Crop Science 42:1441-1446 (2002)
© 2002 Crop Science Society of America
CROP BREEDING, GENETICS & CYTOLOGY
Progress in Breeding Wheat for Yield and Adaptation in Global Drought Affected Environments
Richard M. Trethowan*,
Maarten van Ginkel and
Sanjaya Rajaram
Wheat Program, International Maize and Wheat Improvement Center (CIMMYT), Apdo. Postal 6-641, 06600 Mexico DF, Mexico
* Corresponding author (r.trethowan{at}cgiar.org)
 |
ABSTRACT
|
|---|
The impact of wheat (Triticum aestivum L.) germplasm from the International Maize and Wheat Improvement Center (CIMMYT) in the highly productive environments of the developing world has been significant. However, 55% of the area sown to wheat in these countries is periodically affected by drought and the impact of CIMMYT germplasm on productivity in these areas is not clear. Our objective was to measure rates of yield improvement for the period 1979 through 1998 using yield data from CIMMYT's Elite Spring Wheat Trial (ESWYT) and Semi-Arid Wheat Yield Trial (SAWYT). The mean yield of the five highest yielding entries from each site was expressed as a percent of the trial mean (%TM). The trial mean yield (TM) and the mean yield of locally adapted check cultivars (LC) were used to provide estimates of the productivity of each environment. Measuring rates of progress by means of %TM, TM, and LC were favored to the use of mean yield alone as variable annual rainfall and subsequent fluctuations in productivity influence yield in dry environments. Yearly rates of progress were determined by measuring change in %TM and change in TM. In environments yielding less than 4 Mg ha-1 the respective increases in %TM and TM for SAWYT were 4.38 and 0.09% yr-1. The equivalent rates for ESWYT were 0.34 and 0.19% yr-1. In environments yielding 4 Mg ha-1 or more the respective rates of progress in %TM and TM for SAWYT were 0.85 and 2.87% yr-1 compared with 0.26 and 0.494% yr-1 observed in the ESWYT. Comparisons of %TM, TM, and LC for representative locations in Mexico, Argentina, Syria, and Portugal demonstrated progress in %TM ranging from highly significant to stable, regardless of local fluctuations in productivity.
Abbreviations: CIMMYT, International Maize and Wheat Improvement Center • %TM, mean yield of the five highest yielding lines expressed as percentage of the trial mean • TM, trial mean • LC, mean of the locally adapted check cultivar • ESWYT, Elite Spring Wheat Yield Trial • SAWYT, Semi-Arid Wheat Yield Trial
|
INTRODUCTION
|
|---|
THE WHEAT PROGRAM of the International Maize and Wheat Improvement Center aims to develop germplasm for most of the spring wheat production areas of the developing world. These areas include some 40 million hectares of irrigated and generally high input wheat and close to 45 million hectares of wheat suffering periodically from drought stress (Byerlee and Moya, 1993; Ginkel van et al., 2000). Many authors have found significant progress in yield potential in highly productive environments recording rates of annual yield increase of between 0.7 and 1.3% (Fischer and Wall, 1976; Morris et al., 1992; Byerlee and Moya, 1993; Sayre et al., 1997). Estimates for drought affected areas are on average half of these values (J. Hamblin, 2001, personal communication). Measuring progress in breeding for grain yield in dry environments is frequently confounded by highly variable seasons year to year (Derera et al., 1982). Therefore, measuring yield advances in drought affected areas by plotting average yield against time is likely to be confounded by seasonal fluctuations in rainfall.
Trethowan et al. (2001) examined the relationship among various locations using yield data from the SAWYT. This analysis examined the way in which different environments differentiate CIMMYT wheat germplasm. However, it is not clear what progress in yield has been made in these areas over time. The objective of our study was to examine rates of progress in yield and adaptation in drought affected areas using data from CIMMYT's international yield testing network. We evaluated yield performance of the best germplasm using a combination of yield of the best lines, the trial mean, and the performance of a locally adapted check cultivar from each site to achieve a better-balanced interpretation.
|
MATERIALS AND METHODS
|
|---|
The ESWYT and SAWYT were used to examine trends in yield performance over time. These two yield trials contain germplasm tested at a large number of international locations between 1979 and 1998 (Table 1)
. ESWYTs 1 through 20 (1979–998) contained materials developed primarily for highly productive, irrigated environments, whereas SAWYTs 1 through 7 (1991–1997) represent germplasm bred specifically for drought prone areas. ESWYTs 1 through 20 and SAWYTs 1 through 7 were sown in both irrigated and rainfed locations. Each nursery consisted of between 30 and 50 genotypes arranged in a randomized complete block design with two replicates. After 1990, the experimental design was changed to a two replicate alpha-lattice design (Barreto et al., 1997). A local check cultivar, representing the best locally adapted germplasm at each location, was included by the cooperating organization. The local check varied among locations and sometimes varied with time at the same location. Seed of entries distributed in each year by CIMMYT was always produced at the same location in northern Mexico. Trials were packaged and randomized in Mexico and sown globally with local agronomic practices used. Yield was determined at each location and analyzed by the PROC MIXED procedure in SAS (SAS Institute, 1997) to produce means within locations for each genotype. Genotypes were considered fixed effects and replicates and sub-blocks within replicates (for alpha-lattice designs) as random effects.
View this table:
[in this window]
[in a new window]
|
Table 1. Nursery name, year of dissemination, and the total number of sites within each year returning yield data.
|
|
Using trial means, we then subdivided sites into high yielding (greater than or equal to 4 Mg ha-1) and low yielding (less than 4 Mg ha-1) groups (Table 1). Twenty-one low yielding sites were removed from the comparison because their low mean yield was not related to drought stress, but to severe disease epidemics under well watered conditions. This provided a total of 1041 sites over the 20-yr period of which 483 were low yielding and 558 high yielding (Table 1). The mean of the five highest yielding genotypes from each site and year were calculated and expressed as a percentage of the trial mean (%TM). We then regressed the %TM, TM, and LC against year to assess gains in yield over time using linear regression analysis. Estimates of the mean productivity of sites sown in each year were obtained by regressing TM and LC against time.
|
RESULTS
|
|---|
Progress in Breeding for Grain Yield Across All Sites and Years
Regression of %TM against year indicated that between 1979 and 1998 the mean advantage of the five highest yielding lines over the TM in the ESWYT increased at a rate of 0.34% yr-1 for environments yielding less than 4 Mg ha-1 (Fig. 1A)
. There was a 5.3 kg yr-1 (0.19% yr-1) increase in the TM calculated from the regression equation from 1979 to 1998. Yield progress in the SAWYT between 1991 to 1997 in low yielding environments was significantly greater than that observed for ESWYT (Fig. 1B); the %TM increased by 4.38% yr-1 while the TM increased only marginally (2.1 kg yr-1 or 0.09% yr-1).
View larger version (27K):
[in this window]
[in a new window]
|
Fig. 1. (A) The mean of the five highest yielding entries from each site and year expressed as a percentage of the trial mean ( %TM) compared with change in the trial mean (TM) and local check mean (LC): ESWYT sites yielding less than 4 Mg ha-1. (B) The mean of the five highest yielding entries from each site and year expressed as %TM compared with change in the TM and LC: SAWYT sites yielding less than 4 Mg ha-1.
|
|
In high yielding environments (those yielding 4 Mg ha-1 or more), the rate of yield advance for the ESWYT was 0.26% yr-1 for %TM and 0.49% yr-1 (28.2 kg yr-1) for TM (Fig. 2A)
. However, yield advances were much greater in the SAWYT under high yielding conditions between 1991 and 1997 (Fig. 2B). The respective rates of progress in %TM and TM for the SAWYT were 0.85 and 2.87% (141.8 kg) yr-1. The TM and LC were highly correlated across all 20 yr of the ESWYT in both low yielding (r = 0.76, P < 0.001) and high yielding (r = 0.87, P < 0.001) environments. Similar trends were recorded for the 7 yr during which the SAWYT was sown. The LC and TM either remained static or increased with time between 1979 and 1998 and 1992 and 1998 for the ESWYT and SAWYT, respectively (Fig. 1 and 2).
View larger version (27K):
[in this window]
[in a new window]
|
Fig. 2. (A) The mean of the five highest yielding entries from each site and year expressed as a percentage of the trial mean ( %TM) compared with change in the trial mean (TM) and local check mean (LC): ESWYT sites yielding 4 Mg ha-1 or greater. (B) The mean of the five highest yielding entries from each site and year expressed as %TM compared with change in TM and LC: SAWYT sites yielding 4 Mg ha-1 or greater.
|
|
Progress in Breeding for Grain Yield at Some Selected Locations
Four locations were chosen to examine progress at individual sites over time (Fig. 3)
. Each of these four locations represents a general trend observed across years at many locations. Each figure compares the regression of %TM over time with fluctuations in the yield of TM and LC. To measure yield progress in time before and after the introduction of the stress-targeted SAWYT, a time scale including yield of ESWYT before 1991 and yield of SAWYT after 1991 was used. Progress in %TM at the Centro de Investigaciones Agricolas del Noroeste (CIANO), CIMMYT's primary drought testing location was significant between 1991 and 1997 (Fig. 3A). At this site, %TM increased by 2.5% yr-1. No comparisons before 1991 could be made because all ESWYT trials were optimally irrigated and yield levels did not represent moisture-stressed conditions.
View larger version (70K):
[in this window]
[in a new window]
|
Fig. 3. (A) The mean of the five highest yielding lines expressed as a percentage of the trial mean (%TM) compared with the trial mean (TM) and the mean of the local check (LC): CIANO drought trials. (B) The mean of the five highest yielding lines expressed as %TM compared with TM and the mean LC: Argentina, Marcos Juarez. Fig. 3. (C) The mean of the five highest yielding lines expressed as %TM compared with TM and the mean of LC: Syria, Aleppo. (D) The mean of the five highest yielding lines expressed as %TM compared with TM and the mean of LC: Portugal, Elvas.
|
|
Yield at Marco Juarez, Argentina, increased by 1.8% yr-1 (R2 = 0.30) during the period 1981 through 1996 (Fig. 3B). The increase in %TM continued regardless of the stress level experienced in any one year or period. The period 1992 through 1996 recorded significantly lower TM and LC yields, indicating higher levels of stress in the test environments, yet %TM continued to increase. Aleppo, Syria, also demonstrated a significant increase in %TM of 0.96% yr-1 (R2 = 0.55) for the period 1979 through1997 (Fig. 3C). TM and LC fluctuated considerably during the period highlighting the variable nature of this rainfed test location. The site at Elvas, Portugal, showed no progress in %TM (R2 = 0.02) for the period 1980 through 1994 (Fig. 3D). However, the advantage of the top five cultivars did not decrease throughout this period, regardless of the significant downward trend in TM and LC after 1988.
|
DISCUSSION
|
|---|
Most studies addressing progress in yield over time have used yield of historical genotypes grown in the same environment (Fischer and Wall, 1976; Sayre et al., 1997) or have used mean yield to examine progress over time in highly productive environments (Calderini et al., 1995; Kulshrestha and Jain, 1982). However, measuring rates of progress in drought affected environments by TM can be misleading as yield is greatly affected by rainfall patterns in any given year. This is obvious from the fluctuations in TM and LC in Fig. 3. We decided to compare the mean of the five highest yielding entries in each nursery as the majority of lines in the ESWYT and SAWYT trials are not expected to be adapted at any specific location. These multilocational trials differ in scale to similar trials conducted by breeders working in specific geographic regions. For example, prevailing conditions in the Southern Cone of South America are substantially different to those in North Africa (Trethowan et al., 2001). For this reason we expect that only a small proportion (5–10%) of the genotypes included in the trial will be adapted in any given region (S. Rajaram, 2001, personal communication). The germplasm entering SAWYT and ESWYT is assembled on the basis of broad genetic variability, whereas most regional multilocational trials contain elite germplasm representing much narrower genetic variation. Because of the significant lag phase (approximately 2 yr) between assembly of the new annual nursery in Mexico and receipt of the yield data from different parts of the world, the opportunity to include the highest yielding lines from the previous year is limited. Instead, new high yielding lines identified by testing in Mexico are included in the new yield trials for international distribution. Genotypes selected from these trials by regional researchers usually enter their own regional multilocational trials in the following year and are often not reincluded in the CIMMYT yield trial. Therefore, the relatively small increase in TM with time is not unexpected and, unlike the regional multilocational trials conducted by most breeders, %TM is not expected to approach TM with time. Nevertheless, regression of %TM against time could be misleading. For this reason, we compared %TM with both TM and LC from each site, thereby providing an estimate of the stress level experienced in any given year. Because the LC correlated strongly with fluctuations in the TM, we can discount any fall in TM as being related to the possible inclusion of less adapted lines in later years. Progress in %TM can then be estimated in the context of the prevailing growing conditions. We decided to calculate %TM instead of using the mean of the top five entries expressed as a percentage of the local check cultivar because the local check cultivar varied among sites and at the same site in time. In some instances, the pedigree of the local check was not recorded. It was therefore difficult to measure progress in yield over time by methods such as those described by St. Martin and McBlain (1991) for multilocational yield trials. These authors used a set of check cultivars repeated in more than 1 yr to estimate genetic gain and its associated error.
Clearly, there has been significant progress in the development of germplasm adapted to both low and high yielding conditions over the 20 yr examined in this study. The significantly greater rates of progress in yield over time recorded for the SAWYT compared with ESWYT for both high and low yielding environments reflects, to some extent, the slightly higher average TM yields recorded for ESWYT. Nevertheless, the significant advance in yield of the SAWYT over time in low yielding environments (R2 = 0.63) indicates that selection of germplasm for stress adaptation in Mexico influences performance in dry areas globally. All germplasm entering the SAWYT has been bred in Mexico expressly for adaptation to drought. Breeding for drought adaptation involved selection in alternating drought and nondrought environments as out lined by Trethowan et al. (2001). The increase in %TM and the TM over time recorded for the SAWYT in environments yielding in excess of 4 Mg ha-1 indicates that yield responsiveness to improved growing conditions has been maintained through this shuttle process. The lower rates of yield improvement for the ESWYT in low yielding environments compared with the SAWYT is not unexpected as the ESWYT germplasm was selected for high yield performance under non-limiting conditions. The relatively modest gains in ESWYT yield in the more productive environments is a reflection of the intense selection pressure applied for yield potential at CIMMYT over many decades (Rajaram and van Ginkel, 1996) and the slightly higher trial mean yields compared with SAWYT. In contrast, the targeted breeding effort for drier areas began more recently at CIMMYT (Rajaram et al., 1996).
When individual sites were examined, the difficulties of using mean yield to estimate progress with time are evident. At Marcos Juarez such an interpretation would indicate significant yield depression with time; however, progress measured as %TM was significant. The close correlation of the LC and the TM indicated higher stress levels, and hence lower productivity, in the later years of this study. Other sites such as Elvas, Portugal, indicated very little progress in %TM. However, given the extremely variable productivity of these environments year to year, a flat response in time for %TM indicates that some progress has been made in the development of germplasm suitable for these areas.
Received for publication October 1, 2001.
|
REFERENCES
|
|---|
- Barreto, H.J., G.O. Edmeades, S.C. Chapman, and J. Crossa. 1997. The alpha lattice design in plant breeding and agronomy: Generation and analysis. p. 544–551. In G.O. Edmeades et al. (ed.) Developing drought and low N-tolerant maize. Proceedings of a Symposium, El Batan, Mexico. 25–29 March 1996. CIMMYT, Mexico D.F.
- Byerlee, D., and P. Moya. 1993. Impacts of international wheat breeding research in the developing world, 1966–90. CIMMYT, Mexico D.F.
- Calderini, D.F., M.F. Dreccer, and G.A. Slafer. 1995. Genetic improvement in wheat yield and associated traits. A re-examination of previous results and the latest trends. Plant Breed. 114:108–112.
- Derera, N.F., G.M. Bhatt, and F.W. Ellison. 1982. Two decades of wheat improvement work at Narrabri. J. Aust. Inst. Agric. Sci. 48:131–140.
- Fischer, R.A., and P.C. Wall. 1976. Wheat breeding in Mexico and yield increases. J. Aust. Inst. Agric. Sci. 42:139–148.
- Ginkel, van M., R. Trethowan, and B. Cukadar. 2000. A guide to the CIMMYT bread wheat program. Wheat Special report No. 5. CIMMYT, Mexico D.F.
- Kulshrestha, V.P., and H.K. Jain. 1982. Eighty years of wheat breeding in India: Past selection pressures and future prospects. Z. Pflanzenzuchtg 89:19–30.
- Morris, L.M., H.J. Dubin, and T. Pokhrel. 1992. Returns to wheat research in Nepal. CIMMYT Economics Working Paper 92-04. CIMMYT, Mexico D.F.
- Rajaram, S., H.J. Braun, and M. van Ginkel. 1996. CIMMYT's approach to breed for drought tolerance. Euphytica 92:147–153.
- Rajaram, S., and M. van Ginkel. 1996. Yield potential debate: Germplasm vs. methodology, or both. p. 11–18. In M.P. Reynolds et al. (ed.) Increasing yield potential in wheat: breaking the barriers. CIMMYT, Mexico D.F.
- SAS. 1997. SAS Institute Inc., Cary, NC.
- Sayre, K.D., S. Rajaram, and R.A. Fischer. 1997. Yield potential progress in short bread wheats in northwest Mexico. Crop Sci. 37:36–42.
- St. Martin, S.K., and B.A. McBlain. 1991. Procedure to estimate genetic gain by stages in multistage testing programs. Crop Sci. 31:1367–1369.[Abstract/Free Full Text]
- Trethowan, R.M, J. Crossa, M. van Ginkel, and S. Rajaram. 2001. Relationships among bread wheat international yield testing locations in dry areas. Crop Sci. 41:1461–1469.[Abstract/Free Full Text]
This article has been cited by other articles:
|
|
|
|
 
R. M. Trethowan and A. Mujeeb-Kazi
Novel Germplasm Resources for Improving Environmental Stress Tolerance of Hexaploid Wheat
Crop Sci.,
July 1, 2008;
48(4):
1255 - 1265.
[Abstract]
[Full Text]
[PDF]
|
|
|
|
|
|
|
S. Dreisigacker, P. Zhang, M. L. Warburton, M. Van Ginkel, D. Hoisington, M. Bohn, and A. E. Melchinger
SSR and Pedigree Analyses of Genetic Diversity among CIMMYT Wheat Lines Targeted to Different Megaenvironments
Crop Sci.,
March 1, 2004;
44(2):
381 - 388.
[Abstract]
[Full Text]
[PDF]
|
|
|
TOP
ABSTRACT
INTRODUCTION
