Some studies have estimated that by 2025, 60-70% more food will be needed from water resources similar to those available today, and more non-suitable land will have been incorporated into agricultural production (Edmeades et al., 1996). Global warming today is a fact. It is predicted to increase crop water use, and it may be accompanied by increasing variability in weather and recurrent droughts (Curry et al., 1995). In Colombia, these trends are compounded by the progressive displacement of staple food crops from well-favored environments by higher value crops or by non-agricultural activities. In the Cauca Valley-Colombia the food crops, including maize, were displaced by sugar cane and fruit crops for exportation. In the same way, cotton is displacing maize in the Sinu river's valley (Colombian Atlantic Coast).

Maize, however, is a major food crop widely grown in Colombia and it was grown on 679,000 hectares in 1996. In 1996, the national average grain yield of Colombia was 1.87 t ha-1 and in the same year, Colombia imported about two million tons (FENALCE, 1995). One of the main reasons for the yield difference between developing and developed countries is that more than 80% of the maize in the developing countries, including Colombia, is grown on tropical soils with poor fertility. In developing countries, a considerable proportion of maize is produced under low nitrogen conditions. Fertilizer use in these areas, however, is currently increasing at a rate of about 8% per year (CIMMYT, 1992).

Edmeades et al. (1992) indicated that in the developing world, where more than 60% of the world's maize is sown, annual yield losses due to drought may approach 24 million tons, equivalent to 17% of normal year's production. In the tropics, drought and low soil fertility (mainly N deficiency) frequently occur together, since the risk of economic loss when fertilizers are applied in drought-prone environments is often considered too great by small-scale farmers to justify rates of N application of more than 70 kg ha-1 (Banziger et al., 1997a). Some causes for low use of N include high price ratios between fertilizer and grain, its limited availability (long distance and bad roads between distribution and production centers), and the low purchase power of the poor farmers. (Banziger et al., 1997).

"Successful strategies to develop drought and low N tolerant maize require knowledge of the plant, environment, and the multitude of interactions between the two" (Beck et al., 1996).

Grant et al., 1989 documented very well that maize is specially sensitive to drought during flowering. At CIMMYT, selection for tolerance at flowering has received special emphasis, although recently they have initiated research on tolerance during crop establishment (Banziger et al., 1997c). The low nitrogen stress, in contrast to drought, is more predictable. We can find data of the soil N status, and it may be easily adjusted through controlled fertilizer applications (Beck et al., 1996). The N in the fields may be depleted by continuously growing high density extractor crops and removing the biomass after each crop. Vasal et al. (1997) stated that "the development of both drought and low N tolerant maize is difficult and complex and both require effective strategies that control or modify stress environments in order to obtain good estimates of genotype by environment interactions."

The strategies of maize breeding must considerate the target product. In the developed world, this product is a hybrid, but for developing countries it could also include open-pollinated varieties or synthetic cultivars. The CIMMYT has basically used four strategies. The first one, includes breeding directly under stress (drought or low N). The second one, includes data analysis of multi-location test progeny. The third one, evaluates secondary traits in order to select for stress. The fourth one, involves maize improvement for "general" stress tolerance through high density planting (Beck et al., 1996).

Genetic variation and selection response for drought tolerance has been reported (Edmeades et al., 1992; Byrne et al., 1995; Bolanos and Edmeades, 1996; Edmeades et al., 1997a). Edmeades et al., (1992) began selection for drought tolerance in the population Tuxpeno Sequia in 1975 and late this was expanded to four other tropical populations. Improvement was either by a recurrent full-sib or by S1 selection with a selection intensity from 8 to 30%. Drought stress, intermediate (IS) or severe (SS), was applied at flowering and at grain filling. All progenies were evaluated to determine yield potential under normal conditions (well watered). Grain yield in a progeny test showed a strong genetic correlation with anthesis-silking-interval (ASI) and ears per plant. They concluded that selection for drought tolerance by exposing progenies to a water deficit managed to coincide with flowering was effective at increasing grain yield under drought as well under well- watered conditions.

CIMMYT began recurrent selection for improved populations under low N in the experimental variety, Across 8328. Results of three selection cycles were reported by Lafitte and Edmeades (1994 a,b). Banziger et al. (1997) practiced selection among full-sib families of Across 8328; after five cycles of selection, they reported gains of 4.5% per cycle under low N and 2.2% per cycle under high N. Edmeades et al. (1997) evaluated lines from Across 8328 BN C5 developed after five cycles under low N and high N regimes. They found higher frequency of low N-tolerant topcrosses than for lines extracted from a sister population (improved for three cycles under high N only). The primary yield component that changed with selection was the number of grains per ear. These researchers also defined an ideal plant for low-N environment. It should have high yield under low and high N, vigorous vegetative growth maintaining low plant height, good flowering synchrony, delayed leaf senescence, no barren plants, and proper time at maturity.

Banziger et al. (1997) valuated fourteen-replicated trails grown under low and high N at CIMMYT, Mexico, between 1986 and 1995. They were analyzed to estimate broad-sense heritability of grain yield and its genetic correlations with other traits. They concluded that maize breeding programs for low N tolerance should include low-N selection environments in order to maximize selection gains.

The Colombian Atlantic Coast is a tropical lowland region. This Colombian Natural region is situated by the Caribbean sea and has boundaries with Panama and Venezuela. Its climate has two different seasons. The dry season (total rainfall less of 100 mm) is usually between December 15 and April 15 (four months). The rainy season is between April 15 and December 15 (eight months), but there are some dry periods in June or July. The rainy season has two cycles to grow maize (April-August and September-December), the cycles are called A and B, respectively. The distribution of precipitation in the rainy season is usually unpredictable, especially at north latitude. The IDEAM (1997), Instituto de Hidrologia, Meteorologia y Estudios Ambientales, analyzed the last six months (May to October ) for drought in the Caribbean Region. They determined any recovery of the humidity levels in the greater part of the litoral and Cordoba South. Nevertheless, in the Cesar valley (North) and the center region, the situation has been very variable and it has a deficit tendency (see appendixes A). The soils have an average organic matter content between 1.5 to 2%. The soils are considered to have low N content.

This region produces about 40% of the Colombian production of maize. The grain color is important for the Colombian people. The white grain color is used for food and the yellow grain color is usually used to feed animals. The Atlantic Coast production included 70% white and 30% yellow grain in 1996.

At the Colombian Corporation of Agricultural Research (CORPOICA), our work focuses on increasing the productivity and sustainability of crop farming systems. In this case for maize, CORPOICA has collaborative projects with the International Maize and Wheat Improvement Center (CIMMYT). In the Atlantic Coast CORPOICA (Regional 2) has two research centers. The first one, the National Research Center, Turipana, is situated in Cerete, Cordoba State ( 9o N, 76o W, 13 m elevation, average temperature 28o Celsius, 1,076 mm annual precipitation, and 83% RH). The second one, the Regional Research Center, El Carmen, is situated in Carmen de Bolivar, Bolivar State. It is 150 km north of the Turipana research center. The Agricultural Regional Program of CORPOICA Regional 2 in the Turipana research center, has a special project with CIMMYT. Its objective is to develop maize populations with drought and low N tolerance simultaneously.

Selection of parents

In 1996, 256 white genotypes and 256 yellow genotypes were evaluated simultaneously in three environments: normal N well-watered, low N well-watered, and normal N drought (Tables 1 and 2). The genetic materials included the best genotypes from previous studies in Turipana and CIMMYT, the best varieties, synthetics, base populations, and inbred lines from ICA, and the best accessions from Latino-American Maize Project (LAMP). The best experimental varieties also included inbred lines, and pools evaluated for drought and low N at CIMMYT ( such as; Pool 21, Pob 21 Tuxpeno Seleccion Sequia, La Posta Seleccion Sequia for drought, and Pob 28 for low N). Also, included were accessions, varieties, synthetics and S1 lines evaluated previously in Turipana for low N ( Garces-G, R, et al., 1996).

The materials were planted at the Turipana research center Cerete, Colombia. A 16 x 16 lattice design was used with two replicates in each environment. Plot size was one row, 2.5 m long, 0.8 m between rows, and two plants per hill spaced 0.5 m within the row (12 plants). Planting and harvesting were done by hand. Grain yield (t ha-1) was calculated as 80% of the ear weight adjusted to 150 g kg-1 moisture. Analysis of variance were computed for each environment and combined across environments. From these analysis 64 of the genotypes (25%) for each environment were phenotypically selected to form the base populations. Average performance for each genotype across the three environments was obtained from the combined analysis. Among the white grain genotypes, 25 were presented in the three environments, 39 in two environments (normal-drought), 34 in normal-low N, and 33 in drought-low N. Among the yellow grain genotypes, 23 were presented in the three environments, 37 in two environments (normal-drought), 36 in normal-low N, and 30 in drought-low N.

The best 64 genotypes per color/environment/germplasm source are listed in Tables 3 to 8. They were recombined in isolation fields in 1996B in Turipana Cerete-Colombia. In each isolation field the genotypes were planted in single rows, 5 m long, 0.8 m between rows, and two plants per hill spaced 0.50 m within the row (22 plants). Each row was detasseled (female). The male parent (bulk of equal number of seeds from each female parent) was planted between each block of four female rows. At the harvest, the best 45 rows (best plant per row) in each environment were selected to obtain 135 ears for three environments by grain color. Two populations were formed, one white grain color and other yellow grain color.

In 1997A these populations were planted ear-to-row for second cycle of recombination in isolation fields. The female parents were the 135 ears and the male parent was a bulk of equal number of seeds from each female parent. The final germplasm composition for the white and yellow grain population per germplasm source were as follows: The white grain population included individuals from 4 base populations, 3 synthetic cultivars, 6 open-pollinated varieties, 36 lines from CIMMYT, 12 lines from Pob 21, 12 entries from LAMP, and 5 commercial inbred lines. The yellow grain population included individuals from 5 lines Pob 28, 18 families from IPTT 28, 22 lines from CIMMYT, 3 open-pollinated varieties, 4 entries from LAMP, and 6 commercial inbred lines.

Development of Base Populations

General conditions

1. Two populations will be developed: white population (WP) and yellow population (YP).

2. Recurrent selection based on S1 lines combined with ear-to-row method will be used in both populations.

3. Turipana research center will have the nursery field. Turipana and El Carmen Center will be locations for evaluations. Additional evaluation locations will be in Tierralta ( A farmer field situated 100 km south of Turipana) and Repelon ( A farmer field situated 250 km north of El Carmen).

4. A 13x13 lattice design (169 entries; 165 progenies and four checks) will be used at each location. Plot size will be a single row, 2.5 m long, 0.8 m between rows, and two plants per hill spaced 0.50 m within the row (12 plants = 10 effective plants). Planting and harvesting will be done by hand. Anthesis-Silking-Interval (ASI), plant-ear height, stalk-root lodging, ears per plant and grain yield data will be recorded.

5. We will use 165 progenies. The selection intensity will be 5%. Each selection season the best 33 families (20%) and the best 5 individuals (25%) within families will be intermated. The ‘best' family or individual includes lowest ASI, lowest plant-ear height, least stalk-root lodging, greatest number of ears per plant, and greatest grain yield.

6. We will assume 30% loss of seeds planted (combine=germination+pest+mechanical damage). Therefore, we will need 16 seeds for planting per replication.

7. We also will assume an average of 200 seeds per ear.

8. We could grow three cycles per year: the first one December-15 (dry season), the second April-15 (cycle A), and the third August-15 (cycle B). The vegetative cycle of maize is 110 days.

9. Evaluations under controlled drought stress will be done only in dry season (Dec-15) at both research centers (Turipana and El Carmen). Evaluations under controlled low-N stress will be done in any cycle at both research centers ( there are fields with depleted N). The Carmen and Repelon locations will be used for evaluations under natural drought only cycles A ( April-15) and normal cycles B (August-15).

Season 1 (Date=August 15/97).

Two plots with 1000 plants each were planted in the nursery at Turipana. The best 500 plants were self-pollinated. At the harvest the best 165 S0:1 lines of each population are harvested individually. The S0:1 seed will be advanced by inbreeding (see Development of hybrid Cultivars), intercrossing selected lines, progeny testing, and for reserve ( cold room). See Development of inbred lines.

Season 2 (Date=December 15/97). Progeny testing S1 families.

Progeny testing of S1 families and recombination field are planted. The S0:1 lines are evaluated under drought stress and low N(two locations, Turipana and El Carmen) with one replication per environment/location = 4 replications. At the Turipana Center an isolation field under normal conditions ( well-watered and normal N)with two replications is planted for evaluation under normal conditions. In the same field three replications are planted: two for recombination (female) and one is bulked to produce seed of male pollinator. The field layout will be a ratio of four female rows (detassel) and one male row. A total nine replications will be used (144 seeds). Some plants may be either self pollinated or sib mated to obtain seed for the future use. From the analysis of data and visual evaluation of families in the recombination field, the 33 best families are selected and the 5 best plants within each family are selected (165 half-sib families). Each plant is harvested individually. The male is harvested in bulk and it represents recombination of all S1 plants.

Season 3 (Date=April 15/98). Progeny testing half-sib families.

Progeny testing for half-sib families and one field to produce S1 families are planted. The half-sib families are evaluated under farmer drought stress (two locations, El Carmen and Repelon) and low N(two locations, Turipana and El Carmen) with one replication per environment/location = four replications. At the Turipana Center in the nursery field under normal conditions four replications also are planted; two replications are evaluated under normal conditions and two for S1 seed production (all plants must be bagged and self-pollinated). At Tierralta two replications are planted under the farmer normal conditions. A total 10 replications will be planted (160 seeds). The analysis of data for each environment and across environments will provide information for the selection at the 33 best families and the 5 best plants within each family selected (165 self-families). Each plant is harvested individually.

Season 4 (Date=August 15/98). Progeny testing S1 families.

Progeny tests of S1 families and a recombination field are planted. The S0:1 lines are evaluated under low N(two locations, Turipana and El Carmen) with one replication per environment/location = two replications. At the Turipana Center in isolation field under normal conditions ( well-watered and normal N), two replications are planted for evaluation under normal conditions. In the same field, three replications are planted: two replications are for recombination (female) and one is bulked to produce seed of male pollinator. The field layout will be a ratio of four female rows (detassel) and one male row. Three replications are available for evaluation in normal and/or drought stress conditions. A total 10 replications will be used (160 seeds). Some plants may be either self or sib pollinated to obtain enough seed for the future use. From the analysis of data and visual evaluation in the recombination field, 33 best families are selected and 5 best plants within each family are selected (165 half-families). Each plant is harvested individually. The male is harvested in bulk, and it represents recombination of all S1 plants (Cycle 1 of selection).

Season 5 (Date= December 15/98). Progeny testing half-sib families.

Progeny testing of half-sib families and a recombination field are planted. The half-sib families are evaluated under drought stress and low N(two locations, Turipana and El Carmen) with one replication per environment/location = four replications. At the Turipana Center an isolation field under normal conditions ( well-watered and normal N) with two replications are planted for evaluation under normal conditions. In the same field, three replications are planted: two for recombination ( female) and one to produce seed of male pollinator. The field layout will be a ratio of four female rows (detassel) and one male row. A total of nine replications will be used (144 seeds). Some plants may be either self or sib pollinated to obtain enough seed for the future use. From the analysis of the data and visual evaluation in the recombination field, the 33 best families are selected and the 5 best plants within each family are selected (165 half-families). Each plant is harvested individually. The male is harvested in bulk and it represents recombination of all S1 plants. The procedure of seasons 2,3,4,5 is used for each cycle of selection.

Recurrent selection methods are flexible, and the methods may be adjusted as the needs of the farmers change. The methods of recurrent selection provide the opportunities to develop inbred lines for possible use in hybrids, improved open-pollinated cultivars (the populations themselves), and population crosses (cross of two populations) if the heterosis of the crosses increase with cycles of selection. The goal is to develop cultivars that have consistent performance across the range of environments experienced by the maize producers of Colombia.

Development of inbred lines for use in Synthetics or Hybrid Cultivars

At the present time, Colombia has no commercial hybrids from public sector in the market. The last hybrid was released in 1976 but it is not being presently produced. At the Colombian Corporation of Agricultural Research (CORPOICA), our work focuses on the development of Base Populations and open-pollinated varieties for the poor farmers. The market for hybrid seed is only about 30% of the total area each year (679,000 hectares in 1996). The commercial hybrids from private sector had been developed in other countries and they have some adaptation problems. The results of this project should permit the release of two best cultivars (one of white grain and other of yellow grain) as commercial hybrids from the public sector. These hybrids could be good alternatives for farmers on the Colombian Atlantic Coast.

"The development of inbred lines from a segregating population has six aspects: (a) formation of a segregating population, (b) inbreeding of the population to an adequate level of homozygosity, (c) evaluation of the performance of a line per se, (d) evaluation of the general combining ability of a line, (e) evaluation of a line in potential commercial hybrids, and (f) preparation of breeder seed of an inbred line." Fehr (1987).

Season 1. The best S1 lines (Progeny testing S1 families in December/97 ) will be advanced by inbreeding in the Turipana's breeding nursery.

The reserve of 33 best S1 lines (from the analysis of data and visual evaluation in season 2 of the development of base population program) are planted in April 15, 1998. Plot size will be a single row, 2.5 m long, 0.8 m between rows, and two plants per hill spaced 0.50 m within the row (12 plants = 10 effective plants). All plants are self pollinated. The best 5 plants within each family are selected (33 x 5=165 S2 families), and they are harvested separately (the pedigree method will be applied) . The S2 seed is obtained. Season 2 (Date=August 15/98). Early-generation testing will be conducted in S2 generation.

CORPOICA has two heterotic patterns with inbred lines. ICA L-225 and ICA L-266 (form the female of three way commercial hybrid ICA H-211) will be testers for the yellow lines. This pattern has been used in previous studies to define heterotic groups. ICA L 237 and ICA L 238 (form the single cross commercial hybrid ICA H-260) will be testers for the white lines. At the Turipana Center in the nursery field under normal conditions the 165 S2 families (from each population) are planted. Plot size will be a single row, 2.5 m long, 0.8 m between rows, and two plants per hill spaced 0.50 m within the row (12 plants = 10 effective plants). Within each line all plants must be bagged. They are self-pollinated (S3 seed) and the bulk remnant pollen is used to cross to two testers, respectively, to obtain testcross seed.

Season 3 (Date=December 15/98). The testcross seed from each S2 family is used to plant a replicated test

The testcross seeds (S2 lines x tester) are evaluated under drought stress and low N(two locations, Turipana and El Carmen) with one replication per environment/location = 4 replications. At the Turipana Center a field under normal conditions ( well-watered and normal N)with two replications is planted for evaluation under normal conditions. A total of six replications will be used (96 seeds). The ‘best' family or individual includes lowest ASI, lowest plant-ear height, least stalk- root lodging, greatest numbers of ears per plant, and greatest grain yield. The 33 S2 families with superior tescross performance are identified (superior combining ability). The S3 progeny from each S2 family are grown, and S3 plants are self-pollinated to obtain S4 seed. The evaluation for combining ability in S0:1 or S1:2 eliminates inferior genotypes before expensive testcross trials are conducted (Fehr, 1987).

. Season 4 (Date=April 15/99)

S4 progeny of S2 families with superior combining ability are grown, inbreeding is continued, and experimental hybrids between the best lines are produced and tested after initial tests for combining ability. Large inbreeding depression (ID), ranging from 27 to 58%, for lowland tropical germplasm has been reported for several authors; thus, it is important to improve tropical maize germplasm for tolerance to ID in order to accelerate hybrid breeding programs in developing countries (Vasal et. al., 1995). For the reason above, we have usually used inbred lines with low inbreeding (S3 or S4). Vasal et. al., (1995) practiced, recurrent selection in four tropical maize populations; after two cycles of selection, they reported that ID for grain yield across populations decreased from 39% in C0 to 35% in C2 and the selection was effective in improving grain yield and reducing ID. These method could be applied in the development of base population.

The production of the experimental hybrids and synthetic cultivars will use the S4 lines as parents. It must be included: 1) Diallel crosses among the best 10 lines for each heterotic group. 2) The best identified hybrids in early generation testing will be formed with S4 lines and testers in each heterotic group. 3) Synthetic cultivars with homogeneous lineas. Plot size will be a 10 rows, 2.5 m long, 0.8 m between rows, and two plants per hill spaced 0.50 m within the row (100 effective plants). It should produce 2 kg. of each experimental cultivar. We could have in this time 169 entries including several checks ( 13 x 13 lattice design).

Season 5 (Date=August 15/99). The 1st yield tests of the experimental hybrids and synthetics cultivars.

In this season will begin the production of breeder seed (10 lines for each heterotic group/grain color=40 lines). At the Turipana Center in the nursery field under normal conditions these lines are planted. Plot size will be a 10 rows, 2.5 m long, 0.8 m between rows, and two plants per hill spaced 0.50 m within the row (100 effective plants). Plant-to-plant crosses will be used. Off- type plants in female and male parents will be removed before flowering.

The experimental hybrids and synthetics cultivars are evaluated under low N(two locations, Turipana and El Carmen) with two replications per environment/location = 4 replications. At the Turipana Center in a field under normal conditions ( well-watered and normal N), two replications are planted for evaluation under normal conditions. The ‘best' cultivars includes lowest ASI, lowest plant-ear height, least stalk-root lodging, greatest numbers of ears per plant, and greatest grain yield.

Season 6 (Date= December 15/99). The 2nd yield tests of the experimental hybrids and synthetics cultivars.

The yield tests are planted under drought stress and low N(two locations, Turipana and El Carmen) with two replication per environment/location = eight replications. At the Turipana Center a field under normal conditions ( well-watered and normal N) with two replications are planted for evaluation under normal conditions. At the Turipana Center in the nursery field under normal conditions 20 best line parents are planted. Plot size will be a 20 rows, 2.5 m long, 0.8 m between rows, and two plants per hill spaced 0.50 m within the row (200 effective plants). Plant-to-plant crosses will be used. Off-type plants in female and male parents will be removed before flowering.

Season 7 (Date=April 15/00)The 3rd yield tests of the experimental hybrids and synthetics cultivars.

The yield tests are evaluated under farmer drought stress (two locations, El Carmen and Repelon) and low N(two locations, Turipana and El Carmen) with four replications per environment/location = 16 replications. At the Turipana Center in the nursery field under normal conditions four replications also are planted. At Tierralta four replications are planted under the farmer normal conditions. A total 24 replications will be planted (384 seeds). The analysis of data for each environment and across environments and years will provide information for the selection at the best two hybrids and two synthetics (each with different grain color). Two cycles evaluations and two locations/cycle are required by the Colombian Agricultural Institute (ICA) before the release of any cultivar. At the Turipana Center in the nursery field under normal conditions 4 best line parents and two best cultivars are planted. Plot size will be 0.5 ha in isolation. Off-type plants parents will be removed before flowering

Season 8 (Date=August 15/00). The 4th yield tests of the experimental hybrids and synthetics cultivars.

At the Turipana Center in the nursery field under normal conditions 4 best line parents and two synthetic cultivars are planted in isolation plots. Plot size will be five hectares. Off-type plants will be removed before flowering (Foundation seed is produced). Those hybrids and synthetic cultivars that have superior performance are considerate for possible use by the farmers. Ultimate production and release of hybrids will depend on consistently good performance of the hybrid and if the parent lines of better hybrids can be used to produce the hybrid seed.

The decision to release a cultivar at CORPOICA is made after a seminary of presentation for the new cultivars. At this seminary are invited researchers from public and private sector, farmers, commercial growers, and seed producer companies.

Seed multiplication and distribution

The Colombian Corporation of Agricultural Research (CORPOICA) is a public institution that conducts breeding programs in many crops, including maize. Guidelines for the release and multiplication of cultivars have been developed by ICA ( Colombian Agricultural Institute). CORPOICA and ICA, both, depends of the Colombian Agricultural Ministry. These institutions have been producing foundation and certified seeds for about of 30 years. CORPOICA has been licensed by ICA to sell seeds to commercial growers and it is responsible for the multiplication and marketing of the released cultivars. At CORPOICA, the policy has been to provide a cultivar to farmers at the lowest price. Other strategy is to distribute seed stock (foundation seed) to one o more private companies without any royalty. CORPOICA has not obtained plant variety protection for any cultivars, up to the present time.

Mass selection and progeny testing are usually used to produce breeder seed in CORPOICA Isolation field and mass selection are use to produce foundation and certified seeds. In these fields, the off-type plants are removed before flowering and harvest. Each these fields must be certified by ICA. This process includes the inspection of the crop in field and of the samples of the harvested seed.

References

Bazinger, M., and H.R. Laffite. 1997. Efficiency of secondary traits for improving maize for low- nitrogen target environments. Crop Sci. 37:1110-1117.

Bazinger, M., F.J. Betran, and H.R. Laffite. 1997a. Efficiency of high-nitrogen selection environments for inproving maize for low-nitrogen target environments. Crop Sci. 37:1103-1109.

Bazinger, M., G.O. Edmeades, and S. Quarrie. 1997c. Drought stress at seedling satege: are there genetic solutions? In G.O. Edmeades, M. Banziger, H.R. Mickelson and C.B. Pena-Valdivia (eds.) Developing Drought and Low-N Tolerance Maize. CIMMYT, El Batan,Mexico (in press).

Beck, D., J. Betran, M. Banziger, G. Edmeades, J.M. Ribaut, M. Willcox, S.K. Vasal, and A. Ortega. 1996. Progress in developing drought and low soil nitrogen tolerance in maize. Paper presented at the 51st Corn & Sorghum Conference, Chicago IL., Dec., 10-11,1996.

Bolanos, J., and G.O. Edmeades. 1996. The importance of the anthesis-silking interval in breeding for drought tolerance in tropical maize. Field Crops Res. 31:233-252

Byrne, P.F., J. Bolanos, G.O. Edmeades, and D.L. Eaton. 1995. Gains from selections under drought versus multilocation testing in related tropical maize populations. Crop Sci. 35:63-69.

CIMMYT. 1992. Enduring design for change: An account of CIMMYT's research, its impact and its future directions. Mexico, D.F.

Curry, R.B., J.W. Jones, K.J. Boote, R.M. Peart, L.H. Allen, and N.G. Pickering. 1995. Response of soybean to predicted climate change in USA. P. 163-182. In Climate Change and Agriculture: Analysis of Potential International Impacts. ASA Special Publication No. 59. ASA, Madison, WI.

Edmeades, G.O., J. Bolanos and H.R. Lafitte. 1992. Progress in breeding for drought tolerance in maize. P. 93-111. In: D. Wilkinson (ed.) Proceedings of the 47th Ann. Corn and Sorghum Ind. Res. Conf. 1992. ASTA, Washington.

Edmeades, G.O., J. Bolanos, M. Banziger, H.R. Lafitte, J.M. Ribaut, J.W. White, and M.P. Reynolds. 1996. Improving crop yields under water deficits in the tropics. Paper presented at the International Crop Science Cong., Delhi, India, Nov., 17-21, 1996.

Edmeades, G.O., J. Bolanos, S.C. Chapman, H.R. Lafitte, and M. Banziger. 1997a. Selection for tolerance to mid/late season drought in tropical maize. I. Gains in grain yield, biomass and harvest index. Submitted to Crop Science.

Edmeades, G.O., J. Bolanos, and S.C. Chapman. 1997b. Value of secondary traits in selecting for drought tolerance in tropical maize. In G.O. Edmeades, M. Banziger, H.R. Mickelson and C.B. Pena-Valdivia (eds.) Developing Drought and Low-N Tolerance Maize. CIMMYT, El Batan, Mexico (in press).

Fehr, W.R. 1987. Principles of Cultivar Development. Vol. 1. Theory and Technique. Macmillan Publishing Co., New York.

FENALCE. 1995. Hoja Economica Cerealista . Estadisticas de la produccion de sorgo y maiz en Colombia. No. 32 Mayo-Junio, 1995.

Grant, R.F., B.S. Jackson, J.R. Kiniry, and G.F. Arkin. 1989. Water deficit timing effects on yield components in maize. Agron J. 81:61-65.

Garces-G, R., A. Humanez-A, A. Navas-A, J.C. Perez y U. Caraballo. 1995. Evaluacion agronomica de lineas autofecundadas (S1) obtenidas a partir de nueve cultivares de maiz para condiciones normales y de bajo contenido de N en el suelo. CORPOICA, Colombia.

IDEAM. 1997. Instituto de Hidrologia, Meteorologia y Estudios Ambientales. Colombia. Internet address http://www.ideam.gov.co.

Lafitte, H.R., and G.O. Edmeades. 1994. Improvement for tolerance to low soil nitrogen in tropical maize. II. Grain yield, biomass production, and N accumulation. Field Crops Res. 39:15-25.

Vasal, S.K., H. Cordova, D. Beck, and G.O. Edmeades. 1997. Choices among breeding procedures and strategies for developing stress tolerant maize germplasm. In G.O. Edmeades, M. Banziger, H.R. Mickelson and C.B. Pena-Valdivia (eds.) Developing Drought and Low-N Tolerance Maize. CIMMYT, El Batan, Mexico (in press).

Vasal, S.K., B.S. Dhillon, G. Srinivasan, S.D. McLean, J. Crossa, and S.H. Zhang. 1995. Effect of S3 Recurrent Selection in Four Tropical Maize Populations on Their Selfed and Randomly Mated Generations. Crop Sci. 35:697-702.