloniaceae (both asterid II), Montiniaceae, Tribelaceae (both asterid I), Tetracarpaeaceae, Iteaceae (including Pterostemonaceae), and Celastraceae (Brexiaceae) (all rosids).

Synonymy: Ribesiaceae Marquand

SAXIFRAGACEAE Jussieu, nom. cons.   Back to Saxifragales

Largely herbs; cork also pericyclic; young stem with separate bundles [pseudosiphonostele]; nodes also 1:1, 1:2 and 3<:3<; petiole bundles also annular (with medullary or adaxial bundles); hairs (uni-)multiseriate with multicellular glandular head; leaves (opposite), (bases sheathing), ptyxis variable, (margins entire), 2ndary veins usu. palmate, colleters +; A 10, obdiplostemonous, pollen colpate, colporate, or 6-9-porate, nectariferous disc + (0); G superior to inferior, (free), apical parts quite often free, orientation variable, placentation parietal to axile, many ovules/carpel, styluli long [no compitum], stigmas spatulate to capitate, wet or dry; fruit also a follicle; exotestal cells with outer wall (radial walls) ± thickened, inner pigmented layer +, (endotegmen crystalliferous); endosperm moderate, embryo medium (large); n = (5-)7(+); chloroplast rpl2 intron 0.

Ca 33[list]/540 - two groups below. Mostly N. temperate and Arctic (S. temperate, tropical mountains) (map: from Hultén 1958, 1971; Meusel et al. 1965 - incomplete). [Photos - Collection]

1. Saxifraga s. str.

(Flowers obliquely monosymmetric).

1/370-500. Mostly Arctic and tropical montane, Saxifraga magellanica grows down the Andes.

2. The Rest

(Leaves palmately or ternately compound; stipules basal or sub-basal on petiole, persistent, or cauline [Astilbe]); flowers (3-10)-merous; (hypanthium 0), (C laciniate or toothed; 0), (A 3-15; obhaplostemonous - Chrysosplenium), pollen colpate, colporate, or 6-9-porate, disc + (0); G [(3-5)], ovules unitegmic; (endotegmen thick-walled [Heuchera, Tolmeia]).

Ca 30/170: Micranthes (70), Chrysosplenium (55), Heuchera (35). Mostly N. temperate (tropical montane and Arctic).

• Saxifragaceae can be recognised by their herbaceous habit and gland-toothed leaves; stipules, when present, are often borne on the petiole near the base. The flowers often have 2-3 carpels that are basally connate, although the styluli are clearly free, a nectariferous disc is common, and the fruit is a follicle or septicidal capsule.

Evolution. The moth Greya (Prodoxidae), related to Tegeticula, of yucca moth fame, is both a seed predator and pollinator of some Saxifragaceae (Segraves & Thompson 1999); this association is being studied in considerable detail, clarifying the diversification of both plant and pollinating seed parasite (Rich et al. 2008 and references). It has even been suggested that the general lability of ovary position in the family is connected to selection by such pollinators (Soltis & Hufford 2002); protected, i.e. inferior, ovaries will be favoured. Mitella (and a few other Saxifragaceae) is very largely pollinated by fungus gnats, and this association seems to have evolved in parallel. Since taxonomists here, as elsewhere, used the floral features of pollination syndromes to distinguish genera, Mitella has turned out to be polyphyletic (Okuyama et al. 2008): Flowers pollinated by fungus gnats are often more or less broadly saucer-shaped and the petals have very narrow lobes.

The seeds of a number of forest-dwelling Saxifragaceae are dispersed by rain, whether by a splash cup mechanism, as in Mitella, or by the seeds being thrown from the fruit as it moves violently after being hit by a drop of water, as in Heuchera.

Short-cycle Puccinia rusts (Uredinales, basidiomycetes) are frequently found on Saxifragaceae s. str. (Savile 1979a, b).

Introgressive hybridisation in the Heuchera clade is extensive and there are various combinations of chloroplast and nuclear genomes, for example, the chloroplast genome of Tellima is also found in Mitella (e.g. Soltis et al. 1993).

Chemistry, Morphology, etc. The distribution of druses and acicular crystals is of systematic interest (Gornall 1987); only Saxifraga s.l. has been studied in detail. Over 50 vascular bundles may enter the petiole base in some taxa. Hydathodes are common. In at least some species of Saxifraga, and in Astilbe and Rodgersia, the two carpels are oblique, but in the latter two this is associated with inverted floral orientation, the odd K being abaxial. Many other taxa have median carpels (Eichler 1878; Engler 1930a; Eckert 1966). Variation in ovary position within the family is extreme, even occuring within genera and between the different morphs of heterostylous flowers (e.g. Kuzoff et al. 2001; Soltis & Hufford 2002). In Chrysosplenium one carpel is open, the other closed. Darmera has only a single integument and it is 4-6 cells thick (Gornall 1989).

In Saxifraga s. str. the diploid chromosome number varies from 12-ca 200, in Micranthes from 10-120.

Some information is taken from Morf (1950) and Spongberg (1972), details of vegetative anatomy from Thouvenin (1890) and Gornall (1998) and details of floral anatomy from Bensel & Palser (1975b).

Phylogeny. There are two major clades in Saxifragaceae, Saxifraga s. str., largely arctic-alpine, and the Heuchera clade, the rest of the family and predominantly temperate in distribution. The inflorescence of Saxifraga s. str. often has cauline bracts, however, that of Darmera, including species that used to be included in Saxifraga, is scapose, and there are pollen and testa surface differences between the two. Members of the Heuchera clade contain the bulk of the floral variation in the family (Soltis et al. 2001), but generic limits are unclear (Soltis et al. 1996, and refs.; Okuyama et al. 2008).

Classification. McGregor (2008) provides a useful and well-illustrated summary of the ornamentally important Micranthes and Saxifraga.

Previous Relationships. In the past, genera "intermediate" between the variable Saxifragaceae and other families tended to be included in Saxifragaceae. This was because the inclusion of more odd genera in Saxifragaceae would have little effect on the family description since there was already so much variation included in it. However, if placed in the more homogeneous Crassulaceae, for example, such genera would greatly affect the description of that family and hence make it less discrete. Many woody, tenuinucellate and unitegmic genera that used to be included in Saxifragaceae have turned out to be entirely unrelated either to other members of that family or to each other, indeed, there was clearly a division between Saxifragaceae + Grossulariaceae, with their petals that remain very small for quite some time during development, and Hydrangeaceae, with their relatively faster-developing petals (as is common is asterids) and septicidal capsules (Gelius 1967). Of Saxifragaceae in the old and broad sense, the woody Escallonia (Escalloniales, asterid II/campanulid), Hydrangea and relatives (Cornales: some species are more or less herbaceous) and many other woody taxa are asterids, while the herbaceous Parnassia is in Celastraceae-Celastrales (a conclusion in agreement with data from floral anatomy - e.g. Bensel & Palser 1975a, d). However, the unitegmic Darmera is properly to be retained in Saxifragaceae (Gornall 1989). Fehrenbach and Barthlott (1988) included Dichroa (Hydrangeaceae-Cornales), Francoa (Francoaceae-Geraniales), Vahlia (unplaced asterid I/lamiid), as well as Penthorum (Penthoraceae) and Saxifragaceae s. str. in their survey of cuticle waxes, but no significant differences between the species in these very different groups were noted.

Saxifragaceae can be confused with Rosaceae; Astilbe (Saxifragaceae) and Aruncus (Rosaceae - Rosales) are particularly similar. However, the carpels of former are usually two and basally connate rather than several to many and free, and the flowers have five or ten rather than fifteen to many stamens. Saxifragaceae and Rosaceae are not close.

Synonymy: Brachycaulaceae Panigrahi & Dikshit, Pectiantiaceae Rafinesque

CYNOMORIACEAE Lindley, nom. cons.   Back to Saxifragales

Echlorophyllous root parasite; vessel elements?; cork?; hairs 0; leaves spiral; plant monoecious, inflorescence capitate; flowers minute; P (1)4-5(-8), basally connate or not; staminate flowers: A 1, adnate to P, pollen colporate, ?nectary-stylodium +; carpellate flowers: staminodia 0; G 1, inferior, 1 pendulous straight unitegmic ovule, integument 5-7 cells thick, nucellar cap +?, style long, channeled; fruit an achene; testa ca 7 cells across, persistent, cells little thickened; endosperm cellular, copious, thick-walled, embryo undifferentiated; n = 12, size strongly bimodal.

1[list]/2. Mediterranean to C. Asia (map: from Jalas & Suominen 1976; Jäger et al. 1985; Hansen 1986). [Photo - Habit © D. L. Nickrent.]

Evolution. In the Mediterranean area the host is often a member of Cistaceae or Amaranthaceae, elsewhere Cynomorium parasitizes Amaranthaceae, Tamaricaceae, Nitrariaceae, etc. (Jäger at al. 1985).

Chemistry, Morphology, etc. The root has root hairs. Perfect flowers are also known. The perianth is less well developed in pistillate than in staminate flowers, and there is debate as to its morphological nature. The pistillode in staminate flowers may be superior or inferior, according to Hooker (1856), clearly a matter that should be cleared up. The channeled style has two vascular bundles, together a rather odd combination.

For details of seed anatomy, see Takhtajan (2000), for morphology, see Weddell (1860), for ovule, etc., see Juel (1902) and Teryokhin et al. (1975), for some chemistry, see references in Zhang et al. (2009), and for general information, see the Parasitic Plants website (Nickrent 1998 onwards) and also Heide-Jørgensen (2008).

Previous relationships. Cynomoriaceae have usually been included in Balanophoraceae or Balanophoranae (e.g. Cronquist 1968; Takhtajan 1997), or their position has seemed to be completely uncertain (see above).