Wheat and Barley Production in Rainfed Marginal Environments of ...

ContentsivvForewordAcknowledgements1 Part 1: Wheat and Barley Production in Rainfed Marginal Environments of the Developing WorldI What Is a "Marginal Environment"?2 Distinguishing Types of Rainfed Marginal Environments3 Seven Examples of Rainfed Marginal Environments for Wheat6 Extent of Rainfed Marginal Environments in the Developing World7 Production Trends in Rainfed Marginal Environments11 Rainfed Marginal Environments of West Asia and North Africa19 Rainfed Marginal Environments of South Asia23 Toward a Research Strategy for Marginal Environments29 Part 2: The Current World Wheat Situation29 Wheat Production29 Wheat Utilization29 International Wheat Prices30 Outlook for Wheat32 Part 3: Selected Wheat Statistics33 Eastern and Southern Africa34 West Africa35 North Africa36 West Asia38 South Asia39 Southeast Asia and Pacific40 East Asia41 Mexico, Central America, and Caribbean43 Andean Region, South America44 Southern Cone, South America45 Eastern Europe and USSR46 Developed Market Economies48 Regional Aggregates49 ReferencesAnnex 1: Regions Delineated for This Studyiii

ForewordTwo years ago in the previous issue ofWorld Wheat Facts and Trends, we tooka close look at how the "revolution" inwheat production had spread throughoutthe world in the twenty-odd years sinceit began in Mexico and South Asia. Ourreport revealed that growth in ThirdWorld wheat production had beenimpressive; during the 1980s, wheatyields and production rose almost asrapidly as in the first extraordinary yearsof the Green Revolution. But we alsoobserved that the effects of the wheatrevolution were not equally felt amongthe different production environments inthe developing world. In concluding ourreport we noted that "a continuingchallenge...is to exploit the potential ofdrier environments where the pace ofchange has been slowest," and indicatedthat the next issue of Wheat Facts andTrends would address that topic ingreater detail.To assess the potential for improvingcereal crop production in drierenvironments of the developing world,we decided to focus on two majorgeographical areas where suchenvironments predominate: West Asiaand North Africa (WANA), where barleyas well as wheat is important in rainfedcropping systems; and the areas of SouthAsia where wheat is grown on residualmoisture after the monsoon rains. Oncethe geographical scope of the report wasdetermined, CIMMYT approached theInternational Center for AgriculturalResearch in the Dry Areas (lCARDA)about the possibility ofjointlydeveloping the study to incorporateICARDA's considerable experience inthe WANA Region. This study isstronger because of that collaboration. Inaddition, our colleagues with the IndianCouncil of Agricultural Researchcontributed valuable information to thesections of this report on South Asia.By focusing our attention on wheat andbarley production in the Third World'smarginal rainfed environments, we havehighlighted some of the most acutedilemmas confronting national andinternational research in recent years.The fact that agricultural change hasbeen slow in marginal areas, whereenvironmental degradation is often saidto be proceeding most rapidly, haspresented research with an urgentimperative: to effect positive change inagriculture and incomes in less favoredareas without endangering theirecological stability.As the reader will see, our studydemonstrates that this challenge is farfrom simple. The tradeoffs in decidinghow to allocate research resources to thegreatest benefit of marginal areas areinevitably complex, since they involveweighing the potential benefits that maycome from devoting research resourcesto favored areas (such as lower foodprices or increased employmentopportunities) against the benefits ofresearch focused specifically onmarginal areas. Furthermore, althoughthe following pages strongly advocategreater attention to research for rainfedmarginal areas, the research strategiesthat met with success in more favoredareas are not necessarily best suited tothe rainfed marginal areas where wheatand barley are produced. More work oncrop and resource management relativeto plant breeding research will be a keyto mitigating problems in marginal areaswhose agriculture is characterized byintercropping, crop rotations andfallowing, and crop-livestockinteractions. Even so, it must berecognized that, in many cases,formidable agroclimatic constraints willlimit the potential for improvingproductiVity despite strong contributionsfrom research.But research does not work in isolation.If research is to have an impact inmarginal areas, extension will need to beable to present farmers with a range oftechnical options. Farmers will requiregreater assistance from extension toacquire the managerial skills necessaryto adopt increasingly complextechnologies with success.Policy makers will also have to becommitted to strengthening agriculturein marginal areas, perhaps throughpolicies designed to reduce the riskinessof wheat and barley production. Theextreme diversity, variability, andvulnerability of marginal environmentsmake the need for strong collaborationamong research, policy, and extensionmore urgent. We believe that this reportprovides information that will be usefulin promoting collaboration among allwho are interested in effecting positivechange in agriculture in the world'srainfed marginal areas.Donald L. Winkelmann,Director General, CIMMYT~asrat FaddaDirector General, ICARDAiv

AcknowledgementsNumerous individuals contributed to thisedition of World Wheat Facts andTrends. Michael Morris (CIMMYT),Abderrezak Belaid (ICARDA), andDerek Byerlee (CIMMYT) wrote thefeature report on wheat and barleyproduction in rainfed marginalenvironments (Part I). They wereassisted by several colleagues whocontributed original material, includingMark Bell of CIMMYT (crop simulationmodelling), Elizabeth Bailey ofICARDA (farming systems of theWANA region), Tony Fischer ofCIMMYT (the box on factors affectingwater use efficiency), and MitchRenkow of CIMMYT (the box oninterregional effects of technologicalchange). The information on India couldnot have been compiled without theassistance of colleagues at ICAR whogave generously of their time andexpertise, especially J.P. Tandon of theAll India Coordinated Wheat Project, aswell as many wheat scientists in Centraland Southern India, especiallyR.P. Hanchinal, B.S. Jadon,R.P. Sheopuria, A.K. Singh, andV.S. Tomar. Laura Saad, Daphne Taylor,and Stefan Keyler of CIMMYT providedvaluable research assistance.Part 2 was written by Michael Morris.Part 3 was compiled by Laura Saad usingdata provided by FAO, by colleagues inCIMMYT and ICARDA, and bycollaborators throughout the world: D.Adrien, Z. Ahmed, J.M.H. Al Bader, R.G.Annes, N. Banisado,Z. Barabas, M.T. Barradas,A. Benbelkacem, B. Borghi, J. Brennan,0. Chicaiza, J. Davidson, J.A. Dieseth, R.Downey, D. Flores, D. Gogas,R.W. Gomez Quintanilla, A. Gul,M. Hanis, R. Hassan, He, Z.,C. Heustone, J.H. Helm, K. Hjortsholm,A.L. Katunsi, E.!. Kivi, C. Kling,B. Kristiansson, A. de Leon, B.Libereita,J. Lodga, G. Lohan, M. Maina,P.C.J. Maree, J.A. Martin, J. MartInez,B.M. Maumbe, J.M. McEwan,M. McLelland, A. Meinel, D. Metzger, F.Montes, J.H. Nam, J.E. Nisi,R. Pedretti, R.B. Proud, N. Przulj,A. Razzaque, K. Raoudha, !. Rodriguez,M. Saade, M. Salazar, M.M. Saulescu,W. Sowa, L. Tanash, J.P. Tandon,T. Tesemma, M. Trottet,and W.W. Wagoire.We are grateful to the numerousscientists from both CIMMYT andICARDA who reviewed drafts of thereport, including Don Winkelmann,Tony Fischer, Eugene Saari, HansBraun, Bent Skovmand, Roger Rowe,Maarten van Ginkel, Guillermo Ortiz­Ferrara, Byrd Curtis, Sanjaya Rajaram,Edmundo Acevedo, Renee Lafitte, Aartvon Schoonhoven, Rick Tutweiler, HugoVivar, and Elizabeth Bailey. In addition,we would like to thank Kutlu Somel ofthe World Bank for serving as externalreviewer.The publication was edited by KellyCassaday of CIMMYT. Design andproduction were done at CIMMYT byJose Luis Delgado and Miguel Mellado,with the assistance of Rocio Vargas andMa. Concepcion Castro.v

Part 1: Wheat and Barley Productionin Rainfed Marginal Environments of the Developing WorldMichael L. Morris, Abderrezak Belaid, and Derek ByerleeThe events that precipitated the GreenRevolution in wheat are by now wellknown. During the late 1960s and early1970s, improved high yielding wheatsdeveloped in Mexico (referred to in thisreport as modem varieties, or MVs) wereintroduced into a number of the world'smost populous developing countries,including India, Pakistan, and Turkey.When grown with increased levels offertilizer and an assured water supply,MVs performed significantly better thanolder materials, making possibledramatic increases in wheat productionand higher incomes for the millions ofsmall-scale farmers who adopted thetechnology.Despite these successes, the agriculturalresearch institutes credited withlaunching the Green Revolution cameunder attack from critics who suggestedthat its major beneficiaries were farmerslocated in favored productionenvironments. Indeed, some evidencesuggests that farmers in marginalproduction environments who did notadopt the new technology may havebeen disadvantaged when rapidproduction increases in favoredenvironments depressed cereal prices(Lipton 1989). Since these negativeprice effects were not offset byproduction gains in marginalenvironments, argued the critics,interregional inequalities in rural incomedistribution were actually increasedrather than reduced by the GreenRevolution.One problem in attempting to evaluatethis criticism is that the empirical recordis incomplete. Although numerousstudies have documented the impacts ofimproved seed-fertilizer technologies infavored production environments whereearly and widespread adoption tookplace, relatively little research hasfocused on less favored environments.The issue is complicated by the fact that1 Throughout this report, the tenns "marginalenvironments" and "marginal areas" are usedinterchangeably. "Dryland production" refers torainfed production taking place in marginalenvironments.the adoption of technology is often agradual process spanning many years,so that it is sometimes difficult to knowwhen adoption is still in a relativelyearly stage or has completely run itscourse.This Facts and Trends feature reportpresents an up-to-date picture of wheatproduction in marginal environments,with emphasis on the past and future roleof technical change in these areas. IBarley, a close substitute in productionin rainfed marginal areas, is alsoincluded in the analysis. The focus onmarginal environments is motivated bythe fact that a large number of poorpeople currently depend on theseenvironments for their survival, and bythe concern that marginal environmentspresent a special challenge for thedevelopment of production technologiesthat will be sustainable in the longer run.In the following pages, we examinedryland wheat and barley production at anumber of different levels. First, wecharacterize marginal environments forwheat and barley in terms of agroclimaticcircumstances and describe theimportance of these environments in themajor wheat- and barley-producingcountries of the developing world.Second, we provide an overview of theprincipal cereal-based farming systemsin two regions where large amounts ofwheat and barley are produced undermarginal conditions-West Asia andNorth Africa (the WANA Region), andSouth Asia. Third, we pose threequestions that are addressed in theremainder of the report:I) What has been the impact oftechnological change in marginalenvironments over the past threedecades?2) What types of research promise todeliver the most rapid future gains inwheat and barley production inmarginal environments?3) Should a larger share of researchresources be allocated to addressingthe special challenges of cerealproduction in marginal environments,and if so, is a change in researchstrategy needed to address thespecific needs of marginal areas?This report forms part of a larger effortby CIMMYT and ICARDA to betterdefine the extent of marginalenvironments for wheat, barley, andmaize in the developing world; tocharacterize the conditions under whichthese cereals are grown in theseenvironments; and to identify promisingresearch opportunities.What Is a"Marginal Environment"?Although agricultural scientistsfrequently distinguish between "favored"and "marginal" productionenvironments, in fact there is littleconsensus on precise definitions forthese terms. This is perhapsunderstandable, since the growthrequirements of different plant speciesvary considerably, and an environmentthat is favorable for one species may bemarginal for another.Most agriculturalists would probablyagree that a marginal environment forcrop production is characterized byabiotic stresses which severely inhibitplant growth, such as extreme levels ofmoisture (drought or waterlogging),extreme temperatures (heat or cold),severe imbalances in soil fertility(absence of essential nutrients orpresence of toxicities), unfavorable soilphysical characteristics (shallow depth,degraded condition), and/ormeteorological conditions which cancause physical damage to growing plants(high winds, hail). However, thepresence of one or more of these stressesdoes not necessarily mean that anenvironment is marginal. Many of thefactors causing abiotic stresses aresubject to manipulation; indeed, modern

2crop production technologies haveenabled some of the world's mostinhospitable regions to be transformedinto highly productive agriculturalzones.In this report, we adopt a conceptualdefinition of marginal environmentsused by the CIMMYT Wheat Program:marginal environments for wheat andbarley are defined as those in whichirremediable climatic or soil conditionslimit yields to less than 40% of potentialyields as defined by available solarradiation (CIMMYT 1989b). Conditionsare considered irremediable when thecost of ameliorating them is prohibitive,i.e., beyond the ability of a farmer ornation to pay, except perhaps in the verylong run (Fischer 1988).Based on this definition, most marginalenvironments for wheat and barley arecharacterized by severe drought, oftencombined with extreme temperatures(heat or cold). Other types of stresses,such as unfavorable soil chemical andphysical conditions, can also definemarginal environments for wheat andbarley, but the major focus of this reportwill be on moisture stress. Generallyspeaking these environments are locatedin regions that receive less than 350 mmrainfall in the growing season (sometimesmuch less, as in areas where wheatand barley are grown on residualmoisture).Distinguishing Types of RainfedMarginal EnvironmentsIn rainfed cereal cropping systems, waterused by growing plants comes from twoprincipal sources: rain falling during thegrowing season, and residual moisturestored in the soil at planting time. Therelative importance of these two sourcesof water varies depending on precipitationpatterns, soil characteristics, andfarmers' management practices. Twobasic types of rainfed marginal environmentscan be distinguished based on theseasonal distribution of precipitation. Inwinter rainfall environments, moistureavailability is usually closely related torain falling during the growing season(although sometimes significant amountsof water may be stored in the soil atplanting time as the result of fallowingand other practices that facilitatemoisture storage). In theseenvironments, evapotranspiration oftenexceeds rainfall at the beginning and endPrecipitation (mm)of the growing season, when rainfall islower and also highly variable (FigureI). By contrast, in residual moistureenvironments, very little rain falls duringthe growing season, and the crop derivesvirtually all of its water requirementsfrom moisture stored in the soil from thepreceding monsoon rainfall period. In500 ...,.---------------------------,Aleppo 'yria (winter rainran spring habit)400300200Harve lingHarvestingPrecipitation (mm)500 ...,.--------------------------,Ankara. Turkey (winter rainfall. winter habit)PlanungPlanting1000L-~~~~~~Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug400300H

3these environments, potential evapotranspirationgenerally exceeds rainfallthroughout the season (Figure I).The winter rainfall environments areconcentrated in the WANA Region,where they can be further broadlyclassified into Mediterranean climates(in which average minimum temperaturesdo not usually fall below ODC in thecoldest month) and Continental climates(in which, because of elevation anddistance from the coast, a significantproportion of precipitation occurs assnow). Not all winter rainfall environmentsare marginal. Those classified asmarginal usually lie in the zonereceiving between 200 and 400 mm ofprecipitation annually.The residual moisture environments areconcentrated in South Asia. Here annualprecipitation is much higher (often over1,000 mm), but usually less than 100mm falls during the growing season.It is important to distinguish betweenthese two basic types of marginalenvironments, because they areassociated with very different levels ofproduction risk. In all low rainfallenvironments, crop production dependson uncertain and highly variable rainfall.However, where crops are grown onresidual moisture, the amount ofavailable water is already known at thetime of planting, which means thatfanners can modify the area planted andtechnologies used (including use ofpurchased inputs and crop husbandry). Incontrast, where crop production dependson rain falling during the growingseason, the amount of available water isnot known at the time of planting, andfarmers must therefore select technologiesbased on expectations which mayor may not be fulfilled. (For a discussionof how drought stress influences theyield of cereal crops, see the box, p. 5.)Other, less extensive marginalenvironments for wheat and barley arefound elsewhere in the developing world2 For documentation of the EPIC model, seeWilliams et al. (1990) and Sharpley andWilliams (1990).and feature somewhat different rainfalland temperature patterns. In SouthAmerica. a significant amount of wheatis sown in areas where little rain fallsduring the early part of the growingseason, although late-season rainfall isoften adequate or even excessive. Innortheast China, wheat is grown in areaswhere winters are so severe that onlyspring wheats can be planted; in many ofthese areas, rainfall is often light in theearly part of the season, and wheatgerminates using residual moisture fromsnow melt. Rainfall during the secondhalf of the crop cycle can be excessive,however.Seven Examples ofRainfed MarginalEnvironments for WheatWheat breeding at CIMMYT is targetedtoward particular mega-environments,which are defined as "broad, notnecessarily contiguous areas, usuallyTable 1. Types of rainfed marginal environments for wheatMega- Moisture Temperature Wheat . owing Mainenvirnmnent regime regime type date locationME4A Rainfall, Temperate Spring Autumn North Africa,late drought habit West AsiaME4B Rainfall, Temperate Spring Autumn Southearly drought habit AmericaME4C Residual Temperate Spring Autumn Southmoisture to hot habit AsiaME5B Residual Hot Spring Autumn Southmoisture (dry) habit AsiaME6B Growing season Moderate Facultative, Autumn Westrainfall cold winter habit AsiaME6D Growing season Severe Facultative, Autumn Westrainfall cold winter habit AsiaME7 Growing season Severe Spring Spring Northeastrainfall cold habit ChinaSource:IMMYT Wheat Program.international and frequentlytranscontinental, with similar biotic andabiotic stresses, cropping systemrequirements, and consumerpreferences" (CIMMYT 1989b). Of theseven mega-environments recognized forwheat, three include areas generallyconsidered marginal because of moisturestress. When these three are subdividedaccording to the rainfall patternsdiscussed above, seven rainfed marginalenvironments can be distinguished(Table I).To illustrate the similarities anddifferences between these seven rainfedmarginal environments and highlight thesevere limiting effect of moisture stresson yields, a representative site wasselected for each environment, and theEPIC crop model was used to simulatepotential wheat yields over 50 years. 2Potential yields are defined by moistureand temperature conditions and are notconstrained by other abiotic and bioticstresses, such as soil fertility problems,

4Table 2. Simulated potential wheat yields in seven marginal mega-environments(rainfed vs. irrigated)Averag Averagepotential potential Pot ntialrainred irrigated rainfed/Mega- Dominant yield A yield A potentiaJenviron- Representative wheat (t/hs) (tJha) irrigatedment _ite type( ) ( (CV) yieldME4A Kasba Tadla, Autumn sown, 1.09 8.17 0.13Morocco spring habit (44) (8)ME4B Marcos Juarez, Autumn sown, 2.21 10.38 0.21Argentina spring habit (52) (7)ME4C Mirzapur, Autumn sown, 1.11 6.05 0.18India spring habit (14) (3)ME5B Sagar, Autumn sown, 2.29 6.72 0.34India spring habit (25) (6)ME6B Diyarbakir, Autumn sown, 2.99 6.07 0.49Turkey facultative (27) (8)ME6D Ankara, Autumn sown, 2.05 7.82 0.26Turkey winter habit (41 ) (4)ME7 Harbin, Spring sown, 3.64 6.95 0.52China spring habit (28) (12)alculaled frOID EPIC model simulation results (SO-year imul:i1ion).Cumulative probability (0/0)100 T---:::=::-:~~P"'""""-~::;;--::;,-----:----,9080diseases, insects, and weeds; thus,simulated potential yields will always behigher than actual recorded yields. J Todemonstrate the limiting effect of lowand uncertain rainfall, the EPIC modelwas run twice for each site: once under arainfed production scenario in which thecrop received only natural precipitation, ,and once under an irrigation scenario inwhich water was added to the soilwhenever the crop began to experiencemoisture stress. Average potential yieldsand coefficients of variation (CY) foreach of these sites under both rainfedand irrigated scenarios appear inTable 2. 4 Cumulative probabilitydistributions of the simulated potentialyields under the rainfed scenario appearin Figure 2.The simulated potential yield data revealconsiderable differences betweenmarginal environments, dependingamong other things on temperaturesduring the growing season, as well as onthe amount and distribution of rainfall.In temperate to hot rainfed megaenvironments(4A, 4B, 4C, 5B), averagepotential yields are generally low, withyield variability higher in climateswhere the crop depends on rainfallduring the growing season (4A, 48) ascompared to climates where the crop isgrown on residual moisture (4C, 58). Incold mega-environments (6B, 6D),average potential yields are roughlycorrelated with the amount of availablemoisture; yield variability in thesemega-environments likewise isinfluenced by the distribution of rainfall,showing greater stability where rainfall70605040302010~Diyarbaki:.-.Harbino~~k~~=--....,......-----.--------r-~0.01.02.03.0 4.05.0Yield (t/ha)Figure 2. Cumulative probability distributions for simulated wheat yields in sevenrainfed marginal environments._3 In these simulations, the EPIC model wasadjusted for each site to take into account soiltype, wheat variety, and local crop rotations.4 Following conventional practice, throughoutthis report the coefficient of variation aroundtrend (CV) is used as a measure of yieldvariability. It should be noted that the CV is arelative measure of variability, not an absolutemeasure. Thus, a decline in yield variability as6.0 indicated by a decrease in CV may resull froman increase in the mean yield. from a decreasein the standard error term. or from acombination of the two.

Factors Affecting Water Use EfficiencyThe word "drought" conjures up visionsof stunted plants; wilted, rolled, orprematurely withered leaves; andheavily constrained yield components(e.g., low spike numbers, fewer grainsper spike, and reduced grain size). Whilethese symptoms are indeed the predictableconsequences of severe drought,there exist as well a whole range of Iedramatic physiological effects caused bylack of water which can reduce grainyield in cereal crops.Water is fundamental to crop growth,because CO 2uptake by leaves duringphotosynthesis is necessarily accompaniedby H 20 vapor loss throughtranspiration. In addition to water usedby the plant in transpiration, moisture islost directly from the soil throughevaporation - generally about 75-120mm for typical cereal crops (the exactamount depends largely on rain frequency).Evapotranspiration (ET), theconventional measure of crop water use,is the sum of both types of water loss,i.e., transpiration plus evaporation.The relation 'hip between ET and grainyield is generally speaking direct andlinear, although the mechanisms relatingthe two are far from simple. The ratio ofphotosynthesis to transpiration, oftenreferred to as the transpirationefficiency, is usually fairly stable for agiven crop 'pecies (although it isinversely related to the prevailing drynessof the air). A stable transpirationefficiency means that crop dry matterproduction is more or I directlyrelated to total transpiration and hence,assuming a given amount of evaporation,to ET. Since the ratio of grain yield tototal dry matter yield (known as theharvest index) also tends to be fairlystable, then the relationship of grainyield to ET is also basically linear. Putsimply, for each ton of grain produced, acertain fixed amount of water must beused for ET.In temperate environments, modern highyielding spring wheat varieties grownwithout water limitations under excellentmanagement have a (potential) ET ofaround 500 mm, giving a potential orwater-unlimited yield of around 7 t/ha.In drought-prone environments, yield isreduced by lack of water in proportion tothe amount by which actual E1' fallsbelow potential ET. Farmers' yields areusually well below the ET-determinedmaximum potential yield because of lessthan-perfectcrop management.The performance of a crop in moisturelimitedenvironments is often assessed interms of water use efficiency (WUE),defined by agronomi t a grain yielddivided by ET. WUE i influenced byavailable water supply, defined as totalwater supply minus losses due to soilevaporation, runoff, and drainage. Whilethe amount of available water i mostcritical, it distribution relative to thecrop cycle can also be importliDI. sinceyields may fall below the theoreti almaximum if water shortage is concentratedsharply around flowering, or ifrains come too lnte in grain filling. Infavorable production environments, acrop that uses all potential ET achieves amaximum WUE of around 14 kg/ha/mm(7,000 kg/ha yield divided by 500 mmwater).Average farmers' yields in southernAustralia exhibit the exp ct d response ofyield to ET, but they also reveal a largeeffect of crop management on WUE(Corni h and Murray 1989, French andSchultz 1984). WUE i low relative topotential, e pecially in years of high ET,becRu e of inadequate fertility, poor weedcontrol, poor disease control, poor cropstand, incorrect sowing date, and soildiseases (see Figure). Average WUE (asdefined) for the whole of Australia isestimated at around 6 kg/ha/mm (Fischer1988); WUE levels achieved by farmersin developing countries growing rainfedwheat may be even les .In an attempt to determine whetherprogr S8 has been made in breeding forincreased WUE, Laing and Fischer(L977) explored the r lationship betweenvarieties and water stre S. using meansite yield at a large set of minfed sites asa surrogate for ET amount. Older tallvarieties were found to have a loweryield potential of around 4-5 t/ha withunlimited wliter, and yield fell withreduced IT much a expected. Thedifference between th older tall varietiesand first-generation semidwarfs wasthat the relative rate at which yields fellwith reduced ET was less with the olderwheats, presumably due to their droughtre i 'lance mechanisms. At severe stresslevels (plot yields below 1.5 t/ha), theolder varieties actually yielded morethan the semidwarfs. For many years, agreat deal of controversy surrounded thequestion as to the exact level of stre s atwhich older varieties began to show theiradvantage over first-generation semidwarfs.This controversy has now abated,as the second-generation semidwarfsreleased during the 1980s have clearlymoved the MY yield line upwards,apparently at all ET levels. Recent workby Perry and D'Antuono (1989) sugge tsthat MYs now outperform older varietieseven under very dry Mediterraneanconditions.Grain yield t/ha)7-.-----------------------........---.......65432Fanner averageo-t------y&o-----,-----,----"'T""'"----r-------'o 100500200 300 400Crop evapotranspiration (ET, mm)Relationship of grain yield to crop evapotranspiration for spring-habit wheatcrops in temperate regions in Australia with best management and with averagefarmer management.

6is more reliable. Finally, in the highlatitudemega-environment (7),relatively abundant rainfall results inhigh yield potential compared to theother rainfed mega-environments.When the results achieved under therainfed production scenarios arecompared with the results achievedunder the irrigation scenarios, strikingdifferences emerge. Under simulatednatural precipitation, at most of the sitesdrought stress depresses potential yieldsto less than half the potential determinedTable 3. Wheat production in rainfed marginal mega-environments in developingcountries, mid-1980sBread wheatDurum wheatArea Yield Production Area Yield ProductionMega-environment (mill ha) (tlha) (mill l) (mill ha Wha) (mill t)ME4A Low rainfall, 5.40 0.83 4.50 4.70 0.94 4.41temperate,late droughtME48 Low rainfall, 3.15 1.27 4.00 0 0temperate,early droughtME4C Low rainfall, 4.34 1.09 4.75 0 0temperate,residual moistureME58 Low rainfall, 3.17 0.90 2.84 1.49 0.81 1.20hot, residualmoistureME68 Low rainfall, 4.48 0.88 3.92 0 0moderate coldME6D Low rainfall, 5.91 1.48 8.73 1.21 1.40 1.69severe coldME7 Variable rainfall, 0.99 1.51 1.49 0 0severe coldDeveloping country total 28.46 1.09 31.04 7.46 0.98 7.35Percent of developingcountry total or average (32%) (50%) (16%) (72%) (74%) (53%)Source: IMMY mega-environment. database.by available solar radiation (as revealedby the irrigated simulation runs).5 Inaddition, yield variability underirrigation as measured by the CV is verylow, confirming that moisture stress isthe major factor accounting for highyield variability in these environments(any remaining variability in yields isdue to year-to-year temperaturefluctuations).While simulated potential yields do notreflect yields actually achieved byfarmers, the simulation results are usefulin that they graphically display the largeinfluence of a single climatic factor(moisture) in determining yield potentialand variability across a set of representativerainfed sites. (A historical analysisof yield variability in marginal environmentsis presented in the box, p. 8).Extent of RainfedMarginal Environmentsin the Developing WorldApproximately one-quarter of the areaplanted to bread wheat in developingcountries and nearly three-quarters of thearea planted to durum wheat is locatedin rainfed marginal environments(Table 3).6 Because of low yields, theseenvironments account for a less-thanproportional(but still significant)amount of total production; rainfedmarginal environments produce about16% of the bread wheat and 53% of thedurum wheat grown in the developingworld. (Because much more bread wheatis produced in the world than durumwheat, these percentages are somewhatmisleading. Even though a relativelysmall proportion of bread wheat isgrown in marginal environments, thetotal amount of bread wheat produced inmarginal environments is still over fourtimes as large as the amount of durumwheat produced in marginalenvironments.) Similar data are not5 Strictly speaking, two of the sites may not betruly marginal, since the ratio of rainfed toirrigated potential yields exceeds 40%.However, the sites which clearly conform to the40% criterion are located in the megaenvironmentson which this report focuses.6 The data on wheat production in marginalenvironments presented in Tables 3 and 4 arebased on subjective estimates made during themid-1980s by knowledgeable wheat scientistsfrom national agricultural research programsand CIMMYT. While the estimates are believedto be reasonably accurate, occasionally theydiffer from published government statistics orFAO production data (on which all subsequentanalysis contained in this report is based). Theestimates are nevertheless reproduced herebecause they represent the only comprehensiveset of production data disaggregated by megaenvironment.

7available for barley, but the proportionof barley area located in rainfedmarginal environments is undoubtedlyeven higher, since barley is consideredthe more drought-resistant of the twocrops and is frequently planted in drierproduction zones.?Table 4 presents additional data on thearea sown to wheat in low rainfall zones,broken down by region. Althoughrainfed marginal environments arewidely distributed throughout thedeveloping world, it can be seen thatwheat production in these environmentsis concentrated in the WANA Regionand South Asia.Production Trends in RainfedMarginal EnvironmentsEvidence of divergent rates of progressbetween favorable and marginalproduction environments exists atseveral levels. When a broad distinctionis made between developing countries inwhich most wheat is grown under wellwateredconditions and developingcountries in which most wheat is grownunder low rainfall, it becomes apparentthat wheat area and yields in the lowrainfall group of countries have grown atonly half the rates recorded in the wellwateredgroup (Figure 3).Similar discrepancies are apparentwithin individual countries. Perhaps themost detailed picture of productiontrends in favorable vs. marginalenvironments is available for India,where the past three decades have seen adramatic shift in wheat production fromrainfed to irrigated areas (Figure 4).Between 1961-65 and 1984-86, rainfedwheat area in India declined from about9 million hectares to 6 million hectares,while irrigated wheat area rose from 4million hectares to 17 million hectares.Hence the conventional wisdom thatcrop cultivation is expanding into7 However. in several major barley producingcountries such as Turkey. Tunisia, and India,over half of all barley is produced in irrigated orhigh rainfall zones.Average annual rate of growth,1961-65 to 1987-89 (%)6.....------------.-.Proclll lion ( .4)54Table 4. Area planted to wheat in low rainfall zones ofselected developing countries,mid-1980sA percent Low As percentLow rainfall of total rainfall of totalbread wheat bread wheat durum durumarea area area area(000 ha) in region (000 ha) in regionWest Asia/North Africa 12,464 63 5,913 74South Asia 6,785 23 1,500 94East Asia 4,730 16 0 0South America 3,225 38 0 0Developing country total 28,459 32 7,463 72Source: CIMMYT mega- nvironmen dalaba e.3Vi Id ( .0)Producti n (2.5)Million ha20-r------------------------.......------,2Yi Id (2.0>15Img'lIed areaoCountries that Countries thatgrow wheat grow wheat mostlymostly under underdryJand 5well-watered conditions conditionsJOFigure 3. Growth of wheat area and yieldsin two groups of developing countriesdistinguished by moisture regime, 1961-65to 1987-89.O+...........--'T___,.___..__""T"'"_._...........~___,.___..__""T"'"_._ ...........~___,.___..__""T"'"_._"""T"~___,.___~1963 1966 1969 1972 1975 1978 1981 1984Figure 4. Trends in rainfed and irrigated wheat area in India, 1963-86.

Variability in Wheat Yields in Marginal EnvironmentsIn low rainfall environments, wheatyields are nol only low, but they tend tobe highly variable. Variability in yieldsis ofconcern both at the farm level(since farmers seek stability in foodproduction and income), as well as at thenational level (since national food securityis a major goal of policy makers).Singh and Byerlee (1990 recentlyanalyzed wheat yield variability in 57countries over 35 years (1951-86). Yieldvariability was measured by calculatingcoefficients of variation (CY) of yieldsaround linear trend. Singh and Byerleeconcluded that rainfall is thepredominant factor influencing yieldvariability: countries in which at leasthalf the wheat area is own in drylandconditions experience twice as muchyield variability as countries in whichwheat is mostly grown under wellwateredcondition (see Figure). Therelationship j particularly trong insmall countrie ; in large countrie ,drought in one region may be off et byaverage or above-average rainfall inanother region, leading to lower overallyield variability ( ee Figure). Yieldvariability also tends to be higher inwarmer, subtropicaJ countries because ofheat stress. ize of national wheat area,percentage of national wheat areasubject to moisture, tress, and locationof the country in the warmer tropicsexplain over two-thirds of the variationin CYs of yield across countries.Significantly, and somewhaturprisingly, technologi al variablessuch as use of MYs and fertilizerexplained none of the variation in CYs.ot all rainfed marginal environmentsare characterized by high yieldvariability. In the residual moistureenvironment ofcentral and southernIndia, wheat yields are generally quitestable; in fact, yields in this environmentare no more unstable tban in adjacentirrigated area' where irrigation watersupplies tend to be unreliable (Byerlee1991). However, among countries in theWANA Region the variability in cerealproduction (mo tly wheat) averages 25­30%, two to three times higher than inmost of Asia and Latin America. Thehigh production variability in theWANA Region poses special challengesto policy makers charged with managingnational fi od upplies.Singh and Byerlee also analyzedchanges over time in wheat yieldvariability in order to evaluate the claimthat yields have become more unstableas a result of widespread adoption ofGreen Revolution technologies,particularly input-responsive MYs. Inthe case of wheat, no evidence wasfound to upport this hypothesis. Indeed,in the irrigated and well-wateredprodu tion zones where the GreenRevolution had its greatest impact, yieldvariability has declined sharply in recentyears, especially in the Indian Punjab(see Table). Even in dryland areas,wheat yield variability seems to havedeclined slightly during the most recentdecade, the period when seed-fertilizertechnology began to spread into moremarginal environments. These findingsconfirm that yield risk is higher inmarginal areas, but they do not supportthe assertion that this risk is exacerbatedby the use of improved technology.CV of wheat yields (%)50.,..--------------------------,• • • Dry countries CY = 43.2 - 3.01 Ln (WA)R'=0.S7(4.13)***n= IS40302010o•••-.e.-..• a..A.-.A.Wet countries CY = 14.3 - .64 Ln (WA)R'= 0.11 (2.16)** n = 36.A. Tropical countries•• • -.,-.-. - .. •-•• •••Relationship between coefficient of variation (CV) of wheat yields and wheat area(WA) by climatic regime, 1966-86.Changes in the variability of wheat yields in selected groups of countries, 1951-86Percent changeCoefficient of variation inCV,Number of around trend (%) 1951-65 tocountries 1951-65 1966-75 1976-86 1976-86Irri[ated andweI -watered temperate 36 10.6 8.5 6.4 - 38Dryland temperate 15 16.2 [5.2 14.4 - 9Tropical 6 22.6 23.6 18.9 - 16All countries 57 13.3 11.8 10.1 - 23Indian Punjab 18.9 12.6 6.2 - 6712

9marginal areas is certainly not valid forwheat in India; on the contrary, asrainfed production zones have beenconverted to irrigation, the proportion ofdryland wheat area has declined from67% to 25%.An examination of yield trends in Indiahelps explain the increasingconcentration of production in irrigatedzones (Figure 5). Wheat yields inirrigated areas averaged 2.2 t/ha in 1985,more than double the 0.9 t/ha achievedin rainfed areas. Significantly, thedifference between yields of irrigatedand rainfed wheat has been widening;from 1972 to 1985, yields of irrigatedwheat grew at an average annual rate of2.8%, compared to only 1.4% for yieldsof rainfed wheat. This combination of adeclining proportion of rainfed area anda higher growth rate in yields of irrigatedwheat has caused the share of rainfedwheat production to fall sharply in India,from around 50% in 1960 to 15% in1986.Production data from a number ofWANA countries provide furtherevidence that progress has been slowerin dryland areas. For example, cerealyields in irrigated areas of Syria havegrown at more than double the rate ofyields in rainfed areas (Table 5).Farmers in Syria have responded to themore difficult production conditions inrainfed areas by planting less wheat andmore barley. The decrease in rainfedwheat area has been accompanied by asteady expansion in irrigated wheat area,primarily through the installation oftubewells. As a result of these yield andarea trends, the percentage of total wheatproduction grown under irrigatedconditions increased from around 15% inthe late 1960s to over 40% by the late1980s (Figure 6). Similar trends leadingto the concentration of wheat productionin higher rainfall regions are evidentelsewhere in the WANA Region,including Morocco and Tunisia (Belaidand Morris 1991).The picture is slightly different inTurkey, the country with the largestwheat and barley areas in the WANARegion. Wheat and barley production inYield (t/ha)2.5 -.----------------~-------__,21.5•.... -------­•Rainfed(growth = 1.4%/yr)0-+-------,.-----.,-------.-----.,.------119651970197519801990Figure S. Trends in rainfed and irrigated wheat yields in India, 1968-86.Table S. Cereal production data for Syria by moisture regime, 1968-87Wheat1985-871970Wbeatrca and yield1968-70Barley to1985-87 19711-861972197419761978Growth rates1978·80to1985-8719801968-70to1978·801982Barley1978-80to1985-87Area (000 ha) (%/yr)Rainfed 946 1,490 2.0 -4.0 3.5 3.9Irrigated 236 12 8.5 3.8 0.5 -3.4Yield (t/ha) (%/yr)Rainfed 1.12 0.53 4.9 1.4 1.6 -4.7Irrigated 3.07 1.85 7.9 3.0 5.2 0.9S ume: Calculated from data provided by Iheof Agriculture (I 8 ).yrian Mini tryMillon t2.4 -r----------------------------,2.22.01.81.61.41.21.00.80.60.40.2O-+--,.---,r---r-.....,...--r-r--..,...-r---T-r-"'T"""-,---r-T--r--,.---,r--~ ...1968Figure 6. Production of rainfed and irrigated wheat in Syria, 1968-87.1984 1986

10Turkey are distributed across bothfavored areas (e.g., the well-wateredtemperate zones along the Mediterraneancoast and in Thrace) andmarginal areas (e.g., the dry, coldAnatolian Plateau) (Figure 7). Highyielding semidwarf spring wheatvarieties and fertilizer were adoptedrapidly in the favored areas beginning inthe mid-1960s (Demir 1976), leading todramatic increases in wheat yieldsthroughout these environments(Table 6). Although MVs were notadopted nearly as rapidly in the marginalproduction zones of Central Anatolia,significant progress nevertheless wasachieved in raising yields throughimprovements in crop management,including improved tillage practices,increased fertilizer use, and better weedcontrol.Evidence from a number of countriesthus appears to confirm the widespreadperception that the rate of growth inwheat yields achieved in rainfedmarginal environments has not kept pacewith that achieved in favoredenvironments (including irrigated areas).Farmers have responded by increasingthe area planted to wheat in favoredzones and by concentrating relativelymore production in these zones.In an attempt to uncover the reasons forthese apparent differential rates of yieldgrowth, the next two sections of thisreport examine in some detail theexperiences of two regions containingextensive dryland wheat and barleyproduction zones-the WANA Region (awinter rainfall environment), and centraland southern India and parts of Pakistan(a residual moisture environment).Table 6. Historical growth in wheat yields in four production zones of Turkey,1967-87Average wheatAnnual growthyield (tlha) in yield (%)ProduCtion zone a 1965·69 1974-78 1983-87 1967-76 1976-85Favored environmentsMediterranean Coast 1.68 2.71 3.28 5.3 2.1Thrace 1.23 2.80 2.96 9.2 0.6Marginal environmentsCentral Anatolia 1.24 1.69 1.98 3.5 1.7Eastern Anatolia 0.88 1.01 1.15 1.6 1.5a Based n data from the Howing pro inee : Adana, Hatay (Mediterranean Coast);Edirne. Kicklareli, Tekirdag (Thrace); Ankara. Eskesehir (Central Anatolia); Artvin,Kars. Erzurum, Agri. Mus, Billi , Siirt. Hakkari. Van (Eastern An3tolia).D Winter- facultative wheat areaFigure 7. Wheat production zones in Turkey.

11Rainfed Marginal Environmentsof West Asia and North Africa 8By far the greatest concentration ofdryland wheat and barley production inthe developing world is found in theWANA Region, where approximately 25million hectares are planted to wheat, ofwhich only about 20% is irrigated _(mostly in Egypt and Saudi Arabia). Ofthe rest. nearly three-quarters is locatedin areas receiving less than 500 mmannual rainfall. An additional 12 millionhectares are planted to barley in theWANA Region, of which an even higherproportion is located in low rainfallzones.Types of Fa~mingSystems in the WANA RegionA broad overview of the rainfedagricultural systems of WANA, ignoringlocal variations imposed by topography,shows a transition from crop cultivationin wetter areas to livestock-based8 This section draws on Belaid and Morris (1991).systems in drier areas. In general, coastalareas are wettest and support intensivehorticulture and agriculture. Rainfalldeclines with increased distance fromthe coast, so that moving inland one firstencounters rainfed systems dominatedby wheat and then systems dominated bybarley. Where annual rainfall dropsbelow 200 mm, rainfed farmland givesway to large expanses of steppe thatprovide grazing for nomadic ortranshumant flocks of small ruminants.Deserts predominate in the interior ofthe region, where only irrigatedagriculture is possible (Figure 8).Rainfall in the dryland wheat and barleyzones of the WANA Region is not onlylow but highly variable. In addition tovarying from year to year, rainfall alsovaries through space; steep gradients inprecipitation frequently occur acrossshort distances, with changes sometimesas large as 3-4 mm/km. This markedspatial and temporal variability inrainfall, and the correspondingvariability in potential growing season,is the most important factor contributingto risk and uncertainty in rainfed cropproduction.Like rainfall, temperatures also fluctuateconsiderably from year to year in theW ANA Region, causing the length ofthe growing season to vary by as muchas to three to four weeks at any givenlocation (Cooper and Bailey 1990).Given the sharp increase in evaporativedemand due to a rapid rise intemperatures in spring and earlysummer, wheat and barley crops areusually stressed during grain set andgrain filling.In higher rainfall areas of the WANARegion (350-600 mm average annualrainfall), the main crop is usually wheat.Both bread wheat and durum wheat aregrown, with the proportion between thetwo determined partly by climaticfactors and partly by utilization patterns.Bread wheat tends to predominate at theweller end of the rainfall spectrum,while durum wheat tends to predominateat the drier end.Throughout much of the WANA Region,a fallow-wheat rotation was traditionallywidespread, but in wetter areas and onbetter soils wheat is increasingly beingMOROCCO--," """"'"" ALGERIA" """'"' '-'-,~-LIBYAIIIIIEGYPTII1~ ------------,., I",-.)_1250-500 mm< 50{) mmFigure 8. Distribution of rainfed marginal environments, defined by precipitation, in the WANA Region.

12grown in two- or three-crop rotationswith food or feed legumes and summercrops such as melon, maize, sesame,sunflower, and cotton. Since most wheatfarmers own both sheep and cattle, theyalso grow some barley, using the grainand straw for feed, together with othercrop residues such as wheat and legumestubble.At the drier end of the rainfall spectrum,on land adjacent to the dry steppe (200­350 mm average annual rainfall),livestock production is the dominantenterprise, providing most of the incomeof farm families. Animal production(principally sheep and goats) is based onannually sown feed crops, notablybarley, with both grain and straw used aslivestock feed. In some years barley maybe lightly grazed as a spring pasture, butits main purpose is to provide grain andMetabolizable energy requirements (Ms/d)20161284ostraw for winter feeding. Since barley isthe most important source of feed,farmers attempt to sow at least enougharea to be self-sufficient.Because of the production risks inherent,in the environment, barley production inthese drier areas is frequently based onminimum-input technologies.Throughout the region, a barley/fallowrotation predominates, although fallowmanagement practices varyconsiderably. In some areas farmerscultivate fallow land to reduce weeds,increase water infiltration, and maximizemoisture storage. Elsewhere, a morecommon practice is to leave fallowsuncultivated (weedy fallow), sinceweeds and volunteer crops provide avaluable source of livestock feed.However, as demand for livestock feedrises, fallow is increasingly regarded asan inefficient use of land, andcontinuous barley production isbecoming more common.The important role played by livestockin the farming systems of WANAcountries directly influences the cropproduction strategies of most cerealfarmers, particularly in drier areas wherebarley/livestock systems predominate. Inthese areas, the largest share ofagricultural income (i.e., excludingincome from off-farm employment)typically is derived not from crops, butrather from animals and animalproducts, including milk, cheese, andyoghurt. Studies in the dry rainfed zonesof Syria and Algeria, for example,indicate that the sheep enterprise aloneprovides up to 82% (Thomson, Bahhady,and Nordblom 1982) and 77%(Boutonnet 1989) of agricultural income,respectively.9In addition to being an important sourceof income, livestock act as a financialbuffer against yearly fluctuations in cropproduction caused by rainfall variability.In drought years when crops are poorand harvesting would be uneconomic,farmers graze animals on their crops andalso may reduce their flocks throughsales, thereby generating income and atthe same time reducing feedrequirements in the following winterwhen stored barley grain and straw willbe in short supply because of the poorharvest.The straw of cereal crops, particularlybarley, is one of the most importantfeedstuffs, utilized either as grazing insitu or as a component of supplementarydiets in winter (Figure 9). In drier areas,particularly in dry years when theharvest index is low, the total economicvalue of cereal and legume straw perhectare can exceed that of grain(Table 7).DryAWinterFigure 9. Livestock feeding cycle in the WANA Region.Source: Cocks and Thomson (1988).LactationSpringM M J9 In bom of Ihese studies, the value of feed cropsproduced for use on the faml as an input inlo melivestock enterprise was not taken into account;therefore, the economic importance of cerealswas understated. Nevertheless, even when thevalue of cereals produced for feed use on thefann is included, the sheep enterprise clearlyremains of major importance.

13The integration of barley and livestockmanagement is crucial to the survival ofmany farmers in dry areas. Few cropdecisions are made without consideringthe feed requirements of livestock, andfew livestock decisions are madewithout considering feed availability(Thomson, Bahhady, and Nordblom1982). Strategies for improving cropproduction, such as better fallowmanagement, weed control throughherbicide use, or the introduction of newvarieties, must take account of theseinteractions with livestock productionand not jeopardize vital sources of feed.Use of Improved TechnologyVarietal development and diffusion­Throughout most of the WANA Region,adoption of modern varieties (MVs) hasproceeded much more slowly than inSouth Asia, where in countries such asIndia and Pakistan over 60% of thewheat area was already sown to MVs in1975, rising to about 90% by 1990. Inthe W ANA Region the area sown toMVs was estimated at 17% in 1976-77,rising to an estimated 42% by 1985-86(Dalrymple 1978, CIMMYT 1989c).Although few comprehensive adoptionstudies have been undertaken in recentyears, avai lable evidence suggests thatMVs have been adopted quiteextensively in favored environments andmuch less extensively in marginalenvironments. For example, in 1989 lessthan 7% of low rainfall wheat area inSyria was sown to MVs, compared tonearly 57% of high rainfall and irrigatedarea (Syria Ministry of Agriculture,unpublished data). Similarly, in 1983only I % of the low rainfall area inTunisia was planted to modern durumwheats, compared to 58% of the highrainfall area (Johnson, Ferguson, andFikry 1983).Broadly speaking, in the WANA Regionuse of MVs is highest for spring breadwheats, followed by spring durums,winter bread wheats, and facultativebread wheats. Adoption of improvedwinter durums, facultative durums, andbarley has been less common. Thispattern, which roughly corresponds topast availability of MVs, notcoincidentally also reflects the rainfallgradients under which these differentcereal types grow. Spring bread wheatsare grown in the wettest areas, followedby spring durums and winter breadwheats. Winter durums and barley aregrown in the driest areas.Differences in adoption patternsaccording to wheat type and rainfallregime also have been evident inTurkey. Spring bread wheats movedrapid ly into the well-watered coastalzones of Turkey, and by the mid-1970s,less than 10 years after improvedvarieties first appeared in the country,the majority of farmers in these zoneshad adopted them. Meanwhile, adoptionrates in the marginal productionenvironments of Central Anatolia, wherewinter bread wheats and durum wheatsare grown, remained extremely low.The pattern of adoption of MVs inTurkey is consistent with the number ofvarietal releases coming out of thenational breeding program. Turkishbreeders, working with colleagues frominternational research institutes, havebeen very successful in breedingvariet~es for favorable productionenvironments, as evidenced by the highproportion of spring wheats released forthese environments. Less success hasbeen achieved in breeding for rainfedmarginal environments, where winter(and facultative) durum and bread wheatproduction is concentrated. During theI980s, for example, only a smallproportion () 0%) of the varietiesreleased were winter wheats specificallytargeted for the rainfed areas.To some observers, the fact that MVshave diffused relatively slowly into lowrainfall areas is puzzling. Even thoughmany breeding programs in the WANARegion have been slow in releasingvarieties recommended specifically formarginal areas, most of the materialsdeveloped for favorable productionconditions offer a yield advantage in allbut the harshest environments, evenwhen grown under low rainfallconditions. Thus, it seems odd thatfarmers in low rainfall zones have beenslow in adopting MVs. One possibleexplanation is that the yield gainsachieved from adopting improvedgermplasm alone tend to be modest indrier areas, and frequently the extrayield may not be enough to offset pricepremiums commanded by traditionalvarieties. However, since crop and soilmanagement practices that conservemoisture interact positively with thehigher yield potential of MVs,continuing improvements in cropmanagement practices can be expectedto further stimulate the adoption of MVs.If improved wheat varieties are nowavailable that offer limited yield gains inrainfed areas, improved barley varietiesTable 7. Relative value of wheat grain and straw under different levels of rainfall inBaluchistan, Pakistan, 1985-88Poor car Average year yood earGrain yield (kglha) 16 146 412Straw yield (kglha) 134 335 1,115Grain price (Rs/kg) 2.0 2.0 2.0Straw price (Rs/kg) 1.0 0.5 0.6Gross benefits (Rs/kg) 166 460 1,493Percent of gross benefits from straw 81 38 45urce: Rees el aI. (1991).

14that yield more than local materials arestill not common, especially in drierareas. In a series of multilocational yieldtrials in Syria, Cooper (1985) found thatmost improved barley varieties did notperform significantly better than thelocal check. More recently, however, asnational barley breeding programs havebegun to mature, improved barleymaterials have been developed whichperform well even under extremely dryconditions (for example, the improvedvariety Rihane tested and selected inTunisia) (JNRAT, unpublished data).Improved management practices andinputs- In the WANA Region,aggregate fertilizer consumption hasexpanded considerably during the pasttwo decades, apparently paralleling theadoption of MVs. The experience ofTunisia typifies this pattern: fertilizeruse on wheat in Tunisia doubled from20 kg nutrient/ha in 1970 to 40 kgnutrient/ha in 1980. Over the wholeregion, farmers now apply an average of40-50 kg nutrient/ha to wheat (CIMMYT1989c), although fertilizer use on barleyremains negligible. As expected,fertilizer is used more extensively in theirrigated and well-watered rainfed areaswhere MVs are concentrated and wherecrop response is highest. In these areas,Table 8. Nitrogen-to-wheat price ratios in selected countries, 1989-90Farm level njtrogen "arm level wheat it rogcn-to-wbeatountr prjce ( 8$/1) price f ,$/1) price ratioAfghanistan 216 180 1.2Algeria 360 394 0.9Syria 989 664 1.5Turkey 337 208 1.6India 305 2m" 1.5Pakistan 330 110 3.0Argentina 559 74 7.6Brazil 680 152 4.5Mexico 306 180 1.7Australia 562 104 5.4Canada 530 100 5.3USA 413 96 4.3"urce.: 'IMMYi urvey.Price in I.Iryland central India.fertilizer application rates often exceed100 kg nutrient/ha. Less fertilizer isapplied in drier zones where barleypredominates, on shallower soils, orwhen a legume crop precedes the cerealcrop in the rotation (Kukula andDakermanji 1986, Belaid and Morris1991). In areas receiving less than350 mm rainfall, only a small percentageof farmers use fertilizer.Many countries in the WANA Regiononce provided price incentives for usingfertilizer, either by directly subsidizingfertilizer or by introducing favorableprices for cereals. The effects of thesepolicies are evident in the low nitrogento-grainprice ratios prevailing in manyWANA countries, which are generallybelow those found in other major wheatproducing countries (Table 8). Nonetheless,recent policy reforms may haveadversely affected overall fertilizer usein the WANA Region by raisingfertilizer prices.Extensive research on fertilizer use indryland production zones of the WANARegion has led to the conclusion thatyields of wheat and barley grown underrainfed conditions can be considerablyimproved by using nitrogenous andphosphorous fertilizers (for example, seeBolton 1979, 1981; Harmsen 1984;Jones, Matar, and Pala 1988; ICARDA1989a, 1989b; Mazid 1990). Moreover,fertilizer can substantially enhance theefficiency of moisture use. In oneextensive series of on-farm experimentsin dry areas of Syria, yields of fertilizedbarley plots (measured in yield per unitof growing season rainfall) were threetimes that of unfertilized plots(Figure 10).However, crops do not benefitautomatically from fertilizerapplications, because response tofertilizer in dryland environments variessubstantially through time and space as aresult of variability in climaticconditions. The amount and temporaldistribution of rainfall is the dominantinfluence on crop response to fertilizer,especially to nitrogen. On average, atprevailing low fertilization levels of 40­50 kg/ha of nutrient, a 100-mm increasein rainfall increases the grain-to-nitrogenresponse ratio by about five (Somel,Mazid, and Hallajian 1984).'0Given the strength of this relationship,rainfall variability greatly augments therisk of using fertilizer in drylandproduction zones. Many farmers attemptto protect against this risk by applyingsplit applications of nitrogen. By varyingthe number and level of fertilizerapplications according to rainfall duringthe course of the growing season,farmers fine-tune doses depending on theperceived crop potential for thatparticular year. However, the strategy isnot completely foolproof, and yieldsmay be reduced by excessive nitrogen ifrainfall at the end of the season turns outto be inadequate.Response to fertilizer is influenced by alarge set of interacting agroclimatic, soil,and management factors, many of whichare beyond the control of the farmer. Forpractical purposes, the importantquestion is whether or not it is profitablelOA similar response can be observed in theMediterranean-type dryland environments ofAustralia (Byerlee and Winkelmann 1980).

15to use fertilizer in low rainfall zones.Economic analysis of the extensiveSyrian data set generated by ICARDAand the Syrian Soils Directorate led tothe conclusion that using nitrogenousfertilizer on wheat tends to be profitableonly under certain conditions, such aswhen application levels are modest (40­80 kg N/ha) and the land is continuouslycropped. Likewise in the dryland Settatarea of Morocco, results from on-farmtIials indicate that the profitability offertilizer use varies as a function ofrainfall, crop rotation, and soil type.Depending on the minimum acceptablerate of return on fertilizer required bythe farmer (itself a function of thefarmer's risk preference), investment infertilizer might or might not beconsidered attractive.Barley grain yields (t!ha)432aa, ,,~~.~~100200 300Seasonal precipitation (mm)These results from Syria and Moroccoillustrate the difficulty of translatingexperimental data on crop response tofertilizer into blanket recommendationsfor farmers in marginal areas. Based oneconomic analysis, investment in amodest level of nitrogenous fertilizerwould appear to be wise under a widerange of rainfall outcomes for farmersengaged in continuous cereal cropping,but the returns to fertilizer use becomefar less certain when cereal crops aregrown after a legume crop or afterfallow. Because the profitability offertilizer thus depends on a number ofinterrelated factors, efforts to promotefertilizer use must concentrate oninforming farmers about the complexnature of fertilizer response functionsand letting them decide for themselveswhat practice to adopt in light of theirown attitudes toward risk.Figure 10. Relationship between barley grain yield and seasonal precipitation undertwo levels of fertilizer, northern Syria, mid-1980s.Source: Cooper et al. (1988).400500Mechanical weed control prior toplanting is widely practiced throughoutthe WANA Region. Adequate tillage andproper seedbed preparation can beimportant in minimizing weedinfestation, reducing the need to resort toexpensive manual or chemical controlmethods once the crop emerges. Farmersusually wait until weeds emerge after theearly rains to destroy them with a lighttillage before planting. However, thispractice carries its own risk, since yieldlosses resulting from late planting mayactually offset the yield gains fromimproved weed control.Hand weeding of cereal crops iscommon in areas where family labor isavai.lable and is often performed bywomen and children. Hand weeding issometimes preferred to other weedcontrol methods because of the higheconomic value of weeds as greenfodder for livestock.Farmers in many areas supplementmanual weed control with use ofchemical herbicides, particularly inhigher rainfall areas where fallowing hasbeen eliminated and weed infestationcan be severe. For example, more than80% of the bread wheat area in MeknesProvince of Morocco (a favorablerainfed zone) is sprayed with herbicides(Serghini 1986). In contrast, theproportion of wheat area treated withherbicides does not exceed 6% in ElJadida Province (an intermediate rainfallzone). Likewise in Tunisia less than 20%of the wheat area is habitually sprayedwith herbicides, and in drought yearschemical weed control is practiced onless than 5% of the total cereal area(Tunisia Ministry of Agriculture 1989).Chemical weed control is not alwayseffective. Weed-induced yield losses areoften exacerbated by the nearlyexclusive reliance on phenoxy acidherbicides (e.g., 2,4-0), which controlonly broadleaf weeds. Chemicalseffective on wild oats and other grassyweeds (e.g., Suffix and IIIoxan B) arenot extensively used because of limitedsupply, lack of information, and muchhigher costs.

16Given the ability of weeds to depresscereal yields, the importance of adequateweed control cannot be overemphasized.Chemical weed control is credited withincreasing wheat yields by as much as60% in northern Tunisia (Daaloul 1985),from II % to 68% in central Turkey(Avci et al. 1988; Durutan et al. 1990),and from 22% to 33% in Syria (Pala etal. 1987). At prevailing prices, yieldincrements of this order usually justifythe cost of chemical weed control. Datafrom Tunisia and Turkey indicate thatthe total cost of applying 2-4,D(including costs of capital andequipment) is equivalent toapproximately 100 kg of wheat (Byerleeand Winkelmann 1980).Mechanization has spread rapidlythroughout the WANA Region inresponse to increased labor costsresulting from competition both withinas well as outside the agricultural sector.In dryland production zones,mechanization has enabled cerealfarmers to expand total cultivated areaand to increase cropping intensity.Perhaps the most striking example ofthis trend has occurred in Turkey, wherethe growth of the national tractor fleethas been accompanied by a steadydecrease in fallow area (Figure II).Mechanization has also significantlyreduced the time required for criticaloperations such as sowing andharvesting, leading to higher yields (byallowing more timely standestablishment) and reduced crop losses(by minimizing grain shattering becauseof delayed harvesting).Mechanized tillage based on tractorrental is the norm throughout mostWANA countries, although animal drafttillage is still common in areas wheretopography makes the use of tractorshazardous. Mechanized harvesting hasalso spread rapidly but remains lesswidespread than mechanized tillage.Mechanization of other crop operationshas not been as extensive. For example,mechanical seed drilling remainsuncommon, despite its potential toimprove yields by placing seed andfertilizer where they can better utilizethe available moisture. Farmers stillbroadcast wheat and barley seed by handand make use of any of a variety ofimplements to cover the seed.Farmers in the WANA Region have hadlittle experience with reduced orconservation tillage. In dryland wheatareas of Australia, Canada, the USA, andother industrialized countries, a growingnumber of wheat farmers practiceconservation tillage based on chemicalweed control (chemical fallowing) andestablishment of a protective soil mulchcomposed of crop residues. Suchpractices not only greatly increase theefficiency of moisture conservation inthe fallow period (Greb 1979), but theycan also play an important role inreducing soil erosion. In the WANARegion, conservation tillage may havepotential for conserving moisture andslowing soil erosion, but the complexityof managing conservation tillagepractices, combined with the high localvalue of crop residues and weeds used asfodder, will likely slow adoption.Fallow area milli n hll)In drier areas of the W ANA Region,most wheat and barley is produced in acereal-fallow rotation. Fallowing iswidely regarded as essential forconserving moisture and stabilizingyields, especially in dry years. However,the efficiency of moisture conservationthrough fallowing is often reduced bythe practice of maintaining weedy fallowfor feeding livestock. Recent research onthe role of fallow has cast doubt on thebenefits of fallowing in some marginalenvironments. For example, in Syria it isestimated that less than 10% of the rainthat falls during the fallow seasonremains in the soil profile by plantingtime for cereal crops (Harris 1989).Similarly, in areas of Turkey receivingless than 350 mm of annual rainfall andwhere soil depth does not exceed 90 cm,fallowing probably is not beneficial interms of accumulating moisture (GuIerand Karaca 1988).Introducing changes in traditionalfallowing practices is not always easy.Even where clean fallowing can improvemoisture storage or facilitate desirableearly sowing, managing a clean fallowsystem requires changes in the methodand timing of tillage which may disruptII "T"""-------------------------.,1098765432Tractor fleet (00000 units)O+-........-..,..............-..,..............-..,..............-..,..............-..,..............-..,..............--r---.--r---.,....--!1970 1972 1974 1976 1978 1980 1982 1984 1986 1988Figure 11. Growth in national tractor fleet and associated decrease in fallow area inTurkey, 1970-88.Source: Turkey Ministry of Agriculture (1990).

17livestock enterprises. For example,research on early tillage in the AnatolianPlateau of Turkey revealed that theincreased value of higher grain yieldsachieved through early tillage may notoffset the loss of pasture suffered whenweedy fallow or stubble is plowed under.In recent years, rising demand for foodand feed, coupled with the rapid spreadof mechanization, has led to significantincreases in cropping intensity throughreduced fallowing. Unfortunately, insome areas where fallowing is declining,lack of crop diversification poses seriousproblems, since the fragile soils whichcharacterize many marginalenvironments will not withstandcontinuous intensive cultivation ofcereal crops. Researchers have thusbegun to look for more sustainablerotations.One way of reducing weedy fallow andpromoting more sustainable croprotations is to practice cereal-legumerotations for providing food or cash orfor increasing the production of fodderthrough forage legumes. Maintaining (orincreasing) legume production isespecially important, since the area sownto legumes tends to decline in drierareas. One successful example of cropdiversification is the fallow replacementproject undertaken in the northernAnatolia region of Turkey, where foodlegumes (lentils and chickpeas) and feedlegumes (mostly Hungarian vetch) havebeen introduced on a large scale in therotation as substitutes for fallow, alongwith an improved input packageinvolving fertilizer, herbicide, improvedvarieties, and mechanization. Thissuccessful experience, which isprogressively being extended to otherregions of Turkey where fallowpredominates, shows that there is scopefor technological change in rainfedareas.Another alternative to fallowing whichhas been widely tested in the WANARegion is the so-called ley farmingsystem, in which annual self-reseedinglegume pastures (mainly medicago) aregrown in rotation with wheat. Thissystem is popular in areas of southernAustralia characterized by aMediterranean-type climate and hasplayed a major role in increasing theproductivity of wheat/sheep farmingsystems. However, when implemented inthe WANA Region the system has failedto yield significant results, primarilybecause local land use customs impedeprecise control of the timing andintensity of grazing pressure, but alsobecause of the system's intensivemanagement requirements.The Policy Environmentfor Wheat and BarleyBecause wheat is the primary staple foodthroughout most of the WANA Region,wheat prices are politically sensitive aswell as economically important. In aneffort to reconcile sometimes conflictingeconomic and political goals, manygovernments have chosen to pursueindependent producer and consumerprice policies. For example, in Moroccoreal consumer prices for wheat productsdecreased by 30% between 1970 and1987, while real producer prices forwheat increased by 19%. In general,domestic wheat prices are supportedabove import prices (Gardner and Skully1986, Boutonnet 1989), althoughovervalued exchange rates often reduceproducer incentives in real terms. I IAlthough direct producer price supportfrequently has been extended to wheat,the same cannot be said about barley.Official producer prices for barley areusually published along with those forwheat, but barley prices are rarelysupported through extensive governmentpurchases of barley grain. This usuallyresults in low barley prices relative towheat prices. On the other hand, in somecountries government lack of interest inbarley has paradoxically allowed freemarketbarley prices (which are largelyunregulated) to rise above official wheatprices (which are strictly enforced). InAlgeria, for example, the strong demandfor feed barley in recent years has led tothe emergence of a thriving parallelmarket in which unofficial barley pricesoften are 2-3 times higher than theofficial prices paid for wheat by thegovernment grain procurement agency.Of course, in deciding how to allocateresources among alternative enterprises,farmers' primary concern is not absoluteprofitability but rather relativeprofitability (i.e., the returns they arelikely to earn on resources invested inone enterprise relative to the returns theywould earn by investing the sameresources in the most profitablealternative enterprise). In irrigatedzones, cereal crops have a hard timecompeting with high value crops such asvegetables, oilseeds, and fruit, whichmeans cereal prices would have to beraised to unreasonable levels for cerealsto displace these crops. In rainfed zones,attractive alternatives to cereals in cropproduction are more limited, but giventhe close integration of cereal crop andlivestock enterprises, crop-to-livestockprice relationships are often critical tofarmers' management decisions. This isparticularly true for barley. In fact, anyattempt to understand historical changesin the profitability of barley productionby examining barley grain prices missesthe point. To begin with, the value ofbarley straw often equals or evenexceeds the value of barley grain, sograin prices tell only part of the story. Inaddition, since most barley in theWANA Region is marketed indirectly inthe form of livestock, in order tounderstand the changing profitability ofbarley it is necessary to consider longtermmovements in livestock prices.In most WANA countries, prices oflivestock and livestock products havedemonstrated strong growth relative toprices of wheat, aided by rapid growth inconsumption averaging 4% per year(Oram 1988) and explicit or implicittrade barriers which have effectivelyrestricted imports. For example, in1I In a number of countries, large differencesbetween domestic prices and impon prices,combined with high consumer subsidies, haveactually led governments 10 favor impons overdomestic production in order to reduce fiscaloutlays for food subsidies.

18Algeria the ratios of mutton prices towheat prices more than doubled between1970 and 1987, while the price of barleygrain relative to wheat grain changedonly slightly (Table 9). Significantly,meat-to-grain price ratios in Algeria are40 times higher than similar ratios inAustralia, a major exporter of bothproducts and therefore representative ofworld price trends. Southern Australiancereal/livestock production systems havebeen established under climaticconditions nearly identical to those inmuch of the WANA Region, although(not surprisingly) stocking rates tend tobe considerably higher in the WANARegion than in southern Australia. InAustralia relative prices strongly favorcereals over livestock and provideincentives for adoption of improved cropproduction technologies that might notbe viable in the very different economicclimate of WANA countries. 12dramatically, with direct humanconsumption now restricted to a fewrural areas of Algeria, Morocco, andTunisia. However, feed use of barley hasrisen rapidly because of higher demandfor livestock products, driven by sharpincreases in incomes associated with theoil boom of the 1970s. At the regionallevel, feed use of barley has risen from65% of total production in 1960 to anestimated 90% today.With growth in cereal consumptionoutstripping production gains, imports ofboth food and feed grains have beennecessary to meet demand. Over the pastthree decades, wheat imports into theWANA Region have risen sharply(Figure 12). The region currentlyimports over 20 million tons per year,representing over 30% of totalconsumption requirements. In an effortto slow growth in wheat consumption, anumber of countries raised bread pricesconsiderably during the 1980s. Risingprices precipitated protests byconsumers in some countries andoccasionally led to violentdemonstrations, as in Tunisia, Morocco,and Egypt. These experiencesdemonstrate how policies favoringwheat consumption can provokesituations in which policy makers havelimited leeway to implement muchneededeconomic reforms. As a result,the WANA Region remains the onlypart of the world where most countriescontinue to subsidize wheatconsumption heavily.In most WANA countries, wheat'simportance as the leading staple foodand a primary wage good is recognizedthrough direct government participationin wheat marketing. Retail prices of bothbread wheat and durum wheat areuniversally regulated and usually highlysubsidized, with the goal of keepingwheat affordable for consumers,especially poor urban consumers.Subsidization of consumer wheat priceshas contributed to falling bread pricesand high levels of wheat consumptionthroughout the WANA Region, whereaverage per capita utilization is currentlyfar higher than in all other regions of thedeveloping world (Table 10).As a result of demographic changes(e.g., urbanization) and policy influences(e.g., consumer subsidies), there hasbeen marked substitution among cerealsin recent years throughout the WANARegion. Food use of barley has fallenTable 9. Grain-to-livestock price ratios in Algeria and Australia, 1970-72 and 1985-87I)roducer Ratio in Ilatio in PCTcentagcpr-ice ratio 1970-72 1985-87 chan~AlgeriaBarley: bread wheat 0.7 0.7 +3Mutton: bread wheat 23.2 45.5 +96AustraliaMutton: wheat 4.3 1.0 -77(lUrce: alcullllt:d from BQUlo'nl1l:t (19 9) an I dllla pr(lvided by the AU~LrnliaTI Bureau ( rgri ulLlHal and Resource Eoon\)mi s.Table 10. Per capita utilization of wheat by developing country region, 1987-89West AsiaNorth Africa1987-89 wheal uliiization(kg per capita)220204nnmd growth rate %)1961·63 to 1987-890.41.812 The higher quality of Australian wool. a jointproduct with mutton. also influences the relativeprofitability of livestock as compared to cereals.Sub-Saharan AfricaSouth AsiaSoutheast Asia and PacificEast AsiaLatin AmericaSour

19Rainfed MarginalEnvironments of South Asia 13The second major marginal environmentin which wheat is grown is the extensivedryland wheat area of South Asia, wherewheat production depends largely onresidual moisture from the monsoonrains. The rainfed wheat area of SouthAsia is concentrated in central andsouthern India, especially the northernpart of Madhya Pradesh (Figure 13), aswell as in northern Pakistan.Types of FarmingSystems in South AsiaThe rainfed wheat production zones ofSouth Asia include a range ofenvironments, some of which receiveabundant rainfall and therefore cannot becharacterized as marginal. For example,in the higher rainfall areas of northernIndia, rainfed wheat production systemsresemble adjacent irrigated systems,which are often based on a rice-wheatrotation. However, in central andsouthern India, dryland farming systemsare very distinct from those in adjacentirrigated areas. The following discussionfocuses mainly on these more marginalrainfed zones in central and southernIndia. Reference is also made to the dryrainfed (baroni) areas of northernPakistan, which receive somewhat morerainfall in the growing season but arecritically dependent on residual moistureand thus share many of the samecharacteristics as the systems of centralIndia.In central and southern India, drylandwheat is grown almost exclusively ondeep black vertisols. Because of theirhigh clay content, vertisols areexceedingly sticky when wet; on dryingthey develop large, deep cracks. Thismakes them very difficult to manageduring the monsoon period. However,vertisols have a high moisture retentioncapacity. In the deep vertisols ofnorthern Madhya Pradesh, 400 mm ormore of moisture may be stored in thesoil profile at the end of the monsoon, atleast half of which is available for winter(rabi) season crops such as wheat.Because of these special characteristics,the dominant rotation is fallow in themonsoon (kharij) season followed bywheat grown on stored moisture in therabi season, when rainfall is very low(usually under 100 mm). Over two-thirdsof the moisture available to the wheatcrop comes from stored moisture. Giventhat most rain falls during the kharifseason, however, the fallow-wheatsystem is relatively inefficient in usingavailable moisture, since only about one-quarter of the annual rainfall is used forcrop evapotranspiration.In addition to limited moisture. the othermajor cl imatic constraint on wheatproduction in this zone is hightemperatures. Practically the entiredryland wheat area of central andsouthern India lies in the tropical zone(defined as having mean Januarytemperatures above 17.S°C). Figure 14illustrates why the relationship betweenrainfall and temperature is critical fordryland wheat farmers. At the end of themonsoon in September, stored soilmoisture is at its maximum level.However, the temperature of the soil isalso high and will impede germinationand seedling emergence if wheat is sownearly. Over time, temperature declines,allowing germination and seedlingemergence rates to increase. However,farmers must be careful not to delayplanting too long; late planted wheatmust develop in a steadily recedingmoisture profile and runs the risk ofsuffering from severe drought duringflowering and grain filling. Theoptimum time for planting currentvarieties in these environments is lateOctober; harvesting usually occurs inlate February or March. This means thatthe critical period of heat stress is in theseedling stage.Metric t (millions)70-r----------------------------,605040302011 ul'llpli nProducllonIn the main dryland wheat belt of centralIndia, 70-90% of total area is left fallowduring the kharif season before beingsown to wheat in the rabi season,resulting in an overall cropping intensityof about 100% (Byerlee 1991). Pulses,chickpeas, and lentils are also importantin the rabi cycle and tend to dominate inareas where soils are shallow and lessfertile. In medium depth, better drainedsoils, soybeans have been found to besuitable for the kharif season and havespread quite rapidly in recent years insome areas. This has implications forwheal: where soybeans are grown in thekharif season, land will typically be left1O+"'T""".,......T'"""'1r_'T....,."""T'"_r...,....,_r-r1r-r~__r...,....,_.,......T'"""'1r_'T....,."""T'"_r"'T"""'T'"""'I1961 63 65 67 69 71 73 75 77 79 81 83 85 87 89Figure 12. Wheat production and consumption in the WANA Region, 1961-89.13 This section draws on Byerlee (1991),Hanchinal (1990), Sheopuria (1990),and JARJ (1990).

20x 40,000 ha rainfed wheat• 40,000 ha irrigated wheatfallow in the following rabi season,since sufficient moisture is available forwheat after soybeans in less than one outof every three years (Pandey 1986).Much of the expansion in soybean areahas occurred on larger farms where theopportunities for substitution betweenwheat and soybeans are greater. Incontrast, farmers with less land placeconsiderable priority on meeting theirsubsistence needs for grain and fodderand hence favor wheat production.Farther south where there is no risk offrost, farmers replace wheat with rabisorghum as the primary food crop andusually grow cotton as the main cashcrop.Throughout the dryland wheat areas ofSouth Asia, wheat is frequentlyintercropped. In central India, commonintercrops include chickpeas, lentils, andlinseed, whereas in southern India wheatis intercropped between rows ofsafflower. In Pakistan's barani tract,farmers commonly intercrop wheat withmustard, which is removed as a greenfodder before wheat flowering. Theseintercrop combinations appear to bemore profitable than sole cropping andare also likely to reduce risk (Hobbs eta!. 1985, Byerlee 1991).Figure 13. Distribution of rainfed and irrigated wheat area in India.Soil Soiltemp. moisture(0C) (%)at 5 cm at 6-10 cm Emergencedepth depth (seedlings/m 2 )47- 27 17046- 26 Oermtnullon45- 25 16044- 2443- 2315042- 2241- 21il moi. lure 14040- 2039- 1913038- 1837- 17 Soil temperature12036- 16 11035- 158 Oct 15 Oct 20 Oct 25 Oct 30 Oct 5 NovBecause livestock are important in thedryland farming systems of South Asia,straw is a valuable by-product fromwheat production. Straw typicallycontributes one-third of the total value ofthe wheat crop (at post-harvest prices),and in very dry areas or very dry years,the total value of straw may equal oreven exceed that of grain. Withincreasing tractor use and the spread ofirrigation, however, the value of strawmay be decreasing in some areas.Nonetheless, for small-scale farmerswho continue to use bullocks, productionof wheat straw for fodder is an importantobjective.Figure 14. Effect of soil temperature and moisture on germination of wheat, Indore,India, 1986-88.Source: lARI. Indore (1990).

21Use of Improved TechnologyVarietal development and diffusion­Traditionally a large part of the drylandwheat area of central and southern Indiawas sown to durum wheats, which wereused to make chapatis as well as manyspecialized dishes. Durum wheats stillpredominate in the driest and hottestareas (e.g., Gujarat, Maharashtra, andKarnataka), but most of the wheat sownis now bread wheat.Over the past three decades, India hasreleased numerous wheat varieties forrainfed production conditions. However,the physical characteristics of thesevarieties have varied according torainfall zone. Many of the varietiesreleased in the relatively well-wateredrainfed zones in northern India andPakistan have been semidwarf varietieswhich are also recommended foradjacent irrigated areas. In contrast,most of the varieties recommended forthe dryland areas in central and southernIndia have tended to be tall. Thisdifference reflects the much harsherconditions prevailing in central andsouthern India, where varietalcharacteristics such as drought tolerance,a long coleoptile (allowing the seedlingto emerge from where it has been deeplyplanted to benefit from residualmoisture), and the ability to germinate inhot conditions are more important thanthe input-responsiveness associated withsemidwarf materials. To date, wheatbreeders have been unsuccessful infinding semidwarf materials that performwell under these conditions. In recentyears, the proportion of varietiesreleased in India specifically for rainfedproduction conditions has declined. Thisis consistent with the decline in the shareof rainfed wheat production.Throughout India, over 50% of rainfedwheat area is now planted to MVs, upfrom only 15% in 1976. Varietaladoption data suggest that use ofimproved materials in India has beenclosely correlated with availability ofirrigation. Beginning in the late 1960s,the first Green Revolution varietiesmoved into irrigated districts. Adoptionin rainfed areas lagged, despite the factthat traditional land races grown indryland areas were highly susceptible toleaf and stem rust, which in some yearsresulted in severe losses (Sheopuria1990).Adoption of MVs in dryland areas ofIndia has been slowed by the fact thatyield gains in varieties released for theseareas have been relatively modest(Hanchinal 1988) or in some casesnegligible. For example, experimentsconducted over five years at Indorefound that local varieties yielded morethan the "improved" varietiesrecommended for the area. The slightlybetter performance of the local varietiesappeared to be due to their highergermination rate in hot soils.Nonetheless, despite their lowergermination rates, some improvedvarieties have been widely adoptedbecause they possess better diseaseresistance and excellent droughttolerance.Another critical factor in adoption ofMVs is quality. Many traditionalvarieties grown in rainfed areas producehigh quality grain, which allows them tofetch premium prices in the market.Since the quality of many MVs currentlygrown in irrigated areas is lower, thesupply of high quality wheats has tendedto decline, causing the price premiumfor grain quality to widen. Farmers indryland areas have responded byspecializing in quality wheats to takeadvantage of this price premium.Improved varieties developed fordryland areas therefore must possessqual ity characteristics that meet farmers'criteria. In Gujarat, for example, thesuccessful durum variety GW-I(released in 1986) showed only a modestyield advantage under drylandconditions over the farmers' varietyA-206 (released in 1954) (WheatResearch Institute, Vijapur, unpublisheddata). Nonetheless, GW-I has replacedA-206 over a significant area because ofits improved rust resistance and higherquality (and hence higher price).Experiences are simi lar in the haranitract of Pakistan. In the higher rainfallareas (>500 mm), MVs were adoptedrapidly beginning in the mid-1970s withthe release of the semidwarf varietyLyallpur-73. Although originallyintended for irrigated areas, Lyallpur-73also performed well under rain fedconditions. During the 1980s, Lyallpur­73 was largely replaced by anothervariety recommended for irrigated areas,Pak-81 (based on material fromCIMMYT's spring x winter wheatcrossing program). However, in lowerrainfall areas «500 mm), local varietiescontinue to dominate (Table II).Varietal trials indicate that the firstgeneration semidwarfs such as Lyallpur­73 yield only 5-10% more than localvarieties under improved management;since the local varieties enjoy a pricepremium of roughly 15%, this yieldadvantage is insufficient to encourageadoption. The second generation MVssuch as Pak-81 show a yield advantageof 16% and have been adoptedsomewhat more widely. Given theseTable 11. Varietal adoption by rainfall zone, northern Punjab, Pakistan, 1990Higher rainfall (> 500 rom)1% area)Local varieties 15First generation semidwarfs(e.g., Lyallpur-73) 17Second generation semidwarfs(e.g., Pak-81) 68Low nunrall « 500 mm)(% area)73819Source: Ahmad and Ahmed (1991).

22experiences, it remains to be seenwhether Pakistani breeders' strategy ofrelying exclusively on semidwarfmaterials eventually will succeed inproducing varieties suitable for thedry areas.Improved management practices andinputs-For all India, fertilizer use onwheat now averages around I 10 kgnutrient/ha, compared to only 25 kgnutrient/ha in 1969. As with MVs,adoption of fertilizer began later andproceeded more slowly in dryland wheatareas than in irrigated and high rainfallareas. By the mid-1970s, fully 70% ofthe irrigated wheat area was fertilized, atan average rate of close to 80 kgnutrient/ha. By contrast, only about 10%of rainfed wheat area was fertilized, atan average rate of only 35 kg nutrient/ha(Desai 1982).14 While the extent andlevel of fertilizer use in rainfed wheatareas are still low, substantial progresshas been made over the past decade.Currently perhaps half the dryland wheatarea in central and southern India isfertilized, and increased fertilizer usemay account for much of the modestincrease in dryland wheat yieldsrecorded since 1970 (Byerlee 1991).Fertilizer use on rainfed wheat variesdepending on numerous factors,including prevailing moisture conditions,soil types, and cropping patterns. Thegeneral recommendation is to apply40-20-0 kg NPK/ha, but farmersapproximate these doses only in areas ofbetter rainfall and deep soils. In drierareas, farmers who apply fertilizer useonly 10-20 kg nutrient/ha (equallydivided between Nand P 20J However,in the few years benefiting from timelyrainfall in the growing season, farmersmay top-dress nitrogen to boost yields.Variation in fertilizer use reflectsexpected crop response to fertilizer.At low application levels, the averagegrain-to-nutrient response ratio for14 These data. which represent an average for all ofIndia. are likely to overestimate the use offertilizer in drier areas of cenrral and southernIndia.rainfed wheat in Madhya Pradesh isabout 10 kg wheat!kg nutrient (FertilizerAssociation of India 1985, Rao 1976).However, in drier areas the response isconsiderably lower; for example, Kohli(1976) reports a response ratio of 5.5: 1in Maharashtra. In extremely dry areas,the response may even be negligible. Noresponse at all to fertilizer was reportedin over six years of fertilizer trials inGujarat's dryland wheat areas (Maliwal1989).Moisture conservation is critical to thesuccess of dryland wheat production inSouth Asia, since the wheat cropdepends on stored soil moisture. Duringthe monsoon season, most farmerstherefore attempt to maintain a cleanfallow that improves water infiltration~nd reduces the amount of moisture lostto weeds. Many farmers use a shallowscraper-harrow to kill weeds and leave ashallow soil mulch in the top 5 cm of thesoil surface that helps to conservemoisture.Another widely used soil conservationtechnique is to level and bund fields,which allows monsoon rainfall to standin the field for up to two months. Thistechnique, locally called the havelisystem, achieves several objectives: itprevents runoff and hence soil erosion; itconserves moisture, since 50% ofmonsoon rainfall may be lost to runoff(Singh and Raje 1984); and it helpsprevent the germination of weeds in thekharifseason. The main disadvantagesof the haveli system are the high capitalcosts of levelling the field andconstructing bunds, as well as the loss inflexibility that results from the need todedicate a bunded field to rahi cropping(farmers therefore cannot substitutebetween rabi and kharifcrops). Despitethese drawbacks, the practice hasbecome widespread in the main drylandwheat belts of central India.Improvements in the physical structureof soils may also help conservemoisture. For example, promisingexperimental results are being obtainedfrom rotating wheat with deep-rootedcrops like castor, which help to improvethe physical structure of soils (IARI1990). Organic manure, such as farmyard manure and green manure, alsoimproves moisture conservation.Farmers in the dryland wheat area ofPakistan concentrate farm yard manureon fields close to the village. Theimproved moisture-holding capacity ofthese fields not only boosts wheat yieldsbut allows double cropping. Research inthe same area has also demonstrated thatdeep tillage with a moldboard plowgreatly increases moisture storage androot penetration, providing an averageincrease in wheat yields of 25%. Duringthe past decade, farmers in dry areashave rapidly adopted this practice.By concentrating crop production in thedry season, the current fallow-wheatsystem is inefficient in utilizingavailable moisture; on average, less than1 kg/ha of grain is produced for each1 mm of annual rainfall. One logicalstrategy for using monsoon rainfall moreefficiently would be to plant kharif landthat is currently left fallow. TheInternational Crops Resarch Institute forthe Semi-Arid Tropics (lCRISAT) hasdevoted a decade to research on atechnology to allow kharifcropping ofdeep vertisols in a double croppingsequence (Foster et al. 1987, Walker andRyan 1990). The ICRISAT technology,which involves the construction of broadbeds and furrows across the slope tofacilitate drainage and moistureconservation, has been extensivelydemonstrated over several years incentral India, with only partial success.Farmers tend to adopt elements of thepackage, but the key component-abullock-drawn wheeled tool carrier-hasnot been adopted because of high initialinvestment, high operating costs,increased availability of tractors, and theuncertainty associated with doublecropping. Low adoption is alsoexplained by farmers' strong desire tomaintain a considerable proportion ofarea under wheat both for subsistencegrain and fodder production. As noted

23earlier, double cropping by plantingwheat after soybeans greatly decreasesthe probability that sufficient moistureremains for the wheat crop and augmentsthe risk of crop failure in the rabi season(Pandey 1986).Investment in irrigation facilities is themost widely used strategy to protectagainst drought in central and southernIndia. Rural electrification facilitatedrapid growth in the number of tubewellsover the past two decades, so that over40% of total wheat area is now irrigated.However, chronic overexploitation ofgroundwater in many areas has loweredwater tables, raised the cost of irrigation,and in some cases, led to serious watershortages in dry years. Given thesetrends, the rate of expansion of irrigatedwheat area is likely to decline in thefuture.In most areas, scarce tubewell water isallocated to wheat before other crops.Nonetheless, tubewell water is oftendepleted before the end of the growingseason, so the wheat crop receives only alimited number of irrigations. However,most studies show that even oneRatio1.51.41.31.21.11.00.90.80.70.60.50.40.30.20.1oFine white bread wheat(Sagar)irrigation at planting will increase yieldsby 0.5-1.0 tfha. Hence an importantpriority for research is to developappropriate recommendations for thelarge area of wheat grown under limitedirrigation.The Policy Environment for WheatSince 1965, real prices received byIndian farmers for wheat have steadilyfallen, except for a short period duringthe world food crisis of the mid-1970s(Sidhu and Byerlee 1991). Differentgrades of wheat command differentprices in India's wheat markets: pricesare higher for bread wheat produced inrainfed areas than for bread wheatproduced in irrigated areas, reflecting apremium for the higher quality ofrainfed wheats. As mentioned earlier,farmers in rainfed areas have capitalizedon this market opportunity byspecializing in premium quality durumand bread wheats. Durum wheat carriesa significant price premium over breadwheat. This price premium, althoughapparent even before the GreenRevolution, had increased by the 1980sto about 50% (Figure 15). Consequently,Durum wheat(central and southern India)1954-60 1963-67 1968-74 1981-85--although all Indian wheat farmers haveseen real wheat prices decline since theGreen Revolution, this decline has beenless marked in rainfed areas than itwould have been if these areas did notenjoy a comparative advantage inproducing premium quality wheats.Toward a Research Strategy forMarginal EnvironmentsThe evidence reviewed in this reportsuggests that researchers have onlypartially succeeded in developingimproved technologies for wheat andbarley in rainfed marginal environments.Not only have yields in marginal areasremained low, but the rate at whichyields have grown has lagged comparedto favorable areas. These macro-leveltrends have been borne out by microlevelevidence on technology adoption,which shows that, in dryland areas, MVsand improved crop managementpractices have been adopted relativelyslowly, less extensively, and with lessdramatic results.But the less pronounced impact ofimproved production technologies inmarginal environments does notnecessarily mean that the institutionsresponsible for developing technologyhave somehow failed. At least threerelated factors help explain the relativelyslow rate of progress in marginalenvironments. First, the climate indryland production zones severelyconstrains the yield potential of cerealcrops, so the impact of improved seedfertilizertechnologies is bound to be lessdramatic than in the more favored areasof the world where these technologiesare so successful. Second, investment inagricultural research targeted at rainfedareas has been modest, in part becausesuch research was perceived (correctly)as having a lower potential payoff.Third, largely because of the first twofactors, many countries have been slowto implement policies that wouldpromote cereal production in rainfedareas, such as policies to developFigure 15. Ratio of wholesale prices of bread wheat in Sagar and durum wheat incentral and southern India to prices in Ludhiana, Punjab, India, 1954-85.

24Research Resource Allocationto Marginal Environmentstransportation infrastructure, improvedinput delivery systems, and marketingfacilities. Thus, it is not really correct tosay that research institutions have beenunsuccessful in developing improvedproduction technologies for rainfedmarginal environments. Rather, it wouldseem that some important conditions forsuccessful development and adoption ofimproved technologies have simply beenlacking.Policy makers in many countries mustdecide what proportion of availableresearch resources should be allocated tores arch targeted at marginalenvironments. Although this complexissue cannot be resolved here, it seemsappropriate briefly to describe onesimple approach that can be used toinform research planning. Undercongruency analysis, re earch resourcesare allocated among regions orenvironments in relation to their share inthe value of production. For example, iffavorable and marginal areas account for70% and 30%, respectively, of the totalvalue of wheat production, then wheatresearch resources should be allocated tothe two types of environment in a 70:30ratio (Barker 1988, Scobie 1994 .Research managers both in theinternational centers and within nationalprograms must now decide whether it istime to devote more resources tomarginal areas. A comprehensive answerto this question is beyond the scope ofthis report, but preliminary analysissuggests that further information may beneeded before any decision is taken toshift significant amounts of researchresources from favored to marginalareas, at least in the case of wheatbreeding (see box, this page). Onereason for this conclusion is thatalthough marginal environments accountfor a significant proportion of totalcereal area, their share in total cerealproduction is much lower. Since mostpolicy makers would agree thatproduction is the single most importantconsideration in allocating researchresources (certainly more important thanarea per se), caution should be exercisedbefore allocating research resources onthe basis of area alone.On the other hand, further considerations(especially the desire to alleviatepoverty) might justify increasedallocations to research targeted atmarginal areas. Without a doubt, manypeople who live in marginalenvironments are poor. However, theissue is complicated by the fact thatmany of the poorest people in marginalenvironments are net food purchasers;thus, the most efficient way to alleviatepoverty in marginal environments mayactually be to increase productivity infavored areas, jf that reduces the realprice of food. Alternatively, poverty inmarginal environments may beTable A. Indices of wheat area, production, and weighted productionby mega-environmentEstimated Estimated WeightedMega-environment area production production(characteristics) (%) (%) (%)Bread wheat 89.5 92.4 92.1Spring habitME 1 (irrigated, low rainfall, temperate) 36.1 42.7 43.4ME 2 (high rainfall, temperate) 8.5 10.4 9.5ME 3 (acid soil, high rainfall, temperate) 1.9 1.3 0.4ME 4A (low rainfall, temperate, winter rain) 6.1 2.3 2.14B (low rainfall, temperate, winter drought) 3.6 2.1 0.84C (low rainfall, temperate, stored moisture) 4.9 2.5 2.1ME 5A (high temperature, low humidity) 4.4 4.9 5.75B (high temperature, low humidity) 3.6 1.5 1.2ME 7 (severe winter, spring sown, high latitude) 6.2 6.8 5.8Subtotal (spring habit) 75.3 74.5 71.1Facultative/winter habitME 6A (moderate cold, high rainfall) 5.1 9.8 10.96B (moderate cold, low rainfall) 7.4 2.0 2.46C (severe cold, high rainfall) 6.7 9.2 11.560 (severe cold, low rainfall) 5.5 4.6 4.2Subtotal (facultative/winter habit) 24.7 25.5 29.0Bread wheat total 100.0 100.0 100.0Durum wheat 10.5 7.6 7.9Spring habitME 1 (irrigated, low rainfall, temperate) 3.6 7.9 6.9ME 2 (high rainfall, temperate) 23.0 33.6 46.6ME 4A (low rainfall, temperate, winter rain) 45.6 32.0 23.54C (low rainfall, temperate, stored moisture) 14.5 8.7 5.0Subtotal (spring habit) 86.7 82.1 81.9Facultative/winter habitME 6C (severe cold, high rainfall) 1.6 5.6 6.760 (severe cold, low rainfall) 11.7 12.3 11.4Subtotal (facultative/winter habit) 13.3 17.9 18.1Ourum wheat total 100.0 100.0 100.0Total bread wheat and durum wheat 100.0 100.0 100.0

Congruency analysis was recently usedat IMMYT in preparing the Center'sfive-year budget plan (CIMMYT 1989a).To a sist in determining an appropriatepattern of research resource allocationbetween the different environments andwheat types on which CIMMYT's wheatbreeding focuses, an index wasconstructed of wheat production indeveloping countries. Production datawere disaggregated by megaenvironmentand weighted according toa set of criteria deemed important inlight of CIMMYT's mandate: I) theexpected value of wheat production,2) the per capita income of wheatproducers and consumers, 3) theperceived ability of national researchprograms to carry out effective researchon different types of wheat, and4) expected rat s of progress due towheat breeding in different megaenvironment(Table A).The distribution (by type of wheat andby mega-environment) of weightedglobal wheat production was thencompared to the current allocation ofresearch re Ource within the CIMMYTWheat Program. The objective was todetermine whether the resourcesallocated to different wheat researchactivities were congruent with theweighted value of production, takinginto account the factors deemedimportant in terms of CIMMYT'smandate. Thus, bread wheat was foundto make up approximately 92% ofweighted production, while durum wheatwas found to make up only 8%. Only10% of weighted bread wheat productionwas found to take place in low rainfallenvironments, as compared to 90% formoderate and high rainfallenvironments. In contrast, fully 40% ofweighted durum wheat productionoccurs in low rainfall environments.These results suggest that caution shouldbe exercised before resources devoted tobread wheat improvement are shiftedfrom favorable to marginal areas, butthat in durum wheat improvementconsiderable emphasis should be placedon research targeted at marginal areas.In a more focused application ofcongruency analysis, Byerlee (1991)examined whether the current allocationTable B. Summary of congruency analysis of allocation of Indian wheat researchresources by zoneIrrigatedRainfedNorthwest Northeast Central Centralplains plains and South North and SouthPercent of area a 38.6 25.4 9.9 10.1 16.1Percent of value of production b 49.4 24.5 11.5 6.3 8.2Percent weighted production" 31.8 39.5 18.6 4.5 5.6of Indian research resources to wheatbreeding for rainfed areas is congruentwith the value of production of rainfedwheat. Indian wheat production wasdivided into three major zones and twomoisture regimes. Table B presents dataon area planted to wheat and a weightedindex of production in each zone (usinga weighting procedure similar toCIMMYT's), as well as a measure of theallocation of research resources (asindicated by the number of varietalreleases for each zone). In general, theshare of varieties released for rainfedenvironments corresponds well torainfed environments' share in theweighted value of production. Althoughdryland areas of central and southernIndia account for 16% of total wheatarea, they account for only 6% of theweighted index of production. Thisdifference is due to the very low yieldsin dryland areas, as well as the slow rateof yield gains (0.5% per year) expectedto be obtained through breeding in thiszone. On the other hand, rural incomelevels in the dryland areas wereestimated to be only half of those in theirrigated northwest, and this increasedthe weight assigned to the dryland areas.However, in order to justify increasedallocation of research resources todryland wheat breeding, planners need tobelieve that research progress in the eareas will be much more rapid than inthe past, or they need to attach a veryhigh weight to alleviating poverty.Both of these examples of congruencyanalysis consider only the researchresources invested in breeding. Giventhe evidence presented in this report, thepotential to increase wheat productivitythrough research on crop and resourcemanagement may by relatively high.Hence a congruency analysis ofallocation of resources devoted to cropand resource management research maygive quite a different picture.Percent of varieties released:1960-85 26.9 21.3 31.5 7.4 13.91976-85 31.0 22.4 36.2 6.9 3.4Source: Byerlee (1991).a Current area and production.b Based on projected production and prices.C Weighted by reciprocal of agricultural income per capita and rates of research progress.

26alleviated most effectively by creatingopportunities for work off of the farm.(For a more detailed discussion of thisand related issues, see the box, p. 27).Despite these caveats, it seems clear thatmore effort will be needed to improvecereal productivity in marginalenvironments where yields remain wellbelow potential. However, prospects forfuture gains in productivity, as well aspromising research strategies, differsomewhat between the WANA Regionand South Asia.In the WANA Region, cerealproductivity in very low rainfall zonesshows little potential for improvement,given the adverse agroclimaticconditions and high levels of productionrisk, which combine to limitprofitability. In view of the long-termdownward trend in world wheat prices,coupled with structural adjustmentreforms currently being implemented inmany WANA countries, high-cost localwheat production is unlikely to receiveadditional subsidies, thus lessening theincentives to raise productivity in wheat.Barley will continue to have relativelygreater importance in low rainfall zones,especially for use on the farm as an inputinto the livestock enterprise, but it seemsunlikely that livestock prices-alreadyhigh by global standards-will rise tolevels capable of inducing significanttechnical change in barley productionpractices. Meanwhile, in higher rainfallzones and irrigated areas, both wheatand barley are likely to face increasingcompetition from higher valued crops.Eliminating the driest and wellest areasof the WANA Region leaves a relativelynarrow belt of land, characterized byroughly 300-400 mm of annual rainfall,where cereal yields are low butconsiderable potential exists to increaseproduction. Within this belt,opportunities for expanding arable landare limited, so production gains are mostlikely to be realized through increasedyields, as well as through increasedcropping intensity achieved by bringingfallow land into cultivation. This beltwill be important in meeting the region'sgrowing demand for grain, which givencurrent trends is projected to result in animport deficit of 60 million tons by theyear 2000.In South Asia's dryland wheat zones, thepotential for increasing wheat yieldsunder residual moisture conditionsappears to be much more limited. Innormal years, maximum yield potentialis only around 2 tfha, and in a goodgrowing season when rainfall is timely,maximum yield potential rarely exceeds3 tfha. Nevertheless, wheat will continueto occupy an important place in thefarming systems of these environments.Not only is it a preferred subsistencefood and fodder crop, but it is alsoregarded as a secure crop. In higherrainfall areas, the level of residual soilmoisture is nearly always sufficient toguarantee at least a modest harvest; inlower rainfall areas, farmers protectagainst crop failures by adjusting areaplanted according to the availablemoisture. Input use is low, as wheatproduction requires only minimumtillage, seed, and sometimes a small doseof fertilizer; after planting, no weeding isneeded, and the only other necessaryoperation is harvesting. On average,farmers can recoup variable costs with ayield of only 200-300 kg/ha. In the maindryland wheat belt, the probability isvery high that farmers can obtain yieldsof at least that level. Because of thesecurity it provides, wheat will notdiminish in importance in the rainfedmarginal environments of South Asia,even though only modest gains in .productivity can be expected.Where will future productivity gainscome from in marginal environments?Given that moisture is the key yieldconstraint, three strategies offer the mosthope for increasing and stabilizingyields: I) increasing the water supplythrough irrigation, 2) improving moistureconservation, and 3) utilizing moisturemore efficiently. Of the three options,the first is clearly the most effectivesolution. For this reason, expansion ofirrigated area has long been the majorstrategy employed in India to improveproduction in marginal areas; morerecently, many WANA countries havefollowed suit. However, aside from thefact that irrigation is technically feasibleonly in some areas, pUlling rainfed landunder irrigation is extremely costly.Also, even where irrigation has beensuccessful (such as in central India), itslong run sustainability is uncertain,given limited groundwater supplies.Therefore, the two remainingstrategies-improving moistureconservation practices and enhancing theefficiency of moisture use-should notbe overlooked. Research has consistentlyhighlighted the critical role of improvedcrop and soil management practices inconserving moisture and raising theefficiency with which moisture is used inmarginal areas (Cooper et al. 1988;Durutan, Yilmaz, and Kiziltan, 1988;Durutan et al. 1989; Harris and Pala1988; Byerlee and Winkelmann 1980).Although MVs may play some role inboosting wheat and barley yields inmarginal areas, germplasm will usuallynot be the main stimulus for rapidtechnical change, as was the case infavorable areas. On the contrary, inmarginal environments improvements incrop and soil management practices willoften precede changes in variety (as hasalready happened in Turkey). Thissuggests a larger role for research oncrop and soil management relative tobreeding research, as well as somereorganization of research and extensionstrategies.Traditional commodity-based researchprograms focusing exclusively on cerealcrops are likely to have limited successin marginal environments, given theimportance of intercropping, croprotations and fallowing, and croplivestockinteractions. Instead, a moreintegrative approach to crop and soilmanagement is needed, in whichresearchers give allention to systemwideimplications of new technologies.The role of extension will also be criticalto diffusing new technologies. The"technology package" approach sowidely employed in favorableenvironments is likely to be less relevant

Interregional Effects of Technological Change:Evidence from Favored and Marginal Environments of PakistanIt is now generally accepted thatimproved seed-fertilizer technologieshave succeeded in raising the incomes ofadopting farmers in a wide range ofproduction environments. Morecontroversial is the question of how thegains attributable to technologicalchange have been shared among varioussocioeconomic groups and betweenregions. If MVs and associated inputshave disproportionally benefitedwealthier farmers and/or people living infavorable production zones, then thetechnology may have widened theincome gap between the rich and thepoor, as well as between inhabitants offavored and marginal areas.This issue was addressed by a recentstudy (Renkow 1991) carried out inPakistan, a country with well-definedfavorable (irrigated) and marginal(rainfed) wheat production zones. SinceMVs were introduced in the mid-1960s,most of the improved germplasmdeveloped for use in Pakistan has beenbetter suited to irrigated productionconditions. Following a lag of about 10years, MVs eventually spread intorainfed areas as well, although theresulting yield increases weresignificantly smaller than in irrigatedareas (average wheat yields in rainfedareas are currently less than half tbose ofirrigated areas).Interestingly, even though yieldincreases in irrigated areas haveconsistently outpaced those in rainfedareas, rural incomes in rainfed areashave grown more, particularly since themid-1970s. In both rainfed and irrigatedareas, large farm households haveconsistently enjoyed the highest incomesfrom all sources (including cropproduction, agricultural labor, andnonagricultural employment). However,during the past 25 years poorer smallfarm and landless households registeredfaster real income growth (mainlybecause of relatively greater increases inoff-farm income) (see Figure).Particularly striking is the rapid growthof the incomes of poorer households inrainfed areas.production technology. Aside fromdirectly influencing the level ofproduction, technological change canalso produce benefits that are transmittedindirectly, possibly with differentconsequences for various socioeconomicgroups. For example, by increasingproduction of a commodity such aswheat, a new technology may help lowerthe price of the commodity, making itcheaper than it would have been in theabsence of technological change. For animportant staple food such as wheat inPakistan, net consumers of the foodbothnon-producers (such as urbandwellers) and rural producers for whomproduction fails to meet householdconsumption requirements-stand tobenefit from lower wheat prices.Moreover, because poor consumers tendto spend a greater proportion of theirincomes on food, this price effect islikely to benefit poorer consumers to agreater extent than the more well-to-do.It is important to note, however, thatincreased wheat production will directlylead to a lower wheat price only if theprice is determined by the forces ofdomestic supply and demand. InPakistan, as in most other developingcountries, the government plays anactive role in setting producer andconsumer prices for wheat and flour, andthese prices may remain largeJyunaffected by domestic supply andReal income (Rs/yr)200016001200800400Irrigated areasRainfed areas1-_..1demand conditions. Thus, whileproducer and consumer prices of wheatand wheat products have been decliningover the past 25 years, it is not clear thatthese price changes can be attributeddirectly to technological change inwheat production.In addition to increasing productivity,technological innovations such as MVsusually require greater amounts of laborfor harvesting, threshing, and crop care.As long as agricultural labor is notavailable in unlimited supply, increasedlabor requirements put upward pressureon agricultural wage rates (as wel'l as theimplicit return to family labor ofsubsistence producers). Higher wagesobviously have positive effects on theincomes of poorer rural dwellersdependent on agricultural labor for alarge share of their incomes. Moreover,such wage increases need not beconfined to the areas in which laborintensivetechnologies are adopted ifinterregional differences in wage ratescause agricultural laborers to migratefrom low wage to high wage areas.In Pakistan, real agricultural wages inboth rainfed and irrigated areas haveincreased steadily over time, supportingthe idea that the increased laborrequirements of the Green Revolutiontechnologies have benefited agriculturallaborers in both favorable and marginal(continued, p.2S)Renkow's study attempted to unravel themany factors influencing income growthin an attempt to isolate the effectsattributable to changes in wheat1965 66/71 76/82 84/87Landless households1965 66/71 76/82 84/87Small- farm householdsRelative real incomes in the Punjab, Pakistan, 1965-87.1965 66/71 76/82 84/87Large-farm households

production environments. However,wages in nearly all sectors of theeconomy have risen dramatically sincethe mid-1970s. This period coincidedwith the beginning of large-scalemigration to the Middle East, an activitythat created labor shortages in many keysectors of Pakistan's economy (includingthe agricultural sector).Foreign migration appears to be themain explanation for the pattern ofchanges observed in the incomes of ruraldwellers in Pakistan's rainfed areas sincethe mid-1970s. Particularly for landl 5Sand smaIl farm households, anincreasingly large hare of totalhousehold income has come fromnonagricultural sources of whichforeign remittance are most important).Between 1965 and 1971, about 65% ofthe income of small farms came fromagricultural activities; between 1984 and1987, nearly 90% of the e households'in omes came from nonagriculturalsources. The data for landIe, . and largefarm households in rainfed areas indicatesimilaT although somewhat Ie'dramatic) trends in th composition ofhousehold income.Meanwhile, since 1965 the compositionof household income for all types ofhouseholds in irrigated areas hasremained relatively constant. In thesearea " the fortunes of rural dwellersappear to have been tied more directly tothe profitability of agriculture.Thus it appear' that rural income growthin rainfed areas of Pakistan has been duemainly to the ability of inhabitants ofthose areas to take advantage of incomegeneratingopportunities out:ide theagricultural sector. That the rate ofincome growth in rainfed areas hasexceeded that of irrigated areas over thepast 15 years indicates that thosenonagricultural opportunities proved tobe even more remunerative thanagricultural opportunities in the irrigatedareas. An interesting implication of thisfinding is that differences in wheatproductivity across productionenvironments may have provided animportant incentive for the inhabitants ofPakistan's less productive rainfed areato broaden their income-generatingactivities out 'ide of the agriculturalsector.in marginal areas, where the high degreeof agroclimatic variability over spaceand through time means that farmersmust be able to select among an array oftechnological options to meet therequirements of a specific field orseason. This also implies the need forbetter education of farmers to ensure thehigher skill levels needed to manageincreasingly complex technologies.Likewise, an appropriate policyenvironment is also critical to encouragethe adoption of improved crop and soilmanagement practices, especially sincemany marginal areas lack effective inputdelivery systems and marketinginfrastructure. Since economicincentives in many marginal areas stillfavor livestock production, price policyreforms may be necessary to increaseexpected returns to cereal enterprisesand/or reduce their riskiness. Preciselybecause they are characterized by a highlevel of climatic variability, dry landenvironments will require a strongerjoint technology/policy effort if cerealproduction practices are to changesignificantly.Too, there is a need to remain aware ofthe potential effects of technologicalchange on fragile ecologies. Higheryields and increased crop-ping intensityare desirable goals only if they can bemet without seriously degrading theenvironment. Until significantproductivity increases are achieved,rising demand for food will likely forcefarmers to further expand the area sownto wheat and barley into even morefragile ecologies. Thus, as farmers,researchers, and policy makers confrontthe formidable challenge of raisingproductivity and incomes in rainfedmarginal environments, they must payclose attention to the longer termimplications of short- and medium-termproduction strategies.One final point emerging from thisreport concerns the extreme diversity ofmarginal environments. The two majorenvironments reviewed here-the winterrainfall areas of the WANA Region, andthe residual moisture areas of SouthAsia-are characterized by tremendousvariability in temperature and rainfallpatterns, soils, cereal types, andsocioeconomic circumstances. Thisvariability poses a special challenge forresearchers seeking to developappropriate technologies capable ofincreasing productivity and, at least inthe case of the WANA Region,stabilizing production. Recognizing thediversity of marginal environments alsounderlines the importance of identifyingthe different types of marginalenvironments in which wheat and barleyare grown as a basis for developingappropriate research strategies for theyears to come.

Part 2: The Current World Wheat SituationIgnoring short-run fluctuations causedby weather variability and policychanges, world wheat productioncontinues to rise along a long-termupward trend. However, the sources ofgrowth in global wheat production havechanged over time (Figure 16). Boostedby increasing use of improvedgermplasm, fertilizer, and otherpurchased inputs, wheat yields at theglobal level have grown strongly at2-3% per year throughout the past fourdecades. In contrast, the rate of areaexpansion has slowed dramatically, withtotal area planted to wheat actuallydeclining during the decade of the1980s. Much of the contraction in areahas occurred in the industrializedcountries, where chronic wheatsurpluses have led to policies encouragingfarmers to take land out ofproduction.World wheat production in 1990-91 isestimated at a record 589 million metrictons (MT), an increase of nearly 10%over the previous year. The enormousworld wheat harvest of 1990-91 can beattributed primarily to two factors. First,area planted expanded considerably inresponse to high prices at planting time(a legacy of reduced North Americanproduction in 1987 and J988, whichresulted when significant area was keptout of production due to policyincentives to set aside farm land).Second, favorable weather in all majorwheat-producing zones of the northernhemisphere enabled wheat farmers toachieve record yields.Despite a significant increase in feed useof wheat, world wheat production in1990-91 is projected to surpassutilization by 5%, leading to the biggestincrease in stock volume since the early1980s. This indicates that efforts by themajor exporting countries to rein inchronic overproduction have notsucceeded, and it suggests thataggressive surplus-disposal efforts andexport competition will likely continuein the near future.Wheat ProductionIncreased plantings, combined withrecord yields, led to all-time high levelsof production in many of the world'sleading wheat producers, including theUSSR, China, USA, and Canada. Whilenot reaching record levels, productionnevertheless increased significantly inthe EC-12, Eastern Europe, Australia,and Argentina. India and Pakistan alsoenjoyed good harvests, while Egyptincreased production by 25% on thestrength of yields averaging over 5.0 t/ha(all wheat in Egypt is produced underirrigation). The only major producer tosuffer a significant production declinewas Brazil, which experiencedextremely unfavorable growing-seasonweather.Wheat UtilizationBuoyed by demographic growth, wheatfood use at the global level is projectedto grow 3% in 1990-91 despite efforts inmany developing countries to curtailconsumption of imported wheat. Inaddition, falling prices during late 1990and early 1991 encouraged increasedfeed use of wheat as livestock producerssubstituted large amounts of wheat forother relatively more expensive feedgrains such as maize and sorghum.Increased feed use helped to raise totalwheat utilization in 1990-91 to anestimated 563 MT, up nearly 5% fromthe previous year but still well short ofworld production.International Wheat PricesAverage annual growth (%)6-r------------------------..,5432o-IProduct! n (43)Production ( .0)Production (3.2)Produ ·ti n (1.8)-2 ....1..- ...1World wheat prices fell sharply in 1990­91 as a result of the record harvest. Themain international reference price forwheat (No.2 Hard Red Winter, F.O.B.Gulf Ports) decreased from US$ I69/t inJanuary 1990 to US$ I 14/t in January1991, a decline of nearly 33%. Thissteep decline pushed real wheat pricesback below their long-term trend level,which they had regained only two yearsearlier in response to weather-induceddeclines in production, as well as policyadjustments made in a number ofimportant exporters (Figure 17).Although the US export price is widelyused as a barometer of global pricetrends, it is important to note that thisprice does not necessarily reflect theprices at which transactions in the global1950s1960s1970s1980sFigure 16. Sources of growth in world wheat production, 1950s-80s.

30market actually take place. In recentyears, leading wheat producers such asthe USA and the EC-12 have reliedheavily on export promotion measures todispose of surplus production. Becauseof these export promotion measures,which include concessional sales,subsidized credit, and food aid, asignificant portion of the wheat movingthrough global markets in fact is sold atprices well below commonly citedinternational reference prices based onUS export prices. Figure 18 presents datarelating to US wheat exports in 1987(the latest year for which such figuresare available) showing that only onequarterof all US wheat exports weresold at the commercial export price.Subsidized credits provided through theExport Enhancement Program and otherprograms reduced the effective price fornearly one-third of US exports to around70% of the commercial export price,while concessional sales providedthrough P.L. 480 and other food aidprograms reduced the effective price onnearly one-quarter of US exports to only10% of the commercial export price. Alarge portion of the wheat exported bythe EC-12 similarly is sold at prices wellbelow international reference prices.Since the proportion of the total worldwheat trade affected by exportpromotion programs has been increasingsteadily, the long-term decline incommonly cited international referenceprices based on the US export price infact understates the decline in effectivewheat prices actually prevailing inglobal markets. The International WheatCouncil estimates that traded prices in1990 ranged from US$ 68/t toUS$ 85/t, the lowest levels seen since1972 (International Wheat Council1990).Outlook for WheatConsiderable uncertainty continues tosurround the long-term outlook for theworld wheat market. The sharp rise inproduction recorded in 1990-91 and theresulting increase in world wheat stockssuggests that the industrialized countrieshave done little to curtail overproduction.While importing countries continue tobenefit from depressed world prices,policy makers in many developingcountries have begun to express concernover rising levels of dependency onimported wheat and wheat flour. Despitelow world prices, strong consumptiongrowth has caused the foreign exchangecost of wheat imports to escalatedrastically in many developing countries,prompting governments to initiate policy(J 985 us $/1)40035030025020015010050Real U export pric .ClfwheaJ#2. Gulf pon.)1976 1981Figure 17. Evolution of real world wheat prices, 1961-90.10090807060504030201001961 1966 1971Percentage paid of export price (%)00102030 40 50 60 70Percentage of US wheat exportsreforms aimed at reducing consumption.These policy reforms have begun toproduce results in a number of Africanand Latin American countries, whichhave seen wheat imports level off oreven decline.In the short run, current high stock levelsand continuing support to wheatproducers in many industrializedcountries suggest that wheat will remainplentiful in global markets at affordableprices.80901986100Figure 18. Effective price of US wheat exports, fiscal year 1987.Source: SkuJly (1990)

31Over the longer term, prospects are moreuncertain. Future developments in theworld wheat market are likely to beinfluenced by at least three factors. First,much will depend on the ability ofexporting countries to negotiatereductions in the level of supportextended to wheat producers. Theproblems arising during the UruguayRound of the GAIT negotiations(largely over the issue of reducingsupport to agriculture and liberalizingagricultural trade) indicate that seriousdifferences remain between the majorplayers, although the re-opening of tal kssuggests that all parties are seriouslyinterested in reducing the current highlevels of support. Second, much willdepend on future developments in theUSSR, the world's largest producer andconsumer of wheat. Should currenteconomic and political difficulties in theUSSR be resolved quickly, wheatproduction levels could increasedramatically. But if the Soviet economycontinues to struggle, wheat productioncould be seriously impaired, possiblyleading to a large increase in importdemand from the world's largest wheatimporter. Third, increasing concern overthe environment in many industrializedcountries may make it more difficult forwheat farmers to respond to short-runprice signals by increasing area planted.Recently enacted farm legislation in theUSA, Canada, and the EC-12 containsnumerous provisions designed to protectagainst soil mining and landdegradation, with the result that wheatfarmers may find it difficult to put landback into production withoutjeopardizing their participation in farmprograms.What is likely to happen to the worldwheat market should some sort ofagreement be reached to reduce supportlevels to cereal producers and liberalizeagricultural trade? This question hasbeen addressed by a number ofresearchers working with econometricsimulation models. While much woulddepend on the precise nature of thereforms enacted, several consequencesseem fairly certain. First, assuming ageneral reduction in the level of supportextended to wheat farmers inindustrial ized countries, productionwould almost certainly shift out ofEurope (where wheat production isheavily subsidized) to countries such asArgentina, Australia, USA, and Canada(which have the capacity to be low-costproducers). Second, assuming areduction in the current high levels ofexport subsidies and an abatement in thetrade wars, world wheat prices wouldalmost certainly rise by 15-30%.What would be the welfare implicationsof these changes for developingcountries in general, and in particular forthose which participate actively in theworld wheat trade? Here, differences ofopinion arise among the modelers. lsNearly all agree that developing countrywheat producers which are already ableto compete in the world market (e.g.,Argentina and Turkey) would stand togain significantly from a strengtheningin global prices, since they are clearlylow-cost producers. But for developingcountry importers, the prospects aremore uncertain. Some of the modelspredict that such countries will suffersignificant welfare losses as the result ofhaving to pay higher prices for wheatimports. However, other models assumethat higher international food prices willinduce technological change in foodproduction in developing countryimporters, leading ultimately toincreased domestic food productioncapacity and net welfare gains over thelonger term.15 For an excellent review of the major trademodels and a summary of lheir principalconclusions, see Runge (199)).

Part 3: Selected Wheat StatisticsThe tables that follow present 44statistics related to wheat and barleyproduction, trade, utilization, and prices,as well as some basic economicindicators. The statistics were selected toprovide the latest available information.Countries listed in the tables areclassified either as wheat producers orconsumers. Wheat consumers includedeveloping countries consuming over100,000 t of wheat per year anddeveloped countries consuming morethan one million tons of wheat per yearfrom 1987 to 1989. Wheat producersinclude developing countries in whichwheat production exceeded 100,000 t/yrfrom 1987 to 1989 or accounted for 50%of total wheat consumption, anddeveloped countries in which wheatproduction exceeded one million tonsper year from 1987 to 1989 or accountedfor 50% of total wheat consumption.Unless otherwise indicated, the regionalaggregates given in the last table includeall of the countries of a particular region(see Annex 1, inside back cover). Notethat barley imports and exports (variable31) gi ven in that table do not add up;this is because they do not add up in thesource from which CIMMYT obtainedthe information.All prices reported in the tables wereconverted to US dollars at officialexchange rates.Notes on the VariablesVariable 1: This information is drawnfrom the FAO diskettes of populationstatistics (1990).Variables 2-3: The source of thesevariables is the World Bank WorldDevelopment Report (1990). Forcountries that are not World Bankreporting members, the source is theUnited Nations Population Fund, TheState ofWorld Population 1990.Variables 4-27: Data for these variablesare taken from the FAO diskettes ofproduction statistics (1990). Growthrates were calculated using the standardformula for annual percentagecompound growth:Where:x, = X o[I + (g/IOO)]'x, = average of data for endingperiod;X o= average of data for base period;t = number of years from themidpoint of one period to thatof the other; andg = average annual percent growthrate.In the calculation of annual growth rates,figures were rounded to the nearest tenthof a percent, so the sum of the growthrates of harvested area and yield may notnecessarily equal the productiongrowth rate.Variables 28-33: The source of thesedata is the FAO diskettes of tradestatistics (1990). Net imports werecalculated as imports minus exports.Negative numbers indicate that thecountry is a net exporter. Consumptionwas calculated as production plus netimports. Growth rates were calculatedusing the standard formula given above.Variables 34-39: These data (which arefor 1989-90) were collected through ageneral country survey of knowledgeablewheat scientists. Some data wereestimated by CIMMYT staff. Regionaltotals and regional averages in someinstances were based on data from asubset of countries in the region.Regional data are reported only wheninformation was available for at least50% of the area in the region (or 50% ofwheat production, depending on thevariable).Variables 40-44: These data werecollected through a general countrysurvey of wheat scientists andeconomists. Data for the majority of thecountries refer to the wheat cropharvested in 1989-90, although in somecases 1988-89 is the reference year. Thewheat and barley prices are the postharvestprices received by farmers. Thenitrogen price is usually the price paidby farmers for the most commonnitrogenous fertilizer. In some countries,the price of compound fertilizer onlywas available, and the variable refers tothe price of nutrient only, whether it isN, PP5' and/or Kp.

Eastern and Southern Africa 33~ I. Estimated population, 1990 (millions) 47.8 24.7 25.1 27.2 9.6 15.8 7.5 225.302. Estimated growth rate of population, 1988-2000 (%/yr) 3.3 3.4 2.7 3.4 2.7 3.1 3.1 32.~"0 3. Per capita income, 1988 (US$) 120 370 480 160 650 100 170 272.~4. Per capita cereal production, 1987-89 (kg/yr) 131 139 129 163 248 33 82 129~l: 5. Growth rate of per capita cereal production, 1961-65 to..0 1987-89 (%/yr) -0.8 -1.5 -0.2 1.8 -0.4 -3.6 0.8 -0.76. Wheat area harvested, 1987-89 (000 hal 657 154 142 55 44 4 4 1,1257. Wheat yield, 1987-89 (t/ha) 1.29 1.53 1.38 1.51 5.74 1.518. Wheat production, 1987-89 (000 t) 845 236 196 83 253 1,7049 Growth rate of wheat area, 1967-69 to 1977-79 (%/yr) -4.2 -1.0 10.8 2.4 17.6 -1.110. Growth rate of wheat area, 1977-79 to 1987-89 (%/yr) 2.8 2.6 -5.6 2.0 0.6 1.0." II. Growth rate of wheat yield, 1967-69 to 1977-79 (%/yr) 1.9 0.0 -0.4 2.1 2.8 2.7-a2 12. Growth rate of wheat yield, 1977-79 to 1987-89 (%/yr) 2.8 -0.3 3.0 0.8 3.0 2.1~ 13. Growth rate of wheat production, 1967-69 to 1977-79 (%/yr) -2.4 -1.0 10.4 4.5 20.9 1.7'ii 14. Growth rate of wheat production, 1977-79 to 1987-89 (%/yr) 5.7 2.4 -2.8 2.8 3.7 3.1"0c

West Africa,-- - - - ~ -- ..- -III• I', .. ~'~.~-\-,;-, ".... '.' .- .. ,L - - - - - ..r. I. Estimated population. 1990 (millions) 10.0 11.4 12,5 15.0 2.0 112.1 7.4 35.8 268.50U 2. Estimated growth rate of population, 1988-2000 (%/yr) 3.0 3.2 3.8 3.0 2.7 3.1 3.2 3.0 3.0'ii '" 3. Per capita income. 1988 (US$)" 1,010 770 400 480 290 650 170 365.54. Per capita cereal production, 1987-89 (kg/yr) 36 77 95 83 80 115 143 35 109e... c 5. Growth rate of per capita cereal production, J96J -65 toOJ0 1987-89 (%/yr) -4.3 -1.7 -0.3 1.6 -0.5 -1.7 -1.1 1.1 -1.06. Wheat area harvested, 1987-89 (000 hal 3 1 0 0 I 35 0 28 747. Wheat yield, 1987-89 (t/ha)".. ..".. 1.43 1.14 1.288. Wheat production, 1987-89 (000 t)" " ..".. 50 "32 949. Growth rate of wheat area, 1967-69 to 1977-79 (%/yr)".. .. " "-0.9"4.7 5.310. Growth rate of wheat area, 1977-79 to 1987-89 (%/yr)" " .. .." 13.3 18.0 -0.8.. II. Growth rate of wheat yield, 1967-69 to 1977-79 (%/yr) ..".. .. 1.8 .. -2.8 4.8

North Africa 35IibiIfe8r;~ 1. Estimated population, 1990 (millions) 25.2 54.0 4.6 25.1 8.2 117.0S,~.,.c 16. Barley area harvesled, 1987-89 (000 h) 921 45 137 2,404 390 3,896.., 'iii 17. Barley yield, 1987-89 (t/ha) 0.75 2.77 0.72 1.11 0.68 0.99.c~ 18. Barley produclion, 1987-89 (000 t) 693 124 98 2,665 267 3,847....0 19. Growth rate of barley area, 1967-69 to 1977-79 (%/yr) 0.9 -3.4 -1.8 1.0 5.01.0l:.2 20. Growth rate of barley area, 1977-79 to 1987-89 (%/yr) 2.2 -0.1 -8.0 0.4 -0.1 0.2U£ '"21. Growth rate of barley yield, 1967-69 to 1977-79 (%/yr) -2.8 3.9 2.5 -2.9 -0.1 -2.322. Growth rate of barley yield, 1977-79 to 1987-89 (%/yr) 4.1 0.2 6.6 3.3 4.0 3.623. Growth rate of barley production, 1967-69 to 1977-79 (%/yr) -1.9 0.3 0.7 -2.0 4.9 -1.424. Growth rate of barley production, 1977-79 to 1987-89 (%/yr) 6.4 0.1 -1.9 3.7 3.8 3.825. Barley area as percent of total cereal area, 1987-89 (%) 39 2 32 45 38 3526. Average yield of all cereals, 1987-89 (t!ha) 0.7 5.1 0.7 1.2 0.9 1.727. Growth rate of yield of all cereaJs, 1967-69 to 1987-89 (%/yr) 0.2 1.6 4.5 1.1 1.1 1.528. Net imports of wheat, 1987-89 (000 t) 4,055 7,105 733 1,566 1,151 14,60929. Per capita total wheat consumption, 1987-89 (kg/yr) 207 194 218 210 232 204l:.g'" 30. Growth rate of per capita wheat consumption, 1967-69.~:::i 101987-89 (%/yr) 1.5 2.8 2.3 1.2 1.3 2.0'"0Ṣ 31. Nel imports of barley, 1987-89 (000 I) 395 17 443 -87 240 1,008.., ., 32. Per capita total barley consumption, 1987-89 (kg/yr) 46 3 127 108 65 44f 33. Growth rate of per capita barley consumption, 1967-69 to!-1987-89 (%/yr) 1.2 -1.6 2.8 -1.7 3.5 -0.434. Percent of total bread wheat area irrigated, 1989-90 I .. .. .. 8 ..~ 35. Percent of total durum wheat area irrigated, 1989-90 1 .. .. .. ..0

West Asia,. . - ..- ---- - .- - - .--- - - --- -- -- - - - _. -- - - - --,-- -- -- -- r-=-~~---:-~~-~.I .... - ~~i.. .." >:" " 'f'-~'. -. Y- - - - -_ . ...J... . J,~ I. Estimated population, 1990 (millions) 17.0 56.5 18.9 13.1 12.5 55,8 8.19u 2. Estimated growth rate of population, 1988-2000 (%/yr) .. 3.1 3.4 3.8 3.6 2.0 3.6os'i:i 3. Per capita income, 1988 (US$) .. .. .. 6,200 1,680 1,280 640.Se4. Per capita cereal production, 1987-89 (kg/yr) 228 209 107 255 250 517 102

West Asia, continued 37r---------------------- ~ --:-------- -=I. Estimated population, 1990 (millions)2. Estimated growth rate of population, 1988-2000 (%/yr)3. Per capita income, 1988 (US$)4. Per capita cereal production, 1987-89 (kg/yr)5. Growth rate of per capita cereal production, 1961-65 to1987-89 (%/yr)3.33.61,50035-7.72.02.813,40022.910-6.0[,53.95,000)-5.91.62.315.77032.53.043049-0.4197.13.02,531277-0.56. Wheat area harvested, 1987-89 (000 ha)7. Wheat yield, 1987-89 (!/ha)8. Wheat production, 1987-89 (000 t)9. Growth rate of wheat area. 1967-69 to 1977-79 (%/yr)10. Growth rate of wheat area. 1977-79 to 1987-89 (%/yr)II. Growth rate of wheat yield. 1967-69 to 1977-79 (%/yr)12. Growth rate of wheat yield. 1977-79 to 1987-89 (%/yr)13. Growth rate of wheat production, 1967-6910 1977-79 (%/yr)14. Growth rate of wheat production, 1977-79 to 1987-89 (%/yr)15. Wheat area as percent of total cereal area, 1987-89 (%)890.7870-5.0-3.1-8.28.0-12.74.760 21.~ 22.23.24., 25.Barley area harvested, 1987-89 (000 h)Barley yield, 1987-89 (!/ha)Barley production, 1987-89 (000 t)Growth rate of barley area, 1967-69 to 1977-79 (%/yr)Growth rate of barley area, 1977-79 10 1987-89 (%/yr)Growth rate of barley yield, 1967-69 to 1977-79 (%/yr)Growth rate of barley yield, 1977-79 to 1987-89 (%/yr)Growth rate of barley production, 1967-69 to 1977-79 (%/yr)Growth rate of barley production, 1977-79 to 1987-89 (%/yr)Barley area as percent of total cereal area, 1987-89 (%)550.5731-2.91.3-11.59.1-14.010.637

South Asia..~ I. Estimated population, 1990 (millions) 115.0 849.2 41.6 19.2 122.3 17.2 1,166.19

Southeast Asia and Pacific 39~ I. Estimated population, 1990 (millions) 5.8 181.1 17.3 61.8 2.7 55.6 66.8 411.09 2. Estimated growlh rate of population, 1988-2000 (%/yr) 0.9 1.7 2.2 1.9 1.0 1.3 2.0 1.7:a '"3. Per capita income, 1988 (U5$) 9,220 440 1,940 630 9,070 1,000 876.5]4. Per capita cereal production, 1987-89 (kg/yr) 0 273 104 225 453 275 273c: 5. Growth rate of per capita cereal production, 1961-65 toOJ1987-89 (%/yr) -30.4 2.5 -0.9 1.0 0.3 0.1 l.l"6. Wheat area harvested, 1987-89 (000 ha) 0 0 0 0 0 0 0 07. Wheat yield, 1987-89 (t/ha)8. Wheat production, 1987-89 (000 t)9. Growth rate of wheat area, 1967-69 to 1977-79 (%/yr)10. Growth rate of wheat area, 1977-79 to 1987-89 (%/yr)II. Growth rate of wheat yield, 1967-69 to 1977-79 (%/yr).. '"12. Growth rate of wheat yield, 1977-79 to 1987-89 (%/yr)S v 13. Growth rate of wheat production, 1967-69 to 1977-79 (%/yr)'ia 14. Growth rate of wheat production, 1977-79 to 1987-89 (%/yr)"0 c:f'S 15. Wheat area as percent of total cereal area, 1987-89 (%) 0 0 0 0 0 0 0 0»~.2 16. Barley area harvested, 1987-89 (000 h) 0 0 0 0 0 0 0 0J

_ ~.=E'"c:...oEast Asia1---r------------------------1~r:: I. Estimated population, 1990 (millions)1,109.59OJ 2. Estimated growth rate of population, 1988-2000 (%/yr)1.3:.c "3303293. Per capita income, 1988 (US$)4. Per capita cereal production, 1987-89 (kg/yr)5. Growth rate of per capita cereal production, 1961-65 to1987-89 (%/yr)_=~T-';~~~L ~_ ·:?i~.- -----:--==---..---- 11'1--I---'~ii'~1'.' .".- . ',' •./ r'- .....';' I.' .-'- :;~)l~~4,~ _.~~~~ ~.~ -=.;:~.I~~£~22.9 2.2 43.4 19.8 1,197.82.2 3.1 0.9 .. 1.3.. .. 3,600 .. 454540 374 204 144 3252.0 1.7 1.0 -0.5 -2.2 1.96. Wheat area harvested, 1987-89 (000 hal7. Wheat yield, 1987-89 (t/ha)8. Wheat production, 1987-89 (000 t)9. Growth rate of wheat area, 1967-69 to 1977-79 (%/yr)10. Growth rate of wheat area, 1977-79 to 1987-89 (%/yr)II.Growth rate of wheat yield, 1967-69 to 1977-79 (%/yr)~ 12. Growth rate of wheat yield, 1977-79 to 1987-89 (%/yr)'" 13. Growth rate of wheat production, 1967-69 to 1977-79 (%/yr):; 14. Growth rate of wheat production, 1977-79 to 1987-89 (%/yr)] 15. Wheat area as percent of total cereal area, 1987-89 (%)~] 16. Barley area harvested, 1987-89 (000 h)Z 17. Barley yield, 1987-89 (!/ha)~ 18. Barley production, 1987-89 (000 I)....~ 19. Growth rate of barley area, 1967-69 to 1977-79 (%/yr).9 20. Growth rate of barley area, 1977-79 to 1987-89 (%/yr)i 21. Growth rate of barley yield, 1967-69 to 1977-79 (%/yr)If 22. Growth rate of barley yield, 1977-79 to 1987-89 (%/yr)23. Growth rate of barley production, 1967-69 to 1977-79 (%/yr)24. Growth rate of barley production, 1977-79 to 1987-89 (%/yr)25. Barley area as percent of total cereal area, 1987-89 (%)29,1343.0087,4241.40.15.15.16.65.2339773.073,000-3.8-4.26.63.32.5-1.0I2144.018601.63.92.25.43.99.582602.436312.93.9-2.31.80.55.8104861.316391.42.12.16.43.58.6771010.9910019.31.62.25.221.96.916

Mexico, Central America, and the Caribbean 41- - . -- - --- ----•. e-...._. -...- - ----..... ', .-' ., - - - -- - - - -I. Estimated population, 1990 (millions) 88.1 3.0 10.3 7.1 5.2 9.25U 2. Estimated growth rate of population, 1988-2000 (%/yr) 1.9 2.0 0.9 1.8 2.1 2.8.::'"3. Per capita income, 1988 (US$) 1,760 1,690 .. 720 940 900.E4. Per capita cereal production, 1987-89 (kg/yr) 249 110 59 88 148 179f.. g 5. Growth rate of per capita cereal production, 1961-65 to0 1987·89 (%/yr) 0.1 0.3 2.0 2.2 1.1 0.66. Wheat area harvested, 1987-89 (000 hal 950 0 0 0 0 307. Wheat yield, 1987-89 (t!ha) 4.20 .. .. .. .. 1.648. Wheat production, 1987-89 (000 t) 3,993 .. .. .. ,. 509. Growth rate of wheat area, 1967-69 to 1977-79 (%/yr) -1.6 .. .. .. .. 4.910. Growlh rate of wheat area, 1977-79 to 1987-89 (%/yr) 3.3 .. .. .. .. -5.3I I. Growth rate of wheat yield, 1967·69 to 1977-79 (%/yr) 3.1 .. .. .. 1.1'"~ 12. Growth rate of wheat yield, 1977-79 to 1987-89 (%/yr) 1.4 .. .. .. 4.1ol... 13. Growth rale of wheat production, 1967-69 to 1977-79 (%/yr) 1.4.. .. .. -- 6.0-.: 14. Growth rate of wheat production, 1977-79 10 [987-89 (%/yr) 4.8 .. .. .. .. -1.4'0§ 15. Wheat area as percent of total cereal area, 1987-89 (%) 10 0 0 0 0 4>..."516. Barley area harvested, 1987-89 (000 h) 274 0 0 0 0 IIT.QoJ 17. Barley yield, 1987-89 (t/ha) 1.76 .. .. .. .. ....c:: '-'~ 18. Barley production, 1987-89 (000 t) 482.. .. .. .. .......0 19. Growth rate of barley area, 1967-6910 1977-79 (%/yr) 0.8.. .. .. .. ..c 2 20. Growth rate of barley area, J977-79 10 1987-89 (%/yr) 0.3 .. .. .. .. ..U ::> 21. Growth rate of barley yield, 1967-69 to 1977-79 (%/yr) 6.0 .. .. ..e 22. Growth rate of barley yield, 1977-79 to 1987-89 (%/yr) 0.8 .. .. .. .."'" 23. Growth rate of barley production, 1967-69 to 1977-79 (%/yr) 6.8 .. .. .. ..24. Growth rate of barley production, 1977-79 to 1987-89 (%/yr) 1.1 .. .. .. .. ..25. Barley area as percent of total cereal area, 1987-89 (%) 3 0 0 0 0 026. Average yield of all cereals, 1987-89 (t/ha) 2.2 2.4 2.6 3.7 1.8 1.927. Growth rate of yield of all cereals, J967-69 to 1987-89 (%/yr) 2.2 2.2 3.7 3.0 1.5 3.52.8. Net impons of wheat, 1987-89 (000 t) 549 140 1,425 219 121 178,2 29. Per capita total wheat consumption, 1987-89 (kg/yr) 54 49 140 32 24 26,§ 30. Growth rate of per capita wheat consumption, 1967-69::; to 1987-89 (%/yr) 1.4 0.5 2.4 1.4 1.8 1.5::I"C31. Net impons of barley. 1987-89 (000 t) 44 0 44 0 0 0c..'"32. Per capita total barley consumption, 1987-89 (kg/yr) 6

Mexico, Central America, and the Caribbean, continued~ I. Estimated populalion, 1990 (millions) 6.5 5.1 2.5 1.3 147.0g 2. ESlimated growth rale of population, 1988-2000 (%/yr) 1.9 2.9 0.5 1.4 1.9.~." 3. Per capila income, 1988 (US$) 380 860 1,070 3,350 1,538.64. Per capila cereal production, 1987-89 (kg/yr) 58 123 2 8 189~ 5. Growth rate of per capila cereal production, 1961-65 to~ 1987-89 (%/yr) -3.0 -1.2 -3.7 -2.4 0.36. Wheat area harvested, 1987-89 (000 hal 0 0 0 9827. Wheat yield, 1987-89 (l/ha) 4.128. Wheat production, 1987-89 (000 t) 4,0449. Growth rate of wheat area, 1967-69 to 1977-79 (%/yr) -1.310. Growth rate of wheat area, 1977-79 to 1987-89 (%/yr) 2.9"'- II. Growth rate of wheat yield, 1967-69 to 1977-79 (%/yr) 2.8-;12. Growlh rate of wheat yield, 1977-79 to 1987-89 (%/yr) 1.7~u 13. Growth rate of wheat production, 1967-69 to 1977-79 (%/yr) 1.5-;14. Growth rate of wheat production, 1977-79 to 1987-89 (%/yr)."4.6:a 15. Wheal area as percent of lolal cereal area, 1987-89 (%) 0 0 0 0 8>.....~.D 16. Barley area harvested, 1987-89 (000 h) 0 0 0 0 275ol 17. Barley yield, 1987-89 (l/ha) 1.76.t:~ 18. Barley production, 1987-89 (000 t)483"-0 19. Growlh rate of barley area, 1967-69 to 1977-79 (%/yr)0.8l:.g 20. Growth rate of barley area, 1977-79 to 1987-89 (%/yr) 0.4g." 21. Growth rate of barley yield, 1967-69 to 1977-79 (%/yr) 6.0~22. Growth rate of barley yield, 1977-79 to 1987-89 (%/yr) 0.823. Growth rate of barley production, 1967-69 to 1977-79 (%/yr) 6.824. Growth rate of barley production, 1977-79 to 1987-89 (%/yr) 1.225. Barley area as percent oftola1 cereal area, 1987-89 (%) 0 0 0 0 226. Average yield of all cereals, 1987-89 (t/ha) L1 1.1 I.d 2.6 2.127. Growth rate of yield of all cereals, 1967-69 to 1987-89 (%/yr) -0.1 0.0 1.1 -0.1 2.228. Nel imports of wheal, 1987-89 (000 t) 181 III 165 130 3,528l:.9 29. Per capila tolal wheal consumption, 1987-89 (kg/yr) 29 23 68 104 53.~30. Growth rate of per capila wheat consumption, 1967-69~to 1987-89 (%/yr) 6.0 1.9 -1.2 0.2 1.3." 31. Net imports of barley, 1987-89 (000 t) 0 0 0

Andean Region, South America 43- .- -- - ..,'...r.; L Estimated population, 1990 (millions) 22.2 7.3 3L6 10.7 19.6 92.7(;0ti 2. Estimated growth rate of population, 1988-2000 (%/yr) 2.1 2.7 1.6 2.2 2.2 2.0'" 3. Per capita income, 1988 (US$) 1,300 570 1,180 1,120 3,250 1,597~4. Per capita cereal production, 1987-89 (kg/yr) 112 115 115 132 117 121e~ ;:: 5. Growth rate of per capita cereal production, 1961-65 to"0 1987-89 (%/yr) 0.1 0.0 0.7 0.6 L9 0.6-6. Wheat area harvested, 1987-89 (000 hal 112 88 41 37 I 2787. Wheat yield, 1987-89 (t/ha) 1.32 0.76 1.76 0.89 .. 1.158. Wheat production, 1987-89 (000 t) 147 67 72 33 .. 3209. Growth rate of wheat area, 1967-69 to 1977-79 (%/yr) -3.0 3.3 -8.5 -9.3 .. -3.510. Growth rate of wheat area, 1977-79 to 1987-89 (%/yr) 0.6 -0.3 2.7 1.2 .. 0.6IL Growth rate of wheat yield, 1967-69 to 1977-79 (%/yr) 1.3 0.4 -0.3 0.3 .. -0.2~::12. Growth rate of wheat yield, 1977-79 to 1987-89 (%/yr) 2.6 1.4 4.3 -L3 ..2.4~'"'13. Growth rate of wheat production. 1967-69 to 1977-79 (%/yr) -1.7 3.7 -8.8 -9.0 .. -3.7-;;14. Growth rate of wheat production, 1977-79 to 1987-89 (%/yr) 3.2 LO 7.1 -0.2 ..3.0."= 15. Wheat area as percent of total cereal area, 1987-89 (%) 12 14 3 4

Southern Cone, South AmericaI, ,- -- - --~ , I- - --.......... --- .. ,'. _.' ....- - - - ::.---=.~ I. Estimated population, 1990 (millions) 32.2 149.7 13.1 4.3 3.1 202.40t> 2. Estimated growth rate of population, 1988-2000 (%/yr) 1.1 1.8 1.3 2.7 0.6 1.7~ 3. Per capita income, 1988 (US$) 2,520 2,160 1,510 1,180 2,470 2,160..s4. Per capita cereal production, 1987-89 (kg/yr) 654 302 229 408 404 358-ạ,5. Growth rate of per capita cereal production, 196 I-65 toct3 1987-89 (%/yr) -0.5 1.4 0.8 6.0 1.2 0.66. Wheat area harvested, 1987-89 (000 hal 4,891 3,418 598 216 190 9,3127. Wheat yield, 1987-89 (t/ha) 1.83 1.68 3.00 1.85 2.09 1.858. Wheat production, 1987-89 (000 t) 8,933 5,731 1,791 400 398 17,2549. Growth rate of wheat area, 1967-69 to 1977·79 (%/yr) -2.3 11.8 -2.0 2.6 -3.3 1.010. Growth rate of wheat area, 1977-79 to 1987·89 (%/yr) 0.9 0.5 0.1 19.1 -4.0 0.8II. Growth rate of wheat yield, 1967-69 to 1977-79 (%/yr) 3.0 -1.3 0.4 0.8 0.3 0.8'"~ 12, Growth rate of wheat yield, 1977-79 to 1987-89 (%/yr) 1.3 7.9 5.5 5.3 8.7 3.8.. " 13. Growth rate of wheat production, 1967-69 to 1977-79 (%/yr) 0.7 10.4-1.6 3.5 -3.0 1.8... 14. Growlh rate of wheat production, 1977-79 to 1987-89 (%/yr) 2.2 8.4 5.6 25.5 4.3 4.5'g'"15. Wheat area as percent of total cereal area, 1987-89 (%) 55 15 73 25 36 27>., .,~ 16. Barley area harvested, 1987-89 (000 h) 137 108 22 0 78 345.D~ 17. Barley yield, 1987-89 (t/ha) 2.26 1.68 .. .. 2.23 2.14.c:t 18. Barley production, 1987-89 (000 t) 310 182 .. 174"737'0 19, Growth rate of barley area, 1967-69 to 1977·79 (%:yr) -4.8 9.7 .. ,. 3.5 -2.1,§ 20. Growth rate of barley area, 1977-79 to 1987-89 (%/yr) -7.7 1.9 .. .. 4.9 -3.7U.a21. Growth rate of barley yield, 1967-69 to 1977-79 (%/yr) 1.8 4.0 .. .. 1.9 1.60ll:: 22. Growth rate of barley yield, 1977-79 to 1987-89 (%/yr) 5.2 2.9 .. ..6.8 4.223. Growth rate of barley production, 1967-69 to 1977-79 (%/yr) -3.1 14.1 .. .. 5.5 -0.624. Growth rate of barley production, 1977-79 to 1987-89 (%/yr) -2.9 4.9 .. .. 12.1 0.425. Barley area as percent of total cereal area, 1987-89 (%) 2

Eastern Europe and USSR 45.e0 2. Estimated growth rate of population, 1988-2000 (%/yr) 0.1 0.0 0.0 -0.2 0.5 0.5 0.6 0.6 0.5:0 '" 3. Per capita income, 1988 (US$) 2,460 1,860 2,520.S4. Per capita cereal production, 1987-89 (kg/yr) 919 763 636 1,400 684 871 689 653 717~ .,5. Growth rate of per capita cereal production, 1961-65 to8'"1987-89 (%/yr) 1.8 2.6 2.5 3.0 1.3 1.6 0.9 0.7 1.2~ I. Estimated population, 1990 (millions) 9.0 15.6 16.7 10.6 38.2 23.3 286.0 23.8 426.46. Wheat area harvested, 1987-89 (000 hal 1,135 1,230 763 1,274 2,169 2,500 47,459 1,495 58,2227. Wheat yield, 1987-89 (t/ha) 4.20 5.16 4.90 5.06 3.69 2.80 1.81 3.85 2.218. Wheat production, 1987-89 (000 t) 4,764 6,352 3,738 6,444 7,995 7,000 86,086 5,749 128,7499. Growth rate of wheat area, 1967-69 to 1977-79 (%/yr) -1.2 2.0 2.5 -0.1 -0.5 -2.4 -0.9 -2.0 -0.910. Growth rate of wheat area, 1977-79 to 1987-89 (%/yr) 2.0 0.1 0.7 0.1 2.2 1.2 -2.5 -0.8 -1.9II. Growth rate of wheat yield, 1967-69 to 1977-79 (%/yr) 3.2 2.9 1.2 4.1 2.2 4.0 2.9 3.0 3.0.. '"t! 12. Growth rale of wheat yield, 1977-79 to 1987-89 (%/yr) 1.4 2.6 1.3 2.6 2.2 0.7 0.9 1.9 1.6g..13. Growth rate of wheat production, 1967-69 to 1977-79 (%/yr) 2.0 5.0 3.7 4.0 1.7 1.5 1.9 0.9 2.114. Growth rate of wheat production, 1977-79 to 1987-89 (%/yr) 3.4 2.7 2.0 2.8 4.4 1.9 -1.6 l.l -0.4"0~ L5. Wheat area as percent of total cereal area, 1987-89 (%) 55 49 31 45 26 39 44 36 42>..,'i:.rJ '" 16. Barley area harvested, 1987-89 (000 h) 333 793 887 250 1,237 633 29,332 226 33,705;iII>17. Barley yield, 1987-89 (t/ha) 3.97 4.42 4.77 4.42 3.25 3.05 1.76 2.69 2.03.

Developed Market Economies.-.-~=-~~- -=~- - - - -- - ~ --- - ~ .' -~ ~ --~- ------- . .:_---".'-l.' '-. .' .. ~.,. _-::.~1-~~~-".'-~~t:: I. Estimated population, 1990 (millions) 16.8 7.6 10.3 26.4 5.1 56.3 61.4 10.0 57.69 2. Estimated growth rate of population, 1988-2000 (%!yr) 1.4 0.1 0.0 0.9 0.0 0.4 0.0 0.2 0.1.~'0 3. Per capita income, 1988 (US$) 12.340 15,470 14,779 16,960 18,450 16,090 18,480 4,800 13,330a.54. Per capita cereal production, 1987-89 (kglyr) 1,307 673 214 1,740 1,563 993 419 477 3065. Growth rate of per capita cereal production, 1961-65 to~ 1987-89 (%!yr) 1.1 3.1 0.2 0.6 0.9 2.5 2.0 1.9 0.4,6. Wheat area harvested, 1987-89 (000 hal 8,930 297 202 13,370 384 4,923 1,732 887 2,9717. Wheat yield, 1987-89 (!/ha) 1.52 4.91 6.35 1.65 6.59 6.02 6.33 2.41 2.788. Wheat production, 1987-89 (000 t) 13,541 1,458 1,283 22,124 2,529 29,636 10,973 2,134 8,2479. Growth rate of wheat area, 1967-69 to 1977-79 (%!yr) 0.6 -0.8 -1.1 -0.9 2.1 0.3 1.0 -1.0 -2.510. Growth rate of wheat area, 1977-79 to 1987-89 (%!yr) -1.6 0.6 0.3 2.5 12.6 1.8 0.7 -0.9 -0.9II. Growth rate of wheat yield, 1967-69 to 1977-79 (%!yr) 2.2 1.0 1.8 2.0 1.3 2.6 1.6 3.8 0.912. Growth rate of wheat yield, 1977-79 to 1987-89 (%!yr) 0.9 2.9 2.8 -1.2 2.4 2.5 2.7 0.2 1.0.. "..u 13. Growth rate of wheat production, 1967-69 to 1977-79 (%!yr) 2.9 0.2 0.7 1.1 3.4 2.9 2.7 2.8 -1.7:a14. Growth rate of wheat production, 1977-79 to 1987-89 (%!yr) -0.7 3.4 3.1 1.3 15.2 4.4 3.5 -0.7 0.1"0i 15. Wheat area as percent of total cereal area, 1987-89 (%) 66 31 54 62 24 53 37 63 64~"C...Q 16. Barley area harvested, 1987-89 (000 h) 2,274 292 133 4,953 1,033 1,913 1,813 229 457j 17. Barley yield, 1987-89 (t!ha) 1.47 4.53 5.61 2.41 4.74 5.30 5.13 2.36 3.63:t 18. Barley production, 1987-89 (000 t) 3,340 1,322 747 11,947 4,898 10,128 9,305 540 1,658...0 19. Growth rate of barley area, 1967-69 to 1977-79 (%!yr) 5.1 3.6 0.3 1.8 2.4 0.1 3.6 1.3 5.3c9 20. Growth rate of barley area, 1977-79 to 1987-89 (%!yr) -1.7 -1.9 -2.6 1.5 -4.1 -3.9 -0.6 -4.6 4.4U:I 21. Growth rate of barley yield, 1967-69 to 1977-79 (%!yr) 4.0 0.7 2.1 2.2 0.3 1.3 1.4 2.3 4.9l 22. Growth rate of barley yield, 1977-79 to 1987-89 (%!yr) 1.6 2.4 1.9 0.0 1.6 3.3 1.9 0.7 3.523. Growth rate of barley production. 1967-69 to 1977-79 (%!yr) 9.2 4.3 2.4 4.0 2.7 1.4 5.1 3.6 10.524. Growth rate of barley production, 1977-79 to 1987-89 (%!yr) -0.1 0.5 -0.7 1.6 -2.6 -0.8 1.4 -3.9 8.025. Barley area as percent of total cereal area, 1987-89 (%) 17 30 36 23 64 20 39 16 1026. Average yield of all cereals, 1987-89 (t!ha) 1.6 5.3 5.9 2.1 5.0 5.9 5.5 3.4 3.827. Growth rate of yield of all cereals, 1967-69 to 1987-89 (%!yr) 2.0 2.2 2.1 0.9 1.2 2.6 2.0 3.4 1.9c,228. Net imports of wheat, 1987-89 (000 t) -12,581 -579 528 -18,143 -564 -16,521 -1,791 -304 2.93829. Per capita total wheat consumption, 1987-89 (kglyr) 59 116 177 153 383 235 150 \83 195.~30. Growth rate of per capita wheat consumption, 1967-69~to 1987-89 (%!yr) -8.9 -0.9 1.3 -4.6 7.3 0.7 1.1 0.0 0.0'0 31. Net imports of barley, 1987-89 (000 t) -1,723 -110 623 -4,020 -1,104 -4,208 -459 125 862a~$32. Per capita total barley consumption. 1987-89 (kg/yr) 99 160 134 306 739 106 144 67 4433. Growth rate of per capita barley consumption, 1967-69 to1987-89 (%!yr) 0.5 1.0 0.8 0.2 -1.5 -I.I 1.6 0.5 3.134. Percent of total bread wheat area irrigated, 1989-90 2 0 0

Developed Market Economies, continued 47~ I. Estimated population. 1990 (millions) 35.3 39.4 8.5 57.6 250.1 123.6 14.9 10.392. Estimated growth rate of population. 1988-2000 (%/yr) 2.3 0.4 0.4 0.3 0.8 0.4 0.5 0.4.~] 3. Per capita income. 1988 (USS) 2.290 7,740 19.300 12.810 19.840 21.020 14.520 3.6504. Per capita cereal production. 1987-89 (kg/yr) 375 546 610 379 1.046 116 84 163f~ 5. Growth rate of per capita cereal production. 1961-65 to~ 1987-89 (%/yr) 0.3 2.8 0.6 2.2 0.6 -2.2 -2.7 -0.46. Wheat area harvested. 1987-89 (000 hal 1.914 2.283 287 1.995 23.107 279 121 3197. Wheat yield. 1987-89 (I/ha) 1.64 2.59 5.37 6.26 2.34 3.43 7.28 1.618. Wheat production. 1987-89 (000 t) 3.135 5.923 1.539 12,494 54.024 957 881 5129. Growth rate of wheat area. 1967-69 to 1977-79 (%/yr) 3.7 -4.0 1.7 3.0 1.5 -9.8 -1.8 -5.010. Growth rate of wheat area. 1977-79 to 1987-89 (%/yr) 0.6 -1.6 -0.5 4.9 -0.8 9.2 -0.7 0.5II. Growth rate of wheat yield. 1967-69 to 1977-79 (%/yr) 0.7 2.2 0.4 2.8 1.3 1.5 2.7 -4.1.. '"e 12. Growth rate of wheat yield. 1977-79 to '1987-89 (%/yr) 4.6 4.9 2.4 2.0 0.8 0.4 2.1 6.9u13. Growth rate of wheat production. 1967-69 to 1977-79 (%/yr) 4.5 -1.8 2.0 5.9 2.8... -8.4 0.9 -8.8~ 14. Growth rate of wheat production. 1977-79 to 1987-89 (%/yr) 5.3 3.2 1.9 7.0 0.0 9.6 1.4 7.5..,a 15. Wheat area as percent of total cereal area. 1987-89 (%) 28 29 22 51 39 II 63 31~~ 16. Barley area harvested, 1987-89 (000 h) 92 4.278 521 1,790 3.507 113 54 81J:Jiu17. Barley yield. 1987-89 (I/ha) 2.50 2.43 3.63 4.82 2.53 3.32 5.00 0.89.s::~ 18. Barley production, 1987-89 (000 t) 231 10.405 1.892 8,632 8,876 374 272 72...19. Growth rate of barley area, 1967-69 to 1977-79 (%/yr)0 7.0 6.5 1.6 -0.2 -0.7 -11.2 -4.4 -3.0~tl20. Growth rate of barley area, 1977-79 to 1987-89 (%/yr) -0.6 2.2 -2.3 -2.7 -0.2 1.6 -2.0 0.8"~21. Growth rate of barley yield, 1967-69 to 1977-79 (%/yr) 5.2 1.3 1.6 1.6 1.1 0.7 1.8 -2.722. Growth' rate of barley yield, 1977-79 to 1987-89 (%/yr) 7.7 1.8 0.6 1.3 -0.2 0.2 0.7 4.923. Growth rate of barley production. 1967-69 to 1977-79 (%/yr) 12.6 7.9 3.2 1.4 0.3 -10.6 -2.7 -5.624. Growth rate of barley production. 1977-7910 1987-89 (%/yr) 7.1 4.0 -1.7 -1.5 -0.3 1.8 -1.3 5.725. Barley area as percenl of total cereal area. 1987-89 (%) I 55 40 46 6 4 28 826. Average yield of all cereals. 1987-89 (I/ha) 1.9 2.7 3.9 5.5 4.3 5.6 6.4 1.627. Growth rate of yield of all cereals. 1967-69 to 1987-89 (%/yr) 2.3 2.9 1.2 2.1 1.3 0.4 2.3 1.7c028. Net impons of wheat. 1987-89 (000 t) -484 89 -454 -1.793 -37.293 5.188 1.118 50829. Per capita total wheat consumption. 1987-89 (kg/yr) 79 154 128 187 68 50 135 100." ~30. Growth rate of per capita wheat consumption, 1967-69:; to 1987-89 (%/yr) 1.0 0.8 1.2 1.4 -2.8 0.1 1.6 -0.2..,c 31. Net impons of barley, 1987-89 (000 t) 15 -1,027 -78 -2,770 -2.065 1.303 576 84u"'32. Per capita total barley consumption. 1987-89 (kg/yr) 7 240 215 102 28 14 57 15'0;. 33. Growth rate of per capita barley consumption, 1967-69 to1987-89 (%/yr) 7.3 4.4 0.7 -2.0 -2.2 -0.7 2.3 2.734. Percent of total bread wheat area irrigated. 1989-90 18 10 0 29 2~ 35. Percent of total durum wheat area irrigated, 1989-90 100 I 09Q36. Percent of total barley area irrigated, 1989-90 0 10 2 0 2 0~2:- 37. Percentage of total bread wheat area which receives'> fenilizer (%) 80 95 100 100 10 100'CU:038. Percentage of total durum wheat area which receives£ fenilizer (%) 100 95 10039. Percent of total barley area which receives fenilizer (%) 80 95 100 100

..Regional Aggregates- -..-'. ,-II- - - Lrl I. Estimated population, 1990 (millions) 4.025.0 819.4 426.4 5,270.70c:; 2. Estimated growth rate of population, 1988--2000 (%/yr) 2.0 0.6 0.5 1.6:0 3. Per capita income, 1988 (US$) 677 16,427 .. 3,551.64. Per capita cereal production, 1987-89 (kg/yr) 249 653 717 352~t> 5. Growth rate of per capita cereal production, 1961-65 to§0 1987-89 (%/yr) 0.8 1.0 1.2 0.66. Wheat area harvested, 1987-89 (000 hal 98,982 64,486 58,222 221,6917. Wheat yield, 1987-89 (t/ha) 2.18 2.69 2.21 2.348. Wheat production, 1987-89 (000 t) 216,151 173,401 128,749 518,3019. Growth rate of wheat area, 1967-69 to 1977-79 (%/yr) 1.6 0.2 -0.9 0.410. Growth rate of wheat area, 1977-79 to 1987-89 (%/yr) 0.5 0.2 -1.9 -0.3J> II. Growth rate of wheat yield, 1967-69 to 1977-79 (%/yr) 3.5 1.8 3.0 2.6

ReferencesAhmad, Z., and M. Ahmed. 1991. FactorsAffecting Adoption ofSemidwGlf Wheats inMarginal Areas: Evidencefrom RainfedNorthern Pun;ab. PARC Research Report.Islamabad: Pakistan Agricultural ResearchCouncil (forthcoming).Avci, M., N. Durutan, M. Karaca, and M. Guier.1988. Agronomic Practices in DifferentCropping Systems o.fCentral Anatolia.Ankara: Field Crops Improvement Center.Barker, R. 1988. Methods for SellingAgricultural Research Priorities: Report ofa Bellagio Conference. Working Paper inAgricultural Economics No. 88-3. Ithaca,N. Y.: Cornell University.Belaid, A., and M. L. Morris. 1991. Wheat andBarley Production in Rainfed MarginalEnvironments of WANA Region: Problemsand Prospects. CIMMYT EconomicsWorking Paper 91/02. Mexico, D.F.:CIMMYT.Bolton, F.E. 1979. Agronomic yield constraintsin rainfed cereal production systems. Paperpresented at the Fifth Regional CerealsWorkshop, Algiers.Bolton, F.E. 198 I. Optimizing the use of waterand nitrogen through soil and cropmanagement. In J. Monteith and e. Webb(eds), Soil, Water, and Nitrogen inMediterranean Type Environments. TheHague: Martinus Nijhoff/Dr. W. JinkkPublishers,Boutonnet, J.P. 1989. La speculation ovine enAlgerie: Un produit de la cerealiculture.Serie Notes et Documents, No. 90. Paris:Institut National de la RechercheAgronomique.Byerlee, D. 1991. Dryland Wheat in India: TheImpact ofTechnical Change and FutureResearch Challenges. CIMMYTEconomics Program Working Paper.Mexico, D.F.: CIMMYT (forthcoming).Byerlee, D., and D. Winkelmann. 1980.Accelerated Wheat Production in SemiaridDeveloping Regions: Economic andPolicy Issues. CIMMYT EconomicsProgram Working Paper 80/2. Mexico,D.F.: CIMMYT.CIMMYT. 1989a. CIMMYT's Five-Year Budget:1990-1994. Mexico, D.F.: CIMMYT.CIMMYT. 1989b. Toward the 21st Century:CIMMYT's Strategy. Mexico, D.F.:CIMMYT.CIMMYT. 1989c. 1987-1988 CIMMYT WorldWheat Facts and Trends. The WheatRevolution Revisited: Recent Trends andFuture Challenges. Mexico, D.F.:CIMMYT.Cocks, P.S., and E.F. Thomson. 1988.Increasing feed resources for smallruminants in the Mediterranean Basin. InE.F. Thomson and F.S. Thomson (eds.),Increasing Small Ruminant Productivity inSemi-Arid Areas. Aleppo: ICARDA.Cooper, PJ.M. 1985. The potential of improvedvarieties. In Farming Systems AnnualReportfor 1984. Aleppo: ICARDA.Cooper, PJ.M., and E. Bailey. 1990. Livestockin Mediterranean farming systems, atraditional buffer against uncertainty: Nowa threat to the agricultural resource base.Paper presented at the World BankSymposium, "Risk in Agriculture," 9-10January 1990, Washington, D.e.Cooper, PJ.M., M. Jones, H. Harris, and A.Matar. 1988. Agroecological Constraintsto Crop Production in West Asia and NorthAfrica, and Their Impact on Fertilizer Use.Aleppo and Muscle Shoals, Alabama:ICARDA and International FertilizerDe"velopment Center.Cornish, P.S., and G.M. Murray, 1989. Lowrainfall rarely limits wheat yields insouthern New South Wales. AustralianJournal ofExperimental Agriculture 29:77-83.Daaloul, A. 1985. Recherches agronomiques surles cereales en Tunisie: Situation actuelleet perspectives. Paper presented at the"Seminaire sur les Productions Agricolesen Mediterranee," 5-8 March 1985, InstitutAgronomique et Veterinaire Hassan II,Rabat, Morocco.Dalrymple, D. 1978. Development and SpreadofHigh-yielding Varieties of Wheat andRice in the Less Developed Nations.Foreign Agricultural Economic Report No.95. Washington, D.e.: United StatesAgency for International Development,Bureau for Science and Technology.Demir, N. 1976. The Adoption ofNew BreadWheat Technology in Selected Regions ofTurkey. Mexico, D.F.: CIMMYT.Desai, G.M. 1982. Sustaining Rapid Growth inIndia's Fertilizer Consumption: APerspective Based on Composition ofUse.IFPRI Research Report 41. Washington,D.e.: International Food Policy ResearchInstitute.Durutan, N., M. GuIer, M. Karaca, K. Meyveci,A. Avcin, M. Avci, and H. Eyuboglu.1989. Effect of various components of themanagement package on weed control indryland agriculture. Paper presented at theWorkshop, "Soil and Crop Managementfor Improved Water Use Efficiency," 15­19 May, 1989, Ankara.Durutan, N., K. Meyveci, M. Karaca, M. Avci,and H. Eyuboglu. 1990. Annual croppingunder dryland conditions in Turkey: A casestudy. In A.E. Osman et al. (eds.), The RoleofLegumes in the Farming Systems oftheMediterranean Areas. Aleppo: ICARDA.Durutan, N., B. Yilmaz, and M. Kiziltan. 1988.Small grain and food legume productionimprovement in Turkey. In J.P. Srivastava,M.e. Saxena, S. Varma, and M. Tahir(eds.), Winter Cereals and Food Legumesin Moutainous Areas. Aleppo: ICARDA.Fertilizer Association of India. FertilizerStatistics (various issues). New Delhi: FA!.Fischer, R.A. 1988. Plant breeding and somebroader issues in agriculture. Paperprepared for the 9th Australian PlantBreeding Conference, 27-31 June, 1988,Wagga Wagga.Food and Agricultural Organization. 1990.Diskettes of population statistics. Rome:FAO.___-,---,---=_. 1990. Diskettes of tradestatistics. Rome: FAO.Foster, J.H., K.G. Kshirsagar, MJ. Bhende, V.Bhaskar F-ao, and T.S. Walker. 1987. EarlyAdoption of1mproved Vertisol TechnologyOptions and Double Cropping inBegumgunj, Madhya Pradesh. EconomicsGroup of Resource ManagementProgramme, Progress Report 77.Patancheru: ICRISAT.French, RJ., and J.E. Schultz. 1984. Water useefficiency of wheat in a Mediterraneantypeenvironment. Australian Journal ofAgricultural Research 35: 743-775.Gardner, G., and D. Skully. 1986. The ConductofWheat Marketing in North Africa. StaffReport No. AGES 860808, InternationalEconomics Division, Economic ResearchService. Washington, D.e.: United StatesDepartment of Agriculture.Greb. 1979. Technology and wheat yields in theCentral Great Plains: Commercialadvances. Journal ofSoil and WaterConservation 34: 264-268.Guier, M., and M. Karaca. 1988.Agroclimatological criteria for determiningthe boundaries of fallow practice. In J.P.Srivastava, M.e. Saxena, S. Varma, and M.Tahir (eds.), Winter Cereals and FoodLegumes in Moutainous Areas. Aleppo:ICARDA.Hanchinal, R.R. 1988. Development of wheatvarieties for rainfed conditions. Draftpaper. Dharwad, India: University ofAgricultural Sciences.Hanchinal, R.R. 1990. Background Informationon Wheat in Karnataka. University ofAgricultural Sciences, Dharwad, India.Mimeo.Harmsen, K. 1984. Nitrogen fertilizer use inrainfed agriculture. Fertilizer Research5:371-382.Harris, H. 1989. Crop water use and water useefficiency in contrasting barley rotations.In Farm Resource Management ProgramAnnual Reportfor 1988. Aleppo: ICARDA.Harris, H., and M. Pala. 1988. Tillage andstubble management effect on soilconservation, crop establishment and yield,and economics of production. In FarmResource Management Program AnnualReportfor 1987. Aleppo: ICARDA.

50Hobbs, P., A. Razzaq, N.!. Hashmi, M. Munir,and B.R. Khan. 1985. The effect ofmustard grown as a mixed intercrop on theyield of wheat. Pakislan Journal ofAfiricullural Research 6(4): 241-247.IAR!. 1990. Indian Agricultural ResearchInstitute Regional Station (WheatBreeding). Indore: IAR!. Mimeo.ICAR. 1986. TlVenly Five Years ofCOOl'dinaledWheal Research. New Delhi: IndianCouncil of Agricultural Research.ICARDA. 1989a. Col/ahoralive ResearchProjecr on Ferlilizer Use on Wheal inNorthern Syria, 1986-1988. Reportprepared with the Syrian Arab RepublicMinistry of Agriculture and AgrarianReform Soils Directorate. Aleppo:ICARDA.ICARDA. 1989b. Col/ahoralil'e ResearchProject on Fertilizer Use on Barley inNorthern Syria, 1984-1988. Reportprepared with the Syrian Arab RepublicMinistry of Agriculture and AgrarianReform Soils Directorate. Aleppo:ICARDA.International Wheat Council. J990. GrainMarket Report (December).Johnson, W.F., C.E. Ferguson, and M. Fikry.1983. Tunisia: The Wheal DevelopmenlProfiram. A.!.D. Project Impact EvaluationNo. 48. Washington, D.C.: United StatesAgency for International Development.Jones, M., A. Matar, and M. Pala. 1988. AStrategy for Fertilizer Research in VariableRainfed Environments: Cereals in Syria, aCase Study. In Farm ResourceManafiement Profiram Annual Reportfor1987. Aleppo: ICARDA.Kohli, A.K. 1976. Yield per unit of land andinputs, quantity and quality-wise inrelation to other cereals with reference toMaharashtra. In Wheat in the ChanfiirlfiSocial Conlexl: ProceedinfiS ofa Seminar.New Delhi: Government of India.Kukula, S., and A. Dakermanji. 1986. Wheatproduction practices in northwestern Syria.In Farminfi Syslems Profiram AnnualReporl for 1985. Aleppo: ICARDA.Laing, D.R., and R.A. Fischer. 1977. Adaptationof semidwarf wheat cultivars to rainfedconditions. Euphytica 26: 129-139.Lipton, M., with R. Longhurst. 1989. New Seedsand Poor People. London: Unwin Hyman.Maliwal, G.L. 1989. Bhal and CoastalAgroclimatic Zone: Important ResearchAccomplishments. Arnej Station, GujaratAgricultural University. Mimeo.Mann, C. 1980. The effects of governmentpolicy on income distribution: A casestudy of wheat production in Turkey. InOzbudun and Ulusan (eds.), PoliticalEconomy ofIncome Distribution in Turkey.New York: Holmes and Meier.Mazid, A. 1990. Economic analysis ofresearcher managed trials. In FarmResource Management Program AnnualReporlfor 1989. Aleppo: ICARDA.Oram, P. 1988. Agricultural production andfood deficits in West Asia and NorthAfrica: Future prospects and the role of thehigh-elevation areas. In J. P. Srivastava, M.C. Saxena, S. Varma, and M. Tahir (eds.),Winrer Cereals and Food Legumes inMounlainous Areas. Aleppo: ICARDA.Pala, M., D. Tully, A. Rassam, A. Mazid, and P.Cooper. 1987. Fertilizer and herbicideeffects on farmers' wheat production inNorthwest Syria. Paper presented at theFarming Systems Research Symposium,18-21 October, 1987, University ofArkansas, Little Rock.Pandey, S. 1986. Economics of WaterHarvesting and Supplementary Irrigation inthe Semi-Arid Tropics of India: A SystemsApproach. PhD dissertation. Armidale,Australia: University of New England.Perry, M.W., and M.F. D'Antuono. 1989. Yieldimprovement and associated characteristicsof some Australian spring wheat cultivarsintroduced between 1860 and 1982.Australian Journal ofAgriculluralResearch 40(3): 457-472.Rao, M.V. 1976. Wheat production prospects inIndia. In Wheal in Ihe Changing SocialConrexl: Proceedings ofa Seminar. NewDelhi: Government of India.Rees, D.1., M. Islam, A. Samiullah, F. Rehman,S.H. Raza, Z. Qureshi, and S. Mehmood.1991. Rain-fed crop production systems ofupland Balochistan: Wheat (Triticumaeslivum), barley (Hordeum vulgare), andforage legumes (Vicia species).Experimenlal Agriculture 27: 53-69.Renkow, M. 1991. Technological Change andIncome Distribution in Pakistan's Favoredand Marginal Production Environments.CIMMYT Economics Paper No.4. Mexico,D.F.: CIMMYT.Runge, C.F. 1991. The Developing Counlriesand Ihe Uruguay Round. Staff Paper P91-9.St. Paul: University of MinnesotaDepartment of Agricultural and AppliedEconomics.Scobie, G.M. 1984.lnvestmenr in AgriculturalResearch: Some Economic Principles.CIMMYT Economics Working Paper.Mexico, D.F.: CIMMYT.Serghini,I.H. 1986. Le cerealiculteur Marocainface au developpement technique. InOplions MMilerraneennes: Cereales I!Iproduits cerealiers. Montpellier: InstitutAgronomique Mediterraneen.Sharpley, A.N., and J.R. Williams. 1990.EPlC-Erosion/Productiviry ImpactCalwlalor: I. Model Documentalion.USDA Technical Bulletin No. 1768.Washington. D.C.: United StatesDepartment of Agriculture.Sheopuria, R.R. 1990. A Write Up on WheatFacts and Trends in MarginalEnvironments of Madhya Pradesh, India.Krishi Vishwa Vidyadala ZonalAgricultural Research Station,Powerkheda. Mimeo.Sidhu, D.S., and D. Byerlee. 1991. TechnicalChange and Wheal Productivity in IheIndian Puniab in the Post-GreenRevolution Period. CIMMYT EconomicsWorking Paper. Mexico, D.F.: CIMMYT(forihcoming).Singh, A.1., and D. Byerlee. 1990. Relativevariability in wheat yields across countriesand over time. Journal ofAgriculturalEconomics 41(1): 23-32.Singh, R.P., and S.R. Raje. 1984. Optimizingproduction in rainfed vertisols: A casestudy of an operational research project.Fertilizer News 29(4): 59-71.Skully, D.W. 1990. The International WheatMarket: A Public Choice Perspective.Ph.D. dissertation. Fairfax, Virginia:George Mason University.Somel, K., A. Mazid, and M. Hallajian. 1984.Survey ofBarley Producers in NorthernSyria. 1981/82. Vol. 3: DescripliveStatistics. ICARDA Research Report No.12-III. Aleppo: ICARDA.Thomson. E.F., F. Bahhady. and T.L. Nordblom.1982. The role of livestock in the farmingsystem. FSRP Research Reporr(December). Aleppo: ICARDA.Tunisia Ministry of Agriculture. 1989. Lacerealiculture en Tunisie. Mimeo.United Nations Population Fund. 1990. TheState o/World Population 1990. NewYork: UNPF.Walker, T.S., and J.G. Ryan. 1990. ViI/age andHousehold Economics in India' s Semi-AridTropics. Baltimore and London:Johns Hopkins.Williams.J.R.. P.T. Dyke, W.W. Fuchs, V.W.Benson, O.W. Rice, and E.D. Taylor. 1990.EPIC-ErosionlProductivity ImpactCalculalor: 2. User Manual. USDATechnical Bulletin No. 1768. Washington,D.C.: United States Department ofAgriculture.World Bank. 1990. World Devleopmell/ Reporr1990. New York: Oxford Univresity Press.

Annex 1: Regions of the WorldEastern and North Africa Philippines PeruSouthern Africa Algeria Samoa SurinamBotswana Egypt Singapore VenezuelaBurundi Libya Solomon IslandsComoros Morocco Thailand Southern Cone,Djibouti Tunisia Tokelau South AmericaEthiopia Tonga ArgentinaKenya West Asia Tuvalu BrazilLesotho Afghanistan Vanuatu ChileMadagascar Bahrain Vietnam ParaguayMalawi Cyprus Wallis and Futuna Islands UruguayMauritius Iran Falkland IslandsMozambique Iraq East AsiaNamibia Jordan China Eastern Europe and USSRRwanda Kuwait Korea, North AlbaniaSeychelles Lebanon Korea, South BulgariaSomalia Oman Mongolia CzechoslovakiaSudan Qatar Taiwan Germany. EastSwaziland Saudi Arabia HungaryTanzania Syria Mexico. Central America, PolandUganda Turkey and the Caribbean RumaniaZambia United Arab Emirates Antigua USSRZimbabwe Yemen Arab Republic Bahamas YugoslaviaYemen DemocraticBarbadosWestern and Republic Belize Developed MarketCentral Africa Bermuda EconomiesAngola South Asia Cayman Islands AustraliaBenin Bangladesh Costa Rica AustriaBurkina Faso Bhutan Cuba Belgium-LuxembourgCameroon India Dominica CanadaCape Verde Maldives Dominican Republic DenmarkCentral African Republic Myanmar EI Salvador Faeroe IslandsChad Nepal Grenada FinlandCongo Pakistan Guadeloupe FranceCote d'Ivoire Sri Lanka Guatemala Germany. WestEquatorial Guinea Haiti GreeceGabon Southeast Asia Honduras GreenlandGambia and the Pacific Jamaica IcelandGhana American Samoa Martinique IrelandGuinea Brunei Mexico IsraelGuinea Bissau Cook Islands Montserrat ItalyLiberia East Timor Netherlands Antilles JapanMali Fiji Nicaragua MaltaMauritania French Polynesia Panama NetherlandsNiger Guam SI. Christopher and Nevis New ZealandNigeria Hong Kong SI. Lucia NorwayReunion Indonesia SI. Pierre and Miquelon PortugalSao Tome Kampuchea Republic SI. Vincent Grenadine South AfricaSenegal Kiribati Trinidad and Tobago SpainSierra Leone Laos UK Virgin Islands SwedenSI. Helena Macau US Virgin Islands SwitzerlandTogo Malaysia United KingdomZaire Nauru Andean Region United StatesNew CaledoniaBoliviaNiueColombiaNorfolk IslandEcuadorPacific IslandsFrench GuianaPapua New GuineaGuyana