EXTANT SEED PLANTS

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

MAGNOLIOPHYTA

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

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

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

ERICALES [ASTERID I + II]: ovules tenuinucellate.

ASTERID I + II: Ellagic acid 0, proanthocyanidins not common; inflorescence cymose; C forming a distinct tube; A epipetalous, = and opposite sepals or P [polyandry (secondary) very uncommon indeed].

ASTERID I: G [2]; loss of introns 18-23 in d copy of RPB2 gene.

BORAGINACEAE + VAHLIACEAE + GENTIANALES + LAMIALES + SOLANALES: (8-ring deoxyflavonols +); vessel elements with simple perforation plates; C tube initiation late [sampling!]; [vascularised] nectary at base of G; style long.   Back to Main Tree

Morphology, Anatomy, etc. There are several morphological characters of potential phylogenetic interest in this group. Protein crystals in nuclei are apparently not known from Avicennia, etc. (Speta 1977, 1979), and information is needed for groups recently moved to Lamiales. Whether or not such crystals characterise both Lamiales and also Boraginaceae s. str. (see also Wagstaff & Olmstead 1997) needs to be confirmed. Although some Boraginaceae do have protein bodies in their nuclei, they are of two very different kinds, and many Boraginaceae lack them; there is also variation within Lamiales. Lamiales and Boraginaceae have in common similar hydroxycinnamic acid derivatives, i.e. disaccharide esters of rosmarinic/lithospermic/caffeic acids (Mølgaard & Ravn 1988). Boraginaceae s. str. have callose plugs in the pollen tube, as do Solanales, but they are both present and absent in Hydophylloideae, and absent in Heliotropioideae, Cordioideae, and monosymmetric-flowered Lamiales (Cocucci 1983). Oleaceae, Lamiaceae, and Solanaceae have (arabino)xyloglucans and some (galacto)xyloglucan hemicelluloses in the cell wall; the plesiomorphic condition for seed plants is to have (fuco(galacto))xyloglucans (O'Neill & York 2003; Harris 2005), but the sampling is very poor. Wu et al. (2006) suggest that the ancestral chromosome number of Solanales and Gentianales (specifically Solanaceae and Rubiaceae) was x = 11 or 12; no palaeopolyploidization seemed to have occured in this general area.

Phylogeny. Vahlia is placed sister to Lamiales, but with only 63% bootstrap support by Albach et al. (2001b), or is associated more specifically with Boraginaceae in other analyses (Lundberg 2001e; B. Bremer et al. 2002). Boraginaceae (plus Vahliaceae) may be associated with Lamiales; only in ndhF analyses is there some support for a linkage with Solanales (Olmstead et al. 1999, 2000; see also Savolainen et al. 2000a; Lundberg 2001e). In general, relationships between Gentianales, Lamiales and Solanales are unclear (Albach et al. 2001b; B. Bremer et al. 2002; Janssens et al. 2009, Boraginaceae s.l. not included). Even the analysis of all 79 protein-coding plastid genes and four mitochondrial genes did not clarify relationships in this area (Moore et al. 2008).

Unplaced:  Back to Main Tree

BORAGINACEAE Jussieu, nom. cons.

Isokestose and higher inulin oligosaccharides as storage; cork superficial to mid-cortical (pericyclic); (vessel elements with scalariform perforation plates); sieve tubes with nuclear non-dispersive protein bodies; nodes 1:1; petiole bundle(s) arcuate; plant often roughly hairy, hairs with a basal cystolith or cystolith-like body, and/or walls calcified; inflorescences terminal, cyme scorpioid, (bracteoles 0); K free, C tube formation late; anther placentoid 0; (pollen with pseudocolpi), nectary not vascularized; ovary with secondary ["false"] septae, (heterostyly +), stigma dry; ovules 2/carpel, pendulous, epitropous; K persistent in fruit; micropylar endosperm haustoria +.

148[list]/2740 - in six groups below. Temperate to tropical. [Photos - Collection]

1. Boraginoideae Arnott

Herbs (shrubs); pyrrolizidine alkaloids, alkannin, prenylated naphthoquinones [roots with reddish or purple dye], gamma-linolenic acid +; (pits vestured); one (or both) bracteoles 0; corolla often rotate, mouth with fornices [inpushings of corolla tube, but morphology variable] (0), anther connective produced or not, pollen trinucleate, (with pseudocolpi; 4- or more colporate), tube with callose, placental obturator +, (nucellar cap +), endothelium 0, style gynobasic, hollow, (styles divided), stigma punctate to capitate; fruit a schizocarp (capsule - Wellstedia), with (1-)4 nutlets, exocarp sclerified; (testa vascularised), exotestal cells with outer walls thickened and lignified (other patterns of thickening, or unthickened); endosperm oily or 0, also nuclear, endosperm 0, haustoria 0; n = (4-)12(-13).

112/1600: Onosma (150), Cryptantha (150), Myosotis (90), Cynoglossum (75), Paracaryum (70), Plagiobothrys (70), Lithospermum (70: used in dyeing), Echium (60), Amsinckia (50), Mertensia (50), Trigonotis (50). Largely (warm) north temperate, some on mountains in the tropics (map: see Meusel et al. 1978; Hultén & Fries 1986; Böhle et al. 1996; Långström & Chase 2002; FloraBase 2005: incomplete).

Codon (Codonoideae Retief & A. E. van Wyk) has a bicarpellate gynoecium with many ovules; the fruits are loculicidal. Subshrubs; plant prickly-hairy; calyx, corolla and androecium are 10-12-merous, K deeply linear-lobed (Retief & van Wyk 2005; Retief et al. 2005). 1/2. S.W. Africa (map: see Retief et al. 2005). The combination of a many-lobed corolla and a bicarpellate gynoecium is a little odd in the asterid I + II groups, but c.f. Dialypetalanthus (Rubiaceae).

Wellstedia (Wellstedioideae Pilger) is a small, shrubby plant with peltate glands and the flowers are arranged in two serried ranks along the stem. The flowers are four merous, the corolla lobes are three-veined, the corolla tube, eight veined; the pollen is 12-25 x 8-15 µm, the exine being perforate to reticulate, with pseudocolpi, and there is a coarse-reticulate mesocolpium (cf. hydrophylloids); there is no nectary; the ovary is 2-locular with one pendulous, epitropous ovule/carpel, the style is terminal, and the stigma is bifid and wet; the fruit is a loculicidal capsule in which the septae may separate from the walls; the seed has a fringe of downwardly-pointing hairs from near the apex; the embryo is curved, with accumbent cotyledons, and there is no endosperm. 1/5. S.W. and N.E. Africa (map: see Thulin & Johansson 1996; Reteif & van Wyk 2008).

Synonymy: Anchusaceae Vest, Buglossaceae Hoffmannsegg & Link, Cerinthaceae Martynov, Echiaceae Rafinesque, Onosmaceae Martynov, Wellstediaceae Novák

Hydrophyllaceae [Heliotropioideae + Cordioideae + Ehretioideae + Lennooideae]: (plant smells unpleasantly); endothelium +; single layer of transfer cells [cells with labyrinthine ingrowths of the wall] in the testa; micropylar (and chalazal) haustoria +.

2. Hydrophylloideae Burnett

Herbs (shrubs); tannins 0, inulin?, alkaloids?; leaves (opposite; compound), margins lobed, toothed (entire), 2ndary veins pinnate to palmate; cymes scorpioid, bracts and bracteoles usu. 0; (K free); A usu. with small scales at each side of the base, pollen also colpate, tubes with callose (0 [Phacelia]), nectary usu. 0; G with parietal placentation, (seminferior), 2-many (crassinucellate; pleurotropous) ovules/carpel, integument 5-10 cells across, styles +/0, stigma punctate; fruit a loculicidal (+ septicidal) capsule (indehiscent); (seeds ruminate by inpushings of the exotestal cells), exotestal cells thickened on inner and radial walls, (largely disappearing - Nemophila), endotestal cells persistent, walls esp. the inner periclinal ± thickened; endosperm also nuclear, copious to scanty, no haustoria, or chalazal and micropylar haustoria +, with lateral projections, (reserve hemicellulose), embryo (short), green or white; n = 5, 9-13, 19.

17[list]/225: Phacelia (150), Nama (45). New World, esp. drier areas of western North America. [Photo - Undetermined Flower.]

Synonymy: Ellisiaceae Berchtold & J. S. Presl, Eutocaceae Horaninow, Hydrophyllaceae R. Brown, nom. cons., Sagoneaceae Martynov

Heliotropioideae + Cordioideae + Ehretioideae + Lennooideae: bark oxidises; pollen tubes lacking callose; transfer cells in funicle and placenta also; fruit with a multilayered lignified endocarp.

3. Heliotropioideae Arnott

Trees or lianes to herbs; pyrrolizidine alkaloids +; cambium not storied; pericyclic sheath 0 [?always]; pits not vestured (+); petiole bundles arcuate; (tetrahedral crystals +); leaves usu. conduplicate [Tournefortia]; C imbricate or with involute margins; A connate at apex via papillae, connective not produced, (pollen with pseudocolpi; 4 colporate), ovules weakly crassinucellate, nucellar cap +, obturator +, integument ca 8 cells across, style + (0), stigma receptive only basi-laterally, discoid, then conical and ± bilobed at sterile apex, or hemispherical, with a ring of hairs, wet; fruit (drupe with 4, 1-seeded stones), schizocarp; seed exotestal; endosperm 0 at maturity, embryo curved or straight, cotyledons large, suspensor long; n = 5, 7-9, 11-14, etc.

5/405: Heliotropium (250), Tournefortia (150, salicylic acid: polyphyletic - see Diane et al. 2002a). Tropical to warm temperate (map: see Gottschling et al. 2004; Flora Base 2005). [Photo - Flower.]

Synonymy: Heliotropiaceae Schrader, nom. cons.

4. Cordioideae Link

Trees, shrubs or lianes; terpenoid-based quinones +; cambium storied; 2ndary phloem stratified; pits barely vestured; nodes 3:3 [Cordia, but see below]; petiole bundles (invaginated) annular and with (cortical and) rib bundles; crystal sand and prismatic or columnar crystals +; leaf margins toothed to entire; bracteoles usu. 0; K valvate to open, persistent to accrescent, C contorted (imbricate), anther connective usu. not produced, pollen often spiny, pseudocolpi 0, (3-porate - Varronia), ovules straight, tenuinucellate, nucellar cap [Cordia] +, styles + (divided), stigmas punctate to capitate; fruit often a drupe, stones 1 (2, 4) (K accrescent [an anthocarp!]); testa vascularised [Cordia], 3-4 layers of transfer cells; endosperm haustoria?, 0, cotyledons plicate, toothed; n = 9, 14-16, 19.

3/330: Cordia (200+: esp. West Indies, heterostylous), Varronia (100+). Tropical, especially South America (map: see Gottschling et al. 2004; Flora Base 2005). [Photo - Flower, Fruit.]

Synonymy: Cordiaceae Dumortier, nom. cons., Hoplestigmataceae Engler & Gilg, Sebestenaceae Ventenat

5. Ehretioideae Arnott

?Smell; trees or shrubs (herbs); inulins?; cambium storied; (pits vestured - Rochefortia); petiole bundle arcuate; leaf margins toothed (entire); bracteoles?, flowers (4-)5-merous; K (5-15; valvate), C imbricate or inwards-folded, connective usu. not produced, (pollen 3-porate), mesocolpium coarse-reticulate, placentation apical to axile, ovules (weakly) crassinucellate (1/carpel), styles ± well developed, stigmas capitate or elongate; K accrescent or not, fruit a drupe, stone (1-)2-(4-)seeded, or a schizocarp; endosperm copious to 0, embryo curved, suspensor long; n = 5, 7-11, 13, 16, etc.

8/170: Ehretia (75), Bourreria (50). Mostly tropical (map: see Gottschling et al. 2004; Flora Base 2005).

Synonymy: Ehretiaceae Martius, nom. cons.

6. Lennooideae Craven

Echlorophyllous, herbaceous root parasites; chemistry?; cork?; sieve tubes with nuclear non-dispersive protein bodies?; stem with cortical bundles; nodes?; leaves spiral, reduced to scales; inflorescences congested, flowers 5-10-merous; K long, narrow, free, pollen tube callose?, nectary 0; G [5-16], placentation axile, loculi subdivided, ovule ?type, style stout, stigma capitate or lobed; fruit a drupe, irregularly schizocarpic (± circumscissile), K and C persistent; exotesta with fine reticulate thickening; endosperm starchy, ?development, ?haustoria, embryo minute, undifferentiated; n = 9.

3[list]/7. S.W. U.S.A. to N. South America (map: from Brummitt 2007).

Synonymy: Lennoaceae Solms-Laubach, nom. cons.

• Rough to hispid, spiral, exstipulate leaves occur throughout Boraginaceae, the inflorescence is cymose, and the flowers usually are radially symmetrical and with an often clearly 4-lobed and 4-ovulate bicarpellate gynoecium. Major groupings within the family are easy enough to recognise.

• Boraginoideae - the "typical" borage - are usually herbs that can be recognised by their often spiral leaves the hairs on which have distinctive white bases and by their helicoid cymose inflorescences. The flowers lack one or both bracteoles and often have rotate corollas; the style is gynobasic, and the fruit has four nutlets which may be ornamented with various kinds of hooks and other projections.

• Heliotropioideae are often vegetatively like Boraginoideae and also have similar inflorescences and flowers, but the style is terminal and the stigmatic head is almost plug-like. The fruit is often a schizocarp. The stem may oxidise, the slash turning reddish.

• Cordioideae and Ehretioideae are both woody, and sterile members can be hard to assign to this area, although quite often the bark is fibrous, the stem is oxidising, and the twigs are whitish in color. Both subfamilies have cymose inflorescences, but the flowers of Cordioideae can often be recognised by their branched styles and capitate stigmas; the style in Ehretioideae is once-divided. The fruits are drupes, surrounded by a persistent and more or less accrescent calyx.

• Hydrophylloideae are mostly herbs, again often hispid and with helicoid cymes, but the corolla is more open than in Boraginoideae, the style is terminal, the placentation parietal, and the fruit is a capsule.

• Lennooideae are echlorophyllous root parasites with congested inflorescences; the flowers are rather boraginoid, although 5-10 merous.

Evolution. The diversification of primarily woody taxa may have taken place in the mid-Cretaceous, some 90 million years ago, in South America (Gottschling et al. 2004); Wikström et al. (2001) estimate divergence of the whole Boraginaceae 81-77 million years before present, i.e. rather later and some time in the Ypresian-Thanetian ([Vahliaceae + Lamiales] are the sister clade), with Hydrophylloideae and Boraginoideae separating 59-56 million years before present. Moore and Jansen (2006) also suggest rather later dates, with very end Cretaceous 67-63 million years ago being a date for the diversification of the woody taxa. Divergence within Boraginoideae {Ogastemma + Lithospermeae, etc.] is some 43 million years (56-30) ago (Weigend et al. 2009). Magallón and Castillo (2009) suggest an age of ca 77.5 million years for both relaxed and constrained penalized likelihood stem datings for this clade - but note topology; Boraginaceae are included in its sister clade, Vahliaceae excluded.

Echium is noted for the distinctive secondarily woody sometimes monocarpic species that have evolved on Macaronesia within at most twenty million years (Böhle et al. 1996). Divergence within Echium may have begun some 20.6 million years ago, but the Macaronesian diversification - there seem to have been two reversions to the herbaceous habit - can be dated to a mere (6.3-)3.9(-1.5) million years ago (García-Maroto et al. 2009: see Mansion et al. 2009 for other Mediterranean insular endemic Boraginoideae). Moore and Jansen (2006) and Moore et al. (2006) outline speciation and distribution in the amphitropical disjunct desert genus, Tiquilia (Ehretioideae), in some detail; it may have originated in the Palaeocene ca 58 million years before present but began to diversify only rather later, some 33-38 million years before present, and with repeated dispersal from W. North America to W. South America.

Boraginaceae (including Hydrophylloideae) are not often eaten by caterpillars of butterflies (Ehrlich & Raven 1964). However, wilting plants and/or flowers of Boraginoideae and Heliotropoideae, especially the latter, are visted by adult butterflies and moths belonging to Ctenuchidae, Arctiidae, Danainae and Ithomiinae; the pyrrolizidine alkaloids the plants contain are used in the pheromones of these lepidoptera or are distasteful to other animals (see also Crotalaria and some Apocynaceae and Asteraceae-Asteroideae: Edgar et al. 1974; Ackery & Vane-Wright 1984 [considerable detail for the danaines]; Boppreé 1986; Brown 1987). Cordia (Cordioideae) is also sometimes visited, although it is not known to contain these alkaloids.

Holm (1979) noted that some Boraginoideae and Hydrophylloideae have similar rusts.

Leaves subtending branches are sometimes displaced up the branches they subtend in Tournefortia. The growth pattern of some species of Cordia, including the myrmecophilous species, is distinctive: the apex of the stem aborts, some branches are plagiotropic, while one becomes orthotropic and forms the renewal shoot. The inflorescence can also be oddly placed.

The holoparasitic Lennooideae may be sister to - or even a clade within - Ehretioideae (Smith & dePamphilis 1998; Smith et al. 2000; Olmstead & Ferguson 2001), so members of Lennoaceae appear to be parasitizing their close relatives. More work is needed to clarify the evolution of this distinctive parasitic clade.

Floral evolution shows much of interest in this group, although lacking both a comprehensive phylogeny as well as developmental studies of the more distinctive flowers, details must remain poorly known. Thus several taxa have flowers that are not the normal 5-merous core eudicot flower. These include Codon (10-12-merous, ex Hydrophyllaceae; the only African member), Hoplestigma, and Lennooideae. The first two of these have many more stamens and petals than normal, but a simple bicarpellate gynoecium; Lennooideae also show an increase in carpel number. Both the petals and stamens of Hoplestigma are described as being in several series (Goldberg 1986).Boraginoideae are well known as a group that has a gynobasic style, but the terminal style of Heliotropoideae, at least, is also likely to be derived from a gynobasic style. This would explain the fact that in Heliotropoideae the transmitting tissue proceeds to the base of the gynoecium (e.g. Hanf 1935).

Pollination in predominantly by insects, and in Boraginoideae the corolla often shows color changes on aging, pink to blue, yellow to pink to blue, yellow to white; p.H. changes of the cell sap are involved.

The relationships of the several-seeded capsules of Hydrophylloideae to the fruits of the rest of the group - usually few-seeded and indehiscent - are unclear, but parallelisms and/or reversals would seem to be the order of the day. In the large genus Cryptantha, especially diverse in West North and South America, some species have basal, cleistogamous flowers and non-dispersing fruits (Grau 1983).

Chemistry, Morphology, etc. The hydroxycinnamic acid depside, rosmarinic acid, is known from Boraginoideae and Hydrophylloideae (as well as some Lamiaceae: Mølgaard & Ravn 1988). Taxa with root trichoblasts in radial files are quite common in Boraginaceae s.l. There is considerable variation in ovule and endosperm development, conveniently summarized by Khaleel (1985). A variety of septal structures are produced in this group (see Gottschling 2004). There is controversy as to whether the ovules in some Boraginaceae s.l. are crassinucellate or tenuinucellate. Given that ovules with even a single layer of parietal tissue are crassinucellate by the definition followed here, Boraginoideae, for example are tenuinucellate, while Ehretioideae and Heliotropoideae are crassinucellate, Hydrophyllaceae s. str. may have both conditions (e.g. Di Fulvio 1981, 1987; cf. Gottschling 2004; Berg 2009). The style in at least some Boraginoideae and Hydrophylloideae apears to be hollow (Guéguen 1901). Suspensor size in the embryo varies considerably in this group; I have not attempted to see if variation correlates with clades. There is variation in the presence of the mitochondrial coxII.i3 intron (two species sampled...).

Within Boraginoideae, Trichodesmis has opposite leaves and the pendulous and campanulate flowers are borne in a congested inflorescence at the end of a rather long peduncle. Echium has obliquely monosymmetric flowers, while the flowers of some species of Nonea may be vertically monosymmetric, the abaxial stamen being much longer than the others (the flowers of Echiochilon are also monosymmetric - I would guess obliquely so); Cerinthe has an oblique plane of symmetry (Selvi et al. 2009), although the flowers are functionally polysymmetric. With pollen grains at ca 4.1 µm long, Cryptantha clevelandii has about the smallest grains of any flowering plant (Hargrove & Simpson 2003). Amsinckia has very strongly bilobed cotyledons. The corolla in Boraginaceae s.l. usually has many veins diverging in the lobes, but not in Wellstedia. Thulin and Johansson (1996) described the capsule of Wellstedia as being septifragal, Reteif and van Wyk (2008) as being loculicidal; the latter appear to be correct. The ovule has a distinctive shape caused by some kind of chalazal projection (e.g. Reteif & van Wyk 2008).

There are acicular protein bodies in the nucleus of sieve elements in Hydrophylloideae. The mesocolpium is coarse-reticulate, as in some Boraginaceae s. str. (Wagenitz 1992). Di Fulvio (1989a) described the ovules of Nama as being crassinucellate, but tenuinucellate condition seems to be common (e.g. Berg 2009). For details of embryogenesis, seed coat, etc., see also Berg (2009 and references) and di Fulvio (1987, 1990). Tournefortia astrotricha (Heliotropoideae) lacks inulin. Details of embryology are taken from DiFulvio (1978). The central foliar trace of Cordia (Cordioideae) is at least sometimes inverted C-shaped when it joins the central stele, but I do not know how widespread this character is (see Neubauer 1977 for nodal anatomy of Cordia myxa - also modified 1:3). Ehretioideae are heterogeneous in wood anatomy; both Antrophora and Lepidocordia have vessels in radial groups, apotracheal parenchyma, and fiber tracheids with bordered pits (Gottwald 1982). In embryology Ehretioideae are perhaps closest to Heliotropioideae (Diane et al. 2002a). The embryology of Lennooideae is largely unknown.

For further information, see Gürke (1891: general), Svensson (1925: embryology), K. A. Wilson (1960: Hydrophyllaceae, general), Khaleel (1977: testa anatomy), Prósperi and Cocucci (1979: callose, in Boraginoideae, variable in Hydrophyllaceae s. str.), Gottwald (1982), Fisher et al. (1989: sieve tubes with nuclear non-dispersive protein bodies), Heubl et al. (1990: esp. Cordia), Al-Shehbaz (1991: general), Thulin and Johansson (1996), Di Fulvio (1997: protein inclusions in nucleus), Gunstone (1992) and Velasco and Goffman (1999: fatty acids, esp. gamma linolenic acid - but note that this is by no means restricted to Boraginoideae, see Gunstone 1992), Förther (1998: esp. Heliotropioideae), Diane et al. (2002b: transfer cells), Buys and Hilger (2003: inflorescence morphology), Gottschling et al. (2004: diversification), and Aniszewski (2007: alkaloids). For floral morphology in Hydrophylloideae, see Hilger (1987) and Hofmann (1999). For Boraginoideae, see Y. Heslop-Harrison (1981) and Bigazzi and Selvi (2000) for the morphology of stigma papillae, Hilger (1985) for fruit development, Ovczinnikova (2007) and Simpson and Hasenstab (2009: Cryptantha, see the cover photograph!) for fruit morphology, and Bigazzi and Selvi (1998) for pollen morphology of Boragineae. For pollen morphology of Boraginoideae, see Weigend et al. (2009), for that of Cordioideae, see Nowicke and Miller (1990), and for that of Ehretioideae, see J.-X. Liu et al. (2003). See The Parasitic Plant Collection for general information about Lennooideae. I am grateful to Mark Gottschling for discussion, and also the tree used here.

Phylogeny. Boraginoideae are sister to rest of the group, and may include Codon (10-12-merous, ex Hydrophyllaceae, but the only African member of that family) and Wellstedia (also African) as clades successively sister to the remainder. However, the position of Wellstedia - and Codon - must be confirmed; the former did unambiguously link with the other two Boraginoideae in the analysis in Moore and Jansen (2006). Two groups of ex Hydrophyllaceae also are successive sister taxa within the other major clade in some reconstructions, hence loculicidal capsules may be plesiomorphic for the whole group, and Nameae in particular may be sister to the Cordioideae + Heliotropoideae + Ehretioideae (Olmstead & Ferguson 2001; Moore & Jansen 2006, low bootstrap but high posterior probabilty; Luebert & Wen 2008). Nameae include the only woody and tropical members of Hydrophyllaceae s. str. (i.e., excluding Hydrolea, see below). However, Gottschling et al. (2001), looking at the secondary structure of the ITS1 transcript, found Hydrophyllaceae s. str. to be monophyletic (the position of Codon is uncertain) and sister to [Heliotropoideae [Cordioideae + Ehretioideae]] - in the discussion above I follow this topology, but not with much conviction. Boraginaceae and Hydrophylloideae also have meroterpenoids, rusts (see above) and very variable endosperm development in common. See also Ferguson (1999) for relationships.

Långström and Chase (2002) discuss tribal relationships within Boraginoideae. Within Lithospermeae, the limits of Lithospermum itself are to be extended (Weigend et al. 2009; Cohen & Davis 2009), while Lithodora is polyphyletic (Thomas et al. 2008); for the phylogeny of insular Echium, see García-Maroto et al. (2009).

Generic limits in Heliotropoideae need attention (Diane et al. 2002a; Hilger & Diane 2003); Craven (2005) suggests that the whole lot are best placed in Heliotropium - a suggestion not without merit. For genera of Ehretioideae, see Miller (2003),

For generic limits in Cordioideae, see Gottschling et al. (2003); Coldenia is sister to the rest of Cordioideae and Saccelium in included in Cordia s. str. Gottschling et al. (2005) equivocate as to whether Coldenia should be included in their Cordiaceae - if it is, as here, the clade has few apomorphies. Finally, Miller and Gottschling segregate Varronia from Cordia, although since the two are sister taxa, strictly speaking this is not necessary, although the split is supported by morphology. The distinctive West African Hoplestigma, deciduous trees with perulate buds, a connate, irregularly-splitting calyx, 11-14 basally connate petals and 20-35 stamens that are adnate to the base of the corolla, belongs here morphologically. It also has intrusive parietal placentation and long styles that are bent outwards, then inwards, and are terminated by U-shaped, expanded stigmas. Hoplestigma has been placed near Boraginaceae because of its scorpioid cymose inflorescence, absence of bracts, pollen with pseudocolpi (e.g. as in Bourreria and Ehretia - Ehretioideae: Nowicke & Miller 1989), and gynoecium - two carpels with parietal placentation, perhaps as in Hydrophyllaceae (see also Takhtajan 1997). Molecular data place it close to Cordia (K. Wurdack, pers. comm.). The embryology, etc., of the genus are largely unknown, but there appear to be three vascular traces in the base of the petiole; is its nodal anatomy similar to that of Cordia?

Lennooideae are often associated with Boraginaceae and/or Hydrophyllaceae (Cronquist 1981; Takhtajan 1997; esp. Yatskievych et al. 1986). rps2 data even suggest a position within Ehretioideae. Both Ehretia + Lennoaceae have a shared intron in the mitochondrial gene cox1, and Tiquilia in particular is sister to Pholisma (Smith & dePamphilis 1998; Smith et al. 2000; Olmstead & Ferguson 2001). (Tiquilia and immediate relatives, mostly plants of American deserts, are unusual within Ehretioideae in being herbs, or shrubs that flower very quickly, and their inflorescences may be crowded. Their fruits are dry, and secondary veins in the leaves go to the sinuses [Richardson 1977]). However, until there is further resolution of phylogenetic relationships, I have retained Lennooideae as separate from Ehretioideae. For other relationships within Ehretioideae, see Gottschling and Hilger (2001).

All in all, it may be useful to recognize more than one family here, given appropriate support and morphological distinctions, or at least five subfamilies...

Previous Relationships. Boraginaceae have sometimes been associated with Lamiaceae (Lamiales) because both often have gynobasic styles and fruits with four separate nutlets, but the latter have iridoids, opposite leaves, square (not rounded) stems, monosymmetric flowers usually with 4 stamens, endosperm with haustoria and embryo with a long suspensor, and the two are not close. Furthermore, the radicle in Boraginaceae points upwards in fruit, while in Lamiaceae it faces downwards. No putative relative of Boraginaceae s. l. has alkannin or the carcinogenic pyrrolizidine alkaloids found in that clade. Pteleocarpa, with vestured pits, is usually included in Boraginaceae s.l., but it is here placed in Gelsemiaceae (Gentianales). Hydrolea, ex Hydrophyllaceae, but with axile placentation, is here to be found in Solanales. Hoplestigmataceae were included in the Violales by Cronquist (1981), perhaps because of their parietal placentation. Taktajan (1997), however, placed them in Boraginales, in which they have even been included in the past (e.g. Hallier 1912).

Synonymy: Boraginales Dumortier, Cordiales R. Brown, Echiales Lindley, Ehretiales Martius, Hydrophyllales R. Brown, - Boraginanae Doweld

VAHLIACEAE Dandy   Back to Unplaced

Herbs or subshrubs; iridoids +; nodes 1:1; petiole bundle arcuate; hairs uniseriate-glandular; leaves opposite; flowers small; K valvate, C free, G [2(-3)], inferior, disc +, 1-locular, placentae apical, many bitegmic ovules/placenta, outer integument ca 3 cells across, inner integument 2 cells across, but more at micropyle, micropyle endostomal, styles distinct, diverging, stigmas ± capitate; fruit a septicidal capsule, K persistent; exotestal cells ± elongated, outer wall or all walls thickened, endotesta a layer of cells with well-developed U-thickenings, neither lignified, raphe disappears; endosperm slight; n = 6, 9.

1[list]/8. Africa and Madagascar to India (map: from Bridson 1974).

• Vahliaceae are densely glandular-pubescent herbs or subshrubs with opposite entire leaves and rather spreading basically cymose inflorescences bearing small flowers with inferior ovaries; they look rather like some Rubiaceae. However, there are no stipules, the relatively large, persistent calyx is valvate, the petals are free, and the gynoecium has a single loculus with apical placentae. The fruit dehisces at the apex inside the sepal lobes.

Evolution. Magallón and Castillo (2009) offer estimates of ca 83 million years for relaxed and constrained penalized likelihood datings for stem group Vahliaceae - but note topology.

Chemistry, Morphology, etc. Although the ovary has only a single locule, dehiscence of the fruit occurs between the styles, and this is normally equivalent to dehiscence along the septal radii.

Some information is taken from Thouvenin (1890), Mauritzon (1933: ovule), and Cutler and Gregory (1998: anatomy).

Classification. Bridson (1974) monographed the family.

Previous relationships. Vahliaceae were included in Saxifragales by Cronquist (1981) and Takhtajan (1997).

Synonymy: Vahliales Doweld - Vahlianae Doweld

GENTIANALES + LAMIALES + SOLANALES: ?

Evolution. Magallón and Castillo (2009) offer estimates of ca 81 million years for both relaxed and constrained penalized likelihood datings for this clade - but note topology; Boraginaceae are included, Vahliaceae excluded.

GENTIANALES Lindley  Main Tree, Synapomorphies.

Iridoids [different pathways!], monoterpene indole alkaloids, (O-methylated) flavones and flavonols +, myricetin rare; glandular hairs 0; pits vestured; nodes?; petiole bundle(s) arcuate; branching from current flush; leaves opposite, joined by a line across the stem, (stipules +), colleters +; (corolla swollen at the apex in bud), pollen orbicules +, many ovules/carpel, endothelium 0; endosperm nuclear. - 5 families, 1118 genera, 16637 species.

Evolution. Gentianales may be some 89-83 million years old (Wikström et al. 2001), but timing of diversification within the clade is unclear since Dialypetalanthus is shown sister to the rest of the clade in this study - here it is included well within Rubiaceae. Janssens et al. (2009) date stem group Gentianales to 101±7.9 million years ago and the crown group to 79±10.2 million years (comparable figures are 108 and 78 million years in Bremer et al. 2004). Magallón and Castillo (2009) suggest an age of ca 81 million years for both relaxed and constrained penalized likelihood datings for stem Gentianales - but note topology; Boraginaceae are included in its sister clade, Vahliaceae excluded.

Wink (2008) noted that the enzyme strictosidine synthase, a key intermediary in the formation of the monoterpene indole alkaloids commonly found in this clade, is in fact quite widely distributed in flowering plants.

Chemistry, Morphology, etc. The monoterpene indole alkaloid camptothecin is scattered through Gentianales, e.g. it is found in Opiorrhiza (Rubiaceae), Mostuea (Gelsemiaceae) and Ervatamia (Apocynaceae) (see Lorence & Nessler 2004). The colleters consist of secretory palisade surrounding an elongated axis, so they are "proper" emergences. However, in many Gentianaceae they have a simpler structure and are "true hairs". Colleters are frequently borne on leaves, the calyx or corolla (Renobales et al. 2001), or even in a ring below the leaves and encircling the stem, as well as in their normal axillary position. Tapetal variation is considerable, plasmodial tapeta being known from some species of Gentianaceae, Rubiaceae and Apocynaceae (Furness 2008a). Most families have taxa with bi- or trinucleate pollen grains. There is substantial variation in the presence of the mitochondrial coxII.i3 intron in this clade.

Some information is taken from Rogers (1986: general), Erbar and Leins (1996: corolla development), Conn et al. (1997: general), Jansen and Smets (1998: wood anatomy, 2000: vestured pits), and Vinckier and Smets (2002a, c: orbicules).

Phylogeny. Struwe et al. (1995) suggested that Loganiaceae, even when more narrowly circumscribed, were extremely paraphyletic, with clades including about 1,300 genera and 15,500 species (Rubiaceae, Gentianaceae, Apocynaceae + Asclepiadaceae) coming from within them; they delimited families accordingly. However, B. Bremer (1996a), Potgeiter et al. (2000) and Backlund et al. (2000) found rather different relationships - Rubiaceae are sister to Loganiaceae, Gentianaceae, Gelsemiaceae and Apocynaceae. Relationships between these last four families are still not clear. M. Endress et al. (1996) found the relationships Gelsemiaciaceae [[Strychnaceae + Geniostomaceae - well supported] Apocynaceae], see also B. Bremer and Struwe (1992). In other analyses, as in the tree below, there is weak support for a relationship between Gelsemiaceae and Apocynaceae (Backlund et al. 2000; Jiao & Li 2007: for the tree, see also B. Bremer 1999; Rova et al. 2002).

Previous Relationships. The circumscription and relationships of Loganiaceae are a key to understanding the both the past and present circumscription and relationships of Gentianales. Loganiaceae seemed to show relationships with many sympetalous groups, and Bentham (1856) compared a broadly delimited Loganiaceae to a less-wooded area from which obvious forests representing more distinctive families such as Rubiaceae, Solanaceae, etc., had been removed. Loganiaceae in this sense (e.g. Leeuwenberg 1980) had ca 22 genera and 310 species. However, Buddleja s.l. and Androya (not immediately related), Peltanthera and Sanango (not immediately related), Plocospermum, Nuxia and Retzia and Polypremum, all members of a broadly circumscribed Loganiaceae, are in five separate clades in Lamiales (Scrophulariaceae, Gesneriaceae, Plocospermataceae, Stilbaceae and Tetrachondraceae respectively), while Desfontainia (Desfontainiaceae) is in Bruniales, a campanulid/asterid II (for references, see those families) - hence it is not surprising that Loganiaceae seemed to be such a "central" family.

Includes Apocynaceae, Gelsemiaceae, Gentianaceae, Loganiaceae, Rubiaceae.

Synonymy: Apocynales Bromhead, Asclepiadales Dumortier, Cinchonales Lindley, Chironiales Grisebach, Galiales Bromhead, Loganiales Lindley, Lygodiseales Bartling, Rubiales Dumortier, Strychnales link, Theligonales Nakai, Vincales Horaninow - Gentiananae Reveal - Asclepidopsida Brongniart, Coffeopsida Brongniart, Rubiopsida Bartling

RUBIACEAE Jussieu, nom. cons.   Back to Gentianales

Plant tanniniferous; (cork cambium deep-seated); true tracheids +; nodes 1(-3 or more):1(-3 or more); crystal sand +/0; secretory sacs widespread; stomata paracytic; leaves usu. with flat ptyxis, stipules interpetiolar (sheathing; intrapetiolar; two pairs; toothed or not), innervated from circumferential vascular ring; flowers often aggregated, 4- or 5-merous, (heterostyly +), K small, aestivation open (free; 0), C with early tube formation, (valvate), ovary inferior (largely superior in fruit), nectary on top, placentation axile (to parietal), ovules (apotropous; crassinucellate), archesporium multicellular, style usually well developed, stigma wet or dry; fruit baccate, drupaceous or capsular, variously dehiscent; seeds 1-many, (pachychalazal; ruminate), exotesta alone persisting, papillate/short hairy or not, cells variously thickened, (mesotestal cells thickened); endosperm cellular or nuclear, often hemicellulosic, embryo straight to curved, suspensor haustorium +; n = 11 (10-16).

611[list]/13150 - in four groups below. World-wide, but especially tropical (map: from Hultén 1958, 1971; Brummitt 2007). [Photo - Flower]

1. Rubioideae Verdcourt

Commonly herbs; (cylcotide proteins +), shikimic-acid derived anthraquinones +, plants Al accumulators [esp. in woody taxa]; (root with superficial cork cambium - Paederia); raphides + [square in transverse section]; hairs articulated; C valvate, heterostyly esp. common; loss of atpB promoter.

Psychotria s. str. (1400: ?inc. Myrmecodia, Hydnophytum [55], Cephaelis [100]), Galium (400), Spermacoce (inc. Borreria: 275), Palicourea (250), Oldenlandia (250, but polyphyletic), Hedyotis (200), Argostema (100), Morinda (80), Gaertnera (70). Worldwide. [Photo - Fruit]

Synonymy: Aparinaceae Adanson, Asperulaceae Spenner, Cynocrambaceae Endlicher, Galiaceae Lindley, Hedyotidaceae Dumortier, Houstoniaceae Rafinesque, Hydrophylacaceae Martynov, Lippayaceae Meisner, Lygodisodeaceae Bartling, Nonateliaceae Martynov, Operculariaceae Perleb, Pagamaeaceae Martynov, Psychotriaceae F. Rudolphi, Theligonaceae Dumortier, nom. cons.

Luculia, etc. + Cinchonoideae + Ixoroideae: plants woody; route II carboxylated iridoids +, indole alkaloids; hairs mostly cylindrical; secondary polen presentation common; exotestal cells with perforations[?].

2. Luculia [Acranthera + Coptasapelta]

(Raphides +).

3/53: Acranthera (35). Himalayas, China, to Malesia.

3. Cinchonoideae Rafinesque

Corynanthean and complex indole alkaloids +; (raphides + - Hamelieae); C left contorted, ovules often numerous; fruits usu. dry.

Timonius (150), Guettarda (80). Predominantly Neotropical.

Synonymy: Catesbaeaceae Martynov, Cephalanthaceae Rafinesque, Cinchonaceae Batsch, Coutareaceae Martynov, Guettardaceae Batsch, Henriqueziaceae Bremekamp, Naucleaceae Wernham

4. Ixoroideae Rafinesque

(Calycophylls +), C contorted or cochleate, (stigma not bilobed - e.g. Gardenia), ovule number variable; fruits often fleshy; embryo short [?distribution].

Pavetta (400), Ixora (300), Mussaenda (200), Randia (100), Coffea (110: includes Psilanthus), Tricalysia (90), Gardenia (60), Bertiera (55). Pantropical.

Synonymy: Coffeaceae Batsch, Dialypetalanthaceae Rizzini & Occhioni, nom. cons., Gardeniaceae Dumortier, Randiaceae Martynov, Sabiceaceae Martynov

• Members of the family can be recognised in the vegetative state by their opposite, sometimes whorled, and entire leaves that often dry blackish and interpetiolar stipules that have axillary colleters. The ovary is inferior, the flowers are polysymmetric, often with a narrow corolla tube and spreading lobes, and are quite often rather small and aggregated into heads.

Evolution. Divergence within Rubiaceae began a mere (68.8-)66.1(-63) million years ago, with the minimum age of Cinchonoideae being (54.6-)51.3(-47.8) million years; Antonelli et al. (2009) suggest other dates for divergence within the family, especially within Isertieae and Cinchoneae. Bremer and Eriksson (2009) also provide a comprehensive dated phylogeny for the family, but with rather older dates; they suggest ages of some 90.4 million years for stem and about 85 million years for crown Rubiaceae (divergence of Rubioideae from the rest); the split of Ixoroideae and Cinchonoideae was approximately 73 million years ago, with subsequent divergence within those two subfamilies beginning some 59.6 and 34.7 million years ago respectively.

Molecular evolution in the herbaceous Rubioideae seems to be speeded up compared to that in the woody members (Rydein et al. 2009b: note the laggard woody Dunnia within the herbaceous clade!).

Rubiaceae are not often eaten by caterpillar larvae of butterflies (Ehrlich & Raven 1964), although some sphingids (Semanophorae) do prefer members of the family (Forbes 1956).

Myrmecodia, Hydnophytum and related genera placed in Hydnophytinae are vegetatively highly modified epiphytic Malesian ant plants. The ants live in chambers in the grossly swollen stem (hypocotyl) base, and nutrients from the material brought in by the ants and stored in chambers of distinctive morphology are taken up by the plant. Some species have more or less branched root spines arising from the stem, inflorescence, end even the torn and eaten surfaces of the leaves (Huxley 1978; Jebb 1991; see Huxley & Jebb 1991 for the taxonomy of the group). Myrmecophily has arisen several other times in Rubiaceae. Thus there are several myrmecophytic Naucleeae, the ants living in hollowed stems in plants with a much more "normal" appearance than that of many Hydnophytinae; interestingly, some myrmecophytic clades of Naucleeae seem to have diversified notably slowly and/or have very limited distributions (Razafimandibison et al. 2005). Bacterial leaf nodules are known from some African species of Psychotria, and their presence seems to be correlated with development of distinctive colleters (Lersten 1974 and references). Bacteria of the ß-proteobacterium Burkholderia have been isolated from the nodules (van Oevelen et al. 2004). Burkholderia is known to be a nitrogen-fixing symbiont in the root nodules of some Faboideae, however, early studies failed to detect fixation of nitrogen in Psychotria (Miller 1990), and there seems to be nothing more recent on the subject. Pavetta and Sericanthe also have leaf nodules.

There are hundreds of species of Psychotria s.l. throughout the Pacific, including on Hawaii (Nepokroeff et al. 2003; Givnish et al. 2008b). The genus is abundant throughout Malesia and in New Guinea (Nepokroeff et al. 1999; esp. Andersson 2002), and a Pacific-Malesian clade of Psychotria includes the ant plants Myrmecodia and Hydnophytum mentioned above (Andersson 2002). Psychotria is largely divided up into Old and New World clades, and in a subclade of the latter (= Psychotria subgenus Psychotria) it has been shown that older species tend to occupy larger areas than younger species (Paul et al. 2009; cf. J. C. Willis's "age and area" [e.g. Willis 1922] but the conceptual baggage presumably very different in the latter).

Although Rubiaceae can usually be readily recognised by a distinctive combination of vegetative characters (see above), vegetative variation in the family is actually quite extensive even apart from distinctive morphologies of the myrmecophytic taxa just mentioned. Herbaceousness has evolved in Rubioideae, where some Spermacoceae may be secondarily woody while Knoxeae are both primarily and secondarily woody (Lens et al. 2009a, b). In Galium there are whorled "leaves" whose exact nature is a matter of some dispute, but there are only two, opposite branches per node, suggesting that the basic construction of the plant is of paired, opposite leaves; this is confirmed by the pattern of vascularization in most species (e.g. Neubauer 1981), but not in taxa like Galium rubioides which has four large leaves at the node all of which are directly vascularized from the stele (Rutishauser 1999). Most taxa in Rubiaceae have 1:1 nodal anatomy, only some twelve genera being trilacunar (Neubauer 1981); how this fits in the context of the family phylogeny is unknown. Unilacunar nodes are unusual in taxa with stipules (Sinnott & Bailey 1914), but lateral branches of the traces diverge and form a vascular collar around the stem from which the stipular bundles themselves diverge (Majumdar & Pal 1958). Stipule morphology and position in general shows considerable variation; there are sometimes two pairs of stipules, one more or less intrapetiolar, the other interpetiolar. Genipa and Posoqueria have a deeply lobed lamina; "latex" is not uncommon. Anisophylly is well known in the family, occuring also in taxa like Theligonum and Argostemma, being especially marked in taxa like A. humilis.

Many Rubiaceae have rather small flowers but make them attractive to pollinators in a variety of ways. Large, coloured calyces (calycophylls) are particularly common in Ixorodinae and help to attract the pollinator. The Old World Mussaenda is an example with its its large petaloid calyx lobes usually occuring singly on a few flowers in the quite lax inflorescence (the genus may be polyphyletic - see Alejandro et al. 2005), while in the New World Warsewiczia has similarly conspicuous individual sepals. The flowers are often more or less closely aggregated, and Claßen-Bockhoff (1996a) has surveyed the even more flower-like inflorescences that are quite common in the family. Taxa like Cephalanthus and Naucleeae have flowers aggregated into spherical heads, and Cephaelis (Rubioideae) has a condensed inflorescence immediately subtended by paired and coloured inflorescence bracts. In some species of Spermacoce the apices of the corolla lobes become incurved and extensively modified, and pollination may be explosive (Vaes et al. 2006); palynological variation in this genus is extreme, being almost equivalent to that in the whole of the rest of the family (Dessein et al. 2002). Explosive pollination is also known from Posoquerieae, pollen being catapulted onto the pollinating insect; the flowers are monosymmetric (and may be inverted: Delprete 2009). Nilsson et al. (1990), Puff et al. (1996) and de Block and Igersheim (2001) discuss secondary pollen presentation, which is notably common in Cinchonoideae; pollen is presented on tips of the styles. Theligonum is monoecious and has very distinctive flowers. However, the differences between its flowers and those of other Rubiaceae are those that might be expected of a wind-pollinated plant, i.e., breeding system, reduced and modified perianth, many stamens, and only a single-seeded fruit (see Wunderlich 1971 for details of the flowers). Henriquezia has a monosymmetric corolla, and it, too, was placed in a separate family. Hundreds of species, especially members of Rubioideae, are heterostylous. Dioecy occurs in Vanguerieae, and there have been apparent reversals to hermaphroditism (Razafimandimbison et al. 2009). Taxa like the sister genera Gaertnera (see Malcomber 2002) and Pagamea have secondarily superior ovaries (Igersheim et al. 1994).

Chemistry, Morphology, etc. Petiole anatomy is variable, although the bundles are often open; in some Ixoroideae they are circular (Martínez-Cabrera et al. (2009). A few Rubiaceae, including Theligonum and Dialypetalanthus, have more than twice the number of stamens as perianth/sepals/petals (e.g. Endress 2003a). For pollen variation in the paraphyletic Spermacoce, see Dessein et al. (2005b). There is considerable variation in ovule morphology (Maheshwari 1950). Ronse Decraene and Smets (2000) discuss floral development in the family, emphasizing variation in the relative development of a stamen-corolla tube with a common meristem and a corolla tube sensu stricto; filaments may fuse postgenitally with the corolla tube.

For additional information on Rubiaceae, see Lloyd (1902: esp. suspensor, very variable), Verdcourt (1958: general), Koek Noorman (1977: wood), Rogers (1984: Gleasonia, etc.), Rutishauser (1984: stipules), Robbrecht (1988, 1993: general), B. Bremer and Eriksson (1992: fruits and dispersal), Robbrecht et al. (1996: general), Huysmans et al. (1997: orbicules in Cinchonioideae), Vinckier et al. (2000: orbicules in Ixoroideae), Jansen et al. (2002a, 2003: Al accumulation), Delprete (2004: morphology), Rogers (2005: useful summary), Dessein et al. (2005a) and Verellen et al. (2007), both pollen, not really a great help taxonomically, Aniszewski (2007: alkaloids), Gamalei et al. (2008: phloem), and de Toni and Mariath (2008: ovules). For the odd cylcotide proteins, scattered in Rubioideae, see Gruber et al. (2008). I am grateful to Elmar Robbrecht for help with the synonymy.

Phylogeny. Bremer (2009) summarizes phylogenetic work on the family. The basic phylogenetic structure in the family is [Rubioideae + the rest], and in the latter relationships are [[Luculia [Acranthera + Coptasapelta]] [Cinchonoideae + Ixoroideae]] (see B. Bremer 1995, 1999; Rova et al. 2002; Robbrecht & Manen 2006; Rydin et al. 2009a; etc.). The positions of Luculia and Coptosapelta - and now Acranthera - seem moderately well supported (but Bremer & Eriksson 2009 suggest that the three may not form a single clade). Indeed, Luculia and Coptasapelta differ considerably in major characters so assigning polarity to, say, raphide presence in the family becomes rather tricky. As Robbrecht and Manen (2006) emphasize, these two genera differ in having, as in Coptosapelta (or not, as in Luculia) raphides, accumulating (or not) aluminium, T-shaped hairs (not), the corolla is contorted to the right, and in having pororate, acolumellate (tricolporate) pollen grains and distyly or secondary pollen presentation (Verellen et al. 2004; for pollen presentation, see Puff et al. 1996). Although Acranthera is sister to Coptasapelta (see also Bremer & Eriksson 2009), again the two have little morphologically in common (Rydin et al. 2009a).

Within Rubioideae, relationships are becoming clarified. Opiorrhiza has the atpB promoter region that is lacking in all other Rubioideae (Manen & Natali 1996; Natali et al. 1996) and is sister to the rest of the subfamily. Rubieae are well supported as being monophyletic by both morphological and molecular data (Rogers 2005 for literature). Kårehed et al. (2008) investigated the phylogeny of Spermacoceae; they suggested that Hedyotis was to be resticted to Asian taxa; relationships within Spermacoceae have been investigated further by Groeninckx et al. (2009a, esp. 2009b) and Rydin et al. (2009b). Rydin et al. (2008) discuss the placement of some small and little-kmown genera of Rubioideae; they considerably affect our understanding of evolution and diversification of that clade. The Psychotria and Palicourea complexes are sister taxa, and the latter, often having caducous stipules, is largely divided up into Old and New World clades. A Pacific-Malesian clade of Psychotria includes myrmecophytes as well as genera like Amaracarpus (Andersson 2002, note that the important optimisation of marginal preformed germination slits on the pyrenes seems questionable on the tree presented); nomenclatural adjustments are in order. For further information on the phylogeny of Rubioideae, see Andersson and Rova (1999), B. Bremer and Manen (2000: phylogeny and classification in Rubioideae), Backlund et al. (2007), Kårehed and Bremer (2007: Knoxieae), Smedmark (2008: Urophylleae), and Razafimandimbison et al. (2008: relationships around Psychotrieae, many-seeded carpels evolve from one-seeded carpels in Schradereae). Morinda, with its apparently distinctive capitate inflorescences, is in fact polyphyletic (Razafimandimbison et al. 2009b).

For phylogenetic relationhips within Cinchonoideae, see Andreasen and Bremer (1996) and Rydin et al. (2009), for those within Vanguerieae, see Lantz and Bremer (2005) and references and Razafimandimbison et al. (2009). For relationships in Sabiceeae, see Khan et al. (2008). [Naucleeae + Hymenodictyeae] are a well supported clade sister to the only moderately (jacknife) supported remainder of Cinchonoideae (Andersson & Antonelli 2005: Luculia and Coptasapelta excepted). For relationships in Naucleeae, see Razafimandimbison & Bremer (2002).

Within Ixoroideae, Ixora (Ixoridinae) is paraphyletic (Mouly et al. 2009a, b, see also Andreasen & Bremer 2000). For the circumscription of Coffeeae, see A. Davis et al. (2007); see Maurin et al. (2007) for a phylogeny of Coffea. Tosh et al. (2009) have adjusted the limits of the African Tricalysia, Cortés-B. et al. (2009) looked at Retiniphylleae, Razafimandimbison et al. (2009a) have studied the phylogeny of dioecious Vanguerieae, and Kainulainen et al. (2009) have investigated Alberteae.

Two genera are perhaps particularly odd. Dialypetalanthus has K free, C free, opposite K, both in two decussate pairs; A (8-)16-17(-25), not epipetalous, basally connate, basifixed, porose; G [2] (see Piesschaert et al. (1997). Oil glands are described as being pervasive (Goldberg 1986). RbcL data suggest that it is to be included in Ixoroideae (Fay et al. 2000b: Cinchonoideae here). Theligonum has iridoids and raphide sacs; the plant is monoecious, anemophilous, the staminate flowers are paired, opposite, not subtended by leaves; P tubular, ridged, splitting into 2-5 segments; A 2-30, development variable, pollen 4-8-zonoporate, G pseudomonomerous, bilocular, 1-locular by suppression, with 1 campylotropous ovule; testa single-layered, walls thin, endosperm with starch, and the embryo is curved (Wunderlich 1971; Rutishauser et al. 1998). Vegetatively Theligonum is a good match with Rubioideae, where it was included by B. Bremer (1996).

For other phylogenetic studies, see B. Bremer (1996b).

Classification. Robbrecht and Manen (2006) and B. Bremer (2009) should be consulted for a detailed discussion on taxon limits, formal classification, etc. Bremer (2009) notes that of the ca 611 genera in the family, 1/3 are monotypic. Generic limits are problematic in a number of places. Oldenlandia appears to be wildly polyphyletic, and Spermacoce very paraphyletic (Kårehed et al. 2008; Groeninckx et al. 2009a, esp. 2009b). Generic limits around the polyphyletic Morinda have been adjusted (Razafimandimbison et al. 2009). A Pacific-Malesian clade of Psychotria includes myrmecophytes like Myrmecodia and Hydnophytum as well as a number of genera like Amaracarpus (Andersson 2002); substantial changes in generic limits are needed. The limits of Ixora have been formally adjusted (Mouly et al. 2009a, b), while Coffea is to include Psilanthus, see Maurin et al. (2007). Generic and tribal limits are diffficult around Rondeletieae and Guettardeae (Rova et al. 2009); Guettarda itself is polyphyletic (Achille et al. 2006).

Previous Relationships. Cronquist (1981) placed Dialypetalanthaceae adjacent to Pittosporaceae in his Rosales, while Theligonaceae was in his Rubiales. Both families are also maintained as separate by Takhtajan (1997) but are included in Rubiales.

Gentianaceae + Loganiaceae + Gelsemiaceae + Apocynaceae: route I secoiridoids +; internal phloem + [bicollateral vascular bundles]; C tube formation late, syncarpy postgenital.

GENTIANACEAE Jussieu, nom. cons.   Back to Gentianales

Herbs to shrubs (trees), mycorrhizal (and echlorophyllous); (plants Al-accumulators), no starch, but oligosaccharides +, monoterpene indole alkaloids, tannins 0; plants glabrous; cork?; (vessel elements with scalariform perforation plates); rays often 0; parenchyma septate; nodes 1 or more:1 or more; mucilage cells + (0); (stomata anisocytic); leaves sessile, usu. connate basally, ptyxis variable, 2ndary veins ± palmate (pinnate), (stipules +); flowers 4-5(-16)-merous, "disc-like" structure between K and C, C right-contorted, marcescent, (tube formation intermediate); A (extrorse; placentoids +; tapetum amoeboid), basally connate, disc 0; G ?transverse, integument 2-20 cells across, outer epidermal cells early massive, antipodal cells diploid to polyploid, stigma broadly 2-lobed (capitate), wet; fruit a septicidal capsule, calyx often prominent; testa multiplicative, exotestal cells (± elongated), inner walls variously thickened (not), other layers ± disappeared; embryo white or green; 100 bp deletion in trnL gene.

87[list]/ca 1655 - 6 tribes below. World-wide, but especially temperate (map: from Gillett 1963; Hultén 1958, 1971; van Steenis & van Balgooy 1966; Klackenberg 1985; Ho & Liu 2001; Struwe & Albert 2004). [Photo - Flower, Flower.]

1. Saccifolieae Struwe, Thiv, V. A. Albert & Kadereit

(Echlorophyllous mycoheterotrophic herbs; shrubs); chemically unknown; (leaves spiral); flowers (heterostylous), (4-)5(-6)-merous, placentation parietal; (endosperm cellular - Voyriella; cotyledons 0); n = 10-14.

4/19. tropical South America, Panama.

Synonymy: Saccifoliaceae Maguire & Pires

2. Exaceae Colla

(Plant myco-heterotrophic); (flowers monosymmetric/enantiostylous - Exacum, Orphium), K connate or not, endothelium + [Exacum]; anticlinal walls of exotestal cells sinuous or not; x = 7, n = 9, 11, 15, etc.

6/165: Sebaea (60), Exacum (70, inc. Cotylanthera). Africa, esp. Madagascar, Indo-Malesia, and to Australia and New Zealand (some Sebaea).

Chironieae [Helieae + Potalieae + Gentianeae]: xanthones, L-(+)-bornesitol +; placentation parietal.

3. Chironieae Endlicher

(Shrubs); distinctive 6-substituted xanthones; flowers (2-)4-5(-12) merous, K connate, (pollen in tetrads); n = 10, 13-15, 17, etc.

23/159: Centaurium (50). Tropics and warm N. temperate.

Synonymy: Chironiaceae Horaninow, Coutoubiaceae Martynov

Potalieae [Helieae + Gentianeae]: nectary +.

4. Potalieae Reichenbach

Trees to lianes or herbs; C-glucoflavones +; nodes 5 or more:5 or more; cortical sclereids + [Fagraea]; flowers 3-16(-24)-merous, K basally connate; (fruit a berry); n = ?

13/154: Fagraea (75), Lisianthus (30). Pantropical.

Helieae + Gentianeae: ?

5. Helieae Gilg

(Shrubs); (inter/intrapetiolar sheaths, stipules +); flowers (4-)5(-6)-merous, (corona at adaxial base of A), pollen usu. in tetrads, polyads, style often long, twisted and flattened when dry; n = ?

22/205: Macrocarpaea (110), Symbolanthus (30). Tropical Central and South America, Caribean.

6. Gentianeae Colla

Distinctive xanthones, C-glucoflavones +; (nectary on C - Swertia et al.), style often short or 0; n = ³5, very variable.

17/950: Gentiana (360), Gentianella (250: polyphyletic), Halenia (80), Swertia (135: ?polyphyletic). North temperate, to the Celebes (some Tripterospermum) and Africa and Madagascar (some Swertia).

Synonymy: Obolariaceae Martynov

• Gentianaceae often lack vegetative indumentum; many are bitter-tasting herbs. Their almost sessile, entire, opposite leaves are joined at the base and have ascending venation, and their rather large flowers (not the saprophytes) have a relatively large, persistent calyx, an often marcescent corolla with right-contorted lobes and a bicarpellate gynoecium with parietal placentation. The placentae are deeply bilobed and protrude into the ovary cavity.

Evolution. Paris-type endomycorrhizae involving Glomeromycota are common in Gentianaceae, including the myco-heterotrophic members (Imhoff 1999; Franke et al. 2006). Both "normal" and myco-heterotrophic species are found in genera like Sebaea and Exacum. The unplaced Voyria is another myco-heterotroph. Here root hairs are generally absent, but are present in V. primuloides and also where V. aphylla roots abut roots of other plants and also litter; fungal penetration occurs in the former situation (Imhof 1999). The association of Voyria aphylla roots with roots of other plants is interesting; even if the latter are described as being decomposed, this is apparently only locally, and there is the possibility of carbon exchange (Imhoff 1999), so perhaps parasitism occurs.

Swertia and other Gentianeae sometimes have fimbriate appendages and nectaries on the petals; Halenia has flowers with five nectar spurs, uncommon in angiosperms. Sebaea s. str. has a distinctive secondary stigma (Kissling et al. 2010), apparently unique in the angiosperms.

Chemistry, Morphology, etc. Flavone-O-glycosides are known from two species of Exacum, alone in the family. The plants are often bitter-tasting because of the iridoids they contain. Plants of Exacum may be foetid. 1:3 nodes are scattered through the family, e.g. Exacum, Gentiana and Sabatia. Although Gentianaceae do not have stipules, Potaliinae, especially Fagraea, have inter/intrapetiolar sheaths and auriculate structures at the nodes; for the diversity of stipule-like structures in Macrocarpaea (Helieae), see Grant and Weaver (2003). Corolla tube formation for some taxa has been described as being late-early. Exacum has an imbricate corolla. Chassot and von Hagen (2008) discuss variation in pollen in Swertia. Cotylanthera (= Exacum) has straight, ategmic ovules (cf. some Voyria). Gentiana has ovules over almost the entire inner surface of the loculus. An endothelium is reported from Gentianaceae, (Kapil & Tiwari 1978), however, as Shamrov (1996) describes this, it consists of one or two layers of cells of the integument that are elongated periclinally, not more or less anticlinally enlarged, as is common. Hakki (1999) clearly described the multiplicative integument of Orphium frutescens; it is about 6 cells across at anthesis. The considerable variation in integument thickness in the family needs to be put into a phylogenetic context.

Potalieae-Potaliinae are timber trees (secondarily woody?) with multilacunar nodes and cortical sclereids; the flowers of Anthocleista and Potalia are up to 16-merous, the corolla is deciduous, the pollen porate, carpellary fusion is congenital, and the fruits are berries. Young plants of Anthocleista have leaves over 2 m long. The group is palynologically heterogeneous (Nilsson et al. 2002).

For general embryology, see Hakki (1997), ovule development, see Shamrov (1996), for orbicules, see Vinckier and Smets (2000a), for a summary of pollen variation, see Nilsson (2002) and Nilsson et al. (2002), for chemistry, see Jensen and Schripsema (2002: the first two tribes mentioned above need particular study), and for general information, see Struwe and Albert (2002) and Gentian Research Network.

Phylogeny. Relationships between many of the tribes are still not well supported (Molina & Struwe 2009), and basic anatomical and developmental knowledge of Saccifolieae and Exaceae needs to be improved if the evolution of Gentianaceae is to be understood. The Guayanan Saccifolium, with its distinctive ascidiate leaves, is in a clade with the mycoheterotrophic Curtia, Voyriella, etc., that is sister to all other Gentianaceae (Thiv et al. 1999; Struwe et al. 2002; cf. Struwe et al. 1998); the glandular bodies in the leaf axils of Saccifolium are best interpreted as colleters. For the phylogeny of Exaceae, see Yuan et al. (2003) and Kissling et al. (2009: Sebaea s.l. is paraphyletic). For relationships in neotropical Helieae, see Struwe et al. (2009). Within Gentianeae, Gentianella seems to be polyphyletic (von Hagen & Kadereit 2001), while Swertia, too, may be polyphyletic (Chassot et al. 2001; Kadereit & von Hagen 2003).

Classification. For an infra-familial classification of Gentianaceae, see Struwe et al. (2002), also the Gentian Research Network. Kissling et al. (2009) divide Sebaea s.l. into three genera.

Previous Relationships. Anthocleista, Fagraea, and Potalia used to be in Loganiaceae (Struwe & Albert 2000), but details of their iridoid chemistry are very gentianaceous and they are well embedded in the family (e.g. Backlund et al. 2000). Emblingia has been placed here (Savolainen et al. 2000a), but a position in Brassicales in its own family (q.v.) is justified on both molecular and morphological grounds.

Voyria, a genus of small, echlorophyllous mycoheterotrophs, is unplaced. The axis may be rhizopherous, without nodes but bearing roots, etc. Both roots and shoots may be exogenous, or both endogenous. There are colleters in some species. The pollen is 1-6-porate with smooth to scabrate exine and there are no orbicules; species with straight ovules lack an integument and those with anatropous ovules have a single integument; there is an endothelium and a nucellar cap; the endosperm is cellular or nuclear and almost absent, and the embryo is undifferentiated. There are sometimes two almost stamen-like appendages on either side of the ovary. N = 16-20 (For information, see Wood & Weaver 1982; Maas & Ruyters 1986; Albert & Struwe 1997 [morphological phylogenetic analysis]; Struwe et al. 2002; Franke 2002: Voyria flavescens)). Voyria is found in America, but one species, V. primuloides, occurs in Africa (map: from Maas & Ruyters 1986; Raynal-Roques 1967). Although there is evidence from pollen, etc., that Voyria is not particularly similar to the myco-heterotropic Voyriella, etc., its immediate relationships are unclear, although it has bicollateral vascular bundles like many other Gentianales in this part of the tree.

Loganiaceae + Gelsemiaceae + Apocynaceae: ?

LOGANIACEAE R. Brown, nom. cons.   Back to Gentianales

Annual herbs to shrubs or lianes; tryptophane-derived alkaloids, quercetin, kaempferol +; nodes 1:1 or 3:3 (split laterals); stomata?; leaves ± flat, (2ndary veins palmate), joined by line or stipule sheathing; flowers 4- or 5-merous, (median K abaxial - Logania; monosymmetric - Usteria); K basally connate or not, C (valvate), often hairy at the mouth, (A 1, abaxial - Usteria), disc 0, poorly developed, or nectary on walls of G; G often partly inferior and partly apocarpous (carpels congenitally syncarpous; postgenitally connate apically only), placentation axile, 1-many ovules/carpel, styles 0, or styles +, stigma capitate, long-clavate, 2-lobed, or punctate; fruit a follicle, loculicidal and/or septicidal capsule, drupe or berry; seeds often embedded in pulp/placentae swollen, exotestal cells papillate or hairy, ± thick-walled and lignified except outer wall; endosperm starchy or hemicellulosic, horny; n = 10, 12, 16; seedings epigeal and phanerocotylar.

13[list]/420: Strychnos (190), Mitrasacme (55), Geniostoma (55). Pantropical, esp. Australia and New Caledonia (map: from Leenhouts 1962; van Steenis & van Balgooy 1966; Leeuwenberg 1969). [Photo - Flower, Fruit]

Chemistry, Morphology, etc. The wood of Strychnos has included phloem; the plant has branch tendrils. Colporate pollen without lateral extensions at the endocolpus is reported to be a character unique (in this group) to Strychnos and its immediate relatives. The ovules of Mitrasacme have an endothelium and the endosperm is "intermediate" in development. Mitreola oldenlandioides has straight ovules and cellular endosperm (Reddy et al. 1999). The herbaceous Spigelia is distinctive. Its leaves are often pseudoverticillate; the inflorescence is a cincinnus; it has late corolla tube formation; and its fruit is a septicidal+loculicidal capsule, the valves all falling off.

For information, see Hasselberg (1937: nodes and stipules), Leeuwenberg (1980: general), Keller (1996: "stipules"), Hakki (1998: Usteria), and Aniszewski (2007: alkaloids).

Phylogeny. Frasier and Struwe (2009) outline...

Synonymy: Antoniaceae Hutchinson, Gardneriaceae Perleb, Geniostomaceae Struwe & V. Albert, Spigeliaceae Martius, Strychnaceae Perleb

Gelsemiaceae + Apocynaceae: nodes 1:1; seeds ± flattened.

Phylogeny. Backlund et al. (2000) note that C17 indole alkaloids, the number of tapetum layers, and cytology support this relationship - and the presence of quercetin and kaempferol, imbricate corolla, and horny (starchy) endosperm might support a close relationship between Gelsemiaceae and Loganiaceae.

GELSEMIACEAE Struwe & V. Albert   Back to Gentianales

Shrubs or lianes; quercetin, kaempferol +; true tracheids +; stomata?; leaves connected by a ridge, or stipules 2, interpetiolar or short sheathing; flowers heterostylous; A extrorse [Gelsemium] or latrorse, style twice bifid; fruit a loculicidal and/or septicidal capsule, K persistent; seeds winged or hairy; testa?; endosperm horny, starchy; n = 8, 10.

2[list]/11. ± tropical; South East Asia, Africa, America (map: from Leeuwenberg 1961; van Steenis & van Balgooy 1966). [Photo - Gelsemium Collection © M. Dirr, Flower (Pin), Flower (Thrum).]

Evolution. For the phylogeny and biogeography of the family (Pteleocarpa not included), see Jiao and Li (2007)

Chemistry, Morphology, etc. Pteleocarpa, a poorly known and rather odd west Malesian genus in which the leaves as well as the flowers are yellow when the flowers are open, is probably to be placed here; it was previously included in Boraginaceae s.l., but as an anomalous genus (Olmstead & Ferguson 2001). It has not been incorporated into the description; it has i.a. spiral leaves, two ovules/carpel, and a single-seeded, more or less orbicular samara; when in flower, the whole branch is yellow, the leaves having become a color similar to that of the yellow corolla. Brummitt (2007) recognized this as a separate family.

Vascular pits in Gelsemium are not vestured (Rogers 1986), those of Pteleocarpa are. The latter also has mainly apotracheal parenchyma in unilateral, uniseriate bands and fiber tracheids with bordered pits (Gottwald 1982).

APOCYNACEAE Jussieu, nom. cons.   Back to Gentianales

Lianes to trees (herbs); tryptophane-derived and steroidal alkaloids, cardenolides, route II decarboxylated iridoids +, tanniniferous; (cork cambium deep-seated - Rhazya); pericyclic fibers 0 [always?]; (vessel elements with scalariform perforation plates); non-articulated (articulated) laticifers +; (petioles also with adaxial bundles); stomata also paracytic (actinocytic); leaves (spiral), usu. flat or conduplicate, ("stipules" cauline); K with basal adaxial colleters, C left-contorted, postgenital connation forming the upper [above the insertion of the A] tube, corona from C; A ± connivent, filament short; pollen transported in foam; (nectary on outer wall of ovary, 0); G (-8), (connate - Allamanda, Carissa; transverse), styles elongated, usu. apex alone postgenitally syncarpous, stylar head swollen, wet or dry; (ovules hemitropous); (seeds rounded); testa multiplicative, exotestal cells with all walls thickened, (flattened mesotestal crystalliferous cells); extensive polyploidy including triploids, protein crystalloids in the nuclei; also sporophytic incompatibility system present.

415[list]/4555 in five groups below. Largely tropical to warm temperate (map: from ). [Photo - Flower] [Photo - Fruit]

1. "Rauvolfioideae" Kosteletzky

Indole alkaloids +/0; (calycine colleters + - Tabernaemontaneae), anthers connivent, (with lignified basal appendages - Tabernaemontaneae), (pollen porate), G apocarpous or not, placentation then axile or parietal; fruit berry, drupe or follicle; (seed with coma - Haplophyton; aril - Tabernaemontaneae s. str.); n = (9 - esp. Alyxieae) 10, 11, (23).

84/980: Alyxia (120), Tabernaemontana (100), Melodinus (75), Rauvolfia (110), Landolphia (60). Tropical.

"Apocynoideae" + Periplocoideae [Secamonoideae + Asclepiadoideae]: iridoids 0 [this level?]; C right-contorted, anthers with lignified basal appendages [guide rails], adnate to style head, pollen porate; G apocarpous, stylar head radially differentiated; fruit a follicle; seeds comose.

2. "Apocynoideae" Burnett

(C left-contorted; valvate); A inserted well below bases of corolla lobes; n = (6-)10, 11 (12).

77/860: Parsonsia (120), Mandevilla (115), Prestonia (65), Forsteronia (50). Largely tropical.

Synonymy (first two subfamilies): Carissaceae Bertolini, Cerberaceae Martynov, Ophioxylaceae Perleb, Pacouriaceae Martynov, Plumeriaceae Horaninow, Vincaceae Vest, Willughbieaceae J. Agardh

[Periplocoideae [[Asian clade + New World clade] [Baisseeae [Secamonoideae + Asclepiadoideae]]]: few wide vessels and several narrow vessels grouped; vasicentric tracheids +; pollen in tetrads, tetrads T-shaped and tetragonal [microsporogenesis successive].

Chemistry, Morphology, etc. For tetrad morphology, see Omlor (1998); Periplocoideae, Secamonoideae, and Asclepiadoideae - and Fockeeae - are reported to have tetrads of this type.

3. Periplocoideae Endlicher

C (valvate), tube formation intermediate; A inserted at top of C tube (etc.), (staminal feet erect, connate, forming tube around ovary), anthers without guide rails, pollen in tetrads, collected on translator [spoonlike structure + basal sticky viscidium]; exotestal cells unthickened [Periploca]; embryo color?; n = 11 (mostly).

31/180. Old World, tropics to dry temperate (map: Good 1952).

Synonymy: Periplocaceae Schlechter, nom. cons.

Secamonoideae + Asclepiadoideae: monoterpene indole alkaloids 0; C tube formation intermediate; A inserted well below bases of corolla lobes, corona usually staminal, filament 0, staminal feet erect, connate, forming tube around ovary, anthers inserted on top; pollen grains inaperturate, psilate, aggregated into pollinia, pollinia of one pollinarium from half anthers of adjacent stamens, orbicules 0, translator [retinaculum] of hardened resinous secretion [mostly stigmatic] and apical adhesive corpusculum; nectaries behind guide rails; endosperm nuclear (cellular), embryo green; n = 11.

4. Secamonoideae Endlicher

C also left-contorted, pollinia 4, lacking outer wall, translator arms [caudicle] 0, granular layer of exine thick.

9/170: Secamone (100). Old World, esp. Madagascar, tropics to temperate.

5. Asclepiadoideae Burnett

(Latex clear - Marsdenieae and esp. Ceropegieae); (included phloem +); (leaves spiral); C also valvate, anthers bisporangiate [1 sporangium from each theca is lost], pollinia 2, caudicle + (0 - Fockeeae), anther secreting wall around pollinium, pollen tetrads linear during development, monads, (in tetrads when adult, no pollinium wall - Fockea; orbicules + - Riocreuxia), granular layer of exine thin; nectar usually accessible at or near base of guide rails; (exotestal cells with outer walls unthickened - Hoya); n = (9-14).

214/2365: Cynanchum (200), Matalea (180), Ceropegia (160), Asclepias (100), Gonolobus (100), Marsdenia (100), Brachystelma (100), Hoya (90[-200+]), Oxypetalum (90), Dischidia (80), Ditassa (75), Stapelia (70), Orbea (55), Huernia (50), Tylophora (50), Caralluma (47). Tropics to temperate, drier areas esp. in Africa (map: see Good 1952). [Photo - Flower, Flower, Flower, Fruit.]

Synonymy: Asclepiadaceae Borkhausen, nom. cons., Stapeliaceae Horaninow

• Apocynaceae usually have copious latex and the leaves are often opposite and with colleters; the venation of the dry lamina is not very prominent. A number of taxa are climbers. The corolla is often contorted and the stylar head is swollen and closely associated with the more or less connivent anthers. The two carpels are usually held together only by the stylar head; in fruit, they often diverge (if both are fertilised); follicles containing seeds with tufts of hairs at either end are common. The inflorescence may appear extra-axillary, being borne in the axils of scale leaves themselves borne after a reduced internode, or they may be lateral-axillary, as in many Asclepiadoideae (but Leptadenia is axillary!).

Evolution. Within "Apocynoideae" genera in major clades usually come either from the Old or the New World, with little overlap between the two (Livschultz et al. 2007). There is also substantial geographical structuring of clades within groups like e.g. Asclepiadoideae (Goyder 2006 for a summary).

Caterpillars of Nymphalidae-Danainae relish members of this family (Ehrlich & Raven 1964; see Ackery & Vane-Wright 1984 for a comprehensive treatment), but they also eat other plants with latex, including several Moraceae. Apocynaceae are likely to be the ancestral host family for Danainae, caterpillars of all three main clades of which are found on plants of this family (Janz et al. 2006 [?where]; cf. Wahlberg et al. 2009, as Gentianales), althiough many New World Ithomiini eat Solanaceae (q.v.); the danaine clde diverged from other butterfly clades ca 89 million years ago (Wahlberg et al. 2009). The cardenolides of Apocynaceae are noxious and may protect both larva and adult (e.g. Malcolm 1991). Brightly-colored danaine caterpillars and bright orange aphids are to be found on Asclepiadoideae in both North America and southern Africa; aphids may on occasion induce cardenolide production (Martel & Malcolm 2004). Pyrrolizidine alkaloids are known from some Apocynaceae, so they also attract adult Danainae and Ithomiinae which use these compounds as the basis of their pheromones (Boppreé 2005; Brehm et al. 2007); these alkaloids also occur in Crotalaria, some Boraginaceae and Asteraceae-Asteroideae. Interestingly, Ithomiinae preferentially visit bait with withered flowers, while Arctiidae moths prefer crushed roots; Brehm et al. (2007) suggest that the pyrrolizidine alkaloids of Prestonia (Echiteae) may have been originally involved in the pharmacophagous behaviour by Ithomiinae butterflies which are now commonly found on Solanaceae (Wilmott & Freitas 2006). Hosts of seed-eating bugs of Hemiptera-Lygaeidae-Lygaeinae are concentrated in the old Apocynaceae (Slater 1976).

For the possible co-evolution of the longicorn beetle Tetraopes with Asclepias, see Farrell and Mitter (1998), and for a study of the different defence syndromes of Asclepias, see Agrawal & Fishbein (2006). Although the resprouting ability of the plant - or simply its tolerance of herbivory - may be an effective defence against specialist herbivores in particular in Asclepias (Agrawal & Fishbein 2008), the situation is complex; production of a variety of phenolics also changed, showing an overall increase as cardenolide production decreased (Agrawal et al. 2009a). However, one interpretation of the initial diversification in North American Asclepias is that this was accelerated by reduced investment in defensive traits like latex and cardenolide production (Agrawal et al. 2009b).

Although cymose inflorescences of some sort are the norm in the family, some Marsdenieae in particular have very long-lived inflorescences with single flowers or whorls of flowers opening at intervals over a year or more (Meve et al. 2009). In some perhaps plesiomorphic Apocynaceae the stigma is swollen and sticky polysaccharide-terpenoid material is secreted to which the pollen adheres; there is no spatially-localized receptive and secretory areas on the stigma. Such taxa are found in basal grade of "Rauvolfioideae", however, it is possible that these stigmas have been derived more than once, and that the differentiated stigma described below is plesiomorphic for the whole family (Simõs et al. 2007a). In taxa with spatial differentiation within the stigmatic head, pollen is deposited onto the apex of the swollen head where sticky material is secreted (again, this is secondary pollen presentation). An annulus around the middle of the head aids in the removal of the pollen from the proboscis of the pollinator, scraping it off, while the receptive stigma itself is a ring around the base of the head. The gynostegium, formed by the post-genital fusion of anthers and stigma, develops in some "Apocynoideae" when the connective tissue of the anther becomes adnate to the stigmatic head (the staminal retinacle of Simões et al. 2007b). Pollen from thecae of adjacent anthers mixes, whereas pollen from the thecae of the same anther is more or less prevented from mixing by the zone of adnation of the connective, that is, the basic arrangement of the androecium is the same as that in Asclepiadoideae.

Nectar may be secreted by distinct nectaries at the base of the gynoecium, but in Secamonoideae and Asclepiadoideae in particular it is secreted in "staminal outgrowths" from the petaloid staminal feet (i.e. the corolla + filament tissue below the anther) that completely surround the gynoecium. Kunze (e.g. 1991, 1997 and references) show that it is secreted beneath the guide rails, although it may be accessable to the pollinator only elsewhere in the flower (Fahn 1979 suggested that the stigma itself might secrete nectar in Asclepias). The diversity of form produced by tissues from the corolla, stamens and staminal feet is considerable and has been used to delimit genera and species (see Liede & Kunze 1993 for terms used). Indeed, flowers of Apocynaceae in general develop a variety of petaloid appendages. These may develop from the corolla tube (Nerium, Allamanda) or from the apices of the anthers (Adenium, Nerium). Strophanthus can have appendages at the apices of the anthers or more or less bilobed appendages in the angles of the corolla lobes, while the corolla lobes themselves terminate in a narrow, pendulous appendix that in some species is almost 30 cm long; there is no nectary. Within Asclepiadoideae, the corona develops very late and is clearly staminal, and this is consistent with the pattern of gene expression in the flower (Livshultz & Kramer 2009), although Kunze (2005b) suggests that it is an organ sui generis. There is much variation in details of the gynostegium in Asclepiadoideae. Thus Kunze and Wanntorp (2008a) discuss corona and anther skirt evolution while Kunze and Wanntorp (2008b) puzzle over the morphologically distinctive gynostegium of the molecularly unremarkable Hoya spartioides. Nectar is a reward in such flowers, and is usually accessible at or near the base of the guide rail (Kunze 1991, see also Kunze & Wanntrop 2008b); a variety of floral volatiles have been characterized (Jürgens et al. 2009).

The close association of the androecium and gynoecium to form the gynostegium is postgenital. Within Periplocoideae, pollinia seem to have evolved at least three times (Ionta & Judd 2007). Since it is unlikely that Periplocoideae are sister to [Secamonoideae + Asclepiadoideae] (esp. Livschultz et al. 2007), details of the evolution of the pollinarium that characterize the two latter is unclear. Variation in the pollinarium of Fockeeae (which includes both Fockea and Cibirhiza), sister to all other Asclepiadoideae, further confuses the issue (see Verhoeven et al. 2003). In any event, the old idea of the evolution of the pollinia of Asclepiadaceae via Periplocaceae/Periplocoideae as some sort of intermediate needs to be revised (see also Ionta & Judd 2007), since it is unlikely that Periplocoideae are sister to [Secamonoideae + Asclepiadoideae] (Potgeiter & Albert 2001; Sennblad & Bremer 2002; esp. Livschultz et al. 2007).

In general, pollination is a rather precise process. Hairs on the anthers or stigmatic head, or lignified guide rails (Secamonoideae, Asclepiadoideae) at the bases of the anther, are all involved in pollen presentation and guiding the mouth parts of the pollinator so that effective pollination occurs (e.g. Fallen 1986). Particularly when the nectar is physically associated with the guide rails just mentioned, the proboscis of the insect is guided to the viscidium, which thus attaches the pollinia to the pollinator (e.g. Kunze 1991). Fly pollination has been studied in detail in Ceropegia (Vogel 1961) where it is very common; about 60% of the some 60 species studied were each pollinated by a single genus of flies (Ollerton et al. 2009b: see below for the phylogeny of the genus).

Wyatt and Lipow (2007) suggest that the evolution of pollinia and the secondary apocarpy in Asclepiadoideae and Apocynaceae s.l. is connected with the post-zygotic incompatibility system that characterises Apocynaceae (?all) and at least some other Gentianales. At least some Asclepiadoideae have no compitum developed at all, one of the reasons for the frequent occurence of fruits in which only a single carpel has developed.

Although Paris-type mycorrhizae are found in Gentianaceae, Loganiaceae, and Rubiaceae, Arum-type endomycorrhizae, or intermediates, are common in Apocynaceae, especially in Asclepiadoideae (Imhoff 1999).

Chemistry, Morphology, etc. For distinctive fatty acids in the seed, see Badami and Patil (1981). The cambium is occasionally storied. The leaves of Asclepiadoideae and many genera more basal to them have flat ptyxis (Cullen 1978). Genera with opposite leaves such as Hoya can have seedlings with spiral leaves; Absolmsia, previously segregated from Hoya, has spiral leaves even at the flowering stage. In taxa such as Vallesia there are cauline "stipules", apparently colleters in a stipular position; Vallesia also has spiral leaves, while Mandevilla has a distinctive ring of large, radiating, almost fleshy projections immediately below the leaves. A variety of stipule-like structures is also found in Stapelia and relatives; in Edithcolea grandis the "stipule" is represented by a single hair (Bruyns 2000, see also 2004). In other Apocynaceae like Alstonia there is an adaxial excavation at the base of the petiole in which the axillary bud is enclosed, often surrounded by secretions from the colleters.

The "lateral" inflorescences of at least some stapeliads may be displaced-terminal (Bruyns 2004). There is some variation in the direction of overlap of the corolla lobes (Eichler 1874); note that how the corolla tube is initiated in Periplocoideae is unknown. There is quite a lot of variation in pollen morphology in Apocynaceae, independent of pollinarium formation (see above). Alyxieae in particular have grains with large pores, and some species have remarkable barrel-like pollen grains, the pores forming the ends of the barrel (M. Endress et al. 2007a). There has been considerable discussion about the nature of the paired nectaries of Vinca, and because of their vasculature, etc., it has been suggested that they are carpels (Fahn 1979 for literature); this would seem unlikely. When the carpels are connate, placentation may be axile or parietal. Syncarpy seems to have evolved more than once in the family, it may be congenital (Acokanthera) or postgenital (Allamanda) (Sennblad & Bremer 1996). The carpels may be collateral (Spichiger et al. 2002). An endothelium has been reported in the ovules of Apocynaceae s. str. (Kapil & Tiwari 1978; Cronquist 1981), but it is absent according to Rohwer (1996).

Apocynaceae are a much-studied group, see Leeuwenberg (1983: Plumerioideae), Liede and Weberling (1985) and Steck and Weberling (1982), both inflorescence, Kunze (1990: corona), Nilsson et al. (1993: pollen), P. Endress (1994: much floral morphology), Kunze (1996: stamen), Sennblad (1997: general), Sennblad and Bremer (1996: rbcL phylogeny, morphology), Civeyrel et al. (1998: comparison of molecular phylogeny with variation in the pollinaria), Omlor (1996: translator structure in Periplocoideae and Secamonoideae, 1998: floral morphology and testa anatomy), M. Endress and Bruyns (2000, much general information), Albers and Meve (2001: karyology), Verhoeven and Venter (2001: pollen, Asclepiadoideae and Secamonoideae), Vinckier and Smets (2002b: orbicules), Kunze (2005a: morphology and development of the corona), Aniszewski (2007: alkaloids), Van der Ham et al. (2001: pollen of Alyxieae, variable, asymmetrical, many with remarkable nexine), Van de Ven and Van der Ham (2006: pollen of Melodinus, etc.), Demarco et al. (2006: laticifer type), Lens et al. (2009c: wood anatomy, much still to be integrated), and Meve et al. (2009: floral morphology of Marsdenieae). There is much information on Periplocoideae, Secamonoideae, and Asclepiadoideae at a site run by S. Liede-Schumann and U. Meve while F. d'Alessi and L. Viljoen provide information on stapeliads in particular. I thank M. Endress for comments.

Phylogeny. Note that both Rauvolfioideae and Apocynoideae are paraphyletic (Sennblad & Bremer 2002); see Livshultz et al. (2007) for the phylogeny of "Apocynoideae" and Simões et al. (2007) for that of "Rauvolfioideae". "Rauvolfioideae" are particularly highly paraphyletic. The relationships of seven tribes are more or less clear, but those of the remaining five tribes remain to be established; there is good support for Aspidospermeae and Alstonieae as successively sister to all other Apocynaceae (Simões et al. 2007), and both tribes more or less lack uniseriate rays, a feature uncommon in other "Rauvolfioideae" (Lens et al. 2008b). Of the other tribes, Tabernaemontaneae, etc., have an androecium with lignified guide rails and a stigmatic head complex like that of "Apocynoideae" in particular. Alyxieae have irregularly-shaped 2-3-porate pollen grains the ectoapertures having thickened margins (Lens et al. 2008b and references); all tribes can be more or less readily distinguished by a combination of features in their wood anatomy (Lens et al. 2008b). For a phylogeny of Tabernaemontaneae, see Simões et al. (2006a), of Alyxieae, M. Endress et al. (2007) and of "Rauvolfioideae" as a whole, see Simões et al. (2007a).

For phylogenetic relationships in "Apocynoideae"-Mesechiteae, see Simões et al. (2004, 2006b). A small African clade, Baisseeae, has strong support as sister to [Secamonoideae + Asclepiadoideae], and below Baisseeae is a polychotomy including an Asian clade of "Apocynoideae", Rhabdadenia, Periplocoideae, etc. (Livshultz et al. 2007; see also Lahaye et al. 2007). Within Periplocoideae Phyllanthera is sister to the rest; pollinia seem to have evolved at least three times (Ionta & Judd 2007; see also Venter & Verhoeven 2001). Parsonsia and Echites (see Apocynoideae above) may be part of this whole clade, whether or not they have a different tribal name (Echiteae: Sennblad & Bremer 2002). Thus Periplocoideae are not immediately related to [Secamonoideae + Asclepiadoideae], indeed, relationships in this area seem to be [Periplocoideae [[Asian clade + New World clade] [Baisseeae [Secamonoideae + Asclepiadoideae]]], with the position of Rhabdadenia unclear; Baisseeae are to include Dewevrella, which has coiled filaments (Livshultz 2009).

Relationships within New World Asclepiadoideae are being clarified by Liede-Schumann (2005) and Rapini et al. (2007). Hoya is a particularly speciose genus; see Wanntorp et al. (2006a, b, 2009) for phylogenies, Wanntorp and Forster (2007: floral morphology in general), Kunze and Wanntorp (2008a: corona and anther skirt) and Kunze and Wanntorp (2008b: the morphologically distinctive gynostegium of the molecularly unremarkable Hoya spartioidea). Hoya and Dischidia (Marsdenieae) appear to be separable by morphological features, possibly apomorphies, such as pollinium morphology, inner guide rail absence, and nectary position (Wanntorp & Kunze 2009); estimates of species numbers in these two genera in particular vary widely. For relationships within Asclepias, definitely polphyletic, see Goyder et al. (2007); within Ceropegieae relationships are complex, Ceropegia itself occuring all over the tree (Meve & Leide 2001, 2002; Meve & Liede-Schumann 2007). For the Gonolobus area, see Krings et al. (2008).

Classification. The classification here is based on that of M. Endress et al. (2007a; see also Endress & Bruyns 2000), however, because of the problems discussed below, any classification of the entire family can for now only be highly provisional (e.g. M. Endress 2004). In general, generic limits need attention (see e.g. Liede & Täuber 2000; Liede et al. 2002; Rapini et al. 2003; Goyder et al. 2007l; Meve & Liede-Schumann 2007). Generic limits in Asclepiadoideae seem to be particularly difficult, perhaps because floral characters used to delimit genera are often suspect in asterids, and genera here are distinguished on floral minutiae. Asclepias itself is definitely polphyletic, the New World clade including the type of the genus (Goyder et al. 2007). Ceropegieae are another very difficult area, Ceropegia itself occuring all over the tree (Meve & Leide 2001, 2002; Meve & Liede-Schumann 2007: see de Kock & Meve 2007 for a checklist). For genera in the Asclepiadoideae-Marsdenieae, see Omlor (1998) and Meve et al. (2009). For a magnificent revision of southern African stapeliads, see Bruyns (2004).