II. Sensitivity to Drought at Various Reproductive Stages

62



HARGURDEEP S. SAINI AND MARK E. WESTGATE



co, 1983; Schoper et al,. 1986; Westgate and Boyer, 1986b). A much lower level

of sensitivity at this stage, generally evident only under severe stress, has also been

observed in wheat, barley, and oats (Aspinall et al., 1964; Brocklehurst et al., 1978;

Fischer, 1973; Sandhu and Horton, 1977; Wardlaw, 1971).

In rye, millet, and sorghum, which are also known to be sensitive to drought during the reproductive phase, the precise stages of sensitivity have not been determined (Craufurd et al., 1993; Lewis et al., 1974; Mahalakshmi and Bidinger,

1985a; Mahalakshmi et al., 1987; Salter and Goode, 1967).

Once the grain has been initiated, there is a gradual decline in stress sensitivity

as the grain develops (Aspinall, 1984).



III. NATURE OF INJURY

The nature of injury to the structure and function of reproductive organs, and

the underlying mechanisms involved, depends on the stage of development at

which water stress occurs. This section examines injury during flower initiation,

gametophyte development, pollination and grain initiation, and grain growth.



A. FLOWER INITIATION AND DEVELOPMENT

Appropriate matching of the timing of flowering and the pattern of inflorescence

development to the temporal variation in water availability is recognized as one of

the most important traits conferring adaptation to drought (Bidinger et al., 1987;

Passioura, 1996). However, the effects of drought on flower initiation and early

development are among the least understood aspects of crop reproductive development under water-limited conditions. Apical morphogenesis in cereals is quite

sensitive to water deficit during vegetative and floral development (Husain and Aspinall, 1970; Nicholls and May, 1963; Skazkin and Fontalina, 1951). Water stress

during vegetative development or during flower induction and inflorescence development in cereals slows the rate of inflorescence development, leading to a delay or even complete inhibition of flowering (anthesis) (Angus and Moncur, 1977;

Craufurd et al., 1993; Derouw and Winkel, 1998; Mahalakshmi and Bidinger,

1985a,b; Mahalakshmi et al., 1987; Winkel et al., 1997; Wopereis et al., 1996).

similar effects have also been observed under salinity and may be partially attributable to osmotic stress (Khatun et al., 1995). In Pennisetum and Sorghum, floral

initiation is delayed by water stress (Craufurd and Peacock, 1993; Mahalakshmi

and Bidinger, 1985b; Matthews et al., 1990). Very few studies have been done to

determine the effects of drought on the process of floral induction in cereals per



REPRODUCTION DURING DROUGHT



63



se, which is difficult to separate from the postinduction floral development in most

cereals. However, experiments with Lolium temulentum, a long-day plant with a

simple single-cycle photoperiodic response, and Pharbitis nil, a single-cycle shortday plant, clearly show that flower induction is also inhibited by water deficit. The

mechanism by which water stress inhibits flower induction remains obscure, but

a role for an increased abscisic acid (ABA) level in L. temulentum has been suggested (King and Evans, 1977).

In sweet corn, a female inflorescence bud is first initiated at node 7 and then successively at the lower nodes prior to the initiation of the terminal male inflorescence (Damptey and Aspinall, 1976). Normally, only the bud at node 7 develops

into a mature inflorescence, but if plants are subjected to water stress during terminal inflorescence initiation, the plants produce two to three axillary inflorescences at the lower nodes, whereas the growth of the terminal male inflorescence

is impeded (Damptey and Aspinall, 1976). Because removal of the terminal male

inflorescence prevents the development of additional female inflorescences under

water stress and excision of the bud at node 7 has the opposite effect, Damptey et

al. (1978a) concluded that the promotion of axillary inflorescence formation in

stressed plants is mediated through an effect on the terminal male inflorescence.

Certain parallels between the effects of water deficit and ABA application, including increases in endogenous ABA concentrations of terminal and axillary buds,

suggest that ABA may play a role in this stress response (Damptey et al., 1978b).



B. GAMETOPHYTE DEVELOPMENT

Gametophyte development is impaired when cereals experience water deficit

during meiosis. The most common damage is a loss of pollen fertility. The effect

has been observed in wheat (Saini and Aspinall, 1981; Skazkin, 1961), barley

(Skazkin and Zavadskaya, 1957; Zavadskaya and Skazkin, 1960), oats (Novikov,

1952), rice (Sheoran and Saini, 1996), and maize (Downey, 1969). In all of the

self-pollinated cereals, the increase in pollen sterility leads to a decline in grain set

and yield (Salter and Goode, 1967, and references cited therein). In wheat and rice,

a large proportion of the water stress-affected anthers are small, shriveled, unable

to dehisce, and contain sterile pollen (Saini and Aspinall, 1981; Sheoran and Saini, 1996). A proportion of the pollen in apparently normal-looking anthers of wheat

is also sterile (Saini and Aspinall, 1981). Such pollen grains have dilute cytoplasm

and are devoid of starch (Dorion et al., 1996; Saini and Aspinall, 1981; Saini et

al., 1984; Sheoran and Saini, 1996), which is a conspicuous constituent of fertile

cereal pollen (Franchi et al., 1996).

Details of microsporogenesis during water stress have been determined only in

wheat. Anther development following transitory meiotic-stage stress continues



64



HARGURDEEP S. SAINI AND MARK E. WESTGATE



normally until about first pollen grain mitosis (PGM-1), when microspores lose

contact with the tapetum, dislodge from their normal peripheral location, and fail

to develop further (Saini et al,. 1984). In about half of such anthers the filament

degenerates at the same time. The disoriented pollen grains do not accumulate

starch and have dilute cytoplasm and virtually no intine. However, the exine develops normally. This pattern, observed with an Australian cultivar, differs somewhat from that in a Canadian wheat cultivar, where the first symptoms of developmental disruption were observed at or soon after meiosis (Lalonde et al., 1997a).

The latter included degeneration of some meiocytes, loss of orientation of microspores, and abnormal vacuolization of the tapetal cells.

Comparatively few attempts have been made to determine if female infertility

also contributes to the decline in yield in response to meiotic-stage drought. Reciprocal crosses between stressed and unstressed wheat plants showed that female

fertility was not affected by a water stress treatment that caused complete male

sterility in approximately 40% of the florets (Saini and Aspinall, 1981). The leaf

water potential (␺) of these stressed plants declined to approximately Ϫ2.3 MPa

(control ␺ ϭ Ϫ0.8 MPa), which verges on being a severe stress for wheat. A similar absence of effect on female fertility in wheat was also reported by Bingham

(1966). In oats, female fertility remained unaffected even under severe drought

(soil moisture ϭ 13% of field capacity), unless the stress was prolonged, which

caused a degeneration of the antipodal cells and then the entire embryo sac to a

withered strand lacking functional elements (Skazkin and Lukomskaya, 1962).

Nucellar cells filled the space left by the degenerating embryo sac. In corn, water

stress produced a variety of lesions in the embryo sac, including a complete suppression of development; depending on the severity of stress, 15 to 43% of the embryo sacs were affected (Moss and Downey, 1971). Grain set in these plants was

reduced severely, despite hand pollination with fertile pollen, indicating that the

structurally abnormal embryo sacs were also sterile. Together, these observations

indicate that the development and fertility of the male gametophyte are much more

drought sensitive than those of the female gametophyte. This situation is probably

common to a variety of stresses because temperature extremes, which also reduce

male fertility severely, have no or only minor effect on female fertility (Brooking,

1976; Dupuis and Dumas, 1990; Hayase et al., 1969; Saini and Aspinall, 1982a;

Saini et al., 1983; Satake and Yoshida, 1978). The greater stress tolerance of the

female gametophyte has an adaptive significance because the potential impact of

male sterility on grain set in the field can be partially mitigated by cross-pollination, whereas a reduction in female fertility cannot be overcome.

Meiotic-stage drought also causes other floral abnormalities, which could affect

the number of grains formed. These include the stunting of panicle and the production of immature, small, and discolored “blasted” spikelets in rice (Namuco

and O’Toole, 1986; Sheoran and Saini, 1996). Stress of similar intensity does not

cause the latter effect in wheat (Dorion et al,. 1996; Saini and Aspinall, 1981), but



REPRODUCTION DURING DROUGHT



65



extreme drought can lead to spikelet death (Morgan, 1971; Westgate et al., 1996).

Slight shortening of spike length has also been reported in wheat (Bingham, 1966).



C. POLLINATION, FERTILIZATION, AND GRAIN INITIATION

This stage of reproductive development is particularly sensitive to drought in

rice and corn. In rice, water stress during flowering can reduce the harvest index

by as much as 60%, largely as a result of a reduction in grain set (Garrity and

O’Toole, 1994; Hsiao, 1982; O’Toole and Moya, 1981). Similar yield reduction

also results from desiccation caused by dry wind (Ebata and Ishikawa, 1989). Panicles in stressed plants fail to fully exsert (emerge) from the flag leaf sheath, flowering is delayed, and the percentage of spikelets that open at anthesis is reduced

(Ekanayake et al., 1989; O’Toole and Namuco, 1983). The failure of panicle exsertion alone accounts for approximately 25 to 30% of spikelet sterility because the

unexserted spikelets cannot complete anthesis and shed pollen, even when development is otherwise normal (Cruz and O’Toole, 1984; O’Toole and Namuco,

1983). Rice spikelets lose water easily at this stage, which can result in bleaching

and death of lemma and palea and shriveling of anthers (Ekanayake et al., 1989,

1993; O’Toole et al., 1984; O’Toole and Namuco, 1983). In addition, water stress

reduces the number of anthers that dehisce and lowers the amount of pollen shed

and in vivo pollen germinability (Ekanayake et al., 1990). Presumably, these abnormalities lead to a failure of fertilization. Grain abortion at the early stages following fertilization also accounts for a part of the reduction in grain number

(O’Toole and Namuco, 1983), although the relative contribution of the failure of

fertilization and seed abortion is not known.

The loss of grain number in maize stressed at this stage can be attributed to any

of several causes. The rapid expansion of reproductive structures, particularly the

silk (stigma/style), is required for successful seed set. Water stress just prior to anthesis inhibits ear and silk growth more than tassel growth. This difference causes asynchrony between pollen-shedding and silk emergence, and thus a failure of

pollination (Du Plessis and Dijkhuis, 1967; Herrero and Johnson, 1981; Kisselbach, 1950; Moss and Downey, 1971; Westgate and Boyer, 1985a). The inhibition

of tassel emergence or anther exsertion also prevents pollination (Herrero and

Johnson, 1981). Water stress during anthesis does not affect pollen viability or its

capacity to affect fertilization (Hall et al., 1982; Herrero and Johnson, 1981;

Schoper et al., 1986; Westgate and Boyer, 1986b), but it can cause a decline in silk

receptivity if pollination is delayed (Bassetti and Westgate, 1993). Even when gamete and floral development proceed normally, and pollen is not limiting, grain

number can be reduced by only a few days of dehydration at flowering (Schoper

et al., 1986; Westgate and Boyer, 1985a, 1986b). The failure to set seed is not due

to an inhibition of pollen germination, pollen tube growth, or fertilization, but re-



66



HARGURDEEP S. SAINI AND MARK E. WESTGATE



sults from the failure of the newly formed zygote to survive beyond 2 or 3 days

(Westgate and Boyer, 1986b). Cell division in the proembryo and endosperm is

initiated, but the proembryos fails to develop beyond the globular stage and the

seed coat does not differentiate (Sass, 1977; Westgate and Boyer, 1968b). This

form of zygotic abortion occurs even if droughted plants are rewatered prior to pollination so that pollen germination and fertilization occur at high ␺ (Westgate and

Boyer, 1986b). Evidently, some aspect of zygote development within the pistillate

flowers is disrupted irreversibly by low ␺.



D. KERNEL GROWTH AND MATURATION

The general pattern of kernel development can be divided into three Phases (Fig.

1). Phase I, often referred to as the “lag phase,” is an active period of cell division

and differentiation and is marked by a rapid increase in kernel fresh weight. Enlargement is primarily the result of water influx driven by a rapid accumulation of

solutes (Barlow et al., 1980; Westgate and Boyer, 1986c). In wheat, phase I typically extends from 14 to 20 days after anthesis (Gleadow et al., 1982; Jenner et al.,

1991), when potential kernel size is determined by the number of endosperm cells

and sites for starch synthesis formed therein (Jenner et al., 1991; Jones et al.,

1996). Overlapping and following is phase II, the period of grain filling, which is

marked by a rapid gain in kernel dry weight as a result of the deposition of reserves, predominantly starch (Bewley and Black, 1985). Kernel fresh weight remains relatively stable, as water is displaced by the accumulating reserves within



Figure 1 Typical pattern of cereal kernel development in terms of fresh weight (FW), dry weight

(DW), and water content (WC). For convenience, kernel development is divided into three phases: (I)

cell division, differentiation, and expansion; (II) rapid reserve accumulation; and (III) maturation.

Adopted from Bewley and Black (1994).



REPRODUCTION DURING DROUGHT



67



the cells of the embryo and endosperm. During phase III, dry matter accumulation

ceases and the kernel undergoes maturation drying and approaches a “quiescent

state” (Bewley and Black, 1985). A loss in fresh weight reflects a continued, and

sometimes more rapid, decline in water content.

1. Inhibition of Cell Division and Expansion

Lack of water during phase I (from 5 to 11 days after pollination) exerts its effect primarily on kernel sink potential (Artlip et al., 1995; Mambelli and Setter,

1998). In this case, kernels grow sufficiently to allow some accumulation of starch

and zein, but they abort development prematurely and remain incompletely filled

(NeSmith and Ritchie, 1992). The sink potential of cereal grains is determined during phase I of development. It is a function of both the number of cells and the

number of starch grains initiated in the endosperm, i.e., the number of sites for

starch deposition (Gleadow et al., 1982; Jones, 1994). Although maximal kernel

sink potential is determined genetically, the actual kernel capacity established is a

function of competition for space or assimilate supply and is the growth environment prevailing during the early stages of development ( Jones et al., 1985). The

endosperm cell number is sensitive to environmental conditions during cell division, and hence, the reduction in yield under drought or high temperature during

early kernel development is due mainly to a lesser number of endosperm cells and/

or amyloplasts initiated. Thus drought reduces the capacity of the endosperm to

accumulate starch (Brocklehurst, 1977; Wardlaw, 1971). The decrease in the number of cells and starch granules ultimately affects both the rate and the duration of

dry matter accumulation (Brocklehurst et al., 1978; Jones, 1994; Jones et al., 1985;

Nicolas et al., 1984).

2. Premature Cessation of Grain Filling

Once sink potential has been established and the kernel begins to accumulate

starch and protein reserves (phase II), drought can decrease final kernel size by

limiting the rate and duration of reserve deposition. Tissue desiccation and high

temperatures that accompany drought can both have an impact on the process of

kernel filling. In barley (Aspinall, 1965; Brooks et al., 1982), wheat (Brooks et al.,

1982) and maize (Jurgens et al., 1978; Ouattar et al., 1987a; Westgate, 1994),

drought during seed filling causes physiological maturity to occur earlier, thus

shortening the duration of kernel filling, which reduces the final kernel size. Typically, water deficit has little impact on the rate of kernel growth (Brooks et al.,

1982; Ouattar et al., 1987a; Westgate, 1994), whereas high temperature often increases the kernel growth rate (Egli, 1994; Wardlaw et al., 1980). Both factors

cause premature cessation of kernel filling, which may result from a lack of assimilate supply associated with leaf senescence (de Souza et al., 1997; Jurgens et



68



HARGURDEEP S. SAINI AND MARK E. WESTGATE



al,. 1978), a decreased capacity for assimilation within the embryo or endosperm

(Jenner et al., 1991; Savin and Nicolas, 1996), or premature desiccation of the kernel (Brooks et al., 1982; Egli, 1994; Sofield et al., 1977a; Westgate, 1994).



IV. WATER RELATIONS OF REPRODUCTIVE TISSUES

AND THEIR INFLUENCE ON YIELD

A. STRESS DURING FLOWER INITIATION AND DEVELOPMENT

The turgor of the cereal apex is remarkably resistant to the change in response

to water stress during the reproductive phase prior to grain growth (Morgan,

1980a, 1984). At the initial stages of floral induction and differentiation, the apex

can survive at ␺ as low as Ϫ6 MPa, which are lethal to leaves (Barlow et al.,

1977). The ␺ of apex declines in parallel with that of leaves, but the accumulation of solutes, such as sucrose and amino acids, maintains apex turgor (Munns

et al., 1979). These solutes are evidently imported from vegetative tissues. Water stress during floral initiation reduces spikelet number in the “indeterminate”

type inflorescence of barley (Husain and Aspinall, 1970). Although the extent of

osmoregulation in barley inflorescence was not determined, it probably does occur because it is a common feature of water-stressed expanding tissues, including cereal apex (Meyer and Boyer, 1981; Michelena and Boyer, 1982; Morgan,

1984; Munns et al., 1979). The decrease in spikelet number at low ␺w suggests

that osmoregulation does not give full protection against water stress during floral initiation.



B. MEIOTIC-STAGE STRESS

Even moderate water deficit at this stage can be quite damaging to grain set. In

wheat subjected to 3- to 4-day episodes of water stress between 15 and 5 days prior to ear emergence (a period that includes meiosis and gametophyte development), the threshold xylem ␺ for a reduction in grain was Ϫ1.2 MPa (Fischer,

1973). At lower xylem ␺, the grain set declined linearly with a decline in ␺, reaching zero at ␺ values approaching Ϫ2.4 MPa. When stress was timed precisely to

coincide with pollen meiosis, a decline in leaf relative water content to 67% and

␺ to Ϫ2.3 MPa, compared to the control values of 93% and Ϫ0.8 MPa, respectively, reduced grain set by approximately 35% (Saini and Aspinall, 1981). A more

rapid water stress of a similar magnitude was slightly more damaging (Dorion et

al., 1996). Meiotic stage stress can reduce grain set by 35 to 75% in different cultivars of rice (Namuco and O’Toole, 1986; Sheoran and Saini, 1996), which dis-



REPRODUCTION DURING DROUGHT



69



play wide genotypic differences in susceptibility to water stress during reproductive development (Garrity and O’Toole, 1994).

Wheat and rice plants resist changes in the water status of inflorescence prior to

its emergence. The ␺ of flowers and floral organs of wheat plants stressed during or

close to meiosis either remains unaffected (Saini and Aspinall, 1981) or declines

much less than that of the leaf (Dorion et al., 1996; Morgan and King, 1984; Westgate et al., 1996). Relative water content of the spikelets also remains constant during stress (Morgan, 1980b). Similarly, ␺ of rice panicle changes little diurnally or in

response to water deficit during meiosis, but varies markedly with evaporative demand after panicle emergence (Tsuda and Takami, 1993). The resistance of wheat

and rice inflorescences to water loss during meiosis, which occurs about 7 to 10 days

prior to inflorescence emergence (Saini and Aspinall, 1981; Sheoran and Saini,

1996), may be due partly to limited transpiration within two or more enclosing leaf

sheaths. Moreover, xylem discontinuity between the floral stalk and the pericarp

probably contributes to the apparent hydraulic isolation (Zee and O’Brien, 1970).

Even when the ␺ of spikelets, anthers, or ovaries does decline in response to water stress, the decline is fully matched by a reduction in osmotic potential (␺s), and

hence, the spikelet turgor does not change despite a drop in leaf turgor to zero (Morgan, 1980b; Morgan and King, 1984; Westgate et al., 1996). Even an increase in the

turgor of floral parts during stress has been reported (Westgate et al,. 1996). Thus, a

reduction in grain set in response to meiotic-stage water stress does not correlate with

the water status of the reproductive structures. Further, the grain set declines only after the leaf turgor falls to zero (Morgan, 1980b; Morgan and King, 1984). The implications of this dichotomy between the water status of the vegetative and the reproductive structures for regulating floret fertility is discussed in Section VA.



C. STRESS DURING ANTHESIS

Rice panicles have a very low diffusive resistance to transpirational water loss.

Therefore, they are generally very poor at preventing water loss after they emerge

(Ekanayake et al., 1993; O’Toole et al., 1984; Tsuda and Takami, 1993). Certain

upland adapted cultivars of rice, however, are better able to prevent water loss from

panicle and suffer less sterility during drought (Ekanayake et al., 1993). This correlation suggests that the deleterious effects of water stress during the flowering

of rice could be attributable to the desiccation of the floral parts. However, limiting transpirational water loss by covering the panicles does not prevent sterility

(Garrity et al., 1986). Thus other factors probably act in concert with desiccation

to cause sterility. Insufficient turgidity caused by excessive water loss could also

be responsible for the inhibition of spikelet opening (Ekanayake et al., 1989),

which is driven by the swelling of floral structures on water uptake (Parmar et al.,

1979).



70



HARGURDEEP S. SAINI AND MARK E. WESTGATE



Pollen ␺ in well-watered plants of maize is about Ϫ2.0 MPa, but it does not decline further under water stress (Westgate and Boyer, 1986a). In contrast, female

reproductive organs of maize are prone to water loss, especially at anthesis (Westgate and Thomson Grant, 1989). As soil water is depleted, ␺ declines throughout

the plant, but unlike leaf, root, and stem, silks are unable to lower their ␺s and thus

maintain turgor (Westgate and Boyer, 1985b). The failure of silks to maintain turgor, which is essential for cell enlargement (Cosgrove, 1981, 1993; Lockhart,

1965), could be responsible for the delay in their growth and emergence under water stress (Herrero and Johnson, 1981; Moss and Downey, 1971). Although asynchrony between male and female development can prevent pollination, the decline

in silk ␺ per se does not prevent pollen germination because pollen ␺ always remains lower than that of silks, allowing the pollen to draw water even from quite

dry silks (Westgate and Boyer, 1986a,b). In addition, pollen viability is maintained

at low ␺ during anthesis (Barnabas and Rajki, 1981; Schoper et al., 1986, 1987;

Westgate and Boyer, 1986b). Thus, pollen desiccation is not a factor in limiting

seed set under water stress nor does water stress prevent fertilization, but grain development may be aborted (Westgate and Boyer, 1986b). Moreover, drought-induced zygotic abortion occurs even in droughted plants that are completely rehydrated prior to pollination (Westgate and Boyer, 1968b). The fact that reproductive

failure in maize cannot be attributed to a direct effect of low ␺ at the time of pollination and fertilization implies that the effects of low ␺ on metabolism and cellular differentiation may persist long after the stress has been relieved.



D. STRESS DURING GRAIN FILLING AND MATURATION

Barley, wheat, and maize exhibit little, if any, change in grain ␺ when drought

occurs during rapid grain filling (phase II), whereas other plant structures undergo large decreases in ␺ (Barlow et al., 1980; Brooks et al., 1982; Ouattar et al.,

1987b; Westgate, 1994; Westgate and Thomson Grant, 1989). Various anatomical,

physicochemical, and theoretical models have been proposed to explain this apparent hydraulic isolation of the kernels from other structures of the plant. These

include vascular discontinuities within the caryopsis (Brooks et al., 1982; Zee and

O’Brien, 1970), large hydraulic discontinuities within the grain and stable xylem

␺ within the stem (Ouattar et al., 1987a), osmotic regulation in the apoplast (Westgate and Boyer, 1986c), and the presence of specialized tissues within the vasculature that control osmotic potential of the apoplast (Bradford, 1994). Although

each of these possibilities is supported by some experimental observations, none

has been tested rigorously. Nonetheless, maintenance of a favorable water status

within the kernel presumably permits metabolism to continue despite severe water deficits in the vegetative tissues. If this were true, grain growth during drought

would be limited only by the capacity of the plant to supply assimilates. This ap-



REPRODUCTION DURING DROUGHT



71



pears to be the case only when plants are stressed severely and rapidly ( Jurgens et

al., 1978; Kobata et al., 1992; Nicolas and Turner, 1993; Westgate and Boyer,

1985a).

Although several studies have shown that kernel ␺ remains fairly constant during rapid kernel filling (Ouattar et al., 1987b; Westgate, 1994; Westgate and Thomson Grant, 1989), drought during grain filling causes a decrease in kernel water

volume (Westgate, 1994). Drought also causes a premature decline in kernel ␺ and

␺s late during filling (Brooks et al., 1982; Westgate, 1994). The decrease in kernel

water content (Westgate, 1994) indicates that drought does indeed alter the water

status of the developing kernel. Because endosperm and embryo desiccation may

ultimately limit the metabolism of incoming assimilates (see discussion later), the

hydraulic response of the kernel to late-season drought could have a direct impact

on the duration of kernel filling.

Egli and TeKrony (1997) proposed that the cessation of cell expansion ultimately determines the subsequent timing of seed desiccation and maturation. Cell

expansion during phase I is driven by osmotic water uptake, and the maximum

seed water content establishes the cellular volume that can be filled by storage materials. Deposition of storage reserves during phase II replaces cell water, effectively desiccating the seed (Ray, 1978), which eventually triggers seed maturation.

Therefore, kernel water volume determines final kernel size (dry matter) in two

ways: (1) it sets the maximum volume for storage reserve accumulation and (2) it

establishes the maximum duration for grain filling (Brooks et al., 1982; Nicolas et

al., 1984). Water stress initiates the process of grain desiccation prematurely so

that kernels on water-stressed plants cease to accumulate dry matter sooner after

anthesis (Egli and TeKrony, 1997; Sofield et al., 1977a; Westgate, 1994). This

analysis presumes that there is a minimum moisture content below which dry matter accumulation ceases. This possibility is discussed further in Section VC.



V. PHYSIOLOGICAL AND METABOLIC BASES FOR

REPRODUCTIVE FAILURE UNDER DROUGHT

The information on control processes in the effects of drought on flower initiation and early flower morphogenesis is scant. Therefore, we will focus on the

events including and following meiotic division in the reproductive organs.



A. FAILURE OF POLLEN DEVELOPMENT

Early work aimed at understanding the reasons for the failure of pollen development under water stress found that chromosomal pairing and separation during



72



HARGURDEEP S. SAINI AND MARK E. WESTGATE



meiosis in PMCs was not always normal in rather severely stressed barley plants

(Skazkin and Zavadskaya, 1957). More recently, an increase in meiotic abnormalities was also noticed in water-stressed rice plants (Namuco and O’Toole,

1986). These included an increase in univalents, lagging chromosomes, noncongression of bivalents in metaphase, and micronuclei formation. The abnormalities

started to increase at relatively moderate stress (leaf ␺ Ϫ1.1 to Ϫ1.9 MPa) and

peaked at ␺ around Ϫ2.2 MPa. Under severe stress (leaf ␺ Ϫ3.5 MPa), the entire

meiotic process was arrested. Whether any of these abnormalities contributed to

pollen sterility is not known, although the cessation of meiosis under severe stress

would certainly do so, unless it is reversed on rewatering. In contrast, no such abnormalities were noticed in wheat PMCs (Saini, 1982), which always complete

meiotic division under stress (Lalonde et al., 1997a; Saini et al., 1984). Meiosis is

apparently also completed in the PMCs of moderately stressed rice because microspores are produced, although their subsequent development fails (Sheoran

and Saini, 1996). Microspores of wheat, with a few exceptions (Lalonde et al.,

1997a), continue to develop normally for several days before their development is

arrested (Lalonde et al., 1997a; Saini et al., 1984). These observations indicate that

a more subtle lesion rather than a catastrophic failure of meiosis is the probable

cause of male sterility. At least in wheat, the direct desiccation of the sporogenous

tissue is not responsible for this developmental arrest (Saini and Aspinall, 1981;

Westgate et al., 1996).

Developmental anatomy of stress-affected anthers gives some promising clues

about the metabolic events that may be linked to the failure of pollen development.

Cereal pollen grains are rich in starch (Franchi et al., 1996). Starch accumulates

late during pollen development and is then used to support pollen germination and

pollen tube growth (Clément et al., 1994; Franchi et al., 1996; Miki-Hirosige and

Nakamura, 1983; Pacini and Franchi, 1988). Pollen grains rendered sterile by

drought or other stresses fail to accumulate starch (Ito, 1978; Saini and Aspinall,

1981, 1982a; Sheoran and Saini, 1996). Water stress also changes the pattern of

starch distribution in anthers and inhibits intine development in pollen grains

(Lalonde et al., 1997a; Saini et al., 1984). The timing of the inhibition of starch

deposition coincides with the appearance of structural lesions during anther development (Lalonde et al., 1997a; Saini et al., 1984), suggesting that a disturbance

in carbohydrate availability and/or metabolism may be involved in this developmental failure. This view is also supported by the observation that increased sucrose uptake increases the grain set in wheat spikes cultured in nutrient solution

(Waters et al., 1984). The addition of ABA to the nutrient solution decreases sucrose uptake and grain set, and both effects are reversed by supplementing the

medium with additional sucrose (Waters et al., 1984). By extrapolating these observations to intact plants, one could argue that because water stress inhibits the

rate of photosynthesis and export of assimilate from leaves (Boyer and McPherson, 1975; Hanson and Hitz, 1982), the supply of sucrose to anthers could be the