Our current world population of 7.1 billion is projected to burgeon to as much as 10 billion by 2050, an increase of 2 to 3 billion in approximately 40 yr. This increase, coupled with changing diets that include more animal products, will result in a 70% increase in food demand (UNESCO, 2012). Dryland agriculture must play an increasingly important role toward meeting this challenge (Stewart et al., 2006). Peterson et al. (2012) stated that this will be necessary for two reasons. First, since about 1960, most increases in the world’s food supply resulted from increasing crop yield as cropland area remained nearly constant. While additional lands remain that can be brought into production, they tend to be less productive and more environmentally sensitive. Second, and perhaps more important, the world’s supply of fresh water for irrigation is limited and increasingly the object of competition.
Dryland agriculture and dryland farming are universally used terms but the meaning of the terms can only be discovered by understanding the context in which they are used. These terms are also often used interchangeably with rainfed agriculture, but they are vastly different. Rainfed agriculture includes dryland farming, but dryland farming is generally defined as agriculture in regions where lack of moisture limits crop production for part of the year. Stewart and Burnett (1987) stated that dryland farming emphasizes water conservation in every practice throughout the year. It is within this context that dryland farming is used in this article.
Drought is a recurring phenomenon in dryland farming areas, and particularly in the Great Plains of North America. While drought is defined in various ways, drought in dryland farming areas is almost constant and varies only in severity from year to year. Agricultural drought links various meteorological characteristics to agricultural impacts; the key issues are: precipitation shortages, differences between actual and potential evapotranspiration, soil water deficits, reduced groundwater supplies and/or reservoir levels. Agricultural droughts occur in almost all areas, but the impact of drought in a dryland farming area is usually more severe because precipitation amounts in dryland areas are considerably less than the potential evapotranspiration. This is self-evident when you consider that dryland farming occurs primarily in semiarid areas where annual precipitation is <25 to 50% of the potential evapotranspiration demands. Weeks and even months can pass without meaningful precipitation so the soil not only becomes depleted of plant available water, but can even lose significant amounts of water below wilting point. For example, at the end of the cropping year many agricultural soils contain as much water in the soil profile at the wilting point, as the plants were able to extract from the soil when it was at field capacity (field capacity is the maximum amount of water the soil can hold against gravity). Even though plants cannot extract the water remaining in the soil at the wilting point, it can evaporate. This means that to return the soil to field capacity, an amount equivalent to the “plant available water” plus the water lost below the wilting point must be added. This is vastly different than in more humid regions where droughts also occur. In these more favored areas, plants may become water stressed, but in most cases the soils in these areas do not lose significant amounts of water below the wilting point; thus when a precipitation event occurs, essentially all of the precipitation retained by the soil will be available to the plants. In summary, a smaller percentage of the rainfall retained in the soil in dryland farming areas is subsequently available to plants compared to more humid areas.
Dryland farming occurs mainly in semiarid regions. Thornthwaite (1941, 1948) was an early pioneer in classifying climates. He stated that one knew what to expect of the climate in a desert and could plan accordingly and the same was true for humid regions. However, Thornthwaite (1941) said “Men have been badly fooled by the semiarid regions because they are sometimes humid, sometimes desert, and sometimes a cross between the two. Yet it is possible to make allowances for this too, once the climate is understood. The author argues that the semiarid regions are now understood well enough to do a good job with them and avoid the failures and tragedies of the past.” Time has shown that Thornthwaite was mostly right in his analysis, but dryland farming in semiarid regions remains challenging with highly variable outcomes.
Green water is that fraction of rainfall that infiltrates into the soil and is available to plants. It includes the plant available water held at field capacity and the continual replenishment of reserves by rainfall (Ringersma et al., 2003). When soil drier than the wilting point absorbs precipitation, the deficit between the actual water percentage and the wilting point percentage must be replaced before any of the precipitation is available for plants. Therefore, dryland areas not only receive less precipitation than more favored areas, they also lose a higher proportion of the rainfall by evaporation leaving a smaller proportion as green water. This fact offers great potential for increasing green water use in dryland areas, but the constraints also are great. The purpose of this article is to address these constraints and present strategies for increasing and managing green water for dryland crop production.
DROUGHT MANAGEMENT AND AVOIDANCE
Bowden (1979) provides four management keys unique to semiarid lands. It is imperative that these keys be understood and applied as scientists develop new strategies and practices.
Key 1. No growing season is or will be nearly the same in precipitation amount, kind, or range, or in temperature average, range, or extremes, as the previous growing season. Although this key is critical in any rainfed system, it requires absolute attention in dryland farming. Crop cultivation requires an adjustment every year, which leads to the second key.
Key 2. Crops cannot be planned or managed the same manner from season to season. Most of the world’s agricultural practices in either humid or arid areas have some predictability on an annual basis. In semiarid climates, however, even highly mechanized, technically advanced, commercial farms such as those in the High Plains of North America or the outback of Western Australia do not have sufficiently stable production for the individual or government to count on a given production figure for the following season.
Key 3. The soil and water resource does not remain the same for any extensive period of time once agriculture is introduced. This key is based on the premise that soils of most semiarid lands developed under grass on relatively flat topography. The competition for water and nutrients to produce crops requires removal of the protective grass cover. Because the crops are annual and dependent on precipitation, severe drought often leaves the soil highly vulnerable to wind erosion. Severe drought also leaves the soil highly vulnerable to water erosion from high intensity rainfall, even though this tends to be a more isolated and less frequent event than wind erosion.
Key 4. There is abundant sunshine due to many cloud-free days. This key has potential benefit and is shared with most arid climates. Abundant sunshine means higher temperatures that induce rapid growth, but it also creates a situation that demands careful management of soil water. Warm seasons, high sun, and cloud-free conditions stimulate growth, but also increase evaporation and transpiration. It is possible for a grain crop to mature rapidly due to several weeks of sun-drenched, rainless conditions and desiccate just days before ripening. It is equally possible for a few millimeters of precipitation to occur at almost the last moment and produce a good grain crop.
Rockström et al. (2010) stated that the world is facing a water crisis with little room for further expansion of large-scale irrigation. This dramatically increases the need for improved water management in rainfed agriculture. Rockström et al. (2010) further state that in semiarid and dry subhumid zones, it is not the amount of rainfall that is the limiting factor of production as much as the extreme variability of rainfall, with high rainfall intensities, few rain events, and poor spatial and temporal distribution. They concluded that the large gaps between actual and attainable yields in rainfed agriculture in many regions of the world suggest a large untapped potential for yield increases.
A graphical model of the water balance for a wide range of climatic conditions (Ponce, 1995) is presented in Fig. 1. This model was based on the range of climates in the Sertäo region of Brazil, but it exemplifies important differences in how water is used over a range of climatic zones. Although Fig. 1 is only conceptual, and ignores drainage regardless of the amount of precipitation, it does illustrate important points. Applying this to the colors of water, the ET in the graph would be “green” water, the runoff “blue water” and the water that evaporates goes toward meeting the “unmet potential ET.”
While the concept shown in Fig. 1 supports the view of Rockström et al. (2010) that there is a large untapped potential for using water more efficiently, it also points out the challenges of successfully tapping this potential. The greatest untapped potential is increasing the percentage of precipitation captured by the soil that is subsequently used for transpiration relative to that which is lost by evaporation. The unmet potential for evaporation or transpiration is substantial for semiarid ecosystems while close to zero for humid ecosystems. The ratio of precipitation consumed in transpiration compared to that lost by evaporation decreases greatly with increasing aridity. Water lost to evaporation after it has been stored in the soil is a loss of green water. Water used for transpiration is intrinsic to the development of plant biomass, while that lost by evaporation is not.
There are three basic strategies for increasing crop yields in dryland cropping systems. The first is to increase the capture of precipitation by reducing runoff and to store it in the soil profile for later use by the crop for ET. The second is to increase to the fullest extent feasible the portion of ET that is used for transpiration relative to that lost by evaporation from the soil surface. The third is to ration plant water use so that there is water available during the reproductive and grain-filling periods, particularly for grain crops.
Developing Cropping Systems
The yield of a grain crop can be expressed by the following equation.where GY is kg ha–1 of dry grain yield; ET is kg ha–1 of evapotranspiration (water use by evaporation from soil surface and transpiration by the crop between seeding and harvest); T/ET is the fraction of evapotranspiration transpired by the crop; TR is the transpiration ratio (number of kilograms water transpired to produce 1 kg of aboveground biomass); and HI is the harvest index (kg dry grain/kg aboveground dry biomass). While this equation applies to all situations where grain crops are produced, the ranges of values for each of the components become considerably greater and more variable in dryland farming areas. Each of these components will be discussed for dryland regions and strategies for their improvement will be presented. Grain sorghum [Sorghum bicolor (L.) Moench] and corn (Zea mays L.) will be used as example crops.
Grain yields under dryland farming conditions vary widely from year to year. The average farmer GYs for grain sorghum in the southern Great Plains is approximately 2000 kg ha–1 (Bandaru et al., 2006). For any given year, however, the GY can be as low as 0 or as high as 6000 kg ha–1. It is common in dryland farming areas for yields of major crops to range from zero to about three times average yields. Therefore, the various components shown in Eq.  also vary greatly from year to year.
In dryland farming systems ET is calculated as the sum of growing season precipitation plus water extracted from plant available water in the rooting zone portion of the soil profile. For grain sorghum in the southern Great Plains, growing season precipitation is generally <50% of the potential ET even though about 50% or more of the average annual precipitation occurs during the growing season (Bandaru et al., 2006; Jones and Johnson, 1996; Stone and Schlegel, 2006). Therefore, successful grain production is greatly dependent on having a substantial amount of plant available water stored in the soil profile at the time of seeding to supplement growing season precipitation. Jones and Johnson (1996) showed in a 9-yr study that grain sorghum at Bushland, TX, used an average of 84 mm (22% of the ET) of stored soil water during the growing season. Therefore, most successful cropping systems in these areas have a fallow period when no crop is grown to accumulate precipitation water in the soil profile for use by a subsequent crop. Historically, a common cropping system was winter wheat (Triticum aestivum L.) every 2 yr with a 14 to 16 mo fallow period between crops. Although the amount of fallow period precipitation that was actually stored in the soil profile for the subsequent wheat crop was often <20%, it was essential for a successful crop. In recent years, cropping systems that use shorter fallow periods combined with conservation tillage and no-tillage have greatly increased the capture and storage of precipitation during fallow periods (Peterson et al., 2012; Unger et al., 2006, 2012; Stewart et al., 2010). The storage of 40% or more of the precipitation occurring during the fallow periods is now possible, which can result in an increase of 70 to 100 mm ET for the succeeding crop. In addition to increasing yield, these conservation systems reduce wind and water erosion and enhance or maintain soil organic matter content.
The portion of ET that is used as T is affected by many variables and varies greatly within short periods as well as over the season. Ritchie and Burnett (1971) quantified the effect of leaf area index (LAI) on T/ET when ET was not constrained by available water in the root zone. They concluded that T/ET when the soil surface was dry ranged from 0 for LAI of 0, to about 0.5 for LAI of 1, and to approximately 1 for LAI of 3 and greater. Since dryland agriculture is totally dependent on precipitation and the source for most of the ET is rainfall during the growing season, the soil surface is frequently wetted and considerable water can be lost as evaporation, particularly during the early part of the growing season and even later in the season for dryland crops because of low LAI. Bandaru et al. (2006) reported grain sorghum LAI values in the range of 1.5 of plants 60 d after planting at Bushland, TX. Therefore, the average seasonal T/ET for dryland crops in semiarid regions is likely to be in the range of 0.5. Lascano and Baumhardt (1996) reported that the combined seasonal soil and cotton (Gossypium hirsutum L.) canopy evaporation (ET) was approximately 330 mm over a 100-d period and the T portion was 164 mm when the soil was bare except for the growing cotton. Thus, the T/ET was 0.50. In contrast, when the cotton was grown on soil covered with wheat residues, the net irradiance at the soil surface was reduced and evaporation of water from the soil surface was also reduced leaving more water available for T by the cotton plants. The total ET for cotton growing on the mulch covered soil was also 330 mm, but the amount used by T was 223, so the T/ET was 0.74. Lascano et al. (1994) reported the lint yield of cotton increased from 613 kg ha–1 for bare soil to 830 kg ha–1 with residue. When divided by the corresponding T amounts, both treatments averaged a constant conversion of 0.38 Mg m–3 water. The corresponding water-use efficiency values for the 330 mm ET were 0.19 and 0.26 Mg m–3 for the bare and residue-covered soils, respectively, which was a 37% increase. Earlier, Unger and Jones (1981) conducted a 3-yr study of the effect of a growing-season mulch on grain sorghum. They found that water-use efficiency was increased 19% when the soil was covered with 8 Mg ha–1 of wheat residue, which was less than they expected based on earlier experiments and observations. They concluded that soil shading by the plant canopy largely substituted for the beneficial effect of mulch during the growing season, and that the most important role of mulch in dryland cropping systems is to enhance water storage during the fallow period which increases the ET for subsequent crops as discussed in the Evapotranspiration section. Even though the benefit of mulch during the growing season was less than Unger and Jones (1981) anticipated, at a later point we will discuss the synergistic effects of changing the components of Eq. .
Accumulation of plant biomass is directly related to water availability (Sinclair, 2009a). Sinclair states that the difference in vapor pressure inside and outside the leaf (VPD) controls the water loss through the stomata. The VPD in arid regions is large because the vapor pressure of the atmosphere is very low relative to humid areas. For a given environment, the VPD cannot be controlled—it is what it is. Sinclair (2009b) stated “Despite claims that crop yields will be substantially increased by the application of biotechnology, the physical linkage between growth and transpiration imposes a barrier that is not amenable to genetic alteration.” While many plant scientists disagree, there is little evidence to date to repute it (Gurian-Sherman, 2012).
Sinclair and Weiss (2010) state that the TR of C4 crops, like corn and grain sorghum, is about 220 g water d–1 for each gram of biomass produced when growing in a transpiration environment of 2 kPa which they classify as somewhat “average”. For an arid region with a transpiration environment of 2.5 kPa, they state that the TR for C4 crops increases to about 280 g for each g biomass, but decreases to about 160 g when the crop is growing in a humid transpiration environment of 1.5 kPa. For C3 crops like wheat, they state the TR is about 1.5 times greater than for C4 crops. Obviously a much larger amount of water is required to produce a unit of biomass in a dryland production area relative to more humid zone. Many factors affect the T environment of a plant, but in simple terms, it depends on: (i) how hot it is, (ii) how sunny it is, (iii) how windy it is, and (iv) how dry the air is. The reader should be aware that the T environment is determined by the difference in the vapor pressure inside and outside the leaf surface, and while temperature, humidity, radiation, and wind speed are commonly measured at a weather station and reported, the measurements do not necessarily represent the conditions at the leaf surface. For example, the T environment of a dryland grain sorghum crop with 50,000 plants ha–1 will be very different from an irrigated crop of grain sorghum that has 150,000 plants ha–1 growing in the same environment. The microclimate largely determines the T environment, and in general, the closer plants are to one another, and the more the soil surface is shaded, the more favorable the microclimate. In dryland areas, because water is so limited, plant densities must be reduced significantly resulting in an increased VPD that requires more g water/day to produce a gram of biomass. Nevertheless, it is well documented that there is a linear relationship between water transpired by plants and the production of aboveground biomass, and that the slope of the line represents the T efficiency of the crop growing in that T environment. Figure 2 illustrates this relationship for three different T environments and shows that the t ha–1 biomass produced from 400 mm of water can be almost two times more for plants growing in a humid Tenvironment relative to an arid T environment.
Harvest index, the ratio of grain to aboveground dry matter, is a species-related parameter and is sometimes recommended for screening cultivars. Prihar and Stewart (1990) proposed that the slope of a line beginning at the origin and passing along the upper-bound points of a grain yield vs. dry matter plot approximates the genetic HI of a crop species. This is because the highest grain yields plotted against given dry matter yields represent the least-stressed and/or stress adapted plants, and passing through the origin is necessary to satisfy the definition of HI. Prihar and Stewart (1990) reported that upper HI values stated in the literature for sorghum and corn ranged between narrow limits of 0.48 and 0.53 and 0.58 and 0.60, respectively. They concluded these values approximated the genetic HI and would only be achieved under stress-free conditions. As stress conditions increase, particularly water stress, the HI values decrease and this relationship is well documented. Steiner et al. (1994) estimated crop residues remaining in the field for several crops after harvesting and estimated HIs for grain sorghum of 0.32, 0.38, 0.42, 0.44 and 0.45 for grain yields in semiarid regions of 1, 2, 4, 7, and 9 Mg ha–1, respectively. Estimated HIs for corn were 0.21, 0.32, 0.43, 0.46, 0.53, and 0.56 for grain yields of 1, 2, 4, 7, 9, and 12 Mg ha–1, respectively. Although the HI value generally increases as yield increases, this is not always the case. The HI value of a low yielding dryland crop can be relatively high if the plant density is low and there is an adequate available water supply during the reproductive and grain-filling growth stages. The HI value is dictated by the stress level rather than yield level, but in most cases low yields are caused by water stress, so there is usually a close correlation between HI and grain yields.
Since the equation GY = ET × T/ET × 1/TR × HI is linear, increasing any one of the factors by a certain percentage, while leaving the other factors unchanged, will increase the GY by that same percentage. For example, Jones and Johnson (1996) summarized a 10-yr study showing that continuously grown grain sorghum averaged 352 mm ET and produced 2050 kg ha–1 grain (adjusted to 0% water). Based on the discussions above, T/ET is assumed to be 0.50 and HI to be 0.38. Solving for TR, the resulting value is 326 suggesting that 326 kg of water was transpired for each kilogram of aboveground biomass produced. The TR of 326 is realistic, but indicates relatively arid growing conditions, based on the values proposed by Sinclair and Weiss (2010) and presented in Fig. 2. In the same study, Jones and Johnson (1996) reported a yield of 2735 kg ha–1 for grain sorghum growing in a wheat–sorghum–fallow (WSF) system where the ET was 376 mm, an increase of 6.8% over the 352 mm for the continuous grain sorghum. If the other factors remained the same as for the continuous sorghum treatment, the yield would have increased only to 2190 kg ha–1 instead of the 2735 kg ha–1 reported. Assuming the T/ET value increased to 0.55 and the HI increased to 0.40, and the TR decreased to 302, the calculated yield would be 2739 kg ha–1. This example suggests that by increasing the ET by 6.8%, the T/ET increased 10%, the HI 5%, and the TR 9%, and increased grain yield more than 30% demonstrating a significant synergistic effect. Changing one factor will likely always change others, but the ET factor will always be dominant and will be the driver for making the other factors more positive. Obviously these factors are not going to change in direct proportion to one another, but the point is that changing one factor almost always changes the others.
To gain insight about how the various factors in Eq.  might interact under field conditions in the central and southern U.S. Great Plains, GY–ET relationships from the literature were used to determine GY and ET, and then T/ET ratios and HI values were estimated based on discussions above in the Transpiration Ratio and Harvest Index sections, and TR values were calculated as unknowns using the equation. The results are presented in Table 1, and while most of the values look reasonable, some appear questionable. The T/ET and HI values were assumed to increase linearly as ET amounts increased. The grain sorghum studies used were primarily dryland or with only limited irrigation, while the corn studies included full-irrigation treatments. The TR values were generally in the range of 225 to 275. Sinclair and Weiss (2010) stated that C4 crops require 220 kg of transpired water for each kilogram biomass produced when the average transpiration environment was 2 kPa and would require 280 kg transpiration for a kilogram biomass in an arid climate of 2.5 kPa. The calculated TR values were somewhat similar for corn and grain sorghum as suggested. The calculated TR values for corn were slightly lower for Bushland, TX, than for Nebraska as would be expected. However, the Bushland TR values for grain sorghum were less than those for Kansas which was not as expected. The interesting point is that this analysis suggests that the TR values remain more stable than any of the other factors. Changing the TR values in a linear fashion and then solving for either T/ET or HI by Eq.  results in unreasonable values. The grain sorghum results for Tribune, KS, are of particular interest. Stone and Schlegel (2006) developed separate GY–ET relationships for conventional till and no-till cropping systems and the no-till GYs were considerably higher for similar amounts of ET (Table 1). This resulted in significantly increased T/ET values, which strongly suggests that there was less evaporation from the soil surface leaving more water for transpiration. This result is similar to what Lascano and Baumhardt (1996) reported regarding straw mulch increasing the T/ET ratio from 0.5 to 0.74 for a 100-d growing period of cotton.
|Crop and location||Grain yield†||ET||T/ET||TR||HI|
|kg ha–1 dry wt.||mm ha–1||kg H2O kg biomass–1||kg grain kg biomass–1|
While only the ET and GY results presented in Table 1 are based on actual data, the values shown for each level of ET when placed in Eq.  will result in the corresponding GY. These results may be very useful in evaluating strategies for managing green water in dryland areas.
Evaluating Strategies for Managing Water in Dryland Agriculture
To this point we have clearly shown that water is the most limiting factor in dryland agriculture and that various strategies can be used to improve its management. Water is generally most limiting during the reproductive and grain-filling growth stages for grain crops in dryland areas because plant available water from the soil profile is often depleted during the vegetative periods. Strategies designed to save some of the soil water for use during the latter growth stages are common. The information presented in Table 1 shows that the four factors that determine grain yield are closely intertwined and changing one factor in a positive direction often leads to changing another in a negative direction. Therefore, it is important to consider the maximum effect that addressing one factor may have on the others.
Maximizing Evapotranspiration Amounts
Without question, the amount of ET is the most important factor determining GY, and not only is the amount highly variable from year to year, but the time the precipitation occurs during the growing season is also erratic. Having an adequate supply of plant available water stored in the soil at the time of crop planting is vital to supplement the amount of ET and to buffer the supply between precipitation events. Remarkable progress in increasing the amount of water stored in the soil before crop planting has been made in dryland regions in the past few decades. The amount of water stored in the soil profile before planting the crop has been increased by 75 to 100 mm (Peterson et al., 2012; Unger et al., 2006, 2012). The water stored in the soil at the time of seeding a crop can account for 30 to 50% of the ET during years when small amounts of growing season precipitation are received and can be an important source between precipitation events even in years when growing season precipitation is above average. Precipitation amounts in dryland regions often range from <50% of average to more than 200% of average, and similar extremes occur during the growing season. Consequently, it is common in dryland areas for grain yields to range from zero to three times average. The amount of stored water at the time of seeding greatly reduces risk and increases yield. It is important to have a good estimate of the amount of plant available water in the soil profile at time of seeding because this information combined with the probability of receiving a certain amount of growing season precipitation allows one to estimate potential GY.
Because water is the limiting factor for GY essentially every year in dryland areas, the ideal situation is to use as much of the stored soil water and the growing season precipitation as is feasible. However, this is not always possible because substantial amounts of precipitation late in the season do not benefit the crop, but it partially recharges the soil for the next crop cycle. Nevertheless, the goal should always be to maximize ET.
Adjusting the number of plants per unit area is perhaps the most common water management strategy used in dryland areas. Populations of grain sorghum and corn can be as high as 200,000 and 100,000 plants ha–1, respectively, when water is adequate. Under dryland conditions, a common strategy is to reduce the plant density to limit growth and prevent early depletion of plant available water from the soil profile. Corn and grain sorghum populations are often reduced to populations as low as 25,000 and 50,000 plants ha–1, respectively, or even lower. Referring to Eq.  and Table 1, plant density affects each factor differently. The amount of ET will generally be highest when plant densities are high because more of the stored soil will be extracted. In general, soil water extraction is maximized by decreasing row width and increasing plant density (Stewart and Steiner, 1990) so the amount of ET is maximized. Decreasing plant population will also tend to decrease the T/ET factor because lower plant density results in less shading of the soil which will increase evaporation from the soil. A low population density also stimulates the formation of tillers, particularly when growing conditions are favorable early in the growing season. While the formation of tillers can be beneficial, tillers can decrease grain yield in dryland areas because they increase water use during the vegetative period and often fail to produce grain when water becomes limiting during latter growth stages. Also, TR will likely be affected negatively because the microclimate of plants further apart from one another will result in a higher vapor pressure deficit. Because the amount of biomass is equal to (ET × T/ET × 1/TR), decreasing the plant density will usually result in decreased biomass production. The only factor positively affected by decreasing plant density is HI, and it is difficult to increase the HI factor enough to offset the decreases of the others. Many times the expected benefits from reducing plant populations are negated. The main benefit of reducing plant density is reducing risk rather than increasing water use efficiency. Lyon et al. (2003) conducted a computer model simulation study to supplement field results from western Nebraska and reached a similar conclusion.
The use of skip-rows is another strategy used in dryland areas. However, this strategy has many of the same shortcomings as reducing plant density. As already mentioned above, the amount of ET is generally maximized by reducing row width and increasing plant population. Keeping plant population constant, and using a skip-row system reduces space between plants within the planted rows, but the distance between the planted rows is substantially increased depending on whether the row geometry is plant 1-skip 1, plant 2-skip 1, plant 2-skip 2, or some other configuration. Skip-row systems tend to lower the amount of ET because the extraction of stored soil water between the wider row spacing is less than for the every row planted configuration. The hypothesis is that plants cannot reach that water until late in the season when the water is needed for grain filling. The primary factor that is affected positively is the HI, although the TR may be affected somewhat positively because the plants are closer to one another in the planted rows which may improve the microclimate and reduce the vapor pressure deficit. The T/ET factor, however, is likely to be significantly lower because the amount of soil exposed to the sun and wind is large. This will certainly be true if there are frequent precipitation events because much of the water falling on the area without plants will be lost by evaporation. Therefore, the most likely benefit of skip-rows is reducing risk when water is extremely limited rather than increasing water use efficiency for grain production. Lyon et al. (2009) conducted trials with corn at multiple locations in Kansas and Nebraska using plant 2-skip 2, plant 1-skip 1, and plant 2-skip 1 configurations using 75 cm rows. They recommended plant 2-skip 2 for risk-averse growers where GYs were likely to be less than about 4700 kg ha–1, and plant 1-skip 1 for growers with moderate risk-aversion, and when GYs were likely to be between 4700 and about 6000 kg ha–1. They found little or no interaction between population levels and skip-row configurations. Skip-row configurations were not recommended when expected GYs were above 6000 kg ha–1.
Theoretically, use of mulch during the growing season should increase the T/ET ratio, and potentially have a positive effect on TR. This concept is supported by the data in Table 1 for no-till grain sorghum in Kansas where no-till and conventional tilled systems were compared and showed GY increases in the range of 10 to 25%. Lascano and Baumhardt (1996) also reported positive results with cotton. Unger and Jones (1981), even though they showed a positive response to mulch, suggested that the primary benefit to mulch was increasing soil water storage in the soil profile before planting and theorized that canopy shading of the soil surface negated most of the positive effect of mulch during the growing season. However, mulch should increase the T/ET factor but the amount of increase will vary depending on time and frequency of precipitation events.
Growing Plants in Clumps
A relatively new strategy for dryland agriculture is growing grain sorghum or corn plants in clumps rather than equally spaced in rows (Bandaru et al., 2006; Kapanigowda et al., 2010: Krishnareddy et al., 2010). This strategy attempts to lessen the negative effects associated with reducing plant density, or increasing distance between rows like in skip-row configurations, but still capture some of the benefits that those strategies seek. The clumps usually consist of three or four plants in a clump and with clumps spaced in a row to achieve same population density as in evenly spaced plant environments. The clumps in adjacent rows are off-set so the configuration is essentially checkered. The hypothesis is that plants close together will reduce formation of tillers and improve the microclimate to reduce the VPD. Also, the clumps are close enough to one another that the plant available soil water would be depleted to approximately the same extent during the season as evenly spaced plants. Theoretically, the ET would not be reduced and the TR would be impacted positively. The downside is that the T/ET factor can potentially be negatively impacted because of less shading of the soil surface compared to evenly spaced plants. Combining this strategy with the use of mulch could potentially increase all of the factors shown in Table 1.
Forage as an Alternative to Grain
A careful analysis of Eq.  and Table 1 suggests that growing forage as an alternative to grain should be strongly considered, particularly when water is severely constrained. It is well established that the production of biomass is directly linked to T. For forage production, Eq.  remains the same except the HI factor becomes essentially 1 and is constant regardless of amount of ET, and GY is replaced with forage yield (FY). With forage production, the issue of plant population is greatly diminished, and a higher plant density increases the T/ET factor because of increased soil shading, and the TR is potentially increased because of an improved microclimate associated with plants being closer together to reduce the effects of wind, radiation, temperature, and low humidity that increase VPD. Unlike strategies for grain production that in every case discussed resulted in having a negative effect on at least one of the factors listed in Table 1, forage production tends to impact all of the factors positively. The level of anticipated ET for a cropping season when forage would likely be the best strategy will be influenced by factors other than those listed in Table 1 such as grain and forage prices, but using processes such as those presented in Eq.  and Table 1 could improve the management of water in dryland systems.
Green Water Amounts Increase Disproportionally with Increasing Annual Precipitation
Improving precipitation capture as green water becomes increasingly more beneficial as one goes from regions of low annual precipitation to high precipitation regions. The discussion above focused entirely on dryland areas with relatively low average annual precipitation. Even though the aridity varies from year to year, the area has an average aridity level. The factor values shown in Table 1 represent varying amounts of ET with an average annual precipitation of 500 mm and growing crops either completely dryland or with supplemental irrigation. In Table 2, hypothetical factor values are shown for areas of increasing annual precipitation and corn production without irrigation. The numerical values are assumed, not measured, but are thought to be realistic. They were derived from the conceptual model shown in Fig. 1, and on ET and HI values from the literature for corn growing regions in the United States. The important point is that the proportion of annual precipitation that is actually consumed as green water, and the productivity of that water, increases disproportionally with increasing amounts of precipitation. Every factor shown in Table 2 becomes more positive with increasing precipitation resulting in a synergistic effect. As average annual precipitation doubles from 500 to 1000 mm in this hypothetical example, the yield of corn grain increases almost fourfold. Therefore, even though there is theoretically a lot of water in dryland areas that can be captured and used as green water, the reality is that the constraints are difficult to overcome, particularly for grain crops.
|TR, kg water/kg biomass||270||258||246||234||222||210|
|GY, kg ha–1 dry wt.||2,860||3,860||5,160||6,820||9,090||10,580|
Water Use Efficiency
Water use efficiency (WUE) is a commonly used term but can be somewhat ambiguous because it is not always used in the same manner. Perhaps the most widely used definition, and the one that is used in this article, is the amount of harvestable product produced for each unit of water consumed by the crop as ET. For grain crops, it is usually expressed as kilograms harvested grain per cubic meter of water. The water content of the grain is generally adjusted, or assumed, to be that for suitable storage. Examples are 15.5% for corn, 13.5 for grain sorghum, and 14 for wheat. In the simplest terms, for grain crops, WUE = kg harvested per cubic meter of water consumed by the crop by transpiration and for evaporation from the soil surface between seeding and harvesting of the crop. Rhoads and Bennett (1990) reported that WUE values for well-watered corn ranged from 1.2 kg m–3 for the Texas High Plains to as high as 2.2 kg m–3 for the southern Negev region of Israel. Zwart and Bastiaanssen (2004) reviewed 84 Lature sources and found globally measured values averaged 1.8 kg m–3 for corn and ranged from 1.1 to 2.7 kg m–3.
Grassini et al. (2011) summarized several studies from Nebraska to develop average, and upward bound, relationships between water use and grain production (Fig. 3). Using these relationships, we calculated WUE values for increasing levels of yield. At the highest yield levels, the value calculated from the best fit line through all the data points was about 1.7 kg m–3 and about 2.4 for the upper bound relationship that represented the highest efficiency achieved in any of the studies in Nebraska. These values are quite similar to those reported above by Zwart and Bastiaanssen (2004).
In Developing Cropping Systems section, we presented Eq.  as GY = ET × T/ET × 1/TR × HI. The equation can be easily modified to calculate WUE and understand where there are opportunities for increasing the WUE value, and what the limitations are. Since WUE = GY/ET, the equation can be changed to WUE = T/ET × 1000/TR × HI. The 1/TR in Eq.  becomes 1000/TR in the WUE equation because the WUE value is reported for cubic meters. Also, GY in Eq.  is dry weight, whereas it is 15.5% for calculating WUE for corn. The highest yield shown in Fig. 2 is 16 Mg ha–1. At this level, the HI would be high because water stress would be minimal so an assumed value of 0.55 would be reasonable. The T/ET values can vary significantly depending on several variables such as frequency and amount of rainfall, method of irrigation, and numerous other factors, but a value of 0.7 for a high yielding corn crop is reasonable. Making these assumptions, the TR value can be calculated using the equation WUE = T/ET × 1000/TR × HI, and using the assumed values for a yield level of 16 Mg ha–1, WUE (1.7) = T/ET (0.7) × 1000/TR x HI (0.55), so TR = 227 which indicates that 227 kg of water were required to produce 1 kg of dry matter. This is very close to the 220 value that Sinclair and Weiss (2010) stated was required for a 2 kPa T environment. For a WUE value of 2.4 kg m–3 that was the highest value achieved from the studies reported in Fig. 3, a combination such as 0.80 for T/ET, 0.58 for HI, and 193 for TR would be required. Extremely favorable conditions would be required to achieve these values as evidenced by the fact that a WUE value of 2.4 was only achieved once in the Nebraska studies. Recent studies in northwest China (Yi Liu et al., 2010; Ling-duo Bu et al., 2013) have reported WUE values of 3.5 or greater for corn. Although such values are theoretically possible based on the equation, a combination such as 0.90 for T/ET, 0.62 for HI, and 160 for TR would be necessary to achieve a WUE value of 3.5 kg m–3. Some of the treatments in the Chinese studies did use plastic film to cover the soil surface to minimize evaporation of soil water.
Under typical growing conditions, a T/ET value of 0.65, a HI of 0.5, and a TR of 220 seems reasonable. Such a combination would produce a WUE value of 1.48 kg m–3. The WUE value can only be increased by increasing the T/ET and HI factors, or by decreasing the TR value. The TR value is controlled primarily by the vapor pressure deficit of which the producer has little control. Changes in row spacing, plant population, and plant geometry have some potential for altering the microenvironment but benefits will likely be minimal. There is also limited potential for increasing the HI above 0.55 because this is near the genetic potential reported to be about 0.60 (Prihar and Stewart, 1990). Therefore, the greatest potential at present to increase WUE is to increase the T/ET factor. Reducing row width so that ground shading can be achieved more quickly, using mulch to cover the soil surface, and less frequent irrigation to reduce the time the soil surface is wetted are potential ways to increase the T/ET factor. In parts of China, plastic mulch is widely used and this has been a very effective practice for increasing the T/ET factor. In Gansu Province, there were 896,000 ha of maize in 2012 and 92% was produced using plastic mulch (Dr. Fan Tinglu, Gansu Academy Agricultural Science, Langzhou, personal communication, 2013). Although this practice has been widely adopted in parts of China, it has not been extensively used in other countries and is generally not considered feasible for large commercial farms. However, there is considerable potential for increasing the T/ET factor by the adoption of conservation agriculture that (i) minimizes soil disturbance and uses direct seeding; (ii) establishes permanent soil cover by plant residues and cover crops; and (iii) rotates crops.
There is potential for increasing capture and storage of precipitation in dryland areas, and for utilizing it more efficiently for growing crops. During the past few decades, major advances have been made in capturing and storing more water, but there have been fewer advances in using the water more efficiently. Because of the low and highly variable amounts and erratic distribution of precipitation, growing grain crops annually is marginal at best so a common practice is to grow one crop every 2 yr, two crops in 3 yr, or three crops in 4 yr, to store water in the soil to supplement the rainfall during the period when a crop is grown. The more intensive systems allow more crop production during the time when precipitation occurs and shorten the long summer fallow periods that lose great amounts of water by evaporation. The use of minimum-till and no-till systems has in many cases doubled soil water storage during the fallow periods and increased yields significantly and greatly reduced risk. Even so, the amount of total precipitation used as ET is in the range of 50 to 55%, and only 55 to 60% of the ET is used for T. Therefore, the potential for improvement is enormous, but challenging. A shift to forage crops as an alternative to grain crops appears the most promising, but additional research is needed on manipulating plants during the growing season and to use mulch more effectively.