EMBRYOPSIDA Pirani & Prado

Gametophyte dominant, independent, multicellular, not motile, initially ±globular; showing gravitropism; acquisition of phenylalanine lysase [PAL], microbial terpene synthase-like genes +, triterpenoids produced by CYP716 enzymes, phenylpropanoid metabolism [lignans +, flavonoids + (absorbtion of UV radiation)], xyloglucans in primary cell wall, side chains charged; plant poikilohydrous [protoplasm dessication tolerant], ectohydrous [free water outside plant physiologically important]; thalloid, leafy, with single-celled apical meristem, tissues little differentiated, rhizoids +, unicellular; chloroplasts several per cell, pyrenoids 0; glycolate metabolism in leaf peroxisomes [glyoxysomes]; centrioles/centrosomes in vegetative cells 0, microtubules with γ-tubulin along their lengths [?here], interphase microtubules form hoop-like system; metaphase spindle anastral, predictive preprophase band + [with microtubules and F-actin; where new cell wall will form], phragmoplast + [cell wall deposition centrifugal, from around the anaphase spindle], plasmodesmata +; antheridia and archegonia jacketed, surficial; blepharoplast +, centrioles develop de novo, bicentriole pair coaxial, separate at midpoint, centrioles rotate, associated with basal bodies of cilia, multilayered structure + [4 layers: L1, L4, tubules; L2, L3, short vertical lamellae] (0), spline + [tubules from L1 encircling spermatid], basal body 200-250 nm long, associated with amorphous electron-dense material, microtubules in basal end lacking symmetry, stellate array of filaments in transition zone extended, axonemal cap 0 [microtubules disorganized at apex of cilium]; male gametes [spermatozoids] with a left-handed coil, cilia 2, lateral; oogamy; sporophyte multicellular, cuticle +, plane of first cell division transverse [with respect to long axis of archegonium/embryo sac], sporangium and upper part of seta developing from epibasal cell [towards the archegonial neck, exoscopic], with at least transient apical cell [?level], initially surrounded by and dependent on gametophyte, placental transfer cells +, in both sporophyte and gametophyte, wall ingrowths develop early; suspensor/foot +, cells at foot tip somewhat haustorial; sporangium +, single, terminal, dehiscence longitudinal; meiosis sporic, monoplastidic, MTOC [MTOC = microtubule organizing centre] associated with plastid, sporocytes 4-lobed, cytokinesis simultaneous, preceding nuclear division, quadripolar microtubule system +; wall development both centripetal and centrifugal, 1000 spores/sporangium, sporopollenin in the spore wall laid down in association with trilamellar layers [white-line centred lamellae; tripartite lamellae]; nuclear genome size [1C] <1.4 pg, main telomere sequence motif TTTAGGG, LEAFY and KNOX1 and KNOX2 genes present, ethylene involved in cell elongation; chloroplast genome with close association between trnLUAA and trnFGAA genes [precursors for starch synthesis], tufA gene moved to nucleus; mitochondrial trnS(gcu) and trnN(guu) genes +.

Many of the bolded characters in the characterization above are apomorphies of subsets of streptophytes along the lineage leading to the embryophytes, not apomorphies of crown-group embryophytes per se.

All groups below are crown groups, nearly all are extant. Characters mentioned are those of the immediate common ancestor of the group, [] contains explanatory material, () features common in clade, exact status unclear.


Abscisic acid, L- and D-methionine distinguished metabolically; pro- and metaphase spindles acentric; class 1 KNOX genes expressed in sporangium alone; sporangium wall 4≤ cells across [≡ eusporangium], tapetum +, secreting sporopollenin, which obscures outer white-line centred lamellae, columella +, developing from endothecial cells; stomata +, on sporangium, anomocytic, cell lineage that produces them with symmetric divisions [perigenous]; underlying similarities in the development of conducting tissue and of rhizoids/root hairs; spores trilete; shoot meristem patterning gene families expressed; MIKC, MI*K*C* genes, post-transcriptional editing of chloroplast genes; gain of three group II mitochondrial introns, mitochondrial trnS(gcu) and trnN(guu) genes 0.

[Anthocerophyta + Polysporangiophyta]: gametophyte leafless; archegonia embedded/sunken [only neck protruding]; sporophyte long-lived, chlorophyllous; cell walls with xylans.


Sporophyte well developed, branched, branching apical, dichotomous, potentially indeterminate; hydroids +; stomata on stem; sporangia several, terminal; spore walls not multilamellate [?here].


Vascular tissue + [tracheids, walls with bars of secondary thickening]; stomata involved in gas exchange.


Sporophyte with photosynthetic red light response, stomata open in response to blue light; plant homoiohydrous [water content of protoplasm relatively stable]; control of leaf hydration passive; plant endohydrous [physiologically important free water inside plant]; sporophyte with polar auxin transport, PIN [auxin efflux facilitator] involved; (condensed or nonhydrolyzable tannins/proanthocyanidins +); xyloglucans with side chains uncharged [?level], in secondary walls of vascular and mechanical tissue; lignins +; stem apex multicellular, with cytohistochemical zonation, plasmodesmata formation based on cell lineage; vascular development acropetal, tracheids +, in both protoxylem and metaxylem, G- and S-types; sieve cells + [nucleus degenerating]; endodermis +; leaves/sporophylls spirally arranged, blades with mean venation density ca 1.8 mm/mm2 [to 5 mm/mm2], all epidermal cells with chloroplasts; sporangia adaxial, columella 0; tapetum glandular; ?position of transfer cells; MTOCs not associated with plastids, basal body 350-550 nm long, stellate array in transition region initially joining microtubule triplets; suspensor +, shoot apex developing away from micropyle/archegonial neck [from hypobasal cell, endoscopic], root lateral with respect to the longitudinal axis of the embryo [plant homorhizic].


Sporophyte endomycorrhizal [with Glomeromycota]; growth ± monopodial, branching spiral; roots +, endogenous, positively geotropic, root hairs and root cap +, protoxylem exarch, lateral roots +, endogenous; G-type tracheids +, with scalariform-bordered pits; leaves with apical/marginal growth, venation development basipetal, growth determinate; sporangium dehiscence by a single longitudinal slit; cells polyplastidic, MTOCs diffuse, perinuclear, migratory; blepharoplasts +, paired, with electron-dense material, centrioles on periphery, male gametes multiciliate; chloroplast long single copy ca 30kb inversion [from psbM to ycf2]; mitochondrion with loss of 4 genes, absence of numerous group II introns; LITTLE ZIPPER proteins.


Sporophyte woody; stem branching lateral, meristems axillary; lateral root origin from the pericycle; cork cambium + [producing cork abaxially], vascular cambium bifacial [producing phloem abaxially and xylem adaxially].


Plants heterosporous; megasporangium surrounded by cupule [i.e. = unitegmic ovule, cupule = integument]; pollen lands on ovule; megasporangium indehiscent, megaspore germination endosporic [female gametophyte initially retained on the plant].


Plant evergreen; nicotinic acid metabolised to trigonelline, (cyanogenesis via tyrosine pathway); microbial terpene synthase-like genes 0; primary cell walls rich in xyloglucans and/or glucomannans, 25-30% pectin [Type I walls]; lignin chains started by monolignol dimerization [resinols common], particularly with guaiacyl and p-hydroxyphenyl [G + H] units [sinapyl units uncommon, no Maüle reaction]; root stele diarch to pentarch, xylem and phloem originating on alternating radii, cork cambium deep seated; stem apical meristem complex [with quiescent centre, etc.], plasmodesma density in SAM 1.6-6.2[mean]/μm2 [interface-specific plasmodesmatal network]; eustele +, protoxylem endarch, endodermis 0; wood homoxylous, tracheids and rays alone, tracheid/tracheid pits circular, bordered; mature sieve tube/cell lacking functioning nucleus, sieve tube plastids with starch grains; phloem fibres +; cork cambium superficial; leaf nodes 1:1, a single trace leaving the vascular sympodium; leaf vascular bundles amphicribral; guard cells the only epidermal cells with chloroplasts, stomatal pore with active opening in response to leaf hydration, control by abscisic acid, metabolic regulation of water use efficiency, etc.; axillary buds +, exogenous; prophylls two, lateral; leaves with petiole and lamina, development basipetal, lamina simple; sporangia borne on sporophylls; spores not dormant; microsporophylls aggregated in indeterminate cones/strobili; grains monosulcate, aperture in ana- position [distal], primexine + [involved in exine pattern formation with deposition of sporopollenin from tapetum there], exine and intine homogeneous, exine alveolar/honeycomb; ovules with parietal tissue [= crassinucellate], megaspore tetrad linear, functional megaspore single, chalazal, sporopollenin 0; gametophyte ± wholly dependent on sporophyte, development initially endosporic [apical cell 0, rhizoids 0, etc.]; male gametophyte with tube developing from distal end of grain, male gametes two, developing after pollination, with cell walls; female gametophyte initially syncytial, walls then surrounding individual nuclei; embryo cellular ab initio, suspensor short-minute, embryonic axis straight [shoot and root at opposite ends; plant allorhizic], cotyledons 2; embryo ± dormant; chloroplast ycf2 gene in inverted repeat, trans splicing of five mitochondrial group II introns, rpl6 gene absent; whole nuclear genome duplication [ζ - zeta - duplication], two copies of LEAFY gene, PHY gene duplications [three - [BP [A/N + C/O]] - copies], 5.8S and 5S rDNA in separate clusters.


Lignans, O-methyl flavonols, dihydroflavonols, triterpenoid oleanane, apigenin and/or luteolin scattered, [cyanogenesis in ANA grade?], lignin also with syringyl units common [G + S lignin, positive Maüle reaction - syringyl:guaiacyl ratio more than 2-2.5:1], hemicelluloses as xyloglucans; root cap meristem closed (open); pith relatively inconspicuous, lateral roots initiated immediately to the side of [when diarch] or opposite xylem poles; origin of epidermis with no clear pattern [probably from inner layer of root cap], trichoblasts [differentiated root hair-forming cells] 0, hypodermis suberised and with Casparian strip [= exodermis]; shoot apex with tunica-corpus construction, tunica 2-layered; starch grains simple; primary cell wall mostly with pectic polysaccharides, poor in mannans; tracheid:tracheid [end wall] plates with scalariform pitting, wood parenchyma +; sieve tubes enucleate, sieve plate with pores (0.1-)0.5-10< µm across, cytoplasm with P-proteins, not occluding pores of plate, companion cell and sieve tube from same mother cell; ?phloem loading/sugar transport; nodes 1:?; dark reversal Pfr → Pr; protoplasm dessication tolerant [plant poikilohydric]; stomata brachyparacytic [ends of subsidiary cells level with ends of pore], outer stomatal ledges producing vestibule, reduction in stomatal conductance with increasing CO2 concentration; lamina formed from the primordial leaf apex, margins toothed, development of venation acropetal, overall growth ± diffuse, secondary veins pinnate, fine venation hierarchical-reticulate, (1.7-)4.1(-5.7) mm/mm2, vein endings free; flowers perfect, pedicellate, ± haplomorphic, protogynous; parts free, numbers variable, development centripetal; P +, ?insertion, members each with a single trace, outer members not sharply differentiated from the others, not enclosing the floral bud; A many, filament not sharply distinguished from anther, stout, broad, with a single trace, anther introrse, tetrasporangiate, sporangia in two groups of two [dithecal], each theca dehiscing longitudinally by a common slit, ± embedded in the filament, walls with at least outer secondary parietal cells dividing, endothecium +, cells elongated at right angles to long axis of anther; tapetal cells binucleate; microspore mother cells in a block, microsporogenesis successive, walls developing by centripetal furrowing; pollen subspherical, tectum continuous or microperforate, ektexine columellate, endexine lamellate only in the apertural regions, thin, compact, intine in apertural areas thick, pollenkitt +; nectary 0; carpels present, superior, free, several, ascidiate [postgenital occlusion by secretion], stylulus at most short [shorter than ovary], hollow, cavity not lined by distinct epidermal layer, stigma ± decurrent, carinal, dry; suprastylar extragynoecial compitum +; ovules few [?1]/carpel, marginal, anatropous, bitegmic, micropyle endostomal, outer integument 2-3 cells across, often largely subdermal in origin, inner integument 2-3 cells across, often dermal in origin, parietal tissue 1-3 cells across, nucellar cap?; megasporocyte single, hypodermal, functional megaspore lacking cuticle; female gametophyte lacking chlorophyll, not photosynthesising, four-celled [one module, nucleus of egg cell sister to one of the polar nuclei]; ovule not increasing in size between pollination and fertilization; pollen grains land on stigma, bicellular at dispersal, mature male gametophyte tricellular, germinating in less than 3 hours, pollen tube elongated, unbranched, growing towards the ovule, between cells, growth rate (20-)80-20,000 µm/hour, apex of pectins, wall with callose, lumen with callose plugs, penetration of ovules via micropyle [porogamous], whole process takes ca 18 hours, distance to first ovule 1.1-2.1 mm; male gametes lacking cell walls, ciliae 0, siphonogamy; double fertilization +, ovules aborting unless fertilized; P deciduous in fruit; mature seed much larger than fertilized ovule, small [<5 mm long], dry [no sarcotesta], exotestal; endosperm +, cellular, development heteropolar [first division oblique, micropylar end initially with a single large cell, divisions uniseriate, chalazal cell smaller, divisions in several planes], copious, oily and/or proteinaceous, embryo short [<¼ length of seed]; plastid and mitochondrial transmission maternal; Arabidopsis-type telomeres [(TTTAGGG)n]; nuclear genome size [1C] <1.4 pg [mean 1C = 18.1 pg, 1 pg = 109 base pairs], whole nuclear genome duplication [ε/epsilon event]; ndhB gene 21 codons enlarged at the 5' end, single copy of LEAFY and RPB2 gene, knox genes extensively duplicated [A1-A4], AP1/FUL gene, palaeo AP3 and PI genes [paralogous B-class genes] +, with "DEAER" motif, SEP3/LOFSEP and three copies of the PHY gene, [PHYB [PHYA + PHYC]]; chloroplast chlB, -L, -N, trnP-GGG genes 0.

[NYMPHAEALES [AUSTROBAILEYALES [[CHLORANTHALES + MAGNOLIIDS] [MONOCOTS [CERATOPHYLLALES + EUDICOTS]]]]]: wood fibres +; axial parenchyma diffuse or diffuse-in-aggregates; pollen monosulcate [anasulcate], tectum reticulate-perforate [here?]; ?genome duplication; "DEAER" motif in AP3 and PI genes lost, gaps in these genes.

[AUSTROBAILEYALES [[CHLORANTHALES + MAGNOLIIDS] [MONOCOTS [CERATOPHYLLALES + EUDICOTS]]]]: phloem loading passive, via symplast, plasmodesmata numerous; vessel elements with scalariform perforation plates in primary xylem; essential oils in specialized cells [lamina and P ± pellucid-punctate]; tension wood + [reaction wood: with gelatinous fibres, G-fibres, on adaxial side of branch/stem junction]; tectum reticulate; anther wall with outer secondary parietal cell layer dividing; nucellar cap + [character lost where in eudicots?]; 12BP [4 amino acids] deletion in P1 gene.

[[CHLORANTHALES + MAGNOLIIDS] [MONOCOTS [CERATOPHYLLALES + EUDICOTS]]] / MESANGIOSPERMAE: benzylisoquinoline alkaloids +; sesquiterpene synthase subfamily a [TPS-a] [?level], polyacetate derived anthraquinones + [?level]; outer epidermal walls of root elongation zone with cellulose fibrils oriented transverse to root axis; P more or less whorled, 3-merous [?here]; pollen tube growth intra-gynoecial; extragynoecial compitum 0; carpels plicate [?here]; embryo sac bipolar, 8 nucleate, antipodal cells persisting; endosperm triploid.

[MONOCOTS [CERATOPHYLLALES + EUDICOTS]]: (extra-floral nectaries +); (veins in lamina often 7-17 mm/mm2 or more [mean for eudicots 8.0]); (stamens opposite [two whorls of] P); (pollen tube growth fast).

[CERATOPHYLLALES + EUDICOTS]: ethereal oils 0.

EUDICOTS: (Myricetin, delphinidin +), asarone 0 [unknown in some groups, + in some asterids]; root epidermis derived from root cap [?Buxaceae, etc.]; (vessel elements with simple perforation plates in primary xylem); nodes 3:3; stomata anomocytic; flowers (dimerous), cyclic; protandry common; K/outer P members with three traces, ("C" +, with a single trace); A ?, filaments fairly slender, anthers basifixed; microsporogenesis simultaneous, pollen tricolpate, apertures in pairs at six points of the young tetrad [Fischer's rule], cleavage centripetal, wall with endexine; G with complete postgenital fusion, stylulus/style solid [?here]; seed coat?

[PROTEALES [TROCHODENDRALES [BUXALES + CORE EUDICOTS]]]: (axial/receptacular nectary +).

[TROCHODENDRALES [BUXALES + CORE EUDICOTS]]: benzylisoquinoline alkaloids 0; euAP3 + TM6 genes [duplication of paleoAP3 gene: B class], mitochondrial rps2 gene lost.

[BUXALES + CORE EUDICOTS]: mitochondrial rps11 gene lost.

CORE EUDICOTS / GUNNERIDAE: (ellagic and gallic acids +); leaf margins serrate; compitum + [one position]; micropyle?; γ whole nuclear genome duplication [palaeohexaploidy, gamma triplication], x = 21, PI-dB motif +; small deletion in the 18S ribosomal DNA common.

[ROSIDS ET AL. + ASTERIDS ET AL.] / PENTAPETALAE: root apical meristem closed; (cyanogenesis also via [iso]leucine, valine and phenylalanine pathways); flowers rather stereotyped: 5-merous, parts whorled; P = calyx + corolla, the calyx enclosing the flower in bud, sepals with three or more traces, petals with a single trace; stamens = 2x K/C, in two whorls, internal/adaxial to the corolla whorl, alternating, (numerous, but then usually fasciculate and/or centrifugal); pollen tricolporate; G [5], G [3] also common, when [G 2], carpels superposed, placentation axile, style +, stigma not decurrent; compitum +; endosperm nuclear; fruit dry, dehiscent, loculicidal [when a capsule]; RNase-based gametophytic incompatibility system present; floral nectaries with CRABSCLAW expression; (monosymmetric flowers with adaxial/dorsal CYC expression).

[DILLENIALES [SAXIFRAGALES [VITALES + ROSIDS s. str.]]]: stipules + [usually apparently inserted on the stem].


[VITALES + ROSIDS] / ROSIDAE: anthers ± dorsifixed, transition to filament narrow, connective thin.

ROSIDS: (mucilage cells with thickened inner periclinal walls and distinct cytoplasm); if nectary +, usu. receptacular; embryo long; chloroplast infA gene defunct, mitochondrial coxII.i3 intron 0.

ROSID I / FABIDAE / [ZYGOPHYLLALES [the COM clade + the nitrogen-fixing clade]]: endosperm scanty.   Back to Main Tree

[the COM clade + the nitrogen-fixing clade]: ?

[FABALES [ROSALES [CUCURBITALES + FAGALES]]] / the nitrogen-fixing clade / fabids: (N-fixing by associated root-dwelling bacteria); tension wood +; seed exotestal.

Age. Wikström et al. (2001) dated this node to (96-)94, 89(-87) m.y., but other estimates are a little older - Moore et al. (2010: 95% highest posterior density) suggested ages of (107-)104(-100) m.y. and Bell et al. (2010) ages of (107-)99(-91) m. years. Stem-group ages for Fabaceae (the ages just mentioned) offered by H. Wang et al. (2009), Magallón and Castillo (2009) and Foster et al. (2016a), are not much different, nor, at (119-)115, 111(-109) m.y., are those of of H.-L. Li et al. (2015), while Xue et al. (2012) suggested ages of a mere 68.6 or 62.2 m.y. for this clade and Hohmann et al. (2015) an age of 109.1 m.y., although in all the internal topologies are different from that used here. (100.2-)94.6, 70.6(-61.6) are ages suggested by Pfeil and Crisp (2008) and around 96.8 or 86.3 m.y. is suggested by Naumann et al. (2013); by far the oldest estimate, at ca 132 m.y., is that of Z. Wu et al. (2014).

Evolution: Divergence & Distribution. See D. W. Taylor et al. (2012) for possible apomorphies of the whole clade.

Nitrogen fixation in this group of four orders is a classic example of a "tendency" or "predisposition" (e.g. Soltis et al. 1995b; Kistner & Parniske 2002; Vessey et al. 2004; Marazzi et al. 2012); Werner et al. (2014) introduced a different set of terms, but the principle is the same. The molecular reasons for the restriction of these diverse bacterial associations to the N-fixing clade are being dissected. A number of the genes involved in the establishment of the symbioses with both gram-negative α and β proteobacteria including Rhizobium and the gram-positive actinomycete Frankia are the same as those involved in arbuscular mycorrhizal (AM) associations, the "SYM" (symbiosis) or CSSP (common symbiotic signalling pathway) pathway being involved in all (Markmann & Parniske 2008; Bonfante & Genre 2010; Hocher et al. 2011; J. J. Doyle 2011; Svistoonoff et al. 2013, 2014; Martin et al. 2017). There are connections between the signalling genes involved in AM symbioses and rhizobial Nod factors that allow the bacterium to be recognised by the plant (Maillet et al. 2010; Op den Camp et al. 2010; Streng et al. 2011; Young et al. 2011; Roberts et al. 2012), and the rhizobia may acquire the nod genes by horizontal transfer within Fabaceae (Suominen et al. 2001). Both the AM establishment and nodulation pathways involve the recognition of lipochitooligosaccharides (N-acetylglucosamine units - see chitin - with a fatty acid at one end and 2-O-methyfucose at the other), part of the common symbiosis signalling pathway (CSSP), and nuclear Ca2+ oscillations ("spiking") occur in both (Garcia et al. 2015; Kawaharada et al. 2017). The CSSP may be involved in the establishment of actinorhizal associations, with nuclear Ca2+ oscillations in both (Barker et al. 2017). There are further connections to pathogen resistance, which involves chitooligosaccharides (Liang et al. 2014); for oomycete resistance and AM absence, see Delaux et al. (2014). Other flowering plants respond to the rhizobial Nod factor by suppressing MAMP (microbe-associated molecular pattern) immunity - and this is the initial response even in rhizobial infection in legumes (Liang et al. 2013). Interestingly, Nod factors are not found in many actinorhizal associations (Normand et al. 2007) and even the legume Aeschynomene may not need them (Giraud et al. 2007). It appears that (see also A. Taylor & Qiu 2016).

The symbiosis receptor-like kinase gene exists in a particularly distinctive form in the N-fixing clade - and also in Tropaeolum, but not rice, tomato or poppy. This gene in the three latter genera rescued mycorrhizal formation in defective forms of the gene in Fabaceae, but not nodule formation, whereas the Tropaeolum gene restored the ability to form nodules (Markmann et al. 2008; see also Chen et al. 2007, 2009; Gherbi et al. 2008; Yano et al. 2008; Markmann & Parniske 2009). There is notable similarity at the transcriptional level in the genes involved in the establishment of N-fixing endosymbioses in Casuarina, Alnus, Rhamnaceae (all with actinorhiza), legumes, etc. (Hocher et al. 2011), and many of these genes have been coopted from other pathways that are widely functional in land plants, including hornworts and liverworts (Svistoonoff et al. 2013). Werner et al. (2014) suggested that a state precursory to nitrogen fixation was necessary, and that this had been lost 16< times, however, this idea seems to be somewhat notional, although testable.

Members of the nitrogen-fixing clade, practically alone among land plants, have associations with N-fixing bacteria - there have been some nine independent establishments of symbioses with Frankia alone (H.-L. Li et al. 2015: three chloroplast genes, very good sampling), and at least eight more with the proteobacteria, particularly rhizobia, most in Fabaceae (e.g. Swensen 1996; Clawson et al. 2004; J. J. Doyle 2011; Santi et al. 2013; Werner et al. 2014: c.f. topology; J. J. Doyle 2016).

A final wrinkle is the possibility of an ancient hybridization involving the N-fixing clade and the malvids, the COM clade being the result, but since much of the nuclear genome of the COM clade seems to be malvid in origin, N-fixing in the COM clade would not be expected (Sun et al. 2015; see also discussion under the rosid clade).

Ecology & Physiology. H.-L. Li et al. (2015: n.b. stem-group ages for N-fixing clades - they are sometimes very stemmy!) suggested that associations with Frankia were established during two periods in the Late Cretaceous and Eocene when both global temperatures and atmospheric CO2 concentrations were high; by their estimates, no new Frankia relationships have been established since the beginning of the Oligocene. The ability to fix N has evolved several times in Fabaceae, and is certainly not an apomorphy for the family; ages when N fixation became established are

Nitrogen-fixing members of this clades grow in both tropical and temperate regions of the world and are very important in the global N cycle, moving N from the atmosphere into biological cycles; denitrifying bacteria accomplish the reverse. Thus over 100 kg N ha-1 y-1 can be added to the system (Carlsson & Huss-Danell 2003 and references), similarly, estimates of biological N fixation in terrestrial environments, much of which is by bacteria associated with Fabaceae, are 90% of the 100-140 Tg [= 1012g) of the total N fixed per annum (Gage 2004 and references). Once N has entered biological cycles, the activities of Metarhizium and related ascomycete fungi in moving N from insects eating plants back to plants should be taken into account; they may be another important element in the global N cycle (Behie & Bidochka 2014), along with nitrogen fixation in thunderstorms and denitrification in the soil.

Nitrogen-fixing plants are usually not members of closed lowland tropical rainforest communities, often growing in more open vegetation, even in early successional communities. Frankia-associated plants are woody, and favour high light and soils with low available N, and are early successional for the most part (H.-L. Li et al. 2015). There are major geographic patterns in the kinds of N fixation. Thus Frankia associations are generally obligate and are commoner in cooler areas of the globe, while rhizobial N fixation in Fabaceae is generally facultative and N-fixing fabaceous trees are uncommon north of 35oN; the costs of constructing, maintaining and regulating the N-fixing pathway interact with temperature and the obligate/facultative difference (Menge et al. 2014). Such ecological rather than evolutionary constraints may be the limiting factors in the distribution of N fixation (Menge & Crews 2016). Interestingly, bacteria related to Rhizobium, but which do not fix N, dominate in coniferous forests where N fixing plants are vanishingly uncommon (VanInseberghe et al. (2015).

N-fixing plants in general - although Fabaceae make up the majority of these - have very high concentrations of nitrogen in the leaf, and high leaf nitrogen has been linked with high levels of both carbon fixation and photosynthesis (see M. Adams et al. 2016). However, Adams et al. (2016) did not confirm these correlations, rather, water use efficiency was linked with leaf nitrogen in woody nitrogen-fixing plants. There is no evidence that Fabaceae in general have a high demand for nitrogen, moreover, inoculation of plants with crushed rhizobia affected plamt nitrogen concentrations independently of any fixation, which suggests a rather complex interaction between the plant and bacterium (Wolf et al. 2016). Haemoglobin is intimately involved in helping preserve the largely oxygen-free micro-environment the bacteria need for nitrogen fixation - nitrogenase is inactivated by oxygen; a variety of haemoglobins are involved, including haemoglobin synthesized by Frankia (Vessey et al. 2004).

The ability to fix nitrogen is uncommon elsewhere in seed plants, although nitrogen-fixing blue-green algae are associated with Gunneraceae and Cycadales. In vitro nitrogen fixation by Azotobacteria vinelandii endophytic in Mammillaria has been recorded (Lopez et al. 2011), and Burkholderia, a genus which can fix nitrogen in Fabaceae, forms associations in leaf nodules with some Primulaceae-Myrsinoideae and Rubiaceae, and may even fix nitrogen in sugar cane (de Carvalho et al. 2011).

Species forming ectomycorrhizal (ECM) associations are also common in the N-fixing clade, and this association has evolved here at least seven times (and also in other seed plants), but apparently not in Cucurbitales. Plants with ECM fungi rarely fix nitrogen, although Casuarinaceae are an exception. Interestingly, ECM associations also involve a perturbation of the nitrogen cycle in that nitrogen may move directly from humus to the fungus, and then to the plant, rather than moving in inorganic form from the soil into the root hairs (see also below).

Plant-Animal Interactions. There are associations of particular groups of butterflies and plants (as food sources for caterpillars) within Fabales and Rosales in particular (Ackery 1988, 1991). Indeed, it has been suggested, as by Scott (1985) and Janz and Nylin (1998; see also Braby & Trueman 2006) that the ancestral food plant for larvae of butterflies as a whole may perhaps have been somewhere in the N-ixing clade or the malvids or their (immediate) ancestors (Nylin et al. 2014), however, caterpillars are common in the latter only on Fabaceae, a fairly young, mostly Caenozoic group (see below); stem Fabales are rather older (see above). Malvales are another possibility (Ackery 1991), as are Rosids as a whole (e.g. Powell 1980; Berenbaum & Passoa 1999).

Bacterial/Fungal Associations. Jeong et al. (1999) and Clawson et al. (2004) compared phylogenetic relationships within Frankia with those of its hosts. Clawson et al. (2004) suggested that all three clades of Frankia that they recognised might have diverged before the evolution of the angiosperms, while Jeong et al. (1999) thought that the Frankia clades had diverged after the plant clades, being about one third the ages of the latter. However, their plant ages were extraordinarily old, thus the [Rosales [Fagales + Cucurbitales]] clade was estimated to be 429-199 m.y.o. (Jeong et al. 1999). Using age estimates of Bell et al. (2010), J. J. Doyle noted that the common ancestor of the N-fixing clade was about 100 m.y.o., but the first symbiosis in extant clades was likely to be at most ca 70 m.y.a. - a 30 m.y. lag. Particularly old N-fixing clades may be Datiscaceae and Elaeagnaceae, but since both have very long stems exactly when N-fixing actually evolved there is anyone's guess. For the considerable gene/genome divergence within Frankia, see Normand et al. (2006).

Intercellular penetration of the root epidermis by Frankia occurs in both Rosales and perhaps in Cucurbitales; in Fagales infection is through root hairs (Clawson et al. 2004; c.f. Santi et al. 2013); more derived kinds of entry are via cracks in the epidermis or by root hairs with the formation of infection threads (Svistoonoff et al. 2014). However, given the absence of strong phylogenetic structure in the group, details of how infection patterns map on to phylogeny are unclear (see also Soltis et al. 2005a; J. J. Doyle 2011). The situation is made more complicated because of the diversity of bacteria that have "independently" (but see below: gene casettes, symbiosis islands, etc.) become involved in nitrogen fixation in Fabaceae alone. Thus in Mimosa and some Fabaceae-Faboideae, at least, ß-proteobacteria like Burkholderia phymatum and Cupriavidus form nodules that fix nitrogen. The α-proteobacteria, rhizobia, that form nodules in other Fabaceae occur in four separate clades, over a dozen "species" being involved (J. J. Doyle 1998; Moulin et al. 2001; Sprent 2002; Elliott et al. 2007; Bontemps et al. 2010; Sprent et al. 2013).

Although there is considerable variation in nodule morphology, this does not correlate with bacterium type. In general in Fabaceae the nodules arise from the cortex and have peripheral vascular tissue with bacterial infected cells in the center; they are perhaps shoot-like structures (Hirsch & Larue 1997; Franche et al. 1998; Couzigou et al. 2012). All other nodules, whether associated with Frankia or rhizobia, are modified lateral roots, although they lack a root cap - the vascular tissue is central, initiation of the nodule is pericyclic, and it is the cortical cells that contain bacteria (Gualtieri & Bisseling 2000; Raven & Edwards 2001; Vessey et al. 2004). However, recent work shows that nodule origination in Faboideae occurs where lateral roots develop, although cortical cells may also be involved (op den Camp et al. 2011). Indeed, in nodule development in Faboideae there seems to have been co-option of genes originally involved in lateral root origination after a genome duplication event ca 54 m.y.a. (op den Camp et al. 2011; J. J. Doyle 2011). N-fixing clades such as Chamaecrista lack this duplication, although they may have another (Cannon et al. 2010, 2014), and overall any causal connection between duplication and nodulation is unclear (Cannon et al. 2014).

Ectomycorrhizal (ECM) associations are also common in the N-fixing clade, for instance, in Fabaceae-Detarioideae and Fagales, and ECM associations have been reported from a number of taxa which also harbour Frankia (e.g. Rose 1980).

Chemistry, Morphology, etc. Whether or not the presence of stipules is plesiomorphic in the clade depends in part on its phylogeny, but there is clearly a variety of structures borne at the node - and nodal anatomy is also variable (see below: de Aguiar-Dias et al. 2011).

Phylogeny. For the limits of the N-fixing clade, a rather unexpected group, see Chase et al. (1993, 1999), Savolainen et al. (1997), Soltis et al. (1995b, 1997, 1998), and Swensen (1996). Although it was not recovered by some analyses of the complete chloroplast genome (Bausher et al. 2006), the poor sampling - no other fabids/rosid I taxa were included - may well be reponsible; the nitrogen-fixing clade was also found not to be monophyletic in Duarte et al. (2010) or in the genome-level analysis of Burleigh et al. (2011). The clade had little support in the mitochondrial matR analysis of Zhu et al. (2007), but support was much strengthened when two chloroplast genes were added; it was also monophyletic in the mitochondrial analysis of Qiu et al. (2010; see also Z.-D. Chen et al. 2016). For further discussion of the relationships of the N-fixing clade, an ancestor of which was possibly involved in an ancient hybridization with the malvids, see the Zygophyllales page, however, much of the nuclear genome of the COM clade, the possible product of this hybridization, seems to be malvid in origin.

Relationships within the clade are still somewhat unclear (e.g. Qiu et al. 2010; Xue et al. 2012). Although Zhu et al. (2007) could not even recover a monophyletic Rosales using the mitochondrial matR gene, the other three orders were, however, all strongly supported, and in most other analyses all orders are well supported. Sytsma et al. (2002) recovered a topology [Cucurbitales [Fabales [Fagales + Rosales]]], while in Zhu et al. (2007: four genes) and Lee et al (2011: focus on nuclear genome) relationships were [Fab [C [Fag + R]]], albeit there was little support for the topologies found (see also Bell et al. 20100. However, Ravi et al. (2007) examining data sets including 61 protein-coding genes (for only three orders) and just four genes (Fagales included) found good support for [Fab + R] and some support for the broader grouping [C [Fag [Fab + R]]]. However, apart from Fabales (three Fabaceae-Faboideae included), the other orders were represented by single exemplars. A [Fab + R] clade was also obtained by Jansen et al. (2007) and Moore et al. (2007), but no Fagales were included. In other analyses there is some support for a [C + Fag] clade (see Chase et al. 1993; Setoguchi et al. 1999; Schwarzbach & Ricklefs 2000; Soltis et al. 2000, 2003a; Zhang et al. 2006). However, the support for the topology in the Summary Tree is quite strong (Jeong et al. 1999: no Fabaceae included; Moore et al. 2008, 2011; H. Wang et al. 2009; Soltis et al. 2011; Z. Wu et al. 2014; Foster et al. 2016a: support weak), but confirmation after e.g. increased taxon sampling would be comforting. Indeed, H.-L. Li et al. (2015) recovered the relationships [Fab [R [C + Fag]]] with very good generic sampling while L. Zhao (2016) found strong support for the topology [[Fab + Fag] [R + C]] in a large-scale nuclear gene analysis, although there of course sampling was rather exiguous.

FABALES Bromhead  Main Tree.

Ellagic acid 0; vessel elements with simple perforation plates; wood often fluorescing; nodes?; styloids +; K initiation helical; C clawed; tegmen ± crushed/disintegrating, embryo chlorophyllous. - 4 families, 754 genera, 20140 species.

Age. Wikström et al. (2001) date crown-group Fabales to (83-)79, 74(-71) m.y.a.; other estimates are (90-)87(-84) or (75-)72(-69) m.y. (two penalized likelihood dates), Bayesian relaxed clock estimates being slightly older, to (107.1-)104, 101.7(-91.6) m.y. ((H.-L. Li et al. 2015), 100 m.y. (Hengcheng Wang et al. 2009), ca 90.3 m.y. (Koenen et al. 2013) and ca 71.1 m.y. (Tank et al. 2015: Table S2).

Note: Boldface denotes possible apomorphies, (....) denotes a feature common in the clade, exact status uncertain, [....] includes explanatory material. Note that the particular node to which many characters, particularly the more cryptic ones, should be assigned is unclear. This is partly because homoplasy is very common, in addition, basic information for all too many characters is very incomplete, frequently coming from taxa well embedded in the clade of interest and so making the position of any putative apomorphy uncertain. Then there are the not-so-trivial issues of how character states are delimited and ancestral states are reconstructed (see above).

Evolution: Divergence & Distribution. Fabales contain ca 9.6% eudicot diversity (Magallón et al. 1999), of which the bulk is made up of Fabaceae. If [Quillajaceae + Surianaceae] are sister to Fabaceae, then a strong slow-down in the rate of evolution occurred in the former clade, although if Duparquetia is sister to the rest of Fabaceae, things get more complicated.

Bello et al. (2012) suggested a number of apomorphies for Fabales; Krameria was used as the outgroup because people in the past had suggested similarities between it and Polygalaceae, so most of the apomorphies listed are likely to be plesiomorphies. Despite the floral differences between Polygalaceae and Fabaceae, there are some developmental similarities between them (Prenner 2004d), however, the keel flowers in the two may have arisen independently (Bello et al. 2012). Indeeed, given the uncertainty over the relationships between the four families, optimisation of characters is a particularly fraught enterprise here.

Ecology & Physiology. About a quarter of all records of extra-floral nectaries come from members of this clade (Weber & Keeler 2013).

Chemistry, Morphology, etc. The distribution of a number of features may be of systematic significance in Fabales, but sampling is poor. Variation in nodal anatomy within Surianaceae is correlated with presence/absence of stipules. Although styloids, along with druses, are reported from Surianaceae, Quillajaceae and Fabaceae, details of their distribution within Fabaceae are unclear; they are certainly quite common in Faboideae (Lersten & Horner 2005), apparently less so in the rest of the family. The rpl22 gene is in the nucleus in Polygalaceae and Fabaceae (i.e. it is absent from the chloroplast), but other members of the order have not been studied and its presence in other angiosperms is sporadic (J. J. Doyle et al. 1995; Su et al. 2014; Dugas et al. 2015). Pollen grains of Quillajaceae and some Surianaceae have exine protruding at the apertures, and these and some Fabaceae-Cercidoideae (although perhaps derived within that group?) have striate pollen (Banks et al. 2003; Claxton et al. 2005); these features are unlikely to be high-level apomorphies. It would be nice to know if Surianaceae or Quillajaceae had starchy endosperm, and more details of their chemistry are needed. Many Fabaceae-Faboideae have lost the rps16 gene, and it is also absent from Polygala (Downie & Palmer 1992: again, sampling).

Phylogeny. Fabales as circumscribed here were rather unexpected, but they are quite strongly supported (Morgan et al. 1994; Källersjö et al. 1998; etc.). Although Hilu et al. (2003) found Larrea (Zygophyllaceae) to be weakly associated with Fabaceae, the latter were the only Fabales included in their rbcL analysis.

Within Fabales, Persson (2001) suggested the relationships [Polygalaceae [Surianaceae [Quillajaceae + Fabaceae]]], but with little support (this tree was used in versions 1-6 of this site), although elements of it were recently obtained by Cannon et al. (2014: Surianaceae not sampled) and Z.-D. Chen et al. (2016: Quillajaceae not sampled). Forest et al. (2002, see also Qiu et al. 2010) found weak support for the topology of the tree below, and Banks et al. (2008) also found strong support for the relationship [Q [the rest]]. However, Wojciechowski et al. (2004: ?sampling) suggested the possibility of a [S + Q] grouping. The unrooted topology in Bruneau et al. (2008a) is [P [Q + S] F]. Bello et al. (2009) in a careful analysis of matK and rbcL data, preferred the relationships [P [F [S + Q]]] obtained in a maximum parsimony analysis, however, support was poor, and if anything was still poorer for any relationships obtained in Bayesian analyses of the same data (see also M. Sun et al. 2016). Wang et al. (2009) did not obtain well supported relationships in Fabales in their twelve-gene analysis of the rosids, while in a megaphylogeny of angiosperms Smith et al. (2011) found some support for a clade [Q + F]. Relationships remained unclear in the study by Soltis et al. (2011). Bello returned to the problem and included morphological data with a two-gene data set. In various analyses [S + Q] were usually at least moderately supported, but support for Fabaceae and Polygalaceae as being successively sister to that pair was weak (Bello et al. 2012); the topology recovered by Bello et al. was also preferred by Cardoso et al. (2013b) and Koenen et al. (2013), although they focussed on Fabaceae. Cardoso et al. (2013b) also placed Vauquelina (Rosaceae!) in this area as part of a tritomy. That genus has indeed been associated with Quillaja in the past, but some time ago in an ITS phylogeny it grouped with Rhaphiolepis and Eriobotrya within Rosaceae-Amgydaloideae (Campbell et al. 1995). H.-L. Li et al. (2015, 2016) obtained the groupings [[S + Q] [P + F]], although support for the latter family pair was not good, and L. Zhao (2016: Surianaceae not included) found the relationships [Q [P + F]]. The bottom line: The topology of the tree used here may be wrong - a [S + Q] clade is quite often recovered!

Includes Fabaceae, Polygalaceae, Quillajaceae, Surianaceae.

Synonymy: Caesalpiniales Martius, Cassiales Link, Mimosales Link, Polygalales Berchtold & J. Presl, Quillajales Doweld, Surianales Doweld - Fabanae Reveal, Polygalanae Doweld - Polygalopsida Endlicher

[Quillajaceae + Surianaceae]: ? [If the clade exists.]

Age. The crown-group age of this clade is ca 56.8 m.y. (Koenen et al. 2013).

QUILLAJACEAE D. Don   Back to Fabales


Small evergreen tree; saponins, proanthodelphinidin, flavone C-glycosides +; cork cambium ?deep-seated; storying?; styloids in phloem +; nodes 1:3; petiole bundles arcuate, pericyclic fibres 0; mucilage cells +; hairs warty; leaves spiral, blade vernation conduplicate, margins toothed [hydathodal?], (entire), stipules petiolar; inflorescence terminal, cymose; hypanthium +; K valvate, nectary on lower half of K/hypanthium, C contorted, spathulate; A unidirectional in initiation, 5A opposite sepals above nectary + 5A opposite petals below nectary; pollen striate; G [5], deeply longitudinally ridged, opposite K, stigmatic zone elongated down short style branches; ovules several/carpel, apotropous to pleurotropous, in two marginal rows, micropyle?, outer integument ?3 cells across, inner integument?; fruit strongly asymmetrically lobed, follicular, opening down both surfaces of the lobes, K moderately accrescent; seeds winged; testa with 3 outer layers thickened, sclerotic; endosperm type?, cotyledons investing radicle, conduplicate; n = 14, 17, nuclear genome size [1C] ca 0.42 pg.

1[list]/2. Temperate South America, not Peru (map: from Donoso Z. 1994; Luebert 2013). [Photo - Flower, Fruit.]

Genes & Genomes. There is a genome duplication here (Cannon et al. 2014).

Chemistry, Morphology, etc. The leaves are amphistomatous. The flowers of Quillajaceae, with the distinctive arrangement of nectary and androecium, may be interpreted as having a hypanthium. Androecial development is unidirectional and is rather like that of Fabaceae (Bello et al. 2007/8); the carpels are definitely connate axially, but are largely free laterally, c.f. earlier versions of this site. There appear to be only three traces to each carpel, although Sterling (1969) noted that there were also "intermediate" bundles. Robertson (1974) noted that n = 17. Embryologically the family is poorly known.

See also Culham (2007) and Kubitzki (2006b) for general accounts, Hegnauer (1973, 1990, as Rosaceae) for chemistry, Marchiori et al. (2009) for wood anatomy (intercellular canals, included phloem), Lersten and Horner (2005) for vegetative anatomy, Péchoutre (1902, as Rosaceae) for seed morphology, Sterling (1969) and Kania (1973) for gynoecial morphology,. Additional data from: Aronson 7897 (anatomy, embryo).

Phylogeny. Cardoso et al. (2013b: matK phylogeny) placed Vauquelina in this area.

Previous Relationships. Quillaja was included in Rosaceae-Quillajoideae (Takhtajan 1997) or, more usually, in Spiraeaoideae, e.g. as Quillajeae (Robertson 1974). It is indeed superficially quite similar to the South American Kageneckia (Spiraeaoideae), but wood anatomical data, etc., suggest that it should be removed from Rosaceae (Lotova & Timonin 1999; c.f. Zhang 1992).

[Fabaceae [Surianaceae + Polygalaceae]]: suspensor persistent, also connected to wall of embryo sac [?Surianaceae].

Age. Crown group ages of this clade are (82-)79, 74(-71) m.y. (Wikström et al. 2001) or ca 70.6 m.y. (Naumann et al. 2013).

[Fabaceae + Surianaceae]: ? (if this clade exists)

Age. The crown-group age is ca 70 m.y. (Tank et al. 2015: Table S1, S2).

FABACEAE Lindley, nom. cons. // LEGUMINOSAE Jussieu, nom. cons. et nom. alt.   Back to Fabales

Trees or lianes to annual herbs; (coumarins/furanocoumarins), non-protein amino acids, esp in seeds (0), (cyanogenic glucosides) +, lectins [haemagglutinins] and gums esp. in seeds, 5-deoxyflavonoids, C-glycosylflavonoids, pinitol [cyclitol] +; cork also in outer cortex; (wood storied); secretory cells common, sieve tube plastids with protein crystals (and/or starch, or simply starch); parenchyma (diffuse), (terminal), (rays heterocellular); nodes 3:3; cuticle wax platelets as rosettes; stomata various; branching from previous flush; colleters +, hairs often uniseriate (mesifixed); leaves compound, apex of petiole and petiolules pulvinate, leaflets opposite (alternate), blade with conduplicate vernation, (secondary veins palmate), (stipellate), stipules +, lateral; inflorescence racemose, bracteoles often v. small; flowers monosymmetric, resupinate [median sepal abaxial], (3-)5(-6)-merous, floral developmental sequence K-G-C-outer whorl A-inner whorl A [G initiation/development much advanced]; K ± connate to free, (adaxial-median C with patterning/different in colour from the other C); A unidirectional in initiation, 10, heteranthy common, filaments connate to free, anthers basifixed to dorsifixed; tapetal cells bi(multi)nucleate; exine columellate; G 1, stipitate, stylulus long, (hollow), young stylulus adaxially curved (straight), stigma wet; compitum necessarily 0; ovules 1-several/carpel, one-ranked, micropyle zig-zag, outer integument 2-10 cells across, inner integument 2-3 cells across, parietal tissue to 5 cells across, nucellar cap 2-3 cells across, hypostase +, funicle long; (megaspore mother cells several), antipodal cells persistent; chalazal embryo haustorium +; fruit follicular and dehiscing abaxially also (indehiscent); seed symmetric, with radicular projection, raphe and antiraphe ± same length, vascular bundle in antiraphe; exotesta palisade, linea lucida + [line separating much thickened outer anticlinal walls from the thinner inner walls], area of cells with a deep-seated linea lucida (0), mesotesta of stellate/hourglass cells; endosperm cells thick-walled, with galactomannans [= Schleimendosperm]; embryo ± straight, cotyledons investing radicle, accumbent, venation palmate, (cell walls thick), (with amyloid), (starch +); x = 6, 7; plastid rpl22 gene transferred to nucleus, rps19 pseudogene present.


766[list, all except Faboideae, q.v.]/19,580 - discussed under six main groups below. World-wide.

Age. Wikström et al. (2001, 2004) date the crown group to (71-)68(-65) or (59-)56(-53) m.y.a.; Bruneau et al. (2008a, b; slightly younger estimates in Bello et al. 2009) thought that Fabaceae began diversifying in the Palaeocene ca 64 m.y. ago. Crown Fabaceae are dated to ca 59 m.y.a. by Lavin et al. (2005), (77-)63, 61(-47) m.y. by Bell et al. (2010), (87.1-)80.6, 56.8(-48) m.y. by Pfeil and Crisp (2008), and ca 92.9 m.y. by Hohmann et al. (2015). These ages are based on Cercidoideae being sister to all other Fabaceae; Duparquetia was not included. Magallón et al. (2015) date stem-group Fabaceae (Surianaceae are sister) to ca 92.1 m.y. ago, while an estimate for the legume-common tetraploidy event is ca 56.6 m.y.a. (J. Wang et al. 2017) or ca 58 m.y. (Schmutz et al. 2010).

1. Cercidoideae Legume Phylogeny Working Group

Trees, shrubs, lianes climbing by branch tendrils, (prickles, spines +); (distinctive secondary thickening), (intraxylary phloem/bicollateral vascular bundles +); (plant with prickes or spines); leaves basically even pinnate, leaves apparently simple, (bilobed or not), or bifoliolate, leaflets opposite, with single pulvinus; (flower papilionate); hypanthium +, ± elongated (0); (K connate), C with adaxial-median member innermost [ascending cochleate]; (filaments partly connate), anthers dorsifixed; pollen various, (in tetrads), (3-6-colpate), etc., surface variable; (G stipe adnate to hypanthium); (funicle short), parietal tissue to 20 cells across, (nucellar beak +), nucellar cap to 10 cells across; (fruit samara), seeds 1-several/fruit, (asymmetric), (post-chalazal vascular bundle 0), hilum apical, crescentic, (circular - Cercis), lens inconspicuous; mesotesta lacking stellate/hourglass cells; (post-chalazal vascular bundle 0); (endosperm with galactoglucomannans - Cercis), (embryo curved); n = 7, 12-14, etc.

12/335: Bauhinia s.l. (250), Schnella (47). Pantropical (temperate) (map: from Meusel et al. 1965; Sales & Hedge 1996; Trop. Afr. Fl. Pl. Ecol. Distr. 3. 2008). [Photo - Bauhinia, Cercis, © D. Kimbler.]

Age. The age of crown-group Cercidoideae is estimated at ca 34 m.y. (Lavin et al. 2005), ca 47.3 m.y.a. (Bruneau et al. 2008a), or ca 62.7 m.y. (Meng et al. 2014).

See Meng et al. (2014) for a list of fossils of Bauhinia.

Synonymy: Bauhiniaceae Martynov

2. Detarioideae Burmeister

Trees (shrubs); plant ectomycorrhizal; (resins +, with bicyclic diterpenes); pits vestured; leaf phloem transfer cells + (0); leaves equal pinnate, leaflets opposite or alternate, (crater-like extra-floral nectaries on the abaxial surface), stipules intrapetiolar, connate or not, (lateral); inflorescences branched; bracteoles large, ± surrounding the bud, ± connate or ± adnate to hypanthium, valvate or imbricate, (small); hypanthium +, ± elongated (0); (flowers polysymmetric); K (petal-like), (4), ((2 adaxial) + 3), C (0-7) [adaxial C the last to go], adaxial-median member outermost [descending cochleate]; A (2-many), initiation time of the two whorls overlapping, (ring meristem +), (filaments partly connate), anthers dorsifixed or basifixed; pollen surface variable, (pectic substances below aperture - Zwischenkörper/oncus); G (stipe adnate to hypanthium), young stylulus abaxially curved; (seed arillate); seed coat undifferentiated [= overgrown, "seeds have the nature of tumours"]; endosperm 0, cotyledons walls commonly thick, amyloid +, with xyloglucans; x = (8, 10, 11) 12, etc..

84/760: Cynometra (85), Macrolobium (75), Crudia (55), Copaifera (35). Tropical.

Age. The crown age of this clade is estimated to be ca 29.2 m.y. (Lavin et al. 2005) or ca 53.6 m.y., but only ca 17.3 m.y. when there were no constraints (Bruneau et al. 2008a; c.f. de la Estrella 2017 - 66 m.y.), or 68-64 m.y. (de la Estrella 2017).

Synonymy: Detariaceae J. Hess, Tamarindaceae Martinov

3. Duparquetioideae Legume Phylogeny Working Group

Liane; ?chemistry; wood not storied; leaves odd pinnate, leaflets opposite; inflorescence terminal; hypanthium 0; K 4 [adaxial member missing], free, ± petal-like, lateral K bilobed, C development simultaneous, enclosing the rest of the flower in bud, fringed with glands, adaxial-median member outermost [descending cochleate], two abaxial strap-like; A 4, opposite K, free, anthers basifixed, postgenitally connate, porose, thecae apically aristate; pollen asymmetric, ectoaperture encircling the equator, with two endoapertures; G initiation not advanced relative to that of other organs [= development "normal"]; 2-5 ovules/carpel; endocarp hairy; seeds 2-5/fruit; ?galactomannans, radicle ± obscured by cotyledons; n = ?

1/1: Duparquetia orchidacea. Tropical W. Africa.

Age. The stem-group age of Duparquetia is ca 62.1 m.y. (Koenen et al. 2013: c.f. topology).

4. Dialioideae Legume Phylogeny Working Group

Trees (shrubs); ?chemistry; (pits usu. not vestured); (stomata paracytic); (leaf phloem transfer cells +); leaves (2-ranked - Poeppigia), odd pinnate, (leaflets usu. alternate); (stipules 0); inflorescences thyrsoid, with cymose branches, (racemose), (flowers single, axillary); relative timing of organ formation variable; (flowers papilionate), (syammetrical); (hypanthium usu. 0), (receptacle broad, flattened); K (3-4), free (connate - Poeppigia), C (0-4), (usu. not clawed), with adaxial-median member innermost [ascending cochleate]; A (2-)5(-10) [whorl opposite K present], free, anthers usu. basifixed, usu. porose; (G 2), (not stipitate), style glabrous, (complex, petaloid), stigma ± punctate (peltate); ovules (1-)2(-many)/carpel; fruit indehiscent, (dehiscent), usu. drupe or (narrowly winged) samara; seeds 1-2(-more)/fruit; (seed coat undifferentiated); n = 14.

17/85: Dialium (28). Pantropical.

Age. The crown-group age of this clade is ca 34 m.y. (Bruneau et al. 2008a).

[Caesalpinioideae + Faboideae]: (roots with indeterminate, branched, N-fixing nodules, fixation threads + [or Caesalpinioideae?]), lacking hypodermis [?level]; pits vestured; (stipules 0), stipels +/0; (hypanthium 0); anthers basifixed or dorsifixed; (fruit a drupe, samara, schizocarp, etc.); chalazal endosperm haustoria + [?level]; (seed arillate).

Age. This node has been dated to (62-)59, 53(-50) or (36-)34(-31) m.y.a. (Wikström et al. 2001), (67-)50, 49(-30) m.y. (Bell et al. 2010), ca 61.3 m.y.a. (Bruneau et al. 2008a), or ca 55 m.y. (Lavin et al. 2005; quite similar ages in Bouchenak-Khelladi et al. 2010b).

5. Caesalpinioideae de Candolle

Shrubs, trees, lianes (herbs); sieve tube plastids also with fibres; extrafloral nectaries common, on petiole/rhachis; leaves bicompound (not), usu. even pinnate, leaflets usu. opposite; hypanthium cupular, (G adnate to side of hypanthium); C with adaxial-median member innermost [ascending cochleate]; (stigma porose/cup-shaped), ± punctate; ovules usu. campylotropous [up a level?], outer integument with vascular strand; seed (aril 0), funicle long and thin to stout and thick, pleurogram + [fracture line in xeotesta], ± O-shaped [closed], (0, several); whole nuclear genome duplication, sucrose synthase gene duplicated.

148/ca 4,400. Predominantly tropical, esp. Africa and America. [Photos - Collection.]

Age. The [Umtiza group + The Rest] clade has been dated to ca 58.6 m.y.a. (Bruneau et al. 2008a).

[Umtiza + Ceratonia]: ? leaves pinnate.

Age. The age of the Umtiza group is ca 58-56 m.y. (Bruneau et al. 2008a).

5a. Umtiza clade


Age. The crown Umtiza group is 55-52 m.y.o. (Bruneau et al. 2008a).

5b. Ceratonia clade

(Leaves bipinnate - Acrocarpus); seed coat undifferentiated, pseudopleurogram + {no fracture of exotesta]; genome size [1C] ca 0.57 pg.


Age. The age of this clade is around 45 m.y. (Bruneau et al. 2008a).

Synonymy: Ceratoniaceae Link

[[Cassia + Caesalpinia] [mimosoid clade and things]]: ?

[Cassia + Caesalpinia]: ?

5c. Cassieae Bronn

(N-fixing nodules +, rhizobia usu. in infection threads and symbiosomes - Chamaecrista); (vestured pits 0 - Labicheinae); (inflorescence cymose - Chamaecrista); micropyle zig-zag, outer integument 3-9 cells across, inner integument 2-3 cells across, parietal tissue ca 6 cells across, hypostase +; ± elliptic lens, (pleurogram 0, several); suspensor poorly developed.

Senna (295-350), Chamaecrista (265), Pterogyne, Vouacapoua.

Age. Cassieae are around 53 m.y.o. (Bruneau et al. 2008a).

Synonymy: Cassiaceae Vest

5d. Caesalpinieae Reichenbach

(Prickles, spines, or glandular hairs), (foliar glandular idioblasts +); leaves bipinnate (pinnate), with terminal pinna or not; flowers monosymmetric; abaxial K often somewhat modified, adaxial C often patterned; A ± surrounding G.

27/. Stuhlmannia, Mezoneuron.

Age. The Caesalpinia group is about 56 m.y.o. (Bruneau et al. 2008a).

Synonymy: Caesalpiniaceae R. Brown

[Mimosoid clade and things]: ?

Age. This clade is about 56.3 m.y.o. (Bruneau et al. 2008a).

The following four groups form a polytomy: bracteoles 0; K imbricate; C imbricate.

5e. Tachigalieae Nakai

N-fixing nodules + [with fixation threads].

Tachigali (60).

5f. Peltophorum clade

Bussea, Delonix.

5g. Dimorphandreae Bentham / Dimorphandra Group A

(N-fixing nodules + [with fixation threads]); median sepal adaxial; C protective in bud; staminodes antesepalous.

6/39 (?7/57): Dimorphandra (26). Tropical. Burkea, Dinizia, Erythrophloeum, Mora. [N-fixing - Campsiand, Jacqueshub, Melanoxyl, Moldenhau]

5h. The mimosoid clade (= Mimosoideae de Candolle)

mimosoid clade

Shrubs or trees (herbs); N-fixing nodules with infection threads, rhizobia in membrane-bounded symbiosomes (0); albizziine [non-protein amino acid] +, exudates mostly gums; sieve tube plastids also with fibres; (septate fibres +; aliform axial parenchyma +); rays usu. 20< cells high; petiolar extrafloral nectaries common; inflorescences dense, usu. ± capitate, flowers opening together, organ initiation in all flowers of the one head beginning simultaneously; flowers rather small, polysymmetric, hypanthium often 0; K connate, median sepal adaxial, also often valvate, (much reduced), C protective in bud, (not), initiation simultaneous, usu. valvate, connate (free), not clawed, not patterned; A often connate, (heteranthy +), (many, from ring primordium), (adnate to C), anther with terminal gland (0); endothecial cells with base plate, tapetal cells uninucleate; pollen tetrads/polyads common; (nectary 0); (G 1< [Inga; if 5, opposite K - Archidendron lucyi]), stigma (dry - one record), cup-shaped, (peltate); (nucellus apex exposed); seed (arillate), funicle long, thin; testa with vascular strand, pleurogram U-shaped [= open], (0); suspensor vestigial at cotyledon stage, detached from wall, cotyledons ± cover radicle.

82/3335: Acacia s. str. (1030), Mimosa (450), Inga (350), Vachellia (161), Calliandra (150), Senegalia (85), Abarema (45), Prosopis (45), Pithecellobium (40). Esp. tropical and warm temperate, esp. Africa and America (map: from Vester 1940; Maslin et al. 2003; Trop. Afr. Fl. Pl. Ecol. Distr. 3. 2008a). [Photos - Collection.]

Age. This mimosoid clade has been dated to (49.5-)42.4, 40.5(-31.3) m.y. (Lavin et al. 2005: inc. Pentaclethra; ages in Bouchenak-Khelladi et al. (2010b) are (61-)59.5(-58) m.y., ca 46 m.y.a. in Bruneau et al. (2008a), while (62.7-)51.4(-15.0) m.y. is the estimate in Miller et al. (2013).

Synonymy: Acaciaceae E. Meyer, Mimosaceae R. Brown

6. Faboideae Rudd / Papilionoideae de Candolle, nom. alt.


N-fixing nodules + (0), growth usu. determinate; isoflavonoids [pterocarpans and isoflavans], prenylated flavonoids, (indolizidine and quinolizidine alkaloids) +, exudates mostly gums; wood often ring porous, vessels with helical thickenings, vascular tissue storied (not); sieve tubes with spindle-shaped non-dispersive protein bodies [forisomes]; (cork cambium deep seated); leaves odd or even pinnate, leaflets opposite or alternate; flowers usu. papilionoid, adaxial-median member outermost [ascending cochleate]; ?tapetal cells; pollen (porate), (colpate), (endexine ± 0), (exine granular); stigma semi-dry; seed 1-many/fruit, vascularization various, raphe shorter than the antiraphe, hilum complex, long, with hilar groove, micropyle conspicuous [discoloured] (not) at one end, lens at the other; counter palisade +, tracheid bar in subhilar tissue, exotestal cells lacking deep-seated linea lucida; embryo curved, (straight), radicle long, cotyledons do not cover radicle; whole nuclear genome duplication.

503[list: in progress]/ca 14,000. World-wide, esp. (warm) temperate (map: from Vester 1940; Meusel et al. 1965; Hultén 1971; Trop. Afr. Fl. Pl. Ecol. Distr. 3. 2008a, 4. 2008b). [Photo - Flower, Fruit, Collection.]

Age. Crown group Faboideae may be about 59 m.y.o. (Lavin et al. 2005: age rather sensitive to the age of Fabaceae as a whole), ca 45 m.y. (Bruneau et al. 2008a), or ca 55 m.y.o. (Cannon et al. 2014).

6a. {Angylocalyceae [Dipterygeae + Amburaneae]] / ADA Clade: ?

6a1. Angylocalyceae (Yakovlev) Cardoso et al.

K, hypanthium enlarged, C thickened, red; A exserted..

5/21. Africa, northern South America, Castanospermum eastern Australia, western Pacific,

[Dipterygeae + Amburaneae]: ?

6a2. Dipterygeae Polhill

Flowers papilionaceous (not Monopteryx); 2 adaxial K much enlarged, petal-like, 3 abaxial K small teeth.

4/25. Neotropical.

6a3. Amburaneae Nakai

(Plants with balsam, resin); (leaflets glandular-punctate).

8/30. Neotropical.

6b. Swartzieae de Candolle


Trees, shrubs; nodule growth?; (leaflets alternate), stipels +/0; hypanthium 0; K completely connate, opening irregularly, C 1 (0, 2); A many, from ring meristem, development centripetal or centrifugal, heteranthy usu. notable, (basifixed); exine of tectum only; nectary 0; (G 2-4), long-stipitate; ovules anatropous, micropyle zig-zag, outer integument 6-8 cells across, inner integument 5-6 cells across, ?parietal tissue; seed arillate or not; (coat thin, cracking); (embryo straight); n = 14.

8/213: Swartzia (180). Mostly Central and South America; Bobgunnia Africa and Madagascar (Map: from Cowan 1967; Kirkbride & Wiersema 1997).

Age. The age for a crown group around here is 48.9±2.8 m.y. (Lavin et al. 2005: inc. Ateleia) or ca 45 m.y.a. (Bruneau et al. 2008a: inc. Lecointea).

Synonymy: Swartziaceae Bartling

Cladrastis + 50 kb inversion clade: ?

6c. Cladrastis s.l.

Flowers papilionaceous.

1-3/17. East Asia, W. and E. North America.

50 KB inversion clade.

Herbs, vines (lianes, trees, shrubs); (N-fixing nodules 0); (ectomycorrhizae +); (rotenone +), (quinolizidine alkaloids +); (styloids +); (leaves 2+ compound), (palmate); hypanthium usu. 0; K, C, A with unidirectional [abaxial to adaxial] initiation; K connate, adaxial-median C outermost [= decending cochleate], 2 abaxial C connate; A connate [e.g. 9 + 1], (not); tapetal cells uninucleate; (pollen porate); ovules usu. campylotropous, (endothelium +), (nucellar endothelium +), funicle short; seed asymmetric, not arillate; testa (multiplicative); cotyledons not investing radicle, cotyledon areole +, (xyloglucans +, starch in embryo), (endosperm 0); 50 kb inversion in trnL intron in chloroplast LSC, (rps16 gene absent; ORF184 absent), duplication of CYC gene; (one much-enlarged pleiomorphic N-fixing bacterium in symbiosome).

6d. Exostyleae Nakai

Leaflets serrate/spinescent (leaves unifoliate); flowers poly-/bisymmetric; A basifixed; fruit drupaceous.

6/21. Neotropical.

6e. Vataireoids

Leaves in groups at the ends of the twigs; fruit a samara; 400bp deletion in trnl-f intergenic spacer.

4/27: Neotropics.

6f. Genistoids

Quinolizidine alkaloids + (0).

Ormosieae Yakolev: 6/140: Ormosia (130).

Brongniartieae Hutchinson

N-fixing nodules with fixation threads; colleter-like glands in axils of stipules or on leaflet pulvinuli; flowers papilionate; K bilabiate; A diadelphous, anthers distinctly dimorphic [short dorsifixed, long basifixed]; (seed 1).

15/155: Brogniartia (65), Hovea (33). ± tropical, America (inc. Cuba), Australia, Haplormosia W. Africa.

Leptolobieae (Bentham) Cardoso et al.

Colleter-like glands in axils of stipules or on leaflet pulvinuli; flowers weakly or not papilionate; A free.


Camoensieae (Yakolev) Cardoso

Lianes; flowers huge [for a pea; to ca 20 cm across], hypanthium long; petals crimped, ± spreading; stamens free.

1/2. Africa (Gulf of Guinea).

Sophoreae de Candolle

Shrub or tree; quinolizidine alkaloids; exine of tectum only; antiraphe bundle 0; 14/122: Sophora (50).

Podalyrieae Bentham: 9/130: Amphihales (42).

Synonymy: Inocarpaceae Berchtold & J. Presl, Sophoraceae Berchtold & J. Presl

Crotalarieae + Genisteae]: ?

Crotalarieae Hutchinson

Pyrrolizidine alkaloids +; (stigma wet - Crotalaria).

16/1,225: Crotalaria (700), Aspalathus (280), Lotononis (90), Lebordea (51).

Synonymy: Aspalathaceae Martynov

Genisteae Bronn

αpyridone alkaloids, 5-0-methylgenistein [isoflavone] +; (nodes 1:1. - Ulex s. str.); K bilabiate; anthers dimorphic, filaments connate; aril 0/on short side of seed[?].

25/618: Lupinus (275), Genista (90), Argyrolobium (80), Cytisus (65). Mostly North Temperate.

Synonymy: Cytisaceae Berchtold & J. Presl

Andira clade

Leaves in fascicles; flowers papilionoid; fruit indehiscent.

2/46:Andira (29). Neotropical.

g. Dalbergioids s.l.

Synonymy: Geoffroeaceae Martius

Amorpheae Boriss.

8/247: Dalea (150).

Synonymy: Daleaceae Berchtold & J. Presl

Dalbergieae de Candolle

/1,370: Adesmia (240-425), Aeschynomene (250), Dalbergia (250), Machaerium (130), Zornia (75), Arachis (70), Stylosanthes (48).

Synonymy: Dalbergiaceae Martinov

[Baphieae + NPAA Clade]

h. Baphieae Yakovlev

7/57: Baphia (47).

Non-protein amino acid accumulating clade / NPAAA clade: alkaloids 0; non-protein amino acid accumulation [e.g. canavanine]; whole genome duplication?

Age. The age of this clade is around 61 m.y. (Snak et al. 2016) or 54.3 ± 0.6 m.y.a. (Lavin et al. 2005). A genome duplication in this area has been dated to ca 54 m.y.a. (op den Camp et al. 2011; Q.-G. Li et al. 2013), (67-)63.7, 57(-56) m.y.a. (Vanneste et al 2014a), or ca 53 m.y.a., although exactly where this duplication is to be placed is unclear (Murat et al. 2015b: x= 12).

MIRB: (giant antipodal cells +).

Indig. + Mill + Rob. + IRL Clade :

Age. The age of this clade is ca 59.1 m.y. (Koenen et al. 2013) or ca 67 m.y. (Snak et al. 2016); a rather remarkable suggestion is the 102 m.y. in Z. Wu et al. (2014).

MILL: inflorescence racemose, or pseudoracemose with 2 or more flowers/node; suspensor very large [100< cells], club-shaped [Phaseolus], (embryo straight); x = 10, 11.

If a Phaseoleae (Bronn) de Candolle - Vines (shrubs); canavanine 0 (+) [core Phaseoleae]; leaves trifoliolate, (stipules peltate); 89: 1,600. Desmodieae sister.

Age. The crown age of MILL is estimated to be around 44.3 m.y. (H. Li et al. 2013), (51.5-)48.5(-42.3) m.y. (Sirichamorn et al. 2014a), or ca 55 m.y. (Snak et al. 2016: Ind + rest).

Pultenaea (470: MIRB), Daviesia (135), Gastrolobium s.l. (110), Jacksonia (75), Bossiaea (60).

Old World Clade [inc. millettioids s.l., not baphiioids]: homoglutathione + [= γ-glutamyl-cysteinyl-βalanine], antiraphe bundle 0 [Canavalia]; albumen-1 gene.

Age. The age of this clade - but note topology - has been estimated at around 79 m.y. (Hohmann et al. 2014) (Lavin et al. 2005).

Indigofera (750: aff. MILL, mesifixed hairs), Tephrosia (350), Desmodium (300), Rhynchosia (230), Eriosema (150), Millettia (150), Erythrina (110), Mucuna (105), Lonchocarpus (100), Vigna (90+), Phaseolus (75), Psoralea (70), Canavalia (60), Clitoria (60), Dolichos (60), Galactia (60), Derris (55).

Synonymy: Galedupaceae Martynov, Phaseolaceae Martius

Hologalegina clade / temperate herbaecous group / [[Loteae + Robineae] IRLC clade]: x = 7-8.

Age. Hologalegina have been dated to 56±0.9 m.y. (Lavin et al. 2005).

foliaceous cotyledons. Lotus (inc. Coronilla: 125: ROB), Sesbania (60: ROB).

Synonymy: Coronillaceae Martynov

Age. Crown robinioids are some 48.3±1 m.y.o. (Lavin et al. 2005).

Synonymy: Lotaceae Oken, Robiniaceae Vest

Inverted Repeat Loss Clade / IRL clade: Annual to perennial herbs, (vines; woody lianes); (non-cyanogenic β- and γ-hydroxynitrile glucosides [Lotus] +); bacterioid differentiation irreversible [they cannot divide]; leaves (trifoliolate, palmate), pulvini 0, stipules adnate to the petiole; FLO/LFY genes expressed also in the leaf; CA primordia +; A initiation bidirectional, overlap in the timing of C, A, and G initiation; (stigma wet); tapetal cells uninucleate; outer integument 3-5 cells across, endothelium +; chloroplast inverted repeat lost, (rps16 gene, rps12 intron and clpP intron 1 absent).

Ca 45/4500. Astragalus (2910) Oxytropis (300), Trifolium (240), Vicia (215), Hedysarum (160), Lathyrus (150), Onobrychis (130), Caragana (100), Medicago (87), Swainsonia (85), Ononis (75), Lessertia (50). Especially northen and temperate.

Age. The IRL clade is estimated to be 39±2.4 m.y.o. (Lavin et al. 2005).

Synonymy: Astragalaceae Berchtold & J. Presl, Ciceraceae W. Steele, Hedysaraceae Oken, Lathyraceae Burnett, Trifoliaceae Berchtold & J. Presl, Viciaceae Oken

Synonymy: Papilionaceae Giseke

Evolution: Divergence & Distribution. For the fossil record of Fabaceae, see Herendeen and Dilcher (1992). For fossils of Cercis, see Jia and Manchester (2014); the oldest, from Oregon, date from about 26 m.y. ago. Bruneau et al. (2008a, b) thought that the major clades in Fabaceae had separated by 58-55 m.y.a.; the crown ages of the major clades are 56-34 m.y., but note that constrained (given above) and unconstrained ages differed considerably, the latter sometimes being one seventh of the former (Bruneau et al. (2008a). Ages for nodes throughout the family are given by Lavin et al. (2005) and Koenen et al. (2013).

Fabaceae are a notably speciose clade, particularly the Caesalpinioideae (esp. the mimosoid clade) and Faboideae (esp. the inverted repeat loss clade) (Magallón & Sanderson 2001), and contain ca 9.4% of eudicots. Koenen et al. (2013) note that by practically any measure of success, Fabaceae are successful - which is pretty much a fair statement, as we shall see. In general, the great diversity of the family can be explained by number of separate radiations rather than any great One Thing. Thus even the very speciose Astragalus is not particularly notably so when considered in the context of its immediate relatives (Sanderson & Wojciechoswki 1996).

Floral morphology and development is particularly variable in Fabaceae, and the numerous papers by Shirley Tucker are an essential starting point for any understanding here (e.g. Tucker 1987, 1989, 2003a; Tucker & Douglas 1994: phylogeny, for more general accounts). Clarifying basal relationships in the family will be particularly important here. Mirror image flowers are scattered in non-mimosoid Fabaceae (Tucker 1996b). Details of hypanthial evolution within Fabaceae are unclear; it seems to have become much reduced and lost several times. The "normal" (for flowering plants) floral orientation of the mimosoid clade with the median sepal adaxial and the median petal abaxial is secondary; in some 4-merous members of the mimosoid clade the median petal is adaxial (Prenner 2011). Although the normal orientation is also found in Caesalpininioideae like Ceratonia, the inverted orientation occurs in Cercidoideae (see Tucker 1989; Herendeen et al. 2003; Luckow et al. 2005), Duparquetia, many Caesalpinioideae, and Faboideae. Duparquetia, near basal in the family, has highly derived flowers, while in the Faboideae there are a number of near basal clades that include taxa with polysymmetric flowers and numerous stamens and taxa with papilionoid flowers (e.g. Cardoso 2012a, b), and near-basal in the 50 kb inversion clade is Aldina, with polysymmetric flowers, sister to [Andira + Hymenolobium], with papilionoid flowers (Ramos et al. 2015). Dialioideae vary considerably in the neumbers of their parts and in floral symmetry (Zimmerman et al. 2017 and references). All this suggests an early "experimental phase" in the evolution of floral morphologies in the family with a lack of canalization in floral development (Prenner & Klitgaard 2008; see also e.g. Polhill et al. 1981; Tucker 2001, 2003a, c; Ramos et al. 2015; Zimmerman et al. 2017). Over half the species in the family are in the Faboideae-PAAA clade (mostly papilionoid flowers) and the Caesalpinioideae-mimosoid clade (polysymmetric brush flowers), both deeply embedded in the phylogeny; for the great diversity of floral morphology in other parts of the family, the images in L.P.W.G. (2017) are a good introduction. Endress (2012) suggested that floral asymmetry was a key innovation in Phaseoleae.

One way to think about the diversification of Fabaceae and their distribution is in terms of vicariance of biomes rather than of classic geographical areas (Lavin et al. 2004; Schrire et al. 2005). Indeed, a Gondwanan age for Amherstieae (= Detarioideae), and so a proportionally older age for the family as a whole, suggested because of their amphiatlantic distribution and common possession of ectomycorrhizae (Henkel et al. 2002), seems unlikely. Changes in diversification rates of clades can quite frequently be linked to biome shifts (Koenen et al. 2013). Fabaceae, many of which are deciduous, are very important components of the vegetation of seasonally dry tropical forests (= the succulent biome) and savannas (Oliveira-Filho et al. 2013; see also Schrire et al. 2015; Pennington & Lavin 2016: stem ages of species tend to be old in such habitats), indeed, in the neotropics species diversity of Fabaceae is correlated with temperature (Punyasena et al. 2008). Long distance dispersal has also often been invoked to explain distributions in the family. There are a number of transoceanic disjunctions within the family, and 51/59 of those listed by Schrire et al. (2005) are only 1-22 m.y. old (see also Bouchenak-Khelladi et al. 2010b; Vatanparast et al. 2013: Dalbergia); the North Atlantic land bridge may have been important in the Caenozoic dispersal of the family (Lavin et al. 2000). One particularly interesting connection is in Hymenaea, whose resin forms valuable amber deposits. The majority of the species are New World, but there is a single Old World species, H. verrucosa. Both the Mexican and Dominican ambers seems to have been produced by extinct species whose immediate relationships are likely to be with H. verrucosa (Poinar & Brown 2002). Schaefer et al. (2012) estimated that there had been as many as 7 long distance dispersal events per 10 million years in Fabeae.

Diversification in Cercis began ca 35 m.y.a. and spread was from east to west - and across the Atlantic (Fritsch & Cruz 2012). Cercidoideae may be Tethyan in origin, initially favouring seasonally dry habitat (Sinou et al. 2009, c.f. Meng et al. 2014, but sampling exiguous, note dates). Marazzi and Sanderson (2010) suggest an age of 53-47.5 m.y. for stem group Senna, (47-)45(-41.7) m.y. for the speciose crown group. Early divergence within the mimosoid clade seems to have occurred in Africa (Bouchenak-Khelladi et al. 2010b). The age of the crown-group clade [Ingeae + Acacieae] is estimated at (24.5-)21(-17.2) m.y. by de Souza et al (2013), q.v. for other dates in the Calliandra area. Inga, with some 350 species, has diversified in the New World LTRF that it prefers only within the last 2 m.y. (Richardson et al. 2001b; Pennington et al. 2009; Dexter et al. 2010: thoughts on species limits; see also Iganci et al. 2015 for Abarema); it and the detarioid Tachigali, also quite diverse there, have notably short generation times (Baker et al. 2014). For diversification in Mimosa, a two-step affair, see Koenen et al. (2013). Estimates of the age of crown-group Acacia, some 1,000+ species, range from 26.6-3.3 m.y.a., the older age being driven by recent fossil discoveries (Miller et al. 2013, q.v. for dates of the other ex Acacia genera, etc.). González-Orozco et al. (2013) plotted the distribution of Australian Acacia (practically the whole genus) in the context of various environmental factors, and they thought that climatic variables were correlated with the largest-scale patterns; Mishler et al. (2014) looked at patterns of endemism taking into account endemism and age of clades, and found areas of palaeoendemism scattered in the continent, with neoendemism particularly apparent in soutwestern Australia in particular. The very close similarity between A. koa, from Hawaii, and A. heterophylla, from Réunion Island and some 18,000 km distant, is surprising. It has been explained by movement by Austronesians of a species like A. melanoxylon from eastern Australia (Brown et al. 2012). However, given the structure of relationships - A. heterophylla is embedded in A. koa, the combined clade being sister to A. melanoxylon - long distance dispersal from Hawii to Réunion, implausible although it may seem, is more likely (Le Roux et al. 2014).

Extrafloral nectaries are common in Fabaceae (Weber & Keeler 2013, see also below). They characterise a major clade within South American Senna, where they may be a "key innovation" that is involved in the diversification of that clade, much more speciose than its sister clade (68 species lacking nectaries versus 282 species with nectaries) and it has also speciated significantly faster. The extrafloral nectary-bearing clade may have moved in to new habitats that became available after the Andean uplift and speciated there (Marazzi et al. 2006, 2013b; Marazzi & Sanderson 2008, esp. 2010; Weber & Agrawal 2014). Marazzi and Sanderson (2010) suggest a crown-group age for this clade of some 40.8-30.6 m.y., that is, somewhat before the Andean uplift (ca 30 m.y.a.). But Marazzi and Sanderson (2010) also noted that Simon (2008) had found that the loss of extra-floral nectaries characterized a large clade of Mimosa that was far more speciose than its sister clade; the ratio of species in the sister taxa with and without extrafloral nectaries there is 15:515 (Simon et al. 2011). Furthermore, a less conspicuous and organized kind of extrafloral nectary characterizes a separate clade of Senna (Marazzi et al. 2013b).

Of the ca 116 genera of Faboideae in clades below the non-protein amino acid accumulating (NPAAA) clade, most (ca 70%) are small, with 10 or fewer species. However, floral morphology there is very diverse, many genera having other than papilionoid flowers (Cardoso et al. 2013a, b, 2015). Shoemaker et al. (2006) and Soltis et al. (2009) think, although with some hesitation, that diversification in Faboideae may be connected to a genome duplication immediately basal to the split between mirbelioids and the rest; this is the clade that is notable for the occurrence of the non-protein amino acid canavanine. This duplication has also been implicated in nodule formation in Faboideae, perhaps helping to explain nodule diversity there (Q.-G. Li et al. 2013). See also rate shifts in Smith et al. (2011) and discussion below under Ecology & Physiology.

Within the NPAAA clade, there has been much diversification of Indigofereae (aff. millettioids) in succulent biomes, clades growing there tending to be geographically more restricted than those common in grassland biomes; ages of crown groups of the four clades into which species of Indigofera fall is ca 15.5 m.y. or less (Schrire et al. 2009). Apparent independent increases in the rate of diversification in the Mirbelieae and Podalyrieae clades (mirbelioids) may rather be the results of extinctions caused by cooling climates and increased seasonality ca 32-30 m.y.a. in the early Oligocene (Crisp & Cook 2009), and Crisp and Cook (2007) date the development of SW/SE Australian disjunctions to vicariant events caused by the development of aridity in the Nullarbor plane some 14-13 m.y.a. (other climatic events could also be implicated). The Australian Mirbelieae, with 675 species, include two thirds of all Australian Faboideae, and 470 of those species are included in Pultenaea s.l., which seems to have undergone a rapid radiation 25-20 m.y.a. (Crisp et al. 2004; Orthia et al. 2005a, b). On the other hand, for Schnitzler et al. (2011) diversification of the ca 128 species of Podalyrieae in the Cape Flora began ca 33 m.y.a. at the end of the Eocene and was connected with shifts in how the plants survived fires. Divergence of woody clades in the Old World phaseoloids (millettioids) (crown group age ca 28.6 m.y.a.) has been associated with Late Oligocene warmness and aridity, and of herbaceous members with tropical arid climates in the Early Miocene (H. Li et al. 2013). Robinieae may have been diversifying for some 30 m.y. in the neotropical seasonally dry tropical forest (Pennington et al. 2009; Pennington & Lavin 2016).

Schaefer et al. (2012) suggest that Fabeae (IRLC), believed to be of eastern Mediterranean origin and originally with an annual habit, has moved at least 39 times into Eurasia alone, as well as to the New World, the Atlantic islands, and elsewhere. The speciose Astragalus, with around 2,865 species, characterises drier areas of both hemispheres, growing in woodlands and steppe habitats (Azani et al. 2017), and a number of taxa have leaf rhachis spines. It separated from Oxytropis (the latter has a beak on its keel) 16-12 m.y.a., and its diversification is still more recent. In particular, radiation in the speciose aneuploid New World neoastragalus clade (ca 500 species) started ca 4.4 m.y.a. (Wojciechowski 2004), with two invasions of west South America - there are over 100 species there - timed at a mere 2.07-1.62 m.y.a. (the larger group) and 1.23-0.79 m.y.a. (the smaller group) being followed by very high diversification rates in both (Scherson et al. 2008); see also Koenen et al. (2013). Annuals have evolved several times in both the Old and New Worlds (Azani et al. 2017: see below).

Several major clades that are correlated with geography have been detected in Lupinus, another IRLC group (Aïnouche & Bayer 1999, support not very strong; Aïnouche et al. 2004). Within the North American perennial clade there has been a recent (ca 2.7 m.y.) Central American/Andean radiation that is now represented by over eighty species, some 56 of which grow in the páramo, and the rate of diversification increases as the genus moved into Andean South America from Central America (Moore & Donoghue 2009; see also Silvestro et al. 2011; Sklenár et al. 2011). Lupinus in the Andes, and in alpine North America, where there had been an earlier burst of diversification, are largely perennials (Drummond 2008; Drummond et al. 2012). The annual habit is plesiomorphous, and the alpine taxa in South America in particular show much variation in habit, etc., ranging from tussocks to shrubs 6 m tall (Hughes & Atchison 2015). Perennials also diversified in eastern South America ca 6.5 m.y.a.; they are separately related to east North American annuals (Drummond et al. 2012). This South American radiation may also be connected with the movement of bumble bees, pollinators of Lupinus, from North to South America some six m.y.a. (Hughes & Eastwood 2006). All told, there are ca 171 species of Lupinus in South America versus ca 90 species in the whole northern hemisphere (von Hagen & Kadereit 2003), and the evolution in the South American Andean species, at least, has been adaptive (Nevado et al. 2016). Ree et al. (2003) studied aspects of LEGCYC gene evolution in the context of variation of floral morphology in the genus. However, as Givnish (2015b) notes, understanding the reasons for diversification s.l. in Andean Lupinus is not easy.

Bello et al. (2012) suggest a couple of apomorphies for the family and for clades within it. Zimmerman et al. (2017) provide an apomorphy scheme for Dialioideae. Poeppigia, sister to the rest of the family, has a number of plesiomorphic features when compared with these other members of the subfamily, hence many features that are common in the subfamily are drived within it (Zimmerman et al. 2017, see also above), although again, clarifying the relationships of the subfamilies will clarify this variation. For the direction of curvature of the young style, see Prenner and Cardoso (2016).

Ecology & Physiology.


Fabaceae often dominate in deciduous arid and semi-arid woody vegetation (Lewis et al. 2005; Schrire et al. 2005; M. Adams et al. 2016). They also grow in closed lowland tropical rainforest, but they are also often members of more open vegetation, including early successional communities, in both tropical and temperate regions of the world. Fabaceae include more tree species (= single stem >2 m tall, of if >2 stems, one erect stem >5 cm d.b.h.), ca 5,400, than any other family, for what that is worth (Beech et al. 2017). Faboideae-Robinieae are an important component of the neotropical seasonally dry tropical forests (Pennington et al. 2009). In the Cape fynbos vegetation there are some 250 species of Aspalanthus alone and in the West Australian kwongan over 135 species of Acacia (Cowling et al. 2000). Australia is unusual in that woody Fabaceae, especially Acacia, are notable components of most vegetation types, at least for part of the succession (Orians & Milewski 2007). There has been much diversification within a number of geographically-restricted clades of Indigofereae that grow in succulent biomes (Schrire et al. 2009).

Perhaps 16% of all woody species in neotropical l.t.r.f. are members of Fabaceae, especially Caesalpinioideae and Detarioideae (Burnham & Johnson 2004), and they are notably commoner in Guiana than in western Amazonia - to 49 vs 10% of all trees (ter Steege et al. 2006); they are #1 in individuals, Sapotaceae are next, and then Lecythidaceae. They are overwhelmingly the most common family in Amazonian forests in terms of numbers of species and individuals with stems 10 cm or more across, althouth they do not have a proportionally high number of locally abundant species (Gentry 1988; Hedin et al. 2009; Table 1; ter Steege et al. 2013); those species which do have high abundance values tend to be ECM plants (see below). They make up 1/4 of the 20 species with most above-ground woody biomass (5 species, 2 known to be ECM, = 4.96% of the total biomass), and 6/top 20 species ranked by productivity (Fauset et al. 2015); most of these are probably AM plants (Béreau & Garbaye 1994) - note that Dinitzia excelsa is #931 in abundance, #25 in above-ground biomass (Fauset et al. 2015)!

In Africa "acacia" (Senegalia, Vachellia) is a major component of the ca 2.5 x 106 km2 of microphyllous dry forests on relatively richer soils (Kalahari, E. Africa, the latter forests also with Commiphora) and in the drier Sudanian woodlands to the north (Timberlake et al. 2010). African savanna is also physiognomically distinctive because Senegalia and Vachellia trees, not very tall, have very broad crowns, having almost the shape of an open umbrella (Troll's model), and so this savanna looks different to savannas from other continents (Moncrieff et al. 2014). In eastern Africa there is a strong correlation between the diversity of "acacia" and the mammals that browse it, and mammalisn diversity is thought to drive the diversity patterns of the plants (Greve et al. 2012). Various Detarioideae dominate the ca 2.6 x 106 km2 of Miombo woodland (Timberlake et al. 2010; see also below), indeed, ca 43% of all Detarioideae occur in Africa and Madagascar alone, and some Faboideae clades are also quite diverse there, Detarioideae perhaps originating there (or in South America) (de la Estrella et al. 2017).

Nitrogen Fixation and Nitrogen Metabolism.

There is a substantial increase in leaf N at the crown-group Fabaceae node, along with a preference for less seasonality in precipitation (Cornwell et al. 2014). High N in the leaves has been linked to increased water use efficiency in woody N-fixing plants in general, rather than increased levels of photosynthesis, and legumes in particular have a complex nitrogen metabolism. N moves around the plant during growth, and extrapolating from agricultural legumes, it is likely to be remobilized from the leaves, ending up in large quantities in the seed. In general, N-containing compounds are involved in plant defence and in the water relations of the plant (Wink 2013; M. Adams et al. 2016), thus in Inga, a very speciose (ca 350 spp.) Neotropical mimosoid genus, up to 43 species may coexist at a single site, and this may in part be possible because species, even sister species, differ considerably in antiherbivore defences (Kursar et al. 2009; Endara et al. 2015; see also Richardson et al. 2001b; Pennington et al. 2009; Sedio et al. 2017).

Associations with N-fixing bacteria are very common in Fabaceae (see also Bacterial/Fungal Associations below). Estimates of biological N fixation in terrestrial environments, much of which is by bacteria associated with Fabaceae, are 90% of the total of 100-140 Tg [= 1012g) N fixed per annum (Gage 2004 and references: humans have doubled this). Substantial amounts of N can be fixed as NH3 and moved to the plant, and then to the community when the legume dies or is eaten (e.g. Batterman et al. 2013: Panama). However, the role of legumes in the N cycle of tropical forests is unclear, although non-nodulating species are proportionally less common in the humid tropics (Simonsen et al. 2017). In Fabaceae N fixation is to a certain extent facultative in that the amount a plant fixes can change (Menge et al. 2014, see also below), and in old, relatively nutrient-rich forests, N fixation by legumes (in the example, species of Inga) decreases when compared with species growing in seasonally flooded forests and in light gaps (Barron et al. 2011; Hedren et al. 2009 for the "leaky nitrostat" model); other species of Fabaceae also have low fixation rates in mature forests (Högberg 1986; Barron et al. 2011 and literature; Batterman et al. 2013: Sprent et al. 2013 for a summary). In Costa Rican secondary forests legumes grew more quickly, died more slowly, and recruited more poorly (?seed size) than plants that did not fix N (Menge & Chazdon 2015). Interestingly, nodulation in the family in Amazonia is inversely correlated with its dominance, and so is proportionally less in the poorer soils of the Guianan Amazon where Fabaceae are abundant (ter Steege et al. 2006). The effectiveness of nodulation is often reduced in more acid soils, although Lupinus and Mimosa in Brazil, at least, may be exceptions (Lin et al. 2012). N-fixing tropical forest legumes may have a competitive advantage over non-nodulators as atmospheric CO2 concentration increases (Cernusak et al. 2011), which might affect how one thinks of the history of the family. Interestingly, N-fixing fabaceous trees are very uncommon north of 35%oN (Menge et al. 2014). See also Terpolilli et al. (2011) for the efficiency of nitrogen fixation.

Interactions between legumes and their N-fixing bacteria go far beyond any simple association of the two and resultant fixation of nitrogen. In Acacia mangium nodulation and leaf nitrogen are increased if the plants are ectomycorrhizal (Diagne et al. 2013), while in endomycorrhizal Fabaceae phosphorus uptake by the fungus may affect N fixation (Walker et al. 2013 and references), for instance enabling growth of nodulated seedlings in the nutritionally harsh conditions of sand dunes (van der Heijden et al. 2015b). Nodules in Medicago can be formed both by N-fizing or non-N-fixing ("exploitative") rhizobia, and plants with mixture of exploitative and non-exploitative rhizobia did less well than plants with only N-fixing rhizobia under conditions of no herbivory, but the two performed equally poorly if there was herbivory (Simonsen & Stinchcombe 2014). There are also interactions with endomycorrhizal fungi here, with complex interactions at the level of gene expression (Afkhami & Stinchcombe 2016).

There may be a trade-off between extrafloral nectaries and N fixation. Godschalx et al. (2015) found that N-fixing rhizobia in Phaseolus lunatus reduced the amount of nectar produced in these nectaries and hence the attractiveness of the plant to ants. N fixation utilises much of the host's photosynthate which otherwise could be used to produce nectar, but the nitrogen fixed was used in the production of another form of defence, cyanogenetic compounds (Godschalx et al. 2015). Finally, synthesis of pyrrolizine alkaloids occurred in Crotalaria as a result of the reprogramming of the plant genome that occurs during nodulation (Benedito et al. 2008), and although details of the synthetic pathway are unclear, at least the initial enzyme involved was from the plant; if N levels were high, there was no nodulation and no alkaloid production (Irmer et al. 2015). However, details of such interactions between plant and bacterium are poorly known.

Indeed, Fabaceae have a very distinctive nitrogen metabolism (see M. Adams et al. 2016 and references). Non-protein amino acids are common (see e.g. Fowden et al. 1979), and nitrogen in the xylem sap is transported as a mixture of amino acids, amides, and sometimes also ureides; very little is transported as nitrate. Wojciechowski et al. (2003, 2004, see also Wojciechowski 2003) note that the distribution of some non-protein amino acids are systematically interesting. L-canavanine, which can be taken up in place of the normal amino acid L-arginine, may have potent effects on herbivores, although some can detoxify it (e.g. Rosenthal 1990, 2001 and references; Kergoat et al. 2005b; Huang et al. 2011: non-protein amino acids in general). L-canaline is rather like the amino acid ornithine; both L-canaline and L-canavanine serve as nitrogen reserves for the plant. A number of Fabaceae have cyanogenic glucosides. Here plant-insect interactions have been studied in detail, for instance, those between the cyanogenic host, Lotus corniculatus, and caterpillars of the burnett moth, Zygaena filipendulae; the latter can also synthesize the cyanogenic compounds itself (Zagrobelny & Møller 2011: references to other systems; see also Takos et al. 2011). Alkaloid glucosides are known, e.g. from Vicia, and Pentzold et al. (2014) discuss ways insects have of getting around such defences. Interestingly, canavanine and alkaloid production are mutually exclusive, while within the genistoids quinolizidine and pyrrolizidine alkaloids are similarly mutually exclusive (Wink 2008). For the toxic indolizidine alkaloid, swainsonine, found in a few members of the IRLC clade, see below under plant-fungal relationships.

When and where the associations between Fabaceae and their nitrogen fixers developed is unclear (but see J. J. Doyle 2011; Sprent et al. 2013). The first association presumably occurred some time after the Late Cretaceous crown-group age of the family, so ca 65 m.y.a. is the very earliest. Conservative estimates are that N-fixation arose 6-7 times in the family (J. J. Doyle 2011). In Faboideae, nodulation may involve the cooption of genes from a genome duplication event estimated at ca 54 m.y.a. (op den Camp et al. 2011; Q.-G. Li et al. 2013). Indeed, legumes in which the bacterial associations persist over evolutionary time, stable nitrogen fixers, are commonest in areas with cooler temperatures, whether associated with altitude or latitude, and are particularly frequent in Faboideae, often herbaceous plants and the commonest legumes in these conditions (Werner et al. 2015; see also Menge & Crews 2016 for latitudinal patterns in N fixation, also above). Free-living bacteria closely related to N-fixing rhizobia are very common indeed in conifer-dominated forests, in North America at least (VanInsberghe et al. 2015; for further details, see below).

Ectomycorrhizae (ECM).

Some Fabaceae - most records are from the Old World, particularly African, Detarioideae - are ectomycorrhizal (ECM) and do not fix nitrogen (e.g. Onguene & Kuyper 2001; for further details see clade asymmetries); the ECM habit has evolved at least four times in the family (M. E. Smith et al. 2011). Estimates of the number of Detarioideae involved range from 250 (Brundrett 2009) to 450 (B. Mackinder pers. comm. viii.2012) species. Some 36 of the ca 82 genera included in Detarioideae are reported to be at least locally dominant (e.g. Letouzey 1968; Mackinder 2005), 11 of these genera being in the rather small Macrolobieae/the Berlinia clade of only 16 genera (see also Wieringa & Gervais 2003). Some Detarioideae are endomycorrhizal/vesicular arbuscular mycorrhizal (AM) plants (see Cynometra below).

African Detarioideae

All told, Detarioideae occupy ca 2.8-3.3 m km2 in the Zambezian region alone (estimated from White 1983). In Miombo forests they represent 20-90% of the trees, 30-96% of the basal area, and with biomass estimates in the range of 35-97 Mg ha-1 (Högberg & Piearce 1986; Frost 1996). Genera like Isoberlinia and Brachystegia are very important components of the widespread deciduous Miombo forests which grow on often rather poor soils over some 2.7 million km2 or more of central Africa, and these forests are the centre of diversity of Brachystegia (e.g. White 1983; Högberg 1990; S. E. Smith & Read 2008; Timberlake et al. 2010). Many other species, both of trees and herbs, also grow in these communities, and the trees include other ECM taxa like Monotes (Dipterocarpaceae) and Uapaca (Phyllanthaceae) (White 1983). The AM Colophospermum mopane, an important food for elephants, dominates ca 380,000 km2 of mopane woodland (e.g. Högberg & Piearce 1986; Torti et al. 2001; Mackinder 2005; Newbery et al. 2006), these, along with Miombo forests, form a major part of the Zambezian Region. The ECM Isoberlinia is a major component of Sudanian Woodland (White 1983) which forms an interrupted band south of the Sahara from Mali to Uganda (White 1983; map: very approximate, from White 1983). In savannas, shallower-rooted grasses may obtain water obtained at some depth by trees (by hydraulic lift), which helps the savanna community to persist (K. Yu & D'Odorico 2015). This forest is biogeographically closest to Miombo woodlands among other African vegetation types (Linder et al. 2012).

In Africa, even LTRF can include a substantial element of ECM Fabaceae. Indeed, a caesalpinioid Biafran forest subtype has been recognised, and here, of 34 genera recorded, 28 are members of Detarioideae, and 11 of these are described as being characteristically gregarious (Letouzey 1968; see also Bâ et al. 2011a). Microberlinia dominates Guineo-Congolian forests in Cameroon, and other Detarioideae dominate parts of the coastal forest from Sierra Leone to western Gabon, and again in the periphery of the Zaire basin (White 1983). Newbery (1997) noted that Detarioideae in Cameroon LTRF grew on poor soils and seemed to be able to control the flow of phosphorus through the ecosystem to their own benefit, however, fertilization with phosphorus had little effect on their growth (Newbery et al. 2002); Van der Burgt and Eyakwe (2008) give information about a ca 35 km2 caesalpinioid-dominated area in the Korup Forest. Gilbertiodendron dewevrei, whose foliage, seeds, and perhaps even mycorrhizae are major food sources for elephants, etc. (Blake & Fay 1997), dominates ca 10,000 ha in the eastern Congo (see also Bastin et al. 2015). There it has above-ground biomass of 394-411.1 Mg ha-1, about 74% of the total (Makana et al. 2011); it may occupy 88% of the total basal area in the forests it dominates (Hart et al. 1989). Torti et al. (2001) discussed the factors that seemed to facilitate dominance of Gilbertiodendron, including the dense shade it cast, deep leaf litter, handling the potentially low availability of nitrogen, etc.. However, Peh et al. (2011a) found little difference in the soil of Gilbertiodendron forests in Cameroon when compared with that of adjacent forests, but they noted that there might be differences in such forests elsewhere and that ECM fungi might be involved in the dominance relationships of these forests.

In the New World, the ECM Aldina (Faboideae: 50 kb inversion clade) and the coppicing Dicymbe (Detarioideae) dominate forests in the Pakaraima Mountains in the central Guiana Shield region (ter Steege et al. 2006; McGuire 2007b; M. E. Smith et al. 2011). The latter in particular supports a rich community of fungi (Smith et al. 2011; Henkel et al. 2012) and has a remarkably high basal area of 38.4-52.5 m2 per hectare, around 25(-40) m2 being more normal figures (Henkel 2003). Aldina is also common in Amazonian rainforst (ter Steege et al. 2013; see also Ramos et al. 2015). The ECM Peltogyne (Detarioideae) is one of the rare monodominants of the northern Amazon, where it occupies ca 53% of the basal area of trees 10 cm or more in d.b.h. on Maraca Island, Roraima, a figure that increases in proportion in larger trees (Nascimento & Proctor 1997; Nascimento et al. 1997). The Guianan upland ECM/AM Eperua falcata, along with E. leucantha (mycorrhizae?), also Detarioideae, are 50% more abundant (usually far more) than any other non-palm of the 20 most abundant hyperdominant trees (10 cm d.b.h. or more) of the Amazonian rainforest (Peh et al. 2011b; ter Steege et al. 2013; Fauset et al. 2015, see also above; E. falcata perhaps AM - e.g. Béreau & Garbaye 1994).

Not all monodominant legumes are ECM (Torti & Coley 1999; Torti et al. 2001; Peh et al. 2011b). Legumes like the AM Mora excelsa (it also harbours endophytes), Caesalpinioideae so not immediately related to Detarioideae, dominates ca 37,000 hectares in Trinidad (Beard 1946; Hart et al. 1989). Interestingly, like ECM plants, both foliar and litter nitrogen contents are low, there is foliar resorbtion of nitrogen, soil nitrate concentrations are low, litter decomposes slowly and accumulates - perhaps the roots get the nitrogen the plant needs from this surface layer (Brookshire & Thomas 2013). Two other species of AM Mora may also be monodominants, as is the AM (and clonal) Pentaclethra and Prioria copaifera, Pentaclethra macroloba representing 16-18% of the above ground biomass in some Costa Rican forests (Torti et al. 1997; Henkel 2003; Peh et al. 2011b; Menge & Chazdon 2015). The Old World Cynometra (Detarioideae) also appears to be AM (e.g. Connell & Lowman 1989; Peh et al. 2011b), and Cynometra alexandri dominates ca 11,000 ha in Uganda (Eggeling 1947) and in some forests in the eastern Congo where it comes close to dominating (ca 31.5% of above-ground biomass) the ECM Julbernardia secretii adds another ca 12% (Makana et al. 2011: the next species adds 4.8%). Cynometra forests do not accumulate large amounts of litter, but that is also true of the ECM Julbernardia (Torti et al. 2001). Talbotiella gentii (?= Hymenostegia) forms monodominant stands in dry forests in Ghana, but the soil composition in unremarkable; seedlings tolerate shade and there is some vegetative reproduction (Swaine & Hall 1981), and although its mycorrhizal status is unknown, that of its immediate relatives is usually AM (Bechem et al. 2014).

Thus although the relationship between mono- or oligodominance of legumes in tropical forests in particular and the nature of their fungal associations is not simple, the combination of phylogeny, monodominance and mycorrhizal type in legumes is strong. Of the nine dominant tropical species listed by Hart et al. (1989), five are caesalpinioids, of which three are ECM Detarioideae (two more are ECM dipterocarps), while all the dominants listed by Connell and Lowman (1989) were legumes. 12/22 of the tropical classical monodominants listed by Peh et al. (2011b) are legumes, of which 4/9 whose mycorrhizal status is known are ECM plants (the others are Cynometra, Pentaclethra and three species of Mora). Of the other ecological attributes of these legumes, all had poor dispesal, masting occurred in all but two (two of the three species of Mora), and the seeds were medium-sized to large, being 5-117 g in weight (Peh et al. 2011b). All told, these dominant tropical legumes are characterized by a syndrome of ecological features that includes growing on poor soils, having large seeds with poor dispersal, accumulating litter and accessing organic nitrogen, tolerating shade, etc. (e.g. Alexander 1989b; Hart 1990; Torti et al. 2001; Bâ et al. 2011b). However, Newbery (2005, see Hogberg 1986) noted that mast fruiting had not been recorded from East African forest dominated by caesalpinioids, while Nascimento and Proctor (1997) noted that there was little difference between the soils on which the ECM Peltogyne dominated and those in adjacent more diverse forests. Interestingly, Gentry (1993) compared the ecological role of ECM dipterocarps in the Old World with that of caesalpinioid legumes in the New World, particularly on poor soils and under seasonal climates.


N-fixing taxa like some species of Lupinus, e.g. Lupinus albus, Stylosanthes and Aspalanthus, may also form root clusters of varying morphologies, and in some cases these have been shown to facilitate phosphorus uptake in P-poor soils (e.g. Shane et al. 2004b and references; Lambers et al. 2006, 2012b). The overall appearance of such roots is rather regular and dauciform, but this carrot-like shape is made by the dense, spreading lateral roots, not by root hairs as in dauciform roots proper (Shane & Lambers 2005; Shane et al. 2006). Species of Lupinus with such roots (especially common in Old World), and some with ordinary roots (from the New World) release massive amounts of carboxylates (organic anions) such as citrate into the soil, sometimes in amounts up to 23% of dry mass, and these can replace organic or inorganic P on soil particles and so mobilize it (Lambers et al. 2013 and references). Development of cluster roots has been linked to low foliar P concentration (X. Wang et al. 2015). Since Lupinus can also fix nitrogen, it can be an aggressive pioneer on volcanic and other skeletal and nutrient-poor soils (Lambers et al. 2013). Interestingly, both L. albus and some other species of the genus that lack the ability to form these cluster roots are also unable to form mycorrhizal associations (Delaux et al. 2014).

Fabaceae are perhaps the second most important family of lianes, both ecologically and in terms of number of species, in the New World, with about 720 climbing species recorded from there alone (Gentry 1991). Fabaceae are also prominent in drier forest types in both America and Africa (Lewis et al. 2005; Schrire et al. 2005), and lianes also are prominent in such forests (Gentry 1988; Schnitzer 2005). Fabaceae quite commonly have positive root pressures, which may reduce the susceptibility of liane stems to irreversible cavitation (Fisher et al. 1997, but c.f. Knipfer et al. 2016). Vessels in the stem of Entada are up to 700 μm across (Ewers et al. 2015)! For additional information about leguminous lianes, see Angyalossy et al. (2015; also Schnitzer et al. 2015).

About 120 species of Astragalus are annual, and this habit that has evolved ca 8 times in the Old World, the plants tending to grow in drier or even xeric areas, also in disturbed vegetation; the annual habit has also evolved independently in both North and South America (Azani et al. 2017). In Trifolium, too, the annual habit has evolved several times (Ellison et al. 2006).

About 27 species of Astragalus in North America are hyperaccumulators of the sulphur-antagonist selenium (Se) (El Mehdawi et al. 2012; Schiavon & Pilon-Smits 2016), which is stored as γ-glutamyl-methyselenocysteine; this may be associated with nodulation and also help protect the plant from herbivores (Alford et al. 2014). Se-accumulation seems to have evolved more than once here, and Astragalus includes the largest complex of Se accumulators in seed plants (White 2016); endophytic bacteria are involved in Se uptake, which can be up to ca 1.5% dry weight, 30% elemental Se concentration (Lindblom et al. 2013; Sura-deJong et al. 2015).

Knoblauch et al. (2001) discuss the function of the distinctive spindle-shaped non-dispersive protein bodies (= forisomes) that are common in the sieve tubes of Faboideae (e.g. Behnke 1981b; Behnke & Pop 1981; Peters et al. 2010; Müller et al. 2014). The forisomes can block the pores of the sieve plates when turgor pressure changes, and they can change their shape and volume very quickly depending on the concentration of Ca2+ ions; ATP is not needed for this shape change (Peters et al. 2007, 2008, 2010). Within Faboideae, forisomes are absent from a number of Galegeae, many members of which - including those species of Astragalus studied - also lack calcium oxalate crystals (Peters et al. 2010).

Pollination Biology & Seed Dispersal.

Pollination Biology.

Although one often thinks of the monosymmetric pea or papilionoid flower and its variants as characterising the family - bar the mimosoids - as a whole, this much underestimates the great variation in floral morphology in many "Caesalpinioideae" and basal clades of Faboideae in particular. As Bruneau et al. (2005, p. 201) noted of caesalpinioid legumes, "zygomorphy is expressed as a multitude of homoplasious morphs". Variation in pollen morphology is also considerable in basal "Caesalpinioideae" (Graham & Barker 1981; Banks et al. 2003), and Banks and Rudall (2016) speculate on the functional significance of this variation. Detarioideae in particular show much floral variation, with the complete loss of organs or whorls of organs often being associated with major dislocations of developmental patterns (Tucker 2000; Bruneau et al. 2014), while corolla development in the speciose Inga is notably labile, the median petal being adaxial or abaxial, depending on the species (Paulino et al. 2017). Within Faboideae (= Papilionoideae), papilionoid flowers are found only in a subset, albeit a large subset, of the clade and are probably derived within it. Thus the flowers of Swartzia, perhaps sister to all other Faboideae, are very different from those of all other Fabaceae in their single banner petal, numerous free, mostly infertile and very dimorphic stamens, and absence of nectar (Tucker 2003b), and there is considerable floral variation - but rarely including papilionoid flowers - in other basal Faboideae (e.g. Mansano et al. 2002, 2004). Heteranthy, as in Swartzia, is notably common in Fabaceae, although less so in the mimosoid clade (Vallejo-Marín et al. 2010), and Paulino et al. (2016) describe how this works in some Faboideae. In Lupinus the different stamen morphs, members of different whorls, produce different kinds of pollen, the inner whorl, opening later, producing sterile pollen that contains starch but no lipids, while in Cytisus the pollen from the two morphs - both found in anthers of the same whorl - is deposited on different places on the pollinator, and although pollen from both types of anthers germinated in vitro, this was not so in vivo (Paulino et al. 2016).

The monosymmetric papilionoid flower has a more or less erect banner petal, two wing petals, and paired interlocking keel petals enclosing the stamens; bee pollination is the norm here (for pollination in keel flowers in general, see Westerkamp 1996, 1997). The micromorphology of the epidermis differs between different petals of a single flower. The outer surfaces of the wing petals in particular are variously sculpted (Tewari & Nair 1979), with pockets and folds that afford footholds for the pollinator (Le Roux & van Wyk 2012) in Faboideae, at least (Stirton 1981; Ojeda et al. 2009). Colour patterning of the corolla is conspicuous, as in Lupinus, also Caesalpinia, and Bauhinia, and is always on the adaxial banner petal; these petals may also absorb ultraviolet light. The colour patterning on the standard of Hardenbergia violacea (Faboideae) may even mimic an anther (Lunau 2006). Although the flowers of Cercis are only superficially similar to those of Faboideae (Tucker 2002a), they, too, have keels and are similar functionally, although Cercis lacks the sculpturing on the wing petals that is common in Faboideae. In some Caesalpinia s.l. the abaxial sepal is coloured and more or less functions as a keel.

Hardly surprisingly, Fabaceae attract a diversity of pollinators that visit the flowers for various rewards; Lewis et al. (2000) summarize what is known about pollination in Fabaceae, especially in "Caesalpinioideae". Bat pollination is quite common in tropical members of the family (Fleming et al. 2009), bats visiting flowers with a variety of morphologies. Bee pollination is particularly common, Xylocopa bees alone visiting some 52 genera (Leppik 1966; Hurd 1978; Kalin Arroyo 1981; Lewis et al. 2000). Pollination in "Caesalpinioideae" is predominantly by polylectic bees, while oligolectic bees are commoner pollinators of the mimosoid clade and some Faboideae. Oligolectic bees are more speciose in (warm) temperate regions, especially Mediterranean climates (Michener 1979; Kuhlmann & Eardley 2012), and that is where the latter two subfamilies are particularly common. Even if papilionoid flowers are not as widespread in Fabaceae as one might think, flowers with banner petals are indeed common in non-mimosoid Fabaceae. Faboideae are pollinated mostly by polylectic bees and in a variety of ways, and the plants are a major source of both nectar and pollen (Goulson 2010: U.K. species). Keel flowers are pollinated by medium-sized to large bees, since they can "work" the flowers more easily, and large bees also pollinate Faboideae with inverted keel flowers (Amaral-Neto et al. 2015, q.v. for floral modifications involved); inverted keel flowers are found in genera like Canavalia. Interestingly, within the tropics bees seem to be commonest in the New World (Michener 1979), and woody Fabaceae are especially diverse there.

The nature of the reward for the bee varies, as is how the pollen is presented and where any nectar is to be found. When the androecium is monadelphous, i.e. the filaments of all the stamens are connate, there is no nectar and the reward for the pollinator is often pollen. If the androecium is diadelphous - nine stamens are connate and a single adaxial stamen is free - the reward is often nectar, the nectary lying between the filament tube and gynoecium and being quite variable in morphology (see also Leite et al. 2015 for access to nectar); Indigofera even has short, nectar-secreting spurs (see also Vogel 1997; Davis et al. 1988).

Secondary pollen presentation occurs in Fabeae and a number of other Faboideae in particular. Here the pollinator may pick up the pollen from the pollen brush at the end of the style; this mechanism has evolved perhaps eight times (e.g. Lavin & Delgado-S. 1990). In the asymmetrical flowers characteristic of Phaseolinae the labellum is twisted, forming a tube rather like an elephant trunk; this is why Vigna caracalla is called the corkscrew vine (c.f. Pedicularis: Orobanchaceae). When the bee lands on the keel, it depresses it and the style then forces pollen out of the tip of the keel - the end of the tube - in a tooth paste-like strand (Delgado-Salinas et al. 2011) - a pump-type secondary pollen presentation device. Other genera like Lupinus have a pump-type mechanism, although here apex of the keel is not twisted. Taxa like Cytisus, Medicago, Desmodieae and Indigofera have explosive pollination. Here the style, held under tension, is released when the pollinator lands on the keel and it as it were breaks it, pollen being scattered over the insect (e.g. Polhill 1976); such flowers can be visited only once (c.f. Polygala with explosive pollination - see below).

Buzz pollination is quite common in Fabaceae. It occurs throughout Cassia, Chamaechrista and Senna ("Caesalpinioideae": Lewis et al. 2000). Here the anthers have four different modes of development (Tucker 1996b). The flowers of Senna are often enantiostylous (enantiostyly is likely to have been acquired once, although also subsequently lost); the anthers are basifixed and porose. There are three (or more) stamen morphs: three abaxial staminodes, four middle medium-sized stamens from which pollen is taken by the bees, and three longer abaxial stamens pollen from which is actually involved in pollination (see Tucker 1996b; Marazzi & Endress 2008; Marazzi et al. 2007). Cassia s. str., with dorsifixed anthers, also has three stamen morphs. The filaments are curved and the anthers have slits or basal pores. Finally, the largely herbaceous Chamaecrista is also enantiostylous; the stamens have two morphs in different whorls, the filaments are short, and the basifixed anthers dehisce by pores and have a velcro-like line of hairs (Tucker 1996b). Westerkamp (2004) suggests that in some species of both Senna and Chamaecrista the orientation of the anthers is such that the pollen ejected when the flower is vibrated initially misses the bee entirely, but it bounces off the petals and then lands on the bee's back - whence it is removed by the stigma; enantiostyly is an integral part of this remarkable pollination mechanism.

Bird pollination is scattered in Fabaceae. The some 105 species of the pantropical/warm temperate Erythrina are pollinated by perching (sun) and hovering (humming) birds. Both floral morphology and how the flowers and inflorescences are held varies according the requirements of these different visitors, and there may have been four or more shifts from passerine to hummingbird pollination in the genus (Bruneau 1997; Mabberley 2008 for summary). Bird pollination has originated ca 20 times in the Australian bacon-and-eggs peas ([Mirbelieae + Bossiaeeae]) alone, although the bird-pollinated clades are rather smaller than their bee-pollinated sister clades; evolution of the meliphagid pollinator and of the plants involved may have been contemporaneous (Toon et al. 2014: Dollo model of evolution preferred, no strong evidence from either phylogeny or morphology that any bee-pollinated flowers are derived).

The mimosoid clade has flowers that are very different from those of most other Fabaceae. Numerous often small, polysymmetric flowers are grouped together in a dense raceme or head, all flowers opening at about the same time, and this is the unit of attraction. The pollen grains are frequently aggregated into polyads which are caught in a cup-shaped stigma that is of the appropriate size for the polyad of that species, and there are also about as many ovules in the ovary as there are pollen grains in the polyad (Kenrick 2003: implications of this pollination mechanism for the breeding system; Banks et al. 2010). Banks et al. (2011) note that all the pollen grains of a polyad form a single harmomegathic unit. Anthers in many of the mimosoid clade have terminal glands that are variable in morphology/anatomy, although little is known about any functions they might have (Luckow & Grimes 1997; de Barros & Teixera 2015) - perhaps they produce an odour (Tybirk 1997)? For nectaries, which are on the inside of the staminal tube/hypanthium, see Ancibor (1969); they are, for example, absent from Acacia s.l., although found in taxa now placed in Senegalia (e.g. Pedley 1986; Tybirk 1997). In Calliandra s. str. the polyads sometimes have an associated sticky mucilage body by which they are attached to the pollinator, but in this case the stigma is much larger and capitate and the polyads adhere to its surface (Prenner & Teppner 2005; Greissl 2006, c.f. in part Teppner 2007). Bats may also be pollinators, as in Parkia (Bumrungsri et al. 2008 and refs). For more information on locellate anthers (scattered in the clade), polyads and pollen morphology, anther dehiscence, etc., see Guinet (1981 and references: pollen), Prenner and Teppner (2005), Teppner (2007) and Teppner and Stabentheiner (2007, 2010) and references.

North temperate megachilid osmiine bees like Hoplitis species of the Annonosmia-Hoplitis group collect pollen from concealed-pollen flowers of the family and/or members of Boraginaceae; polylecty is derived in this group of bees (Sedivy et al. 2013). The bees may visit flowers of these two groups - which have very different morphologies - because both have pyrrolizidine alkaloids and/or particular nutrients that are essential for the growth of the bee larvae (Sedivy et al. 2013).

There is considerable variation in stigma morphology; the stigmas are porose to crateriform and may have an exudate (Dulberger et al. 1994; Marazzi et al. 2007; Costa et al. 2014). How the pollen gets to the receptive surface in stigmas with very small pores is poorly known.

Seed Dispersal.

The legume s. str. is a single-carpellate fruit that dehisces explosively, the two valves of the carpel twisting as they separate. Such fruits have two layers of lignified fibres, at least one of which is oblique to the long axis of the fruit (Fahn & Zohary 1955). A legume so defined is common in European-North American Faboideae, but also in Bauhinia, Duparquetia, Detarioideae (where seeds of emergent Tetraberlinia moreliana may be thrown some 60 m - e.g. van der Burgt 1997), and so on. The fruits of Cercis are not explosively dehiscent, but are otherwise similar - and they are also typologically rather similar to the fruits of Myristica, the nutmeg! Overall there is a great diversity of fruit morphology in the family, including variously winged fruits, fleshy fruits, arillate aeeds, fruits breaking up into single-seeded units in different ways, and fruits modified for external animal transport with spines and hooks, for example, the velcro-like hooks on Desmodium (hence its common name, beggar's ticks) and the indehiscent, spiny, spirally-coiled fruits of many species of Medicago (for their evolution, see Fourquin et al. 2013). In Trifolium the calyx and corolla are both involved in fruit dispersal. Arillate seeds are common and are eaten by birds in particular, and legumes are an important food source for frugivorous birds in Africa and Southeast Asia-Malesia in particular (Snow 1981). However, seeds of Abrus precatorius and Pithecellobium have red and black colour patterns on the coat and mimic the colour contrasts of these arillate seeds, and mimicry, of either an aril (seed bicoloured) or a fleshy fruit, occurs in Erythrina, Rhynchosia and Ormosia, all sizeable genera (McKey 1975; Foster & Delay 1998; c.f. transference of function, Corner 1958) - but chewing the seeds of Abrus precatorius is not recommended. Close to a thousand Faboideae are myrmecochorous, myrmecochory occuring in many species of Acacia s. str. and other Australian Fabaceae (Mckey 1989; Orians & Milewski 2007; Lengyel et al. 2009, 2010); independent evolution of this dispersal mode has occurred several times within the family. Dispersal of seeds by water has evolved more than once in Canavalia (Snak et al. 2016).

In many taxa, especially those with explosively dehiscent fruits, the seed coat is very hard and may need scarification for germination to occur (for fruits and seeds, see Corner 1951; van der Pijl 1956; Kirkbride et al. 2003; etc.). For other aspects of the ecology of seed coats, see Souza and Marcos-Filho (2001).

Plant-Animal Interactions. Herbivory by foliovorous insects is often quite marked on Fabaceae, despite the diversity of their defensive secondary metabolites, herbivores in general preferring to feed on nitrogen-rich plants, which those of Fabaceae certainly are (Simonsen & Stinchcombe 2014 and references). Cassia fistula is sometimes almost defoliated by caterpillars of pierid butterflies (pers. obs.), while up to one third or more of the developing foliage of species of Inga may be eaten by herbivores (Kursar et al. 2009).

Associations of insects and Fabaceae are often quite close. Caterpillars of Lycaenidae-Curetinae, -Riodininae-Riodinini and especially -Lycaeninae-Lycaenini butterflies are often found on Fabaceae (Ehrlich & Raven 1964; Fiedler 1991, 1995), in part because of the close association between many lycaenid caterpillars and ants, and ants are often conspicuously active on Fabaceae (see also Pierce et al. 2002). Similarly, larvae of the some 260 species in 15+ genera of Coliadinae and Dismorphiinae (Pieridae) butterflies, are found here (Braby & Trueman 2006: a quarter of the records, see also Brassicales and Santalales; the African Pseudopontiinae are recorded from Acanthaceae and Opiliaceae, Robinson et al. consulted vii.2015), and Fabaceae may be the original food plants of Pieridae (Braby & Trueman 2006; Wheat et al. 2007; Fordyce 2010). Caterpillars of the skipper group Eudaminae are also quite common on Fabaceae (Warren et al. 2009). The diversity of caterpillars, especially that of "basal" butterfly groups, including Baronia, sister to all Papilionidae whose caterpillars eat Acacia (Heikkilä et al. 2011; see also Ehrlich & Raven 1964), on Fabaceae is such that Janz and Nylin (1998) and Braby and Trueman (2006) suggested that Fabaceae might be the springboard for host-plant diversification of butterflies feeding on angiosperms in general (see also the introduction to Fabales). That would then mean that diversification of these butterflies would be very largely Caenozoic, given suggestions for the age of Fabaceae (see above).

Some 70% of the 1,700 seed beetles, bruchids (Chrysomelidae-Bruchinae [used to be Bruchidae]), are associated with Fabaceae. Two clades, made up largely of New World Acanthoscelides and predominantly Old World Bruchidius (neither monophyletic), dominate, and their hosts are mostly members of the mimosoid clade and a diversity of Faboideae (and in Acanthoscelides some Malvaceae, and Spermophagus Malvaceae and Convolvulaceae in particular). In Faboideae they detoxify the non-protein amino acid, L-canavanine (Kergoat et al. 2005b, 2006). Divergence of the beetles is estimated at around 60 m.y.a. near the time of origin of Fabaceae, and of Acanthoscelides and Bruchidius in particular ca 49.5 m.y.a. (Kergoat et al. 2005b, 2011). Stem Bruchinae are rather older, and may initially have eaten members of Arecaceae (q.v.); from Faboideae, probably their original host within Fabaceae, they moved on to other groups following the chemistry of the plants (esp. Kergoat et al. 2005a, b; also Johnson 1989, 1990: Acanthoscelides; Birch et al. 1989: chemistry of the interaction; Janzen 1969: the complexity of the association; Kato et al. 2010: importance of female oviposition preferences; Kergoat et al. 2015: Spermophagus). The larvae are specialized seed-eaters, and particular groups of bruchids may be associated with particular groups of Fabaceae (e.g. Kergoat et al. 2006: Bruchus and Vicieae, Sennius on Cassia, 2011 and references). For estimates of when particular groups of bruchids diversified on particular clades of Faboideae, see Kergoat et al. (2011). The pattern of association of bruchid groups with mimosoids is interesting and often quite specific (Kergoat et al. 2007); individual bruchid genera tend to be found on adajcent pectinations of the mimosoid phylogenetic tree (Kergoat et al. 2007), but in Spermophagus, which does not eat Fabaceae, there is less specificity (Kergoat et al. 2015).

For the diversification of Psyllidae (jumping plant lice; Hemiptera-Sternorrhyncha) on Macaronesian Faboideae-Genisteae, see Percy et al. (2004). Psyllids are quite common there, and the diversification of psyllids has been dated to around 3 m.y.a., well after that of their hosts which is dated at ca 8 m.y.a. (Percy et al. 2004).

Six subtribes including about 60% of the almost 1,000 species in the straight-snouted weevil Brentidae-Apioninae-Apionini are found on Fabaceae-Faboideae, perhaps moving there from the [[Frankeniaceae + Tamaricaceae] [Plumbaginaceae + Polygonaceae]] clade no later than the Upper Cretaceous (Winter et al. 2016: numerous dates for weevil diversification). Pinzón-Navarro et al. (2010) discuss the weevils found on Inga; for diversification of this genus, see above.

In another variant of insect-plant relationships, the flowers of Crotalaria are visted by Danainae butterflies and Ctenuchidae (arctiid moths) because the pyrrolizidine alkaloids they contain are used as the basis of the pheromones of these lepidoptera (also Asteraceae, and wilting plants of some Boraginaceae and Heliotropaceae: Edgar et al. 1974; Pliske 1975; Boppreé 1986). Crotalaria is also associated with arctiid moths such as Utetheisa, its secondary metabolites providing defence for the young, etc. (Eisner & Meinwald 1995; Hartmann 2009; other articles in Conner et al. 2009; Zaspel et al. 2014 for the evolution of pharmacophagy in Arctiinae).

Rather generalized legume/ant associations are very common in Fabaceae and are mediated by the extrafloral nectaries that are widespread here (Weber & Keeler 2013). The nectar secreted, which usually contains sucrose, attracts ants that can protect the plant against herbivorous insects; a variety of ants may be attracted to the nectaries of a single legume species (Schemske 1983). Such nectaries have possibly evolved some 35 times, and have also subsequently been lost and even regained (e.g. Marazzi et al. 2011, 2015; see also McKey 1989). They are notably common in the mimosoid clade (perhaps only one origin) and are found towards the base of the petiole, on the rachis, and sometimes on petiolules. They are less common in caesalpinioids, although they may be represented by tufts of hairs (Pascal et al. 2000), and still less common in Faboideae, although they have originated several times even there (Marazzi et al. 2014). Marazzi et al. (2012) suggested that in a clade making up most of Caesalpinioideae (Gleditsia/Chamaechrista/the mimosoid clade) and dated to 63 or more m.y.a., the origin of extrafloral nectaries could be thought of as a facilitating "deep homology" manifest in the numerous subsequent independent acquisitions of the nectaries there. Interactions between the nature of rhizobial infections, nectaries, and the effects of herbivory are discussed above.

There are also more intimate ant/legume associations where generally but a single species of ant is involved in each association (Schemske 1983). The close association of Pseudomyrmex ants with some members of the old Acacia subgenus Acacia (= Vachellia), including American swollen-thorn acacias such as V. sphaerocephala, are well known. The plant provides protein-rich Beltian bodies at the ends of each leaflet (the leaves have many leaflets, even for Acacia s.l.) as food for the ants, the ants also taking nectar from the petiolar extrafloral nectaries, and the swollen stipular thorns serve as their homes; the ants patrol the plants, even clearing the ground around the host (c.f. devil's gardens) (Janzen 1966, 1967b, 1974b; Webber & McKey 2009). Interestingly, Vachellia produces new leaves (and thus Beltian bodies) even during the dry season; the biomass locked up in the swollen thorns, which persist on the plant and do not break down quickly, is considerable (e.g. Janzen 1966). The nectar is sucrose-free unlike normal extrafloral nectar, invertase in the nectar breaking the sucrose down into glucose and fructose (Kautz et al. 2009). Furthermore, the invertase the ants need to break down the sucrose is inhibited by chitinase, a major protein in the extrafloral nectar (Heil et al. 2014). Complex interactions between protease inhibitors in the plant and proteases in insects may help deter other than the mutualist ants from eating the Beltian bodies (Orono-Tamayo et al. 2013; see Rubin and Moreau 2016 for interaction between molecular evolution and mutualism in Pseudomyrmex). The ages of Pseudomyrmex and the main clade of Vachellia that it inhabits is about the same, ca 5 m.y.a., and there seems to have been but a single gain of domatia there (Chomicki et al. 2015; Chomicki & Renner 2015), although there may have been two origins of this association in Pseudomyrmex (Ward & Branstetter 2017). Perversely, species of Vachellia with low nectar rewards are derived from those with higher rewards, and species that offer only low rewards are often colonized by exploiter ants that do not defend the plant (Heil et al. 2009), while specialized ant/domatia systems can be exploited by parasitic ants recently derived from generalist rather than specialist relatives (Chomicki et al. 2015). The three species of ants commonly found on the African V. drepanolobium harbour distinctively different fungal communities in their domatia (but Chaetothyriales, commonly associated with ants, are not prominent), although what these fungi might do is unclear (Baker et al. 2017). In V. erioloba, at least, two or even more species of ants are commonly found inhabiting different domatia on the one plant (Campbell et al. 2015).

There are several other examples of close ant/plant relationships in Fabaceae (McKey 1989 for a list). There have been at least 15 gains of ant domatia in the New World legumes Platymiscium (Faboideae-Dalbergieae - ?4 gains: see Chomicki & Renner 2015, origins ca 13.4 m.y.a., 6.1 m.y.a., and more recently) and Tachigali (Caesalpinioideae: 9 origins; see Fonseca 1994). Evolution of ant and plant seems to have been more or less contemporaneous in the latter, but with later bouts of colonization by the same and different ant clades (Chomicki et al. 2015). In the common African Leonardoxa africana (Detarioideae) a third party, an ascomycete (Chaetothyriales - see Vasse et al. 2017), is central to the relationship. Nitrogen from the ants initially moved more into the fungus than the plant (Defossez et al. 2010), then ant larvae ate the fungi (Blatrix et al. 2012: the ant is Petalomyrmex phylax); over time the nitrogen became distributed about equally among all three partners (Defossez et al. 2010). For plant-ant signalling in this association, see Vittecoq et al. (2011); Brouat et al. (2001) thought that the size and shape of the prostoma, the more or less unlignified area of the stem through which the ant entered the plant, could in some cases be linked with comparable attributes of the head of the ant involved in the association. There are similar plant-ant-fungus associations in Tachigali (Blatrix et al. 2012: the ant is Pseudomyrmex penetrator). Finally, bacteria living off colony debris are eaten by rhabditid nematodes that are possibly in turn eaten by ants living in the southeast Asian Saraca thaipingensis (Detarioideae: Maschwitz et al. 2016).

Bacterial/Fungal Associations.

Bacteria and N-Fixation.

Many Fabaceae have a close association with nitrogen-fixing bacteria which grow inside irregular, pinkish-coloured nodules on the roots (Sprent 2009 and Sprent et al. 2017 for summaries), the α-proteobacterium Rhizobium is the best-known genus involved. Nodulation is especially widespread in Faboideae and the mimosoid clade, but it is sporadic in more basal Caesalpinioideae, occuring in Chamaecrista and several other genera (Manzanilla & Bruneau 2014). In general, nodulation is very common indeed outside the tropics, less common in the humid tropics (Sprent et al. 2013). Although most Faboideae are nodulators (Sprent 2000, 2001, 2007) and the nodulating Swartzia (derived nodule morphology) and immediate relatives may form a clade sister to the rest, or most of the rest, of the subfamily (see below: Ireland et al. 2000; Pennington et al. 2000; Lavin et al. 2005), Faboideae in other clades basal to that including the bulk of the subfamily do not nodulate. Root nodule morphology may help delimit groups of genera in Faboideae (Wojciechowski 2003; Sprent et al. 2013); nodule morphology is independent of the particular bacterium involved and is controlled by the plant (Angus et al. 2013; Agapakis et al. 2014). Some aquatic legumes in both the mimosoid clade (Neptunia) and Faboideae (Aeschynomene) form stem nodules, albeit sometimes also associated with adventitious roots. In Aeschynomene the ability to form such nodules has evolved more than once, and members of a clade of this genus can form nodules with Bradyrhizobium even though they have lost Nod-genes (see below: the bacteria enter via cracks in the stem), furthermore, several strains of Bradyrhizobium are involved and some are photosynthetic (Chaintreuil et al. 2013); for more on Bradyrhizobium, see below.

Rhizobia fix nitrogen only when in association with their host, and one of the components of the cofactor of the nitrogenase which actually fixes the nitrogen comes from the host legume (Hakoyama et al. 2009). Nodule formation is initiated by the exudation of flavonoids, isoflavonoids and other bacterial attractants by the host and the infection of a root hair by a bacterium. The plant may also become infected through cracks in the epidermis, perhaps the plesiomorphic condition, as in Faboideae-Dalbergieae and -Genisteae (Cannon et al. 2010; Vessey et al. 2004; Okubo et al. 2012; Terpolilli et al. 2012; Chaintreuil et al. 2013); nodulation near lateral roots also occurs (Sprent et al. 2013 for a summary). Infection in Sesbania rostrata is both through cracks and via root hairs, the particular mode depending on the interaction of ethylene and water conditions in the soil (Oldroyd & Downie 2008).

Nodule formation often involves the production of Nod factors (NFs) by the bacterium, although in a few nodule-forming α-proteobacteria such as the photosynthetic Bradyrhizobium canonical nodABC genes are absent (Giraud et al. 2007). NFs are lipochito-oligosaccharides (LCOs), made up of a chain of 3-5 N-acetyl-D-glucosamine units, variously substituted and with a 16-20 C fatty acid attached. However, Gourion et al. (2015) emphasize the fact that NFs are not always involved in nodulation and that during infection there may be cell death as in the hypersensitive response to pathogens, and that in general, modulation of the normal immunity defence reactions of the legume are important in the initial establishment of the plant/bacterium association. There is a common symbiosis pathway in both baterial and mycorrhizal pathways, with a number of genes being shared by the two (Kawaharada et al. 2017 for details). Initial stages of nodulation are accompanied by the reprogramming of the host legume's genome, and the expression levels of large numbers of genes changes, including the repression of genes which might be involved in defence (Benedito et al. 2008).

The plesiomorphic infection morphology is that of persistent fixation threads, where the N-fixing bacterioids are retained within cell-wall bounded structures, two or more bacterioids per cell, and the nodules are long-lived, the bacterioids being able to divide (see also Parasponia [Cannabaceae]). More derived is the absence of fixation threads, rather, there are infection threads, the bacteria being released into membrane-bounded symbiosomes (Sprent et al. 2017), also the bacteria-infected cells may undergo mitosis, there is a single bacterioid per cell, and the nodule has a short life span (de Faria & Sprent 1995; Corby 1988: survey of nodule morphology; Sprent 2005: nodule distribution, see also J. J. Doyle 1994, 1998; J. J. Doyle et al. 1997; Lavin et al. 2001; Oono et al. 2010; Terpolilli et al. 2012). Thus Lotus has determinate nodules and infection is usually by root hairs [desmodioid nodules] - J. J. Doyle (2011). Infection threads are invaginations of the wall of the root hair - and how these threads develop is not that dissimilar from how root hairs develop - through which bacteria reach the apoplastic area beneath the epidermis. The bacteria synthesize exopolysaccharide, Epr3 being the receptor gene in the plant, and they move down the thread in part by successive division, and the development of the thread results from quasi-independent events in successive cells, so reaching more deeply into the root, the thread eventually branching in the nodule primordium, the bacteria being released (for details, see Gage 2004; Oldroyd and Downie 2008; Kawaharada et al 2017). Persistent infection threads and nodules of indeterminate growth are found in "Caesalpinioideae" and also in some Faboideae like Pisum and Vicia (Naisbitt & Sprent 1992; Rae et al. 1992). There the bacteria eventually move by endocytosis into the nodule cell in which they will reside; one or a few bacteria (now known as bacteroids) are enclosed by a host-derived membrane, the whole forming an organelle-like symbiosome (Rae et al. 1992; Gage 2003; Streng et al. 2011, summary; Sprent et al. 2013). Symbiosomes are formed in cells that are no longer meristematic, the meristem continuing to produce a supply of cells that are serially infected by the persistent infection threads. In determinate nodules such as occur in Phaseolus, but which are usually found in tropical legumes (Gage 2003), symbiosomes form in cells of the nodule meristem, and transmission is by cell division (Rae et al. 1992). In Medicago, at least, repeated endoreduplication of the host cell is needed for invasion of the bacterium to occur and the whole nodulation process to be effective (Maluszynska et al. 2013 and references). The nodules are anatomically rather like stems in having peripheral vascular bundles, the bacteria being in the pith (Franche et al. 1998), however, nodule origination occurs where lateral roots develop, at least in Faboideae (op den Camp et al. 2011). Bacteroid differentiation is either reversible, little morphological change occurring, or non-reversible. In the latter case the bacteroids become swollen and there is only one per cell, but N-fixation is more efficient than that by non-swollen bacteroids (Oono & Denison 2010; Sprent et al. 2013). The plant produces nodule-specific cysteine-rich [NCR] antimicrobial peptides, and these cause endoreduplication of the bacterial genome, and the bacteria stop growing and become dependent on the host for many basic activities; they have been called ammoniaplasts and are effectively plant organelles Gage 2004; Terpolilli et al. 2012). This apparently irreversible process seems to have evolved several times (Oono et al. 2010).

There is a relatively small chromosome-born "symbiosis island" that can be exchanged by lateral gene transfer as a unit (e.g. Agapakis et al. 2014; see also Ormeño-Orillo et al. 2013 for lateral transfer) among bacteria and that enables nodulation to develop, although not all nodulation genes may occur on the island (Sullivan & Ronson 1998; J. J. Doyle 1998), amd there are also similar mobile plasmids the exchange of which can cause symbionts to become pathogens and vice versa (Sprent et al. 2017). Particular legumes may select particular bacterial variants for nodulation, yet all the bacteria may have similar symbiosis islands (Parker 2012). Several nod genes aggregate to form the nodABCIJ operon that is common in α-rhizobia (Aoki et al. 2013, see also below). Such an aggregation of genes enables bacteria to become endosymbionts in a single step (Maclean et al. 2007). In Ethiopia there is both vertical transmission of these gene complexes as well as horizontal transmission of symbiosis islands from introduced strains of nodulating bacteria to native strains (Aserse et al. 2012; see also Beukes et al. 2016). In eastern North America legume hosts select for particular Bradyrhizobium strains, and independent of this particular variants of the symbiosis island are exchanged between the different bacteria (Parker 2012). Strains of Bradyrhizobium may be metabolically quite different, conversely, quite different legume bacterial symbionts may share more genes than Rhizobium leguminosarum, for example, shares with the closely related but non-symbiotic Agrobacterium (Maclean et al. 2007). See also Hirsch and LaRue (1997), Couzigou et al. (2012) and Chaintreuil et al. (2013) for the complex origin of the nodule developmental pathway.

Further complicating the issue, rhizobia do not form resting spores, and how they persist in the soil is unknown - perhaps in biofilms (Hirsch 2010)? It is often suggested that genes involved in nodulation moved from α- to ß-proteobacteria, but it appears that nodIJ genes, at least, may have evolved following a gene duplication in ß-proteobacteria and then moved and became part of the nodABCIJ operon (Aoki et al. 2013). Finally, it has recently been realized that strains of Bradyrhizobium are the dominating bacteria in pine forests across North America, and although they are unable to fix nitrogen, they seem to be able to metabolize aromatic carbon sources (VanInsberghe et al. (2015).

Details of the evolution of nodulation in Fabaceae are still not well understood (J. J. Doyle 2013). In Faboideae, nodulation involves the cooption of genes (originally involved in lateral root origination) after a genome duplication event ca 54 m.y.a. or (67-)63.7, 57(-56) m.y.a. (op den Camp et al. 2011; Q.-G. Li et al. 2013; Vanneste et al. 2014a, 2014b). This happened after the divergence of Faboideae from N-fixing clades such as Chamaecrista which do not have this duplication and probably form nodules in a different way (Cannon et al. 2010), the nodules probably being plesiomorphic in morphology (Vanneste et al. 2014b). Indeed, the ability to nodulate has been acquired several times within the family. Manzanilla and Bruneau (2012: but c.f. Figs 2 and 3) discuss the distribution of nodules in the Caesalpinioideae; they are quite widespread there, although more so in the mimosoid area.

Many nodulating bacteria in legumes are members of the proteobacteria α-2 subclass (Sprent et al. 2017), but they do not form a monophyletic group, and several "species" are involved; Agrobacterium (crown gall tumour) and others are also members of this group (J. J. Doyle 1998; Sprent 2001). Bradyrhizobium, also associated with Fabaceae both in Australia and Africa, is very diverse with 15 or so main clades (the number is increasing), although many species lack names (Stepkowski et al. 2012; Beukes et al. 2016 and references), and there is substantial temperate/tropical differentiation in the bacteria (Stepkowski et al. 2012). Other important N-fixing bacteria associated with Fabaceae include Burkholderia and relatives, which are ß-proteobacteria not at all close to Rhizobium. These are quite common in the tropics, and can tolerate alkaline conditions (Sprent 2007; Angus et al. 2013). Not all fix nitrogen, for instance, some are human pathogens, but those that are effective nitrogen-fixing symbionts form two groups, one involved in symbioses with New World Mimosa and the mimosoid clade and with the nod and nif genes on plasmids, the other, B. tuberum, nodulating African Faboideae-Crotalarieae and -Phaseoleae and with the nod and nif genes on the chromosomes (Bontemps et al. 2010; Agapakis et al. 2014). Other ß-proteobacteria form nodules, albeit ineffective, with the mimosoid clade (Moulin et al. 2001; Sprent 2002; Elliott et al. 2007 and references), and some are even pathogenic, but these rarely nodulate. Surprisingly, a Burkholderia growing on Solanum was able to form nodules and fix nitrogen when on Phaseolus vulgaris (Martínez-Aguilar et al. 2013). Sprent et al. (2013, 2017) summarise the relationships of the 14 bacterial genera known to be able to nodulate legumes; see above for the timing of evolution of these associations.

A single plant may form associations with more than one species of bacteria, and these may be members of the two main groups just mentioned (Sprent et al. 2013 and references). Symbiont specificity tends to be greatest in the IRLC clade (Faboideae), although genera like Astragalus are exceptions (Howard & Wojciechowski 2006; Sprent et al. 2017). At least some pioneer legumes form nodules with a variety of bacteria, thus perhaps enhancing their success as colonizing species (Behm et al. 2014 for references), although legumes associated with rhizobia spread into non-native habitats less readily than do legumes lacking this association, suggestiong that appropriate bacteria can be lacking (Simonsen et al. 2017). while in Mimoseae exactly which bacteria form associations varies according to soil conditions (Sprent et al. 2017). In Medicago truncatula nodule-specific cystein-rich peptides control exactly which bacteria form nodules with the plant by causing early nodule senescence in some bacterial strains (S. Yang et al. 2017). Furthermore, legumes are associated with other endophytic bacteria (Peix et al. 2015).

A number of species of Astragalus in North America are hyperaccumulators of selenium, which is stored as γ-glutamyl-methyselenocysteine; this seems to be associated with nodulation (Alford et al. 2014: see also above).

Venkatheshwaran et al. (2012) discuss a mutation that improves nuclear calcium signalling at an early stage in the development of symbioses (for which, see Barker et al. 2017) - IRLC clade only (Lotus, Medicago)?


There are important fungus-legume interactions. Thus in the mimosoid clade (phyllodinous Acacia, Senegalia), Faboideae such as Aldina, Gleditsia, a group of five genera around Mirbelia, and in particular Detarioideae, are ECM plants, but only very rarely also AM (e.g. Sprent & James 2007; S. E. Smith & Read 2008; Bâ et al. 2011a, b; M. E. Smith et al. 2011; Bennett et al. 2017: see above). In the ECM associations of three species (two Dicymbe, Aldina, Detarioideae and Faboideae respectively) dominating in New World forests there were over one hundred species of mostly basidiomycete fungi involved (M. E. Smith et al. 2011). ECM networks, as in Dicymbe forests in Guyana and in Gilbertiodendron dewevrei forests in Cameroon, can be complex, the network being established as the seed germinates (McGuire 2007b; Michaëlla Ebenye et al. 2017). Interestingly, ECM fungi from African Fabaceae and those on Uapaca (Phyllanthaceae) group together (Tedersoo et al. 2014a), perhaps because the hosts may grow together. For further details of ECM fungi and Fabaceae, see Ecology & Physiology above and clade asymmetries, however, little is known about mimosoid:fungus relationships (Acacia can be AM or ECM - Hayman 1986). AM fungi are known from several N-fixing Fabaceae-Faboideae (Hayman 1986), and have been found in root nodules in several species (but not in Lupinus, see also above), although their importance there is unclear (Scheublin & van der Heijden 2006).

Some species of Lupinus have lost the ability to form AM associations, although they still have the conserved genes that are part of the symbiosis toolkit but have non-symbiotic roles (Delaux et al. 2014, see also above).

In Oxytropis kansuensis the toxic indolizidine alkaloid, swainsonine, is synthesised by the endophyte Undifilum (inc. Embellisia), an imperfect stage of Pleosporaceae, an ascomycete (Pryor et al. 2009; see also Reyna et al. 2012). Swainsonine is also found in some species of Astragalus and of Swainsona itself (Ralphs et al. 2008). In North America these species with swainsonine are often called "locoweeds", and they cause a serious, sometimes fatal, disease in cattle that eat them. Swainsonine was first found to be synthesized by the bacterium Rhizoctonia leguminicola.

Rusts show interesting patterns of distribution on Fabaceae. Uromyces is found predominantly on herbaceous Faboideae, but also on Bauhinia and one or two other non-Faboideae), while Ravenelia is found on woody members of the family, "Caesalpinioideae" (but not Bauhinia) and especially the mimosoid clade (Savile 1976, 1979a, b; El-Gazzar 1979). In some species of Ravenelia the teliospores, thick-walled spores in which nuclear fusion and then meiosis occur, are aggregated into groups, and these telial heads may mimic the groups of pollen grains (polyads) that are common in the mimosoid clade. Stingless Trigona bees pick up both telial heads and polyads as they forage for pollen, so helping disperse the fungus. Interestingly, Ravenelia is only very rarely found on Acacia, and the distributions of rusts, acacias and trigonid bees all break at about Wallace's Line; thus Ravenelia is not native to Australia, Acacia s. str. is centred on Australia, etc. (El-Gazzar 1979; Savile 1979b).

Bacteria and Fungi together.

Fabaceae include most ECM plants that also fix nitrogen, a rather uncommon combination (for Alnus [Betulaceae], see Walker et al. 2013), although relatively little is known about their eco-physiology. Endomycorrhizal Fabaceae are commonly associated with N-fixing bacteria (Bâ et al. 2011b). The establishment of endomycorrhizal associations and a variety of aspects of the nodulation process starting with root hair curling are connected and can be linked to an autophagy related kinase, precursors for all these activities being produced by autophagy (Estrada-Navarrete et al. 2016: Phaseolus vulgaris).

Vegetative Variation. Although most Fabaceae have once- or twice-compound leaves, leaflets that are opposite or alternate and with entire margins, and pulvini associated with leaves and leaflets, there is extensive variation on this theme. Thus palmate leaves occur in Lupinus, serrate leaflet margins in Cicer, Medicago, etc., and unifolioliate leaves are scattered throughout the family. A number of Faboideae (e.g. Vicia, Pisum) are tendrillar vines, the tendrils being modified terminal leaflets; in Lathyrus aphaca the photosynthetic function of the leaf is taken over by the large stipules, the rest of the leaf being an unbranched tendril, while L. nissolia lacks tendrils and has a phyllodinous leaf (see Kenicer et al. 2005 for a phylogeny). In some taxa compound leaves are reduced to a single more or less connate pair of leaflets, as in Bauhinia, named after the botanical brothers Caspar and Jean Bauhin, and Cercis (Owens 2000). The leaves of Tachigali have continuous growth like those of Chisocheton.

In general, angiosperm leaf development is associated with the activity of the KNOX1 gene, and this is also true of plants with compound leaves incuding many Fabaceae. However, in the IRL clade (Faboideae) the KNOX1 gene is not - or rather differently - expressed in the developing leaves, while FLO/LFY genes, normally floral meristem identity genes, are expressed both in the leaves and in the flowers (Hofer et al. 1997; Wang et al. 2008; Peng et al. 2011; Townsley & Sinha 2012). Interestingly, the leaves of the IRL clade lack pulvini, which are present in nearly all other Fabaceae (Champagne et al. 2007; Rosin & Kraemer 2009).

In Acacia s. str. (the old subgenus Phyllodinae), the leaves of the mature plant are much modified undivided structures flattened at right angles to the plane of flattening of a normal lamina; they are often called phyllodes. A long-standing question has been, what "is" this Acacia leaf? In the early development of normal leaves of Fabaceae two rows of adaxial meristems produce the leaflets/pinnae, and these become lateral in position by secondary reorientation. In the phyllodes of Acacia there is a single, broader adaxial meristem that develops into the entire leaf (Kaplan 1980); there is no reorientation, hence the plane of flattening of the phyllode. Complicating the issue, in species like A. verticillata these phyllodes are densely set along the stem, some are associated with stipules and buds and have an extrafloral nectary, but others are simply flattened, needle-like processes (Kaplan 1980); the relationship between the two kinds of phyllodes is unclear. See also Rutishauser (1999) for the morphology of whorled leaves in Acacia, and Rutishauser and Sattler (1986) and Sattler et al. (1988) for the development of A. longipedunculata leaves, where there are only seven traces and 1-3(-4) axillary buds per whorl of up to 27 phyllodes, and colleters and other structures are also involved. Gardner et al. (2008) summarized of the history of this controversy. Seedlings and regeneration shoots can also have normal-looking once or twice compound leaves as well as intermediates. Pasquet-Kok et al. (2010) looked at the complex functional aspects of this switch from regular leaves to phyllodes during the development of the plant in the Hawaiian A koa, and they found i.a. that phyllodes were more drought tolerant and regular leaves might grow faster and tolerate shade better.

Some species of Mimosa and other genera have leaves that are sensitive to touch, stimulus transmission occurring as membrane depolarisation is propagated down the petiole and along the stem (Volkov & Markin 2014 for a summary); folding of the leaf is caused by tugor changes in the cells of the pulvini at the bases of the leaf and leaflets (for the anatomy of the pulvinus, which has an endodermis, see Rodrigues & Machado 2007). Simon et al. (2011) suggest that sensitive leaves have evolved ca six times in Mimosa alone, although a full understanding of the evolution of this feature depends on more extensive studies on this and related phenomena in legume leaves. Thus in Albizzia (Samanea) saman, there are similar movements as the leaflets fold towards the evening when the light is failing, or just when there is heavy cloud cover, this behaviour being responsible (in some tellings of the tale) for the name of this plant, the rain tree. In Desmodium gyrans the single pair of lateral leaflets move intermittently without being touched, the speed of movement increasing with the temperature.

Nodal anatomy is more variable than might be thought - Ulex (= Genista s.l.) has 1:1 nodes, while the pattern of change in nodal anatomy during ontogeny in Vicia is complex, and in adult plants the lateral bundles sometimes arise a full internode below the node they innervate (Kupicha 1975).

Economic Importance. Weeden (2007) discusses the diversity of genetic changes involved in domestication of legumes. For information on the domestication of soybean (Glycine max), common bean Phaseolus vulgaris, pea (Pisum sativum), the azuki bean (Vigna angularis) and relatives, see papers in Ann. Bot. 100(5). 2007, and for these taxa and lentil (Lens culinaris), see Fuller (2007). For the peanut, Arachis hypogea, see Krapovikas and Gregory (1994), for its domestication, see Dillehay et al. (2007) and for its phylogeny, see Friend et al. (2010), for the domestication of the lima bean, Phaseolus lunatus, see Serrano-Serrano et al. (2010), for much information on alfafa, Medicago sativa, see Small (2011), and for the relatives of soybean, see Sherman-Broyles et al. (2014).

Genes & Genomes. A whole genome duplication in Faboideae seems to have occurred prior to the divergence of Dalbergieae (Bertoli et al. 2009; Schmutz et al. 2010; Cannon et al. 2014: Copaifera; see also Young et al. 2011; Wang et al. 2017), although exactly where this duplication is to be placed is unclear and subsequent chromosomal rearrangements have been extensive (Murat et al. 2015b: Glycine, Lotus); see above for dates. There may be additional duplications somewhere in Cercidoideae and in the ancestor of [Chamaecrista + the mimosoid clade] (e.g. Young et al. 2011), although at this stage it is not clear exactly where they occurred and what immediate evolutionary consequences there might be for the clades involved (Cannon et al. 2014: Swartzieae s.l. not sampled). Jiang et al. (2013) examine the fate of duplicated genes in Glycine max, interestingly, there is no evidence of fractionation bias or genome dominance associated with the duplication in Glycine, suggesting it was an autoploid event, while there was evidence of the former, at least, for a duplication in Medicago, suggesting alloploidy there (Garsmeur et al. 2013). Overall, however, there is little polyploidy in Faboideae, and there has been extensive reduction in chromosome numbers after the original duplication. Bertioli et al. (2009) link genome regions that are variable with the presence of retrotransposons.

Transmission of plastids may be biparental (Corriveau & Coleman 1988). The mimosoid plastome is about 10,000 kb larger than that of the Faboideae listed (no caesalpinioids were mentioned), partly because of the extension of the inverted repeat and partly because of tandem repeat expansions (Dugas et al. 2015). Overall, nucleotide substitution rates in Faboidese were higher than those of other Fabaceae (Schwarz et al. 2017).

There have been major changes in chloroplast organisation in Faboideae in particular that, apart from their intrinsic interest, provide a considerable amount of phylogenetic structure. The chloroplast genome is notably labile, both in terms of sequence (Kua et al. 2012) and structure (see Martin et al. 2014 for a summary). Most Faboideae have a 50kb inversion in the trnL intron in the large single-copy region of their chloroplast genome, however, taxa like Swartzieae, Cladrastis, and a few others, lack this inversion (J. J. Doyle et al. 1996, 1997; Pennington et al. 2001; Wojciechowski et al. 2004). The loss of the chloroplast IR characterises a largely temperate, herbaceous and very speciose group, (see above: Wojciechowski 2003 and references). Temperate members of the IRLC lack the clpP intron, while the rps12 intron has been lost from all members of the clade examined except Wisteria, Callerya and Afgekia - but not Glycyrrhiza (Jansen et al. 2008; Wojciechowski et al. 2008; see also Saski et al. 2007; Cai et al. 2008; c.f. Sabir et al. 2014 in part). Nucleotide substitution rates in the IRLC tend to be lower than in taxa with the IR (Schwarz et al. 2017). Genome evolution in taxa of the ILRC has also been considerable (Magee et al. 2010; Sabir et al. 2014; Schwarz et al. 2015). Pisum and Lathyrus have diffuse centromeres/holocentric chromosomes, and this follows a duplication of the CenH3 gene (Neumann et al. 2015). Desmodium and possibly related genera (MILL) have also lost the rps12 intron (it has moved to the nucleus) (Doyle et al. 1995; Bailey et al. 1997; Jansen et al. 2008). ORF 184 has been lost many times, especially in the MILL clade, and accD (= ORF 512, zpfA) has also been lost (Doyle et al. 1995). Both the rps16 and ycf4 genes are lost in many Faboideae (Doyle et al. 1995; Jansen et al. 2007), the former being absent in all members of the IRL clade, but also being lost in four other clades in Faboideae (Schwarz et al. 2015).

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

Chemistry, Morphology, etc. The diversity of secondary metabolites in Fabaceae, perhaps especially in Faboideae, is remarkable (see also above). For instance, about 28% of all known flavonoids and about 95% of the isoflavonoid aglycone structures - over 1,000 alone - have been found there, and isoflavonoids are restricted to Faboideae. Isoflavonoids are both phytoalexins involved in plant defence and may also play a role in nodulation (Hegnauer & Grayer-Barkmeijer 1993; Reynaud et al. 2004). Flavonoids lacking the 5-hydroxy group are characteristic of Fabaceae (Seigler 2003), but I do not know at what level they are apomorphic. Pea albumin, a small sulphur-rich peptide involved in food storage and also with insecticidal properties, is known only from Faboideae, and the albumin-1 gene may be a synapomorphy for the [hologalegina + millettioid] clade (Louis et al. 2007); it is not to be found in some/all robinioids, and has been transferred at least twice to parasites (Cuscuta, Orobanche and relatives - Y. Zhang et al. 2013). Despite the diversity of secondary metabolites in the family, correlation of metabolites with clades is for the most part poor (Wink 2013). Colville et al. (2015) linked the presence of homoglutathione to the genome duplication in Faboideae and to the Old World clade; the function of homoglutathione, unlike that of glutathione, was unclear. The Australian Gastrolobium (Mirbelieae) produces the toxic sodium monofluoroacetate (Chandler et al. 2001).

Characters of caesalpinioid woods include rays that are usually more than 20 cells tall, presence of silica bodies, and axial canals (Evans et al. 2006); I do not know how these fit on to the current tree. Some of the mimosoid clade and Faboideae have leaves that are rich in silica (Westbrook et al. 2009). Colleters have been reported from "Caesalpinioideae", especially Chamaecrista (Coutinho et al. 2015 and refs); Bauhinia has "staminodial" structures at the base of the ovary (Endress 2008c) that may also have something to do with colleters.

The parts of the flowers of most Fabaceae develop in unusual sequences, often sepals-carpels-petals-outer stamens-inner stamens, and there are other distinctive features of their development (e.g. Champagne et al. 2007; Feng et al. 2006; Wang et al. 2008). For variation in general patterns of floral development, see Prenner and Klitgaard (2008b); they emphasized the diversity of developmental patterns even within the corolla whorl, thus although both Duparquetia and Faboideae have the adaxial petal in the outermost position, yet the two get there by developmentally different pathways.

Floral variation is considerable in members of the old Caesalpinioideae (e.g. Prenner & Klitgaard 2008b: p. 1363; see also Zimmerman et al. 2013b, 2017). When there is the complete loss of individual floral structures in evolution, floral development can be greatly changed (Tucker 1988, 2000), although, as Bruneau et al. (2014) noted, in Detarioideae organs in one whorl can be lost without much affecting the development of the rest of the flower - as with the complete loss of a calyx member in Duparquetia and Goniorrhachis (Prenner & Klitgaard 2008b; Prenner & Cardoso 2016). For additional information about floral development in "caesalpinioids", see also Tucker (1996a, b, 2001, 2002b, c, 2003c [all Detarioideae]), Tucker (1998) and Zimmerman et al. (2013a, b), all Dialiinae - complete organ loss common, Mansano and Teixeira (2008: Lecointea clade), Pedersoli et al. (2010: Copaifera, Detarioideae), and Bruneau et al. (2014: Detarioideae). Prenner and Cardoso (2016) note that in Detarioideae a 4-merous calyx may be the result of fusion of two members or, in Goniorrhachis, the absolute suppression of one member, while in Dialioideae it is the result of the loss of one member (Zimmerman et al. 2017).

For the adaxial sepal member in the mimosoid clade, see Ramírez and Tucker (1990); they describe a variety of developmental pathways that result in the connate calyx of that subfamily. For more floral development in the mimosoid clade, see Gemmeke (1982); stamens may develop centripetally on five main primordia. De Barros et al. (2017: useful tables) describe floral development of taxa in and close to the mimosoid clade. The cochlear-descending calyx aestivation, helically-initiated androecium, etc., of Calliandra s. str., are distinctive within this clade (Prenner 2004b).

Prenner (2013a, b: esp. androecium) surveyed floral development in Faboideae and suggested that there was a slight asymmetry in the early development of the androecium (the adaxial median stamen is initiated slightly off the median axis) in more "basal" Faboideae and also in some "Caesalpinioideae" (Prenner 2004c). Variation in floral morphology and development is considerable in clades of "basal" Faboideae, e.g. Cardoso et al. (2013a, b). Pennington et al. (2000) discussed floral evolution in these Faboideae, some of which like Swartzia have flowers with very derived morphologies; Swartzia lacks petal rudiments, but these are to be found in Amburana, of the ADA clade (Leite et al. 2015). Androecial initiation in Swartzia can be both centripetal and centrifugal (Tucker 2003b). The pattern of initiation of the sepals and stamens is variable, by no means always being unidirectional (e.g. Prenner 2004a; de Chiara Moço & de Araujo Mariath 2009; Leite et al. 2015). Floral variation in basal Faboideae is quite extensive - for instance, although the flower of Petaladenium is "faboid", the keel petals are not fused and there are no locking devices on the petals - suggesting that there was an early "experimental" phase in floral evolution in this clade (Tucker 1993; leite et al. 2014; Prenner et al. 2015). Prenner (2013b) found that petal initiation in Abrus, as in some other Faboideae, was simultaneous. There are CA primordia in the flower, A initiation is bidirectional, and there is overlap in the timing of C, A, and G initiation is members of the ILR clade (Naghiloo & Dadpour 2010). The flowers of some Amorpheae have a stemonozone rather than a hypanthium, i.e. the staminal tube is adnate to the petals (McMahon & Hufford 2002). For floral and inflorescence morphology, especially in Faboideae-Loteae, see Sokoloff et al. (2007a). Klitgård et al. (2013) found that polysymmetric flowers, sometimes with long, linear petals, had evolved ca four times in the Pterocarpus clade alone, which otherwise has monosymmetric papilionoid flowers. The calyx of Parkia multijuga (the mimosoid clade) is quite monosymmetric, especially in bud (Pedersoli & Texeira 2015). See also Mansano et al. (2002: Swartzieae s.l.), Song et al. (2011: Clianthus), Paulino et al. (2011: Indigofera, 2013: Swartzia).

Guinet (1981a) outlined some major patterns of pollen variation in the family. Hesse (1986) noted that both Bauhinia and Cercis - and Caesalpinia and Delonix - had pollen-connecting threads made up of something other than sporopollenin. For pollen variation in "Caesalpinioideae", see Graham and Barker (1981), Banks et al. (2003) and Banks and Rudall (2016) and references, also Banks and Klitgaard (2000: Detarioideae), Banks and Lewis (2009: Dimorphandra group), and Banks et al. (2013, 2014: Cercidoideae, diverse); Cercidoideae and Detarioideae are particularly variable, while the pollen of Duparquetia is unique among angiosperms (Banks et al. 2006). Many of the mimosoid clade have polyads, which vary in size and aperture number, etc. (see Guinet 1969, 1981b; Feuer 1987; Banks et al. 2011). The recently-described Afrocalliandra has a 7-celled polyad (de Souza et al. 2013, also literature for mimosoid pollen); I have no idea how this develops. For polyads, anther dehiscence, etc., in some of the mimosoid clade see Teppner (2007) and Teppner and Stabentheiner (2007) and references. Aperture position in the mimosoid clade does not follow Fischer's rule (Banks et al. 2010). Ferguson and Skvarla (1981) discuss pollen of Papilionoideae (see also Kuriakose 2007). The extensive palynological variation needs ro be placed in the context of a full tribal-level phylogeny.

Compared to variation in other parts of the flower, that of the gynoecium is slight: There is nearly always a single carpel with the same orientation. The style is at least sometimes hollow, although the cavity arises in various ways, including by lysigeny (Lersten 2004). The carpels may have five traces and are quite often open during development in "Caesalpinioideae", but not, apparently, in Cercidoideae, the mimosoid clade or Faboideae (Tucker & Kantz 2001). The embryo sac of some Faboideae (?elsewhere) more or less protrudes into the micropyle, as in Archevaletaia (Maheshwari 1950). There is a great deal of variation in the development of the embryo suspensor, even within Faboideae, especially in Vicia where endopolyploidy in the suspensor can reach 8,000C (Lersten 1983; also Tucker 1987; Yeung & Meinke 1993; Rodriguez-Pontes 2008; Endo 2012b; Shi et al. (2015: coleorhiza in some taxa?). Both a true (integumentary) endothelium and a nucellar endothelium may be present in Faboideae (Rodrigues-Pontes 2008 for discussion and references). The two recurrent vascular bundles lateral to the hilum are absent in basal Faboideae (Lackey 2009). D. L. Smith (1981) discussed cotyledon vasculature and anatomy; quite variable. For the aborting plumule in seedlings of Lotus and Coronilla and their relatives, see Dormer (1945).

For additional general information see Polhill and Raven (1981), Ferguson and Tucker (1994), Crisp and Doyle (1995), Doyle and Luckow (2003), and Lewis et al. (2005: geographic distributions, illustrations, etc. of all genera); for much information about Genisteae, see Polhill (1976) and about Dialoideae, see Zimmerman et al. (2017). For chemistry, about which a great deal is known, see e.g. Hegnauer (1994, 1996), Southon (1994), and Hegnauer and Hegnauer (2001), also Frohne and Jensen (1992) and Waterman (1994: secondary metabolites in general), Wink and Waterman (1999: evolution of secondary metabolites), Wink and Mohamed (2003: particularly useful), Dixon and Sumner (2003), and Wink (2003, 2013). For polysaccharides and flavonoids in particular, see Hegnauer and Grayer-Barkmeijer (1993) and Harborne and Baxter (1999), for gums and resins, see Lambert et al. (2009, 2013), for the synthesis of quinolizidine alkaloids (from cadaverine), see Bunsupa et al. (2012), for terpenoids, see Langenheim (1981, 2003), for alkaloids, see Aniszewski (2007), for non-cyanogenic hydroxynitrile glucosides, see Bjarnholt et al. (2008 and references), and for glucosylceramides, see Minamioka and Imai (2009). For starch, see Czaja (1978) and for epidermal wax crystals, see Ditsch et al. (1995).

For wood anatomy, see Baretta-Kuipers (1981), Wheeler and Baas (1992: esp. fossil woods), Gasson et al. (2000 [Faboideae], 2003 ["Caesalpinioideae"], 2009 [Caesalpinieae], and references), Evans et al. (2006: the mimosoid clade), Oskolski et al. (2014: Crotalarieae), for distinctive secondary thickening in Bauhinia, see Carlquist (2013) and Fisher and Blanco (2014), for foliar variation in basal Caesalpinioideae, see Lersten and Curtis (1994), and for the diversity of crystals in the bark of African genistoids and their possible evolution, see Kotina et al. (2015). Luckow et al. (2005) discuss variation in flower and seed in the mimosoid clade; for general floral and inflorescence morphology, see Endress (1994b), Naghiloo et al. (2012: variability), and Prenner (2013a: Faboideae, 2013b: Faboideae, esp. androecial variation), for floral anatomy, see Rao et al. (1958), for floral morphology, see Tucker (1993: ex Sophoreae), Crozier and Thomas (1993: Glycine), Kantz and Tucker (1994: Caesalpinia s.l.) and Pedersoli and Texeira (2015, also references: two mimosoids), for mimosoid anther glands, see de Barros et al. (2016), for carpel development, van Heel (1981, 1983), van Heel (1993: Archidendron, if five carpels, alternate with corolla), Sinjushin (2014: polymerous gynoecia in peas), and Paulino et al. (2014: polycarpellary gynoecia usu. in the mimosoid clade, unknown from Faboideae with keel flowers). For endothecial thickenings, see Manning and Stirton (1994), for tapetum, see Wunderlich (1954: c.f. Caesalpinia) and Buss and Lersten (1975), for embryology, etc., see Guignard (1881), Newman (1934), Dnyansagar (1970), Rugenstein (1983: Cercidieae), Cameron and Prakash (1990: giant antipodal cells in some Bossiaeeae, Indigofereae, Mirbelieae, 1994: Faboideae megagametophyte v. variable) and De-Paula and Oliveira (2012: Chamaecrista ovules). For funicle morphology, see Endo (2012a). For cotyledon areole presence, the size of the areole being linked to the amount of endosperm, see Endo and Ohashi (1998) and Lackey (2011), for endosperm, see e.g. Rau (1953), Johri and Garg (1959) and Rodrigues-Pontes (2008), both haustoria, and for galactoomannans, see Nadelmann (1890), Reid (1985) and Meier and Reid (1982: Lupinus has galactans in its cotyledons), and for xyloglucans, see Kooiman (1960) - the ratio of galactose to mannose in galactomannans may be of phylogenetic interest (Buckeridge et al. 1995, 2000a, b). For information about seed coat morphology and anatomy, see e.g. Pammel (1899), Corner (1976), Gunn (1981, 1984, 2003), Kirkbride and Wiersema (1997), Moïse et al. (2005), and Lackey (2009), for the pleurogram, esp. in Chamaecrista, see De-Paula and Oliveira (2008, 2012), for fruit anatomy in Crotalaria and relatives, see Le Roux et al. (2011), for fruits and seeds in "Caesalpinoideae", see Gunn (1991: features other than those noted in the characterisations may be of systematic interest), also Pfeiffer (1891), Kapil et al. (1980), etc., for seedlings of African Detarioideae, see Léonard (1957), a classic study, and for chromosome numbers, see Goldblatt (1981) and Cannon (2014).

Phylogeny. The Legume Phylogeny Working Group (2013a) provide a good summary of relationships in the family. Fabaceae are monophyletic in both molecular and morphological analyses, although support may not be strong. "Caesalpinioideae" are wildly paraphyletic at the base of Fabaceae, with the mimosoid clade and Faboideae clearly being monophyletic and separately embedded in "Caesalpinioideae". Cercidoideae, Duparquetia, and/or Detarioideae are all candidates for being sister to the rest of the family (Bruneau et al. 2008a, b; Cardoso et al. 2012a: Duparquetia not included; Legume Phylogeny Working Group 2013a). Morphology and anatomy support this: All three lack the vestured pits that are common in the rest of the family (they are also absent in Caesalpinioideae-Cassieae). Duparquetia was found to be sister to Dialioideae by Herendeen et al. (2003b), while Cardoso et al. (2013b) found some support for the topology [Duparquetia [Cercidoideae + Detarioideae] [[Dialioideae [the mimosoid clade + Faboideae]]]. In addition to placing Cercideae as sister to the rest of Fabaceae, within the latter clade Wojciechowski et al. (2004) found that Dialioideae were sister to the remainder. Two main clades made up the rest of the family. One includes the mimosoid clade, to which Ceratonia, Gleditsia, etc., Caesalpinieae, Cassieae, and Cercideae (all Caesalpinioideae) were more or less successively sister taxa, and the other is made up of Faboideae. Bruneau et al. (2008a, b) found a somewhat similar set of relationships - [Detarieae [Duparquetia, Cercideae, [Dialiieae [Faboideae [Caesalpinioideae + Mimosoideae]]]]]. Cercis and Bauhinia may be sister to all other Fabaceae (e.g. J. J. Doyle et al. 2000 and references; Bruneau et al. 2001), although they are also placed sister to Detarioideae, if sometimes with only with moderate support (Wojciechowski et al. 2004; Lavin et al. 2005; Forest et al. 2007b; Cardoso et al. 2013b; Cannon et al. 2014: only one Detarioideae sampled). See also M. Sun et al. (2016) for relationships - they found "Caesalpinioideae" in eleven separate clades - and Z.-D. Chen et al. (2016).

For Duparquetia, see Forest et al. (2002) and Tucker et al. (2002). The genus is highly derived and its pollen is very distinctive (Banks et al. 2006), but although the carpel develops after the stamens are initiated, unlike other Fabaceae (Prenner & Klitgaard 2008a, esp. b), this could be plesiomorphic.

Within Cercidoideae, Cercis is sister to all other members of the clade (Sinou et al. 2009), and within Cercis, C. chungii is sister to the rest, the North American and European species being embedded in the tree (Jia & Manchester 2014). Generic limits around Bauhinia are discussed by Sinou et al. (2008, 2009); Bauhinia s. str. and immediate relatives, but not other Cercidoideae, lack the plastid rpl2 intron (e.g. Lai et al. 1997; Sinou et al. 2009).

Detarioideae (Bruneau 2000; Mackinder 2005: genera) show extensive loss of sepals and/or petals and/or stamens, and linked with the first two changes in particular, increase in size of the bracteoles and in stamen number (Tucker 1992b, 2000, etc.). Thus in Monopetalanthus durandii the flower is surrounded by bracteoles and the floral formula is K 1 (minute), C 1; A 10; G 1, and Brachystegia glaucescens also has large bracteoles, K 5 (all small), C 0; A 10; G 1 (Tucker 2000). Resins found in some Detarioideae contain distinctive bicylic diterpenes, and may be an apomorphy for a subgroup within the tribe (Fougère-Danezan et al. 2007). Pollen is also extremely variable (Banks & Klitgaard 2000; Banks 2003). There is amyloid in the cotyledons (Hegnauer & Grayer-Barkmeijer 1993), also found in Sclerolobieae (= Tachigalieae: see Kooiman 1960; Meier & Reid 1982).

Gigasiphon, with two-ranked unifoliolate leaves, may be sister to the rest of the clade (M. Sun et al. 2016 and references). However, de la Estrella et al. (2017) found [Schotia [Barnebydendron + Goniorrhachis]] to be sister to Detarieae/resin-producing members of the subfamily, and Saraca and Afzelia clades successively sister to the rest of the subfamily. Relationships in Bruneau et al. (2008a) were similar, except there was a basal tetratomy made up of Schotia, Barnebydendron, the resin-producing taxa, and the rest. Redden et al. (2010) examined relationships in the Brownea clade, possible synapomorphies for it being an unchanneled leaf rachis, thread-like stipules, connate bracteoles, four sepals, and introrse anthers. For relationships around Gilbertiodendron, see de la Estrella et al. (2014).

A clade made up of the ex-caesalpinioid Dialiinae seems well supported as sister to the rest of Fabaceae (see also Cardoso et al. 2013b); the monotypic neotropical Poeppigia is sister to the rest of the genera in this clade (Bruneau et al. 2008a; M. Sun et al. 2016; Zimmerman et al. 2017). [[Eliigmocarpus + Badoinia] [Zenia + rest of clade], with a fair number of polytomies complete the relationships as known (Zimmerman et al. 2017).

There are then two large clades.

1. [The mimosoid clade + some of the old Caesalpinioideae] = Caesalpinioideae. Mimosoideae in the old sense (= the mimosoid clade) are monophyletic, and have distinctive small polysymmetric flowers, but immediately basal to them are several clades made up of Caesalpinioideae, in which genera like Caesalpinia, Cassia, and Dimorphandra have large more or less monosymmetic/asymmetric flowers. In Caesalpinioideae, the poorly-supported Umtiza clade may be sister to the rest. This clade includes taxa like Gleditsia, Gymnocladus and Ceratonia, several of which are dioecious and have smallish, greenish flowers sometimes with a poorly differentiated calyx and corolla - not plesiomorphic features (Herendeen et al. 2003a; Forest et al. 2007b; see also Redden & Herendeen 2006: morphological analysis; Fougère-Danezan et al. 2003, 2007, 2010: molecular and morphological studies; Legume Phylogeny Working Group 2013a, 2017; Cardoso et al. 2013b). Other caesalpinioids with small, more or less simultaneously-opening flowers borne close together (e.g. Dimorphandra) are in this area of the tree (Luckow et al. 2003; Wojciechowski 2003; Bruneau et al. 2008a, b). They show considerable similarity to the mimosoid clade in wood anatomy (Evans et al. 2006) and also in pollen, which is rather homogeneous although nearly always in monads (Banks & Lewis 2009). Genera like Pentaclethra are also to be included (c.f. Bouchenak-Khelladi et al. 2010b, but some confusion there?), while taxa like Dinizia, Pachyelasma and Erythrophleum, also with racemose inflorescences and small, more or less polysymmetric flowers with free sepals and petals (and ten stamens), were in a grade immediately basal to the mimosoid clade (Bouchenak-Khelladi et al. 2010b); there is a considerable amount of phylogenetic structure here (Kyalangalilwa et al. 2013). Caesalpinia, Cassia, and relatives are also basal to the mimosoid clade, Dimorphandra, Peltophorum, etc. (Cardoso et al. 2012c). In general, relationships between these ex-caesalpinioids remain unclear (Cardoso et al. 2012c, 2013b; Legume Phylogeny Working Group 2013a, b). In this account I have listed the major clades found by Manzanilla and Bruneau (2012; see also ), although they examined other genera that were not included in any of the groupings above. They found that Pterogyne associated with Caesalpinieae in plastid analyses, but with Cassieae in nuclear analyses (Manzanilla & Bruneau 2012). For relationships in this clade, see also M. Sun et al. (2016).

For Peltophorum and its relatives, see Haston et al. (2005). For relationships in Caesalpinieae, see Simpson et al. (2003). Gagnon et al. (2013) discussed relationships around the old Caesalpinia; the genus is either poly- or paraphyletic (Manzanilla & Bruneau (2012). For phylogenetic relationships within Senna, see Marazzi et al. (2006), and for relationships within Chamaecrista, see de Souza Conceição et al. (2009).

Within the mimosoid clade, there is a large monophyletic clade, Mimoseae I, in which Ingeae, derived, with a valvate calyx and many stamens connate and forming a tube, are embedded in Acacieae (e.g. Clarke et al. 2000; Robinson & Harris 2000; Miller & Bayer 2000, 2001; Luckow et al. 2003; Jobson & Luckow 2007; Brown 2008; Brown et al. 2008; Bouchenak-Khelladi et al. 2010b; Miller & Seigler 2012; Kyalangalilwa et al. 2013). See Richardson et al. (2001b) for the diversification of Inga. The old Acacia subgenus Acacia, which includes the bull's horn acacias, seems to be monophyletic, but Acacia s.l. is highly poly/paraphyletic. Murphy et al. (2000, 2003, 2010 [relationships still only moderately resolved]) and Miller et al. (2003) discuss the phylogeny of Acacia s. str., the old subgenus Phyllodinae. There was little support for basal relationships in the comprehensive analysis of Mishler et al. (2014), although this was improved in a study using complete chloroplast genome sequences (Williams et al. 2016). Brown et al. (2012) focussed on relationships of Acacia s. str. outside Australia. Miller and Bayer (2003) looked at relationships in Vachellia, the old subgenus Acacia, and Senegalia, the old subgenus Aculeiferum (support for the monophyly of this is weak - Miller & Seigler 2012). Kyalangalilwa et al. (2013) looked at the whole complex; there may still be clades of the old Acacia without names. Siegler (2003) summarized the phytochemistry of the complex, Evans et al. (2006) detailed wood anatomy, and Kergoat et al. (2007) noted what bruchids had to say about systematics of Acacia s.l.; see also Muelleria 26(1). 2008, a special issue on Acacia, and also World Wide Wattle website. Albizzia is also in Mimoseae I, and it seems potentially quite polyphyletic (Kyalangalilwa et al. 2013). Mimosa may be monophyletic and sister to Piptadenia (Besseger et al. 2008); this relationships was also found by Simon et al. (2015), whose focus was on Stryphnodendron. Simon et al. (2011) provide an extensive phylogeny of Mimosa, optimizing various characters on the tree. De Souza et al. (2013) provide a comprehensive phlyogeny of Calliandra - largely, if not entirely, New World - and its immediate relatives.

Mimoseae II, named for convenience, make a paraphyletic grade basal to Mimoseae I. They include Parkia, a polyphyletic Neptunia, Leucaena, etc., as well as genera like Pentaclethra (Kyalangalilwa et al. 2013: see also above). Catalano et al. (2008) provide a phylogeny of the ecologically important New World genus Prosopis, also in this area; see also Iganci et al. (2015) for Abarema.

2. Faboideae are monophyletic. The following topology (simplified) aids in the discussion of relationships in Faboideae: [ADA clade, Swartzieae, etc. [Cladrastis, etc., [(= 50 KB inversion clade) Andira et al., Vatairea et al., Exostyleae/Lecointea et al., genistoids: GEN, [Amorpheae + dalbergioids = dalbergioids s.l.: DAL], [baphioids: BAPH [(= Non-Protein Amino Acid Accumulating - NPAAA - clade) mirbelioids: MIRB [[Indigofereae + millettioids: MILL] [robinioids: ROB + Inverted Repeat Loss Clade: IRLC]]]]]]], seems moderately well supported (Wojciechowski 2003; McMahon & Sanderson 2006; Legume Phylogeny Working Group 2013a). The [robinioids + IRLC] clade is the hologalegina clade (see Farruggia & Howard 2011 for possible nuclear markers). The tree here is based largely on Wojciechowski et al. (2004), Peters et al. (2010) and Cardoso et al. (2012c, 2013b); see also the Legume Phylogeny Working Group (2013a), McMahon and Sanderson (2006: a supermatrix analysis of 2228 species) and M. Sun et al. (2016) for further details.

Swartzieae may be sister to other Faboideae, but support is weak (Ireland et al. 2000; Pennington et al. 2000; Lavin et al. 2005). Indeed, Cardoso et al. (2012a, 2013b) found the well-supported [Angylocalyceae [Dipterygeae + Amburaneae]], the ADA clade, to be sister to all other Faboideae, although this position must be confirmed (see also M. Sun et al. 2016). Cardoso et al. (2013b; also M. Sun et al. 2016) found some support for a Swartzieae s.l., i.e. [Swartzieae s. str. + Ateleia, etc.], as sister to remaining Faboideae (see Torke & Schaal 2008 for a phylogeny). For the lecointeoids (= Exostyleae), see Mansano et al. (2004) and for the vataireoids, see Cardoso et al. (2013a). Cardoso et al. (2012a, esp. 2013b) discussed other relationships among Faboideae basal to the NPAAA clade. Cladrastis and relatives are sister to the 50 KB inversion clade (e.g. Cardoso et al. 2013b).

50 KB inversion clade. Relationships at the base of this clade form an extensive polytomy (9-furcate in Cardoso et al. 2013b). The ectomycorrhizal Aldina is placed in a basal clade with Andira and Hymenolobium, all members of which have a drupaceous fruit, an undifferentiated testa, no endosperm, etc. (Ramos et al. 2015).

GEN: x = 9. Crisp et al. (2000) outline relationships in Genisteae s.l., with their distinctive quinolizidine (pyrrolizidine) alkaloids (see also van Wyk 2003 for chemistry). Genista is perhaps to include Ulex and other genera, but relationships between the major groups in the genus are poorly supported; Cytisus is paraphyletic (Pardo et al. 2004). For relationships within Crotalarieae, see Boatwright et al. (2008b, esp. c, 2009) and of Lotononis and relatives in particular, Boatwright et al. (2011: also character evolution). Le Roux et al. (2013) found little support for relationships along the backbone of Crotalaria, although several well-supported clades were recovered. For relationships within Psoraleae, see Egan and Crandall (2008), and in Canary Island Genisteae, see Percy and Cronk (2002), for diversification in Cape genistoids, see Edwards and Hawkins (2007); for relationships in Ononis, see Liston (1995), and in Cytisus, which includes Ulex, see Cubas et al. (2002) and Cristofilini and Troia (2006). Bacterial infection is through the epidermis, nodule morphology is very various (J. J. Doyle 2011).

DAL: Nodules small, oblate, determinate, always associated with a lateral root common [desmodioid nodules = aeschynomenoid nodules - Lavin et al. 2001]; (leaves opposite -= Platymiscium); x = 10. For the phylogeny of dalbergioid legumes, see Lavin et al. (2000). Machaerium is more related to Aeschynomene section Ochopodium (that genus is polyphyletic) than to Dalbergia, so the apparent similarities in habit, fruit, etc., between Machaerium and Dalbergia need re-evaluating (Ribeiro et al. 2007; Chaintreuil et al. 2013). Cardoso et al. (2012a, b) and Klitgård et al. (2013) noted that taxa with polysymmetric flowers had evolved several times in this clade. For relationships within Amorpheae and the floral evolution of the latter (petals may be lost, or all look rather similar; a stemonozone may be developed; etc.), see McMahon and Hufford (2002, 2004, 2005) and McMahon (2005). The large genus Dalbergia is likely to be monophyletic (Vatanparast et al. 2013). For relationships in Arachis, see Krapovickas and Gregory (2007).

MIRB: For relationships in or revisions of Mirbelia s.l., see Crisp and Cook (2003a, b), Gastrolobium, Chandler et al. (2001), Pultenaea, Orthia et al. (2005b), and Jacksonia, Chappill et al. (2007). [Mirbelieae + Hypocalypteae] are around 55 m.y.o. (Schrire et al. 2005).

MILL: For characters of the millettioid clade, see Tucker (1987a); see da Sila (2012) for a phylogeny of part of this clade, Lonchocarpus is split. [Milletieae + Indigofereae] may share early expression of monosymmetry in floral development (Paulino et al. 2011). For the delimitation of Millettieae, see Lavin et al. (1998); Hu (2000) and Hu et al. (2000) studied their phylogeny, and Abrus (with nine stamens) is (near-)basal (Prenner 2013b and references). Kajita et al. (2001: rbcL) also looked at the phylogeny of Millettieae and relatives, and in an extensive study de Queiroz et al. (2015) found the relationships [Indigofereae [Clitorieae [Phaseoleae, etc. [Abreae [Diocleae + Millettieae]]]]], although the relationships obtained depended on the markers used (Egan et al. 2016).

Indigofereae. These have true racemes (Wojciechowski et al. 2004); Desmodium et al. lack the chloroplast rps12 intron (Jansen et al. 2008). Barker et al. (2000) and Schrire et al. (2009) disentangle relationships within the tribe, finding considerable phylogenetic structure (i.a. there are four major clades within Indigofera) that can be linked with both morphology and ecology. Phaseoleae, etc.. Stefanovic et al. (2009: eight chloroplast genes) concentrated on relationships among the some 2,000 species of phaseoloids, finding substantial resolution, i.a. Mucuna sister to Desmodium and its relatives, the combined clade being sister to the rest of the group, which also includes Cajanus, Vigna, Erythrina, etc. (see also H. Li et al. 2013 for a more detailed analysis of Phaseoleae-Phaseolineae, ¾ genera included). For a phylogeny of Phaseolus itself, see Delgado-Salinas et al. (1999, 2006). In a number of phaseolids nodules are determinate, infection usually being by root hairs, i.e. the nodules are desmodioid (J. J. Doyle 2011). Vigna has to be dismembered (Delgado-Salinas et al. 2011), as does Pueraria (of kudzu vine fame), members of which are in five widely separate clades (Egan et al. 2016). For relationships in the pantropical Mucuna, nearly all lianes, see Moura et al. (2015) and for the optimisation of some fruit and seed characters which correlate nicely with the subgenera, see de Moura et al. (2016). The pantropical Erythrina is not monophyletic (de Moura et al. 2011); for the relationships of Psoralea, see Dludlu et al. (2013). Brongniartieae. Thompson et al. (2001) looked at relationships within this Australian-South American clade; see also Cardoso et al. (2016) and de Queiroz et al. (2017) for characters, relationships and circumscription. Diocleae. De Queiroz et al. (2015) looked in some detail at relationships in Diocleae, which includes a number of New World lianes, and found genera like Camptosema and Galactia were very much polyphyletic; for relationships in Canavalia, see Snak et al. (2016: stem, ca 15.8 m.y.o., crown, (11.1-)8.7(-6.7) m.y.). Millettieae. For the phylogeny of Derris and its immediate relatives, see Sirichamorn et al. (2012, 2014a, b).

ROB: For the phylogeny of Robinia and its relatives, which include Lotus and Sesbania, two genera that are quite close, see Wojciechowski et al. (2000), Lavin et al. (2003: Gliricidia area), and Farruggia and Wojciechowski (2009: Sesbania), and of Canary Island Loteae, see Allan et al. (2004). Degtareva et al. (2012) looked at relationships of Anthyllis, and Allan et al. (2003) at those around Lotus, with Hammatolobium being sister to Old World Lotus, and the unrelated New World Lotus perhaps including Ornithopus and other genera in a basal polytomy (see also Degtjareva et al. 2008; Kramina et al. 2016).

IRLC clade. Wojciechowski et al. (2000) outlined relationships in this speciose clade. At the base of the clade relationships are [Glycyrrhiza [Wisteria s.l. + ...]], Wisteria being embedded in Callerya, although C. atropurpurea was tending to wander between Wisteria s.l. and Glycyrrhiza (J. Li et al. 2014). Extensive phylogenetic studies (e.g. Wojciechowski 1993, 2004; Liston & Wheeler 1994; Wojciechowski et al. 1999; Kazempour Osaloo et al. 2003, 2005) show that Astragalus is monophyletic; most New World taxa are aneuploid (n = 11-15) and are also monophyletic, other species are base 8; f; the Old World [A. pelecinus + A. epiglottis] clade - the two are annual species - may be sister to the rest of the genu (Liston & Wheeler 1994; Azani et al. 2017). For general relationships in Old World Astragalus, see Kazempour Osaloo et al. (2003, 2005), Kazemi et al. (2009), Riahi et al. (2011) and especially Azani et al. (2017); infrageneric limits are indeed unsatisfactory. For relationships in New World Astragalus, see Scherson et al. (2005, 2008). Oxytropis is sister to Astragalus; for some relationships in the former, see Archambault and Strömvik (2012). Hedysareae and Galegeae are intermixed (Duan et al. 2015), and might best be combined; Hedysarum itself is polyphyletic. Within the combined clade, Safaei Chaei Kar et al. (2014) looked at relationships within Onobrychis and found that current infrageneric groupings had little support, Amirahmadi et al. (2016) providing an elaborated phylogeny, while for the phylogeny and diversification of Caragana in the context of the Qinghai-Tibetan Plateau uplift, see M. Zhang et al. (2009, 2015), Zhang and Fritsch (2010) and Duan et al. (2016). For the limits of Trifolieae, see Dangi et al. (2015). Within Trifolium the American species form a monophyletic group (Ellison et al. 2006; Liston et al. 2006). Phylogenetic relationships within Medicago have turned out to be highly reticulating (de Sousa et al. 2016 and references); Medicago perhaps including Trigonella, and for its limits, see Bena (2001), Steele et al. (2010) and Dangi et al. (2015: Medicago and Trigonella separate?). A preliminary phylogeny of Lathyrus suggested that the ca 20 South American species might represent a single clade derived from Northern hemisphere ancestors (Asmussen & Liston 1998; see also Kenicer et al. 2005). Relationships in the Vicia/Lathyrus area are complex, with Lenus and Pisum embdedded in the complex, Trigonella perhaps including Melilotus and Vicieae embedded in Trifolieae (e.g. Steele & Wojciechowski 2003; Schaefer et al. 2012). Within Hedysareae Alhagi may be sister to the rest (Amirahmadi et al. 2014). Fabeae are the subject of an extensive study by Schaefer et al. (2012).

Classification. In the past, Fabaceae have usually been divided into three groups, Fabaceae/oideae (= Leguminosae/-oideae), Mimosaceae/-oideae and Caesalpiniaceae/-ioideae, but the latter have turned out to be paraphyletic. See the Legume Phylogeny Working Group (2013b) for a fascinating introduction to the dynamics of the reclassification of the family; relationships at the deeper nodes are poorly known, but the old Mimosoideae are deeply embedded in Caesalpinioideae, so changes in the names for the major elements that make up the scaffolding of the family were to be expected. These have recently been published (Legume Phylogeny Working Group 2017), and i.a. the old Mimosoideae are now to be refered to as "the mimosoid clade", a not unreasonable decision pending clarification of relationships around the old Mimosoideae.

See Lewis et al. (2013) for a linear sequence of legume genera recognised as of March, 2013. Cardoso et al. (2013b) list the early-branching clades of Faboideae, i.e., those below the NPAAA clade, as well as the genera and numbers of species that they contain.

In general, a fair bit of adjustment to generic limits is also needed, but in some cases, although it is clear that there will be changes, sampling is not good enough to know yet what to do (e.g. see Percy & Cronk 2002; Allan et al. 2004; Ribeiro et al. 2007; Cardoso et al. 2012a; Dludlu et al. 2013). For generic limits in Cercidoideae, see Wunderlin (2010; Sinou et al. 2009 for the phylogeny), for those in the Caesalpinia group, initially unclear (Gagnon et al. 2013), see Gagnon et al. (2017), for those around Gastrolobium, see Chandler et al. (2001), for those around Vigna, see Delgado-Salinas et al. (2011), for those around Lotononis, see Boatwright et al. (2011), for those around Calliandra, see de Souza et al. (2013); the limits of Derris have been redrawn (Sirichamorn et al. 2014b). Unique combinations of floral characters can be used to recognize genera around Crotalaria (Le Roux & van Wyk 2012); see Orthia et al. (2005a, b) for the expanded generic limits of Pultenaea. Schaefer et al. (2012) suggest that one solution to the unexpected phylogenetic relationships they found in Fabeae might be to include Lens in Vicia and Pisum in Lathyrus...

The old Acacia subgenus Acacia, which includes the bull's horn acacias, seems to be monophyletic, but Acacia s.l. is polyphyletic. As Maslin (2001) noted sadly of the 955 or so species then placed in Acacia for the Flora of Australia treatment, "we are obliged to present the flora treatment in the absence of a more meaningful classification". However, things had already begun to change, although Pedley's (1986) solution following strict nomenclatural lines was proving only partly acceptab;e, and the argument became what names to use for the bits into which Acacia s.l. was clearly to be divided (Maslin et al. 2003). See Miller and Bayer (2003; also Boatwright et al. 2015) for Vachellia, the old subgenus Acacia, and Senegalia, the old subgenus Aculeiferum; see also Siegler et al. (2006, 2017) for other segregates. However, some remain deeply unsatisified with this nomenclatural solution.

See Pennington (1997) for a monograph of Inga and Dexter et al. (2010) for some species limits there. In Faboideae, for a sectional classification of the neotropical Swartzia, see Torke and Mansano (2009), for that of an expanded Cytisus, see Cristofilini and Troia (2006), for that of the pantropical Crotalaria, see le Roux et al. (2013), and for that of Obobrychis, see Amirahmadi et al. (2016). Thompson (2001) provides a careful study of E. Australian Hovea (Brongniartieae).

Previous Relationships. Fabaceae s.l. have often been placed in their own order, as in both Cronquist (1981) and Takhtajan (1997), and then they are usually divided into three families, Fabaceae/Leguminosae, Caesalpiniaceae and Mimosaceae. They can be confused with Connaraceae (Oxalidales), although the latter lack stipules, their flowers are polysymmetric and have stamens of two distinctly different lengths, and their gynoecium is frequently multicarpellate. However, in both the RP122 chloroplast gene has moved to the nucleus, and the ovaries of both have adaxial furrows (c.f. the ventral slit: Matthews & Endress 2002). Fabaceae have also been linked with Sapindaceae (e.g. Dickison 1981b), here in the malvids, but there is little support other than the common possession of compound leaves and non-protein amino acids for such an association.

Botanical Trivia. There has been as much diversification of the ycf4 protein, involved in photosystem 1 assembly, within Lathyrus as there has been between cyanobacteria and other angiosperms (Magee et al. 2010).

The seeds of Mora megistosperma (Caesalpinieae) are, at ca 18 x 12 cm, perhaps the largest of any broad-leaved angiosperm (Lewis et al. 2005), and the embryo is the largest of all angiosperms.

And when kudzu vine was thought to be a useful ground cover, soil stabilizer, etc., there were homecoming Kudzu Kings and Queens...

[Surianaceae + Polygalaceae]: embryo chlorophyllous.

Age. This node is dated to (71-)68, 66(-63) m.y.o. (Wikström et al. 2001).

SURIANACEAE Arnott, nom. cons.   Back to Fabales


Woody; ellagic acid?; cork also in inner cortex; storying +/0; wood fluorescing?; vessels in radial multiples; (sieve tube plastids with starch grains and protein filaments forming a peripheral shell); nodes 3:3 (1:1 - Suriana); (medullary vascular bundles - Recchia); sclereids +; petiole bundle arcuate to annular; leaves spiral or two-ranked, (unifacial), (pinnate, leaflets alternate, articulated), (stipules 0 - Suriana); inflorescence cymose, usu. terminal; pedicels articulated; K connate basally or not, quincuncial, C (0), contorted, shortly clawed or not; ?receptacular tissue ± forming a ring around the C base; A obdiplostemonous (= and opposite sepals); pollen surface variable (vermiform - Cadellia); nectary 0; (gynophore +, nectariferous - Recchia); G 1-5, when 5 opposite C, styluli separate, ± gynobasic, stigma clavate to capitate; compitum 0; ovules surrounded by mucilage, 1-5/carpel, apotropous [Suriana], campylotropous to amphitropous, unitegmic, integument 3-7 cells across, parietal tissue 4-5 cells across, (nucellar cap +), hypostase +; megaspore mother cells several, antipodal cells ± degenerate; fruit indehiscent, a berry, drupe or nut, endocarp with outer layer of palisade sclereids, other cells apart from the inner epidermis isodiametric, K persistent, accrescent or not; exotestal cells enlarged, cuboidal, tanniniferous, rest crushed [ca 7 cells thick], or seed tegmic; chalazal endosperm haustorium +, endosperm 0, embryo curved or folded, cotyledons incumbent; n = ?; germination epigeal, phanerocotylar.

5[list]/8 Mostly Australian, also Mexico and the Osa Peninsula, Costa Rica (Recchia); Suriana maritima is pantropical (map: from van Steenis & van Balgooy 1966 [blue - Suriana maritima]; FloraBase xi.2010). [Photo - Flower.]

Age. Crown-group Surianaceae are some (50.4-)38.7(-27.0) m.y. old (Bello et al. 2009).

Chemistry, Morphology, etc. The family is vegetatively heterogeneous, although its wood anatomy is quite homogeneous (Webber 1936). Both Cadellia and Recchia have thickened cell walls in the exocarp and sclereids in their bark parenchyma (Crayn et al. 1995).

There is no compitum (Armbruster et al. 2002). The exotesta of Suriana is described as being green (Rao 1970).

For more information, see Gutzwiller (1961) and Weberling et al. (1980), and Schneider (2006), all general, Hegnauer (1973, as Simaroubaceae), chemistry, Jadin (1901) and Boas (1913), both vegetative anatomy, Mauritzon (1939), Wiger (1935), Anantaswamy Rau (1940a), Rao (1970) and Heo and Tobe (1994), all embryology, etc., Gadek and Quinn (1992: pericarp); for floral development, see Bello et al. (2007/8: Suriana only). Additional data from: Cadellia - Benson s.n. = NSW 408528 (anatomy); Stylobasium - Latz 12864 (fruit), Strid 20708 (anatomy).

The vegetatively "atypical" Suriana is the only genus whose embryology has been studied and Surianaceae as a whole are little known chemically.

Phylogeny. [[Recchia + Cadellia] [Suriana [Guilfoylia + Stylobasium]] are suggested relationships in the family (Forest et al. 2007b); c.f. also Crayn et al. (1995) and Bello et al. (2009).

Classification. Although the sieve tube plastids of Stylobasium are distinctive, having sieve starch grains and protein filaments forming a peripheral shell (Behnke et al. 1996), there seems little reason to recognise Stylobasiaceae as a family, i.a. four families for five genera would then be needed.

Previous Relationships. Surianaceae were included in Rosales-Simaroubaceae (here in Sapindales) by Cronquist (1981) and in Rutales as a separate family by Takhtajan (1997).

Synonymy: Stylobasiaceae J. Agardh

POLYGALACEAE Hoffmannsegg & Link, nom. cons.   Back to Fabales

Saponins +; nodes 1:1; styloids 0; (tracheidal/fibrous/sclereidal cells); (stomata other than anomocytic); plant glabrous or with unicellular hairs; branching from previous flush; axillary buds 2 or more; lamina entire, often paired glands [crateriform extrafloral nectaries] or thorns at nodes (elsewhere); flowers monosymmetric, ± keeled; K quincuncial, C 5, keel ± apparent, (not clawed); A 8, ± connate, basally adnate to C, median adaxial A often absent; pollen polycolporate, surface psilate or foveolate; (disc excentric); G connate, style long, stigma dry; ovules epitropous, micropyle zigzag (endo-, exostomal), exostome often long, outer integument 2-6 cells across, inner integument (1-)2(-3) cells across, parietal tissue 1-3 cells across, nucellar cap 2-3 cells across, suprachalazal region ± massive; testa multiplicative, exotesta subsclerotic or otherwise distinct, endotestal cells ± palisade, U-thickened, crystalliferous or not; endosperm 0-copious; rpl22 gene transferred from chloroplast to nucleus [?sampling].

Ca 21[list]/965 - four tribes below. World-wide, except the Arctic and New Zealand. [Photo - Flower.]

Age. Diversification of Polygalaceae began in the Caenozoic (65.5-)57.4(-49.3) m.y.a. (Bello et al. 2009).

1. Xanthophylleae Chodat


Shrubs or trees; growth sympodial, terminal bud aborts; plants Al-accumulators; wood parenchyma apotracheal, diffuse; Petiole bundle annular, with inverted central plate; glands at nodes; (conspicuous domatia on leaf blades); inflorescence indeterminate; K unequal, C contorted, (no keel), (adaxial petals with colour patterning); A (7-10); G [2], placentation parietal, stigma small, bilobed (capitate); ovules 2-8(-20)/carpel, in two rows, outer integument 4-12 cells across; fruit a berry (irregularly loculicidally dehiscent), K deciduous; testa vascularized, strongly multiplicative, (not), (± crushed); hypostase massive [?level]; (endosperm starchy); n = 8 [1 species].

1/95. Indo-Malesia (map: from van der Meijden 1982).

Synonymy: Xanthophyllaceae Reveal & Hoogland

[Polygaleae, Carpolobieae, Moutabeae]: inflorescence cymose; A often monadelphous, anthers opening apically [pores/slits]; ovule 1/carpel; seed hairy (glabrous), exostomal/funicular aril + (0).

More or less world-wide (Map: from Wickens 1976; Frankenberg & Klaus 1980; Trop. Afr. Fl. Pl. Ecol. Distr. 1. 2003; GBIF 2009; Flora of China; Australia's Virtual Herbarium xii.2012; orange from Paiva 1998).

2. Polygaleae Chodat

Polygaleae, etc.

Herbs (echlorophyllous mycoheterotrophs), lianes, shrubs; (ergoline alkaloids +), at least some smell of wintergreen, tannins 0 [Polygala]; (successive cambia +); pits vestured; banded paratracheal parenchyma +; (glands at nodes); petiole bundle arcuate to annular; (flower asymmetrical); (2 abaxial lateral K, minute), two adaxial lateral K = wings, two connate adaxial C = the standard, abaxial C = the keel, often fringed [with crest], 2 abaxial-lateral C minute; (A 2-7), anthers with apical pores; G [2] (adaxial member suppressed), stylar canal + [Polygala], stigma bilobed, ± asymmetric, wet; ovule with (postament - Epirixanthes?), (antiraphe +); fruit an often flattened capsule, berry, drupe or samara, (K persistent, green - Polygala, etc.); hilar/chalazal/exostomal elaiosome + (0); testa ?multiplicative, (mesotesta +); n = 6+, very variable, nuclear genome size [1C] 0.43-1.39 pg.

18/830: Polygala (?350), Monnina (180), Muraltia (120), Securidaca (80). World-wide, except the Arctic and New Zealand (Map: from Wickens 1976; Frankenberg & Klaus 1980; Trop. Afr. Fl. Pl. Ecol. Distr. 1. 2003; GBIF 2009; Flora of China; Australia's Virtual Herbarium xii.2012; orange from Paiva 1998).

Age. The [Polygala + Securidaca] node is estimated to be (42-)40, 28(-26) m.y.o. (Wikström et al. 2001).

The distinctive Paleosecuridaca curtisii, from the Late Palaeocene of North Dakota ca 60 m.y.a., has fruits remarkably like those of Securidaca and seeds with a testa that has a well developed palisade layer, however, there are two seeds per carpel (Pigg et al. 2008b).

3. Carpolobieae Eriksen

Shrub to small tree, liane; (glands at nodes); (C contorted); A (4) 5, anthers with short confluent apical slits; G [3], stigma capitate; fruit?; exotesta fleshy [?both]; endosperm copious; n = 9-11.

2/6. Tropical Africa.

4. Moutabeae Chodat

Woody, (lianes); plants Al-accumulators; successive cambia +; banded apotracheal parenchyma +; petiole bundle annular, with wing bundles; glands on leaves (and at nodes); inflorescence often racemose; K adnate to C, abaxial C not keeled; A (6-10), anthers with short confluent apical slits; G [3-8], stigma capitate; fruit?; funicular aril +; n = 14.

4/15. Tropical America, New Guinea to New Caledonia. [Photos - Flowers, Flower - Close-up, Petioles, Branch, Petioles with ants, Flower with moth]

Synonymy: Diclidantheraceae J. Agardh, Moutabeaceae Pfeiffer

Evolution: Divergence & Distribution. N.B. The names of the genera mentioned below follows the usage in the papers cited. I have not attempted to clean up the nomenclature.

The evolution of elaiosomes in Polygalaceae is dated to (69.9-)54-50.5(-35.2) m.y.a., well after that of the ant clades concerned (by some estimates, at least), and it may have spurred diversification in the family (Forest et al. 2007b; Lengyel et al. 2009). Muraltia, also myrmecochorous and with some 120 species found mostly in the Cape region of South Africa, may have diversified quite recently, mostly within the last ca 10 m.y. (Forest et al. 2007a); Verboom et al. (2009) thought diversification started in the Fynbos (21.4-)18.5(-14.1) m.y.a. and in the Succulent Karoo (4-)2.5(-1.3) m.y. ago

Bello et al. (2012) list a number of apomorphies for the family and of several clades within it.

Ecology & Physiology. Xanthophyllum is one of the five most speciose genera in West Malesian l.t.r.f. (Davies et al. 2005).

Pollination Biology & Seed Dispersal. The "papilionoid" flower in Polygalaceae is quite differently constructed from that of Fabaceae (Westerkamp & Weber 1997, 1999; Bello et al. 2010, but see Prenner 2004d), although quite often both looking and being functionally similar. Note that the flowers of Polygala, which in overall appearance are particularly like those of some Fabaceae, do not represent the plesiomorphic condition of Polygalaceae; overall floral variation in Polygalaceae is considerable.

In many species of Polygala pollen is presented on the sterile lobe of the often rather complex, asymmetric stigma, i.e. secondary pollen presentation on a stylar brush (Weekley & Brothers 1996; see Brantjes 1982; Castro et al. 2008a for further details; Bello et al. 2010 for stigma morphology), other secondary pollination mechanisms known are explosive and pump pollination, and these two may both occur in a single flower - explosive pollination first and then pump pollination (Westerkamp & Weber 1997). The flowers of Polygala are complex, and details of pollination are correspondingly so.

Ant dispersal is quite common in Polygaleae in particular, and hilar/chalazal elaiosomes (the former are called caruncles) may be an apomorphy here; there may have been at least six origins of myrmechochory in the family (Forest et al. 2007b; Lengyel et al. 2009, 2010).

Bacterial/Fungal Associations. Epirixanthes is an echlorophyllous mycoheterotroph associated with glomeromycota (for details of the fungal association, see Imhof 2007; Imhof et al. 2013).

Vegetative Variation. Although genera like Xanthophyllum and some Moutabeae may have paired glands at the nodes, other genera seem to lack anything even faintly like stipules, and where stipules might be lost in this part of the tree is uncertain. De Aguiar-Dias et al. (2011) have recently suggested that the paired nectary glands at the base of the leaf in Polygala laureola are indeed stipules because they receive a vascular bundle from the single foliar vascular trace.

Genes & Genomes. There is a genome duplication somewhere around here (Cannon et al. 2014).

Chemistry, Morphology, etc. Polygala myrtifolia has eight stamens; the two stamens in the median plane, so on opposite sides of the flower, appear to have been lost (Prenner 2004d); see Bello et al. (2010) for other floral diagrams. The degree of connation of the filaments varies, as does that of their sometimes rather slight adnation to the petals. For floral morphology and development of Polygaleae, see Krüger and Robbertse (1988) and Krüger et al. (1988), and for that of the family as a whole, see Bello et al. (2010, 2012). The tricolpate pollen of Balgooya is probably derived; some Polygalaceae such as Heterosamara have asymmetric, almost boat-shaped pollen grains (Banks et al. 2008). Some species of Polygala, at least, have a stylar canal (Castro et al. 2008b). Monnina seems to have a nucellar cap ca 6 cells across, while the inner integument of Securidaca is up to 9 cells across in the endostomal region (Verkeke 1985). In indehiscent fruits the testa is more or less crushed (Rodrigue 1893; but c.f. Verkeke 1984, 1985). Verkeke (1985) distinguished between epitropous-dorsal ovules (Xanthophyllum) and epitropous-ventral ovules (the rest).

Additional information is taken from Eriksen (1993a) and especially from Eriksen and Persson (2006), both general; Chodat (1891, 1893) is still worth consulting. For chemistry, see Hegnauer (1969, 1990), for wood and leaf anatomy of Moutabeae, see Styer (1977), also Banks et al. (2008: pollen morphology and evolution), Manning and Stirton (1994: endothecial thickenings), and Verkeke and Bouman (1980), Verkeke (1991) and Takhtajan (2000), all ovule and seed. Also see Johow (1910) and Merckx et al. (2013a), both Epirixanthes, Paiva (1998: Polygala, especially Africa and Madagascar), and for a monograph of Xanthophyllum, van der Meijden (1982).

Phylogeny. Of the four groups mentioned above, Moutabeae may be paraphyletic (Persson 2001: trnL-F), although adding rbcL data suggests they are monophyletic (Forest, in Eriksen & Persson 2006), and morphology also points in this direction (Eriksen 1993b); the other three groups appear to be monophyletic (although Carpolobieae are only weakly supported). However, all four tribes are strongly supported in a three-gene analysis (Forest et al. 2007b), and Xanthophylleae are sister to the other three tribes; relationships between the three were unclear and remain so in Bello et al. (2012). Polygala and Bredemeyera are grossly paraphyletic (Persson 2001; Abbott 2011; Pastore et al. 2017). See Eriksen (1993b) for a morphological phylogeny.

Classification. Because of the polyphyly of Polygala and Bredemeyera generic adjustments are needed (see Pastore 2012; Abbott et al. 2011, 2013; Pastore et al. 2017).

Previous Relationships. The Polygalales of Cronquist (1981) included seven families, the mutual affinities of five of which were described as being "widely accepted". These include Xanthophyllaceae (= Polygalaceae), Vochysiaceae (Myrtales), Trigoniaceae (Malpighiales) and Tremandraceae (= Elaeocarpaceae, Oxalidales). For Emblingiaceae, another group often included in (e.g. Cronquist 1981; Mabberley 1997) or near (e.g. Takhtajan 1997) Polygalaceae in the past, see Brassicales.