HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Figures Only Full Text (PDF) Free Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in ISI Web of Science Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via ISI Web of Science (39) Citing Articles via Google Scholar Google Scholar Articles by Edmeades, G.O. Articles by Bänziger, M. Search for Related Content PubMed Articles by Edmeades, G.O. Articles by Bänziger, M. Agricola Articles by Edmeades, G.O. Articles by Bänziger, M.

CROP BREEDING, GENETICS & CYTOLOGY

Selection Improves Drought Tolerance in Tropical Maize Populations

I. Gains in Biomass, Grain Yield, and Harvest Index

^a Pioneer Hi-Bred International, Inc., 7431 Kaumualii Highway, P.O. Box 596, Kekaha, HI 96752 USA
^b CIMMYT, Apdo. Postal 6-641, 06600 Mexico D.F., Mexico
^c CSIRO Tropical Agriculture, 306 Carmody Rd., St. Lucia, QLD 4067, Australia
^d IRRI, P.O. Box 933, Manila, Philippines

edmeadgreg{at}hibred.com

	ABSTRACT

TOP ABSTRACT INTRODUCTION Materials and methods Results and discussion Conclusions REFERENCES

 
Drought is common in tropical environments, and selection for drought tolerance is one way of reducing the impacts of water deficit on crop yield. The primary objective of this study was to evaluate biomass, grain yield, and harvest index of maize (Zea mays L.) populations selected for drought tolerance. Three late-maturing tropical maize populations were subjected to three cycles of S₁ recurrent selection (`La Posta Sequía' and `Pool 26 Sequía') or eight cycles of full-sib recurrent selection (`Tuxpeño Sequía') for yield and traits indicative of drought tolerance during flowering and grain filling. Selection gains were assessed in five trials conducted under mid–late season drought and in five trials conducted under well-watered conditions. In water-stressed environments, with average yields of 1.0 to 4.5 Mg ha^-1, yield gains averaged 0.26 Mg ha^-1 (12.6%) cycle^-1 for La Posta Sequía and Pool 26 Sequía and 0.08 Mg ha^-1 (3.8%) cycle^-1 for Tuxpeño Sequía. In well-watered conditions, where mean yields ranged from 5.8 to 10.4 Mg ha^-1, corresponding gains per cycle were 0.12 Mg ha^-1 (1.5%) and 0.04 Mg ha^-1 (0.5%). Total biomass was unaffected by selection. Mean correlated responses to selection observed under drought were -0.11 Mg ha^-1 cycle^-1 in stem biomass and 0.025 cycle^-1 in harvest index (HI), and under well-watered conditions, 0.005 cycle^-1 in HI. Stem biomass and HI were negatively correlated under drought . Improved drought tolerance was attributed to simultaneous selection in well-watered environments and under carefully managed water stress at flowering, resulting in greater partitioning of biomass to the ear and increased harvest index.

Abbreviations: AD, days from sowing to 50% anthesis • ASI, anthesis-silking interval • HI, harvest index • IS, intermediate stress • SS, severe stress • WW, well watered

	INTRODUCTION

TOP ABSTRACT INTRODUCTION Materials and methods Results and discussion Conclusions REFERENCES

 
DROUGHT is thought to cause average annual yield losses in maize of about 17% per year in the tropics (Edmeades et al., 1992), but losses in individual seasons have approached 60% in regions such as southern Africa (Rosen and Scott, 1992). Periods of soil water deficit can occur at any time during the crop season, but maize is particularly sensitive to water stress in the period 1 wk before to 2 wk after flowering (Denmead and Shaw, 1960; Grant et al., 1989). Drought at flowering commonly results in barrenness. One of the main causes of this, though not the only one (Zinselmeier et al., 1995), is thought to be a reduction in the flux of assimilate to the developing ear below some threshold level necessary to sustain grain formation and growth (Westgate and Bassetti, 1990; Schussler and Westgate, 1995). Drought which coincides with this growth period can cause serious yield instability at the farm level, as it allows no opportunities for farmers to replant or otherwise compensate for loss of yield.

Given the damaging effect of drought on the well-being of farm families, CIMMYT began recurrent selection in an elite lowland tropical maize population `Tuxpeño Crema I' (Johnson et al., 1986) in 1975, under conditions of managed drought stress. This population, later renamed `Tuxpeño Sequia', underwent eight cycles of recurrent full-sib selection in the dry winter season at Tlaltizapán, México. Here the timing and intensity of drought stress could be managed by irrigation, one strategy sometimes used by plant breeders to develop drought tolerance. Methods of selection and water management have been described by Fischer et al. (1989) and Bolaños and Edmeades (1993a).

A second strategy for developing drought tolerance is the conventional breeding approach which relies on multilocation testing of progenies in environments representing a random selection of the variation in drought stress in the target environment (Rosielle and Hamblin, 1981). In CIMMYT's tropical maize breeding program, populations are improved by this means, with emphasis on high and stable yields across sites, as well as tolerance to foliar and ear diseases (Pandey and Gardner, 1992).

These two strategies were compared in studies reported by Bolaños and Edmeades (1993a) and Byrne et al. (1995). After eight cycles of full-sib selection in Tuxpeño Sequía, gains in grain yield of 0.11 Mg ha^-1 cycle^-1 were observed under different water regimes in the selection environment at yield levels that ranged from 0.6 to 8 Mg ha^-1. Byrne et al. (1995), in a series of international trials, demonstrated that 83% of these gains could still be observed in the normal rainy season. Yield gains in Tuxpeño Sequía were mainly attributed to selection for decreased barrenness, a short anthesis-silking interval (ASI) in an environment where plant water deficits coincided with flowering, and for yield itself (Bolaños and Edmeades, 1993a; 1996). Similar selection efforts were begun in 1986 in several other tropical populations to determine if these results could be repeated in a broader range of germplasm.

A modification to both strategies involves selecting for putative drought-adaptive secondary traits (Ludlow and Muchow, 1990), either alone or as part of a selection index. Previous studies have clearly shown the importance of ASI as an indicator of barrenness under stress (Bolaños and Edmeades, 1993b, 1996). An upright leaf habit should increase water use efficiency, since, as leaves become more upright, photosynthesis per unit intercepted radiation increases (Duncan, 1971) and heat load on individual leaves decreases. Similarly, selection for reduced tassel size may reduce shading and intraplant competition for assimilates. Bolaños and Edmeades (1993b) observed a concomitant reduction in both tassel size and plant height with selection in Tuxpeño Sequía. Leaf rolling under drought is an indication of leaf water status (Sobrado, 1987) and may identify, therefore, plants with inadequacies in water uptake or turgor maintenance. Drought is known to accelerate leaf senescence (Wolfe et al., 1988; Bolaños and Edmeades, 1993a), so selection for staygreen should increase intercepted radiation and hence grain yields. Relationships between grain yield and these traits at the progeny level, and responses to selection of secondary traits, have been reported by Bolaños and Edmeades (1996) and Chapman and Edmeades (1999). Genotype x environment effects for these traits arising from these studies have been reported in detail elsewhere (Chapman et al., 1997a,b) and will not be examined here.

An alternative to recurrent selection involves the intercrossing of known sources of drought tolerance to form a single population. Two populations of this type, `DTP1' and `DTP2', have been developed at CIMMYT. Some sources were landraces with obvious agronomic defects under high input conditions, while others were elite hybrids, lines, and selections from drought-tolerant populations. Initially, sources were recombined by a half-sib recurrent selection scheme. Later, S₁ recurrent selection was used to improve performance more rapidly (Edmeades et al., 1997a).

The objective of the present study was to evaluate changes in biomass, grain yield, and harvest index resulting from recurrent selection under managed drought stress in three maize populations and to compare these with conventionally selected check entries. A secondary objective was to assess the relative performance of source populations assembled from drought tolerant components or selected only for upright leaves.

	Materials and methods

TOP ABSTRACT INTRODUCTION Materials and methods Results and discussion Conclusions REFERENCES

 
Selection History of Entries
Information on each of the 16 entries included in the evaluation trials is summarized in Table 1 . All are adapted to the lowland tropics and subtropics and are intermediate to late in maturity. Trials consisted of 11 bulks formed from selection cycles of four populations that had undergone recurrent selection for drought tolerance. These populations were La Posta Sequía, Pool 26 Sequía, Tuxpeño Sequía, and its derivative `TS6'. To these were added five open-pollinated check populations, chosen to represent the outcome of alternative selection strategies.

Two populations formed from sources of drought tolerance, DTP1 and DTP2, were also examined. Following its formation from 14 sources in 1987, DTP1 underwent four cycles of mild half-sib selection, followed by S₁ recurrent selection as described above. After evaluation at Tlaltizapán, Morelos, México (19 °N, 940 m elevation), 10 superior S₁ progenies were recombined and advanced to F₂ to form `TL89DTP1 C₅'. This was then crossed with 25 additional drought-tolerant sources to form DTP2, and two cycles of half-sib recombination under very mild selection resulted in DTP2 C₂. An estimated 15% of DTP1 and 11% of DTP2 are comprised of landraces identified from areas in Mexico at an elevation of <1000 m and having <500 mm of annual rainfall (Edmeades et al., 1997a).

Recurrent Selection Methodologies
Recurrent selection was practiced in three unrelated lowland tropical maize populations. Progress from selection for drought tolerance in a fourth population, Tuxpeño Sequía, the progenitor of TS6, was examined in an earlier evaluation (Bolaños and Edmeades, 1993a). Two selections from Tuxpeño Sequía, C₀ and C₈, were included in this study.

The selection history of Tuxpeño Sequía and the creation of cycle bulks is described by Bolaños and Edmeades (1993a). In brief, each of 250 full-sib progenies was grown in single-row plots in three selection environments in the dry winter season at Tlaltizapán. These differed in drought intensity and were as follows: well watered (WW) under regular irrigation; intermediate stress (IS) where water was withheld from approximately 10 d before flowering until harvest; and severe stress (SS), in which water was withdrawn from about 30 d before flowering until harvest. The superior 26 to 32% of progenies were selected on the basis of an index comprised of increased grain yield under SS and IS, a constant grain yield in the WW regime, a short anthesis-silking interval (ASI), low canopy temperature and reduced leaf senescence under drought, and a high relative stem and leaf extension rate in the SS vs. WW treatment. The selection index (Barreto et al., 1991) was a single numerical value per progeny, obtained by summing contributions from each trait used during selection. That contribution was the square root of the product of the following: the trait value expressed in standard deviation units; a target for the trait, expressed in standard deviation units (e.g., 3 standard deviations to increase the value of a trait; -3 standard deviation units to decrease the value of the trait); and a weight that reflected the importance of that trait to breeding goals. Thus a small selection index value indicated a progeny whose phenotype was closest to the target for the traits under selection, while a large index value indicated an undesirable progeny. This procedure did not use correlations among traits to adjust weights automatically. Generally the weight for grain yield under stress was about twice that of traits such as ASI and canopy temperature and about four times the weight for the senescence score and the relative leaf extension rate. Exceptions occurred when grain yield was <0.4 Mg ha^-1 and its heritability low. It was more effective to place a greater emphasis on ears per plant and ASI, which usually exhibit large genetic variation at that yield level (Bolaños and Edmeades, 1996). Weightings were also modified subjectively on the basis of experience, as well as knowledge of relative trait heritabilities and phenotypic correlations among the traits.

Selection methodologies used in La Posta Sequía and Pool 26 Sequía were similar to that for Tuxpeño Sequía, the principal differences being in the progeny structure (S₁ vs. full-sib), selection intensity (8 vs. 26–32%), the number of selection sites (2 vs. 1), and the secondary traits used during selection. A two-stage evaluation of S₁ progenies was used, in which about 600 progenies were screened in single-row plots in an unreplicated nursery under two water regimes exposed to summer heat and drought stress at Cd. Obregón in the Sonoran Desert, México (27°30' N, 30-m elevation). The two regimes were WW and an intermediate level of stress induced by extending the interval between irrigations during flowering and grain filling from the normal 14 to 24 d. Each single-row plot of 10 plants was 3 m long, with 0.75 m between rows, at a density of 40 400 plants ha^-1. Fertilizer was applied at the rate of 150:30:0 kg N:P:K ha^-1. Traits measured were grain yield, ears per plant, time from sowing until 50% anthesis (AD), and ASI. Visual scores were recorded for leaf rolling, leaf senescence, and plant uniformity with a scale of 1 (unrolled, green leaves, or uniform plant type) to 5 (rolled, dead leaves, or very variable plant type). The superior 200 to 250 progenies were identified by the selection index as described. Generally the weight for grain yield under stress was about twice that of traits such as ASI and ears per plant, and around four times the weight of other secondary traits.

The 200 to 250 selected progenies were sown from remnant seed in an alpha (0,1) lattice design (Patterson and Williams, 1976) in two replicates during the dry winter season in Tlaltizapán, under the SS, IS, and WW water regimes. Superior progenies were again identified with an index that favored: increased grain yield under all treatments, a short ASI, increased ears per plant, and reduced leaf senescence and leaf rolling under stress; and small tassels, erect leaves, and resistance to lodging under WW conditions. The AD of the selected fraction was maintained the same as with the original population to avoid selecting escapes. Generally the weights for grain yield, ASI, AD, and ears per plant under stress were twice as large as those for leaf senescence under stress and grain yield under well-watered conditions, and three to four times larger than those for other secondary traits. The 50 selected progenies were crossed in all combinations at Poza Rica in summer, and a new set of S₁ progenies was generated by selfing individual plants within each resulting full-sib.

The fourth population, TS6, was treated slightly differently. Following evaluation of recurrent selection in Tuxpeño Sequía, C₆ was chosen for continued selection because of its vigor, renamed TS6, and S₁ recurrent selection was begun. In this population, only Tlaltizapán and the same three water regimes (WW, IS, SS) were used for evaluating the 220 S₁ progenies, with a selection intensity of 18%. Thus TS6 C₁ can be considered an advanced selection of Tuxpeño Sequía, but TS6 is treated here as a population distinct from Tuxpeño Sequía because of the change from full-sib to S₁ recurrent selection.

Cycle bulks of each population were prepared by sowing a balanced bulk of all newly created S₁ progenies and making plant-to-plant crosses among 300 plants with each plant used as a male only once. Selection at harvest was minimal, and each ear was shelled. A given volume of F₁ seed from each ear was bulked, and the process repeated to generate the F₂ seed which were used during evaluations of progress.

Evaluations
Treatments and Design
Ten trials were conducted at sites that included the relatively cool, dry selection environments of Tlaltizapán in winter, the hot dry selection environment of Obregón in summer, and one evaluation in the heat and humidity of the Poza Rica summer rainy season (Tables 2 and 3) . Daily maximum temperatures during the Tlaltizapán winter were comparable to those in the summer environments, but daily minimum temperatures were on average 7°C cooler. In the five well-watered (WW) trials, irrigation was applied every 10 d if rain was insufficient. The five water-deficit trials were managed by withdrawing or delaying irrigation. In the two winter dry season trials in Tlaltizapán, two moisture regimes were established: SS, with irrigation withdrawn approximately 21 d prior to anthesis; and IS, where irrigation was withdrawn 7 d before the first tassels appeared until harvest. Dates of irrigation withdrawal were estimated from time elapsed from sowing to flowering in previous trials in the same location. In 1992, 70 mm of rain fell several weeks prior to anthesis and alleviated stress during flowering. As well as being dry, the winter environment exposes crops to high maximum daily temperatures (often >33°C), relatively cool nights, and a large radiation load (26 MJ m^-2 d^-1) from early grain filling onwards. At Obregón, drought was imposed by applying irrigation at 20- to 25-d intervals instead of the normal 10- to 14-d intervals. Daily maximum temperatures at Obregón exceeded 35 °C on most days around flowering (Table 3), and reached 44 °C twice during grain filling. Each trial consisted of the 16 entries in an alpha (0,1) lattice design with four plots per sub-block and three replications. In the four trials conducted in 1992 winter season, a different cultivar was substituted for Pool 26 Sequía C₀, and these results are excluded from the analysis.

At final harvest, 34 bordered hills per plot (6.4 m² or 8.4 m² at Obregón) were harvested and plant stand recorded. Ears with at least one filled grain were counted, dried, and shelled. In the three trials conducted in the 1993 winter season, plants in 16 well-bordered hills (3.0 m²) were harvested for biomass and HI estimates. Stems, leaves plus husks, cobs, and grain were separated and dried to constant weight at 80°C in a forced air oven. Weights, including all grain yields, are expressed on an oven-dry basis.

Statistical Analysis
. Following analyses of variance for each variable at each site, the PROC MIXED procedure of SAS (SAS, 1994) with the REML option was used to analyze lattices across sites, considering entries as fixed effects, environments as random effects, and incomplete blocks as random within replicates. When comparing an advanced cycle with the original selection cycle or a check entry in a pre-planned comparison, a one-tailed t-test was used. For La Posta Sequía, and Pool 26 Sequía the entry sums of squares were partitioned into linear, quadratic, and residual components. A non-significant test for the linear component was taken to indicate no significant change with selection. Progress from selection was estimated by regression of entry means on cycle number. Percentage change per cycle was computed as the ratio of the regression slope to the mean of C₀ as estimated by that same regression. For estimating gain in Pool 26 Sequía, C₁ was used as the basis.

	Results and discussion

TOP ABSTRACT INTRODUCTION Materials and methods Results and discussion Conclusions REFERENCES

 
Differences among Entries in Grain Yield
Mean grain yield among the trials ranged from 1.0 to 10.4 Mg ha^-1 (Table 2). Earlier flowering entries (Table 4) generally performed better in water-stressed environments, while later flowering selections had the highest yields under full water supply. In the five drought environments where yields averaged < 5 Mg ha^-1, the earliest flowering entries, TL89DTP1 C₅, DTP2, and advanced cycles of Pool 26 Sequía and Tuxpeño Sequía gave the highest yields. Under the water regimes imposed in this study, water deficits increased with time during the flowering period. Since maize yield is more sensitive to water stress at flowering than at other times in the crop's life cycle (Denmead and Shaw, 1960; Grant et al., 1989), early flowering populations were able to escape the severe stress encountered by their later flowering counterparts during pollination. In addition, these environments were characterized by rising temperatures during grain filling so that later flowering entries experienced a less favorable environment for grain formation and growth. As a consequence, the regression of grain yield on AD for all entries in water-stressed environments showed a change in grain yield of -0.10 Mg ha^-1 d^-1 (P < 0.001). In the WW environments, selections from La Posta Sequía, a late-flowering population, were usually the highest yielding, though DTP2 C₂, one of the earliest maturing entries, also yielded well where foliar disease was not present. Despite these effects of relative maturity on yield, Chapman et al. (1997b) were able to clearly identify groups of drought intolerant (generally original cycles and checks) and tolerant (advanced selection cycles) entries in a principal components analysis of genotype x environment interaction sums of squares and showed that selection resulted in the progressive migration of populations from one group to the other.

Changes in Grain Yield in Response to Selection in Different Environments
In droughted environments, advanced cycles of La Posta Sequía, Pool 26 Sequía, and Tuxpeño Sequía (TS6 C₁) significantly outyielded their original cycles of selection and their respective checks, with the exception of TLWD-ELV (Table 4). In WW environments, however, the frequency of significant differences was lower than under drought. Across WW environments, advanced cycles of Pool 26 Sequía and Tuxpeño had larger yields than their original cycles (one-tailed t-test). A notable exception occurred at Poza Rica, where the yield of La Posta Sequía C₃ was significantly less than C₁. When data were combined across drought or WW environments, the yields of TS6 C₁ were greater than those of Tuxpeño Sequía C₈, and Tuxpeño Sequía C₈ yielded more than Tuxpeño Sequía C₀ (P < 0.05).

Across drought environments, yield gains from selection were always significant (P < 0.05). For La Posta Sequía, Pool 26 Sequía, and Tuxpeño Sequía, gains were 0.23, 0.29, and 0.08 Mg ha^-1 cycle^-1 (Table 4) or 12.4, 12.7, and 3.8% cycle^-1. Under WW conditions gains ranged from 0.04 to 0.18 Mg ha^-1 (0.5–2.3%) per cycle. While the gains for Tuxpeño Sequía from C₀ to C₈ were small in several WW environments, there was often significant improvement from C₀ to TS6 C₁. Despite care taken to avoid changes in flowering date with selection, AD changed with selection at an average rate of 0.7 d cycle^-1 (Chapman and Edmeades, 1999) and a proportion of the yield gains observed can be explained by this change. When AD is used as a covariate for La Posta Sequía and Pool 26 Sequía, gains in yield under drought declined by 29% from an average of 0.26 Mg ha^-1 cycle^-1 to 0.18 Mg ha^-1 cycle^-1. The use of AD as a covariate is questionable, however, since the reduction in AD and the increase in grain yield were not independent of selection, and indeed were exacerbated by selection. For this reason, throughout the rest of the discussion rates of gain for all traits are considered independent of their relationship with AD.

After three cycles of recurrent S₁ selection in La Posta Sequía and two in Pool 26 Sequía, advanced cycles outyielded their initial selection cycles by an average of 0.78 Mg ha^-1 (38%) in water-stressed environments producing 1 to 5 Mg ha^-1, by 0.35 Mg ha^-1 (4%) in WW environments yielding 5 to 10 Mg ha^-1, and by 0.45 Mg ha^-1 (9%) across all environments. After eight cycles of full-sib selection followed by one cycle of S₁ selection, TS6 C₁ outyielded Tuxpeño Sequía C₀ by 0.94 Mg ha^-1 (or 54%) under drought, by 0.71 Mg ha^-1 (or 9%) under WW conditions, and by 0.83 Mg ha^-1 (or 18%) over all water regimes. About 40% of the gains observed in this population, however, resulted from the single cycle of S₁ recurrent selection that occurred after the formation of C₈.

As in previous studies of Tuxpeño Sequía (Fischer et al., 1989; Bolaños and Edmeades, 1993a), improved drought tolerance was accompanied by increases in yield in unstressed conditions. In contrast to previous studies, lower gains were observed in all populations under fully irrigated vs. stress conditions, especially in Tuxpeño Sequía. On average, gains in La Posta Sequía and Pool 26 Sequía declined from 0.26 Mg ha^-1 cycle^-1 under drought to 0.12 Mg ha^-1 cycle^-1 under WW environments. High yielding sites in the present study included one with a mean yield of 10.4 Mg ha^-1, where gains in Tuxpeño Sequía, though non-significant, were negative. This suggests that the reduction in spikelet number per ear observed with selection in this population (Edmeades et al., 1993) may cause a yield reduction in high yield potential environments. Continued selection in the other two populations may eventually produce similar effects, since a decline in spikelet number per ear with selection has already been observed (M. Bänziger, 1994, unpublished data). One possible solution would be to increase the level of prolificacy under WW conditions, thereby increasing the number of competent spikelets per plant at flowering (Otegui and Melón, 1997).

Gains per cycle from S₁ recurrent selection in La Posta Sequía and Pool 26 Sequía were almost three times larger than those from the full-sib recurrent selection practiced in Tuxpeño Sequía (Table 4). This reflected differences in the selection intensity used with the two methods (8% for S₁s versus 26–32% for full-sibs), and the greater heritability for yield among S₁ progenies vs. full-sibs (Hallauer and Miranda Fo, 1981). On an annual basis, however, given that one cycle of S₁ selection requires 2 yr to complete, rather than 1 yr for full-sib selection, gains in the lower yielding environments in this study were only about 50% higher (0.12 Mg ha^-1) under S₁ recurrent selection than under full-sib selection (0.08 Mg ha^-1). Previous studies at yield levels ranging from 0.3 to 8.1 Mg ha^-1 have reported rates of gain per selection cycle in Tuxpeño Sequía of 0.11 to 0.14 Mg ha^-1 (Fischer et al., 1989), 0.11 Mg ha^-1 (Bolaños and Edmeades, 1993a), and 0.09 Mg ha^-1 (Byrne et al., 1995). It is thought that gains measured under water stress in the present evaluation are more accurate than some earlier estimates based on randomized complete block designs, since the relative efficiency of the lattice design was >50% in four of the five water-stressed environments.

Biomass, Partitioning, and Radiation Interception
Biomass was measured only in the TL93A season, under IS, SS, and WW conditions. Since stress levels of IS and SS were similar in severity (biomass yields averaged 6.47 and 7.53 Mg ha^-1), data for these were combined (Table 5) . Under drought AD and biomass production were not associated, and differences among entries in biomass were significant only because TS6 C₁ produced 22% more biomass than Tuxpeño Sequía C₀. Total aboveground biomass in the WW environment was significantly greater for cycles of La Posta Sequía (21.9 Mg ha^-1) than for those of Pool 26 Sequía (19.7 Mg ha^-1). Biomass of other entries fell between these two extremes (Table 5), and was generally similar in rank to AD (Table 4). Selection for improved grain yield under drought did not affect total biomass production, even though grain sink size was increased, a finding in agreement with earlier observations on Tuxpeno Sequía (Bolaños and Edmeades, 1993a).

Among entries, HI was greatest in the earlier maturing TL89DTP1 C₅ and DTP2 C₂ populations and in Pool 26 Sequía C₃ (Table 5). Harvest index under drought increased by 74 to 83% with selection in Pool 26 Sequía, La Posta Sequía, and Tuxpeño Sequía, while the HI of checks did not differ from that of their C₀ counterparts. The regression coefficient of HI on AD (-0.016 d^-1; P < 0.001) implies that around 25% of the observed gains in HI are accounted for by concomitant reductions in AD (Table 4). In the WW environment HI also increased significantly in these populations, though at a lower absolute rate (0.005 cycle^-1 in WW vs. 0.025 cycle^-1 under drought). This contrasts with earlier findings from Tuxpeño Sequía by Bolaños and Edmeades (1993a), who reported gains in HI of 0.005 to 0.007 cycle^-1 in both types of environments. The mean grain yield of WW environments in that study was only 3.8 Mg ha^-1, compared with 8.0 Mg ha^-1 in the present study. While managed drought stress can increase HI in environments yielding up to about 5 Mg ha^-1, gains in higher yielding environments may require the establishment of a larger sink though direct selection for traits such as prolificacy, reduced plant height, and small tassels (Mock and Pearce, 1975).

As expected, a strong positive relationship was detected between HI and grain yield under drought (Fig. 1) . Bolaños and Edmeades (1993a) and Edmeades et al. (1993) found that high HI under drought was associated with rapid early ear growth and suggested that it was an increase in partitioning to the ear that was responsible for increases in HI under all water regimes.

View larger version (21K): [in this window] [in a new window]   Fig. 1 Relationships between grain yield, stem biomass, and harvest index for 15 maize cultivars grown in two low yielding water-stressed environments at Tlaltizapán, Mexico. Data are lattice-adjusted means of three replicates from each trial

Selection Methodology
The advanced cycles of the drought-tolerant populations La Posta Sequía and Pool 26 Sequía outyielded their corresponding conventionally selected check entries in all of the drought environments and in 50% of the WW environments (Table 4). While part of their superiority can be attributed to small differences in AD (Table 4), much may also be due to the management of the intensity and timing of drought stress during selection. In addition, selection in all drought tolerant populations was made with an index that included high yield in drought and irrigated environments, as well as several secondary traits. Evidence from other selection studies under rainfed conditions (Zavala-García et al., 1992), and a study under an N-induced stress (Bänziger and Lafitte, 1997), suggest that index selection can result in a significant improvement in efficiency over selection for yield per se.

A comparison of Pop. 43 C₉ with La Posta Sequía C₀ indicates that little progress has been made in drought tolerance or yield potential through international progeny testing. In a separate study, a direct comparison of Pop. 43 C₆ and C₉ indicated that C₉ outyielded C₆ in a WW trial, but yielded similarly in two water-stressed trials (1994, unpublished data). A comparison of Pool 26 Sequía C₁, vs. Pool 26 C₂₃ that has been developed exclusively at Poza Rica, reveals gains in yield potential through conventional selection, but no improvement in drought tolerance. Similar results were reported by Bolaños and Edmeades (1993a). Byrne et al. (1995) also showed that gains from selection in Tuxpeño Sequía were 61% greater than the conventionally selected population when evaluated across a series of international sites. The value of drought-tolerant selections has been further examined in a series of 18 international trials with CIMMYT collaborators in the lowland tropics. Here La Posta Sequía C₃ outyielded all entries across environments, but Pool 26 Sequía C₃ was highest yielding only in environments characterized by a short and inadequate rainy season (Edmeades et al., 1997b).

There is concern that gains from selection in a dry winter nursery may not be realized in a normal rainy season environment. Limited evidence in the present study, provided from a typical summer production environment in Poza Rica, suggests that the proportion of gains observed at locations other than the selection site is approximately 0.85. Byrne et al. (1995), when evaluating progress in Tuxpeño Sequía at several sites within and outside Mexico, reported this proportion as 0.83 across a range of water availabilities. These data support the hypothesis that improved drought tolerance linked to changes in biomass partitioning (Edmeades et al., 1993) is relatively unaffected by environment.

We conclude that conventional selection of open-pollinated maize populations in generally favorable international environments, as currently conducted, results in little or no improvement in tolerance to mid–late season drought. This is confirmed by analysis of entry x environment effects which showed that the two selection environments are largely independent (Chapman et al., 1997a). This perhaps is not surprising, since the stress used during selection for drought tolerance was quite severe and occurred during every selection cycle. Under these conditions, facultative traits, such as osmotic adjustment and capacity to recover from stress, would have a direct effect on productivity in each selection cycle. At the same time, genetic variation for constitutive traits affecting stress tolerance and productivity, such as partitioning to the ear, is effectively and reliably exposed under managed water deficits. As drought stress intensifies, the flux of assimilate to the ear approaches, and for some cultivars crosses, a threshold level needed to form a fertile ear (Westgate and Bassetti, 1990; Edmeades et al., 1993; Schussler and Westgate, 1995; Otegui and Melón, 1997). Barrenness, or its commonly used equivalent, ears per plant, can then provide an easily observable measure of drought tolerance in the field.

Drought Tolerant Source Populations
The drought tolerant source populations (TL89DTP1 C₅ and DTP2 C₂) displayed good levels of drought tolerance, even though alleles for tolerance were contributed by diverse sources of germplasm rather than being increased in frequency by recurrent selection. The highest yield in this trial series, 11.4 Mg ha^-1, was recorded for DTP2 C₂ in the TL93A WW environment, suggesting that drought tolerance has not diminished yield potential in this population. The performance of DTP1 and DTP2 in international trials has generally been superior to that of Pool 26 Sequía and Tuxpeño Sequía, though slightly inferior to that of La Posta Sequía (Edmeades et al., 1997b).

	Conclusions

TOP ABSTRACT INTRODUCTION Materials and methods Results and discussion Conclusions REFERENCES

 
This study has demonstrated that drought tolerance can be obtained in lowland tropical maize populations, either by recurrent selection to increase frequency of drought-adaptive alleles or by assembling populations from sources in which these types of alleles are already present at a relatively high frequency. Tolerance is associated with an increased partitioning of biomass to the developing ear under drought so that HI and grain yield are increased. It does not result, however, in an increase in total biomass production under drought. Further gains in grain yield under well-watered conditions should be possible by increasing HI, preferably by shortening stems rather than compromising stalk strength.

Gains of 5% per year in grain yield under mid–late season water stress can be reliably attained through the use of recurrent selection, managed drought stress applied during flowering and grain filling, and appropriate selection indices. Recurrent S₁ selection has resulted in annual gains which were about 50% greater than those observed with recurrent full-sib selection, though differences in selection intensity explain part of this response. Recurrent S₁ selection has additional advantages over full-sibs in a pedigree breeding program because superior progenies can be advanced more rapidly to inbred lines. Managing the timing and intensity of water deficit so that it exposes genetic variation in ear growth rates, and hence HI, is a critical factor in obtaining consistent improvement in drought tolerance. When this is coupled with selection in favorable environments, it results in increased grain yield and HI under water stress, and smaller but significant improvements for these traits in unstressed environments. The reverse does not appear to hold—selecting in unstressed environments does not necessarily increase maize grain yields under water stress.SAS Institute 1994

	ACKNOWLEDGMENTS

 
We thank P. Galvez, J.C. Bahena, H. Corrales, and L. Casteñeda and their assistants who conducted these field trials in a competent and careful manner. Thanks are also due to Ings. L. Martinez, R. Salazar and M. Cortes for assistance with selection, seed preparation and data collection. We also thank Dr. A. Ortega C. of CIRNO, INIFAP, Obregón for his willing collaboration. Others involved in the selection of this germplasm include J.A. Deutsch, A.O. Diallo, K.S. Fischer, T.M.T. Islam, E.C. Johnson, S. Pandey, S.K. Vasal, and W. Villena. Helpful comments on earlier versions of this paper by A. Elings, P. Fox, S. Pandey and S. Waddington are gratefully acknowledged. This research was conducted as part of the UNDP project `Development of New Stress-Resistant Maize Genetic Resources' (GLO/90/003).

Received for publication August 27, 1998.

	REFERENCES

TOP ABSTRACT INTRODUCTION Materials and methods Results and discussion Conclusions REFERENCES

 

• Bänziger M., Lafitte H.R. Efficiency of secondary traits for improving maize for low-nitrogen target environments. Crop Sci. 1997;37:1110-1117.[Abstract/Free Full Text]
• Barreto, H.J., J. Bolaños, and H.S. Córdova. 1991. Indice de selección: guía para la operación del software. Manual de capacitación regional. CIMMYT, El Batan, Mexico.
• Bolaños J., Edmeades G.O. Eight cycles of selection for drought tolerance in lowland tropical maize. I. Responses in yield, biomass and radiation utilization. Field Crops Res. 1993;31:233-252 a.
• Bolaños J., Edmeades G.O. Eight cycles of selection for drought tolerance in lowland tropical maize. II. Responses in reproductive behavior. Field Crops Res. 1993;31:253-268 b.
• Bolaños J., Edmeades G.O. The importance of the anthesis-silking interval in breeding for drought tolerance in tropical maize. Field Crops Res. 1996;48:65-80.
• Bolaños J., Edmeades G.O., Martinez L. Eight cycles of selection for drought tolerance in lowland tropical maize. III. Responses in drought-adaptive physiological and morphological traits. Field Crops Res. 1993;31:269-286.
• Byrne P.F., Bolaños J., Edmeades G.O., Eaton D.L. Gains from selection under drought versus multilocation testing in related tropical maize populations. Crop Sci. 1995;35:63-69.[Abstract/Free Full Text]
• Chapman S.C., Crossa J., Basford K.E., Kroonenberg P.M. Genotype by environment effects and selection for drought tolerance in tropical maize. II. Three-mode pattern analysis. Euphytica 1997;95:11-20 a.
• Chapman S.C., Crossa J., Edmeades G.O. Genotype by environment effects and selection for drought tolerance in tropical maize. I. Two mode pattern analysis of yield. Euphytica 1997;95:1-9 b.
• Chapman S.C., Edmeades G.O. Selection improves drought tolerance in tropical maize populations: II. Direct and correlated responses among secondary traits. Crop Sci. 1999;39:1315-1324 (this issue).[Abstract/Free Full Text]
• De León C., Pandey S. Improvement of resistance to ear and stalk rots and agronomic traits in tropical maize gene pools. Crop Sci. 1989;29:12-17.[Abstract/Free Full Text]
• Denmead O.T., Shaw R.H. The effects of soil moisture stress at different stages of growth on the development and yield of corn. Agron. J. 1960;52:272-274.[Abstract/Free Full Text]
• Duncan W.G. Leaf angles, leaf area, and canopy photosynthesis. Crop Sci. 1971;11:482-485.[Abstract/Free Full Text]
• Edmeades G.O., Bänziger M., Beck D., Bolaños J., Ortega C A. Development and per se performance of CIMMYT maize populations as drought-tolerant sources. In: Edmeades et al G.O., ed. Developing drought and low N-tolerant maize. Mexico: CIMMYT, Mexico D.F, 1997:254-262 a.
• Edmeades G.O., Bänziger M., Beck D.L., Chapman S.C., Cortes M. Drought and low N testing networks—past, present and future. In: Edmeades et al G.O., ed. Developing drought and low-N tolerant maize. Mexico: CIMMYT, El Batan, 1997:405-409 b.
• Edmeades G.O., Bolaños J., Hernández M., Bello S. Underlying causes for silk delay in lowland tropical maize. Crop Sci. 1993;33:1029-1035.[Abstract/Free Full Text]
• Edmeades G.O., Bolaños J., Lafitte H.R. Progress in breeding for drought tolerance in maize. In: Wilkinson D., ed. Proc. Annu. Corn and Sorghum Ind. Res. Conf., 47th, Chicago. 9–10 Dec. 1992. Washington, DC: ASTA, 1992:93-111.
• Fischer K.S., Edmeades G.O., Johnson E.C. Selection for the improvement of maize yield under moisture-deficits. Field Crops Res. 1989;22:227-243.
• Fischer K.S., Johnson E.C., Edmeades G.O. Breeding and selection for drought resistance in tropical maize. El Batan, Mexico: CIMMYT, 1983.
• Grant R.F., Jackson B.S., Kiniry J.R., Arkin G.F. Water deficit timing effects on yield components in maize. Agron. J. 1989;81:61-65.[Abstract/Free Full Text]
• Hallauer A.R., Miranda Fo J.B. Quantitative genetics in maize breeding. Ames: Iowa State Univ, 1981.
• Johnson E.C., Fischer K.S., Edmeades G.O., Palmer A.F.E. Recurrent selection for reduced plant height in lowland tropical maize. Crop Sci. 1986;26:253-260.[Abstract/Free Full Text]
• Ludlow M.M., Muchow R.C. A critical evaluation of traits for improving crop yields in water-limited environments. Adv. Agron. 1990;43:107-153.
• Mock J.J., Pearce R.B. An ideotype of maize. Euphytica 1975;24:613-623.[ISI]
• Otegui M.E., Melón S. Kernel set and flower synchrony within the ear of maize. I. Sowing date effects. Crop Sci. 1997;37:441-447.[Abstract/Free Full Text]
• Pandey S., Gardner C.O. Recurrent selection for population, variety, and hybrid improvement in tropical maize. Adv. Agron. 1992;48:1-87.
• Patterson H.D., Williams E.R. A new class of resolvable incomplete block designs. Biometrika 1976;63:83-92.[Abstract/Free Full Text]
• Rosen S., Scott L. Famine grips sub-Saharan Africa. Agric. Outlook 1992;191:20-24.
• Rosielle A.A., Hamblin J. Theoretical aspects of selection for yield in stress and non-stress environments. Crop Sci. 1981;21:43-946.
• Russell W.A. Genetic improvement of maize yields. Adv. Agron. 1991;46:245-298.
• SAS Institute, Inc. Getting started with PROC MIXED. Cary, NC: SAS Institute, Inc., 1994.
• Schussler J.R., Westgate M.E. Assimilate flux determines kernel set at low water potential in maize. Crop Sci. 1995;35:1074-1080.[Abstract/Free Full Text]
• Sobrado M. Leaf rolling: a visual indicator of water deficit in maize (Zea mays L.). Maydica 1987;32:9-18.
• Westgate M.E., Bassetti P. Heat and drought stress in corn: what really happens to the corn plant at pollination?. In: Wilkinson D., ed. Proc. Annu. Corn and Sorghum Res. Conf., 45th, Chicago. 5–6 Dec. 1990. Washington, DC: ASTA, 1990:12-28.
• Wolfe D.W., Henderson D.W., Hsiao T.C., Alvino A. Interactive water and nitrogen effects on senescence of maize. I. Leaf area duration, nitrogen distribution, and yield. Agron. J. 1988;80:859-864.[Abstract/Free Full Text]
• Zavala-García F., Bramel-Cox P.J., Eastin J.D., Witt M.D., Andrews D.J. Increasing the efficiency of crop selection for unpredictable environments. Crop Sci. 1992;32:51-57.
• Zinselmeier C., Westgate M.E., Jones R.J. Kernel set at low water potential does not vary with source/sink ratio in maize. Crop Sci. 1995;35:158-163.[Abstract/Free Full Text]

This article has been cited by other articles:

				R. Venuprasad, H. R. Lafitte, and G. N. Atlin Response to Direct Selection for Grain Yield under Drought Stress in Rice Crop Sci., February 6, 2007; 47(1): 285 - 293. [Abstract] [Full Text] [PDF]

				P. Monneveux, C. Sanchez, D. Beck, and G. O. Edmeades Drought Tolerance Improvement in Tropical Maize Source Populations: Evidence of Progress Crop Sci., December 2, 2005; 46(1): 180 - 191. [Abstract] [Full Text] [PDF]

				J. T. Edwards, L. C. Purcell, and E. D. Vories Light Interception and Yield Potential of Short-Season Maize (Zea mays L.) Hybrids in the Midsouth Agron. J., January 1, 2005; 97(1): 225 - 234. [Abstract] [Full Text] [PDF]

				F. J. Betran, J. M. Ribaut, D. Beck, and D. G. de Leon Genetic Diversity, Specific Combining Ability, and Heterosis in Tropical Maize under Stress and Nonstress Environments Crop Sci., May 1, 2003; 43(3): 797 - 806. [Abstract] [Full Text] [PDF]

				F. J. Betran, D. Beck, M. Banziger, and G. O. Edmeades Genetic Analysis of Inbred and Hybrid Grain Yield under Stress and Nonstress Environments in Tropical Maize Crop Sci., May 1, 2003; 43(3): 807 - 817. [Abstract] [Full Text] [PDF]

				S. Chapman, M. Cooper, D. Podlich, and G. Hammer Evaluating Plant Breeding Strategies by Simulating Gene Action and Dryland Environment Effects Agron. J., January 1, 2003; 95(1): 99 - 113. [Abstract] [Full Text] [PDF]

				M. N. Andersen, F. Asch, Y. Wu, C. R. Jensen, H. Nasted, V. O. Mogensen, and K. E. Koch Soluble Invertase Expression Is an Early Target of Drought Stress during the Critical, Abortion-Sensitive Phase of Young Ovary Development in Maize Plant Physiology, October 1, 2002; 130(2): 591 - 604. [Abstract] [Full Text] [PDF]

				W. B. Bruce, G. O. Edmeades, and T. C. Barker Molecular and physiological approaches to maize improvement for drought tolerance J. Exp. Bot., January 1, 2002; 53(366): 13 - 25. [Abstract] [Full Text] [PDF]

				S.C. Chapman and G.O. Edmeades Selection Improves Drought Tolerance in Tropical Maize Populations: II. Direct and Correlated Responses among Secondary Traits Crop Sci., September 1, 1999; 39(5): 1315 - 1324. [Abstract] [Full Text]

This Article Abstract Figures Only Full Text (PDF) Free Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in ISI Web of Science Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via ISI Web of Science (39) Citing Articles via Google Scholar Google Scholar Articles by Edmeades, G.O. Articles by Bänziger, M. Search for Related Content PubMed Articles by Edmeades, G.O. Articles by Bänziger, M. Agricola Articles by Edmeades, G.O.