Revision of Poaceae (Angiosperm Phylogeny) from Wed, 2014-03-19 12:11

Extracted from Angiosperm Phylogeny Website

By P.F. Stevens and E.A. Kellogg

    Poaceae can be recognised by their leaves that have long, usually open
    sheaths and ligules at the sheath-lamina junction, round and often hollow stems,
    and inflorescences whose basic unit is a spikelet. Individual flowers are small,
    lacking an obvious perianth and with a gynoecium that usually has two plumose
    stigmas and a single ovule. In the dry, achenial fruit (often called a
    caryopsis), the rather large embryo is lateral, lying next to the seed

    The large tribe Andropogoneae, with almost 1,100 species, can be
    distinguished from all other grasses because they have paired

Evolution. Stem-group Poaceae are dated to
ca 89 million years before present, the crown group diverge ca 83 million years
before present (Janssen & Bremer 2004: Streptochaeta included, see
also Bremer 2002; dates in Wikström et al. 2001 are far younger).
Bouchenak-Khelladi et al. (2009, 2010a) suggest that grasses originated ca 90
million years ago, although Bouchenak-Khelladi et al. (2010c) estimated that the
basal split in the family was rather younger, (86-)68(-53) million years, i.e.
in the Late Cretaceous (in their Fig. 1 it is ca 72 million years).
Interestingly, Bouchenak-Khelladi et al. (2010c) suggest that the family
originated in Africa; Bremer () had suggested that its origin was in South
America - either way, it seems to be Gondwanan. The family may initially have
been forest dwellers, and now species-poor clades, as well as the stem PACCMAD
clade, may have diverged by the end of the Cretaceous (Bouchenak-Khelladi et al.
2010c - but Puelioideae not included).

However, Poinar (2004) proposed that Programinis
burmitis, fossil in the Early Cretaceous of Myanmar some 100-110 million
years before present, was an early bambusoid grass type. Although it has some
vegetative features that are common in Poaceae, it does not have distinctive
features of the family and so is unlikely to be included here (Caroline
Stömberg, pers. comm.; Smith et al. 2010). The age of grasses (as well as that
of other monocot groups, not to mention the animals, both vertebrates and
insects, associated with them) is also questioned by the discovery of
well-preserved phytoliths of types to be found in the PACCMAD and BEP clades in
coprolites of sauropod dinosaurs from the Late Cretaceous (71-65 million years
before present) of central India (Prasad et al. 2005: there are grass pollen and
macrofossils also known from this age), and this would date the origination of
the clade to some 85-80 million years ago. This record, too, needs confirmation,
although the enigmatic Late Cretaceous mammalian sudamericid gondwanatherians
also had hypsodont teeth and there is a record of a hadrosaurian dinosaur with
carbon isotope ratios that suggests that it might have been eating C4
plants (Prasad et al. 2005; Bocherens et al. 1993). The bottom line is that
dates from these different lines of evidence are apparently irreconcilably in
conflict (Vicentini et al. 2008).

As to more conventional grasses (the [PACCMAD +
BEP] clade), fossil spikelets assignable to them are known from the
Palaeocene-Eocene boundary, about 55 million years before present (Crepet &
Feldman 1991), and this estimate is broadly in line with an estimate of the age
of a genome duplication in Poaceae (70-50 million years before present: Blanc
& Wolfe 2004; Schlueter et al. 2004; Paterson et al. 2004; Kim et al. 2009)
and other estimates like that of Vicentini et al. (2008) and Bouchenak-Khelladi
et al. (2010a) - (60-)52(-44) million years old. Bouchenak-Khelladi et al.
(2009, 2010a, c) suggested that the BEP clade began to diversify about 55-35
million years ago and the PACCMAD clade rather later some 45-37 million years
ago (see Bouchenak-Khelladi et al. 2010a for other dates; Bouchenak-Khelladi et
al. 2010c suggest (55-)52(-50) and (34-)28(-22) million years respectively).
Bouchenak-Khelladi et al. (2009, see also 2010a, c) suggested that Bambusoideae
diversified only some (39-)32(-24) million years ago in the middle

The Poaceae group of families has been described as
being notably speciose (Magallón & Sanderson 2001), but there is
considerable asymmetry in family size within this clade, with most
species belonging to Poaceae themselves. The second most species-rich family
(Restionaceae) has only some 520 species. Poaceae themselves may be seven times
more speciose than its animal-pollinated sister clade (Kay & Sargent 2009).
But again, even within Poaceae there are three species-poor clades that are
successively immediately sister below the PACCMAD and BEP clades (there are two
more such clades successively sister below the family), so calling the whole
family speciose is an overstatement. If there is a focus on diversification,
this may have occurred within the PACCMAD and BEP clades (see also Linder &
Rudall 2005; Smith et al. 2011; and especially Bouchenak-Khelladi et al. 2010c
for diversification). Such statements can be refined, as by Bouchenak-Khelladi
et al. (2010c).

Hodkinson et al. (2008) discuss increases of
diversification rates in Poaceae in the context of a supertree; there seems to
have been one increase when true spikelets developed, and several others
elsewhere in the family (see also Bouchenak-Khelladi et al. 2010c). The
herbaceous habit and annual life cycle appear to be correlated with species
richness (Salamin & Davies 2004; Smith & Donoghue 2008).

Much has been written on the evolution of
C4 photosynthesis in grasses, e.g. see Kellogg (1999), Giussani et
al. (2001: Paniceae), etc. C4 photosynthesis may have evolved up to
eight times in Panicoideae alone (Giussani et al. 2001); it has also evolved
independently in Micrairoideae, Aristidoideae (twice) and Chloridoideae, for a
total of perhaps ca 18 independent origins (Kellogg 2000 and references;
Christin et al. 2008, 2009a, b; Vicentini et al. 2008; Cerros-Tlatilpa &
Columbus 2009 and Christin & Besnard 2009 [both Aristidoideae]; Sage et al.
2011). Details of the mechanisms of C4 photosynthesis and the
morphologies associated with it are very variable in Panicoideae, and
C4 photosynthesis has evolved more than once there (Kellogg 2000 and
references; see also Giussani et al. 2001; Christin et al. 2007a, 2009a).
Furthermore, taxa like Miscanthus x giganteus carry out this kind of
photosynthesis under decidedly cooler conditions than is common (Wang et al.
2008). It has been suggested that the relatively uncommon C4 PCK
subtype (phosphoenolpyruvate carboxykinase) is basal in Chloridoideae,
being subsequently lost and reacquired (Christin et al. 2009b, but see Christin
et al. 2010a for reversals; Ingram et al. 2011b for a reversal that wasn't).
Indeed, the level of parallelisms here may be at the amino acid, similar changes
occurring independently in the phosphoenolpyruvate carboxylase gene in
grasses (Christin et al. 2007a, esp. b, 2009a), in particular, a mutation to
serine at position 780 seems to have occurred in all plants with C4
photosynthesis (Bläsing et al. 2000; see also Brown et al. 2011 for parallelisms
between grasses and Capparidaceae). The adoption of C4 photosynthesis
is associated with well-known anatomical changes, such as closer spacing of the
veins, the development of a sheath of chloroplast-rich cells around the vascular
bundles, etc., and anatomy is also used to characterize subtypes of
photosynthetic pathways, however, the correlation may not be that good (Ingram
2010), and the typology needs to be revisited (E. A. Kellogg, pers. comm.).

In warmer grasslands C4 grasses now
predominate, and all told C4 photosynthesis accounts for about 18-21%
of terrestrial gross primary productivity (Lloyd & Farquhar 1994; Ehleringer
et al. 1997), while it has been suggested that grasslands - both C3
and C4 species are of course involved - currently account for 11-19%
of net primary productivity on land and 10-30% of soil C storage (Hall et al.
2000). Over half the species with the C4 photosynthetic pathway occur
in Poaceae - in total there are a mere 6,000-6,500 species involved (R. F. Sage,
pers. comm.) of which probably somewhat under 4,600 species are grasses (Sage et
al. 1999).

The balance of the evidence suggests that
C4 photosynthesis in grasses appeared first in the Oligocene, some 32
million years ago as global CO2 levels in the atmosphere declined
(e.g. Christin et al. 2008, 2011b; Vicentini et al. 2008; Bouchenak-Khelladi et
al. 2009 - but see below). It is known from grasses from the Early to Middle
Miocene in both the Great Plains and Africa, some 25-12.5 million years before
present, as C4 photosynthesis became energetically advantageous in
some environments (e.g. Ehleringer 1997 and references; Christin et al. 2008,
2011b). The great expansion of grasses with C4 photosynthesis seems
to have occurred considerably later - i.e. in the late Miocene only 9-4 million
years before present - than the initial evolution of this photosynthetic
syndrome. However, why C4 grasses spread in the way they did is still
not well understood. Increasing temperature, open habitats, and perhaps
especially decreasing precipitation (e.g. Edwards & Still 2007; Edwards et
al. 2007; esp. Edwards & Still 2008 - although by no means all C4
grasses are drought tolerant; Edwards 2009), the high flammability of dry
grasses, and windiness are additional factors that would lead to the increased
occurrence of fires, which would have removed trees - and also nitrogen by
volatilizing it - from some habitats and hence favoured grasses, C4
grasses have a reduced requirement for photosynthetic enzymes and so a lower
nitrogen requirement (Wedin 1995). The relatively low nitrogen content in grass
litter means that this builds up, so making grasslands more susceptibe to fire
(Wedin 1995). Grasses also have dense root masses that would make the
establishment of woody vegetation in grassland difficult. Interestingly, both
origins of and reversals from C4 photosynthesis may be clustered
(Vicentini et al. 2008, for reversals from C4, see also Ibrahim et
al. 2009).

Which of the various factors favoured the initial
expansion of grasslands, and what might cause clustering of origins and losses
of different photosynthetic mechanisms, and which favoured the great spread and
expansion to dominance of late Miocene C4 grasslands, is unclear (see
also Tipple & Pagani 2007; Christin et al. 2008, for the early origins of of
C4 photosynthesis and its subsequent development; Jacobs et al. 1989
and especially Retallack 2001 for the paleoecology of Poaceae; Sage & Kubien
2003; Fox & Koch 2004; Osborne & Beerling 2006; Bond 2008; Osborne 2008
for fires, etc.; Osborne & Freckleton 2009: open habitats, then drier
conditions; Arakaki et al. 2011). Recent work suggests that many C4
origins are correlated with a reduction in annual rainfall - indeed, grasslands
transpire less than the woodlands they seem iniitially to have replaced
(Retallack 2001), and this may be connected with declining late Miocene
temperatures, as Arakaki et al. (2011) point out - which perhaps favoured the
expansion of more open habitats in which these grasses thrived (Edwards &
Smith 2010). Taylor et al. (2010) and Ripley et al. (2010) make ecophysiological
comparisons between C3 and C4 grasses, the latter
sometimes being more sensitive to drought and recovering more slowly from it.
But in the late Miocene declining CO2 in the atmosphere may also have
helped things along again (Arakaki et al. 2011).

Cooler temperate grasslands are dominated by the
derived Poöideae, all of which are C3 grasses, and understanding
diversification in Poöideae entails understanding the evolution of cold
tolerance (Edwards 2009; Edwards & Smith 2010). Core Poöideae evolution may
be linked with a period of cooling at the beginning of the Oligocene ca 33-27
million years ago (Strömberg 2005; Sandve et al. 2008; Sandve & Fjellheim
2010). Gene families implicated in low temperature stress response expanded
prior to Poöideae diversification (Sandve & Fjellheim 2010), and proteins
that inhibit ice recrystallization are known from the group (Sandve et al. 2010;
Tremblay et al. 2005). There is a complex relationship between day length,
freezing tolerance and flowering (Dhillon et al. 2010). Furthermore, although
only low levels of fructan - specifically levans - accumulation have been noted
in many Poaceae, notably high levels are found only in Poöideae, although not in
the taxa of the "basal" pectinations (see Hendry 1993 for taxa involved; Pollard
& Cairns 1991). Storing carbohydrates as fructans may enable these Poöideae
to survive drought or frost better, fructans being implicated in stabilizing
cell membranes at low temperatures (Livingston et al. 2009; Sandve &
Fjellheim 2010). Another factor contributing to the diversification of Poöideae
may be the establishment of vernalization (Preston & Kellogg 2008), although
how widely this occurs outside the subfamily is unclear. Finally, the
establishment of the Epichloë/Poöideae relationship may have been
involved in the spread of Poöideae from shady to sunny habitats in the
predominantly cool-season climates that they favor (Kellogg 2001), the mutualism
aiding the plant's defences against herbivores and drought (Schardl et al. 2008;
Schardl 2010).

Grasses now cover about 20% of the land surface,
about half that area being within the tropics (Hall et al. 2000; Sabelli &
Larkins 2009 for references). Open-habitat grasses - these were mostly
C3 grasses initially - diversified taxonomically in North America in
the early Oligocene ca 34 million years ago and became ecologically dominant in
the late Oligocene to early Miocene 7-11 million years later (Strömberg 2005),
although grasslands of the Great Plains may be late Oligocene ca 24 millions
year in age and Argentinian grasslands older (Edwards et al. 2010). Some
C4 grasses may have originated in the Oligocene ca 33 million years
ago, but they became diverse - and made a corresponding major contribution to
overall vegetation biomass - only in the late Miocene-early Pliocene 9-8 million
years ago, the process being complete as recently as a mere 3-2 million years
ago (Bouchenak-Khelladi et al. 2009; Edwards et al. 2010; Strömberg &
McInerney 2011; McInerny et al. 2011 for North America; Arakaki et al. 2011).
The spread of grasslands may be associated with a CO2 decrease
(Arakaki et al. 2011) perhaps connected with the activities of ectomycorrhizal
taxa (Taylor et al. 2009; see Gerhart and Ward 2010 and Zachos et al. 2008 for
past CO2 concentrations) which made trees less competitive (Pagani et
al. 2009); grasslands seem to have expanded at the expense of woodlands
(Retallack 2001), although McInerey et al. (2011) suggest that the late Neogene
expansion of C4 grasses in North America was at the expense of
C3 grasses rather than woody vegetation. What makes grasslands still
more distinctive ecologically is not just the habit of grasses, but that
relatively few species of grasses are ecologically dominant in any one
place, and most of these seem to be C4 grasses (Edwards et al. 2010).

A perhaps ecologically related feature of a number
of grasses scattered in different subfamilies is their accumulation of glycine
betaines and other compounds commonly associated with allowing plants containing
them to grow in saline conditions (Rhodes & Hanson 1993). Some Poaceae have
allelopathic reactions with other plants, Sorghum roots producing a
quinone (an oxygen-substituted aromatic compound) and Festuca roots
meta-tyrosine, a non-protein amino acid (Bertin et al. 2007).
Benzoxazinoids, cyclic hydroxamic acids, are known from members of Poaceae
including both Panicoideae and Poöideae; they confer resistance to fungi,
insects, and even herbicides, as well as being allelopathic (Frey et al. 1997,
2009); they are very uncommon in other angiosperms (Schullehner et al. 2008).
Sindhu et al. (2008) note that the PACCMAD clade are characterized by a gene
that protects the plant against attack by the the ascomycete Cochliobolus
carbonorum. Finally, leaves of pooid monocots (presumably including sedges)
decompose more slowly than do those of other angiosperms (Cornwell et al. 2008).

There are additional eco-physiological factors to
bear in mind. The dumb-bell shaped stomata of grasses show remarably rapid
stomatal movements, very much faster than those few other stomata have been
examined (Franks & Farquhar 2006). However, the significance of this is
unclear given that quite a number of other Poales have similar stomata; whether
the evolution of these stomata is a major component of th ability of grasses to
spread as climates became drier at the end of the Eocene remains to be seen
(Hetherington " Woodward 2003). Prominent rhizosheaths - mucilage from root cap
cells, soil particles, bacteria, etc., all anchored to root hairs - occur in
many Poaceae (McCulley 1995), especially those growing in drier conditions,
although the distribution of such roots is poorly known - they certainly occut
in other Poales, but are rare in broad-leaved angiosperms. Interestingly,
C4 grasses have roots with long root hairs yet may respond positively
in terms of phosphorus uptake when forming endomycorrhizal associations - long
root hairs and endomycorrhizal associations tend to be thought of as alternative
ways of securing phosphorus supply, etc. (Schweiger et al. 1995 and references).
Finally, it may be worth mentioning that Poaceae, apparently alone in flowering
plants (Römheld 1987), acquire iron through chelation of ferric ions with
siderophores which are then taken up by the roots; iron (and zinc) are commonly
limiting trace elements in alkaline soils (Schmidt 2003; Kraemer et al. 2006).
Ectomycorrhizal plants, also noted for dominating the communities in which they
occur, also produce siderophores......

There was a Miocene radiation of grazing mammals
(Thomasson & Voorhies 1990) that may be associated with the spread of such
prairie and savannah grasses (see also Cerling et al. 1997; Bouchenak-Khelladi
et al. 2009, 2010a, the latter with considerable detail and many dates;
Mihlbachler et al. 2011). These mammals evolved hypsodont dentition, i.e. teeth
with high crowns, enamel extending below the gum lines, and short roots to deal
with the wear caused by eating the abrasive grasses (but see below - Sanson
& Heraud 2010); the persistent dead leaves of most grasses may have
decreased their palatability (Antonelli et al. 2010). C4 grasses may
be less palatable than C3 grasses, having more sclerenchyma because
the veins are closer (see Caswell et al. 1973 in part), although the nitrogen
content of C3 and C4 grasses seems to be similar (Taylor
et al. 2010). However, although prairie grasses expanded in Nebraska in the
Early Miocene ca 23 million years before present, hypsodont ungulates were
already around by then (Strömberg 2004); Bovidae and Cervidae started
diversifying at least 26 million years ago (Bouchenak-Khelladi et al. 2009).
Massive diversification of ungulates is largely a Miocene phenomenon
(Bouchenak-Khelladi et al. 2009), and specialists on C4 grasses seem
to have evolved before those grasses dominated (Edwards et al. 2010).
Bouchenak-Khelladi et al. (2009) noted that the density of silica bodies in the
leaf epidermis seems to have increased in a number of grass groups, perhaps as a
defence against herbivory, however, Sanson and Heraud (2010) suggested that the
silica there might not be in crystalline form and so be unable to cause wear on
the enamel of mammalian teeth. In New Zealand, lacking mammalian herbivores,
Danthonioideae in particular may have tended to become deciduous, increasing
their productivity (Antonelli et al. 2010). Establishing connections between the
evolution and rise to dominance of grasses in some ecosystems and the evolution
of grazing animals needs more work, but it seems to me likley that the two are
linked (see also Retallack 2001 for a good summary).

Clavicipitaceous endophytes (class 1 endophytes:
Rodriguez et al. 2009) are widely distributed among grasses. Leuchtmann (1992)
discussed the distribution and host specificity of grass endophytes in general
(Clay 1990 is a still useful general review; see also Schardl 2010). Some 30% or
more of Poöideae are involved in such associations, and both horizontal and in
particular vertical transmission of these fungi occurs. They can be placed in
ascomycetes-Clavicipitaceae-Balansiae (Clay 1986). This endophyte-grass
relationship is usually described as being one of mutualism, although this may
sometimes, at least, not be so (see Saikkonen et al. 1998; Gundel et al. 2006;
Ren & Clay 2009), and the more that is found out about this relationship,
the more complex it appears to be. One of the most important fungi involved is
Epichloë (Clavicipitaceae), a systemic endophyte restricted to Poöideae;
Neotyphodium is its asexual stage, perhaps hybrids of Epichloë
species (Roberts et al. 2005; Moon et al. 2005). For its phylogeny and evolution
see Schardl (1996, 2002, 2010), Craven et al. (2001), Clay and Schardl (2002),
Jackson (2004 [possible codivergence]), and Gentile et al. (2005), and for the
patterns of infection of the two forms, see Rodgers et al. (2009). There are
four groups of alkaloids that are synthesized by Epichloë: indole
diterpenes, lolines, peramine, and the ergot alkaloids (Fleetwood et al. 2007).
Indeed, a variety of "grass" alkaloids, including loliine (pyrrolizidine) and
ergot alkaloids (ergolines), are in fact synthesized by the fungal member of
this association, which could be ca 40 million years old (Schardl et al. 2004).
These loliine alkaloids are primarily active against insects (Schardl et al.
2007; Zhang et al. 2009), and so the presence of endophytes affects the
palatability of grasses to herbivores and of their seeds to granivorous birds,
the animals eating the infected material sometimes not thriving at all; the
level of aphid infestation and that of their parasites and parasitoids,
infestation by nematodes, and even the pattern and rate of decomposition of dead
grass are also affected (e.g. Madej & Clay 1991 - birds; Popay & Rowan
1994 - general; Omacini et al. 2001 - aphids; Lemmons et al. 2005 -
decomposition; Schardl 2010). Furthermore, the larvae of Phorbia (or
Botanophila) flies live on Epichloë stroma, and the adults
transmit the spermatia in a fashion analogous to insect pollination of flowers
(Bultman 1995), indeed, Epichloë synthesizes unique compounds that
specifically attract the flies (Steinebrunner et al. 2008) and which may also be
toxic to other fungi that secondarily invade the fungal stromata (Schiestl et
al. 2006). However, the equilibrium of such relationships can easily be
disturbed (Eaton et al. 2010).

Large numbers of other apparently symptomless
endophyte species (class 3 endophytes) may grow together on Poaceae, but little
is known about their interactions with the host. Márquez et al. (2007) found
that only when the endophytic fungus (Curvularia) was infected with a
virus was Dicanthelium lanuginosum, the host of the fungus, able to grow
in volcanically-heated soils, suggesting the complexity of such relationships,
while Marks and Clay (2007) discuss growth rate of endophyte-infected and -free
plants under various conditions. For fungal records - very numerous and diverse
- on grasses, see Tang et al. (2007); there are at least 1933 species of fungi
from bamboos alone. Some root-associated endophytic fungi (class 4) are also
coprophilic fungi (Herrera et al. 2009), perhaps aiding in their dispersal.

Bacteria may be endophytes too, and several
bacterial endophytes are implicated in fixing 1/3 to 1/5 the nitrogen needed by
sugarcane in Brazil - the bacteria include Gluconacetobacter
(alpha-Proteobacteria) and Herbaspirillum and Burkholderia
(ß-Proteobacteria, for the latter, see also Fabaceae) (de Carvalho et al.

Poaceae provide food for both adults (as pollen)
and larvae (as roots) of Chrysomelidae-Galerucinae-Luperini-Diabrotica
beetles (Jolivet & Hawkeswood 1995). Caterpillars of nymphalid butterflies,
in particular the browns, Satyrini, and the related Morphini, are common (over
10% of the records); the satyrine group as a whole diverged from other
Nymphalidae some 85 million years ago (Wahlberg et al. 2009), Satyrini
themselves diverging from the rest of the group about 65-55 million years ago
(Peña & Wahlberg 2008; Wahlberg et al. 2009 - age depends on calibration
points used, the position of Satyrini within the satyrine clade differs
greatly). Indeed, larvae of Satyrinae-Satyrini, with some 2,200 species, the
bulk of the clade, almost exclusively eat grasses, and the main lineages within
Satyrini diversified well after the initial divergence of Satyrini and only some
36–23 Myr ago - an age perhaps contemporaneous with the spread of grasses (see
above: Peña et al. 2006; Peña & Wahlberg 2008, dates from a tree where
Satyrini diverge ca 55 million years ago and are not sister to all other
Satyrinae). Galling diptera, especially Cecidomyiidae, are quite common here
(Labandeira 2005); Cecidomyiid gall midges, notably Mayatiola (M.
destructor is the Hessian fly), are quite common on Poöideae in North
America (Gagné 1989). Shoot flies (Diptera - Chloropidae) are gall formers on
monocots, especially grasses, but they are also simple herbivores and have other
life styles (de Bruyn 2005). Chinch bugs of the Hemiptera-Lygaeidae-Blissinae
have been most commonly observed on members of the PACCMAD clade, less commonly
on the BEP clade; Teracrini are also concentrated on Poaceae (Slater 1976). In
some grasses, at least, defence against herbivores is mediated by the production
of volatiles which attract nematodes (to attack Diabrotica larvae) or
parasitic wasps (to attack caterpillars: Degenhardt 2009).

Rusts and smuts are common on Poaceae, and those on
Bambusoideae, Poöideae (inc. Stipa and relatives) are particularly
distinctive (Savile 1979b); two thirds of Ustilaginales (smuts) - close to 600
species - are found on Poaceae (Kukkonen & Timonen 1979; Stoll et al. 2003).
Some seventy species of Berberis are alternate hosts (the aecial stage)
for Puccinia graminis, the black stem rust of wheat and other grain crops
in Poöideae - this species (or complex) infects some 77 genera of mostly pooid
grasses (Abbasi et al. 2005 and references). Cyclic hydroxamic acids are widely
distributed in the family and confer resistance against a variety of fungal and
insect pathogens (Frey et al. 1997).

Poaceae are of course predominantly wind-pollinated
with dangling anthers and protandrous flowers. However, insect pollination is
known from some forest-dwelling grasses, especially smaller Bambusoideae
(Soderstrom & Calderón 1971). Streptochaeta may also be animal
pollinated, since it lacks a plumose stigma and the anthers do not dangle; the
flowers are protogynous (Sajo et al. 2008). Woody bamboos are known for their
synchronized flowering (see below). Lodicules, modified members of the inner
tepal whorl, seem to be involved in the opening of the staminate or perfect
flowers; they can be absent from carpellate flowers (see Sajo et al. 2007;
Reinheimer & Kellogg 2009 for references).

The caryopsis is often described as being a
distinctive fruit type of the Poaceae; it is basically a variant of an achene.
In fact, there is quite a variety of fruit types in the family when it comes to
thinking about how dispersal is accomplished (e.g. Werker 1997). Dispersal is
quite often by animals, and although few Poaceae have true fruits as dispersal
units - an example is Alvimia (Bambusoideae) - there are other structures
attracting animals such as elaiosomes (Davidse 1987), as well as hooks and
spikes by which the diaspores attach to passing animals (Centotheca is a
good example). A number of taxa are wind-dispersed, for example,
Andropogon has long hairs on the awns, while Spinifex and a few
other genera are tumbleweeds. Awns can aid in both wind and animal dispersal;
the surface microstructure on awns - minute retrose bristles - may result in the
achene becoming "planted" in the ground (Elbaum et al. 2007; Humphreys et al.
2010b) or moving along the surface of the ground (Kulić et al. 2009).
This is by a ratchet principle similar to that which operates when you put an
entire inflorescence of Hordeum up your sleeve; the whole inflorescence
migrates up your arm and sometimes also down your back. Davidse (1987) notes a
number of taxa with "creeping diaspores" which can move using this mechanism.
Despite the apparent advantages of having an awn, this has been lost several
times in Danthonioideae, at least, perhaps in association with the adoption of
the annual habit where passive burial of seeds suffices (Humphreys et al.

Woody bamboos are known for their tendency to
dominate the vegetation and their synchronized flowering, even when transported
thousands of miles from their native habitat. Flowering may occur only every 120
years or so, and after a rather protracted period of reproduction, the plant
dies. The fruits are used as food by humans and they also attract animals -
birds, rats, etc. - in very large numbers (Janzen 1976; Judziewicz et al. 1999).
This behaviour is also found in some herbaceous bamboos and, depending on
relationships within Bambusoideae, may even be plesiomorphic for the subfamily.
Water often congregates in the hollow stems of bamboos, and a distinctive fauna
lives there (Kitching 2000). Many bamboos are monocarpic - that is, the plants
dies after flowering - and this has profound effects both on the general
vegetation and all organisms dependent on them.

Relationships within Danthonioideae are complex,
and there seems to have been much reticulation in the past (Pirie et al. 2009),
as is notoriously the case in Triticeae (G. Petersen et al. 2006a; Mason-Gamer
2008; Sun & Komatsuda 2010), polyploidy and introgression further
complicating the picture.

There has been a duplication of the whole genome in
a clade that includes at least Zea, Oryza, Hordeum and
Sorghum, i.e. the PACMAD clade, and this duplication has been dated to
70-50/73-56 million years before present (Schlueter et al. 2004; The
International Brachypodium Initiative 2010). Soltis et al. (2009) suggest that
diversification in Poaceae may be connected with this genome duplication. It has
been suggested that diversification of the groups including the cereals occurred
ca 20 million years later (Paterson et al. 2004, but cf. The International
Brachypodium Initiative 2010). Malcomber and Kellogg (2005) suggest that there
has been duplication of LOFSEP genes within Poaceae, while the duplication of
AP1/FUL gene, apparently in stem-group Poaceae, may be involved in
the evolution of the spikelet (Preston & Kellogg 2006). In general,
developmental gene duplication and subsequent functional divergence seem to have
played a very important role in allowing the development of the baroque
diversity of inflorescences in the family (Malcomber et al. 2006; Zanis 2007).
Indeed, there has been very extensive duplication of genes - API,
AG and SEP families - but not in genes of the AP3 lineage
(Zahn et al. 2005a; see also Saski et al. 2007 for other duplications in the

Salse et al. (2008, 2009a, b) and Abrouk et al.
(2010) discuss genome evolution in the family, suggesting that the base
chromosome number (x) is 5, but in the [PACCMAD + BEP] clade at least x
increased to 12 after a genome duplication (to x = 10) and two interchromosomal
translocations and fusions (to x = 12). n = 12 is still found in rice
(Oryza), for example, while x = 10 in Panicoideae. However, Hilu (2004)
suggested that the base chromosome number for the whole family might be x = 11.
Certainly one or more rounds of genome duplication have occurred, with
subsequent independent reductions in chromosome numbers (Schnable et al. 2009;
Abrouk et al. 2010 and references). Genome size varies considerably and at least
in part independently of chromosome number, both increasing and decreasing
(Caetano-Anollés 2005; Smarda et al. 2008; Schnable et al. 2009). Within
Poöideae there seems to have been independent reduction in chromosome number
from n = 12 (The International Brachypodium Initiative 2010). Overall, there has
been very substantial evolution in the genome of grasses, with genome evolution
in Triticeae (Poöideae) being particularly accelerated (Luo et al. 2009; see
also Messing & Bennetzen 2008; Salse et al. 2009a) - many Triticeae have
massive genomes in part because of changes in base chromosome number (Jakob et
al. 2004). Indeed, comparisons of expressed sequence tags and general genomes
suggest that the genomes of Poaceae are much more different from the genome of
Allium (Alliaceae, Asparagales) than is the genome of Arabidopsis
(Brassicaceae, Brassicales, rosid II) from that of Allium (Kuhl et al.
2004). There has also been substantial evolution in the chloroplast genome
(Guisinger et al. 2010 for literature), although details on where on the tree
(and so when) particular changes occurred await more extensive sampling of the
chloroplast genome in Poales and even "basal" Poaceae, and the rate of plastid
evolution may have since slowed down; these changes are placed at the level of
Poaceae as a whole, althougfh they might more correctly go at the PACCMAD/BEP

Scratchpads developed and conceived by (alphabetical): Ed Baker, Katherine Bouton Alice Heaton Dimitris Koureas, Laurence Livermore, Dave Roberts, Simon Rycroft, Ben Scott, Vince Smith