Sustaining production and productivity of maize with stability under marginal environment like drought is a major challenge for present and coming future. Both, agronomic management and genotype improvement were exploited to improve yield under water deficit conditions.
Cultural practices such as time and method of sowing, planting density, and reduced tillage has been found to be beneficial in reducing drought stress. Date of sowing is crucial for maize production in drier areas. Where length of growing season in short due to duration of rainy season, early planting reduces risk of terminal drought at grain filling stage. Matching the date of planting with the rainfall distribution pattern in a way to avoid exposure to stress at crucial growth phases, i.e.- “Response farming”, is a best approach. However, this needs adequate information, on long-term basis, about rainfall distribution pattern of a particular area. Another management practice is to reduce maize plant population in order to maintain the amount of available water per plant above minimum. For example- in South Africa, relatively late maturing maize grown under an annual rainfall of 500-600mm is often sown at densities as low as 10,000 plants per hectare with a row-to-row distance up to 2 m. cultivars are selected for prolificacy (or tillering) so that in case of good rain they can be fully exploited (Magson,1997). Moisture management through reduced tillage practices can allow to plant at more optimal time along with moisture conservation. Another tillage management i.e.- post-harvest or winter plowing may contribute to earlier planting by reducing the time and energy requirement to prepare the land when the next rainy season begins (Waddingtons, 1995).
Increasing water availability for optimal physiological phenomenon and thereby better per se performance under water deficit condition is one of the major goal of agronomic management. This can be achieved through reducing water loss other than transpiration, through runoff, evaporation from soil and weed infestations. Many management options has been exploited for intercepting a larger proportion of precipitation and directing it towards utilization by crop, such as- tillage, water harvesting, mulching etc.
· Tillage : Tillage has been found useful as it facilitate entry of water deep into soil, better root growth for enabling them to capture stored moisture and reduce weed problem that compete with crop for water. Though, this option vary from soil-to-soil. However, in many semi-arid areas deeper plowing at suitable time can improve maize yield up to 25% by permitting deeper rooting (Willcocks, 1981; Ivy, 1987).
· Water harvesting : By controlling runoff or diverting it towards an area were it can be captured can contribute in yield improvement in under water defit conditions. However, water harvesting in the cropped area should be avoided because maize is highly susceptible to water logging conditions. Tied contour ridges system, diverting excess runoff to top of terrace systems etc. have been exploited successfully for water harvesting in sub-saharan Africa (Waddington, 1995).
· Mulching: In direr arid and semi-arid areas as much as 50% of total evapo-transpiration from a crop can be lost through evaporation from soil surface (Unger and Stewart, 1983). Therefore, mulching can play a crucial role in water conservation in such areas. Crop residue mulching has been found to closely associated with increased water capture and improve maize yield significantly in semi-arid western Mexico (Edmeades et al., 1998).
Though, crop productivity can be sustained, up to some extent, by manipulating different crop management practices in a way to avoid or reduce the impact of drought stress. However, development of improved maize genotypes capable of withstanding the stress condition at critical growth stages may be most reliable and affordable solution, suitable for poor and marginal farmers for drought prone maize areas. Therefore, breeding for drought tolerance in maize using modern tools and techniques needs to be approached.
6.2.1. Conventional vs. modern approach
Most maize breeders use the screening phase to select for yield potential, resistance to diseases and insect pests, and desirable grain and plant type. Only at the advanced testing stage, when relatively few genotypes remain, entries are evaluated as well under abiotic stress. At this stage, selection intensity is customarily low and progress in breeding for tolerance to abiotic stress is therefore poor.
There are several reasons for plant breeders' apprehensions about selecting under abiotic stress at earlier breeding stages:
Heritability and genetic variances for grain yield usually decrease under abiotic stress as yield levels fall. Differences between entries are often non-significant, and expected selection gains are less than under conditions where yields are high.
Because of the high genotype x environment interactions involved, stressed experiments often produce rankings that differ significantly from one experiment to another, making it difficult to identify the best germplasm.
Breeders expect that selection under high-yielding conditions will also increase grain yields under abiotic stress conditions.
In developing countries, farmers in high-yielding, high-input conditions are usually more attractive targets for the private seed sector than the 'average', (often) resource-poor farmer, and commercial sector breeders often ignore abiotic stress-tolerance for this reason. Public sector breeders are influenced by this viewpoint, even though their responsibilities and target environments usually include areas not served by the private sector.
Based on extensive research at CIMMYT, it appears that many of these apprehensions can be overcome, and it is actually possible to make relatively rapid progress in breeding for improved yield under favorable and stress conditions by including extensive screening under abiotic stress conditions. The following arguments underpin this concept:
If the target environment is commonly affected by abiotic stresses, then the fact that selection gains in an unstressed target environment are higher than under stress is of little or no help in improving yield in the target environment. If abiotic stresses are the major feature of the target area, the breeder should aim at improving yields for these conditions. Generally, maize breeding methodologies in the tropics have been influenced strongly by experience from maize breeding in temperate areas. Maize in temperate environments is generally grown under relatively stress-free conditions and on-farm yields approach those obtained on experiment stations. On the contrary, in tropical environments maize is frequently stressed and on-farm yields fall far below those obtained on breeding stations. Thus, in the tropics selection under high-yielding conditions may not be the best way to increase yields in farmers' fields.
No breeder would expect to improve disease resistance in maize by selecting in a virtually disease-free environment, yet breeders routinely expect to increase the tolerance of their varieties to drought by selecting mainly in high potential environments. This strategy might work, if yields under stressed and favorable conditions were determined by the same plant characteristics. However, as stress levels rise from lack of water, different plant characteristics affect yield and genotype-by- stress interactions become significant.
When genotypes are selected under favorable conditions, much useful genetic variation for stress tolerance may be lost. This variation cannot be replaced simply by multi location testing that exposes a few varieties to stressed conditions at latter stages in breeding. In contrast, a method for reliably detecting abiotic stress tolerance in a large collection of maize genotypes will increase chances of making significant progress in breeding for this trait.
Over the past 20 years, researchers at CIMMYT have improved maize germplasm for drought tolerance using an approach that is probably unique. Large populations were screened under carefully managed drought so that genetic variation for tolerance was revealed to the greatest extent possible.
The key to breeding for drought tolerance is to manage stress. In the case of drought this is done by conducting experiments partly or entirely in the dry season and managing the stress through irrigation. The objective of such experiment is to measure the genotypic variability for drought tolerance. The objective of such experiments is not to simulate a farmer's field, but to simulate a clearly defined stress that is relevant in farmers' fields. If we select under random stresses, a combination of stresses or just 'low yields', we may well select for a different stress tolerance mechanism each time, and will likely not make much breeding progress. Timing, intensity, and uniformity of the stress are factors to consider in stress management.
Timing should be such that the growth stages targeted are susceptible to the stress, have a high probability of being affected by that stress in the target environment, and involve tolerance-related traits that can be modified through breeding.
Stress intensity should be severe enough so that traits become important for yield distinct from those which affect yield under non-stressed conditions.
Uniformity: If the stress is uniform over space and time, genetic differences will be easier to observe and progress will be greater.
i) Flowering stage:
Irrigation is designed so that drought at flowering is severe enough to delay silking and cause ear abortion. The components that determine yield are number of ears and kernels per plant. Ideally, ASI should average about 4 to 8 days, ears per plant should average about 0.3 to 0.7, and yields should average around 1-2 t/ha (about 15-20% of well-watered yields). If drought stress at flowering is not severe enough, the accuracy (heritability and genetic variance) with which ASI and ears per plant can be measured decreases (Bolanos and Edmeades, 1996).
ii) Grain filling:
Irrigation is designed so that drought develops directly after flowering and accelerates leaf senescence. The yield component affected in this case is kernel weight, because photosynthesis during grain filling is reduced. Ideally, ASI should not be affected much by this type of a stress, but yields should be reduced to 50% of yield potential at least.
The following considerations should be made before sowing drought stress experiments.
Because it is difficult to divide a field into different stress regimes, whole fields (drought blocks) should be managed with one stress level only, and fields and/or stress levels should be far enough apart to prevent border effects.
Irrigation has to be stopped earlier for early maturing germplasm than for late maturing germplasm, to obtain a similar drought stress intensity at a given growth stage. Genotypes should therefore be grouped in experiments of similar maturity.
Experiments should be grouped so that flowering time coincides for all experiments being subjected to a single stress treatment. This can be done either by grouping experiments of different maturity in different fields and designing specific irrigation schedules for each field, or by planting early maturing experiments later so that flowering coincides with that of late maturing germplasm.
Before the period when drought stress is desired, irrigation intervals and other agronomic measures are designed so that the crop has optimal conditions for establishment and growth. Most important question when managing drought stress is: When should irrigation be stopped so that drought stress is sufficiently intense at the critical growth stage (e.g., at flowering)?
(a) Using a crop water balance module:
A crop water balance is usually used to calculate irrigation intervals for crops that should be irrigated when they show first symptoms of drought stress. It takes about twice this interval to obtain a stress level that reduces maize yields to the extent needed for drought experiments. To calculate a crop water balance for severe drought stress at flowering estimations is desired, as follows:
i) Average anthesis date {AD):
Be aware that the temperature during the trial determines crop development. If the temperature between planting and flowering in drought trials is higher than usual main season temperature, it will take less time for the crop to reach flowering. Calculating the temperature sum (heat units) between planting and flowering can help to determine anthesis date. The temperature sum between planting and flowering is constant for a certain maturity group, provided photoperiod effects can be disregarded. It can be calculated as (Kiniry 1991):
Temperature sum = ∑{(Tmax + Tmin)/2-8}
where Tmax = daily maximum temperature max
if Tmax > 34 then Tmax = 34- 2.6*(Tmax -34)
if Tmax > 44 then Tmax = 34- 2.6*(44-34) = 8
Tmin = daily minimum temperature
if Tmin < 8 then Tmin = 8
∑ = make the sum for the period planting to anthesis
ii) Daily water consumption:
There are various methods for estimating daily water consumption of crops (Doorenbos et al. 1984). They differ in the type of weather data they use. Most experiment stations measure pan evaporation (water evaporation from an open water surface). Daily water consumption (DWC) of maize can be calculated as:
DWC = PE * Kp * Kc
where PE = pan evaporation, as measured in a Class A pan in a standard meteorological station installation
Kp = pan coefficient (determine from Table 2)
Kc = crop coefficient
Maize has a Kc of about 0.25 at germination, 0.50 at 6-leaf stage, 1.10 at flowering stage and 0.40 near maturity. An average Kc of about 0.80 for calculating daily water consumption near flowering will be appropriate. Note that the crop coefficient is influenced by leaf area, stomata opening, and the relative importance of transpiration (from plants) and evaporation (from the soil). Inbred materials have a smaller crop coefficient than full vigor materials, because they have less leaf area. Stressed plants have a smaller crop coefficient than unstressed plants, because they close their stomata. Plants that were stressed during early growth stages have a smaller crop coefficient at later growth stages because their leaf area is reduced.
iii) Soil texture: from Table 3.
iv) Plant-available water (PAW, in mm/10cm depth): from Table 4.
v) Rooting depth (RD) of maize:
The rooting depth of maize is about 10 cm at germination, 30 cm at the 6-leaf stage, and 70-100 cm at flowering, depending on how porous or compacted the soil. Inbred materials generally have fewer roots and shallow root development than full vigor materials.
vi) Amount of water available (W) to the crop until first stress symptoms are visible. Maize shows first symptoms of stress when 55 to 65% of [PAW * RD] is used,
i.e. W = RD/10 * PAW * 0.65
vii) Calculate the time (T1 until maize shows first symptoms of drought stress:
T1 =WA/DWC
viii) Calculate the time of the last irrigation (T2)
T =AD-2*T1
Note: it takes about twice as long to obtain severe drought stress (desirable in a drought experiment) as first visible drought symptoms. Therefore, T1 is multiplied by 2.
Calculation of a crop water balance can allow quantitative insights on factors that affect the length of time after the last irrigation or rainfall until first stress symptoms. This operation will also point up the limits of any predictions: factors that are difficult to determine-such as effective rooting depth-can modify the water balance considerably.
Table 2. Pan coefficient, Kp, for a Class A pan in a standard meteorological station installation. Distance refers to the ‘fetch’, or the distance wind passes over the crop or fallow before it reaches the pan.
Pan placed in short green cropped area Pan placed in dry fallow area
|
|
|
Wind
distance of Relative
humidity (%) Distance
of
Relative humidity (%)
(km/day) crop (m) < 40 40-70 > 70 fallow (m) < 40 40-70 > 70
175 1 0.55 0.65 0.75 1 0.70 0.80 0.85
Light 10 0.65 0.75 0.85 10 0.60 0.70 0.80
100 0.70 0.80 0.85 100 0.55 0.65 0.75
1000 0.75 0.85 0.85 1000 0.50 0.60 0.70
175-425 1 0.50 0.60 0.65 1 0.65 0.75 0.80
Moderate 10 0.60 0.70 0.75 10 0.55 0.65 0.70
100 0.65 0.75 0.80 100 0.50 0.60 0.65
1000 0.70 0.80 0.80 1000 0.45 0.55 0.60
425-700 1 0.45 0.50 0.60 1 0.60 0.65 0.70
Strong 10 0.55 0.60 0.65 10 0.50 0.55 0.65
100 0.60 0.65 0.70 100 0.45 0.50 0.60
1000 0.65 0.70 0.75 1000 0.40 0.45 0.55
> 700 1 0.40 0.45 0.50 1 0.50 0.60 0.65
Very 10 0.45 0.55 0.60 10 0.45 0.50 0.55
strong 100 0.50 0.60 0.65 100 0.40 0.45 0.50
1000 0.55 0.60 0.65 1000 0.35 0.40 0.45
Table 3. Determining soil texture: 1. From a ball of about 3 cm diameter from fine soil; 2. Drip water onto the soil until starts sticking to the hand.
Type Description
Sand The soil remains loose. You cannot form a ball.
Sand loam The soil can be rolled into a short thick cylinder.
Loam The soil can be rolled in a 15 cm cylinder that breaks when bent.
Clay loam As loam, but the soil can be bent into a U.
Light clay As loam, but the soil can be bent into a circle that shows cracks.
Heavy clay As loam, but the soil can be bent into a circle without cracks.
Table 4. Characteristics of various soils.
Field capacity Permanent Plant-available Bulk density
wilting point water
(Vol %) (Vol %) (mm/10 cm depth) (g/cm3)
Sand 15 7 8 (6 -10) 1.65
Sandy loam 21 9 12 (9 - 15) 1.50
Loam 31 14 17 (14 - 20) 1.40
Clay loam 36 17 19 (16 - 22) 1.35
Light clay 40 19 21 (18 - 23) 1.30
Heavy clay 44 21 23 (20 - 25) 1.25
(b) Using a crop simulation model:
Crop simulation models provide a more sophisticated estimate of the crop water balance than the method described above. If they are to be used for managing irrigation in drought experiments, they still rely on an accurate calibration based on site- and crop-specific data, especially with respect to water conditions at various depths in the soil at the start of the simulation period. Note that the timing for stopping irrigation can never be precisely determined. This is because evaporation between the time when irrigation stops and the growth stage when stress should occur is a prediction based on previous years' weather data. The actual conditions for a given season may differ significantly from the long-term average.
Breeders usually underestimate the time it takes to develop severe stress in a maize crop, because they take usual irrigation intervals or the time it takes to first visible drought symptoms as guidelines for terminating irrigation. However, it will take considerably longer than the time to first visible drought symptoms to produce a severely drought stressed maize crop.
(c) Using an experiment:
An experiment where seed of a particular maize genotype is sown at different dates but irrigated at the same time can help to improve drought stress management in following years. Plant 10 rows of maize 5 times at 5-day intervals (that is, a total of 50 rows of maize in 5 sections with 5 different planting dates). Irrigate all on the same day, and apply the last irrigation before flowering when you predict that it will result in ideal drought stress intensity for the 2nd planting date. The first rows sown should exhibit less stress than maize from the 2nd planting date, because the last irrigation is applied relatively later in crop development. By the same token, all rows planted from the 3rd date on should experience greater stress. Determine the planting date for which stress intensity was ideal and calculate the time between the last irrigation before flowering and flowering. Use this time period for scheduling the last irrigation for stress experiments in coming years.
(d) Using two different drought stress levels:
The problem of estimating the time when irrigation should be stopped may as well be solved by managing two drought stress levels, each in a different field where sets of the same trials are planted. The two stress levels create selection environments that are representative of two different, important types of drought stress: flowering stress and grain-filling stress. In both cases, optimal irrigation at regular intervals is applied for germination and crop establishment, until the last irrigation before the stress period.
Severe stress: Irrigation is timed so that severe drought stress is predicted for flowering. An additional irrigation is applied about 14 days after the end of male flowering to ensure that the small amount of grain formed will fill adequately.
Intermediate stress: This treatment receives one irrigation more before flowering than the severe stress treatment, but no further irrigation after flowering or during grain filling. This stress regime targets grain-filling.
If both experiments are planted at the same time, they provide the following management options:
If evapotranspiration and crop development proceed as predicted, the severe stress treatment results in flowering drought stress and the intermediate treatment results in severe grain filling stress.
If evapotranspiration is greater or crop development slower than expected, the intermediate stress treatment will result in drought stress at flowering. The severe stress treatment can be rescued with irrigation near flowering when stress becomes too severe; it thus becomes a grain-filling stress treatment.
If evapotranspiration is much lower or crop development faster than predicted, the severe and intermediate stress treatment will result in two levels of grain filling stress, and there will be no treatment with drought stress at flowering.
(e) Application of irrigation after flowering stress:
After drought stress at flowering, an additional irrigation may be necessary to ensure grain filling. The following guidelines can help.
If the average ASI of the drought stress block is less than 3 days, do not apply any further irrigation after flowering.
If the average ASI of the drought stress block is between 3 to 5 days, apply one irrigation two weeks after male flowering is completed.
If the average ASI of the drought stress block is between 5 to 8 days, apply one week after male flowering is completed.
If the average ASI of the drought stress block is estimated at more than 8 days, apply irrigation when 80-100% of the plots have completed male flowering.
Note: Irrigation should be applied only before silking starts or after male flowering is complete; not curing flowering, when the susceptibility of maize changes rapidly.
Variation in drought stress intensity comes from two sources: variation in soil characteristics and variation in the application of irrigation. Variation in soil characteristics is almost impossible to correct, unless you move to another field. Variation in the application of irrigation can and should be corrected.
Normal experimental precision is required for irrigation and crop management until the last irrigation before the stress period is due to begin. It is vitally important that the last irrigation before the stress period begins is applied as uniformly as possible. To achieve this:
Choose a field that is as level as possible for drought experiments. Try to avoid old river beds, or areas where soil depth or texture is known to vary over short distances.
If using sprinkler irrigation, apply it when there is no wind. Note that wind usually varies over the day and you should choose a time of the day when there is little or no wind.
Make sure that risers are high enough so that water jets do not damage plants near the sprinklers.
Make sure beforehand that the irrigation system is set up properly, that pipe connections are sealed and that sprinkler heads are clean and work properly; replace sprinkler heads that do not work properly; if necessary exchange nozzles.
When the irrigation system is turned on, remove the end cap of the main pipe for a brief period to flush out dirt that may clog sprinkler heads.
Apply irrigation so that, as a minimum, field capacity in all parts of the field is reached; where more water is applied than necessary for reaching field capacity, the water will drain, but the whole field will be at field capacity for one or two days after irrigation, thus leveling differences that might have resulted from non-uniform irrigation.
Use carefully leveled catch cans to measure the amount of irrigation at places in the field where irrigation is expected to be the lowest. If placed systematically in the field, the volume of water collected in the catch cans can be used to adjust the sprinklers for uniformity.
Increased uniformity of water application before stress onset will translate into more uniform drought stress, more uniform plant performance, and increased breeding progress.
Once irrigation is stopped, drought stress increases over time. Later maturing germplasm will be more stressed and therefore lower yielding than early maturing germplasm. A systematic increase in stress intensity with time and a systematic yield decrease with later anthesis dates can be accounted for in data analyses; not so for non-systematic changes in stress intensity, such as an irrigation application or rainfall event during flowering.
Differences between genotypes are usually smaller under stress conditions, and superior genotypes are therefore more difficult to detect (Rosielle and Hamblin 1981). As a consequence, heritability of grain yield decreases. Breeders have traditionally coped with this problem by evaluating genotypes mainly under high yielding conditions, where differences between genotypes, and therefore heritability, for grain yield are larger, ignoring the fact that superior genotypes under high yielding conditions are not necessarily high yielding under stress conditions. Given the need to select under stress, it is a challenge to keep the heritability for yield as high as possible. Since genetic variance under stress cannot be changed, every measure should be used to keep experimental error low.
Experimental error variance can be reduced by:
Using uniform fields and managing them uniformly (discussed above). Increasing the number of replicates in an experiment.
Ensuring that the experiment is bordered uniformly and adequately towards alleys in the field and field borders.
Using improved statistical designs that partly control the variation within a replicate.
Choosing an optimal field layout that reduces the variation within replicates.
Using statistical analysis tools that consider spatial variation.
Note that these principles apply for both stressed and non-stressed experiments, but that the relative importance of reducing error variance is larger under stress.
i) Increasing the number of replicates
Increasing the number of replicates does not necessarily increase the efficiency of a breeding program, because more resources (land, seed, etc.) have to be used or fewer genotypes can be evaluated. Increasing the number of replicates and decreasing the plot size at the same time may increase selection efficiency as long as the total number of plants sampled per genotype does not decrease much. The number of plants sampled for a genotype using smaller plots but more replicates usually does decrease, because plants next to alleys need to be removed in stress experiments where border effects can be extremely large. Decreasing plot size results in larger border effects from the neighboring plots. However, the trade-off of increased border effects between neighboring plots is apparently not as large as the relative gain from increasing the number of replicates (Castleberry 1986). Under stress, field variation is often small-to medium-scale and a smaller plot size allows an entire experiment, replicate, or incomplete block to be fitted into a more uniform area.
Increasing the number of replicates and decreasing plot size is advantageous in advanced yield trials that usually have fewer entries but large plots. For example, a yield trial that is planted with four-row plots and three replicates under non-stress conditions could be planted with six replicates and two-row plots under drought, and more precise information on genotype performance would be obtained, in most cases.
ii) Improved statistical designs
a) Unreplicated experiments
Improved designs are available both for replicated and unreplicated experiments. In the case of unreplicated experiments, checks planted systematically over the field allow the breeder to discriminate between genotype performance and field variation (augmented designs). In general, some 20% of the plot area is planted to one or several check entries. Example: An augmented design may allow a breeder to compare 496 genotypes and 124 check plots in 62 sub-blocks of 10, where 8 plots are genotypes and 2 plots are checks. There are 4 check genotypes, each repeated 31 times. In this same area the breeder could evaluate only 310 genotypes in two replicates. Augmented designs are very valuable during prescreening involving many genotypes.
b) Replicated experiments
In the case of replicated experiments, improved statistical designs are available that provide better control of within-replicate variation, and hence experimental error, than randomized complete block designs (RCBD). Improved designs adjust genotype means for variation occurring within a replicate, with the result that genotype means are no longer equal to the arithmetic average of individual plot data. Improved designs provide a better estimate of the true genotype means. Computer software is available that creates and analyzes improved designs in a routine manner. By using improved, replicated statistical designs instead of an RCBD, a breeder-can increase breeding progress without additional costs.
We discuss here only two of several improved designs available: lattice designs and covariate analysis.
Lattice designs: These designs group genotypes in incomplete blocks within each replicate and adjust genotype means for incomplete block effects; i.e., soil variation among incomplete blocks within a replicate. Compared to other lattice designs, alpha lattice designs (an unbalanced type of lattice design) pose very few restrictions on numbers of treatments, replicates, incomplete blocks, or spatial layout. Scientist has to determine block size. When soil variation is low, lattice designs with an incomplete block size equal to or slightly smaller than the square root of the treatment number are the most efficient. For example, with 240 treatments, a design of 16 blocks with 15 plots per block would be appropriate when soil variation is low. When soil variation is high, incomplete lattices with a smaller block size are more effective. In the case of 240 treatments, 24 blocks with 10 plots per block or 30 blocks with 8 plots per block would be suitable when soil variation is high.
Covariate analysis: Because much of the variation in stress trials is due to inherent differences in soil characteristics, variation consistently shows up at the same place within the field over several seasons. Plot yields of a single maize check sown among trial genotypes at a frequency from 1:1 to 1:5 and measured in only one season could be used effectively as a covariate in the analysis of variance for several following seasons. Lafitte et al. (1997) showed that such a covariate adjustment can reduce error variance to a greater extent than lattice designs, when the number of genotypes evaluated is low and plot size is large (e.g., in advanced yield trials).
Several new designs under development hold even greater promise for controlling error variance. Among these are row and column designs and other types of spatial adjustments in two dimensions. All require the "geographical" (row, column) coordinates of each plot to be entered along with the data.
iii) Field layout
Experiments should be laid out in the field so that:
Replicates and incomplete blocks are as compact (square) as possible.
Replicates are arranged in a manner that they lay at right angles to trends in stress intensity.
Entire experiments, replicates or incomplete blocks are placed within areas of uniform stress.
iv) Border effects from alleys
Plants grown next to an alley have greater access to the factor (water, N) that is causing stress in the rest of the plot since their roots penetrate the soil of the alley where there is no competition for the available N and water. They are therefore less stressed. In a severely stressed drought trial, the plant next to the alley can produce up to half of the entire plot yield. Plants bordering alleys need to be removed before harvest. They are far less stressed, and so may disproportionately affect plot yield while not representing the mean performance of that genotype accurately.
Breeders' primary interest is in grain yield. So why do we need other, secondary traits to assess drought in maize? In a drought-breeding program, secondary traits are valuable for the following reasons.
They can improve the precision with which drought or low N tolerant genotypes are identified, compared to measuring only grain yield under drought or low N stress. This is because under stress the heritability of grain yield usually decreases, whereas the heritability of some secondary traits remains high, while at the same time the genetic correlation between grain yield and those traits increases sharply (Bänziger and Lafitte 1997a; Bolanos and Edmeades 1996).
They can demonstrate the degree to which a crop was stressed by drought or low N. If observed before or at flowering, they can be used for selecting desirable crossing parents. Under this scenario, a crossing block is planted separately and usually slightly delayed to the drought or low N trial and data from the trial are used to select the crossing parents.
If observed before maturity, they can be used for preliminary selection when tum- around time between seasons is short.
Most breeders use secondary traits. A breeder who is interested in disease resistance is not just measuring yield under disease pressure, but assesses disease incidence as well. Many breeders consciously or unconsciously use an ideotype approach (an ideotype can be defined as the target plant the breeder has in mind when selecting). Selection for stay-green, upright leaves, small tassels, dark green leaves, etc., is common, although the additional improvement in genetic progress such traits provide over selecting for grain yield alone has rarely been measured objectively.
1. Deciding the value of secondary traits:
Many reviews on plant characteristics related to drought or low N tolerance have been written (e.g. Hsiao 1973; Ludlow and Muchow 1990; Turner 1986), but few secondary traits have been used in breeding programs and even fewer have proven to contribute to the improvement of drought or low N tolerance in maize.
Edmeades et al. (1998) established that an ideal secondary trait should be:
Genetically associated with grain yield under stress.
Highly heritable
Genetically variable
Cheap and fast to measure.
Stable within the measurement period.
Not associated with a yield penalty under unstressed conditions.
Observed at or before flowering, so that undesirable parents are not crossed.
A reliable estimator of yield potential before final harvest.
Many recommendations on the use of secondary traits have been made to breeders, base on phenotypic correlations between such traits and grain yield. Unfortunately, many such correlations have been calculated from few varieties where outlying values may greatly affect the sign and magnitude of the correlation. In addition, for a breeder it is not sufficient to know that a secondary trait is related to drought tolerance. Rather, it is important to know that breeding progress using grain yield and a given secondary trait in selection is greater than progress using grain yield alone. Thus, not only must secondary traits be identified, but also their value in breeding must be proven. This can be accomplished using:
Analyses of genetic correlations and heritability among progenies of a single population.
Selection indices (Fukai and Cooper 1995).
Divergent selection that creates synthetics or near-isogenic lines that have a similar genetic background but differ for a single, selected trait. Correlated response of grain yield can then be measured.
Analyzing physiological and morphological changes in varieties that have been consistently selected for performance under drought or low N stress.
Simulation models.
2. Important secondary traits for drought tolerance
CIMMYT physiologists have evaluated many secondary traits for their value in a drought breeding program. The following recommendations on the use of secondary traits are based on the results (Banziger and Lafitte 1997a; Bolanos and Edmeades 1993a; 1993b; Bolanos et al. 1993; Edmeades et al. 1993; Lafitte and Edmeades 1994a; 1994b; 1994c).
i) Grain yield:
Heritability: medium under grain filling stress, medium to low under flowering stress.
Relationship with grain yield: high.
Selection: for increased grain yield.
Stress type: to be measured under flowering or grain filling drought stress.
Measurement: shelled, adjusted for grain moisture.
Remarks: Shelling percentage varies considerably under drought. Grain weight, not ear eight, should be used for calculating grain yield.
ii) Ears per plant:
Heritability: high and increasing with stress intensity.
Relationship with grain yield: high under flowering stress.
Selection: for more ears per plant (i.e., less barrenness).
Stress type: to be measured under flowering drought stress; heritability and genetic variance is largest when flowering stress is intense enough so that ears per plant average 0.3 to 0.7 across the entire experiment.
Measurement: count the number of ears with at least one fully developed grain and divide by the number of harvested plants.
iii) Anthesis-silking interval (ASI):
Heritability: medium, maintaining a reasonably high level under severe flowering stress.
Relationship with grain yield: high under flowering stress.
Selection is for a reduced or even negative ASI.
Stress type: to be measured under flowering drought stress; heritability and genetic variance is the largest when flowering stress is intense enough so that ASI averages 4 to 5 days across the entire experiment-
Measurement: determine the number of days from sowing until 50% of the plants have extruded anthers (anthesis date, AD), and the number of days from sowing until 50% of the plants show silks (silking date, SD); calculate: ASI = SD - AD.
iv) Leaf senescence:
Heritability: medium.
Relationship with grain yield: medium under grain-filling stress. . Selection: for delayed leaf senescence (stay-green).
Stress type: grain filling stress.
Measurement: score on a scale from 0 to 10, dividing the percentage of estimated total leaf area that is dead by 10.
1 = 10% dead leaf area 6 = 60% dead leaf area
2 = 20% dead leaf area 7 = 70% dead leaf area
3 = 30% dead leaf area 8 = 800;0 dead leaf area
4 = 40% dead leaf area 9 = 90% dead leaf area
5 = 50% dead leaf area 10 = 100% dead leaf area
Remarks: Leaf senescence should be scored on 2-3 occasions 7-10 days apart during the latter part of grain filling.
v) Tassel size:
Heritability: medium to high.
Relationship with grain yield: medium under flowering stress.
Selection: for a smaller tassel with fewer branches.
Stress type: this is the only trait that can be measured under well-watered conditions but is indicative of drought tolerance at flowering stage
Measurement: score on a scale from 1 (few branches, small tassel) to 5 (many branches, large tassel).
Remarks: advisable only with lines that have an inbreeding degree of at least S1; more difficult to determine with full vigor material. Two independent scores are recommended.
vi) Leaf rolling:
Heritability: medium to high.
Relationship with grain yield: medium to low.
Selection: for unrolled leaves.
Stress type: flowering stress.
Measurement: score plots on a scale from 1 to 5.
1 = unrolled, turgid 4 = rolled leaf rim covers part of leaf blade
2= leaf rim starts to roll 5 = leaf is rolled like an onion
3 = leaf has a the shape of a V
Remarks: to be measured before flowering when leaves are still more upright; leaves are less likely to roll after flowering when they become more lax and thicker. Two to three scores are recommended.
Many other secondary traits for drought tolerance were evaluated at CIMMYT but proved to be of low heritability, among them: leaf and stem elongation rate, canopy temperature, leaf photo-oxidation, leaf chlorophyll concentration, predawn leaf water potential, and seedling survival under drought.
Other traits evaluated by CIMMYT were heritable, but proved to have no relationship with grain yield under drought: osmotic adjustment, leaf erectness.
Selection indices: There are several good reviews on the theory and use of selection indices (Baker 1986; Lin 1978). A selection index summarizes the worth of genotype by making use of information from different traits such as grain yield, ASI, senescence, etc. To add together traits measured in different units, the phenotypic values, Pc are usually standardized, as:
Pi = (xij- mi)si
where mi and si are the mean and standard deviation of trait i in a population, and xij is the value of the trait i measured on genotype j. A selection index I in its simplest form can then be written as:
I = b1P1 + b2P2 + ………bnPn
where Pi is the observed standardized value of the trait i and bi is the weight given to that trait in the selection index. Weights may be chosen based on the relative economic value of each trait or based on the relative value of each trait as an indicator of drought tolerance. The genotype with the largest value for I is the best genotype. When defining weights, one has to be careful that they have the right sign: positive where larger values are desired (e.g. grain yield), negative where lower values are desired (e.g. lodging, ASI). CIMMYT has developed software that calculates such a selection index (ALPHA) using MSTAT data files. AGROBASE® is another program that can calculate selection indices. These programs standardize the data for the traits included in selection. Breeder assigns weights to these traits, based on their variance, their correlation with other traits, the significance of the genotype term in their ANOVA (or their heritability), and their relative importance in contribution to drought tolerance. Traits that are genetically variable and that are known to be relatively more important for assessing drought or low N tolerance are weighted more strongly. The program then calculates the index as a single measure of the drought or low N tolerance of that genotype.
Typical weights allocated in a drought breeding program are:
Grain yield weight = 5 sign = + (increased grain yields)
Ears per plant 3 + (increased no. of ears per plants)
ASI 2 - (decreased ASI)
Leaf senescence 2 - (decreased leaf senescence)
Tassel size 2 - (decreased tassel size)
Leaf rolling 1 - (decreased leaf rolling)
Anthesis date and plant height are often included in the selection index so that the selected fraction of the population does not become either later, earlier, or taller than the original, unselected population. In drought experiments, there is a risk that earlier (drought escaping) genotypes will be selected. Comparing the mean of the selected fraction with the mean of all genotypes being evaluated can help to prevent undesirable changes in the germplasm. With the selection index introduced above, weights are chosen by the breeder. Optimal weights for each trait used in selection can actually be calculated, based on the phenotypic and genotypic covariance between the trait and grain yield. Predicted progress for grain yield can thereby be maximized (Lin 1978). The problem of this mathematically correct approach is that there is no software available that routinely does the calculations, which become especially complex with improved experimental designs.
Developing maize genotypes with tolerance to drought and N stress is complex. This is due to various factors, including the largely polygenic nature of the tolerance, the typically low frequency of tolerance alleles in most maize germplasm, and the difficulties commonly encountered in field evaluations. Important considerations in establishing a selection program for stress tolerance should be whether OPVs, hybrids or both types of products are needed, and what human, financial, and physical resources are available for experimental work. Additional important factors include the choice of germplasm, breeding methodology, selection environments, and essential data to collect.
1. Choice of germplasm
The selection of appropriate germplasm is critical, requiring careful consideration of all available information. A wrong choice cannot be corrected by using sound and efficient breeding methodologies. There are several approaches a breeder can take to develop drought tolerant germplasm.
· Improving locally adapted, elite germplasm for drought tolerance CIMMYT evaluated a wide range of landraces for drought tolerance, but there were few (about 3%) that compared favorably with elite, adapted germplasm for drought tolerance, and even fewer that compared favorably with elite, adapted germplasm under high yielding conditions. Additionally, we have found genetic variability for drought tolerance in all types of elite germplasm. Thus, improving adapted, elite germplasm for drought tolerance is probably nearly always better than working with landraces. A compromise approach would be to create synthetic populations from local landraces and improved, adapted varieties.
· Improving non-adapted but drought tolerant populations for local adaptation CIMMYT-Mexico and much more recently CIMMYT-Zimbabwe and CIMMYT-Kenya have developed germplasm with high levels of drought and/or low N tolerance. A breeder may want to use such germplasm or any other known source of drought tolerance. Such materials may not be well adapted in terms of disease resistance, maturity, etc., to the target environment. Screening for general adaptation by the breeder may be necessary, and only the best adapted among the introduced stress-tolerant materials should be used. Adaptation can usually be further improved through selection, and for many breeders it is easier to select for adaptation (suitable maturity, disease resistance, yield potential) than for drought or low N tolerance.
· Formation of new breeding germplasm through introgression: The third and more complex strategy is to develop a new breeding population by introgressing locally adapted germplasm with source drought or low N tolerant germplasm. This approach raises several issues: which source germplasm should be used? What proportion of local and source material is appropriate? How much recombination is necessary before the population is ready for intense selection? And how can molecular markers facilitate this process?
a) Source population(s)? A breeder should check first the information that is already available for the potential source germplasm:
General adaptation: lowland tropical, subtropical/mid-altitude, highland, temperate.
Grain color and grain texture.
Maturity: it is better to use heat units than calendar days for characterizing the maturity of a genotype.
Disease resistance.
Tolerance to abiotic stresses.
Heterotic pattern and response.
Combining ability: it takes a lot of effort to identify a line with good combining ability, and if possible only lines with proven combining ability should be introduced.
Other value-added traits.
A breeder should introduce only germplasm that matches farmers' preferences (grain color, texture, size) and the environment (disease resistance, maturity, need for acid soil tolerance etc.). If hybrids are the desired product, the heterotic response of an introduced line should be known so that the line can be used in the appropriate heterotic group.
b) Evaluation of source population(s) in the target environment
Once source germplasm is introduced, it should be evaluated in the target environment. Such an evaluation might be of:
Population or line per se: This should be done not only for gathering initial data on maturity, performance, disease resistance, etc., but also for increasing seed of the most promising germplasm.
Self of a population: to select directly the most highly adapted fraction of a population.
Population x local tester topcross combinations.
Line x local tester topcross combinations.
Diallels of local and exotic populations or lines.
The frequency of useful 'exotic' lines is typically low, and a breeder should invest only in germplasm that proves to have as many valuable traits as possible. Even if only germplasm that meets certain basic requirements such as maturity, grain color and texture, and disease resistance is introduced, a breeder should likely invest only in about 10-25% of all germplasm introduced.
c) What proportion of 'local' versus 'source' germplasm?
The proportion of 'source' germplasm will depend on the balance between the adaptation of the source germplasm and the drought tolerance level of the local population(s). If the local and source populations do not differ much in performance, the F2 population should be used as a base foundation population. If the difference in performance between parents is large, one to three backcrosses to the superior germplasm are appropriate.
d) How much recombination in 'local x source' populations?
It is beneficial to recombine 'local x source' populations for 2-3 generations with mild phenotypic selection for general adaptation following the introgression process, before intense inbreeding and selection are initiated (Geadelmann 1984). This breaks up linkages between desirable and undesirable genes. During recombination it is important to maintain an adequate effective population size (usually > 200 successful pollinations) to avoid genetic drift.
2. Breeding schemes
The extent to which selection for stress tolerance can be included in a breeding program depends on the breeding scheme used. A few strategies will be outlined below, though there are many others that can be employed. In any breeding program, however, the following need to be clearly defined:
The type of product to produced: OPV, hybrid, topcross hybrid, etc.
The most important characteristics of the product: maturity, grain characteristics, necessary stress tolerance and disease resistance, etc.
The strategy for developing and deploying the product.
All breeding programs use a step-wise selection procedure to identify the best performing progenies, given limited resources. First, a large number of progenies are evaluated with few replicates and at few sites (screening), then the more successful progenies, or their descendants, are evaluated with more replicates and at more sites (testing).
Screening: If drought tolerance is important breeding goals, evaluation under these stresses should be included in the screening phase, and the results should be combined with results obtained under unstressed conditions. Only the best germplasm (i.e., genotypes that possess stress tolerance and have good yields under optimum conditions) should be advanced to the testing phase.
Testing: The testing phase should include sites that are representative of conditions under which farmers grow maize. If drought and N stress are frequent in farmers’ fields, these conditions must be included. Both managed stress sites and randomly stressed sites (e.g., fanners' fields) during the testing phase should be included. These multi location trials need to be set up in a manner that considers the difficulties of stressed sites. One should not hesitate to increase the number of replicates and reduce the number of entries in such trials.
3. Population improvement schemes
Maize provides a wide array of options with respect to breeding methodologies. One choice is between intra-population and inter-population improvement methods. Within intra-population improvement methods, alternatives are:
Individual plant versus family selection.
Non-inbred families versus selfed progenies.
Per se performance versus test cross performance.
Within inter-population improvement methods, alternatives are:
Test crosses involving individuals versus families.
Half-sib versus full-sib test cross progenies.
Parental versus non-parental testers.
i) Individual plant selection schemes:
Two common procedures are simple mass selection and stratified mass selection (Gardner 1961). The procedures are not recommended for traits with relatively low heritability, such as grain yield under drought stress. They can be quite effective for highly heritable traits, such as selecting for disease resistance after introgressing an exotic stress-tolerant but disease susceptible genotype into well-adapted, disease resistant germplasm. One selection cycle can be completed every season. The experimental area can be stratified to reduce differential environmental effects. Tassels of undesirable plants can be eliminated before flowering to prevent pollen contaminating selected plants. A few cycles of mass selection may successfully eliminate the most susceptible fraction of the population before switching over to a family-based improvement method. This is a good option where human, financial, and physical resources are limited.
ii) Family-based selection: per se
Family-based selection methods result in greater gains when traits under selection are complex and of low heritability, but are more demanding in resources, record keeping and overall management. Progenies such as half-sib, full-sib, S1, S2, etc., are evaluated. Progress can be expected from anyone of these methods. The choice of method will be guided by the availability of off-season test sites, the ability to store remnant seed, the choice of product (variety, hybrids or both), desired traits, heritability, progeny seed quantities, the degree of control over pollination (both parents or only one parent), and the time required to complete a selection cycle.
· Half-sib improvement methods in which detasseled half-sibs (females) are pollinated with pollen from a bulk of all half-sibs (male) are commonly practiced using an unreplicated layout. Because the females are detasseled, ASI cannot be observed. For drought tolerance improvement, it is therefore more desirable to plant replicated trials of half-sib progenies and use remnant seed for recombination of selected families. Heritability of yield from half-sib progenies is lower than that for other types of progenies. However, where resources are limited, this may be the most cost-effective selection scheme.
· Full-sib family recurrent selection has been used extensively at CIMMYT to improve populations for drought tolerance. Replicated trial sets of full-sib progenies are evaluated under drought and well-watered conditions. Selection is made based on performance in all environments and considering other factors such as disease resistance, grain texture etc. A single cycle of selection requires at least two seasons to complete.
· Selfed progenies: When breeding procedures are based on selfed progenies, it takes longer to complete a cycle of selection, but this approach significantly improves tolerance to inbreeding over time. Formation of many S1 or S2 progenies is recommended. These can be prescreened in unreplicated observation nurseries under drought or low N and the selected fraction (perhaps only 30% of the original progenies) can be examined in more detail in replicated evaluations under, say, drought-stressed and well-watered conditions. Where prescreening in the main season is possible, disease susceptible progenies can be eliminated. Seed supplies may become limiting. This can be solved by using selected S2 ear bulk seed developed from each S1 progeny. To maintain population gains over longer periods, it is recommended that no fewer than 20-40 inbred progenies be recombined.
iii) Family based selection: test crosses
Here the test crosses of S2 or more inbred progenies are evaluated. The time required to complete a cycle of selection will thus depend on the materials that are test-crossed. Such schemes are useful when the emphasis is on combining ability, hybrid-oriented germplasm, and the integration of population and hybrid development. They can also be recommended where the need is to identify superior, early generation lines for further inbreeding or improving a population per se. Evaluation for stress tolerance can be emphasized during the formation and prescreening of selfed progenies as well as during test cross evaluation.
iv) Interpopulation improvement alternatives:
Two commonly used methods discussed here include reciprocal recurrent selection/half-sibs (RRS-HS) (Comstock et al. 1949) and reciprocal recurrent selection/full-sibs (RRS-FS) (Hallauer and Eberhart 1970; Hallauer 1973). Such schemes result in improved populations and superior OPV products as well as improving hybrid-oriented features of the two populations by increasing the level of heterosis between them. In addition these schemes allow the extraction of early- generation lines with good general combining ability (GCA), provide a sound basis for recycling early generation lines, identify superior testers on a continuous basis, and may identify new conventional and non-conventional hybrids. The schemes are not particularly suitable if the populations do not tolerate inbreeding, and the per se performance of lines and parent populations is ignored during selection. The original schemes also recommend evaluating So test crosses and recombining the parental S1 seeds of good performing plants. The modified schemes attempt test crosses (HS or FS) on S1 or S2 progenies and also permit selfed progeny evaluation for elimination of undesirable progenies. The RRS-FS schemes have an added advantage over RRS-HS, in that only 50% of resources are spent on test cross progeny evaluation trials. Both original and modified schemes permit selection for drought and low N at one or more stages during the selfed progeny regeneration and evaluation stages and during the evaluation of test cross progenies. These types of inter-population improvement schemes are not a necessary requirement for hybrid development, but from a long-term perspective they should generate useful early generation lines.
4. Development of drought lines and hybrids:
Types of hybrids emphasized (top-cross hybrids, double-, triple-, or single-cross hybrids) will depend on the stage of hybrid development and seed industry infrastructure, but an evolution from non-conventional to conventional and from multi-parent to two-parent hybrids seems logical. Populations improved for drought tolerance is useful sources for extracting drought tolerant inbred lines.
i) Line - hybrid correlations
The relationship between the performance of inbred lines and their hybrids is an important issue in hybrid development. Inbred line information indicative of hybrid performance is desirable to reduce hybrid trial evaluations. Betran et al. (1997) have reported correlations of around 0.4 between S3 per se and top-cross performance for some stress-related traits under drought, indicating that inbred lines insufficiently predict hybrid performance under drought. The practical implication of these findings is that drought evaluations of lines may be justified in early generations when numbers of progenies are yet very large, but the performance of advanced lines is best evaluated in hybrid combination.
ii) Choice of appropriate testers
The choice of testers is a critical yet difficult decision in hybrid development. Appropriate choices will have a strong effect on the outcome of a program designed to identify stress tolerant hybrids. Testers can be inbred, non-inbred, populations, synthetics, or hybrids. The choice involves a blend of theoretical and practical considerations. For example, should one use a broad or narrow genetic base tester, high or low yielding, one with high or low frequency for stress tolerance traits, good or poor GCA, one or several testers, and related or unrelated testers? Testers with a low gene frequency for the selection traits emphasized are theoretically attractive but are not commonly used, particularly regarding yield. For drought selections, they might be more practical, since many conventionally developed testers have never been selected for drought tolerance. A desirable tester must facilitate discrimination among genotypes for combining ability and desirable traits, simultaneously identify useful hybrid products for direct use, and be compatible with a practical maize breeding program (Vasal et al. 1997). For practical purposes, we recommend using the same testers for evaluating combining ability under drought stressed conditions, as they are used for evaluating combining ability under well-watered, well- fertilized conditions.
iii) Dosage effects
Preliminary results on the genetic control and modes of action for drought and low N tolerance show the following:
Lines are more affected by drought stress than hybrids.
As drought stress increases, so does the importance of general combining ability and additive genetic effects.
Dosage effects are important under drought but not under low N stress, suggesting the need for including drought tolerant parents on both sides of the hybrid to achieve acceptable drought tolerance, where stress is severe.
Line-hybrid correlations are generally lower under stress than non-stressed conditions.
iv) Line and hybrid improvement by introgression
Here we discuss strategies to improve line and hybrid performance. Our most important decision is to identify the source germplasm (lines, synthetics, populations, hybrids, etc.) most likely to contribute the most favorable genetic factors for drought tolerance to the elite recipient line or hybrid. Several methods of selection of the donor source have been described by Beck et al. (1997). The objective is to identify source germplasm with the highest frequency of favorable dominant alleles that are not present in an elite hybrid. A detailed discussion of these methods is complex and we refer readers to these sources for further information. A pragmatic approach that is taken by many breeders is first to evaluate source inbred lines for per se adaptation to the target environment. Again, many lines, maybe 60 to 80%, may be discarded in this step, and only lines with desirable disease resistance, maturity, and grain characteristics are then crossed to the local tester lines and their combining ability and heterotic response determined under managed stress and unstressed conditions. Introduced stress tolerant lines may be used directly as one of the parents in a hybrid that is then released. More often, however, stress-tolerant lines need to be introgressed into local germplasm. After the initial cross between source and recipient line, selection and or inbreeding can be initiated either immediately or after one or more recombinations or backcrosses. Repeated recombination before initiating inbreeding increases the chances of obtaining inbreds with stress tolerance and good agronomic performance. Backcrossing is advantageous if one parent has more loci with favorable alleles than the other, if the parents are diverse, or if the level of dominance is high (Dudley 1984).
6.3. Integrated approach
For improved production and productivity under drought-affected areas it is necessary to not to lose any option, whether improvement in plant through genetic approach or stress environment through crop management practices. Conventional wisdom holds that genetic improvement could make up about 15-25% of the yield gap between drought and optimal moisture conditions. Effective cop management practices using available water along with the genetic improvement could possibly close the gap by an additional 15-25% (Edmeades and Banziger, 1997). The remaining 50-70% can only be filled by adding water to the crop. Many observers suggest that in dry areas crop management interventions could play more important role than improved varieties in yield improvement (Waddington and heisey, 1997). This may be the case where moderate rather than severe dry spells are commonly encountered. Genetic improvement may be applicable over a much wider areas than any single crop management practices. Therefore, most of the time improved varieties often provide better alternative, affordable for poor farmers of such areas. As a matter of fact, an integrated approach considering all technological option available both genetic improvement and crop management practices could be more beneficial and effective, rather that either of them.
|
Author: P. Zaidi Date: 2002/07/23 Copyright © CIMMYT, Int. |
