Crop Science Journal of Natural Resources and Life Sciences Education
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF) Free
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via ISI Web of Science (18)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Smale, M.
Right arrow Articles by Crossa, J.
Right arrow Search for Related Content
PubMed
Right arrow Articles by Smale, M.
Right arrow Articles by Crossa, J.
Agricola
Right arrow Articles by Smale, M.
Right arrow Articles by Crossa, J.
Related Collections
Right arrow Crop Genetics
Right arrow Plant Genetic Resources
Crop Science 42:1766-1779 (2002)
© 2002 Crop Science Society of America

PERSPECTIVES

Dimensions of Diversity in Modern Spring Bread Wheat in Developing Countries from 1965

M. Smale*,a, M. P. Reynoldsb, M. Warburtonb, B. Skovmandb, R. Trethowanb, R. P. Singhb, I. Ortiz-Monasteriob and J. Crossab

a International Plant Genetic Resources Institute (IPGRI), Rome, International Food Policy Research Institute (IFPRI), 2033 K St. N.W., Washington, DC 20006
b International Maize and Wheat Improvement Center (CIMMYT), km 45 Carretera Veracruz, Texcoco, Mexico

* Corresponding author (m.smale{at}cgiar.org)


   ABSTRACT
TOP
 NOTES
 ABSTRACT
INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION I:...
 RESULTS AND DISCUSSION II:...
 CONCLUSION
 REFERENCES
 
Diverse and varied crop genetic resources are necessary (though not sufficient) for adequate food production in a rapidly changing world. Since the scientific community first raised public concern several decades ago, modern cultivars have been viewed as the cause of declining diversity in the world's crop genetic resources. This paper tests the hypothesis of increasing genetic uniformity in modern spring bread wheat (Triticum aestivum L.) cultivars from 1965, a year which marks the release of some of the first modern semidwarf cultivars carrying Rht1 and Rht2 genes in the developing world. Results from previously published studies are summarized. Preliminary molecular analyses, and new analyses of cultivar numbers, areas, ages, and genealogies are presented. An estimated 77% of the spring bread wheat area in the developing world today is sown to CIMMYT-related wheats, but this does not imply that they are genetically uniform. The hypothesis of increasing genetic uniformity is tested by assessing changes in the diversity of leading progenitors over three decades, in terms of several dimensions of diversity. Latent dimensions include genetic distance and genealogies. Apparent dimensions include performance with respect to yield potential, maintenance and stability across management (input use), and growing environments. The data are not consistent with the view that the genetic diversity of modern semidwarf wheat grown in the developing world has decreased over time. Moreover, since national programs in developing countries cross CIMMYT lines with their own materials before releasing them, the genetic diversity in their cultivars is at least as great as that present among CIMMYT lines.


   INTRODUCTION
TOP
 NOTES
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
 RESULTS AND DISCUSSION I:...
 RESULTS AND DISCUSSION II:...
 CONCLUSION
 REFERENCES
 
NEARLY 30 YR HAVE PASSED since in a landmark study, the National Research Council (1972) alerted the scientific community and the public about the dangers of restricting crop improvement to a narrow collection of germplasm. In the USA, fears were raised by an epidemic of corn leaf blight that struck the U.S. corn crop in 1970. The epidemic resulted from genetic uniformity in T male-sterile cytoplasm, in which a mutant form of Bipolaris maydis (Nisikado & Miyake) Shoemaker (= Helminthosporium maydis Nisikado & Miyake) found a welcome home. In that same year, the late Jack Harlan applied the term "genetic erosion" (Harlan, 1972) to describe what he viewed as a diminishing global stock of "landraces," or traditional forms of cultivated crop plants still grown in parts of the developing world. By referring to the stock of crop germplasm as resource economists refer to a nonrenewable natural resource, he drew attention to the economic value associated with rare alleles or unique gene complexes that may be found in such landraces.

Genetic erosion became synonymous with the displacement of landraces by modern cultivars. In 1970, Frankel called for urgent collection expeditions to forestall "the loss of ancient patterns of diversity in the Vavilovian centers," since modern cultivars contain "a minimum of genetic variation" and "in many instances have a narrow genetic base" (Frankel, 1970, p. 11). Harlan asserted that the "destruction of genetic resources is caused primarily by the very success of modern plant breeding programs" (Frankel, 1972, p. 212). About a decade later, Hawkes warned that "the breeder, who means well, is destroying by his actions the genetic base for a new generation of varieties" (Hawkes, 1983, p. 109).

Two related economic issues motivated these expressions of concern. One relates to the pattern of cultivars sown by farmers across a crop-producing region and the diversity of the genetic mechanisms that convey their resistance to plant disease. When many farmers sow cultivars that carry the same genetic mechanism of resistance to a plant disease, the crop is vulnerable to an epidemic. Combating epidemics once they occur can be very costly to society both in terms of garnering the resources necessary to control them and the yield losses incurred. A second economic issue relates to the capacity of plant breeders to respond to unforeseen events should they occur, and the loss of potentially valuable alleles in genetic stocks still held on farms. Such alleles can be lost when the seed of cultivars or populations that carry them is sown to diminishing areas. Storing seed samples in ex situ collections is an incomplete solution because these cannot substitute for the cultivars that evolve in the fields of farmers under natural and human selection pressures.

What evidence do we have that genetic erosion has occurred in the case of bread wheat? Brush has argued that the genetic erosion "hypothesis" is "plausible but nowhere documented" (1992, p. 148–149). Assembling conclusive quantitative data on the extent to which genetic erosion has actually occurred and its causes is difficult given definitional problems and the scale of analysis required (Wood and Lenn, 1997, p. 112), especially for highly-bred, widely cultivated crops such as wheat.

What can be said about genetic uniformity in modern plant breeding? This article presents scientific evidence on the scope of the genetic base in modern cultivars of spring bread wheat in the developing world from 1965 to the present, drawing on previously published sources and some new analyses of experimental and survey data. Three decades after the National Research Council published its report, our findings suggest that among the modern spring bread wheat cultivars grown in the developing world, genetic uniformity is not likely to have increased.


   MATERIALS AND METHODS
TOP
 NOTES
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS AND DISCUSSION I:...
 RESULTS AND DISCUSSION II:...
 CONCLUSION
 REFERENCES
 
Hypothesis
Motivated by the assertion that the genetic base of modern cultivars tends to narrow as a result of conventional breeding, we adopted the working hypothesis that genetic diversity in CIMMYT and CIMMYT-related spring bread wheat has decreased over time. Formally, the hypothesis is stated as:

where ßd is a the regression coefficient of a diversity index or indicator on the year of cultivar release. The null hypothesis states that genetic diversity in modern spring bread wheat in developing countries has remained constant or has increased over the past 30 yr. The maintained hypothesis is that it has declined. If the null hypothesis is rejected, the hypothesis that genetic narrowing occurred is maintained. When data are inconsistent with the hypothesis of genetic narrowing, we fail to reject the null hypothesis.

In the first part of Results, we characterize the spring bread wheat cultivars grown in the developing world today in terms of their spatial, temporal, and genealogical diversity. Though we have no continuous time series on the specific cultivars grown in the many wheat-producing areas of the developing world, we are able to draw on extensive experimental analyses of the leading progenitors from which these cultivars were selected. In the second section of Results, we use these data to test the null hypothesis by assessing changes that occurred in major progenitors between 1960 and 1990.

We examine both latent and apparent dimensions of diversity. Dimensions are perspectives, or ways of viewing diversity. By latent diversity, we refer to the genealogical and molecular measurements that are not necessarily expressed in a crop's performance. We call performance across environments and management regimes apparent diversity, gauging it by yield potential and stability, yield maintenance through resistance to one of the major diseases of wheat [leaf rust cause by Puccinia recondita Roberge ex Desmaz. f. sp. tritici (Eriks. & E. Henn.) D.M. Henderson], yield progress in dry and warm environments, and nitrogen use efficiency. In each dimension, we assess or test with a regression parameter (ßd) whether improvements have been made in performance over time, other factors held constant. We do not attempt to establish a direct relationship between changes in performance and changes in latent diversity.

It may be important to note that in the literature on diversity, researchers would typically select a particular index which represents in scalar terms the genetic diversity present in a set of populations or cultivars, comparing values for the index between sets defined by time or location. For the purposes of this research synthesis, we have adopted a less conventional, more conceptual approach.

Definition of Modern (CIMMYT-Related) Spring Bread Wheat
Although over half of the world's wheat crop is produced on the large, mechanized farms of industrialized countries, most of the world's wheat farmers live in the developing world, where the popularity of the "green revolution" cultivars first attracted concern among conservationists in the early 1970s. By green revolution, we refer specifically to the development of semidwarf wheat cultivars that spread rapidly throughout South Asia during the late 1960s and early 1970s. These cultivars contained the Rht1 or Rht2 genes, two of several dwarfing genes in the wheat gene pool. The genes were introduced into Japanese breeders' materials through Daruma, believed to be a Korean landrace (Dalrymple, 1986). Norin 10, a cross descended from Daruma, was introduced into a breeding program at Washington State University (USA) in 1949. In Mexico, N.E. Borlaug and colleagues subsequently bred the dwarf characteristic from Norin 10 into the first green revolution wheat cultivars.

When grown with optimal quantities of fertilizer and a controlled water supply, these semidwarf cultivars responded well and performed significantly better than either the landraces or tall cultivars they replaced. Initially, they spread throughout the irrigated zones most favorable to wheat production, replacing the earlier tall cultivars released by plant breeding programs. Later, more widely adapted descendants of these cultivars spread gradually into less favorable growing environments, including rainfed areas with a relatively modest production potential, where they were more likely to have replaced landraces.

Many of the semidwarf wheat cultivars grown in developing countries today are descendants of those green revolution wheat cultivars. Since CIMMYT was established in 1966, the mandate of its wheat program has been the development of spring wheat (both durum and bread wheat) lines for breeding programs in the developing world. A winter and facultative subprogram was initiated during the 1980s. Nurseries consisting of anywhere from dozens to hundreds of advanced lines are sent routinely by CIMMYT to national agricultural research programs that request them for testing and selection (Fox, 1996).

Each nursery contains progeny of various parental combinations. From these materials, scientists in the national research programs in developing countries choose lines demonstrating the best adaptation to local conditions, select from them or cross them to elite local germplasm, and submit the resulting materials to national trials. Superior genotypes are then released as finished cultivars. The vast array of spring bread wheat cultivars developed in this way and released by national wheat research programs from 1966 to 1997 is referred to in this paper as CIMMYT-related. A CIMMYT-related, spring bread wheat cultivar has at least one CIMMYT ancestor, and will generally carry Rht1 or Rht2 genes. The term CIMMYT-related is therefore synonymous with modern for the spring bread wheat grown in the developing world with the exception of China, where other types of semidwarf spring bread wheat are also grown. Characterizing a modern spring bread wheat cultivar as CIMMYT-related identifies it as the product of a certain type of crop improvement process, and enables us to link the materials released and grown in the fields of farmers in the developing world to experimental analyses of the performance of leading progenitors.

Data Sources
Our analysis of cultivars released by national breeding programs and those grown by farmers is representative of the area sown to bread wheat with spring habit in the developing world, comprising around 72 million hectares in the late 1990s (Heisey et al., 2002). About two-thirds of the total wheat area in the developing world is sown to spring bread wheat, with the remainder sown to durum, winter, and facultative habit wheat.

Sources of data about cultivar numbers and areas by cultivar are the CIMMYT Global Wheat Impacts Surveys of 1990 and 1997. In 1990, CIMMYT's Wheat and Economics Programs conducted a survey of wheat research programs in 38 developing countries (37 countries in full, and three provinces of South China) that accounted for 68% of the wheat area in the developing world. The follow-up survey in 1997 covered 36 countries, representing 97% of the wheat area in developing countries. Representation across countries changed slightly between the two surveys (Table 1) , though the main difference between the two surveys in terms of coverage was that in 1997 information was collected for China as a whole.


View this table:
[in this window]
[in a new window]
  		Table 1. Countries included in the CIMMYT Global Wheat Impacts Surveys and in this analysis, 1990 and 1997, by region.{dagger}

 
Both surveys elicited information on the output of breeding programs, including (i) the names, pedigrees, and origins of all spring wheat cultivars released after 1966 and (ii) the estimated area under individual cultivars in 1990 and 1997. Area estimates were based on annual government surveys in some countries, special surveys conducted at the regional or country level, seed sales in some countries, and estimates by wheat researchers. In the 1997 survey, as compared with the 1990 survey, an effort was also made to differentiate among semidwarf cultivars, tall cultivars with pedigrees, landraces, and cultivars of unknown ancestry. The definition of CIMMYT-related is also broader in 1997, since it includes not only cultivars with CIMMYT parents but also cultivars with any known CIMMYT ancestor. The set of CIMMYT-related cultivars is therefore larger in 1997 than in 1990. Here, we have excluded China from those analyses that require cultivar name and pedigree information, since these data remain less complete for China relative to other countries surveyed. The data for the 1990 survey are summarized in Byerlee and Moya (1993) and data from the 1997 survey are summarized in Heisey et al. (2002).

While some tables include data calculated from all spring bread wheat cultivars listed in the two survey periods, others include those computed from a sample. This sample, though much larger than that used in the experiments reported in subsequent sections, overlaps it. The sample includes (i) all CIMMYT-related, spring bread wheat cultivars sown to over 250 000 ha in 1990 and 1997 and (ii) a systematic random sample drawn from the list of all CIMMYT-related, spring bread wheat cultivars sown to less than 250 000 ha in 1990 and 1997. The sample of minor wheat cultivars was drawn by stacking the regional lists of all CIMMYT-related, spring bread wheat cultivars for which there are identifiers in the Wheat Pedigree Management System, sorting them by descending order of area sown, and drawing 30 cultivars from both time periods. The names of wheat cultivars included in the analysis are displayed in Table 2 .


View this table:
[in this window]
[in a new window]
  		Table 2. Cultivars{dagger} included in sample of major and minor spring bread wheats grown in 1990 and 1997 in the developing world.

 
Data sources for pedigrees of wheat cultivars include CIMMYT's Wheat Pedigree Management System, a part of the International Wheat Information System that includes pedigree information (Skovmand et al., 1998). All wheat cultivars are identified by cross numbers and selection numbers, or landrace codes. A computer program was developed to transform pedigree information for a set of cultivars with known pedigrees into a matrix of genealogical characteristics such as those summarized below.

In addition to the analyses of cultivar numbers, areas, ages, and genealogies, new analyses conducted for this study include the measurement of genetic diversity in a sample of wheat lines using molecular markers and trend analysis of trial data on heat and drought tolerance. Replicated trials established for a set of major CIMMYT cultivars released over a 30-yr period provided the data for previously published work. The historical cultivars included in these trials and other data analyses assembled for this article overlap but are not identical to those employed in the molecular studies (Table 3) .


View this table:
[in this window]
[in a new window]
  		Table 3. Cultivars evaluated in analyses of historical trial data.

 
Methods
We used a simple description of the count and area shares of cultivars, according to their ancestry, as indicators of spatial diversity. Though many indicators are possible (Magurran, 1988), we believe these to be more concise when comparing data covering such a large geographical area at two points in time. Temporal diversity, which Duvick (1984) has called "genetic diversity in time," refers to the turnover in farmers' fields of cultivars that are products of plant breeding programs. An important means of controlling pathogen evolution and maintaining yield, cultivar turnover in modern wheat in some sense substitutes for the spatial diversity characteristic of heterogeneous populations of wheat landraces (Apple, 1977; Plucknett and Smith, 1986). Temporal diversity, like spatial diversity, is jointly determined by a number of economic, agroecological, and technical factors (Heisey and Brennan, 1991). The area-weighted average age (Brennan and Byerlee, 1991) has the advantage of combining information about cultivar diffusion, in the form of area shares, with cultivar age—and is employed here as an indicator of temporal diversity.

The coefficient of parentage (COP) provides a theoretical estimate of the genetic relationship between two cultivars based on their genealogies, and estimates the probability that a random allele taken from a random locus in one cultivar is identical, by descent, to a random allele taken from the same locus in another cultivar (Falconer, 1981; see also Kempthorne, 1969; Malécot, 1948; Wright, 1922). Calculated with detailed pedigrees and Mendelian rules of inheritance, the pairwise COP has been used extensively in the crop science literature as a measure of similarity between two modern cultivars due to inheritance. There its advantages and disadvantages have also been discussed (for some applications in wheat, see Cox et al., 1986; Mercado et al., 1996; Murphy et al., 1986; Souza et al., 1998, 1994; van Beuningen and Busch, 1997). For example, one assumption is that landrace ancestors are assumed to be unrelated (with a pairwise COP of zero). Perhaps the factor that most limits the utility of COP analysis is that the accuracy of the calculations depends heavily on the completeness of the underlying pedigree data.

The matrix of pairwise COPs among a set of modern cultivars has been used to construct indices of latent genetic diversity (Souza et al., 1994) or genetic distance (Skovmand and de Lacy, 1999). The average coefficient of diversity, or one minus the average coefficient of parentage, can be used to summarize diversity in a set of cultivars grown by farmers. When weighted by area shares, the difference between the average COP and area-weighted average COP expresses the effects of the economic, agroecological, policy, or technical factors determining cultivar diffusion, as is the case with spatial and temporal diversity indices described above. Both indexes range between 0 and 1, where 0 is the COP between two unrelated cultivars and 1 is the COP between a cultivar and itself. We have constructed some of these indices by combining our data on cultivar pedigree and area.

To test the hypothesis regarding genetic narrowing in modern spring bread wheat cultivars over time, we have combined excerpts from results previously published in scientific journals, including Crop Science, with some new analyses. Methods used to analyze yield potential, stability and maintenance, as well as nitrogen-use efficiency and heat and drought tolerance, are found in the original articles cited in the Results section. Methods used to analyze molecular data are reported next.

On the basis of their popularity in the developing world or importance as major progenitors, 27 lines were selected for molecular analysis. This set overlaps those included in analysis of historical trials and other analyses (Tables 2 and 3). DNA was extracted from the leaves of 4-mo-old plants according to the method of Saghai-Maroof et al. (1984); 400 mg of leaf tissue that had been ground to a fine powder in liquid nitrogen were used. DNA was quantified with a DU-65 Spectrophotometer (Beckman Instruments, Inc., Fullerton, CA) and diluted to a concentration of 0.3 µg/µL for storage. CIMMYT's standard procedures were followed (CIMMYT, 2001). DNA was subjected to analysis by means of 32 microsatellite (SSR) markers and four amplified fragment length polymorphic (AFLP, primer–enzyme combination) markers. For the SSRs, amplification occurred in a reaction volume of 20 µL under the following concentration of reagents: 4 µL [10 ng/µL] genomic DNA, 1.6 µL dNTPs [2.5 mM each], 2 µL 10x Taq Polymerase Buffer (Mg free), 0.1 units Taq polymerase, and 0.5 µL glycerol [100%]. Amplification conditions were as follows: 1 cycle of 93°C for 1 min; 30 cycles of 93°C for 1 min, X°C for 2 min (where X is the annealing temperature according to Roder et al., 1995), and 72°C for 2 min; and a final extension of 72°C for 5 min. Electrophoresis of gels of 12% (w/v) polyacrylamide:bis-polyacrylamide (29:1) was used to separate the amplified fragments. Gels were run for 2 h at 250 V and stained by a 0.1% (v/v) silver nitrate solution. For AFLPs, the method of Vos et al (1995) was modified for chemiluminescent fragment visualization (Greg Penner, Monsanto, and Enrico Perotti, CIMMYT, 1998, personal communications).

Thirty-two SSR and four AFLP markers were employed. Fragments of the same molecular weight amplified by each SSR primer or AFLP primer–enzyme combination were scored as either present (1) or absent (0). This data matrix was used to calculate similarity coefficients for each pair of individuals in the study. Simple Matching Coefficients were calculated by means of NTSYS pc 1.7 (Applied Biostatistics, Inc., New York). The coefficients were then analyzed by principle components analysis to view the relationships among lines in three dimensions. Next, genetic distance (1-coefficient of similarity) was regressed on 2-yr increments in release years, beginning with 2 and ending with 26 yr.


   RESULTS AND DISCUSSION I: DIVERSITY OF SPRING BREAD WHEAT GROWN IN THE DEVELOPING WORLD DURING THE 1990s
TOP
 NOTES
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION I:...
RESULTS AND DISCUSSION II:...
 CONCLUSION
 REFERENCES
 
To understand the relevance of our genetic narrowing hypothesis, it is essential to identify the cultivars grown today in the wheat fields of the developing world's farmers. Have wheat landraces been replaced by modern wheat cultivars? If so, what is the nature of these cultivars? This section presents data regarding the ancestry and spatial and temporal diversity of the spring bread wheat cultivars sown today in the developing world.

Diversity in Space
Of the area sown to spring bread wheat by farmers in the developing world in 1997, 78% was sown to modern semidwarf types that are CIMMYT-related, 11% (principally in China) was sown to other semidwarf types, 8% was sown to tall wheat cultivars released by breeding programs, and 3% was sown to cultivars recognized as landraces or those with unknown ancestry. In all of the developing world excluding China, an estimated 86% of the spring bread wheat area in 1997 was sown to CIMMYT-related, modern semidwarf cultivars. Only in the West Asia-North Africa region was the estimated percentage under 80%, because of the relatively large area still planted to wheat landraces or wheat varieties whose names and pedigrees were unknown. In China, only about 36% of the spring bread wheat area in 1997 was sown these semidwarf types (Heisey et al., 2002).

Not counting China, where data on cultivar names and pedigrees are less complete for spring bread wheat, 86% of all spring bread wheat area in the developing world in 1997 and 82% in 1990 were sown to CIMMYT-related cultivars (Table 4) . In 1997, the count of named cultivars of spring bread wheat grown in the developing world was 382, as compared with 310 in 1990 (Table 4). In both time periods, the most widely grown cultivar was a CIMMYT-related, semidwarf type. The dominance of the most popular wheat cultivar in 1997 was clearly less than it was in 1990, as measured by total area planted and as a percent of all area sown to spring bread wheat. However, the 6.3 million hectares sown to Sonalika in 1990 were distributed across a number of countries, while the 4.2 million hectares planted to HD 2329 and Inqalab each in 1997 were concentrated in the single countries of India and Pakistan, respectively. The 1997 survey occurred at a point when the area planted to HD 2329 was declining while the area planted to Inqalab was rising.


View this table:
[in this window]
[in a new window]
  		Table 4. Area shares, numbers of cultivars and cultivar dominance by parentage, varieties of spring bread wheat grown in the developing world, 1990 and 1997.{dagger}

 
No time-series data on cultivars grown by farmers over large areas of the developing world have been assembled to permit analysis at the level of detail reported in Table 4 for 1990 and 1997. From data assembled earlier by Byerlee and Moya (1993), we know that the number of cultivars released from the ‘Veery’ cross (cross made in 1974, first cultivar released from a selection in 1981) in developed countries by 1990 was at least twice that of cultivars released from the ‘Mexipak’ cross (cross made in 1957, first cultivar released from a selection in 1966). At the same time, the area planted to all Veery cultivars in that year was only about one-fifth the area once sown to Mexipak. Of all of the cultivars released during the early green revolution period in the 1960s, the highest number were selections from the Mexipak cross and ‘Sonalika’ cross (cross made in 1961, first cultivar released from a selection in 1967). The Veery cross, which widened the gene pool of spring bread wheat through the 1BL/1RS wheat-rye (Secale cereale L.) chromosomal translocation, was the leading cross among cultivars released during the 1980s. The Veery cross also generated the largest proportion of cultivars grown in the developing world during the 1990s of any cross, having been selected and released under many cultivar names by numerous national programs. Sister lines may be diverse, of course, with respect to particular alleles, such as those conferring biotic resistance (Skovmand et al. 1997). Cultivars selected from the crosses Attila (cross made in 1984, first cultivar released from selection in 1995) and Kauz (cross in 1980, first cultivar released from selection in 1988) are now of growing popularity.

Diversity in Time
For the developing world as a whole, the average age of spring bread wheat cultivars is similar when measured at such a small interval of time (Table 5) . In either year, for any category of ancestry and when weighted by area or not, the average age of all cultivars of spring bread wheat grown by farmers was 9 to 11 yr. The percentage of CIMMYT-related cultivars older than the average seems to have decreased slightly between 1990 and 1997. Among regions, in 1990, the highest rates of turnover among cultivars were found in Mesoamerica and the Southern Cone of Latin America, with the lowest in the Andean and South Asia regions. In 1997, Mesoamerica was the least temporally diverse. The oldest cultivars from CIMMYT crosses were found in South Asia in either year (Mexipak and T4). In both years, the oldest cultivar was Florence Aurore, still sown in the region of West Asia and North Africa.


View this table:
[in this window]
[in a new window]
  		Table 5. Temporal diversity of cultivars of spring bread wheat grown in the developing world, 1990 and 1997.

 
Genealogical Diversity
The level of diversity due to inheritance among cultivars of spring bread wheat appears high at this scale of analysis (Table 6) , since all average coefficients of parentage are far below the level often used for sisters from the same cross (0.5625 = 0.75 x 0.75, where 0.75 is the COP between a cultivar and a selection from a cultivar, as in Souza et al., 1994). The average and area-weighted averages are almost identical between the two time periods for each size class, and the differences between size classes in each period are not statistically significant. Nonparametric tests show that the frequency distributions of the pair-wise coefficients of parentage are not significantly different between the major cultivars grown in either of the two years. However, in both 1990 and 1997, frequency distributions differ statistically between major and minor cultivars (data not shown, reported in Table 6).


View this table:
[in this window]
[in a new window]
  		Table 6. Pedigree characteristics and coefficients of parentage for cultivars of spring bread wheat grown in the developing world, 1990 and 1997.

 
The hypothesis that the average coefficient of parentage among cultivars increases with their popularity is tested in Fig. 1 across all area classes ranging from 4.2 million hectares to several hundred hectares. Since averages are sensitive to the dimension of the matrix, area classes are irregular in definition to ensure that each matrix includes the same number of cultivars (20). The null hypothesis that the regression coefficient equals zero cannot be rejected, suggesting that there is no systematic relationship between the midpoint of an area class and its average coefficient of parentage. This result may be explained by the fact that minor cultivars include once popular ones now on their way out of production, newly released materials, and cultivars whose optimal area sown is limited by farmer demand or seed supply. For example, sisters selected from the popular Veery cross are found in each area class.



View larger version (18K):
[in this window]
[in a new window]
  		Fig. 1. Average coefficient of parentage of cultivars of spring bread wheat grown in the developing world in 1997, by area class. Calculated from 1997 Global Wheat Impacts survey data and CIMMYT Wheat Pedigree Management System. Note that classes are defined in order to assure COP matrices of equal size. About 300 cultivars for which pedigree information is known are included.

 

   RESULTS AND DISCUSSION II: DIMENSIONS OF DIVERSITY IN CIMMYT BREAD WHEAT OVER TIME
TOP
 NOTES
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION I:...
 RESULTS AND DISCUSSION II:...
CONCLUSION
 REFERENCES
 
Latent Diversity: Genealogy
Modern spring bread wheat cultivars have a diversity of materials in their genetic background. Over the past 30 yr, sources of the germplasm introgressed by CIMMYT wheat breeders have included breeding programs throughout the world, and parents have included cultivars with pedigrees as well as named landraces. These lines are then distributed to national programs and crossed with local materials. Wide crossing has been used to introduce genes and in the case of synthetic hexaploids, entire genomes from related species (Mujeeb-Kazi and Hettel, 1995, 1996), and the resulting materials entered in the crossing blocks of the CIMMYT wheat program.

The international exchange of breeding materials increases the likelihood of introducing new ancestors to the genetic background of the materials crossed by breeders, though often the genetic contribution of landrace progenitors may be small. First, CIMMYT-related semidwarf cultivars released by national programs in developing countries have larger numbers of different landrace ancestors in their pedigrees than those that are not. Of the 1749 spring bread wheat cultivars released by developing countries from 1965 to 1997, those with pedigree data numbered 1162. Of those with pedigree data, an estimated 87% had at least one CIMMYT progenitors. The average number of distinct landrace ancestors per pedigree was 45 for those with at least one CIMMYT progenitor as compared with only 19 for those with no CIMMYT ancestry. When plotted against year of release, the number of unique landraces per pedigree has increased significantly (P < 0.01) by an average of one landrace each year (Fig. 2) . If breeders were unable to introduce materials with distinct ancestry into their programs and the same ancestors recurred, the slope of this line would be constant.



View larger version (14K):
[in this window]
[in a new window]
  		Fig. 2. Landrace ancestors in spring bread wheat cultivars released by developing countries, 1966 through 1997. Calculations based on CIMMYT 1997 Global Wheat Impacts survey data and pedigree information in CIMMYT Wheat Pedigree Management System. Data were available for 1162 (included here) of the 1749 spring bread wheat cultivars recorded (in the survey data) as released during these years. Coverage is less complete for China and for wheats released in the last few years.

 
Latent Diversity: Molecular Markers
When molecular data from each individual line included in the sample (Table 3) are plotted onto axes that represent the three first principal components, the pattern of genetic relationships among lines released over a 28-yr period emerges (Fig. 3) . Lines tend to cluster together based on pedigree and year of release, with those released within a short time of one another having more similar pedigrees. By grouping lines on year of release and drawing circles around comparable numbers in each group, it is possible to see that older lines form smaller, denser clusters, while those released more recently form very dispersed clusters.



View larger version (32K):
[in this window]
[in a new window]
  		Fig. 3. Principal components analysis of 27 CIMMYT spring bread wheat lines released from 1962 to 1990. The axes are generated from the first three principal components of 32 SSR and four AFLP markers. Data were analyzed with the program NTSYS pc 1.7. Source: CIMMYT Applied Biotechnology Laboratory.

 
The regression of differences in genetic distance on two-yr increments in release year reveals a significant positive correlation (P < 0.05). The greater the difference in release year, the greater the genetic distance between any pair of wheat lines (Fig. 4) . The estimated regression slope indicates an average increment in genetic distance of 0.09% per unit increment in year of release. Over the 28-yr period, genetic distance increased by 2.3%.



View larger version (30K):
[in this window]
[in a new window]
  		Fig. 4. Regression of difference in genetic distance, based on 32 SSR and four AFLP markers, and year of release for 27 CIMMYT spring bread wheat lines released from 1962 to 1990. Source: CIMMYT Applied Biotechnology Laboratory.

 
Both figures are based on a small sample size of 27 major progenitors of the CIMMYT-related wheats described above. The findings are consistent with Skovmand and de Lacy's genealogical analysis (1999) and the molecular and genealogical studies of Almanza-Pinzon (2000), who found that in each decade, CIMMYT lines of spring bread wheat represented distinct genetic groups. A much more extensive review of cultivars and breeding lines is currently underway in the CIMMYT Applied Biotechnology Center with a larger sample of SSR markers, but has not yet been published. Preliminary results with the larger data set do appear to confirm those presented here (Pingzhi Zhang and Marilyn Warburton, 2001, personal communication).

Apparent Diversity: Yield Potential and Stability
Increasing latent diversity, as expressed by genealogical indicators and indices based on molecular data, has occurred at the same time that wheat yield potential and stability has improved. Byerlee and Moya (1993) and Rejesus et al. (1999) surveyed the published literature on global progress in wheat yield potential and reported for the most part positive results. In the latest analysis of progress in yield potential of CIMMYT bread wheat, Sayre et al. (1997) estimated an average annual linear increase of 67 kg ha-1 yr-1, representing a 0.88% rate of progress per year from 1960 to 1990 (Fig. 5) . This rate of progress is similar to that reported in earlier studies summarized by Byerlee and Moya (1993), including two other studies conducted in the Yaqui Valley (Fischer and Wall, 1976; Waddington et al., 1986). Their study was based on 6 yr of data with three to four replications in each year, including eight outstanding cultivars of those included in the historical trials (Table 3).



View larger version (18K):
[in this window]
[in a new window]
  		Fig. 5. Grain yield (6-yr mean) of eight wheat cultivars as a function of year of release. Cultivars were grown under irrigation and optimal management, including disease and lodging protection, in each of six winter growing seasons (1990–1995) at CIANO (Centro de Investigaciones Agricolas de Noreste) experiment station, Sonora, northwest Mexico. Cultivars were derived from CIMMYT germplasm, and are listed in Table 3. Reprinted with permission from Sayre et al. (1997).

 
Sayre et al. (1997) also found that deviations from the regression line tended to decrease with yield progress. Six-year means were very stable for irrigated conditions in northwest Mexico and by inference, other temperate, low-latitude, irrigated environments. Stability was pronounced for the latest cultivars included in the study, Oasis 86 and Bacanora 88.

Sayre et al.'s (1997) findings support earlier research. Ten years ago, Pfeiffer and Braun (1989) presented an analysis based on data from the First to the Fifteenth International Spring Wheat Yield Nurseries (ISWYNs), which were distributed by CIMMYT during the period 1965 to 1980 and grown in sites representative of major wheat-growing environments in the developing world. The ISWYN is a standardized international yield nursery consisting of three replications of 49 spring bread cultivars and advanced lines, plus one local check. Pfeiffer and Braun subdivided the genotypes according to the extent of their CIMMYT ancestry and analyzed their stability using several types of statistical models. The most stable of the groups analyzed by Pfeiffer and Braun included genotypes with initial crosses made by CIMMYT, and at least one further selection made by national programs (p. 164). In other words, the ancestral diversity that is enhanced by the selections and crossing activities of national programs was reflected in greater stability. Neither materials with exclusively local development nor CIMMYT crosses released as cultivars were as stable.

Though yield stability of individual cultivars as analyzed in trial data is related to but does not necessarily imply stability in crop yields across production environments and countries, the evidence regarding wheat production variability is also fairly conclusive. While the coefficient of variation of world cereal production adjusted for trend rose by 21% between the 1960 to 1970 decade and the 1970 to 1980 decade, the coefficient of variation in wheat production decreased by 11.5% (Hazell, 1989, p. 16). Smale found that the coefficient of variation of world wheat yields around trend was less in the most recent decade (1985–1994) than it was during the 1955 to 1964 period (1998, p. 92). Since the 1950s, then, the evidence suggests that wheat yields have become more stable even as mean yields have increased. The same holds true for South Asia (Singh and Byerlee, 1990), and for the major wheat-producing regions of the developing world (Smale, 1998).

Apparent Diversity: Input Use Efficiency
Improved yield potential and stability has not been at the expense of increased input requirements. A common misconception is that early semidwarf wheat cultivars did not perform well in the absence of nitrogen fertilizer (Simmonds, 1979), and farmers would therefore be better off growing their old tall cultivars if no fertilizer was available. The popular viewpoint has evolved because semidwarf cultivars of wheat have a better response to nitrogen and therefore the economically optimum rate of application is higher.

When Pfeiffer and Braun (1989) subdivided the materials in their study by environments, the most important CIMMYT cross made in the 1980s (Veery) was higher yielding across all environments than either the best locally developed cultivar or Siete Cerros, the most popular semidwarf wheat of the early green revolution. On the basis of experimental results comparing Yaqui 50 and Nainari 60 (tall wheats) to Veery and 12 other CIMMYT advanced lines, they found that the later materials produced yields under water stress, zero input, weed-free, and weedy conditions that were at least as high or higher than those given by the tall cultivars. They concluded that semidwarf wheats of CIMMYT origin were both "input efficient and input responsive." (p. 164).

Ortz-Monasterio et al. (1997) have further shown that modern spring wheat cultivars produce higher yields than older, tall cultivars under both high or low N fertility conditions in the Yaqui Valley of Mexico. Other researchers elsewhere have also shown that semidwarf wheat types either yield the same or more than tall cultivars under low nitrogen fertility conditions (Jain et al., 1975; Wall et al., 1984; Entz and Fowler, 1989; Austin et al., 1993). Semidwarf cultivars do not require more nitrogen; in fact they often need less N to produce the same yield than old tall cultivars (Fig. 6) .



View larger version (32K):
[in this window]
[in a new window]
  		Fig. 6. Mean genetic progress of ten bread wheat cultivars, two tall and eight semidwarfs, produced by CIMMYT and released by the Mexican government from 1950 to 1985. Cultivars were planted over 3 yr at four levels of applied N, 0, 75, 150, or 300 kg N per ha at the CIANO (Centro de Investigaciones Agricolas de Noreste) experiment station in Sonora, northwestern Mexico. Cultivar names are listed in Table 3. Reprinted with permission from Ortiz-Monasterio et al. (1997).

 
Breeding under medium to high N fertility conditions, the performance of modern spring wheat cultivars from 1950 to 1985 has improved when they are grown under either high or low N fertility. In semidwarf wheat, this increase in grain yield has been associated with gains in uptake or utilization efficiency, both components of nitrogen use efficiency. At medium to high levels of N fertility in the soil, uptake and utilization efficiency have been improved; at low levels of N fertility, only uptake efficiency has increased (Ortz-Monasterio et al., 1997). In theory, the nitrogen level in the soil could be manipulated together with the genetic diversity of the crop as a breeding tool for the development of wheat cultivars with enhanced uptake and/or utilization efficiency. Genetic variability for nitrogen use efficiency in wheat has been reported (Dhugga and Waines, 1989; van Sanford and MacKown, 1986; Ortz-Monasterio et al., 1997).

Apparent Diversity: Genetic Resistance to Disease
Yield protection in farmers' fields has also improved through advances in genetic resistance to disease. Genetic resistance, rather than use of fungicides, remains the principal means of controlling the rust diseases of wheat—especially in the wheat-producing areas of the developing world. During the past 25 yr, the bread wheat program has emphasized selection for nonspecific resistances, as defined theoretically by Vanderplank (1963) and applied to leaf rust resistance by Caldwell (1968).

Genes conferring race-specific resistance tend to produce resistant reactions, but their effects are overcome in a relatively short period of time. In contrast, genes conferring race-nonspecific resistance to leaf rust in wheat have partial and additive effects, and although the response to infection is essentially susceptible, the rate of disease progress is slowed. Geneticists and pathologists at CIMMYT now believe that adequate levels of nonspecific resistance can limit disease losses to insignificant levels under farmers' conditions. In addition, nonspecific resistance is more likely to endure for many cropping seasons than race-specific resistance. Sayre et al. (1998) demonstrated that while the grain yield potential of modern spring bread wheat has increased significantly over the past 30 yr, progress in protecting this yield potential through the incorporation of genes that confer slow rusting resistance to leaf rust disease has been more dramatic. The average annual progress in grain yield over the four trials planted at normal dates was estimated to be 0.52% in plots protected with fungicides, and 2.07% in unprotected plots (Fig. 7) . Using the data summarized in Sayre et al. (1998), Smale et al. (1998) found that the economic benefits of breeding for nonspecific resistant to leaf rust in bread wheat are likely to have been substantial.



View larger version (32K):
[in this window]
[in a new window]
  		Fig. 7. Relationship between year of release of 15 cultivars and their grain yields under fungicide protected and non-protected conditions for normal planting dates. Cultivars were sown at normal planting dates over four seasons at CIANO (Centro de Investigaciones Agricolas de Noreste) experiment station, Sonora, northwest Mexico. Leaf rust epidemics were established by inoculating spreader rows planted adjacent to plots of cultivars which were not protected by fungicide. Cultivars were derived from CIMMYT germplasm, and their names are listed in Table 3. Adapted with permission from Sayre et al. (1998). g = annual genetic progress.

 
While gene Lr34 is important for slow rusting and is known to be present in number of the wheats included in the Sayre et al. (1998) study, other genes involved in durable resistance (Johnson, 1984) appear also to be present in several of them (Rajaram et al., 1997). One such gene (Lr46) was recently identified in the cultivar Pavon F 76 (Singh et al., 1998). Rajaram et al. (1998) identified at least 12 different slow rusting genes for leaf rust resistance in a set of 10 cultivars, with two or three genes usually present in combination. Singh and Rajaram (1991) identified an additional 13 named race-specific genes in cultivars derived from CIMMYT germplasm that were released in Mexico. Another 10 named race-specific genes have been detected in CIMMYT germplasm. Taken together, these findings suggest that both a large number of gene combinations and a range in types of reaction to rust disease are present among the materials sent through international nurseries to national wheat research programs in developing countries.

Apparent Diversity: Heat and Drought Tolerance
Recent findings are also encouraging with respect to yield performance of modern bread wheat across environments, including warm and dry areas. Regression analysis of average yield performance of selected cultivars across 16 environments on year of release demonstrates that yield potential in warm environments has improved by approximately 40% between 1965 and 1985 (Fig. 8) . More recent cultivars have higher crop biomass and increased photosynthetic rate (Reynolds et al., 1994), indicating that they are physiologically better adapted to higher temperatures, in addition to being well adapted in terms of phenological development. Though the physiological basis of improved heat tolerance is complex and is not completely understood, one of the most interesting traits associated with heat tolerant cultivars is the ability to maintain cooler leaves through increased evapotranspiration rate at critical times of the day when the heat load is highest (Amani et al., 1996). The identification of physiological traits associated with heat tolerance provides a basis for screening germplasm collections for new sources of genetic diversity for the trait (Hede et al., 1999)



View larger version (22K):
[in this window]
[in a new window]
  		Fig. 8. Regression of grain yield of 10 historical CIMMYT spring bread wheat lines averaged for 16 hot environments on year of release. Analysis based on data from trials conducted from 1990 through 1992, published in Reynolds et al. (1994). Cultivars were derived from CIMMYT germplasm, and their names are listed in Table 3.

 
Figure 9 shows the mean performance of the five highest yielding lines of the ESWYT (Elite Spring Wheat Yield Trial) and Semi-Arid Wheat Yield Trial (SAWYT) for the period 1989 through 1997, compared with the mean of the local check cultivar averaged across environments. The local check cultivar is the best performing local cultivar and differs among sites and within sites over time. The SAWYT, containing lines bred specifically for drought conditions, was first deployed globally by CIMMYT in 1992. Before 1992, researchers working in moisture stressed areas grew and selected germplasm from the ESWYT, which contains lines developed specifically for high yield potential environments. Performance data of ESWYT's 11, 12, and 13 were included in the analysis as these represent performance data prior to the introduction of the SAWYT in 1992, and provide a reference point for SAWYT comparisons. These data represent 230 trials each sown to between 30 and 50 genotypes between 1989 and 1997. Genotypes included in each trial were the same within each year but differed between years. A comprehensive analysis of these data, including a greater range of genotypes, years and sites has been conducted by Trethowan et al. (2002).



View larger version (15K):
[in this window]
[in a new window]
  		Fig. 9. Regression against time of the mean yield of the top five entries of CIMMYT-derived germplasm expressed as percent of local check for the Elite Spring Wheat Yield Trial (ESWYT 1989–1991) and Semi-Arid Wheat Yield Trial (SAWYT 1992–1997). Between 30 and 50 genotypes were sown in 230 trials conducted between 1989 and 1997. Average yields are less than 2.5 Mg/ha in low-yielding environments and 2.5 to 4.5 Mg/ha in medium-yielding environments. Adapted from Trethowan et al. (2002).

 
Yield trends suggest that under low yielding conditions (less than 2.5 Mg/ha), there has been a significant improvement in yield over time (r2 = 0.63) following the introduction of the targeted SAWYT nursery. Furthermore, in the less stressed years (average yield between 2.5 and 4.5 Mg/ha) progress in yield over time has also been significant (r2 = 0.39). This is an important observation for yield stability in dry areas. When the environment is favorable, as it is in some years, these cultivars respond to available moisture and fertilizer.


   CONCLUSION
TOP
 NOTES
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION I:...
 RESULTS AND DISCUSSION II:...
 CONCLUSION
REFERENCES
 
In one respect, the predictions of Harlan and Hawkes are correct: the products of modern plant breeding, for spring bread wheat in particular, have today largely replaced "the ancient patterns of diversity" in the fields of farmers in the developing world. Both modern and traditional patterns of variation coexist, but on vastly different scales. Only an estimated 3% of the spring bread area in the developing world in 1997 was sown to landraces (though in an area that is minor on a global scale can be found small niches and pockets with great genetic variation). Not counting China, where other types of semidwarf cultivars are grown but our data are incomplete, 89% of all spring bread wheat area in the developing world was sown to CIMMYT-related cultivars. In China, about 36% of the spring bread wheat cultivars grown in 1997 were CIMMYT-related.

These modern spring bread wheat cultivars are not genetically uniform, however. First, the contribution of any single CIMMYT line can vary considerably in the presence of the many other ancestors in the cultivar's pedigree. Secondly, CIMMYT lines are themselves constituted by genetic recombination of diverse germplasm from throughout the wheat-growing world. No systematic relationship between the popularity of cultivars (area planted) and genetic similarity due to ancestry is evident for the cultivars grown in 1997 in developing countries. The number of distinct landrace ancestors in the pedigrees of over one thousand cultivars of spring bread wheat released by national agricultural research systems in the developing world since 1966 increases positively and significantly with year of release. In addition, CIMMYT-related cultivars released by national programs have a significantly higher number of different landrace ancestors in their pedigrees than those with no known CIMMYT ancestry. These findings reflect the fact that germplasm with diverse genetic backgrounds is continually brought into the crossing blocks of CIMMYT and national partners through an international system, revealing the extent of the international exchange on which the cultivars now grown in the wheat fields of the developing world have depended.

Preliminary analyses suggest that the molecular genetic diversity of major CIMMYT progenitors of these spring bread wheat cultivars has not decreased over the past 30 yr. Since national programs in developing countries cross CIMMYT lines with their own materials before releasing them, the genetic diversity in their cultivars is at least as great as that present among CIMMYT lines.

We then presented evidence that any increases in genetic and genealogical diversity have occurred at the same time that performance has improved for the major CIMMYT progenitors. The data show improved yield potential, maintenance, and stability. While advances in yield potential have continued, advances in yield maintenance through genetic resistance to disease have been greater. Progress has been achieved in tolerance to heat and drought, contributing to yield stability in more difficult growing environments. In each decade, leading cultivars have required smaller and smaller amounts of land and lower rates of nitrogen application to achieve the same level of output.

The data presented here are not consistent with the hypothesis of increasing genetic uniformity in the modern (semidwarf) spring bread wheat grown in the developing world. Although some argue that de novo variation can be generated without new ancestral material (Rasmusson and Phillips, 1997), our consensus view is that improvements in latent diversity and performance over time in modern spring bread wheat have relied on maintaining genealogical and genetic diversity. This has been accomplished through the open exchange of large numbers of diverse materials among the wheat breeding programs of the world, most of which are public.


   ACKNOWLEDGMENTS
 
The authors thank Mireille Khairallah and Martha Isabel Almanza Pinzon for contributions to the molecular work reported here. We thank Tony Fischer, Paul Heisey, David Hoisington, Prabhu Pingali, and Sanjaya Rajaram for guidance and encouragement. The comments of three anonymous reviewers were very helpful in preparing the manuscript. The Global Wheat Impacts Survey data are based on responses from wheat researchers in national agricultural research institutions, and were collected and analyzed by CIMMYT's Wheat and Economics Programs. Paul Heisey, Jesse Dubin, and Mina Lantican organized and implemented the survey in 1997. Derek Byerlee and Piedad Moya implemented the survey in 1990. Mina Lantican assembled some of the 1997 data for the analysis presented in this report, and Kelly Nightingale prepared some of the 1990 data for further analysis. Jesper Norgaard ran the program to calculate many of the coefficients of parentage. The program is part of the International Wheat Information System, a public database. The authors express their appreciation to their colleagues in national agricultural research systems all over the developing world who provided the data and to CIMMYT staff members who were involved in conducting the surveys in 1990 and 1997 (see Byerlee and Moya, 1993, and Heisey et al., 2002, for a complete listing). They also thank Alma McNab, Kelly Cassaday, and Miguel Mellado for their assistance.


   NOTES
TOP
 NOTES
ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION I:...
 RESULTS AND DISCUSSION II:...
 CONCLUSION
 REFERENCES
 
When this research was conducted, all authors were employed by CIMMYT, Texcoco, Mexico.

Received for publication November 30, 2001.


   REFERENCES
TOP
 NOTES
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION I:...
 RESULTS AND DISCUSSION II:...
 CONCLUSION
 REFERENCES