1a. Cercideae - Leaves apparently simple, bilobed or not. - 4-12/265: Bauhinia (250: for generic limits, see Sinou et al. 2007). Pantropical (temperate).

1b. Duparquetia - Floral development acropetal ["normal"], K 4, petal-like, adaxial-median C external; A 4, opposite K, connate, porate; G initiation not advanced. - 1/1: Duparquetia orchidacea. Tropical W. Africa.

The standard shows colour patterning (Prenner & Klitgaard 2008b for details of floral development).

1c. Detarieae - Resins with bicyclic terpenes; leaflets with crater-like glands on the abaxial surface, stipules deciduous; bracetoles decidous.

"Caesalpinioideae" + Mimosoideae + Faboideae: leaves bipinnate; vestured pits +.

2. "Caesalpinioideae" Candolle - 160/1930: Senna (350), Chamaechrista (265), Caesalpinia (100: ?polyphyletic). Predominantly tropical, esp. Africa and America.

3. Mimosoideae Candolle - Sieve tube plastids also with fibres; extrafloral nectaries common; flowers often aggregated into heads and developing together, bracteoles 0; flowers rather small, polysymmetric, K connate, C connate, valvate, odd member abaxial, claws 0; A often connate, polyads common; funicle long, thin, testa with vascular strand, pleurogram +, linea fissura common, U-shaped [open]. - 82/3275: Acacia s. str. (960: the old subgenus Phyllodinae), Mimosa (480), Inga (350), Calliandra (200), Vachellia (161: the old Acacia subgenus Acacia), Senegalia (85: the old Acacia subgenus Aculeiferum), Pithecellobium (40: some polycarpelate, but secondary; red and black seeds). Esp. tropical and warm temperate, esp. Africa and America.

4. Faboideae Rudd - Isoflavonoids [pterocarpans and isoflavans] +; sieve tubes with spindle-shaped non-dispersive protein bodies; median member of C external; pleurogram 0, linea fissura 0, hour-glass mesotestal cells [below palisade exotesta] +, raphe shorter than the antiraphe, hilum with a hilar groove, tracheid bar [group of tracheids just below surface of hilum] +; embryo curved, radicle long, cotyledons not investing radicle, cotyledon areole +. - 476/13855 (abbreviations - BAPH = baphioids, DAL = dalbergioids s.l., GEN = genistoids, IRLC = Inverted Repeat Loss Clade, MILL = Indigofereae + millettioids, MIRB = mirbelioids, ROB = robinioids, SWAR = swartzioids): Astragalus (2400-3270: IRLC), Indigofera (700: aff. MILL, mesifixed hairs), Crotalaria (700: GEN), Mirbelia s.l. (450: MIRB, see Crisp & Cook 2003a, b), Tephtosia (350: MILL), Desmodium (300: MILL), Aspalanthus (300: GEN), Oxytropis (300: IRLC), Adesmia (240: DAL), Trifolium (240: IRLC), Rhynchosia (230: MILL), Lupinus (200 pollen extruded as threads - for Andean diversification, see Hughes & Eastwood 2006), Aeschyomene (160: DALB), Hedysarum (160: IRLC), Lathyrus (160: IRLC), Vicia (160: IRLC, pollen presented on stigmatic brush), Dalea (150: DALB), Eriosema (150: MILL), Lotononis (150: GEN), Millettia (150: MILL), Vigna (150: MILL), Swartzia (140: SWAR)), Daviesia (135: MIRB), Lonchocarpus (MILL), Machaerium (130: DALB), Onobrychis (130: IRLC), Ormosia (130: unplaced), Lotus (inc. Coronilla: 125: ROB, for floral and inflorescence morphology esp. in Loteae, see Sokoloff et al. 2007a), Lonchocarpus (120: MILL), Erythrina (110: MILL), Gastrolobium s.l. (110: MIRB), Mucuna (105: MILL), Pultenaea (100: MIRB, for limits, see Orthia et al. 2005), Genista (90: GEN), Medicago (inc. Trigonella, 85: IRLC), Swainsonia (85: IRLC), Caragana (75: IRLC), Jacksonia (75: MIRB), Ononis (75: IRLC, see Liston 1995), Zornia (75: DALB), Argyrolobium (70: GEN), Arachis (70: DAL - see Krapovickas & Gregory 2007 for a revision), Brogniartia (65: GEN), Cytisus (65: GEN), Bossiaea (60: MIRB), Canavalia (60: MILL), Clitoria (60: MILL), Dolichos (60: MILL), Galactia (60: MILL), Phaseolus (60: MILL, for phylogeny, see Delgado-Salinas et al. 2006), Sesbania (60: ROB), Derris (55: MILL), Trigonella (55: IRLC), Lessertia (50: IRLC), Psoralea (50: MILL), Sophora (50: GEN). Esp. (warm) temperate, but world-wide.

• Fabaceae may be recognised by their compound, stipulate leaves whose leaflets are entire and often have wrinkled pulvini, and flowers with a single carpel, free petals (or two may be connate), and stamens either all similar in length, or if dissimilar, not regularly alternating longer and shorter. The fruit is usually dry and elongated, the often flattened seeds being borne in a single rank.

• "Caesalpinioideae" and Cercis and its relatives are woody plants that may be recognised by their usually once-compound leaves and monosymmetric flowers with the odd petal adaxial and inside the others in bud; the hypanthium is often well-developed.

• Mimosoideae are also mostly woody plants that have leaves that are often bicompound and with extrafloral nectaries. The flowers are rather small and are borne in groups, all flowers opening more or less simultaneously; the long, coloured filaments form the visually attractive part of the inflorescence. Both the calyx and corolla are valvate, and both (and often the bases of the filaments) are connate. The flattened seeds have a fine, U-shaped line, on each surface, although this may be hard to see.

• Faboideae are usually herbs or vines and often have once-compound leaves. The flowers are monosymmetric with the odd petal being adaxial and outside the others; the papilionoid flower consists of adaxial banner, paired wings and two connate petals forming the keel. The stamens have more or less connate filaments and the fruit is often explosively dehiscent, simultaneously splitting down both sides and the two valves twisting.

Fabaceae are a notably speciose clade, particularly the branches with Mimosoideae and Faboideae (Magallón & Sanderson 2001), and contain ca 9.4% of eudicots; it has been estimated that some 16% of all woody species in Neotropical rainforests are members of this family (Burnham & Johnson 2004). Indeed, Fabaceae are the most speciose family in lowland tropical rainforest and also drier forest types in America and Africa (Gentry 1988). They began diversifying ca 60 Ma ago (the stem group is little older), and the major clades had separated by ca 50 Ma. Thus the clade [part of Caesalpinioideae + Mimosoideae] may date to 54 ± 3.4Ma before present and crown group Mimosoideae to 44 ± 2.6Ma before present (Lavin et al. 2005; c.f. Wikström et al. 2001: dates a little older). Although there are several transoceanic disjunctions within Fabaceae, 51/59 of these are only 1-22 Ma old (Schrire et al. 2005). Lavin et al. (2004) and Schrire et al. (2005) suggest that it is more profitable to think of diversification and distribution of Fabaceae in terms of vicariance of biomes rather than of the classic geographical areas; the North Atlantic land bridge may have been important in the Caenozoic dispersal of the family (Lavin et al. 2000).

Within Faboideae, a number of divergences can be dated, including the separation of the speciose Astragalus from Oxytropis 16-12 Ma, although diversification in both is relatively recent; radiation in the speciose aneuploid New World neoastragalus species started ca 4.4 Ma (Wojciechowski 2004). Lupinus has a recent (within the last two My) Andean diversification of over eighty species probably associated with the migration of bumble bees, a major pollinator, into Andean South America at most six Ma (Hughes & Eastwood 2006). Inga (Mimosoideae), with some 350 species and an important component of Neotropical forests, also seems to have diversified within the last two Ma (Richardson et al. 2001b).

There are numerous often quite specific associations of insects and Fabaceae. Caterpillars of Lycaenidae-Riodininae-Riodinini, Lycaenidae-Curetinae and especially Lycaeninae-Lycaenini butterflies are found on Fabaceae (Ehrlich & Raven 1964; Fiedler 1991, 1995), as are pierid butterfly larvae (Coliadinae, Dismorphiinae: some 260 species in 15+ genera, about a quarter of the records - see also Brassicales and Santalales, Braby & Trueman 2006). The diversity of caterpillars - especially that of "basal" butterfly groups - on Fabaceae is such that Janz and Nylin (1998) and Braby and Trueman (2006) suggested that Fabaceae might be the springboard for hostplant diversification of butterflies feeding on angiosperms in general (see also the introduction to Fabales). In another variant of insect-plant relationships, the flowers of Crotalaria are visted by Danainae butterflies and day-flying Ctenuchidae moths because the pyrrolizidine alkaloids they contain are used as the basis of the pheromones of these strikingly-coloured lepidoptera (also Asteraceae, and wilting plants of some Boraginaceae: Edgar et al. 1974; Pliske 1975; Boppreé 1986); Crotalaria is also associated with arctiid moths such as Utetheisa, and its secondary metabolites provide defence for the young, pheromones for the adult, etc., etc. (Eisner & Meinwald 1995). The jumping plant lice Psyllidae-Arytaininae are often found on Fabaceae-Faboideae, especially genistoids, and especially in the Mediterranean-Macaronesian region, while Psyllidae-Acizzinae are associated with Mimosoideae in the Southern Hemisphere (Percy 2003; Percy et al. 2004). In Acacia s. str., well over 200 species of Phlaeothripidae (thrips) form galls and other habitations on species of subgenera Juliflorae and Plurinerves, the taxa involved lacking reticulate venation; galls occur less frequently on species of subgenus Acacia, which has only a single main vein, unlike the others (Morris et al. 2002; Crespi et al. 2004 and references); Fabaceae in general have a large number of asssociations with gall-forming insects (Mani 1964; Gagné 1989). Bruchids (Chrysomeloidea-Bruchidae/Bruchinae), a large group with some 1700 species, have larvae that are specialized seed-eaters. They were perhaps first associated with Faboideae, and then moved on to other groups following the chemistry of the plants involved (esp. Kergoat et al. 2005a, b; see also Johnson 1989, 1990 [Acanthoscelides], Birch et al. 1989 [chemistry of the interaction], and Janzen 1969 [the complexity of the association between plant and weevil]). Two clades, made up largely of New World Acanthoscelides and predominantly Old World Bruchidius, dominate, and they may have radiated contemporaneously with their hosts, largely Fabaceae-Mimosoideae and Faboideae (elsewhere thye are also found on some Malvaceae, in particular); they can detoxify the non-protein amino acid, L-canavanine (see below), that occurs only in a large clade of Faboideae. The pattern of association of bruchid groups with mimosoids is interesting; individual bruchid genera tend to be found on adajcent pectinations of the mimosoid phylogenetic tree (Kergoat et al. 2007).

Finally, less widespread but very well known is the association of ants with some members of the old Acacia subgenus Acacia (= Vachellia). These includes the swollen-thorn acacias such as V. sphaerocephala which provide protein-rich Beltian bodies at the ends of the leaflets (the leaves have many leaflets, even for Acacia s.l.) as food for the Pseudomyrmex ants, and there are swollen stipular thorns that serve as their homes (Janzen 1974b). The ants also take nectar from extrafloral nectaries, and in the case of these close associations, the nectar produced is sucrose free, the ants lacking invertase needed to break down sucrose (Kautz et al. 2009). Extrafloral nectaries are notably common in Mimosoideae, being found towards the base of the petiole and sometimes on petiolules, and the nectar usually has sucrose. Such nectaries are rare in "Caesalpinioideae", or they are represented by tufts of hairs (Pascal et al. 2000). However, they characterise a major clade within Senna, and there it has been suggested that they are a "key innovation" involved in plant defence and in the diversification of that clade, which is much more speciose than its sister taxon (Marazzi et al. 2006; Marazzi & Sanderson 2008).

Fabaceae are well known for their association with nitrogen-fixing bacteria which grow inside irregular, pinkish-coloured nodules on the roots. Although the nodulating bacteria are mostly members of the proteobacteria α-2 subclass, they do not form a monophyletic group; Agrobacterium (crown gall tumour) and others are also members of this clade (Sprent 2001; Moulin et al. 2001). Flavonoids, produced in great diversity by Fabaceae, attract the bacteria to the roots and trigger the synthesis of Nod(ulating) factors by the bacteria (Chen 2007: remarkably, in a few nodule-forming α-proteobacteria such as the photosynthetic Bradyrhizobium there is no nodABC gene producing these factors [Giraud et al. 2007]).The plesiomorphic infection morphology shows persistent infection threads and long-lived nodules (see also Parasponia - Cannabaceae), while more derived may be the absence of infection threads, the mitosis of infected cells, and a short life span for the nodules (de Faria & Sprent 1995; see also Corby 1988 & Sprent 2005 for nodule distribution). Nodulation is especially widespread in Faboideae and Mimosoideae, but less common in "Caesalpinioideae" - although occurring in taxa like Chamaecrista. It appears that the acquisition of the ability to nodulate has occurred more than once, perhaps even several times, with the family. Within Faboideae, the nodulating Swartzia and immediate relatives may form a clade sister to the rest of the subfamily (support is weak, see Ireland et al. 2000; Pennington et al. 2000; Lavin et al. 2005). Some other Faboideae in clades that are also separate from that including the bulk of the subfamily do not nodulate. However, most Faboideae are nodulators (Sprent 2000, 2001, 2007). Generally speaking, symbiont specificity is greatest in the IRLC clade (Faboideae, see below), although genera like Astragalus are exceptions (Howard & Wojciechowski 2006), and overall distribution of nodulation within Fabaceae and variation in nodule morphology are of some systematic significance (see also J. J. Doyle 1994, 1998; J. J. Doyle et al. 1997).

Evolution in nodulation in Fabaceae is still not well understood. Interestingly, a number of Fabaceae, perhaps especially non-nodulated members and including Acacia s. str. and "Caesalpinioideae" like Dicymbe, are ectomycorrhizal (Sprent & James 2007 for literature). Futhermore, Burkholderia, a unrelated ß-proteobacterium, is also an effective nitrogen-fixing symbiont of at least some Faboideae and in Mimosa, and other ß-proteobacteria can form nodules, albeit ineffective, with Mimosoideae (Moulin et al. 2001; Elliott et al. 2007 and references); this kind of association may be quite common in the tropics (Sprent 2007).

Rusts show interesting patterns of distribution on Fabaceae. Uromyces is found predominantly on herbaceous Faboideae, but also on Bauhinia and one or two other woody taxa (being found, along with related genera, on Acacia in Australia alone), while Ravenelia is found on woody members of the family, i.e. "Caesalpinioideae", but also especially Mimosoideae (Savile 1976, 1979a, b; El-Gazzar 1979). In a number of species of Ravenelia the teliospores, thick-walled spores in which nuclear fusion and then meiosis occur, are aggregated into groups, and these telial heads seem to mimic the groups of pollen grains (polyads) that are common in Mimosoideae. Stingless Trigona bees may pick up the telial heads and polyads as they forage for pollen. However, Ravenelia is only very rarely found on Australian Acacia; the distributions of rusts, acacias and trigonid bees all break at about Wallace's Line.

The diversity of secondary metabolites in Fabaceae, perhaps especially in Faboideae, is remarkable; for instance, about 28% of all known flavonoids and about 95% of the isoflavonoid aglycone structures - over 1,000 alone - identified in plants are known from Fabaceae, and isoflavonoids are restricted to Faboideae. Waterman (1994) discusses cost/benefit aspects of the production of these secondary metabolites. Isoflavonoids may be phytoalexins involved in plant defence, and are perhaps also involved in nodulation (Hegnauer & Grayer-Barkmeijer 1993; Reynaud et al. 2004; Samac & Graham 2007). In general Fabaceae have a very distinctive nitrogen metabolism. Non-protein amino acids are common, and nitrogen in the xylem sap is transported as a mixture of amino acids, amides, and sometimes also ureides; very little is transported as nitrate as is common in other plants. Wojciechowski et al. (2003, 2004, see also Wojciechowski 2003) note than the evolution of some non-protein amino acids are taxonomically interesting, thus canavanine production seems to have originated in the ancestor of one major subgroup of Faboideae (it includes mirbelioids, millettioids, robinioids, and the large IRLC clade and may date to 54.3 ± 0.6Ma before present - Lavin et al. 2005: see tree above); canavanine and alkaloid production are mutually exclusive. Synthesis of canavanine may be an efficent way for the plant (the seed in particular) to store nitrogen, but canavanine is no more efficient in this than the ordinary amino acid arginine. However, canavanine is an antimetabolite toxic to a wide variety of animals, although bruchids seem not to be affected by it, and it is sometimes lost, as in Phaseolus (Bell et al. 1978). Pea albumin, a small sulphur-rich peptide with insecticidal properties, is known only from Faboideae where it may be a synapomorphy for the [hologalegina + millettioid] clade, being lost in some/all robinioids (Louis et al. 2007). These and other metabolites are also involved in the associations of insects and fungi with Fabaceae (see above: Wink & Mohamed 2003).

Although most Fabaceae have once or twice compound leaves, leaflets with entire margins, and pulvini associated with leaves and leaflets, there is extensive variation on this theme; palmate lewaves occur in Lupinus. In Acacia s. str. (the old subgenus Phyllodinae), the leaves of the mature plant are much modified and are often called phyllodes, but seedlings and regeneration shoots may have once or twice compound leaves. Kaplan (1980) suggested that these "phyllodes" were not equivalent to the petiole of a compound leaf, but to the leaf as a whole. In early development, there were normally two, adaxial meristems that proceeded to develop the leaflets/pinnae; the leaflets/pinnae became lateral in position by secondary orientation. In phyllodinous leaves there was a single, broader adaxial meristem that developed to produce the entire leaf (Kaplan 1980). There is no possibility for secondary orientation in such leaves, and they are flattened at right angles to the plane of flattening of a normal leaf. In some species of Acacia phyllodinous leaves are densely set along the stem, but only some are associated with stipules and buds, others lacking both. A number of Faboideae (e.g. Vicia, Pisum) are tendrillar vines, the tendrils being modified terminal leaflets; in Lathyrus aphaca the photosynthetic function of the leaf is taken over by the large stipules, the rest of the leaf being tendrillar, while L. nissolia lacks tendrils and has a phyllodinous leaf. The leaves may be reduced to a single more or less connate pair of leaflets, as in Bauhinia, named after the botanical brothers Caspar and Jean Bauhin (seeQwens 2000 and Champagne et al. 2007 for evidence that they are secondarily simple). Leaf development is associated with the activity of the KNOX1 gene, as is common in plants with compound leaves, however, in the IRLC clade (Faboideae, see below) the KNOX1 gene is not expressed, the FLO/LFY gene, normally a floral regulatory gene, being expressed instead; the leaves have no pulvini (Champagne et al. 2007).

Extrafloral nectaries are notably common in Mimosoideae, being found towards the base of the petiole and sometimes on petiolules. Such nectaries are rare in "Caesalpinioideae", or they are represented by tufts of hairs (Pascal et al. 2000). However, they characterise a major clade within Senna, and there it has been suggested that they are a "key innovation" involved in plant defence and in the diversification of that clade, which is much more speciose than its sister taxon (Marazzi et al. 2006; Marazzi & Sanderson 2008).

Some species of Mimosa and other genera have leaves that are sensitive to touch, stimulus transmission occurring as membrane depolarisation is propagated down the petiole and along the stem; folding of the leaf is caused by tugor changes in the cells of the pulvini at the bases of the leaf and leaflets (for the anatomy of the pulvinus, which has an endodermis, see Rodrigues & Machado 2007). In taxa like Albizzia (Samanea) saman, similar movements occur as the leaflets fold towards the evening when the light is failing, or just when there is heavy cloud cover, this behaviour being responsible (in some tellings of the tale) for its name, the rain tree. In Desmodium gyrans the single pair of lateral leaflets move intermittently without being touched, and the speed of movement increases as the temperature increases.

The "pea flower" or "papilionaceous" floral morphology (see below) and its variants are common in "Caesalpinioideae" and Faboideae. The flowers of Cercis are only superficially similar to those of Faboideae (Tucker 2002a), although both are more or less papilionoid and are similar functionally. The papilionoid flower is characterised by the more or less erect ultraviolet-absorbing banner petal which sometimes has colour patterning, the two wing petals, and the paired interlocking keel petals enclosing the stamens. In Caesalpinia the abaxial sepal may be colorful and look like a keel. In Hardenbergia violacea the colour patterning on the standard may mimic an anther (Lunau 2006). Papilionaceous flowers encompass a variety of morphologies; as Bruneau et al. (2005, p. 201) note of caesalpinioid legumes, "zygomorphy is expressed as a multitude of homoplasious morphs". Hardly surprisingly, flowers of Fabaceae attract a diversity of pollinators that visit the flowers for various rewards. Pollination in "Caesalpinioideae" is predominantly by polylectic bees, while oligolectic bees are commoner pollinators of Mimosoideae and Faboideae. However, this may be as much a reflection of where these subfamilies occur, since oligolectic bees are more speciose in (warm) temperate regions, especially Mediterranean climates (Michener 1979), and that is where the other two subfamilies are particularly common. Interestingly, within the tropics bees seem to be commonest in the New World tropics (Michener 1979), and woody Fabaceae are especially diverse there.

There is interesting variation of pollination types within Faboideae. When the androecium is monadelphous, i.e. the filaments of all the stamens are connate, the pollinator reward is often pollen, and this can be delivered by a pump secondary pollen presentation mechanism. The bee lands on the keel, and the style then forces pollen out of the keel in a tooth paste-like strand, as in Lupinus. If diadelphous - nine stamens are connate and a single adaxial stamen is free - the reward is often nectar, the nectary lying between the filament tube and gynoecium. Other taxa like Cytisus have explosive pollination where the style is held under tension which is released as the style curves when the pollinator lands; such flowers can be visited only once. Vicia is another genus that has secondary pollen presentation: The pollen is presented to the pollinator on a stigmatic brush. Erythrina is pollinated by both perching and hovering (humming) birds, and both floral morphology and how the flowers and inflorescences are held varies according the requirements of these different visitors (Bruneau 1997). Flowers of Swartzia (sister to all other Faboideae) have very distinctive flowers. They often have an entire calyx, only a single petal, numerous free and dimorphic stamens, a stalked ovary, and they lack nectar; some taxa have more than one carpel. Pollination here may be by euglossine bees (see Torke & Schaal 2008 for a phylogeny).

The large genus Cassia ("Caesalpinioideae") has been divided into three. Senna is enantiostylous and lacks bracteoles. Its anthers are porose, pollination being by buzz pollination, and there are three stamen morphs. There are three adaxial staminodes, four middle medium-sized stamens from which pollen is taken by the bees, and three longer abaxial stamens pollen from which is actually involved in pollination (see Marazzi & Endress 2008 for development). Cassia s. str. also has three stamen morphs. The filaments are curved and the anthers have slits or basal pores. Finally, the largely herbaceous Chamaechrista is enantiostylous; the stamens have two morphs in different whorls, the filaments are short, and the porose anthers have a velcro-like line of hairs. Buzz pollination is likely here, too. Bats may also be pollinators, as of Parkia (Mimosoideae) (see e.g. Arroyo 1981; Lewis et al. 2000 [esp. "Caesalpinioidae"] for summaries of pollination in Fabaceae).

Many Mimosoideae have a very different floral morphology to that of other Fabaceae. Here numerous small and secondarily polysymmetric flowers are aggregated into attractive units, all flowers opening at about the same time. Pollen grains are frequently aggregated into polyads which are caught in the cup-shaped stigma that is of the appropriate size for the polyad of that species, and there are also about as many ovules as there are pollen grains in the polyad (Kenrick 2003 for references and the implications of this pollination mechanism for the breeding system). In Calliandra s. str. the polyads have an associated sticky mucilage body by which they are attached to the pollinator, but the stigma is much larger and capitate and the polyads adhere to its surface (Greissl 2006, c.f. in part Teppner 2007). For polyads, anther dehiscence, etc., in Mimosoideae, see Teppner (2007) and Teppner and Stabentheiner (2007) and references.

The "normal" (for flowering plants) floral orientation of Mimosoideae with the median sepal adaxial and the median petal abaxial may be secondary. Although the normal orientation is also found in some caesalpinioids like Ceratonia, the inverted orientation occurs in both Cercis and Bauhinia (see Tucker 1989; Herendeen et al. 2003; Luckow et al. 2005), other caesalpinioids, and Faboideae. Polysymmetry in the African Cadia (Faboideae-genistoid), a "reversal", is the result of dorsalization of the flower (the same basic principle as peloria in Antirrhinum: Citerne et al. 2006). Details of hypanthium evolution within Fabaceae are unclear; it seems to have become much reduced and lost several times.

The legume s. str. is a single carpelate fruit that dehisces explosively along both sutures, the two valves twisting as they separate. The legume is common in European-North American Faboideae, but it occurs also in the other subfamilies, including in Bauhinia, the clade sister to the rest of the family. The fruits of Cercis, in the same clade, are not explosively dehiscent, but are otherwise similar; they are also typologically rather similar to the fruits of Myristica, the nutmeg! However, overall there is a great diversity of fruit morphology in the family, including variously winged fruits, fleshy fruits, fruits breaking up into single-seeded units in different ways, and fruits modified for animal transport with spines and hooks, for example, the velcro-like hooks on Desmodium (hence its common name, beggar's ticks). In Trifolium the calyx and corolla are both involved in fruit dispersal mechanisms. Arillate seeds are common, and seeds that have red and black color patterns such as Abrus, Erythrina and the sometimes pluricarpelate Pithecellobium are well known; these mimic the color contrasts of red aril and black seed of some other Fabaceae, and also other plants. There are also seeds with fleshy coats. However, in many taxa, especially those with explosively dehiscent fruits, the seed coat is very hard and may need scarification if germination is to occur (for fruits and seeds, see Corner 1951; van der Pijl 1956; Kirkbride et al. 2003; etc.).

Knoblauch et al. (2001) discuss the possible mode of action of the distinctive spindle-shaped non-dispersive protein bodies (= forisomes), found commonly in Faboideae, in blocking the pores of the sieve plates when turgor pressure changes; the protein bodies change shape depending on the concentration of Ca²⁺ ions (Peters et al. 2007).

Cytisus purpureus forms a well-known graft hybrid with Laburnum anagyroides (+ Laburnocytisus adamii; see Herrmann 1951 for another example); the epidermis alone is Cytisus tissue, and any seeds, being derived from cells from deeper layers, will give Laburnum plants. However, the graft hybrid often breaks down, resulting in branches that are pure Laburnum anagyroides or pure Cytisus purpureus.

Fabaceae are monophyletic based on both molecular and morphological analyses, although in the past the family has sometimes been dismembered into three (e.g. Takhtajan 1997). In all recent reconstructions of phylogeny, Caesalpiniaceae/Caesalpinioideae are paraphyletic, Mimosaceae/Mimosoideae largely monophyletic, and Fabaceae/Faboideae s. str. monophyletic. Within "Caesalpinioideae", [Cercis + Bauhinia s.l.] alone may be sister to the rest, and because of the paraphyly of the subfamily, to all other Fabaceae (e.g. J. J. Doyle et al. 2000 and references; Bruneau et al. 2001), although they are placed sister to Detarieae s.l. (inc. Cynometra) with moderate support in other phylogenies (Wojciechowski et al. 2004; Lavin et al. 2005; Forest et al. 2007b), this combined clade being sister to the rest of Fabaceae. Recent studies (Bruneau et al. 2008a, b) find that Detarieae s.l., Duparquetia, and/or Cercideae are all candidates for being sister to the rest of the family. Morpholology and anatomy support such relationships. Thus all three lack vestured pits (they are also absent in Cassieae), but such pits are common in the rest of the family. Umtiza is excluded from Detarieae, and forms a clade with taxa like Gleditsia, Gymnocladus and Ceratonia, several of which are dioecious and have smallish, greenish flowers sometimes with a poorly differentiated calyx and corolla - not plesiomorphic features (Herendeen et al. 2003a). Extrafloral nectaries are rare in "Caesalpinioideae", or are represented by tufts of hairs (Pascal et al. 2000). However, they characterise a major clade within Senna (Marazzi et al. 2006) and may be involved in the diversification of that clade compared with immediately related but nectary-less clades (Marazzi & Sanderson 2008). Asymmetry (enantiostyly) is likely to have been acquired once within Senna, although it has also subsequently been lost. For additional information on relationships in caesalpinioid legumes, see Herendeen et al. (2003b) and Lavin et al. (2005).

Mimosoideae are very largely monophyletic, however, some ex-caesalpinioids (e.g. Dimorphandra) with small flowers in spikes or panicles all opening more or less together may have to be included with them (Wojciechowski 2003); they show considerable similarity in wood anatomy (Evans et al. 2006). Ingeae are embedded in Acacieae or vice versa (e.g. Clarke et al. 2000; Robinson & Harris 2000; Miller & Bayer 2000, 2001; Luckow et al. 2003; Jobson & Luckow 2007; Brown et al. 2008). What used to be called Acacia subgenus Acacia (now = Vachellia), which includes the bull's horn acacias, seems to be monophyletic, but Acacia s.l. is polyphyletic. As Maslin (2001) observed sadly of the 955 or so species then placed in Acacia for the Flora of Australia treatment, "we are obliged to present the flora treatment in the absence of a more meaningful classification". However, the argument now is over what names to call the bits into which Acacia s.l. is to be divided (see above). Murphy et al. (2000, 2003) and Miller et al. (2003) discuss the phylogeny of Acacia s. str., and Miller and Bayer (2003) that of Vachellia and Senegalia. Siegler (2003) summarized the phytochemistry of the complex, and Evans et al. (2006) detailed wood anatomy. Kergoat et al. (2007) noted what bruchids had to say about the systematics of Acacia s.l., and the disribution of rusts on Acacia s.l. is also of interest (see above).

Faboideae are monophyletic, but the recognition of this subfamily (and Mimosoideae) makes Caesalpinioideae paraphyletic. A topology (simplified) for Faboideae in general including [swartzioids: SWAR [Cladrastis, etc., [genistoids: GEN, [Amorpheae + dalbergioids = dalbergioids s.l.: DAL], [baphioids: BAPH [mirbelioids: MIRB [[Indigofereae + millettioids: MILL] [robinioids: ROB + Inverted Repeat Loss Clade: IRLC]]]]]]] seems moderately well supported (Wojciechowski 2003; McMahon & Sanderson 2006). (For the contractions, see the subfamilial generic list above; note that the [robinioids + Inverted Repeat Loss Clade] clade is called the hologalegina clade.) Within Faboideae, Swartzieae, woody, nodulators, lacking bracteoles, with very variable flowers and arillate seeds, may be sister to the rest, but support is weak and the exact circumscription of Swartzieae is unclear (Ireland et al. 2000; Pennington et al. 2000; Lavin et al. 2005); it may well be largely restricted to Swartzia. Some other Faboideae clades (they include some ex-Swartzieae) separating early from the bulk of the subfamily also do not nodulate (Sprent 2000, 2001). Many Faboideae have a 50kb inversion in their chloroplast genome; Sophora, Myrospermum, Swartzia and their relatives lack this inversion (e.g. see Doyle et al. 1996). There has also been the loss of the 25kb chloroplast inverted repeat; this characterises a largely temperate, epulvinate, herbaceous and very speciose group, although Wisteria is also a member of this clade (the IRLC - the Inverted Repeat Lacking Clade, see Wojciechowski 2003 and references): genera involved have a star in the list above. All members of this clade lack the clpP intron, while the rps12 intron has been lost from all members of this clade examined except Wisteria, Callerya and Afgekia - but not Glycyrrhiza - c.f. the tree above (Jansen et al. 2008). Desmodium and possibly related genera have also lost the rps12 intron as well as the srp12 intron (Doyle et al. 1995; Bailey et al. 1997; Jansen et al. 2008). Both the rps16 andycf 4 genes are absent - presumably lost - in the majority of tribes of Faboideae (Doyle et al. 1995; Jansen et al. 2007). These and other changes in chromosomal organisation provide a considerable amount of phylogenetic structure, especially in Faboideae. Furthermore, alkaloid type, or presence/absence, presence of non-protein amino acids, especially canavanine (see tree above), and of pea albumins all characterise major clades in Faboideae (Wink & Mohamed 2003).

Within Faboideae Dalbergioids often have aeschyomenoid root nodules and glandular punctate leaves; genistoids commonly have quinolizidine alkaloids (for which, see Wink 1992); and Cladrastis and its relatives have inflorescence bracts at the base of the inflorescence. Members of the millettioid clade have a pseudoracemose inflorescence with more than a single flower at each node; Indigofereae, its sister clade, has true racemes (Wojciechowski et al. 2004). Schrire et al. (2009) disentangle relationships within Indigofereae, finding considerable phylogenetic structure (i.a. there are four major clades within Indigofera) that can be linked with both morphology and ecology. For the phylogeny of dalbergioid legumes, see Lavin et al. (2000) and Ribeiro et al. (2007), for that of Amorpheae and their floral evolution (petals may be lost, or all look rather similar; a stemonozone may be developed, i.e. there is a tube formed by the adnation of filaments to the corolla; etc.), see McMahon and Hufford (2005) and McMahon (2005) and references, for that of Robinia and its relatives, see Lavin et al. (2003), of Crotalarieae, see Boatwright et al. (2008c), and for that of Psoraleae, see Egan and Crandall (2008). The IRLC clade is characterized not only by the loss of the inverted repeat and the clpP intron, but the compound leaves lack pulvini and the KNOX1 gene is not expressed early in development, although the (normally floral) FLO/LFY gene is (Champagne et al. 2007). Within the IRLC, Astragalus is an extremely speciose genus usually found in drier areas of both hemispheres, and a number of taxa have leaf rhachis spines. Extensive phylogenetic studies (e.g. Wojciechowski 1993, 2004; Kasempour Osaloo et al. 2004; Scherson et al. 2004) show most New World taxa are aneuploid (n = 11-15) and form a monophyletic group, other species are base 8. Oxytropis is sister to Astragalus. In Trifolium the calyx and corolla involved in fruit dispersal mechanisms, and again the American species form a monophyletic group (Ellison et al. 2006; Liston et al. 2006). Phylogenetic relationships within Medicago have turned out to be highly reticulating (Maureira-Butler et al. 2008).

For variation in general patterns of floral development in Fabaceae, see Prenner and Klitgaard (2008b). The parts of the flowers of many Fabaceae develop in the unusual sequence sepals-carpels-petals-outer stamens-inner stamens, and there are other distinctive features of their development (e.g. Champagne et al. 2007; Feng et al. 2006; Wang et al. 2008).

Fabaceae s.l. are often referred to their own order, as in both Cronquist (1981) and Takhtajan (1997). They can be confused with Connaraceae (Oxalidales), although the latter lack stipules, their flowers are polysymmetric and have stamens of two distinctly different lengths, and their gynoecium is frequently multicarpelate. Fabaceae have also been linked with Sapindaceae, here in the rosid II group and also with compound leaves (e.g. Dickison 1981b and references), but there is little support - and none molecular - for such an association.

For more information see Polhill and Raven (1981), Crisp and Doyle (1995), Doyle and Luckow (2003), and Lewis et al. (2005: well-illustrated summary of geographic distributions, etc., of all genera; some taxa are para/polyphyletic), all general. For chemistry, about which a great deal is known, see Southon (1994), and Hegnauer and Hegnauer (2001) and references, and for the evolution of secondary metabolites, see Wink and Mohamed (2003: particularly useful) and Wink (2003); see also Hegnauer and Grayer-Barkmeijer (1993: polysaccharides) and Harborne and Baxter (1999: flavonoids). For wood anatomy, see Baretta-Kuipers (1981) and Gasson et al. (2003), seed anatomy, Corner (1951), seed and fruit morphology, Gunn (1991) and Kirkbride et al. (2003), embryo suspensors, Lersten (1983) and Tucker (1987), general floral and inflorescence morphology, Endress (1994b), floral development, Tucker (1996 and references, 2003), and cotyledon areoles, Endo and Ohashi (1998).

Surianaceae + Polygalaceae: ?

SURIANACEAE Arnott - Carpels separate, styles ± gynobasic. - 5/8 Mostly Australian, also Mexico, Suriana pantropical.

For relationships, see Forest et al. (2007b); [[Recchia + Lundellia] [Suriana [Cadellia + Stylobasium]] seems to be the cladistic structure in the family. The vegetatively "atypical" Suriana is the only genus whose embryology has been studied and the whole family is little known chemically.

The family is vegetatively very heterogeneous, although quite homogeneous in wood anatomy (Webber 1936). The exotesta of Suriana is described as being green (Rao 1970).

Although the sieve tube plastids of Stylobasium are very distinctive, with both protein and one or a few starch grains (Behnke et al. 1996), there seems little reason to recognise Stylobasiaceae as a monotypic family; Suriana has more ordinary plastids with starch grains alone.

Surianaceae (Suriana was the only genus included) were placed in Rosales by Cronquist (1981) and in Rutales by Takhtajan (1997), and the other genera, and sometimes also Suriana itself, were included in Simaroubaceae (here in Sapindales, which include Rutales).

For more information, see Gutzwiller (1961: general), Gadek and Quinn (1992: pericarp), Ito and Tobe (1994: embryology), Crayn et al. (1995: relationships), Schneider (2006: general), and Bello et al. 2007 (floral development: Suriana only).

POLYGALACEAE Hoffmannsegg & Link - Median abaxial C clawed, ± boat-shaped/keeled; A 8, median adaxial A 0, pollen polycolporate; K caducous; carpels connate; fruit a berry. - Ca 21/940 - four tribes below. World-wide, except the Arctic and New Zealand.

1. Xanthophylleae Chodat - Plants Al accumulators; glands at nodes, (conspicuous domatia on leaves); G [2], placentation parietal, 2 or more ovules/carpel, stigma small, bilobed; testa multiplicative. - 1/95. Indo-Malesia.

The Rest: A ± adnate to petals, variously connate, often monadelphous, anthers opening by short apical slits, 1 ovule/carpel.

2. Polygaleae Chodat - Flowers with two adaxial lateral K = wings, 2 abaxial lateral K, minute, two connate adaxial C = the standard, abaxial C = the keel, often fringed, 2 abaxial-lateral C minute; G [2], stigma complex, asymmetric; fruit dry [an often flattened capsule, samara]; seed with caruncle [around hilum]. - Ca 13/830: Polygala (325), Monnina (180), Muraltia (120), Securidaca (80). World-wide, except Arctic and New Zealand.

3. Carpolobieae Eriksen - A (4) 5; G [3]. - 2/6. Tropical Africa.

4. Moutabeae Chodat - Glands on leaves; G [3-8]; seed with funicular aril. - 4/15. Tropical America, New Guinea to New Caledonia.

• Polygalaceae are woody to herbaceous plants that may be recognised by their often yellowish-drying simple and entire leaves; there are quite frequently paired glands at the nodes or glands or domatia on the leaf blades. The flowers are usually monosymmetric and are more or less papilionaceous. They range from having five rather large sepals and five petals - not strongly papilionaceous - to having two of the sepals much enlarged and coloured (the "wings" of the flower), two of the petals much reduced, and the three remaining petals modified as "standard" and "keel" - strongly papilionaceous. There are two or more carpels and a single style.

The distinctive Paleosecuridaca curtisii has recently been described from the Palaeocene of North Dakota; although in gross morphology its fruits are remarkably like those of Securidaca and the seeds have a testa with a well developed palisade layer, there are two seeds per carpel (Pigg et al. 2008b).

Overall floral variation in Polygalaceae is very considerable. In at least some North American species of Polygala pollen is presented on the sterile lobes of the asymmetric stigma (secondary pollen presentation: Weekley & Brothers 1996; see Castro et al. 2008 for details). In a study of ant dispersal in Polygalaceae, which is quite common in Polygaleae, it seems that caruncles may be an apomorphy of Polygaleae, although chalazal arils have also evolved more than once in this clade, and they and caruncles have been lost, too (Forest et al. 2007b). Evolution of these structures, which are functionally elaiosomes that attract ants to the seeds, is suggested to have occurred (69.9-)54-50.5(-35.2)Ma before present, well after initial diversification of the ant clades attracted to them (Brady et al. 2006 and references). Muraltia, with some 120 species mostly found in the Cape region, appears to have diversified relatively recently, mostly within 10 Ma (Forest et al. 2007a).

Epirixanthes (Polygaleae) is an echlorophyllous mycoheterotroph. Although genera like Xanthophyllum may have paired glands at the nodes, other genera seem to lack anything faintly comparable with stipules, and what stipules "are" and where they are to be placed in this part of the tree is uncertain.

The flower in Polygalaceae is quite differently constructed from that of Fabaceae, although quite often both looking and being functionally very similar (but see Prenner 2004d). However, the flowers of Polygala, particularly similar in overall appearance to those of some Fabaceae, are unlikely to represent the plesiomorphic condition of the family, and overall floral variation in Polygalaceae is very considerable. The tricolpate pollen of Balgooya is probably derived; some Polygalaceae such as Heterosamara have asymmetric, almost boat-shaped pollen grains (Banks et al. 2008).

Of the four groups mentioned above, Moutabeae may be paraphyletic (Persson 2001: trnL-F), although adding rbcL data suggests they are monophyletic (Forest, in Eriksen & Persson 2006), and morphology points in this direction (Eriksen 1993b); the other three groups appear to be monophyletic (although Carpolobieae are only weakly supported). However, the monophyly of all four tribes was strongly supported in a recent three-gene analysis (Forest et al. 2007b), and Xanthophylleae are sister to the other three tribes; relationships between the latter are unclear. Polygala and Bredemeyera are grossly paraphyletic (e.g. Persson 2001; Forest et al. 2007b).

For Emblingiaceae, often included in (e.g. Mabberley 1997) or near (e.g. Takhtajan 1997) Polygalaceae, see Brassicales.

For information, see Verkeke (1985, 1991: ovule and seed), Krüger and Robbertse (1988) and Krüger et al. (1988), floral morphology in Polygaleae), Eriksen (1993a: general), Takhtajan (2000: ovule and seed), and Banks et al. (2008: pollen morphology and evolution).

ROSALES [CUCURBITALES + FAGALES]: 1-2 apical ovules/carpel.

ROSALES Perleb - [Back to Index]

Dihydroflavonols +; sieve tube plastids lacking starch; leaf blade margins with teeth; inflorescence cymose; hypanthium +, nectariferous, K valvate, C clawed, 1 ovule/carpel, styles branched; K and/or hypanthium persistent in fruit. - 9 families, 261 genera, 7725 species.

Rosales contain ca 1.9% of eudicot diversity (Magallón et al. 1999); fossils are known from the Middle Eocene, ca 44Ma before present. Quite a diversity of butterfly larvae - especially caterpillars of "basal" groups and Lycaeninae - feed here (Fiedler 1995; Janz & Nylin 1998), and galls are common (Mani 1964). Ronse De Craene (2003) suggests that loss of the petals may characterise Rosales, with apparent "petals" occupying the position of stamens and their evolution allowing e.g. Rosaceae to diversify; comparing the vasculature of petals and stamens may bear on this morphological hypothesis, and whether or not it has anything to do with diversification is a separate issue. Indeed, if Rosales are sister to Fabales, they would not seem to be a notably diverse group in terms of species numbers, the more so since almost 4,000 species of Rosales are in the Ulmaceae-Urticaceae group, which lack petals.

At least Rosaceae, Rhamnaceae, Elaeagnaceae and Ulmaceae can be ectomycorrhizal (see Malloch et al. 1980; Smith and Read 1997). Sieve tube plastids lacking both starch and protein inclusions are rare outside Rosales, although they occur in some parasites as well as in Crassulaceae and Malpighiaceae (Behnke 1991a).

Relationships within the order remain rather unclear, although Rosaceae may be sister to the rest of the order (sometimes with strong support: Savolainen et al. 2000a), and Ulmaceae and relatives (the old Urticales) and Rhamnaceae and relatives may form two more clades (e.g. Thulin et al. 1998; Savolainen et al. 2000b; Richardson et al. 2000a; Sytsma et al. 2002).

In the past, Urticales (Urticaceae, Moraceae, etc.) were kept well separate from Rosaceae, largely because of the very reduced and usually wind-pollinated flowers of the former group; many of the other families now included in Rosales were usually placed elsewhere yet again.

There is much useful information in Thulin et al. (1999); for wood anatomy, see Jansen et al. (2000b) and Baas et al. (2001), and Kubitzki (2004) provides a summary of the order.

Includes Barbeyaceae, Cannabaceae, Dirachmaceae, Elaeagnaceae, Moraceae, Rhamnaceae, Rosaceae, Ulmaceae, Urticaceae.

ROSACEAE Jussieu - A ³10, carpels separate. - 95/2830. World-wide, but esp. N. hemisphere.

1. Rosoideae Arnott - 2-pyrone-4,6dicarboxylic acid, ellagic acid +; leaves compound; (epicalyx +), carpels usu. many, ovule unitegmic; fruit an achene; x = 7; plant with phragmidiaceous rusts. - Especially temperate (to Arctic) areas.

1a. Filipendula - Plant herbaceous; receptacle enlarged, 2 ovules/carpel. - 1/10. Eurasia.

1b. Rubus - Prickly scrambling shrub; receptacle enlarged; fruit an aggregate of drupelets. - 1/± 250. ± Worldwide, esp. N. temperate.

1c. Colurieae Rydberg - 3/42: Geum (40: Kajewski 1957 for classic cytological work; Smedmark & Eriksson 2006 for development of the stylar hook). Temperate, inc. montane tropics, Chile.

[Rosa + Potentilleae] + Sanguisorbeae: ?

Rosa + Potentilleae: ?

1d. Rosa - Prickly arching shrub; 2 collateral ovules/carpel; hypanthium fleshy, urn-shaped. - 1/100-150: see Bruneau et al. (2007: hybridization) and Wissemann and Cox (2007) for a phylogeny, relationships not easy to disentangle. N. temperate; 1/3^rd spp. in Europe.

1e. Potentilleae Sweet - Receptacle enlarged. - 11/800. Potentilla (500), Alchemilla (270). N. temperate, esp. Europe, tropical mountains (S. temperate). Some species of Potentilla turn out to be more closely related to Fragaria, etc., and most have at one time or another been segregated from Potentilla.

1f. Sanguisorbeae Candolle - Phragmidiaceous rusts 0. - 12/380: Cliffortia (115), Acaena (100). ± Worldwide, few Indo-Malesdia, tropical America.

Dryadoideae + Spiraeoideae: sugar alcohol sorbitol as transport carbohydrate, cyanogenic glycosides +.

2. Dryadoideae Juel - Association with N-fixing Frankia; ovules straight; fruits achenes with hairy styles. - 4/19: Cercocarpus (8). W. North America, circumboreal (Dryas).

3. Spiraeoideae C. Agardh - Ellagic acid 0; G <5, 2< ovules/carpel; fruit a follicle.

3a. Lyonothamnus - Cyanogenic glycosides 0; leaves opposite, compound, stipules deciduous; G seminferior; ca 4 apical ovules/carpel. - 1/1: Lyonothamnus floribundus. California Islands, off S. California.

3b. Niellieae Maximowicz - Ovule single, apical; fruitlets hard, shiny. - 2/24: Niellia (14). E. and W. North America.

3c. Amygdaleae Jussieu - Plant ectomycorrhizal; nectaries on petiole or abaxial lamina; G 1; fruit a drupe; n = 8. - 1/200. Temperate areas, also tropical montane.

3d. Osmaronieae Rydberg - Pith chambered; stipules deciduous; styles lateral; fruit a drupe; n = 8. - 3/9: Exochorda (4), Prinsepia (4). Central to East Asia, W. North America.

3e. Kerrieae Focke - Wart-like projections on lamina; fruit an aggregate, units nut-like. - 4/4. East Asia, W. North America, Alabama.

3f. Sorbarieae Rydberg - Leaves compound (simple: Adenostoma); ovules apical; phragmidiaceous rusts 0. - 4/8: Spiraeanthus (4). Central to East Asia, W. North America.

3g. Spiraeeae Candolle - Vestured pits +; nodes 1:1 [?all]; stipules 0; ³2 unitegmic ovules/carpel. - 8/106: Spiraea (80-100). N. temperate, to Columbia, (S. and) E. Africa, West Malesia.

3h. Pyreae Baillon - Plant ectomycorrhizal; G ± connate, adnate to base of hypanthium, ovules basal; exotesta ± thickened, often mucilaginous, mesotesta thick, sclerotic; Gymnosporangium rust common.

Gillenia - Leaves compound. - 1/2. E. North America.

The Rest - N = 17; four copies of GBSSI [granule bound starch synthase I]. - 33/ca 1000. Mostly N. temperate to Arctic.

Kageneckia + Lindleya - 4 ovules/carpel; Gymnosporangium rust 0. - 2/5. Mexico, Peru, Chile.

Vauquelinia - Tannin-containing cells pervasive; fruit septicidal, carpels opening adaxially and partially abaxially as well. - 1/3. S.W. North America.

Pyrinae Dumortier - Stipules deciduous; ovary at least half inferior; hypanthium fleshy in fruit. - 30/1000: Sorbus (260: generic limits are difficult, with divisions perhaps reflecting the European origin of taxonomy), Crataegus (260, inc. Mespilus), Cotoneaster (260: forms grafts with Crataegus), Pyrus (75), Malus (55). Mostly north Temperate.

• Rosaceae are a heterogeneous family, but they may be recognised by their often compound, stipulate and serrate leaves that are never(?) two-ranked and are only rarely opposite. Their flowers have a valvate calyx, always free and often clearly clawed petals, usually many stamens, a well-developed nectary on the hypanthium or at the base of the stamens, and often free carpels. The fruits are often achenes or drupelets, but the fruit in the broad sense is fleshy in a variety of ways, the hypanthium or receptacle also being involved; follicles are also quite common. Cyanogenesis is common (crush the leaves of Prunus, and smell), but there is more than one pathway involved.

Turonian fossils from some 90Ma before present are assignable to this family (Crepet et al. 2004 for references). The inferior-ovaried clade of Pyreae seems to represent a rapid but ancient radiation (Campbell et al. 2007). For the relationship between polyploidy and diversification in Rosaceae - perhaps direct - see Vamosi and Dickinson (2006).

Savile (1979b; see also Jackson 2004b for possible codivergence) discusses the distribution of phragmidiaceous rusts within Rosaceae-Rosoideae; they occcur on no other Rosaceae, and only rarely on plants from other families. Most of these rusts are autoecious, that is, their entire life cycle occurs on the same species of rosaceous host. In Gymnosporangium rusts the telial stage (the teliospore is a thick-walled resting spore that germinates to produce basidisospores) is common on some Cupressaceae, the aecial stage, which produces thinner-walled binucleate aeciospores, is found on Spiraeaoideae-Pyrodeae. Rosaceae rarely produce phytoalexins, protective compounds induced by e.g. fungal infection (Harborne 1999). Galls caused by cecidomyids are quite common in North America (Gagné 1989).

Apomixis is quite common, as in Alchemilla (Rosoideae) and Amelanchia, Cotoneaster and Crataegus (Spiraeaoideae-Pyrodeae). Some 17 species of Crataegus were recognised in North America in 1896, and 30 years later there were over 1000 species - C. S. Sargent described many of these. Hybridisation occurs in Crataegus, and the hybrids are triploid and apomictic. In general in Rosaceae, apomixis seems to have preceded hybridisation (Dickinson et al. 2007).

Although Rosoideae, Spiraeaoideae, and a number of other clades, e.g. Spiraeoideae minus Lyonothamnus, [Pyrodeae + Sorbarieae] and Pyrodeae are all well-supported, little can yet be said of the larger patterns of relationship in the rest of the family (e.g. Morgan et al. 1994; Potter et al. 2002; Potter 2003; Potter et al. 2007). The position of Dryadoideae is uncertain, other than being a rather "basal" branch in the tree (Potter et al. 2002: Evans et al. 2002, 2007), although a sister group relationship with Spiraeaoideae is perhaps most likely. Eriksson et al. (2003) proposed a phylogeny of Rosoideae where Filipendula is sister to all other members of the subfamily; the limits of Alchemilla are to be expanded to include Aphanes, etc. (Gehrke et al. 2008). See Soják (2008) for a classification of the speciose Potentilleae. Aldasoro et al. (2005) suggest morphological and biogeographic relationships in the inferior-ovaried Spiraeaoideae-Pyreae. Generic limits in Pyreae are difficult, there being little molecular divergence between many of the genera, but considerable divergence within them (Dickinson et al. 2007; Lo et al. 2007, esp. Crataegus s.l.). For relationships within Prunus, see Wen et al. (2008: some conflict between ITS and ndhF). The unfortunately rather overly complex classification above is based on that in Potter et al. (2007).

Fruit types are certainly not as good indicators of relationships in Rosaceae as was for a long time thought, but chemistry, chromosomes, and fungi all support the molecular realignments (especially Morgan et al. 1994), as does developmental work by Evans and Dickinson (1999a, b, 2002). Thus the Spiraeoideae now include a considerable amount of variation and are strongly paraphyletic, including the old Prunoideae/Amygdaloideae and Maloideae with their fleshy fruits, although Spiraeoideae s. str. in the past had been considered a very "natural" group that were characterized by their follicular fruits (e.g. Kalkman 1988). However, tribes in the Spiraeaoideae may represent clades (especially Evans et al. 2002; Potter et al. 2006). As is common, optimisation of characters on the tree presents problems. Potter et al. (2007) used DELTRAN, and as a result being host to Gymnosporangium rusts is not an apomorphy of their Pyrodeae; using ACCTRAN (as here) it is. They divided the presence of sorbitol into two states; it might be present in only small amounts (Dryadoideae), or it was more abundant (Spiraeaoideae), but if presence at any level is an apomorphy, then it would support the clade [Dryadoideae + Spiraeoideae].

A polyderm is common, although perhaps not occurring in Pyreae (Mylius 1913); this is a lateral meristem that develops in ground tissues in the pericycle and cuts off lamellae of paired concentric layers of parenchymatous and endodermal cells externally. 1:1 nodes occur in Spiraea, a genus that also lacks stipules, although normally Rosaceae have stipulate leaves and 3:3 nodes. (This correlation between stipule presence/absence and nodal vasculature is fairly general in eudicots, and so it is interesting that it is evident within Rosaceae, as was early noted by Sinnott and Bailey [1914].) Species of different genera of Rosaceae, one N-fixing (Cowania - Dryadoideae) and the other (Fallugia - Rosoideae-Colurieae) not, can form successful grafts, but when the non-N-fixing genus is the stock it will still not fix nitrogen (Kyle et al. 1986). Even in taxa with inferior ovaries, there is great variation in whether or not the carpels are connate, or what parts are connate, and in whether or not the carpels are adnate to the hypanthium. Thus Cotoneaster, which will form grafts with Crataegus, has an inferior ovary yet more or less separate carpels, c.f. also Pyracantha. Indeed, the odd genus Dichotomanthes, also a member of Pyreae, has a single carpel that is superior in position (Rohrer et al. 1994); its distinctive gynoecial morphology must represent a reversal.

Although Rosaceae as described above are holding together very well despite their morphological heterogeneity, there have been departures. Chrysobalanaceae, often associated with Rosaceae in the past, are part of a distinct clade within Malpighiales, Quillaja rather unexpectedly is an isolated monotypic clade within Fabales, while recently (Oh & Potter 2006) Guamatela has been found to be a correspondingly isolated clade in Crossosomatales.

As is evident above, the diploid Gillenia is sister to Pyreae (Potter et al. 2002; Evans & Dickinson 2002). It had long been suspected that Pyreae were of wide hybrid origin, as their chromosome number might suggest - 9 (Rosoideae) + 7 (Spiraeaoideae) = 16 (some Maloideae), see Evans et al. (1998) for literature. However, Evans and Campbell (2002) suggested that such an hybrid origin was unlikely, polyploidisation with subsequent aneuploidy (9 + 9 = 18, 18 - 1 = 17) of something similar to Gillenia (herbaceous, with compound leaves) being more probable. Note that Gillenia is host to the same Gymnosporangium rusts that are found on many other Pyreae, but no other Rosaceae, but it has only two copies of GBSSI, as is the condition in the rest of the family, Rhamnaceae, etc.

Some general information is taken from Robertson (1974), Judd et al. (2002), and especially Potter et al. (2007), for carpel orientation and general morphology, see Focke (1888), Sterling (1969 and references), Kania (1973) and Weberling (1989), for androecial diversity, see Lindenhofer and Weber (2000 and references); for cork cambium initiation and bark anatomy, see Lotova and Timonin (2002 and references); for 2-pyrone-4,6dicarboxylic acid distribution, see Wilkes and Glasl (2001); for exotesta anatomy, see Frohne and Jensen (1992); for wood anatomy, see Zhang (1992); and for floral development, see Evans and Dickinson (1999a, 1999b).

Barbeyaceae [Dirachmaceae + Rhamnaceae + Elaeagnaceae]: ?

Dense, curly hairs on the abaxial surface of the leaf blade could be a synapomorphy for this group (Sytsma et al. 2002); Qiu et al. (1998) put this feature at the next higher node.

BARBEYACEAE Rendle - Inflorescence fasciculate, bracts and bracteoles 0; flowers small, hypanthium and nectary 0; P uniseriate, anther with prolonged connective; G 1-2(-3), ± separate, style branches and stigmas long; fruit indehiscent; P accrescent. - 1/1: Barbeya oleoides. N.E. Africa, Arabia.

• Barbeyaceae may be recognised by their opposite, entire, exstipulate leaves with dense white indumentum on the undersurface and their shortly pedunculate, axillary, cymose inflorescences. The flowers are small, wind-pollinated and with a uniseriate perianth and the nutlets are enveloped by the accrescent perianth.

Barbeyaceae were included in Urticales by Cronquist (1981), while Takhtajan (1997) included a monotypic Barbeyales in Hamamelididae-Barbeyanae. Testa anatomy is not likely to tell us much about relationships because the fruit is a nutlet.

For information, see Dickison and Sweitzer (1970: morphology), Tobe and Takahashi (1990: hairs and pollen), Friis (1993: general), and Bouman and Boesewinkel (1997: ovule and seed).

Dirachmaceae + Rhamnaceae + Elaeagnaceae: stamens = and opposite C/alternate with P, capsule septicidal; coat multiplicative, exotesta palisade, thick-walled; cotyledons large.

DIRACHMACEAE Hutchinson - Flower single, terminal; C contorted, nectaries on base or on subbasal appendages, lacking stomata, anthers extrorse, long, opening from apex; fruit beaked, segments also opening adaxially, densely wooly inside, with columella. - 1/2. Socotra, Somalia.

• Dirachmaceae are shrubs that may be recognised by their small, strongly-toothed leaves which have subulate stipules; the latter persist along the stem. The flowers have an epicalyx and are relatively large, and the stamens are opposite the petals and have extrorse anthers that clearly open from the top downwards. The capsules are densely wooly inside.

Petal initiation is later than that of the stamens, and until quite late in development the petals are much shorter than the stamens. Dirachmaceae have been described as having a small, funicular aril, perhaps similar to that found in some Rhamnaceae (Ronse De Craene & Miller 2004).

The exotestal seeds with straight embryos suggest that Dirachma is not close to Geraniaceae (Geraniales), with which Dirachma had been linked (Boesewinkel 1985).

For information, see Link (1991b, c.f. 1994: general, esp. nectaries), Baas et al. (2001: wood anatomy), and Bayer (2004: general).

RHAMNACEAE Jussieu - A enclosed by C. - 50/900: Phylica (150), Rhamnus (125), Zizyphus (100-170, paraphyletic, see Islam & Simmons 2006), Ceanothus (55), Gouania (50). World-wide, especially tropics and warm temperate regions.

• Rhamnaceae are recognisable by their toothed, stipulate leaves that often have strong, parallel secondary veins and scalariform tertiary venation. Their flowers have stamens opposite the often clawed petals, a hypanthium with a nectary inside, and a valvate calyx with sepals that are ridged down the middle on their inner surfaces. The capsules are dehiscent, the fruit wall often separating into two layers with the inner woody layer twisting as dehiscence occurs. Since the seeds are basal, there is usually no columella, while on the ouside of the fruit towards the base there is often a conspicuous rim marking the place where the hypanthium was attached. In taxa such as Gouania with largely inferior ovaries the calyx, etc., is persistent. In lianes the shoots often develop somewhat laterally in the leaf axils and may have a bud at the very base.

  The capsules are rather similar to those of Euphorbiaceae, Phyllanthaceae and Putranjivaceae where the capsule also falls to bit as dehiscence occurs; there, too, the fruit wall often separates into two layers, the inner, woody, twisting as dehiscence occurs and hel;ping to eject the seed. However, since the seeds are basal in Rhamnaceae, there is usually no columella, as is so typical of these families, but they lack a hypanthium; Phyllanthaceae and Putranjivaceae have two seeds per carpel.

Fossils from the Cretaceous-Cenomanian, some 94Ma before present, can be identified as belonging to this family (Crepet et al. 2004 for references), although Wikström et al. (2001), followed by Richardson et al. (2004), date stem Rhamnaceae to some 62-64Ma before present. Richardson et al. (2004, see also Richardson et al. 2001a) discuss the diversification of the family in detail, noting i.a. the rapid diversification of the speciose and largely African Phylica within the last ca 8 my.

Ceanothus has N-fixing actinomycetes, as do many Colletieae; together they form a monophyletic group (Richardson et al. 2000b). Ectomycorrhizae have been reported from Rhamnus and Pomaderris (Malloch et al. 1980). Vegetatively, Rhamnaceae are quite variable. Quite a few are xeromorphic, for instance, species of Colletia have intricately branched, thorny stems, and their leaves are minute. New World species of Gouania have glands at the base of the lamina; Karwinskia has pellucid dots in the leaves.

There are three main clades in the family, the rhamnoids, which include Maesops and Ventilago (three tribes), the ziziphoids (five tribes), which include most of the rest of the family (for the Pomaderreae, one of these tribes, see Kellermann et al. 2005), and the ampeloziziphoids (three tribes: three genera: four species) (Richardson et al. 2000a, b).

For information, see Medan and Schirarend (2004: general), Vikhireva (1952: fruit anatomy), Medan (1988: gynoecium), and Medan and Aagesen (1995: flower and fruit morphology).

ELAEAGNACEAE Jussieu - Hairs lepidote or stellate; pits vestured; hypanthium long, C 0; G 1; hypanthium accrescent and fleshy in fruit; pericarp thin; endosperm with chalazal haustorium. - 3/45. North Temperate, warm tropical; Malesia and Australia.

• Elaeagnaceae are often thorny and mostly deciduous shrubs that are recognisable by their exstipulate leaves with entire margins; the whole plant is densely covered with scaly-stellate indumentum, which often gives it a greyish or silvery appearance. The hypanthium is relatively long and there is no corolla. The dry fruit proper is surrounded by the fleshy hypanthium (c.f. some Nyctaginaceae), and so the result is very like a berry or drupe.

All genera are associated with N-fixing Frankia.

The androecium is obdiplostemonous according to Huber (1963). Seed anatomy is rather like that of Rhamnaceae (Corner 1976); Harrison and Beveridge (2002) have clarified fruit and seed anatomy of Hippophae.

Elaeagnaceae have been difficult to place. They were included in Proteales by Cronquist (1981) because of superficial floral similarities, and in Elaeagnales - Rhamnanae, next to Proteanae, in Rosidae, by Takhtajan (1997).

For rust host preferences, see Savile (1979), for general information, Bartish and Swenson (2004).

Ulmaceae [Cannabaceae [Moraceae + Urticaceae]]: plant with ± watery exudate; phloem stratified; cystoliths [globose; usu. CaCO₃] and epidermal and hair cell wall silicification and calcification +; axillary shoots with at least one prominent (sub)basal prophyllar bud; flowers small, C 0, stamens equal and opposite perianth members, pollen porate, nectary 0; G [2], abaxial member only fertile, ovule apical, stigmas sessile, spreading, receptive area extending down adaxial surface and ± confluent; fruit a drupe.

This clade may be some 67-65 Ma old, Ulmaceae diverging 57-55Ma before present, the rest ca 48-42Ma before present (Wikström et al. 2001). Some nymphalid butterfly groups have larvae on members of these families (see also under Urticaceae) - but also on the immediately unrelated Euphorbiaceae (Malpighiales: see Ehrlich & Raven 1964). Similarly, caterpillars of Acraea are found quite commonly on Urticaceae (including Cecropia), but also on Moraceae, etc.; this particular genus is also particularly common on Passifloraceae and their relatives (Malpighiales).

Because of the well-developed prophyllar bud(s), the inflorescences are often paired, with a bud between them, and/or the branches may have a bud on one or both sides at the base. The "stipular buds" of Cannabis (Miller 1970 and references) are really prophyllar buds. The profuse development of shoots from these prophyllar buds in response to galling insects, etc., causes the irregular witches brooms so common on Celtis around here [St Louis, Missouri]. Ulmus does not have prophyllar buds.

It is not clear which taxa have a hypanthium; at least some species of Ulmus and Pilea do, but other species of Ulmus, Zelkova and Trema show no obvious signs of one (see also Bechtel 1921). Whether a hypanthium is present or not, there are no nectaries. Taxa with a perforated testa are quite common in the group, although this feature may have arisen more than once (Kravtsova & Oskolski 2007). Sytsma et al. (2002) note that inflexed stamens and their dehiscence, fruit type, and laticifers need further detailed study within this clade; hypanthium presence and other characters can be added to this list.

The phylogenies suggested by Sytsma et al. (2000) and Song et al. (2001) place Cannabaceae within Celtidaceae - Cannabaceae is the earliest name for the combined group - and Cecropiaceae within Urticaceae, and this set of relationships has been strongly supported by a more comprehensive analysis (Sytsma 2002). The group as a whole is very well characterised; Corner (1952) even included Moraceae in Urticaceae. The copious information on the four families awaits synthesis, although this process has been begun by Sytsma et al. (2002); this paper should be consulted for details of character evolution.

For further information, see Satake (1931: spodograms), Sweitzer (1971: anatomy), Giannasi (1978, 1986: chemistry), Behnke and Barthlott (1983: hairs), Takaso and Tobe (1990: testa), Omoron and Terabayashi (1991: phylogeny), Terabayashi (1991: vernation), Judd et al. (1994: morphological phylogeny), and Tobe and Takaso (1996: hairs).

ULMACEAE Mirbel - C-glycoflavones 0; hairs smooth; seeds flattened. - 6/35: Ulmus (20-30, species limits uncertain). Mostly N. temperate, esp. Asian, but scattered elsewhere except Australia and the Pacific.

• Ulmaceae are usually trees and may be recognised by their usually two-ranked, strongly serrate, stipulate leaves with asymmetric bases and rough surfaces; the pinnate secondary veins proceed to the apex of the teeth (not in Ampelocera). At least some of the small flowers are perfect and the fruits are flattened.

Dutch elm disease, the result of infection by the ascomycete Ophiostoma ulmi complex and carried by bark weevils the native Hylurogopinus rufipes and Scolytus multistriatus (introduced in North America), has devastated native elm populations in North America, and also in Europe.

Ulmus lacks well-developed prophyllar buds and has one of the two stipules intrapetiolar, they are both intrapetiolar in seedlings of some species, and the leaves in seedlings may also be opposite. Nawaschin (1895) suggested that chalazogamy occurred in Ulmus.

The poorly-known Ampelocera was included here by Ueda et al. (1997b); although its hairs are smooth, its leaves have ascending veins (see also Wiegrefe et al. 1998).

For general information, see Todzia (1993).

Cannabaceae [Moraceae + Urticaceae]: C-glycoflavones +; unicellular hairs usu. micropapillate; secondary veins palmate; stipules cauline-intrapetiolar; flowers imperfect; embryo curved.

CANNABACEAE Martynov - 11/170. Worldwide, but not Arctic.

• Cannabaceae may be recognised by their two-ranked, serrate leaves with ascending veins that do not proceed straight into the teeth, more or less intrapetiolar stipules, and prominent prophyllar buds. The fruit is a rounded drupe. However, Humulus and Cannabis have secondary veins running into the teeth, and Humulus lacks buds at the base of the staminate inflorescence...

The distinctive ascending venation of Celtidaceae s. str. was used to distinguish it from Ulmaceae, in which the spreading veins proceed straight to the teeth. However, Cannabis and Humulus have compound leaves, and the venation in Palaeocene fossils attributable to Celtis is like that typical of Ulmaceae (Manchester et al. 2002).

Parasponia is the only non-legume nitrogen fixer that is associated with other than actinomycetes; its rhizobia remain in infection threads.

Whether the laticifers of Cannabis etc., are really similar to those of Urticaceae and Moraceae must be confirmed; they have no milky exudate and they are found throughout the plant. Lozanella, with its opposite leaves and boxy venation, looks rather like Urticaceae, but it has a bilobed stigma. Cannabis and Humulus have an X-autosome balance system determining the 'sex' of the plant.

Pteroceltis, Humulus, and Cannabis are close. Lozanella is sister to Aphananthe or sister to the rest of the family (Wiegrefe et al. 1998; Soltis et al. 2002).

For general information, see Grudzinskaya (1967), Todzia (1993: as Ulmaceae), and Kubitzki (1993b).

Moraceae + Urticaceae: latex system +; stamens inflexed in bud [± abruptly straightening during pollination].

Rates of molecular evolution are likely to have increased at least twice in this clade, e.g. in most Urticaceae and in Dorstenia as compared with other Moraceae (Smith & Donoghue 2008).

It now appears that inflexed stamens are a synapomorphy for Moraceae + Urticaceae (Datwyler & Weiblen 2004). Moraceae with explosive pollen dispersal are in the paraphyletic Moreae that is part of a basal polytomy within Moraceae. For a morphological phylogeny focussing on Urticaceae s.l., see Kravtsova and Oskolski (2007).

MORACEAE Link - Laticifers throughout plant, latex milky; inflorescence compact. - 38/1100: Ficus (750: inflorescence axis concave), Dorstenia (105: inflorescence axis spreading, fruit a dehiscent drupe), Artocarpus (50: inflorescence axis elongated, bread and jack fruits - up to 50 kg), Antiaropsis (inflorescence axis concave at first, but spreading ["dehiscing"] when ripe). Mostly tropical to warm temperate.

• Moraceae are laticiferous plants with stipulate leaves and often boxy fine venation; the flowers are small and densely aggregated; the inflorescence axis is often large, becoming much accrescent and ± fleshy in fruit.

Berg (2005) suggests that diversification in Moraceae occurred on a still physically coherent tropical supercontinent, however, Zerega et al. (2005) advance a more complex hypothesis to explain the distribution and diversification of the family. They date the diversification of crown group Moraceae to at least 79Ma before present, and the divergence of the clade to at least 89Ma before present.

Moraceae, and in particular Ficus, includes a number of (hemi) epiphytes, stranglers and lianes. The family is often the second most speciose family in lowland tropical rainforest in America and Africa (Gentry 1988). Bombyx mori (the silkworm) caterpillars can eat quite widely within Moraceae, but not on most other members of the Ulmaceae group - although they will grow on Ulmus itself (Fraenkel 1959). Caterpillars of danaine butterflies quite commonly use Moraceae as food plants; both Moraceae and their usually preferred Apocynaceae are rich in latex, although Moraceae do not often have the apocynaceous cardenolides (Ackery & Vane-Wright 1984).

Dorstenia, etc., have "dehiscent" drupes (and incurved stamens); here the stone is shot out of the turgid mesocarp, the separation following a line of weakness in the tissue. The flowers are borne on the upper side of a flat inflorescence axis. In other Moraceae such as Morus the flowers are borne on all sides of an erect inflorescence axis; just what part of the plant contributes to the fleshiness of the infructescence varies. Castilleae, sister to Ficeae, have both these kinds of inflorescences. Finally, in Ficus the flowers are borne on the inside of an invaginated inflorescence axis, the outer part often bearing conspicuous bracts.

The intimate and remarkable association between figs and their agaonid fig-wasp pollinators, and the parasites of those pollinators, is well known (Lopez-Vaamonde et al. 2001; Jousselin et al. 2003; Jackson 2004a; Rønsted et al. 2005b, 2008 for references); the beginning of the association/co-divergence dates to 100-60Ma before present. Ficus is included in Ficeae, and they are sister to Castilleae (which in turn include some errant Artocarpeae), and both have urceolate inflorescences and insect pollinators breeding among the flowers (Datwyler et al. 2003; Datwyler & Weiblen 2004). Within Ficus itself, the fig/fig wasp association has been touted as a classic example of an obligate one-on-one association of species of fig with its wasp pollinator. Recent work, however, suggests that there is suprisingly often rather little specificity between fig and wasp (Machado et al. 2005; Jackson et al. 2008), so questioning both the classic idea of an obligate association and also the frequency of occurrence of co-speciation. Furthermore, despite fig pollen being dispersed by tiny agaonid wasps, these wasps can sometimes be transported up to 14 km, the result being that breeding units of figs may be an order of magnitude larger - some 100 square kilometres or more - than those of other plants in the rainforest (Nason et al. 1998). Nevertheless, there is at least a general association between figs and wasps, and in groups like section Ficus sect. Galoglychia cospeciation does seem to occur, even in the non-pollinating wasps (Jousselin et al. 2008). Parasitoid wasps that parasitize the fig wasps may be of considerable importance in preserving the mutualism betweem the agaonid wasp and fig (Dunn et al. 2008; for more information on figs and wasps, as well as gallers and parasitioids, see also the papers in Symbiosis 45, nos 1-3, 2008; Silvieus et al. 2008; especially Herre et al. 2008). Also involved are drosophilid flies that in Africa, at least, have a very close association with figs and oviposit either on the stomium or the exit holes made by the male fig wasps (Harry et al. 1996, 1998).

Figs are also a very important and dependable source of food for frugivores, both birds and mammals, throughout the tropics, and the diversity of their growth forms means they are encountered throughout the forest. Indeed, figs are perhaps surprisingly nutritious (Herre et al. 2008 for references), and individuals of many species will be in fruit throughout the year. It has been found that the species richness of Ficus is correlated with that of their avian frugivores, especially specialised frugivores, and the frugivores also select on particular aspects of the morphology of the fig species (Shanahan et al. 2001; Kissling et al. 2007; Herre et al. 2008). Some bats also eat figs, and epending on whether the bats are New or Old World, they search for food in different ways and have selected figs with different qualities - thus in the New World figs tend to be greenish and odoriferous (Compton 1996 [a whole series of papers]; Korine et al. 2000; Shanahan et al. 2001; Harrison 2005).

The paraphyletic Moreae with their incurved stamens include Morus (fleshy perianth + drupe, a syncarp), Maclura (ditto), and Broussonetia (drupe - tapa cloth). The straightening stamens and reflexing tepals of Morus alba are thought to show the fastest known movement of any plant part, moving at over half the speed of sound (Taylor et al. 2006). Some species of Dorstenia have small, cauline stipules that do not overlap the petiole; chromosome number in this genus is very variable.

For more information, see Rohwer (1993a: general) and Oginuma and Tobe (1995: chromosomes).

URTICACEAE Jussieu - Laticifers only in bark; lignification of vascular tissue delayed; G apparently 1, ovule basal, ± straight. - 54/2625 Pilea (500-600: achenes dispersed by straightening of inflexed filaments of staminodes; see Monro 2006 for phylogeny), Elatostema (300: see Hadiah et al. 2003 for phylogeny), Urtica (80), Cecropia (75), Coussapoa (50). World-wide, but mainly tropical.

• Urticaceae are variable in habit but have stipulate leaves with boxy venation. The small flowers are borne in branched, clearly cymose inflorescences, although carpelate inflorescences may be capitate, and the anthers are often explosively dehiscent. Cecropia and some genera related to it are trees with stout stems, deeply lobed or compound leaves, and densely spicate inflorescences, and have stamens do not dehisce explosively.

It has been suggested that caterpillars of Nymphalini butterflies have a plesiomorphic association with Urticaceae as food plants (Janz et al. 2001).

Cecropia is a fast-growing pioneer tree that is associated with Azteca and some other ants that live in the stems and eat glycogen-rich food bodies (Müllerian bodies) produced by the plant at the abaxial base of the petiole; some beetles eat these food bodies, as well as ant eggs, etc. (Jolivet 1991; Longino 1991; Yu & Davidson 1997). This association may break down, especially on islands and at high altitudes where Cecropia lacks ants (Janzen 1973); Musanga, e.g. M. cecropioides (sic), from Africa, also lacks ants but is otherwise very similar to Cecropia. (Species of Macaranga [Euphorbiaceae] are ecological analogues of Cecropia in Malesian forests.)

Explosive dispersal of the pollen as the filaments abruptly straighten is common in Urticaceae. In Pilea the achenes are similarly dispersed by the inflexed filaments of the staminodes.

Groups of cells even in somewhat older vascular tissue may be unlignified and the pericyclic sheath may also be late in lignifying. Boehmeria has a fleshy perianth. The gynoecium is basically bicarpelate, but one carpel is highly reduced (Eckardt 1937).

Urticaceae minus Cecropiaceae are paraphyletic (Sytsma et al. 2000, 2002 - three genes; Monro 2006 - two genes). However, Datwyler and Weiblen (2004 - one gene) and Zerega et al. (2005) find the reverse relationship, and strong support for Poikilospermum as sister to the rest of the family - [Poikilospermum [Cecropiaceae s. str. [rest of Urticaceae]]]. Either way, Cecropiaceae are best included in Urticaceae; their cystoliths may be circular, unlike the elongate cystoliths of other Urticaceae, or they may be absent.

See Bigalke (1933: cystoliths and hairs), Berg (1978: Cecropiaceae), Kubitzki (1993b: Cecropiaceae), Friis (1993: Urticaceae) and Kravtsova (2003: seed coat anatomy) for additional information.

CUCURBITALES + FAGALES: ovary inferior; fruit 1-seeded, indehiscent.

"Embryo with large cotyledons" may be another synapomorphy (Zhang et al. 2006), also a three-carpelate gynoecium. However, ovary position and fruit characters in particular reverse spectacularly (see also Matthews & Endress 2004; Zhang et al. 2006).

CUCURBITALES Dumortier

Wood rays wide, multiseriate; leaves with palmate secondary veins; K or P valvate, styles separate. - 7 families, 129 genera, 2295 species.

Butterfly caterpillars may be relatively uncommon on members of the order (Ehrlich & Raven 1964).

The two whorls of the perianth tend to be rather similar in texture, etc. More or less lacinate petals (and staminodes) are common in the order (Endress & Matthews 2006b). Zhang and Renner (2003) suggest that the flowers are usually imperfect; flower type varies considerably in Anisophylleaceae, with perfect flowers being known there, so that character may properly be placed at a slightly higher level, or it may be a synapomorphy of [Cucurbitales + Fagales].

The circumscription of Cucurbitales is rather unexpected, including as it does families like Coriariaceae (in the past often placed with other families with separate carpels, a "primitive" family) and Anisophylleaceae (often associated with Rhizophoraceae, in the past itself often associated with Myrtales, but now in Malpighiales). Corynocarpaceae are Celastralean according to Cronquist (1981), isolated, according to Takhtajan (1997), but they find a home here sister to Coriariaceae (see e.g. Setoguchi et al. 1999; Schwarzbach and Ricklefs 2000). Similarities in floral morphology between Anisophyllea and Ceratopetalum (Cunoniaceae - Oxalidales: see Matthews et al. 2001; Endress & Matthews 2006b), although striking, are unlikely to be evidence of immediate close relationships of the two families. The tree here follows that of Zhang et al. (2006: see also Schwarzbach & Ricklefs 2000); Zhang and Renner (2003a) also discuss aspects of morphological evolution and evaluate the extensive variation in breeding system in the clade.

Nickrent et al. (2004) suggested relationships of Apodanthaceae either within Malvales (especially the three-gene analyses and that of nuclear SSU rDNA), or in or near Cucurbitales (analysis of matR); Barkman et al. (2007: support weak, but rather comprehensive analysis) also suggested the latter position. Additional molecular analyses (D. Nickrent, pers. comm.) support the position of Apodanthaceae in Cucurbitales, and this is consistent with their inferior ovary and parietal placentation, both features common in Cucurbitales.

For wood anatomy, see Wagstaff and Dawson (2000), Nandi et al. (1998), and Baas et al. 2000), and for an excellent survey of floral morphology and evolution, see Matthews and Endress (2004, summarized in 2006b).

Includes Anisophylleaceae, Begoniaceae, Coriariaceae, Corynocarpaceae, Cucurbitaceae, Datiscaceae, Tetramelaceae.

ANISOPHYLLEACEAE Ridley - Nodes 1:1; K epidermis with mucilaginous inner walls, C open, ± enclosing groups of A, lobed or laciniate, pollen heteropolar, dicolpate, smooth; seed coat initially thick, vascularized, becoming crushed; endosperm starchy, embryo largely hypocotylar, cotyledons at most small. - 4/34. Pantropical.

• Anisophylleaceae are a rather undistinguished-looking family. The plant may be anisophyllous, the leaves being rather coriaceous, often yellowish drying, entire, and often more or less asymmetric at the base. The flowers are small and the ovary is inferior.

There are no laticifers, the petals are not aristate, and there is no sclerified exotegmen in the seed; in these and other characters Anisophylleaceae differ from Rhizophoraceae, here in Malpighiales (see also Juncosa & Tomlinson 1988a, b).

Diversification in Anisophylleaceae may have begun (107-)85(-67) Ma; Combretocarpus is sister to the rest of the family (Zhang et al. 2007).

See also Vincent and Tomlinson (1983: Anisophyllea architecture), Tobe and Raven (1987c: embryology), Dahlgren (1988: general).

[Corynocarpaceae + Coriariaceae] [Cucurbitaceae [Tetramelaceae [Datiscaceae + Begoniaceae]]]: filaments shorter than anthers in bud, anthers basifixed, receptacular nectaries 0.

Corynocarpaceae + Coriariaceae: ellagic acid +; stomata paracytic; lamina margins entire; flowers small, petals thick, base broad, ovary superior, 1 apical ovule/carpel, vascular bundle extending into the outer integument; cotyledons very large.

CORYNOCARPACEAE Engler - Stipules intrapetiolar; stamens = and opposite and basally adnate to petals, incurved in bud, 5 fringed petal-like staminodes with a basal, adaxial nectary opposite sepals, pollen heteropolar, dicolpate, smooth; G [2], only one fertile. - 1/6. New Guinea to New Zealand, W. Pacific.

• Corynocarpaceae are best recognised by dissecting their flowers. These are quite distinctive, although small: the stamens are opposite the petals and alternate with fringed staminodes, and the ovary has a single style. Vegetatively the family is rather less distinctive, having spirally-arranged leaves with entire margins and intrapetiolar stipules.

The plant is very poisonous, having bitter glucosides. Variants that have two styles are known (Matthews & Endress 2004), so the single, excentrically-placed style at the apex of the ovary may suggest that the gynoecium is basically two carpelate.

Corynocarpaceae are Celastralean according to Cronquist (1981), isolated, according to Takhtajan (1997).

See also Nowicke and Skvarla (1983: pollen) and Philipson (1987a: general).

CORIARIACEAE Candolle - Nodes 1:1; leaves opposite; inflorescence racemose, bracteoles 0; C open, fleshy; G 5, at most only basally connate, styluli slender, stigmatic all around; fruit an achene, surrounded by enlarged and fleshy corolla. - 1/5. Very disjunct: circum S. Pacific to China and Himalayas, Mediterranean.

• Coriariaceae can be recognised by their opposite, sessile, entire leaves with veins coming from the base and at most minute stipules; many branches look like compound leaves, apparently being of limited growth. The inflorescences are racemes and the flowers have a poorly differentiated, biseriate perianth and carpels that are connate at most only basally; the fruits are achenes that are surrounded by the accrescent corolla and the long, subbasal styluli persist on the achenes.

For phylogenetic relationships within the family, see Yokoyama et al. (2000); the Eurasian clade is sister to the rest.

Coriariaceae were placed in Ranunculales by Cronquist (1981) and as a monotypic Coriariales in Rosidae (Takhtajan 1997), largely because of their separate carpels.

Nodal anatomy is taken from Sinnott (1914) and wood anatomy from Yoda and Suzuki (1992).

Cucurbitaceae [Tetramelaceae [Datiscaceae + Begoniaceae]]: perennial herbs; cucurbitacins [triterpenes] +, ellagic acid 0; young stem with separate bundles; leaves with teeth, stipules 0; flowers imperfect; placentation parietal, many ovules/carpel, a roof over the ovary [= styles marginal], stigmas large, elongated, bilobed.

The formation of a "roof" over the ovary from tissue adaxial to the styles (Matthews & Endress 2004) is obvious when it is well developed in that the inidividual styles are widely separate and are borne on the margin of the ovary.

CUCURBITACEAE Jussieu - Plant a vine, (pachypodia [conical, above-ground tubers] +); citrullin [non-protein amino acid] +; tendrils +, bifid, coiled both below and along the branches, lateral to the leaf; young stem with bicollateral vascular bundles; cystoliths +; indumentum rough/prickly, hair cell walls calcified; C connate, (C lobes long-fimbriate), nectary of nectariferous hairs; staminate flowers: anthers extrorse, anthers monothecal, often much bent and coiled; carpelate flowers: placentae intrusive-parietal; fruit baccate, with a thick, hard skin; seeds flattened, pitted, testa multiplicative, exotesta complex, hypodermis of 1 or more layers of lignified cells; chalazal haustorium +. - 129/800. Especially tropical and subtropical.

1. "Fevilleoideae" Burnett - Plants perennial; plant dioecious; stamens inserted on or near gynoecium, (filaments 3, anthers 5), pollen grains striate, ovules usu. ascending to horizontal; fruit opening apically; seeds winged. - 19/60. Tropical.

2. Cucurbitoideae Kosteletzky - Often annuals; additional non-protein amino acids +; (tendril branches alone coiled); staminate inflorescence + carpelate flower + bud + tendril making up axillary complex; plant monoecious; (nectar exuded through stomata); stamens usu. inserted on hypanthium, (2 + 2 + 1), pollen grains porate or colpate, (spinose-echinate), style single, ovules pendulous; fruit fleshy, indehiscent; integument vascularized. - 111/740: Trichosanthes (100), Cucumis (52), Momordica (45). Tropical to warm temperate.

• Cucurbitaceae are easily recognised. They are vines or lianes with often rather roughly hairy exstipulate leaves that have palmate venation. In the leaf axil there is a usually lateral and branched tendril, a vegetative bud, a flower, or a yet more complex organisation. The plants are dioecious or monoecious, and the flowers have a hypanthium, a often complex androecium made up of twisted, monothecal anthers, and an inferior ovary with parietal placentation. The corolla is often more or less yellow, sometimes red. The seeds are more or less flattened, even when large.

Schaefer et al. (2008b) suggest that stem Cucurbitaceae are some (69-)63(-61[properly -58?]) Ma old, i.e. Late Cretaceous, with the current world-wide range of the family being in large part the result of extensive dispersal, Madagascar being colonized an estimated thirteen times and Australia twelve times (the latter currently has only twelve genera and thirty species). Interestingly, the woody Socotran endemic Dendrosicyos is dated to (30-)22(-14) Ma, although Socotra itself is only about ten Ma old, which suggests that the genus was once on the mainland and has since become extinct there (Schaefer et al. 2008b).

.

Low concentrations of the very bitter cucurbitacins elicit a feeding response from diabotricites beetles, rootworm leaf beetles (Chrysomelidae: Galerucinae: Luperini) (Metcalf et al. 1980; Jolivet & Hawkeswood 1995), and adults visit the flowers, feeding on pollen and sometimes other parts, too, from whence they sequester the cucurbitacins (Eben 1999). This capability for feeding on Cucurbitaceae seems to have evolved independently in Old and New World Luperini (Gillespie et al. 2003). The beetles may cut leaf veins, so locally interrupting the translocation of cucurbitacins to the leaf tissue and allowing the insect to eat it (Dussourd & Eisner 1987). However, given that at least some of these beetles will eat pure cucurbitacin crystals, avoidance of the copious sap produced by Cucurbitaceae is a more likely explanation of this feeding behaviour, the more so because the sap, very rich in P-protein, gels within seconds and would gum up the mouth parts, etc., of the beetles (McCloud et al. 1995).

Heliconius, which has diversified in association with Passiflora (Passifloraceae), preferentially pollinates Anguria and other Cucurbitaceae, and also some Rubiaceae. It takes masses of pollen from the flowers and then moistens them with nectar; amino acids are released from the pollen and are taken up by the butterfly (Gilbert 1972). Some Cucurbitaceae, notably Momordica and Thladiantha, are oil flowers, Ctenoplectridae bees collecting material from the oil-secreting hairs (Buchmann 1987; Vogel 1990 for details). More or less long-lacinate corolla lobes are scattered in the family; taxa with such corolla lobes may be pollinated by moths (Schaefer et al. 2008). Several genera of Fevilleoideae, perhaps even including the large-fruited Pteropepon with its massive seeds, have more or less obconoid fruits opening apically, the seeds, sometimes winged, falling out, but other Fevilleoideae appear to have coloured, indehiscent, and apparently animal-dispersed fruits.

The tendrils of Cucurbitaceae are part of a branch complex. Often there is a sublateral tendril + bud + slightly lateral flower associated with each leaf, or a tendril + vegetative bud + carpelate flower + staminate inflorescence, all more or less collaterally arranged, etc. Eichler (1875) and Goebel (1932) suggested that the tendrils were prophylls, and in Bryonia dioica paired tendrils occur on the pedicel of an axillary flower. Non-flowering Zanonioideae have tendrils more or less lateral to vegetative axillary buds. When flowering finally begins, tendrils are replaced by flowers, which are now more or less adaxial to the axillary bud. This produces an inflorescence branch which has an internode below the first prophyllar leaf and this leaf subtends the first flower. Most Cucurbitoideae lack an initial prolonged vegetative period, and the inflorescence branch lacks a basal internode, so the first flower, often carpelate, arises in the leaf axil of the main branch but is sometimes subtended by a branch prophyll; the staminate inflorescence represents the further development of this inflorescence branch (Lassnig 1997, for details). However, Gerrath et al. (2008) found that in Echinocystis lobata tendril, lateral bud, carpelate flower, and staminate inflorescence, were all more or less independent in origin, although the latter two did arise from a common primordium. Acanthosicyos has paired thorns at the nodes.

Cucurbitaceae were particularly important in early agriculture in the Americas, being one of the triumvirate of squash, corn and beans. For discussion of various aspects of the history of cultivation of Lagenaria and Cucurbita in particular, see Teppner (2004). For the domestication of squash (Cucurbita moschata) which began ca 10,000 years ago, see Dillehay et al. (2007); for phytoliths of the family, see Piperno (2006).

Most, but not all, Cucurbitoideae have a disc-like nectary that may even be covered by a flap of tissue; nectar exudes through nectarostomata. The nectary hairs of Fevilleoideae are less localized on the petals (Vogel 1997, see also Bernadello 2007); they are also found in Sicyeae and Cyclanthereae. When the stamens are connate 2 + 2 + 1, the vascular supply shows evidence of this, although there are differences over the interpretation of the apparently bithecal stamens (e.g. de Wilde & Duyfjes 1999). The pollen grains may produce several pollen tubes, and the grains of some Cucurbitoideae-Cucurbiteae are up to 200 µm or so long. The chalazal haustorium of Sechium edule, at up to 19,000 µm long, is apparently the longest in the family, although others are also quite long; only Santalales have longer embryo sacs (Johri et al. 1992). Seedlings (of the whole family?) have a peg, a cortical outgrowth on one side of the seedling axis at the root-shoot transition (e.g. Klebs 1884 for a list of taxa). Some of these characters may turn out to be quite high-level apomorphies in the family.

Renner et al. (2002) suggested that Cucurbitoideae were probably monophyletic, with Thladiantha possibly sister to the rest; Fevilleoideae formed an unresolved basal polytomy. This was largely confirmed by Kocyan et al. (2007) and Schaefer et al. (2008b), although in the latter Indofevillea was sister to other Cucurbitoideae; monophyly of Fevilleoideae (Zanonioideae) was not well suppported, Alsomitra sometimes appearing as sister to Cucurbitoideae. Schaeffer et al. (2008b) also found Fevilleoideae to be paraphyletic and did not recognise the subfamily, however, there were no strongly supported relationships between the tribes that were included in it; it is kept here pending the appearance of the classification proposed by these authors. Characters of Fevilleoideae are likely to be family apomorphies, or perhaps plesiomorphies. Jeffrey's tribes (Jeffrey 2005) have largely been supported, but his subtribes have not (Kocyan et al. 2007). Jobst et al. (1998) found Benincaseae (Cucurbitoideae) to be polyphyletic; Chung et al. (2003) and Schaefer et al. (2008) also looked at relationships within Cucurbitoideae. There are many small genera in Cucurbitaceae, and generic limits need attention, thus Ghebretinsae et al. (2007) adjusted the limits of Cucumis (see also de Wilde & Duyfjes 2006; Schaefer et al. 2008b).

For general information, see Jeffrey (1980) and Bates et al. (1990), for non-protein amino acids, see Fowden (1990), for seed coat anatomy, see Singh and Dathan (1998, 2001) and Teppner (2004), for embryology, etc., see Singh (1970), and for a suprageneric classification, see Jeffrey (2005).

Tetramelaceae [Datiscaceae + Begoniaceae]]: pollen spherical; fruit a septicidal capsule [dehiscing apically adaxial to the perianth]; seed operculate; exotestal cells honeycomb, inner walls strongly thickened and lignified.

Tebbitt (2005) suggests that the seeds of this group have a lid or operculum, but whether this is a synapomorphy or not is unclear. Seeds of Tetramelaceae are apparently little known, and Boesewinkel (1984) found that the opercula of Datiscaceae and Begoniaceae were rather different.

TETRAMELACEAE Airy Shaw - Trees. - 2/2. Indo-Malesia.

Tetrameles wood is known fossil from the Deccan Traps in India ca 70.6-65.5 Ma old (Zhang et al. 2007).

For relationships to Datiscaceae, see Swensen et al. (1994, 1998).

Datiscaceae + Begoniaceae: ?

DATISCACEAE Berchtold & J. Presl - 1/2. W. North America, Crete to India.

For information, see Davidson (1973, 1976).

BEGONIACEAE Berchtold & J. Presl - Herbs; stomata with accessory cells in two rings [helicocytic]; pearl glands +; leaves two-ranked, asymmetric, stipules large, cauline-extrapetiolar; staminate flowers: A centrifugal, basifixed, connective enlarged, pollen colpate; carpelate flowers: borne towards the ends of the inflorescence branches. - 2/1401. Largely tropical.

1. Hillebrandia - Plant with round tubers; P 10 [5 sepals + 5 petals?]; G [5], only partly inferior. - 1/1: Hillebrandia sandwicensis. Hawaii.

2. Begonia - Plant rhizomatous; staminate flowers: P in 2s; carpelate flowers: P 5; G [2-3], placentation axile, styles central; capsule dehiscing down sides, winged. - 1/1400. Largely tropical.

• Begoniaceae may be recognised by their succulent, herbaceous habit and often two-ranked leaves the blades of which have asymmetric bases and well developed teeth. The hairs are often large, flattened, or almost setular. The plants are monoecious, the staminate and carpelate flowers often have different numbers of tepals, both stigmas and anthers are bright yellow, and the fruit is often winged.

Hillebrandia, from Hawai'i,is sister to the pan-tropical Begonia (Clement et al. 2001; Swensen et al. 2001). Recent estimates put diversification of the genus as occurring some time from the Eocene to early Oligocene 45-25 Ma during a period of global cooling (Goodall-Copestake et al. 2009). But how and when did Hillebrandia get to Hawai'i and how has a genus that is now monotypic maintained itself for some 50 Ma (c.f. Psiloxyum on the Mascarenes)? Begonia itself is not known from Hawaii.

Staminate flowers of Begoniaceae produce pollen as a reward, and although carpelate flowers usually have no reward they have bright yellow and anther-like stigmas; deceit pollination is probably involved (Schemske et al. 1996). There are a few ornithophilous species with nectaries at the base of the styles in carpelate flowers only, while other species have no reward at all; various levels of deceit/mimicry are again involved (Vogel 1998b; Renner 2006). Begoniaceae are somewhat unusual among monoecious taxa with cymose inflorescences in that the carpelate flowers are produced only later and the first flowers produced in the inflorescence are staminate, although the derived Symbegonia group shows the reverse arrangement. Tebbit et al. (2006) looked at the evolution of dispersal mechanisms in the speciose Southeast Asian Begonia; taxa with seeds dispersed by animals or that are rain-ballists predominate in a single clade, while other taxa have seeds that are dispersed by the wind.

For phylogenetic studies of Begonia, see Plana (2003) and Plana et al. (2004: African taxa) and Forrest and Hollingsworth (2003) and Forrest et al. (2005).

Hillebrandia has a number of perhaps plesiomorphic features, and some of the features we think of as being characteristic of Begoniaceae as a whole (style position, fruit dehiscence) may in fact be apomorphies for Begonia alone. The inner five tepals ("really" petals?) of Hillebrandia are smaller than the others, indeed, they are sometimes minute. It has been suggested that the perianth of Begonia is to be compared with the sepals of Hillebrandia (see Gauthier 1959), and also that the petals of Hillebrandia are staminodial (Ronse Decraene & Smets 1990). The plesiomorphic tepal number of Begonia may be four in staminate flowers and five in carpelate flowers (Forrest et al. 2005).

For floral development, see Charpentier et al. (1989), for sections, etc., see Doorenbos et al. (1998), for the species of Begonia, see Golding and Wasshausen (2002) and Tebbitt (2005: more horticultural).

APODANTHACEAE Takhtajan - Endophytic stem parasites; plant monoecious or dioecious; P 2 + 4 + 4 or 3 + 6 + 6, nectary +, staminate flowers; gynostemium +; A synandrial, pollen sacs in rings, extrorse, pollen tricolpate, or apertures 0, psilate; carpelate flowers: G ± inferior, ovules tenuinucellate, style very short; fruit baccate; testa thin-walled, exo[?]tegmen massively lignified; endosperm +, embryo undifferentiated. - 3/23(+). New World from California and Florida southwards, Mediterranean and S. W. Asia, S. W. Australia and E. Africa.

Recorded hosts include Fabaceae, Salicaceae, Burseraceae, and Meliaceae.

For information, see Kuijt (1969: general), Takhtajan et al. (1985: pollen), Blarer et al. (2004: floral morphology) and a great deal of information in the Parasitic Plants website (Nickrent 1998 onwards) and also Heide-Jørgensen (2008).

FAGALES Engler - [Back to Index]

Ectomycorrhizae common; leaf with secondary veins proceeding straight to non-glandular teeth; plants monoecious, inflorescences congested, flowers in compact cymose clusters [i.e. more than one flower/main inflorescence bract], very small; P +, pollen ± spinulate, nectary 0, 2 apical unitegmic ovules/carpel, poorly developed at pollination, fertilization delayed, style ± 0, stigmas ± decurrent; fruits dry; testa vascularized, not mechanical, exotesta often enlarged and persisting; cotyledons large. - 8 families, 55 genera, 1877 species.

Fagales are not particularly speciose (e.g. Magallón & Sanderson 2001), but they can dominate the forests in which they grow, as is quite common in plants with ectomycorrhizae, and this is particularly true of Fagaceae.

The pollen of Betulaceae, Rhoipteleaceae and Juglandaceae, and to a lesser extent that of Fagaceae, is rather like that of the distinctive Normapolles type abundant in north temperate regions in the Turonian-Campanian of the Cretaceous, some 98-94Ma before present (Friis et al. 2003a for a summary; Crepet et al. 2004 for a list of early records of Fagales fossils). Normapolles pollen is oblate in shape (i.e. it is a radially symmetric grain in which the polar axis is shorter than the equatorial diameter) and is triaperturate with protruding, elaborate and strongly thickened aperture regions, hence the pollen is triangular in transverse section or when viewed from above. The apertures have internal pores and externally short colpi or pores (details from Friis et al. 2003a). Fossil flowers producing pollen of the Normapolles type are perfect, rarely imperfect (c.f. extant Fagales), and with a simple, undifferentiated perianth. Pollination in some fossil taxa assigned to Fagales (the fossils are ca 84 Ma old - see Herendeen et al. 1999) may have been somewhat different from that of the extant members. Thus Antiquacupula appears to have nectaries at the base of the stamens (Sims et al. 1998; Herendeen et al. 1999). Other details of the the morphology of some of these fossil flowers are also distinctive. Normanthus, from the late Cretaceous of Portugal, has perfect flowers with five perianth members that alternate with the stamens, there are two collateral carpels with separate and quite long style branches, and the placentation is described as being parietal (Schönenberger et al. 2001b). Endressianthus has imperfect flowers, and in the staminate flowers the stamens alternate with the tepals; Dahlgrenianthus has perfect flowers and a superior ovary with more or less separate styles (Friis et al. 2006); while Archaefagacea has a tricarpelate gynoecium, two ovules per carpel and sometimes three-seeded fruits. Schönenberger et al. (2001b) and Friis et al. (2003a) give useful tables comparing the morphology of extant and fossil members of the order.

Wind pollination and monoecy now pervades Fagales. Delayed fertilization, chalazogamy, and also intermittent pollen tube growth (see Sogo & Tobe 2006ad for a summary) are also common, although chalazogamy itself is perhaps unlikely to be plesiomorphic in the order (the situation is unknown for Nothofagaceae, Fagaceae are porogamous). Delayed fertilization is associated with the immaturity of the ovules at pollination and competition between the ovules, indeed, in Corylus avellana the ovules do not even begin to develop until after pollination (Germain 1994).

Lycaenidae caterpillars are quite commonly to be found on members of this order (see Fielder 1991). Phyllonorycter leaf-mining moths (Lepidoptera - Gracillariidae - Phyllocnistinae) are especially speciose on Fagales, about half the known host records being from this group (Lopez-Vaamonde et al. 2003), but their diversification seems to have occurred in the region of 50.8-27.3Ma before present, well after the order itself originated (see above), and after the leaf-miner clade itself evolved, some 76.3-50.3Ma before present (Lopez-Vaamonde et al. 2006). Both Heterobathmidae and Eriocraniidae, clades rather "basal" in the lepidopteran tree, are found on Fagales, on Nothofagaceae, and Fagaceae plus Betulaceae, respectively (Shields 1988). Rusts on Fagales are predominantly to be placed in Pucciniastraceae, also found on ferns (Savile 1979). A particular Frankia clade involved in nitrogen fixation is restricted to Fagales, although members of another clade are also to be found here (Clawson et al. 2004).

In the late 19th century and early twentieth century in particular, a number of botanists thought that nearly all wind-pollinated angiosperms formed a single group, Amentiferae, that were primitive within flowering plants, and the chalazogamy common in the order was even thought to be intermediate between fertilization as it occurred in some gymosperms and the porogamy that characterises most angiosperms (e.g. Nawaschin 1895). The other main hypothesis was that plants like Magnoliaceae with large flowers and many, free, spirally-arranged parts were primitive. Be that as it may, Fagales are the core of the old "Englerian" Amentiferae which have since been comprehensively demolished, other members finding resting places among many otherwise entirely unrelated groups within the eudicots such as Malpighiales (Salicaceae), Proteales (Platanaceae), and Rosales (Ulmaceae and relatives: e.g. Qiu et al. 1998). There are no immediate relationships with hamamelid taxa such as Altingiaceae and Hamamelidaceae (see Saxifragales here) that were previously thought to be intermediate between Amentiferae and more conventional broad-leaved angiosperms. Fagales as circumscribed here comprise Faganae and two and a half other superorders in Takhtajan (1997).

Relationships within Fagales are becoming fairly well resolved (see tree), although the position of Myricaceae is somewhat uncertain. Manos and Steele (1997) show Myrica as immediate sister to Betulaceae, etc., in a matK and combined matK + rbcL analysis, although support was weak, but sister to all Fagales except Nothofagaceae and Fagaceae in a rbcL analysis. The latter set of relationships was also found by Li et al. (2002) using trnL-F sequence data, but with only 61% bootstrap support. Li et al. (2004: six genes, all three genomes) found Myricaceae to be sister to [Juglandaceae + Rhoipteleaceae], although the support still was not strong; the tree here follows the topology in this last paper. Herbert et al. (2006: three genes) find the same set of relationships, but again with little support for the position of Myricaceae.

For additional information, see Abbe (1974: flower and inflorescence morphology), Giannasi (1986: chemistry), Zavada and Dilcher (1986: pollen), Hickey and Taylor (1991: vegetative morphology), Feur (1991: pollen), and Xing et al. (1998: embryology).

Includes Betulaceae, Casuarinaceae, Fagaceae, Juglandaceae, Myricaceae, Nothofagaceae, Rhoipteleaceae, Ticodendraceae.

NOTHOFAGACEAE Kuprianova - Peltate glandular hairs +; stipules peltate, enclosing colleters; fruit surrounded by cupule. - 1/35. New Guinea to South America.

For the biogeography of Nothofagus, a much-discussed subject, see Swenson et al. (2001) and Knapp et al. (2005). The family is perhaps 90 Ma old, and the current distribution of the genus has often been explained by vicariance, i.e. continental drift; fossils of all four subgenera are known from the Late Campanian in Antarctica (Swenson et al. 2001). However, Knapp et al. (2005) suggest that Nothofagus reached New Zealand, at least, by long distance dispersal only ca 30Ma before present (see also Waters & Craw 2006).

The rust parasites of Nothofagaceae are rather different to those of other Fagales (Savile 1979). The inaperturate discomycete Cyttaria is found on Nothofagus in both the Antipodes and in South America, but not in New Guinea. Humphries et al. (1986: this work needs to be re-evaluated in the context of recent ideas of relationships) discuss the parasites and associates of Nothofagus, and suggest the possibility of some co-evolution of the genus with Eriococcus scale insects that grow on it; the moth Heterobathmia, a genus perhaps 125 Ma old, part of a clade sister to all other Lepidoptera, makes its home exclusively on Nothofagus, both as an adult (it has jaws, and eats pollen) and as a larva (see also Futuyma & Mitter 1996; c.f. Winteraceae).

Staminate flowers of Nothofagaceae that apparently have many stamens are interpreted as being the result of fusion of dichasia, so they are pseudanthia.

Further information may be found in Poole (1952: seed development), Philipson and Philipson (1988: classification), Kubitzki (1993b: general, in Fagaceae), Manos (1997: relationships), and Rozefelds (1998) and Rozefelds and Drinnan (1998: stamens and staminate flowers).

Fagaceae [Myricaceae [Juglandaceae + Rhoipteleaceae]] [Casuarinaceae [Ticodendraceae + Betulaceae]]: anthers dorsifixed.

FAGACEAE Dumortier - Hairs often stellate/branched; nests of sclereids containing rhomboidal crystals in bark; fruit surrounded by cupule [if with valves, one more valve than fruit number], ± spiny, individual nuts trigonous, endocarp hairy inside. - 7/670. More or less worldwide.

1. Fagoideae K. Koch - Ellagic acid 0; inflorescence capitate; cotyledons folded. - 1/10. Temperate N. hemisphere.

2. Quercoideae Õrsted - 6/640: Quercus (400), Lithocarpus (120: fruits like those of Quercus), Castanopsis (110). N. temperate, at higher elevations in the tropics, not S. Africa or New Zealand, barely in Australia.

• The family is readily recognisable in fruit. The distinctive cupules enclose one to three fruits, and the seed itself is often surrounded by dense hairs coming from the endocarp. The outer surface of the wood is characteristically furrowed, and the inflorescences are erect to pendant, spicate to catkinate. The leaves are stipulate and usually spirally inserted, serrate, and rather leathery, and the hairs are stellate or branched.

Theclines (Lycaenini) are commonly found on this family (Ehrlich & Raven 1964). Indeed, oaks support the highest diversity of herbivores of all temperate holarctic forest trees (Kelly & Southwood 1999), including an extensive radiation of hundreds of species of the species-rich gall wasps, Cynipini (Csóka et al. 2005; Stone et al. 2009). By some estimates half of all galls in the north temperate region are found on the family (Mani 1964; Abrahamson et al. 1998). These Cynipini show very considerable host plant conservatism, members of major clades usually having associations with similar oaks (Stone et al. 2009).

Castanea dentata, previously the dominant large tree in extensive areas of forest in eastern North America and an important source of food for humans and other animals, has been utterly devastated by the introduced ascomycete fungus Cryphonectria parasitica (Endothia parasitica) over a period of less than forty years. The species may persist for some time after infection because it continues to sucker from collars of the old trees or from stumps, but the suckers practically never reach reproductive age (Schlarbaum et al. 1997).

Fertilization is porogamous, according to Johri et al. (1992), although it is much delayed (Sogo & Tobe 2006d and references; Deng et al. 2008). Quercoideae such as Castanea and Castanopsis are insect pollinated.

Members of Fagaceae are often very common in north temperate areas, but also on hills and mountains in Central America and Malesia, and they produce large numbers of fruits. When their abundance is combined with the tendency of a number of species to show masting behaviour, their effect on the animals that depend on these fruits is considerable. Red oaks (section Lobatae) take one and a half years to mature their fruits and the seeds are high in tannins and lipids; the fruits of white oaks (section Quercus) mature in about six months and their seeds are less rich in tannins and lipids, furthermore, their seeds tend to germinate faster. Much has been written about the behavior of animals that eat and disperse the acorns of these species. Thus squirrels tend to eat the embryos of white oaks before caching the fruits, although in general they prefer to eat acorns of white oaks in the fall and to cache those of red oaks (Wood 2005 for literature).

There has been infinite discussion over the morphological nature of the small protrusions surrounding the ovary, and the whole complex is often interpreted as a modified cymose inflorescence (Oh & Manos 2008 for references). When the cupule has valves, probably the plesiomorphic condition, there is one valve more than the number of fruits. Oh and Manos (2008) suggest that the cup-shaped cupule, more or less smooth to scaly or spiny and enclosing a single, rounded fruit, has evolved more than once within Quercoideae.

The taxonomic history of Trigonobalanus is interesting. It is common on the often-visited Fraser's Hill in Peninsula Malaya, but it was described only some 40 years ago from Mt Kinabalu in Borneo, then found in S. America less than 20 years ago, then from fossils in N. America. The three extant species have been placed in three genera, but they form a single clade (Nixon & Crepet 1989 for information).

Fagus is sister to rest of Fagaceae (Manos et al. 1993); Quercoideae s. str. are paraphyletic, and Trigonobalanus (= Trigonobalanoideae) is sister to the rest of Quercoideae + Castaneoideae, all three of which are here combined as Quercoideae (see Chen et al. 2007 for references to different classifications). Manos and Stanford (2001), Manos et al. (2002) and Oh and Manos (2008) discuss the phylogeny, character evolution, and biogeography of the family. Recent studies (Oh & Manos 2008) suggest that Lithocarpus is polyphyletic, the South East Asian members grouping with Chrysolepis, while L. densiflorus, the single species from West North America (in the California floristic province) groups with Quercus, Castanopsis, and Castanea. For phylogenetic relationships within Quercus, see Manos et al. (1999) and Oh and Manos (2008); New and Old World species are in separate clades.

For information, see Kubitzki (1993b: general) and Govaerts and Frodin (1998: checklist and bibliography for the family).

[Myricaceae [Juglandaceae + Rhoipteleaceae]] [Casuarinaceae [Ticodendraceae + Betulaceae]]: myricetin +; pollen pororate, G [2]; fertilization chalazogamous.

[Myricaceae [Juglandaceae + Rhoipteleaceae]]: peltate glandular hairs +; leaves spiral, stipules 0; one flower/bract; ovule single [per flower], straight.

The evolution of features like stipules, inflorescence type, and ovule morphology is particularly difficult to understand; they could be synapomorphies of the clade as a whole (as above), or be independent apomorphies of Myricaceae and Juglandaceae. Herbert et al. (2006) discuss possible synapomorphies of this clade.

MYRICACEAE Kunth - Ovule basal, fertilization porogamous. - 3/57. ± Cosmopolitan, including New Caledonia but not Australia.

• Myricaceae are woody plants recognisable by the aromatic glands that cover the plant coupled with the spiral, simple, rarely deeply lobed leaves.

Fertilization is porogamous in those Myricaceae studied, but it is delayed as in other Fagales, the growth of the pollen tubes as it were pausing on the nucellar surface; this method of fertilization, described as pseudoporogamy, may be derived (Sogo & Tobe 2006a, b).

Although the ovary appears to be superior, as in Comptonia, it is often so highly reduced that any traces of its inferior construction would be lost, however, in Canacomyrica, from New Caledonia, staminodes are borne on top of the ovary and there is a six-lobed perianth. In some species of Myrica the ovary is invested by tissue from a meristem developing below the flower, even below the bracteoles, which are then borne on the flower.

For the staminate flowers, see Macdonald (1978), for general information, see Kubitzki (1993b), and for wood anatomy, see Carlquist (2002c).

Juglandaceae + Rhoipteleaceae: leaves odd-pinnate; P 4.

JUGLANDACEAE Perleb - intrusive; seeds large, pachychalazal, cotyledons much folded. - 7-10/50. North temperate, S. to Argentina and Malesia.

1. Engelhardioideae Iljinskaya - Leaves even-pinnate; bracts 3-lobed; nuts with a fibrous layer. - 3-4/14. Himalayas to Malesia, Mexico to Colombia.

2. Juglandoideae Manning - Bracts unlobed; nuts with sclereids in shell. - 3-6/35. Temperate N. hemisphere, only 1 sp. in Europe, Central America and Andes.

• Juglandaceae may be recognised by the small, aromatic glands on the young parts of the plant; the compound, estipulate, usually spiral leaves the leaflets of which are often serrate; and the spicate inflorescences with single, axillary flowers. The endocarp of the drupaceous fruit has complex intrusions.

The oldest fossils assignable to Juglandaceae may be from some 98-83Ma before present (Crepet et al. 2004 for references) or 78Ma before present (Manos et al. 2007, based on the age of Caryanthus). However, details of the timing of diversification within the family are unclear, there being great variation of estimates in Juglandoideae in particular (Manos et al. 2007). Several genera are fossil in North America and especially Europe that are not found there now (Manchester 1987, 1991), and there are several extinct genera, some showing very interesting combinations of characters, from early Caenozoic deposits in North America - clearly the family was previously very diverse (Elliott et al. 2006). It seems that taxa with biotic means of dispersal evolved in the early Caenozoic from taxa that were probably dispersed by wind (Tiffney 1986).

Sirococcus clavigignenti-juglandacearum, an imperfect fungus described as recently as 1979, is currently devastating Juglans cinerea, a North American species important both for its fruits and wood (Schlarbaum et al. 1997).

Manos and Stone (2001; see also Gunter et al. 1994) provide a phylogeny and revised classification of the family; adjustments to current generic limits are needed.

For general information, see Stone (1993), for fertilization, Luza and Polito (1991), for possible apomorphies, Judd et al. (2007).

RHOIPTELEACEAE Handel-Mazzetti - Buds lacking scales; leaves two-ranked, stipules +, asymmetrically caudate; 3 flowers/bract, central apparently perfect; pollen colporate, ovary superior, ovule bitegmic, campylotropous. - 1/1: Rhoiptelea chiliantha. China.

• Rhoipteleaceae may be recognised by their stipulate and odd-pinnately compound leaves with serrate leaflets, indumentum of peltate scales, and branched, pendulous inflorescences with flowers in triads, the central flower apparently being perfect and having a superior ovary.

The ovary is presumably secondarily superior.

For information, see Wu and Kubitzki (1993: general) and Sun et al. (2006: the breeding system).

Casuarinaceae [Ticodendraceae + Betulaceae]: stigmas elongate; fertilization chalazogamous.

Sogo and Tobe (2008) suggest that the chalazogamous fertlization that occurs in all families of this clade is similar down to the details of where the pollen tube growth is temporarily delayed.

CASUARINACEAE R. Brown - Rootlets clustered, of limited growth; nodes 1:1; stomata usu. tetracytic, transversely oriented; leaves 4-16-whorled, connate, scale-like, stipules 0; inflorescence capitate-spicate, 1 flower/bract; staminate flowers: P ["inner bracteoles"] 2, A 1, filament incurved in bud, anther ± longer than connective; carpelate flowers: bracteoles large; G naked, only abaxial fertile, ovules straight, bitegmic; fruit a samara, freed as the much accrescent bracteoles separate. - 4/95. South East Asia and Malesia to the S.W. Pacific, esp. Australia.

• Casuarinaceae are easily recognised, looking as they do rather like a conifer. The minute leaves are whorled, and the carpelate inflorescence forms a small cone. The bracteoles that gape widely at maturity to release the samara fruits with their apical wings are very distinctive.

Casuarinaceae fossils are known from Caenozoic deposits in South Africa and Argentina (Coetzee & Muller 1984); material from the Eocene of Patagonia has been identified as Gymnostoma (Zamaloa et al. 2006). Nitrogen fixing is known from the family, and Casuarina plays an important role in agriculture in parts of montane New Guinea, both in providing firewood and in fixing nitrogen.

The texture of what some have called the outer and inner bracteoles is very different; the latter are called the perianth above.

Although the monophyly of Causarina has never been in doubt, it has been split into four genera, themselves probably monophyletic. Gymnostoma is sister to the rest of the family and has many plesiomorphous features (Steane et al. 2003): Both carpels are fertile (although this is likely to be an apomorphy, given the situation in the rest of the order), with 2 ovules/carpel, its stem stomata are not hidden, etc.

For further information, see Johnson and Wilson (1993: general) and Sogo et al. (2001: fertilization).

Ticodendraceae + Betulaceae: nests of sclereids containing rhomboidal crystals in bark; leaves two-ranked, doubly serrate.

TICODENDRACEAE Gomez-Laurito & L. D. Gomez P. - Hairs T-shaped, unicellular, not glandular; stipules encircling the stem; G with divided loculi, ovules with massive integument 20-30 cells across. - 1/1: Ticodendron incognitum. Central America.

• Ticodendraceae are wind-pollinated, evergreen trees that may be recognised by their doubly-serrate leaves that have stipules encircling the stem, and carpelate flowers with two, long stigmas.

Fruits assignable to Ticodendron have been found in Eocene deposits from Oregon and in the London Clay (Manchester & Renner 2005).

For information, see Carlquist (1991: wood), Feuer (1991: pollen), Tobe (1991: floral morphology), Kubitzki (1993b: general), Sogo and Tobe (2008: fertilization) and Govaerts and Frodin (1998: bibliography).

BETULACEAE Gray - Anthers longer than connective, pollen tube branched. - 6/145. North Temperate, to Andes and Sumatra.

1. Betuloideae Arnott - Peltate glandular hairs +; carpelate flowers: P 0; infructescence with woody or scaly bracts, nut separate, small, ± flattened and samaroid. - 2/95: Betula (60). N. hemisphere, to South America; montane in tropics.

2. Coryloideae J. D. Hooker - Staminate flowers: P 0; A hairy; infructescence with leafy bracteoles [from one or two orders of branching] remaining associated with the fruit; nuts large, not or little flattened. - 4/50. N. Temperate, South East Asia, Central America.

• Betulaceae are usually deciduous trees that may be recognised by their stipulate, doubly serrate leaves, catkinate inflorescences (both staminate and carpelate), and fruits are that are either small and flattened or associated with large, leafy bract-like structures.

Alnus fixes nitrogen and is often planted in soil remediation efforts. In Corylus avellana in particular three to five months may elapse between pollination and fertilization, ovules starting to develop about half way through this period. The young nuts are 7-10 mm across at the time of fertilization. If pollination does not occur, the stigma may remain receptive for up to three months (Germain 1994).

The oldest fossils assignable to the family are from 94-83Ma before present (Crepet et al. 2004 and Forest et al. 2005 for references). Normanthus and Endressanthus, from the late Cretaceous of Portugal, may be close to the root of the Betulaceae clade (Friis et al. 2005); the former has has perfect flowers with five perianth members that alternate with the stamens, and the placentation is described as being parietal (Schönenberger et al. 2001b; Friis et al. 2003a).

Alnus has adaxial prophylls. Staminate flowers in Coryloideae are sometimes reported as being single (e.g. Mabberley 1997), however, as Abbe (1935) noted, there are usually three together in the axil of each inflorescence bract; the flower of Ostrya, with some 15 pairs of half stamens, is pseudanthial in origin, being derived from these three flowers (Abbe 1935, 1974; Macdonald in Sattler 1973). In staminate flowers of Corylus the perianth is reduced to a ridge. The carpelate flower is so reduced that ovary does not always appear to be inferior.

Although Li et al. (2004) suggested that Betuloideae were paraphyletic, with Alnus and Betula being successively sister to the rest of the family, Forest et al. (2005), analysing variation in ITS and the 5S spacer, recovered the two subfamilies as monophyletic. The monophyly of Ostrya and Carpinus is unclear, as are many of the basic characters of Coryloideae (c.f. Yoo & Wen 2002, 2007).

See Abbe (1935: flowers, inflorescences), Crane (1989: fossils), Kubitzki (1993b: general), Manchester and Chen (1998: fossils).

MALVIDAE = [GERANIALES + MYRTALES] [PICRAMNIALES [SAPINDALES [HUERTEALES [MALVALES + BRASSICALES]]]]: ?

Based on a study of the genome of Arabidopsis, De Bodt et al. (2005, see also Maere et al. 2005) suggest there was a duplication of the whole genome somewhere in this part of the tree some 109-66Ma before present, although given the uncertainty over the dating of this duplication and relationships within rosids, exactly where this should go on the tree is unclear.

[GERANIALES + MYRTALES]: ellagic acid +; K persistent in fruit.

GERANIALES Dumortier - [Back to Index]

Ellagic acid +; inflorescence cymose; nectary outside A; A obdiplostemonous; K persistent in fruit; seed testal. - 5 families, 17 genera, 836 species.

Geraniales are poorly understood, and although a smallish group, they are morphologically quite heterogeneous, and until recently there was uncertainty over their exact phylogenetic position. This has made thinking about apomorphies difficult (Kubitzki 2006a). Geranium, the only representative of the order studied, was sister to all other rosids except Vitaceae (Zhu et al. 2007, support weak). Their position is unstable in an rbcL analysis of all angiosperms (Hilu et al. 2003). Savolainen et al. (2000a) found the order to be monophyletic, but with only 52% support (see also Savolainen et al. 2000b); Crossosomatales were its sister group, but with still less support.

Geraniaceae and Vivianiaceae have a layer of small hypodermal druse-containing cells in the calyx (Kenda 1956). Indeed Weigend (2006: p. 217) suggested that there might be "a close and possibly exclusive relationship between Geraniaceae and Ledocarpaceae [= Ledocarpaceae and Vivianaceae here]." He lists numerous characters suggesting such a relationship, and these will need to be evaluated critically even if Geraniaceae and Ledocarpaceae are not sister taxa, as seems likely.

Includes Francoaceae, Geraniaceae, Ledocarpaceae, Melianthaceae, Vivianiaceae.

GERANIACEAE Jussieu - 7/805. Temperate and warm temperate.

1. Hypseocharis - Stipules 0; A 15, style filform, stigma capitate; fruit a loculicidal capsule; seeds with scanty endosperm, cotyledons spiral. - 1/1-3. S.W. Andean South America.

2. The Rest - sepals with nectariferous spurs, aristate; 1-2 apical ovules/carpel, true style short, stout, stigma lobed; fruit with upper part of ovary elongating [the "stylar" beak], septicidal, separating into mericarps which curl upwards and separate from columella, whether or not seeds disperse separately; embryo curved, cotyledons longitudinally folded. - 5/805: Geranium (430), Pelargonium (280), Erodium (80), Monsonia (40: inc. Sarcocaulon). Temperate and warm temperate, esp. southern Africa.

• Geraniaceae are often herbs with jointed stems; the stipulate leaves have palmate venation or are palmately or pinnately compound. The inflorescence is cymose, and all parts of the flower apart from the carpels are free. The sepals are aristate, the androecium is obdiplostemonous, the carpels are usually beaked, there are few ovules per carpel, and the embryo is curved.

Struck (1997) and Bakker et al. (2004, 2005) discuss the phylogeny and diversification of Pelargonium in the Cape region; there is striking vegetative and floral variation, and almost 100 species are geophytes (Procheŝ et al. 2006). Diversification of Pelargonium (and Monsonia) occurred ca 30-10Ma before present as aridification set in, while diversification of the Geranium-Erodium clade occurred at roughly the same time in Eurasia and the Mediterranean, perhaps in response to climate change and mountain uplift (Fiz et al. 2008). In Erodium in particular (Fiz et al. 2006 for a phylogeny) there has been substantial dispersal.

Pelargonium, the "geranium" of the window sill, has monosymmetric flowers with patterning on the adaxial tepals, the single, tubular nectary is on the adaxial side of the pedicel. Darwin (1859) noted that the central flower of a Pelargonium inflorescence might lose its adaxial markings and also the nectary; this would be expected of a peloric flower as the adaxial petals become more like the abaxial petals (c.f. the peloric flower of Linaria vulgaris with its five spurs - there is normally one, abaxial in position). Monsonia has an androecium consisting of of five groups of three laterally connate stamens.

Palmer et al. (1987) and Chumley et al. (2006) noted extensive expansion of the chloroplast inverted repeat in Pelargonium, and the chloroplast genome is the largest in all flowering plants. There have been very great increases in the rate of evolution of the mitochondrial gene nad1 throughout this part of the family, and especially in Pelargonium, but not in Hypseocharis (Parkinson et al. 2005; Bakker et al. 2006; see also Palmer et al. 2000). Hypseocharis is sister to the other members of the family (e.g. Price & Palmer 1993). Other relationships are [Pelargonium [Monsonia [Geranium + Erodium]]]; the position of the monotypic Geranium californicum is unclear (Fiz et al. 2008). Whether or not Hypseocharis is recognised as a family is of little importance, however, this is an option in A.P.G. II (2003); there are certainly phenetic differences between it and the Geraniaceae s. str.

Hypseocharis used to be included in Oxalidaceae (Hutchinson 1973; Cronquist 1981), but nectaries, testa anatomy, etc., place it unambiguously here (e.g. Boesewinkel 1988; Rama Devi 1991). Also, its leaflets are not strongly articulated with the petiole, as they are in Oxalidaceae. Geraniaceae used to include taxa like Biebersteinia (Biebersteiniaceae - Sapindales) and Dirachma (Dirachmaceae - Rosales).

For more information, see Narayana (1970: embryology), Link (1994: nectaries), Erbar (1998: floral morphology), Meisert et al. (2001: seed coat), Aldasoro et al. (2001: Monsonia, perhaps paraphyletic), and Albers and Van der Walt (2006: general).

[Melianthaceae + Francoaceae] [Vivianaceae + Ledocarpaceae]: ?

Melianthaceae + Francoaceae: leaves with rather broad insertion; inflorescence terminal, racemose, sterile bract(s) at apex, bracteoles 0; ovary with intrusive placentae, style long; endosperm copious, embryo short.

Obdiplostemony develops after the initiation of the androecium, which is initially diplostemonous. The nectaries are not vascularized.

Erstwhile Greyiaceae + Francoaceae make a strongly supported pair (Morgan & Soltis 1993; Price & Palmer 1993; Soltis & Soltis 1997). Neither family has myricetin, common in Saxifragaceae with which Francoa and Tetilla have often been associated or even included in the past. Melianthaceae and Francoaceae are placed in a single family by Savolainen et al. (2000b); many of the family characters given there may be plesiomorphic. A.P.G. II (2003) suggest as an option separating Melianthaceae and Francoaceae, an option which is followed here.

For information on seed anatomy, see Nemirovich-Danchenko (1995), Danilova (1996) and Corner (1976: Melianthaceae), for anatomy, see Gregory (1998); Linder (2006) provides a general account of the two families under Melianthaceae.

MELIANTHACEAE Berchtold & J. Presl - 3/11. Africa.

• Melianthaceae are at most small trees that may be recognised by their serrate leaves or leaflets, and large, toothed stipules. They have terminal racemes of large, more or less monosymmetric flowers with all the parts (apart from the ovary and sometimes calyx) free; the nectary is outside the stamens.

In Greyia the leaf sheaths are adnate to the stem; the sheath, petiole and lamina are detached together, helped by the activity of a cambium.

For node and petiole anatomy, see Hilger (1978a, b), for general anatomy, see Gornall and Al-Shammary (1998), for floral development, see Ronse Decraene et al. (2001b), for anthers, see Endress & Stumpf (1991), and for flower and fruit, see Doweld (2001a); see also Hideux and Ferguson (1976: pollen), Steyn et al. (1986: embryo), and Steyn and van Wyk (1987) and Ronse Decraene and Smets (1999), both floral development.

FRANCOACEAE A. Jussieu - Herbaceous. - 2/2. Chile.

Francoa: A 8, capsule with loculicidal dehiscence, Tetilla: stamens equal and opposite to sepals, capsule with septicidal dehiscence (both described as diplostemonous and septicidal in Linder 2006).

Vivianiaceae + Ledocarpaceae: leaves opposite, simple, stipules 0, but a line across the stem; inflorescence terminal; K aristate, C contorted, pollen with many pores, style short, stigmatic lobes long; endosperm walls thick, pitted, embryo curved.

The two are placed in a single family by Savolainen et al. (2000b), but the support is not strong. However, given that apomorphies for the two families are hard to come by, combination may be in order.

See Boesewinkel (1997: ovules and seeds) and Weigend (2005: floral morphology and pollination, 2006: general; under Ledocarpaceae).

VIVIANACEAE Klotzsch - Nodes 1:1. - 1 or 4/6. Chile, S. Brasil.

The inflorescence may be represented all or in part by a branched thorn.

See also Lefor (1975: general), Carlquist (1985b: wood anatomy), Narayana and Rama Devi (1995: general).

LEDOCARPACEAE Meyen - 3/12. W. South America, especially the Andes.

The leaves can be deeply lobed. The fruits of Balbisia may be more or less septifragal (with a loculicidal slit, too) or clearly loculicidal.

Rhyncotheca is a morphologically very distinctive genus: it lacks a corolla, its fruit is a pointed, long-ovoid, septicidal capsule with a columella, and it has thin-walled endosperm and a sraight embryo. Note that the fruits lack a true beak, unlike the apparently rather similar fruits of Geraniaceae. Rhyncotheca is possibly wind pollinated.

MYRTALES Reichenbach - [Back to Index]

Flaky bark; methylated ellagic acid +; cork pericyclic; pits vestured; internal phloem +; nodes 1:1; leaves opposite, (small stipules +), colleters +; hypanthium +, nectariferous, K valvate, stamens incurved in bud, pollen with pseudocolpi, style single, long; K persistent in fruit; endotesta crystalliferous; endosperm slight. - 11 families, 380 genera, 11027 species.

• Myrtales may be recognised by their often flaky bark; opposite leaves with colleters, at most small stipules and strong intramarginal veins; flowers with petals that are clawed or at least very narrow at the base, a more or less inferior ovary, and a usually nectariferous hypanthium. The sepals are often valvate and the stamens are usually incurved in bud.

Myrtales contain ca 6% core eudicot diversity. Caterpillars of Lycaenidae butterflies are quite commonly to be found on members of this order, especially on Lythraceae, Myrtaceae, Combretaceae (Fielder 1991, 1995).

There are a number of other characters that are common in Myrtales and may be apomorphies for them. Polyderm (alternating endodermal and parenchymatous layers laid down by a pericyclic meristem) is known from families like Onagraceae, Lythraceae, Myrtaceae, and probably Penaeaceae and Oliniaceae, at least (Mylius 1913), while in roots of some aquatic Lythraceae, Melastomataceae, Myrtaceae and Onagraceae (and Euphorbiaceae and Fabaceae) there is a distinctive cork that has air spaces and comes from a pericyclic cork cambium (Little & Stockey 2006). Opposite leaves predominate in the order and may be an apomorphy for it. Weberling (2000) notes that "true rudimentary stipules" occur in Myrtaceae and most myrtalean families; stipules, when they occur, are indeed generally small and may be developmentally connected with the colleters common in the order. Unlike many other stipulate groups (but like Celastraceae), the nodes are predominantly unilacunar, not trilacunar. Many Myrtales, including some Myrtaceae, have notably narrow petal bases, i.e., the petals are clawed or close to being clawed.

Those taxa in which the stamens are straight in bud have short filaments, and the length of the style is correlated in part with the length of the hypanthial tube. Ovary position is quite variable and may even change during ontogeny, as in Vochysiaceae.

For further information, see Beusekom-Osinga and Beusekom (1975: morphology etc. around Crypteroniaceae), Johnson and Briggs (1985: morphological phylogeny), Van Vliet and Baas (1975, 1985: vegetative anatomy), (Graham et al. 1993: chromosome number), Tobe and Raven (1983a, 1985a, 1985b, 1987a, 1990: embryology), Dahlgren and Thorne (1985: general, and other papers in same place), Boesewinkel and Venturelli (1987: ovule and seed) and Almeda (1997: chromosome numbers).

Includes Alzateaceae, Combretaceae, Crypteroniaceae, Lythraceae, Melastomataceae, Memecylaceae, Myrtaceae, Oliniaceae, Onagraceae, Penaeaceae, Rhynchocalycaceae, Vochysiaceae.

COMBRETACEAE R. Brown - Hairs unicellular, thick-walled, with a basal internal compartment; ovary unilocular, ovules apical; fruits single-seeded, indehiscent, ± flattened and/or winged; seed large. - 14/500. Largely tropical.

• Combretaceae may not be very easy to recognise, although like other Myrtales they often have opposite leaves and the bark is often flaking. Trees quite often look rather pagoda-like, the branching following the Terminalia pattern, the branches consisting of flattened sprays of leaf rosettes. The leaves are entire and may have pellucid dots, surface glands (also on the petiole), or domatia. The indumentum is commonly adpressed and distinctive enough under the microscope; nodes are unilacunar. The flowers are often rather small and massed in spicate inflorescences, the hypanthium can be quite long, but the corolla is small or absent. The ovary is inferior, and the single-seeded fruit is often flattened and/or winged.

Strephonema (stomata paracytic; K 5, C 5, clawed; A 10, pollen lacking pseudocolpi; G half inferior, 2 pendulous ovules; cotyledons large, hemispherical) may be sister to the rest of the family. Lumnitzera and Laguncularia, both mangrove plants (see Rhizophoraceae for more details), are sister taxa, but Conocarpus, also found in similar habitats, is not immediately related (Tan et al. 2002).

Some information is taken from Graham (1964: general), Verhoeven and van der Schijff (1974: anatomy), El Ghazali et al. (1998: pollen) and Stace (2006: general).

Onagraceae + Lythraceae: vessels in groups in the wood; C clawed, ovule with several megasporocytes; seed with fibrous exotegmen; x = 8.

Decodon is the only typically pentamerous genus in Lythraceae (Graham 2006) as is Ludwigia in Onagraceae, and both seem to be sister to the rest of their respective families. Four-merous flowers could be a synapomorphy for the two families, or independently derived within them. Futhermore, many Lythraceae have racemose (indeterminate) inflorescences; genera with such inflorescences include Decodon, Lythrum and Peplis, which in some combination may be sister to the rest of that family (Graham et al. 2005). Hence where a number of characters are to be placed on the tree is unclear.

ONAGRACEAE Jussieu - Raphides +; inflorescence racemose, stamens straight, anther loculi with numerous subdivisions, pollen with viscin threads, exine paracrystalline, beaded, embryo sac monosporic, 4-nucleate, stigma capitate; fruit a loculicidal capsule opening down the sides; endosperm nuclear, diploid. - 18/650. World-wide, esp. America.

1. Ludwigioideae W. L. Wagner & Hoch - Flowers 4-5-merous, hypanthium 0, pollen in tetrads, nectary on top of ovary, ovary with central vascular bundles, megasporocyte 1, style short. - 1/82: World-wide, esp. America.

2. Onagroideae W. L. Wagner & Hoch - Included phloem +; flowers 4-merous, hypanthium long, deciduous, ovary with transseptal vascular bundles [the dorsal placental bundles are not in the central axis but in the ovary walls], no minor stylar bundles, (stigma divided); K deciduous in fruit. - 21/574: Epilobium (165), Oenothera (145: inc. Gaura, etc.), Fuchsia (105), Clarkia (42), Lopezia (22). World-wide, esp. S.W. North America.

• Onagraceae are herbs to shrubs that are recognisable by their usually opposite and quite often toothed leaves with small stipules and the often 4-merous flowers with an inferior ovary and a well-developed hypanthium. The calyx is usually rather thin, reflexed, and deciduous, there are up to eight stamens, and the pollen has viscin threads, so it is conspicuously clumped.

Some caterpillars eat both Vitaceae and Onagraceae (Forbes 1956), and both contain raphides. Different rusts occur on Onagroideae and Jussiaeaoideae (Savile 1979b). For information on floral morphology and its relation to pollination, see details throughout Wagner et al. (2007).

There are some very distinctive floral morphologies in Onagroideae. Thus Circaea has only two petals and two stamens, the latter being opposite the calyx. In the monosymmetric flowers of Lopezia there is only a single extrorse anther, a petal-like staminode, and the two adaxial petals may be recurved, with basal pseudonectaries on their claws. The viscin threads intermixed with the pollen vary considerably in appearance, often being annular-vermiform, but they can also be smooth or irregularly beaded (Skvarla et al. 1976). Gongylocarpus has sessile flowers, as is quite common in Onagraceae, but after pollination the ovary becomes completely enveloped by stem tissue.

In some species of Oenothera all the chromosomes form a ring by a series of permanent translocations, the whole genome thus forming a single linkage unit. Interestingly, development of such rings has occurred many times and is usually not constant within a species (Harte 1993: the contribution of Oenothera to biology; Wagner et al. 2008 and references). Hugo de Vries thought that the abrupt appearance of variants of O. lamarckiana were examples of evolution, which for him was a process in which mutation = major change/speciation, but in which natural selection had little or no role to play. However, it turned out that these morphological variants represented the breakdown of the translocation system (c.f. Linnaeus and Peloria [Plantaginaceae]).

Knowledge of relationships along the backbone of the Onagroideae seems to be stabilising. Hauya, [Fuchsia + Circaea], Lopezia, and Gongylocarpus are successively less basal branches (tribes), while [Epilobeae + Onagreae] form the treminal group (Levin et al. 2003, 2004), however, Ford and Gottlieb (2007) obtained a grouping [Hauya [Fuchsia + Circaea]]. Reflecting the new, but for the most part well supported phylogeny of the family, taxon limits have had to be adjusted, so Oenothera has been expanded and Camissonia very much cut up (e.g. Levin et al. 2004); for appropriate nomenclatural changes, see Hoch and Wagner (2007).

In 1827 Robert Brown recorded the phenomenon that is now called Brownian motion when observing the movement of minute particles (in fact mitochondria, etc.) in the pollen grains of Clarkia pulchella.

For more information, see Eyde (1982: floral anatomy), Tobe and Raven (1986, 1996) and Hoch et al. (1993), all variation in anther septum development and embryology, the Onagraceae website (up-to-date summary of phylogeny, classification, etc.), and Wagner et al. (2007: superb summary, all supraspecific taxa included).

LYTHRACEAE Jaume Saint-Hilaire - C crumpled in bud; testa many-layered. - 31/620: Cuphea (250), Diplusodon (75), Lagerstroemia (55 - A centrifugal), Nesaea (55 - probably to include Ammannia - then = 80), Rotala (45). Tropical, but some temperate.

• Lythraceae are herbs to trees that are recognisable by their flaky bark, opposite entire leaves that lack gland dots, flowers with a ribbed hypanthium or K + C tube, rather thick, valvate sepals, clawed petals that are crumpled in bud, and often 10 or more stamens of unequal lengths. The ovary is more or less superior, and the fruit is usually a capsule.

Taxa whose seeds have mucilaginous hairs are more or less myxospermous. The fruits of Cuphea open by the placenta expanding and moving laterally, breaking through both the thin ovary wall and the hypanthium; the seeds are exposed on the placenta.

The old Sonneratiaceae - Sonneratia plus Duabanga - are not monophyletic (Shi et al. 2000; see also Huang & Shi 2002: one species per genus sampled), and the Southeast Asian-Malesian mangrove genus, Sonneratia itself may be sister to Trapa, a free-floating fresh-water aquatic herb which used to be placed in its own family, Trapaceae. For the evolution of the mangrove habitat, see Rhizophoraceae.

Pollen is notably variable in Lythraceae (A. Graham et al. 1990 and references). The inferior ovary of Punica granatum, the pomegranate, is unique in flowering plants. It appears to have two (or three) superposed layers of carpels, the basal with axile and the others with intrusive parietal placentation; the placentation is fundamentally axile, the appearance of parietal placentation being the result of growth at the base of the ovary (Sinha & Joshi 1959 for vasculature). In the other species of the genus, P. proto-punica, there is an ordinary semi-inferior ovary.

S. Graham et al. (2005) found some support for the topology [Decodon [[Lythrum + Peplis] remainder of the family]]; the genera in the remainder of the family were included in two large clades, inverted mucilaginous hairs in the testa possibly being a synapomorphy. However, support along the whole back-bone was weak, and in some analyses there were two major clades in the whole family, with the three genera just mentioned forming a subclade within one of them. S. Graham et al. (2005) list characters for the major clades that they found.

For additional information, see Weberling (1989 and references: androecial development), S. Graham (2006: general).

[Vochysiaceae + Myrtaceae] [Melastomataceae [Crypteroniaceae [Alzataeaceae [Rhynchocalycaceae [Penaeaceae + Oliniaceae]]]]]: ?

Oil glands are found in the anthers of Myrtaceae and Memecylaceae, and a number of other taxa in the Melastomataceae-Crypteroniaceae clade also have a very much expanded connective. Whether some staminal features - perhaps linked with pollination - are a higher-level apomorphy in this clade awaits further study.

Vochysiaceae + Myrtaceae: hairs simple, 1-2-celled; K and C imbricate, pollen colpi fused [syncolporate]; style depressed in apex of gynoecium.

Sytsma et al. (2004) discuss the age and biogeographic history of the group in some detail.

Wilson et al. (2005: matK only) found Myrtaceae s. str. to have 80% jacknife support, while [Myrtaceae s. str. [Heteropyxidaceae + Psiloxylaceae]] (all together = Myrtaceae here) had ³95% support; a similar set of relationships occurred in Sytsma et al. (2004), although they were not exactly the same in some earlier work (e.g. Conti et al. 1996).

VOCHYSIACEAE A. Saint-Hilaire - Mucilage cells, sclereids +; flowers obliquely monosymmetric, calyx with a single spur; A 1. - 7/190. Lowland tropical America, W. Africa.

• Vochysiaceae are readily recognisable by the combination of a rather ordinary Myrtalean vegetative body - their leaves are opposite, entire, and with lateral veins looping and joining near the lamina margin - with their rather large and distinctive monosymmetric or asymmetric flowers with a spurred calyx. The spur may be curved, and there is a single stamen with a long anther and short filament. Some taxa have large, flat, stipular glands.

The present distribution of Vochysiaceae on either side of the Atlantic is likely to be the result of dispersal, not vicariance (Sytsma et al. 2004).

Erismieae, containing the tropical American Erisma and the West African Erismadelphus and Korupodendron, are monophyletic. They have cortical/subepidermal phellogen, apparently a single carpel with an inferior ovary (perhaps plesiomorphic) and with 1-2 lateral to apical ovules/carpel. The fruit is samaroid, the wing being formed from the persistent enlarged calyx. Vochysieae are probably not monophyletic (Litt 1999).

At least in Vochysia guatemalensis there are conspicuous, symmetrically-arranged mucilage canals in the pith. The exact position of the single stamen varies (Kawasaki 1998; also Litt & Stevenson 2003b). The flowers are basically inferior, the ovary being initiated in this position; the superior ovary in the mature flower is secondarily superior (Litt 1999; Litt & Stevenson 2003a).

Although anatomy and molecular data strongly suggest that Vochysiaceae are to be included in Myrtales, at first sight the flowers appear unlike those of the other members of the order. Hardly surprisingly, Vochysiaceae have often been associated with other families. Thus Vochysiaceae, often including Euphronia (see Euphroniaceae in Malpighiales here), were placed in Celastrales by Cronquist (1981), and in Vochysiales, along with families like Malpighiaceae (here in Malpighiales) and Krameriaceae (in Zygophyllales), by Takhtajan (1997).

See also Sajo and Rudall (2002: anatomy) and Kawasaki (2006: general).

MYRTACEAE Jussieu - Aromatic compounds [terpenes, etc.] +; leaves with gland dots. - 131/4625. Tropical and (warm) temperate.

1. Psiloxyloideae R. Schmid - Leaves spiral; plant dioecious; petals clawed, staminate flowers: stamens erect in bud, anthers usu. with 4 locules opening separately; carpelate flowers: G [3], embryo sac bisporic, 8-nucleate; endotesta cells periclinally elongated. - 2/4. Southern Africa, Mascarenes.

1A. Heteropyxideae Harvey - stamens = and opposite sepals (+ 2 or 3 opposite petals); G fully superior [base narrow]. - 1/3. S.E. Africa.

1B. Psiloxyleae A. J. Scott - Style 0; fruit a berry. - 1/1: Psiloxylum mauritianum. Mascarenes.

2. Myrtoideae Sweet - A many, anthers with terpenoid-producing glands. - 129/4620: Eugenia (1115, to include Hexaclamys), Syzygium (1045), Eucalyptus (680 - Corymbia if separate 115), Myrcia (400: limits unclear), Melaleuca (220), Verticordia (100), Calyptranthes (100: limits unclear), Psidium (100), Campomanesia (80), Leptospermum (80), Calytrix (75), Myrcianthes (50), Metrosideros (50: limits unclear), Darwinia (45), Xanthostemon (45), Tristania (40). Tropical, also temperate, esp in Australia.

• Myrtaceae are recognisable by their flaky bark, their entire, punctate leaf blades, and their imbricate calyx, often rather fugaceous corolla, and androecium with numerous stamens each often with an apical gland. In fact, in leaf, flower and fruit there is considerable variation. Many taxa have opposite leaves with a well-developed intramarginal vein; there is a group of genera with spiral leaves that dry a distinctive, dark, almost purplish color. Many genera have inferior ovaries and baccate fruits, while another large group of genera have more or less superior ovaries and capsules.

Psiloxylon, known only from the Mascarenes which are only some 8 Ma old, may have been hopping about on islands in the Indian Ocean for almost 40 Ma (Sytsma et al. 2004). This might seem a little improbable (but c.f. Hillebrandia [Begoniaceae] on Hawai'i,also other references in Sytsma et al. 2004 and Smedmark & Anderberg 2007), although there are a number of other submerged volcanoes produced by the mantle plume that currently lies under Réunion.

Tortricine moths (Epitymbiini) larvae feed on Myrtaceae leaf litter in Australia, and some other moth groups are also foliovores on this family there (Powell et al. 1999). About half the galls on Australian plants have been recorded from Myrtaceae (Mani 1964); thus galls are commonly formed from a possibly mutualistic association of a nematode Fergusobia and the dipteran Fergusonina, which the nematode parasitises for part of its life cycle (Ye et al. 2007). In many Myrtaceae the numerous stamens with their brightly coloured filaments are the part of the flower than visually attracts pollinators. In a number of taxa such as Callistemon, the bottle-brush, such flowers are aggregated into inflorescences all the flowers of which open simultaneously; it is then the whole inflorescence that attracts pollinators. The anther glands produce oils, sometimes perhaps to attract pollinators, but also to help in the attachment of pollen to stylar hairs that are involved in the secondary pollen presentation that occurs in Verticordia, for example (Ladd et al. 1999).

Dry-fruited Myrtoideae are ectomycorrhizal (e.g. Smith & Read 1997). A number of Myrtoideae produce gums.

Myrtaceae quite commonly have a more or less elaborated basal portion of the hypocotyl bearing root hairs, however, in Angophora there is a broad, disc-like structure apparently devoid of such hairs (Baranov 1957). Calothamnus has monosymmetric flowers, the upper lip being formed by a group of connate stamens (Westerkamp & Claßen-Bockhoff 2007). The two whorls of the perianth of some relatives of Eucalyptus are undifferentiated (Bohte & Drinnan 2005), and circumscissile abscission of the hypanthium may occur, but in various ways. Androecial variation is extreme, even in quite closely-related taxa; even when the stamens are apparently opposite the petals, this situation is developmentally derived from an androecium that is fundamentally oppositisepalous (see e.g. Carrucan & Drinnan 2000; Drinnan & Carrucan 2005). The low stamen number of Heteropyxis may be an apomorphy for Heteropyxideae.

Myrtaceae s. str. were traditionally divided into Leptospermoideae - leaves spiral to opposite; fruit dry, dehiscent - and Myrtoideae - leaves opposite; terpenoid-containing glands in the apex of the connective, stigma dry; fruit fleshy, indehiscent. Phylogenetic work suggests that the latter are largely derived and monophyletic, although the large and mostly Indo-Malesian genus Syzygium (in the past sometimes even synonymised under Eugenia), along with Acmena and other segregate genera, represent an independant acquisition of fleshy fruit from that in Eugenia itself and the bulk of Myrteae. Myrteae are a largely New World group (Myrtoideae in old classifications, see also Johnson & Briggs 1984), and generic limits there are problematic (Lucas et al. 2005, 2007). Biffin et al. (2006; see also Biffin et al. 2007) suggest that Syzygium should be delimited broadly (see Parnell 2007 for general information on Syzygium s.l.), which seems reasonable. The capsular Leptospermoideae are paraphyletic (Sytsma et al. 1998; Wilson et al. 2001, 2005 [classification]; Salywon et al. 2002). However, if Leptospermoideae are paraphyletic, then stamens with with terpenoid-producing glands on the anthers may not be an apomorphy for Myrtoideae. It is not surprising that testa anatomy correlates with fruit type - capsular fruits have exotestal seeds; baccate fruits have seeds with a generally sclerotic testa.

See also Briggs and Johnson (1979: inflorescence structure), Schmid (1980: general), EUCLID (2006: identification of Eucalyptus), and Basinger et al. (2007: fossils).

Melastomataceae [Crypteroniaceae [Alzataeaceae [Rhynchocalycaceae [Penaeaceae + Oliniaceae]]]]: inflorescence cymose; connective abaxially much expanded [least in Rhynchocalycaceae], nectary disc 0.

Nodes other than simple unilacunar (one trace from one gap) are quite widespread in this part of Myrtales, however, a survey of nodal anatomy, particularly that of Melastomataceae, is much needed. For relationships in this area, see Conti et al. (2002).

Melastomataceae + Memecylaceae [= Melastomataceae s.l.]: included phloem +; secondary veins 2 or 4, strong, from (near) the base; C contorted, anthers with branched trace, pollen with pseudocolpi; radicle bent.

Included phloem may be homoplasious in Pternandra and Memecylaceae (Clausing & Renner 2001). For leaf anatomy, see Mentink and Baas (1992), and for vasculature of the stamens, see Wilson (1950 - ignore the explanation!).

?name of clade - Lamina with veins diverging at the base, tertiaries scalariform; stamens held on one side of the flower, connective basally thickened and with a ± prominent appendage. - 182/4570. Largely tropical and subtropical, esp. South America.

1. Kibessioideae - Hairs uniseriate; endothecium localized to part of inner sporangium wall, placentation parietal; capsule fleshy. - 1/15 (Pternandra). Southeast Asia.

2. Melastomatoideae - Included phloem 0; leaf veins lacking fibrous sheath; endothecium ephemeral.

2A. Astronieae - Hairs as peltate scales; placentation basal to basal-axile. - 4/150: Astronidium (70), Astronia (60). Indomalesia and Pacific.

2B. The Rest - 177/4305: Miconia (1000), Medinilla (400), Tibouchina (245), Sonerila (?180), Leandra (175), Clidemia (120), Gravesia (105), Microlicia (100), Tococa (50). Largely tropical and subtropical, esp. South America.

• Melastomataceae are recognisable by their opposite, entire, exstipulate leaves with three or more strong veins ascending from at or near the base; tertiary veins are conspicuous, and run at right angles to the midrib. Hair types are notably diverse. The flowers with their contorted and often somewhat stridently pink petals and porose anthers which invert during development and have an elaborate, expanded connective, are also distinctive. In old fruits, the walls often distintegrate, leaving vertical vascular bundles behind.

Morley and Dick (2003) were inclined to think that the broad outlines of diversification in the family could be linked with major tectonic (drift) events, but Renner (2004 and references) suggests that dispersal is more likely, with separate Miocene dispersal events populating Madagascar, for example.

Epiphytes are quite common in the family, and some species are scramblers, whether with hook-shaped roots, or climbing with roots that attach to the support; some of the latter taxa have pseudo2-ranked leaves, one leaf of the pair being much reduced (Clausing & Renner 2001a). Some species of Tococa live in close association with the ant Myrmelachista which creates mono- or oligospecific "devil's gardens" by injecting formic acid into the leaves of the surrounding vegetation, which is thus suppressed (Morawetz et al. 1992; Frederickson et al. 2005). Petiolar or laminar ant domatia are quite common in Blakeeae and Miconieae. In Maieta guianensis, one of these species, ca 80% of the host plant's nitrogen is derived from the waste deposited by the ant Pheidole minutula that lives in the domatia (Solano & Dejean 2004). The petals are usually more or less widely spreading, and buzz pollination is common, flowers of the family being visited by many species of bees in search of pollen (Renner 1989; Harter et al. 2002). However, some taxa have nectariferous anther connectives, or nectar is produced on the corolla (Medinilla), hypanthium, or even on the top of the ovary (some Miconia), and in these cases the contorted petals form a tube, rather than spreading widely as is common in the family and pollination may be by birds (Vogel 1997; Varassin et al. 2008: for necariferous taxa, see also Renner 1989).

Sonerila may look rather different vegetatively from the rest of the family, the leaf blades sometimes lacking the typical venation and having toothed margins, but it has similar anthers (although sometimes there are only three stamens), fruits, etc. In some species the fruits open by the placenta breaking through the wall - c.f. Cuphea, in Lythraceae (Cellinese 1997).

For the phylogeny and morphology of Melastomataceae, see Clausing and Renner (2001b), and for that of the fleshy-fruited and speciose, but paraphyletic, Miconieae, see Michelangeli et al. (2004) and Martin et al. (2008); members of the one genus are considerably intermingled with others. Generic limits in Merianeae are also difficult (Schulman & Hyvönen 2003). A tribal classification awaits further sampling of New World taxa in particular, but it cannot be based simply on fruit types. Capsular fruits are linked with superior ovaries and fleshy fruits with inferior ovaries; fleshy fruits have arisen more than once (for the fleshy fruit of Miconia, see Cortez and Carmello-Guerreiro 2008), and a few taxa with capsules have inferior ovaries (Michelangeli et al. 2008a). Bécquer-Granados et al. (2008) and Michelangeli et al. (2008b) discuss the complex biogeography of the speciose Antillean melastomes. Stone (2006) provides a phylogeny of Olisbeoideae; the six morphologically-based genera all have molecular support.

Memecyloideae - Sclereids +[?]; anthers with terpenoid secreting gland. - 6/435: Memecylon (300), Mouriri (85). Tropical.

• Memecyloideae have opposite, entire leaves and swollen nodes; the venation is variable, but indumentum is absent or consists of uniseriate hairs. The blade can be minutely punctate when held up to the light, or the surface may be minutely lineate - this is because of the sclereids common in the family (the surface may also dry minutely bumpy). The rather small flowers have inferior ovaries, but the parts are otherwise free, the calyx is valvate and there are twice as many stamens as petals; the anthers dehisce by slits and there are glands on the connectives. Most species are plants of the understory.

Genera of Apidae and Anthophoridae collect material from the terpenoid-producing anther glands, and they also collect pollen by buzz pollination. Flowers with these glands (overwhelmingly predominating in the subfamily) are blue, flowers without them are white (Buchmann 1987).

For information, see Morley (1976: Memecyloideae) and Renner (1993: general).

PENAEACEAE Guillemin - Stipules minute; stamens = and opposite petals, endothecium ephemeral; fruit a capsule; exotestal cells periclinally elongated, endotegmen not fibrous - 12/35. Scattered: South East Asia, Malesia, Sri Lanka, southern and eastern Africa, Costa Rica to Peru.

Crypteronioideae van Beusekom - 3/10. South East Asia, Malesia, Sri Lanka.

Crypteronioideae possibly rafted from Gondwana (Africa) to Asia via India, with an origin any time from the Early to Late Cretaceous, a later date being favoured (Rutschmann et al. 2007 and references).

Penaeaoideae - Style stout. - 9/25. Southern and eastern Africa, Costa Rica to Peru.

Relationships within this clade are only weakly supported (see Schönenberger & Conti 2003 for phylogeny and floral evolution of the group).

Alzatea - C 0, pollen pseudocolpi 0 [faint], placentation parietal and with an incomplete septum, embryo sac bisporic, eight-celled. - 1/?2. Costa Rica to Peru.

Rhynchocalyx [Penaea et al. + Olinia]: leaves with glandular tips; x = 10.

Rhynchocalyx - Anthers with 4 loculi opening separately. - 1/1: Rhynchocalyx lawsonioides. South Africa.

Penaea et al. + Olinia: hypanthium well developed, pollen grains smooth.

Penaea et al. - K petal-like, C 0; embryo sac tetrasporic, 16-celled. - 7/20. South Africa, the Cape.

Olinia - ?Epicalyx +, small, K ± spatulate, C concave, thick, hairy, pollen heteropolar; fruit drupaceous. - 1/5. Africa, St. Helena.

The stems of Penaeaceae are square, sometimes ridged, and the nodes are swollen. Indeed, there is often a general similarity with other Myrtales; the epithet "lawsonioides" of Rhynchocalyx refers to the general similarity of Rhynchocalyx to Lawsonia in Lythraceae.

Nectar is secreted from the base of the hypanthium of Penaea and relatives.

Van Beusekom-Osinga and van Beusekom (1975) included Alzataeaceae and Rhynchocalycaceae in their expanded Crypteroniaceae, indeed, combination of all these small families may well be in order. Penaeaceae is the oldest name.

Crushed or broken plant parts of Olinia smell of almonds because of the presence of the cyanogenic glucoside prunasin. The interpretation of the perianth of this genus is a matter of some dispute; here I follow the interpretation offered by Schönenberger and Conti (2003), although the exact nature of the outer whorl of very small appendages, described as a possible epicalyx above, is still unclear; they might even be colleters.

For pollen of much of this group, see Muller (1975), for anatomy - very variable - see van Vliet and Baas (1975: detailed comparison within Myrtales), for gross morphology, see van Beusekom-Osinga and van Beusekom (1975), and for perianth morphology, see Schönenberger and von Balthazar (2006). Some information on Crypteronia and relatives is taken from Tobe and Raven (1983b, 1987b) and Renner (2006b: general). See Graham (1985, 2006) for general information on Alzatea; Schönenberger (2006) for general information about Rhynchocalyx; for the embryo suspensor of Penaea and relatives, see Ross and Sumner (2005), and for general information, see Schönenberger et al. (2006); and for general information about Olinia, see von Balthazar and Schönenberger (2006), and for flowers, see Schönenberger and Conti (2003).

[CROSSOSOMATALES [PICRAMNIALES [SAPINDALES [HUERTEALES [MALVALES + BRASSICALES]]]]]: ?

CROSSOSOMATALES Reveal - [Back to Index]

?Nodes; lamina margins toothed; hypanthium +, with nectary [disc]; G stipitate, ovules with both integuments 4 or more cells across, styluli free, postgenitally connate at their apices; fruit?; seed testal, walls of outer layers much thickened. - 6 families, 12 genera, 66 species.

There is some support for placing Crossosomatales as sister to the malvid group (Huerteales, Sapindales, etc.: see Zhu et al. 2007). Crossosomatales are a surprising group, but they are likely to be monophyletic, even if they currently lack strong molecular support. For details of some of the relationships in the tree like [[[Crossosomataceae + Stachyuraceae] Staphyleaceae] Geissolomataceae], see the rbcL tree in Nandi et al. (1998); see also Savolainen et al. (2000a), Sosa and Chase (2003) and Oh and Potter (2006). Although some relationships within the group remain unclear, and Crossosomatales as a whole are not very strongly supported (but 1.0 p.p. in Soltis et al. 2007a), the similar and distinctive seed coat anatomy of all the included members is a morphological feature uniting it. Furthermore, Matthews and Endress (2005a, also 2006b for a summary) emphasise the number of distinctive morphological features of Crossosomatales, even if exactly where some of these are to be placed on the tree is unclear. They also note that there is much more in common between members of Crossosomatales (as delimited below) than in the three families then included in narrow circumscription of the order, Crossosomataceae, Stachyuraceae, Staphyleaceae (Matthews & Endress 2005a, 2006b). Recently the distinctive Guametala has been added to Crossosomatales (Oh & Potter 2006: three genes); it has strong support as sister to [Stachyuraceae + Crossosomataceae]. All in all, however, this group is remarkably heterogeneous containing as it does twelve genera that are placed in seven families, as Oh and Potter (2006) mentioned as they rather reluctantly decide to place Guamatela in a monotypic family. Ultimately, however, some merging of families will be in order.

For a survey of wood anatomy, with a listing of the features - possibly plesiomorphic - common to the whole order, and also possible apomorphies of several families, see Carlquist (2007c).

Includes Aphloiaceae, Crossosomataceae, Geissolomataceae, Guamatelaceae, Strasburgeriaceae , Stachyuraceae, Staphyleaceae, Strasburgeriaceae.

Staphyleaceae [Guamatelaceae [Stachyuraceae + Crossosomataceae]]: ?

STAPHYLEACEAE Martynov - Leaves opposite, odd-pinnate, with small "glands" at the articulations; stamens = and opposite sepals. - 2/45. North temperate, tropics in America (to Bolivia) and Malesia.

• Staphyleaceae are shrubs to trees that may be recognised by their opposite, odd-pinnately compound leaves with serrate leaflets and little "glands" at the articulations (the latter often collapse when dry), the stipules are cauline and are sometimes interpetiolar. The flowers have rather petal-like sepals and usually two to three carpels.

The limits of the two genera currently accepted in the family are newly circumscribed (Simmons 2006).

Staphyleaceae were placed in Sapindales by Cronquist (1981) and Takhtajan (1997). Tapiscia and Huertea have long been recognised as being rather different from other members of the family, i.a. having spirally-arranged leaves (see Dickison 1987 for a table of differences), and molecular data suggest that they are not at all closely related and here they are placed in Tapisciaceae, in Huerteales.

See Carlquist and Hoekman (1985: wood anatomy), Dickison (1986: floral morphology, 1987a: pollen, 1987b: vegetative anatomy), Ramp (1987: gynoecium), Danilova (1996: seed), and Simmons (2006: general).

Guamatelaceae [Stachyuraceae + Crossosomataceae]: funicular aril +.

GUAMATELACEAE S. Oh & D. Potter - Leaves opposite, venation palmate; pollen membranes protruding through apertures; endosperm scanty. - 1/1 (Guamatela tuerckheimii). Mexico to Honduras.

• Guamatelaceae are sprawling shrubs with opposite, stipulate, serrate leaves that have palmate venation and a densely whitish tomentose lower surface. The flowers have sepals and petals almost equal in size, ten stamens, and three carpels connate only by the tips of their styles. The fruit are follicles and are almost enclosed by the persistent sepals.

The anatomy, embryology, and chemistry of this genus are largely unknown.

Guamatela used to be included in Rosaceae, and it does indeed look faintly Rubus-like.

Stachyuraceae + Crossosomataceae: anthers X-shaped; whole testa sclerotic.

STACHYURACEAE J. Agardh - Pith well developed; stomata paracytic; inflorescence long-racemose; flowers 4-merous, hypanthium 0, nectary at base of ovary; G [4], placentation axile, style +; fruit a berry. - 1/5-10. South East Asia.

• Stachyuraceae are small trees or shrubs with closely serrate, stipulate leaves and rather long, pendulous, racemose inflorescences. The sepals and petals of the four-merous flowers are similar, and the fruit is a berry.

Stachyuraceae were included in Theales by Takhtajan (1997).

For embryology, see Mathew and Chaphekar (1977) and Kimoto and Tokuoka (1999); see Schneider (2006) for general information and Zhu et al. (2006) for a phylogeny.

CROSSOSOMATACEAE Engler - Vessel elements with simple perforation plates. - 4/12. W. North America.

• Crossosomataceae are rather small-leaved, much-branched deciduous shrubs that are otherwise vegetatively nondescript. The flowers have separate sepals, petals and carpels and a more or less well developed hypanthium. The fruits are follicles and the seeds have a fimbriate aril.

Crossosomataceae were included in Rosales by Cronquist (1981).

For information, see Kapil (1970: embryology, etc.), DeBuhr (1978: wood anatomy), Tatsuno and Scogin (1978: chemistry), Thorne and Scogin (1978: inclusion of Forsellesia), Rzedowski and Rzedowski (1997: new genus), Sosa and Chase (2003: phylogeny) and Sosa (2006: general).

Aphloiaceae [Geissolomataceae + Strasburgeriaceae]]: conspicuous protrusions from pollen apertures [?position on tree].

APHLOIACEAE Takhtajan - Cork pericyclic; tracheids dimorphic; rays of two different widths; stomata anisocytic; inflorescence fasciculate; P ca 7, ?spiral; A many, pollen surface striate; G 1, stigma sessile, annular-peltate; fruit a berry. - 1/1 or more. E. Africa, Madagascar, the Mascarenes and Seychelles.

• Aphloiaceae are undistinguished evergreen shrubs or small trees with 2-ranked, serrate leaves and often persistent stipules. The axillary flowers have a hypanthium, an undifferentiated perianth, many stamens, and a single carpel. The fruit is a berry. Herbarium specimens have a yellowish, crusty material associated with them, rather like some Thymelaeaceae.

The hypanthial region is broad and spreading and its surface may be nectariferous.

Aphloia used to be included in Flacourtiaceae; it was included in Violales by Takhtajan (1997).

For information, see Takhtajan (1992: seed coat, in part) and Kubitzki (2006b: general).

Geissolomataceae + Strasburgeriaceae: hairs single-celled and T-shaped; pollen ± smooth; G not stipitate, with abaxial ["dorsal"] ribs, two ovules side by side/carpel; fruit a capsule, terminated by persistent style; hilum on seed large.

GEISSOLOMATACEAE Endlicher - Nodes 1:1; flowers axillary, 4-merous, with 3-4 pairs of basal bracts; P 4, petal-like, basally connate, stamens adnate to base of perianth. - 1/1: Geissoloma marginatum. Cape Province, South Africa.

Geissolomataceae were included in Celastrales by Cronquist (1981), in Ericanae by Takhtajan (1997).

For information, see Forest (2006: general) and Danilova (1996: seed coat).

STRASBURGERIACEAE Solereder - Acicular crystals +; leaves margins gland-toothed; flowers large; calyx spiral, petals clawed, filaments flattened, anthers ³3mm long, ovules sessile, style long, persistent in fruit; x = 25. - 2/2. New Zealand, New Caledonia.

Ixerba - Stipules 0; stamens = and opposite sepals, pollen 4-5-colporate; carpels also splitting adaxially, valves woody, reflexed. - 1/1: Ixerba brexioides. New Zealand.

Strasburgeria - Stipules intrapetiolar, connate basally; flowers axillary; hypanthium 0, nectary on base of ovary, one ovule/carpel; fruit rather dry, indehiscent; hilum the length of the seed. - 1/1: Strasburgeria calliantha. New Caledonia.

• Ixerba is an evergreen tree with pseudoverticillate and strongly toothed but estipulate leaves. The inflorescences are terminal and the flowers are rather large and have five spreading sepals and petals, ten stamens with thick connectives, and a conspicuous ovary gradually tapering into the style and with a spreading, lobed nectary at its base. The pointed capsular fruit has a few shiny black seeds with conspicuous red arils. Strasburgeria is also an evergreen tree that may be recognised by its rather large, spiral and toothed leaves that have intrapetiolar stipules. The flowers are large and axillary and have spiral sepals, five petals and ten stamens with long anthers attached their lengths to the thick connective. The stoutly-pointed fruit is indehiscent.

Caenozoic pollen like that of Strasburgeria has been found from Australia, Tasmania, and New Zealand. Strasburgeria calliantha is a decaploid, with n = 250 (Oginuma et al. 2006).

Glands like those of lamina margin of Ixerba are in a cauline stipular position in seedlings.

Ixerba was placed in Grossulariaceae by Cronquist (1981), in Brexiaceae (now = Celastraceae) by Airy Shaw (1966: note the specific epithet), and in Rosidae-Brexiales by Takhtajan (1997). As Amaral (1991) suggested, Strasburgeria is to be excluded from Ochnaceae, where it was often included.

For additional information, see Cameron (2003), for anatomy of Ixerba, see Gregory (1998: in Escalloniaceae) and Hils (1985) and for general information, see Schneider (2006); for Strasburgeria, see also Dickison (1981a: morphology and anatomy, 2006: general) for information.

[PICRAMNIALES [SAPINDALES [HUERTEALES [MALVALES + BRASSICALES]]]]: ovules 2/carpel, apical.

The position of the character of ovule number on the tree is unclear. However, taxa with one or two ovules/carpel are common in the rosid II clade, and the ovules are usually apical.

PICRAMIACEAE Fernando & Quinn - C₁₈ acetylenic acids [tariric, petroselenic] +; leaves odd-pinnate, stipules 0; stamens = and opposite petals; 2 ovules/carpel. - 2/46. Neotropical.

1. Picramnioideae Engler - Ovules apical; fruit a berry. - 1/41. S.E. USA, Central and South America, Caribbean.

2. Alvaradooideae Liebm. - Staminodes opposite sepals, only 1 carpel fertile, ovules basal; fruit a winged capsule. - 1/5. Florida, Central America, Bahama, esp. the Greater Antilles, Bolivia to Argentina.

• Picramniaceae are often rather small trees with bitter-tasting bark and spirally-arranged, compound leaves; the leaflets alternate along the rachis. The small flowers are born in a raceme, and the stamens are equal in number to and borne opposite the petals.

The family is very poorly known.

In the past Picramniaceae have often been included in Simaroubaceae (e.g. Cronquist 1981; Takhtajan 1997). Fernando et al. (1995) suggested that they were to be excluded, although exactly where they are to be placed remains unclear. Jacobs (2003) emphasized the distinctive nature of their secondary metabolites, some being quite unknown from other flowering plants.

For general information, see Fernando and Quinn (1995) and Kubitzki (2006b).

Huerteales are strongly supported as being monophyletic (Peng et al 2003: three gene analysis, sampling sketchy). However, relationships within the rosid II group have been somewhat uncertain. Relationships [Brassicales [Malvales + Sapindales]] are possible (Soltis et al. 2000; Peng et al 2003), but the support is weak, Endress and Matthews (2006) note some features perhaps more common in these last two families than elsewhere, and there is some support for this clade in Soltis et al. (2007a). However, Bausher et al. (2006) in an analysis of whole chloroplast genomes found strong support for the relationships [Sapindales [Brassicales + Malvales]], but only one species from the three orders, and no Huerteales, were included. Alford (2006) suggested that [Huerteales + Brassicales + Malvales] might be sister to Sapindales; Perrottetia was not included in this study. In line with these latter studies, Worberg et al. (2007b) recovered the relationships [Sapindales [Huerteales [Brassicales + Malvales]]], with strong support, and also for the monophyly of each of the four orders.

SAPINDALES [HUERTEALES [BRASSICALES + MALVALES]]: ?

SAPINDALES [HUERTEALES [BRASSICALES + MALVALES]]: ?

SAPINDALES Dumortier - [Back to Index]

Interesting secondary compounds +; silicified wood or wood with SiO₂ grains; leaves spiral, odd-pinnately compound; intrastaminal nectariferous disc well developed; seed coat?. - 9 families, 460 genera, 5670 species.

Sapindales contain ca 3% eudicot diversity. Associated with the frequent accumulation of noxious secondary metabolites in this group, specialised herbivores are found on many of them. Thus the hemipteran Calophya eats largely Anacardiaceae, Burseraceae, Simaroubaceae and Rutaceae (Burckhardt & Basset 2000). Galls are quite common, perhaps especially on Sapindaceae and Anacardiaceae (Mani 1964).

Gums and resins occur in both the Rutaceae-Meliaceae-Simaroubaceae and Burseraceae-Anacardiaceae groups (Nair 1995). Stratified phloem may be quite widespread (in some Meliaceae, Burseraceae and Simaroubaceae, at least: M. Ogburn, pers. comm., also Sapindaceae). Stipules, when present, are sometimes clearly modified leaflets. They have been described as pseudostipules or metastipules; the latter have been defined as having the morphology of true stipules, yet there was good reason to believe that they had been derived from pseudostipules (Weberling & Leenhouts 1965). The flowers are often imperfect, but since staminate and carpelate flowers often have well-developed pistillodes and staminodes respectively, the flower types can be difficult to distinguish. The style in at least some Rutaceae and Sapindaceae is hollow (Lersten 2004).

Includes Anacardiaceae, Biebersteiniaceae, Burseraceae, Kirkiaceae, Meliaceae, Nitrariaceae, Rutaceae, Sapindaceae, Simaroubaceae.

BIEBERSTEINIACEAE Endlicher - 1 apical pendulous ovule/carpel, embryo sac tetrasporic, 16-nuclear, styles separate, impressed, apically connate, stigma capitate; fruit a schizocarp, columella persisting. - 1/5. Greece to Central Asia.

• Biebersteiniaceae are perennial, glandular-hairy herbs that have odd-pinnate leaves with lobed to compound leaflets and petiolar stipules. Ihe inflorescence is erect and racemose and the flowers are large. The fruit has a single seed per carpel and a persistent columella.

At least some species are foul-smelling.

Although previously associated with Geraniaceae (Geraniales), molecular data place it in Sapindales, albeit with a long branch (Bakker et al. 1998). The herbaceous habit is unusual in the order, but ethereal oils, single ovule/carpel, etc., are all congruent with a position here.

For information, see Greenham et al. (2001: chemistry), Boesewinkel (1997: seed anatomy) and Tzakou et al. (2001: fatty acids).

NITRARIACEAE Berchtold & J. Presl - 3/16. Usu. ± arid regions in N. hemisphere, Australia.

This is a very heterogeneous group. Nitraria often has two or three leaves at a node, and some species have scarious stipules, while others appear to lack them; the leaves are occasionally toothed or lobed. Peganum also may have two much-divided leaves at a node, in this case also with ± foliaceous and divided stipules; it has foliar raphides. Takhtajan (1997) says that that there are no stipules in Tetradiclis, however, they are present, if small. The androecium of Peganum harmala is described as being obdiplostemonous by Eckert (1966); it has fifteen stamens, and these may be in groups of three opposite the sepals, or there may be paired stamens opposite the petals (Ronse Decraene & Smets 1991a, 1992; Ronse Decraene et al., 1996). Only the seeds in the central locelli of Tetradiclis are released when the capsule opens. Basically, just about all features need careful examination throughout the group.

Molecular data suggest the relationships [[[Peganum + Malacocarpus] Tetradiclis] Nitraria] (Sheahan & Chase 1996), and assign the group to Sapindales.

Phenetically, and also in features like wood anatomy, and perhaps also chemistry (Nag et al. 1995), there are similarities to Zygophyllaceae, in which Nitrariaceae used to be included. Both families grow in dry and warm habitats, which may account for some of these similarities. Takhtajan (1997) included this family as three separate families in his Zygophyllales. However, Zygophyllales here include only Zygophyllaceae s. str. and the monogeneric Krameriaceae and are not remotely close to Nitrariaceae.

See Batygina et al. (1985) for endosperm development, etc., Danilova (1996), for seed anatomy, and Sheahan and Cutler (1993), for anatomy.

Kirkiaceae [Anacardiaceae + Burseraceae]: 2 ovules/carpel; fruits with 1 seed/carpel.

KIRKIACEAE Takhtajan - Inflorescence dichasial, flowers small, 4-merous, staminate flowers: stamens = and opposite sepals, pollen syncolpate; carpelate flowers: G [8 (always?)], 1 ovule/loculus, tip of axis in ovary swollen, glandular, stigma with post-genital fusion, stigmas covered with large uniseriate hairs; fruit a schizocarp, mericarps pendulous from columella on slender strands. - 2/6. Tropical and S. Africa.

• Kirkiaceae are woody plants that have often rather closely-set odd-pinnate leaves with toothed leaflets. Ihe inflorescence is dichasial and the flowers are small and 4-merous. The fruit is a ridged or angled schizocarp with persistent if slender columella strands.

See Bachelier and Endress (2008b) for floral development.

Kirkiaceae were previously placed in Simaroubaceae, but they lack the bitter quassinoids of that family.

Anacardiaceae + Burseraceae: biflavones +; vertical intercellular secretory canals in phloem; (A obdiplostemonous); G [3]; fruit a drupe, endocarp cells lignified, not oriented.

Note that Anacardiaceae like Pachycormus have thin, brown, flaking bark and vegetatively look quite like Burseraceae.

The flowers are obdiplostemous when parts of the calyx, corolla and androecium have the same multiple of parts. The basic endocarp condition for this clade seems to be that of an unoriented mass of sclerified and often crystalliferous cells (Wannan & Quinn 1990). Such an endocarp characterises Anacardiaceae-Spondiadoideae and Burseraceae and it is also found in Buchanania (perhaps sister to all other Anacardioideae), Campnosperma and Pentaspadon (also near-basal Anacardioideae). It has been suggested that an operculum may be derived twice in Anacardiaceae (Pell & Urbatsch 2001), but it is also found in fruits of Burseraceae and perhaps is plesiomorphic within the whole clade.

For floral morphology, see Bachelier and Endress (2007a, 2008b).

ANACARDIACEAE R. Brown - 1 ovule/carpel. - 70/985. Tropical, also temperate.

1. Spondiadoideae Link - G [4-5(-12)] and as many locules, ovule pendulous; exocarp thick. - 19/. Tropical.

2. Anacardioideae Link - Leaves simple (compound); antesepalous carpel alone fertile; exocarp thin, epidermis lignified, endocarp with up to three layers of palisade lignified sclereids, internal to these a crystalliferous layer [= stratified]. - 58/ : Semecarpus (60). Largely tropical, also temperate.

• Anacardiaceae may be fairly readily recognised because of their often black and/or rather resinous-smelling exudate; the leaves often have black discolorations and the blade may dry a distinctive grey colour. In herbarium specimens, the dry, black exudate may entirely cover the cut surface of the twig, or it may form a distinctive black ring around the periphery. The leaves are often odd-pinnate and the leaflets are opposite to alternate. The flowers are small, and the fruits are often flattened, operculate drupes with excentric style or styles.

There is a variety of winged fruit types in Anacardioideae. These include fruits adnate to broad bracts (Dobinea); more or less conventional samaras (Loxopterygium); an inflorescence with a flattened peduncle that forms the wing (Amphipterygium); much enlarged sepals (Parishia); or even much enlarged petals (Swintonia). The evolution of these fruit types seems to be correlated with the adoption of a drier habitat (Pell & Mitchell 2007), as is also found in other groups.

Anacardiaceae - more accurately, some Anacardioideae - are noted for the sometimes extremely violent allergenic reactions their exudates cause; catechols, resorcinols and other types of phenolic compounds, often in a mixture, as in urushiol, are involved. A vivid memory from the Western Highlands of Papua New Guinea is of quite large black caterpillars - geometrids, I think - congregating arond the black exudate oozing from the slash of an anacardiaceous tree. About a quarter of the genera, all Anacardioideae, have such compounds. Aphids (Fordinae) that form distinctive galls are closely associated with species of Pistacia (Inbar 2009), while a gall-forming jumping plant louse, the hemipteran Calophya, is notably common on Schinus (Burckhardt & Basset 2000).

In Pistacia, and perhaps other genera, chalazogamy occurs, the pollen tube moving from the funicle to the chalaza and thence to to the embryo sac (Martínez-Pallé & Herrero 1995). Pistacia and Amphipterygium (the latter has sometimes been placed in its own family) both have unitegmic ovules with a massive funicle, etc. (Bachelier & Endress 2007).

Spondiadoideae - the clade includes Spondiadeae and some Rhoeeae (Pegia and Tapirira, etc.: Aguilar-Ortigosa & Sosa 2004; Pell 2004, the latter with a list of included genera) - are sister to the rest of the family (Anacardioideae: Pell & Urbatsch 2000; see Pell 2004 and Mitchell et al. 2006 for a list of genera); wind-dispersed taxa in this latter subfamily form a clade (Pell & Urbatsch 2001). Buchanania is quite well supported as sister to Anacardioideae in some analyses (Aguilar-Ortigosa & Sosa 2004; Wannan 2006), consistent both with its chemistry, endocarp anatomy (see above), carpel number of 4-6, and different position of the fertile carpel, but in other analyses its position is uncertain (Pell & Mitchell 2007). Note that Campnosperma, not included in any recent study, has an endocarp similar to that of Buchanania and the fruit is sometimes two-locular. The morphology of neither of these genera is accounted for in Anacardioideae as characterised here; if they are included, the subfamily will probably lack much in the way of apomorphies. Furthermore, since Beiselia (Burseraceae) has a 9-12-carpelate gynoecium, an apomorphy for [Anacardiaceae + Burseraceae] could be an increased carpel number with subsequent independent reduction in both families, although this is rather unlikely. For the limits of Rhus, which seem best narrowly drawn, i.e., there are ca 35 species in the genus, excluding Toxicodendron, etc., not some 200 species, see Yi et al. (2006, and references).

In Anacardium the single stamen is on an oblique plane of symmetry; more generally, the position of the fertile carpel, when single, suggests that the flower is obliquely symmetric.

For further information, see Ding Hou (1978: general), Wannan and Quinn (1991: floral morphology), Young (1976: chemistry), and von Teichman (1991, 1994, and references: seed).

BURSERACEAE Kunth - C valvate. - 18/550. Tropical, but esp. America and N.E. Africa.

1. Beiselieae Thulin, Beier & Razafimandimbison - Leaf ?bases persistent, with a spine; G [9-12], ovules superposed; fruit septifragal, pericarp splitting septifragally separately, columella with deep flanges, mericarps apically winged. - 1/1: Beiselia mexicana. Mexico.

2. Bursereae DC, Garugeae Marchand, Protieae Marchand - Stigmatic head formed by postgenital fusion; cotyledons folded, palmately lobed. - 17/550. Tropical, but esp. America and N.E. Africa.

• Burseraceae are recognisable by their often scented resinous exudate and often smooth, flaking bark; compound, odd-pinnate leaves with opposite, often long-petiolulate and more or less pulvinate leaflets that have prominent fine venation especially when dry; and petioles, petiolules and rachis that tend to be brownish and more or less scurfy. The swollen leaf base is often concave adaxially. The calyx is often connate and 3- or 4-lobed, and the angled drupes have a wrinkled surface when dry. The resinous exudate, which gives myrrh and frankincense, often smells of almonds, and in some genera the leaflets are strikingly symmetric.

Weeks and Simpson (2006) suggest that divergence of Commiphora from the clade now represented by the East Asian Bursera tonkinensis occurred some 53-41.6Ma before present (Eocene). However, Commiphora itself did not diversify until 32.3-23.2Ma before present, Neogene aridification of Africa occurring more or less at that time (see Weeks et al. 2005 and Weeks & Simpson 2006 for further details on the complex biogeography of the family). For possible coadaptive relationships between Burseraceae, especially Bursera itself, and herbivorous chrysomelid beetles (Blepharida), and how the latter deal with the toxic terpene-containing resins the plants contain, see Becerra (2003 and references) and Becerra et al. (2001: particularly interesting). Depending on the terpene composition, the resin is more or less fluid, sometimes being ejected in a toxic spray when the vein is cut; in such cirumstances, the beetle may cut the veins so draining that part of the blade of resin and making it palatable.

The relatively recently described Beiselia (B. mexicana is the only species) has unlobed cotyledons. Molecular studies suggest that it is sister to the rest of the family (e.g. Clarkson 2002); this has considerable implications for character evolution, including cotyledon morphology and carpel number. For additional information on relationships within the famnily and a tribal classification, see Thulin et al. (2008).

There can be foliaceous stipules; these are usually interpreted as being reduced leaflets of the compound leaf. A few genera (e.g. Garuga) have a well-developed hypanthium

For general information, see Leenhouts (1956) and Forman et al. (1989), and for pollen morphology, see Harley et al. (2005: considerable variation).

SAPINDACEAE Jussieu - Cyclopropane amino acids +; C with various adaxial appendages/folds, disc outside A; A 8³; G [3] [?level], ovules sessile; testa vascularized; embryo curved, the radicle in a pocket of the testa. - 135/1580. ± World-wide.

1. Xanthoceroideae Thorne & Reveal - Leaves deciduous; disc with golden, horn-like glands, 7-8 ovules/carpel. - 1/1: Xanthoceras sorbifolia. N. China.

Hippocastanoideae + Dodonaeoideae + Sapindoideae: flowers often strongly obliquely or vertically [Aesculus] monosymmetric, 2 ovules/carpel.

2. Hippocastanoideae Burnett - Leaves opposite, palmate. - 5/130: Acer (110: for a phylogeny, etc., see Li et al. 2006). N. temperate, some tropical, usu. mountains.

3. Dodonaeoideae Burnett - 16/135: Dodonaea (70). Pantropical-warm temperate, esp. Australia/Southeast Asia.

4. Sapindoideae Burnett - 1 (2) basal ovules/carpel. - 111/1185: Serjania (215), Paullinia (195), Guioa (65), Cupaniopsis (60). Pantropical.

• Sapindaceae are recognisable by their often spiral, pinnately-compound leaves that often have subopposite leaflets and a terminal rhachis tip, although many taxa have ternately/palmately-compound leaves. The base of the petiole is quite strongly swollen and the stem is ridged. The leaflets may be coarsely serrate and the fine venation is often prominent when dry. The inflorescences are paniculate, and the flowers are often borne in congested groups along the axis. The flowers are often rather small and are conspicuously hairy inside; there are usually eight stamens and complex folds or scales on the petals. The fruits have only one or two seeds per carpel and are often deeply lobed, or a single carpel only may develop, the other one or two carpels persisting at the base; samaras of various types are common. The radicular pocket in the testa is distinctive.

Finely striate Cupaniopsis-type pollen (Sapindoideae) is distinctive and is widespread in the fossil record, including several sites in Africa, although Sapindaceae with such pollen are no longer found there (Coetzee & Muller 1984). The largely Neotropical Paullineae (Sapindoideae), with 8 genera including Serjania and Paullinia, contain one third of the species in the family. Many are vines, and Sapindaceae, along with Bignoniaceae, are major components of the viny vegetation of the Neotropics.

The evolution of floral symmetry (vertical versus oblique monosymmetry) needs to be fixed more accurately on the tree. The petals of Sapindaceae are often rather complex - they are clawed and often hairy, with ridges and folds - and they have a similarly complex set of terms used to describe them. However, in genera like Acer the petals are secondarily more or less flat and simple. The fruit looks like a follicle when only one carpel develops; dehiscence is, however, down the abaxial side and is loculicidal. In many Sapindaceae (and some Anacardiaceae!) the pericarp grows much faster than the seed, so what seem to be almost mature fruits can contain seeds that are still very small.

Phylogenetic studies suggest that Xanthoceras, with simply 5-merous, polysymmetric flowers, several ovules arrranged in parallel (see also Magonia), and complex, golden nectaries borne outside the eight stamens, may be sister to all other Sapindaceae. Genera included in the erstwhile Aceraceae and Hippocastanaceae are monophyletic sister taxa, the combined clade being sister to the rest of the family (Harrington et al. 2005 and references).

For wood anatomy, see Klaassen (1999) and Agarwal et al. (2005), for embryology, Tobe and Peng (1990), for chromosome numbers, Lombello and Forni-Martens (1998), for floral development of Koelreuteria, Ronse Decraene et al. (2000b) and of Handeliodendron, see Cao et al. (2008), and for fruits of Paullineae, see Weckerle and Rutishauser (2005).

Simaroubaceae [Rutaceae + Meliaceae]: alkaloids, pentanortriterpenes +; inflorescence branches cymose.

The triterpenoid limonoids, meliacins, cneorids and quassinoids found in this clade are all biosynthetically related (e.g. Connolly et al. 1970; Evans & Taylor 1983) and often have a bitter taste.

SIMAROUBACEAE Candolle - Quassinoids [triperpenoids] +; carpels united by styles alone, 1 ovule/carpel; fruitlets separate, drupes. - 19/95. Largely tropical; a few (e.g. Ailanthus) temperate.

• Simaroubaceae can be recognised because the twigs have conspicuous pith - herbarium specimens are often notably light - and the bark is bitter, even when dry; the leaflets of the odd-compound leaves often have rather coarse, blunt, teeth and flat glands on the lower surface, sometimes close to the margin, and the rhachis collapses at the nodes. The flowers are rather small, the petals usually barely exceeding the sepals. The fruits are often separate drupelets.

Picrasma, Holacantha and Castela form a clade with strong support that is sister to the rest of the family, and Ailanthus is sister to the remainder; much of the phylogenetic structure along the backbone of the family is also well supported (Clayton et al. 2007b). Leitneria is well embedded in the family.

Although the carpels may appear to be more or less free, there is often only a single style. There are reports of other than porogamous fertilization in the family (see also Anacardioideae: Rao 1970). The fruits may also be samaras (e.g. Ailanthus).

Simaroubaceae have been very difficult to delimit, and molecular data (Fernando et al. 1995) suggest the excision of Suriana and its relatives (to Surianaceae - Fabales), Harrisonia (to Rutaceae) Picramnia and Alvaradoa (to Picramniaceae - unplaced rosid: Fernando & Quinn 1995). At the same time Leitneriaceae, previously thought to be the only family endemic to the U.S.A., are clearly apetalous Simaroubaceae (see Abbe 1974 for floral morphology/anatomy).

For information, see Fernando and Quinn (1992: pericarp anatomy), and Webster (1936: wood anatomy).

Rutaceae + Meliaceae: tetranortriterpenes +; stigma capitate.

RUTACEAE Jussieu - Leaves punctate. - 161/1815: Zanthoxylum (250: prickles in the stipular position), Melicope (150). Largely tropical.

• Rutaceae are recognisable because of their often opposite, compound and glandular-punctate leaves; their cymose inflorescences; their flowers, with the stamens often arranged in a ring, thick and sometimes strongly flattened filaments, conspicuous disc at the base of the ovary, and expanded stigma, are quite distinctive. Although the fruits are variable, many have a glandular-punctate pericarp, while others are dry, deeply lobed, and dehisce to reveal the black and shiny seeds.

For dates of diversification within Rutaceae, especially Aurantioideae, see Pfeil and Crisp (2008; c.f. in part Muellner et al. 2007); the family is relatively young, and distributions are unlikely to be much affected by continental drift events.

Rutaceae have very diverse secondary metabolites, some of which (essential oils, coumarins, etc.) are similar to those in Apiaceae (Hegnauer 1971), while their alkaloids are like those found in some magnoliids - and are produced via nine or more different biosynthetic pathways. Caterpillers of Papilionidae-Papilioninae butterflies are notably common on Rutaceae, indeed, ca 1/3 of the records are from this family, and 80% of the ca 550 species of Papilio are found here (Zakharov et al. 2004). Like the magnoliids, on which closely related Papilioninae are also found, plants of Rutaceae have a variety of alkaloids that attract the butterflies. It is possible that Rutaceae were the original food plants for Papilio, since even those caterpillars which now eat Magnoliales will eat Rutaceae if they have to.

Ca 250 species of Diosmeae are restricted to South Africa, largely to the Cape Floristic Region (Trinder-Smith et al. 2007).

Azadirachta indica (Melia azadirachta) is the neem tree; the wood of Swietenia spp. is the prized mahogany.

Rutaceae are also rather variable in flower and fruit, perhaps less so vegetatively since gland dots are ubiquitous except in the Cneorum group. Kallunki (1992) illustrates the flowers of Erythrochiton fallax as having the median sepal adaxial, but the exact orientation of the flowers and how they are held is unclear since the inflorescence can be up to 1.5 m long and pendent. The flowers of Galipeinae, to which Erythrochiton belongs, may have only two stamens plus staminodes, and the flowers form a tube in various ways - the corolla or the filaments may be connate, or the petals are postgenitally adnation to the broad filaments, the tube being formed by both corolla and filaments (Pirani & Menezes 2007). Peltostigma has a floral formula K3 C3 A9 G [?5], and looks almost lauraceous; Pilocarpus has an erect raceme and the calyx is reduced to a rim.

General relationships in a two-gene analysis suggested that there were two main clades - [most of Rutaceae] [[Spathelia + Dictyoloma] [[Cneorum + Ptaeroxylum] Harrisonia]], both genus pairs with good support, but the position of Harrisonia, with sequences from only a single gene, is less clear (Chase et al. 1999). Taxa of the latter clade are chemically sometwhat different from other Rutaceae, often lacking alkaloids, but they have limonoids (Simaroubaceae do not), and all have chromones, unique to them in the context of Sapindales. Jadin (1901) noted that anatomically Harrisonia was rather different from other Simaroubaceae its its heterogenous pith and lack of medullary secretory canals. Although it does not seem to have pellucid foliar gland dots, Fernando and Quinn (1992) found secretory cavities in the fruits and Fernando et al. (1995) suggested that its removal from Simaroubaceae was justified on both molecular and morphological grounds.

Some of the fruit characters used to distinguish the classic subfamilies in Rutaceae are proving unreliable. Within core Rutaceae, [Aurantioideae (fibrous exotegmen) + [Chloroxylon + Ruta]] form a well-supported group. Salvo et al. (2008) found that Dictamnus is widely separate from the other members of Ruteae; most other Rutoideae are also immediately unrelated (Poon et al. 2007). Subfamilies other than Aurantioideae are interspersed in another group, within which relationships are uncertain. Zanthoxylum may be sister to the rest (see also Muellner et al. 2007); it is certainly part of a well-supported clade with distinctive isoquinoline alkaloids that may be sister to Ptelea. Euodia and relatives form a moderately supported clade, but other than that, relationships are unclear (Poon et al. 2007). Although Aurantioideae seem to be monophyletic (e.g. Morton et al. 2003: distinctive fruit with fleshy hairs in the loculi; endosperm 0; x = 9), even there genera need overhauling (see also Scott et al. 2000; Samuel et al. 2001), as in Galipeinae (Kallunki & Groppo 2007) and the largely South African Diosmeae (Trinder-Smith et al. 2007). For relationships within Aurantioideae, see Pfeil and Crisp (2008).

Cronquist (1981) included Cneorum in Sapindaceae; Hegnauer (1990) included Ptaeroxylum in Meliaceae, although he noted it was chemically more similar to Rutaceae, and although Harrisonia has also been included in Rutaceae, as by Thorne (1992: no reasons given) it has often been included in Simaroubaceae. As noted above, all these genera are part of the same smallish clade.

For information, see Gut (1966: gynoecial morphology), van der Ham et al. (1995: general), White and Styles (1966: general), Caris et al. (2006b: floral development of Cneorum), and Mulholland et al. (2000: "Ptaeroxylaceae" in particular), Waterman and Grundon (1983), Da Silva et al. (1988), and Adsersen et al. (2007) - all chemistry.

MELIACEAE Jussieu - Filaments connate. - 52/621. Pantropical, but largely Old World; plants of the lowlands.

1. Melioideae Arnott - Seeds with arillode or sarcotesta. - 37/571. Aglaia (110), Trichilia (85), Dysoxylum (80), Chisocheton (50), Guarea (50). Pantropical, but largely Old World.

2. Swietenioideae Kostel. - Buds perulate; 3-many ovules/carpel; fruit a septifragal capsule, valves falling off, columella persisting. - 15/50: Entandophragma (15). Pantropical, but largely Old World.

• Meliaceae are recognisable by their usually spiral and odd-compound leaves with well-developed leaf buttresses, i.e. the leaf bases are swollen and more or less elongated vertically. The leaflets are often opposite and dry thin and brittle, with little in the way of tertiary venation evident. The bark may smell - sweet, like garlic, etc. - and milky exudates occur sporadically. The flower buds are often longer than broad, with petals much longer than the sepals, and most members have a staminal tube and a large stigma. The fruits and seeds are usually quite large.

Muellner et al. (2006) discussed the biogeography of the family, suggesting that it originated in Africa and subsequently dispersed to achieve its curent range; the family is ca 85-76 Ma old, diversification beginning ca 75-67.5Ma before present.

Most Meliaceae have a well-developed floral tube which is formed by the connation of the filaments - a rather uncommon way of forming the tube. The pistillode in staminate flowers is well developed, the result being that staminate and carpelate flowers are very similar functionally, although the former are often somewhat narrower. The whole apex of the style may be massively swollen and involved in secondary pollen presentation, as in Vavaea (Ladd 1994).

Guarea (tropical America) and Chisocheton (Malesia), both Melioideae, have indefinitely growing leaves (see Fukuda et al. 2003 for a phylogeny of the latter). In Guarea the apical part of the leaf is shoot-like in its gene expression (Tsukaya 2005), while in species of Chisocheton such as C. pohlianus the inflorescence may be epiphyllous (Fisher & Rutishauser 1990), flowers appearing between the leaflets (specimens have been misidentified as Rubiaceae!). In the latter genus the leaves can be rooted, and then they continue to grow for a long time, although I do not know of any case where a fertile tree has been produced from a leaf. Capuronianthus (Swietenioideae) has opposite, compound leaves, while the simple-leaved Vavaea and Turraea (both Melioideae) look rather unlike other Meliaceae, except when in flower; the leaves of the latter genus may even be two-ranked and lack articulations.

The filaments of Vavaea are largely free, as are those of Cedrela and Toona (Swietenioideae). Indeed, Cedreleae are rather different florally from other Meliaceae, but features found there such as more or less free stamens may be derived, not plesiomorphous (c.f. Gouvêa et al. 2008).

For a three-gene molecular phylogeny of the family, see Muellner et al. (2003), for a more detailed rbcL phylogeny, see Muellner et al. (2006). Although Swietenioideae and Melioideae can be separated chemically based on limonoid types, recent work on Quivisianthe suggests that the distinction may not be that simple (Mulholland et al. 2000). Two distinctive Malagasy genera, Quivisianthe and Capuronianthus, previously segregated as separate monogeneric subfamilies are embedded in Melioideae and Swietenioideae respectively (e.g. Muellner et al. 2003, 2006). For relationships in Aglaia, see Muellner et al. (2005) and Muellner et al. (2008b) for its biogeography; for Neotropical Cedreleae, see Muellner et al. (2009).

For more information, see Mulholland et al. (2000: chemistry), Nair (1970: embryology, etc.), Pennington and Styles (1975: generic monograph), and Pennington (1981: monograph of Neotropical taxa).

HUERTEALES [BRASSICALES + MALVALES]: ?

HUERTEALES Doweld

Flowers small, stamens equal and opposite sepals, ovules 1-2/carpel; embryo at most medium in size. - 3 families, 5 genera, 23 species.

None of the four genera in these three families is at all well known. Nor is Perrottetia, previously in Celastraceae, if only rather uneasily so; molecular data suggest that it is to be placed near Tapiscia (M. Simmons, in Matthews & Endress 2005b), a placement with which its morphology is in general agreement. Given the possibly superficial but still striking similarity of Gerrardina eylesiana (Gerrardinaceae) to Perrottetia in particular, and perhaps a similarity in seed coat anatomy of the two genera, and the rather similar flowers of Huerteales and Gerrardinaceae (but note that their androecial arrangement differs), the latter are included in Huerteales.

Includes Dipentodontaceae, Gerrardinaceae, Perrottetia, Tapisciaceae.

TAPISCIACEAE Takhtajan - Stomata paracytic; leaves compound, articulated, with small "glands" or stipels at the points of articulation; pollen colpate. - 2/5. China, West Indies and N. South America.

• Tapisciaceae can be recognised by their spiral, odd-pinnately compound or trifoliolate leaves with glands at the point of articulation of the petiolule with the rhachis; the leaflets have toothed margins.

Leaves of Tapiscia can have foliaceous stipels. There is quite a prominent disc in Huertea, but it is not vascularized (Dickison 1986; Danilova 1996). The embryo is up to about half the length of the seed and is embedded in copious endosperm. In Tapiscia sinensis, at least, the fruit takes about eighteen months to develop (Liu et al. 2008).

The two genera that make up the family were long included in Staphyleaceae (see Dickison 1987b for a table of differences separating them from the rest of Staphyleaceae), here in Crossosomatales, and placed in Sapindales by Cronquist (1981), while a segregated Tapisciaceae were still included in Sapindales by Takhtajan (1997).

See also Dickison (1987a: pollen morphology, 1987b: vegetative anatomy), Wei et al. (2002: pollen), and Kubitzki (2002d: general).

GERRARDINACEAE Alford - Hypanthium +, stamens equal and opposite petals, ovary unilocular, placentae apical; embryo minute. - 1/2. Eastern Africa.

See Alford (2006) for what little is known about the family.

Perrottetia - 2 basal erect ovules/carpel; K + C persistent; exotegmen palisade, ridged; embryo medium. - 1/ca 15. China to Malesia and N.E. Australia, Mexico to Peru (map: from Ding Hou 1962, New World rather approximate).

For general information, see Ding Hou (1962), for floral morphology, etc., see Matthews and Endress (2005b).

DIPENTODONTACEAE Merrill - Inflorescence umbellate; K and C similar. - 1/1: Dipentodon sinicus. S. China and adjacent Burma, India and Vietnam.

• Dipentodontaceae can be recognised by their two-ranked, serrate leaves and lobed stipules that are borne on the stem adjacent to the leaves. The simply umbellate inflorescences with small flowers and fruits that have almost indistinguishable and persistent sepals and petals are distinctive.

Dipentodontaceae were included in Santalales by Cronquist (1981) because of apparent similarities in gynoecial structure, but the toothed lamina margin and large, lobed stipules suggested that this was incorrect. Dipentodonaceae were placed by Takhtajan (1997) in his Violales (= Malpighiales), while Wu et al. (2002, 2003) recognised a Dipentodontales C. Y. Wu et al. that they included in Dilleniidae, where it was placed between Passiflorales (which included Caricaceae) and Violales (Cucurbitales were next).

For general information, see Merrill (1941), for pollen, see Lobreau-Callen (1982).

BRASSICALES + MALVALES: ?

BRASSICALES Bromhead - [Back to Index]

Myrosin cells +, glucosinolates produced; endoplasmic reticulum with dilated cisternae; tannins 0; leaves spiral; inflorescence racemose; (petals clawed); G [3]; seed coat? - 17 families, 398 genera, 4450 species.

Brassicales contain ca 2.2% eudicot diversity (Magallòn et al. 1999); the earliest fossil known is from the Turonian, ca 89.5Ma before present; the stem group age is estimated to be 90-85Ma before present, crown group age 71-69Ma before present (Tropaeolum, etc., not included: Wikström et al. 2001).

Caterpillars of Pieridae-Pierinae (ca 360 species recorded, in 33+ genera; about one third of all records, ca 840 species in the clade) are commonly found on Brassicales (Fraenkel 1959; Ehrlich & Raven 1964; Braby & Trueman 2006; Braby et al. 2006), including Bretschneidera; however, they are abundant only in the Capparaceae-Cleomaceae-Brassicaceae clade. Pierinae may have moved to Brassicales from an original host in Fabaceae (Braby & Trueman 2006) some (90-)85(-60) Ma, within ten Ma or so of the origin of Brassicales (Wheat et al. 2007). This is associated with the evolution of a novel glucosinolate detoxifying mechanism in these butterflies, which are more diverse than clades without this mechanism (Wheat et al. 2007); the enzyme concerned produces nitriles rather than toxic isothiocyanates on the hydrolysis of glucosinolates (Fischer et al. 2008). Relationships between herbivores and plants (nearly all information comes from Brassicaceae) are complex, but specialised feeders - and their hymenopteran parasites - may be attracted by isothiocyanates (Hopkins et al. 2009 and references). Some chrysomelid beetles favor Brassicales, for example, Phyllotreta (Alticinae - see Jolivet & Hawkeswood 1995), while the dipteran leaf miner Liriomyza brassicae is found on Resedaceae, Cleomaceae, Tropaeolaceae and Brassicaceae (Spencer 1990).

Nearly all the glucosinolate-producing families of flowering plants are in this clade (c.f. Kjær 1974; Dahlgren 1975). Glucosinolates are mustard oil glycosides, and mustard oils themselves are esters of isothiocyanic acid, which contain a R—N=C=S group, and have a pungent smell and a sharp taste. Specialised, protein-rich myrosin cells contain enzymes like thioglucoside glucohydrolase (a ß-thioglucohydrolase, or myrosinase), that breaks down glucosinolates into glucose and toxic isothiocyanates, thiocyanates (mustard oils), nitriles (not necessarily toxic) and other compounds when plant tissue is damaged by a herbivore. This is the "mustard oil bomb" or "Senfölbombe" (Rask et al. 2000; Wittstock et al. 2003; Grubb & Abel 2006 for details, most of which have been worked out in Brassicaceae; Bones & Rossiter 2006 particularly for glucosinolate degradation). Not all taxa producing mustard oils have myrosin cells, and even in Brassicaceae, which do, such cells may be absent in the young plant (Maile 1980).

Includes Akaniaceae, Bataceae, Brassicaceae, Capparaceae, Caricaceae, Cleomaceae, Emblingiaceae, Gyrostemonaceae, Koeberliniaceae, Limnanthaceae, Moringaceae, Pentadiplandraceae, Resedaceae, Salvadoraceae, Setchellanthaceae, Stixaceae, Tovariaceae, Tropaeolaceae.

Akaniaceae + Tropaeolaceae: bracteoles 0; flowers quite large, obliquely monosymmetric at least in bud, calyx/corolla tube +, C clawed; A 8, ovules apical, style long; testa vascularized.

It is possible that Tropaeolaceae have basically pinnate leaves (Endress 2003c), a matter that may be cleared up by further developmental studies. This may yield another synapomorphy for the clade. Although both families have a nectary, it is extrastaminal in Tropaeolaceae and intrastaminal in Akaniaceae. Furthermore, exactly which stamens are reduced, details of the plane of asymmetry of the flowers, and perhaps also development of a hypanthium differ between Tropaeolum and Bretschneidera (Ronse Decraene & Smets 2001; Ronse Decraene et al. 2002a). Carlquist and Donald (1996) give additional characters of wood anatomy that unite these two families.

AKANIACEAE Stapf - Leaves pinnate; testa vascularized, multiplicative, exotestal cells palisade, thick-walled, mesotesta ± thick, cells thick-walled, endotesta thickened. - 2/2. S.W. China, adjacent Vietnam, Formosa, E. Australia.

A relationship with of Akaniaceae with Sapindales has often been suggested (e.g. Carlquist 1997a), but note that in this case Sabiaceae were included in the latter. An analysis of morphological data also suggested similar relationships (Ronse de Craene & Haston 2006), indeed, Bretschneideraceae do look rather sapindaceous with their obliquely symmetric flowers with eight stamens and fruits rather like those of Aesculus.

For information, see Bayer and Appel (2002).

TROPAEOLACEAE Berchtold & J. Presl - Rather fleshy vines; venation palmate; flowers monosymmetric, with adaxial nectariferous calycine spur, ovules tenuinucellate; fruit a septicidal schizocarp. - 1/95. New World.

• Tropaeolaceae are rather fleshy vines with twining petioles. The broad leaf blades have palmate venation and may be palmately-lobed, palmate or peltate. The flowers are large, strongly monosymmetric, and have an adaxial calycine spur, five clawed petals, and eight stamens. The fruit is usually a three-lobed schizocarp, and each unit contains but a single seed.

For a phylogeny, see (Andersson & Andersson 2000); there is only a single genus in the family.

The nectary ends up in the spur; it is outside the stamens (Ronse Decraene et al. 2002). The median carpel is actually slightly off the median (Eichler 1878; Ronse Decraene & Smets 2001); two antepetalous stamens are suppressed (Ronse Decraene et al. 2002).

For information, see Walker (1947: embryology), Carlquist and Donald (1996: anatomy), and Bayer and Appel (2002: general).

[Moringaceae + Caricaceae] [Setchellanthaceae [Limnanthaceae [[Koeberliniaceae [Bataceae + Salvadoraceae]] [Emblingiaceae [Pentadiplandraceae [Gyrostemonaceae + Resedaceae] Tovariaceae [Cleomaceae [Capparaceae + Brassicaceae]]]]]]]: several [³6] ovules/carpel.

Moringaceae + Caricaceae: woody, stems stout; venation palmate, stipules as glands; flowers whitish, ovary longitudinally ridged, placentation parietal, many ovules/carpel, style hollow; testa multiplicative, mesotesta ± lignified.

Whether or not the thickened mesotesta of the two families is comparable needs to be confirmed, certainly there are substantial anatomical differences in the seed coat. For general information and relationships in this clade, see Olson (2002a).

MORINGACEAE Martynov - Schizogenous gum canals +; leaves odd-pinnate, with glands at the articulations and at the base; stamens equal and opposite petals, declinate, bisporangiate [unithecal], staminodes opposite sepals; fruit a long, slender capsule; seeds 3-angled. - 1/12. India to Africa, Madagascar.

• Moringaceae are woody, often quite stout-stemmed shrubs or trees whose foliage often smells unpleasantly when crushed. They are recognisable by their spiral, deciduous, odd-pinnate, 1- to 3-compound leaves that have conspicuous glands at the articulations and at the base; often clearly obliquely monosymmetric and papilionoid flowers (the median petal is adaxial) with five stamens; and long, slender, tricarpelate, capsular fruits.

All species have rather fleshy roots/rootstock and are usually to be found in more or less arid habitats. Olson and Rosell (2006) suggest that heterochrony is involved in the evolution of the various life forms in the family; the bottle-tree growth form is probably plesiomorphic, the tuberous shrub growth form is probably derived (see also Olson 2006 for wood anatomy). Even although flowers of all species are slightly monosymmetric early in development (Olson 2002b), flowers at anthesis vary from polysymmetric to strongly monosymmetric. When flowers are monosymmetric, they have the median petal in the adaxial position, and when they are polysymmetric the median petal is in the normal abaxial position (Olson 2003).

For further information, see Olson and Carlquist (2001: wood anatomy), Olson (2000b: phylogeny, etc.) and the Moringa website, and Kubitzki (2002d: general).

CARICACEAE Burnett - Anastomosing articulated laticifers +; leaves palmately-veined or strongly palmately-lobed (palmate), glands on adaxial surface at base; plants di(mon)oecious; staminate flowers: C connate, stamens adnate to corolla, nectary on pistillode; carpelate flowers: C often free, nectary 0; G [5], styles ± separate; fruit a berry; sarcotesta +, mucilaginous, mesotesta tanniniferous, with lignified ribs. - 4(-6)/34. Mostly tropical America; Africa (Cylicomorpha only).

• Caricaceae are stout-stemmed often prickly trees, rarely vines, with flowing, latex-like exudate and palmately-compound or -lobed leaves. The inflorescences are axillary, cymose, and the flowers - always imperfect - are moderate in size. The fruit is a rather large berry and contains numerous seeds borne on five parietal placentae and each surrounded by mucilage.

Jacaratia, at least, has carpelate flowers with white, spreading stigmas mimicking staminate flowers (Bawa 1980).

Note that the staminate flowers have a corolla tube, while the carpelate flowers have separate petals.

Kyndt et al. (2005) discuss phylogenetic relationships within Caricaceae: The African Cyclimorpha is sister to the rest of the family.

For information, see Singh (1970: embryology), Carlquist (1998c: wood anatomy), Ronse Decraene and Smets (1998b: floral development), and Kubitzki (2002d: general).

Setchellanthaceae [Limnanthaceae [[Koeberliniaceae [Bataceae + Salvadoraceae]] [Emblingiaceae [Pentadiplandraceae [Gyrostemonaceae + Resedaceae] Tovariaceae [Cleomaceae [Capparaceae + Brassicaceae]]]]]]: nodes 1:1; extended 3' terminus of rbcL gene.

SETCHELLANTHACEAE Iltis - Hairs T-shaped, on multicellular podium; flowers 6-merous, K opening irregularly; capsule septifragal, columella persistent. - 1/1: Setchellanthus caeruleus. Mexico.

• Setchellanthaceae are pungent-smelling shrubs that may be recognised by their large, blue, axillary flowers with parts usually in sixes and an irregularly-splitting calyx. Vegetatively the plant is rather undistinguished, although it has T-shaped hairs and the rather small, entire, exstipulate leaves have secondary veins arising from near the base and lack much in the way of a petiole. The three-carpelate fruit is septifragal and there is a persistent columella.

Setchellanthus used to be included in Capparaceae (Brassicaceae-Capparoideae), but it comes out near Limnanthaceae in molecular phylogenies (Karol et al. 1999, see also Rodman et al. 1997), although just basal to them in Chandler and Bayer (2000).

For information, see Carlquist and Miller (1999: anatomy), Iltis (1999: general), Tobe et al. (1999: floral anatomy), Tomb (1999: pollen), and Kubitzki (2002d: general).

Limnanthaceae [[Koeberliniaceae [Bataceae + Salvadoraceae]] [Emblingiaceae [Pentadiplandraceae [Gyrostemonaceae + Resedaceae] Tovariaceae [Cleomaceae [Capparaceae + Brassicaceae]]]]]: ?

Distinctive vertical files of root hairs are known from Limnanthaceae and some families in the Emblingiaceae-Brassicaceae clade, they have not been observed in Tropaeolaceae, but other taxa have not been studied (Dolan & Costa 2001).

LIMNANTHACEAE R. Brown - 1 basal tenuinucellate unitegmic ovule/carpel, embryo sac tetrasporic, 4- or 6-nucleate, style gynobasic; fruit a schizocarp, mericarps ± spiny; seed coat pachychalazal, thick, with vascular bundles, otherwise undistinguished; cotyledons cordate. - 1(2)/8. Temperate North America.

• Limnanthaceae are rather soft-stemmed herbs with spiral, exstipulate, deeply-lobed or compound leaves and rather open flowers with a valvate calyx, contorted corolla, and gynobasic style or styles. The fruit is a schizocarp separating into shortly spiny, one-seeded portions (mericarps).

For information, see Link (1992a: nectaries), Carlquist and Donald (1996: wood anatomy), and Bayer and Appel (2002: general).

[Koeberliniaceae [Bataceae + Salvadoraceae]] [Emblingiaceae [Pentadiplandraceae [Gyrostemonaceae + Resedaceae] Tovariaceae [Cleomaceae [Capparaceae + Brassicaceae]]]]: ovules campyltropous, style/styles short to absent; fruit indehiscent; seeds exotegmic, exotegmen fibrous [?level]; embryo curved.

There is extensive variation of floral merosity in the Emblingiaceae-Brassicaceae group, and 4-merous flowers could be an apomorphy at this rather deep level - but with plenty of reversals. A fibrous, if unlignified, exotegmen may well be another apomorphy; it has, for example, been reported from Koeberliniaceae, Salvadoraceae, Resedaceae, Cleomaceae, etc. (see also Tobe & Raven 2008).

Koeberliniaceae [Bataceae + Salvadoraceae]: myrosin cells 0; flowers 4-merous; G [2]; exotestal cells well developed.

KOEBERLINIACEAE Engler - Plant thorny, leaves minute; pits vestured; ovule tenuinucellate, style long[!]. - 1/1: Koeberlinia spinosa. C. and S.W. North America, Bolivia.

• Koeberliniaceae are thorny, much-branched shrubs of dry regions; the prophylls are basal on the branches/thorns. The flowers are small and undistinguished, and the fruit is a few(1-2)-seeded berry terminated by the persistent style.

Canotia, sometimes placed here (e.g. Hutchinson 1973), is included in Celastraceae.

For information, see Gibson (1979: anatomy), Mehta and Moseley (1981: floral anatomy), Tobe and Raven (2008: embryology)and Kubitzki (2002d: general).

Bataceae + Salvadoraceae: rays wide, multiseriate; nodes 1:2; stomata paracytic; leaves opposite, with secondary veins ascending from at or near base; A 5 [alternating with K], pollen smooth; G (2), 2 basal non-campylotropous ovules/carpel; tegmen non-fibrous; endosperm 0, embryo ± straight.

The two families are very similar anatomically (Carlquist 2002a). Ronse de Craene and Haston (2006) list a number of other features the two families share including colleter-like stipules, a sepal tube, etc.

BATACEAE Perleb - Pits vestured; leaves fleshy; inflorescence more or less strobilate; flowers imperfect; carpelate flowers: P 0; G 4-locular [the loculi of the two carpels are divided]; seed coat membranous. - 1/2. N. Australia and S. New Guinea, tropical America, and the Galapagos.

• Bataceae are halophytic subshrubs with opposite, small, entire, fleshy leaves that have minute "stipules". The flowers are staminate or carpelate and aggregated into dense, spicate, more or less strobilate, inflorescences. The carpelate flowers lack bracts or perianth and are fused together to form a compound fruit.

The morphological nature of the structure enveloping the staminate flowers is most obvious in Batis argillicola, but there is controversy over its interpretation. In B. maritima, Ronse De Craene (2005) describes it as being derived from four sepals, although he notes that it has only a single vascular trace; it is possible that some of the lobing of the tubular structure might be caused by pressure from other parts of the developing flower. There is also controversy over the nature and position of the stipules; they could be colleters.

For information, see Batygina et al. (1985: ovules), Tobe and Takahashi (1995: pollen), and Bayer and Appel (2002: general).

SALVADORACEAE Lindley - 3/11. Africa (inc. Madagascar) to South East Asia and West Malesia, often in drier regions.

• Salvadoraceae are rather small-leaved shrubs, sometimes scrambling, or small treees with opposite, rather leathery leaves; small, 4-merous flowers with connate sepals and two carpels with a short style; and a fleshy fruit.

For wood anatomy, see Carlquist (2002a), for general information, see Kubitzki (2002d).

Emblingiaceae [Pentadiplandraceae [Gyrostemonaceae + Resedaceae] Tovariaceae [Cleomaceae [Capparaceae + Brassicaceae]]]: cisternae of endoplasmic reticulum dilated and vacuole-like; inflorescence terminal, racemose, bracteoles 0; floral development open, C clawed, disc/nectary outside A; endotesta crystalliferous. - [Back to Index]

For ovule type and the different mechanisms by which the ovule becomes campylotropous within this group, see Boesewinkel (1990) and Bouman and Boesewinkel (1991), for bracteoles, see Ronse Decraene (1992). Ronse de Craene and Haston (2006) suggest more characters that may be synapomorphies for this clade, including floral symmetry and embryo development. Note that nectary morphology and absence/presence/position is very variable in Brassicales outside this core group.

Phylogenetic relationships in this group of Brassicales have been partly resolved in a three-gene study by Hall et al. (2004). However, Ronse de Craene and Haston (2006) found that Emblingiaceae moved to a position sister to [Bataceae + Salvadoraceae] in a combined morphological-molecular study, but there was little support for this position and many data were missing for Emblingia in particular.

EMBLINGIACEAE Airy Shaw - Flowers axillary, monosymmetric, resupinate, K connate, lobed, deeply divided adaxially, C 2, ?not clawed, abaxial, connate by epidermis, slipper-shaped, nectary abaxial, androgynophore curved abaxially; A 8, median members absent, 4 abaxial fertile, 4 adaxial staminodial. - 1/1: Emblingia calceoliflora. West Australia.

• Emblingiaceae are low-growing herbs with axillary, ebracteolate and strongly monosymmetric flowers. The flowers are held upside down, the large, connate calyx is slit adaxially, there are two, abaxial, slipper-shaped petals, and there is a curved androgynophore with an abaxially-positioned nectary at the base.

The plant dries yellowish. There has been considerable disagreement over the floral structure: is the flower resupinate or not? Is the hood petal-like or not? Are there one or three carpels? I largely follow Melville's interpretation (in Erdtman et al. 1969), see also Mueller (1860). Detailed studies of all aspects of this plant are needed, nevertheless, its embryo is curved, its flowers monosymmetric, and its nectary is between the petals and stamens, all features appropriate for a position around here.

Emblingia was included in Polygalaceae by Cronquist (1981), Polygalales by Takhtajan (1997), and in Gentianales by Savolainen et al. (2000b).

For general information, see Kubitzki (2002d).

Pentadiplandraceae [Gyrostemonaceae + Resedaceae] Tovariaceae [Cleomaceae [Capparaceae + Brassicaceae]]: ?.

PENTADIPLANDRACEAE Hutchinson & Dalziel - K 5, valvate, C 5, in close contact at enlarged, concave base, limb flat; carpelate flowers: style long; fruit a berry; 1 seed/loculus, coat with layer of white, wool-like, elongated cells towards outside. - 1/?1: Pentadiplandra brazzeana. Tropical W. Africa.

The fruit contains the sweet-tasting protein, brazzein.

Ronse de Craene and Haston (2006) suggest that the floral morphology of Pentadiplandra is close to the ancestral form of that of the core Brassicales.

Pentadiplandra is poorly known, but see Ronse Decraene (2002: floral anatomy) and Bayer and Appel (2002: general).

Gyrostemonaceae + Resedaceae: hairs unicellular; styluli +; seeds arillate.

The composition of this clade and relationships within it are currently uncertain. Some Stixeae (ex Capparaceae), the Asian Tirania and the New World Forchhammeria, come out around here (Hall et al. 2002, 2004). Their flowers tend to be 3-, 5- or 6-merous, there is no differentiation between the two perianth whorls, and the placentation is axile (see also Kers 2002). Tirania has six sepals and petals and axile placentation. Forchhammeria has two carpels, as well as an irregular number of sepals (see Gyrostemonaceae), no petals, parietal placentation and one ovule/carpel, only one ovule/fruit usually developing. Stixeae (the name Stixaceae Doweld is available) are all Indo-Malesian climbers with successive cambia (Carlquist 1988) and are certainly out of place in Capparaceae, however, further studies are needed to place them accurately in the phylogeny.

GYROSTEMONACEAE Endlicher - P uniseriate, connate, pollen colpate; G borne around the central axis in whorl(s). - 5/18+: Gyrostemon (12). Australia.

• Gyrostemonaceae are yellowish-drying (including the twigs) shrubs with rather small, often quite narrow but otherwise featureless leaves. The rather small flowers with a persistent, connate perianth whorl and many stamens and/or carpels clearly arranged in a whorl are quite distinctive.

For information, see Tobe and Takahashi (1995: pollen), Goldblatt et al. (1976: general), Hufford (1996: floral development) and George (2002d: general).

RESEDACEAE Berchtold & J. Presl - C ligulate at junction of claw and limb, limb fringed. - 3/75: Reseda (70). Warm temperate and dry subtropical, esp. Mediterranean-Middle East-North African, also southern Africa, S.W. North America and S.W. China (all Oligomeris).

• Resedaceae are yellowish-drying sometimes somewhat shrubby herbs with terminal, racemose inflorescences and rather small, monosymmetric flowers. The leaves are quite variable, although often there are small stipules. The petals are free and clawed; the adaxial petals are usually largest and the limb is often fringed. The extrastaminal disc/nectary is often conspicuous. The dry fruits often have an apical opening persisting between the styles (a quasi follicle) and the seeds are often arillate.

Ochradenus has C 0, A many; G [3], the carpels are ultimately closed, and the fruit is berry-like - c.f. Gyrostemonaceae (Hufford 1996). The androecium of Reseda luteola may be in 3-4 whorls (for references, see Abdallah 1978), c.f. again Gyrostemonaceae. However, given the phylogenetic structure in the family, neither case suggests direct similarities with Gyrostemonaceae, since both taxa are derived within Resedaceae.

Martín-Bravo et al. (2007) discuss the phylogeny and biogeography of the family; given the phylogenetic structure that they found, three genera should be recognised - [Caylusea [Sesamoides + Reseda]].

For stipules, see Weberling (1968), for general information, see Abdallah and de Wit (1979 and references) and Kubitzki (2002d), and for wood anatomy, see Carlquist (1998a) and Schweingruber (2006).

TOVARIACEAE Pax - G [(5-)6(-8)]. - 1/2. Tropical America.

• Tovariaceae are herbs that can be recognised by their trifoliate, stipulate leaves, terminal, racemose inflorescence, and 6-9-merous, polypetalous flowers with a short style and spreading stigmas.

For general information, see Appel and Bayer (2002).

Capparaceae [Cleomaceae + Brassicaceae]: methyl glucosinolates, erucic acid [fatty acid] +; cisternae of endoplasmic reticulum organelle-like; flowers 4-merous; A from 4 primordia, centrifugal, longer than the petals, gynophore +, placentation parietal; K deciduous in fruit; seeds with invagination of the coat.

The divergence between Capparaceae and Brassicaceae has been dated to ca 41Ma before present (Schranz & Mitchell-Olds 2006, but c.f. Wikström et al. 2001). Pierid caterpillars (Pieridae-Pierinae) are notably common on members of this group. For details of the interactions of butterflies and plants, see Courtney (1986) and Chew (1988) and references. Ca 1,000 species of these plants are susceptible to pseudoflower-forming rust fungi (Puccinia spp.), especially in Brassicaceae (see Roy 1993, 2001).

Although Cruciferae/Brassicaceae s. str., cabbage and mustard, have always been considered one of the most natural plant families, their recognition makes Capparaceae s.l. (= Capparaceae s. str. + Cleomaceae) paraphyletic. So the alternatives are to have one family (Brassicaceae s.l.), three families, or two families, a Capparaceae and a Brassicaceae including Cleomoideae; the second option is followed here.

True blue or red flowers are very rare in the whole group.

The distribution of methyl glucosinolates and of root hairs and variation patterns in seed coat anatomy are a little odd. Both Brassicaceae and Resedaceae have glucosinolates derived from elongated amino acid chains (Kjær 1973; esp. Fahey et al. 2001).

CAPPARACEAE Pax - Fruit indehiscent. - 16/480: Capparis (250), Maerua (100), Boscia (37), Cadaba (30). Largely tropical.

• Capparaceae are woody, the inflorescence is racemose, the flowers have a gynophore and often many stamens with long filaments, and the fruit is indehiscent.

A paraphyletic Crateva is strongly supported as being sister to the rest of the subfamily; Capparis is probably diphyletic (Hall et al. 2002; Hall 2008). Hall (2008) discusses generic limits in Capparaceae, which are in need of substantial work, thus New World Capparis will need a new name.

Forchhammeria and perhaps other Stixeae that used to be placed here are properly to be placed in or near Resedaceae. Setchellanthus (see Setchellanthaceae) and Koeberlinia (Koeberliniaceae, both still Brassicales) also used to be included here.

For some information on floral development, see Ronse Decraene and Smets (1997), and for general information, see Kers (2002).

Cleomaceae + Brassicaceae: herbaceous annuals; inflorescence ± corymbose; C 4, clawed; A 6; G [2]; fruit septicidal, persistent woody placentae +.

Since Aethionema, with a more or less sessile gynoecium, is sister to all other Brassicaceae, similarities of Stanleya, etc., which form a clade well embedded in Brassicaceae, to Capparaceae and Cleomaceae (e.g. long gynophore) are presumably reversals (Galloway et al. 1998; c.f. Judd et al. 2002).

For floral development, see Leins (2000, and references).

CLEOMACEAE Horaninow - Anthers linear, coiled at dehiscence. - 10/300: Cleome (275: including Podandrogyne). Tropical and warm temperate, esp. America.

• Cleomaceae are herbaceous or shrubby plants with palmately compound leaves, the inflorescence is racemose, the flowers have four clawed petals, six stamens, and two carpels; the gynoecium has a gynophore, the stamens have long filaments, and the dehiscent fruit has a thin, persistent, loop-like woody placenta that remains on the plant after the fruit valves have fallen off.

For a phylogeny of Cleome and its immediate relatives, see Inda et al. (2008b). Podandrogyne (probably = Cleome) is monoecious; rather unusually, the staminate flowers are basal on the inflorescence. However, relationships within Cleomaceae have rather little support, even if there are suggestions that Cleome itself is widely scattered on the tree (Hall 2008).

The small-flowered Dipterygium, placed in Capparaceae in a subfamily by itself by Kers (2002), is possibly to be included here. It has six stamens that are all equal in length, a filiform style, a fruit that is a winged, 1-seeded nut, and an embryo with incumbent cotyledons; methylglucosinolates are recorded from the plant (Hedge et al. 1980).

For general information, see Kers (2002: in Capparaceae), for anther dehiscence, see Mitchell-Olds et al. (2005).

BRASSICACEAE Burnett - Methyl glucosinolates 0; roots lacking mycorrhizae; the two outer stamens shorter than the four inner, all about as long as petals, commissural septum +, stigma commissural; cyst-like structure in the endosperm at the chalazal end of the seed, embryo folded, but no invagination of the coat. - 338/3710. World-wide, esp. N. temperate (but less E. North America).

Aethionemeae Al-Shehbaz, Beilstein & E. A. Kellogg - Plant glabrous; fruit flattened at right angles to the plane of the septum [angustiseptate]. - 1-2/70. The Mediterranean and Europe to Afghanistan.

The Rest - Genome duplication. - 335/3640: Draba (365), Cardamine (2006), Erysimum (225), Lepidium (230), Alyssum (195), Arabis (120), Boechera (ex Arabis - 110), Physaria (105: inc. Lesquerella), Rorippa (85), Heliophila (80), Isatis (80), Noccaea (80), Thlaspi (55), Biscutella (55), Matthiola (50), Descurainia (50), Hesperis (45), Sisymbrium s. str. (45). World-wide, esp. N. temperate (but less E. North America, and even more so humid lowland tropics).

• Most Brassicaceae are herbs, and a number of taxa have distinctive small, branched, eglandular hairs. The ebracteate inflorescence is corymbose, the flowers have four, clawed petals (the alternate name 'Cruciferae' for this family is taken from the cross-like arrangement of these four petals), six stamens with relatively short filaments (so the anthers barely protrude), the gynoecium has two carpels and a short style, and the dehiscent fruit has both persistent, almost woody placental strands and a membranous septum.

Stem group Brassicaceae are estimated to be 22-18 Ma old (Wikström et al. 2001), some 41 Ma old (Schranz & Mitchell-Olds 2006), or some 50 Ma old, with divergence of Aethionemeae from the rest ca 40Ma before present (Al-Shehbaz et al. 2006). Initial diversification was in the Old World. The relationships between many of the tribes are still unclear, and there seems to have been a rather rapid radiation of the clades that encompass most of the diversity of the family today (Beilstein et al. 2006; Al-Shehbaz et al. 2006; Bailey et al. 2006a, b). Kiefer et al. (2009) discuss the phylogeographic structure of the speciose largely North American Boechera; divergence seems to be a Pleistocene phenomenon with considerable hybridization and apomixis.

Duplication of the whole genome seems also to have occurred fairly soon after the split from the Capparaceae and has been dated to 34-25Ma before present (Vision et al. 2000; Blanc & Wolfe 2004; Schranz & Mitchell-Olds 2006); there is also extensive genome duplication within the family (Kellogg & Bennetzen 2004), and this has happened more than once (Vision et al. 2000; Blanc & Wolfe 2004). As a result, "diploid" species like Brassica oleracea, with n = 9, are hypothesised to be ancestral hexaploids (Lysack et al. 2005; Mitchell-Olds et al. 2005). Simillion et al. (2002) and others suggest that there has been a total of three genome duplications in Arabidopsis. Karyotype evolution in the family is of considerable interest. One major clade (lineage I) has x = 8, perhaps the ancestral number of the family (Lysack et al. 2006), while another major clade, that includes lineage II (Brassiceae, etc.), has x = 7, but with subsequent increases (Mandáková & Lysack 2008).

According to Medve (1983), any mycorrhizal associations with the roots of Brassicaceae are at best weak and facultative, and although arbuscular mycorrhizae have recently been reported from Thlaspi, it is doubtful if an effective symbiosis results (Regvar et al. 2003). Indeed, glucosinolates can depress vesicular-arbuscular mycorrhizal activity, a possible element in the invasive capabilities of Alliaria petiolata in parts of North America (Wolfe & Klironomos 2005). As is well known, glucosinolate diversity in the family is considerable, as is variation in content between different species in the same community. The presence of particular glucosinolates may induce oviposition by Pieris butterflies, whether or not the crucifer with that glucosinolate is edible or kills the caterpillar (Chew 1979, see also 1988). Caterpillars of cabbage white butterflies (Pieris spp.) and the diamond-back moth (Plutella xylostella) are able to convert the glucosinolates that are produced when they damage plant tissue in the course of eating into non-toxic substances such as nitriles (Wittstock et al. 2004; Ratzka et al. 2002 - see also introduction to Brassicales above). Insects such as the cabbage aphid, Brevicoryne brassicae, sequester the glucosinolates, in this case even producing its own myrosinases that break the glucosinolates down and helping to deter predators (Kazana et al. 2007).

The association between the pseudoflower-forming Puccinia rust and host has been much studied in Brassicaceae (Roy 1993 [particularly fine photograph], 2001). Insects come to the pseudoflowers, attracted both by colour and floral fragrances, these latter different from those of both of the crucifer host and the other plants in the general area (Raguso & Roy 1998). They pick up the rich fructose nectar secreted by the fungus along with the fungal spermatia in the nectar; they then fly to another "flower", deposit the spermatia and pick up more. On combination of spermatia of the appropriate mating types, diploid aecia are produced, the "flower" stops producing nectar, and the leaves turn green.

Most of the nickel and zinc hyperaccumulators known in flowering plants occur in Brassicaceae (Grennan 2009 for a review).

The flowers of Iberis amara are monosymmetric, although the corymbose inflorescence is functionally more like a single polysymmetric flower (Busch & Zachgo 2007). The commissural stigmas of Brassicaceae have been supposed to indicate that the gynoecium is basically 4-carpelate, however, such stigmas are common in groups with parietal placentation (c.f. Dickison 2000; Brückner 2000). Another controversy concerns the six stamens - did they arise by dédoublement or by reduction? There is also considerable variation in nectary morphology and position. Nectaries may be either outside or inside the stamens (Bernadello 2007); the latter position is likely to be derived.

The border cells of the root cap dissociate in rows (Driouich et al. 2006), unlike the irregular dissociation in other angiosperms; other Brassicales should be examined for this character. There are rarely glandular stipules in the inflorescence and elsewhere (see Weberling 2006 for a summary, also Bowman 2006). The chalazal endosperm cyst may be involved in the movement of metabolites into the developing seed since it is surrounded by transfer cells (Brown et al. 2004).

For information, see Vaughan and Whitehouse (1971: ovules and seed coat), Bouman (1975: ovules and seed coat), Erbar and Leins (1997a, b: floral development), Appel and Al-Shehbaz (2002), Koch et al. (2003), Hurka et al. (2005) and Mitchell-Olds et al. (2005) for general information and references, Warwick and Al-Shehbaz (2006: chromosome numbers), Grubb and Abel (2006: glucosinolate metabolism), Bowman (2006: morphology and developmental genetics), Beilstein et al. (2006: trichome evolution), Koch and Kiefer (2006: biogeographic studies), Warwick and Al-Shehbaz (2006: chromosome numbers), Marhold and Lihová (2006: hybridization and polyploidy), and Warwick et al. (2006b: species checklist).

Phylogeny. Of late, substantial progress has been made in providing a phylogenetic framework for the family (e.g. Koch et al. 2001: Koch 2003; Beilstein et al. 2006) and realigning taxa accordingly; Al-Shehbaz et al. (2006) recognise 25 tribes in which they place 260 of the 338 genera of the family (see also Beilstein et al. 2008). For a study of Alysseae and related tribes, see Warwick et al. (2008).

Genera in the past were often based on fruit and embryo differences and their limits are often turning out to be unsatisfactory; there has been parallel or convergent evolution of just about all the morphological features used to distinguish genera (Koch 2003; Al-Shehbaz et al. 2006; Bailey et al. 2006a, b; Beilstein et al. 2008). Hence the major disagreements over generic limits, and although intergeneric hybridisation is quite common (see summary in Warwick et al. 2006), this is of uncertain significance given the problems with generic circumscriptions. Although genera such Draba and Lepidium are monophyletic or largely so on both morphological and molecular grounds, others, such as Brassica, are not (e.g. Mitchell-Olds et al. 2005 and references). Although Arabidopsis thaliana is perhaps the most important model vascular plant in biology, the limits of the genus Arabidopsis are only now being established (see Clauss & Koch 2006 for a discussion of its immediate relatives). Koch and Mummenhoff (2001) and Meyer (2003) discuss generic limits surrounding Thlaspi, a polyphyletic genus; the sometimes invasive Alliaria petiolata, with very different fruit morphology, is included around here. Sisymbrium s. str. is restricted to the Old World, the New World taxa being unrelated and mixed in with Theylpodieae (Warwick et al. 2002, 2006a). There is extensive polyploidy and hybridization in Cardamine (Lihová & Marhold 2006).

MALVALES Dumortier - [Back to Index]

(Cyclopropenoid fatty acids +), myricetin +; mucilage cells +; stipules +; C contorted, nectarifeous disc 0, style long; exotegmen much thickened and lignified, palisade. - 10 families, 338 genera, 6005 species.

• Many members of this order can be recognised by their spiral, pulvinate leaves that often have palmate venation, or at least a pair of lateral veins leaving from the very base of the blade; stellate or fasciculate hairs and their derivatives are common. The bark is often very fibrous and tough because of the stratified phloem. The calyx is sometimes valvate, the corolla often contorted, the stamens usually numerous, the capsule sometimes more or less spiny, and the seeds or inside of the carpel walls sometimes conspicuously hairy.

Malvales contain ca 3.2% eudicot diversity (Magallón 1989). Timing the separation of clades in this group is of considerable interest given the distributions of many of the families (see Ducousso et al. 2004). Taxa with ectomycorrhizal associations are common in this order, e.g. Malvaceae (Tilioideae and ex Sterculiaceae), Dipterocarpaceae and Cistaceae (Smith and Read 1997), and Sarcolaenaceae (Ducousso et al. 2004); it will be interesting to see just how widely distributed such mycorrhizae are.

Species with stratified phloem always have wedge-shaped phloem rays, Sarcolaenaceae perhaps excepted (Kubitzki & Chase 2002). The androecium is possibly basically oligomerous; polyandry is derived, and arises in various ways. Starchy endosperm may be an apomorphy for the group. For the distribution of cyclopropenoid fatty acids, which are also scattered outside Malvales, see Bayer et al. (1999). However, as Kubitzki and Chase (2002: Table 1) show, the evolution of this and other features common in the order are difficult to understand given out current knowledge both of phylogeny and of basic morphology.

Elaeocarpaceae, previously usually included in or near Malvales, are placed unambiguously in Oxalidales. Most Malvales as delimited here are included in Takhtajan's (1997) Malvanae.

Some clades within this order are quite well established, but relationships between them, as well as the positions of one or two families, remain unclear (Fay et al. 1998a; Bayer et al. 1999). These may be represented as Muntingiaceae [[[Thymelaeaceae + Sphaerosepalaceae] [Neuradaceae [Sarcolaenaceae [Dipterocarpaceae + Cistaceae]]]] [Bixaceae + Malvaceae]], but few of these relationships have much support, even after successive weighting. Other studies suggest that Neuradaceae may be sister to the rest of the order (Soltis et al. 2007a; Nickrent 2007). Relationships between Sarcolaenaceae, Dipterocarpaceae, and Cistaceae are uncertain (Ducoussu et al. 2004). Relationships of Cytinaceae with Malvales have been suggested (Nickrent 2002; esp. Nickrent et al. 2004). Since the only Malvales included in Nickrent et al. (2004) were Cistaceae and Malvaceae, the placement of Cytinaceae was somewhat provisional; however, recently Nickrent (2007) found that Cytinaceae were sister to the poorly-known Muntingiaceae, support being moderate to strong. Both Cytinaceae and other Malvales have exotegmic seeds, and aspects of the perianth of Cytinaceae and Malvaceae can perhaps be compared. Apodanthaceae, here placed in unplaced eudicots, may also best be included in Malvales. They both have a connate androecium and extrose anthers, inferior ovary, parietal placentation, etc. (Nickrent et al. 2004, but see also Cucurbitaceae). In any event, the tree here is conservative.

For information, see Nandi (1998b) and Kubitzki and Chase (2002), both general; Gaydou and Ramanoelina (1983: cyclopropenoid fatty acids), and den Outer and Vooren (1980) and den Outer and Schütz (1981), both wood anatomy.

Includes Bixaceae, Cistaceae, Cytinaceae, Dipterocarpaceae, Malvaceae, Muntingiaceae, Neuradaceae, Sarcolaenaceae, Sphaerosepalaceae, Thymelaeaceae.

NEURADACEAE Link - Pollen with 3 or 4 pores at each end, not all carpels fertile, styles separate. - 3/10. Africa to India, dry or desert areas.

• Neuradaceae are rather small herbaceous or subwoody plants growing in the Saharo-Indian desert areas with small, toothed, exstipulate leaves and large flowers the petals of which dry a distinctive color; there are 5-10 separate styles.

The plant probably lacks stipules, the stipule-like structures that are sometimes seen in fact being prophyllar (Bayer 2002). The family is poorly known.

Neuradaceae have previously placed been elsewhere, as in Rosales in Cronquist (1981) and Takhtajan (1997), Hutchinson (1973) even included it in the Rosaceae, and the floral anatomy of the two is similar (Ronse Decraene & Smets 1996b). However, the corolla dried a distinctive purplish colour, as in some Malvaceae (Airy Shaw 1966) - see Huber (1993a).

For general information, see Bayer (2002).

Thymelaeaceae, Sphaerosepalaceae, Bixaceae [Cistaceae + Sarcolaenaceae + Dipterocarpaceae] [Muntingiaceae + Cytinaceae] Malvaceae: pits vestured; phloem stratified, phloem rays wedge-shaped.

THYMELAEACEAE Jussieu - Pericyclic fibres 0; crystal sand +; epidermal cells (massively) mucilaginous; stomata cyclocytic. - 46-50/891. World-wide, esp. trop. Africa and Australia.

1. Tepuianthus - Scales outside androecium, styles bifid. - 1/7. Guayana highlands, N.E. South America.

Octolepidoideae + Thymelaeoideae: P [= K?] with appendages [scales; ligules], C 0, pollen oligo- to polyporate, minutely spinulose, style single, stigma ± capitate; endotegmen with stripes on the inner surface.

2. Octolepidoideae Gilg - Secretory cavities + [leaves punctate]; glandular scales 4-40, anthers usu. recurved and hippocrepiform, connective well developed; seed arillate, with swollen funicle, or angled at the raphe. - 8/49: Gonostylus (20). Tropical Africa and Madagascar (Octolepis), Malesia to Australia (Queensland), New Caledonia, Fiji.

3. Thymelaeoideae Burnett - Phorbol ester diterpenes +; internal phloem +; vascular bundles bicollateral; inflorescence often capitate; hypanthium long-tubular, petal-like appendages +/0, pollen crotonoid [with supratectal processes, i.e. 5-6 capitate columellae forming a ring]; G 2; fruit indehiscent. - 37/690: Gnidia (140), Pimelea (110), Daphnopsis (55), Daphne (ca 100, inc. Wikstroemia, see Halda 2001), Lachnaea (40). World-wide, esp. trop. Africa and Australia.

• Thymelaeaceae are recognisable by their very fibrous bark and entire, exstipulate often opposite leaves which frequently have rather close, parallel venation and leave prominent raised scars on the stem after they fall off; the indumentum is of unicellular hairs and is often silky-adpressed. In many taxa, the twigs are notably flexible, and plants often smell unpleasantly. The flowers are quite often in heads, and the hypanthium is often long; any corolla lobes are usually at most small. The plants are often toxic.

The inclusion of Tepuianthus in this clade makes eminent morphological sense (Wurdack & Horn 2001). It has a number of apomorphies of other Thymelaeaceae, while the features in which it differs are mostly plesiomorphies, i.e., they are similarities with other Malvales. Tepuianthus has both a well-developed corolla and glandular scales outside the androecium, suggesting that the "corolla scales" of Gonostylus, also glandular, and perhaps those of the rest of the family, are homologous with those of Tepuianthus and would be a synapomorphy of the family. The palisade mesophyll of the leaf of Tepuianthus is massively mucilaginous, resulting in a clear subepidermal layer which appears to be hypodermal; this is found in other Thymelaeaceae such as Solmsia (New Caledonia: Octolepidoideae). Finally, the well developed parallel venation of Tepuianthus is very like that of other Thymelaeaceae, and Solmsia is vegetatively remarkably similar to Tepuianthus down to details of the base and mucronate apex of the lamina.

A number of taxa scattered through the family produce gaharu or agarwood. This is developed from the heartwood often after wounding and perhaps also infection by fungi; gaharu is much esteemed as source of incense and medecines, and some species that produce it have been decimated in the wild (Eurlings & Gravendeel 2005 for references and a phylogeny of some of the most commercially-important taxa).

The vasculature of the perianth was studied by Heinig (1951). The structures inside the calyx, whether they were paired and more or less opposite the sepals or single and in the petaline position, were supplied by vascular bundles that came from lateral branches of the sepal vasculature. Dicranolepis (Africa: Octolepidoideae) has large "petals" that are variable in number but paired and opposite the petals, and they are sometimes serrate or laciniate; its anther connectives are massive.

The pollen of many Thymelaeaceae-Thymelaeoideae is similar to that of Euphorbiaceae-Crotonoideae, and the chemistry is also similar to that of Euphorbiaceae (Seigler 1994). Microsemma (= Lethedon - Gonostyloideae) has cyclopentenoid cyanogenic glycosides; Spencer and Seigler (1985) suggested that because of this, it should be placed in Flacourtiaceae (see Achariaceae-Malpighiales). Heinig (1951) compared Thymelaeaceae with well over twenty families when discussing its relationships; almost by chance one might expect a member of Malvales to be one of these, and she did note similarities in wood anatomy with Tiliaceae...

For more information, see Domke (1934: general), Ding Hou (1960: general), Evans and Taylor (1983: phorbol esters), Wurdack and Horn (2001: Tepuianthus), van der Bank et al. (2002: phylogeny), Kubitzki (2002: Tepuianthaceae), Herber (2002a: palynology, 2002b: general).

Sphaerosepalaceae + Bixaceae [Cistaceae + Sarcolaenaceae + Dipterocarpaceae] [Muntingiaceae + Cytinaceae] Malvaceae: hairs often stellate; stipules well developed; A many, developing centrifugally, from 5 or 10 (15) bundles, when 5 often opposite the petals, several [³6] ovules/carpel, micropyle bistomal.

SPHAEROSEPALACEAE Bullock - 2/18. Madagascar.

• Sphaerosepalaceae are a small family, but with a great deal of variation. The stipules are broad, intrapetiolar, and more or less encircling the stem. The flowers are usually four merous, and the sepals are imbricate in pairs; the fruits or the individual carpels of deeply lobed fruits are single seeded.

For information, see Capuron (1962: general), Huard (1965: anatomy), Bayer (2002: general) and Horn (2004: morphology).

BIXACEAE Kunth - Plant with secretory canals; resin-filled cells outside veins; hairs glandular [not tufted or stellate]; flowers large; K imbricate, anthers porate, many ovules/carpel, funicles long; exotegmen curved inwards in chalazal region, chalazal plug present; embryo long, cotyledons spatulate, radicle short, stout. - 4/21. Pantropical.

In Cochlospermum vitiifolium the median sepal is abaxial, i.e. the floral orientation is inverted, there are no bracteoles, and the sepals are of unequal size (another interpretation could be that there are three "true" sepals + two bracteoles - see also Cistaceae below). Amoreuxia has monosymmetric flowers, the adaxial stamens being much shorter than the abaxial ones; flowers of Cochlospermum are monosymmetric in bud, and the floral vasculature is monosymmetric.

The orange colouring of Bixa orellana, annatto, is used as a food colouring, e.g. for margarine.

For information, see Keating (1970, 1972 and references: anatomy of Cochlospermaceae s. str.), Poppendieck (1980: monograph of Cochlospermaceae s. str., 2002: general), Dickison (1988: Diegodendron anatomy), and Bayer (2002: Diegodendron, general).

Cistaceae + Sarcolaenaceae + Dipterocarpaceae: ectomycorrhizae +; ellagic acid +; plant with secretory canals; K imbricate, two outer members often different from the rest, filaments not articulated; exotegmen curved inwards in chalazal region, chalazal plug present; endosperm starchy.

The similarity in seed coat anatomy between this group and Bixaceae is striking; in both cases, the chalazal plug has a core and annulus astructure. Corner (1976 1: 97) described Cistaceae as being "little more than variations on a single generic theme", and noted similarities between Cistaceae and Bixaceae (see also Takhtajan 1997).

CISTACEAE Jussieu - Stigma with multicellular multiseriate papillae. - 8/175. More or less worldwide, often temperate or warm temperate, esp. Mediterranean region.

• Cistaceae are aromatic shrubs growing in open, sunny areas often on a sandy or chalky substrate. They often have opposite leaves with broad, even connate bases. The inner three sepals are sometimes described as being contorted, rather, the aestivation is extreme-quincuncial, the two outer sepals being much smaller than the others. The petals are free and are contorted in a direction opposite to that of the three large sepals and are crumpled when in bud; there are numerous stamens.

Many Cistaceae have stamens that are sensitive to touch, moving and dusting the insect with pollen.

Fumana and Lechea are successively sister to the remainder of the family; the former in particular has a number of features that are plesiomorphic in the family (Arrington 2004). Both have only three petals, the former, alone in the family, has staminodes.

For the absence of root hairs, at least in seedlings, and the fungal associations of the plant, see references in Arrington and Kubitzki (2002). Dickie et al. (2004) also mention ectomycorrhizae in Helianthemum.

For information, see Nandi (1998a, b: flower and seed development), Arrington and Kubitzki (2002, revised in Arrington 2004: general), and the Cistus and Halimium website.

Sarcolaenaceae + Dipterocarpaceae: stipules usu. well developed.

SARCOLAENACEAE Caruel - Involucre of varying morphology subtending flowers; disc extrastaminal, pollen in tetrads. - 8/60. Madagascar, mostly E. and C.

• The flowers, which have a massive involucre and sometimes also an enclosing bract and an extrastaminal disc, are distinctive. The underside of the lamina of Sarcolaena has parallel lines. The hairs are stellate.

Fossil pollen of Sarcolaenaceae is known from the Caenozoic of South Africa (Coetzee & Muller 1984).

DIPTEROCARPACEAE Blume - Mucilage canals in pith; K enlarging equally, thinnish, anther connective with terminal appendage; testa vascularized. - 17/680. Tropical, overwhelmingly W. Malesian in species diversity.

1. Monotoideae Gilg - Adaxial gland at base of lamina; androgynophore +. - 3/30: Monotes (26). Africa, Madagascar, South America.

2. Pakaraimaeoideae Maguire, Ashton & de Zeeuw - Included phloem +; C imbricate, shorter than K. - 1/1: Pakaraimaea roraimae. The Guianan Highlands, South America.

3. Dipterocarpoideae Burnett - Resin ducts pervasive in wood; no mucilage canals in pith; anthers ± basifixed, pollen tricolpate; fruit usu. 1-seeded nut, endocarp hairy, K enlarging unequally and thickish. - 13/650: Shorea (360), Hopea (105: these two should perhaps be merged), Dipterocarpus (70), Vatica (60). Seychelles, Sri Lanka, India, South East Asia to New Guinea, but mostly W. Malesian.

• Dipterocarpaceae are trees that may be recognised by their often two-ranked, coriaceous leaves with strong, parallel, if not particularly close, secondary veins and scalariform tertiaries; the hairs are often fasciculate or stellate. The inflorescences are often distinctively monochasial, and the small flowers are pointed in bud and have a conspicuously contorted corolla. Distinctive fruits with strongly accrescent but often unequal sepals surrounding single-seeded nuts are common in the family, but there are other fruit types.

Dipterocarpaceae are large trees that are often dominant in the tropical lowland forests from India and Sri Lanka to West Malesia, where they grow both in seasonal and everwet forests and are the most diverse family in the latter, at least (Gentry 1988); elsewhere in the range of the family members are also often described as being abundant. Individual species tend to be mast fruiters, but in the everwet forests of S.W. Sri Lanka and West Malesia all taxa in the family tend to flower and especially fruit at the same time, apparently in response to climatic changes induced by El Niño events; this kind of behaviour is very uncommon in the tropics. The ectomycorrhizal habit of the family has been implicated in this distinctive phenological behaviour, and/or predator satiation (Janzen 1974a, no mention of fungi; Ashton 1982, 2002). Dipterocarpaceae lack parasitic rust fungi (Uredinales), unlike many of the other ectomycorrhizal groups (Malloch et al. 1980).

Pollination of species of Shorea section Mutica and of other dipterocarps in Sarawak is by chrysomeild and curculionid beeltes; in peninsula Malaysia species of section Mutica are apparently pollinated by thrips (Sakai et al. 1999).

Dipterocarpaceae lack parasitic rust fungi (Uredinales), unlike many of the other ectomycorrhizal groups (Malloch et al. 1980); this observation should be confirmed - or not - for other members of this clade; for mycorrhizae in Pakaraimaeoideae, see Moyersoen (2006).

See Dayanandan et al. (1999) and Morton et al. (1999) for relationships within Dipterocarpaceae. Generic limits within Dipterocarpoideae need attention (Yulita et al. 2005), Neobalanocarpus possibly even being an intergeneric hybrid (Shorea sp. X Hopea sp., see Kamiya et al. 2005).

For additional information, see Gottwald and Parameswaran (1966: wood anatomy), Ashton (1982: general, 2002: general), Kajita et al. (1998), Appanah (1998: ectomycorrhizae), and Kocyan (2005: floral development), Maguire et al. (1977) and Maguire and Ashton (1980), both Pakaraimaea, and Catalina Londoño et al. (1995), Morton (1995), both Monotoideae.

Cytinaceae + Muntingiaceae: fruit a berry.

For general information about this family pair, see Nickrent (2007). Synapomorphies for the clade depend on more detailed knowledge of all aspects of the poorly-known Muntingiaceae and a comprehensive phylogeny of both families.

CYTINACEAE A. Richard - Echlorophyllous endophytic root parasites; vessels 0; inflorescence racemose; flowers imperfect; P 4-9, basally connate; staminate flowers: A 6-10, extrorse, connate, monothecate, nectariferous cavities between the stamens; carpelate flowers: nectary near base of style; G [8-14], inferior, placentation intrusive parietal, many straight tenuinucellate ovules/carpel; exotegmic cells thickened all around; embryo undifferentiated. - 2/10. Central America, Mediterranean, South Africa and Madagascar.

• Cytinaceae are echlorophyllous parasites with racemose inflorescences, sometimes capitate, in which the individual flowers are moderately sized and easy to make out. The more or less spreading basally connate perianth is in a single whorl, the stamens are extrorse (staminate flowers) and the ovary inferior and the style long and expanded towards the apex (carpelate flowers).

Cytinaceae are quite often parasitic on Cistaceae (same order!) in the Mediterranean region, but on a variety of non-Malvalean families in Africa. The seeds become embedded in mucilaginous material derived from the placentae and funicles (Nickrent 2007).

For general information see the Parasitic Plants website (Nickrent 1998 onwards) and also Heide-Jørgensen (2008), de Vega et al. (2008) for population differentiation in the western Mediterranean and de Vega et al. (2007) for endophyte anatomy.

MUNTINGACEAE C. Bayer, M. W. Chase & M. F. Fay - Prophylls basal on axillary shoot, heteromorphic, stipule-like, stipules ?0; inflorescence extra-axillary, fasciculate; C imbricate, crumpled in bud, pollen small, ca 10 µm, triporate. - 3/3. Tropical America.

• Muntingiaceae can be recognised by their two-ranked, toothed, leaf blades with asymmetric bases and heteromorphic stipule-like prophylls. Their flowers are in extra-axillary fascicles, the calyx is valvate, the corolla shortly clawed and crumpled in bud, and the stamens are numerous.

Although Muntingiaceae appear to have stipules, this may not to be the case. Dicraspidia has strikingly asymmetric prophylls; on the adaxial side of the branch they are orbicular, foliaceous and persistent, while on the abaxial side they are linear, thin and caducous. In Muntingia only an adaxial prophyll is present, and it is narrow (Karima Gaafar, pers. comm.: the situation in Neotessmannia is unknown). Some characters of Muntingiaceae (lack of stipules; pits not vestured) might suggest that the family may be rather basal in Malvales. Petenaea (?Elaeocarpaceae) may also be associated with this clade (Bayer et al. 1999); the genus is described as having minute stipules (Bayer 2002).

Muntingia has a superior, ovary, caducous calyx, and pendulous placentae, the two other genera have inferior ovaries, laminar placentation, and a persistent calyx (?: Neotessmannia). Stamens, etc., of Muntingia are borne on a massive, almost disc-like structure towards the inside of which are dense hairs; the inner side of this disc as it faces the ovary seems to be a nectariferous region.

Muntingiaceae were placed in Tiliaceae - Neotessmannioideae by Takhtajan (1997).

Some information is taken from Venkata Rao (1952: embryology), Bayer (2002: general), Bayer et al. (1998c: new family), and Carlquist (2005a: anatomy).

MALVACEAE Jussieu - Inflorescence made up of modified cymose units ["bicolor units" - three bracts, two of which subtend flowers or cymose part inflorescences]; nectary of multicellular hairs borne adaxially on the K bases [?this level]; A fundamentally obdiplostemonous, anthers extrorse. - 243/4225+. Largely tropical, also temperate.

Grewioideae + Byttnerioideae: ?

1. Grewioideae Hochreutiner - K free. - 25/770: Grewia (290), Triumfetta (150), Corchorus (40-100), Microcos (60). Pantropical (warm temperate).

2. Byttnerioideae Burnett - petal base expanded, concave [cucullate], limb clawed. - 26/650: Byttneria (135), Hermannia (100), Ayenia (70), Melochia (55: A 5), Theobroma (20). Pantropical, esp. South America.

Sterculioideae + Tilioideae + Dombeyoideae + Brownlowioideae + Helicteroideae [Malvoideae + Bombacoideae]: 21 bp deletion in ndhF gene.

3. Sterculioideae Burnett - Plant monoecious, inflorescence axillary, paniculate, lacking cymose units, epicalyx 0; C 0, androgynophore +, staminodia 0; G largely free; fruit a follicle. - 12/430: Sterculia (150), Cola (125). Pantropical.

4. Tilioideae Arnott - 2(?3)/50: Tilia (23). N. temperate.

5. Dombeyoideae Beilschmied - Pollen spiny. - 21/381: Dombeya (225), Melhania (60). Old World tropics, St Helena, esp. Madagascar and Mascarenes.

6. Brownlowioideae Burret - Anthers arrow-like [thecae diverging at base, converging at apex]. - 8/68: Pentace (25). Tropical, esp. Old World.

7. Helicteroideae Meisner - C often with lateral constrictions, androgynophore +. - 8-10/95: Helicteres (40), Durio (27). Tropical, esp. Southeast Asia and W. Malesia.

Malvoideae + Bombacoideae: K connate, adaxially at base with multicellular clavate nectariferous hairs [?level], C adnate to base of A, staminal tube +.

8. Malvoideae Burnett - Staminal column with apical teeth [?level], (anthers bisporangiate/monothecal), pollen spiny. - 78/1670: Hibiscus (550, inc. Abutilon, etc. - see Pfeil & Crisp 2005), Sida (200), Pavonia (150), Nototriche (100: epiphyllous inflorescences), Cristaria (75), Gossypium (40). Temperate to tropical.

9. Bombacoideae Burnett - Trunk often stout, with parenchymatous water-storage tissue, bark thin; leaves palmately compound; stamens usu. in bundles [fasciculate], (anthers bisporangiate/monothecal), pollen ± flattened, triangular, nucellar cap +; cotyledons folded. - 16/120: Pachira (24), Pseudobombax (20). Tropical, esp. America and Africa.

• Malvaceae are usually readily recognisable even when sterile by the combination of fibrous bark, alternate, stipulate leaves with toothed margins, ± palmate venation, and stellate to lepidote indumentum. Mucilage is common (c.f. Elaeocarpaceae!). The flowers and fruits are also distinguishable by a combination of characters. The flowers often have a valvate, connate calyx with a nectary at the base inside and a contorted corolla; the stamens are often numerous and variously connate and/or fasciculate. The inside wall of the fruit and/or the surface of the seed is often hairy.

Pollen of Bombacoideae is known from deposits 69-65 Ma old (Pfeil et al. 2002 for references).

Caterpillars of the nymphalid Acraea are quite commonly found on members of Malvaceae, as are members of Lycaeninae (Fielder 1995). Acanthoscelides, whose larvae eat seeds, are bruchids (Chrysomeloidea-Bruchidae-Bruchinae) that have diversified on Malvaceae s.l., esp. on Malvoideae, more than on any other group outside their primary hosts, members of Fabaceae (Kergoat et al. 2005). Seed-eating bugs of the Hemiptera-Lygaeidae-Oxycareninae are also concentrated on Malvoideae (Slater 1976), and galls are quite commonly found on plants of Malvaceae (Mani 1964).

In most Malvaceae studied the nectary is made up of multicellular glandular hairs covering glandular tissue beneath; the same morphology even occurs in the foliar extrafloral nectaries in Triumfetta (Leitão et al. 2005). The nectaries that provide rewards for pollinators are commonly found on the inside of the calyx and the corolla is fused at most basally, the petal lobes soon separating and leaving a space through which the pollinator can get at the nectar. These two features are probably connected, because access to the nectar must be permitted, and if the corolla were completely connate there would be no easy way for the pollinator to reach the nectar while simultaneously pollinating the flower (for nectary morphology, etc., see Sawidis et al. 1989 and references; Vogel (2000; Leitão et al. 2005). However, overall there is considerable variation in nectary - and staminodium - position and type in the family. Nectaries are borne on the petals in some Grewioideae, e.g. Grewia and Luehea, and on the androgynophore and leaf in e.g. Triumfetta (Leitão et al. 2005), while nectar in some Byttnerioideae, at least, is secreted through nectarostomata. Byttnerioideae often have remarkably complex if sometimes quite tiny "basket" flowers that are pollinated by small flies and the like (see Westerkamp et al. 2006). The petal often has a concave lower portion more or less enclosing the rest of the flower, the limb of the petal may dangle and twist in the wind, as in Abroma, and the prominent staminodia are opposite the sepals. Pollination by small insects is likely here. For pollination in Bombacoideae, see Janka et al. (2008) and references.

Distinctive tile cells occur in wood rays. They are best observed in radial section and are of two or three main types (Manchester & Miller 1978; Carlquist 1988b; Tang et al. 2005b). However, any correlation of tile cell "type" with phylogeny awaits a more completely resolved tree, thus the Durio type occurs in Helicteroideae, Malvoideae and Byttnerioideae.

Several taxa have palmate leaves. In Brachychiton and Adansonia there is comparable variation within a flush - the first leaf/leaves have a very short petiole and long, narrow ?phyllode, while later leaves are palmate; any intermediates have long, winged petioles and a few leaflets.

The inflorescence in most taxa is made up of "bicolor units" - a terminal flower with three bracts, two of which may subtend flowers or cymose part inflorescences that themselves have normal bracteole number and arrangement. The epicalyx seems to be made up of these three bracts, and so a flower with an epicalyx represents a highly reduced "bicolor unit" (Bayer 1999); such an epicalyx has evolved several times in the family. Although the inflorescence of Sterculioideae seems to be very differently constructed from that of other Malvaceae, it, too, may be made up of much modified bicolor units (Bayer 1999).

Von Balthazar et al. (2006) suggest an interpretation of the androecium of the [Malvoideae + Bombacoideae] clade, and extend their findings to propose a basic androecial structure for Malvaceae as a whole. They suggest that each androecial unit consists of an antesepalous primordium with its own vascular supply and which remains sterile (usually). It is flanked on either side by a single primordium, each derived from a separate antepetalous primordium and that each give rise to a sessile, elongated theca. Each theca is supplied by a branch from an antepetalous vascular bundle. The androecial unit thus consists of [half anther + sterile primordium + half anther]. Further details are given by Janka et al. (2008), focussing on Adansonia and relatives; Ceiba has only five alternipetalous stamens, supplied by branches of the oppositipetalous traces. For androecial development in Malvaceae s.l. compared to that in other Malvales, see Nandi (1998b) and von Balthazar et al. (2006). The basic androecial structure in this clade may be obdiplostemonous, with stamens developing in one or both whorls; anther dehiscence is extrorse (von Balthazar et al. 2006).

Not only are the carpels of Sterculioideae secondarily free, but in Firmiana they open early in development, exposing the developing seeds on the carpel margins; the ripe fruits with seeds attached are dispersed by wind. However, there is a compitum even in these apocarpous Sterculioideae; it develops after the postgenital connation of part of the styles (Jenny 1988). Wilkie et al. (2006) discuss evolution of fruit types, dispersal mechanisms, etc., within Sterculioideae; leathery follicles seem to be the basal condition in the group.

For information on relationships in the extended family, see Nyffeler et al. (2005 and references). [Grewioideae + Byttnerioideae] are probably sister to the rest of the family (see also Soltis et al. 2007a), while Sterculioideae are perhaps sister to the well supported [Malvoideae + Bombacoideae]. Other relationships between the subfamilies are unclear. Dombeyoideae and Tilioideae are sometimes weakly associated (Alverson et al. 1999), but the former may be sister to all other taxa in the big polytomy in the family (Nyffeler et al. 2005). Wilkie et al. (2006) suggest relationships within Sterculioideae. Sister to the rest - or almost so - in the [Malvoideae + Bombacoideae] clade may be things like Fremontodendron, etc. (ex Sterculiaceae - C 0), and Quararibea, etc. (ex Bombacaceae) (Alverson et al. 1999). However, Baum et al. (2004) suggested that Quararibea, etc., are best placed in Malvoideae, [Ochroma + Patinoa] and Septotheca are unplaced, and [Fremontodendron + Chiranthodendron] might be sister to all other members of this clade. For groupings within Malvoideae, see Judd et al. (2002). Within Malveae, Tate et al. (2005) found that presence or absence of an epicalyx correlated very well with two major clades recognizable on analysis of ITS sequence data, but a subsequent study using this gene and four others found that Malva, at least, was polyphyletic (García et al. 2009). There generic limits had in the past been based mainly on the number of parts of the epicalyx and their fusion, but fruit characters seemed to be more useful features to use when characterising clades. For relationships within Hibisceae, see Pfeil et al. (2002), Pfeil and Crisp (2005) and Koopman and Baum (2008: Malagasy taxa); generic limits around Hibiscus are especially difficult - s. str. or s. lato? Diversification and molecular evolution pick up almost together in Malvoideae (Baum et al. 2002, 2004).

For a discussion of groupings in the extended family, Robert Brown's comments over 150 years ago (Brown 1814) on family limits in the Malvales (= Malvaceae here) are a good starting point. Malvaceae + Bombacaceae + Sterculiaceae + Tiliaceae make a well circumscribed group, but apart from Malvaceae, all the other families in the old Malvales are highly para- or polyphyletic. The clades within it are mostly difficult to distinguish, even with flowers, so combination seems sensible (e.g. Judd & Manchester 1997; Alverson et al. 1999; Bayer et al. 1999); Cheek (2007), however, opts for dismemberment into ten families which would seem to be hard to characterize.

For more information, see van Heel (1966: androecial morphology and development), Jenny (1988: gynoecial diversity in "Sterculiaceae"), Manchester and Miller (1978: wood anatomy), while for general information, see the Malvaceae Pages website and Bayer and Kubitzki (2002).

[SANTALALES [BERBERIDOPSIDALES [CARYOPHYLLALES + ASTERIDAE]]] / ASTERIDS ET AL. / SUPERASTERIDS : ?

SANTALALES Dumortier - [Back to Index]

Mycorrhizae absent; polyacetylenes [triglycerides with C18 acetylenic acids]; pericyclic fibres 0; petiole/mesophyll with astrosclereids; lamina entire; inflorescences cymose; K often reduced, C valvate, A opposite petals, anthers basifixed, pollen colpate or porate, disc +, 1 pendulous tenuinucellate ovule/carpel, style single; fruit a drupe, 1-seeded, K persistent; seed coat crushed; endosperm cellular, chalazal haustorium +, embryo small/minute. - 8 families, 151 genera, 1985 species.

Caterpillars of some Pieridae-Pierinae (ca 440 species recorded, but in only 9+ genera; 2/5 of all host-plant records) and Lycaenidae-Lycaeninae-Iolaini in particular are found on members of this order, Loranthaceae (especially), "Olacaceae" (= Ximenia, Olax, Lycaeninae), Opiliaceae and Santalaceae being recorded as hosts (Ehrlich & Raven 1964; Fiedler 1991, 1994; Congdon & Bampton 2000). The ancestors of the pierine butterflies seem for the most part to have eaten members of Brassicales and their initial santalalean hosts may have been Loranthaceae; there are no reports of caterpillars on free-living Santalales (Braby 2005; Braby & Trueman 2006: I thank M. F. Braby for information). Interestingly, a number of the adults of these butterflies have warning colouration on the undersides of their wings, and some caterpillars may also have warning coloration. However, it is unclear what compounds the insects might pick up from their santalalean hosts that would discomfit potential predators (Braby & Trueman 2006).

It is clear from the phylogeny of the order that the plesiomorphic condition in it is the free-living habit, and hemiparasitism by attachment to the roots of the host is derived (Malécot 2002; Malécot et al. 2004). Aerial hemiparasitism has been derived five times (Nickrent 2002), intermediate taxa perhaps being both root and stem parasites (Vidal-Russell & Nickrent 2008); photosynthesis in some aerial hemiparasites is largely in their stems. Aerial hemiparasites may have a single point of attachment to their host, or roots running over the surface of the bark may form both additional points of attachment and additional plantlets (as in Loranthaceae: Vidal-Russell & Nickrent 2006, 2008). Some Santalales are endophytes whether or not the part of the plant that finally appears is chlorophyllous (Santalaceae, e.g. Viscum), of the plant is a holoparasite (Balanophoraceae: see Nickrent et al. 2005). Some Loranthaceae are even hyperparasites, parasitizing other members of that family. For some information on the physiological details of parasitism in Santalales, see Stewart and Press (1990).

Although little is known about mycorrhizae in Santalales, the few taxa studied largely lack them (Landis et al. 2002), exceptions being Ongokea, Coula and Strombosia, scattered in ex-Olacaceae (Malécot 2002 for references). The absence of mycorrhizae and also of root hairs (?reference) which have been reported for the group may be connected with the adoption of the parasitic habit.

Malécot (2002) provides a phylogeny based on the analysis of variation in four genes that emphasises members of the old Olacaceae; he discusses variation of morphological characters in the context of this phylogeny and also provides analyses of a combined morphological-molecular data set, although there is still little well-supported structure along the spine of basal Santalales (= "Olacaceae" s.l.: see also Michaud 1966; Malécot et al. 2004, a morphological phylogenetic analysis). Malécot et al. (2004) found some support for Erythropalaceae, Ximenia plus some other genera, and most of the rest of the old Olacaceae (the latter a very weakly supported clade) as successively sister to the rest of Olacales in a morphological analysis. Malécot and Nickrent (2008) have recently shown that the old Olacaceae to form about eight clades basal to other Santalales, but relationships between these clades were unclear. Molecular data place Brachynema, a genus that is so morphologically distinctive that its inclusion in the order was in some doubt (e.g. see versions 8 of this site and earlier), close to Maburea, in Erythropalaceae s. str. (K. Wurdack, pers. comm.). Within the Loranthaceae clade, Misodendraceae may be sister to [Schoepfiaceae + Loranthaceae], especially in some analyses that include many taxa, although when the number of taxa is reduced they may be sister to Schoepfia in particular (Malécot 2002), and this relationship has also been confirmed by more recent studies (see especially Der & Nickrent 2008; Vidal-Russell & Nickrent 2008) and is followed here. Der and Nickrent (2008) show that Santalaceae are polyphyletic, a few genera being removed to Opiliaceae and Schoepfiaceae, while within Santalaceae there are eight well supported clades, relationships between which are not yet well understood. Finally, Mystropetalum, Dactylanthus, and Hachettia, three members of Balanophoraceae, form a clade with 100% posterior probability that was sister to a clade made up of Schoepfia, Dendrophthoe and Santalum (almost 100%), the combined group having 100% support (Nickrent et al. 2005), so Balanophoraceae are included in Santalales, but without any particular position. Su and Hu (2008) suggested that Balanophoraceae were basal or near basal (with respect to the other families) in the clade. Further work is needed to resolve details of all these relationships. Balanophoraceae are the only Santalales known to be holoparasitic, and being holoparasitic, they lack most or all of the distinctive vegetative features of other Santalales.

Santalales have often been compared with Icacinaceae (e.g. Takhtajan 1997), but there is little evidence suggesting such a relationship. Icacinaceae are scattered through asterids, being near Garryales (asterid 1) or in Aquifoliales or Apiales (asterid 2).

For general information, see Barlow (1997), Calder and Bernhardt (1983), and Reed (1955), for anatomy, see Baas et al. (1982), for hemi-parasitism, see Fineran (1991), for additional information on relationships, see Nickrent and Duff (1996), Nickrent et al. (1998), and for pretty much everything, see Parasitic Plants website (Nickrent 1998 onwards) and also Heide-Jørgensen (2008).

Includes Balanophoraceae, Erythropalaceae, Misodendraceae, "Olacaceae", Opiliaceae, Loranthaceae, Santalaceae, Schoepfiaceae.

ERYTHROPALACEAE Miquel, nom. cons. - 13/65. Pantropical, SE Asia, but not Madagascar or New Guinea and to the SE.

"Olacaceae" [[[Misodendraceae + Schoepfiaceae] Loranthaceae] [Opiliaceae + Santalaceae]]: root parasites; vessel elements with simple perforation plates; rays ³7 cells wide; nodes 1:1; stomata paracytic; silicification of some foliar mesophyll/epidermal cells.

It is difficult to assign change in ovary position to a particular place on the tree. "Olacaceae" have a superior ovary, and so do [Exocarpos + Omphalomeria] "a basal clade" in Santalaceae (Der & Nickrent 2005), and Opiliaceae (the latter has recently been expanded to include some Australian ex-Santalacaeae with superior ovaries - see Der & Nickrent 2005), Loranthaceae are inferior, Schoepfiaceae are half inferior. Either there are independent origins for the characters of inferior ovary, or reversals, or both. Ovule, embryo sac and embryo development of many Santalales, but probably not Erythropalaceae, is more or less remarkable. Distinct ovules may not be recognizable, the embryo sacs being borne in a spherical body, the mamelon; this consists of a basal placenta and ovular tissue, although neither can be recognized as such. Individual embryo sacs may elongate greatly and approach the apex of the mamelon or even the stigma at the end of the long style. This may be direct competition between female gametophytes for male gametes given that their normal spatial constraints - i.e., being enclosed individually in an organised ovule - are absent (Haig 1990). In any event, after fertilization the embryo is "planted" back down at the base of the mamelon by the development of a long suspensor (Johri et al. 1992; Shamrov et al. 2001, and references).

"OLACACEAE" R. Brown - 14/103. Pantropical (subtropics).

[[Misodendraceae + Schoepfiaceae] Loranthaceae] [Opiliaceae + Santalaceae]: ovary not septate, ovules ategmic; testa 0.

[Misodendraceae + Schoepfiaceae] Loranthaceae: petiole astrosclereids 0; K minute; G [3].

Misodendraceae + Schoepfiaceae: ?

MISODENDRACEAE J. Agardh - Stem parasites; stem apex aborting; plant dioecious, vegetative and reproductive shoots sharply distinguished; staminate flowers without a perianth; fruit a nut with three persistent plumose staminodes. - 1/8. Cool temperate South America.

• Misodendraceae are dioecious stem parasites (its hosts are Nothofagaceae) that are recognizable by their stems, which are often thick, even when young, "giving the leafy species a rather elephantine appearance". The leaves may be reduced, and those on flowering shoots are different from those on vegetative shoots; the two kinds of shoots are sharply distinguished. The flowers are small, and the small, nut-like fruits are very distinctive with their three, long, plumose attachments.

Misodendraceae may be ca 80 Ma old (Vidal-Russell & Nickrent 2008), diversification being dated to ca 75Ma before present, much before that in the more speciose branch-parasitic Loranthaceae (Vidal-Russell & Nickrent 2006, 2007, but c.f. 2008 - 27 Ma). Aerial parasitism may be the condition for the stem of this clade (Vidal-Russell & Nickrent 2008).

For general information, see Kuijt (1969), for seeds and seedlings, see Kuijt (1982: corrections to earlier work), for details of wood anatomy, see Carlquist (1985c), and for a phylogeny, see Vidal-Russell and Nickrent (2007).

SCHOEPFIACEAE Blume - Bract immediately below and surrounding G; flowers distylous, C tubular, with adaxial hairs. - 3/55: Quinchamalium (25). Central and South America, a few species in tropical South East Asia-Malesia.

That Schoepfia is rather different from "Olacaceae" has often been remarked (e.g. Metcalfe & Chalk 1950; Sleumer 1984). There are prominent bracteoles immediately below its flowers which look not unlike those of Loranthaceae, and the calyx and pollen of Schoepfia are also similar to those of Loranthaceae. However, pollen aperture development in Schoepfia follows Garside's Rule, in pollen grains with tetrahedral tetrads, these apertures are at points where three grains are in contact, i.e. there are four groups of three apertures (see Blackmore & Barnes 1995). Arjona, ex Santalaceae, is sister to Schoepfia (Malécot 2002), and Quinchamalium, another ex Santalaceae, is also to be included in this clade (Vidal-Russell & Nickrent 2006; Der & Nickrent 2008). Der and Nickrent (2008) characterise the clade.

LORANTHACEAE Jussieu - Stomatal orientation transverse to long axis; collenchymatous zone below the embryo sacs; fruit with viscin; endosperm compound, i.e. formed from more than one initial triploid cell. - 68/950. World-wide.

1. Nuytsia - Fruit dry, winged; cotyledons 3-4, foliaceous. - 1/1: Nuytsia floribunda. S.W. Australia.

2. The Rest - Stem parasites, forming burl at point of attachment and with epicortical roots running over the surface and often forming secondary burls; embryo ± plug-shaped, no radicle. - 67/950: Tapinanthus (250), Psittacanthus (120), Amyema (95), Struthanthus (50), Phthirusa (40). World-wide.

• Loranthaceae are usually stem parasites that form secondary points of attachment along the stem. They are recognizable by their more or less opposite, often rather thick and brittle (especially when dry) leaves with entire margins and inconspicuous venation born on stems that are not jointed. Their 4-5-merous flowers are quite long, narrowly tubular and brightly colored - often red and/or orange - and lack much in the way of a calyx; the ovary is inferior and the fruits are fleshy. The flowers often have a bract/bracteoles at the base and immediately to one side of the ovary.

Dicaeidae, the passerine flowerpeckers, are involved in the pollination of Malesian Loranthaceae. In some taxa the flower opens only if it is pecked by the bird, and explosive opening of flowers is sporadic in the family. Birds are very common pollinators in the family, and the groups involved and the evolution of bird-pollinated flowers is discussed in detail by Vidal-Russell and Nickrent (2008b). How loranthaceous seeds are dispersed, whether after moving through the gut, or by regurgitation, varies; the behaviour of the disperser is sometimes modified in distinctive ways. Dicaediae are also involved in dispersal, the seeds passing very quickly through the gut and being deposited on a branch, the bird sometimes showing distinctive behaviour as it defecates, swinging its body parallel to the branch so the seed lands on the branch (Docters van Leeuwen 1954 in particular; Polhill & Wiens 1998 and references). The viscid covering of the embryo aids in its attachment to the branches; the fruits contain rubber and are used in bird lime. Loranthaceae are the major hosts for caterpillars of pierid and lycaenid butterflies in Santalales (see above for literature). Within Iolaini (Lycaenidae) caterpillar preferences show quite a good agreement with the classification of Pohill (1998), in particular, most Iolaini are found either on tapinanthoid or on taxilloid genera, the two main African groups of the family, but not on both (Congdon & Bampton 2000). Caterpillars of the pierid Delias occur on Malesian Loranthaceae (Docters van Leeuwen 1954).

The root parasitic habit, as in Nuytsia, is probably plesiomorphic in the family (Vidal-Russell & Nickrent 2005). It is also known from Atkinsonia (S.E. Australia) and Gaiodendron (Central and South America), other genera that may also be near the base of the phylogeny, hence assigning the stem parasitic habit to The Rest above may be overly optimistic. It is estimated that the stem/branch parasitic habit evolved ca 28-40Ma before present (Vidal-Russell & Nickrent 2006, 2008a), about when the family started diversifying (c.f. Misodendraceae), even although the clade itself is possibly much older, ca 81 Ma (Vidal-Russell & Nickrent 2008a). In such parasites, the host-parasite junction may be much swollen, producing wood roses in the host; wood roses are the broadened point of attachment of the parasite to the host, the vascular elements of the latter producing a radiating structure. Details of the morphological nature of the association between stem parasite and hosts varies, and the epicortical roots that run along the host's branches - which may be plesiomorphous in the aerial parasites - form either sympodial or monopodial systems (e.g. Polhill & Weins 1998; Calvin & Wilson 2006; Wilson & Calvin 2006b); Vidal-Russell and Nickrent (2008b) suggest that the aerial parasitic habitat arose once. Some Loranthaceae are hyperparasites, parasitizing other Loranthaceae; the endophytic Tristerix aphyllus can almost be considered holoparasitic, only the infloresence appearing on the cacti it inhabits (Amico et al. 2007). In Australia, the shapes of loranth leaves and of the eucalypt host on which they grow are similar; explanations for this vary and this phenomenon is at most uncommon elsewhere (Barlow & Wiens 1977).

Mitochondrial genes from a presumably root-parasitic member of Loranthaceae seem to have been acquired by the fern Botrychium, perhaps via a common mycorrhizal associate (Davis et al. 2005b). However, in general mycorrhizae are not very common in Santalales and if the gene transfer took place in Asia, as is suggested, the apparent absence of root-parasitic Loranthaceae from that area is notable - perhaps they have gone extinct...

The embryo sac in Moquiniella is some 48 mm long, the longest in the angiosperms; it grows up the style and then may grow back downwards after reaching the stigma (this is sometimes called an embryo sac haustorium - see Mikesell 1990), and other members of the family have embryo sacs nearly as long. In many Old World Loranthaceae the cotyledons are connate, but not basally; the plumule emerges through the basal slit between the unfused portions of the cotyledons (Kuijt 1969).

Loranthaceae and Viscaceae (the latter here included in Santalaceae) have often been considered to be close, even being put in a single family, but there are numerous features separating the two; Kuijt (1969), Raj (1970) and Polhill and Wiens (1998) provide useful tables of differences. Thus Viscaceae s. str. are usually monoecious plants, they lack the bract and bracteoles immediately under the ovary, the perianth is 3-4-merous and very short, the anthers open by pores, there are two "ovules" per ovary with bisporic, 8-celled (Allium-type) embryo sacs, the embryo suspensor is short, the viscous covering of the fruit is incomplete and is found inside the vascular bundles, being made up of polysaccharide threads and mucilage, and there is a thin endocarp and green endosperm. Loranthaceae have larger, usually perfect flowers subtended by a bract and bracteoles, the anthers open by slits, there are more "ovules", the embryo sac being the common Polygonum type, unisporic and eight-nucleate, the fruits contain rubber which is outside the vascular bundles, and the endosperm is white. Although Loranthaceae such as Phthirusa have very small flowers and congested inflorescences, they may still readily be separated from Viscum and its relatives in the field. Plants of the former are often darker green in color, they have roots running over the surface of the host, the stems are not articulated, the flowers are often five-merous, and the endosperm is white, not green, etc.

For embryology, etc., see Raj (1970) and Johri and Bhatnagar (1972), for stomata, see Butterfass (1987), for seedlings of Neotropical Loranthaceae, see Kuijt (1982), and for foliar anatomy, see Kuijt and Lye (2005: terminal tracheids and cristarque cells common).

Opiliaceae + Santalaceae: P as a single whorl only.

The single perianth whorl is probably equivalent to the corolla of other members of the order, where the calyx is often small; Hiepko (1984) called this single whorl the perianth in Opiliaceae only because there was no obvious calyx.

OPILIACEAE Valeton - Ovary superior. - ? Pantropical.

• Opiliaceae may be recognised by their leaves, which are often two-ranked, lack much in the way of venation, and may have small bumps because of the cystoliths. The flowers are small and are borne in inflorescences which are often almost catkin-like when young; there is only one perianth whorl. The ovary is superior.

The Australian Anthobolus, ex Santalaceae, may belong here (Der & Nickrent 2005, 2008); but other Anthoboleae are close to Santalum in Santalaceae. The ovary of Opiliaceae may be secondarily superior.

For embryology, see Ram (1970).

SANTALACEAE R. Brown - Cuticular epithelium +; stomatal orientation transverse to long axis. - 41/910: Thesium (325), Phoradendron (235 - see Kuijt 2003), Viscum (65), Dendrophthora (65), Exocarpus (26), Santalum (sandalwood: see Harbaugh & Badlwin 2007, perhaps two migrations from Australia → Hawai'i,also two migrations out of Hawaii). World-wide, esp. tropics.

The mistletoe Viscum minimum, from the eastern Cape, has aerial stems about 3 mm long only and with but a single internode (Don Kirkup, pers. comm.); these arise from the endophytic portion of the plant. Viscum crassulae is a pleasing horticultural subject with its bright red fruits when growing on Portulacaria afra, its preferred host and whose leaves its leaves are supposed to mimic; it must be about the only stem parasite so grown. It is also an indicator of elephant grazing; since it is succulent and nutrient-rich, the animals preferentially eat it (Germishuizen et al. 2007, and references). Some mistletoes like species of Arceuthobium (e.g. A. americanum - the genus used to be placed in a separate family, Viscaceae) are major pests on conifers, causing extensive witches' brooms and the death of the host. The seeds take two years to mature, the embryo lacks a plumule but has a well developed radicle, and seed discharge is explosive (Hawksworth & Wiens 1996; Ross Friedman & Sumner 2009 for embryo development). As Der and Nickrent (2008) and Vidal-Russell and Nickrent (2008) note, the branch-parasitic habit seems to have evolved three times in Santalaceae, in the Viscum, Eremolepis and Dendromyza groups - the latter two are in the Santalum and Amphorogyne clades respectively. The Viscum clade itself may be some 72 Ma old, perhaps being branch parasitic for much of that time; molecular evolution has greatly speeded up (Vidal-Russell & Nickrent 2008).

Some species of Exocarpus lack expanded leaves; their branches are flattened and photosynthetic. When a cuticular epithelium develops, there is no cork cambium. Subepidermal cells divide irregularly, the outer walls being cutinised; the cuticular epithelium lacks both suberin (c.f. cork) and lenticels (Wilson & Calvin 2003). It is present in the old Eremolepidaceae and Viscaceae, as well as some, but possibly not all Santalaceae s. str. Stomata in the stem and of leaf are commonly tranversely oriented (Butterfass 1987). Exocarpus and some other genera have a superior ovary.

Der and Nickrent (2008) found eight well supported clades in Santalaceae, but relationships between these clades are poorly understood; there may be a clade including most of the family except the Cervantesia, Thesium and Comandra clades. Der and Nickrent (2005) proposed to recognise all major clades as families, and it would be necessary to recognise eight families here if Santalaceae were maintained as a family. The two genera of Eremolepidaceae, very often kept separate before, that were included in this study are well embedded in the Santalum clade (Der & Nickrent 2008).

For more information, see Wiens and Barlow (1971: cytology), Hawksworth and Wiens (1972: Arceuthobium), Leins (2000: floral morphology of Viscum), and Wilson and Calvin (2003: anatomy).

BALANOPHORACEAE L. C. & A. Richard - Echlorophyllous root parasites from underground tuber-like structures, these rupture and leave a collar-like structure at ground level; roots 0; stems endogenous; inflorescence racemose, terminal; flowers small; anthers extrorse, usu. connate; carpelate flowers: P 0 or minute; embryo small to undifferentiated. - 7/50: Balanophora (15). Mostly tropical.

• Balanophoraceae are echlorophyllous yellow to red or brown root parasites whose aerial stem arises from a collar-like structure borne at ground level. The dense inflorescence is racemose, bearing many minute flowers.

It is very difficult to understand the morphology of the flower, whether the ovary is superior or inferior, etc., in many Balanophoraceae, indeed, development of the gynoecium of Balanophora is perhaps best compared with that of the nucellus, hence the morphology of the "fruit" is unclear. For cautionary comments on attempts to determine ovule type and orientation in the family, see Holzappel (2001) - but problems are hardly surprising, perhaps, if Balanophoraceae are to be compared with the hemiparasitic Santalales. Dactylanthus lacks any seed coat, and one of the cells produced by the first division of the endosperm never divides further - this is perhaps to be compared with the chalazal haustorium in other Santalales. In Helosis the inner layer of cells of the seed coat is massively thickened on the inner and anticlinal walls, while in Balanophora it is the outer layer that is thickened; the seed coat of taxa like Hachettia is more complex.

Takhtajan (1997) linked Balanophoraceae with Cynomoriaceae (here Saxifragales), Rafflesiales (here Malpighiales) and Hydnoraceae (here Piperales), including them all in his Rafflesianae, made up entirely of parasitic plants; within Balanophorales he included seven other families here all included in Balanophoraceae.

For general information, see Takhtajan (1988, 1997); see also Kuijt (1969).

[BERBERIDOPSIDALES [CARYOPHYLLALES + ASTERIDAE]]: ?

BERBERIDOPSIDALES Doweld - [Back to Index]

Crystals +; stomata cyclocytic; filaments stout; style +; seed endotestal; endosperm development?, embryo? - 2 families, 3 genera, 4 species.

There is good support for this clade in a three-gene tree (e.g. D. Soltis et al. 1999). For possible synapomorphies in wood anatomy, see Carlquist (2003b).

Includes Aextoxicaceae, Berberidopsidaceae.

AEXTOXICACEAE Engler & Gilg - Pith heterogeneous; peltate scales +; bracteoles enveloping flower; K thin, deciduous; A 5, opposite sepals, nectaries staminodial; fruit a drupe; seeds ruminate; embryo curved, transverse. - 1/1: Aextoxicon punctatum. Chile.

• Aextoxicaceae are dioecious evergreen trees that can be recognised by their opposite, entire exstipulate leaves that are covered with peltate scales and by their pendulous, racemose inflorescence. The petioles are almost pulvinate at the apex and the base and the lamina appears to be minutely peltate. The flower buds are covered by bracteoles and when they fall off, the almost scale-like sepals fall off with them.

Sclereids are found in all vegetative parts of the plant; those of the leaf blade are about half the thickness of the blade in length. The pith is notably heterogeneous with large and small cells sharply distinguished. Although the stigma is bilobed, the bicarpellary construction of the gynoecium should be confirmed; the endocarp appears to split particularly readily along two vertical lines.

Aextoxicaceae were included in Euphorbiales (Takhtajan 1997), and have also been linked with Saxifragales (Qiu et al. 1998).

For a general account, see Kubitzki (2006b).

BERBERIDOPSIDACEAE Takhtajan - Filaments short, anthers inserted along connective, connective with apical prolongation, placentation parietal; fruit a berry. - 2/3. Chile, E. Australia.

• Berberidopsidaceae are woody scramblers that can be recognised by their exstipulate, sometimes spiny-toothed leaf blades with palmate venation; the veins run straight into the spines, when present. The fruit is a berry crowned by the persistent style base.

There are only two genera in the family, but they are morphologically very different. Berberidopsis has more or less spirally-arranged floral parts, no clear distinction between sepals and petals, many stamens, and a nectary; Streptothamnus has whorled parts, distinct sepals and petals, few stamens, and the nectary is absent, or perhaps to be found between the stamens and the gynoecium. The pollen is tricolpate or tricolporate.

For the floral development of Berberidopsis corallina, a "link" in the evolution of the flower of core eudicots, see Ronse De Craene (2004, 2007); the link is at best a parallelism, or even a connection in the reverse direction.

Berberidopsidaceae were included in Flacourtiaceae by Cronquist (1981) and in Violales by Takhtajan (1997).

For other information, see van Heel (1979, 1984: seed, pollen), Baas (1984: anatomy), Takhtajan (1992: seed), and Kubitzki (2006b: general).

[CARYOPHYLLALES + ASTERIDAE]: embryo long.

CARYOPHYLLALES Perleb - [Back to Index]

Not mycorrhizal; leaves entire; pollen colpate, tectum spinulose, styles/style branches long; fruit a loculicidal capsule. - 33 families, 692 genera, 11155 species.

Caryophyllales contain ca 6.3% of eudicot diversity (Magallón et al. 1999) and may date to the Albian, 111-104Ma before present, although Rhabdodendraceae do not split off until 90-83Ma before present (Wikström et al. 2001); Anderson et al. (2005) suggests figures of 116-114Ma before present for stem group Caryophyllales, 102-99Ma before present for the crown group. Some chrysomelid beetles - Alticinae, Cassidinae+Hispinae - seem notably more common on this clade than others (Jolivet & Hawkeswood 1995).

Given the variation in carpel number in the clade, it is with some hesitation that three carpels is suggested as the plesiomorphic condition. There are many other unusual characters here, but their phylogenetic significance is unclear, partly because of sampling problems, e.g., Rhabdodendraceae, Simmondsiaceae and Physenaceae of the basal pectinations of the core Caryophyllales clade are poorly known. Many families are tolerant of saline/arid conditions and/or have a distinctive habit (e.g. climbers with grapnel organs) or physiology (carnivory, C₄ pathway, CAM) or both. It is interesting that Landis et al. (2002) found that members of the two major clades, the Droseraceae-Polygonaceae and core Caryophyllales clades, commonly lack mycorrhizae, although there are some exceptions (e.g. some Polygonaceae, Nyctaginaceae and Amaranthaceae). Flavonol sulphates occur in Plumbaginaceae, Polygonaceae, and Amaranthaceae (Chenopodiaceae s. str.), and sulphated betalains in Phytolaccaceae; this is perhaps associated with the distinctive physiology/habitat preferences of these plants. Anomalous secondary thickening - separate cambia initiated successively - is widespread. The outer stamens are often initiated in pairs, especially in core Caryophyllales, but also elsewhere in the order (Ronse Decraene & Smets 1993); petal plus adjacent antepetalous stamen form developmental units in Plumbaginaceae and Caryophyllaceae (Friedrich 1956; Ronse Decraene et al. 1998). Carpels that are open in development are known both from Polygonaceae and core Caryophyllales (Tucker & Kantz 2001). Purple-spored smuts and Uromyces rusts parasitize several families, including Plumbaginaceae, Polygonaceae and core Caryophyllales, a strong sign that the groups were close according to Savile (1979b).

Rhabdodendraceae are sister to all other Caryophyllales in the rbcL analysis of Fay et al. (1997b) and in the Bayesian analysis of Soltis et al. (2007a). Cuénoud et al. (2002) found that Simmondsia was grouped with Rhabdodendron in a matK analysis, but with only weak support, but in two- and four-gene analyses (with poorer sampling) it was associated with core Caryophyllales; in trees shown by Drysdale et al. (2007) and Brockington et al. (2007) a position of Rhabdodendron as sister to the rest of core Caryophyllales was found in most analyses, and is adopted here (cf also Hilu et al. 2003: matK analysis alone). Asteropeiaceae and Physenaceae form a well supported pair, in turn showing a well-supported sister group relationship to core Caryophyllales (Källersjö et al. 1998; see also D. Soltis et al. 2000). The other main clade includes families like Plumbaginaceae, Polygonaceae, Nepenthaceae, and Droseraceae and is well supported (Morton et al. 1997b). It includes a total of four carnivorous families (see also Albert et al. 1992; Meimberg et al. 2000; Cuénoud et al. 2002; Cameron et al. 2002) and also other families with distinctive vegetative morphologies (see also Heubl et al. 2006). The tree below is based largely on those presented by Meimberg et al. (2000), Cameron et al. (2002: 4 genes) and Cuénoud et al. (2002: matK alone), for the position of Achatocarpaceae, see Müller and Borsch (2005).

For wood anatomy, see Carlquist (2002b, 2003a), for variation in nectary morphology, which is considerable, see Bernadello (2007), and for a summary of variation in the order, see Cuénoud (2002a, b).

Includes Achatocarpaceae, Aizoaceae, Amaranthaceae, Anacampseros, etc., Ancistrocladaceae, Asteropeiaceae, Barbeuiaceae, Basellaceae, Cactaceae, Caryophyllaceae, Didiereaceae, Dioncophyllaceae, Droseraceae, Drosophyllaceae, Frankeniaceae, Gisekiaceae, Halophytaceae, Limeaceae, Lophiocarpaceae, Molluginaceae, Montiaceae, Nepenthaceae, Nyctaginaceae, Physenaceae, Phytolaccaceae, Plumbaginaceae, Polygonaceae, Portulacaceae, Rhabdodendraceae, Sarcobataceae, Simmondsiaceae, Stegnospermataceae, Talinaceae, Tamaricaceae.

Droseraceae group [Tamaricaceae group + Polygonaceae group]: acetogenic naphthoquinones [plumbagin and related compounds] +; endosperm starchy.

It is unclear where the character of starchy endosperm is to be put on the tree. The condition is not known for the clades just below core Caryophyllales, Rhabdodendraceae, [Physenaceae + Asteropeiaceae] and Simmondsiaceae, and while Netolitzky (1926) noted that the core Caryophyllales lack starchy endosperm, Narayana and Lodha (1963) recorded starch in the young endosperm of Orygia (Amaranthaceae) and Corbichonia (Lophiocarpaceae). In the Flora of China, several families of core Caryophyllales are reported to have starchy endosperm (e.g. Dequan & Gilbert 2003), but this must be a mistake; starchy perisperm must have been intended.

Droseraceae [Nepenthaceae [Drosophyllaceae [Ancistrocladaceae + Dioncophyllaceae]]]: plants carnivorous; vascularized multicellular glands +; inflorescence cymose; C contorted, anthers extrorse, ovary unilocular.

Meimberg et al. (2000) and Cameron et al. (2002) discuss relationships in this clade; the acquisition of carnivory may have happened more than once, or it happened once and then was lost (perhaps more likely given the topologies found). For a synapomorphy scheme for the whole group, see Albert and Stevenson (1996), Meimberg et al. (2000: the floral characters listed are mostly plesiomorphies), but especially Heubl et al. (2006). The latter suggest that fly-paper traps are the plesiomorphic condition for the group, but note that where features like this or the possession of spirally coiled (circinate) leaves and pollen tetrads are placed on the tree will depend on how characters are optimised.

For acetogenic quinones and alkaloids, see Bringmann and Pokorny (1995) and Bentley (1998). For general information on carnivory, see Lloyd (1942), Juniper et al. (1989), and the Carnivorous Plants Database.

DROSERACEAE Salisbury - Herbs; leaves adaxially circinate, with glandular hairs, sensitive; pollen in tetrads, porate, placentation parietal. - 3/115: Drosera (110). World-wide.

• Droseraceae are insectivorous herbs usually with adaxially coiled or folded leaves and moderate-sized flowers; there are three carpels usually terminated by a three-branched style.

Evolution: Several episodes of dispersal have to be invoked to explain the current distribution of Drosera (Rivadavia et al. 2003). Indeed, some species disperse readily, and may even be found forming large patches on otherwise almost bare scoria on the flanks of a volcano (the Father: New Britain). Drosera is exceptionally diverse in S.W. Australia, which has about one third of the species in the whole genus; diversification may be linked to the onset of the Mediterranean climate there some 15-10Ma before present (). Species of Drosera are not only rosette formers; they may be climbers, and/or have stolons, tubers, or other perennating structures.

The free-floating aquatic Aldrovandra and he terrestrial Dionaea both have snap-traps, multicellular trigger hairs, etc. (Cameron et al. 2002; Rivadavia et al. 2003). The traps (leaf blades) of Dionaea close in about 100 ms, the movement being aided by the leaf blades changing from concave to convex (Forterre et al. 2005); Volkov et al. (2008 and references) provide physiological details of the mechanisms involved. Digestion may take up to a week, and then the trap reopens. In Drosera, insects become entrapped by the stalked glands, and the leaf blade slowly closes around the animals; digestion is carried out by enzymes produced by sessile glands (Juniper et al. 1989).

Phylogeny. Aldrovandra and Dionaea (both n = 6) may be sister taxa; both have snap-traps (Cameron et al. 2002; Rivadavia et al. 2003).

Chemistry, Morphology, etc.. Metcalfe and Chalk (1950) describe a distinctive vasculature in the inflorescence axis (a ring of centric vascular bundles) and petiole (two centric vascular bundles) of Drosophila binata.

See Boesewinkel (1989) for ovule and seed anatomy, Cutler and Gregory (1998) for general anatomy, Conran et al. (2007) for gland morphology, and Kubitzki (2002d) for general information.

Phylogeny. The free-floating aquatic Aldrovandra and Dionaea (both n = 6) may be sister taxa; both have snap-traps, multicellular trigger hairs, etc. (Cameron et al. 2002; Rivadavia et al. 2003). The traps (leaf blades) of Dionaea close in about 100 ms, the movement being aided by the leaf blades changing from concave to convex (Forterre et al. 2005); Volkov et al. (2008 and references) provide physiological details of the mechanisms involved. Digestion may take up to a week, and then the trap reopens. In Drosera, insects become entrapped by the stalked glands, and the leaf blade slowly closes around the animals; digestion is carried out by enzymes produced by sessile glands (Juniper et al. 1989).

Nepenthaceae [Drosophyllaceae [Ancistrocladaceae + Dioncophyllaceae]]: cortical bundles in stem [position unclear]; anthers basifixed.

NEPENTHACEAE Berchtold & J. Presl - Climbers; large spirally-thickened cells in pith, cortex, etc.; leaves with foliaceous basal part, a narrow twining portion, and terminal pitcher covered by a lid; plant dioecious; inflorescence racemose; P uniseriate, with nectariferous glands adaxially; A connate into a central column, pollen in tetrads, placentation axile, style short, stigma broad. - 1/90. Madagascar to New Caledonia.

• Nepenthaceae are insectivorous plants that are easy to recognise because of the lidded pitchers borne on the end of a twining prolongation of the leaf. The leaf base itself is broad, further widening to form a laminar portion that then narrows to form the tendrillar portion. The plants are dioecious, the inflorescences are racemes, the flowers are rather inconspicuous, and the seeds are very small.

The expanded part of leaf is developed from the leaf base, as in many monocots, the pitcher from the rest. How insects are trapped in the pitchers has long been unclear. Recent work suggests that the rim (peristome) of the pitcher is extremely wettable, and insects may aquaplane when they step on it, falling in to the pitcher below where they die and get digested; only when the rim is dry can insects walk on it easily. Nectar secreted by nectaries in the ridges of the mouth of the pitcher is hygroscopic and under suitable conditions takes up water, increasing the possibility of insect capture. In some species insects become trapped when they walk on to the wax-covered inner pitcher walls, slipping on the wax crystals and falling in to the pitcher (Bohn & Federle 2004; Bauer et al. 2008). In yet other species, including Nepenthes rafflesiana, the fluid in the pitcher is viscoelastic, even when much diluted, and insects that fall in drown very quickly (Gaume & Forterre 2007). Interestingly, the ant Camponotus schmitzi lives in close association with Nepenthes bicalcarata, and it can run across even the wetted rim. For the fauna of the pitchers, see Kitching (2000), while Pavlovic et al. (2007) discuss the physiology of lamina and trap.

For the biogeography and relationships of the group, see Meimberg and Heubl (2006). Some analyses suggest that the Malesian Nepenthes (including species from New Caledonia and Australia) are derived from a stock represented by the extant extra-Malesian taxa, but different relationships are suggested by different genes.

For general information, see Cheek and Jebb (2001: almost a monograph) and Kubitzki (2002d), and for anatomy, Pant and Bhatnagar (1977).

Drosophyllaceae [Ancistrocladaceae + Dioncophyllaceae]: ?

DROSOPHYLLACEAE Chrtek, Slavíkovà & Studnicka - Leaves abaxially circinate, with glandular hairs; pollen polyporate; placentation basal. - 1/1: Drosophyllum lusitanicum. Iberian Peninsula, Morocco.

• Drosophyllaceae are subwoody carnivorous plants that may be recognised by their long, abaxially coiled leaves that have sticky, glandular hairs in long lines on their abaxial surfaces; these hairs are not sensitive. The flowers are relatively large and short-lived (they remain open only a single day), but the petals dry and persist around the capsule.

Stem/leaf anatomy would repay investigation; both the cortical and petiole bundles appear to be inverted (Metcalfe & Chalk 1950, as Droseraceae).

Although superficially similar to Droseraceae, with which it has often been included in the past - but note that the curvature of the leaf is in the opposite direction and the lamina is not sensitive - Drosophyllaceae are sister to Dioncophyllaceae + Ancistrocladaceae, with good support, in an analysis of matK sequences (Cuénoud et al. 2002: the position of Nepenthaceae is here uncertain). Drosophyllum lusitanicum grows in rather exposed, calcareous places, unlike the habitats where Droseraceae are to be found.

For information, see Boesewinkel (1989: ovule and seed), and Kubitzki (2002d; general).

Ancistrocladaceae + Dioncophyllaceae: lianes; cortical bundles 0; petiole with sclerenchyma bundles; stomata actinocyclic.

ANCISTROCLADACEAE Walpers - Not carnivorous; modified branch systems with grapnels in series; lamina with glands in pits on abaxial surface; ovary half or more inferior, with 1 basal ovule; fruit a nut, K much enlarged; seed ruminate. - 1/12. Tropical Africa to W. Borneo and Formosa.

• Ancistrocladaceae are lianes that can be recognised by their shortly petiolate, extsipulate, entire leaves with glands in pits on the abaxial surface, and the complex, branched stem grapnels that are often borne opposite the leaves.

In the past Ancistrocladaceae (and Dioncophyllaceae) have often been included in Theales or Theanae (e.g. Cronquist 1981; Takhtajan 1997).

For general information, see Porembski (2002).

DIONCOPHYLLACEAE Airy Shaw - Plant with rosette leaves first, venation closely parallel, then leaves with a short blade and glandular hairs on the abaxial surface of the prolonged midrib, then leaves with paired grapnel hooks at the apex; placentation parietal; fruits open early; seeds flattened, on elongated funicles. - 3/3. Tropical West Africa.

• Dioncophyllaceae are distinctive in their rather long rosette leaves with closely parallel venation. These are followed by leaves with a short blade and glandular hairs on the abaxial surface of the prolonged midrib, and then, in the climbing phase, by leaves with paired grapnel hooks at the apex. The fruits open early and expose the still-developing flattened seeds.

For morphology, see Gottwald and Parameswaran (1968) and Schmid (1964), for chemistry, Spencer and Siegler (1985), for carnivory, Bringmann et al. (2001), and for general information, Porembski and Barthlott (2002).

[Frankeniaceae + Tamaricaceae] [Plumbaginaceae + Polygonaceae]: sulphated flavonols, ellagic acid +; seed exotestal.

Frankeniaceae, Tamaricaceae and Plumbaginaceae all have flat, multicellular glands of subepidermal origin (Conran et al. 2007). This is perhaps an apomorphy here (or still higher), with a loss in Polygonaceae.

Frankeniaceae + Tamaricaceae (= Frankeniaceae group): halophytic; leaves small, with salt-excreting glands; flowers small, 4-6-merous, C with basal adaxial appendages, pollen not spinulose, placentation more or less parietal; exotestal cells bulging or as hairs; endosperm +.

For the phylogeny of the two families, see Gaskin et al. (2004). It is equally parsimonious to assume that petal appendages are apomorphies for the family pair and are subsequently lost as it is to assume that they have evolved independently; in Tamaricaceae the Reamuria clade, members of which have these appendages, is sister to the rest of the family. Seeds with copious endosperm have the same distribution.

FRANKENIACEAE Desvaux - Leaves opposite. - 1/90. ± World-wide in warm, dry areas, but scattered.

For information, see Walia and Kapil (1965: embryology), Whalen (1987: taxonomy Old World Frankenia), Olson et al. (2003: anatomy) and Kubitzki (2002d: general).

TAMARICACEAE Berchtold & J. Presl - Embryo sac tetrasporic, often sixteen-nucleate. - 5/90: Tamarix (55). Eurasia and Africa, esp. Mediterranean to Central Asia.

• Tamaricaceae are woody plants with small leaves and flowers; the style branches are long and are terminated by distinctive expanded stigmas. The capsular fruits release hairy seeds.

Relationships within the family are [[Holachna + Reamuria] [Myricaria + Tamarix]] (Gaskin et al. 2004).

See Czaja (1978: seed storage), Zhang et al. (2001: pollen) and Gaskin (2002: general).

Plumbaginaceae + Polygonaceae (= Polygonaceae group): O-methylflavonols, quinones +; successive cambia 0; cortical and/or medullary vascular bundles +; nodes 3:3; leaf bases broad; ovary with 1 loculus and 1 basal ovule; fruit surrounded by accrescent calyx which forms part of the dispersal unit.

Lycaeninae caterpillars are quite commonly found on this group, probably because of the polyphenolics in their leaves (Fielder 1995).

For sterol composition of Plumbaginaceae-Polygonaceae in comparison to that of core Caryophyllales, see Wolfe et al. (1989), for anatomy, see Carlquist and Boggs (1996).

PLUMBAGINACEAE Jussieu - stamens equal and opposite petals; G [5], ovule circinotropous, funicle long and curved, embryo sac tetrasporic, 4- or 8-nucleate. - 27/836. Predominantly Mediterranean to Central Asia, scattered elsewhere.

1. Plumbaginoideae Burnett - Stigma with receptive areas in bouquet-like aggregations along branch; capsule basally circumscissile. - 4/36. East Asia and Africa.

2. Staticoideae Kostel. - Plumbagin 0, betaalanine betaines +; leaves cartilaginous, with 5-10 marginal rows of whitish cells; C connate, stamens adnate to corolla.

2a. Aegilitideae Peng - Ellagic acid +; successive cambia +. - 1/2. Malesian mangroves.

2b. Staticeae Bartling - Betaalanine betaines +; leaves more or less clustered at base; inflorescence leaves reduced or absent; inflorescence axis channelled; K scarious. - 22/800: Limonium (350), Acantholimon (165), Armeria (90: half on the Iberian Peninsula). Mostly Irano-Turanian (Mediterranean), but also S. Africa, S. South America, and W. Australia.

• Plumbaginaceae are perennial herbs or shrubs, rarely climbers. They can be recognised by their leaves which have broad bases and are frequently only shortly petiolate and have entire margins, and their flowers which have a relatively large, persistent, and sometimes coloured calyx (which can be scarious in fruit) and antepetalous stamens. The single ovules borne at the ends of long, curved funicles are easily enough seen on dissection of the flower.

Members of the family prefer saline and sometimes rather dry conditions. In a number of Staticeae the calyx becomes scarious in fruit and helps in dispersal; in Plumbago the sticky calyx glands persist in fruit. Staticeae are most diverse in the area from the western Mediterranean (Limonium, etc., with hybridization and hundreds of microspecies, some apomictic) to Central Asia (Acantholimon, etc.), diversification in the family beginning perhaps as recently as 18-16 Ma (Lledó et al. 2005).

Lledó et al. (1998, 2001) suggest phylogenetic relationships within the group; these are consistent with the classification here. For a phylogeny focussing on Limonium, see Lledó et al. (2005).

Betaines are quaternary ammonium compounds that are involved in salt excretion; glycine betaine is apparently only scattered in the family (Rhodes & Hanson 1993). For wood anatomy, which may be paedomorphic, the family perhaps having a more or less herbaceous ancestry, see Carlquist and Boggs (1996).

Primulaceae used to be associated with Plumbaginaceae. Both have stamens opposite the petals, common petal-stamen primordia, a more or less connate corolla (the latter especially in Plumbaginaceae-Staticoideae), and placentation that could be considered similar, but the two are not close - for Primulaceae, see Ericales.

Baker (1948, 1953) discussed variation in floral morphology (pollen, stigmas, etc.), de Laet et al. (1995) floral development, and there is much general information in Kubitzki (1993b).

POLYGONACEAE Jussieu - Plumbagin 0; pits vestured; stipule sheathing the stem; P 5, ovule straight; achene angled. - 43/1100. World-wide.

1. Polygonoideae Eaton - 15/590: Polygonum (200+, sometimes split - as in Brandbyge 1993 - and then Persicaria 150, see also Kim & Donoghue 2008), Rumex (200), Calligonum (80). Especially (warm) temperate.

2. Eriogonoideae Arnott - Plant woody[?] - 28/520: Eriogonum (250), Coccoloba (120). Largely tropical, especially America, Eriogonum and relatives mainly W. North America.

• The flowers of Polygonaceae are rather small, all parts (bar the carpels) are more or less free, and the 5 or 6 tepals are persistent or enlarged in fruit; the achene itself is often angled. A sheathing stipule is common and the nodes of the herbaceous taxa in particular are more or less swollen. Eriogonum and relatives lack the distinctive stipule and have a more or less cymose and involucrate inflorescence.

Lycaena and Heliophorus (Lycaenini) are found on this family throughout its extratropical range (Ehrlich & Raven 1964). Eriogonum and relatives are very diverse in the drier regions of southwest North America (with links with southern South America), and may represent a relatively recent radiation (Sanchez & Kron 2008).

Some species of Rumex have an X-autosome balance system determining the 'sex' of the plant. Interestingly in view of the paucity of mycorrhizae in the order, endomycorrhizae are reported from Eriogonum and ectomycorrhizae from Coccoloba (Malloch et al. 1980).

The largely herbaceous Eriogonoideae s. str., i.e. Eriogonum and its immediate relatives, used to be kept separate from Polygonoideae, of variable habit, because the former lacked a sheathing stipule, their inflorescence was cymose and with an involucre, while the latter had a sheathing stipule and a racemose inflorescence that lacked an involucre. However, studies show a division of the family into two moderately supported clades, largely woody and largely herbaceous respectively. The former includes Eriogonoideae s. str., and this is derived from within the woody clade, while Coccolobeae appear to be paraphyletic (e.g. Cuénoud et al. 2002; Lamb Frye & Kron 2003); [Antigonon + Leptopus], both lianes, are sister to the rest of the woody clade (Sanchez & Kron 2008). Eriogonum itself is paraphyletic and includes taxa like Chorizanthe and Dedeckera (Sanchez & Kron 2006, 2008). Polygonoideae s. str. are more or less herbaceous. Generic limits around Polygonum are difficult, and the mostly viney Muehlenbeckia (nestled in Brunnichia) is also included in Polygonoideae (Schuster & Kron 2008).

Sieve tube plastids with protein fibres are reported from Triplarieae (Eriogonoideae) (Behnke 1999). The climber Antigonon has leaf tendrils and successive cambia. It has been suggested that the flowers of Polygonaceae are basically 3-merous. Flowers with five tepals were derived from those with six, perhaps by fusion of two of the members. Recent work, however, suggests that the basic condition for the family is to have five tepals (Lamb Frye & Kron 2003). The exact nature of the funicle is unclear; it might be a reduced basal placenta.

For information, see Haraldson (1978: general), Brandbyge (1993: general) and Carlquist (2003a: wood anatomy).

Rhabdodendraceae [Simmondsiaceae [[Asteropeiaceae + Physenaceae] [Caryophyllaceae etc., Nyctaginaceae etc., Cactaceae, etc.]]]: styles or style branches stigmatic/papillate their entire length, stigmas not expanded; endosperm slight.

The morphology, etc., of members of the basal pectinations of this clade seem rather different from those of the core Caryophyllales, however, the former have been relatively little studied. Single-seeded fruits could be an apomnorphy of the whole clade, becoming multi-seeded in core Caryophyllales; taxa with "long" anthers are also common in these basal pectinations.

RHABDODENDRACEAE Prance - Sieve tube plastids with protein crystalloids and starch; nodes multilacunar; secretory cavities with resin; fibre-like sclereids in the leaves; hairs peltate, fringed; leaves revolute; A many, filaments very short, anthers very long, pollen not spiny; G 1, with 1 (2) basal campylotropous ovules, style basal; fruit a drupe, pedicel apex swollen. - 1/3. Tropical South America.

• Rhabdodendraceae are evergreen trees that may be recognised by their fringed-peltate hairs, their rather large, entire and exstipulate leaves, flowers with stamens that have short filaments and long anthers, and a single carpel with a basal style. The fruit is distinctive, the drupes being surrounded by the persistent calyx and swollen pedicel apex.

I have not seen stipules (c.f. Prance 1972c), but the rather broad petiole base can be confused with them. The ovule is often described as being unitegmic (e.g. Nandi et al. 1998, following Puff & Weber 1976; but see Tobe & Raven 1989).

The position of Rhabdodendraceae has long been uncertain. They were placed in Rutales by Takhtajan (1997), but a position in this general area was suggested by Prance (1968).

For general information, see Prance (2002).

Simmondsiaceae [[Asteropeiaceae + Physenaceae] [Caryophyllaceae etc., Nyctaginaceae etc., Cactaceae, etc.]]: nodes 1:1; C 0.

SIMMONDSIACEAE van Tieghem - Petiole articulated; staminate plant: anthers extrorse, longer than connective; carpelate plant: flowers axillary, 1 apical pendulous ovule/carpel, styles papillate all around; fruit with persistent columella; seed 1, testa multiplicative, vascularized, exotestal cells palisade, walls thickened. - 1/1: Simmondsia chinensis. S.W. North America.

• Simmondsiaceae are evergreen shrubs that may be recognised by their opposite leaves that are clearly articulated near the stem and have flat, rather thick blades with entire margins. The staminate flowers are small and borne in terminal inflorescences while the carpelate flowers are single and axillary; the calyx is much enlarged in fruit.

For testa anatomy, etc., see Tobe et al. (1992), for general information, see Köhler (2002), and for wood anatomy, see Carlquist (2002b).

[Asteropeiaceae + Physenaceae] [Caryophyllaceae, Nyctaginaceae, Cactaceae, etc.]: ?

Asteropeiaceae + Physenaceae: successive cambia 0; fruit single-seeded.

Although there may be more than ten stamens in Physenaceae, at least, there appear to be basically twice as many stamens as perianth members. Morton et al. (1997b) discussed the affinities of these two families and Carlquist (2006) compared their wood anatomy.

ASTEROPEIACEAE Reveal & Hoogland - Pedicels with many bracteoles; corolla present, deciduous, stigma continuous across ovary; cotyledons spirally twisted. - 1/8. Madagascar.

• Asteropeiaceae are evergreen trees that may be recognised by their shortly petiolate leaves whose blades are entire, exstipulate and have rather indistinct venation. The flowers are small and in terminal panicles, the fruit has a single seed, and the calyx is much enlarged and spreading.

Asteropeiaceae were previously often included in Theaceae or Theales (Cronquist 1981; Takhtajan 1997), but are very different in wood anatomy (Baretta-Kuipers 1976); the rays are uniseriate.

For information, see Beauvisage (1920: anatomy), Schatz et al. (1999: revision), and Kubitzki (2002d: general).

PHYSENACEAE Takhtajan - Plant dioecious; anthers long; seed large, coat vascularized, 16-20 layers thick, walls not notably thickened. - 1/2. Madagascar.

• Physenaceae are rather undistinguished with their more or less two-ranked, entire, exstipulate and coriaceous leaves. The flowers lack a corolla, the stamens have a very thin filaments and long anthers, and the styles are long. The fruit is inflated and contains a single seed which apparently has a vascularized seed coat.

The petiole is often described as being articulated; it commonly breaks transversely above the base.

Physenaceae were included in Urticales by Cronquist (1981) and placed in a monotypic Physenales in Dilleniidae by Takhtajan (1997).

For information, see Dickison (2002: general).

Caryophyllaceae etc. [Stegnospermaceae [Limeaceae [Nyctaginaceae etc. + Cactaceae, etc.]]], "Core Caryophyllales": plant herbaceous; CAM [especially pervasive in succulents] and C₄ photosynthesis common; ferulic acid bound to unlignified cell walls which fluoresce under UV light; betalains [chromoalkaloids] +, tannins 0; sieve tube plastids with a ring of proteinaceous filaments and a central angular crystalloid; inflorescence cymose; nectaries on adaxial bases of stamens, ovary 1-locular, ovules campylotropous, (funicles long); seeds with exotestal and endotegmic cells thickened, bar-like thickenings on endotegmic cells; endosperm 0, perisperm +, starchy, embryo curved.  - [Back to Index]

Core Caryophyllales contain ca 5.3% of eudicot diversity (Magallón et al. 1999). Taxa growing in saline and/or dry conditions are noticeably well represented here, and taxa that can grow on gypsum (hydrous calcium sulphate) are scattered throughout the clade (Douglas & Manos 2007). Succulents are common, and many taxa have C₄ photosynthesis, CAM, their variants. Core Caryophyllales are little liked by butterfly caterpillars (Ehrlich & Raven 1964).

Stomata on the stem and/or leaf are often oriented transverse to the long axis of the organ (Butterfass 1987). Any "corolla" present, as in Caryophyllaceae, is usually described as being of staminal origin (e.g. Ronse Decraene & Smets 1993; Leins et al. 2001); it arises at the same time or after the androecium, not before it, the "petals" and stamen(s) opposite them may form a developmental unit, and the corona - in Lychnis viscaria, at least - arises from two bulges on the adaxial side of the corolla, perhaps equivalent to anther thecae. When the stamens are equal in number to the perianth members they are opposite to them. When there are many stamens the initial primordia either alternate with the perianth members (Aizoaceae), or continue the spiral of the tepals (Pereskia - Cactaceae: see Leins & Erbar 1994); development of the stamens is centrifugal. Infrafamilial variation in pollen aperture type and number is considerable, and polyporate grains occur in a number of taxa. Seeds of a number of taxa have an operculum, although not necessarily identical in morphology (Bittrich & Ihlenfeldt 1984).

Achatocarpaceae + Amaranthaceae + Caryophyllaceae form a moderately well supported clade, the rest of the core Caryophyllales another (Källersjö et al. 1998), however, although 13 families were included, sampling within them was poor (see also Savolainen et al. 2000a; D. Soltis et al. 2000). For relationships I largely follow those suggested by Cuénoud et al. (2002), which are mostly similar to those found by previous workers, and with rather better if still sketchy sampling. Cuénoud et al. (2002) found two quite well supported clades within core Caryophyllales (see tree). Aizoaceae are monophyletic, albeit with only slightly better than marginal (52%) support in an analysis of matK sequences, the only gene for which they had moderately good sampling. Gisekia and Sarcobatus tended to be in different positions in different analyses (c.f. Cuénoud et al. 2002). For relationships around Cactaceae and Portulacaceae, which have been much studied recently (e.g. Nyffeler 2006), see below.

The course of evolution of betalains and anomalous secondary thickening in this group has long been uncertain. However, given the current phylogeny it seems unlikely that the presence of anthocyanins is plesiomorphic, rather, it may represent a reversal (Cuénoud et al. 2002; Cuénoud 2002a; see also Clement & Mabry 1996), and "normal secondary thickening" may also be apomorphic.

Most of this group was included in the old Centrospermae (so named because of the basal or free-central placentation that is common in the clade) or Caryophyllales in the strict sense; however, Cactaceae were sometimes somewhat separated because of their distinctive growth habit.

For information, see Zandonella (1977: nectaries), Rutishauser (1981: "stipules" and similar structures), Wolfe et al. (1989) and Patterson & Zu (1990: sterols), Steglich and Strack (1990: betalains), Barthlott (1994: waxes), Behnke (1994a: sieve tube plastids, phytoferritin), Gibson (1994: stem anatomy), Hofmann (1994: floral morphology), Nowicke (1994: pollen), Werker (1997: seed coat), Jansen et al. (2000c: successive cambia), and Muhaidat et al. (2007: C₄ pathways).

Caryophyllaceae [Achatocarpaceae + Amaranthaceae] (= Caryophyllaceae etc.): ?

CARYOPHYLLACEAE Jussieu - Anthocyanins +, betalains 0; "C" present; endotegmic cells lacking bars. - 86/2200: Silene (720), Dianthus (300), Stellaria (175), Gypsophila (150), Arenaria (150: polyphyletic), Minuartia (120: polyphyletic), Paronychia (110), Cerastium (100), Acanthophyllum (75), Drynaria (50). Mostly temperate.

• The family can be recognised by its herbaceous habit, opposite, entire leaves that often have little in the way of a petiole and are joined by a line at the base; geotropic adjustments occur at the swollen nodes. The inflorescence is cymose and the flowers appear to have a clearly distinguishable calyx and corolla, the latter being bilobed and/or clawed; there are usually 5 or 10 stamens. The fruit is a capsule.

Silene and its relatives may be pollinated by moths which at the same time lay eggs on the ovary, rather like the yucca - moth association (Kephart et al. 2006 for literature, also other papers in that issue of the New Phytologist [169(4)]).

The old tripartite division of the family into Silenoideae, Alsinoidae and Paronychioideae is not confirmed by any recent work (e.g. Nepokroeff et al. 2002; Smissen at al. 2002; Fior et al. 2006). Paronychioideae (often with colpate pollen) are a basal grade, with Corrigioleae (Telephium, Corrigola) sister to the rest of the family. Dicheranthus, Polycarpon, etc. may be the next clade, Paronychia, etc., the next. Drymaria and Pycnophyllum, both morphologically distinctive taxa, may be sister. In the erstwhile Alsinoideae (polyporate pollen) the calyx is free and the corolla has ± open venation. Alsinoideae for the most part break down into two groups: one, including Cerastium, Stellaria, etc., has capsules with split valves, and the other including Minuartia, etc., is very diverse, but has capsules with entire valves; the corolla is often bilobed. For Moehringia, the evolution of its strophiole, and its allies, see Fior and Karis (2007 and references). The old Caryophylloideae (polyporate pollen) have a connate calyx and clawed corolla with more or less closed venation and adaxial appendages (ligules). Sileneae and Caryophylleae are perhaps monophyletic, and together are sister to or form a polytomy with part of Arenaria (in the old Alsinoideae: Nepokroeff et al. 2002; Fior et al. 2006); for Silene, perhaps not monophyletic, and its relatives, see Erixon and Oxelman (2008).

Caryophyllaceae and Amaranthaceae at first sight seem to be very different, but the calyx of some Caryophyllaceae can be very scarious, as in Brachystelma, while herbaceous Amaranthaceae also often have swollen nodes (and opposite leaves), so the gross morphological distinctions between the two families are less striking than you might at first think.

Weberling (1989 and references) discusses the placentation in Caryophyllaceae, which varies from axile to free central to the single, basal ovule of Uebelinia - which is getting close to a circinotropous basal ovule (c.f. Plumbaginaceae).

For information, see Bittrich (1993: general), Zibareva et al. (2003; phytoecdyosteroids), Jia et al. (2004: cyclopeptides), and Schweingruber (2007: stem anatomy).

Achatocarpaceae + Amaranthaceae: pollen porate.

ACHATOCARPACEAE Heimerl - Pollen 4-6-porate; fruit a berry. - 3/7. S.W. USA to South America.

This group has been little studied. Some information is taken from Bittrich (1993).

AMARANTHACEAE Jussieu - Sieve tube plastids lacking protein crystalloid; pollen polyporate. - 174/2050: Chenopodium (100: C. quinoa), Atriplex (300), Gomphrena (120), Salsola (100), Alternanthera (100), Iresine (80), Amaranthus (60), Celosia (45). ± World-wide, esp. warm and dry temperate and subtropics, esp. saline habitats.

• Amaranthaceae are more or less succulent herbs often with swollen nodes; they prefer dry and/or saline habitats. The small flowers often have 1-3 carpels and 1-2 basal ovules. The fruits either have a circumscissile capsule and a chartaceous to scarious perianth and bracts or are indehiscent and are closely invested by the more or less fleshy perianth and bracts.

There are several types of C₄ photosynthesis with ca 17 different kinds of leaf anatomy and probably 10+ independent acquistions of the C₄ photosynthetic pathway, or four within Amaranthaceae s. str. (Kadereit et al. 2003; Sage et al. 2007; for details of the evolution of enzymes involved in Alternanthera, see Gowick et al. 2006). In three Suaedeae s.l. all the different aspects of C₄ photosynthesis occurs within a single cell (i.e. there is no Kranz anatomy), and this has evolved independently at least twice (Kapralov et al. 2006). Partitioning of the organelles within the cell is maintained by the distinctive organization of the cytoskeleton (Chuong et al. 2006), although there can be plasticity induced by the light-environment (Lara et al. 2008). With some 800 species with C₄ photosynthesis, Amaranthaceae are by far the largest BLA group with this photosynthetic pathway. Many taxa are halophytes (Jacobs 2001; Sage 2002), being among the first angiosperm colonizers of tidal mudflats or sand, and because of phenotypic plasticity, polyploidy, morphological parallelism, etc., species limits in some of these are extremely difficult to establish (Kadereit et al. 2007).

Many taxa are halophytes (Jacobs 2001; Sage 2002), and in erstwhile members of Chenopodiaceae in particular the accrescent perianth may envelop the fruit, being variously winged or spiny (e.g. Cabrera et al. 2009). Although the family is apparently largely without mycorrhizae, vesicular-arbuscular mycorrhizae have been reported from chenopods in the Red Desert of Wyoming - but only on native taxa and under undisturbed conditions (Miller 1979).

Relationships within this rather large group still need much more study, but it seems best to combine the old Chenopodiaceae (cuticle waxes with platelets; bracts and P ± fleshy; fruit rarely circumscissile; 300 bp deletion in chloroplast DNA inverted repeat) with Amaranthaceae s. str. (cuticle waxes lacking platelets; scarious bracts and perianth, filaments often connate; fruit often circumscissile). However, a good phylogenetic study with full generic representation is badly needed. Cuénoud et al. (2002) found Amaranthaceae s. str. to be monophyletic, with very strong (97%) support, Chenopodiaceae were perhaps monophyletic, but the branch collapsed in a strict consensus tree; the sampling was moderately good, but only one gene - matK - was sequenced and analysed. In an extensive rbcL analysis, much of the old Chenopodiaceae were monophyletic, but with little bootstrap support, ditto the old Amaranthaceae (incl. Polycnemoideae), Betoideae were paraphyletic (G. Kadereit et al. 2003). Other studies suggest paraphyly of Chenopodiaceae and sometimes even potential polyphyly of Amaranthaceae (Pratt et al. 2001). In a recent analysis of matK/trnK sequences, Müller and Borsch (2005b; see also Müller & Borsch 2005c), Polycnemum and Nitrophila (100% support) were sister to the rest; they have ordinary secondary thickening, imperfect flowers, basally connate filaments, and unithecate anthers. The rest of the Amaranthaceae + Chenopodiaceae had <70% support (and still lower posterior probabilities), while Amaranthaceae s. str. had 100% support and Chenopodiaceae s. str. again <70% support, but this time 100% posterior probabilities. Some of the extreme halophytic genera are morphologically much modified, and generic limits are difficult; in the reduced perianth of the Australian Tecticornia (Salicornioideae) the odd member is abaxial (Shepherd et al. 2004, 2005a, esp. Shepherd & Wilson 2007, also nomenclatural changes). See also Schütze et al. (2003), G. Kadereit et al. (2005: Australian chenopods, 2006: Salicornioideae) and Akhani et al. (2007: Old World Salsoleae) for studies on groups of Chenopodiaceae sensu stricto.

For more information, see Beck et al. (1982: cortical vascular system), Robertson (1981: general), Kühn (1993), Townsend (1993: general), Blunden et al. (1999: betaine distribution, very common and widespread), Judd and Ferguson (1999: general), Rajput (2002: anatomy), Carlquist (2003c: anatomy), Shepherd et al. (2005b; seed anatomy), and Sukhorukov (2007: fruit wall anatomy).

Stegnospermataceae [Limeaceae [[Lophiocarpaceae [Barbeuiaceae [Aizoaceae [Gisekiaceae [Nyctaginaceae, etc.]]]]] [Molluginaceae, Cactaceae, etc.]]]: ?

STEGNOSPERMATACEAE Nakai - "C" (2-)5, 1 basal ovule/carpel; seeds arillate. - 1/3. Central America, the Antilles.

• Stegnospermaceae are more or less scandent or sprawling plants with rather fleshy leaves and racemose inflorescences that have apparently petal-like flowers with a more or less rotate corolla.

For information, see Bedell (1980: general), Horak (1981: secondary thickening), Narayana and Narayana (1986: embryology) and Rohwer (1993a: general).

Limeaceae [[Lophiocarpaceae [Barbeuiaceae [Aizoaceae [Gisekiaceae [Nyctaginaceae, etc.]]]]] + [Molluginaceae, Cactaceae, etc.]]: ?

Wide-band tracheids are scattered through this group, occurring especially in the very succulent members.

LIMEACEAE Shipunov - Anthocyanins +. - 2/23: Limeum (20). Southern Africa to Ethiopia, S. Asia, also Australia.

For information, see Hoffmann (1973: flower, growth), Behnke et al. (1983b: Macarthuria), Endress and Bittrich (1993: general, as Molluginaceae), and Hassan et al. (2005a: seed coat).

[[Lophiocarpaceae [Barbeuiaceae [Aizoaceae [Gisekiaceae [Nyctaginaceae, etc.]]]]] [Cactaceae, etc.]]: sieve tube plastids with globular crystalloids

Lophiocarpaceae [Barbeuiaceae [Aizoaceae [Sarcobataceae [Phytolaccaceae + Nyctaginaceae]]]] (= Nyctaginaceae etc.): ?

LOPHIOCARPACEAE Doweld & Reveal - 2/6. Africa, esp. SW Africa.

See Hoffmann (1977: general), Rohwer (1993a: general, under Phytolaccaceae) and Carlquist and Schneider (2000: anatomy).

Barbeuiaceae [Aizoaceae [Gisekiaceae [[Sarcobataceae + Phytolaccaceae + Nyctaginaceae]]]: ?

BARBEUIACEAE Nakai - Sieve tube plastids with polygonal crystalloids. - 1/1: Barbeuia madagascariensis. Madagascar.

See Hoffmann (1977: general), Rohwer (1993a: general, under Phytolaccaceae) and Carlquist and Schneider (2000: anatomy).

Aizoaceae [Sarcobataceae + Phytolaccaceae + Nyctaginaceae]: soluble oxalate accumulation; raphides +.

For soluble oxalate accumulation, see Zindler-Frank (1976).

AIZOACEAE Martynov - Wood rayless; epidermis with bladder-like cells; hypanthium +; P colored internally, often with subapical abaxial appendage ["horn"]; A centrifugal, many, primordia 5, nectary as ring, ovary septate. - 123/ca 2020. Mostly southern Africa, also Australia, etc., tropical and subtropical, arid.

1. Sesuvioideae Lindley - Stipules +; prophylls often prominent; inflorescence usu. distinct from vegetative plant; A primordia opposite perianth; capsule circumscissile; seeds arillate. - 5/36: Trianthema (17), Sesuvium (12). Tropics and Subtropics.

Aizooideae [Mesembryanthemoideae + Ruschioideae]: inflorescence not distinct from the rest of the plant; A primordia alternate with perianth; fruit a hygrochastic capsule.

2. Aizooideae - Bladder hairs with large terminal cell and multicellular stalk[?]; inflorescence leafy; seeds upright [?]. 7/135: Tetragonia (85). Drier parts of S. Africa, also Australia (Gunniopsis), few N. Africa and Asia Minor, N. America, etc.

Mesembryanthemoideae + Ruschioideae: leaves very succulent; P sepal-like, "petals" [staminodial] many; G more or less inferior, nectary interrupted.

3. Mesembryanthemoideae Ihlenfeldt, Schwantes & Straka - Cortical bundles +; nectary hollow or shell-shaped [coilomorphic], placentation axile; capsule with purely septal expanding keels. - 1/100. S. Africa, also a few species from W. South America, Australia, N. Africa, the Mediterranean and the Near East.

4. Ruschioideae Schwantes - Bladder-like epidermal cells 0; inflorescence often distinct; filaments papillate, nectaries crest-like [lophomorphic], placentation basal or parietal; capsule with expanding keels more or less restricted to valves. - Ca 110/ca 1765: Ruschia (290-350), Conophytum (87-290: see Opel 2005a for leaf anatomy, 2005b for a morphological analysis), Lampranthus (180-220), Delosperma (155-165), Phyllobolus (150), Drosanthemum (100-110), Psilocaulon (65), Antimima (6-60). Southern Africa, esp. the western coastal Succulent Karroo.

• Aizoaceae are more or less prostrate herbs that may be recognised by their fleshy, opposite leaves the bases of which usually surround the stem, and the bladder-like epidermal cells. The flowers either have five perianth members that are green outside and colored inside or five green perianth members and numerous linear "petals" (these are staminodial in origin). There are usually many stamens, a more or less inferior and septate ovary, and often a fruit that opens when wetted as specialised tissue swells.

Aizoaceae, in particular Mesembryanthemoideae and Ruschioideae, dominate much of the Succulent Karoo of southwestern Africa, making up more than 50% of the species and up to an astounding 90% of the biomass. Klak et al. (2004) suggest that the radiation in Ruschioideae in S.W. Africa, at least, is both recent (3.8-8.7Ma before present) and very fast, indeed, the "meganiche" dominated by the whole family there - arid winter rainfall area with moderate temperatures - is only some 5 Ma old (Ihlenfeldt 1994). Edaphic specialization - soils can vary considerably locally - may also be involved in the diversification of Aizoaceae (Ellis & Weis 2006).

Variation in growth characters - leaf size and shape, internode elongation, etc. - is considerable (Ihlenfeldt 1994). Although a distinction is sometimes made between plants with foliaceous bracts or bracteoles in which the infloresence is not distinct from the rest of the plant, and plants with smaller bracts and distinct inflorescences (e.g. Hartmann 1993), it is unclear to me what the real growth characters are and where they go on the tree. Plants in several genera consist of paired leaves whose surfaces are more or less flush with the surface of the ground and almost invisible in the stony habitats in which they grow, being greyish or brownish and looking like pebbles themselves - hence names like Lithops and "flowering stones". The exposed surfaces of the leaves sometimes have distinctive "windows", more or less irregular areas of green within grey "frames". These leaves are bracteoles, the flower is terminal, and shoots - consisting solely of paired prophylls or bracteoles - develop in the axils of the bacteoles so giving rise to the next flowering unit (Hartmann 2004, 2006 for a summary). An extreme is seen in some species of Conophytum the leaves of which are almost completely connate except for a slit across the top; out of this the flower, etc., appear. The bladder-like cells on the leaf surface ("idioblasts") may be involved with water uptake from dew or mist; other taxa may have massively-thickened outer walls that contain layers of calcium oxalate crystals (e.g. Ihlenfeldt & Hartmann 1982). In addition, individual cells may be variously papillate or the surface otherwise sculpted and/or with epicuticular waxes, the stomatal openings may be deeply sunken, etc. (e.g. Ihlenfeldt & Hartmann 1982; Hartmann 2002; Opel 2005a).

Hartmann (1988) described the intricate morphology of the capsules of Mesembryanthemoideae and Ruschioideae in particular. These are often hydrochastic, that is, they open only when wet; indeed, seed dispersal is by "jet action" that uses the kinetic energy of falling raindrops to eject the seeds (Parolin 2006). There is considerable variation in dispersal and establishment "strategies" in the family.

Studies of the perianth in Sesuvioideae show that the petal-like basal part is equivalent to a sheathing leaf base, and the apical "horn" the rest of the leaf, rather as in monocot leaf development; B floral genes were not expressed as might have been expected, although they were in the petal-like staminodes of Ruschioideae (Frohlich et al. 2007).

I follow Klak et al. (2003) for basic groupings in the family. For a phylogeny of Sesuvioideae, see Hassan et al. (2005b). Tetragonia is embedded within Aizooideae (Klak et al. 2003), however, i.a. it has rays in the wood and lacks the bands of fibres found in other Aizoaceae (Carlquist 2007). Within Ruschioideae, Apatesieae and Dorotheantheae are successively sister to the remainder of the subfamily, and they are not very speciose. The much more speciose core Ruschioideae, including Drosanthemeae and Ruschieae, have a crest-like nectary, hygrochastic capsules with a distinctive anatomy (for which, see Kurzweil 2006) that release only a few seeds at a time, wide band tracheids are common (absent in sister taxa), and cylindrical or trigonous and succulent (rather than more or less flattened) leaves (Klak et al. 2004; Chesselet et al. 2004) that are without the bladder-like epidermal cells of the rest of the family. They also have lost the chloroplast rpoC1 intron (Thiede et al. 2007) - c.f. Cactoideae!

Both species and generic limits are difficult. In the early twentieth century Mesembryanthemum included the whole of the Ruschioideae and Mesembryanthemoideae, and until recently the Mesembryanthemoideae, by far the smaller of the two subfamilies, was divided into numerous genera. However, Klak et al. (2007) in a comprehensive study of the subfamily, obtained quite detailed phylogenetic resolution within it. Mesembryanthemum itself, although quite a small genus, was polyphyletic, and any attempt to maintain current genera would have caused the recognition of numerous and often poorly characterised taxa, so only one genus, Mesembryanthemum, was recognised. Species limits, too, are difficult, partly because of the horticultural interest in variation in the small, succulent members of the family. Hammer in 1993 noted that there were then about 1,800 known populations of Conophytum (Ruschioideae) - for which there were 450 names.

Some information is taken from Schwantes (1957: esp. fruit dehiscence), Hofmann (1973: morphology), Hartmann (1993: general), Landrum (2001: wide band tracheids), Chesselet et al. (2002: nectaries, etc.) and Interactive Mesembs (general). The books edited by Hartmann (2001a, b) include thousands of photographs.

Gisekiaceae [Sarcobataceae + Phytolaccaceae + Nyctaginaceae]: 1 basal ovule/carpel; 210 bp deletion in chloroplast genome.

Family limits in this area may need adjustment. Douglas and Manos (2007) found only moderate support for the monophyly of Nyctaginaceae and vanishing little support for the monophyly of Phytolaccaceae (including Sarcobataceae). Similar relationships were found by Brockington et al. (2009), with Gisekia sister to the whole clade.

GisekIACEAE Nakai - Anthocyanins +; raphides +; inflorescence leaf-opposed; G pseudapocarpous; fruit muricate. - 1/5. Africa, Asia.

Sarcobataceae + Phytolaccaceae + Nyctaginaceae: ?

SARCOBATACEAE Behnke - Bracteoles 0; staminate inflorescence catkinate, flowers with peltate scales ["bracts"], anthers long, pollen with several pores, pore margins raised; carpelate flowers single; P tubular, bilobed [?bracteoles]; G [2], ovule 1. - 1/2. S.W. North America.

Sarcobatus used to be in Chenopodiaceae, but sieve tube plastids with globular inclusions, etc., suggest that it goes somewhere here (Behnke 1997). Carlquist (2000a) discusses the relationships suggested by its wood anatomy.

Phytolaccaceae + Nyctaginaceae: 1 basal ovule/carpel.

The pollen of this clade is notably variable.

PHYTOLACCACEAE R. Brown - Inflorescence racemose; fruit a berry. - 18/65: Phytolacca (25). Tropical and warm temperate, esp. America.

• Phytolaccaceae form a rather heterogeneous group that can be recognised by its racemose inflorescences of apetalous flowers which have many, connate carpels - or a single carpel - each with a single basal ovule and a style; the fruit is often a berry. The plants are herbs, trees, or lianes, and the leaves tend to be rather thick and fleshy, lacking prominent fine venation.

This is a very heterogeneous family, but pending further phylogenetic studies I have circumscribed it broadly.

For information, see Hoffmann (1977: general), Rohwer (1993a: general; he also includes Microtea - see Amaranthaceae), and Carlquist and Schneider (2000: anatomy).

NYCTAGINACEAE A.-L. de Jussieu - P connate, petal-like, lobes induplicate-valvate or contorted; G 1, ovule 1, basal, funicle short; fruit achene or nutlet. - 30/395. Tropical to warm temperate.

1. Leucastereae Bentham & Hooker - Indumentum ± stellate; style thick/0, stigma crest-like. - 4/5. S.E. South America, esp. Brasil.

Boldoeae + The Rest: style long, slender.

2. Boldoeae Heimerl - Stigma inconspicuous. - 3/3. Mexico to Bolivia, the West Indies.

3. The Rest - Leaves opposite; P bipartite [tube rather thick, limb thin], pollen with several pores scattered over the surface; fruit proper closely enveloped by the accrescent basal part of perianth tube. - 25/390: Neea (85), Guapira (70), Boerhavia (50), Mirabilis (45). Tropical to warm temperate, esp. herbs and shrubs in arid southwestern North America and arborescent genera in the Neotropics and Caribbean.

• Nyctaginaceae may be readily recognised. The family includes herbs with swollen nodes, as well as lianes and trees. The leaves are often opposite and the wood is oxidising, i.e. it quickly turns orange/red-brown when cut. The leaves often dry dark grey and the venation is indistinct; the indumentum of the terminal bud is rich brown in color. The tubular, petal-like perianth with its induplicate-valvate or contorted/plicate aestivation is distinctive. The fruit proper, an achene, is covered by the usually accrescent basal part of the perianth. The result may a fruit that looks like a drupe, and this may be covered by glands that produce a very sticky secretion.

The xerophytic North American clade, especially common in S.W. North America, are noted for their abundance in dry or desert conditions; a number of species tolerate gypsum-rich soils, and taxa that flower in the evening or night (hence the common name, the "four o'clock family") are also quite common (Douglas & Manos 2007). Especially in fruits of the xerophytic North American clade, subepidermal cells of the perianth produce mucilage when the fruit is wetted; in species like Pisonia the pericarp in fruit is viscid and very sticky indeed and may be used as bird lime to catch birds. Pisonieae may form ectomycorrhizae with various basiomycetes (Haug et al. 2005).

The single-flowered inflorescences of some species of Mirabilis can look remarkably like individual flowers: The green inflorescence bracts appear to be the calyx, and the brightly-coloured connate perianth then appears to be a sympetalous corolla. Some Nyctaginaceae (Boerhavinae, Nyctagineae) have pollen ca 200 µm long, about the largest in angiosperms outside the aquatic Cymodoceaceae. Abronia has only a single well-developed cotyledon, while the cotyledons of Pisonia and its relatives are unequal in size (and the embryo is straight).

The basic phylogenetic structure of the family is becoming established. The South American Leucastereae and Mexican-Central American Boldoeae are successively sister taxa to the remainder of the family, positions that have moderate to strong support. Within the remainder of the family a North American xerophytic clade had very strong support, and in the whole group Bougainvilleae (paraphyletic), Pisonieae and Abronieae were embedded in a highly paraphyletic Nyctagineae (see also Levin 2000 for a more limited study).

See Vanvinckenroye et al. (1993) for floral development and Bittrich and Kühn (1993) for general information.

Molluginaceae [Halophytaceae + Didiereaceae + Basellaceae + Montiaceae [Talinaceae [Portulacaceae [Anacampserotaceae + Cactaceae]]]]: ?

MOLLUGINACEAE Rafinesque - Anthocyanins +; leaves pseudoverticillate. - 9/87: Mollugo (35), Pharnaceum (20). Largely S. Africa, a few ± tropical to warm temperate.

• Molluginaceae are herbs or subshrubs with pseudoverticillate leaves that often have scarious "stipules", flowers that may be quite showy but which generally lack any corolline whorl, and seeds that lack an obviously long funicle.

This is another poorly-known group, thus confirmation that betalains are absent is needed and the nature of the stipule-like structures will bear examination.

The limits of Molluginaceae have long been unclear. Most Molluginaceae as circumscribed in M. Endress and Bittrich (1993) are included here, but some genera (Limeum and relatives - Limeaceae; Corbichonia - Lophiocarpaceae) are elsewhere in core Caryophyllales. Polpoda, although also core Caryophyllales, is not incorporated in any description. Gisekia is probably to be placed with Phytolaccaceae-Rivinioideae (see Cuénoud et al. 2002). Nepokroeff et al. (2002) found that Mollugo and relatives and Pharnaceum and relatives each formed a well-supported clade, but the two were only weakly linked; both are included in Molluginaceae here.

For information, see Adamson (1960: general), Bogle (1970: general), Richardson (1981: flavonoids), M. Endress and Bittrich (1993: general) and Vincken et al. (2007: saponins).

Halophytaceae + Didiereaceae + Basellaceae + Montiaceae [Talinaceae [Portulacaceae [Anacampserotaceae + Cactaceae]]]: plants succulent; (CAM +); secondary growth normal; Ca oxalate crystals in stem epidermis; outer pair of bracteoles in the median plane [lacking subtending buds, and ± enclosing the flower]; P petal-like.

Pozner and Cocucci (2006) illustrate the bracteoles of Halophytaceae as being in the median plane, although they do not comment on this; this is the same position as the bracteoles of most Montiaceae + Portulacaceae + Didiereaceae + Basellaceae, etc.... Depending on the morphological interpretation of these structures and the topology of this part of the tree, bracteole position could be a synapomorphy for a clade that includes Halophytaceae (see above). The transverse bracteoles have axillary flowers, the median bracteoles always lack them. When there is the same number of stamens as perianth members, their positions relative to the carpels varies.

HALOPHYTACEAE A. Soriano - Vascular rays 0; plant monoecious, flowers sessile; staminate flowers: inflorescence densely spicate; P 4, valvate-decussate, anthers extrorse, porose, anther dehiscence by contraction of the connective, pollen cuboid, hexaporate; carpelate flowers: inflorescence fasciculate; P 0; G unilocular, with 1 basal ovule; fruit a nutlet, several becoming enclosed by inflorescence axis. - 1/1: Halophytum ameghinoi. Argentina.

For information, see Bittrich (1993: general), and Pozner and Cocucci (2006: staminate flower).

Didiereaceae + Basellaceae + Montiaceae [Talinaceae [Portulacaceae [Anacampserotaceae + Cactaceae]]]: mucilage cells +; leaves amphistomatic [?Basellaceae]; median P abaxial, pollen with several colpi scattered over the surface of the grain; G 3.

Nowicke (1996) described a number of pollen characters that are shared in the group (her Portulacinae). For CAM in Portulacaceae s. l., i.e., effectively for most of this clade, see Guralnick and Jackson (2001). CAM cycling is common; this occurs when plants do not completely shut their stomata during the day, and carbon is fixed at night not from atmospheric but from respiratory CO₂.

Relationships between members of this clade are complex, but not well to be understood through the lens of previous classifications, which generally recognised Cactaceae (with "Pereskia" s.l. as sister to the rest of the family), a broad Portulacaceae, Didiereaceae s. str. (i.e., Malagasy taxa only), Basellaceae, and the Antipodean Hectorellaceae. For general information about the old Portulacaceae, see especially Carolin (1987 [also a phylogenetic analysis], 1993). Recent work is beginning to clarify relationships. Hershkovitz and Zimmer (1997) studied relationships in much of the group. They found that if Cactaceae were recognised, "Portulacaceae" would be paraphyletic (see also Appelquist & Wallace 1999, 2001). The [Talinaceae [Portulacaceae [Anacampseros, etc. + Cactaceae]]] clade is rather well supported (Hershkovitz & Zimmer 1997; Appelquist & Wallace 2001; esp. Nyffeler 2007: three genes, two compartments), although details of some of the relationships within it are still somewhat unclear.

The taxa recognised below are consistent with the topology suggested by Nyffeler (2007). Basellaceae and Didiereaceae remain distinct, although a few African genera of Portulacaceae have recently been associated with the latter family because of molecular data; morphology is largely consistent with their new positions (see Didiereaceae for details). Portulacaceae are strongly paraphyletic, and erstwhile members occupy three pectinations on the clade immediately basal to Cactaceae while Pereskia itself is also paraphyletic. A sizeable group of genera that used to be in Portulacaceae comprise Claytoniaceae, which also includes Hectorellaceae. Nevertheless, Cactaceae, Didiereaceae and Portulacaceae remain a closely entwined complex (Appelquist & Wallace 2000). Indeed, they can all be intergrafted (Anderson 1997), although natural grafts of such unlikely subjects as Boscia (Capparidaceae) with Colophospermum (Fabaceae - Caesalpinioideae) and Combretum (Combretaceae) are also reported (Wissels & Potgieter 1997)!

The patterns of morphological and anatomical variation within this clade is complex (see also Nyffeler 2007, especially Ogburn 2007; Ogburn & Edwards 2009; the work of Chorinsky 1931 is still valuable). In at least some species of Anacampseros the median bracteoles are in the same plane as the bud-subtending bracteoles (Vanvinckenroye & Smets 1999: c.f. Talinum - at right angles). Portulaca has an androecial ring primordium, and this is found in some species of Anacampseros, sometimes also with centrifugal initiation of stamens, other species have fewer stamens, which may be initiated in pairs (facing each other!) opposite the perianth members, or as single stamens alternating with them (Vanvinckenroye & Smets 1999).

Didiereaceae + Basellaceae: stomata paracytic; ovary with single basal ovule; fruit indehiscent.

DIDIEREACEAE Radlkofer - 7/16. Madagascar, South Africa, E. Africa.

• Didiereaceae are usually more or less stout-stemmed plants with often "spiny" short shoots; the flowers are usually imperfect and the one-seeded fruits are usually indehiscent.

  This clade includes bizarre-looking plants from Madagascar that have often been segregated as Didiereaceae s. str. However, immediately basal to them are some African ex-Portulacaceae, and the phylogenetic structure is [[Ceraria (Africa; looks like Didiereaceae s. str.!) + Portulacaria (both have tricolpate pollen)] [Calyptrotheca (polyporate) + Didiereaceae (in the old sense: polycolpate)]] (Appelquist & Wallace 2000, 2003: the latter has a tribal classification).

  There are questions as to the morphology of the "thorns" of Didiereaceae s. str. Rauh (1983) calls them spines, being either leaves on short shoots or paired and stipular. However, Alluaudia has leaves subtending an axillary thorn, and later there are paired apparently prophyllar leaves developed from an axillary bud below the thorn; the stem here seems to be modified and form a thorn.

  See Rauh and Schölz (1965: growth, morphology, etc), Kubitzki (1993b: general), Schatz (2001: generic descriptions), and Erbar and Leins (2006: floral ontogeny).

  BASELLACEAE Rafinesque - Vines with swollen rhizomes or tubers; successive cambia +; inflorescence racemose; pollen hexacolpate/porate; perisperm scanty. - 4/19. Mostly New World, some in Africa, but widely cultivated.

  • Basellaceae are more or less succulent and viney plants with small flowers and a single seeded fruit surrounded by the persistent perianth.

  Sperling (1987) reports both bracteoles and large, paired structures immediately surrounding the perianth (see also Eriksson 2007). Basella itself has distinctive, cuboidal pollen, but several other pollen types occur in the family.

  For general information, see Bogle (1969), Sperling (1987), and Eriksson (2007); the latter includes a synopsis of species in the family. For wood anatomy, see Carlquist (1999), and for successive cambia, see Jansen et al. (2000c).

  Montiaceae [Talinaceae [Portulacaceae [Anacampserotaceae + Cactaceae]]]: stomata parallelocytic.

  For information on Portulacaceae that are now scattered throughout this group, see Carolin (1993: general) and Nilsson (1967: pollen).

  MONTIACEAE Rafinesque - Annual to perennial herbs, often with swollen roots and basal leaves, or subshrubs; leaves often with clasping bases; outer wall of exotesta thickened and with stalactite-like projections. - ca 10/: Claytonia (27). Especially Western North and South America, also the Antilles and the Subantarctic Islands.

  West American members of the old Portulacaceae placed here include Montia, Lewisia, Phemeranthus, etc. (e.g. Hershkovitz 2006; Hershkovitz & Zimmer 2000). The New Zealand-Antarctic Hectorellaceae, previously of uncertain relationships, also belongs to this clade. Support for the clade is strong, as it is for a number of groupings within it (Applequist et al. 2006). Although flower position (axillary) and bracteole and stamen position of Hectorellaceae differ from that of Montiaceae (e.g. the stamens are alternate to the "petals"), there are no betalains recorded, the ovary is unilocular, and the anatomy of the two is very similar (Carlquist 1998b). O'Quinn and Hufford (2005) outline the phylogeny of Claytonia (tricolpate) and its sister taxon, Montia (several colpi, all over the surface).

  Claytonia virginiana shows extreme infraspecific variation in chromosome numbers - 2n = 12-ca 191 (Bogle 1969).

  For information, see also Philipson (1993: general, as Hectorellaceae).

  Talinaceae [Portulacaceae [Anacampserotaceae + Cactaceae]]: plant mucilaginous; roots ± tuberous; pericarp two-layered, exocarp ± caducous.

  TALINACEAE Doweld - Shrubs; leaves with paired axillary scales; pericarp epidermis papillate. - 2/22. America and Africa.

  See Vanvinckenroye and Smets (1996) for floral development.

  Portulacaceae [Anacampserotaceae + Cactaceae]: cork cambium cortical; loss of pericyclic fibres; sclereids in stem cortex; leaves with axillary uni- or biseriate hairs/scales +.

  PORTULACACEAE Jussieu - Succulent herbs; leaf stomata transverse; leaves subterete; capsule circumscissile, pericarp undifferentiated. - 1/40-100. Worldwide, but especially tropical and subtropical North and South America.

  • Portulacaceae are succulent herbs with capitate inflorescences, flowers with two "sepals" and usually five "petals", and circumscissile capsules.

  Anacampserotaceae + Cactaceae: A many.

  ANACAMPSEROTACEAE Eggli & Nyffeller - Subshrubs; leaf stomata transverse; leaves subterete; stigma receptive on both surfaces; exotesta ± separate, thin walled, unlignified; embryo slightly curved, not surrounding poorly developed perisperm. - 3-8/32. C. and S. Australia, Somalia to South Africa (most species), S. South America, N. Mexico and S.W. U.S.A.

  • Anacampserotaceae are often rather small, fleshy plants with two "sepals" surrounding five "petals", rather complex fruits in which two layers of the pericarp separate periclinally, and pale-coloured seeds in which two layers of the seed coat also separate pericilinally.

  Generic limits are difficult, since the six species of Grahamia included in the analysis of Nyffeler (1997) formed a perfect basal pectination, and at least some nodes have good support... On the other hand, this helps in reconstructing the character states that are basal in the clade.

  For general information, see Gerbaulet (1992a) and Rowley (1994), for biogeography, see Gerbaulet (1992b), and for ecology, Gerbaulet (1993).

  CACTACEAE Jussieu - Short shoots producing spines and hairs; median bracteoles 0; P several-numerous, spiral, sepaline outside and petaline inside [modified bracts], ovary inferior, placentation parietal; fruit, and also funicles, fleshy, pericarp not two-layered. - 100/1500. Mostly New World, esp. arid conditions, but also rain forest climbers and epiphytes.

  1. Rhodocactus - 1/7. Mexico and the Caribbean, Brasil.

  Pereskioideae [Cactoideae [Opuntioideae + Maihuenioideae]]: delayed bark formation; stomata in stem epidermis.

  2. Pereskioideae Engelmann - 1/9. Andean and S. South America.

  Cactoideae [Opuntioideae + Maihuenioideae]: stems succulent; outermost cortical layers ["hypodermis"] collenchymatous, intercellular spaces +, cortical chlorenchyma forming mesophyllar tissue; leaves reduced [lacking well-developed blade and petiole]; inflorescences axillary, flowers single; hypanthium +.

  3. Cactoideae Eaton - Plant leafless [in fact, leaves up to 1.5(-2.5) mm long when mature], ribbed stems common; testa interstitially pitted or cratered, outer periclinal wall much thickened. - 92/1250: Mammillaria (150), Echinopsis (50-100), Echinocereus (50), Gymnocalycium (50), Rhipsalis (50). New World, (S. Canada to S. W. U.S.A. southwards. Rhipsalis also with a few (?native) spp. in Africa and Madagascar.

  Opuntioideae + Maihuenioideae: leaves terete.

  4. Maihuenioideae P. Fearn - Bark formation not delayed; leaves with cylindrical reticulum of bundles, the external xylem surrounding central mucilage reservoir; pollen tricolpate. - 1/2. Argentina and Chile.

  5. Opuntioideae Burnett - Areoles containing glochids with retrorse barbs; leaves ± deciduous; pollen porate; seeds surrounded by bony funicular "aril". - 5/225: Opuntia (200). S. Canada to Patagonia.

  • Cactaceae can be recognised by their usually stout and very fleshy stems, axillary areoles of spines and hairs, and flowers that usually have an inferior ovary and many stamens and petals; the fruits are fleshy.

  Diversification in Cactaceae is estimated to have occurred in the mid-Caenozoic ca 30 Ma (Hershkovitz & Zimmer 1997, which see for other estimates), so [Maihuenioideae + Opuntioideae + Cactoideae] may be a rather young group. Although Cactaceae are pre-eminently characteristic of arid climates in the New World, but they also include a number of taxa that grow in more or less humid forest as lianes and epiphytes. Animal dispersal of the fruits is universal in the family; in some Cactoideae in particular the seeds may germinate while still in the fruit, a form of vivipary (Cota-Sánchez et al. 2007).

  .

  Edwards and Donoghue (2006; see also Edwards 2006 and Edwards & Diaz 2006) discuss the eco-physiological evolution of Cactaceae (see also Nobel 1988 and references). They emphasize that the leafy Pereskia and Rhodocactus clades have high photosynthetic water use efficiency, very high minimum leaf water potentials, and conservative stomatal behaviour (stomata open only when there is available water, that is, at night or after rain). Other features of potential functional interest include the production of much water conducting tissue relative to leaf area, and perhaps also CAM-type photosynthesis. This latter is poorly developed in Pereskia, etc., but is well developed in the stem of succulent cacti (Martin & Wallace 2000). Most taxa have a shallow rooting system that allows quick uptake of water after rain. In some Cactaceae-Cactoideae, at least, the primary root is determinate in growth, perhaps facilitating the rapid development of lateral roots (Rodríguez-Rodríguez et al. 2003). In many Cactaceae there are "rain roots", water-absorbing roots that develop quickly after rains, and dying when the soil dries up; a skeletal root system of perennial, cork-covered roots persists (Gibson & Nobel 1986). Diversification of the "leafless" Cactaceae (i.e. Cactaceae other than the two leafy clades that are basal to the rest of the family) may be as much connected with the development of a cauline water storage system as with the evolution of the other ecophysiological features just mentioned (and of course one would like to know much more about the physiology and anatomy of the clades immediately basal to Cactaceae...).

  Calcium oxalate metabolism in Cactaceae and relatives is potentially interesting. There is variation in crystal type, i.e., in the degree of hydration, and the distribution of the two crystal forms, whewellite (CaC₂O₄.H₂O) and weddellite (Ca₂O₄.2H₂O), may be systematically interesting, with the latter form predominating in Cactoideae alone (Hartl et al. 2007). Some Cactaceae accumulate positively massive amounts of calcium oxalate crystals - e.g. ca 85% of dry weight in Cactus senilis.

  Fleshy, water-storing roots are scattered in Cactaceae, including Pereskia. The taxa involved are usually small plants, and although the tissue involved varies considerably, suggesting the independent origin of such structures, it is modified secondary vascular tissue (Stone-Palmquist & Mauseth 2002). However, in Opuntioideae these swollen roots - see, for example, the appropriately-named Maiheuniopsis clavarioides - seem to be particularly common in the taxa of the basal pectinations (Griffith & Porter 2009).

  The leaves of Pereskiopsis are large, petiolate and bifacial, those of Quiabentia are terete, unifacial but also persistent, in other Opuntioideae they are terete and deciduous. The more complex leaf types may in fact be derived here (Griffith & Porter 2009). Although one commonly thinks of Cactoideae in particular as being leafless, Mauseth (2007) has shown that most do have leaves that are up to 1.5(-2.5) mm long when mature, although these are still mostly shorter than the terete leaves found in Opuntioideae. Despite their small size, some of these leaves have a rudimentary lamina with vascular tissue, stomata, etc.

  The inferior ovary of Cactaceae is a text-book example of receptacular epigyny in angiosperms (Boke 1964), where the tissue investing the ovary is of axial origin. This is clear in genera like Opuntia where areoles cover the inferior ovary; it is as if the ovary has sunk into the stem. The hypanthium so conspicuous in some Cactoideae, etc., is a development of this axial tissue; an obvious hypanthium is absent from Cactoideae like Rhipsalis.

  Phylogenetic relationships within Cactaceae are still rather unclear, with chloroplast and nuclear genes sometimes suggesting different major clades. A recent study by Nyffeler (2002) found rather weak support for the subfamilies and perhaps rather distressingly Pereskia was not clearly monophyletic! Edwards et al. (2005) confirm that Pereskia s.l. is probably paraphyletic, which allows them to shed new light on the evolution of the cactus habit. For relationships in Opuntioideae, see Griffith (2002), Wallace and Dickie (2002) and especially Griffith and Porter (2009). The latter found the well-supported set of relationships [Maihueniopsis et al. [Pterocactus [terete-stemmed species + flat-stemmed species]]] Within Cactoideae, Blossfeldia liliputana (= Blossfeldioideae Crozier) is perhaps sister to all other Cactoideae (Crozier 2004), however, there is some controversy over its phylogenetic position (see also Gorelick 2004; Mauseth 2006b).

  Wallace and Dickie (2002) suggest that Opuntia should be dismembered. Indeed, depending on the author, the number of genera in the family varies by a factor of ten, the species by a factor of two... The situation in Opuntioideae is in fact a mess, as is clear from the recent study by Griffith and Porter (2009); Hunt (2002) had proposed the recognition of eight broadly-delimited genera, roughly equivalent to tribes, which does make sense pending sorting out the phylogeny of the group as a whole - and might also be a sensible final solution. Whether or not the "stakeholders" (Griffith & Porter 2009) would agree might be another matter. Cactoideae are in a similar mess. A mere sixteen genera encompassed the species in the subfamily 1n 1903, but now as many as 116 genera may be recognized (Hunt 2002).

  For general information, see Barthlott and Hunt (1993), Anderson (2001) and Nobel (2002), for Pereskia s.l., see Leuenberger (1986: general), and Butterworth and Wallace (2005: phylogeny), for pollen, see Leuenberger (1976), for placentation, see Leins and Schwitalla (1988), for seeds of Cactoideae, see Barthlott and Hunt (2000), for general anatomy, see Terrazas and Arias (2003 - esp. Cactoideae), for structure-function relationships, see Mauseth (2006a), for wide-band tracheids, see Mauseth (2004), for subfamilies, see Crozier (2004) and especially Edwards et al. (2005).

  [Back to Index]

  Back to Main Tree