PHYSIOLOGY OF HIGH YIELDING CORN AND SOYBEANS Mark E. Westgate1 I. Introduction Canopy Photosynthesis and Grain Yield a. Corn as a sink limited crop b. Soybeans as a source limited crop II. Managing for Efficient Use of Assimilates a. Cross pollination of maize Reproductive Development – Seed formation b. Asynchrony and abortion c. Assimilate supply and carbohydrate metabolism III. Reproductive Development -- Seed Development a. Rate and duration of seed filling IV. Concluding Remarks References Department of Agronomy, Iowa State University, Ames, IA 50011, EUA. E-mail: westgate@iastate.edu This paper is a contribution from the Iowa State University Experiment Station. Experiment Station Journal Number _____________. I. INTRODUCTION This review will outline the plant factors that contribute to high grain yield in maize (Zea mays L.) and soybean (Glycine max L.) canopies under field conditions. Because grain yield is determined by the number of grains per unit area and the average weight per grain, the discussion will necessarily focus on physiological factors controlling grain formation and development. The extent to which yield formation is limited by the capacity of the maize and soybean canopies to produce photoassimilate and deliver it to the developing flowers and grains will be considered in detail. This discussion will lead to realistic conclusions about how the maize and soybean canopies should be managed for optimum use of available sunlight and grain production. Average maize and soybean yields typically are only about one-third to one-fourth of those obtained in ‘maximum yield’ experiments. In most cases, the primary environmental factor preventing these crops from producing much greater grain yield is lack of soil moisture. Therefore, this review also will consider the impact of water deficits during reproductive development on the success of grain formation and grain growth. The discussion will focus on the importance of maintaining floral development prior to pollination and the physiological factors contributing to increased grain abortion commonly observed during drought. It will also consider how drought alters the rate and duration of grain development leading to the production of smaller grains. Finally, several possible ‘physiological strategies’ to limit abortion and maintain grain growth under water limited conditions will be discussed. II. CANOPY PHOTOSYNTHESIS AND GRAIN YIELD It seems intuitive that grain yield is somehow related to the production of photosynthate during the season. But there is only limited evidence in the literature to provide a ‘physiological connection’ between the primary productivity of the maize and soybean canopies, and their capacity to produce a large number of grains per unit area. In this section, we will review some of the physical factors limiting photosynthate production by a crop canopy, we will examine the documented relationships between seasonal canopy photosynthesis and grain yield in maize and soybean, and we will build upon this information to determine whether 2 managing these crops for increased photosynthetic capacity is a realistic approach to improve grain yields. As the term implies, “photosynthesis” is the physico-chemical process plant canopies use to convert solar energy into useable biochemical energy. Obviously, capturing solar radiation in the ‘photosynthetically active range (PAR: approx. 400-700 nm)’ is a primary consideration. It has been estimated that about 5% of the incident solar energy is captured by plants. The value is generously high, and might apply to a rapidly growing canopy (eg. maize, or sugarcane) at high light intensity. Most crops are much less efficient. Also, the rate of photosynthesis for individual leaves can be saturated for light at a fraction of the light intensity typical of field conditions. Assuming that more photosynthesis is needed to increase grain yield, it is reasonable to consider ways to alter light interception by the canopy to improve the efficiency and use of incoming solar radiation. For convenience, reflection of light by the canopy and soil surface are ignored, and the amount of PAR intercepted by the canopy (IPAR) is approximated by the difference between the incoming PAR above the canopy and the transmitted PAR at the soil surface (TPAR). The overall efficiency of the canopy for light interception is described by a canopy extinction coefficient, k, which relates the amount of light intercepted per unit leaf area, IPAR = PAR exp-kLAI. According to this relationship, the amount of light intercepted – IPAR – should increase exponentially with an increase in the number of leaf layers (leaf area index = LAI), or with an increase the efficiency of each leaf layer, k. Does managing the maize and soybean canopies for increased IPAR lead to an increase in grain yield? Maize -- a sink limited crop Countless row spacing and population studies have sought to identify the optimum management combination for maximizing grain yield. In theory, decreasing spacing between rows and increasing the distance between plants should increase light interception between the rows, and minimize competition between plants for water and nutrients within rows. The combination of 38 cm row spacing and 7.4 plants m-2 provides a nearly equidistant pattern of plants. But a general consensus from these ‘row pop’ studies is that maximum maize yields are achieved at an intermediate row spacing (about 50 cm), and moderately aggressive plant population densities (8-10 plants m-2). A recent study by Flenet et al. (1996) concluded 3 that the efficiency of light interception of maize canopies increases linearly with decreasing row spacing. If light interception is limiting, why doesn’t yield continue to increase with closer rows (eg. 38 cm), which provides an equidistant spacing between plants? A study by (Westgate et al., 1996) may provide the answer. In this example, there was no discernible difference in yield between at 38 cm and 76 cm row spacing. Grain yield increased with plant population density, however, up to about 9 plants m-2. Increasing plant population had a large effect on maximum LAI and light interception (LI) achieved by the canopies – and presumably the total amount of canopy photosynthesis for the season. There was no difference in light interception between the narrow (38 cm) and wide (76 cm) row spacings. Evidently, the greater spacing between plants in the 38-cm rows decreased TPAR between the rows, but increased TPAR between the plants within the rows. When canopy LI is plotted vs. LAI for all combinations of row spacing and plant population, it becomes evident that altering row spacing and plant population density does not necessarily improve the efficiency of light interception of the individual leaf layers, k. Greater plant density improves light interception primarily by adding more leaf layers. Importantly, decreasing row spacing beyond 76 cm had relatively little impact on overall canopy light interception. From a yield formation perspective, there is an optimum amount of light interception for the maize crop in terms of the maximum achieved by the canopy, the rate of canopy closure, and the total IPAR accumulated prior to anthesis. Canopies that can intercept about 95% of incident PAR when maximum leaf area is achieved at flowering achieve maximum grain yield. A canopy that ultimately intercepts more than 95% PAR (possible at high population density), for example, may have a photosynthetic advantage early in the season because of early canopy closure. But this advantage does not translate into increased grain production, presumably because of a low growth rate per plant during flowering (discussed below). Such results imply that maximizing light interception of the canopy in an attempt to increase photosynthate production will not necessarily lead to yield increases. Apparently, the maize canopy produces more photosynthate during the season than can be utilized by grain, even when less than 100% of the incident PAR is captured. This implies that grain yield in 4 maize is limited by the number and activity of ‘sinks’ for photosynthate (grains), rather than the ‘source’ of photosynthate for those sinks. Studies documenting source-sink relations of maize canopies have utilized large ‘canopy photosynthesis chambers’ to monitor photosynthate production during the entire growing season (Christy and Porter, 1982). These chambers enclose a large number of plants without disturbing the canopy structure, and quantify the rate of photosynthesis by CO 2 depletion within the chamber. Photosynthetic rates are measured on clear days at high light intensity, and are corrected for ‘soil respiration’ from bare soil nearby. Thus, the rates provide an estimate of the potential rate of CO2 fixation for the canopy during the season. Typically, canopy rates increase with LAI to a maximum at flowering, and decrease thereafter. By integrating the area under the curve, it is possible to relate ‘seasonal canopy photosynthesis’ with grain yield for a range of crop management options and weather conditions. When similar treatments are compared across years, so that environment was the main treatment variable, there was a poor correlation between yield and seasonal canopy photosynthesis (Christy and Williamson, 1985). There was wide variation in available assimilate during the four years of this study that showed little correspondence to yield levels. Likewise, variation in grain yield produced across a wide range of plant population densities was not related seasonal canopy photosynthesis. But yield varied directly with the capacity of the canopy to convert available photosynthate into grain, a term Christy and Williamson coined Photosynthetic Conversion Efficiency (PCE = grain yield/seasonal photosynthesis). Apparently, the PCE of the maize canopies in these experiments peaked at a plant density of about 10 plant m-2, which also was the most efficient canopy for light interception in other studies (Westgate et al. 1996). Explaining variation in grain yield through differences in PCE is somewhat biased because grain yield is included in the calculation of PCE. Nonetheless, this approach underscores the conclusion that high grain yield in maize is not solely the result of high levels of seasonal photosynthate production. Physiological factors controlling the efficiency of converting available photosynthate into grain will be considered later in this review. Although grain yield in maize generally is not ‘source limited’, current assimilate supply is critical during flowering and early kernel development. Shade treatments applied during pollination have a large negative impact 5 on kernel numbers, even under well-watered conditions (Christy et al., 1986). But the impact of decreasing PAR by 50% during this period on final grain yield varies with prevailing environmental conditions. In fours years of study, the variation in grain yield in the unshaded control plots was greater than the loss in yield caused by shading during pollination or grain filling. The highest yield was achieved in a relatively cool year with high seasonal canopy photosynthesis. This was also the year that showed the greatest effect of shading during reproductive development. Lowest yields were obtained in hot years with low to medium seasonal canopy photosynthesis. These results suggested that plant stress associated with higher than average temperatures affected the capacity of the maize canopies to utilize the available photosynthate, and decreased yield potential below that expected from seasonal assimilate production. This possibility was taken into account by adjusting the seasonal canopy photosynthesis by a heat stress factor to create a PHS-Stress Index. Remarkably, the PHS-Stress Index accounted for nearly all the variation in grain yield across the four yield environments in this study. Most of the variation in grain yield from year-to-year is due to differences in grain number per unit area. Grain size is relatively more stable, but can contribute to lower yields when stress occurs late in the season (see below). It is well established that the number of grains that develop on a maize plant is determined by the amount of PAR intercepted per plant growth during a two-to-three week period around silking. Recent studies confirm that the relationship between kernel number (KN) and IPAR is curvilinear for each ear. A minimum IPAR plant–1 of about 0.5 MJ plant-1 d-1 is required to set any kernels. And about 1.5 MJ plant-1 d-1 is required to set kernels on a second ear. These results help explain why barrenness increases at high plant densities, and why second (or third) ears set grains at low plant densities. In both cases, IPAR plant -1 determines the potential plant growth rate, which is a critical factor for establishing the number of pollinated flowers that continue to develop into grains. In summary, the results from a number of field studies lead to the conclusion that grain yield in maize is ‘sink’ rather than ‘source’ limited. Therefore, establishing growth conditions that provide high rates of canopy photosynthesis, while necessary, are not sufficient to ensure high grain yield. Management strategies that increase seasonal canopy photosynthesis will only be beneficial if they do not interfere with the establishment of reproductive sinks. The challenge for achieving high grain 6 yield is to e determine optimum plant population density needed for 95% LI yet ensure a rapid growth rate per plant during silking, which is critical high grain set. Soybeans -- a source limited crop As was the case for maize, canopy photosynthesis rates for the soybean canopy increase with LAI < 4, reach a maximum coincident with flowering and pod set, then decrease during grain filling. In contrast to maize, however, maximum rates of canopy photosynthesis for soybean are considerably lower, even at comparable LAI. This raises the possibility that seasonal canopy fixation might limit grain production. Row spacing and plant population studies confirm that managing the crop for earlier canopy closure can increase grain yield. Research from many year/locations in Iowa, for example, indicate that planting soybeans in narrow (<76 cm) rows and at high population densities (>350k plant ha-1) provide maximum yield performance in most environments. Measurements of light interception by soybean canopies planted at various row spacings confirm that the narrowrow canopies are indeed more efficient at light interception. But is the difference between light extinction coefficients for 76 cm rows (k = ~0.43) and 38 cm rows (k = ~0.51) sufficient to explain the yield advantage? In reality, the advantage in terms of increasing IPAR is small, on the order of 3 to 5 %. As with maize, the major advantage for increasing light interception occurs in response to plant population and increased leaf area. In fact, a less efficient canopy (with a lower k) is more likely to achieve higher growth rates because it has a greater LAI at a given level of light interception. Planting date studies also show that extending the length of the growing season also can increase grain yields. Soybean varieties adapted for Iowa, for example, typically achieve maximum grain yield when planted in late April or early May. Yields decrease dramatically with later plantings. The earlier planting extends the duration of both the vegetative (VE-R1) and reproductive (R1-R8) stages of development. Together, these crop management studies suggest that increasing the integral of seasonal canopy photosynthesis is an effective means to increase grain yield in soybeans. Studies relating seasonal canopy photosynthesis and grain yield confirm that this is indeed the case (Christy and Williamson, 1985). In contrast to maize, there is a very close relationship between seasonal 7 canopy photosynthesis and grain yield in soybean — a very clear indication that yield of soybeans is ‘source limited.’ When viewed from the perspective of canopy photosynthetic rates, the advantage of planting at populations and row spacings that promote early canopy closure is obvious. Canopies that enter the flowering and pod setting stage with high rates of photosynthesis produce the greatest yields. The variation in yield is associated almost entirely with the number of grains m-2. In the example above, altering plant density caused yield variation. But the dependence of yield on grain number holds true for a wide range of growth conditions. In the Midwest US, soybeans typically are exposed to a number of stresses at the same time, such as herbicides, nematodes, and water deficits. Experiments are currently under way to test how soybean yield is affected by multiple environmental stresses. Preliminary results indicate that, regardless of the combination of stresses imposed, variation in grain yield reflects the capacity of the canopy to form seeds. In this experiment, the highest yielding genotype/treatment combinations were those that had the greatest biomass and canopy growth rate during flowering (R1-R3). Treatments that decreased growth rate caused a corresponding and predictable decrease in seed number, and yield. It is clear from the results of this study, and those of Egli and Zhen-wen (1991), that seeds m-2 (and therefore, potential grain yield) depends directly on canopy growth rate during flowering and pod set. Therefore, it is essential to manage the soybean crop to achieve its maximum growth rate during this critical period. Shading studies confirm this conclusion. Christy et al. (1986) decreased incident PAR by 50% using shade cloth during vegetative, flowering-pod set, and pod filling to determine the dependence of seed number and seed size on current assimilate supply. Decreasing incident PAR by 50% decreased canopy photosynthetic rates about 35% on average, indicating that the soybean canopy was saturated for light during most of the season. As expected for this source-limited crop, decreasing canopy photosynthesis continuously resulted in a corresponding decrease in grain yield, caused entirely by a loss of seed numbers. Shading during flower/pod set and seed filling caused a similar decrease in seed number, but yield loss was greater in the pod-fill treatment. This result indicates three things about the ‘indeterminate’ soybean canopy. First, seed number m-2 is the yield component most sensitive to the decrease in current assimilate supply. Second, final seed number is not determined until well 8 after seed filling has begun. And third, once seed number is established, some compensation in seed size is possible, if assimilate is available. Yield losses occur only when compensation is no longer possible. It is interesting to note that removing the shade after the vegetative period had a positive impact on seed number. This likely reflected an increase in growth rate of the canopy in response to the 50% increase in incident PAR as the plants reached the flowering stage. In summary, grain yield in soybean is ‘source limited.’ Therefore, management strategies that increase seasonal canopy photosynthesis have the potential to increase grain yield. As was the case in maize, seeds m-2 is the primary determinant of yield; and this yield component is closely coupled to the rate of crop growth during flowering and pod set. Maximizing crop growth rate during this period is essential for maximum grain yield. Optimum row spacing and plant population can improve the light harvesting efficiency of the canopy. But providing optimum conditions for plant growth early in the season likely will have a greater impact on seasonal canopy photosynthesis – and therefore, on maximizing grain yield. III. MANAGING FOR EFFICIENT USE OF ASSIMILATES POTENTIAL YIELD ADVANTAGE FROM OUT-CROSSING MAIZE HYBRIDS To this point, managing the maize and soybean crops for high yield has focused on maximizing light interception and crop growth rate during the period seed numbers are being determined. This approach seeks to increase grain set by providing the maximum level of photosynthate possible to the recently fertilized ovaries. The right combination of row spacing, population, date of planting and genotype might achieve the goal of increasing assimilate production, but which combination ensures that assimilates produced by the crop are used most efficiently for seed formation and development? To address this question, it is essential to examine the physiological factors that control seed formation and seed growth. Because soybean flowers are predominately self-pollinated, maximizing crop growth rate at flowering may be the best way to maximize grain yield on a field scale. The pistillate flowers of maize, however, are largely cross-pollinated – i.e. with pollen produced by other plants. Maize 9 breeders have long taken advantage of cross-pollination of genetically dissimilar inbreds to produce highly productive hybrids – a phenomenon known as heterosis. Cross-pollination occurs naturally in the field of maize plants as well, but in a typical monoculture, all the plants are genetically related. All the seed are the product of sib- or self-pollination. So, instead of realizing a yield gain from heterosis, there is actually a slight yield penalty associated with the first generation of inbreeding depression. It may be possible to eliminate this inbreeding depression (and thereby realize a yield gain) by adopting a planting strategy that ensures a high degree of crosspollination between hybrids in the same field. Several years of large-scale field trials have shown that planting hybrids from different companies together in an alternate row pattern provided a 4 bu/a yield advantage over the same hybrids grown in monoculture in adjacent plots. The yield advantage resulted from greater seed number and increased seed size. There was no yield advantage to mixing two hybrids from the same company. Obviously, careful hybrid selection is the key to this strategy. Hybrids must flower (shed pollen and exsert silks) at the same time. And they also must not have a common parent. The first requirement is relatively easy to meet. A comparison of the flowering characteristics of 75 locally adapted hybrids from Western Minnesota showed it quite possible to identify 10 to 12 hybrids that flowered within 25 GDU (1 day) of one another. The second requirement is more difficult because seed companies do not share pedigree information readily. But selecting hybrids from different companies, though not foolproof, is a reasonable starting point, as our field trials have clearly shown. In summary, this field study shows that out-crossing maize hybrids can provide a ‘free’ yield advantage beyond that expected from maximizing crop growth rate during flowering. The yield advantage results from an increase in sink demand (grain number and size), which is consistent with the conclusion that maize is a ‘sink limited’ crop. IV. REPRODUCTIVE DEVELOPMENT – SEED FORMATION Under most growing conditions, grain yield in maize and soybean is only a fraction of the maximum yields recorded for these crops. The formation and development of the reproductive sinks, which are the economically valuable parts of the crop, are highly vulnerable to stresses, particularly water stress. Lack of soil moisture during flowering and early seed formation decrease the number of grains that develop. And moisture 10 stress during seed filling results in smaller grains. In both soybean and maize, the period of seed formation is the most sensitive to drought in terms of yield losses. The impact of drought decreases dramatically as the grain approaches physiological maturity. Early reproductive development is particularly vulnerable to drought because much of the development involves expansion growth, which is inhibited severely by lack of water. In maize, the inhibition of growth leads to two important problems – asynchrony in male and female flower development, and abortion of newly formed zygotes with the ovaries. The next sections describe the yield losses that occur in maize due to these developmental problems, and discusses the central role of assimilate supply in improving grain set under drought conditions. Asynchrony in flower development and kernel abortion in maize Evidence from long-term breeding trials at CIMMYT in Mexico clearly underscore the importance of maintaining synchrony between male and female flower development to improve yield performance of maize under severe drought conditions (Edmeades et al., 2000). Recurrent selection for improved grain yield under severe drought produced genotypes with a shorter anthesis-silking interval (ASI) and more rapid ear growth rate during drought. There was no yield advantage under irrigated conditions, but selection increased the proportion on plants in the population that produced a seed-bearing ear. Presumably, the shorter ASI increased the number of successful pollinations of exposed silks. Breeders routinely select for close synchrony between pollen shed and silk emergence in the hopes that all the silks will emerge when there is sufficient pollen being shed to ensure their pollination. As drought inhibits ear and silk growth, an increasing number of flowers fail to become pollinated. Is the solution to this dilemma of asynchrony an increase in the amount or duration of pollen shed, or a change in the pattern of silk emergence? The distinction between these two possibilities is important because they lead to very different selection strategies. To answer this question, Bassetti and Westgate (1994) established a number of asynchronous plots within a large field, which provided a single source for pollen shed. Asynchrony included silk emergence prior to pollen 11 shed (protogyny) as well as silk emergence after pollen shed (protandry). As expected, maximum kernel set occurred in plots whose silks began to emerge within one or two days of anthesis (50% of plants shedding pollen. Kernel set decreased dramatically in plots whose silk emergence started after the peak rate of pollen shed. By examining the pattern of kernel set on ears collected from the various asynchrony plots, they could determine whether silk emergence or pollen shed limited kernel set. There ware perfect kernel set only on ears whose silks began to emerge within one day of anthesis. Ears collected from plants whose silks began to emerge later (3, 5, 7, 9, 11 days) showed progressively greater loss of kernels at tip positions, ultimately progressing to the base of the ear. But ears with 3, 5 or 7 days of anthesis all had perfect kernel set in the lower portion of the ear (positions 5-15). This result indicated that the intensity of pollen shed was sufficient up 7 days after anthesis to ensure pollination of all exposed silks. At this point, the rate of pollen shed was about 125 grains cm-2 d-1. The loss of kernels at apical ear positions must have resulted from lack of silk emergence. With asynchrony beyond 7 days, loss of kernels resulted from lack of silk emergence and lack of sufficient pollen. The results of this study indicated that late emerging silks were at risk of not being pollinated even under well-watered conditions. Any delay in ear development or silk growth due to drought would only worsen this situation. Bassetti and Westgate (1994) also observed that plots whose silks began to emerge in advance of pollen shed by as many as six days also had a high level of kernel set and grain yield. The success of these protogynous plots suggests that one possible strategy to overcome the negative effects of drought on the synchrony between silk emergence and pollen shed is to select for genotypes that naturally initiate silk emergence prior to pollen shed. The immediate advantage of this shift in development in favor of earlier silk emergence is that any stress-induced delay in silk emergence would actually improve the synchrony in flowering. A subtler, but equally important, benefit of selecting for protogynous plants is the increased priority for ear growth prior to anthesis. These results also highlighted the need to define the flowering process in maize on a more rigorous quantitative basis than that currently being used to define dates of silking and anthesis. Silk emergence on an individual ear is a progress process, which may take up to 10 days to complete. Likewise, the process of pollen shed from an individual tassel progresses in intensity over the course of several days. The process of 12 pollination itself depends not on the percentage of the population shedding pollen or silking, but on the actual density of pollen grains being shed, and the number of silks that are exposed to that pollen each day. By exposing receptive silks on different plants to a wide range of pollen densities occurring naturally in the field, Bassetti and Westgate (1994) discovered that >95% of all exposed silks were successfully pollinated when the density of pollen shed reaching the silks was >125 grains cm-2 d-1. At lower densities of pollen shed, the percent kernel set decreased rapidly. This minimum density of pollen shed was nearly identical to that observed in the asynchrony experiments described earlier. With these basic quantitative measures of flowering dynamics and a curve relating pollen density to kernel set, it is now possible to predict in a quantitative manner the potential number of kernels set each day in the field, and the potential number of kernels ha-1 set by the end of pollen shed. In the example provided, which is taken from actual field data, the greatest intensity of kernel set occurs 2 to 4 days after anthesis. Afterwards, kernel set decreased dramatically because silk emerge too late to become pollination. This approach, of course, provides potential kernel numbers because stress-induced abortions have not been taken into account. Even if pollination is successful, fewer kernels may develop under drought conditions because a higher percentage of zygotes abort a few days after fertilization (Westgate and Boyer, 1986). For plant growing in the field, a short-term water deficit during pollination can decrease kernel number per ear by 50% (Schussler and Westgate, 1994); the same level of water stress imposed on the same genotype grown in a growth chamber can cause a complete failure of kernel set (i.e. 100% abortion) (Schussler and Westgate, 1991). Results from a large number of related studies aimed at understanding the physiological basis for this increase in kernel abortion have lead to the conclusion that abortion is directly related to the inhibition of ovary growth. Regardless of genotype, growth conditions, or stress level, inhibition of ovary growth lead to a decrease in kernel number per ear. In all cases, plants were hand pollinated with abundant pollen. Measurements of ovary weight in maize populations selected for short ASI during drought support this conclusion. Plants with short ASI under drought conditions have larger ovaries at the time of pollination. The practical outcome of these studies is that maintenance of rapid ovary growth is essential for high kernel set when stress occurs during pollination. 13 Assimilate supply and carbohydrate metabolism The decrease in kernel numbers in response to shading, decreased kernel numbers per plant at low IPAR, and the severe inhibition of photosynthesis at low leaf water potential, all imply that the inhibition of ovary growth during drought is somehow linked to the current supply of photosynthate. The model relating ovary growth to photosynthate supply proposed by Zinselmeier et al. (1999) suggests that rapid ovary growth occurs when there is an ample supply of photosynthate from current photosynthesis and from reserves. Under mild stress conditions, photosynthesis is inhibited and ovaries depend on mobilized reserves to continue growing. As stress becomes more severe, lack of current photosynthate and depletion of reserves cause the ovaries to abort. If this model is accurate, it should be possible to maintain the supply of photosynthate to the ovaries under stressful environments. First, and most obvious, is maintaining the production of photosynthate at low leaf water potential. To the author’s knowledge, there are no published reports indicating this is possible. Small improvements associated with osmotic adjustment have not been sufficient to sustain reproductive development. A second possibility is to increase the contribution of assimilates remobilized from temporary storage tissues, such as the stem internode, and rachis. Third, sustaining metabolic activity within the water stressed ovaries could increase partitioning of available assimilate to the ear. To be successful with the latter two strategies, a number of ‘physiological barriers’ need to be overcome. First is the fact that the developing ovaries are competing with a large number of well-established sinks. These sinks have priority for assimilates from both current and reserve sources. Second is the fact that the temporary storage tissues (stem internodes, rachis) are still active sinks during pollination. They contain very high concentrations of sugars, but the sugar is primarily glucose, which cannot be remobilized. The third barrier to overcome is the inhibition of carbohydrate metabolism within the ovary itself. This inhibition of metabolism makes this weak sink an even poorer competitor for assimilates. To determine which of these ‘barriers’ was most important to overcome, it was necessary to resolve whether the failure of kernels to develop during drought was due entirely to the lack of current assimilate supply. This was done by comparing the kernel set in plants whose 14 photosynthesis rates were inhibited to the same extent and duration by low leaf water potential or by low light intensity. Presumably, both treatments would cause the same reduction in kernel set if current assimilate supply controlled ovary abortion. The results of this experiment indicated that lack of current photosynthate accounted for about 70% of the kernel loss observed in droughted plants. The additional loss of kernels was due to another, as yet unknown, factor (Schussler and Westgate, 1991). Sucrose feeding experiments, in which supplemental sucrose was supplied by stem infusion, lead to the same basic conclusion. Adding enough sucrose to account for the sugar that would have been produced had photosynthesis been able to continue at low leaf water potential, recovered about 70% of the kernels lost in droughted plants without stem infusion. Related experiments proved that sucrose was the active agent causing the recovery of kernel numbers (Zinselmeier et al. 1995, 1999). These experiments clearly demonstrated that maintaining the current supply of photosynthate was critical for improving kernel set during drought. But it is not realistic to infuse sucrose into plants in the field! A possible alternative is to increase the contribution of reserve sugars when photosynthesis is inhibited. To test this possibility, three treatments were established in the field designed to vary the level of reserve sugars in the plant prior to imposing a water deficit during pollination. A 50% shade treatment decreased sugar levels (and concentration) relative to the control plants in full sun. And a widely spaced treatment increased sugar levels, presumable to the genetic potential for the hybrids in the study. At pollination, a water stress of similar intensity and duration was imposed on all three treatments and plants were hand pollinated with abundant pollen. The inhibition of photosynthesis was the same in all three treatments as well. The control plants at a normal plant density in full sun set about 50% fewer kernels when water stressed during pollination. Shaded plants with a lower level of reserves prior to pollination set even fewer kernels. But the increased accumulation of reserve sugars in the spaced plants did not render them less sensitive to the water stress. Thus, is was possible to make plants more sensitive by shading, but it was not possible to decrease their sensitivity by increasing the level of reserves (Schussler and Westgate, 1994). Examination of the type of sugars accumulating in the stem internodes and rachis (cob) at the time of pollination, revealed why altering sugar levels along would not be sufficient to increase assimilate availability 15 to the developing ovaries. At pollination, well over 30% of the weight of the cob is in the form of sugars. But only about 2% of that sugar is in a form than can be remobilized, i.e. sucrose. Likewise, the sugars in the ear stalk (shank) are predominantly reducing sugars (glucose + fructose); sugars in the stem internode below the ear node are about equally distributed between reducing sugars and sucrose. Thus, none of these structures is an ideal source of sucrose for remobilization. In fact, just the opposite occurs when water is withheld; sucrose accumulates rapidly in the cob and shank, and sucrose levels remain fairly stable in the stem internodes. Evidently, these structures continue to function physiologically as sinks for assimilate during pollination. High levels of acid invertase (AI), which is typical of growing tissues, in the cob at this time support the accumulation of glucose. And the inhibition of AI in droughted plants is consistent with the accumulation of sucrose. It will be exceedingly difficult to alter the pattern of sucrose accumulation in the stem internodes, cob and shank tissues given our current state of knowledge about mechanisms of sugar metabolism and storage in these tissues. The mechanism(s) of sucrose import and export are not known. The biochemistry of carbohydrate metabolism is not fully characterized. Tissue- and temporally-specific promoters are lacking for possible transgenic approaches. And, most importantly, the pattern of sucrose accumulation is coupled to tissue development. It likely will be necessary to alter development of these structures in order to accelerate or enhance their capacity for exporting sucrose during pollination. The fact that not all of the kernel loss in water stressed plants could be accounted for by lack of assimilate supply implied that other factors also contributed to the increase in kernel abortion. The failure of kernels to develop when pollinations occurred at low ovary water potential suggested that kernel set might depend on conditions within the pistillate flowers themselves. Preliminary measurements on isolated ovaries showed that water decreased the capacity of ovaries to utilize sucrose in vitro (Schussler and Westgate, 1991). A developmental profile of the carbohydrate status of ovaries samples from well-watered and water-stressed plants revealed that a brief water deficit at silk emergence dramatically altered the reducing sugar, sucrose, and starch levels within the ovaries. Most notable was a large increase in the amount and concentration of sucrose relative to the control plants. The rapid decrease in sucrose content at silk emergence in the control plants may serve to maintain sucrose unloading from the phloem 16 into the pedicel region of the ovary. If so, the high concentrations of sucrose that develop in water stressed ovaries may limit the capacity for continued phloem unloading, thereby decreasing ovary ‘sink strength.’ This possibility is supported by estimates of sugar concentrations in the ovary apoplasm (Westgate, unpublished), which clearly indicated a large decreased in reducing sugar/sucrose ratios in the apoplasm of waterstressed ovaries. This result suggested that the capacity for sucrose hydrolysis had been impaired. Measurements of ovary acid invertase activity indicated that, indeed, extractable acid invertase activity within the ovaries was dramatically inhibited at low ovary water potential. The physiological basis for the inhibition of invertase activity is not fully understood, but evidence from Anderson et al. (2000) shows that transcription of the soluble form of acid invertase (Ivr2) decreases during water stress. Parallel analyses of the insoluble form of acid invertase (Ivcw) have not been reported. But the activities of both the soluble and insoluble forms of invertase are inhibited to the same extent by water deficits (Zinselmeier et al., 1995). Importantly, the level of (insoluble) invertase activity is directly related to the rate of ovary growth. The recovery of kernel numbers in response to stem infusion of sucrose corresponds to a recovery of ovary acid invertase activity (Zinselmeier et al., 1999). These results provide strong support for the central role of carbohydrate metabolism in establishing the young kernels as a sink for photosynthate. Water stress caused a lesion in carbohydrate metabolism within the ovaries, which must be overcome to increase kernel set during drought. Could this ‘lesion’ in metabolism also explain the more rapid loss of starch within water-deficient ovaries? Typically during silk emergence and pollination, ovary starch levels are increasing slowly (actually declining on a concentration basis). But the level and concentration of starch decrease dramatically when water is withheld, even though sucrose concentrations are well above control levels. Evidently the sucrose that accumulated, whether from starch hydrolysis or phloem unloading, is not being metabolized by the ovaries. Sucrose infusion into the stems of water stressed plants reverses the drought-induced loss of starch. And as was the case for invertase activity, the recovery of ovary starch level is closely correlated with kernel set (Zinselmeier et al. 1999). Water stress had little impact on activities of enzymes in the pathway from glucose to starch, such as ADP-glucose pyrophosphorylase or starch synthases. The most sensitive enzymes were the invertases. These results indicate that loss of invertase activity apparently is the key ‘metabolic’ lesion caused by a water 17 deficit. It explains the rapid accumulation of sucrose and the loss of starch in the ovaries. And the subsequent decrease in reproductive sink strength when drought occurs during pollination. In summary, grain yield in maize remains highly vulnerable to water deficits during pollination in large part because ovary carbohydrate metabolism is inhibited. This inhibitions disrupts the flux of assimilates to the ovaries, and slows ovary growth. The decrease in ovary growth translates directly to an increased probability of reproductive failure – i.e. kernel abortion. Current and future strategies to maintain zygote (kernel) development during drought include developing a greater understanding of the molecular controls of sucrose and starch accumulation in storage tissues prior to anthesis; determining the molecular basis for the inhibition of ovary invertase activity at low ovary water potential, and identifying genes whose activity is related to the inhibition of ovary growth under drought conditions. V. REPRODUCTIVE DEVELOPMENT - SEED DEVELOPMENT Once the number of seeds has been established, the only reproductive adjustment possible when canopy photosynthesis is inhibited by adverse environmental conditions is to limit the extent of seed development. Fortunately, the sensitivity of grain yield to drought decreases as reproductive development progresses towards physiological maturity in both maize and soybeans. Drought during seed filling generally results in the production of smaller seeds. Possible physiological explanations for producing smaller seeds include loss of current assimilate supply due to accelerated leaf senescence, decreased sink capacity possible resulting from fewer cells or less cell volume, or inhibition of storage product synthesis due to premature desiccation. The remainder of this review will consider the evidence supporting these possibilities for the maize and soybean crops. 18 Rate and duration of seed filling Under well watered and fertilize conditions, maize kernels typically achieve their maximum volume early in development. Kernel moisture remains fairly stable during linear grain filling, but begins to decrease while kernels are rapidly filling. Maximum dry weight is achieved during a period of ‘terminal’ desiccation. In a greenhouse study, Westgate and Grant (1989) showed that a brief water deficit during linear filling had little impact on dry matter accumulation of maize kernels, but did cause a substantial decrease in kernel water content. The change in water content of the kernels was not reflected in kernel water potential, which did not change despite a large decrease in leaf water potential. The remarkable stability of kernel water potential in the water-deficient plants implies that developing grains are hydraulically isolated from the water status of the plant. The actual mechanism that makes this possible is not known. In maize and soybean, there are no direct vascular connections between the maternal tissue and the embryos. And special ‘apopolastic barriers’ have been proposed (Bradford, 1994), but direct evidence for such a barrier is lacking. Regardless of the mechanism, the maintenance of stable water potentials within the developing seed should allow metabolism to continue even under severe drought conditions. These results of this short-term experiment indicated two things about grain development during drought. First, the smaller grains that typically develop under drought conditions probably result from a shorter grain filling period, rather than a decrease in grain fill rate. Second, drought did indeed alter the water status of the seed, by decreasing its water content, even though the water loss was not evident from measurements of grain water potential. This suggested that there might be connection between grain moisture content and the duration of grain filling. To test this possibility, a long-term water deficit was imposed on plant grown in the field. Irrigation water was withheld after seed number was established, so that any yield adjustments must result from a decrease in grain weight. The water deficit decreased ear-leaf water potential over a period of several weeks, and caused the leaves to senesce prematurely. Leaf photosynthesis was inhibited at leaf water potentials below –1.2 MPa. After this point, grain filling depended primarily on mobilization of reserves. Periodic measurements of grain mass showed that kernels on water deficient plants 19 continued to fill at the control rate, confirming the results of the short-term greenhouse study. The water-deficient plants produced smaller grains because they ceased filling sooner after anthesis. The decrease in grain yield from 11.1 to 9.1 Mg ha-1 in this experiment was due entirely to the decrease in grain size. The embryos within the kernels showed the same general response. Water deficient plants produced smaller embryos because of a shorter duration of filling. What was the physiological basis for the shorter duration of filling? One possibility was lack of assimilate. It is well known that carbohydrate reserves in the stem are remobilized to support kernel growth, and that under severe stress, grain growth will continue until these reserves are nearly depleted (Westgate and Boyer, 1985; Simmons and Jones, 1985). Measurements of non-structural carbohydrate in leaves, vegetative stalk, and reproductive stalks revealed that assimilate levels decreased in the water-stressed plants relative to levels in the well-watered controls. But carbohydrates in the vegetative stalk were not completely depleted when grain filling ceased (about 50 days after anthesis), and carbohydrates began to accumulate thereafter. Also, there was no indication that carbohydrate accumulation in the grain was altered due to a lack of photoassimilate. These data support the conclusion that assimilate supply was not the primary cause of a shorter filling period in the water deficient plants. The fact that carbohydrates continue to accumulate in the vegetative stalk of well-watered plants throughout grain filling supports this conclusion. Measurements of grain water status showed that kernels on the water-deficient plants reached the same maximum water content as did the well-watered plants, but began to lose moisture sooner (but not faster) after anthesis. This premature desiccation was consistent with the decrease in moisture content observed by Westgate and Grant (1989) earlier. The fact that the kernels on the water-deficient plants achieved the same maximum water volume as did kernels on well-watered plants indicated that the potential grain size was the same in both treatments. But the premature desiccation raised the possibility that the change in grain water status early in filling actually affected the duration of the grain filling process. If so, there should be a consistent relationship between grain moisture status and grain development, regardless of the treatments imposed. When compared on a tissue moisture basis, the developmental pattern of kernel and embryo water potential were nearly identical for the well-watered and waterdeficient treatments. Likewise, data examined from other studies shows a 20 very consistent pattern of dry matter accumulation when development is normalized on a kernel moisture basis. Regardless of genotype or environment, kernels ceased to accumulate dry matter at about 30 % moisture (Egli and Tekrony 1997, Westgate and Boyer, 1986). The same is true of kernel development when water deficit treatments are imposed at different times during grain filling. Withholding water from plants at the kernel blister stage (rapid water uptake), dough stage (early grain filling), or dent stage (late grain filling) had dramatically different effects on final kernel dry matter. But kernels in all treatments ceased to accumulate dry matter at about the same moisture content – approx. 30%. Evidence from a host of sources including in vitro measurements of enzyme activities at low moisture contents (Rupley et al., 1983), enzyme activities extracted from dry seeds (Muhitch, 1991), and the physical properties of water in dry seeds (Vertucci, 1989) all point to the conclusion that the osmotic conditions in the endosperm and embryo late in grain filling do not directly inhibit the activity of enzymes responsible for storage product synthesis. Rather, the decrease in storage product synthesis probably reflects a decreased capacity for protein synthesis (Bewley, 1981; Kermode et al., 1989). Desiccation itself may be a developmental queue for terminating transcription and translation of synthetic proteins and initiating the synthesis of proteins required for desiccation tolerance, eg. dehydrins. If so, the premature desiccation caused by drought during grain filling causes smaller kernels to be produced because they reach the minimum moisture content that supports metabolism sooner after anthesis. Maintaining grain filling then requires modifications in grain development that either increase the maximum water volume early in filling, or prevent premature desiccation from occurring. Current studies are investigating the importance of increasing maximum cell volume as a means to achieve this goal. The soybean embryo undergoes a similar pattern of development, but maximum dry weight is achieved coincident with maximum water content (and therefore, maximum fresh weight). The embryo desiccates rapidly thereafter. As was the case in maize, drought during grain filling shortens the duration of filling, resulting in smaller seeds (Meckel et al., 1984). Water deficits severe enough to inhibit leaf photosynthesis completely had little effect on the rate of dry matter accumulation by the embryo – at least on the short term (Westgate et al., 1989). Part of the reason embryo could continue to develop was that they exhibited the same 21 type of hydraulic isolation from changes in plant water status that was observed in maize. There also was an increased rate of mobilization of carbohydrate and reduced N from the vegetative tissues and pod wall. Interestingly, the embryos from water deficient plants also exhibited an increased capacity for sucrose uptake from the surrounding free space, which apparently compensated for the decrease in apoplast sucrose concentration being by the maternal tissues. This study indicated that embryo growth rate is maintained in soybean by a coordinated shift in metabolism among the vegetative and reproductive tissues. This coordinated adjustment allows embryo growth to continue until reserves are depleted. The fact that soybean embryos reached maximum water volume and dry weight at the same time suggest that embryo water status might also be an important determinant of grain fill duration. In vitro culture studies show that soybean cotyledons will continue to accumulate dry matter as long as can continue to expand in water content and volume. Embryo removed from pods early in grain filling achieve much greater mass than their in planta counterparts (Egli, 1990). This type of evidence implies that embryo mass in planta is limited, at least in part, by the surrounding pericarp (pod) walls. This possibility was tested by placing small tubes around developing pods to restrict their expansion. The results clearly show that embryos with limit maximum water volume develop into smaller seeds. Related studies indicate that restricting volume shortens the grain fill period, and that the ‘restriction’ is reversible during most of the grain filling period. Whether water deficits during grain filling limit maximum embryo volume in a similar manner has not been examined. But it is a reasonable possibility since numerous studies confirm that, regardless of treatment or growth conditions, grain filling in soybean ceases at about 60% moisture. In summary, embryos that store a large amount of protein in vacuole (eg. soybean embryos, maize embryos), protein accumulation continues until maximum water volume is achieved. In seeds that store predominately starch (egg. maize, wheat) grain fill duration is determined by the maximum cell volume established early in seed development, and by the onset of desiccation later in development. Drought shortens the duration of grain filling by causing premature desiccation and/or by limiting the maximum cell volume. 22 VI. CONCLUDING REMARKS These studies have raised a number of questions regarding the control of grain filling and the impact that adverse environmental conditions have on the capacity of developing seeds to reach their potential size. How do environmental conditions early in grain filling regulate the duration of the filling process? Under what conditions does lack of assimilate supply limit the duration of grain filling, rather than seed water status? And finally, how is seed composition (eg. protein, oil, starch content) maintained under conditions that alter the rate and duration of grain filling? Because lack of available soil moisture is the primary environmental limitation for high grain yields, answers to these questions will lead to rational targets for genetic modification, physiological selection, and practical management strategies to improve grain production across a range of environments. REFERENCES ANDERSON, M.N.; ASCH, F.; JENSEN, C.R.; NAESTED, H. Drought regulates invertase expression and carbohydrate levels in maize ovaries. In: PROC. INTL. CONGRESS OF PLANT MOL. BIO., 6., Quebec, Ontario, CA, 2000. BASSETTI, P.; WESTGATE, M.E. Floral asynchrony and kernel set in maize quantified by image analysis. Agronomy Journal, v.86, p.699703, 1994. BEWLEY, J.D. Protein synthesis. In: PALEG, L.G; ASPINALL, D. (eds.). The Physiology and Biochemistry of Drought Resistance in Plants. Sydney: Academic Press, 1981. p.261-282. BRADFORD, K.J. Water stress and the water relations of seed development: A critical review. Crop Science, v.34, p.1-11, 1994. CHRISTY, A.L.; PORTER, C.A. Canopy photosynthesis and yield in soybean. In: GOVINDGE (ed.). Photosynthesis: Development, Carbon Metabolism and Plant Productivity. Academic Press, 1982. v.2, p.499-511. 23 CHRISTY, A.L.; WILLIAMSON, D.R. Characteristics of CO2 fixation and productivity of corn and soybeans. In: LUDDEN, P,W.; BURRIS, J.E. (eds.). Nitrogen Fixation and CO2 Metabolism. Elsevier Science Pub. Co., 1985. CHRISTY, A.L., WILLIAMSON, D.R.; WIDEMAN., A.S. Maize source develoment and activity. In: SHANNON, J.C.; KNIEVEL, D.P.; BOYER, C.D. (eds.). Regulation of Carbon and Nitrogen Reduction and Utilization in Maize. Amer. Soc. Plant Physiol., 1986. EDMEADES, G.O.; BOLANOS, J.; ELINGS, A.; RIBAUT, J.M.; BANZIGER, M.; WESTGATE, M.E. The role and regulation of the anthesis-silking interval in maize. In: WESTGATE, M.E.; BOOTE, K.J. (eds.). Physiology and Modeling Kernel Set in Maize. Crop Sci. Soc. America and Amer. Soc. Agron., Madison, WI., 2000. EGLI, D.B. Seed water relations and the regulation of the duration of seed growth in soybean. Journal of Experimental Botany, v.41, p.243248, 1990. EGLI, D.B.; ZHEN-WEN, Y. Crop growth rate and seeds per unit area in soybean. Crop Science, v.31, p.439-442, 1991. EGLI, D.B.; TEKRONY, D.M. Species differences in seed water status during seed maturation and germination. Seed Science Research, v.7, p.3-11, 1997. FLENET, F.; KINIRY, J.R.; BOARD, J.E.; WESTGATE, M.E.; REICOSKY, D.C. Row spacing effects on light extinction coefficients of corn, sorghum, soybean, and sunflower. Agronomy Journal, v.88, p.185190, 1996. KERMODE, A.R.; OISHI, M.Y.; BEWLEY, J.D. Regulatory roles for desiccation and abscisic acid in seed development: A comparison of the evidence from whole seeds and isolated embryos. In: STANWOOD, P.C.; M.B.; McDONALD, M.B. (eds.). Seed Moisture. Madison: Crop Sci. Soc. Amer., 1989. v.14, p.23-50. 24 MECKEL, L.; EGLI, D.B.; PHILLIPS, R.E.; RADCLIFFE, D.; LEGGET, J.E. Effect of moisture stress on seed growth in soybeans. Agronomy Journal, v.76, p.647-650, 1984. MUHITCH, M.J. Tissue distribution and developmental patterns of NADHdependent and ferridoxin-dependent glutamate synthase activities in maize (Zea mays L.) kernels. Physiologia Plantarum, v.81, p.481488, 1991. RUPLEY, J.A.; GRATTON, E.; CARERI, G. Water and globular proteins. Trends Biochem., v.8, p.18-22, 1983. SCHUSSLER, J.R.; WESTGATE, M.E. Maize kernel set at low water potential. II. Sensitivity to reduced assimilates at pollenation. Crop Science, v.31, p.1196-1203, 1991. SCHUSSLER, J.R.; WESTGATE, M.E. Increasing assimilate reserves does not prevent kernel abortion at low water potentials in maize. Crop Science, v.34, p.1569-1576, 1994. SIMMONS, S.R.; JONES, R.J. Contributions of pre-silking assimilate to grain yield in maize. Crop Science, v.25, p.1004-1006, 1985. VERTUCCI, C.M. The effect of low water contents on physiological activities of seeds. Physiologia Plantarum, v.77, p.172-176, 1989. WESTGATE, M.E.; BOYER, J.S. Carbohydrate reserves and reproductive development at low leaf water potentials in maize. Crop Science, v.25, p.762-769, 1985. WESTGATE, M.E.; FORCELLA, F.; REICOSKY, D.C.; SOMSEN, J. Rapid canopy closure for maize production in the northern US corn belt: Radiation-use efficiency and grain yield. Field Crops Research, v.49, p.249-258, 1996. WESTGATE, M.E.; THOMSON GRANT, G.L. Water deficits and reproduction in maize. Responses of the reproductive tissues to water deficits at anthesis and mid-grain fill. Plant Physiology, v.91, p.862867, 1989. 25 WESTGATE, M.E.; BOYER, J.S. Reproduction at low silk and pollen water potential in maize. Crop Science, v.26, p.951-956, 1986. WESTGATE, M.E.; SCHUSSLER, J.R.; REICOSKY, D.C.; BRENNER, M.L. Effect of water deficits on seed development in soybean. II. Conservation of seed growth rate. Plant Physiology, v.91, p.980-985, 1989. ZINSELMEIER, C.; WESTGATE, M.E.; SCHUSSLER, J.R.; JONES, R.J. Low water potential disrupts carbohydrate metabolism in maize (Zea mays L.) ovaries. Plant Physiology, v.107, p.385-391, 1995. ZINSELMEIER, C.; LAUER, M.J.; BOYER, J.S. Reversing drought-induced losses in grain yield: Sucrose maintains embryo growth in maize. Crop Science, v.35, p.1390-1400, 1995. ZINSELMEIER, C.; BYEONG-RYONG, J.; BOYER, J.S. Starch and the control of kernel number in maize at low water potential. Plant Physiology, v.121, p.25-35, 1999. 26