CIMMYT - IBSA

Editing: Gene HettelAbstract translations: Alma McNabDesign and layout: Miguel MelladoE .. Rafael de la Colina F .. JoseManuel Fouilloux B .. BerthaRegalado M.Typesetting: Silvia Bistrain R..Maricela A. de RamosTyping: Maria de la Cruz Villag6mezand Maria Teresa Pedroza P.

BREEDING STRATEGIESFOR RESISTANCETO THE RUSTS OF WHEAT

The International Maize and Wheat Improvement Center (CIMMYT) is an internationallyfunded. nonprofit scientific research and training organization. Headquartered in Mexico.the Center is engaged in a worldwide research program for maize. wheat. and triticale. withemphasis on food production in developing countries. It is one of 13 nonprofit internationalagricultural research and training centers supported by the Consultative Group onInternational Agricultural Research (CGIAR). which is sponsored by the Food andAgriculture Organization (FAO) of the United Nations. the International Bank forReconstruction and Development (World Bank). and the United Nations DevelopmentProgramme (UNDP). The CGIAR consists of 40 donor countries. international and regionalorganizations. and private foundations.CIMMYT receives support through the CGIAR from a number of sources. including theinternational aid agenCies of Australia. Austria. Brazil. Canada. China. Denmark. FederalRepublic of Germany. France. India. Ireland. Italy. Japan. Mexico. the Netherlands.Norway. the Philippines. Saudi Arabia. Spain. Switzerland. the United Kingdom and theUSA. and from the European Economic Commission. Ford Foundation. Inter-AmericanDevelopment Bank. International Development Research Centre. OPEC Fund forInternational Development. Rockefeller Foundation. UNDP. and World Bank. Responsibilityfor this publication rests solely with CIMMYT.Correct Citation: CIMMYT 1988. Breeding Strategies for Resistance to the Rusts of Wheat.Mexico. D.F. CIMMYT.ISBN 968-6127-23-2

PrefaceAn international workshop onBreeding Strategies for Resistance tothe Rusts of Wheat was held atCIMMYT/Mexico June 29-July I,1987. It brought together an eminentpanel of internationally recognizedspecialists charged with the task ofreviewing the current situation anddefining a broad breeding strategythat could be implemented in thefuture to incorporate the necessarylevel of resistance to control leaf,stem, and yellow rusts (Pucciniarecondita Rob. ex Desm. f.sp. triticiEriks, Puccinia graminis Pers. f.sp.tritici, and Puccinia striiformisWestend) in wheat.In the first seven chapters,authorities from Australia, Canada,Europe, and the United States focusexclusively on the three rusts anddiscuss among other points the roleof specific genes, diversity, and theuse of polygenic, partial, and durableresistances.Chapter 8 addresses diseasemanagement through varietymixtures, based on experience withbarley powdery mildew-a conceptthat has obvious application to therusts. Chapter 9 outlines currentapproaches at CIMMYT in breedingwheat for rust resistance.Chapter 10 summarizes thediscussions in the precedingchapters in the context of a definedstrategy. The outcome represents ageneral consensus on future breedingstrategies that should be employedto incorporate resistance to thesethree serious wheat diseases.We hope that this document shedsnew light on wheat rust resistancebreeding and, as a result, will be avery significant contribution to thescientific literature.We should like to thank theInternational Development ResearchCentre (IDRC) of Canada for thesupport it provided to make thisworkshop possible.Norman W. SimmondsSanjaya RajaramTechnical editors

Table of ContentsPrefaceChapter I-The Role of Specific Genes in Breeding for DurableStem Rust Resistance in Wheat and Triticale. RA. McIntosh. PlantBreeding Institute. Castle Hill. AustraliaChapter 2-Resistance to Leaf and Stem Rusts in Wheat, A.P.Roelfs. Cereal Rust Laboratory, U.S. Department of AgricultureResearch Service and the University of Minnesota. St. Paul,MinnesotaChapter 3-Pathogenicity Analysis of Yellow (Stripe) Rust ofWheat and Its Significance in a Global Context. RW. Stubbs.Research Institute for Plant Protection, Wageningen, TheNetherlandsChapter 4-Using Polygenic Resistance to Breed for Stem RustResistance in Wheat, D.R Knott, Department of Crop Science andPlant Ecology, University of Saskatchewan, Saskatoon. CanadaChapter 5-Strategies for the Utilization of Partial Resistance forthe Control of Cereal Rusts. J.E. Parlevliet. Department of PlantBreeding. Agricultural University, Wageningen, The NetherlandsChapter 6-Durable Resistance to Yellow (Stripe) Rust in Wheatand Its Implications in Plant Breeding, R Johnson, Plant BreedingInstitute, Cambridge, EnglandChapter 7-Current Thinking on the Use of Diversity to BufferSmall Grains Against Highly Epidemic and Variable FoliarPathogens: Problems and Future Prospects, J.A. Browning.Department of Plant Pathology and Microbiology. TexasAgricultural Experiment Station, College Station. TexasChapter 8-The Use of Variety Mixtures to Control Diseases andStabilize Yield, M.S. Wolfe, Plant Breeding Institute. Cambridge.EnglandChapter 9-Current CIMMYT Approaches in Breeding Wheat forRust Resistance, S. Rajaram, RP. Singh. and E. Torres, WheatProgram, CIMMYT. MexicoChapter lO-Synthesis: The Strategy of Rust Resistance Breeding,N.W. Simmonds. Edinburgh School of Agriculture, Edinburgh,Scotlandiv110233948637691101119Abstracts in Spanish 137Abstracts in French 144

Chapter 1The Role of Specific Genes in Breeding forDurable Stem Rust Resistance in Wheat andTriticaleR.A. McIntosh. Plant Breeding Institute. Castle Hill. AustraliaAbstractBreeding for resistance to stem rust in wheat has been successful. Theresistances deployed in agriculture have depended on single identifiablegenes and combinations of identifiable genes. The adult plant resistancegene, Sr2, has contributed to durable resistance in many areas. Stem rustresistant wheats for the northeastern wheat areas of Australia havedepended on the use of resistances which are replaced following thedetection of virulent pathotypes. The deployment ofgenes in this waydepends on relevant pathogenicity surveys, a knowledge of the genes presentin wheat cultivars, and industry cooperation in rapid cultivar replacement.Genetic vulnerability to stem rust in the CIMMYT triticale program can bereduced by using information generated in Australia. The narrow geneticbase for resistance can be widened by the use ofEuropean triticales, rye,and wheat. However, genetic diversity between wheat and triticale should bemaintained.IntroductionDisease resistances and, inparticular. resistance to one or moreof the three rust diseases of wheat(and triticale) represent a small partof the genotype package that mustbe delivered by the wheat breeder.Whereas it is not difficult to find orto produce rust resistant materials, itis difficult to combine high levels ofresistance to multiple diseases withother desired characters. One onlyhas to refer to recent issues of theCIMMYT Review in order to gain anappreciation of what these charactersare and some indication of what theymay involve. Clearly. many of themare more elusive and more difficultto achieve than rust resistance. Thusit is imperative that the rustresistance objective be kept relativelysimple.Breeding wheat. and presumablytriticale. for resistance to rusts isrelatively easy. The problems comewith the genetic plasticity of thepathogens. So often. we no soonerhave resistance when virulentpathotypes increase in frequency andeither render the resistant cultivar(s)vulnerable to disease or actuallycause crop losses. Thus. we get the"boom and bust" cycles of which weare often reminded. The recentoccurrence of stem rust in triticale inAustralia provided a timely reminderof the potential of this disease. Onthe other hand. we shouldacknowledge the success of wheatbreeders in reducing the fear ofwidespread stem rust epidemics.This has been achieved by acombination of disease escapemechanisms such as earlier maturityand alternate host eradication as wellas genetic resistance. This successhas been supported by thecontinuing research effort that hasbeen devoted to stem rust in contrastto many other disease problemsaddressed by intermittent effort asindividuals and fundingorganizations make short-term

2contributions and then abandonthem as researchers gainpromotions, or are distracted, orretire.Breeding for rust resistancecontinues to be largely experimentalin approach and it involves:• The identification of potentialsources of resistance.• Assessment of their effectivenessover sites, seasons, andpathotypes.• Incorporation of them into cultivars. • Introduction of them to agriculturewith cultivar release andrecommendation.The eventual test of a successfulresistant cultivar will involve somemeasure of how widely it is grown,and the time for which it remainsresistant.Features of ResistanceThere are two important criteria ofresistance. namely durability anddiversity.DurabilityThe time for which a cultivar willmaintain its resistance in aparticular area, or at a particulartime, cannot be predicted. Variousstrategies to extend the period ofeffectiveness of a resistance sourcehave been suggested but, on a globalbasis. experience seems to vary. Ifwe could identify a source ofresistance with adequate effectivenessand total durability. then onlythat source of resistance would berequired by all wheat breeders.DiversityThere is a diverse range of rustresistance genes. Many of them havebeen used, and continue to be used,in various ways. In practice, geneticdiversity is used as insurance againsta lack of durability and thus as ameans of reducing geneticvulnerability.On the assumption that we have nodurable sources of stem rustresistance giving adequate protectionunder all conditions. we mustattempt to increase or prolong theeffectiveness of the resistances thatwe have. This will be supported by:• A knowledge of the epidemiologyof the pathogen in relation to theprevailing agricultural practices.• A relevant pathogenicity survey.• A continuing research effort onhost resistance.EpidemiologyStem rust occurs in variousepidemiological regions in thewarmer wheat-growing areas.Epidemics are contingent onconditions of favorable moisture,high levels of initial (source)inoculum (by implication. virulentpathotypes). and by susceptiblehosts. Epidemics can be continentalor local.Pathogenicity SurveysOver the last 60 years. it has beencustomary for rust workers toconduct pathogenicity or racesurveys. There is no doubt that thesehave proved useful in epidemiologicaland evolutionary studies, butthey have been of only limited valueto breeders.Several factors influence thedistribution and frequency ofpathogen genes and genotypes in aparticular area.

3Migration/introductionNew genes can be introduced fromoutside the area. Clearly distinctivepathotypes of Puccinia graminis f.sp.tritici were found in Australia in1926. the early 1950s. and in 1968(15). These presumed exoticintroductions became established indifferent regions. but subsequentlyspread to other areas. Comparativestudies of the putative 1968introductions with collections fromsouthern Africa. and meteorologicaldata. indicated that uredospores hadbeen wind-transported to Australiafrom Africa (16).The evidence for periodicintroduction to Australia of newpathotypes is supported by variousintroductions of new diseases. Thesehave included Puccinia graminis f.sp.secalis in the 1950s (14) andPuccinia striiformis f.sp. tritici in1979. Once the new diseases arrivedin Australia. they becameestablished and spread quite rapidly.and evolved to form new pathotypes.Within a year of its occurrence inAustralia. yellow rust was found inNew Zealand where it subsequentlyformed pathotypes different fromthose found in Australia. However.yellow rust has not been reported inWestern Australia.The original yellow rust introductioncame from Europe (8). presumablytransported to Australia by man (17).Thus both natural factors and manare Significant elements in themovement of diseases or particularpathotypes to new areas.MutationSurveys worldwide have providedample evidence indicating the role ofmutation in the origin of newpathotypes. This has beendemonstrated in Australia for thevarious historic groups of the wheatrust pathogens (8. 15). Mutations arethe most likely and most predictableevents that contribute to short-termrust pathogen variability. Suchchanges can be anticipated providedthere is genetic knowledge ofresistance and can be simulated bymutation experiments in thelaboratory.RecombinationThe sexual cycles of rust pathogenshave obvious implications both forthe evolution of new pathotypes andfor the seasonal carry-over ofinoculum. However. in the absenceof alternate hosts. there areestablished mechanisms of asexualvariation. In Australia. somatichybridization between P. graminisf.sp. tritici and P. graminis f.sp.secalis was probably involved in theorigin of a group of rusts commonlyfound on Agropyron scabrum andbarley (Hordeum vulgare). Inaddition. both pathogenic (6) andisozyme data (1) point to a somatichybridization origin for oneevolutionary pathway of P. graminisf.sp. tritici characterized bypathotype 34-2.1l. However. somatichybridization probably plays only aminor role in the evolution of rustpathogens.SelectionThe genotypes of the predominantcommercia1 cultivars as well as thoseof the wheats. barleys. and grasseson which rust survives betweenseasons will inhuence the pathogengenotypes that survive.ChancePathogen populations go throughmammoth boom and bust cyclesbetween crop seasons or. in thelonger term. between epidemics. Thesurvival of pathotypes in particularareas and between seasons isinfluenced by seasonal and

4agronomic factors. After Cook wheatwas infected by P. graminis f.sp.tritici pt. 343-1.2.3.4.5.6 in 1984. itwas rapidly withdrawn fromcultivation. The frequency of this"Cook" pathotype also quicklydeclined. despite the fact it had thepathogenic abilities of its widespreadprogenitor. 343-1.2.3.5.6. Was thisdecline due to chance because it wasrelatively localized and failed toestablish or was the mutation eventcontributing to its origin associatedwith reduced fitness such that it hada survival advantage only whenpresent on Cook wheat?Uses of pathogenicity surveysFor breeding purposes. a relevantpathogenicity survey should do thefollowing things:• Indicate what pathotypes arepresent. where they occur and.with cautious interpretation. inwhat frequencies. The informationwill be in pathotype codes or aspathogenicity formulas fromwhich avirulence/virulencefrequencies for single genes orgene combinations can bedetermined. Single genefrequencies are not adequate.• Indicate and/or confirm whenpathogenic changes relate tocommercial cultivars. Thus thevalue of surveys will be enhancedby a knowledge of the genesdeployed in those cultivars.• Act as an early warning toextension and advisory servicesinvolved in cultivarrecommendation. There is usuallya lag period between the detectionof a new pathotype and theoccurrence of crop losses as aconsequence.• Provide the pathotypes to be usedin the breeding nursery. A newpathotype can be used in thebreeding nursery before it causesdamage in agriculture.With a relevant pathogenicity surveyand a reasonable knowledge of thegenetics of resistance. induced fieldepidemics required for testing can bebased on one or few releasedpathotypes. The use of a mixture ofall available pathotypes is aninsurance against ignorance. Multipathotypenurseries have problemsin that pathogen componentsprobably do not increase toequivalent levels; the breeder is thenuncertain as to which componentsare present. In practice. eachpathotype in the field nursery shouldbe increased on a host genotype towhich it is specialized.Genetics of ResistanceAt any particular time. breedershave access to resistant cultivars.resistant materials at various stagesof development and potentialresistance sources. The primaryobjective of a genetics program is tounderstand the expression andinheritance of resistance and toknow the range of genetic diversitythat is present in agriculture and inbreeding programs. Most resistancebreeders will demand resistance thatis sufficiently effective and stableand that can be selected by means ofa single assessment in the diseasenursery.I recognize two types of resistance.that effective at the seedling stageand that coming into effect at postseedlingstages. Obviously. potentialresistance sources must conferresistance at growth stagescorresponding to those when damageis likely to occur in the fieldsituation. Some genes that areeffective in seedlings do not confer

5adequate levels of adult plantresistance (e.g. Sr8a. Sr25. andpossibly Sr13).There appears to be somemisunderstanding of the processes ofselection and the use of geneseffective at the seedling stage. Geneseffective in this respect usually alsoconfer adult plant resistance. Genesfor adult plant resistance cannot bedetected in standard seedling tests.However. the breeding approach isdifferent. The breeder is interested inresistance sources. These are initiallydetected or confirmed in field diseasenurseries. If such sources alsodisplay seedling reSistance and. ifgenes responsible contribute toresistance at both growth stages.there seems no reason for notcombining seedling and adultassessments in the breedingexercise. It is usual to manipulateadult plant resistances in fieldnurseries. although they could beselected in the greenhouse.Methods for EstablishingGenetic DiversityLong-term resistance to stem rust isdependent on a continuingavailability of resistance sources.Various procedures assist inestablishing that potential resistancesources carry new or different genesfor resistance.PedigreeWhile useful for postulating geneticdiversity. the breeder must be awarethat apparently unrelated sourcesmay carry the same gene(s) and thatpedigrees might not be as stated.ResponseDifferent seedling or adult plantresponses are indicative of differentgenes. Experienced rust workers canfrequently recognize individual genesand can accurately postulate genesfrom an array of low infection types.SpecificityA practical method for identifyinggenes in resistance sources is multipathotypetesting. Genes can bepostulated from the correlation of theresponses of selected resistancesources with those of controls.However. this classic application ofthe gene-for-gene relationship hasinvolved problems which includemisinterpretation of the concept.miSinterpretation of the data becausegenetic control lines developed inone geographic location were notapplicable in another. use of poordata. and the use of pathotypesinappropriate for breeding purposes.Multi-pathotype tests oftenaccurately identify those geneswhich are useless for breedingbecause the important fieldpathotypes are virulent. However.they may not identify the potentiallyuseful genes because these willprovide resistance to the entirearray. Moreover. the tests are usuallyperformed on seedlings andimportant adult plant resistancesmay not be recognized. Despite suchshortcomings. multi-pathotype tests.combined with relevant fieldmonitoring'; remain reliable andefficient means of distinguishingamong potential resistance sourcesfor use as parents in breedingprograms.Genetic studiesThe most accurate way ofdemonstrating diversity is byconventional genetic analysis.However. this method is timeconsuming because up to fourgenerations may be required. Inaddition. tests of allelism may benecessary. Known associations of

6genes can be important in aiding theidentification and manipulation orresistance genes. For example.wheats with Sr24 will always carryLr24 and. if derived from cv. Agent.will be red-seeded; hexaploid wneatswith Sr9 usually will carry Yr7;wheats with Sr31 will carry Lr26and Yr9. will display two instead offour chromosome satellites in mitoticchromosome preparations. can beidentified by a unique isozymepattern. and cannot have brownchaff. These associations can beuseful when attempting to identify ormanipulate the individualcomponents of multigene resistances.Genes of greatest potential will bethose that are effective against all ormost pathotypes in the area ofinterest. Some genes likely to be ofcurrent worldwide interest arediscussed below:• Sr24-The "Agent" gene producesLIT"2 =" to "2" . Australian whiteseededwheats possessingSr24/Lr24 were derived from Dr.E .R. Sears' 3D/3Ag transfers Nos.3 and 14. Isolates of P. graminisf.sp. tritici virulent for Sr24appeared in South Africa in 1984(4).• Sr26-This gene on chromosome 6A/6Ag produces LIT"; 1". It is present in several Australian wheats and has been deployed over a large area since 1967. • Sr30-The Webster/Festiguay genewith LIT"2" to "3" becameineffective in Australia afterseveral years of use. Howevervirulence in the pathogen hasdeclined and certain of the newerwheats including Banks. Vulcan.and Sunstar. probably possess thisgene. Singh and Mcintosh (10)identified it in Klein Cometa andits presence is suspected in certainCIMMYT-produced wheats such asInia 66. Pavon. and Cheel.Virulence for Sr30 has beenrelatively common in South Africa(5).• Sr31-The Kavkaz/Aurora gene inchromosome IBLlIRS conferringLIT"; 1-" is very common inEuropean winter wheats and inspring wheats developed byCIMMYT. It occurs inapproximately 60% of linesdistributed in the 17thInternational Bread WheatScreening Nursery. Virulence in P.graminis f.sp. tritici has not beenreported. This gene is present inone Australian biscuit wheat butis unlikely to be exploited furtherbecause of fear of the doughstickiness problem that appears tobe associated with the presence ofchromosome 1 RS.Several other genes conferringresistance to a wide array ofpathotypes are being investigated.These include Sr22. Sr32. Sr33.Sr35. and a gene present in VPMland its English derivative.Rendezvous. It appears that fewadditional highly effective resistancegenes will be found in hexaploidwheat but. if required. further genescould be obtained from nonhexaploidWheats and related genera.Adult plant resistanceProbably the most important andmost durable source of stem rustresistance in hexaploid wheat is thattransferred from tetraploid wheatresulting in Hope and H44. Manystudies in the 1930s and 1940s. plusmore recent work (2. 3) showed thatboth seedling and adult plantresistance genes were involved. Theseedling resistance genes Sr9d andSr1 7 provided resistance only tocertain pathotypes. On the otherhand. the adult plant resistance geneSr2. although less effective. proved

7to be a durable source of resistancefor many parts of the world. Sr2 ispresent in many spring wheats andsome winter wheats and its presenceis shown by its association with headand stem melanism known as falseblack. or pseudo-black. chaff. Thismelanism can become excessive insome environments and may beconfused with other diseaseproblems.Sr2 is very common in wheatsdeveloped by CIMMYT. Theseinclude Sonalika. Inia 66. LermaRojo. the Bluebird series. Pavon. andthe Veery series. Consequently. it ispresent in wheats grown throughoutthe world. Sr2 is recessive and itsslow-rusting response permits thedevelopment of variable levels ofdisease. There appears no doubt thatSr2 provides a desirable genetiCbackground into which moreeffective. but less durable resistancegenes can be placed.Multiple gene resistanceDuring the 1960s. Rajaram (9)conducted a genetic study of severalwheats displaying low coefficients ofstem rust infection on a worldwidebasis. Resistance in these wheats toAustralian pathotypes wasdetermined by combinations ofknown and unknown genes. Morerecently. Singh and McIntosh (11)reported that Kenya Plumepossessed eight genes (Sr2. Sr5. Sr6.Sr7a. Sr8a. Sr9b. Sr12, and Sr17).Its field resistance to thepredominant Australian fieldpathotype was determined by Sr2and possibly by an interaction ofSr7a and Sr12. These authors (12)also attributed field resistance inChris wheat to Sr7a:Sr12 interaction.In order to test this conclusion, theAustralian National Wheat RustControl Program has initiated thetransfer of Sr7a to wheats known tocarry Sr12.While wheats with low coefficients ofinfection in multilocational testsfrequently possess multiple genes forresistance, the number of genes ortype of gene interaction operating ateach site has not been determined.However. information of this typewill be essential if breeders are toassemble a diversity of suchresistance as a means of achievinggreater stability.Stem Rust on Triticalein AustraliaGeneral observationsStem rust first appeared on triticale(x Triticosecale Wittmack) inAustralia in 1981 but it was notuntil 1982 that we showed that aunique P. graminis Lsp. triticipathotype was involved. The genepresent in the affected triticales wasSr27 which had originated fromImperial rye. Comparative studieswith selected P. graminis f.sp. triticicultures showed that 67% of entriesof the 12th International TriticaleScreening Nursery (lTSN) possessedthis gene. Moreover, Australiancommercial cultivars and CIMMYTlines with Sr27 were extremelysusceptible to pt. 34-2.12 ("-.12"refers to virulence on seedlings of cv.Coorong) as adult plants (7). In 1984.a further mutational changeresulting in pt. 34-2.12.13 ("-.13"refers to virulence on Satu) causedsevere rusting on Satu and Toort andmoderate rusting on Ningadhu(Drira). Venus (Beagle), Currency.and Samson (Ram). In seedling tests,the IT"12" response of Ningadhuwas distinguishable from IT"1 + 3 + ..displayed by Venus, Currency, andSamson and the IT"3 +" of Satu andToort.In 1984, a grant from the AustralianRural Credits Development Fundenabled the appOintment of Dr. S.J.Singh to investigate the genetics ofrust resistance in triticale, to provide

8rust screening services to triticalebreeders. and to initiate abackcrossing program to transferresistance genes to rust susceptiblegenotypes. The genetic findings arelisted below:• The genes Sr27 in Coorong andSrSatu in Satu are allelic (13).• A second allelic series involvesgenes in Tejon-Beagle (IT"; 1 + N").Ningadhu (IT" 12"). and Juanillo100 (IT''23'').• A gene in 14th ITSN No. 64 and15th ITSN No. 99 (IT; 1-") and agene in 14th ITSN No. 122(IT"; 1-2-") appear to beindependent of the above groups.• A highly effective gene occurs in17th ITSN No. 78 (IT";"). Thisgene may be derived from a Polishtriticale. Other studies haveindicated high levels of resistancein European triticales to pt.34-2.12.13.• Wheat genes Sr9b (13th ITSN No.33) and Sr36 (University of NewEngland) have been identified.• There was no evidence for the presence of Sr3l in triticale. The adult plant responses to pt.34-2.12.13 in the field closelycorrespond to the seedling responses.Comparisons of chromosomallysubstituted and complete triticaleslisted separately in the 17th ITSNdemonstrated that terminal rustseverities for the complete groupwere generally lower than those forthe substituted entries. While thisundoubtedly reflected the higherfrequencies of the Ningadhu andVenus genes in the completetriticales. other genes may bepresent. There was no evidence foradult plant resistance genes intriticale.Additional studies have shown thatSatu triticale carries a leaf rustresistance gene. LrSatu (IT; I")showing 11 % genetic recombinationwith SrSatu. Preliminary evidencesuggests that this gene contributesto the adult plant leaf rust resistanceof Satu and many CIMMYT lines.Whereas CIMMYT materials aregenerally resistant to leaf rust inAustralia. European triticales areoften very susceptible. Thus. the useof European triticales as sources ofstem rust resistance will requireclose monitoring in order to preventloss of resistance to leaf rust.ConclusionDespite the spectacular break-downof stem resistance in triticale. severalfurther genes for resistance areavailable in CIMMYT lines andmaterials developed elsewhere. Anawareness of the high frequencies ofSr27 and SrSatu in CIMMYTtriticales and the knowledge thatvirulent pathotypes are present inAustralia should enable rapidprogress in broadening the geneticbase for stem rust resistance in theCIMMYT triticale population.Although some wheat resistancegenes occur in triticale. geneticdiversity for resistance between thetwo crops should be maintained.AcknowledgementsFinancial support from theAustralian Wheat Industry ResearchCouncil and the Rural CreditsDevelopment Fund is gratefullyacknowledged.References1. Burdon. J.J.. D.R. Marshall. N.H.Luig. and D.J.S. Gow. 1982. Isozymestudies on the origin and evolution ofPuccinia graminis f.sp. tritici inAustralia. Australian Journal ofBiological SCiences. 35:231-238.

110.2. moderately resistant = 0.4.mixed = 0 .6 . moderately susceptible== 0.8. and susceptible == 1.0. Thus.a 60MR (60 x 0.4 = 24) and 20S (20 x1.0 == 20) give a similar averagecoefficient of infection value. Arethese equal? Combining severity andhost response can thus obscuredifferences in susceptibility andresistance. Also. differences in hostresponse are based on lesion sizeand characteristics. not on sporeproduction potential. The severity ona resistant line in a nursery maydepend more on its neighbors thanon its genotype for rust resistance(23). Perhaps it is time to redesignour nurseries for some types ofresistances by blocking similarmaterial together in the nursery orinserting a border row between testlines or planting three-row plots andlimiting notes to the center row. Forsome types of resistance it will benecessary to redesign completely ournurseries for disease evaluation (seeRowell and McVey (26)). Thus. totest for resistance due to a longerlatent period or lower receptivity(fewer lesions) it may be necessary toinoculate uniformly a nursery over ashort period (1-7 days) and thenscore host response 14 to 21 dayslater. To be effective this nurserymust be isolated from otherinoculum sources.Components of ResistancePerhaps the first stage in breedingfor resistance Is the carefulobservation of the proposedresistance. We must ask why do wewant this as a resistant parent?Under what conditions were we ableto detect the resistance? Then wemust design experiments ornurseries (using the same or similarconditions) to allow for the detectionof this particular resistance inprogenies. It may be easier to do thisif we look at the mechanisms(components) of resistance. Fourfactors that can be measured are thenumber or lesions per unit of leaf orstem area (receptivity). size ofsporulating area of the uredium.length of latent period (time frominfection to sporulation). and lengthof sporulating period. The genes forresistance that have been studied indetail may affect one or more ofthese components. Sr2 reduces thenumber of lesions but not evenlythroughout the plant life span nor onall host tissues (8. 31). Sr8a reducesthe size but not the number oflesions. Sr36 lengthens the latentperiod and. with most cultures. alsoreduces the number of pustules (24.25). Resistances. such as Sr23. thatare expressed with chlorosis ornecrosis often have shorter periods ofsporulation for a given uredium.Early telia formation is anotherexample of this type of resistancemechanism. Thus. knowing how theresistance is expressed should makeit easier to design the proper testand to follow it through a breedingprogram. Breeding for resistancemay not be made easier; in fact.more complicated notes may berequired but the end result may bebetter. Notes may have to be takenat different times or in different waysdepending on the cross. A uniformtest or level of resistance across allcrosses may be neither desirable norpossible.Gene-for-Gene Relationship. -.. .: .. ' .Although all resistances may not beon a gene-for-gene basis. many are.Understanding the complexity of theinteraction is important in breedingfor disease resistance. It is oftenassumed that genes for resistanceare dominant. Our experience wouldindicate that most resistance genesare expressed as incompletedominants; a few are almostrecessive. Thus. in a breedingprogram the heterozygousindividuals will probably be moresusceptible than the resistant parent.Temperature. inoculum density. and

12host growth stage usually have someeffect on the expression of resistance.Genes for virulence. likewise. areassumed to be recessive but this isnot always the case. The typicalgene-for-gene interaction is shown inTable 1. The representation is oversimplifiedIn that the heterozygousgenotypes of both host and pathogenare omitted. Additionally. the diseaseresponse is characterized only as ahigh or low infection type. Lowinfection types result (asswningresistance is dominant andavirulence is recessive) fromhomozygous and heterozygousavirulent pathogen cultures witheither a homozygous or heterozygousresistant host genotype. Thehomozygous virulent culture withany host genotype or any pathogenwith the susceptible host results inhigh infection types (13).Table 1. The gene-for-geneinteraction as frequently shownwith infection types represented ashighs and lowsHostRrrpLowHighPathogenppHighHighThe simple gene-for-gene model inTable 1 is expanded in Table 2.showing the interaction betweenSr7b in the host and thecorresponding virulence/avirulencelocus in the host. Note that. in Table2. when the heterozygous pathogenor heterozygous host is involved inthe interaction. detectable change inthe low infection types occur (21).This is probably the usual case. Achange in infection type often occurswhen the temperature is altered (4).The P7bP7b pathogen genotypeinteraction with the three possiblehost genotypes in a segregating F2population could be fitted to a 1 :2: 1or 3: 1 ratio. The P7bP7b pathogengenotype interaction could also bescored as a 1:2: 1 or 3: 1 ratio;however. the 3 ± and 4 infectiontypes might be combined giving a1:3 ratio of resistant to susceptibleplants. This variation in infectiontype due to heterozygous individualscould cause the apparent loss orreduction in effectiveness ofreSistance when a heterozygous hostgenotype was evaluated or when aheterozygous culture was substitutedfor the homozygous avirulentculture. A range of phenotypes in theF2 is assumed to indicate polygeniccontrol. However in a F2 populationwith a single host-pathogen genepair responding to the pathogen. atleast five, different infection typesTable 2. The gene-for-gene interaction for Sr7b and p7bPathogenP7bP7bP7bP7b2 23 4Host 2+ 3± 44 4 4

13usually occur. If the test wereperformed under field conditions.additional phenotypes would beinduced by an environmental effecton the host-pathogen interaction.The range of phenotypes is evenmore evident when two genes inboth the host and the pathogeninteract (Table 3). Many wheatcultivars in the rust-prone areas ofthe world have many genes forresistance. e.g. Marquis released in1911 (Sr7b. 18. 19. 20. X). Selkirk in1953 (Sr6. 7b. 9d, 17, 23. 2) andCenturk in 1972 (Sr6. 8a, 9a. 17);thus a single cultivar can result in awide range of infection types andinteractions if a wide spectrum ofcultures is used.Interaction betweenGenes and GenomesMany accessions of wheats or wheatrelatives with lower ploidy levelshave often been looked to aspotential sources of resistance torust. Derivatives of some of thesesources have been very useful. e.g.Thatcher and Hope to stem rust.However. many such attempts havebeen disappointing. As the resistanceis transferred to successively higherploidy levels. the expression of theresistance decreases (6). This wouldseem to be a 'dilution' effect.Perhaps this can be overcome in thefuture by transferring the resistanceto another homoeologous pair so thatTable 3. The gene-for-gene interaction for Sr6 and Sr7bPs P7b Ps P7b Ps P7b Ps P7b Ps P7b Ps P7b PS P7b PS P7b PS P7bPs P7b Ps P7b PS P7b PS P7b PS P7b PS P7b PS P7b Ps P7b PS P7bR6 R7bR6 R7bR6 R7bR6 r7bR6 f7bR6 f7bR6 R7br6 R7bR6 R7br6 r7bo· 0; 0; ; 1 ; 1 ; 1 2 2­ 40; 0 ; o· ; 1 ; 1 ; 1 23 32 4o· 0 ; 0; ; 1 ; 1 ; 1 4 4 41+1 +1 +1 +1+1 +x­x­x­x­x­,x­2­23232441+ 1 + x­ x­ x­ 4 4 4r6 R7br6 R7br6 R7br6 f7br6 r7br6 r7b2­ 2 4 2­ 2 4 2· 2 423 32 4 23 32 4 23 32 44 4 4 4 4 4 4 4 4

14four alleles of a resistance gene couldbe present in a cultivar. The'dilution' effect is especiallyimportant for wheat leaf rust wherebroadly effective genes for resistanceare few in hexaploid wheats andhigh levels of resistance exist inTriticum monococcum and T.durum . The resistance in thesespecies is apparently effective anddurable.Recently, Dyck (unpublished) hasdescribed a suppressor gene forwheat stem rust on the 7Dchromosome. This suppressor geneappears to be the same as, or veryclosely linked to, the Lr34 locus.Fortunately, the Lr34 resistanceallele is an indication of the presenceof the non-suppressed stem rustgenotype. A great deal of work needsto be done in this area. Are thereother suppressors? To what extentdo they affect leaf rust resistance?Are they primarily on the D genomewhere they affect leaf rust resistancetransferred from T. durum and T .monococcum?When Lr13 (adult plant resistance)was combined with Lr16 (resistant atall stages), cultures virulent to Lr16were avirulent on seedling plantswith Lr13 and 16 (29). In fact thelow infection type produced appearedto be similar to that of a cultureavirulent to Lr16. The enhancementof Lr16 resistance by Lr13 is anexciting discovery and should lead tothe evaluation of many specific genecombinations. Generally,combinations of resistance genesresult in an effect similar to the moreresistant of the genes in thecombinations. However,combinations of a gene that producefew pustUles, i.e. Sr2 combined witha gene that produces small pustUles,i.e. Sr24 results in fewer and smallerlesions.Durable ResistanceDurable resistance is that resistancewhich has been adequate against thedisease for a number of years over arange of environments and pathogencultures. It should not be assumedthat it will always be adequate in thefuture nor that it will be effectiveagainst all cultures. However, theuse of a resistance that has beeneffective over a range ofenvironments, cultures, and years iscertainly more likely to lead to aresistant cultivar than untestedresistance and certainly more apt tosucceed than resistances that areknown to have failed. In case of stemrust, there are several knownsources of durable resistance relatedto a 'single' gene while, for wheatleaf rust, most durable resistance isassociated with gene combinations.Wheat Stem RustSr2-This is an adult plantresistance (not effective until aroundthe boot stage) which was derivedfrom Yaroslav emmer by McFadden(17) and is generally availablethrough the cultivars Hope and H-44and their derivatives (Table 4). InNorth America, the spring wheatSelkirk and the hard red winterwheat Centurk have this gene incombination with others. This genedoes not provide immunity and,under high inoculum densities, isoften characterized by susceptibletypelesions near the nodes and inthe spike and awns (8, 31). Cultivarswith Sr2 in combination with othergenes have been grown on millionsof hectares in the northern GreatPlains of North America withoutserious disease for nearly 30 years.Eagle, a hard red winter wheat fromKansas with Sr2, was grown on overa million hectares for 5 yearswithout stem rust losses. Incombination with other effectiveresistances, Sr2 is difficult to follow

15in progenies. The brown necrosisassociated with Sr2 has often beenused to follow the resistance.Sr26-This resistance. which wasderived from Agropyron elongatum.has been an effective resistanceagainst cultures obtained worldwide.It has been widely used in Australiancultivars (Table 4) which have beengrown on a million hectares annuallyfor over 10 years (14). Thisresistance should be easy to followusing standard breeding techniques.Table 4. Cultivars with selected genes for resistance to wheat stem rustSr2 Sr31 Loerie ViriEagle (Ks) Advokat Lovrin 10 WeiqueHartog Agra Lovrin 12 WentzelKalkee Almus Lovrin 13 WinnetouKenya Page Alondra Lovrin 24 Yan 7770-4Kenya Plume Aurora Magister Cebeco Yi 78-4078Lancer Balkan Mamut ZorbaLawrence Benno MerkurLerma Rojo 64 Bezostaya 2 Mildress Sr36Madden Burgas 1 Millaleau Inia AbeNewthatch Burgas 2 Mironovskaya 10 ArthurNuri 70 Cebeco 97 Nizkoroskava Arthur 71Ottawa Clement Nautica CookPembina Cordillera Odessa DipkaRedman Danubia Odessa 4 FlaminkRenown Disponet Odesskaya 66 GouritzRescue Feldkrone Odilo HandScout Feng Kang 2 Orlando Idaed 59Scout 66 Feng Kang 8 Pakistan 81 KenoshaSelkirk Feng Kang 15 Perseus MendosSonalika Fundulea 29 Poleskaya 71 MengaviSonglen Fundulea 262 Predgornaya 2 OasisSuneca Gamtoos Roxana Ramona 64Sunkota Genaro F 81 Sabina RoughriderGlennson M81 Saladin ShortimSr26 Goetz Salzmunde )4/44 SonglenAvocet Granada Selekta TimgalenBass Hamlet Seri 82 TimsonBlade Helios Seric TimveraEagle (Aust) Huenufen Shtorm TimwinFlinders Iris Siouxland WigleHarrier Istra Skorospelka Zaragosa 75Hybrid Titan Jing Dan 106 Skorospelka 35Jabiru Jugoslaviaya SlaviaKing Kaloyan SolarisKite Kavkaz SutjeskaSunelg Kronjuwel TransilvaniaTakari Licanka Ures T81Terra Lima 1 UrbanLinosVeery'S'From Bartos (1). Hu and Roelfs (10). Luig (14). and McVey (unpublished)

16screening either seedling or adultplants for resistance. Itsdisadvantage is the narrowagronomic background of thecultivars in which the resistancecurrently exists.Sr31-This gene was derived fromImperial rye. It is currently widelyspread in the world population inmany wheats (Table 4) . This gene ison the 1B/1R translocation whichalso carries Yr9 and Lr26. as wellhaving a sticky dough. a poor mixingcharacteristic. Although theresistance is useful. further work isneeded to remove the undesirablecharacter.Sr36-This gene has beensuccessfully used in much of theUnited States but failed once inAustralia (15). It reduces the numberof lesions and increases the latentperiod (24. 25). This resistance losesits effectiveness at or near maturity.Sources are shown in Table 4.Thatcher-This resistance wasderived from Iumillo durum byHayes and others (9). In Thatcher aresistance exists in addition to thatprovided by the combination of Sr5,9g, 12. and 16. Brennan (2) thoughtthat was due to two recessive genes.Nazareno and Roelfs (19) indicatedthat the resistance was oftenTable 5. Known host genes for resistance to wheat stem rust and theirresponses to P. graminis f.sp. triticiResponse to an avirulentculture alSeedlingSr infection Adult Chromosomegene Sourcebl typec/ plantdl locationel Effective2 T. dicoccon 4 S (few uredia) 3BS worldwide5 T . aestivum 0.0; R 6Da6 T. aestivum ;. ; 1­ R 2Da7a T . aestivum lC.23C MS 4BL7b T. aestivum 2 .2+ MS 4BL8a T. aestlvum 2 MS 6Aa8b T. aestivum X MS 6Aa9a T. aestivum 2.23 MS 2BL9b T. aestivum 2+ MS 2BL9d T. dicoccon ; 1.2­ R-MR 2BLge T. dicoccum ; 1.2­ R-MR 2BL9f T. aestivum ;2 ? 2BL9g T . durum 2 MR 2BL10 T. aestivum ; 1 +N.23CN MR-MS ?11 T. aestivum ;2-.2+3­ MR 6BL12 T. durum O;X­ MR 3BS13 T. durum 2-.2+ MR 6Ab worldwide14 T. durum 12C.23C MR-MS IBL15 T. aestivum ;1 +N.XN MS-S 7AL16 T. aestivum 2-.2+ MS 2BL17 T. dicoccon ; IN.X-N MR 7BL18 T. aestivum ;2 ? lDL19 T. aestivum 1 ? 2BL20 T. aestivum 2 ? 2BL21 T. monococcum ; 1.2­ MR 2AL22 T. boeoticum 0;.2­ R-MR 7AL worldwide23 T. aestivum 23C.l +C MS-S.MR 4A

17associated with Sr12 in lines derivedfrom Thatcher. This resistance ismore effective in the field under lowto moderate rust epidemics where itis generally rated susceptible, with alow (5-30%) severity on the modifiedCobb scale; however in severelydiseased nurseries, severities of80-90% are common. On seedlings,it produces an 'X' or 'X-' infectiontype at 18°C but is susceptible at20°C. This resistance has beeneffective in the northern Great Plainssince Thatcher was released in 1934although. in the 1953 and 1954epidemics. it was damaged whengrown in conjunction with the moresusceptible Lee and durum wheats.It is probably present in most of theThatcher derivatives grown in thisarea. Chris. Era. Neepawa, andColumbus may be examples ofcultivars with this resistance. Manyof the CIMMYT cultivars from CIANO67 onwards may have someresistance from Thatcher. but thishas not been proved.In Table 5 the designated genes forstem rust resistance are listed. alongwith infection types produced onseedlings by avirulent cultures,chromosomal locations, sources ofresistance, and where the gene iseffective against the naturalpopulation of P. graminis.Table 5. (continued)Response to an avirulentculture aJSeedlingSr infection Adult Chromosomegene Source bl typecl plantdl locationel Effective24 A. elongatrim 2-,2+ MS 3DL not in S. Africa25 A. elongatum 2·,2+ MS 7DL worldwide26 A. elongatum ;2· R-MR 6Ab worldwide27 S. cereaJe 0; R 3A not in Australia28 T. aestivum R 2BL South Asia29 T. aestivum 2+,2· MR 6Db worldwide?30 T. aestivum 2-,2 MR 5DL worldwide31 S. cereaJe 0;,2­ R IBL worldwide32 Ae. squarrosa 2· MR 2AS worldwide33 Ae. squarrosa 2 MR lDL worldwide34 T. aestivum 3C.l +C MS-S,MR 2A35 T. monococcum 0; R 3Aa36 T. timopheevll 0,0;1 + ,40; R.S (few uredia) 2BS37 T . timopheeviJ 0; R4Ab worldwideKt'2' T. aestivum 2 ? 2BLUfHgT. aestlvumT. dlcoccon21CN,1 +C2CMRMSal Updated from Roelfs and McVey (22). LUig (14). Roelfs (20)bl Triticum = Too Agropyron = A.. AegiJops = Aeoo 5ecaJe = S.cl Infection types at 18°C (plants with Sm. 10. 12. 15. and 17 are more susceptible athigher temperatures. whereas plants with 5r13 are more resistant): variation isencountered with host genetic background and plOidy level also (Knott. 11: LUig andRajaram. 16)dl Many host resistances are less effective at high temperatures. high inoculum densities.and at plant maturity. Variations also occur with different host genetic background. R =resistant, MR = moderately resistant. MS = moderately susceptible and S = susceptibleel Updated from McIntosh (18)fI Gene from H-44. other than 5r7b. 9d. and 17. 5rH was the cause of the differences inCanadian and United States survey data in the 1970s (7)2D?

18Wheat Leaf RustLeaf rust is probably the mostimportant disease of wheat on aworldwide basis (28). Durableresistance to leaf rust is thought tobe more difficult to obtain than withstem rust but some successes havebeen recorded. Leaf rust is morediverse for virulence than stem rust.This diversity may be the result ofone or more factors. First. thepopulation that survives betweenwheat crops probably is much largerfor leaf rust. Second. the pathogenpopulation size is currently muchlarger during the crop season. Third.resistance deployed against leaf rusthas often been a single gene at atime. Thus population sizes arelarge. which results in a greaterprobability of mutants (30) and agreater probability that a greaterdiversity of virulence/avirulencecombinations can survive the nonwheatgrowing period. The use ofcultivars with single effective genesfor resistance permits mutations atsingle pathogen loci to renderresistances ineffective. The sexualcycle of P. recondita f.sp. tritici is notgenerally thought to have a majoreffect (5. 27). Evidence thatparasexual recombinants survive innature. if they occur at all. islacking. Although P. recondita has awide range. the host range of P.recondita f.sp. tritici seems to belimited to Triticum and perhaps afew very closely related genera.However. additional research isdesirable on the role of non-Triticumspecies on survival of P. reconditaf.sp. tritici. especially during thenon-wheat growing period.Leaf rust resistance provided by thesingle genes (Table 6). with theexception of Lr19. are inadequate bythemselves. Lr19 resistance has yetto be tested on commercial acreageso its durability is not proven.Unfortunately. Lr19 has usually beenassociated with a yellow flour color.The other genes listed are usefulonly in combinations of two or more.Durable Resistanceto Leaf RustThe most durable resistance to leafrust is associated with a few genecombinations (Table 7). It appearsthat Lr13 and perhaps Lr12. bothadult plant resistances, incombinations with Lr34, are thebasis of most of this resistance. Theoriginal source of these genes isunknown but apparently Lr13 and34 were present in Alfredo Chaves, aland cultivar found in Brazil about1921. Americano 440 was selectedin 1918 from a land cultivar inUruguay. We have not determinedits genotype for resistance, but itprobably includes Lr12 and/or Lr13and Lr34. Thus, these two landcultivars. which may be very Similar,have been the resistance source ofmost of the durably resistantcultivars. It is assumed that theseland cultivars had a southernEuropean prigin but no Europeancultivars are known to have thislevel of resistance. Resistanceconditioned by Lr13 or Lr12 issometimes inadequate underconditions that are very favorable forthe disease, in nurseries in whichinoculum levels are very high. inareas where wheat is grown at hightemperatures, and in cultivarswithout other resistance.

19Table 6. Known host genes for resistance to wheat leaf rust and theirresponse to P. recondlta f.sp. trltlclLrgeneSourceblResponse to an avirulentculture aJSeedlinginfection Adult Chromosometypecl plantdl locationelEffective1 T . aestivum2a T. aestivum2b T . aestivum2c T . aestivum3a T . aestivum3(bg) T . aestivum3(ka) T. aestivum9 Ae. umbellulata10 T. aestivum1112T. aestivumT . aestlvum13 T. aestivum14a T. dicoccon14b T . aestivum15 T. aestivum1617T. aestivumT. aestivum18 T. aestivum19 A. intennedium20 T. aestivum21 Ae. squarrosa22a Ae. squarrosa22b T . durum23 T. aestivum24 A. elongatum25 S. cereale26 S. cereale27 T. aestivum28 Ae. speJtoides29 A. intennedium30 T . aestivum31 T . aestivum32 Ae. squarrosa33 T. aestivum34(T2) T. aestivum0; R 50L0; R 20a; 1 R 20aO;lN R 20a;C R 6BL;C.23 MR-MS 6BL:C.12C MR 6BL0; R 6BL;,2 R.MS lASY MR 2AR4AR2BSX MS 7BLX MS 7BL:C R 20a:IN MS-MR 4A;1 +N MR-MS 2AS2± MS 5BL0 R 70L0; R 7AL0 R lDLR 20aR 20a0 ;1 MR 2B0: R 30L;N R 4Ab0; R IBL0: MR 3BS0; R 4BL0 ; R 70S; 1 R 4BLMR 4Ab: 1 + MR 3D1+ MR IBL23C MS 70in combinationsin combinationsin combinationsworldwide?In combil1ationsonly with Lr31only with Lr27in combinationsat Updated from Browder (3) and Long (unpublished) bl Triticum = T .. Agropyron = A .. Aegilops = Ae .. and Secale = S cl Infection types are at 20°C. can be more or less resistant at other temperatures (4)dl Many host resistances are less effective at high temperatures. high inoculum densities.and at plant maturity. Variations also occur with different host genetic background.R = resistant. MR = moderately resistant. MS = moderately susceptible. andS = susceptibleel Updated from McIntosh (18)

20Table 7. Bread wheat cultivars with durable leaf rust resistanceProbable sourceName Habit Source Released of resistance Lr gene(s)Americano 44D spring Uruguay 1918 land race ?Frondosa spring Brazil 1934 Alfredo Chaves (land race) 13.+Fronteira spring Brazil 1934 Alfredo Chaves (land race) 13.+Surpreza spring Brazil 1934 Alfredo Chaves (land race) 13.+Frontana spring Brazil 1943 Frondosa 13.34.T3La Prevision 3 spring Argentina 1935 Amerlcano 44D 13.34. +La Prevision 25 spring Argentina 1937 Amerlcano 44D 13.34. +La PreviSion 28 spring Argentina Amerlcano 44D 13.34. +La Prevision 32 spring Argentina 1935 Americano 44D 13.34. +Klein Aniversario spring Argentina 1945 Americano 44D 13.3ka. +Klein Cometa spring Argentina 1942 Americano 44D 13.+Klein Lucero spring Argentina 1950 Americano 44D 17 +Klein Progreso spring Argentina 1937 Amerlcano 44DKlein Rendidor spring Argentina 1954 Americano 44D13.+13.+Klein Sinvalocho spring Argentina Americano 44D 13.+Klein Titan spring Argentina 1925 Americano 44D 13.+Klein Vencedor spring Argentina 1925 Amerlcano 44D 13.+Ciano 67 spring CIMMYT 1967 Chris 13.+Pavon F76 spring CIMMYT 1976 Ciano 67'S' 13.+Minter winter USA 1949 ?Sturdy winter USA 1960 Klein Sinvalocho 10.12.34Gage winter USA 1963 ? 3. +Redcoat winter USA 1960 Surpreza 13.+Atlas 66 winter USA 1948 Frondosa 13.+Chris spring USA 1965 Frontana 13.34. +Era spring USA 1970 Frontana 10.13.34. +References4. Browder, L.E., and M.G.Eversmeyer. 1987. Influence of1. Bartos. P. 1984. Genes for stem temperature on development ofrust resistance in European wheats. Puccinia recondita with TriticumProceedings of the European and aestivum 'Suwon 85'.Mediterranean Cereals Rusts Phytopathology, 77:546-554.'tConference, 6:29-32.5. Chester, K.S. 1946. The Nature2. Brennan, P.S. 1975. General and Prevention of the Cereal Rusts resistance in wheat (Triticum as Exemplified in the LeafRust of aestivum L.) to stem rust (Puccinia Wheat. Chronica Botanica, Waltham, graminis Pers. f.s.p tritici Erik. andHenn.) Ph.D. Thesis, University of Saskatchewan, pp. 142.Massachusetts. 6. Dyck, P.L., and E.R. Kerber. 1985. Resistance of the race-specific-type.3. Browder, L.E. 1980. A The Cereal Rusts Vol. II. Diseases.compendium of information aboutnamed genes for low reaction toPuccinia recondita in wheat. CropSCience, 20:775-779.Distribution, Epidemiology andControl. A.P. Roelfs and W.R.Bushnell (eds.), 469-500, Orlando,Florida, Academic Press.

217. Green. G.J.. and P.L. Oyck. 1979.A gene for resistance to Pucciniagraminis f.sp. tritici that is presentin wheat cultivar H-44 but not incultivar Hope. Phytopathology.69:672-675.8. Hare. RA.. and RA. McIntosh.1979. Genetic and cytogeneticstudies of durable adult-plantresistances in 'Hope' and relatedcultivars to wheat rusts. Zeitschriftfur Pflanzenzuchtung. 83:350-367.9 . Hayes. H.K.. E.R Ausemus. E.C.Stakman. C.H. Bailey. H.K. Wilson.RH. Bamberg. M.C. Morkley. RF.Crim. and M.N. Levine. 1936.Thatcher wheat. Bulletin of theMinnesota Agricultural ExperimentalStation. 325. pp. 36.10. Hu. C.C.. and A.P. Roelfs. 1986.Postulation of genes for stem rustresistance in 13 Chinese wheatcultivars. Cereal Rusts Bulletin.14:68-74.11. Knott. O.R 1981. The effects ofgenotype and temperature on theresistance to Puccinia graminis triticicontrolled by the Sr6 gene inTriticum aestivum. CanadianJournal of Genetics and Cytology.23:183-190.12. Loegering. W.g. 1968. A secondgene for resistance to Pucciniagraminis f.sp. tritici in Red Egyptian20 wheat substitution line.Phytopathology. 58:584-586.13. Loegering. W.g.. and H.R Powers Jr. 1962. Inheritance of pathogenicity in a cross of physiological races III and 36 of Puccinia graminis f.sp. tritici. Phytopathology. 62:546-554. 14. Luig. N.H. 1983. A Survey ofVirulence Genes in Wheat StemRust, Puccinia graminis f.sp. tritici.Berlin. Parey.15. Luig. N.H. 1985. Epidemiology inAustralia and New Zealand. TheCereal Rusts Vol. II. Diseases.Distribution. Epidemiology andControl. A.P. Roelfs and W.RBushnell (eds.). 301-328. Orlando.Florida. Academic Press.16. Luig. N.H .. and S. Rajaram.1972. The effect of temperature andgenetic background on host geneexpression and interaction toPuccinia graminis tritici.Phytopathology. 62:1171-1174.17. McFadden. E.S. 1930. Asuccessful transfer of emmercharacters to vulgare wheat.Agronomy Journal. 22:1020-1034.18. McIntosh. RA. 1979. A catalogueof gene symbols for wheat.Proceedings of the Fifth InternationalWheat Genetics Symposium 1978.2 : 1299-1309.19. Nazareno. N.RX.. and A.P.Roelfs. 1981. Adult plant resistanceof Thatcher wheat to stem rust.Phytopathology. 71: 181-185.20. Roelfs. A.P. 1985. Wheat and ryestem rusts.' The Cereal Rusts. Vol. II.Diseases. Distribution, Epidemiologyand Control. A.P. Roelfs and W .RBushnell (eds.) 3-37. Academic Press.Orlando. Florida.21. Roelfs. A.P .. and J.V. Groth.1987. Puccinia graminis f.sp. tritici.Advances in Plant PathologJ], Vol. VI.Genetics of PathogeniC Fun~. G.S.Sihdu (ed). Orlando. Florida.Academic Press.

2222. Roelfs. A.P .. and D.V. McVey.1979. Low infection types producedby Puccinia graminis f.sp. tritici andwheat lines with designated genesfor resistance. Phytopathology.69:722-730.23. Roelfs. A.P.. D.V. McVey. D.L.Long. and J.B. Rowell. 1972. Naturalrust epidemics in wheat nurseries asaffected by inoculum density. PlantDisease Reporter. 56:410-414.24. Rowell. J.B. 1981. Relation ofpostpenetration events in Idaed 59wheat seedlings to low receptivity toinfection by Puccinia graminis f.sp.tritici. Phytopathology. 71:732-736.25. Rowell. J.B. 1982. Control ofwheat stem rust by low receptivity toinfection conditioned by a singledominant gene. Phytopathology.72:297-299.26. Rowell. J.B.. and D.V. McVey.1979. A method for field evaluationof wheats for low receptivity toinfection by Puccinia graminis f.sp.tritici. Phytopathology. 69:405-409.27. Saari. E.E.. and J.M. Prescott.1985. World distribution in relationto economic losses. The Cereal RustsVol. 11 Diseases, Distribution,Epidemiology and Control. A.P.Roelfs. and W. R. Bushnell (eds).259-298. Academic Press. Orlando.Florida.28. Samborski. D.J. 1985. Wheat leafrust. The Cereal Rusts Vol. 11.Diseases. Distribution. Epidemiologyand Control. A.P. Roelfs and W.R.Bushnell (eds.), 39-59. AcademicPress. Orlando. Florida.29. Samborski. D.J.. and P.L. Dyck.1982. Enhancement of resistance toPuccinia recondita by interactions ofresistance genes in wheat. CanadianJournal ofBotany. 37:1153-1155.30. Schafer. J.F.. and A.P. Roelfs.1985. Estimated relation betweennumbers of urediniospores ofP:;ccinia graminis f.sp. tritici andrates of occurrence of virulence.Phytopathology. 75:749-750.3l. Sunderwirth. S.D .. and A.P.Roelfs. 1980. Greenhouse evaluationof the adult plant resistance of Sr2 towheat stem rust. Phytopathology,70:634-637.

Chapter 3Pathogenicity Analysis of Yellow (Stripe) Rustof Wheat and Its Significance in a GlobalContextR.W. Stubbs. Research Institute for Plant Protection. Wageningen. TheNetherlandsAbstractYellow (stripe) rust of wheat (Puccinia striiformis Westend. f.sp. tritici) isstudied by the Research Institute for Plant Protection (lPO) on aninternational scale. Races (virulences) are identified on seedlings ofa broadset of 'old' and 'new' differential cultivars with some known, but mostlyunknown. resistance genes under controlled conditions. Virulence related torace-specific mature plant resistance is analyzed in race nurseries (separatefield plots). The relationship between the distribution ofpathogen virulencesand host resistances is evident but yet under-studied. Results are presentedofan analysis ofyellow rust infecting triticales and cultivars with resistancederived from rye. The zonal distribution ofyellow rust races in Europe.Africa. Asia. and South America is described. A continuous survey ofchanges in pathogenicity in race populations is highly recommended withregard to breeding for resistance and the evaluation ofhost resistance underdifferent environments.Introduction23Yellow (stripe) rust. caused byPuccinia striiformis Westend.. is oneof the major rust diseases of wheat;it also attacks barley. rye (Secalecereale). and other grasses. As far asis known. the fungus does not attackoats (Avena sativa). rice (Oryzasativa). or maize (Zea mays). Theform infecting wheat is referred to asP. striiformis Lsp. tritici and the oneinfecting barley as P. striiformis Lsp.hordei. Yellow rust has no alternatehost and mutation and somaticrecombination are the mechanismsof variability. It is assumed thatTranscaucasia is the center of originof the fungus. From this center.yellow rust dispersed in alldirections. reaching Australia only in1979 (20). Yellow rust is consideredto be a low-temperature pathogenand is serious in areas in which cool.moist weather prevails. as innorthwestern Europe andmountainous regions of SouthAmerica and East Africa. Theminimum. optimum. and maximumtemperatures for spore germinationare ooe. 9-12°e. and 20-26°e.respectively (24). Yellow rust is alsocharacterized by its systemic spreadin the leaf. Severe leaf and headinfections may cause total losses inyield. Of the three wheat rusts.yellow rust appears to be the mostsensitive to environmental factors.such as air pollution. which reducesgermination of urediospores (25).Resistance of the host is muchinfluenced by temperature and lightwhich. in turn. influence diseaseassessment of the infected plant (26).Since 1956. the Research Institutefor Plant Protection (lPO) inWageningen. The Netherlands. hasbeen engaged in studying thepathogenicity of yellow rust on aninternational scale. The presentcollection of yellow rust specimenscontains 5000 cultures from 60countries.

24Pathogenicity AnalysisMethodology in the seedling stageTechniques of handling thepathogen-The techniques ofhandling yellow rust in theglasshouse in the IPO weredeveloped by Zadoks (41). Thesetechniques, slightly modified, are asfollows. Yellow rust cultures aregrown separately in plastic cages ina glasshouse held at 15 + 2°C.Daylength is kept constant at 16hours by supplementing light withfluorescent tubes giving a lightintensity of 7500 lux. Yellow rust isdifficult to grow in late autumn andwinter, due to light deficiency whichreduces sporulation intensity (18)and to air pollution being relativelyhigh at the time. Spores, which arenormally viable when samples arereceived within 2 weeks of collection,are transferred to seedlings of asusceptible cultivar. Older samplesare placed in moist Petri dishes in arefrigerator to induce resporulationof the fungus. After inoculation, theseedlings are kept for 48 hours in adew chamber held at 9°C and 16hours light of intensity of about 7500lux and 8 hours dark. Urediosporesappearing 10-14 days afterinoculation are collected every 2days, dried, and then stored in glassampules in liquid nitrogen (-196°C).In contrast to leaf and stem rusts,urediospores of yellow rust stored inliquid nitrogen do not need to bereactivated by thawing in a waterbath at a temperature of 40°C. Nodifferences were found ingermination when spores werethawed between 5 and 40°C.The wheat cultivars used for growingP. striiformis f.sp. tritici are MichiganAmber, Taichung 29, or Morocco,depending on the origin of thepathogen. Triticum dicoccum var.tricoccum, being highly receptive toinfection (36) and being compatiblewith P. striiformis f.sp. tritici, P.striiformis f.sp. hordei, and yellowrust on grasses (15), is commonlyused for transferring the rustsamples. However, it should be notedthat this species is resistant to P.striiformis f.sp. hordei race 57,prevalent in the Indian Subcontinent.The generation time (time betweendate of inoculation and date ofsporulation) can be race-dependentas observed by Fuchs (8) whogrouped races into slow, normal, andfast. The difference between the firstand last group is 4 days.Inoculation of large numbers ofcultivars is done by atomizing sporessuspended in mineral oil (Soltrol170) which is somewhat toxic tobarley. For barley, therefore, theurediospores are mixed with sporesof Lycopodium. As the temperaturein the pre-inoculation phaseinfluences resistance expression (4),the plants are grown in the pre- aswell as in the post-inoculation phasein growth chambers under aday/night regime of 18°/15°C and 18hours/6 hours. The light intensity isaround 20,000 lux produced by acombination of high-pressuremercury lamps and high-pressuresodium lamps. Depending on thegeneratioq time of the races, theinfection fypes are observed 14-17days after inoculation and scoring isdone on the 0-9 scale (19).Identification of races(virulences)-The study of thephysiologic races of yellow rust isnot yet far advanced. A limitingfactor is the inadequate informationon resistance genes in the hostwhich are necessary for identificationof virulence genes in the pathogen.Attempts have been, or are being,made to develop isogenic lines ofknown resistance genes.

25Physiologic specialization was firstdemonstrated in 1930 (1). Gassnerand Straib (9) introduced a system ofrace identification and nomenclaturewhich was revised by Fuchs (8), andwas used until a new system wasproposed in 1972 (13). The oldsystem is still in use in the SovietUnion (27, 38) and Iran (2). The setof differential cultivars introduced byGassner and Straib consisted of 11wheat cultivars, six barleys, one rye,and one Triticum dicoccum var.tricoccum. In the nomenclature, nodistinction has been made in thespecialized forms. The reaction typesof 66 races on those differentialshave been summarized by Stubbs(35). Noteworthy is the inclusion ofPetkus rye which contributedresistance to yellow, stem, and leafrusts in chromosome lR (42) tomany wheat cultivars presentlygrown or being developed. Straib(28) found only one race (race 34),giving a susceptible reaction onPetkus rye, but a resistant reactionon all wheat differentials. This racemay have belonged to f.sp. secalis.Petkus rye unused for many years isnow included in our study of thepathogenicity of yellow rust ontriticale.In our routine work on raceidentification, wheat differentials areused, as proposed by Johnson et al.(13). 'Old' differential cultivars suchas Ble Rouge d'cosse and HolzapfelsFriih have been abandoned, which isregrettable because they possessresistance genes that may still bepresent but unrecognizable incultivars presently grown or beingdeveloped. In recent tests, virulenceand avirulence for Ble Rouged 'Ecosse was clearly shown by non­European yellow rust cultures. Thequestion is whether this virulence is'necessary' or 'unnecessary' in therust population. The same applies tothe virulence to the cultivars whichhave supplemented the standard setof differentials. These cultivars are,among others, Mexipak =Kalyansona and Giza 155 (31) andAnza (40), among others.Registration of these virulences isimportant in the evaluation of hostresistance. However, theconsequence of adding newsupplementals is that hundreds ofold yellow rust cultures haveincomplete virulence formulas.At present, much attention is givento the analysis of samples fromtriticale and from wheat cultivarswith resistance derived from rye(IB/IR substitution or translocation).The resistance gene Yr9 is a ryegene and is present in differentialcultivars Riebesel 47/51 (Criewener104/Petkus rye) and Clement(Yr9 + 2?). The resistance of the latterwas overcome in The Netherlands inthe second year of its cultivation(1974). Riebesel47/51 is susceptiblein the seedling stage but so farshows a high degree of mature plantresistance. Table 1 exemplifies theanalysis of samples from wheatswith resistance derived from rye andfrom triticale. Considerablecomplexity is evident. Race 6EO wasfound in Ecuador in 1985, infectingboth wheat with Yr9 and triticale (P.Fox, pers. comm.). Race 6E150severely attacked triticale in Rwanda(E. Torres, 'pers. comm). Race140E12 did the same in Zambia.Race 134E150 was also found inKenya on triticales (D. Danial. pers.comm.) which were much lessinfected than those in Rwanda (E.Torres, pers. comm.). Race 234E171overcame the resistance of Granada(SR/GB//Triticale/Thatcher/Taca//Jubilar)in The Netherlands in 1985, inthe second year of its cultivation.Race 191E206 infected Lovrin 13 inChina (Yang Hua-an, pers. comm.).All these races are virulent for Yr9but react differently on Clement,Granada, and Delphin. The latter

26two may possess resistance genesdiffering from Yr9. In field tests, thetriticale cultivars Mapache andRosner showed resistance to Dutchraces with avirulence for Yr9 andsusceptibility to races virulent forYr9. This indicates the possiblepresence of Yr9 in triticale. Thusmuch useful information can beacquired from maintenance of a"bank" of pathogen samples.Methodology in the matureplant stageThe analysis of pathogenicity ofyellow rust in the post-seedling stageis done by the IPO in race nurseries(41). The nurseries are inoculated inthe second half of April and,depending on the diseasedevelopment, 3-5 observations aremade at intervals of 8-10 days. Theseverity of infection is assessed on a0-100% scale and the infection typeon a 0-9 scale (19). The diseaseprogress in each race-or isolatecultivarcombination is expressed asa compatibility index (41). Thecompatibility index varies from 0 to100 and, from the epidemiologicalpoint of view, a compatibility indexabove 15 is considered to bedangerous (41). The value100-compatibility index is used as aparameter of the so-called 'restresistance' which is assumed to berace-non-specific (32, 41). Thesusceptible checks are MichiganAmber and ItanaJPI 178383 selectionIII (SIll) having one temperaturesensitiveminor gene (26). The latteris used to detect change in infectiontype and, consequently, change inthe disease progress with change oftemperature in the race nurseries.In Table 2, examples are given ofisolates that perform similarly on thestandard set of differential cultivarsin the greenhouse or growth room,but differently on cultivars havingmature plant resistance. Accordingly,the isolates have been named'greenhouse races' and 'field races,"respectively (41). Field races areusually named after the cultivarsfrom which they were obtained. Mostof the cultivars hardly show theirrace-specific mature plant resistancein the seedling stage or, at least. doTable 1. Seedling reactionsaJ of wheat cultivars with Yr9, and triticaleafter inoculation with yellow rust races with virulence to Yr9RaceblVirulenceformulaclOriginClement9+21Wheat Fed.4 TriticaleGranada Kvk.dl Fink'sel Salvo91+1 9 9 1Delphin16EO 6.7.9.A Ecuador R S S S R6EI50 2.6.7.8,9 .A Rwanda R R S S S140E12 3 .6 .9 .A Zambia S R S S S S134EI50 2.6.7.8 ,9.A Kenya S S S S R234EI71 2.3.4.7.8.9.Su.C5 Netherlands S S S S S S175EI42 1.2.3.6,7.9.P Chlnafi S R S S S Ra/ S = susceptible (types 7·9); R = resistant (types 1/6); = no databl Nomenclature after Johnson et aJ. (13)cl According to resistance genes Yr 1.2.3 etc" A = Avocet (39). Su = Suwon 92. C51 Carsten"S" V.dl Federation4/Kavkaz developed by the Plant Breeding Institute. Castle Hill. Australiael Data obtained by D.L. Danial (CIMMYT. Kenya) In the IPOfl Data obtained by Yang Hua·an (Institute for Plant Protection. Beijing) In the IPO

Table 2. Infection spectrum of races of P. striiformis f.sp. tritici on wheat cultivars in the matureplant stageRace32E128Cultivar 32EOai 32EO 32E128 Heine's7 32E128 36E132 lO6E139(genotype) 32EO Albabl Triumph Heine'57 -Heine's4 Leda 36E132 Flamingo lO6E139 LelyCheck 100 100 100 100 100 100 100 100 100 100Heine's VII (2 +?)cl 0 0 0 48 49 50 34 40 50 50H, Kolben (6 + ?)cl 0 0 0 0 3 73 64 0 0Alba (Ab +?) 0 54 7 0 31 8 6 0 400, Triumph (OT +?) 3 63 3 5 3 7 0 0 0Heine's IV (H4+?) 0 0 68 0 43 46 0 39 0 0Leda (2+Ab+H4) 0 3 2 0 33 0 3 0 0Flamingo (2 + 6 + H4) 0 0 0 0 3 10 39 0 0Lely (2 + 7 + Ab) 0 0 0 0 0 0 0 0 5 55at Nomenclature based on seedling reactions of the standard set of differential cultivars (13) bl Trivial name O. cl Cultivar of the standard set of differential cultivars ~

28not do so at the temperature used foridentification of races. A fewcultivars may do so at othertemperatures which indicates thattheir resistance is temperaturesensitiveas well as race-specific (29).It is worthwhile to mention here thecauses of the break-down of theresistance of Lely which had a life of6 years, relatively long as comparedto other cultivars (which often lastonly 1 or 2 years). Lely had twoeffective genes (Yr7 and YrAb) whenfirst cultivated. Then, in France, race106E139 appeared and was virulentto Yr7 but avirulent to YrAb. InFrance, this race overcame theresistance of those cultivars havingYr7 as the only protection againstthe prevailing race 104E137. Theprogeny of race 106E139, the Lelyrace,completed the break-down ofresistance of Lely. The question iswhether the life time of Lely wouldhave been prolonged if those Frenchcultivars had not been grown withinthe epidemiological zone of yellowrust in northwestern Europe.Virulence to Alba has also beenobserved outside Europe and itspresence there is probably due tocultivars possessing the resistancegene(s) of Alba. Cultivars with thesame specific resistance as Alba(such as Mado and Falco) have onecommon ancestor, Juliana (33),which is a progeny of a cross withWilhelmina, also a common ancestorof many cultivars now or formerlygrown outside Europe. Investigatingthe history of resistance is a matterof time as well as of money, butworthwhile to do. In this connection,Gaumann (10) may be quoted: "theforms persisting to the present daymight well be regarded as livingfossils." Yellow rust is one of thesefossils, not only revealing its ownhistory but also that of the resistanceof its hosts.Geo~raphical Distributionof Yellow Rust RacesIn Europe before World War II,studies of the geographicaldistribution of races of P. striiformiswere mostly done by Gassner andStraib in Braunschweig, WestGermany. Starting in 1955, thesestudies were resumed by Fuchs (7)in cooperation with Zadoks (41) whoconducted the Yellow Rust TrialsProject. These trials, now named"trap nurseries," were sown inalmost all European countries and ina few places outside Europe. A rustsurvey on a worldwide basis wasinitiated in 1968 when a resolutionwas taken at the First InternationalCongress in London to conduct aworldwide survey of virulence inplant pathogens. This survey is nowbeing made through CIMMYT'sInternational Disease Trap Nurseries.In 1932 Gassner and Straib (9)stated that the distributions of racesare related to the local host cultivarsand that changes in cultivars will befollowed by changes in racecomposition. Their statementremains valid to this day. Thefollowing description of racedistribution therefore presents amomentary snapshot of a changing,or co-evolutionary process of twobiological entities.The world distributions of yellowrust virulences have been reviewedelsewhere (35) and will now betreated by areas.Europe (Figure 1)Based on data collected in1932-1955, Oort (21) distinguishedthree zones in Europe, one of whichextends from England to Turkey, adistance of about 2500 km. Thelatter zone has been omitted in

29Figure 1 as many races in centraland eastern Europe have originatedin northwestern Europe. An exampleof this is race 104E137 which wasobserved in England in 1969 for thefirst time (5) and became dominantin northwestern Europe in thefollowing 4 years. It graduallydisappeared from the west butmoved to the east and becameprevalent in eastern Europe in thelate 1970s. The same race travelledto Australia in 1979 (20).Northwestern Europe is consideredas a source of new races andvirulences because it is the scene ofintensive breeding for resistance.Almost all resistance genes. eithersingly or variously combined incultivars have been overcome by thepathogen. The evolution of P.striiformis f.sp. tritici innorthwestern Europe has beendescribed by Stubbs (35).The races of southwestern Europe(Spain and Portugal) have beendesignated by Zadoks (41) as theIberian population, based on theirperformance in the trap nurseries. Acharacteristic was the absence of Yr7(Hope/Timstein)-virulence. Thisvirulence is now present as a resultof the immigration of race 6E16 fromnorthwestern Africa. It is prevalentin zone 2 and differs from other6E16 races by being avirulent to YrA5040.......::"'"1 ,.' ! .._"\(/"\~oFigure 1. Distribution of races of yellow rust in Europe.

30(Anza), but virulent to Sonalika alsopossessing YrA. Races in zones 1and 5 are avirulent to both Anza andSonalika. Data on the races of zone 2have been used in the analysis of the1978 epidemic of yellow rust inSpain (17).The rusts of zone 3 have beendesignated by Zadoks (41) as theGrecian population having a similarinfection spectrum to that of theIberian population. Yr7 virulence isnow also present but the data onraces now prevalent in Greece areinsufficient to link zone 3 to zone 4.Rust in zone 4 has been designatedas the Levantine population, differingfrom the previous two populationsby virulence on Hope/Timstein (Yr7)and Selkirk (41). The dominatingrace of zone 4 is 6E16 with virulenceor avirulence to Anza, Sonalika, Giza155 (previously resistant in Egypt).or Miriam (previously resistant inIsrael). Yr10 (Moro)-virulence, beingabsent in zones 1. 2, 3, and 5, isrepresented by race 82E16.YrlO-resistance is indigenous in thiszone (35) and was used in breedingfor resistance in the USA but wassoon overcome by the pathogen.Zone 5, comprising Norway, Sweden,Finland, and the western part of theSoviet Union, has not been describedpreviously because the first sampleswere only received in 1979. Theraces identified, 4EO among others,are similar to those described byShchekotkova (26) and by Tsikaridzeet al. (38) in the Soviet Union. Acharacteristic of these races is thevirulence for Yr6 which is commonin spring wheats such as thosegrown in Finland (Stubbs,unpublished). Zone 5 overlaps zone 7(shown in Figure 3), as both zoneshave the same races, 6E16 amongothers.East Africa (Figure 2)The data of Zadoks (41) showed fewdifferences in pathogenicity betweenthe Levantine and the Kenyanpopulations. A commoncharacteristic was their compatibilitywith Selkirk, being resistant inEurope. The data on races identifiedin Wageningen and Braunschweigalso revealed few striking differences.The two popUlations, however, dodiffer in virulence for Yr9, somethingwhich has not yet been observed inzone 4. Race 134E150, havingYr9-virulence (Table 1), iswidespread in Kenya and is alsopresent in Ethiopia. It may beexpected that race 134E150 willmigrate to zone 4. For Kenya, theevolution of virulence in relation tohost resistance has been studied byBonthuis (3) in Wageningen. Therelationship of Yr9, 7, 6 and/or2-virulence in races with resistancegenes in wheat cultivars selected inKenya has been shown by D.L.Danial (pers. comm.). Yr10-virulencerepresented by race 82E16 is presentin zones 4 and 6 . According to D.L.Danial (pers. comm.) the cultivarKenya Popo possesses Yr10, but theorigin of this gene may be differentfrom the one in the USA cultivar,Moro. The presence of race 6E150only in Rwanda and of race 140E12in Zambia (Table 1), both unknownelsewhere. and both infecting thetriticale Delphin, indicate a separatedivergent evolution of the rust inzone 6.Asia (Figure 3)Zones 4, 5, 7, and 8 overlap becausethey have a few races, (6E16, 38E16,and 70E16, among others) incommon. The west (zone 4) to east(zone 8) movement of virulences, andits causes have been described bySaari and Prescott (22) andNagarajan (16). At present, zone 8 ischaracterized by the widespreadpresence of race 7E150 in Pakistan,

31......("1---' ......... : I ~ ...... , " ,, ,,20Figure 2. Distribution of races of yellow rust in Africa.

32India (S. Nagarajan, pers. comm.),and Nepal. According to IPO's racedata (unpublished), race 7E150appeared in Afghanistan in 1981 andthen moved eastwards. This race,which is new to zone 8, infects thecommercially grown cv. Sonalika.which appears to be moresusceptible at high elevations than atlow elevations, as observed by H.J.Dubin and R.W. Stubbs in Nepal in1986. Race 7E150 seems to occurmore often than other races whichare also virulent on Sonalika. Thefirst race may be more aggressive orvirulent, or may have a widertemperature adaptation than theothers.Cultivars Sonalika, Inia 66, andAnza =WW 15 =Karumu have geneYrA (39), but the last-named cultivarmay possess an additional resistancegene, as suggested by tests withcultures from North Africa. In Iran.the resistance of Inia 66 wasovercome by race B20A2 (2), whichis not identical to race 7E150.Zone 9 is an isolated spot in theIndian Subcontinent with a separateevolution of the pathogen.Data recently collected by Yang Huaan(Institute for Plant Protection,Beijing. China) in the IPO indicatethat yellow rust in China (zone 10)has evolved in some isolation fromthe rest of Asia. The difference isFigure 3. Distribution of races of yellow rust in Asia.

33mainly due to the fact thatindigenous as well as foreignresistance genes have been utilizedin breeding. The indigenous genesare ineffective in China but effectivein other parts of the world as shownin tests with Chinese and non­Chinese yellow rust races (Yang Huaan,pers. comm.).The Americas (Figure 4)According to Humphrey et al. (11),yellow rust entered the Americancontinents by way of the Aleutiansand Alaska. The western mountainranges of the two continentsprovided the route to southern Chile.The Andean-Patagonian valleys arethought to have been the paths toArgentina (39). In South America,the route described was followed byP. striiformis Lsp. hordei race 24when it was introduced to Colombiain 1975 (6). Most of the barleysgrown at that time were susceptibleto race 24, which was ideal forregistering the route and the speedof dispersal of the pathogen.Rajaram and Campos (22) suggestedsix epidemiological zones for wheatrusts in the Western Hemisphere.Yellow rust is important in two ofthese zones, namely, the PacificNorthwest in the U.S. and theAndean countries (Figure 4). Thepathogenicity of the fungus in thefirst zone is being analyzed by theRegional Disease Laboratory inPullman, Washington, and that inthe second zone principally by theIPO in Wageningen. Regrettably,uniformity of procedures with regardto race nomenclature and the use ofdifferential wheat cultivars (34) hasnot been achieved. A few races fromthe U.S. have also been identified bythe IPO but the data are too limitedto indicate whether zone 1 overlapszone 2 (Figure 4). However, it can besaid that virulence on the resistancegenes Yr2, 3, 6, 7 and on thedifferential cv. Suwon 92/0maroccurs in both zones. So far,virulence on Yr9 has been reportedin zone 1 but it is interesting to notethat the U.S. race CDL-21 wascollected from rye and triticales aswell as from wheat (14). The U.S.differential Riebesel 47-51 possessingYr9 is resistant to this race but, likeClement, it does not always indicatethe presence of virulence on Yr9 inthe races (Table 1). In Mexico thisvirulence is represented by race138ElO.In the review ~iven on thedistribution 01virulences in SouthAmerica, the epidemiological zonefor yellow rust has been divided intotwo sub-zones (35) (as shown inFigure 4). Zone 1 includes southernMexico a:ld Guatemala (32), whichoverlaps zone 2 because the sameraces occur (OEO and 8EO amongothers). These races may havetravelled from one zone to the other.but they may also have developedseparately in the two zones. Zones 2and 3 are distinguishable as shownin Table 3. The zones all overlap inPeru. where races of zone 1 as wellas those of zone 2 have beenobserved. This fact suggests a northto south as well as a south to northmovement of urediospores.According to Tollenaar (37),southerly winds predominate incentral Chil~ but are not veryeffective in carrying urediosporesfrom south to north. However, thisview is somewhat contrary to therace data from Chile and Peru.Differences in races between zones 2and 3 may be related to differencesin host resistance genes. In Chile,western European wheat cultivarsare, or have been, grown, forexample, Vilmorin 27 (Yr3) ,Intermedio = Orca (Yr3c + 2 + 6),Manella (Yr2+Ab) and CappelleDesprez (Yr3a + 4a). The latter.possessing a durable resistanceagainst yellow rust (12)' was severely

34oFigure 4. Distribution of races of yellow rust in North and South America.

35attacked by race 104E9. Evidently.the durable resistance of CappelleDesprez is environmentally bounded.as it was adequate in En~land. butinadequate in Sweden (12). Race108E141 was found on Intermedio.The appearance of race 110E143. aprogeny of race 108E141. can berelated to the introduction ofcultivars with Yr7. as Victoria=Pavon 76 (Yr6 + 7) (40). Race236E141 was found in Clement(Yr9+2?). although this cultivar wasnot grown on a commercial scale.Race 134EO. a progeny of race 6EO.appeared in Ecuador and Colombiain 1985. and infected cultivars withYr9 (P. Fox. H.J. Dubin. pers.comm.). As in Chile and in othercountries. virulence for Yr9 wasreadily present in the pathogenpopulation to infect cultivars withthis resistance gene.The presence of race 104E9 in twocontinents and that of race 108E141in three continents are a typicalexample of a parallel evolution ofraces.General Remarksand RecommendationsThe data presented in this paper arebut a fraction of those collected bythe IPO In the past 15 years. Theypresent a situation which wasdescribed in 1972 as follows: "thedistribution of the factors ofvirulence of yellow rust in the worldis directly related to the factors ofresistance of the cultivated hostvarieties. either indigenous orforeign. and that the evolution ofyellow rust develops in the samestages and places as the man-guidedevolution of resistance" (29). Thisstatement is a variant of the oneTable 3. Distribution of races of P. str11formls f.sp. trltlcl in SouthAmerica. Europe, and AustraliaRace(virulenceformula)2EO (7)64EO (Su)40EO (3)6EO (6.7)134EO (6.7.9)104E9(Su.3.4)108E141(Su.2.3.4.6)Country/Continent-- --- ---Colombia Ecuador Peru Chile Argentina Uruguay Brazil Europe Australia+ ++ + + + + + + + + + + + + ++ + + + +l09E141(Su.l.2.3.4.6)110E141(Su.2 .3.4.6.7)236El41(Su.2.3.4.6.9)+ ++ + + ++ +

36presented in 1932. We may expectthat other variants will be stated inthe future but may hope that it willbe possible to add that the evolutionof the rust had been better guidedthan in the past.In breeding for resistance. whethernationally or internationally.knowledge of the stages of evolutionof the pathogen is essential andshould be continuously updated. Thestage of evolution of the rust differsfrom place to place and therecognition of these differences is abasis for an effective evaluation ofhost resistance under differentenvironments. Internationalcooperation is clearly essential.It is relevant to the objectives of theworkshop to enquire whether. in themany wheat collections screened inthe IPO. cultivars have been foundwith resistance that is effectiveagainst the rust. There are a few. butit is uncertain whether theirresistance is determined by a singleunidentified gene or by combinationsof known genes. In the former case.the cultivars could be classified asnew sources of resistance and usedas such in breeding.References1. Allison. C .. and K. Isenbeck.1930. Biologische Specialisierungvon Puccinia glumarum triticiErikss. und Henn.Phytopathologische Zeitschrift.2:87-98.2. Bamdadian. A. 1984. Variationin pathogenicity and evolution ofPuccinia striiformis West. in Iranduring 1980-1984. Proceedings ofthe 6th European and MediterraneanCereal Rusts Conference. 153-156.3. Bonthuis. H. 1985. Survival ofstripe rust (Puccinia striiformis onwheat in the Kenyan highlands andthe consequences for virulence.Mededelingen van de Faculteit voorLandbouwwetenschappen.Rijksuniversiteit Gent. 50/3B.1109-1117.4. Brown. J.F.. and E.L. Sharp.1969. Interactions of minor hostgenes for resistance to Pucciniastriiformis with changingtemperatures. Phytopathology.59:999-1001.5. Doodson. J.K.. and N.H.Chamberlain. 1972. Physiologicraces of Puccinia striiformis onwheat in the U.K.. 1966-1971.Proceedings of the European andMediterranean Cereal RustsConference. Prague. 89-93.6. Dubin. H.J.. and R.W. Stubbs.1986. Epidemic spread of barleystripe rust in South America. PlantDisease. 70: 141-144.7. Fuchs. E. 1956. Der Stand derRassenspezialisierung beim GelbrostPuccinia glumarum (Schm.) Erikss.et Henn. in Europa. Nachrichtenblattdes Deutschen Pflanzenschutzdienstes.8:87-93.8. Fuchs. E. 1965. Untersuchungentiber die physiologischeSpezializierung des Weizengelbrostes(Puccinia striiforrnis West. f.sp. triticiErikss. et tienn.) in den Jahren1959-1964 und tiber dasAnfallingkeitserhalten einigerWeizensorten. Nachrichtenblatt desDeutschen Pflanzensch u tzdienstes.17:161-176.9. Gassner. G.. and W. Straib.1932b. Die Bestimmung derBiologischen Rassen desWeizengelbrostes (Pucciniaglumarum f.sp. tritici (Schm.) Erikss.u. Henn.). Arbeiten aus derBiologischen Reichsanstalt fur Landund Forstwirtschaft. 20: 141-163.10. Gaumann. E.A. 1952. The Fungi.New York. Hafner.

3711. Humphrey. H.B.. C.W.Hungerford. and A.G. Johnson.1924. Stripe rust (Pucciniaglumarum) of cereals and grasses inthe United States. Journal ofAgricultural Research. 29:209-227.12. Johnson. R 1981. Durabledisease resistance. In Strategies forthe Control of Cereal Disease. (J.F.Jenkyn and RT. Plumb. eds.).55-64..0xford. Blackwell.13. Johnson. R. RW. Stubbs. E.Fuchs. and N.H. Chamberlain. 1972.Nomenclature for physiologic racesof Puccinia striiformis infectingwheat. Transactions of the BritishMycological SOCiety. 58:475-480.14. Line. RF. 1980. Pathogenicityand evolution of Puccinia striiformisin the United States. Proceedings ofthe Fifth European MediterraneanCereal Rusts Conference. 93-96.15. Manners. J.G. 1950. Studies onthe physiologic specialization ofyellow rust (puccinia glumarum(Sch.) Erikss & Henn.) in GreatBritain. Annals of Applied Biology.37:187-214.16. Nagarajan. S .. and L.M. Joshi.1985. Epidemiology in the IndianSubcontinent. In The Cereal Rusts(A. P. Roelfs and W.R Bushnell.Eds.). 371-402. Orlando. Florida.Academic Press.17. Nagarajan. S .. J. Kranz. E.E.Saari, G. Seiboldt. RW. Stubbs. andJ .C. Zadoks. 1984. An analYSis ofthe 1978 epidemic of yellow rust onwheat in Andalusia. Spain.Phytopathologische Zeitschrift.91 :159-170.18. McGregor. A.J.. and J.G.Manners. 1985. The effect oftemperature and light intenSity ongrowth and sporulation of Pucciniastriiformis on wheat. PlantPathology. 34:263-271 .19. McNeal. F.H.. C.F. Konzak. E.P.Smith. W.S. Tate. and T.S. Russell.1971. A uniform system forrecording and processing cerealresearch data. United StatesDepartment of Agriculture.Agricultural Research Service.34-121. 1-42.20. O·Brien. L.. J.S. Brown. RM.Young. and T. Pascoe. T. 1980.Occurrence and distribution of wheatstripe rust in Victoria andsusceptibility of commercial wheatcultivars. Australasian PlantPathology. 9: 14.21. Oort. A.J.P. 1955. Verspreidingen verwantschap van physiologischerassen van gele roest (Pucciniaglumarum) van tarwe in Europa.Tijdschrift voor Plantenziekten.61 :202-219.22. Rajaram. S .. and A. Campos.1974. Epidemiology of wheat rusts inthe Western Hemisphere. CIMMYTResearch Bulletin 27. 1-27.23. Saari. E.E.. and J.M. Prescott.1985. World distribution in relationto economic losses. In The CerealRusts (A. P. Roelfs and W .RBushnell. Eds.). 259-298. Orlando.Florida. Academic Press.24. Schroder. J .. and K. Hassebrauk.1964. Unte'rsuchungen tiber dieKeimung der Uredosporen desGelbrostes (Puccinia striiformis)West. Zentrablatt fur Bakteriologie.Parasitenkunde. Infektionskrankheitenund Hygiene. II. 118:622-657.25. Sharp. E .L. 1967. Atmosphericions and germination of urediosporesof Puccinia s triiformis. Science.156: 1359-1360.26. Sharp. E.L.. and RB. Volin.1970. Additive genes in wheatconditioning resistance to stripe rust.Phytopathology. 60:1146-1147.

3827. Shchekotkova. T.V. 1975. Racialcomposition of Puccinia striiformison wheat in the European territoriesof the USSR. Vsesoyuznyi InstitutFitopatologii, BoJ'shie Vyazemy.Publ. Moscow. USSR, Kolos.124-128.28. Straib. W. 1937a.Untersuchungen tiber dasVorkommen physiologischer Rassendes Gelbrostes (Puccinia glumarum)in den Jahren 1935-1936 und tiberdie AgressivWit einiger neuerFormen auf Getreide und Grasern.Arbeiten aus der BiologischenReichsanstalt fur Land undForstwirtschaft. 22.:91-119.29. Stubbs. R.W. 1966. Recentaspects of the physiologicalspecialization of yellow rust in theNetherlands. Proceedings of theThird Cereal Rust Conference. 47-54.30. Stubbs. R.W. 1972. TheInternational Survey of Factors ofVirulence of Puccinia striiformisWestend. Proceedings of theEuropean and Mediterranean CerealRusts Conference. Prague. 283-288.31. Stubbs. R.W. 1973. Significanceof stripe rust in wheat productionand the role of the EuropeanMediterranean cooperative program.Proceedings of the FAG/RockefellerFoundation Wheat Seminar.258-266.32. Stubbs. R.W. 1977. Observationson horizontal resistance to yellowrust (Puccinia striiformis f.sp. tritici).Cereal Rusts Bulletin. 5:27-32.33. Stubbs. R.W. 1978. Gele roest intarwe. een internationaal probleem.Stichting Nederlands GraanCentrum, VoordrachtenGraanziektendag. 1978:41-50.34. Stubbs. R.W. 1980.Environmental resistance of wheat toyellow rust (Puccinia striiformisWestend. f.sp. tritici). Proceedings ofthe Fifth European MediterraneanCereal Rusts Conference. 77-81.35. Stubbs. R.W. 1985. The CerealRusts (II) (A.P. Roelfs and W.R.Bushnell. eds.), Orlando. Florida.Academic Press. 61-101.36. Stubbs. R.W .. and J.M.Plotnikova. 1972. Uredosporegermination and germ tubepenetration of Puccinia striiformis inseedling leaves of resistant andsusceptible wheat varieties.Netherlands Journal of PlantPathology. 78:258-264.37. Tollenaar. H. 1985. Observationson the annual development of somecereal rusts in south central Chile.Agro-Ciencia 1 (2): 171-177.38. Tsikaridze. O.N .. G.L. Tsereteli.Z.L. Tsikaridze. L.U. Akhvlediani,T.l. Gogava. and M.G. Gogava. 1977.Racial composition of pathogens ofwheat rusts in Georgia.Soobshcheniya Akademii NaukGruzinskoi SSR. 88:197-200.39. Vallega. J. 1955. Wheat rustraces in South America.Phytopathology 45:242-246.40. Wellings. C.R. 1984. Geneticbasis of seedling resistance to wheatstripe rust. Annual Report. PlantBreeding Institute, University ofSydney. Sydney. Australia. 34-35.41. Zadoks. J.C. 1961. Yellow ruston wheat studies in epidemiologyand physiologic specialization.Tijdschrift voor Plantenziekten.67:69-256.42. Zeller. F.J. 1973. IB/IR wheat-ryechromosome substitutions andtranslocations. Proceedings of theFourth International Wheat GeneticsSymposium. Columbia.

39Chapter 4Using Polygenic Resistance to Breed for StemRust Resistance in WheatD.R. Knott, Department of Crop Science and Plant Ecology, University ofSaskatchewan, Saskatoon, CanadaAbstractMultigenic resistance to stem rusts has been known for many years. Nonspecificresistance to disease has been hypothesized but it is difficult toprove. Partial resistance and slow rusting are often controlled by severalgenes having small effects and are sometimes thought to be non-specific. Instudies at Saskatoon, lines of wheat were developed that lacked seedlingresistance to race 15B-1 but had good field resistance to the same race.Their resistance proved to be controlled by three to five recessive genes,each having a small effect. The genes reduced the latent period and pustulenumber and size. Resistance that is controlled by several genes having smalleffects is likely to be relatively durable, regardless of whether it is specific ornon-specific. Polygenic resistance is difficult to use in wheat breedingprograms but its use could be very worthwhile.IntroductionA polygenic character is one that iscontrolled by a number of geneseach having a small effect. Just howmany genes should be involved tomake a character polygenic is notclear. For the purpose of this paper, Iwill assume that if a character iscontrolled by several genes and it isdifficult or impossible to identify theeffects of individual genes, then thecharacter is polygenic.The occurrence of polygenicresistance to the rusts has beenknown for many years. In 1946,Ausemus et al. (1) cited a number ofreports of multigenic resistance toleaf rust (P. recondita Lsp. tritici) andstem rust (P. graminis f.sp. tritici).However, in 1971, in an extensivesurvey of genetiC studies on hostparasiteinteractions, Person andSidhu (10) found that 875 papersreported that resistance to variouspathogens was due to major genesand only 50 reported resistance thatwas due to minor genes orpolygenes. Much of the early workwas concentrated on genes that hadmajor effects and usually proved tobe race-specific. The situation haschanged. As a result of the stimulusprovided by Vanderplank (11) in thelast 20 years, there have been manystudies on types of resistance thathave proved to be complex ininheritance-partial resistance, slowdisease development, etc.Non-specific Resistance--;.".-....­ .Vanderplank (11) first hypothesizedthat there are two distinct types ofdisease resistance, vertical andhorizontal, now more commonlycalled specific and non-specific. Hepresented evidence for the existenceof the two types, particularly for thepotato-late blight system (Solanumtuberosum-Phytophthora infestans).Basically, specific resistance iseffective against only certaingenotypes of the pathogen while nonspecificresistance is effective againstall genotypes.Since then, the concept hasundergone various modifications.First, in 1968 Vanderplank (12)concluded that. if data from the

40interaction of a set of host andpathogen genotypes were analyzedby an analysis of variance, thepresence of a significant meansquare for the interaction betweenhost and pathogen genotypesindicated the operation of specificresistance. The presence ofsignificant main effects due todifferences among host genotypesand among pathogen genotypessupposedly indicated the presence ofnon-specific resistance. However, itcan easily be demonstrated thatresistance that results solely fromthe action of genes for specificresistance can generate significantmain effects (Table 1). The data inTable 1 are based on field tests ofnear-isogenic lines of Marquis withfour races of stem rust. Typically, agene gives the same rust severitywith each race to which it isresistant. Although the resistance isentirely due to the action of genesfor specific resistance, there aresizeable mean squares for the twomain effects which result fromdifferences among lines and amongraces.Although the original term,horizontal resistance, derived fromthe fact that resistance was uniformor horizontal against all pathogengenotypes, in 1978 Vanderplank (13)stated that resistance was horizontalif a set of host genotypes gaveconstant rankings when tested withdifferent pathogen genotypes. Just ashost genotypes could differ in theirlevel of horizontal resistance, so alsothe pathogen genotypes could differin aggressiveness. A model for sucha system with four genes forresistance, all having equal andadditive effects, and four genes foraggressiveness, all having equal andadditive effects, gives someinteresting results (Table 2). Thevalues in the table vary dependingon the disease severity assumed forthe combination of the host with nogenes for resistance and thepathogen with no genes foraggressiveness (assumed to be 50%in the example). In Table 2A, themodel shows constant ranking,although six combinations show100% severity and six show O. As aresult of these limitations of theTable 1. Theoretical results from testing Marquis and three near-isogeniclines of Marquis with four races of stem rust (rust severity in percentbased on actual field tests but adjusted so that Marquis is rated as 100%in each case)Cultivar orRaceline 56 15B-1 29-1 11-1 MeanMarquis 100 100 100 100 100.0Marquis-Sr6 10 10 100 10 32.5Marquis-Sr7 100 40 100 40 70.0Marquis-Sr9a 30 100 30 100 65.0Mean 60.0 62.5 82.5 62.5 66.9Analysis of variance (mean square [DFIJ : lines 3056 (3), Races 440 (3). interaction1473 (9)

41Table 2. A model for non-specific resistance in which the five hosts carryfrom 0 to 4 genes for resistance. each reducing disease severity by 25%.and the five pathogens carry 0 to 4 genes for aggressiveness. eachincreasing disease severity by 25% (A). and the genes each affect diseaseseverity by only 10% (B)A.Pathogens and genotypesaJI II III IV V+ + + + + + + + + + Host and genotype* Effect 0 +25 +50 +75 + 100 MeanA - 0 50 75 100 100 100 85B + -25 25 50 75 100 100 70C + + -50 0 25 50 75 100 50D + + + -75 0 0 25 50 75 30E + + + + -100 0 0 0 25 50 15Mean 15 30 50 70 85 50Analysis of variance (mean squares [DF]): Pathogens 4063 (4); Hosts 4063 (4);Interaction 156( 16).B.Host and genotype * Effect 0 +10 +20 +30 +40 MeanA 0 50 60 70 80 90 70B + -10 40 50 60 70 80 60C + + -20 30 40 50 60 70 50D + + + -30 20 30 40 50 60 40E + + + + -40 10 20 30 40 50 30Mean 30 40 50 60 70 50Analysis of variance (mean squares [DF]): Pathogens 1250 (4);Hosts 1250 (4); Interaction 0(16).aJ A + indicates the presence of a gene for either aggressiveness or resistance.respectively. in homozygous condition

42range of severity. there is aninteraction mean square. However. ifthe effect of each gene is reduced to10% so that the range is only from10 to 90% severity. then theinteraction mean square is zero(Table 2B). A host genotype with nogenes for resistance shows aconsiderable range in diseaseseverity. as do all host genotypes.Resistance is no longer horizontal inVanderplank's original sense.Similarly. a pathogen genotype withno genes for aggressiveness shows arange in disease severity. as do allpathogen genotypes. Interestingly.there seems to be no theoreticalreason why a pathogen cannotdevelop increasing levels ofaggressiveness to the point where allhost genotypes are fully susceptible.On the other hand. it might bepossible for the host to developresistance of a type that thepathogen could not overcome.A major problem with the concept ofnon-specific resistance is thedifficulty in proving its occurrence. Aresistance may appear to be nonspecificor effective against all racesuntil a race of the pathogen isdiscovered to which it is susceptible.In other words. resistance is nonspecificuntil it is found to bespecific.Slow Rusting andPartial ResistanceIn 1968 Caldwell (2) drew theattention of wheat breeders togeneral resistance to rusts.emphasizing its durability. Later hewas particularly interested in slowrusting and pattern rusting in leafrust in which the rust was heavyonly on certain areas of the leaf. e.g.the leaf tip.In recent years. many rust workershave emphasized slow rusting orpartial resistance. These types ofresistance involve host-pathogencombinations in which the rustdevelops slowly and never reaches ahigh degree of severity. As a result.damage is slight. The degree ofresistance is often measured as thearea under the disease progresscurve (AUDPC) when the diseaseseventy has been measured severaltimes during the development of theepidemic. There is now ampleevidence that minor genes forresistance can affect rustdevelopment at various stages. forexample: receptivity. length of thelatent period. pustule size. and sporeproduction. Each gene has arelatively small effect but whenseveral of them are combined.satisfactory resistance can result.Parlevliet (7. 8) has done a thoroughanalysis of partial resistance to leafrust (Puccinia hordei) in barley.particularly of the inheritance of thelength of the latent period. He foundthat up to five genes were involved.Their total effect was to increase thelatent penod from 8 to 16 days andthe cumulative effect on a leaf rustepidemic in the field was sufficient toreduce considerably the final rustseventy. Parlevliet (8) also found thatthe effects of at least one of thegenes was specific.Studies on Adult PlantResistqce at SaskatoonSome years ago. I made four crosseseach involving four parents that hadbeen selected because it was thoughtthat they had resistance to stem rustthat was not due to major genes forspecific resistance. The F2 progenyof the crosses were selected forseedling susceptibility to race 15B-1.This should have eliminated anymajor genes for specific resistance torace 15B-1. The progeny of thesusceptible F2 plants were thenselected for several generations foradult plant resistance to the samerace in the field. Lines with good

43resistance to race 15B-1 were easilyobtained (4). Although only the onerace had been used in the selection.the lines proved to be resistant tomulti-race mixtures and resistantalso in tests throughout NorthAmerica. Of 20 lines tested in the1976 International Spring WheatRust Nursery. 17 were susceptible atat least one location while three hadat least some resistance at alllocations. Thus. it appeared thatsome degree of specificity wasinvolved.Woodend (15) tested five of the lineswith race 15B-1 in the field.Compared to a susceptible check.they showed (with one exception)reduced rust severity. pustule size.and areas under the disease progresscurve and reduced apparent infectionrates per day. In tests undergreenhouse. growth chamber. andfield conditions. the lines showedlonger latent periods and reducedpustule sizes compared with thesusceptible check. Resistanceincreased as plants got older.When the resistant lines werecrossed to a susceptible. the F1progeny were in most cases similarto the susceptible check inpercentage rust readings in fieldtests. The F2 populations gavedistributions that were fairly normalwhen the epidemic was moderate.However. if the epidemic was heavy.the distributions were skewed. withthe plants concentrated at thesusceptible end of the curve. Ineither case. few were as resistant asthe resistant parents. The resultssuggested that the resistance wasdue to several recessive genes having'small. cumulative. or perhapsmultiplicative. effects.Recently. Padidam (6) completed agenetic study of seven of the lines.Each line was crossed to asusceptible parent and a single seeddescent procedure was used toproduce a random set of F5 plantsfrom each cross. The seed from theseplants was increased and F5-derivedF7 lines were tested with race 15B-1in field nurseries from 1982 to 1984.The results varied somewhatdepending on the severity of the rustepidemic from year to year.However. the readings on the F7lines showed Wgh correlations fromone year to the next. The results forresistant line 91 will be used as anexample (Figure 1). In 1982. aNumber of Lines ~40------------~1~9~8~2~--------~--Number of Lines30----------~1~9=8~4------------CI)30------------------------~20------------------------~20----------------~.!

44susceptible check showed 81 % rustseverity; line 91 showed 13% andonly 7 of 135 F7 lines wereconsidered to be similar to it. In1984 a susceptible check showed66% rust severity. line 91 showed11 %. and 20 of the 135 lines wereSimilar to line 91. In 1982. for six ofthe seven resistant lines. estimates ofthe number of genes involved inresistance ranged from three to five.The seventh line proved to besegregating for Sr6 as well aspolygenes. The F5-derived F7 lineswere tested for seedling resistance toother races to see if such resistancemight be related to the adult plantresistance in the field. No associationwas found.The results confirm that resistance isdue to several genes. but not a largenumber. The frequent negativeskewness of the distributionssuggests that individual genes tendto have little effect by themselvesbut that their effects aremultiplicative. There is little doubtthat these genes are similar to thosethat have been reported by otherworkers to control resistancedescribed as partial resistance orslow rusting. They can provideadequate field resistance. In fact. thelines tended to give better resistanceagainst multi-race mixtures thanagainst 15B-l alone. Undoubtedly.this was because they carriedadditional genes for specificresistance to some races in themixture.Durability of PolygenicResistanceA key question is whether resistanceof this type will be durable and. if so.why. Resistance can be durable foronly two reasons. First. the pathogencannot develop a highly virulent oraggressive race or. if one isproduced. it is not competitive. Or.second. a highly virulent oraggressive race. for whatever reason.does not come into contact with theresistant host.In North America. under normalcircumstances. there Is little reasonto think that. if a virulent racedevelops for a particular type ofresistance that is in common use. itwill not eventually becomeestablished. Nevertheless. it ispossible that a virulent race maydevelop where a particular resistanceis being used but never becomeestablished in the overwinteringarea.In general. however. it must beassumed that. if a virulent racedevelops. it will become established.If this is so. then durability mustdepend on the inability of thepathogen to develop virulence. It iswell-known that genes for specificrust resistance are rapidly overcomeby the pathogen (although there canbe exceptions). Gene Sr26 derivedfrom Agropyron elongatum has beenused in Australian wheats since1970. It is present in at least sevencultivars and has remained effective.Luig (5) reported that all attempts tofind a susceptible infection in thefield or to produce one by mutationhad failed but he later foundvirulence .In a laboratory culturefrom the GSA. Vanderplank (14)states that. "resistance genes thatthe pathogen cannot match are morelikely to be found in foreign species."1 doubt that resistance genes in therelatives of wheat will somehow bephysiologically different from thosein wheat itself. Resistances derivedfrom relatives of wheat havefrequently been overcome.An important question is whetherresistance controlled by polygenes issomehow different physiologicallyfrom resistance controlled by major

45genes for specific resistance. For thebarley-barley leaf rust system,Clifford et al. (3) argue that there areseparate mechanisms governing thetwo types of resistance. I agree thatthey are genetically separate, but itis not clear that they arephysiologically different. Many genescontrolling specific resistance haveintermediate effects and produceresults similar to slow rusting. Is themechanism different from thatgoverned by polygenes which resultsin slow rusting? Even if themechanisms are different. does thismean that one type can be overcomeby the pathogen and the othercannot? Parlevliet (9) argues thatgenes for partial resistance dooperate on a gene-for-gene basis. Iwould be surprised if they did not.but the question is still open.The final question is whetherselection pressures are different onresistance controlled by major p;enesfor specific resistance compared toresistance controlled by polygenes.ConSider, for example, two cases: acultivar with five genes for specificresistance compared to one with fivepolygenes for resistance. In the firstcase, a pathogen genotype with nogenes for virulence would requirefive mutations to overcome theresistance. The probability of thesimultaneous occurrence of fiveseparate mutations is essentiallyzero. Unfortunately, what usuallyhappens is that the five genes arereleased singly in cultivars over aperiod of time. The presence of anyone single resistance gene exertsstrong selection pressure on thepathogen to overcome it. A virulentmutation can attack the cultivar, theavirulent genotype cannot do so or,at least. can only do so much lesseffectively. Thus, as the genes arereleased singly, the pathogenovercomes them by stepwisemutation.In the second case, a cultivar withfive polygenes, the situation isdifferent. If the genes are speCific, apathogen mutant that overcomes oneof them would undoubtedly increase.As additional mutations occur, theywould also increase slowly infrequency. Eventually, the resistancewould be overcome but the processcould be much slower than forspecific resistance. If. on the otherhand, the resistance were nonspecific,there is no reason to thinkthat it could not eventually beovercome by mutations foraggressiveness in the pathogen. Itreally goes back to the originalquestion, "Are some types ofresistance actually uniform orhorizontal against all races of apathogen?" Certainly there is goodevidence in a few cases of resistancethat have been effective over longperiods.For both stem and yellow rust (P.striiformis) of wheat, there isincreasing evidence that polygeniCresistance is recessive and it maytake several genes to produceappreciable resistance. The geneeffects appear to be multiplicativerather than additive, as follows:First gene-3% reduction; 97%severitySecond g~e-6% reduction; 91 %severityThird gene-12% reduction; 79%severityFourth gene-24% reduction; 55%severityFifth gene-48% reduction; 7%severityIn this model, only combinations offour or five genes would have easilydetectable effects. If a host carried allfive genes. then a pathogen mutantthat overcame one of them wouldhave a sizable effect. increasing rustseverity from 7% to 55%. It would

46have a considerable selectiveadvantage. Additional mutantswould have less selective advantage,but still might appear over time.Some degree of resistance couldremain for a long time.The major difference between thetwo situations that I have discussedis that major genes for specificresistance act independently of eachother but that polygenes actadditively or multiplicatively. Withgenes for specific resistance, avirulent mutant in the rust attacksonly those cultivars that carry thematching gene for resistance and noother unmatched gene. Cultivars areeither resistant or susceptible to themutant. With polygenes, a mutant inthe rust overcomes only part of theresistance of a cultivar regardless ofwhether the resistance is specific ornot. Polygenic resistance is muchmore likely to be durable.Using Polygenic Resistancein BreedingThe durability of polygenicresistance makes it of considerableinterest in wheat breeding,particularly in areas in whichspecific resistance is usually rapidlyovercome by the pathogen. As yet.however, polygenic resistance hasbeen studied more than it has beenused. There is no doubt thatpolygenic resistance is difficult touse in a routine breeding program.First, it is impossible to select for ifgenes for specific resistance to theraces being used are present.Second, the frequency of resistantplants in crosses is low and selectionmust be carried out over severalgenerations and finally on a familybasis. If breeders want to usepolygenic resistance, they must beprepared to put considerable effortinto it.The masking effects of genes forspecific resistance can be overcomein two ways:1) A race can be used that is virulenton all of the genes for specificresistance present in the parentsbeing used in the program (assumingthat such a race is available). Ithought of transferring resistancefrom some of my lines to our mostimportant cultivars such as Manitouand Neepawa. Unfortunately no racevirulent on either Manitou orNeepawa was available.2) The alternate is to test seedlingsof a cross with a highly virulent raceand eliminate all resistant plants.The race can then be used to selectfor field resistance in the progeny ofthe seedling susceptible plants.Because of the expected lowfrequency of resistant plants incrosses involving polygenicresistance, the breeder will have toconcentrate on a few, well plannedcrosses. If the source of polygenicresistance is poorly adapted to thebreeder's area, then probably at leastone backcross will have to be made.Since some evidence suggests thatthree to five genes may be involved,it should be possible to make thecross and the backcross, and thenstart selecting for resistance. Fairlysizable populations will be neededand several generations of selectionwill be required. Since moderatelevels of resistance may beoverwhelmed by heavy spore loads,particularly in single plants, apedigree system involving the testingof families will probably be mostsuccessful.Breeders have probably shied awayfrom using polygenic resistancebecause it would requireconsiderable effort and a substantialchange in breeding procedures.However, the effort could well bevery worthwhile.

47References1. Ausemus. E.R, J.B. Harrington.L.P. Reitz. and W.W. Worzella. 1946.A summary of genetic studies inhexaploid and tetraploid wheats.Journal of the American Society ofAgronomy. 38: 1082-1099.2. Caldwell. RM. 1968. Breedingfor general and/or specific plantdisease resistance. Proceedings of theThird International Wheat GeneticsSymposium. 263-272. Canberra.Australian Academy of Science.3. Clifford, B.C .. T.L.W. Carver. andH. W. Roderick. 1985. Theimplications of general resistance forphysiological investigations. In:Genetic Basis ofBiochemicalMechanisms of Plant Disease. (J.V.Groth and W. R Bushnell. eds),43-84. St. Paul, Minnesota. AmericanPhytopathological Society.4. Knott. D.R 1982. Multigenicinheritance of stem rust resistance inwheat. Crop Science. 22:393-399.5. Luig. N.H. 1983. A Survey ofVirulence Genes in Wheat StemRust. Puccinia graminis f.sp. tritici.Berlin and Hamburg. Verlag PaulParey.6. Padidam. M.R 1985. Genetics ofadult plant resistance to stem rust inseven wheat lines and in their eightparents. Ph. D. thesis. University ofSaskatchewan.8. Parlevliet. J.E. 1978. Furtherevidence of polygenic inheritance ofpartial resistance in barley to leafrust. Puccini hordei. Euphytica,27:369-379.9. Parlevliet. J .E. 1985. Resistanceof the non-race-specific type. In: TheCereal Rusts II (A. P. Roelfs and W.R Bushnell. eds). 501-525. Orlando.Florida, Academic Press.10. Person. C.. and G. Sidhu. 1971.Genetics of host-parasiteinterrelationships. In: MutationBreeding for Disease Resistance.31-38. Vienna, International AtomicEnergy Agency.11. Vanderplank, J.E. 1963. PlantDiseases: Epidemics and Control.New York and London. AcademicPress.12. Vanderplank. J.E. 1968. DiseaseResistance in Plants. New York.Academic Press.13. Vanderplank. J.E. 1978. Geneticand Molecular Basis of Plan tPathogenesis. Berlin and Heidelberg.Springer Verlag.14. Vanderplank. J.E. 1982. Host­Pathogen Interactions in PlantDisease. New York. Academic Press.15. Woodeno. J.J. 1979. Slow stemrustingin wheat. M.Sc. theSis.University of Saskatchewan.7. Parlevliet. J.E. 1976. Partialresistance of barley to leaf rust,Puccinia hordei. III. The inheritanceof the host plant effect on latentperiod in four cultivars. Euphytica25:241-248.

48Chapter 5Strategies for the Utilization of PartialResistance for the Control of Cereal RustsJ.E. Parlevliet. Department of Plant Breeding. Agricultural University.Wageningen. The NetherlandsAbstractIn cereals all resistance to cereal rusts is of the species-specific type. i.e. theresistance is effective to one rust species only. Against each rust pathogentwo types of species-specific resistance can be recognized: i) A major gene,hypersensitive type ofresistance, characterized by low infection types, racespecificityand lack of durability; ii) a quantitative type ofresistance (partialresistance), characterized by a reduced rate of epidemic build-up despite ahigh, susceptible infection type. by absence of large race-specific effects(although small ones do occur) and by durability. In the absence ofmajorgenes. selection for partial resistance is easy. Even a mild selection againstsusceptibility, ifapplied consistently. is highly effective in accumulatinggenes for partial resistance. This mild selection enables the breeder to selectfor other characteristics at the same time. If one wishes to increase partialresistance in the presence ofmajor genes that have not been fullyneutralized by the pathogen, the efficiency of selection is considerably less. Ifpossible, the breeder should expose the host population to a single race ofthe pathotype, a race that neutralizes a maximum number ofmajor genes.In this host population. the breeder should remove the most susceptiblegenotypes at each stage of selection and also those genotypes that show alow infection type. Ifit is too difficult to score infection types reliably, thebreeder should remove the most resistant genotypes together with the mostsusceptible ones as the former are assumed to carry major genes. In somesituations, the pathogen population to which the host is exposed cannot becontrolled and exists as a mixture of races. Selection for partial resistance isvery difficult in this case. Continuous removal of the most susceptible linestogether with those lines that are nearly unaffected will tend to favor partialresistance, but the progress may be slower than hoped for.IntroductionWheat. barley, oats. and rye arehosts for several different rustspecies. These rusts collectivelyrepresent the biggest disease threatfor these crops worldwide.The most important rusts are thestem rusts of wheat. oats, rye. andbarley (Puccinia graminis f.sp. tritici,avenae, and secalis and P. graminis).leaf (brown) rust of wheat (P.triticina = P. recondita fsp. tTitici).rye (P. recondita = P. recondita fsp.secalis). and barley (P. hordei),yellow (stppe) rust of wheat andbarley (P. striiformis fsp. tritici andhordei), and crown rust of oats (P.coronata).Resistance to these rusts has beenwidely used and is often of the majorgene type. Characteristic is the largenumber of major genes that havebeen found-over 40 Sr-genes (wheatstem rust). over 30 Lr-genes (wheatleaf rust), over 40 Pc-genes (oatscrown rust)-and the numerousraces of each pathogen. With fewexceptions these major, race-specific

49resistance genes are not durable;when exposed over large areas forlong periods. races develop thatneutralize the effect of the resistancegenes. In order to obtain longerlasting resistance. two approachesare possible. Either another type ofresistance is looked for and used orthe non-durable resistance genes areused in ways that reduce the threatsof new races. This chapterconcentrates upon the former. i.e..on partial resistance.TerminologyThe terminology of host-pathogensystems is far from consistent. Inorder to be clear. the most importantterms and concepts used in thispaper are discussed and defined.The cereals are hosts to several rustpathogens. These rusts are able toinvade the host plant. The tissuesinvaded can be indicated as thetissues affected. The affected areascan be recognized as sporulating anddiscolored areas. High growth andreproduction rates of the pathogen,measured as rapid expansion of thehost tissue area affected, areindicative of a high aggressiveness ofthe pathogen. or of a highsusceptibility and thus lowresistance of the host or of both.With an increase in resistance or adecrease in aggressiveness, the rateof tissue area expansion diminishes.With immunity, complete resistanceor with non-aggressiveness. the rateof fungal growth and/or reproductionis zero. Incomplete resistance (withrusts often designated slow rusting)or reduced aggressiveness allowssome growth and reproduction of thepathogen. Partial resistance is a formof incomplete resistancecharacterized by a susceptible orhigh infection type. Despite thissusceptible infection type, the tissuearea affected remains less than thatof a very susceptible genotype; theepidemic is retarded. The infectiontype describes a resistance reactionto the rust pathogen on a 0-9 scale.where 0 = no sporulation. tinynecrotic flecks and 9 = largeurediosori. abundant sporulation.and no host tissue reaction aroundthe sori. Low infection types indicatethe presence of one or more genesfor hypersensitivity. Nearly all majorgene resistances belong to thiscategory of hypersensitive or lowinfection type resistance.Resistances can be effective against abroad range of parasites, forexample. tannins in various crops orglucosinolates in Cruciferae. This isbroad resistance. If the resistance iseffective to one parasite species only.it is species-specific resistance (17).All known forms of resistance incereals to the rusts have to beclassified thus. Parasites in turn mayhave a parasitic ability directed to awide range of host species.generalists. or to a narrow range ofhost species. specialists. The cerealrusts are specialists.Resistance is often sub-divided intorace-specific and race-non-specificresistance. Race-specific resistance isresistance that is effective only tocertain rac~s of the pathogen. Thereare host cuhivar-pathogen genotypeinteractions and the ranking order ofthe host cultivars for resistancedepends on the race of the pathogenused. Races of the pathogen at thesame time vary in their cultivarspecificaggressiveness (= virulence).Race-nan-specific resistance operatesagainst all genotypes of thepathogen.

50The resistance level may varydepending on the cultivar-nonspecificaggressiveness· of thepathogen, but there are nointeractions between host andpathogen genotypes and the rankingorder of the cultivars for resistance isindependent of the pathogengenotype used. Race-nonspecificresistance is almost invariably usedwithin the context of resistances thatare of the species-specific types as inthe case of the cereals/cereal rusts.Whether it is rightly used in thisgroup of host pathogen systems isdiscussed later (see section onSpecificity below).Partial ResistanceAs defined above partial resistance ischaracterized by a reduced rate ofepidemic development despite a highor susceptible infection type. It istherefore not identical with slowrusting, as all incomplete resistanceto rusts results in slow rustingincluding resistances withintermediate infection types.The reduced rate of epidemicdevelopment is a result of thecombined effects of reduced infectionfrequencies, longer latent periods,and reduced rates of sporeproduction per urediosorus (16, 20).• In the literature. Virulence.aggressiveness. and pathogenicity areused inconsistently. If virulence is thecultlvar-speciflc and aggressivenessthe cultlvar-non-specific counterpart ofrace-specific and race-non-speclficresistance. one needs a counterpart ofresistance. Pathogenicity is notsuitable as it means more thanaggressiveness. Therefore the termsused here are proposed to replace thepresent usage of virulence andaggressiveness. Virulence could beused to indicate symptom-Inducingability of pathogens such as viruses.which incite true diseases.Measuring partial resistancePartial resistance must be evaluatedin the field. Basically. the proportionof host tissue affected is measured,either once near the end of epidemicdevelopment or several times duringthe development of the epidemic.The former is assumed to representthe cumulative result of thecomponents of partial resistance overtime (26). The latter makes itpossible to calculate the area underthe disease progress curve (AUDPC)(36) or the apparent infection rate. r(34). This r-value approach. however.is clearly inferior to the other twomethods mentioned (20, 29, 30).Assessment of partial resistance isnormally done on cultivars grown insmall plots adjacent to one another,whereby cultivars differing inearliness and in susceptibility areoften exposed to unusually highlevels of inoculum. This situation,quite different from that of thefarmer, may result in whatVanderplank (34) called"representational errors." Theseerrors may result in under- oroverestimating the partial resistanceor even ranking the cultivarswrongly.Inoculum pressure-The use ofsmall plots exposed to spreader rowsof a very s1,lsceptible cultivar tendsto reduce apparent differences inpartial resistance considerably. It is aform of interplot interference.Interplot interference-In the caseof windborne pathogens such as thecereal rusts, differences in partialresistance can be considerablyreduced. Highly susceptible cultivarsproduce far more spores than thepartially resistant ones in the trialand many of these are exported toadjacent plots. In general, thepartially resistant cultivars receivemore spores from the surroundingplots than they export. Partial

51resistance is best-measured in thecentral parts of not-too-small plotsthat are sufficiently separated fromothers to reduce the import ofinoculum to insignificant levelsrelative to their own sporeproduction. The partial resistancemeasured in small adjacent plots canbe greatly underestimated. Forbarley leaf rust. the underestimationvaried from 5 to 16 times if theadjacent plots were 4.5 m wide. from14 to 30 times in case of 1.5-m plotsand from 75 to 130 times for singlerowplots (27). Partial resistance inwheat to leaf rust in small adjacentplots appeared to be underestimatedto a similar extent (Broers. pers.comm.). In barley. partial resistanceis strongly underevaluated in smalladjacent plots but the rank orderremains constant (26). Measurementsfrom adjacent plots can easily betranslated into representative valuesif one includes cultivars thatrepresent a wide range of partialresistance (27).Earliness-If the genotypes to becompared vary greatly in maturityand time of heading. partialresistance may be difficult toevaluate. If observed on the sameday. early genotypes tend to beunderestimated and late genotypesoverestimated. If one observes theamount of rust present at the samedevelopmental stage. the partialresistance of the late cultivars isunderestimated because of infectionfrom the early genotypes. Thisproblem can be solved by plantinggenotypes of similar earlinesstogether in the same block andevaluating the blocks at differenttimes. Also. one should not assessthe flag leaf only. but at least thethree upper leaves.Time of evaluation-The best timeto assess infection is when the mostsusceptible genotypes in the trial arenot yet fully affected. Laterassessments reduce the differencesbetween genotypes while earlierevaluations tend to be morelaborious or less accurate. If theepidemic has not developed verywell. one should assess it when atleast three leaf layers are still green.Presence of major genes-If thegenotypes to be assessed containmajor genes and the rust populationis a mixture of races to which thesemajor genes are only partiallyeffective. it may be very difficult todiscern partial resistance (Table 1)(19). Where possible one shouldavoid using race-mixtures. Oneshould use a single race. the onewith the highest number of cultivarspecificaggreSSiveness factors.Components of partial resistanceFor detailed reviews the readershould see references 16 and 20. Thebasic components are infectionfrequency. latent period. and sporeproduction per urediosorus. Of thelast component. the spores producedearly in the life of the pustule areespecially important for thedevelopment of the epidemic. Theimportance of the components mayvary with the rust species. Latentperiod is the most importantcomponent in barley and wheat leafrusts. pathogens with little systemicactivity. In yellow rust. partiallysystemic within the leaf it infects.infection frequency and sporeproduction may be the mostimportant components. Thecomponents tend to vary in anassociated way. Partially resistantcultivars tend to have reducedinfection frequencies. longer latentperiods. and reduced sporulationrates compared with moresusceptible cultivars.Generally it is assumed (though notproved) that these components arecontrolled by different genes. In

52barley. however. the partialresistance to barley leaf rust seemsto be largely controlled by minorgenes with pleiotropic effects oninfection frequency. latent period.and spore production (1. 21). As towheat leaf rust. the components areless strongly associated with oneanother and with partial resistance(Jacobs. Broers. pers. comm.).GeneticsThe genetics of slow-rusting incereals were reviewed by Wilcoxson(35). In many of the studiesreviewed. slow rusting and partialresistance were taken to be identical.Often. progenies of crosses betweenhighly susceptible cultivars andpartially resistant ones wereinvestigated. The segregationpatterns were quantitative in natureand transgression was oftenobserved. The number of genesassumed to be involved varied froma feW-in maize/Po sorghi (7), inwheat/Po triticina (8) and in oats/P.coronata (10)-to several-inwheat/Po triticina (5) and P. graminisf.sp. tritici (33) and in barley/P.hordei (14. 15.23).In wheat. temperature-sensitiveminor genes against P. striiformishave been reported. They actadditively and together give a lowinfection type. They appear to havedurable effects (31).SpecificityPartial resistances to the cerealrusts. even when they are typicallypolygenically inherited are speciesspecific(17); 1. e. the partialresistance genes are effective to onlyone Puccinia species. Partialresistance is not only species-specificbecause race-specific effects havealso been reported for severalpathogens. including barley leaf rust(20). These effects tend to be small.or at least insufficiently large to be ofuse in identifying races. It isquestionable whether true race-nonspecificresistance occurs withinspecies-specific resistance.DurabilityPartial resistance is considered to bedurable. but pertinent information ishard to obtain. The wheats Thatcherand Lee have been known to rustslowly for 55 and 30 yearsTable 1. Percentage of host tissue affected if cereal cultivars carryingdifferent race-specific resistance (R) genes were exposed to a mixture ofrust races. where the races vary in cultivar-specific aggressiveness(= virulence) genes (a-genes)a-genes of rust racesala2 ala3 ala4 a2a3a4 Percentage of hostR-genes (30) (40) (25) (5) tissue affected+ bl + + + 70Rl + + 50R3 + +R2R4 +4020aJ Percentage of each race in the initial inoculum in parenthesesbl A + sign indicates that the host cultivar is susceptible for that race, a - sign that itis not

53respectively (35). Red Rustproof. anoat, has been reported to be slowrusting to crown rust for over 100years (9), while the wheat Knox hasbeen partially resistant to wheat leafrust for over 20 years (12).Most western European barleycultivars carry low to fair levels ofpartial resistance to barley leaf rust.There are no indications that thispartial resistance, although exposedon large areas over many years, hasdiminished to any significant extent(6, 18, 20). The wheat cultivarsgrown in The Netherlands before1930 all showed durable (mostlypartial) resistance to yellow rust (4).Data of this type are also availablefrom a wide range of other hostpathogensystems. All theseobservations together suggest thatpartial resistance, provided it is notclearly monogenically inherited, is ofa durable nature.Selection for partial resistanceIf the aim is to increase partialresistance, various approaches arepossible depending on the situation.The most important restrictive factoris the presence of effective resistanceof the major gene, low-infection type.Also important is the degree ofassociation between the componentsof partial resistance. If thisassociation is high and of apleiotropic nature (as in barley toleaf rust) selection for one of thecomponents of partial resistance canbe as effective as field selection. If.however. the association betweencomponents is low then it is notadvisable to use selection for a singlecomponent. On the other hand, onecould try to improve the plantmaterial for each of the componentsseparately. which would increase thepartial resistance strongly.A third aspect is the expression ofpartial resistance in differentenvironments. The reaction of barleyto its leaf rust appears to be ratherindependent of environment.Resistance is expressed under a widerange of temperature conditions andthe ranking order of the cultivars forpartial resistance does not vary overyears or test locations (13. 25, 26).This stable expression of partialresistance over a wide range ofenvironments, however convenientfor the breeder, is not necessarilycharacteristic of all cereal-rustpathosystems. In wheat leaf rust, forinstance, the important componentof partial resistance, latent period,appeared highly temperaturesensitive.The long latent periods ofmost partially resistant cultivarswere best expressed at lowtemperatures and hardly at all athigh temperatures (Broers, pers.comm.). However, when tested intotally different environments (TheNetherlands, two soil types; southernBrazil; and Mexico), the partiallyresistant cultivars ranked in thesame order and the partial resistancewas well expressed in all fourenvironments (Broers, pers. comm.).Partial resistance in the absenceof major genesAbsence of effective major genes incommercial cultivars is not common.Within the small-grain cereals,barley with respect to barley leafrust is the only clear-cut example(25). Maize-Puccinia sorghi andpeanut-Po arachidis are otherexamples.How to proceed in such a case canbe shown by reference to the barleyleafrust system. All cultivars grownin western Europe appear to carrysome partial resistance (25). Evensome such as Akka, which areconsidered to be extremely

54susceptible. carry a little resistance.Most cultivars. however. carryconsiderably more and it is this thatprevents the rust from becoming amajor pathogen in western Europe.Parlevliet et al. (25) showed that onecan select for this resistance in anystage of a breeding program, and atany stage of the barley plant. Theyselected with success in the seedlingstage in the greenhouse by takingseedlings which had slightly longerlatent periods and reduced infectionfrequencies. The latent periods andinfection frequencies were notmeasured; from each box ofseedlings, the 10% with relativelylong latent periods and the fewesturedia were selected. Selectionamong genetically diverse genotypesin the field was successful wheneach genotype was represented by asingle plant and even moresuccessful if the genotypes wererepresented by small plots. In bothcases, the selection consisted oftaking the 5% or 25% of units(single plants or plots) with thelowest level of leaf area affected.Either selection intensity gaveexcellent selection response.In the field. the plants or plots to beselected received their inoculumfrom spreader rows. These consistedof plants of a highly susceptiblecultivar and infection was initiatedseveral weeks before the seasonalepidemic was expected to begin. Thesingle plants were well spaced andtwo rows were separated by onespreader row. The spreader rows inthe case of small plots ranperpendicularly to the plots on bothsides. The inoculum used in allexperiments consisted of a singlerace.This experiment clearly indicatedthat both selection in thegreenhouse, based on components.and selection in the field, based onlevels of leaf rust, were effective inraising the level of partial resistance.The following two experimentsconfirmed this.Selection for one component inthe greenhouse-Partial resistancein barley is strongly correlated withlatent period in the adult plant stage(25. 26). However. the Europeancultivars seem to carry much thesame minor genes for latent period(15). To increase partial resistance,other minor genes must be added tothose of the European barleys. It wasobserved that the fairly primitivecultivar Cebada Capa carried, behindthe hypersensitivity gene Pa7, a highlevel of polygenic partial resistance.Cebada Capa was therefore crossedto Vada and the F2 seedlingssegregated in a 3: 1 ratio for the Pa7gene. All plants carrying Pa7.characterized by a low infectiontype, were removed and the 25% ofthe plants with a susceptibleinfection type were grown on. Theadult plants were re-inoculated. Theyshowed a relative latent periodranging from about 150 to over 260(L94 = 100, Vada = 185). The plantswith the longest latent period werekept and F3 lines evaluated the nextyear. From the selected F3 lines,plants with the longest latent periodwere again kept and so on. In thisway F6 lin~s were obtained with arelative latent period in the adultplant stage approaching 300 (23).These lines were evaluated in thefield and the results were beyondexpectation. The partial resistance ofthe selected lines (Table 2) was 100times higher than that of Vada andsome 5000 times higher than that ofAkka. The prevailing conditions wereextremely favorable for the rust;there was still some interplotinterference and the epidemics in theindividual plots had started early.Under normal western Europeanfarming conditions. these lines wouldremain virtually free of rust.

55This procedure was very effectivebut it placed no pressure onagronomic characters and istherefore suitable only for selectingparental material or for improvingbase populations from which furtherbreeding will be done.Selection in the field-Asmentioned above, it is not sufficientto show that one can increase partialresistance. Agronomic charactersmust be, at least, maintained. Tostudy this, an experiment was set upto see whether partial resistancecould be improved while selecting foragronomic traits as well. Twogenetically variable populations,quite different from each other, weretaken as the starting point. Onepopulation (A) was produced byintercrossing eight two-rowedEuropean spring barley cultivarswith each other for the threeconsecutive generations torecombine their genes thoroughly.The eight cultivars carried no knownmajor genes for rust and varied inpartial resistance from hardly any(Mamie, as susceptible as Akka) togood (Vada). The other population(B) was taken from Composite Cross(CC) XXI (32). This extremelyvariable population had beenmultiplied in isolation for more than10 years. This resulted in apopulation consisting of a mixture ofwidely different, fairly homozygouslines, most of which were four- tosix-rowed. For partial resistance torust. this population was also quitevariable (11). The Idea was toaccumulate resistance genes, whileselecting for agronomic traits as wellin both populations, using arecurrent selection procedure. Theselected lines of each populationwould then be crossed to try to getrenewed response to selection.The selection practiced consisted ofdiscarding in each cycleapproximately the 30% mostsusceptible plants or lines. Amongthose remaining, selection fordesirable agronomic traits (grainyield, 1000-grain weight, lodgingresistance, earliness, and partialresistance to powdery mildew) wascarried out. The selection for partialresistance to the two fungi wastherefore very weak.Table 2. Leaf area affected (percent) and number of uredia per tiller ofthree barley cultivars and three lines selected for high partial resistanceto barley leaf rust about 6 weeks after the start of the epidemic. Therewere plots of 1.0 m 2 , separated by 4 m of rye. The latent periods givenare relative to those of the very susceptible L94 (after 24)Cultivar Leaf area affected No. of uredia Relative latent periodor line on 4/7/1983 per tiller (L94 = 100%)Akka 40 5000 113Sultan 10 1200 137Vada 1.0 100 18526-6-11 0.011 1.1 29117-5-9 0.009 0.9 28117-5-16 0.004 0.4 281

56The experiment started in 1976 with5000 plants of each population. Afterremoving approximately 30% of theplants most affected by rust and the30% most affected by mildew. theremainder was selected for the otheragronomic traits. From eachpopulation. 400 plants were thusselected. The selection applied in thesubsequent generations wasessentially the same. The 800progenies of these plants (SI) weregrown and evaluated in the nextyear. together with various controls.Of each population the 12 best F3lines (S2) were selected and crossedin as many combinations as possiblewithin each population. In 1979.again. 5000 plants of eachpopulation were evaluated. as in1976. In 1980. the 400 best F3 linesof each population (S3) wereevaluated as in 1977. together withsuitable controls. The field appearedvery heterogeneous and thereforeonly very weak selection waspracticed. About one third of thelines were retained and retested asF 4 lines in 1980. The 12 best F 4lines of each population (S5) wereselected and used to recombine thegenes of both populations. The 12lines of population A were crossed inas many different combinations withthe 12 lines of population B aspossible. Of the resulting F2. 10.000plants were grown in 1983 andselection carried out as in 1976 and1979. In this population. two-rowedand four- to six-rowed plantsoccurred. Selection was carried outindependently within these twophenotypes. As in the earlier cycles.the 400 best plants were selectedand their F3 lines grown andevaluated in 1984. From the F3lines. the best 20 within eachphenotype were selected and thisend product of selection (S7) wascompared with a series of checksderived from the various stages ofthe recurrent selection procedure.This evaluation was done at twosites near Wageningen. on a sandysoil and a clay soil.Results are summarized in Table 3.The data are averaged overreplications and sites. The selectionfor resistance resulted in an almostidentical response in bothpopulations. The starting level wasthe same. a level slightly below thatof the fairly susceptible cultivarSultan. At the end of the secondcycle (S5), the amount of rust wasgreatly reduced. to less than 10%.After the S5. selected lines wereintercrossed and from thispopulation S7 lines were selectedwith partial resistance significantlybetter then that of Vada. Due to thetesting situation (fairly smalladjacent plots). there was aconsiderable interplot interference.The 60-fold increase in partialresistance in this experiment (Table3) from the SO to the best linesunderestimates the real progress.Corrected for interplot interferencethe real progress is estimated atabout 900-fold. This is comparablewith a gain from the level of a verysusceptible cultivar to the level of acultivar sufficiently resistant toprevent any significant yield damagein western Europe. even in yearsconducive,to rust (28).Very interesting was the ease withwhich lines could be obtained thatwere considerably more resistantthan Vada. which represents thehighest level of partial resistanceamong commercial cultivars. Onlymild selection against susceptibilitywas needed and the remarkable gainin resistance was accompanied by again in yield as well (Table 3).Partial resistance in the presenceof major genesIn cereal-rust systems. major genesare frequent and we have alreadyseen that their presence seriously

57confounds the identification and useof partial resistance. As a rule. onecan state that. if major and minorgenes are both present. to selectstrongly will tend especially to selectthe major genes. while mild selectionfavors both types (18). If breederswish to select exclusively for theminor genes. they should ascertainthat they first remove the majorones. This implies the need todistinguish clearly between the tworesistances. though this is.unfortunately. not always easy.In the cereal rust pathosystems themajor gene resistance is of thehypersensitive type. characterized bylow infection type. If such genescaused only very low infection types(say 0 to 3) and partial resistanceonly very high infection types (say 7to 9). it would be easy to classify theobserved resistances. However. thedistinction is rarely so clear. Moreoften there is a more or lesscontinuous distribution of infectiontypes and it is then difficult toseparate one type of resistance fromthe other. This continuousdistribution is caused by variousfactors. as follows:• Hypersensitive resistance genesmay have, even under optimalconditions. different infectiontypes. Major genes withintermediate infection types arenot rare.• The genetic background of amajor gene may affect its infectiontype.Table 3. Numbers of barley leaf rust uredia per tiller and relative grainyields of two barley populations, A and B, at three stages of selection.Two barley cultivars that went into population A are included asstandards (after 28)UrediaYield aJGeneration A B Mean A B MeanSO, unselected 5200 5200 5200 114 87 100S5, after 2-3b/ 350 500 425 139 107 123S7, after 3-2c/ 275 250 260 127 126 126

58• Gene dosage can be of importance.It is often reported that majorgenes are inherited as dominants.If, however, the infection types ofhomozygotes (double-gene dosage)and heterozygotes (single-genedosage) are compared, there isoften a difference, indicatingincomplete dominance. Thus, Pa7in an heterozygous condition giveson average a higher infection type(range 2 to 6) than in anhomozygous condition (range I to4) (22).• The development of the plant is ofimportance as well. The infectiontype may become lower at moreadvanced development stages.• The race may affect the infectiontype.• The environmental conditions areof great importance in theexpression of these major genes.Browder (2) says that, for theexpression of the active event,incompatibility resulting in lowinfection type, a specificenvironment is required. In otherenvironments, the activity (and sothe expression) is less. Examplesof resistance genes giving lowinfection type at a giventemperature and a higher one atother temperatures are numerous.• The moment of assessment canaffect the infection typeassessment. After infection, ittakes some time for the fullexpression of the infection type.With a high infection type, theoptimal moment to assess is oftena few days after the lesions startto sporulate. If one waits, theinfection type tends to becomelower probably because ofexhaustion of the tissue directlysurrounding the lesion. So, as arule, one should not wait too long.• With non-hypersensitiveresistance, the infection type isnot always uniform. Actually, thegreater the level of partialresistance the lower the infectiontype tends to become. Inseedlings, the partial resistancetends to be only weakly expressed,so the infection type remains high.But. in adult plants, where thepartial resistance is fullyexpressed, the infection type maybe somewhat reduced. So, withhigh levels of partial resistance,the infection type can be in thesame range as the infection typeof some major genes.Clearly it is impossible todiscriminate unambiguously betweenmajor gene and partial resistance onthe basis of infection type. But onecan discard at least some of themajor genes by assessing theinfection type. How efficient this isdepends on the host-pathogensystem. In the barley leaf rustsystem, most major genes can berecognized in the seedling stage. If agenotype has an infection type lowerthan 8, one should assume thepresence of a major gene. In adultplants it is less easy to discriminateas the infection types may overlap.However, in case of an intermediateinfection type with a very long latentperiod, oqe may think of partialresistance, while an intermediateinfection with a moderate latentperiod may suggest a major gene.In other crop-pathogen systems, it isless easy. Yellow rust in wheat, forinstance, shows a fully continuousspectrum of infection type that iswell correlated with a resistancemeasured as spore production (3). Inthis pathosystem it is very difficultto use infection types in order toidentify partial resistance.

59How should one proceed if majorgene and partial resistance aredifficult to discriminate? A seedlingscreening with a single race, whichcan neutralize as many major genesas possible, is the best approach.Very susceptible and moderatelysusceptible seedlings serve asstandards. Seedlings with low orintermediate infection types can beremoved and, among the ones with ahigh infection type, those which areless affected or carry less and fewerlesions that are slightly smallercould be selected. Major genes thatare expressed only in the adult plantstage, such as the wheat Sr2 andwheat Lr12. Lr13, and Lr22 cannotbe removed in this type of test.In many cases, the screening is donesolely in the field, in which case onlygeneral guidance is possible. Muchdepends on the host, the pathogenand the local conditions. If thebreeder thinks he can obtainreasonable infection typeassessments, the adVice is to selectthose plants or lines that have aninfection type as high as possiblewith an amount of tissue affected aslow as possible. Plants or lines thatare virtually clean of the pathogenshould not normally be consideredas carrying an extremely high levelof partial resistance, but of probablycarrying an effective major gene.Table 4 tries to show this. The lines2, 5, 18, and 20 are assumed tocarry a major gene. Those withinfection types of 7 or more and arelatively low leaf area affected areconsidered to carry partial resistance(lines 3, 8, 17, and 22).If no infection type data are collectedat all, the possibility of selectingpartial resistance declines. To givepartial resistance in such a case afair chance, one should select mildlyTable 4. Infection types (IT) and percentage leaf area affected (LAA) of aseries of wheat lines and two control cultivars. a highly susceptible one(CI) and a moderately susceptible one (C2). in plots adjacent to oneanother and exposed to a single race of yellow rustLine IT LAA Line IT LAA Line IT LAA7 60 C2 8 35 16 9 752 2 (O)a! 9 6 15 C2 8 253 .. 25 10 9 50 17 7 154 6 5 11 7 55 18 1 (O)a!C2 9 60 12 7 30 19 8 605 3 5 Cl 9 70 20 4 56 7 30 13 9 80 Cl 9 507 9 70 14 5 10 21 7 408 8 25 15 8 50 22 8 20at (0)trace

60against susceptibility (the lines 1. 7,lO. 11. 13. 15, 16. and 19), but alsoagainst too strong levels of resistance(the lines 2, 4, 5, 14, 18, and 20). Inthis way, at least some of the majorgenes are removed. while someaccumulation of partial resistance isensured.In short, the previous discussionshows that it is difficult to select forpartial resistance in the presence ofmajor gene resistance but not, ingeneral, impossible. The experimentsummarized in Table 3 illustratesthe matter. The two barleypopulations were selected forresistance to two pathogens. rustand mildew. For mildew. populationB carried various major genes thatwere partly effective, while thenatural mildew population to whichthese populations were exposedconsisted of mixtures of races thatvaried in composition over the years.Although the same selectionprocedure was practiced as againstleaf rust, progress in resistance tomildew was small compared withprogress in resistance to leaf rust.Progress in the former was probablydue to both major gene and partialresistance while progress in thelatter was solely due to partialresistance.References1. Arntzen. F.K., and J.E. Parlevliet1986. Development of barley leafrust, Puccinia hordei, infections inbarley. II. Importance of early eventsat the site of penetration for partialresistance. Euphytica, 35,:961-968.2. Browder. L.E. 1985. Parasite:host: environment specificity in thecereal rusts. Annual Reviews ofPhytopathology, 23:201-222.3. Dehne. D. 1977. Untersuchungenzur Resistenz von Sommer-weizengegenber Gelbrost (Pucciniastriiformis West.) im Feld und unterkontrolierten Bedingungen wahrendder Ontogenese der Wirtsplanze. Ph.D. Dissertation. University ofG6ttingen.4. Dijk, P. van, J.E. Parlevliet,G.H.J. Kema, A.C. Zeven, and RW.Stubbs. 1988. Characterization of thedurable resistance to yellow rust inold winter wheat cultivars in TheNetherlands. EuphyUca, 37. In press.5. Gavinlertvatana, S., and R D.Wilcoxson. 1978. Inheritance of slowrusting of spring wheat by Pucciniarecondita f. sp. tritici and hostparasite relationships. Transaction ofthe British Mycological SOciety,71:413-418.6. Habgood, RM., and B.C. Clifford.1981. Breeding barley for diseaseresistance: The essence ofcompromise. In J.F. Jenkyn andRT. Plumb (eds.) Strategies for theControl of Cereal Disease. Oxford.Blackwell, 15-25.7. Kim. S.K., and J.L. Brewbaker.1977. Inheritance of generalresistance in maize to Pucciniasorghi Schw. Crop Science,17:456-491.8. Kuhn, RC., H.W. Ohm, and G.E.Shaner, 1980. Inheritance of slowleaf-rusting resistance in suwon 85wheat. Crop SCience. 20:655-659.9. Luke. H.H., W.H. Chapman, andRD. Barnett. 1972. Horizontalresistance of red rustproof oats tocrown rust. Phytopathology62:414-417.

6110. Luke. H.H., R. D. Barnett. andP.L. Pfahler. 1975. Inheritance ofhorizontal resistance to crown rust inoats. Phytopathology , 65:631-632.11. Niks, R.E .. and J .E. Parlevliet.1978. Variation for partial resistanceto Puccinia hordei in the barleycomposite XXI. Cereal RustsBulletin. 6:3-10.12. Ohm, H.W., and G .E. Shaner.1976. Three components of slow leafrustingat different growth stages inwheat. Phytopathology.66: 1356-1360.13. Parlevliet. J.E. 1975. Partialresistance of barley to leaf rust,Puccinia hordei. I. Effect of cultivarand development stage on latentperiod. Euphytica, 24:21-27.14. Parlevliet, J .E. 1976. Partialresistance of barley to leaf rust,Puccinia hordei. III. the inheritanceof the host plant effect on latentperiod in four cultivars. Euphytica,25:241-248.15. Parlevliet, J .E. 1978. Furtherevidence of polygenic inheritance ofpartial resistance in barley to leafrust, Puccinia hordei. Euphytica,27:369-379.16. Parlevliet, J .E. 1979.Components of resistance thatreduce the rate of epidemicdevelopment. Annual Review ofPhytopathology, 17:203-222.17. Parlevliet. J.E. 1981. Racenonspecificdisease resistance. InJ.F. Jenkyn and R.T. Plumb (eds.).Strategies for the control of cerealdisease. Oxford. Blackwell. 47-54.18. Par1evliet, J.E. 1983a. Durableresistance in self-fertilizing annuals.In F. Lamberti. J .M. Waller and N.A.v.d. Graaff (eds.) Durable Resistancein Crops, New York and London.Plenum Press, 347-362.19. Parlevliet. J .E. 1983b. Modelsexplaining the specificity anddurability of host resistance derivedfrom observations on the barley­Puccinia hordei system. In F.Lamberti, J.M. Waller and N.A. v.d.Graaf (eds.). Durable Resistance inCrops. New York and London,Plenum Press. 57-80.20. Parlevliet. J .E. 1985. Resistanceof the nonrace-specific type. In A.P.Roelfs and W.R. Bushnell (eds.). TheCereal Rusts II. New York, AcademicPress, 501-525.21. Parlevliet. J.E. 1986. Pleiotropicassociation of infection frequencyand latent period of two barleycultivars partially resistant to barleyleaf rust. Euphytica, 35:267-272.22. Parlevliet. J .E .. and H.J. Juiper.1977. Resistance of some barleycultivars to leaf rust, Pucciniahordei; polygenic. partial resistancehidden by monogenichypersensitivity. NetherlandsJournal of Plant Pathology. 83:85-89.23. Parlevliet, J.E., and H.J. Kuiper.1985. Accumulating polygenes forpartial resistance to barley leaf rust,Puccinia hordei. I. Selection forincreased latent periods. Euphytica.34:7-13.24. Parlevliet, J .E., M. Lejin. and A.van Ommeren. 1985. Accumulatingpolygenes for partial resistance inbarley to barley leaf rust, Pucciniahordei. II. Field evaluation.Euphytica. 34:15-20.

6225. Parlevliet, J.E., W.H. Lindhout,A. van Ommeren, and H.J. Kuiper.1980. Level of partial resistance toleaf rust, Puccinia hordei in West­European barley and how to selectfor it. Euphytica, 29:1-8.26. Parlevliet, J.E., and A. vanOmmeren. 1975. Partial resistance ofbarley to leaf rust, Puccinia hordei.II. Relationship between field trials,micro plot tests and latent period.Euphytica, 24:293-303.27. Parlevliet, J.E., and A. vanOmmeren 1984. Interplotinterference and the assessment ofbarley cultivars for partial resistanceto leaf rust, Puccinia hordei.Euphytica, 33:685-697.28. Parlevliet, J.E., and A. vanOmmeren 1988. Accumulation ofpartial resistance to barley leaf rustand powdery mildew throughrecurrent selection againstsusceptibility. Euphytica, 37, inpress.29. Rees, RG., J.P. Thompson, andR Mayer. 1979. Slow rusting andtolerance to rusts in wheat. I. Theprogress and effects of epidemics ofPuccinia graminis tritici in selectedwheat cultivars. Australian Journalof Agricultural Research, 30:403-419.30. Rees, RG., J .P. Thompson, andE.A. Goward. 1979. Slow rustin~ andtolerance to rusts in wheat. II. Tbeprogress and effects of epidemics ofPuccinia recondita tritici in selectedwheat cultlvars. Australian Journalof Agricultural Research, 30:421-432.3l. Robbelen, G., and E.L. Sharp.1978. Mode of inheritance,interaction and application of genesconditioning resistance to yellowrust. Advances in Plant Breeding(Supplement 9), Berlin, Parey.32. Suneson, C.A., and G.A. Wiebe.1962. A 'Paul Bunyan' plantbreeding enterprise with barley. CropScience, 2:347-348.33. Skovmand, B., RD. Wilcoxson,B.L. Shearer, and RE. Stucker.1978. Inheritance of slow rusting tostem rust in wheat. Euphytica,27:95-107.34. Vanderplank, J.E. 1968. DiseaseResistance in Plants. New York,Academic Press.35. Wilcoxson, RD. 1981. Geneticsof slow rusting in cereals.Phytopathology, 71 :989-993.36. Wilcoxson, RD., B. Skovmand,and A.H. Atif. 1975. Evaluation ofwheat culJ:ivars for ability to retarddevelopment of stem rust. Annals ofApplied Biology, 80:275-28l.

Chapter 6 Durable Resistance to Yellow (Stripe) Rust inWheat and Its Implications in Plant BreedingR. Johnson, Plant Breeding Institute, Cambridge, EnglandAbstractYellow (stripe) rust, caused by the obligate parasite, Puccinia s triiformis. isfound wherever wheat is grown in cool climates. Several race-specificresistance genes effective in wheat seedlings have been identified but moreremain to De identified. Such resistance genes may be dominant or recessiveand some are strongly influenced in expression by environment and geneticbackground. Resistance developing after the seedling stage is also frequentlyrace-specific. Combining together race-specific resistance genes has not beensuccessful in controlling yellow rust in Britain. However, some resistancedeveloping after the seedling stage does not show race-specificity even afterprolonged and widespread testing. Such durable resistance can only bedistinguished from race-specific adult plant resistance by prolonged testing.Although it may be under complex genetiC control, such resistance can beused in breeding programs as described here but the durability of resistanceproduced in such programs cannot be guaranteed. All new resistantcultivars, whatever the breeding method used, should therefore be monitoredfor eVidence ofpathogen races with matching pathogenicity.The Distribution ofYellow Rust on WheatYellow (stripe) rust is potentially adamaging disease in all cool climatesin which wheat is grown.Surrounding or contiguous withmost of these areas are other wheatgrowing zones where the climate ismarginal. usually too warm and dry.for yellow rust. Occasionally thedisease spreads into such areas andits distribution may be thought to beextending. Such events aresometimes attributed to climaticchange or, perhaps, to adaptation ofthe pathogen to higher temperatures.There are several reports ofexperiments on optimumtemperatures for spore production,germination and infection by thepathogen Puccinia striiformis. Theseshow that many environmentalfactors can influence the viabilityand germinability of theurediospores, including thetemperatures, light conditions, andhumidity in which they areproduced, as well as those in which63germination and infection take place(4). It has been suggested that thetemperature at which the spores areproduced can influence the optimumtemperature for germination (20). Ifso, it is perhaps not surprising tofind that different temperatureoptima have been reported forgermination of P. striiformis indifferent studies. What is less clear iswhether such differences indicate thespecializatj,on of the pathogen intoraces genetically adapted to differenttemperatures. Dennis (3) suggestedthe possibility that races of P.striiformis capable of surviving hightemperatures in the Australiansummer might have hightemperature optima for infection.However, he concluded that a race ofP. striiformis that had evolved inAustralia showed similar responsesto temperature to those reported inother countries, with an optimum forinfection of between 7°C and 10°C,maximum about 18°C andminimum below OOC . He concludedthat these temperatures wereprobably typical of P. striiformis.

64In general. I believe that occasionalextended distribution of yellow ruston wheat usually occurs due towidespread cultivation of a highlysusceptible cultivar in an area withan environment marginal for thedisease, sometimes assisted byunusually favorable weatherconditions, rather than adaptation ofthe pathogen to new climates.Specificity in ResistanceAs with the other rust pathogens ofwheat, P. striiformis has been knownsince the 1930s to be specialized intoraces that differ in pathogenicitytowards individual wheat cultivars,as well as to related species andgenera. It is assumed that thespecificity of pathogenicity towardsindividual wheat cultivars operatesaccording to the gene-for-genehypothesis. Resistance hasfrequently been demonstrated inwheat seedlings and eleven loci withsuch genes have been described,using the Yr symbol (Yr1 to Yr10and Yr15) (13,14, 15). Other suchgenes have been detected but remainto be identified and designated. Ofthe Yr genes so far identified,including those such as YrB and Yr9from alien sources, all have provedto be race-specific, except therecently introduced and not yetexploited gene Yr15 from Triticumdicoccoides (C. van Silfhout, pers.comm.).The infection types of the differentgenes expressed in seedlings includesome which produce very minutechlorotic flecks such as Yr1, YrB,and Yr10, others that produceextensive necrosis with or withoutsome sporulation such as Yr7, andothers that give less consistentreactions sometimes ranging from anonsporulating reaction toconsiderable sporulation and onlyslight chlorosis. I include in thisgroup such genes as Yr2 and Yr6.which may vary with environmentand also in response to the geneticbackground in which they occur.Some of the named genes aredominant but several are recessiveat least in some crosses, includingYr2, Yr6, and Yr9.In addition to specificity of resistancedetected in seedlings, it is critical tonote that. in the interaction of wheatwith P. striiformis. race-specificity ofresistance can also be found inresistance that develops after theseedling stage and is most readilydetected in adult plants (8, 16, 18,25). Four resistance genes (Yr11 toYr14) that provide such resistancewere recently identified (15) and it isclear that other race-specificresistances detected in adult plantsreported by Stubbs (21) are differentfrom those controlled by these fourgenes (9). Specificity of resistance inadult plants is not unique to yellowrust. having certainly been observedalso in resistance to brown leaf rustof wheat (2). However, I emphasizethis aspect of resistance to yellowrust because of the widespreadassumption implicit in somepublications that resistance that isincomplete or moderate, rather thancomplete, is race-non-specific.Distribution of Races ofp. strllformls in the WorldStubbs (2i) concluded fromworldwide surveys of P. striiformisthat "yellow rust in all parts of theworld possesses the same geneticbackground of pathogenicity." Thiswas based on observations of thedistribution of pathogenicity forrecognized race-specific resistancegenes in many parts of the worldand particularly where suchresistance genes were deployed.Despite these observations, it shouldnot be assumed that the distributionof pathogenicity is uniform. andsome of the variation may beimportant in trying to propose ageneral approach to breeding for

65resistance. The following examplessuggest some significant differencesin the distribution of specificpathogenicity. The gene Yr5 fromTriticum speJta var. album waseffective against all naturallyoccurring races in Europe for manyyears. but was matched in Australiawithin 4 years of the arrival of thepathogen there (C.R. Wellings andR.A. McIntosh. pers. comm.); it wasnot used widely in Europe nor in anycommercial cultivar in Australia.The resistance gene YrlO derivedfrom a Turkish wheat PI No. 178383was transferred to the cultivar Moro.where it was soon matched whenthis cultivar was used commerciallyin the USA (1). Pathogenicity wasalso reported from the Middle East(21) but. so far. no pathogenicity forthis gene has been detected in theUK or Australia. It has not been usedin commercial cultivars in eithercountry but its present performancein Australia contrasts with the thatof Yr5. Pathogenicity for Yr8 fromAegilops comosa was found inEngland from three different sources.including natural occurrence in thefield. although the gene was not usedin any commercial cultivar (7).Pathogenicity for Yr8 is found insome other areas. but noteverywhere. A race-specificresistance recognized in theAustralian wheat cultivar Avocet wasdetected shortly after theintroduction of P. striiformis toAustralia (23). Attempts to recognizethis specificity using UK races haveshown that the race-specificresistance of the Australian Avocet ismore effective against some UKisolates than others but. of the racesso far tested. none completelymatches the gene in the way shownby the race 104E137A + fromAustralia (Table 1).Table 1. Infection types (scale: 0 resistant to 4 susceptible) ofselections of Australian wheats lacking (-) or possessing ( + ) race-specificresistance first identified in the Australian cultivar Avocet (YrAv)Race and Isolate Number104E137A+ aJ 104E137A- aJ 106E139 45E140Line YrAv Leaf WYR 85-25 WYR 85-24 WYR 81-24 WYR 75-23Avocet I 3+·4' 4' 4 42 4 4Avocet + I 4' 3·3+ 3·3+ 3'·4'2 ON·3· ON·I·N 2N·3·Banks I 3'·3+ 3+·4 4'·4 42 4'Banks + I 3'·4' ON·3 ON·I·N 1+ N·3 +2 4' ON·2·N ON·2·N 1+·3'Strubes Dickkopf 3+·4 3+ 3+ 4(Susceptible)aJ Isolates from Australia via R.W. Stubbs, possessing (A+) or lacking (A·) pathogenicity for YrAv(UK MAFF license PHF 48/A80(76))• No data

66These aspects of the distribution ofvariation in pathogenicity of P.striiformis are referred to in thediscussion of breeding for durableresistance.Durable Resistanceto Yellow RustIn the climate of the UK, which isfavorable to the pathogen in mostyears, combining as many as fourrace-specific genes, whether of thetype effective in seedlings andthroughout the life of the plant. oronly in the adult stage, has not sofar been a successful breedingstrategy. Table 2 indicates the fate ofsome of the cultivars in which suchgenes were used. Combining racespecificgenes, particularly those forwhich matching pathogenicity israre, might provide a moresatisfactory control of disease inclimates that are less favorable tothe pathogen.Despite these examples of rapidlymatched resistance, there wereduring the same period othercultivars that remained adequatelyresistant. Similar experiences arereported from the northwesternUnited States (l2). Some cultivarsthought to have displayed durableresistance to yellow rust in the UKand elsewhere are listed in Table 3.The list is not complete but all thecultivars induded possess a type ofincomplete resistance in thepresence of races with pathogenicitymatching their known race-specificgenes. As yet there seems to be nosimple test for distinguishing thephenotype presented by theseTable 2. Races of P. striiformis with pathogenicity matching cultivarswith race-specific genes and combinations of genes in the UKaIYear Race Cultivar/Line Yr genes in cultivarbl1966 37E132 Rothwell Perdix I, 2(seg)cl1968 41E136 Maris Templar 1. Cdl1969 104E137 Maris Beacon 2, H el1971 104E137(2) Joss Cambier 2, 1141EI36(2) Joss Cambier1974 41E136(3) Maris Nimrod 2, C, 131975 232E137 Clement 2?f/, 91979 41EI36(4) CWW 916/26 I, C, 141980 171E138 CWW 1771 1,91982 169E136 CWW 1645 1. 9, 13a/ All cultivars were too susceptible for further usebl Postulated from interactions with racescl Segregatingdl C indicates specificity like that of Cappelle Desprezel H indicates specificity like that of Hybrid 46fI ?, probably present but not demonstrated with existing races

67Table 3. Wheat cultivars that probably possess durable resistance to yellow rust, maximum level of infection to any race in a field trial 26 June 1980, and probable race-specific genes CultivarOriginMaximumpercentinfectionYr genesAnza(WWI5)(Kararnu)AtouH. de Bersee Bouquet Cappelle Desprez Champlein Elite Lepeuple Flanders Flinor Holdfast Hybrid 46 Jubilar Little Joss Luke Maris Huntsman Nugaines Starke II Vilmorin 27 Maris Widgeon Yeoman Desprez 80 (susceptible)USAAustraliaNew ZealandEurope (E)EEEEEEEEEEEUSAEUSAEEEEE8 Ava}AvAv4 C bl5 C 25 C, 14 37 C 38 C 13 2 17 1 (seg), C 2 7 cl 6 7 25 Hdl13 7 3 7 22 7 '1.72, C, 13 33 7 17 7 12 C 4 C 20 13 75 C a/ Av specificity similar to Australian Avocet bl C, specificity similar to CappeUe Desprez cl 7, no recognized specificity dl H, specificity similar to Hybrid 46

68durably resistant cultivars fromcultivars that possess incompleteresistance that subsequently provesto be race-specific. For example,those we have identified do notappear to have a high infection typecombined with a low diseaseincidence. as is sometimes indicatedfor slow rusting resistance or partialresistance that may be inferred to berace-non-specific to other rustpathogens. It is reported from theUSA that the resistance of durablyresistant cultivars such as Nugainesis temperature-sensitive, being moreeffective in adult plants at hightemperatures (12). In our experience,however, high temperatures are notcritical for the expression ofresistance in cultivars that weconsider to have durable resistance.Nevertheless, it is probable that themajority of the resistances, includingthose known to be race-specific, areto some degree sensitive totemperature. Thus temperaturesensitivity per se cannot beconsidered to be diagnostic fordurable resistance. This raises theproblem of whether there are anysimple tests that could help toindicate potentially durableresistance.One necessary component of anactive breeding program is thetesting of selected lines at manylocations to assess the stability ofperformance. This method maysometimes indicate vulnerability todiseases, thus helping to avoid risksin using a new cultivar. However,the method cannot reliably indicatethe potential durability of resistancein a cultivar to variable pathogenssuch as P. striiformis because thetotal area of the cultivar that can actas a selective screen for pathogenicraces remains small in such testscompared with the area in largescalecommercial exploitation.Similarly, it is sometimes consideredthat successful performance in multilocationaltesting can help to ensurethat resistance is under complexgenetical control. Combined withrecycling of breeding materialthrough crossing, as in CIMMYTprograms it may help to permit theformation of longer-lastingcombinations of resistance genes.However, it is also possible for asingle gene, especially a newlyintroduced one, to give resistance inall test nurseries. Tbe example ofresistance to stem rust in triticalegiven by McIntosh (Chapter 1)illustrates this problem. Thealternative, of testing new cultivarswith many pathogen races suffersfrom similar limitations. Even wherethe cultivar remains resistant. thisdoes not provide a strong test forpotential durability of resistance, andit does not necessarily indicate itscomplex genetical control.Lastly it should be noted that slowdevelopment of disease, that whichmay be referred to as slow rusting ordilatory resistance, can be due to oneof at least three possible causes: 1)race-specific, adult plant orincomplete resistance; 2) a lowfrequency of pathogenicity for a racespecificgene in a mixed populationof races, so that cultivars possessingthe gene receive a low frequency ofmatching infection; 3) slow rusting ofa durable, apparently race-nonspecifictype.For these reasons I consider that thisstill leaves the difficulty of being ableto identify cultivars with durableresistance only after they have beenwidely grown. If this is true, suchcultivars are all the more valuableboth as potential sources of durableresistance and for studies of thegenetical basis and mechanism oftheir resistance, as indicated forstem rust and leaf rust of wheat byMcIntosh (Chapter 1) and Roelfs(Chapter 2).

69Genetical Basis of DurableResistance to Yellow RustMany of the cultivars identified ashaving displayed durable resistance.such as Cappelle Desprez and MarlsHuntsman. show evidence ofpossessing some race-specificcomponents. while others. such asHoldfast and Little Joss. have notdisplayed any race-specificity to P.striiforrnis in the UK. However. theyhave not been tested with exoticraces of the pathogen. The questioncan be posed as to whether theresistance of the latter two cultivarsand the residual resistance whenrace-specific components have beenmatched in the former class are racenon-specific.I think that there canbe no absolute proof of nonspecificityand. particularly. of thenon-specificity of the geneticalcomponents of which their resistancecan be shown to consist. Forpractical purposes. however. theresistance has not displayed racespecificityeven after prolonged andwidespread testing. whichencourages the hope. though not thecertainty. that it will not prove to berace-specific if used further or eventransferred to a new cultivar byappropriate breeding.For cultivars that possess durableresistance (and contain genes thatare race-specific). it is worthenquiring whether those genesprovide the residual resistanceshown by the cultivars to races thatare considered to have matchingpathogenicity to the race-specificgenes. I have no proof that they donot provide such residual resistancebut the data of Table 3 show thatcultivars with the same complementof race-specific genes can have verydifferent levels of resistance to theraces that have pathogenicitymatching those genes. I interpretthis to mean that the cultivars carrygenes. other than the recognizedrace-specific genes. that providethese levels of resistance. Thus. inCappelle Desprez and other similarcultivars. I do not consider that theirdurable adult plant resistance is. toany measurable degree. a residual or'ghost' effect of their known racespecificgenes.There have been few detailedgenetical analyses of cultivarsidentified as possessing durableresistance to yellow rust. However.some cytogenetical studies havebeen carried out on Cappelle Desprezand other cultivars at the PlantBreeding Institute (PBl). UsingCappelle Desprez as an example. it isknown to possess race-specificresistance to certain races and wasdescribed as possessing genes Yr3aand Yr4a by Lupton and Macer (13).In addition. Cappelle Desprezpossesses at the adult stage amoderate level of resistance. thatremained effective during widespreaduse of the cultivar for almost 20years in the UK (5). Evidence wasobtained in many tests for a smallrace-specific component in this adultplant resistance (6). Races used toanalyze the cytogenetic control ofthe adult plant resistance were thosegiving the highest known levels ofinfection on Cappelle Desprez. It wasshown that a chromosomedesignated as 5BS-7BS carried animportant component of theresistance since plants nullisomic forthis chromosome were much moresusceptible than the euploid and thatthis resistance was controlled by the5BS arm (Table 4) (11). Familiesconsisting of a mixture ofmonosomic and euploid plants.derived from plants monosomic foreach of the chromosomes exceptchromosome 5BL-7BL. deviated fromthe euploid in their resistance. Inparticular. plants monosomic forchromosomes 5A and 5D were moreresistant than the euploid (11). Theseresults corresponded with the results

70of studies with other cultivars (17)and suggested that the long arms ofchromosomes of homoeologousgroup 5 often promote susceptibilitywhile the short arms promoteresistance. Also. using the techniqueof reciprocal backcross monosomics.Worland and Law (24) identified agene (Yr16) on chromosome 2D ofCappelle Desprez that controlled partof its adult plant resistance.These data indicate that theresistance of Cappelle Desprez isgenetically complex; it includescomponents identified as racespecificand also other componentsthat contribute to the level ofresistance observed in adult plants.the effects of which have not beenshown to be race-specific. Theseinclude various elements that bothincrease and decrease resistance. theobserved resistance being theresultant product of thesecomponents. Given this complexity.it is not possible to conclude whichare the particular geneticcomponents that are critical for thedurability of resistance displayed byCappelle Desprez.It is not suggested that thisdescription of the differentcomponents of resistance to yellowrust in Cappelle Desprez is a modelfor other cultivars that possessdurable resistance. However. nosingle genes or gene complexes haveyet been identified that can beimplicated in durable resistance andthe methods I have proposed totransfer durable resistance inbreeding programs therefore dependupon trying to ensure that resistanceis derived from a known durablesource and to include as many of thecomponents of resistance from thatsource as possible.Breeding for DurableResistance to Yellow RustAt the PBI I have tried to encouragethe breeders to use sources ofdurable resistance. such as Hybridede Bersee and Cappelle Desprez. toprovide resistance to yellow rust innew wheat cultivars. As noted above.however. some of the durablyresistant wheats possess identifiedrace-specific genes and manymodern cultivars that breeders wishto use in crosses also contain racespecificgenes. In order to select themore important components ofresistance. selection should becarried out using a race of thepathogen with pathogenicitymatching the known race-speCificgenes. If this is done. no specialTable 4. Percentage of flag leaf infected with yellow rust in CappelleDesprez, euploid and various aneuploids involving chromosome 5BS-7BSalLineEuplOidMonosomic 5BS-7BSNullisomic 5BS-7BSDitelosomic 5BSD!telosomic 7BSPercentage infection12.617.348.87.747.5aJ Law et ai. (11)

71techniques are required, since merevisual selection for resistancesuffices. However, it is possible, insome crosses, for recognized racespecificgenes to come together innew combinations for which we haveno currently available matching race.This can include combinations ofgenes effective in the seedling withthose race-specific genes that areeffective only after the seedlingstage. These combinations giveresistance, sometimes at a higherlevel than that derived from thedurably resistant parent. It istherefore proposed that suchcombinations should be prevented,either by controlling the crossinp; sothat they cannot be formed or, ifthey do occur, by eliminatingwhichever is the simplest gene todetect so that a race withpathogenicity matching the knownremaining race-specific genes can beused for screening. An example ofthe method is indicated in Table 5.In practice, breeders are reluctant touse old cultivars in crossingprograms, and also to limit the rangeof crosses merely to prevent theformation of new combinations ofrace-speCific resistance genes. Mycolleagues, Mr. A.J. Taylor and Mr.G.M.B. Smith, have thereforeinitiated a project to introduceresistance to yellow rust fromdurably resistant varieties intoTable 5. Pedigree of cultivar Bounty showing how resistance wasinherited from the cultivar Ploughman. thought to possess durableresistance to yellow rust derived from Maris WidgeonPloughman x Durin(durable resistance)(susceptble in fieldto races with matchingpathogenicity)Yr genes: Hal 1, CblRace Seedling Adult Seedling Adult4IEI36 rC r s s104EI37 sd r r rBountyYr genes: 1, C, 13Race Seedling Adult4IEI36(3) s r104EI37 r rNote: Combination of YrH with Yrl would have given complete resistance to all racesavailable at the time. so YrH was eliminated and selection carried out in presence ofrace 41E136(3). The origin of Yr13 in Bounty is not clear, but it may have come fromPloughman. in which it was not detected due to the presence of YrH.al H, race·specificity like Hybrid 46bl C, race-speCificity like Cappelle Desprezcl Resistantdl Susceptible

72advanced but susceptible lines fromthe main PBI winter wheat breedingprogram. In this project. selectionhas not been limited to resistance toyellow rust but has includedresistance to other diseases andselection for agronomic characters(Chapter 5). Selection for resistanceto rust takes place in the presence ofraces with pathogenicity for all thepossible combinations of race-specificgenes that could occur in thecrosses. In order to limit the possibleproduction of effective newcombinations of race-specific genes.the opportunity has been taken toeliminate as many as possible of therace-specific genes that can bedetected in seedlings. Owing to therecessive nature of some of thegenes. this has not been Simple. Noattempt has been made to eliminatethe race-specific genes effective inadult plants. owing to the difficultyof detecting them when races withmatching pathogenicity are frequentin the pathogen population presentin the breeding nurseries (9). Thelogistics of this small breedingprogram are quite different fromthose of the main winter wheatprogram. and it is most unlikely thatany of the selected lines couldcompete as successful cultivars withthose from the main program. It isintended only that they will besuffiCiently attractive for thebreeders to consider them ascrossing parents. thus keeping in theprogram as much as possible of thegenetic components derived fromsources of resistance that arethought to be durable.As noted. some of the lines from thisspecial breeding program willcontain genes for race-specific adultplant resistance. probably includingYr12. Yr13. and Yr14 for which wepossess races with matchingpathogenicity. However. thedistribution of pathogenicity forthese genes in the world is notmonitored and. if the lines were sen~to breeders in other countries. it maybe that they would observe theresistance due to those race-specificgenes. Likewise. the cultivar Anzaand related (or synonymous)cultivars appear to possess a durableresistance to yellow rust. However.they also possess the race-specificcomponent first recognized in theAustralian Avocet (C.R. Wellings andR.A. McIntosh. pers. comm.) (Table3). In Britain. the resistance weobserve in these lines is probablydue to this race-specific component.rather than to the genes controllingthe durable resistance expressed inAnza in the presence of races thathave matching pathogenicity.Therefore. if we attempted to useAnza as a source of durableresistance to yellow rust in ourprograms. we might only besuccessful in selectin~ the racespecificcomponent of the resistanceand producing cultivars withresistance similar to that of theAustralian Avocet which becamehighly susceptible when its racespecificcomponent was matched.Because of the extensive nature ofthe breeding programs at PBI, whichare typical of many wheat breedingprograms. it cannot be expected thatcrosses will be limited. or evenhandled conSistently in the waydescribed. to those that will ensurethat resistance to yellow rust in allnew cultivars has been derived fromthe appropriate components of asource of durable resistance. It musttherefore be expected that manyvulnerable race-specific genes willsurvive in breeding populations andvarieties.Transgressive Segregation forResistance to Yellow RustNumerous authors have reportedthat resistance at a higher level thanin either parent can be selected inthe progeny from crosses betweensusceptible or moderately resistant

73parents (e.g., 10, 22). One suchexample concerns a cross betweenMaris Huntsman and CappelleDesprez. These two cultivarspossessed a similar level ofresistance to a race of P. striiformisthat overcrune the recognized racespecificcomponents of both andCappelle Desprez was in the pedigreeof Maris Huntsman. Surprisingly,despite their close relationship, therewas wide segregation for resistancein the F2 progeny, indicating thatthe parents differed genetically incomponents of resistance (Table 6).This was confirmed in the F3generation (22). These and otherexamples indicate that transgressivesegregation for resistance to yellowrust can be selected in the progenyfrom many different wheat crosses. Itcould therefore be a useful way ofincreasing resistance by crossesamong locally adapted wheats,rather than trying to transferresistance from unadapted sourcesthought to possess durableresistance. However, I believe thatthe assumption that resistanceaccumulated in this way will be racenon-specificis not warranted. Thereis no proof that a race matching allpotentially race-specific genes can befound to screen the segregatingprogeny, as proposed by Robinson( 19). Furthermore, transgressivesegregation for resistance could arisefrom interactions or additive effectsof race-specific genes. or from thetransfer of race-specific genes fromsuppressive to an expressivebackground (22). It could also arisefrom accumulation of resistancegenes of the type associated withdurable resistance. It would not bepossible in advance to predict whichof these possibilities had beenachieved.ConclusionsThere is much resistance to yellowrust among existing wheat cultivarsand even more is available fromalien sources but the genetic basis ofmuch of this resistance is notknown. There are numerousexamples of resistance that haveproved. on exploitation. to be racespecificand as many again that havedemonstrated great durability. Muchmore data should be accumulated onthe scale of cultivation of cultivarsand the evidence of durability oftheir resistance to rust diseases. It isproposed in this chapter that thebest way to enhance the probabilityof achieving durable resistance inTable S. Frequency distributions of parents and progeny from crossesbetween Cappelle Desprez and Maris Huntsman infected with race41E13S(3) of P. strllformis. (Modified from Wallwork andJohnson (22))Percent of flag leaf infectedLines 10 20 30 40 50 SO 70 80 90 100 MeanCappelle Desprez 9 11 3 1 26Maris Huntsman 14 9 1 23Fl 4 22 14 70F2 5 29 33 41 27 21 28 31 18 6 50

74new cultivars is to transfer resistancefrom sources already identified asdurable. As yet there is no simpleway to identify the precise geneticcomponents that are associated withthe durable resistance to yellow rustof wheat. so the method is designedto retain as much of the resistanceas possible from durably resistantsources. Resistance could also beaccumulated from crosses betweenmoderately susceptible wheats; thiscan readily be achieved withinpedigree programs and does notreqUire resort to recurrent massselection methods. Although there isno guarantee that such resistancewould be durable. the method seemsmuch more promising than theintroduction of single major genes orcombining already known racespecificgenes into new groups. Theaccumulation of resistance fromcrosses of locally adapted cultivarscould sometimes be attractive as analternative to attempting to derivedurable resistance from known butunadapted sources.Because of the large-scale nature ofmajor wheat breeding programs. it isunlikely that either of these methodscould be applied conSistently to allcrosses. Furthermore. although themethods might increase theprobability of producing cultivarswith durable resistance. they are notsuch as to guarantee futuredurability. I have also indicated myopinion that no currently availabletest. including multilocation testing.can determine the potentialdurability of resistance. Under thesecircumstances. it remains vitallyimportant to monitor the diseaseresistance of all new cultivars and tomaintain systems for detectingchanges in pathogen populations.The risks of genetic uniformityshould be continually kept in viewand systems to reduce widespreaddependence on single uniformcultivars should be encouraged.Referencesl. Beaver. RG.. and RL. Powelson1969. A new race of stripe rustpathogenic on the wheat varietyMoro. CI 13740. Plant DiseaseReporter. 53:91-93.2. Browder. L.E. 1980. Acompendium of information aboutnamed genes for low reaction toPuccinia recondita in wheat. CropScience. 20:775-779.3. Dennis. J.I. 1987. Temperatureand wet-period conditions forinfection by Puccinia striiformis f.sp.tritici race 104E137A+.Transactions of the BritishMycological Society. 88: 119-12l.4. Gopalan R. and J.G. Manners.1984. Environmental and otherfactors affecting germination ofurediniospores of Pucciniastriiformis. Transactions of theBritish Mycological Society.82:239-243.5. Johnson. R 1968. Practicalbreeding for durable resistance torust diseases in self-pollinatingcereals. Euphytica. 27:529-540.6. Johnson, R 1984. A criticalanalysis of durable resistance.Annual R(;:view of Phytopathology.22:309-330.7. Johnson. R, RH. Priestley. andE.C. Taylor. 1978. Occurrence ofvirulence in Puccinia striiformis forCompair wheat in England. CerealRusts Bulletin. 6:11-13.8. Johnson. R. and A.J. Taylor.1977. Yellow rust of wheat. InAnnual Report of the Plant BreedingInstitute for 1976. 106-108.9. Johnson. R. A.J. Taylor. G.M.B.Smith. 1987. Yellow rust of wheat.In Annual Report of the PlantBreeding Institute for 1986. 89-93.

7510. Krupinsky. J.M.. and E.L. Sharp.1979. Reselection for improvedresistance of wheat to stripe rust.Phytopathology, 69:400-404.11. Law. C.N., RC. Gaines, RJohnson, and A.J. Worland. 1978.The application of aneuploidtechniques to a study of stripe rustresistance in wheat. Proceedings ofthe 5th International Wheat GeneticsSymposium, New Delhi, 1978,427-436.12. Line, R 1983. Resistance toPuccinia striiformis in United Statesof America. In The Current Status ofStripe Rust Research, 48-49,Victorian Crop Research Institute,Horsham, Australia.13. Lupton, F .G.H., and RC.F.Macer. 1962. Inheritance ofresistance to yellow rust (Pucciniaglumarum Erikss. & Henn.) in sevenvarieties of wheat. Transactions ofthe British Mycological Society,45:21-45.14. McIntosh, RA. 1983. A catalogueof gene symbols for wheat (1983edition). Proceedings of the 6thInternational Wheat GeneticsSymposium, Kyoto, Japan,1197-1254.15. McIntosh, RA. 1986. Catalogueof gene symbols for wheat: 1986supplement. Cereals ResearchCommunications, 105-115.16. Manners, J.G. 1950. Studies onthe physiologic specialization ofyellow rust. (Puccinia glumarum(Schm.) Erikss. and Henn. in GreatBritain. Annals ofApplied Biology,37:187-214.17. Pink, D.A.C., F.G.A. Bennett,C.E. Caten, and C.N. Law, 1983. Theeffects of homoeologous group 5chromosomes on disease resistance.Zeitschrift fUr Pflanzenziichtung.91 :278-294.18. Priestley, RH. 1978. Detection ofincreased virulence in populations ofwheat yellow rust. In Plant DiseaseEpidemiology. P.R Scott and A.Brainbridge (eds.J. 63-70. Oxford.Blackwell.19. Robinson, RA. 1976. PlantPathosystems. Berlin etc.. Springer­Verlag.20. Sharp. E.L. 1965. Prepenetrationand postpenetration environmentand development of Pucciniastriiformis on wheat. Phytopathology55:195-203.21. Stubbs, RW. (1985). Stripe Rust.In The Cereal Rusts, II, A.P. Roelfsand W.R Bushnell (eds.J. 61-101.London. Academic Press.22. Wallwork. H.. and R Johnson.1984. Transgressive segregation forresistance to yellow rust in wheat.Euphytica.33:123-132.23. Wellings. C.R 1983. Stripe ruststrain survey-Australia. In TheCurrent Status of Stripe RustResearch. 41-47. Victorian CropsResearch Institute. Horsham,Australia.24. Worland. A.J.. and C.N. Law.1986. Genetic analysis ofchromosome 2D of wheat. I. Thelocation of genes affecting height.day-length insensitivity. hybriddwarfism. and yellow-rust resistance.Zeitschrift fUr Pflanzenziichtung. 96.331-345.25. Zadoks. J.C. 1961. Yellow ruston wheat. Studies in epidemiologyand physiologic specialization.Tijdschrift over Plantenziekten.67:69-256.

76Chapter 7Current Thinking on the Use of Diversity toBuffer Small Grains Against Highly Epidemicand Variable Foliar Pathogens: Problems andFuture ProspectsJ.A. Browning, Department of Plant Pathology and Microbiology, TexasAgricultural Experiment Station, College Station, TexasAbstractIndigenous co-evolved populations of small grains progenitors and theirobligate parasites are stabilized by epidemiologic dilatory resistance. This iseffected genetically by general resistance in which are embedded major-effectgenes for specific resistance in different spaciogenic configurations. Thegeneral resistance protects the plant; the specific resistance protects thepopulation by minimizing the pathogen's aggressiveness by adjusting thefinely tuned feed-back interaction of the interlocked host-pathogen geneticsystems. Both general and specific resistance should be used in agriculturalsystems consistent with their evolutionary origin. But often agriculture hasdepended on genes for specific resistance to protect not the population, as innature, but the plant from highly epidemic foliar pathogens; this has beenexperienced when the resistance gene was used in cultivars grown inhomogeneous stands over extensive areas. In contrast. programs inColombia. England, The Netherlands, India, and the USA (Iowa andWashington) have deployed genes for specific resistances in diverse standsand experienced population protection with as little as 1/3 resistance. Thesmall amount ofresistance has stabilized stands in a diverse naturalecosystem; thus, the figure seems real. Corroborating data from natural andagroecosystems suggest that a holistic gene-management system thatincludes diversity. as in nature, portends truly durable resistance. In spite ofits many benefits. relatively little diversity has been used. primarily becauseofi) a paradigmatic problem and ii) a cryptic error of experimentation thathas frequently underestimated the potential value of resistance anddiscouraged its use. A suitable compromise between the legitimate needs foragronomiC uniformity and the benefits of diversity are three-cultivarmixtures. .IntroductionThe USA's southern com leaf blight(SCLB) epidemic of 1970 resulted inthe greatest biomass loss of anybiological catastrophe. The cause ofthe epidemic is commonlyconsidered to have beenHelminthosporium maydis race T,but it was not. It is obvious that H.maydis race T was present and thatthe environment was favorable butthis was a man-made epidemiccaused by excessive homogeneity ofthe USA's tremendous maizehectarage (23, 33). A few plantpathologists and plant breeders hadbeen concerned about geneticvulnerability to disease, arthropods,and other causes of plant stress priorto 1970. But the birth of the conceptof genetic vulnerability in thepopular consciousness stemmedfrom the SCLB outbreak and wascertified by the National Academy ofSciences' (NAS) 1972 tome "GeneticVulnerability in Major Crops." Itstated, "If uniformity be the crux ofgenetic vulnerability, then diversityis the best insurance against it" (23,p.295).

77Diversity is considered desirable,even necessary. in many aspects oflife. including the world of finance.but not in the crops and animals onwhich mankind depends. With them.most effort has been directed tominimizing diversity. rather than toreconciling the benefits of diversitywith legitimate requirements foruniformity. Over a decade after thedevastating SCLB epidemic and thewarning of the NAS (23), breeders offive major crops (cotton. corn.sorghum. soybeans. and wheat)deemed USA cultivars not to begenetically vulnerable. The plantbreeders. in a survey conducted byDuvick (12). assessed vulnerabilityrelative to their perceived ability toadjust in a reasonable time to athreat detected anywhere in theworld and announced expeditiouslyover their commodity's internationalnetwork. They are probably correct ifone has the luxury of "time toadjust."Significantly. wheat breeders wereleast optimistic in their response tothe survey. possibly because theywork on an autogamous crop that,with other small grains. has a longand continuing history ofvulnerability to disease. Wheat andother vulnerable crops were inRobinson's (26. p.3) mind when hepleaded for a new. holistic approachto meeting disease challenges. Hestated: "After some 70 years ofscientific breeding of plants fordisease resistance. we are no nearercontrol of many of those diseasesthan when we started. Ironically. wehave been least successful with thediseases that have been mostintensively studied. We still suffersavage epidemics of wheat leafrust..." Recurringly. to the presenttime. small grains have been thrownout of genetic balance with theirpathogens and into an epidemic. Theintractable pathogens we have beenunable to keep in genetic balancecause crowd diseases (diseases ofdense populations) and they are ourmajor concern. These are highlyvariable pathogens in which virulentand aggressive variants can increaseat rates of up to 50% per day onnew. previously resistant cultivarsthat give them opportunity. Tomanage these pathogens. we mustmanage holistically all aspects of thehost population they mirrorgenetically. Truly. a new. moreeffective gene-management isindicated if we are to establish andmaintain a yield-stabilizing geneticbalance. This chapter is about genemanagement to thwart crowddiseases in crop populations. Idiscuss: 1) genetic and epidemiologicconcepts of resistance, concentratingon resistance principles that impactits management in diversepopulations to achieve populationbalance and, therefore. portenddurability; and ii) constraints againstthe use of diversity.Duvick (12) took pride in plantbreeders' ability to effect futureadjustment. As a small grainspathologist. my concern is not justfuture adjustment but also protectingthe current crop: Texas's 3 millionha of wheat sustained a 28%reduction in yield in 1985 from leafrust. and the loss was 50% in600.000 ha,of its most productivecentral area (21). The question is notjust how this could have beenprevented in 1986 and thereafter(important as that question is). buthow it might have been preventedbefore the 1985 crop was sown inthe fall of 1984.Theoretically. there are two possiblestrategies that are not mutuallyexclUSive-fungicides anddeployment of genetic host-plantresistance. Fungicides carry a cost tothe producer and to the environmentand pathogens may developinsensitivity to them. If used. they

78are completely miscible withresistance. If resistance is adequateand well managed, however, no foliarfungicide should be needed.Theoretically, seed-appliedsystematic fungicides may beeconomical, environmentally safeand epidemiologically promising.Applied to the seed of some, but notall, components of a mixedpopulation on a rotational basis asadvocated by Wolfe (36), theypromise to augment genemanagementprograms in a uniqueway. Deployment of genetic hostplantresistance is the main thrust ofthis paper.Genetic and EpidemiologicConcepts of ResistanceIn this chapter, the term hostreaction expresses, simultaneously,reaction (Le. resistance/susceptibility)of the host, avirulence/virulence andaggressiveness of the pathogen, andfavorableness of the environment.Host reaction has been described indiverse ways. Vanderplank (35)advanced resistance theory andpractice by dividing resistance intotwo contrasting types, vertical andhorizontal. He said vertical resistanceis that which is effective againstsome but not all races whereashorizontal resistance is thatoperative against all races. Then hereasoned that vertical resistance actsepidemiologically by decreasing theeffective amount of incominginoculum but not the rate of increaseof virulent races, while horizontalresistance acts, not by selectingdifferentially among strains, but byreducing the rate of increase of allstrains.Thus, Vanderplank (35) not onlydefined vertical and horizontalresistance genetically, but heprojected the consequencesepidemiologically. The apparentcorrelation between the type ofgenetic resistance in an individualplant and the epidemiologicalconsequences in a populationfrequently holds, but there areenough exceptions that confusionresults. In attempts to help correctthis, the terms and concepts havebeen analyzed and redefined.Unfortunately, some authors choseto retain the terms but with newdefinitions, which only increasedconfusion. This was reviewed byBrowning et al. (9) and Browning (5).The former (9) saw the need not justto clarify terms but also to separategenetic and epidemiologic conceptsof resistance. They retained"specific" and "general" resistanceas having the right of priority todescribe genetic concepts ofresistance. They coined the terms"discriminatory" and "dilatory" todescribe epidemiologic concepts.Discriminatory means "todistinguish and treat differently."dilatory means "to delay," andepidemiology deals with disease inpopulations. Thus, discriminatoryresistance or susceptibilitycharacterizes a population of hostplants that affects the epidemic bydiscriminating among pathogenstrains, Le., favoring or rejectingcomponents of the pathogenpopulation. Dilatory resistancecharacterizes a host population thataffects th.e epidemic by reducing therate of development of the pathogenpopulation. Tolerance, another usefulterm, was retained separate fromresistance to describe a plant or apopulation that is rated susceptiblevisually but that is damaged less bythe epidemic than anothersusceptible plant or population. Thepathogen's counterpart to the host'sgenetic resistance/susceptibility isavirulence/virulence. Thecounterpart in the pathogenpopulation to dilatory resistance inthe host population isaggressiveness.

79Management of HostPlant Resistance to BufferCrop Populations AgainstPathogen PopulationsThe natural ecosystem of Israelabounds in progenitors of cultivatedwheat, oats, and barley. They areattacked by the same arrays ofpathogens that plague these crops inagroecosystems, yet epidemics arerare; the co-evolved populations havereached a dynamic balance and croptraits (multiple disease resistance,large seed, excellent grain quality,high protein percentage, etc.), highlydesired by breeders, have evolved.Such ecosystems yield not onlyvaluable germplasm, but knowledgeof how plants in natural ecosystemsprotect themselves. As these are thesame biologic species as theirrespective cultivated counterparts, acentury of literature aids immediateinterpretation and application toagroecosystems. This has beenreviewed by Browning (4, 5) and bySegal et al. (29).Conclusions are that naturalecosystems have identifiableprotective mechanisms. These areprimarily the pathological-ecologicalphenotype of the host population'spolygenic general resistance ortolerance, and of its necessarycounterpart, the pathogen'spolygenic general pathogenicityaggressiveness.OligogeniC specificresistance/susceptibility andavirulence/virulence probablybecame superimposed duringevolutionary time over the basicpolygeniC systems primarilyresponsible for host-pathogenhomeostasis in indigenouspopulations. Oligogenic systems cancontribute in a Significant way tomaintaining homeostasis where bothpolygeniC and oligogenic resistanceoccur and where a protectivespaciogenic (D.S. Marshall, TexasAgricultural Experiment Station,Dallas, Texas, pers. comm.)population structure has co-evolved.Nature uses one resultant type ofepidemiologic resistance-dilatorytoprotect the population, but it usesmany types of genetic resistance andpopulation structures to achieve it.This had been called the protectionof indigenousness (4) and it can beemulated in agroecosystems. Thechallenge that faces agriculturalscience is to develop agroecosystemsthat utilize specific and generalgenetiC resistance to create cultivarsand/or spaciogenically stablepopulations with dilatoryepidemiologic resistance or tolerance.This was illustrated (25) with anepidemiological test of differentdegrees of tolerance and of differenttypes of resistance and populationstructures, which would tend tosimulate a naturally diversepopulation. Politowski and Browning(25) trapped urediospores of Pucciniacoronata at the periphery of large.isolated oat plots and plottedcumulative spore yields as a functionof time for their measurement ofrelative disease progress curves(Figure 1). They included: i) asusceptible check (C649. which isthe recurrent parent of themidseason Multiline M73 and of theisoline X421-1) that carried two genesfor specific resistance. both of whichwere match'ed by races in theexperiment; ii) an immune check(X421-1) that only measuredbackground noise because it carriedan effective gene for specificresistance from the wild oat. A venasterilis; iii) a cultivar (Portage) said tohave "partial resistance." possiblypolygenically inherited; iv) sevenpureline cultivars thought to havedegrees of polygenic tolerance tocrown rust; and v) two multilinecultivars. M73 and E74. The last twohad the protection of population

80-6 C649480---6 GRUNDY---0 O'BRIEN--.. OTEE400~ C237-891V---6 OTTER...;;;: ---{) NODAWAY 70 P'-(:1'0320 ,,rJJ...- - -0 CHEROKEECl)l:f...., I Pj , ,I------A MULTILINE M730 ?i p­O , I , ,......___ MULTILINE E74,enCl) ~ p'... 2400 , ...II0.. ---e PORTAGE~'rJ)P• ,,Cl),.:: ...., ~ X421-I /.­ ,ro"3 .­•E~E 160 ,::l , ~U, ~~4 8 12 16 20Days of CollectionFigure 1. Disease progress curves of Pucclnla coronata increase on eachof the 12 oat cultivars in the 1975 field experiments. Cumulative sporecounts per 100 liters of air are an average of three replications fitted tothe logistic equation. From ref. (25).

81buffering from crown rust by way ofthe heterogeneity characteristic ofmultilines (14). The results showedthat the multiline cultivars werehighly mixed in reaction to crownrust (as in Israel's indigenouspopulation that was the source ofseveral of the genes for specificresistance. some plants supportedconSiderable crown rust while otherswere free); X421-I was immune;Portage was moderately resistant;and all other cultivars appearedsusceptible. The epidemiologicresponse. as shown by the diseaseprogress curves of all cultivars. bothpure lines (except C649 and X421-I)and multilines. however. was that ofa continuum of different degrees ofdilatory resistance-a slowing of therate of epidemic developmentrelative to the susceptible check.Several types of genetiC resistance(pureline specific resistance/susceptibility. multiline specificresistance/susceptibility. anddifferent levels of general resistance)were included in the experiment(25). Epidemiologically. the resistantand susceptible checks showed.respectively. discriminatoryresistance/susceptibility against. orin favor of. the rust races in theexperiment; all other cultivars(pureline and multiline) displayedonly different degrees of dilatoryresistance.In several respects. this experiment(25) simulated aspects of a naturalecosystem and illustrated theprinciple that nature uses primarilya single type of epidemiologicreSistance-dilatory resistance-toprotect populations. but that it usesmany different genetic systems andspaciogenic structures to achieve it.Specific resistance/susceptibilitycontributes to dilatory resistance innatural systems. as does the basicsystem for general resistance. Thelatter probably is the system mostresponsible for maintaininghomeostasis in indigenousecosystems. Specific resistance.superimposed on the generalresistance system. provides a feedbackmechanism and contributes tohomeostasis in a major way (5. 24).Much of our disillusionment withfrequently ephemeral specificresistance has come frommisunderstanding its role in natureand its epidemiological manifestationas discriminatory resistance. Hence.we have managed it ineffectively.Dilatory resistance is a common andeffective phenomenon in nature.Discriminatory resistance. however.so widely used in agriculturalsystems. is an artifact of agriculturalsystems. Natural populations do notnormally occur in the largehomogeneous areas required eitherto generate or receive largequantities of homogeneousinoculum. Mixed stands in nature. ora multiline cultivar. do filter withdiscrimination the small amounts ofincoming "seed" inoculum. but themain effect on pathogen increase andspread will be that of dilatoryresistance in the naturallyheterogeneous plant populations.What then is the role of specificresistance/susceptibility in naturalsystems? Parlevliet (24) said that itsrole is not to protect the plant-thatis the role of general resistance-butto prevent the pathogen frombecoming too aggreSSive.Vanderplank (35) considered thatpathogen aggreSSiveness is thecounterpart of host horizontalresistance. An epidemiologic concept.aggreSSiveness is also thecounterpart of dilatory resistance.Thus. in Figure 1. host cultivarMultiline M-73 can be described ashaving more dilatory resistance than.say. Otee. Also the pathogenpopulation can be described as being

82less aggressive on Multiline M-73than on Otee. Like resistance.aggressiveness is relative. Thedisease progress curves in Figure 1give the ecological carrying capacityof the different cultivar entries.Measured epidemiologically. Oteehad a carrying capacity for a largerpathogen population than hadMultiline M-73; Otee enabled thepathogen population to express itsaggressiveness more than MultilineM-73. regardless of the differentgenetic systems involved. The feedbackmechanism of using the finelytunedspecific resistance/susceptibility system to reduceaggressiveness in the pathogen. asillustrated in Figure 1. undoubtedlyco-evolved in the host-pathogensystem and contributed survivaladvantages to both the host and thepathogen in a geneticallyhomeostatic equilibrium.How should the host-pathogengenetic systems that result In theexpression of specific and generalresistance be managed? Clearly.specific resistance should be used. asin nature. only in diversepopulations and. preferably.embedded in an effective system forgeneral resistance (4. 24). How muchspecific resistance is needed toprotect a population? Our work (J.A.Browning and M.E. McDaniel.unpublished) with a I-resistant:2-susceptible. two-cultivar mixtureshowed that as little as one-thirdresistance is adequate to protect apopulation even in the long SouthTexas disease season. In anenvironment in which thehomogeneous susceptible cultivarwas killed prematurely by rust. In acoastal environment, even morefavorable for the disease. one-thirdresistance to an individual race stillseemed adequate. but the populationrequired more diversity. In thisexperiment. the susceptibleparent was killed by crown rust. butthe two Iowa multilines seemeduntouched even though half theircomponents were susceptible tostrains of the undefined pathogenpopulation.Similarly. dramatic protection ofwinter barley from powdery mildewwas experienced in the long diseaseseason in England with three-cultivarmixtures. making each populationtwo-thirds resistant (37). Clearly. notnearly as much resistance isrequired to protect a population fromhighly epidemic. variable. airborne.foliar pathogens as was forrnerlyandis still very commonly-thought.That these disease data arecorroborated by data from thenatural ecosystem (29) gives megreat confidence that the one-thirdfigure is real (4). The degree ofresistance desired can be obtained.as in nature. by deploying genes forspecific resistance in diversepopulations and. preferably. byembedding them in an effectivesecond basic system-that whichtends to result In general resistance.Use of DiversityBorlaug's (3) early wheat multilinedevelopment for the RockefellerFoundation Agricultural Program inMexico (th,at later was incorporatedinto CIMMYT) readied two backcrossseries for release in the early 1960s.However. they were withheldbecause the agronomic type wasoutclassed by CIMMYT's new highyielding.semidwarf, photoperiodinsensitivewheats. The firstmultilines used commercially werethe highly successful wheatsMiramar 63 and Miramar 65 releasedby the Rockefeller FoundationAgricultural Program in Colombia.Meanwhile. the semidwarfs becamewidely accepted and. by 1970.CIMMYT Cross 8156 had beenproved in commercial productionworldwide. To continue its multiline

83interest and diversify 8156. CIMMYTdeveloped mixture componentsbased on the 8156 genotype for useby national wheat breedingprograms. Wheat multilines areresearched in at least three suchprograms in India and MultilineKSML3 was released in the Punjabin 1979. Unrelated to CIMMYT or8156. yellow rust-resistant multilinesTumult and Crew were released in.respectively. The Netherlands in1979 and in Washington and Idaho.USA in 1982 (22).Our Iowa. USA. program released 13multiline cultivars of oats in twomaturity classes. These were grownon up to 0.4 million ha in Iowa andcontiguous states without report ofdamage from crown rust even insevere crown rust years. Tested alsoin the long disease season of theTexas coastal plain and in Israel.they responded Similarly. In1983-84. Iowa released a new oatmultiline on a higher yieldingrecurrent parent with neededresistance to the barley yellow dwarfvirus (13. 22).Wolfe (36. 37) has advocated andresearched three-cultivar mixtures ofbarley in the UK to buffer primarilyagainst powdery mildew.In the light of the foregoing. aquestion obviously arises as to therelative usefulness of multilines andbroader (eg. varietal) mixtures.Multilines. as in Iowa. giveagronomic uniformity. as industryrequired when the first Iowamultilines were released. But thesewere developed also for researchpurposes; there are so many benefitsfrom diversity (16. 30) that auniform agronomic background wasnecessary to research the role ofdiversity per se itself in controllingdisease. Since that question has beenanswered. I see no additionalbiological advantage to growingmultilines. They have excessiveuniformity. some of it for merelycosmetic traits where diversity wouldbenefit the crop population. Asuitable compromise between thelegitimate needs for agronomiCuniformity and the benefits ofdiversity against biotic and abioticsources of stresses are the threecultivarmixtures developed andresearched in England (36. 37).Constraints on the Useof DiversityDiversity per se has been reviewedadequately (7. 30) and more recently(22. 36). Each review established aclear need and rationale for diversity.The Rockefeller Foundation (27)recognized that "The multilinetheory for the production ofcomposite varieties is one of thetruly new concepts of the century inbreeding self-pollinated crops." For along time. diversity has been knownto buffer against all of the mosthighly epidemic foliar pathogens ofsmall grains (7) except Pyriculariaoryzae that causes rice blast. FinallyChin (11) showed that rice blast. too.could be controlled with mixtures ifthe right host genotypes werechosen. This illustrates that. to effectcontrol of any crowd disease. theresistance genes used must befunctional (28) and relevant (36) tothe pathogen population. Diversityhas been shown to be effectiveagainst a non-specific pathogen(Septoria nodorum) of wheat (15) andagainst a soilborne pathogen of oats(2). It has protected soybeans fromthe cyst nematode (10) and wheatfrom Hessian fly (32). It has evenbenefited yield when different barleycultivars were mixed that carried thesame major-effect gene againstpowdery mildew (J. K. M. Brown.Plant Breeding Institute. Cambridge,pers. comm.).

84After the 1970 SCLB pandemic. theNAS (23) targeted uniformity as the"crux of genetic vulnerability" andUllstrup (33. p. 46) warned"Diversity must be maintained inboth the genetic and cytoplasmicconstitution ofall important cropspecies" (his emphasis). The sectionon "Genetic and EpidemiologyConcept of Resistance" above built acase for the use of diversity in someform if it is indicated. i.e. if diseaseor other sources of stress are seriousenough to justify it, and the genesfor the job are available. The sectionthat followed compared diverseagricultural populations with anindigenous ecosystem anddemonstrated its epidemiologicalnaturalness. which portends trulydurable resistance. Experience, as inIowa and Colombia, has proved itssuccess.With the abundance of theory, of"bio-Iogic," of data, and ofexperience from both naturalsystems and agroecosystemssupporting the use of diversity. whyhas so little diversity actually beenused? What are the problems? Isuggest that they are these three:• A paradigmatic one.• Our failure to communicateconvincingly that diversity canbuffer a crop population againstdisease or another source of stressand especially to explainadequately and logically itsmechanism of action.• Other, including unjustified fear ofpotential superraces and optimism(some of it justified) for theeventual success of current effortsto effect durable resistance byaccumulating minor-effect genes,pyramiding major-effect genes.etc.Space will only allow me to discussthe first two constraints to the use ofdiversity. which I will do from theperspective of a professional careeras one of its advocates. Of the threeconstraints, the first may be themost significant and the least readilysolvable.The paradigmatic problemThe collective thinking of a peergroup of scientists or the state of theart of an industry may be describedas their paradigm (19). and it is verydifficult to move beyond it. Thus, aparadigm is much more than a clearexample or pattern. It is anunderlying basic idea of what we aretrying to understand. And ourparadigm. our underlying basic idea.may be so strong and fixed that itmay actually prevent us fromunderstanding what we purport to betrying to understand. Our idea of thenature of our subject prompts us tomake certain inquiries, not to makeothers, and to interpret all from theperspective of our paradigm. AsKuhn (19) wrote. even in thesciences, fundamental ideas aboutthe way things work gUide ourseeing, including what we see fromthe results of our experiments,rather than simply emerging fromwhat we have seen.To take a, now-ridiculous example, aperson convinced that the world isflat (or round) would ask questionsand interpret answers from theparadigm of his underlying basicbelief about that world. If a personand his peers are sufficientlyconvinced, albeit subconsciously,that the world is flat, say, evenserious inquiry is probably going toresult in reinforcing that belief. Theindividual must take a determined.conscious. sometimes painful effortto keep this from being true. TheUSA's creation-evolution controversycontinues to be a current andrelevant example. In both examples,

85two irreconcilable paradigms operate.What Kuhn (19) called a paradigmconversion is indicated.Our view of diversity versusuniformity per se is another exampleof irreconcilable paradigms in whicha paradigm conversion is indicated. Iexperienced a personal and veryrelevant paradigm conversion in myearly professional years. I was asenamored of a pure-culture. purelinephilosophy as any other plantpathologist or plant breeder. Then.while still in graduate school atCornell University. plant breeder NeilJensen (16) and phytopathologistand major professor George Kentchallenged this. A long. slow. andpainful transition started and had torun its course before I could acceptdiversity in principle. Before that, allmy inquiries reinforced myunderlying basic idea of the need forgenetic purity; subsequently. in theneed for diversity. For many yearswhile I was on the Iowa StateUniversity faculty researching anddeveloping multiline cultivars as amember of the oat project and.especially after I was privileged tostudy disease development in Israel'sindigenous populations. I referred towhat I called an "agroecosystembias" or mind-set that helpedprevent many excellent scientistsfrom seeing the need and potentialbenefits that we in Iowa (and a fewothers) saw in favor of more geneticdiversity in agricultural crops.Nowadays. I call that anagroecosystem paradigm. Theagroecosystem paradigm is very realand a number of concepts basic toplant pathology. such as the fitnessand formae speciales concepts. havebeen expressed as we know thembecause of the agroecosystemparadigm. These would have to beexpressed very differently to beconsistent with observations onpathogen development in bothnatural ecosystems andagroecosystems (6). Thus. theagroecosystem paradigm governsmost thinking in phytopathology andin plant breeding. It also governsmost thinking and activity aboutwestern-type agriculture andagricultural research in general.including what inquiries are made.what are the current grant-fundingareas. what grant proposals arerecommended to be funded by peerreviewpanels. and what we see as aresult of our inquiries. I believe thatthe agroecosystem paradigm ofuniformity for uniformity's sake isthe greatest constraint toresearching. understanding. andusing diversity in agroecosystems."The lesson to be learned seems tobe that society in general. andscientists in particular. must realizethat some changes of attitudes andemphasis will be required..... (23.p. 13).How diversityprotects a populationProbably the most sustained andsuccessful program developingheterogeneous cultivars is Iowa's oatmultiline program (8. 13). Yet,according to current thinking. Iowa'smultilines cannot effect control of anepidemic foliar pathogen in Iowa andthey cannot possibly do what theyhave done! This illustrates the extentto which ~e scientists have failed todiscern and communicate the meansby which diversity helps to controldisease. I will discuss this.As analyzed by Vanderplank (35), anepidemic is controlled by reducingdisease either at the outset (Xo) orlater by reducing its apparent rate ofincrease (r). Recent reviewers (22.36) stated. for example: "if onecompares a mixed population withthe mean of its components grownseparately and exposed to the samepathogen population. there should beno reduction in Xo" (22. p. 533).And. relatively to r. Wolfe (36) added

86that ..... with rust diseases of cerealsin ... the northern USA, epidemiclevels may be determined almostdirectly by the exogenous inoculumfrom further south: the number ofpathogen generations duringdevelopment may be only three orfour. The mixture may thus providelittle more protection than by simplediversification, with almost norestriction in disease spread." Thus,eliminating Xo and r leaves noknown mechanism to explain thedramatic success of the Iowamultilines, and little reason toencourage others to try diversity.Obviously, either the data have notbeen obtained, or the correctconcepts have not been formulated,or communication has failed.That the above quotation fromMundt and Browning (22) could notbe correct defies logiC. The statementuses the same criterion for yield ofthe pathogen or the amount ofdisease that is the standard measureof success of diversity, namely:compare data from the mixture withthe mean of the components in purestands.But the matter is academic. If afarmer grows, say, a three-cultivarmixture, he probably doesn't grow orobserve the components: a commoncomplaint by scientists is thatfarmers generally won't grow checktreatments.If the cultivars havedifferent resistances andsusceptibilities, the incominginoculum will be only 113 as effectiveas in a pure stand of one of thecomponent cuitivars. The number ofearly subfoci are Similarly reducedfrom the outset and this willcontribute to reducing r and x(amount of disease) and increasingprofit. And, as Wolfe (36) hasemphasized for yield, the farmercould not have predicted which, ifany, cultivar grown in a pure standwould have been protected. In thethree-way mixture, all benefited fromthe outset.For a diverse population to outyieldthe mean of its components (at leastwith near-isogenlc lines), there musthave been less disease, doing lessdamage, on the susceptible plants inthe mixture than in pure stands: thewhole becomes an entity that isgreater than the sum of the parts.But the amount of disease iscommonly deemed proportional tothe percentage of susceptible plants.This traces at least from Leonard(20) who found that "The amount ofrust in mixtures... in noninoculatedcheck plots was proportional to thepercentage of (susceptible) plants inthe mixture" (20, p. 1846). As thisresult does not mesh with myexpectation and is inconsistent withmy observations on the Iowamultilines, I am compelled to ask"why?" It seems to me that dataleading to this conclusion are anunrecognized artifact ofexperimentation and that they resultwhen one tests the spread of highlyepidemic pathogens in mixtures ofhost genotypes and 1) uses smallplots and extraneous inoculum, ii)moves and works among the plantsoften repeatedly, unwittinglyspreading inoculum with one'sclothing apd hands, and/or, iii)inoculates the mixtures withunnaturally large quantities ofinoculum.Take movement through plots andhandling of plants to count uredia,for instance. Fingers are recognizedas excellent for making greenhouseinoculations. But when they are usedto handle plants in the field to takedata on rust increase that ispresumed to result from naturalspread of the pathogen, they cangive spurious results. Once when myproject personnel were counting oatstem rust uredia repeatedly in purestands and mixtures of oat isolines, I

87observed that "workers handlingplants supplemented natural aircurrents in disseminating spores.Resistant plants served as effectivebarriers to air dissemination ofinoculum but not to workerdissemination. This was mostobvious near the foci and along entrypaths where worker activity wasgreater..... (7. p. 363-364). Theincreased amount of rust was visiblyand dramatically greater wherepeople worked through the plots. Wehad no choice but to discard thosedata. That was the last year weentered plots to take data on rustincrease in tests of diversity.Thereafter. we used Rotorod SporeSamplers outside the plots. trappedspores 2 hours daily. counted thespores. and plotted cumulative sporecounts/lOO liters of air as ourquantitative measure of the carryingcapacities of different populationsand the effectiveness of diversity inbuffering against an airbornepathogen (17). The spore yieldcurves in Figure 1. which are ourmeasure of disease progress. resultedfrom using this technique (25).The effect of each of the above threepractices is to telescope time.hastening in an unnatural way thetime taken by inoculum to reachsusceptible plants. so that inoculumquickly ceases to be limiting. Withnaturally windborne inoculum suchas urediospores of the rust fungi(pathogens with r of up to 50% perday) one cannot influence time andexpect a fair assessment of thefinely-tuned buffering effect ofdiversity. Both Browning and Frey(7) and Wolfe (36) analyzed theimportance of time in thefunctioning of diversity. Telescopingtime gives a monomolecular growthcurve. not a sigmoid curve.Experimental techniques that give amonomolecular curve are exceIlentfor the qualitative identification ofspecific resistance in a diseasenursery. But one must achievesigmoid growth curves to assessdilatory resistance. horizontalresistance in the sense ofVanderplank (35). or the benefits ofmixtures. Yet more authors continueto concern themselves withminimizing inter-plot contaminationthan with other aspects of managingexperiments to test diversity; one ormore of the three practices are stillused. and with the same conclusionof proportionality (1). Vanderplank(34) cautioned against using fieldtechniques that would lead to the socalled"cryptic error" and thusunderestimate the value of horizontalresistance. Experimental designs andtechniques can lead also to crypticerrors that underestimate the valueof diversity which epidemiologically.can be considered "synthetichorizontal resistance." Threeexperimental techniques that shouldavoid the cryptic error and evaluatediversity fairly are: i) estimatepathogen yield by trapping spores atthe periphery (or down mowedwalkways) of large. isolated plots(17); ii) estimate disease remotelywith multispectral sensors (D.S.Marshall. pers. comm.); and iii)estimate host yield conventionally.ConclusionsDiversity is the only defense againstthe unkno}Vn. as against a futuredisease threat. Use of adequatediversity in some form is also theonly way to justify the prediction ofgenetic protection of an extenSivelygrown crop in the current year. Withthe ease of mixing genotypes ratherthan breaking linkage groups.combining many polygenic systems.or pyramiding major-effect genes.diversity is also an easy and effectiveway of effecting future adjustmentsuch as Duvick (12) anticipates.Furthermore. diversity also offers ameans to confront exotic pathogenswith high hitchhiking potential andto ensure against an epidemicfollowing their introduction (23. 31).

88Clearly. diversity can benefit smallgrains greatly but more scientistsmust discover this for themselves inexperiments that avoid crypticerrors. This will contribute also toovercoming the serious paradigmaticconstraint to using diversity. Grainyield response of mixtures of nearisogeniclines of wheat or oatsaffected by disease is commonlynonlinear when compared to themeans of the components in purestands. Similarly. the yield of sporesor the amount of disease should benonlinear. This would be expectedbecause disproportionately less rust,causing disproportionately lessdamage. is generally present onsusceptible plants in mixtures thanin pure stands. If this is not so andthe amount of disease is strictlyproportional to the percentage ofsusceptible plants. I fear it may haveresulted from an artifact ofexperimental technique. This mayresult especially if "the observerinescapably became part of theobserved system" (18). as bysuperimposing a spore distributionsystem different from that which onepurports to be studying. The all-toofrequentreporting of linear responseis probably the second majordeterrent to using diversity.A degree of diversity whereby onethirdof the population is resistantseems to effect adequate protection.This goal is safely exceeded withsimple three-cultivar mixtures.Cultivar mixtures offer a bettercompromise than multilines betweenthe needs for uniformity and thebenefits of diversity and are. inprinciple. recommended. Diversityportends an extended useful lifeexpectancy for a valuable naturalresource-genes for specificresistance. Of great currentimportance. diversity portends thesame for the even more valuablespecific resistance genes that oneday will be incorporated viabiotechnology. Diversity offersconsiderable additional benefits topathology and breeding programsalike. at least if containing a highlyepidemic disease is a resourceconsumingobjective of the programs(8). Once this economy of genes andother resources is recognized. Ipredict a greatly expanded use ofdiversity. Diversity is stronglyindicated for centers of cultivarimprovement. "One cannot helpbeing thrilled at their prospects forhelping to feed a hungry world. noralarmed at their potential for gUidingthe evolution of major pathogens ona global scale" (7). Thus. diversityshould be among the genemanagementstrategies in therepertoire of responsible breedingprograms of international centersand of multi-national corporations.References1. Alexander. H.M .. A.P. Roelfs. andG. Cobbs. 1986. Effects of diseaseand plant competition on yield inmonocultures and mixtures of twowheat cultivars. Plant Pathology.35:457-465.2. Ayanru. D.K.G .. and J .A.Browning. 1977. Effect ofheterogeneous oat populations on theepiphytotic development of Victoriablight. New Phytologist. 79:613-623.3. Borlaug. N.E. 1959. The use ofmultilineal or composite varieties tocontrol airborne epidemic diseases ofself-pollinated crop plants.Proceedings of the First InternationalWheat Genetics Symposium.Manitoba 12-27.4. Browning. J.A. 1974. Relevanceof knowledge about naturalecosystems to development of pestmanagement programs for agroecosystems.Proceedings of theAmerican Phytopathological Society.1: 191-199.

895. Browning. J.A. 1980. Geneticprotective mechanisms of plantpathogenpopulations: their coevolutionand use in breeding forresistance. In: M. K. Harris (ed.).Biology and Breeding for Resistanceto Arthropods and Pathogens ofAgricultural Plants. TexasAgricultural Experiment StationMiscellaneous Publication.1451:52-75.6. Browning. J.A. 1981. The agroecosystem-naturalecosystemdichotomy and its impact onphytopathological concepts. In: J.M.Thresh (ed.). Pests. Pathogens. andVegetation: The Role of Weeds andWild Plants in the Ecology ofCropPests and Diseases. London. Pitman.159-172.7. Browning. J.A.. and K.J. Frey.1969. Multiline cultivars as a meansof disease control. Annual Review ofPhytopathology. 7:355-382.8. Browning. J.A.. and K.J. Frey.1981. The multiline concept intheory and practice. In: J.F. Jenkynand RT. Plumb (eds.). Strategies forthe Control of Cereal Diseases.Oxford. Blackwell. 37-46.9. Browning. J.A., M.D. Simons.and E. Torres. 1977. Managing hostgenes: epidemiologic and geneticconcepts. In: J.G. Horsfall and E.B.Cowling (eds.l. Plant Pathology: AnAdvanced Treatise. New York,Academic Press. 191-212.10. Caldwell, S.P., and O.J. Myers.1980. Seasonal field mapping ofsoybean cyst nematode populationsas influenced by cultivars andblends. In: F.T. Corbin (ed.), WorldSoybean Research Conference II,Boulder, CO.11. Chin, K.M., and A.M. Husin.1982. Rice variety mixtures indisease control. Proceedings of theInternational Conference on PlantProtection in the Tropics. KualaLumpur, 241-246.12. Duvick. D.N. 1984. Geneticdiversity in major farm crops on thefarm and in reserve. EconomicBotany. 38: 161-178.13. Frey. K.J. 1982. Breedingmultilines. In: I.K. Vasil, W. RScowcroft, and K. J. Frey (eds.l.Plant Improvement and Somatic CellGenetics. New York. Academic Press.43-72.14. Frey. K.J., J.A. Browning, andM.D. Simons. 1977. Managementsystems for host genes to controldisease loss. Annals of the New YorkAcademy of Sciences, 287:255-274.15. Jeger, M.J.. D.G. Jones. and E.Griffiths. 1981. Disease progress ofnon-specialized fungal pathogens inintraspecific mixed stands of cerealcultivars. II. Field experiments.Annals ofApplied Biology.98: 199-210.16. Jensen, N.F. 1952. Intra-varietaldiversification in oat breeding.Agronomy Journal, 44:30-34.17. Jowett: D., J.A. Browning, andB.C. Haning. 1974. Nonlinear diseaseprogress curves. In : J. Kranz (ed.).Epidemics of Plant Diseases:Mathematical Analysis and Modeling.Berlin, Springer-Verlag, 115-136.18. Kroon. RP. 1987. The observedand the observer. AmericanScientist. 75:229-230.19. Kuhn. T.S. 1970. The Structureof Scientific Revolutions. Chicago.University Press (2nd ed.).

9020. Leonard, K.J. 1969. Factorsaffecting rates of stem rust increasein mixed plantings of susceptible andresistant oat varieties.Phytopathology, 59: 1845-1850.21. Marshall, D.S. 1987.Characteristics of the 1984-85 wheatleaf rust epidemic in Central Texas.Plant Disease (in press).22. Mundt, C.C., and J.A. Browning.1985. Genetic diversity and cerealrust management. In: A.P. Roelfsand W.R Bushnell (eds.), The CerealRusts, II. New York, Academic Press,527-560.23. National Academy of Sciences.1972. Genetic Vulnerability of MajorCrops. U.S. National Academy ofSciences, Washington.24. Parlevliet, J.E. 1981. Diseaseresistance in plants and itsconsequences for plant breeding. In:K.J. Frey (ed.), Plant Breeding II.Ames, Iowa, University Press,309-364.25. Politowski, K., and J.A.Browning. 1978. Tolerance andresistance to plant disease: Anepidemiological study.Phytopathology, 68: 1177-1185.26. Robinson, RA. 1976. PlantPathosystems. Berlin, Springer­Verlag.27. Rockefeller Foundation. 1963.Annual Report of Program inAgricultural Sciences, 1962-63.28. Schmidt, RA. 1978. Diseases inforest ecosystems: The importance offunctional diversity. In: J.G. Horsfalland E.B. Cowling (eds.), PlantDisease: An Advanced Treatise, 2.New York, Academic Press, 287-315.29. Segal, A., J. Manisterki, G.Fishbeck, and 1. Wahl. 1980. Howplant populations defend themselvesin natural ecosystems. In: J. G.Horsfall and E.B. Cowling (eds.)Plant Disease: An Advanced Treatise,5. New York, Academic Press,75-lO2.30. Simmonds, N.W. 1962.Variability in crop plants, its use andconservation. Biological Reviews ofthe Cambridge Philosophical Society,37: 422-465.31. Simons, M.D., and J.A.Browning. 1983. Buying insuranceagainst exotic plant pathogens. In:C.L. Wilson and C.L. Graham (eds.).Exotic Plant Pests and NorthAmerican Agriculture. New York,Academic Press, 449-478.32. Suneson, C.A. 1960. Geneticdiversity-a protection against plantdiseases and insects. AgronomyJournal, 52:319-321.33. Ullstrup, A.J. 1972. The impactsof the southern corn leaf blightepidemics of 1970-71. AnnualReview ofPhytopathology, lO:37-50.34. Vanderplank, J.E. 1963. PlantDiseases: Epidemics and Control.New York, Academic Press.35. Vanderplank, J.E. 1968. DiseaseResistance in Plants. New York,Academic Press.36. Wolfe, M.S. 1985. The currentstatus and prospects of multilinecultivars and variety mixtures fordisease resistance. Annual Review ofPhytopathology, 23:251-273.37. Wolfe, M.S., and J.A. Barrett.1980. Can we lead the pathogenastray? Plant Disease, 64:148-155.

Chapter 8The Use of Variety Mixtures toControl Diseases and Stabilize YieldM.S. Wolfe. Plant Breeding Institute. Cambridge. EnglandAbstractThe major factors leading to loss ofeffectiveness of disease resistance andfungicides in current European agriculture are briefly described. Among theoptions available to the breeder and farmer to improve the situation. the useof variety mixtures is discussed in some detail. Considerable evidence fromfield trials points to the advantages of disease control, yield increase, andyield stability from this simple system. which can be added to any othermethod of disease control.IntroductionWithin the advanced agriculturalsystem of northwestern Europe.there is continuous erosion of theeffectiveness of qualitative andquantitative host resistance and offungicides by the cereal mildewpathogens. The main reasons for thisare summarized in the followingparagraphs. First. there areuncontrollable factors intrinsic to thesituation. The disease is endemicthroughout most of the area. thepathogen has a regular andfunctional sexual cycle. and thespores are widely distributed by thewind system. Then there are severalcontrollable factors listed as follows:• Agronomic features-Monocultureof cereals. particularly ofwheat. tends to be predominant.and is based mostly on thecultivation of very few. highlypurified varieties. often in largefields. Large amounts of inorganicnitrogen fertilizer are used. greatlyincreasing the potentialsusceptibility of the uniform crops.The general trend to earliersowing of winter wheat alsoencourages most diseases.• Resistance breeding-Breedersdepend largely on rapid replacement of qualitative resistances that are rapidly 91overcome because of the factorsnoted above. Quantitativeresistance is also used but thedurability of such resistance isoften unknown simply because ofthe technical difficulty ofconfirming its erosion by thepathogen.• Fungicide use-Because of theproblem of obtaining andmaintaining effective hostresistance, farmers have beenpersuaded to use fungiCidesextensively. Unfortunately. thereis little diversity among thechemicals available and little morein prospect. As a result. they. too.are steadily losing effectivenessunder intense selection.To progresS from this depressingEuropean picture. the followingdiscussion first summarizes therange of options available for diseasecontrol. The second part considersspecifically the option of using mixedvarieties in cereal cropping.Current Optionsfor Disease ControlOptions for the breederCereal breeders have to reconcilemany different breeding objectives.Within such a framework. the

92resources available for improvingdisease resistance are limited andthe following methods prevail.Reserve strategy-The continuousintroduction of simply inheritedqualitative resistances probablyrepresents the simplest and mostreliable procedure for the breedersince the resistance is easy torecognize and causes leastdisturbance to the achievement ofother breeding objectives. Thisstrategy depends on a ready supplyof different resistance genes of whichmany are available for control ofbarley mildew. but not for wheatmildew. It also contributes to yieldvariation for the farmer because ofpathogen response to the resistances.Durable resistance-Among thesimply inherited genes for qualitativeresistance to mildew. the ml-o genein barley has proved to be unusuallydurable. A pathogen response to thegene was not detected in the fielduntil 1986 (8). about 8 years after itsintroduction. Even then. it wassmall; apparently. several genes areneeded in the pathogen to overcomethe single host gene. Most othersingle resistance genes have begunto lose effectiveness within 2 to 5years of introduction.Partial or quantitative resistance.usually in the form of adult plantresistance. may have been morevariable in durability. For example.several wheat varieties with Pm2mildew resistance became highlysusceptible at the seedling stageshortly after introduction. but theystill provide adequate resistance atthe adult stage. There are manyother instances of varieties withquantitative resistance that becamemore susceptible within a shortperiod of general use. Unfortunately.there are no rigorous tests availableeither of the mode of inheritance ofquantitative resistance or of anypathogen response to it.Re-cycling resistance genes­Except in rare circumstances. recyclinga variety that has becomesusceptible is unlikely to besuccessful. However. if the resistanceis hybridized into a new variety witha genetic background selected in thepresence of the matching virulence.then the combination of theresistance and the new backgroundmay provide useful protection (10).This assumes that. at the time ofintroduction of the new variety withthe re-cycled resistance. thefrequency of the matching virulencehas declined to a low level.Combining resistance genes-Theprocedure in Australia and NorthAmerica. of releasing varieties withcombined resistance based onknowledge of the structure of thepathogen population. can be highlysuccessful. The procedure of simplyrecombining defeated resistancegenes that have been used separatelyis disastrous (11). In this case. thematching virulence genes are alreadyin adapted backgrounds. so that onlya simple recombination is needed forthe pathogen to overcome thecombined host resistance and tocause increased infection of thevarieties with single resistancecharacters.In summ.ary. the methods describedare used to produce varietiesintended for large-scale use inintensive agriculture. The first. thereserve strategy. assumes that theresistances will not be durable. butthat they may last as long as theexpected commercial lives ofvarieties. However. a system ofvariety use that prolongs theeffectiveness of these resistancesmay be more economical ofresources and of greater benefit tothe farmer in terms of stability ofproduction.

93The other methods are used in thehope that resistance will persist.However. we are still unable topredict the durability of resistance ofany variety ahead of the only knowntest for this character. of majorexposure of the variety for a longperiod (4). A system of varietal usethat protects the variety from thisstringent test will therefore help tomaintain the effectiveness of itsresistance.Options for the farmerThe breeder produces varieties foruse in existing systems of agricultureon which he has little directinfluence. The systems that are usedtend to arise as compromisesbetween conflicting needs; they varyconsiderably in their influence ondiseases:Reserve strategy-This depends ona readily available supply of resistantvarieties and the assumption thatnone is likely to remain incommercial production for more thana few years. However. resistancegenes may be in short supply. as forwheat mildew. and the claims thatnew varieties are always better thanolder ones may sometimes bedubious. Constant change is notliked by the farmer. partly becauseof cost and partly because he wishesto learn how best to grow aparticularly variety in order torestrict the variation in itsperformance. This process may takeseveral years.Crop rotation-This can be aneffective strategy. particularly forsoilborne diseases. The longestpossible rotation will providediversification of cropping both intime and space. particularly if fieldsize is small. thus limiting the size ofpotential sources of inoculum.Unfortunately. demand for differentcrops is uneven and. where a majorcereal such as wheat or rice becomesdominant. this restricts thepossibilities for rotation.Interestingly. the average yield ofwheat in Europe has risen to a levelthat now allows the possibility forimproved rotations; it is not certain.however. that this opportunity willbe exploited.Varietal diversification (betweencrops)-Whether or not rotations arefeasible. a farmer should growdifferent varieties of any crop speciesin different fields. ensuring as far aspossible that the varieties differ inresistance to a particular disease.This form of diversification has theobvious merit of insurance; it isunlikely that a new race of apathogen able to overcome severaldifferent varieties will emerge andincrease immediately.It is also argued (7) thatdiversification among fields will slowdown epidemic development relativeto cultivation of a single variety. butthe extent of such an effect isdebatable. It has been argued thatinteraction between pathogenpopulations in adjacent fields occursonly during the early period ofestablishment of an epidemic (12).However. more recent data suggestthat migrating spore clouds mayhave a considerable influence over alarge area d\}ring the whole cropcycle (5). .Varietal diversification (withincrops)-Mixing varieties withdifferent levels and kinds ofresistance offers a simple method ofensuring diversification andinteraction between neighboringplants on disease progress. Themethod can be added to any of theother forms of disease control toobtain the benefits of complementaryinteraction. This option is discussedin the section below on "The VarietyMixture Option."

94Use of fungicides-Farmers areadvised not to use fungicidesprophylactically but. because theyare averse to risk and commonly findit difficult to spray on the correctday. they often do not follow theadvice. Unfortunately. the tendencyto prophylactic treatment increasesselection for fungicide insensitivity inpathogen populations. Farmers arealso encouraged to diversify theirtreatments among the availablefungicides but: a) there are very fewdistinct fungicide groups amongwhich to diversify; b) the majortriazole group has a broad spectrumof activity thus encouraging repeateduse and continuous selection forinsensitivity; and c) manufacturersnow encourage the use of mixturesof fungicides that have already beenwidely used separately. This maylead to rapid selection of pathogengenotypes with complexcombinations of fungicideinsensitivity characters (15).However. farmers are becoming moreconscious of the cost of pesticidesand of public concern about theiruse. The consequences of theseattitudes could be to relax selectionfor fungicide insensitivity and toplace a higher premium on inheriteddisease resistance.T~e _Variety _ .~Ii~ture OptionSome of the work on varietymixtures and multilines has beenreviewed recently (9). This sectionhighlights some of the moreimportant features of the review.relevant to disease control inmixtures and their potential use inwheat cultivation.The mechanisms of diseasecontrol in mixturesTo restrict the spread of an airbornepathogen in a variety mixturerelative to the mean spread in thecomponents grown as pure standsrequires only a difference inresistance between the components.Such comparisons should involvegeometric rather than arithmeticmeans (3). but even then. spread inthe mixture can be reduced. Asimple explanation is that thereaction of a more resistantcomponent is less affected byvariation in inoculum density thanthat of a more susceptiblecomponent. Thus. the moresusceptible component becomes lessinfected to a greater degree than themore resistant component becomesmore infected. The net effect ofreduced infection will become morepronounced as the distance betweenthe susceptible plants is increased.The phenomenon can be exploitedeven more effectively in restrictingpathogens that exhibit specificadaptation to resistant hosts,assuming that the pathogen tends toadapt singly to different hosts ratherthan to combinations of hosts (14).Under these conditions, thesusceptibility of host A to onefraction of the population isrestricted by the presence of host B,which is resistant to that fraction.Conversely, the susceptibility of hostB to a different fraction is restrictedby the presence of host A which isresistant to that fraction. Theaddition of other resistant hostsincreases the effect, principallybecause of the increased separationof plants with the same genotype.Each addition, however, has asmaller effect and there is probablylittle to be gained from increasingthe number of components beyondfive.Disease controlThe net effect of interaction betweenseveral hosts can be considerable;disease may be reduced to less than5 % of expectation under favorableconditions (16). Commonly, in a

95three-component mixture of springbarley varieties, the level of powderymildew infection is about half that ofthe mean of the components grownalone.Trials with spring barley and mildewinfection indicated the necessity forintimate mixing of the components.Indeed, there were indications that ahigh seed rate was desirable for themixture so as to reduce the averagetiller number and thus increase theinteraction between neighboringplants of different resistancegenotype. More recently, it has beensuggested (6) that the appropriatespatial arrangement of thecomponents of a mixture depends onwhether or not the pathogen initiatesthe epidemic in separate foci. If itdoes so, implying a slow start fromfew initial points in the field, thenthe spatial arrangement of the hostsis not critical; relatively large hostunit areas will be as effective assmall areas. If foci are not evident, asis true of barley mildew, implying arapid start to the epidemic frommany pOints, then intimate mixingof the host varieties is essential.Since the focal nature of epidemics islikely to be environmentally variableand therefore unpredictable, thesafest procedure may be to opt forintimate mixing.Most of the evidence for diseasecontrol in mixtures has come fromtrials with crown rust of oats (2) andpowdery mildew of spring barley(13). However, evidence fromelsewhere suggests that some controlof airborne pathogens may beexpected from appropriate mixturesof any crop.Mixtures and durabilityof disease controlIt has often been argued that varietymixtures or multilines would selectrapidly for recombinants in thepathogen population that wouldovercome all components. This maybe true for multilines in which thecomponents are near-identical andthe resistance genes have beenexposed previously to the pathogenpopulation. However, if the pathogenrace adapted to all components is atsome competitive disadvantage on aparticular component against therace adapted only to thatcomponent, then the outcome is notpredictable (1). Indeed, there has sofar been no field evidence for aconsistent increase in a complex raceon a variety mixture (10).Furthermore, if a single race able toinfect all components does becomepredominant, it is unlikely to attackall components to the same degree;there may be residual resistance insome. In this case, the level ofinfection will tend towards that ofthe most resistant component.To help to guard against a possibleloss in effectiveness of a widelygrownmixture, it is stronglyrecommended to diversify thecomposition of mixtures. Thisrequirement adds to the desirabilityof using mixtures with only fewcomponents since these can bematched easily for quality andharvest maturity and newcomponents can readily beintroduced.The effect of disease control onyield: mean yieldThere are two options for comparingthe yield of a mixture with that ofpure crops: use either the mean ofthe components or the yield of thebest component. The argument forthe latter is supposedly practical; thecomparison should be maderigorously between mixture yield andthe farmer's best available option.However, this presupposes that it ispossible to predict which componentwill be the best yielding in a future

96environment. Statistical analysis,supported by practical experience,shows that this is generally notpossible, even if large bodies of trialsdata were available. For most areasof the world, such data are notavailable, so that the only sensibleoption is to compare mixture yieldwith the mean of the componentsgrown as pure stands.During 11 years of trials at the PlantBreeding Institute, more than 150mixtures, mostly with threecomponents, were compared withtheir components (17). Among these,122 were described as well-chosenmixtures in that they providedeffective control of powdery mildew(other diseases occurred onlyoccasionally and in small amounts).These mixtures provided a meanyield increase of 8 %, with most ofthe individual yields distributedaround those of the best components(Table 1).The poorly chosen mixtures weredefined as those composed of highlysusceptible components such thatthe disease control that did occurwas insufficient to protect themixture from heavy infection. Evenunder these conditions, the yielddistribution of the mixtures wasbetter than that of the individualpure stands.The effect of disease control onyield stabilityStability of yield is often said to be ofmore importance to a farmer than ishigh yield. It is therefore necessaryto consider the stability of the highyields of mixtures. Among the trialssummarized in Table I, there weresome in which the same mixturesand their components werecompared over differentenvironments, mostly among years(17). From these data, it was shownthat the yields of three-componentmixtures were about as stable as themean yields of their componentsgrown as sets of three pure cultures,equivalent to the strategy of varietydiversification. Mixing anddiversification provided more stableyields than pure culture of singlevarieties (Figures 1 a-c). A similardistribution of results was obtainedfor a set of 12 winter wheat mixturesin 1986 even though there was littledisease. Individual trial effects werenot significant but, in allcomparisons, the mixture yieldexceeded that of the mean of thecomponents.The lack of data on yield stability ofa single mixture grown in diverselocations provided part of thestimulus for a project that is beingundertaken by the author and Dr.H.J. Dubin. The intention is to runtrials at many sites internationallyover 3 years, using widely adaptedTable 1. Comparisons of yield data for well-chosen and less well-chosenmixtures of three varieties; yields in t/haChoiceMixtureNo. Av. yieldAv. yield ofpure components% gainMixture better thanO 1 2 3componentsGood 122 5.61 5.20 8*** 0 12 50 60Poor 30 4.66 4 .58 2* 4 6 13 7

97CIMMYT wheat varieties that differin their disease resistancecharacters.32 --Claret (Cl ........ Egmont (El ----- Goldmarker (Gl1- - Triumph (Tl1980 1981 1982 1983 1985 1986Figure lao Mean yields (tlha) eachof four spring barley varietiesgrown as pure stand in each of sixtrial years.Summary of advantagesof variety mixturesThe principal advantages of varietymixtures in relation to diseasecontrol, mean yield and yieldstability have been described above.In addition, it should be pOinted outthat mixtures may provide someadvantage relative to pure stands inresponse to non-airborne pathogensand abiotic stresses if there is somevariation in the mixture with respectto the stress. This is becauseneighboring plants less affected bythe stress may be able tocompensate in terms of yield forthose that are more affected; thiscannot happen in a pure standunless the stress only affects randomindividuals.76542 --CEG, ..•..···CET. 1 -----CGT--EGT--CEG,·..·····CET,-----CGT--EGT1980 1981 1982 1984 1985 1986 1980 1981 1982 1984 1985 1986Figure 1b. Mean yields (tlha) of fourspring barley varieties grown aspure stands but expressed as theaverage for each of the fourpossible sets of three permutatedfrom them. Each line thusexpresses the effect of diversifyingamong three varieties.Figure lc. Mean yields (t/ha) ofeach of the four possible mixturesof three components that can bederived from four varieties.Note: The horizontal line in Figs. 1 a-crepresents the mean yield over the wholeperiod.

98In relation to airborne pathogens, thecomposition of a mixture can bearranged to provide control of arange of diseases in a way that isdifficult to achieve in a single hostgenotype (Table 2). From this table,the three-component mixturesappear to be especially suitable formildew control because they are allreasonably resistant and theresistances are derived from differentgenes. Against each of the rustdiseases, a different pair ofcomponents is resistant, so that themixture should be highly resistant toboth rusts.By using mixtures, and particularlyif a range of different mixtures isprovided, the farmer is forced intovariety diversification with littleeffort on his part. The greater thearea that is occupied in such astrate~y, the larger will be theoveral1 benefit, relative to large-scalepure culture, as the size of eachpathogen population is dampeddown.The potential advantages ofcultivating mixtures on a large-scaleappear to have been recognized firstin the German Democratic Republic(GDR) where 60% of that country'smalting barley crop is now producedfrom mixtures. The popularity of thisapproach has been influenced by thehigh cost of fungicides in the GDR;for a similar reason, it is likely thatPoland will soon have a large area ofvariety mixtures. Interest has beenslower to develop in western Europe,although the area of mixtures inDenmark is considerable, helped bypublic concern against the over-useof pesticides.Summary of disadvantages ofvariety mixturesOther than the reluctance,particularly on the part of the seedtrade, to change any feature of theagricultural system, the maindisadvantage of mixtures lies in theacceptance of the product by largescaleusers of grain. However, thetechnical reasons are often notTable 2. Disease resistance characters of three winter wheat varieties. Ahigh number denotes resistanceDisease Brock Norman RendezvousPowdery mildew 8 7 7Yellow rust 7 4 9Leaf rust 4 8 9Septoria nodorumblotch 7 5 7Septoria tritici blotch 6 4 6Eyespot 4 5 8Data from the NIAB leaflet. Recommended List of Cereal Varieties, 1987

99strong and may be removed if alimited range of mixed grain were tobe produced in large quantity; thisappears to be the case in the GDRAgronomically, there are fewdifficulties. It is important to ensurethat the range of harvest maturity ofthe mixture components is less thanthe potential range of harvest dates;where herbicides are used, thecomponents must have the samerange of sensitivities.Unfortunately, it is usually necessaryto reconstitute the mixtures after oneor two cycles of production becauseof drift in their composition. Suchchanges often result from thecompetitive action of the mostvigorous component, frequentlyunrelated to disease response.References1. Barrett, J.A. 1980. Pathogenevolution in multilines and varietymixtures. Zeitschrift fiirPflanzankrankheit undPflanzenschiitz, 87:383-396.2. Browning, J.A., K.J. Frey, M.E.McDaniel, M.D. Simons, and I. Wahl.1979. The bio-logic of usingmultilines to buffer pathogenpopulations and prevent disease loss.Indian Journal of Genetics and PlantBreeding, 39:3-9.3 . Jeger, M.,J, D.G. Jones, and E.Griffiths. 1981. Disease progress ofnon-specialized pathogens inintraspecific mixed stands of cerealcultivars. II. Field experiments.Annals of Applied Biology,98: 199-210.5 . Limpert, E., and G. Fischbeck.1987. Distribution of virulence andof fungicide resistance in theEuropean barley mildew population.In: Integrated Control to Reduce theDamage Caused by Cereal Mildews.Eds., M.S. Wolfe and E. Limpert.CEC Workshop, Weihenstephan,9-20.6. Mundt, C.C., and K.J. Leonard.1986. Analysis of factors affectingdisease increase and spread inmixtures of immune and susceptibleplants in computer-simulatedepidemics. Phytopathology,76:832-840.7 . Priestley, RH., and RA. Bayles.1982. Evidence that varietaldiversification can reduce the spreadof cereal disease. Journal of theNational Institute of AgriculturalBotany, 16:31-38.8. Schwarzbach, E., 1987. Shifts toincreased pathogenicity on ml-ovarieties. In: Integrated Control ofCereal Mildews: Monitoring thePathogen. Eds., M.S. Wolfe and E.Limpert. CEC Workshop,Weihenstephan 5-7.9 . Wolfe. M.S . 1985. The currentstatus and prospects of multilinecultivars and variety mixtures fordisease resistance. Annual Review ofPhy topathology. 23:251-273.10. Wolfe. M.S. 1987. Trying tounderstand and control powderymildew. In: Populations of PlantPathogens: their Dynamics andGenetics. Eds.. M.S. Wolfe and C.E.Caten, pp. 253-273. Oxford:Blackwell Scientific Publications.4. Johnson, R 1984. A criticalanalysis of durable resistance.Annual Review of Phytopathology,22:309-330.

10011. Wolfe, M.S ., and J.A. Barrett.1976. The influence andmanagement of host resistance oncontrol of powdery mildew on barley.Proceedings of the 3rd InternationalBarley Genetics Symposium,Garching, 433-439.12. Wolfe, M.S., and E .Schwarzbach. 1978. Patterns of racechanges in powdery mildews.Annual Review of Phytopathology,16:159-180.13. Wolfe, M.S., J .A. Barrett, andJ.E.E. Jenkins. 1981. The use ofcultivar mixtures for disease control.In: Strategies for the Control of ",Cereal Diseases. Eds., J . F. Jenkynand R.T. Plumb, pp 73-80. Oxford:Blackwell Scientific Publications.14. Wolfe, M.S., J.A. Barrett, andS .E. Slater. 1983. Pathogen fitness incereal mildews. In: DurableResistance in Crops, Eds., F.Lamberti, J .M. Waller, and N.A. Vander Graaff, pp 81-100. New York:Plenum Press.15. Wolfe, M.S., S.E. Slater, and P.N.Minchin. 1987. Annual Report of theCereal Pathogen Virulence Survey.1986.16. Wolfe. M.S., P.N. Minchin, andS.E. Slater. 1985. Annual Report ofthe Plant Breeding Institute.1984:93-95.17. Wolfe, M.S., A.J. Wright. andP.N. Minchin. 1987. Deployment ofcereal varieties: the farmer'sproblem. Journal of AgriculturalScience (in press).

Chapter 9Current CIMMYT Approaches inBreeding Wheat for Rust Resistances. Rajaram, RP. Singh, and E. Torres, Wheat Program, CIMMYT, MexicoAbstractWheat varieties derived from CIMMYT germplasm are grown on more than50 million ha in the developing world. Because these materials are grown onsuch a large area which probably will increase in the future, the CIMMYTbreeding policy has been to maintain and enhance diversity of rustresistance in wheat germplasm. International multilocational testing (IMT)has made an important contribution in ascertaining this genetic diversity. Inaddition to IMT, CIMMYT employs some genetic analyses in its improvementstrategy. CIMMYT believes that utilization of major gene, pathotype-specific,vertical resistance (VR) as described by Simmonds (Chapter 10) could lead toprecarious situations; as an alternative to the endless incorporation ofresistance genes or gene combinations, CIMMYT has pursued andrecommends breeding for what Simmonds describes in Chapter 10 aspolygenic, pathotype-non-specific horizontal resistance (HR) which portendsdurability of resistance. In a global context. durable resistance (or stability)and genetic diversity are ofparamount importance in CIMMYT's breedingprogram. The ideal situation would be to identify a gene or set ofgenes thatmay prove to have provided durable resistance as a foundation and thencontinually combine additional genes for resistance to ensure genetiCdiversity. Stem rust resistance (Sr2 complex) derived from the variety Hopeand leaf rust resistance (Lr13 complex) derived from the variety Fron tanaare the foundation ofresistance durability to these two diseases in CIMMYTgermplasm. For yellow (stripe) rust. the CIMMYT-bred variety Anza has beenreported to have durable resistance to the disease. CIMMYT routinelyidentifies lines with partial resistance (slow rusting) in the field and believesthis endeavor has made important contributions to sustainability of wheatyields worldwide. CIMMYT has pursued multiple major gene-based resistanceonly as a supplementary strategy and is actively looking into the revival ofthe development and use of multiline composites/cultivar mixtures.IntroductionThe CIMMYT wheat improvementprogram produces high-yielding,broadly adapted, rust-resistantgermplasm for the less developedcountries (LDCs). In 1967-68 highyieldingsemidwarf wheats weregrown on about 5 million ha in theLDCs; by 1982-83 this area hadincreased to 50.7 million ha (3). Mostof these wheats are either CIMMYTmaterials or are lines derived fromcrosses with CIMMYT germplasm innational programs.101Both yield potential and productivityof wheat have steadily increased inmany agroclimatic environmentswith genetic advance for diseaseresistance playing a principal role. Inthe southern part of Sonora. Mexico,yield potential for bread wheats inthis favorable environment doubledfrom 3000 kg/ha to 6000 kg/habetween 1950 and 1960. From 1960to 1980, yield potential increasedfurther by at least 100 kg per year.Most varieties released in the 1980shave yielded more than 8000 kg/hain the Yaqui Valley of Sonora. In afavorable year. it is not unusual torealize as much as 9000 kg/ha.

102Due to those yield gains andsubsequent increases in production.especially in optimum growingenvironments. the sustainability ofyield is a central and continuingobjective of the CIMMYT wheatbreeding programs. The potentialgenetic vulnerability to mutablepathogens. particularly the rustfungi. warrants top priority forresearch. especially in regard towhich type of resistance should bechosen and which breedingmethodology will be used to achievethat resistance. Although there havebeen sporadic. isolated epidemics ofleaf rust on semidwarfs. it isnoteworthy that major rustpandemics. phenomena thatfrequently occurred in the past onwheat land races in developingcountries. have not been reportedsince the large-scale introduction ofsemidwarf varieties 20 years ago.The durable resistance to stem rustand leaf rust has been achieveddespite the fact that more than 50million ha of semidwarfs have beenplanted in an environment that isvery conducive to the developmentof these diseases. Furthermore. inthe Indian Subcontinentapproximately 7 million ha aregrown to one variety. Sonalika. thatis resistant to stem rust. Prior to theintroduction of the semidwarfs. fewobservers would have predicted thatSonalika would have displayed thedurability of resistance that it hasconsidering that it occupies such alarge area. Sonalika was introduced20 years ago in India. Sincesemidwarfs now occupyapproximately 50% of the wheatacreage in the developing world. it isopportune to evaluate the durability(stability) of rust resistance insemidwarfs (especially in regard tostem rust and leaf rust associatedwith Mexican-CIMMYT semidwarfbread wheats). This chapterhighlights the critical factors thathave provided stable rust resistancein CIMMYT germplasm worldwide.Incorporationof Genetic DiversityRecognizing that wheat varietiesderived from CIMMYT materials aregrown on such a large area and areexposed to different pathogens underconditions that may favor diseasedevelopment. CIMMYT breedingpolicy has been to utilize sources ofgermplasm that are as diverse aspossible for rust resistance. Thecurrent array of varieties inCIMMYT's bread wheat crossingblock consists of varieties and lineswith the following characteristicsand geographic origins:1) Stem rust- and leaf rustresistantgermplasm from theSouthern Cone countries ofSouth America.2) Yellow rust- and leaf rustresistantgermplasm from theAndean region of SouthAmerica.3) Rust-resistant germplasm fromCentral America. includingMexico.4) Rust-resistant wheat lines fromNorth' America:a) Yellow rust-resistant linesfrom the Pacific Northwest.b) Stem rust- and leaf rustresistantlines from the GreatPlains of the USA andCanada.5) Stem rust- and leaf rustresistantlines from the IndianSubcontinent.6) Stem rust- and yellow rustresistantlines from the easternhighlands of Africa. includingKenya.

1037) Rust-resistant lines andvarieties from North Africa, theIberian Peninsula, and MiddleEast, including the Nile Valleyregion.8) Yellow rust-resistant lines fromwestern Europe.9) Stem rust-resistant lines from southern Europe. 10) Stem rust-resistant lines fromAustralia and New Zealand(Oceania).The flow of germplasm to and fromthe bread wheat improvementprogram is continuous and CIMMYTscientists are in contact withnational program scientists to ensurethis germplasm exchange. Thesources of rust resistance in CIMMYTgermplasm have, by intent, beenkept very diverse, first by theeXChange of germplasm and secondby its use in the breeding program.When a new yellow rust race wasintroduced into Australia in the early1980s, most of the locally developedvarieties without CIMMYTgermplasm in their pedigrees werehighly susceptible, but CIMMYTderivedgermplasm was highlyresistant (R.A. McIntosh, pers.comm.). This situation permittedAustralian plant breeders to useCIMMYT germplasm as a principalsource of yellow rust resistance intheir breeding programs.International MultilocationTesting: A System That Aidsthe Enhancement of GeneticDiversity in CIMMYTGermplasmAlthough multilocation testing is nota perfect system for identifyingresistance sources, evidenceaccumulated in CIMMYT over manyyears would indicate thatmultilocational testing has greatlyfacilitated the confirmation of theexistence of genetic diversity inCIMMYT germplasm. Indeed, theinitiation of international testingcame about because three peopleenvisioned its use for identifyingdiverse and durable sources ofresistance. Dr. E.C. Stakmanproposed the USDA InternationalRust Nursery in 1950; Dr. N.E.Borlaug proposed similar testing forCIMMYT in 1958-59; and Dr. JohnNiederhauser initiated such testingfor the Rockefeller Foundation'sPotato Program in the 1960s. Thistesting system is now fully in placefor major crops whose improvementis dealt with at other InternationalAgricultural Research Centers(lARCs) and at major agriculturaluniversities in the United States withinternational programs.It should be emphasized thatinternational multilocation testing(lMT) is complementary to traditionalgenetiC analysis and both should beparts of an overall improvementstrategy. Some genetiCists/plantpathologists have criticized thesystem perhaps without giving dueconsideration to its benefits, whichhave included the development ofgermplasm with the combined traitsof high yield potential, broadadaptation, and resistance to thethree rusts. CIMMYT has employedboth IMT and genetic analysis in itsimprovement strategy, as willbecome clear in this chapter.Genetic studies have suggested thatwheat genotypes that are resistant toa given rust disease in manydissimilar locations-as indicated bylow average coefficients ofinfection-often contain multiplefactors for resistance (14).Irrespective of whether some of thisresistance is race-non-specific, a linethat contains several functioningresistance genes has a better chanceto have stable resistance against achanging pathogen than one with asingle gene resistance. By testing

104lines at a number of epidemiologicallydissimilar sites andexposing the lines to the greatestpossible range of virulence factors.the probability of identifying linesthat may prove to have had durableresistance should. in principle. beincreased.CIMMYT recognizes that. in certainsituations. a single gene. such asLr19. will give a low averagecoefficient of infection because novirulence has developed for that ofinfection gene. But such a situationis rare and should be treated as anexception. The resistance of mostlines showing low average coefficientof infections in internationalnurseries is polygenic (14).A low average coefficient of infectionmay be associated. but notnecessarily. with the presence ofbroad-based resistance. Analyses ofpatterns of reaction to diseases atdiverse sites that possess differentcombinations of virulence factorsallow grouping of lines with distinctsets ofresistance genes. Althoughthe individual genes cannot beidentified thus. the method doesprovide a simple. rapid means ofidentifying lines with differentresistance genes for use in thebreeding program (4). Table 1provides a list of the bread wheatnurseries that comprise part ofCIMMYT's multilocation testingsystem.In Table 2. the genetic diversity of188 advanced lines of bread wheat isshown for resistance to stem rust.leaf rust, and yellow rust. The linesare classed into eight arbitrarygroups. Such noticeable differencessuggest the existence of differentgroups of varieties where response torust is under distinct genetic control.Table 1. Major international bread wheat nurseries distributed by CIMMYTfor the evaluation of disease resistance. Different locations have differentpathogen populationsNurseryNo. Sets Approximatedistributed no. of(1986-87) entries a/International Bread Wheat Screening Nursery (IBWSN) International Septoria Observation Nursery (ISEPTON) Helminthosporium Resistance Screening Nursery (HRSN) Scab Resistance Screening Nursery (SRSN) International Disease Trap Nursery (IDTN) Barley Yellow Dwarf Virus Screening Nursery (BYDVSN) Drought Screening Nursery (DSN) Aluminum Tolerance Screening Nursery (ATSN) 186 25090 10084 10061 80245 20054 15050 15050 150aJ Germplasm included in each nursery is different and serves different mega-environments;rust evaluation is made on each entry and ACI calculated on the basis of IMT data

105Additional genetic analysis iscritically needed to interpret thegenetic makeup of these groups butis not absolutely essential for plantbreeding. The CIMMYT WheatProgram does perform some analyseson selected groups of genotypes, butthis work should be expanded,preferably conducted as collaborativeresearch with other centers ofexcellence.It should be emphasized that "hotspot" locations (those locations withmaximum variability of a pathogenand/or severity of a disease) arecarefully selected for inclusion in theIMT program. Table 3 lists locationscurrently used as hot spots. Hot spotlocations for rusts exist in Kenya.Ecuador. and Mexico.Durability of Resistanceand Genetic DiversityGeneral featuresIn a global context. durable rustresistance. along with geneticdiversity for thwarting geneticvulnerability. is of paramountTable 2. Genetic diversity in 188 advanced lines of 18th InternationalBread Wheat Screening Nursery classified into average coefficient ofinfection (ACI) in respect to the three rusts in InternationalMultilocation Testing (IMT)Disease No. of No. of entries in ACI classeslocations0-2 2.1-5 5.1-10 10.1-15 15.1-20 20.1-30 30.1-40 40.1-60Leaf rust 31 38 33 44 29 28 9 7 0Stem rust 12 21 67 44 25 14 14 2Yellow rust 24 18 30 59 29 27 17 2 6Table 3. Hot spot locations for various diseases currently utilized byCIMMYT for shuttle breeding and testingLocationNjoro. KenyaQUito. EcuadorCd. Obregon. MexicoRio Bravo. MexicoPoza Rica. MexicoHoletta. EthiopiaToluca. MexicoNanjing. ChinaDiseaseStem rustYellow rustLeaf rustLeaf rustHelminthosporium leaf blotchSeptoria tritici blotchBYDVFusarium head scab

106importance in CIMMYT's breedingprogram. It is essential thatgermplasm with different geneticmakeup for rust resistance isavailable to breeders for deployment.However. as has been discussed inthe literature. it is quite possible thata particular genetic resistance maysuccumb to a new biotype.Whenever this happens. the breederdevelops materials with another kindof resistance or combination ofresistances. until that is overcomealso. The availability of geneticallydiverse materials has so far allowedbreeders to keep on utilizing thoseresistances. CIMMYT believes thatthis major gene. pathotype-specific.vertical resistance (VR) as defined bySimmonds in Chapter 10 is aprecarious situation and. in theevent of the development of newpathotypes. could result incatastrophic epidemics.An alternative to the continuousdevelopment of new sets ofresistance gene combinations is toattempt to breed for resistance thathas a better probability of beingdurable. In the literature. there are anumber of instances of rustresistance associated with durability.Breeding exclusively for one set ofresistance genes. even if they are ofa stable nature. is likely to result ina narrowing of genetic variabilitythisis not satisfactory in a globalcontext. The ideal situation would beto identify a gene or set of genes forprobable durability of resistance tobe used as a backbone. and thencontinually combine various othersets of genes to provide geneticdiversity. This situation would resultin germplasm that combined genesfor durability of resistance with otherresistance genes. We need theseother genes even though we alreadyhave durable resistance genesbecause. in many instances. the geneor genes for durable resistanceperfonn better in the presence ofother genes. Historical evidence ofdurable resistance is an indication ofpolygenic. pathotype-nonspecific.horizontal resistance (HR) asdescribed by Simmonds inChapter 10.Use of the Sr2 complex inCIMMYT germplasm for controlof stem rustThe stem rust resistance derivedfrom the variety Hope (Sr2 complex)seems to have provided thefoundation for durable resistance tostem rust in CIMMYT germplasm.Durable resistance in CIMMYTgermplasm to stem rust undoubtedlyfacilitated adoption of semidwarfgermplasm in many developingcountries. Before the Sr2 complexwas bred into the semidwarfs. stemrust created periodic havoc in SouthAmerica. Asia. and Africa. Analysisof comparative stem rust infectiondata on local land race, improvedtall. and semidwarf varietiesindicates that semidwarfs are. ingeneral, more resistant than the localland races and unimproved tallvarieties (Table 4). This finding iscontrary to the opinion of critics ofthe green revolution.As far as can be ascertained. theenhanced resistance to stem rust inCIMMYT semidwarfs is associatedwith the.Sr2 gene complex (Sr2 genein combination with various othergenes), derived from the varietyNewthatch. a Minnesota release.which inherited it from the varietyHope (Figure 1). It is unlikely thatSr2 alone would have imparted thisdurable resistance. It is CIMMYT'scontention that the Sr2 gene incombination with other genes isresponsible for this durableresistance. Dr. A.P. Roelfs. Universityof Minnesota (unpublished) haspostulated that the resistance ofmost of the CIMMYT semidwarfs tostem rust is associated with Sr2;however. he ascertained that there

107were additional genes (sometimes germplasm have remained resistantthree to four) in the background. The to stem rust in worldwide testing. itSr2 complex is an excellent example is suggested that the Sr2 complex inof the combination of a durable CIMMYT wheats is a typical exampleresistance gene plus an array of of the HR character referred to byadditional genes which hasSimmonds in Chapter 10. not VR. Itcompound durable resistance to is important to note that stem ruststem rust in CIMMYT germplasm. resistance in CIMMYT germplasmSince varieties derived from CIMMYT has not been conferred through theTable 4. Bread Wheat Regional Disease Trap Nursery (RDTN) data(average coefficient of infection) for stem rust (P. gramlnls f.sp. trltici)from approximately 50 locations in 30 countriesYear1978 1979 1980 1981MeanLocal 2l.8 2l.8 2l.1 19.6 21Improved Tall 9.2 8.6 9.9 9.3 9Semidwarf 7.8 4.7 5.1 6.2 6Source: CIMMYT. unpublished.Genaro 81* Papagos 86*Opata 85*Figure l. Major wheat cultivars of Mexico (*) from 1950 onwards thathave remained resistant to popUlations of stem rust under fieldconditions. Their resistance is probably derived from the cultivarNewthatch (Sr2).

108presence of a single gene (Le .. Sr2)but through a complex of genes inwhich Sr2 plays a principal role. Drs.R.A. McIntosh. University of Sydney.and A.P. Roelfs (both contributors tothese proceedings. Chapters 1 and 2respectively) have confirmed this inpersonal communications.As already mentioned. the originalsource of Sr2 was the variety Hope(and the related line H44-24) whichhas remained resistant to stem rustfor the last 70 years. University ofMinnesota breeders used Hope intheir program and producedNewthatch. Dr. N.E. Borlaug ofCIMMYT used Newthatch in theMexican/Rockefeller Foundationbreeding program in the early 1950s;since then this complex has entereda substantial amount of germplasm.Knott (9) and Green and Dyck (6)have described genes Sr2. Sr7b.Sr9d. Sr17. and Sr18 in the varietyHope. Hare and McIntosh (7) locatedSr2 in Hope and Hope derivatives inthe short arm of chromosome 3B.These authors also found that theamount of rust development onvarieties with Sr2 was variable andmay have been modified by alleles atadditional loci.CIMMYT (16) and Minnesota (18)studies have also provided someconfirmatory evidence that Sr2 alonegives only a slow rusting responsewhen tested with races in Mexico;however. when combined with otherresistance genes. this has resulted inenhanced levels of durableresistance.Over the last 20 years (1965-1985).the CIMMYT breeding program hasattempted to incorporate aiversity inconjunction with Sr2. Most of thegenetic combinations displayed inthe international nurseries sent toCIMMYT cooperators have Sr2 plustwo to four additional genes (A .P.Roelfs. pers. comm.). Theseadditional genes. which mayor maynot be modifiers. include Sr5. Sr6.Sr7a. Sr7b. Sr8a. Sr9b. Sr9d. Srge.Sr9g. SrlO. Sr11. Sr12. Sr17. Sr24.Sr26. Sr30. Sr31 . and Sr36.Distribution of the Lr13 complexin CIMMYT germplasm as asafeguard against leaf rustThe South American varietyFrontana has been judged to be oneof the best sources of durableresistance to leaf rust (A.P. Roelfs.pers. comm.). The variety was firstused in the Mexican-RockefellerProgram in the 1950s. The geneticanalysis of this variety by Dyck et al.(5) has indicated the presence ofLr13 and other genes. The currentgeneral opinion. especially of Drs.D.J. Samborski. Canada Departmentof Agriculture. Winnipeg. A.P. Roelfs.and R.A. McIntosh (pers. comm.) isthat the Lr13 gene. in combinationwith the other genes. may impart ahigh degree of durable resistance toleaf rust. Under Mexican conditions.this gene complex shows a slowrusting characteristic-similar to theSr2 complex phenomenon discussedabove (i.e.. pathotype-non-specific.horizontal resistance or HR asdescribed by Simmonds in ChapterlO. CIMMYT recognized theimportans::e of this gene complex inthe early -1970s when it wastransferred. along with other genes.into many wheat varieties. Table 5provides a partial list of thesevarieties.Again. it should be emphasized thatLr13 alone confers only a measure ofresistance. at least in Mexico. andonly in conjunction with other genesdoes it provide a degree of resistancewith high probability of beingdurable. The mode of action of the

109Lr13 complex in the Mexican­ Dr. R.A. McIntosh suggest that mostCIMMYT Program is non-specific CIMMYT lines have the Lr13resistance (HR as described by complex (unpublished data).Simmonds in Chapter 10. and themost important aspect is that Lr13 An analysis of the resistancemust be combined with other genes. spectrum to leaf rust in local landVarieties such as Genaro 81. Pavon race types. tall improved. and76. and Torim 73 have remained semidwarf wheats. compared over aresistant to leaf rust in Mexico for 6. 4-year period indicates that the11. and 14 years respectively. semidwarfs are more resistant thanAnalyses of CIMMYT advanced lines the other two (Table 6). This reflectsin the rust laboratories at the effective deployment of genetiCUniversity of Minnesota by Dr. A.P. resistance to leaf rust on a worldwideRoelfs and at Castle Hill Australia by basis.Table 5. CIMMYT varieties and CIMMYT derivatives from India andPakistan with Lr13 aJCountryCountryofofVariety Adoption Variety AdoptionSonalika India Kea'·S" Mexicolnia 66 Mexico Prl"S" MexicoTobari 66 MexiCO Myna"S" MexicoNuri 70 Mexico Chilero"S" MexicoYecora 70 Mexico Kauz"S" MexicoZaragoza 75 Mexico Prl"S"lVee"S" MexicoPavon 76 Mexico Punjab 81 PakistanTonichi 81 Mexico Pari 73 PakistanGenaro 81 MexiCO LyaUpur 73 PakistanUres 81 Mexico Kohinoor 83 PakistanGalvez 87 Mexico Sandal 73 PakistanTrap No.1 Mexico Sarhad 82 PakistanGaruda·'S" Mexico Zargoon 79 PakistanaJ Source: R.A. Mcintosh. University of Sydney and R.P. Singh, CIMMYT

110Advances made in durabilityof resistance to yellow(stripe) rustData on yellow (stripe) rust from theinternational nurseries for land race.improved tall. and semidwarf wheatsindicate that there has beenimprovement for yellow rust insemidwarfs. but not to the extentexhibited for stem rust and leaf rustresistance (Table 7). The dataindicate that local land race typesare more susceptible than improvedtalls and semidwarfs.In its crossing program during themid-1960s. CIMMYT used varietiesfrom the Andean region. which werehighly resistant to yellow rust suchas the Colombian variety Andes. TheCalifornian variety Anza was bred byCIMMYT. and has been reported tobe durably resistant to yellow rustby Johnson (Chapter 6). Anza wasderived from the crossLRfNIOBI13* ANE and has beenreleased in North Africa. Sudan.South Africa. and New Zealand. It isthought that the durable resistanceof Anza to yellow rust is derivedfrom the Andes; this hypothesisshould be tested by genetic analysis.Although the genes conferringresistance in Anza are likely to bedifferent from those in CappelleDesprez and Little Joss. the effect ondurability of resistance is the same.In both cases there is a need toTable 6. Bread Wheat Regional Disease Trap Nursery (RDTN) data(average coefficient of infection) for leaf rust (P. recondita f.sp tritici)from approximately 50 locations in 30 countriesYear1978 1979 1980 1981MeanLocal 39.9 31.1 41.6 28.1 35Improved Tall 19.1 11.4 14.1 7.8 13Semidwarf 12.1 7.1 8.7 6.0 8Source: CIMMYT, unpublished.Table 7. Bread Wheat Regional Disease Trap Nursery (RDTN) data(average coefficient of infection) for yellow rust (P. striiformis) fromapproximately 50 locations in 30 countriesYear1978 19'19 1980 1981MeanLocal 11.9 18.3 22.4 18.6 181m proved Tall 7 .9 6.5 9.2 6.9 8Semidwarf 9.2 8.6 10.0 9.9 9Source: CIMMYT, unpublished.

111conduct genetic analysis to elucidatethe genetic control of resistance.Durable resistance derived from thevariety Anza is widely dispersed inCIMMYT advanced lines. and hotspot testing will be continued toverify the durable resistance toyellow rust in those advanced linesthat have Anza in their pedigrees.CIMMYT also plans to continuecollaboration with Dr. R. Stubbs ofIPG. Wageningen. The Netherlands.who has a worldwide collection ofyellow rust isolates in his laboratoryto test CIMMYT germplasm againstselected yellow rust races.To accelerate the incorporation of awide spectrum of yellow rustresistance genes into CIMMYTsemidwarfs. a cooperative shuttlebreeding program has been initiatedinvolving CIMMYT and nationalprograms in Kenya. Ethiopia.Ecuador. and Peru -each as anequal partner. This largeinternational partnership venture inbreeding for yellow rust should bebeneficial to all the parties involved.Breeding for PartialResistance (Slow Rustin~)General featuresPartial resistance is a manifestationof a host:parasite interaction inwhich infection occurs. but in whichone or more steps in the infectionprocess take place with lesserefficiency than in a susceptible host.As such. partial resistance has longbeen observed and utilized by potatobreeders against the late blightfungus (10. 19). The first convincingcase of partial resistance (slowrusting. general-type resistance) towheat leaf rust was documented in1968 (2).The components of partial resistanceare difficult to detect. However. theeffects of partial resistance can bequantified by precise observations ofthe steps affected. and these datamay be utilized for breedingpurposes (11). Even when specificeffects are not quantified. partialresistance can be detected by amarked reduction in the rate ofepidemic development (smaller areaunder disease curve) (4. 20).CIMMYT wheat breeders andpathologists have maintained thatthe crucial test for durability ofresistance can only be conducted inthe field over time (12). Generalresistance of a partial nature to stemrust in the adult plant stage wasobserved in a number of wheatvarieties such as Yaqui 50. Bonza55. and Penjamo 62 (12). Sartori etal. (16) showed in a genetic analysisof Penjamo 62. Hopps. and Mengavithat these varieties have general.partial resistance to stem rust. andthat in Mengavi this resistance is dueto a lesser receptivity to infection bythe stem rust fungus. Skovrnand eta1. (17) in a different study found asimilar situation in the varietyMengavi as reported by Sartori et a1.(16).Identification of components ofpartial resistance (slow rusting)at CIMMYTAs stated earlier. longer latentperiod. smaller pustule (uredium)size. and pustules per unit area allplay strong roles in retarding diseasedevelopment. Tables 8 and 9 listwheat varieties showing differencesfor these components of partialresistance to leaf rust. In Table 8.the varieties Juzco. Katahdin. andPavon 76 differed significantly fromInia 66 in terms of pustule size andpustule number in the seedlingstage; and for pustule number in theadult plant stage.

112In Table 9, the varieties Opata 85,Pavon 76, Genaro 81, Seri 82,Myna"S", and Kauz"S" arecompared with the susceptiblechecks Morocco and Siete Cerros.The partially resistant varietiesdiffered significantly from bothchecks for latent period, days to fullinfection, and in receptivity. It isnoteworthy that this approach issimilar to that described byParlevliet (Chapter 5), i.e., selectionfor types of resistance characterizedby a reduced rate of epidemicdevelopment.Based on the above results, Figure 2compares Pavon 76 and Genaro 81to Inia 66 under field conditions. Thepattern of infection rates of Pavon 76and Genaro 81 (both partiallyresistant) has been shown now for 8consecutive years. Both varietiescarry the Lr13 complex (Table 10).The procedure to identify partialresistance in the field (Figure 2) andstudies of components (Tables 8 and9) are routine activities at CIMMYT.It is believed that the studies of thisnature, which have been going on forthe last 17 years, have madeimportant contributions in achievingsustainability of wheat yields on aglobal basis.Percent infection100y---------------------~---SO60+-------------~L-----------4020+-------~~----~~--~~---Figure 2. Slow rusting resistance ofGenaro 81 and Pavon 76 to leafrust when compared to Inia 66 (Cd.Obregon 1984-85).Table 8. Size and number of pustules in greenhouse trials (seedling andmature plant) with isolates 82060-B and 82076-A of P. recondlta f.sp.trltlcl, EI Batan, 1983GenotypeSeedlingMature plantPustulesize(mm2) aJNo. ofpustulesper cm2 aJPustulesize(mm2) aJNo. ofpustulesper cm2 aJJuzco 0 .1605a 4.2SaKatahdin 0.202Sa 7.14aPavon 76 0.209Sa 9.28aDove 0.236Sb 4.00aInia 66 (check) 0.2625b 20.2Sb0.1055a 3.4Sa0.1052a 2.23a0 .1033a 2.S0a0.1062a 3 .01a0.1194a 11.51bSource: Huerta-Espino and Rajaram (in press).aJ Genotypes followed by different letters are statistically different at P = 0.05(Dunnett Test)

113Breeding for MultipleMajor Gene ResistanceAlthough the longevity of resIstancebased on multiple major genes (Le.,VR) may be limited and stepwisemutation can eventually lead tosusceptibility, this strategy has beensuccessfully employed in Australiafor stem rust. However, givenCIMMYT's global mandate and thecorresponding difficulty in deployinggenes effectively in an internationalcontext, CIMMYT has pursued thisas a supplementary strategy.If multiple major gene resistance isto be pursued as a strategy,pathogenicity analysis is aprerequisite for maintaining control.The spectrum of virulence/avirulencegenes for stem rust and leaf rustidentified in Mexico are given inTables 11 and 12, respectively.Resistance gene analysis can beconducted utilizing these races.When the varieties have complexcombinations of genes, known orunknown, however, this method isnot adequate. In that situation,continual genetiC analysis is requiredTable 9. Analyses of slow rusting components such as latent period.infection period. and receptivity to leaf rust in eight varieties of wheatswhen tested with isolates 87.34A or 87.40BIsolateusedDays tolatentperiod atDays tofullinfection alNo. ofpustules(10 cm2) alMorocco (check) 87.34A 5.00a1O.00a228.geMorocco (check) Siete Cerros (check) Slete Cerros (check) Opata 85 Pavon 76 Genaro 81 Seri 82 Myna"S" Kauz"S" 87.40B 5.00a87.40B 6.00b87.34A 6.00b87.34A 7.60c87.34A 7.65cd87.40B 8.19d87.40B 8.80e87.40B 8.05cd87.40B 9.45f1O.00a 204.6e1O.00a 123.9d10.40b 119.8d12.85d 74.0cd12.00c 80.5cd14.00e 22.6ab14.00e 8.7a13.00d 61.1 bc13.00d 7.5aLSD0.53Source: M. L. Vargas and R. P. Singh, CIMMYT. 1987 (unpublished).0 .19 47.40aJ Different letters denote significant difference at P=O.05 (Duncan's New MultipleRange Test)

114to study diversity and allelicrelationships of resistant varieties.Such a study for leaf rust resistancein 10 varieties is presented in Table13. The data indicate that thesevarieties possess as many as foureffective genes and as many as ninedifferent genes may be involved inconferring leaf rust resistance. Sucha study on genetic diversity isimportant for pyramiding resistancegenes and ultimately achievingmultiple gene combinations.Revival of theMultiline ApproachAnother way to achieve stableresistance to diseases caused byobligate parasites is through the useof multiline composites as originallyproposed by Jensen (8) and Borlaug(1) . This means of manipulating rustresistance and the methods used byCIMMYT are discussed in greaterdetail elsewhere (13, 15). Multilinebreeding is a very conservative andslow approach in regard to yieldbecause newer varieties may rapidlysupersede the recurrent parent. Themultiline approach may offerconsiderable merit to maintainingyield stability, especially in areas athigh-risk from disease. CIMMYT iscurrently in the process ofgenerating multilines using the highyieldingvarieties Seri 82 and Genaro81 for stabilizing leaf rust resistancein the northwestern Mexican statesof Sonora and Sinaloa. Resistancegenes Lrg. Lrlg. Lr24, and othersyet to be identified are being used.Only a moderate allocation ofCIMMYT resources is currentlydevoted to the multiline approach.However. this may be increased inthe future depending on the progresswith multilines and newcollaborative research that isunderway on varietal mixtures inconjunction with Dr. M. Wolfe at PBIin Cambridge. England.Table 10. Genetic constitution of certain CIMMYT/INIFAP varietiesreleased since 1976 in Mexico in relation to P. recondlta f.sp. triticiCultivarNacozari 76Pavon 76Ciano 79Tonichi 81Genaro 81Opata 85Papago 86Cucurpe 86Source: R. P. Singh.Year ofrelease19761976197919811981198519861986LR genesLrlO + slow rusting genesLrl. LrlO. Lr;13 + slow rusting genesLr16 + 1 geneLrlO. Lrl7 + 2 adult plant genesLr3. Lr13. Lr26 + slow rusting genesLrlO + 1 adult plant geneLr16 + 1 geneLrlO + 2 adult plant genes

115Ideal Situation:The above approach has beenCombining Durable Resistance pursued to develop resistance to leafand Partial Resistance with rust in the set of varieties listed inMajor GenesTable 10. The varieties Pavon 76.Ciano 79. Tonichi 81. Genaro 81.CIMMYT's preferred scenario Opata 85. Papago 86. and Cucurperequires that the principal thrust 86 have shown a combination ofmust be a durable resistance these desirable and different types ofexpressed as partial resistance in leaf rust resistance. No leaf rustconjunction with major genes that epidemic has occurred in Mexicoconfer additional security. For since 1978 where the above varietiesexample the Sr2. Lr13. and Anza­ have been widely adopted. Largetyperesistances are used as principal scale epidemics of stem rust werecomponents to establish durability thwarted wherever Sr2 plus otherthrough partial resistance ingenes for stem rust resistance haveconjunction with other genes.Table 11. Virulence/avirulence combinations of P. gramlnls f.sp. trltlclidentified in Mexico during 1984-86Sr5. 9a. 9d. 36/Sr7a. 7b. 8. 9b. ge. 10. 11. 13. 24. 25. 26. 27. 30, 37Sr5. 8. 9a. 9d. 36/Sr7a. 7b. 9b. ge. 10. 11. 13. 24. 25, 26. 27, 30. 37Sr5. 7b. 8. 9a. 9d. 36/Sr7a. 9b. ge. 10. 11. 13. 24. 25. 26. 27, 30. 37Sr5. 8. 9a. 9b, 9d. 36/Sr7a. 7b. ge. 10. 11. 13.24. 25. 26. 27. 30. 37Sr5. 8. 9a. 9d. 11. 36/Sr7a. 7b. 9b. ge. 10. 13. 24. 25. 26. 27. 30. 37Sr5. 7b. 8. 9a. 9d. 11 . 36/Sr7a. 9b. ge. 10. 13. 24. 25. 26. 27. 30. 37Sr5. 9. 9a. 9b. 9d. 11. 36/Sr7a. 7b. ge. 10. 13. 24. 25. 26. 27. 30. 37Sr5. 7b. 8. 9a. 9b. 9d. 11. 36/Sr7a. ge. 10. 13. 24. 25. 26. 27. 30. 37Source: R. P. Singh.Table 12. Virulence/avirulence combinations of P. recondlta f.sp. trltlclidentified in Mexico during 1984-87Lrl. 3. 3bg. 10. 13. 17. 27 + 31/Lr2a. 2b. 2c. 15, 23, 24. 26Lr2c. 10. 17/Lrl. 2a. 2b. 3. 3bg. 13. 15.23.24.26.27+31.Lr1. 3, 3bg. 10. 13. 15.23.24.26. 27 + 31/2a. 2b. 2c. 17Lr1. 2c. 10. (17), 23. 26. (27+31). /2a. 2b. 3, 3bg. 13, 15.24Lr1. 2a. 2b. 2c. 3. 3bg. 10. 13. 15. 17. 27 + 31/23. 24. 26Lrl. 2c. 10. (17), /2a. 2b. 3. 3bg, 13. 15.23.24.26.27+31Lrl. 2a. 2b. 2c, 3. 3bg. 13. 15. 26/10. 17.23. 24Lrl, 2a. 2b, 2c, 3, 3bg. 13. 15,23,26.27+31/10.17,24Lr2a. 2b. 2c, 3, 3bg. 10. 13. 15, 23. 27 + 31. /Lrl. 17, 24. 26100% virulence for genes: 14a, 14b, 18. 20. and 28100% aVirulence for genes: Lr3Ka, 9. 11. 16. 19.21. 25. 29. 30. and 33Genes in parentheses indicate intermediate virulenceSource: R. P. Singh.

116been used; likewise. wherever Lr13plus other genes for leaf rustresistance were used. large-scaleepidemics of leaf rust have beenprevented. When Lr13 has been usedon its own. as in the varietySonalika. there has beensusceptibility. Where Lr13 has beenabsent. as in Mexico in 1977 withthe variety Jupateco 73. there was aserious epidemic.CIMMYT also contends that thewidespread adoption of semidwarfson more than 50 million haworldwide is due. in part. to durableresistance against rust diseases. Thehigh yield potential of semidwarfswould have been short-lived if stableresistance to rust diseases had notbeen simultaneously bred in. In thefuture. we must find new stable genecombinations to complement Sr2and Lr13.References1. Borlaug. N.E. 1953. Newapproach to the breeding of wheatvarieties resistant to Pucciniagraminis tritici. Phytopathology.43:467 (Abstract).Table 13. Genetic relationship and genetic diversity present in 10 leafrust-resistant parents in field conditions when crossed to susceptiblevariety Siete CerrosCrossed to varietyVarietyNo. ofgenes whencrossed withSiete CerrosKea"S" D D D D D D (D) 1Parula"S" D D D D (D) 3Bow"S" D D D D (D) 4Tonichi"S" D S D 2Yaco"S" D 2Bucbuc"S" D D 2Chilero"S" 2Junco"S" D 2Myna"S" D 2Hahn"S" 1Source: R.P. Singh. Fl. F2. and F3 data. In body of Table: D = Completely different genetic basis. in the two varieties; (D) = Partially different genetic basis; S = Same genetic basis.

1172. Caldwell, RM. 1968. Breedingfor general and/or specific resistance.Proceedings of the 3rd InternationalWheat Genetics Symposium,Canberra. Australian Academy ofSciences, 263-272.3. Dalrymple, D.G. 1986.Development and Spread of High­Yielding Wheat Varieties inDeveloping Countries. Agency forInternational Development,Washington, D.C., Publication No.86-71247.4. Dubin, H.J., and S. Rajaram.1981. The strategy of theInternational Maize and WheatImprovement Center (CIMMYT) forbreeding disease resistant wheat: aninternational approach. In: Strategiesfor the Control of Cereal Diseases(J.F. Jenkyn and RT. Plumb, eds.J,pp. 28-35. Blackwell ScientificPublications, Oxford.5. Dyck, P.L., D.J. Samborski, andRG. Anderson. 1966. Inheritance ofadult-plant leaf rust resistancederived from the common wheatvarieties Exchange and Frontana.Canadian Journal of Genetics andCytology, 8:655-671.6. Green, G.J., and P.L. Dyck.1979. A gene for resistance toPuccinia graminis f. sp. tritici that ispresent in wheat cultivar H-44 butnot in cultivar Hope.Phytopathology, 69:672-675.7. Hare. RA., and RA. McIntosh.1979. Genetic and cytogeneticstudies of durable adult-plantresistance in 'Hope' and relatedcultivars to wheat rusts. Zeitschriftfur Pflanzenzuchtung, 83:350-367.8. Jensen, N.F. 1952. Intervarietaldiversification in oat breeding.Agronomy Journal, 44:30-34.9. Knott, D.R 1968. Theinheritance of resistance to stem rustraces 56 and 15B-IL (Can.) in thewheat varieties Hope and H-44.Canadian Journal of Genetics andCytology, 10:311-320.10. Niederhauser, J.S., J. Cervantes,and L. Servin. 1954. Late blight inMexico and its implications.Phytopathology, 44:406-408.11. Parlevliet, J.E. 1979.Components of resistance thatreduce the rate of epidemicdevelopment. Annual Review ofPhytopathology, 17:203-222.12. Rajaram, S. 1972. Method fordetection and evaluation of adultplant resistance to Puccinia graministritici in wheat. Proceedings of theEuropean and Mediterranean CerealRusts Conference, Praha,Czechoslovakia. Pp. 203-208.13. Rajaram, S., and H.J. Dubin.1977. Avoiding genetic vulnerabilityin semidwarf wheats. In: The GeneticBasis of Epidemics in Agriculture(P.R Day, ed.). Annals of the NewYork Academy of Sciences,287:243-2~4.14. Rajaram, S., and N.H. Luig.1972. The genetic basis for lowcoefficient of infection to stem rust incommon wheat. Euphytica,21 :363-376.

11815. Rajaram. S .. B. Skovmand. H.J .Dubin. E. Torres. RG. Anderson.A.P. Roelfs. D.J. Samborski. and LA.Watson. 1979. Diversity of rustresistance of the CIMMYT multilinecomposite. its yield potential andutilization. Indian Journal ofGenetics and Plant Breeding.39:60-71.16. Sartori. J .F .. S. Rajaram. Ma. deLourdes del I. de Bauer. 1979. Basespathologicas y geneticas relacionadascon la resistencia general del trigo aPuccinia graminis Pers. f. sp. triticiEriks. et Henn. Agrociencia. 34:3-16.17. Skovmand. B .. A.P. Roelfs. andRD. Wilcoxson. 1978. Therelationship between slow rustingand some genes specific for stemrust resistance in wheat.Phytopathology. 68:491-499.18. Sunderwirth. S.D. and A.P.Roelfs. 1980. Greenhouse evaluationof the adult plant resistance of Sr2 towheat stem rust. Phytopathology.70:634-637.19. Thurston. H.D. 1971.Relationship of general resistance:late blight of potato. Phytopathology.61 :620-626.20. Wilcoxson. RD.. B. Skovmand.and A.H. Atif. 1975. Evaluation ofwheat cultivars for ability to retarddevelopment of stem rust. Annals ofApplied Biology. 80:275-281.

119Chapter 10 Synthesis: The Strategy of RustResistance BreedingN.W. Simmonds. Edinburgh School of Agriculture. Edinburgh. ScotlandAbstractThe purpose of breeding for disease resistance is to protect biomass andhence crop yield. Resistance is simply a state of "less disease"; and nodisease (immunity) is rarely a realistic objective. Four kinds ofresistancemay usefully be recognized. namely: 1) major gene. pathotype-specific.vertical resistance (VR); 2) polygenic. pathotype-non-specific. horizontalresistance (HR); 3) pathotype-non-specific, major gene resistance (NR); and 4)interaction or mixture resistance (IR). VR is often effective against immobilepathogens. but is generally non-durable against mobile. airborne ones;continued disease control is sometimes possible. however. by the use ofmultiple (pyramided) VR genes deployed under tight genetical andpathological management. HR is durable and generally fairly highlyheritable; most diseases are controlled thus in many crops. NR is valuable ifavailable. but is rare. IR. due essentially to heterogeneous VR elements. ispoorly understood. but probably more valuable than is yet generally realized.Of the wheat rusts, stem rust has been well controlled worldwide for yearsby pyramided VR genes. but must be judged to retain potential for epidemicoutbreak if tight genetic/pathological control were to lapse. Leafrust isprobably the most damaging of the three rusts at present. the genetic VRbase is narrow and more epidemics must be expected. Yellow rust differsfrom the other two in that VR seems to have wholly failed. so that breedersin Europe have begun to abandon VR and effective HR is recognized andbeginning to be exploited.CIMMYT breeding has so far been mainly concentrated upon pyramiding VRfor all three rusts. but hardly under tight genetic control. It is arguable thatthis is a risky strategy because small farmers in less developed countries canill afford epidemics that farmers in rich countries could well tolerate. It issuggested that a reasonable/feasible shift of strategy would be a movetowards research on and exploitation ofHR and IR; in the context of stemand leaf rusts. both are greatly under-researched but must. with allreasonable certainty. have much potential. The shift from VR to HRlIRemphasis would not come rapidly but is of very great long-term potentialand practical importance. A parallel shift of the grand strategy of the Centerfrom breeding wheat varieties to strategic (not basic) research is implied andseems to be consonant with the policy of the Consultative Group forInternational Agricultural Research.IntroductionThis meeting was primarilyconcerned with the three rusts ofwheat on a worldwide basis; therewas. of design. some emphasis onthe CIMMYT context and the lastpart of the meeting. indeed.concentrated upon this. Theauthorities in CIMMYT thought that.though nearly everyone present wasan internationally-respected experton some aspect of the cereal rusts. itwould do no harm to have someonepresent to sum up. someone who

120had rather wide experience of cropsand disease resistance breeding butno specialized knowledge of therusts. Hence my presence. Theonlooker. they say. sometimes seesmost of the game. I was specificallycharged by the CIMMYT authoritiesto formulate some conclusions as tothe Center's breeding strategy. This Ihave done and thank my colleaguesboth in CIMMYT and abroad for theirmany comments; in the end. though.since there was rather less than totalagreement. those views had better beattributed to me. But I hope that noone will feel that they are seriouslyunrepresentative of the general trendof discussion.The material for this chapter largelycomes from the preceding chaptersand I have found it convenient torefer to them briefly thus: [5].meaning Chapter 5 by J.E.Parlevliet. Few other referencesseemed necessary since we aremostly dealing with well known factsand arguments. A relevant referenceto the general strategy of diseaseresistance breeding is (8) and avaluable compendium of papers onhorizontal resistance will be found in(4). The tropical agricultural contextof disease resistance is emphasizedin two FAO publications (1. 2).General Context ofDisease ResistanceThe paramount objective of all plantbreeding is timely high yield in achosen environment. True. yield asan objective is sometimes qualifiedby quality considerations and thereare trade-offs. But. at a given qualitylevel. then yield remainseconomically dominant. Yield (Y) isprocured by enhancing biomass (B)(dry matter per unit area) andpartition to desired product (adimensionless fraction. P) such thatY = BP. This is a perfectly generalrelation and the importance of plantdiseases is that they act by reducingB and hence Y. Disease resistancebreeding is therefore a yieldenhancingprocedure that does so byprotecting biomass. Diseaseresistance is not an objective in itsown right. though it sometimesseems to be thought of thus. Itmatters only insofar as it protectsyield. Ideally. one would like to haveformal proof that disease resistanceis economically worthwhile beforeembarking upon breeding. For thewheat rusts. of course. such proof ishardly necessary; but the point isrelevant elsewhere because somediseases can look bad but do littledamage.Resistance is simply a state of lessdisease; it is not a state of .'nodisease" and. if one meansimmunity. one should say so.Susceptibility is simply thecomplement of resistance but is notobviously bounded because a state oftotal susceptibility is hardlydefinable. Resistance scales oftenpresent difficulties at the bottom endand. as a general practical pOint. it isalways helpful to have verysusceptible standards in allexperiments to help to define thelower bound.In practice. most crop diseases areonly partially controlled by onemeans or another. genetic.agronomic or chemical. Absolutecontrol is rare so we generally haveto live with a moderate level of anyparticular disease: annoying perhapsbut economically tolerable. A state ofno disease is rarely a realisticobjective. Furthermore. it is notnecessarily even a desirable one. Theyield-disease intensity curve is rarelylinear so that low. even moderate.levels of disease are often found tohave effects so small as to beunmeasurable or economicallytrivial. Therefore "enough resistanceis enough" is a good practical

121maxim; how much is enough will ofcourse, vary from place to place,according to disease intensity.Finally. we should recall the point(that emerged several times in themeeting), that a good plant varietyrepresents a balanced package ofcharacters of which diseaseresistance is but one component.The farmer judges on a sort ofweighted index and there are manyexamples of favored varieties thatbreeders or pathologists haveproclaimed to be too susceptible; alsoof resisters that farmers did not like.Again. enough is enough, and howmuch that is has to be judgedagainst that elusive quality ofgeneral worth.Kinds of ResistanceGeneralFour kinds of resistance-Themain features of the four broad kindsof resistance are summarized inTable 1. I shall now run throughthem briefly but avoid theterminological (sometimes almosttheological) niceties that tend toobtrude in this area. Before doing so,it is well to point out that the list isbroadly applicable to all classes ofpathogen: virus, bacterial, fungal,and animal. But it would.admittedly, be hard to cite reallyclear cases of all the 4x4combinations.Table 1. The four main kinds of resistance (8, 9)Kind ofresistanceSpecificityGeneticsDurability1. Pathotypespecificorvertical.VR/SRvery higholigo­(I) mobile pathogens.durabilityusually bad(2) immobile pathogens,durabilitymay be good2. Pathotypenon-specificmajor genereSistance. NRniloligo­good but NRrather rare orspecialized3. General orhorizontalresistance.HRlGRnil/lowpoly-high4. Interaction ormixture resistance,IRlMRsomeheterogeneousoligo-a!probably goodaJ Some authors (7J are inclined to attribute some weight to heterogeneity forpolygenic systems. but this matter seems to me to be undeCided

122First. vertical resistance (VR,Vanderplank's term) or specificresistance (SR) is due to major geneshighly specific to matchingpathotypes (gene-for-genecorrespondence). VR genes are oftendominants and often control seedlinghypersensitivities but not always;semi-dominance (heterozygousexpression). recessiveness and adultplant resistance are fairly frequent(1.21. VR genes. when looked for,are numerous. so numerous thatestimates seem to be largelybounded by the patience ofinvestigators. Given specificity andpathotype adaptability (by migration.mutation. recombination under hostselection pressure (Ill, VR often fails.Against highly mobile pathogens(rusts. mildews. downy mildews.many ascomycetes, some insects),failures have been very numerousbut some instances of long-livedperSistence of a vulnerable VR geneare known and some successes havebeen achieved by the use of multipleand/or sequentially deployed VRgenes (see below). By contrast.against immobile (soil-inhabiting)pathogens. VR systems have oftenbeen very successful (as againstpotato wart. Synchytrium, and cysteelworm. Globodera rostochiensis).Thus VR is typically (but not quitealways) non-durable against mobilepathogens but may have very usefuldurability against immobile ones.Second (Table I), NR is that rareevent. a pathotype-non-specificresistance oligogene. Such genes arefew and the only examples I can callto mind relate to a few fungi andseveral viruses; thus there arecomprehensive resistances (NR) topotato viruses X and Y as well asseveral VR genes effective onlyagainst specific strains. I shall ignoreNR hereafter though perhaps oneshould note that Sr2 in wheat [1. 2,9 and see below] seems to havesomething of this character;however. Knott (pers. comm.) isdoubtful of this in the Canadiancontext.Third (Table I), horizontal resistanceof Vanderplank (HR) or. as preferredby many writers. general resistance(GR), or field resistance (not favorednowadays). or partial resistance (ofParlevliet [5ll are all terms that meanat least nearly the same thing.namely: a polygenic resistance thatis pathotype-non-specific and acts,typically, by a rather complexmixture of inhibition of infection.long latent period. slow lesiongrowth, and reduced sporulation. HRis typically highly durable (4). It isthe rule rather than the exceptionand most plants are protected thusfrom most diseases. a fact whichoften becomes apparent only by wayof new-encounter events: a varietycarried to a new place and meetingan unfamiliar disease or a newdisease introduced to a population ofunadapted varieties. Genetic andoperational features of HR arediscussed further below.Fourth, interaction or mixtureresistance OR for the presentpurpose) (8) occurs when aheterogeneous pathotype populationmeets a heterogeneous crop and theresult is a,damping interaction thatreduces overall epidemic intensity [7.8] (and see (5) . (6)) . The effect hasbeen likened to the damping thatseems to be characteristic of(heterogeneous) wild populations [7].Typically. an effective IR looks likean HR epidemiologically and mightbe expected to be durable (thoughthis can hardly be said to have beenthoroughly tested).Complications-The simplicity ofTable 1 needs some qualification(Figure 1). The intersections of thefigure make several useful points.First. if polygenic HR is based on

123relatively few genes (which isprobably often true because polydoes not have to mean many [4. 5)).then a monogenic NR is the limitingcase. Second a weak VR gene thatallows some infection andsporulation (as with diverse Sr. Lr.Yr genes in wheat [1. 2. 4. 6]) canlook very like an HR untilinvestigated genetically and.conversely. there can be VRcomponents of seeming HR that maybe difficult to detect [6]. Third. if aVR gene remains effective for a longtime and no virulent pathotypeappears it will look like an NR and.indeed. had better be functionallyclassified as an NR even if theappropriate pathotype is known toexist but fails to multiply. forwhatever (usually unknown) reason.Genetic featuresMajor genes-The major genes.whether VR or NR offer no unusualfeatures: they segregate in normalMendelian fashion and numerousexamples are available in theliterature (e.g .. (7)). For the wheatrusts we have an example in thisvolume (leaf rust. one or two genesVRis very oftennon-durablea surviving VRlooks like NRa weak VRgene givespartial resistanceand may look likeHR.IR/MR•polygenes maybe few'nature's way'?- heterogeneouspopulation. withinteractionsmost resistancesto most diseaseshereFigure 1. Relation between the four main kinds of disease resistance (seeTable 1). Modified from (7).

124segregating [9]) and hundreds ofother such cases have been recorded.Linkages. of course. sometimes turnup [1. 2J and may be useful or not.depending on circumstances (e.g. theblack chaff associated with Sr2 [1])and the yellow flour associated withLr19 [2J. Sometimes, the very refinedcytology available in wheat permitsexact allocation of genes tochromosome arms (e.g. for Yr genes[6J: and see also [2J Tables 2 and 3for Sr and Lr genes).Polygenic inheritance-Analysis ofpolygenic HR has. until recently.been rather little studied, as Knott[4J points out. However, with theincreasing recognition of itsimportance in the past 20 years.studies multiply and examplesdemonstrating the key features willbe found in (4. 7. 8, 9). Thosefeatures are: continuous variation insegregating generations.environmental variation inexpression (statistical error).necessity for biometrical rather thanMendelian analysis. evidence of asignificant genetic component ofvariance (Le. heritability) and.finally, the need (if starting frominbred lines at least) for fourgenerations if the general nature ofinheritance is to be unambiguouslystated (see [4, 5]).Two examples will illustrate thesepOints. The first (Figure 2) is fromdurum wheat in North Dakota, USA.I have selected one cross out of threefor illustration. It is between asusceptible variety A and a resistantone P. The F2 mean resistance (R)was almost exactly intermediatebetween the parents and thedistribution continuous and verynearly normal. So far, we could bef40--~----------------------------~~-----------------------P120------------.-------~~__-----.----------~--~----~------i i i40 60 80Figure 2. Horizontal resistance to stem rust in durum wheat in NorthDakota. USA. Ordinate f is frequency. abscissa R is resistance on a scaleo (very susceptible) to 100 (immune). Parent A (Akrona) susceptible.parent P (Pentad) resistant. For details see the text. From ref. (10).R

125dealing with a semidominant gene(segregating 1 :2: 1) plus minor genesand error. However. the F3 was alsocontinuously distributed. with notrace of the discontinuity that couldsurely have been detected had amajor gene been segregating. Thereis no means of estimating thenumber of genes concerned; ingeneral. there is no biometricallysatisfactory way of doing so. Asremarked above. poly does not haveto mean many. In this caseheritability is obviously high. asevidenced by: the calculated geneticrange in F2 (R ±2dG). the top end ofwhich is high but not transgressiveof variety P; and the offspring-onparentregression (RF3 = 23 + 0.66RF2) which shows that excellentprogress would be made by selectingamong the F2. Progress by selectingamong F3 lines would probably beeven better.The other example (Figure 3) refersto barley mildew and it showsbroadly the same features. namely:hybrid generations with meansintermediate between parents.R16-----------+-------------------------------------------------­14 -------G~V~------GVlE~-----------------1 -=I-===-=----=-t.~.12--------+----P~R~~~~~~=====-+--------+~·~·.--10--------------------------------------~~~------------~··-----...' .' '8-----------+----------~~-----------------+----~------------KU6----------~----------------------------~··--------------------. . ........ :!:2Cfp ......······ .'4-------------------------------------------------------------­IPIF3Figure 3. Horizontal resistance to mildew in spring barley in the UK.Ordinate R is resistance (Parent KU very susceptible, parent PR middling,parent GV resistant). Parental and F3 means and ranges with indicationsof genetic ranges. KU, Kuusamo; PR, Proctor; GV, Gloire du Vellay, Fromref. (3).

126continuous. more or less normal F3distributions. evidence of widegenetic spans in the F3 and of goodheritability. potential for roughlyequaling the resistant parent (GV)but not transgressing it. These wereUK barleys and variety PR has anHR roughly at the threshold ofacceptability; clearly. GV and thebetter F3 lines from both crosseswould be acceptable in respect ofresistance.The above examples were chosen forheuristic reasons as showing. withtextbook clarity. the main features ofmore or less additive polygenicsystems. Complications are possibleof course: chosen scales maythemselves be non-linear andtherefore might need transformation;if appropriate scales were chosen.dominance/recessiveness would tendto produce skewness at F2 and latergenerations but the point aboutcontinuity still holds (see. forexample [4. Figure 1]); somedistributions defy any simpleinterpretation (e.g. [6. Table 5]). butstill leave the fact of polygenicinheritance and heritability ratherplain. (The data of [6. Table 5] looklike additive recessive resistancesfrom both parents giving asusceptible Fl and transgressive F2·)StrategiesGeneralThe general objective of all diseaseresistance components of plantbreeding programs must be toprovide adequate levels of resistancethat are reliable over the years. Ifvarieties are individually expected tosurvive in cultivation for manyyears. the need for durability isimplied and this would suggest theneed for HR or. perhaps. IR; ifimmobile pathogens were concerned.then reasonable durability couldsometimes be provided by VR. Formobile pathogens. VR wouldsometimes simply be a foolishchoice. as against leaf diseases ofevergreen perennial crops in the wettropics (2). Even for annual crops.VR is. in general a risky choice. ifdurability. in the sense of longsurvival of specific genotypes. is theobject. The accumulation of severalVR genes together (pyramiding) hassometimes been useful but often ithas not (as in potatoes againstblight. in rice against blast. and inother inbred cereals against rusts(e.g.. [6]). It may take several yearsfor a new virulent pathotype toappear but it is prudent to assumethat it will appear. however manyVR genes have been accumulated.and that it will do so sooner ratherthan later.Thus VR per se. even pyramided.cannot be expected to providedurability but it may still beoperationally fairly reliable ifdeployed effectively over time. newpathotypes being anticipated byappropriately resistant varieties. Todo this implies a high degree ofpathological and geneticalunderstanding and control. as isevidently available for wheat rusts inAustralia [I], but is by no meansuniversally feasible. One should notethat. although skillful deployment ofVR genes,.can. in favorablecircumstances. give long-continuedprotection of the crop at large. thissituation cannot be described as anexample of durability.VR in single. specific genotypesremains. in general. non-durable. VRalone thus has both uses andlimitations. If its use has to beabandoned or supplemented or if. asis true of many host-pathogencombinations. it is simply notavailable. what remains? Theprimary recourse is to HR and. asremarked above. this is the meansby which. in fact. most minor plant

127diseases are kept down to minorstatus and many major ones arecontrolled well enough. Despitemuch speculation to the contrary.there is no clear evidence ofsignificant erosion of polygenic HR,that is. of pathogenic adaptation orincrease in aggressiveness. There areinnumerable examples of varietiesthat continue at an unchangedmoderate to good level of resistancefor decades. thus satisfyingJohnson's [6] practical criterion ofdurability. namely survival.Polygenic HR is a perfectly generalsource of durability that VR per secannot offer (4. 7. 8. 9).IRlMR [7. 8] remains an attractivepossibility which is yet too littleinvestigated to allow judgment as togeneral utility. Obviously. it offers achance of using VR genes whichwould otherwise be useless or. atbest. of transient utility. But betterunderstanding is essential withparticular reference to the followingfeatures (5. 6): optimal (or necessary)numbers of resistance genes; theirdisposition between and withincomponent lines; the relativeimportance of VR and HR amongcomponents; durability; andagricultural features. such as seedsupply. Whatever the problems.though. the prospects are exciting.If durable resistance be the object.therefore. the general strategy isfairly clear: prefer polygenic HR asthe basic approach and use VR perse only if (as against immobilepathogens) there is a clear prospectof durability. Consider thedevelopment of populations thatexploit IRlMR as a supplement to abasic HR. For airborne pathogens.this means. in effect. that HR isfundamental and that VR should beavoided or deployed only underrigorous scientific control orexploited as an element of IRIMR.ParticularFeatures of VR/SR breeding-Aprogram committed to this approachwould have to accept the risksincurred. That these risks are real inrespect of the rusts of inbred cerealsis abundantly documented in severalchapters in this book (e.g.. [1] forstem rust of triticale in Australia. [6]for yellow rust in Europe. [7] forrecent leaf rust outbreaks in Texas).The breeder would try to pyramidVR genes. recognizing the need forexact genetic control of host genesand good knowledge of pathotypes ifthis were to be done efficiently andsafely [1. 2]. He would. further. try toenhance the genetic backgroundagainst which the VR genes work.though this is. in general. a difficult.sometimes Virtually impossible task[5.6]. The breeder would make useof a battery of methods. includingshuttle breeding selection.multilocation testing. hot spots.glasshouse tests. and so forth but hewould recognize that. in the limit.none of these can guaran teeprolonged survival of the products.Further. the breeder would hope toimpose some discipline upon theusers of his products. avoiding.above all. the deployment one-at-atimeof VR genes which might bemuch better pyramided (see examplein [3]). Fin

128term (which is likely), the breederwould take care to use appropriatevirulent pathotypes for testing (aspotato breeders have been doing formany years; see also examples inthis book [4, 5, 6]). The breederwould accept slowness of progressrelative to VR in the initial stagesbut he would recall that theheritability of HR usually turns out,when studied, to be high, even veryhigh, so progress would be fasterthan many workers might imagine.High heritability is documented in(4, 7, 9), in Figures 2 and 3 above,and in the striking selectionexperiment on barley leaf rustdescribed here by Parlevliet [5]. Onerecalls also that maize breeders havelong recognized the high heritabilityof HR to the leaf diseases (rusts andHelminthosporium) with which theyhave to deal. Once several cycles ofHR have been accomplished, thedisease resistance problem declines,because parents with fair HR andmore or less additive geneticvariance jointly ensure good averagelevels of HR in progeny. In theabsence of VR (or with itsnullification by the use of virulenttester-pathotypes) breeding thensimply becomes a matter of throwingaway the worst in each generation,as emphasized by Parlevliet [5] andas has long been known tosugarcane breeders (2). Thus thelonger term condition tends tostability, with only small resourcesbeing devoted to resistance per se.This is not the least attraction of awell established HR system.The breeder exploiting HR wouldhave to accept the idea of enoughresistance rather than immunity asthe objective and his farmercustomerswould usually have toaccept the presence of at least somedisease as being normal. In practicethis is true of very numerousdiseases and too much emphasiscan, I believe, be laid on the farmer'sdesire for an absolutely clean crop.Immunity is very rare. The breederwould generally have to be aware ofthe need for adaptation to varyinglocal levels of disease, of the use (andmisuse) of hot spots (it is all too easyto over-select), and of interactionsbetween neighboring plantings suchthat variation of disease scoresbetween genotypes tends to bediminished, the extremes ofsusceptibility and resistance beingdamped. But errors tend to beenhanced, thus depressingheritabilities.Features of IR/MR breeding­Little strategic thinking about thissubject is yet possible because ourknowledge is but rudimentary,relating mainly to oat multilines inthe USA and barley mixtures inEngland [7, 81. The very idea of nonuniformityof crop varieties, thoughnot new, has only recently begun togain wide acceptance, howeversensible it might seem biologically[7]. Superficially, then, it looks asthough the breeder of multilineswould have to have access to diversestocks at a good level of fieldperformance, canying diverse VRgenes. He would have to accept thatthe backcrossing programs to formconstituent lines would take time,might have to be extended to usenew VR genes, and that multilineswould therefore be liable to beovertaken in yield potential by laterpure lines. He would also have toaccept that there might be some(maybe not great) problems ofmaintenance and seed supply.The breeder exploiting varietymixtures [8] would face somewhatdifferent, and perhaps slightly lesser,problems. He would have to haveaccess to good varieties canyingdiverse VR genes, so such a program

129would fit naturally into a programthat already had a large stock offailed VR material (as forbarley/mildew and wheaUyellow rustcombinations in Europe [6, 8]). Thebreeder would have to ensure thatconstituents of mixtures werecompatible as to maturity andquality (though the latter sometimesmight not matter), would face heavyexperimentation in choosing goodcombinations, and would have towork out maintenance and seedproduction methods (which are yetunclear).In general. IRlMR looks particularlyattractive as a means of exploitingVR genes that have already lost theireffectiveness; thus it would helpgreatly in effecting the transitionfrom a VR system to an HR one. Butfurther speculation is vain. Thefundamental requirement is forbetter understanding (5, 6). forserious research, especially inagriculture at low latitudes andparticularly in regard to durability. Ifone had to guess as to a choicebetween mixtures and multilines, Isuspect (with Browning [7]) that theformer might be preferred because:first, diverse mixtures could often becomposed straight out of a breedingprogram without the necessity ofmuch laborious backcrossing;second, there might well be merit inheterogeneity per se besides thedisease control element; and, third,there could sometimes beopportunity to control two or morediseases simultaneously [8].potentially a feature of greatimportance with the wheat rusts. Inconnection with the second pointabove, one notes that, in Wolfe'swork with barley [8]. the proportionof the average yield gain of 8% thatis to be attributed to mildew controlis undeterminable (though possiblylarge).Wheat Rust ContextIn this section I attempt tosummarize, from the informationpresented in preceding chapters,what I take to be the main featuresof the three wheat rusts in relationto their control by breeding.Stem (black) rust (P. graminisf.sp. trltlcl)Ecologically, stem rust ischaracteristic of areas where wheatmatures in hot, dry conditions. Ithas, on occasion, been devastatingbut, in the last few decades,epidemics have been pretty wellcontrolled (sometimes even averted[lD by the systematic use of VR (Sr)genes built up over many years by ahuge body of research in manycountries. STR is probably the bestresearched of all plant diseases.Many of the 30-40 Sr genes knownhave failed, some of them more orless immediately, but a few havelasted longer. Sr2, isolated decadesago in Hope from an emmer cross,seems to be exceptional in havingpersisted usefully for a very longtime (I.e. seems to be durable) [1. 2];it has an incomplete adult plantresistance type of reaction and isvery widely spread (e .g., in CIMMYTwheats [9]). It has something of anNR about it and is presumably not asingle locu~ but a linked block. It isnormally not strong enough by itselfbut is valuable in a pyramidedbackground. Of the other Sr genes,Sr24, 26, 30, 31, and 36 have beenmore than averagely useful [1, 2].Many Sr genes are exotic, havingbeen transferred from eight speciesother than T. aestivum [2], a sign ofthe superb cytogenetic controlpossible in wheat, unmatched in anyother crop.The multiple Sr system has workedpretty well, by virtue of excellentgenetics and critical pathotypesurveys especially well developed in

130Australia, even to the point ofidentifying the next pathotype beforeit happened [1]. This is far fromrandom assortment of pyramided VRgenes. Surveys show that multiplepathogenicities are common in rustisolates: for example up to eight inMexico in 1984-86 [9, Table 11].Mere pyramiding. of itself, is clearlyhelpless in only the slightly longerrun against stem rust (and indeed forthe other two rusts. as well). Thatseemingly good resistances can gothrough extensive internationaltesting successfully and still fall toan unexpected new pathogenicity isevident from recent Australianexperience with Sr27 in triticale [1].There is no evident reason to mewhy the same should not happen todurum and bread wheats and moresuch episodes must surely beexpected.The question for the future iswhether continuation of thehistorically pretty successful VRprogram will be appropriate. Will thebreeders run out of Sr genes? Or canone foresee an indefinitecontinuation of the currentprocedure. adding new Sr genes(often from alien sources)successively. as dictated bypathotype surveys? An alternative.not yet seriously entertained. I think.would be to initiate a move towardsthe use of HR. This would involve amajor re-orientation but mightdevelop into an important provisionfor the future. HR against stem rustis not yet deeply researched butthere is clear evidence that it can bedeveloped [4] and also a strong apriori presumption to that effect.Leaf(bro~) rust(P. recondita)Leaf rust is a disease of warm ratherthan hot places. worldwide indistribution. and generally judged tobe. nowadays. economically the mostimportant of the three rusts [2].Some 30-40 Lr genes are known;they resemble the Sr genes incoming from diverse sources (sevenspecies other than T. aestivum andthree genera other than Triticum[2]). Pathotypic complexity seems tobe as great as for stem rust [2. 9). Ofthe Lr genes in use. Lr13 associatedwith several other elements(especially Lr34) is prominent (9) andLr19 is is thought by some workers(but not all) to be promising [2].though not yet much exploited. TheCIMMYT workers tend to regardLr13 in much the same light as Sr2.having something of an NR about it.The control exercised by breedingappears to be less securelyestablished than for stem rust andthe longer term durability ofpyramided Lr genes must beregarded as dubious. It isunfortunate that this meetingaccorded relatively little attention tothe disease despite its importanceand despite the fact that there is agood deal of information availableabout it. especially from NorthAmerica (McIntosh. pers. comm.).Yellow (stripe) rust(P. striiformis)This disease is characteristic ofcooler places than the other tworusts. It is worldwide in temperatelatitudes and at high altitudes in thetropics. but it only reached Australiain 1979 [3]. Somewhat fewer VRgenes are known than for the otherrusts; Yrl-15 have been identified [6]but no doubt many more could bepicked up or brought in from aliensources if required. The Yr geneshave a long history of failure [3. 6]and no signs of either durability orsustained control by pyramiding areapparent [6]. The race structure iscomplex and. as for all the rusts.constantly shifting; it is repeatedlyapparent that the new pathotypesare evoked by the Yr genes in thecultivars grown [3].

131So complete has been the failure ofVR breeding against yellow rust thatserious attention has been paid toalternatives. The historical evidencein favor of durable resistance inwheats (mostly but not exclusivelyEuropean) is set out by Johnson [6]and is, In my opinion, persuasive.The genetic nature is not wellestablished but a substantial HRelement seems clear, probablyaccompanied by miscellaneousresidual effects connected with majorgenes. Some of the durableresistance occurs in European landrace materials which would notnormally be favored by breeders asparental material. Johnson [6] makesvaluable observations on simplebreeding strategies whereby suchresistance could be built up into aform usable in modern breedingprograms. He also makessuggestions as to how useless (ornuisance) VR genes can be got rid ofor nullified, the better to exploit HR(see also [5]). Broadly. the prospectfor practical levels of HR routinelyincorporated in excellent varietiesseem good.An analogous situation holds inbarley in relation to its leaf rust.Parlevliet's [5] observation of durableHR in established cultivars and of anoutstandingly good response to jointselection for resistance and yield ([5],Table 3) points the way to managingthe disease in barley and neatlycomplements Johnson's [6]observations on exploiting HR inotherwise unexciting parents.CIMMYT Rust ProgramsOutline of practiceBased on Chapter 9 and referencestherein, the bulk of the CIMMYTeffort is founded upon pyramidingmajor genes (whether hypersensitivitiesor adult plantresistances) rather few of which arestrictly identified by reference topathogenicity tests on standardfungal isolates. The procedures usedare essentially shuttles in Mexico(Toluca-Obregon, with special sitesalso for leaf rust), Internationalshuttles, and multilocational testingof the more advanced stocks. So faras they are available, mixtures ofcomplex pathotypes and spreaderrows are used.Some race surveys are done inMexico for stem and leaf rusts (see[9J Tables 11 and 12) andconsiderable genetic complexity ofthe fungi is thereby revealed; but, asemerged in discussion, there is not,and cannot be, any guarantee thateither shuttling or multilocationaltesting will always reveal all the(sometimes critical) specificpathogenicities. The selection ofmaterials on the basis of low averagecoefficients of infection and theabsence of high average coefficientsof infection, but with little specificknowledge of host genetics orpathogen specificities, means thatthe pyramiding process is essentiallyrandom.Of the three rusts, leaf rust currentlyreceives the most weight; yellow rustis a relatively recent addition to theprogram [9]; and the stem rustsituation at present seems to befairly stable [1] . At least a partialreason for the last point seems to bethat many CIMMYT materials carrySr2 and Lr13 which, unlike most ofthe other genes with which we areconcerned, appear to have somethingof the character of an NR (see stemrust under the "Wheat RustContext" section above) and workwell with other components. As tothe recent history of durability, onerecalls that. in Australia and theUSA, really damaging outbreaks ofstem rust have largely beenprevented in the past 30 years (butsometimes fairly narrowly [I]) bytimely genetic analysis of wheat

132varieties and of fungal races.followed by appropriate varietalsubstitution. There was. however. anoutbreak in southern Australia in1973 due. not to failure of VR. butsimply to growing a susceptiblecultivar (McIntosh. pers. comm.). Onthe whole. control has been rathergood. but the recent [1] experience inAustralia of triticale susceptible tostem rust with pathogenicity forSr27 should surely serve as areminder that much may dependupon unforeseen (usuallyunforeseeable) vulnerability of majorgenes (VR). It is by no means clear(to me. at least) that pyramided VRto stem rust should generate anyconfidence in durability. as distinctfrom some years survival followed byenforced varietal replacement.As to leaf rust. this is clearlyCIMMYT's greatest current preoccupation.As I read the literatureand heard the discussions. resistanceseems to depend much uponcombinations of Lr13 with othergenes (e.g.. Lr34 [2]) The cultivarsCiano 79. Tonichi. Pavon 76. andGenaro 81 which have been standingup well to leaf rust [9] depend uponthe above in part and in part uponother genes such as Lr16 and Lr26(McIntosh. Roelfs. pers. comm.). Thepotential of Lr19 may beconsiderable [2]. but it carries anundesirable association with yellowflour color. As to yellow rust.epidemics since the mid 1970s inwheats in California and Pakistanand in Andean barleys serve asreminders that this rust seems to beas adaptable to host VR as theothers. a repeated experienceelsewhere. as described by Stubbs [3]and Johnson [6].As a complement to the pyramidingprocess. the CIMMYT program alsoseeks slow rusting (termed dilatoryresistance) of a major gene (adultplant resistance) charactermentioned in Chapters 1 and 2.Thus Pavon 76 and Genaro 81 haveat least some VR genes betweenthem [9. Table 10] and also slowrustingcomponents [9. Figure 2].Some effort also goes into attemptsto backcross VR hypersensitivegenes into a slow-rusting (dilatoryresistance) background.There has. in the past. been someeffort to construct multilines (e.g..materials based on backcrosses of Lrgenes into 8156 derivatives such asSiete Cerros) and the work continueswith the present use of Seri 82 andGenaro 81 as recurrent parents [9].So far. varietal mixtures have notbeen exploited in this context but itwas good to learn that a jOintCIMMYTICambridge venture in thisarea ','las in hand ([8] and see alsosection on particular strategies).Alien sources of resistance genes arealso being exploited. as they havebeen very extensively exploited inthe past [2. Tables 5 and 6]. There isno reason to think [6] that they willprovide genes that confer resistanceany more durable than usual VR.ObservationsSocial context-It would generallybe conceded that. for small farmersin the Third World. and even formany not-so-small ones. not onlyhigh yields but yields stable overseasons are important. Occasionaldisasters can be very damaging.Epidemics that the farmers of adeveloped country could bear orcould afford to control by sprayingcould be economically hurtful. evendisastrous. to the small farmer.Furthermore. in developed countries.an unforeseen epidemic can often bequickly met by variety substitutionin the next season. so that the effectis transient. In the Third World.there can but rarely be any

133guarantee that resistant successorvarieties and the means to distributethem will be available quicklyenough. In general, by intensescientific effort and muchexpenditure, the developed countrieshave kept the wheat rusts under fairto good control and when thingshave gone wrong (as they notinfrequently have) rich agriculturescould bear the losses. It is otherwisein the Third World; the sameintensity of scientific effort is notthere and the socioeconomicconsequences of failure are worse.One must conclude, I think, thatdeliberate attention to long-termstability of disease control is welljustified.Technical features-Broadly,CIMMYT breeding produces a goodflow of high-yielding wheats that areresponsive to high inputs, well likedby growers, and reasonably rustresistantat the time of release.However, the resistance breedingstrategy has heretofore rested onpyramided VR genes with littlegenetic control of what genes areused, limited pathotype information,and therefore little chance ofanticipating epidemics in the mannersometimes possible in, say, Australia[1]. Efforts to introduce more durableresistance also rest, as we have seenabove, on the use of adult plantresistance-type major genes which. itwould be prudent to assume, arelikely to be as much pathotypespecificin effect as typical seedlinghypersensitivities [1, 2]. One has toconclude, I think, that risks ofunpredicted, indeed unpredictable,epidemics are real. Some. not so farvery damaging ones, have happenedin the past and more must beassumed possible in the future.Of the strategies that might promotelonger term stability of cropperformance by promoting durabilityof reSistance, one (lRlMR) is receivingsome attention and the other (HR)apparently none.Conclusions-I conclude that, in theinterests of stability of cropping,some shift of emphasis would bedesirable. The following elements areapparent:(1) Commit a substantial effort tobuilding up polygenic HR in aproportion of breeding stocks bydeliberately discarding majorgenes, by using appropriate landrace materials [5, 6], and bytesting with virulent races. Thesemethods have worked in diversecrops (4) and are already beingapplied to wheat in relation toyellow rust [6]. The productswould have to be worked up on abroad genetic base, underenhanced recombination, to alevel of performance at whichthey would be acceptable asparents; that good progress isindeed possible withoutprohibitive labor is well knownin potatoes and neatly illustratedhere by Parlevliet's [5]experiments with barley. Wheatcould be more difficult becauseof the complex background of illdefinedVR genes but cannot beimpossible to handle, asindicated by Knott's preliminaryresults [4].(2) Test the potential of IRlMR forcontrolling wheat rusts at lowlatitudes. Since some work isalready in hand, little is impliedhere beyond an enhanced effortand a very deliberate effort todetermine what has long beenthe subject of speculation butnot yet of test, namely durabilityof resistance. Much research onthis, on principles of constructionof multilines and mixturesand on their deployment andmaintenance is needed.

134(3) Strengthen the genetic controlexercised over the VR genesbecause this would be valuablein three respects. namely: inenabling more criticaldeployment of VR genes inrelation to pathotypes during theperiod (which must extend overa good many years) in which theprogram is still dependent uponpyramided VR; in aSSisting inthe elimination or nullification ofunwanted VR genes in the HRprograms; and in assisting in theconstruction of populations thatexploit VR genes in multilines ormixtures.In summary. the conclusions arethat there should be a move awayfrom the exploitation of pyramidedVR towards research on thedevelopment of HR and IRiMR. Inthe long term. it would be sociallyfavorable but. so great is thecommitment worldwide topyramided VR in wheat. that theshift could not come rapidly; there istherefore all the more reason to startsoon.Wider ContextThe general style of the CIMMYTwheat program has so far beenintensely practical and.overwhelmingly. the biggest outputhas been a stream of excellent newwheat varieties. having. in total. ahuge practical impact. A strongadvance in underlying yield potentialhaving thus been secured. I believethat the time is ripe for a shifttowards enhanced understanding ofwheat breeding strategies. including.of course. a substantial elementrelating to stable disease resistance.The conclusions outlined on thispoint in the preceding section mightseem far-reaching but are not in factat all remote from current CIMMYTpre-occupations [91 with theimportance of stability of resistance.the need for HR. and the exploitationand study of IR.If these arguments be accepted. aconsiderable shift of emphasis. amoving of resources from practicalbreeding towards strategiC research.would be implied: strategiC research.not basic because no agriculturalresearch can ever sensibly bedescribed as basic.Two implications follow (Figure 4).First. the practical breeding effort.the flow of varieties. would declineas resources shifted. leaving thenational systems (Figure 4) withgreater local responsibilities fordomestic progress. This has longbeen foreseen by the internationalcenters as a whole as a natural longtermprogression. As nationalprograms develop. the Centers'responsibilities would move.according to conventional doctrine.away from short-term practicalendeavors towards becoming centersof scientific and training excellence.supportive of the national programs'activities. My suggestion would thusseem to be concordant with officialdoctrine. Second. the scientific tasksare substantial and it cannot beexpected that even a great institutesuch as CIMMYT could cover therelevant field of strategiC research onits own. CIMMYT already hasnumerous collaborative researchprojects with laboratories indeveloped countries. There must bevery manY-SCientists and skills inthose laboratories that could beadapted to CIMMYT's emergentstrategic research needs. Lest I bemisunderstood here. I shouldemphasize that I am not talkingabout molecular biology or geneticengineering which is yet. I believe.irrelevant to plant breeding.whatever the long-term promisemight be. The research. however.could well have biotechnologicalcomponents since this field isstarting to throw up powerfuldiagnostic techniques of greatpotential for routine applications.

135The conclusion is plain: adapt the implied here is that, more often thanexisting network quite explicitly to heretofore, CIMMYT identifies ansupporting and enhancing CIMMYT's area needing study and goes out toin-house research program (Figure find and use the expertise it needs in4). At present, projects often come to a laboratory overseas.ClMMYT from outside; all that isin-house researchat strategic levelresearchinstitutes indevelopedcountriescollaborativeresearchnetworkCIMMYTI outreachI breedingshuttles andmultilocationaltestingIInational systemsin LDCsagriculturaldevelopmentFigure 4. CIMMYT: The general context of breeding and researchprograms.

136The change proposed is not, in fact,profound. Rather it represents a shiftof emphasis, as Figure 4 makesclear. I am confident that, In thelonger run, such a shift wouldbenefit, not only the customers,wheat growers, and consumers, butalso CIMMYT itself, adding scientificluster to a name already famous forgreat practical achievements.References1. FAO. 1984. Breeding for DurableDisease and Pest Resistance. Rome,F AO (F AO Plant Production andProtection Paper, 55, pp. 167) (2nded. 1986).2. FAO. 1986. Breeding for DurableResistance in Perennial Crops. Rome,FAO (FAO Plant Production andProtection Paper, 70, pp. 130).3. Jones, I.T., H. Sethar, andI.J.E.R. Davies. 1981. Genetics ofpartial resistance to barley mildew.Barley Genetics IV. Proceedings ofthe Fourth International BarleyGenetics Symposium. Edinburgh,University Press, 449·457.4. Lamberti, F., J.M. Waller, andN.A. Van der Graaff (eds.). 1983.Durable Resistance in Crops. NewYork and London, Plenum Press(NATO/AS I Series 55).5. Marshall, D.R., and A.J. Pryor.1978. Multiline varieties and diseasecontrol I. The 'dirty crop' approachwith each component carrying asingle unique resistance gene.Theoretical and Applied Genetics.51: 177·184.6. Marshall, D.R.. and A.J. Pryor.1979. Multiline varieties and diseasecontrol II. The 'dirty crop' approachwith components carrying two ormore resistance genes. Euphytica,28: 145·159.7. Simmonds, N.W. 1979.Principles of Crop Improvement.London, Longman.8. Simmonds, N.W. 1983. Strategyof disease resistance breeding. FAGPlant Protection Bulletin, 33:2·10.9 . Simmonds, N.W. 1985. A plantbreeder's perspective of durableresistance. FAG Plant ProtectionBulletin, 33:13·17.10. Smith, G.S., and J.A. Clarke.1933. Inheritance of stem rustreaction and correlation of charactersin Pentad, Nodak, and Akronadurum wheat crosses. USDATechnical Bulletin, 385. pp.27.

137Capitulo 1La funclan de genes especificos en el mejoramlento paraobtener en el trigo y el triticale resistencia durable a la royadeltalloR.A. McIntosh. Instituto de Fitogenetica. Castle Hill. AustraliaResumenHan tenido exito los intentos fitotecnicos para obtener resistencia ala roya.Los tipos de resistencia logrados en la agricultura han dependido de genesidentificables imicos 0 de combinaciones de esos genes. El gen de resistenciaen la planta adulta. Sr2. ha contribuido a la obtencion de una resistenciadurable en muchas zonas. Los trigos resistentes a la roya del tallo para laszonas de cultivo del cereal en el nordeste de Australia se han basado en elempleo de tipos de resistencia que se reemplazan despues de detectarpatogenos virulentos. Un manejo tal de genes requiere encuestas sobre lapatogenicidad, el conocimiento de los genes presentes en las variedades detrigo y la cooperacion de la industria para un reemplazo rapido de lasvariedades. La vulnerabilidad genetica a la roya del tallo en el programa detriticale del CIMMYT se podria reducir utilizando informacion obtenida enAustralia. La estrecha base genetica de la resistencia se podria ampliarrecurriendo a variedades europeas de triticale, centeno y trigo. No obstante.es preciso conservar la diversidad genetica entre el trigo y el triticale.Capitulo 2Reslstencla a las royas de la hoja y del tallo en el trigoA.P. Roelfs. Laboratorio de Royas de los Cereales. Servicio de Investigacionesdel Departamento de Agricultura de los Estados Unidos y Universidad deMinnesota. St. Paul. MinnesotaResumenSi bien las royas de los cereales han logrado dominar un gran numero de lasvariedades resistentes obtenidas en los ultimos 80 aiios, muchas otrasvariedades se han cultivado con exito en grandes extensiones de tierra. Seha combatido la roya del tallo usando combinaciones de resistencia queincluyen el gen Sr2 transferido por McFadden de la escanda a las variedadesHope y H-44 en 1923. Las resistencias conferidas por Sr26 (proveniente deAgropyron elongatum), Sr31 (de Secale cereale) y Sr36 (de Triticum .timopheevii) parecen ser las resistencias causadas por un solo gen maseficaces en todo el mundo. La variedad Thatcher (con resistencia provenientede T. durum), obtenida por Hayes et a1. en 1934, tiene tambien un gradoadecuado de resistencia en la mayoria de las zonas. Se ha combatido conexito la roya de la hoja mediante una combinacion de los genes Lrl3 y 34.Se usaron por primera vez estos tipos de resistencia en las variedadesFrontana (Brasil, 1934) y Americano 44D (Uruguay, 1918). Se continuautilizando esta combinacion de genes en variedades durables recientes comoChris, Era, Ciano 67, Pavon 76, etc. Las suposiciones acerca de la genetica yla durabilidad de algunos tipos de resistencia han obstaculizado la selecciony obtencion de variedades resistentes.

138Capitulo 3Analisis de la patogenicidad de la roya amarilla (lineal) deltrigo y su importancia en el contexto mundialR.W. Stubbs. Instituto de Investigaciones para la Protecci6n de las Plantas.Wageningen. Paises BajosResumenEl lnstituto de Investigaciones para la Protecci6n de las Plantas (IPO) estudiaa nivel internacional la roya amarilla (lineal) del trigo (Puccinia striiformisWestend. f.sp. tritici). En condiciones controladas. las razas (virulencias) seidentifican en las plantulas de un ampIio conjunto de variedadesdiferenciales "antiguas" y "nuevas" con ciertos genes de resistencia. algunosconocidos. pero en su mayoria desconocidos. En viveros para observar lasrazas (parcelas separadas del campo), se analiza la virulencia vinculada conla resistencia de la planta adulta, especffica para cada raza. Aunque todaviano se ha estudiado suficientemente. es evidente la relaci6n entre ladistribuci6n de la virulencia de los agentes pat6genos y los grados deresistencia de los huespedes. Se presentan los resultados de un estudio de laroya amarilla que infecta los triticales y variedades con resistencia derivadadel centeno y se describe la distribuci6n por zonas de las razas de royaamarilla en Europa, Africa. Asia y America del Sur. Se recomienda investigaren forma continua las modificaciones de la patogenicidad en las poblacionesde las razas con el prop6sito de mejorar la resistencia y evaluarla en loshuespedes en distintos medios.Capitulo 4Empleo de la resistencia poligenica para mejorar la resistenciaa la roya del tallo en el trigoD.R. Knott. Departamento de Ciencia de los Cultivos y Fitoecologia.Universidad de Saskatchewan. Saskatoon. CanadaResumenDesde hace muchos alios se conoce la resistencia multigenica a las royas deltallo. Se ha postulado la existencia de una resistencia no especffica a laenfermedad. pero es dificil demostrarla. Varios genes que producen pequenosefectos a menudo determinan la resistencia parcial y la resistencia dilatoria ala enfermedad, que a veces se consideran no especfficas. En investigacionesreaIizadas en Saskatoon. se obtuvieron Iineas de trigo que como plantulascarecfan de resistencia a la raza 15B-1. pero que mostraron buenaresistencia a la misma raza en el campo. Se comprob6 que su resistenciaestaba determinada por tres a cinco genes recesivos. cada uno de los cualestenia un efecto pequeno. Los genes reducian el periodo de latencia y elnumero y tamali0 de las pustulas. Es probable que la resistenciadeterminada por varios genes de efectos pequenos sea relativamenteduradera. sin importar que esa resistencia sea 0 no especffica. Aunque esdificil su empleo. la resistencia poIigenica podria ser de gran utiIidad en losprogramas de fitomejoramiento de trigo.

139Capitulo 5Estrategias para utilizar la resistencia parcial en la luchacontra las royas de los cerealesJ .E. Parlevliet. Departamento de Fitogenetica. Universidad Agricola.Wageningen. Paises BajosResumenEn los cereales. toda resistencia a las royas que los afectan es de tipoespecifico para la especie. es decir. la resistencia es eficaz s6lo en relaci6ncon una especie de roya. Se pueden distinguir dos tipos de resistenciaespecifica para la especie contra cada agente pat6geno de la roya: 1) un tipohlpersensible de resistencia. determinada por genes mayores. que secaracteriza por bajos niveles de infecci6n. la especificidad para una raza y lafalta de durabilidad; ii) un tipo cuantitativo de resistencia (resistenciaparcial). caracterizado por una tasa reducida de acumulaci6n epidemica apesar de ser susceptible y presentar un alto nivel de infecci6n. la ausencia degrandes efectos especificos para la raza (si bien se producen efectospequenos) y la durabilidad. Cuando no intervienen genes mayores. es facil laselecci6n para obtener resistencia parcial. Aun una selecci6n poco rigurosapara elimlnar la sensibilidad resulta. cuando se apJica sistematicamente.muy eficaz para acumular genes que determinan la resistencia parcial. Estaselecci6n poco rigurosa permite al fitogenetista obtener al mismo tiempootras caracteristicas. Cuando se pretende aumentar la resistencia parcial enpresencia de genes mayores que no han sido neutralizados por completo porel agente pat6geno. la eficacia de la selecci6n es conslderablemente menor.Siempre que sea posible. el fitogenetista debe exponer la poblaci6n huespeda una sola raza del pat6geno. una raza que neutralice una cantidad maximade genes mayores. En cada etapa de la selecci6n. el fitogenetlsta debeeliminar de esa poblaci6n huesped los genotipos mas sensibles y tamblenaquellos que presenten un bajo nivel de infecci6n. Cuando es demasladodificil clasificar de manera confiable los niveles de infecci6n. el fitogenetistadebe eliminar los genotipos mas resistentes junto con los mas susceptibles.ya que se supone que los primeros son portadores de genes mayores. Enalgunos casos. no es posible controlar la poblaci6n pat6gena a la que estaexpuesto el huesped porque esta constituida por una mezcla de rozas. Enestas circunstancias es muy dificilla selecci6n para obtener resistenciaparcial. La eliminaci6n continua de las lineas mas susceptibles y de las quecasl no resultan afectadas favorecera la resistencia parcial. pero el progresopuede ser mas lento de 10 esperado.

140Capitulo 6Resistencia durable a la roya amarilla (lineal) en el trigo y susrepercusiones en la fitogeneticaR. Johnson, Instituto de Fitogenetica, Cambridge, InglaterraResumenLa roya amarilla (lineal), causada por el parasito obligado Pucciniastriiformis, se encuentra dondequiera que se cultive el trigo en climasfrescos. 5e han identificado varios genes de resistencia especifica para cadaraza, eficaces en las plantulas de trigo, pero aun resta identificar otros. Esosgenes de resistencia pueden ser dominantes 0 recesivos y tanto el mediocomo el fonda genetico influyen mucho en la expresion de algunos de ellos.La resistencia que se desarrolla despues de la etapa de plantula a menudo estambien especifica para la raza. La combinacion de genes de resistenciaespecificos para cada raza no ha tenido exito como metodo para combatir laroya amarilla en Gran Bretana. No obstante, a veces la resistenciadesarrollada despues de la etapa de plantula no muestra especificidad paralas razas incluso despues de pruebas prolongadas y en diversos lugares. Esaresistencia durable solo puede distinguirse de la resistencia de las plantasadultas especifica para la raza mediante pruebas prolongadas. 5i bien esaresistencia puede estar sometida a un control genetico complejo, se puedeusar en programas de mejoramiento como los descritos en este capitulo; sinembargo, no es posible garantizar la durabilidad de la resistencia producidaen tales programas. En consecuencia, es preciso vigilar todas las variedadesresistentes nuevas, cualquiera que sea el metodo de mejoramiento empleado,para detectar la existencia de razas patogenas con patogenicidad equivalente.Capitulo 7Ideas actuales sobre el empleo de la diversidad para protegerlos cereales de agentes patogenos foliares muy epidemicos yvariables: Problemas y perspectivas para el futuroJ.A. Browning, Departamento de Fitopatologia y Microbiologia, Estaci6nAgricola Experimental de Texas, College Station, TexasResumenLas poblaciones aut6ctonas de progenitores de cereales y sus parasitosobligados, que han evolucionado paralelamente, son estabilizadas por laresistencia dilatoria epidemiologica, producida geneticamente por laresistencia general que incluye genes mayores de resistencia especifica endistintas configuraciones genicoespaciales. La resistencia general protege a la

141planta; la resistencia especifica protege a la poblaci6n al minimizar laagresividad del agente pat6geno ajustando la delicadamente armonizadainteracci6n de realimentaci6n entre los sistemas geneticos del huesped y elagente pat6geno, muy vinculados entre s1. Tanto la resistencia general comola especifica deben usarse en sistemas agricolas coherentes con sus origenesevolutivos. No obstante, con frecuencia la agricultura ha dependido de losgenes de resistencia especifica no para proteger la poblaci6n, como ocurre enla naturaleza, sino para defender la planta de pat6genos foliares muyepidemicos, como ha sucedido cuando se emple6 el gen de resistencia envariedades cultivadas en poblaciones homogeneas en zonas extensas. Por elcontrario, en programas realizados en Colombia, Ingla terra , los Paises Bajos,la India y los Estados Unidos (Iowa y Washington), se han incorporado genesde resistencia especifica en poblaciones diversas y se ha logrado laprotecci6n de la poblaci6n con una reststencia de escasamente un tercio.Esta pequena resistencia ha estabilizado las poblaciones en un ecosistemanatural varia do y, por consiguiente, la cifra parece real. Los datoscorroborativos obtenidos en ecosistemas naturales y agricolas indican que unsistema de manejo global de los genes que incluya la diversidad, comosucede en la naturaleza, vaticina el logro de una resistencia verdaderamentedurable. A pesar de los numerosos beneficios que aporta la diversidad, se hautilizado relativamente muy poco, en esencia a causa de i) un problemaparadigmatico y ii) un error criptico de la experimentaci6n, que a menudo hahecho que se subestime el valor potencial de la resistencia y se desista de suempleo. Con las mezc1as de tres variedades se satisface de manera adecuadala legitima necesidad de conservar la uniformidad agron6mica y al mismotiempo se aprovechan los beneficios de la diversidad.Capitulo 8Uso de mezc1as varietales para combatir enfermedades yestabilizar el rendimientoM.S. Wolfe, Instituto de Fitogenetica, Cambridge, InglaterraResumenSe describen brevemente los principales factores que conducen a la perdidade eficacia de la resistencia a las enfermedades y de los fungicidas en laagricultura europea actual. Se analiza con cierto detalle el empleo demezcIas varie tales , una de las opciones de que disponen el fitogenetista y elagricultor para mejorar la situaci6n. Numerosos datos obtenidos en ensayossobre el terreno revelan las ventajas que este sistema sencillo aporta alcontrol de las enfermedades y al aumento y estabilidad del rendimiento;dicho sistema puede sumarse a cualquier otro metodo de lucha contra lasenfermedades.

142Capitulo 9Metodos actuales del CIMMYT para mejorar la resistencia a laroya en el trigoS. Rajaram, R.P. Singh y E. Torres, Programa de Trigo, CIMMYT, MexicoResumenEn mas de 50 millones de hectareas del mundo en desarrollo se cultivanvariedades de trigo derivadas del germoplasma del CIMMYT. Como estosmateriales se siembran en una superficie tan extensa, que probablementeincrementara en 'el futuro, la politica de mejoramiento del CIMMYT haconsistido en conservar y aumentar la diversidad de la resistencia a las royasen el germoplasma de trigo. Las pruebas internacionales en sitios multipleshan representado una importante contribucion para verificar esta diversidadgenetica. Ademas de esas pruebas, el CIMMYT emplea algunos analisisgeneticos en su estrategia de mejoramiento. En el CIMMYT se piensa que lautilizacion de la resistencia vertical (RV) especifica para el patogenoproducida por genes mayores, como la describe Simmonds (Capitulo 10),podria llevar a situaciones precarias; como una alternativa para lainterminable incorporacion de genes 0 combinaciones de genes deresistencia, el CIMMYT ha in ten tado y recomienda el mejoramiento paraobtener 10 que Simmonds describe en el Capitulo 10 como resistenciahorizontal (RH) poligenica, no especifica para el patogeno, que prometedurabilidad. En el contexto mundial, la resistencia durable (0 estabilidad) yla diversidad genetica tienen una enorme importancia en el programa demejoramiento del CIMMYT. La situacion ideal seria identificar como base ungen 0 un conjunto de genes que quiza haya proporcionado resistenciadurable, y luego combinar continuamente genes de resistencia adicionalespara asegurar la diversidad genetica. La resistencia a la roya del tallo(complejo Sr2) derivada de la variedad Hope y la resistencia a 1a roya de lahoja (complejo Lr13) derivada de la variedad Frontana son la base de ladurabilidad de la resistencia a esas dos enfermedades en el germoplasma delCIMMYT. En cuanto ala roya amarilla (lineal), se ha informado que lavariedad Anza producida por el CIMMYT tiene resistencia durable a laenfermedad. El CIMMYT ordinariamente identifica en el campo lineas conresistencia parcial (avance lento de la enfermedad) y se opina que esteesfuerzo ha contribuido mucho a mantener los rendimientos del trigo entodo el mundo. El CIMMYT ha buscado la resistencia basada en multiplesgenes mayores solo como una estrategia complementaria y ahora investigaactivamente la posibilidad de volver a desarrollar y usar compuestos demultilineas y mezclas varietales.

143Capitulo 10Sintesis: La estrategia para mejorar la resistencia a las royasN.W. Simmonds. Escuela de Agricultura de Edimburgo. Edimburgo. EscociaResumenEl objeto de mejorar la resistencia a las enfermedades es proteger la biomasay. por consiguiente. el rendimiento del cultivo. La resistencia es simplementeun estado de "menos enfermedad " y la ausencia de enfermedad (inmunidad)rara vez constituye un objetivo realista. Para flnes practicos. se puedendistinguir cuatro tipos de resistencia: 1) la resistencia vertical (RV) especificapara el pat6geno, producida por genes mayores; 2) la resistencia horizontal(RH) poligenica. no especifica para el pat6geno; 3) la resistencia producidapor genes mayores no especifica para el pat6geno (RN). y 4) la resistenciacombinada 0 de interacci6n (Rl). La RV a menudo es eficaz contra agentespat6genos inm6viles. pero en general no muestra efectos durables contra losagentes pat6genos m6viles. transmitidos por el aire; no obstante. a veces esposible el control continuo de las enfermedades mediante el empleo de genesmultiples (acumulados) de RV. incorporados con un estricto manejo geneticoy patol6gico. La RH es durable y. por 10 general. en gran medida heredable:en muchos cultivos se controlan las enfermedades de este modo. La RN esvaliosa pero poco frecuente. Se sabe poco acerca de la Rl. causadaesencialmente por elementos heterogeneos de RV. pero es probable quetenga mas valor del que en general se Ie atribuye.En cuanto a las royas del trigo. durante aiios se ha combatido con exito laroya del tallo en todo el mundo empleando genes acumulados de RV. pero espreciso tener en cuenta que aun podrian presentarse brotes epidemicos de laenfermedad si se interrumpiera el control genetico y patol6gico estricto. Laroya de la hoja es probablemente la mas perjudicial de las tres royas en laactualidad; la base genetica de la RV es estrecha y se pueden esperar masepifitias. En el caso de la roya amarilla, a diferencia de 10 que sucede con lasotras dos royas. la RV parece haber fracasado por completo y. enconsecuencia. los fitogenetistas europeos han comenzado a abandonarla. Sereconoce la eficacia de la RH y se ha comenzado a explotarla.La labor fitogenetica del CIMMYThasta ahora se ha

144Chapitre 1Le role de genes specifiques dans l'amelioration de varietes deble et de triticale dotees de resistance durable Ii la rouillenoireR.A. McIntosh, Institut de phytogenetique, Castle Hill, AustralieResumeLa selection conduite pour obtenir une resistance a la rouille noire (ou rouillede la tige) a ete couronnee de succes. Les types de resistance employesjusqu 'alors en agriculture dependaient d'un seul gene identifiable ou decombinaisons de genes identifiables. Dans beaucoup de regions, Ie gene Sr2a contribue a conferer une resistance durable a la plante adulte. Dans leszones cerealieres du nord-ouest de J'Australie, cette resistance a la rouiJIe aete basee sur J'usage de formes resistantes qui sont remplacees au fur et amesure de la detection de pathotypes virulents. Dans ce sens, la selection degenes est etroitement liee aux etudes en matiere de pathogenicite, alaconnaissance des genes presents dans les cultivars de ble et a la cooperationde l'industrie pour remplacer rapidement ces dernieres. La vulnerabilitegenetique a la rouiJIe dans Ie programme triticale du CIMMYT pourrait etrereduite en mettant a profit l'information provenant d 'Australie. L'etroite basegenetique associee aux mecanismes de resistance pourrait etre elargie gracea J'emploi de varietes europeenes de ble et de triticale et de seigle. Toutefois,il est bon de preserver la diversite genetique qui differencie Ie ble et Ietriticale.Chapitre 2Resistance du ble ala rouille brune des feuilles et de la tigeA.P. Roelfs, Cereal Rust Laboratory, Service de recherches du Departementde l'agriculture des Etats-Unis et Universite du Mi.nnesota, St. Paul,Minnesota .ResumeBien qu 'au cours des 80 dernieres annees les rouiJIes des cereales aient suaffecter nombre de varietes supposees resistantes, bien d'autres varietes ontete cultivees avec succes sur de grandes etendues. La rouilIe de la tige a etecontr6lee a la faveur de combinaisons de genes resistants, dont Sr2 transmis

145a Hope et H-44 de J'amidonnier par McFadden en 1923. Sr26 (en provenancede Agropyron elongatum), Sr31 (de Secale cereale) et Sr36 (de Triticumtimopheevii) semblent etre les genes simples de resistance les plus efficacesdans Ie monde. La variete Thatcher (dont la resistance est due a T. durum)produite par Hayes et al. en 1934 offre une resistance satisfaisante dans laplupart des regions. La rouille des feuilles a ete combattue avec succes aumoyen d'une combinaison de Lr13 et 34. Ces genes resistants ont ete utilisespour commencer dans les varietes Frontana (Bresil, 1934) et Americano 44D(Uruguay, 1918). Leur combinaison continue a etre employee dans la culturede varietes offrant une resistance durable, telles que Chris, Era, Ciano 67,Pavon 76, etc. Les hypotheses touchant a la genetique et ala durabilite decertains types de resistance ont fait obstacle a la selection et audeveloppement de varietes resistantes.Chapitre 3Analyse de la pathogenicite de la rouille jaune (striee) du bleet son importance dans Ie contexte mondialRW. Stubbs, Institut sur la protection des plantes (lPO), Wageningen,Pays BasResumeLa rouille jaune (striee) du ble (Puccinia striiformis Westend. f.sp. tritici) faitJ'objet d'etudes au niveau international de la part de l'lPO. En conditionscontro1ees, certaines races (virulences) sont identifiees dans des plantulesd'un large ensemble de varietes differentielJes "anciennes" et "nouvelles"portant des genes de resistance dont quelques uns sont connus, mais enmajorite inconnus. La virulence liee a la resistance des plantes adultes.specifique pour chaque race, est etudiee en parcelles separees. La relationqui existe entre la distribution de la virulence des agents pathogenes et laresistance des plantes hates est evidente, mais elle continue a faire J'objetd'etudes. Sont donnes ici les resultats d'une recherche sur la rouille jaune,laquelJe affecte les tritica1es et 1es cultivars dotees de resistance derivee duseigle, ainsi que la distribution des races de rouille jaune dans diverses zonesd'Europe, d'Afrique, d'Asie et d'Amerique du sud. ' 11 est tres souhaitable queles etudes visant a connaitre les modifications de la pathogenicite dansdiverses populations de races se poursuivent activement afin d'elever Iedegre de resistance et mesurer la capacite de resistance des plantes-hotesdans des conditions environnementales differentes.

146Chapitre 4Utilisation de la resistance polygenique pour accroitre laresistance du ble Ii la rouille noire (rouille de la tige)D.R. Knott. Departement de sciences agricoles et phytoecologie, Universitede Saskatchewan. Saskatoon, CanadaResumeDepuis plusieurs annees deja la resistance multigenique a ce type de maladieest connue. Mais bien que supposee, l'existance d'une resistance nonspecifique est difficile a prouver. Une resistance partie11e et une infection adeveloppement tres lent sont souvent determinees par la presence deplusieurs genes consideres parfois comme non specifiques. Lors derecherches effectuees a Saskatoon, on a pu developpe certains bles nonresistants a la race 15B-1 au stade de plantule, mais qui possedaient unebonne resistance au champ. 11 a ete prouve que cette resistance etaitdeterminee par la presence de 3 a 5 genes recessifs dont chacun joue uncertain role. Ces genes reduisaient la periode latente ainsi que Ie nombre etla taille des pustules. La resistance due a la presence de plusieurs genesayant chacun de petits effets est sans doute relativement durable, que cetteresistance soit ou non specifique. Bien qu'il soit difficile d'avoir recours a laresistance polygenique dans les programmes d'amelioration du ble, ellepourrait etre d'une grande utilite.Chapitre 5Strategies pour l'utilisation de la resistance partielle dans lalutte contre les rouilles des cerealesJ .E. Parlevliet. Departement de phytogenHique, Universite agricole.Wageningen, Pays BasResumeToute resistance aux rouilles qui affectent les cereples est de type specifiqueau regard de l'espece, c'est-a-dire que la resistance n'est efficace qu'a l'egardd 'un seul type de rouille. Contre chaque agent pathogene de la rouille deuxtypes de resistance peuvent etre identifies: 1) un type de resistancehypersensible determinee par des genes majeurs et caracterisee par uneinfection limitee, la specificite se10n la race et une durabilite ephemere; 2) untype quantitatif de resistance (resistance partie11e) caracterisee par un iaibletaux de propagation epidemique, en depit du niveau eleve de J'infestation,par l'absence d'effets specifiques importants pour la race (bien que certainspetits efiets puissent avoir lieu) et par sa durabilite. En l'absence de genesmajeurs. il est facile d'operer une selection pour obtenir une resistancepartielle. Et meme une selection moyennement rigoureuse, matssystematiquement effectuee et visant a elimtner la senstbilite, est tresefficace, accumulant des genes propres a conferer a la variete une resistance

147partielle. Cette selection douce permet a J'ameliorateur de selectionner enmeme temps d'autres caracteristiques. S'il s'agit d 'accroftre la resistancepartielle en presence de genes majeurs qui n 'ont pas ete totalementneutralises par J'agent pathogene, J'efficacite de la selection estconsiderablement diminuee. Dans toute la mesure du possible, J'ameliorateurdevra exposer la population-hate a une seule race d'agent pathogene, capablede neutraliser un nombre maximum de genes majeurs. De cette populationhote,J'ameliorateur devra eliminer les genotypes les plus sensibles a chaqueetape de la selection de meme que les genotypes qui ne presentent qu'uneinfection peu importante. S'il s'avere difficile d'operer cette demiereseparation, J'ameJiorateur devra eliminer les genotypes les plus resistants enmeme temps que les plus sensibles, puisque les premiers sont supposesporter des genes majeurs. Dans certains cas, la population pathogene alaquelle est exposee la population-hate n 'est guere controlable et constitue enfait un melange de races. Il est alors tres difficile d'operer une selection envue d 'obtenir une resistance partielle. L 'elimination constante des lignees lesplus sensibles en meme temps que de celles les moins affectees favoriseraJ'apparition d'une resistance partielle, mais les progres dans ce sens sontplus lents qu'on ne le souhaiterait.Chapitre 6Resistance durable du ble Ii la rouille jaune et sesrepercussions en phytogenetiqueR. Johnson, Institut de phytogenetique, Cambridge, AngleterreResumeLa rouille jaune (striee) causee par le parasite oblige Puccinia striiformis,affecte les cultures de ble dans toutes les regions de c1imat frais. Plusieursgenes de resistance, specifique a chaque race, et efficaces dans les plantulesdu bJe ont ete identifies, mais il en reste encore beaucoup a identifier. Cesgenes de resistance peuvent etre dominants ou recessifs et J'expression decertains d'entre eux est grandement influencee par le milieu ambiant et leurpatrimoine genetique. La resistance acquise quand la plante a depasse lestade de plantule est souvent aussi specifique pour la race. La combinaisonde genes de resistance speciflque selon les races n 'a pas eu de succes enGrande Bretagne dans la lutte contre la rouille jaune. Cependant, laresistance acquise ulterieurement a J'etape de plantule ne fait preuved'aucune specificite pour les races, meme apres des essais prolonges et tresetendus. Cette resistance durable ne peut se distinguer des resistancesspecifiques des plantes adultes qu 'au moyen de tests prolonges. Bien qu 'ellepuisse etre soumise a un controle genetique complexe, cette resistance peutetre mise a profit dans des programmes d'amelioration tels que ceux decritsici, mais la durabilite de la resistance obtenue dans ces programmes ne peutetre garantie. Il faut donc suivre de tres pres J'evolution de toutes lesnouvelles varietes resistantes, quelle que soit la methode d'ameliorationappliquee, pour detecter J'existence de races pathogenes de pathogeniciteequivalente.

148Chapitre 7Idees actuelles sur Ie parti a tirer de la diversite pour protegerles cereales d'agents pathogimes foliaires epidemiques etvariables: Problemes et perspectives d'avenirJ .A. Browning. Departement de phytogenetique et microbiologie, Stationagricole experimentale du Texas, College Station, TexasResumeLes populations autochtones de progeniteurs de cereales et leurs parasitesnaturels, qui ont evolue parallelement, sont stabilisees par la resistancedilatoire epidemiologique produite genetiquement par la resistance generalequi implique la presence de genes d'une grande efficacite pour la resistancespecifique dans des configurations genico-spatiales distinctes. Si la resistancegenerale protege la plante, la resistance specifique protege la population enminimisant J'agressivite de J'agent pathogene par ajustement de laretroaction parfaitement harmonisee des systemes genetiques de J'h6te et deJ'agent pathogene. Resistance generale et resistance specifique devraient etremises a profit dans des systemes agricoles coherents avec leurs originesevolutives. Toutefois, J'agriculture a ete souvent dependante de genes deresistance specifique, non pour proteger la population, comme il en est dansla nature, mais pour defendre la plante d'agents pathogenes foJiairesepidemiques. Ainsi en a-t-il ete lors de J'emploi de genes de resistance dansdes varietes cultivees dans des populations homogenes sur de vastesetendues. Contrairement a cela, dans des programmes realises en Colombie,en Angleterre, aux Pays Bas, en Inde et aux Etats-Unis (Iowa et Washington),ont ete developpes des genes de resistance specifique dans des populationsdiverses et la protection de la population a ete assuree avec une resistanced'un tiers a peine. Cette petite resistance a stabilise les populations dans unecosysteme naturel varie et Ie chiffre paran donc reel. Des donneesconcordantes obtenues dans des ecosystemes naturels et agricoles indiquentqu 'un systeme global de manipulation des genes qui indue la diversite,comme i1 en est dans la nature, permettrait d'obtenir une resistanceveritablement durable. En depit des nombreux avantages qU'offre ladiversite, elle n 'a ete que relativement peu mise a profit en raison: 1) d 'unprobleme paradigmatique et 2) d'une erreur crypUque dansl'experimentation ou la valeur potentielle de la resistance a ete sous-estimee,ce qui a decourage son emploi. Les combinaisons de trois varietesconstituent un compromis adequat entre la legitime necessite d'uniformiteagronomique et les avantages de la diversite.Chapitre 8Le melange de varietes comme moyen de lutte contre les maladies et de stabilisation du rendement M.S. Wolfe, Institut de phytogenetique, Cambridge, AngleterreResumeCe chapitre comporte une description sommaire des principaux facteursconduisant a une perte d'efficacite de la resistance aux maladies et desfongicides dans J'agriculture europeenne. L 'emploi de melanges de varietes

149est expose en detail comme etant J'un des moyens dont disposent lesameliorateurs et les agriculteurs pour ameliorer la situation. Nombre dedonnees obtenues d'essais au champ revelent les avantages d'un tel systemepour lutter contre les maladies, augmenter et stabiliser les rendements; cesysteme peut etre applique parallelement a tout autre methode de luttecontre les maladies.Chapitre 9Methodes du CIMMYT pour ameliorer la resistance du ble a larouilleS. Rajaram, R.P. Singh et E . Torres, Programme bIe, CIMMYT, MexiqueResumeDes varietes de ble provenant des ressources genetiques du CIMMYT sontcultivees sur plus de 50 millions d'hectares dans Ie monde endeveloppement. L'etendue meme de ces cultures appelees a prendre encoreplus d 'extension a J'avenir a amene Ie CIMMYT a orienter sa politiqued'amelioration vers la conservation et J'accroissement de la capacite deresistance aux rouilles du materiel genetique de ble. Les essais entrepris auniveau international dans de multiples sites ont largement contribue aconfirmer cette diversite genetique. De plus, Ie CIMMYT dans sa strategied 'amelioration a recours a des analyses genetiques. Les chercheurs duCIMMYT pensent que l'utilisation de la resistance verticale (RV) specifiquepour Ie pathotype et produite par des genes majeurs comme la decritSimmonds (chap. 10) pourrait conduire a des situations precaires. A titred'alternative a la perpetuelle incorporation ou combinaison de genes deresistance, Ie CIMMYT a essaye et recommande J'amelioration pour obtenirce que Simmonds decrit (chap. 10) comme resistance horizontale (RH) nOIlspecifique pour Ie pathotype et d'origine polygenique, laqueIJe favoriserait ladurabilite de la resistance. Dans Ie contexte mondial, la resistance durable(ou stabilite) et la diversite genetique ont une enorme importance dans Ieprogramme d'amelioration du CIMMYT. L 'ideal serait de pouvoir identifierun gene ou un ensemble de genes dont on pourrait prouver qu'il confere uneresistance durable, et ensuite combiner de fac;on continue d'autres genes deresistance afin d'assurer la diversite genetique. La resistance a la rouille dela tige (complexe Sr2) derivee de la variete Hope et la resistance a la rouilJedes feuilles (complexe Lr13) derivee de la variete Frontana sont la base de ladurabilite de la resistance a ces deux maladies dans Ie materiel genetique duCIMMYT. En ce qui concerne la rouille jaune (striee), selon des informationsfournies a ce sujet, la variete Anza produite par Ie CIMMYT est dotee d 'uneresistance durable a cette maladie. Le CIMMYT procede ordinairement aJ'identification sur place de lignees offrant une resistance partieIJe (avancelente de la maladie) et a, de la sorte, largement contribue a maintenir lesrendements dans Ie monde. Le CIMMYT poursuit ses recherches sur laresistance basee sur de multiples genes majeurs a titre de strategiecompJementaire et cherche activement a obtenir Ie developpement etJ'emploi de composes de plusieurs lignees et de melanges de varietes.

150Chapitre 10Synthese: Strategie visant a accroitre la resistanceaux rouillesN.W. Simmonds, Ecole d'agriculture d'Edimbourg, Edimbourg, EcosseResumeAccroitre la resistance aux maladies equivaut a assurer la protection de labiomasse et, par suite, a elever le rendement des cultures. La resistancen 'est autre qu 'un etat de "moindre maladie" et la suppression de la maladie(immunite) constitue rarement un objectifrealiste. On distingue quatre typesde resistance: 1) la resistance verticale (RV) specifique pour le pathotype,provenant de genes majeurs; 2) la resistance horizontale (RH) non specifiquepour le pathotype et d'origine polygenique; 3) la resistance due a des genesmajeurs, non specifique pour le pathotype (RN) et 4) la resistanced'interaction ou combinee (RI). La resistance verticale (RV) est souventefficace quand iJ s'agit d'agents pathogenes immobiJes, mais peu efficacedans le cas d'agents mobiles, c'est-a-dire transmis par l'air environnant.Tou tefois, il est parfois possible de contr6ler durablement les maladies a lafaveur de genes multiples (accumuJes) de RV obtenus au moyen d'unerigoureuse manipulation genetique et pathologique. La RH est durable et engenerale en grande mesure hereditaire; nombre de maladies sont contr6leesde la sorte dans beaucoup de cultures. La RN est appreciable, mais peufrequente. Mal connue, la RI est due essentiellement a des elementsheterogenes de RV et a probablement plus de merites qu'on ne lui enattribue generalement.De toutes les rouilles qui affectent le ble, la rouille de la tige a tie combattueavec succes dans le monde entier au moyen de genes accumuJes de RV,mais iJ ne faut pas oublier qu'il peut toujours y avoir des pousseesepidemiques de la maladie en cas d'interruption du contr61e genetique etpathologique. La rouille des feuilles est probablement la plus prejudiciable deces trois maladies; la base genetique de la RV est etroite et iJ y a lieu decraindre de nouvelles epidemies. Quant ala rouille jaune, contrairement a cequ'iJ en est des deux autres, la RV semble avoir totalement echoue pour envenir a bout, et cet echec en a motive l'abandon par les ameliorateurseuropeens. L 'efficacite desormais reconnue de la RH fait qu'elle commence aetre mise a profit.

Le CIMMYT dans ses programmes d'amelioration s'estjusqu'8. presentemploye 8. accumuler RV pour lutter contre les trois rouilles, sans toutefoisque le controle genetique ait ete tres rigoureux. Sans doute cette strategiepeut-elle etre jugee dangereuse, car les petits agriculteurs des pays moinsdeveloppes ne sont pas en mesure de faire face au risque d'epidemies quetolerent plus facilement les agriculteurs de pays riches. Cette strategiepourrait etre modifiee au moyen de recherche et d'exploltation desresistances RH et RI qui, jusqu '8. present, n 'ont pas fait l'objet de recherchestres poussees en tant que moyen de lutte contre la rouille de la tige et larouille des feuilles, sans cependent que leur potentiel ne puisse etrelegltimement mis en doute. 11 ne faut pas esperer que l'attention jusqu'alorsaccordee 8. la RV solt rapidement detournee au profit de RH et de RI, bienque ce changement d'orientation offrirait 8. long terrne de grands avantages.Parallelement, le Centre devrait adjoindre 8. sa strategie d'ameliorationgenetique des varietes de ble la recherche strategique (non de base)cofncidant avec la politique du Groupe consultatif pour la recherche agricoleinternationale.151