Incompleteresistanc et ocoffe elea frus t (Hemileia vastatrix)

Aanmij nouder s ParaHeloisa ,Chantr ae Nand a

CENTRALE LANDBOUWCATALOGUS

LAHDBOUW W.h.;.«SCHOO L QQQQ QQQQ g315 Promotor :dr.ir .J .Sneep ,emeritu shoogleraa ri nd elee rva n deplantenveredelin g

Co-pratotor:dr .J.C .Zadoks ,hoogleraa ri nd efytopathologi e fOUO ^!,1S3 A.B. ESKES

INCOMPLETE RESISTANCE TO COFFEE LEAF RUST (Hemileia vastat rix)

Proefschrift terverkrijgin gva nd egraa dva n doctori nd elandbouwwetenschappen , opgeza gva nd erecto rmagnificus , dr.C.C . Costerlee, hoogleraari nd eveeteeltwetenschap , inhe topenbaa rt everdedige n cpvrijda g2 5maar t198 3 desnamiddag st evie ruu ri nd eaul a vand eLandbouwhogeschoo lt eWageninge n

\^>-W= \®1rlr5^-Ori> jj/oo810\;«?33 STELLINGEN

Selectie op polygeen verervende incomplete resistentie tegen ziekteverwekkers, met als doel het verkrijgen van duurzame resistentie, is veelal te ingewikkeld voor de praktische ver­ edeling.

2. Bij het streven naar verhoging van de landbouwproduktiviteit in ontwikkelingslanden wordt het belang van teeltkundige aspecten onderschat en het belang van de veredeling overschat.

De op zich interessante hypothese van Nelson et al. (1970) dat genen voor 'horizontale' en 'vertikale' resistentie identiek zouden zijn, is in feite gebaseerd op een onjuiste uitleg van hun eigen onderzoeksresultaten en van het begrip 'horizontale' resistentie. Nelson, E.R., D.R. MacKenzie &G.L . Scheifele, 1970. Interaction of genes for pathogenicity and virulence in Trichômetasphaeri a turcica with different number of genes for vertical resistance in Zea mays. Phytopathology 60: 1250-1254.

4. De term 'horizontale' resistentie blijft aanleiding geven tot verwarring. De term is onnodig en zou daarom niet verder ge­ bruikt dienen te worden. Dit proefschrift.

5. De door Ou (1980) vermeende instabiliteit van fysio's van Pyricularia oryzas berust grotendeels op een verkeerde inter­ pretatie van het heterogene reactietype. Ou, S.H., 1980. Pathogen variability and host resistance in rice blast disease. Ann. Rev. Phytopath. 18: 167-187.

Het 'interactiemodel' beschreven door Parlevliet en Zadoks (1977) verklaart niet de variatie in pathogeniteit waargenomen op waardplanten zonder resistentiegenen. Parlevliet, J.E. &J.C . Zadoks, 1977. The integrated concept of disease resistance; a new view including horizontal and vertical resistance in plants. Euphytica 26: 5-21.

7. Resistentie verkregen als gevolg van transgressie in nakome­ lingschappen van vatbare ouders kan berusten op additief wer­ kende 'minor' genen, zoals gesuggereerd is o.a. door Krupinsky en Sharp (1979), of op 'major' genen waarvan de expressie beïnvloed wordt door de genetische achtergrond. In het kader van het verkrijgen van duurzame resistentie is onderzoek naar beide mogelijkheden gewenst. Krupinsky, J.M. & E.L. Sharp, 1979. Reselection for improved resistance of wheat to stripe rust. Phytopathology 69: 400-404. Dit proefschrift.

BIBLIO i ,i KH IJKK LAMDBOU"»K<.n,I -SCHOO L VAGÉmBiGEH 8. De schattingen door Skovmand et al. (1978)va n de 'heritability enva nhe t aantal genenbetrokke nbi j de 'slow-rusting'eigen ­ schapvoo r zwarte roestbi j tarwe zijn onjuist enmisleidend . Skovmand,B. ,R.D .Wilcoxson ,B.L .Sheare r& R.E .Stucker ,1978 . Inheritanceo fslo wrustin gt oste mrus ti nwheat .Euphytic a27 :95-107 .

Vegetatievevermeerderin g 'invitro' , antherencultuur ensoma ­ tische hybridisatie vanprotoplaste n biedenperspectie fbi j de koffieveredeling.

10. Het oprichtenva n een internationaal onderzoekcentrum voor overblijvende tropische gewassenverdien t ernstige overweging.

11. Het 'tonisch' effectva nbespuitinge nme t fungiciden, zoals aangetoond bij dekoffieteel t inKeny a onder andere door Van derVosse n (1982), dientnade r onderzocht teworde n ind e koffieproducerende landenva nAmerika . Vande rVossen ,H.A.M. ,1982 .Consequence so fphytotoni ceffect so f fungicidet obreedin gfo rdiseas eresistance ,yiel dan dqualit yi n Coffeaarabic aL. ,J .Hort .Sei .57 : 321-329.

12. Heteffec tva n ziekte op opbrengst is redelijkbekend , terwijl er omtrent het effectva n opbrengst op ziekte veel te weinig bekend is.

13. Hetweglate nva n de dertiende verdieping inAmerikaans e flat­ gebouwenvergroo t de instabiliteit van deveertiend e verdieping.

14. Hetponere nva n een stelling isbelangrijke r dand e juistheid ervan.

15. Er iséé nuitzonderin g op de regel 'geenrege l zonderuitzonde ­ ring', namelijk deze regel zelf,hetgee n de algemenetoepas ­ baarheid vandez e regelbevestigt .

Stellingenbi j hetproefschrif t vanA.B .Eskes ,getitel d 'Incomplete resistance to coffee leaf rust (Hemileia vastatrix)' te verdedigen op2 5maar t 1983 ind eAul a vand e Landbouwhogeschool,Wageningen . CONTENTS page

Introduction 1

Chapters: 4- 12 2

1 Assessmentmethod s forresistanc et ocoffe e leaf rust (Hemileiavastatrix Berk. & Br.). 4 A.B.Eske s& M .Toma-Braghini . PI. Prot.Bull .FA O 29:56-6 6

2 Theuse of leaf disk inoculations in assessing resis­ tance to coffee leaf rust (Hemileia vastatrix). 15 A.B.Eskes . Neth.J .PI . Path.88 :127-14 1

3 Theeffect of light intensity on incomplete resis­ tance of coffee to Hemileia vastatrix. 33 A.B.Eskes . Neth.J .PI . Path. 88:191-20 2

4 Theeffect of leaf age on incomplete resistance of coffee to Hemileia vastatrix. 49 A.B. Eskes& M .Toma-Braghini . Neth.J .PI . Path.88 :215-22 6

5 Qualitative andquantitative variation in patho­ genicity of races of coffee leaf rust (Hemileia vastatrix) detected in the State of Sao Paulo, Brazil. 63 A.B.Eskes . Neth.J .PI . Path.89 :I npress .

6 Variation for incomplete resistance toHemileia vastatrix in Coffea arabica. 82 A.B.Eske s& A .Carvalho . Euphytica 32:I npress . Chapters: page

7 Characterization of incomplete resistance to Hemileia vastatrix in Coffea canephora cv. Kouillou. 98 A.B.Eskes . Euphytica 32:I npress .

8 Characterization of incomplete resistance to Hemileia vastatrix in the Icatu coffee population. Ill A.B. Eskes& W.M .d aCosta . Euphytica32 :I npress .

General discussion 123

References 133

Samenvatting 135

Resumo 138 INTRODUCTION

The twocoffe e species ofeconomi c importance are Coffea arabica L. and C. canephora Pierre.Th e first species isresponsibl e for about 80% ofth eworl d coffeeproductio n (Carvalho et al., 1969). Iti s an allotetraploid autogamous species,growin gbes t intropica lhigh ­ lands. C. canephora is adiploid , allogamous species and growsbes t intropica l lowlands. Coffee leafrus to rorang e coffee rust (Hemileia vastatrix Berk, etBr. )i sa majo r disease ofcoffee .Th e rust arrived inBrazi l in 1970 andha s recently spread tonearl y all coffeeproducin gcoun ­ tries inSout h and CentralAmerica .Al l cultivars (C. arabica) grown inthes e countries are susceptible to race II,th ecommo n raceo f H. vastatrix. Yield losses inBrazi lhav ebee n estimated tob e about 30%, ifn o controlmeasure s aretake n (Monaco, 1977). ForBrazil , thiswoul d represent about 500millio nU.S .dollar s annually inex ­ port revenues. The rustca nb e controlled by chemicals. Inth e State ofSä o Paulo,Brazil ,4 t o 5application s per year are recommended (Mariotto et al., 1979). Thecost s ofth e chemicals needed for full protection of the coffee crop inBrazi l isyearl y about20 0millio n U.S. dollars (Monaco, 1977), which represents 10t o20 %o fth e total production costs.Chemica l control isno tprofitabl e forplantation s with lowyields .Onl y asmal lpercentag e ofth eBrazilia n farmers follows therecommende d spraying schedule.Th e above observations indicate thatchemica l control isrelativel y expensive and thatit s value inth e control ofcoffe e leaf rust isrestricted . Thisempha ­ sizes the importance of genetical control ofcoffe e leafrust . Breeding fordiseas e resistance facesth e challenge of obtaining durable resistance.Mos tpathogen s show greatvariabilit y forpatho ­ genicity, especially biotrophic pathogens such asrus t fungi.Coffe e leafrus tto oha s shown considerable variation inpathogenicity . More than3 0physiologi c races havebee n identified (Rodrigues, 1975). Fivemajo r resistance genespresen t inC . arabica have not given anefficien t control ofcoffe e leafrus t (Eskes,1981 a). Many othermajo r genes arepresen t inC . canephora, whichhav e onlybee n partly identified. Presently, inbreedin g for resistance to coffee leafrus tmuc h attention isgive nt o thehybri dpopulation s 'Icatu' and 'Catimor',derive d from crossesbetwee n C. arabica and C. cane­ phora. Thesepopulation s contain genotypes with complete resistance and otherswit hvaryin g levels of incomplete resistance. Durable resistance iso f special importance forperennia l crops like coffee.Th ebreedin gproces s is slow andcultivar swhic h have lostresistanc e cannotb e rapidly replaced. Thebasi s ofdurabl e resistance hasbee n asubjec to f active scientific debate during thepas t 15year s (Vanderplank, 1968 and 1982,Simons ,1972 , Robinson, 1976,Johnso n andTaylor , 1976,Parlevlie t andZadoks , 1977, Nelson, 1978,Ellingboe , 1981). Thehistor y of resistance breeding suggests thatpolygenicall y inherited incomplete resistance isgenerall ymor e durable thanhig h levels ofresistanc e based on major genes (Caldwell, 1968,Simons , 1972). Vanderplank (1968)sug ­ gested thatdurabl e and temporary resistance arebase d ontw o dis­ tinct types ofresistanc e namedb y him 'horizontal'an d 'vertical' resistance.Horizonta l resistance was said tob e race-non-specific. Subsequent debate inth e literature has involved basic questions such as: a)doe s race-non-specific resistance occur innature , b)ca npolygenicall y inherited race-specific resistanceb e durable, c)ca ndurabl e resistanceb e identified in abreedin gprogramme , and d)ca nmajo r resistance genesb e ofvalu e inobtainin g durable resistance?

Theobjectiv e ofth epresen tresearc hwa s tocharacteriz e incom­ plete resistance ofcoffe et o H. vastatrix andt o assess itspos ­ siblevalu e inobtainin g durable resistance.Th epresen t research includes all types ofresistanc e whichd ono t fully inhibitth e reproduction ofa certai n strain ofth epathoge n (incompleteresis ­ tance). The term incomplete waspreferre d to 'partial'becaus e the latterha sbee nuse dmor e specifically to indicate incompleteresis ­ tancewit h ahig h (susceptible)reactio ntyp e (Parlevliet, 1978) . Theresearc h has resulted from ajoin tprojec t ofth e 'Instituto Agronômicod e Campinas' (I.A.C.), S.P., Brazil and the Food and Agriculture Organization ofth eUnite dNation s (F.A.O.), atth e initiative ofdrs .A . Carvalho andL.C .Monac o (I.A.C.)an ddr . R.A. Robinson (F.A.O.). The researchwa s carried out from 1976t o 1981 atth eExperimenta l Station ofth e I.A.C. atCampinas ,an d received partial support from the 'InstitutoBrasilier o deCaf e (IBC)'. In1981/8 2 theAgricultura l University ofWageninge nha s supported the research, facilitating theelaboratio n ofth e results and thepreparatio n ofth emanuscripts . The author has received co-operation from several researchers mentioned asco-author s in fiveo fth e eightpublications .Dr . AleidesCarvalho ,hea d ofth e coffeebreedin gprogramm e atth e I.A.C.,gav e scientific stimulus andprovide d datao nproductivit y ofth ecoffe egenotypes .En gAgr aWaldirM . daCost aha s assisted in the research onth e Icatupopulation .Mrs .Masak oToma-Braghin i served asa n assistingbiologis tt oth eprojec t from 1979 through 1981.

Acknowledgements forth eco-operatio n receivedb yman y others are made atth e endo feac h chapter. Inaddition , Ilik e to thank Dr. ir.I .Bo s andMr .G .Heemstr a forhel pwit h the statistical elaboration ofth eresult s andMr .J.S .d eBloc k for corrections made inth emanuscript .Als o thanks got oRie tHilhors t andWalte r deMilliano ,whos emor e general support isgreatl y appreciated.

(Forreference s seepag e 132) 1 ASSESSMENT METHODS FOR RESISTANCE TO COFFEE LEAF RUST (HEMILEIA VASTATRIX BERK. &BR.)

Summary

Assessment scales are proposed for quantitative recording of reaction type and infection frequency of coffee leaf rust. AH scales run from 0 to 9. The scale for reaction type includes heterogeneous reactions that can be frequently found in the derivatives of Coffea canephora. The scales designed for infection frequency may help the coffee breeder in making a quick assessment of partial resistance to coffee leaf rust. The advantages of a laboratory method, using leaf disc inoculations for resistance assessment, are discussed.

The use of genetic resistance to coffee leaf rust More durable resistance may be obtained constitutes the ideal control method of this by applying adequate breeding strategies for very important coffee disease. One of the basic vertical resistance, such as gene .deployment , problems for resistance breeding is that new pyramiding of resistance genes in one cultivar races of the fungus, capable of overcoming or using multiline cultivars (Nelson, 1973). resistance, may appear. This type of resistance However, most of these strategies are difficult is called vertical resistance, according to Van to implement, especially when applied to peren­ der Plank (1968), and its durability is often nial crops. related to the extent of commercial use of the An alternative may be the search for hori­ resistant cultivar. zontal resistance, which by definition should be Problems with vertical resistance also exist equally efficient against all races of the pathogen in relation to coffee leaf rust. The five tradi­ (Robinson, 1976; Van der Plank, 1968). The tionally used resistance genes of Coffea arabica concept of this type of resistance is still con­ have not provided efficient control when used troversial and it continues to be widely discussed in cultivars. Hemiteia vastatrix possesses con­ in world literature (Nelson, 1973; Parlevliet siderable variability for virulence (Rodrigues, and Zadoks, 1977; Van der Plank, 1978). Jr, Bettencourt and Rijo, 1975). This was However, a general agreement between research­ demonstrated in Brazil, where, within 10 years ers seems to exist that polygenically inherited of rust presence, 10 different races could be partial resistance confers more durable protection detected in breeding plots of the Instituto to crop pathogens than the monogenic hyper­ Agronomico of Campinas (Eskes, 1981). There­ sensitive type of resistance (Parlevliet and fore, the search for a more durable type of Zadoks, 1977; Parlevliet, 1979; Simons, 1972). resistance for the perennial coffee crop should Since 1976, FAO has been collaborating with be given high priority. the Instituto Agronomico of Campinas, Brazil, to develop a programme on durable resistance to coffee leaf rust. One important aspect of MR ESKES has been with FAO and is at present with this programme was to develop an adequate the Institute for Plant Breeding of the Agricultural methodology for quantitative measurement of University of Wageningen, the Netherlands. Ms TOMA- BRAGHINI is with the Genetics Department of the the level of resistance, including the establish­ Instituto Agronomico (IAC), Campinas, Brazil. The ment of quick assessment methods for routine authors wish to acknowledge the encouragement and useful comments given by Dr J.C. Zadoks of the resistance screening. Agricultural University in Wageningen, the Nether­ In this paper, a series of illustrated as­ lands. This article was prepared as part of the FAO/ sessment scales, from 0 to 9, are presented. IAC project on durable resistance to coffee leaf rust under execution at the Instituto Agronomico of Cam­ The 0 to 9 scale was proposed by McNeal et ai, pinas, SP, Brazil. (1971) for cereal rust research, and has been widely adopted for epidemiological and re­ found also in compatible combinations. Recent sistance observations (Zadoks and Schein, 1979). studies in Portugal confirm that the t reaction This scale has the advantage that the data are is not a specific resistance reaction. It can be computable. A uniform recording system is induced by heat treatment of compatible reac­ necessary for comparing research data from tions (C.J. Rodrigues, Jr, personal communica­ different locations or experiments. tion) and by wash-outs of uredospores of a compatible race (Rodrigues, Jr, Rijo and Me- deiros, 1981). c) Whatever the reaction type, necrosis is REACTION TYPE rarely observed in young lesions, but it may occur when the lesions are ageing. Environ­ THE TRADITIONALLY USED SCALE mental as well as genetic factors may induce FOR REACTION TYPE early necrosis. Therefore, necrosis does not seem to be an entirely suitable criterion for The Centre for Coffee Leaf Rust Research, determining reaction types of coffee leaf rust. CIFC, in Oeiras, Portugal, has adopted a re­ d) As necrosis and tumefaction are less ap­ cording system for disease reaction type that propriate to indicate resistance, the basic cri­ is similar to the traditionally used system for teria for the reaction type of individual lesions cereal rusts (d'Oliveira, 1954-57). This system should be the size of the discoloured area includes the following reaction types: /' (im­ (flecks, 0,0+ reaction types), and the intensity of mune), fl (flecks), t (tumefactions), ; (necrosis), sporulation in proportion to the extension of o (chlorosis), I (rare sporulation), 2 (small the chlorotic area (/ to 4 reaction types). pustules), 3 (medium to large pustules), 4 (large e) In derivatives of C. canephora, a mixture pustules), and X for the heterogeneous reaction. of susceptible and resistant reaction types is It has served to differentiate the more than often found on the same leaf. This occurs 30rus t races known and to choose 24 differential generally on plants with an intermediate re­ coffee genotypes (Rodrigues, Jr, Bettencourt sistance level. Results indicate that this reaction and Rijo, 1975). Often the classification is may not be a result of variation in the fungus, simplified into four types: R (resistant), MR but is rather a response of incomplete resis­ (moderately resistant), MS (moderately suscep­ tance that can be affected by light intensity and tible) and S (susceptible). leaf age (Eskes, Rivera and Da Costa, 1978; The following comments on reaction types of Eskes, Kroon and Van de Weg, 1980; Eskes, coffee leaf rust should be borne in mind: 1981). Consequently, for the description of the resistance of a certain coffee genotype toward a) Reaction type 2, with small-size pustules, a certain rust genotype, it will be necessary to is only rarely found in coffee (d'Oliveira, 1954- include heterogeneous reaction types in the 57; personal observations). It should preferably assessment scale. be considered as an intermediate reaction type ƒ) At CIFC, inoculations are generally made between / and 3. The 1 to 4 reaction types only on young tender leaves, as it is much then present a continuous scale with an in­ more difficult to produce disease on older creasing rate of sporulation in relation to the leaves (d'Oliveira, 1954-57). Work in Campinas diseased or chlorotic area. showed that some types of resistance are better b) Reaction type t (tumefaction), which is expressed in the young or adult leaves, as old frequently observed in Rubiaceae, consists of leaves are more susceptible (Eskes, Rivera and swollen spongy tissue cells (Rijo, 1972). In Da Costa, 1978; Eskes, Kroon and Van de Weg, coffee, it is often associated with incompatible 1980). For a complete appreciation of the combinations and has been called a hyper­ resistance of a plant it would be necessary to sensitive reaction type. The authors have observe the reaction type of leaves of varying observed, however, that tumefactions can be ages. A 0 TO 9 SCALE FOR REACTION TYPES 7 fl, I, o-< As in 6, but with abundant uredospore production. Sporu­ OF COFFEE LEAF RUST lation of up to 95 percent of all lesions. The adoption of a 0 to 9 scale for reaction type t, 2-4 A mixture of lesions with a is proposed, following the international efforts varying degree of sporulation, on uniform reporting of research data (McNeal sometimes associated with a et al., 1971). The proposed scale reflects the few tumefactions. aforementioned observations on reaction types. 9 4 Only lesions with abundant The scale is made to be used for resistance sporulation, without marked determinations of individual leaves or of entire chlorosis at the lesion border. plants. It includes a wide range for heteroge­ neous types of resistance, in view of the frequent occurrence of these types, especially in inter­ Reaction types 4 to 7 indicate intermediate specific hybrids. A description of the proposed types in which increasing sporulation intensity 0 to 9 scale is given in the table below. per lesion is generally associated with an in­ creasing proportion of sporulating lesions. When no computing of the data is needed, the notation can be simplified by using the symbols R, MR, MS and 5 to indicate the CIFC following reaction types: Index classification for Description of reaction types value individual lesions for leaves or entire plants R = 1, 2 and 3 MR = 4 and 5 Immunity, no visible reaction. MS = 6 and 7 S = 8 and 9 fl.-t- Minute "chlorotic spots" often associated with small tumefac­ tions. Sometimes only visible with a hand lenso r when hold­ ing the leaf against the light. ASSESSMENT OF RESISTANCE fl, t, o Larger chlorotic spots, often associated with tumefactions. No uredospore production. DETERMINATION OF THE COMPONENTS fi, t, o, o+ A mixture of various sizes of OF PARTIAL RESISTANCE chlorotic spots, including very large chlorotic areas. Fewer All components of partial resistance behave tumefactions. No uredospore production. cumulatively to decrease the reproductive ca­ pacity of the pathogen. The individual com­ 4 fi, t, o, 1 A mixture ofchloroti c spots of ponents, however, may have different effects various sizes, with some uredo­ spore formation on large and on the epidemiological development of the chlorotic lesions. Sporulation disease in the field (Parlevliet, 1979; Parlevliet of less than 25 percent of all et ai, 1980; Zadoks and Schein, 1979). Com­ lesions. Few tumefactions may occur. Early necrosis of lesions ponent analysis is therefore helpful in deter­ is sometitr.es observed. mining efficient selection criteria for practical breeding work (Parlevliet, 1976; Zadoks and ft, t, o-2 As in 4, but with more uredo­ spore formation. Sporulation Schein, 1979). It is also important in achieving of less than 50 percent of all a better understanding of the underlying re­ lesions. sistance mechanisms. In inoculation experi­ fi, t, o-3 As in 5, but with increased ments with coffee leaf rust, significant variation uredospore production. Sporu­ has been found for all components of partial lation of less than 75 percent of all lesions. resistance, i.e., infection frequency, latency period, sporulation intensity, and duration of assessment scales may also be applied to mea­ sporulation or sporulation period. sure the overall resistance level in monocyclic Infection frequency is the number of lesions experiments. obtained per unit of leaf area. A uniform In specific inoculation experiments, it may inoculation method should be used to determine be of interest to determine the individual com­ genetic differences for infection frequency. The ponents of resistance separately. No assess­ inoculation method of CIFC, by which dry ment scaleca n bedesigne d for measuring latency uredospores are applied first and the leaf is period. Instead of actually counting all sporu­ wetted afterwards (d'Oliveira, 1954-57), is not lating lesions one by one, the observations may suitable to measure differences in infection be simplified by estimating the day on which frequency, as the quantity of spores applied is 50 percent of the apparently "susceptible type" not controlled. A quantitative inoculation lesions that will probably sporulate actually do method, using uniform spore suspensions in so. Alea f or wholeplan t isuse d asa reading unit. water, is to be preferred. Sporulation intensity may be assessed by Latency period is denned here as the number the scale for reaction type, if large differences of days from inoculation until 50 percent of in sporulation are present. If smaller differences the lesions sporulate. If all the lesions do not in sporulation intensity need to be observed, sporulate then the latency period should be spore collecting and counting should be carried calculated in relation to the final number of out. sporulating lesions, and not in relation to the The duration of sporulation is determined total number of all visible infection sites. by leaf shedding or lesion necrosis. Necrosis Sporulation intensity is the number of spores can be induced by the environment (extreme produced per infection unit or per leaf area temperatures, low relative humidity), by the unit. coffee genotype or simply by ageing of the Sporulation period is the time interval from lesions. In certain C. arabica coffee types, leaf completion of the latency period until cessation retention after inoculation may be a very im­ of sporulation (necrosis of lesions or leaf portant factor in determining field resistance shedding). (Eskes et al., 1979). This parameter can be assessed in artificial inoculation experiments by applying an arbitrary scale running from 0 THE USE OF ASSESSMENT SCALES to 9 (see section on greenhouse and nursery FOR PARTIAL RESISTANCE observations). Observations on the individual components of partial resistance are time consuming and FIELD OBSERVATIONS would not be practical for routine resistance screening. Therefore, assessment scales are Epidemiological parameters. Incidence of coffee proposed that serve the breeder in carrying leaf rust has been generally estimated by the out observations more quickly. All scales are following parameters: made for determinations on disease incidence 1 = number of lesions per leaf (number of lesions per leaf area unit) or disease 2 = percentage of diseased leaves severity (proportion of diseased leaf tissue). 3 = number of lesions per diseased leaf. These scales will measure the combined action These parameters correlate well. Observations of all resistance components when applied to on 100 C. arabica cv. Catuai plants showed polycyclic experiments with successive natural disease cycles in the field or nursery. When the following correlation coefficients (/• values), applied to monocyclic inoculation experiments, all significant at P = 0.01. the scaleswil lonl y assess the infection frequency. parameter 2 parameter 3 If infection frequency is well correlated to the logio (parameter 1) .83 other components of resistance, the proposed parameter 2 .56 These data indicate that the number of le­ sions per leaf is the most reliable parameter for measuring disease incidence in the field (see Fig. 1).

Assessment Scales I and II. For measuring re­ sistance of large field populations, two types of disease-assessment scales have been designed. y = 0.30x- 2 Scale I (see Fig. 2) is based on the whole r - 0.87 tree as a reading unit, whereas for Scale II a branch has been taken as a reading unit (see Fig 3). Both scales have been designed to give comparable results. Scale I is especially suited for Brazilian growing conditions, in which the epidemic builds up from the lower 0123456789 part of the tree. Score of Scale I Each point on the 0 to 9 scale should indicate Figure 1. Correlation of the logarithm of real equal differences in resistance. The relationship disease incidence (no. of lesions per leaf) with the between the scale points and actual disease disease assessment values of Scale 1 incidence should follow the logit [In(xjl — x)\ *The lognumbe r of lesionspe r leaf (independant variable) shouldhav ebee n plotted on theadsiss a and the scoreso f curve (Van der Plank, 1963, 1968; Zadoks and scale 1o nth eordinate .Th eformul a for linear regression becomes theny =3.3 3X +5.53 .

Figure 2. Disease assessment Scale I: field incidence of coffee leaf rust Schein, 1979). This curve corresponds to the Description of Scale II. The observation unit normal exponential curve for low values of x. here is the average disease incidence per diseased As the breeder is especially interested in the branch (see Fig. 3). Readings should not lower part of the epidemiological curve, the involve very detailed observations. Each small number of dots chosen for each scale value dot represents a sporulating lesion. Bigger dots follow a 2n exponential curve. For scale value represent a coalescence of two lesions. As, in 9, only 192 dots were placed instead of 256, the field, older leaves are generally more severely as a massive leaf fall is normally induced, which infected than younger ones, more dots have limits the exponential increase of the lesion been placed on the former. number per leaf, at thisleve l of disease incidence. The number of lesions for each scale value When using the scales, only the quantity of and the percentage of leaf area they occupy are sporulating lesions should be considered. The given below. observation can be made either on one side of the coffee shrub or by two readers simul­ Percentage of taneously on both sides. In Brazil, the south­ No. of lesions eastern side of the coffee shrub normally shows leaf area diseased a higher degree of attack than the northwestern side, probably due to differences in microclimate. 0 0 0 1 1 0.06 Description of Scale I. The observation unit 2 2 0.12 of this field scale is the entire coffee shrub. Each dot on the drawing represents the number 3 4 0.24 of diseased branches seen on the shrub at first 4 8 0.47 look (see Fig. 2). More detailed observations are only made to confirm the 0 and 1 readings. 5 16 0.94 6 32 1.89 Scale value Description 7 64 3.78 8 128 7.56 0 Absence of sporulating lesions; confir­ mation necessary by pushing aside branches 9 192 11.51 of the lower canopy and more detailed ob­ servation of the branches inside the shrub. / Presence of one diseased branch at first look on any side of the shrub. Pushing Description aside of lower canopy branches may be necessary for confirmation. On detailed observation, more than one diseased branch 0 Absence of sporulating lesions. More may be found. detailed observation, by removal of lower branches, needed for confirmation. 2 to 8 These values represent a gradual increase in the number of diseased branches per 1 Average number of lesions per diseased tree, seen at first look. Value 7correspond s branch is about 1. with an average attack on susceptible coffee cultivars in Brazil at harvest time. 2-8 Increasing number of lesions per diseased branch. Value 7 is commonly found for 9 This scale value indicates the maximum of normally susceptible cultivars in Brazil at disease incidence, corresponding to the harvest time. level of disease of the very susceptible Harar coffee type in Brazil at harvest time. 9 This value indicates maximum disease in­ Normally, susceptible cultivars may drop cidence, occurring frequently on the Harar leaves before reaching this level. Therefore, coffee type in Brazil at harvest time. An 8- an S-value associated with intensive leaf value that is associated with intensive leaf shedding should be scored as a 9. shedding should be classified as a 9. Experimental results. Scales I and II were tested number of lesions per leaf. Here also the degree during June 1981 on 75 coffee plants with a of agreement between the observers was high. varying degree of disease intensity. In 64 percent of the observations, scoring was Readings were made by two trained observers. equal. In 34 percent of the observations, one The relationships between the average scores, scale value, and in 2 percent of the observations using Scale I, of the two observers were plotted two scale values difference were obtained. against the logarithm of the disease intensity Scale I is recommended for Brazilian condi­ as determined by detailed assessment (see Fig. tions. It is less time-consuming than Scale II. D- The rate of observations using the former is For the individual observers the r2 values about 200 readings per hour for an experienced were 0.86 and 0.85. The degree of agreement observer. between two observers was high, with equal scores obtained in 67 percent of the observa­ AN ASSESSMENT SCALE FOR GREENHOUSE tions, one scale value difference in 30 percent AND NURSERY OBSERVATIONS and two scale values differences in 3 percent. The readings using Scale II were very sim­ Greenhouse and nursery observations are gen­ ilar to those using Scale I, with the follow­ erally made on one-year-old coffee seedlings. ing parameters of regression obtained: y = The degree of natural infection can be assessed 0.31*— 1.69 and r2 = 0.88, with x being the by Scale II, if the drawing is held upside down. score of Scale II and y the logarithm of the Another scale has been designed to measure

Figure 3. Disease assessment Scale II: field incidence of coffee leaf rust

10 infection frequency obtained in artificial inoc­ Percentage of ulation experiments (see Fig. 4). For this No. of lesions diseased area scale, a deviation from the 2n exponential relationship had to be included, to correct for 0 0 0 the overlapping of lesions. The correction was I 1 0.6 made by applying the following formula (Gar­ 2 2 1.3 wood, 1947; Justesen and Tammes, 1960): 3 4 2.5 r = (1— slm)", 4 8 5.2 where r = disease-free area, s = surface of one lesion, m = surface ofth e leaf, and n = number 5 15 9.7 of lesions per leaf. 6 29 18.5 Furthermore, the leaf area vulnerable to 7 52 33.1 infection is considered to be only 90 percent S 84 54.1 of the total area. No lesions develop on the veins and few lesions at the tip and the border 9 118 75.6 of the leaf. Taking these assumptions into account, the following number of lesions and Scale III will be useful when screening a large percentage of diseased area have been calculated number of plants. When few observations are for each scale value: to be made, it will be more accurate to count

Figure 4. Disease assessment Scale III: infection frequency of coffee leaf rust on individual leaves

11 the actual number of lesions per unit leaf area. locations and different inoculation data; plants In addition to assessments of infection fre­ selected in the breeding programme that come quency, it may be of interest to observe leaf from distant locations can be tested by taking retention after inoculation. This parameter leaves to a central testing laboratory; resistance can be important in determining field suscepti­ of field plants may be better determined in bility in some C. arabica coffee types (Eskes et leaf discs than by field inoculations, as a random al., 1979). An arbitrary 0 to 9 scale can be sample of leaves can be used for the inocula­ used, giving a 5 value for the commercial coffee tions; host/pathogen interactions can be studied cultivar. Higher values would indicate a longer very precisely by using discs of the same leaves leaf retention, lower values a shorter leaf reten­ for each combination; and in one experiment tion. Both very high and low values may not a large number of host/pathogen combinations be desirable. High values will give an inoculum can be studied under uniform conditions. build-up and, thus, a high apparent suscepti­ bility, whereas low values may be correlated Assessment scalefor diseaseseverity on leaf discs. with low production capacity. The observation unit for this scale is a group of 10 leaf discs. Scale IV (see Fig. 5) has LABORATORY INOCULATION TESTS been designed for measuring disease severity, using the same mathematical principles as those Leaf disc inoculations. In the past, leaf disc applied for Scale III. The size of one leaf disc inoculations have been sporadically used to is taken to be 2.54 cm2, with the area of the measure vertical resistance (Narasimhaswamy, infection droplet being 0.78 cm2 (1 cm in Narayanan Nambiar and Sreenivasan, 1961; diameter). Each dot on the scale represents a CIFC, personal communication). Recently, this diseased area of about 3.5 mm2. method has been extensively tested at the Insti­ Observations can be made once or repeatedly tute Agronomico of Campinas (Eskes, 1977, during the development of the lesions. Readings 1978; Eskes, Rivera and Da Costa, 1978; may include total area of diseased tissue or Eskes and Toma-Braghini, 1979). total sporulating area. The inoculation method used consists of ap­ The assessment scale has given good cor­ plying a 0.025-ml droplet of a uniform uredo- relations with other parameters for disease pore suspension to the centre of coffee leaf severity. For the number of lesions per disc, discs 1.7 cm in diameter. The discs are incubated the correlation coefficient was r = 0.86 (n = 84) in the dark for 24 h, after which the droplets and, for the percentage of diseased discs, are allowed to dry. The discs are then main­ r = 0.84 (n = 96). tained at 100 percent humidity, a temperature The degree of agreement between two trained of 22 ± 2°C and moderate light intensity. observers was good. No difference in readings The method has given reliable results for was obtained for 68 percent of the observa­ different types of experiments, i.e., determina­ tions, one point difference for 30 percent and tion of vertical resistance and race differentia­ two points difference for 2 percent. tion, determination of partial resistance of field plants, determination of partial resistance of coffee lines grown in the nursery, determination FACTORS AFFECTING of environmental effects on the expression of THE RESISTANCE LEVEL resistance, and determination of the effect of leaf age on the expression of resistance. Partial resistance to coffee leaf rust may be The advantages of the leaf disc inoculation affected by environmental as well as physio­ method, compared with inoculations of intact logical factors. Experiments have shown a plants, are as follows: very little inoculum and marked effect on resistance of productivity, leaf space are needed; the inoculation conditions are age, and light intensity (Eskes, 1977, 1978; controlled, allowing for comparisons between Eskes, Rivera and Da Costa, 1978; Eskes and

12 Toma-Braghini, 1979; Eskes et al, 1979; Eskes, was found for several plants (Eskes, Rivera and Kroon and Van de Weg, 1980). These factors Da Costa, 1978; Eskes, 1981). Generally, old combined may produce substantial seasonal leaves become more susceptible to coffee leaf differences in resistance, and will affect the ef­ rust. Inoculations using young leaves only may ficiency of selection for partial resistance to not be sufficient to indicate the real level of coffee leaf rust. resistance of the plants. The level of natural disease in the field is the Natural disease occurring in the nursery may final test for partial resistance. However, ob­ be very valuable in determining the resistance servations should be repeated over several years, of coffee populations. However, plants should taking into account the biannual bearing habit be placed under moderate shade levels. Full of the coffee tree. Plants will show a higher sunlight may increase susceptibility to coffee degree of attack in a production year than in leaf rust, but it may also kill the young lesions a non-production year. Selected plants should by overheating of the leaf (Eskes, Kroon and have a good yield and a low disease intensity. Van de Weg, 1980). In the C. canephora cv. Kouillou and in The environmental effect on disease severity derivatives of the C. canephora x C. arabica is considerable in greenhouse inoculation ex­ cross, a marked effect of leaf age on resistance periments. The coefficient of variation of a

Figure 5. Disease assessment Scale IV: disease severity of coffee leaf rust on leaf discs

13 genetically homogeneous population of C. ara­ by choosing seedlings ofsimila r height. How­ bica plants mayb e over 100percent . This ever, even then, the efficiency of individual variation canb e reduced by predisposing the plant selection for disease severity is considered coffee seedlings to similar light conditions and low in greenhouse or nursery inoculation tests.

References

D'OUVEIRA, B. As ferrugens do cafeeiro. Centro de competitive organisms on crop yield. T. PI. 1954-57 Investigaçaodas Ferrugensdo Cafeeiro,Oeiras, Ziekten, 66:281-87 . Portugal. Revista do Café Portugués, Separata MCNEAL, F.H., KONZAK, CF., SMITH, E.P., TATE, W.S. n. 3. 61 p. 1971 & RUSSELL, T.S. A uniform system for recording ESKES, A.B. Uso de discos de folhas para avaliar a and processing cereal research data. Washington, 1977 resistencia do cafeeiro a Hemileia vastatrix: efeito DC, USDA, Agricultural Research Service. Doc­ da luminosidade e concentraçào de inóculo. ument ARS/M71/34-121, 42 p. (Use of leaf discs todetermin e resistance of coffee NARASIMHASWAMY, R.L., NARAYANAN NAMBIAR, K.K. to Hemileia vastatrix: effect ofligh t intensity and 1961 & SREENIVASAN, M.S. Report on work of inoculum concentration.) Proceedings V Bra­ testing races of leaf disease fungus on coffee zilian Coffee Research Congress, Guarapari, ES, selection at Coffee Research Station, Balehonnur. Brazil, October 1977,p . 85-86. Indian Coffee, 25(11): 333-336. ESKES, A.B. Avaliaçào do uso de discos de folhas NELSON, R.R., ed. Breeding plants for disease resistance 1978 para medir o nivel de resistencia do cafeeiro a 1973 — concepts and applications. Pennsylvania State Hemileia vastatrix. Proceedings VI Brazilian University Press. 401 p. Coffee Research Congress, Ribeiräo Preto, SP, PARLEVLIET, J.E. Screening for partial resistance in- Brazil, October 1978, p. 114-115. 1976 barley to Puccinia hordei Otth. Proceedings 4th European and Mediterranean Cereal Rusts Con­ ESKES, A.B., RIVERA, F.A. & DA COSTA, W.M. Efeito ference, Interlaken, Switzerland, 1976, p. 153-155. 1978 da idade da folha e luminosidade no campo PARLEVLIET, J.E. & ZADOKS, J.C. The integrated con- sobre a resistencia do cafeeiro a H. vastatrix 1977 cept of diseases resistance, a new view including medido em discos de folhas. (Effect of leaf age horizontal and vertical resistance in plants. and light intensity in the field on the level of Euphytica, 26: 5-21. resistance to H. vastatrix, measured in leaf discs.) Proceedings XXX Annual Meeting of the PARLEVLIET, J.E. Components ofresistanc e that reduce Brazilian Society for the Progress of Science, 1979 the rate of epidemic development. Annual Rev. Säo Paulo, Brazil, July 1978,p . 12. Phytopathology, 17: 203-222. PARLEVLIET, J.E., LINDHOUT, W.H., VAN OMMEREN, A. ESKES, A.B. & TOMA-BRAGHINI, M. USO de discos de 1980 & KUIPER, H.J. Level of partial resistance to 1979 folha do cafeeiro para estimar o nlvel dere ­ leaf rust, Püccinia hordei, andho w to select for sistencia a Hemileia vastatrix. (Useo f coffee it. Euphytica, 29: 1-8. leaf discs to estimate the level of resistance to RIJO, L. Histopathology of the hypersensitive reaction Hemileia vastatrix.) Proceedings VII Brazilian 1972 T (tumefaction) induced on Coffea spp. by Coffee Research Congress, Araxd, MG, Brazil, Hemileia vastatrix Berk. & Br. Agronomia Lu- December 1979,p . 128-129. sitana, 33: 427-431. ESKES, A.B., TOMA-BRAGHINI, M., KROON, K. & HOOG- ROBINSON, R.A. Plantpathosystems. Berlin, Heidelberg, 1979 STRATEN, J. Parametros para medir o grau de 1976 NewYork , Springer Verlag. 184 p. suscetibilidade a Hemileia vastatrix em cultivares RODRIGUES, JR, C.J., BETTENCOURT, A.J. & RIJO, L. e populaçôes de Coffea arabica. (Parameters to 1975 Races of the pathogen and resistance to coffee measure the level of susceptibility to Hemileia rust. Annual Rev. Phytopathology, 13:49-70 . vastatrix in C. arabica cultivars and popula­ RODRIGUES, JR, C.J., RIJO, L. & MEDEIROS, E. Induc- tions.) Proceedings VII Brazilian Coffee Research 1981 tion of flecks andtumefaction s on coffee leaves Congress, Araxd, MG, Brazil, December 1979, by washouts of coffee rust uredospores. Pro­ p. 70-71. ceedings International Conference on Tropical Crop ESKES, A.B., KROON, C. & VAN DE WEG, W.E. Efeito Protection, Lyon, France, July 1981,p . 58. 1980 da luminosidade sobre o nivel de resistencia a SIMONS, M.D. Polygenic resistance to plant diseases H. vastatrix. (Effect of light intensity onth e 1972 and its use in breeding resistant cultivars. J.env. resistance to H. vastatrix.) Proceedings VIII Qual., 1: 232-239. Brazilian Coffee Research Congress, Campos do VAN DER PLANK, J.E. Plant diseases: epidemics and Jordäo, SP, Brazil, November 1980,p . 83-86. 1963 control. NewYork , London, Academic Press. ESKES, A.B. Partial resistance of coffee to Hemileia 349 p. 1981 vastatrix. Proceedings International Conference VAN DER PLANK, J.E. Disease resistance in plants. on Tropical Crop Protection, Lyon, France, July 1968 New York, London, Academic Press. 206 p. 1981, p. 57. VAN DER PLANK, J.E. Genetic and molecular basis of GARWOOD, F. The variance of overlap of geometrical 1978 plant pathogenesis. Berlin, Heidelberg, New 1947 figures with reference to a bombing problem. York, Springer Verlag. 167 p. Biometrica, 34: 1-18. ZADOKS, J.C. & SCHEIN, R.D. Epidemiology and plant JUSTESEN, S.H. & TAMMES, P.M.L. Studies on yield 1979 disease management. New York, Oxford Uni­ 1960 losses. I. The self limiting effect ofinjuriou s or versity Press. 429 p.

14 2 THE USE OF LEAF DISK INOCULATIONS IN ASSESSING RESISTANCE TO COFFEE LEAF RUST (HEMILEIA VASTATRIX)

A.B.Eske s

InstitutoAgronômic o of Campinas,Campinas ,SP ,Brazi l

ABSTRACT

The suitability ofinoculation s of leafdisk s of 1.8 cm diameter in determining resistance ofcoffe e to Hemileia vastatrix, the causal agento fcoffe e leafrust ,wa s studied.Result s obtained by this methodwer e similar tothos e obtained by greenhouse testswit h res­ pectt oreactio n types ofcoffe eplant swit h complete and/ormajo r gene resistance.Th eefficac y ofth emetho d in assessing incomplete resistance was tested on 19plant s of Coffea canephora cvKouillou , whichvarie d in level ofdiseas e inth e field.Fou r series ofinocu ­ lationwer e carried out in four differentmonth s ofth eyear ,an d six components ofresistanc e were assessed. The analysis ofmultipl e correlation, applied to the average data ofth e four series,indi ­ cated that 79%o fth evariatio n indiseas e inth e field couldb e explained by theobservation s inth e leafdis k test.Fo r the indiv­ idual series thispercentag e varied from 58t o 70.Th e coefficients ofcorrelatio nbetwee n the six componentswer e significant andhigh . Thepercentag e of leafdisk swit h sporulating lesionswa s found to be themos t suitable component for assessing incomplete resistance. Thenumbe r of lesionspe r leafdis kwa s affected substantially by thehou r of the day of leafcollectio n andb y light intensity inth e field.N o effectwa s observed ofth e sizeo fth e disks (1t o 2c m in diameter)an d ofth e leafwetnes sperio d after inoculation (24an d 48 h). Resultswer emor e consistentwhe n the inoculum was applied in droplets of0.02 5m l thanwhe n the inoculum was sprayed onto the disks.N o genotypex treatment interactions were observed forth e hour oflea fcollection , forth e size ofth e leafdisk , forth e inoculationmetho d or forth e leafwetnes s period. Iti sconclude d thatth e leafdis kmethod , ifadequatel y stan­ dardized, canb e aver y useful tool inbreedin g forcoffe e leafrus t resistance and also inbasi c research onth e coffee - H. vastatrix relationship.

15 Additional keywords: laboratory tests,greenhous e tests,component s ofresistance , light intensity, leafwetnes speriod , urediospore density.

INTRODUCTION

Coffee leafrus to rorang e coffee rust, Hemileia vastatrix Berk, et Br., has recently invadedman y South andCentra l Americancountries . All Coffea arabica cultivars grown inAmeric a are susceptible to the commonrac e IIo f H. vastatrix. Yield losses inBrazi l owingt oth e disease havebee n estimated at30% ,i fn o controlmeasure s aretake n (Monaco, 1977). Breeding forresistanc e hasbee nhampere d by theap ­ pearance ofne w rustraces .Withi n threeyear s ofrus tpresenc e in Brazil theresistanc e genes S„l, S„2an d S„4los tthei reffective - tl ri rl ness (Monaco, 1977). Byth e joint efforto fth e InstitutoAgronômic o ofCampinas , Brazil, and ofth e Food andAgricultura l Organization ofth eUnite d Nations, aprogra m was started in 1976 to investigate incompletere ­ sistance tocoffe e rust,wit h the aim to obtain durable resistance. Thepresen t studyresultin g from this effortwa s carried out atth e GeneticsDepartmen t ofth e InstitutoAgronomico . Methods todetermin e incomplete resistance aretim e and space consuming and resultsma yvar y according toth e environmental condi­ tions. Inoculations of leafdisks ,performe d under controlledcondi ­ tions,ma y reduce these inconveniences. Detached plantpart s have longbee nuse d todetermin e resistance toplan tpathogen s (Yarwood, 1946). Incomplete resistance tobio - trophicpathogen s hasbee n determined with thismethod , reportedly with success (Hodgson, 1961;Umaeru s and Lihnell, 1976;Verm aan d Pétrie, 1978;Shai n and Cornelius, 1979), although susceptibility ratingsma yb e higher fordetache d plantpart s than for intact plants (Atif andWilcoxson , 1975;Zadoks, 1963) . Retardants of senescence such asbenzimidazole shav ebee n applied to keep the tissues in agoo dstate . Inoculations ofdetache d leaves or leafpart s ofcoffe e have occassionally beenuse d to study thebiolog y of H. vastatrix (Ward, 1882; Nutman andRoberts ,1963 ;Sacca s and Charpentier, 1971)an dt o assessmajo r generesistanc e (Mayne,1932 ;B.d'Oliveira ,persona l communication;Narasimhswam ye t al., 1961). Coffee leafdisk swer e incubated byplacin g them onmois t filterpape r orb y floating them onwater . Costa et al. (1978)foun d that leafdisk s could be easier kept ina goo d state thanentir e detached leaves.Conservatio n of

16 the disks,whe nplace d onmois t filterpaper ,wa s satisfactory for more than10 0 days andn o extra advantage ofth eus eo fbenzimida - zolewa s observed. Thepresen t studywa s undertaken toevaluat e the leafdis kmetho d for its efficacy indeterminin g complete aswel l as incompletere ­ sistance to H. vastatrix. Also studies are reported on factors which may influence theresult s ofth e leafdis ktest .

MATERIALS ANDMETHO D

Definitions. 'Complete resistance' isconsidere d here tob e afor m ofresistanc e that fully inhibits the reproduction ofth epathogen . 'Majorgen e resistance' results from the action ofon eo r afew , generally dominant genes,whic h individually have agrea teffec to n resistance,bu t isno tnecessaril y complete resistance. 'Incomplete resistance' isdefine d here asa formo fresistanc e that allows for atleas t some reproduction ofth epathogen .

Coffee genotypes. Plants of C. arabica cvMund oNovo , C. canephora cvKouillo u and ofth e Icatuhybri dpopulatio n wereused .Th e 'Kouillou' fieldplant swer e about4 5year s old and the Icatuplant s were planted inth e field in1970 .Icat uconsist s ofbackcrosse s of an artificial hybridbetwee n C. arabica (4x = 44 )an d C. canephora (2x =22 )wit h cultivars of C. arabica (Monaco, 1977). Mosto fth e Icatu andKouillo uplant s used showedmediu m or low levelso f attack ofcoffe e rust inth e field, incompariso n toc vMund oNovo . For the evaluation ofmajo r gene resistance thecoffe edifferen ­ tialscarryin g the S„l, S„2, S„3 and S„4gene s ofth eCoffe eRust s Research Center (CIFC)i nOeiras, Portugal ,wer eused , and also two plants of Icatu andKouillo uwit hcomplet e race-specific resistance (Eskese t al., 1981).

Rust material. Unless stated otherwise,greenhous e or field iso­ lates ofrac e II (v5)o f H. vastatrix were used.Thes e isolates were maintained on livingplant s inth egreenhouse . Sporeswer e stored in the refrigerator at 52%relativ e humidity.Germinatio n percentages ofth eurediospore s were checked before each experimentb y placing small quantities ofurediospore s ondroplet s ofdistille dwater . Incubationwa s done indarknes s at2 2 ±2 °Can d germination counts weremad e after 14hours .Onl y sporebatche swit hmor e than 10%ger ­ minationwer euse d inexperiments .

17 Standard method of leaf disk inoculation. Randomsample so ffull - grown leaveswer e obtained from field,nurser y orgreenhous eplants , theus eo fol d and damaged leavesbein g avoided. Leaves collected between 08.00 and 10.00A Mwer e temporarily stored inmois tplasti c bags.Disk swer e cutwit h acor kbore r of1. 8 cmdiamete r early in the afternoon.Thes ewer eplace d inplasti c boxes ofsiz e 48x 64x 10cm ,wit h theuppe r leaf sidedown ,o n aplasti c foam layer of2 c m thickness saturated with tapwater.Th edisk swer e kept moistb y spraying distilledwate runti l inoculation time,a tth een d ofth eafternoon . Uniform inoculumwa s obtainedb y suspending urediospores in dis­ tilledwate r and shaking or stirring during fivet o tenminutes . Thereafter the suspensionwa s kept inmotio nb y amagneti c stirrer, to avoid deposition ofth e spores.Fo r inoculation, samplescontain ­ ing1 0m lwer e taken fromth e suspensionwit h apolyethylen emicro - pipette (Beckton andDickinso n nr 5688 disposable pipette). One dropleto fabou t0.02 5m lwa splace d onto each disk.Du e toth e moist surface ofth e diskth edroplet s spread tocove r anare a of about0. 5 -0. 7 cm2. The coefficiento fvariatio n forth enumbe r of urediospores per droplet,a sdetermine d in4 8 droplets,wa s 10.5percent .Preparatio n ofth e inoculum and inoculation was always carried outunde r low lightintensities . Urediospore densities applied inth e experiments oncomplet e and/ ormajo r gene resistancevarie d from 0.8 to 1.2 mgml " ,an d inth e experiments onincomplet e resistance from 0.1 to 0.5 mgml " ,ac ­ cording to thegerminatio npercentag e ofth e sporebatch .Th e number of sporespresen t inon em g is about 1.5 x 105. After inoculation theboxe swer e closed with glass lids andin ­ cubated inth e dark, at2 2± 2 °C,durin g 20hours .The n the lids were removed inorde r to allowth e inoculum droplets toevaporate . Ina nair-conditione d room thiswa s completed within3 t o 5hours . Carewa s taken to avoid drying outo fth edisks .Afte r evaporation ofth edroplets ,th edisk swer e slightlywette d again and the closed boxeswer eplace d undermoderat e lightcondition s (fluorescent or indirectda y lighto f50 0 -100 0 lux)wit h daily a1 2hou r darkpe ­ riod.Durin g incubation thewate r level inth eplasti cboxe swa s checkedweekl y and thedisk smoistene d slightly. Itwa s avoided that the edgewoul d comeint o contactwit hliqui dwater .Temperatur e was notcontrolle d inth e experiments ofwhic h theresult s are shown in Table 2,3 , 5an d7 . Inth e other experiments temperature wasmain ­ tained at2 2± 2 °Cb y theus eo f anair-conditioner . This improved longevity ofth edisk s andwa s adopted as astandard , althoughade -

18 quateresult shav e alsobee n obtained atles scontrolle d tempera­ tures (Costae t al., 1978).

Greenhouse and field tests. Greenhouse testswer e carried outb y spraying aurediospor e suspensionwit h aSteul a Ipain tsprayer , pressurized with asmal lpump ,ont o the abaxial surface ofhealth y leaves,usuall ybelongin g toth eyounges ttw o leafpairs .Th e spore densities usedvarie d from 0.5 to 1.5 mgurediospore s perml .After ­ wards, theplant swer e incubated inth edar k at100 %relativ e humi­ dity and2 2± 2 °Cdurin g 24t o4 8hours .Fo r field experiments a manual DeVilbis s atomiserwa suse d to applyth eurediospor e suspen­ sion.Th e reservoir ofth e atomiserwa sprotecte d againstth e light by aluminium foil.Afte r inoculation the leaveswer e covered imme­ diatelyb y an innerplasti c bag, containing some liquidwater ,an d anoute rpape rbag .Fiel d inoculationswer e carried out inth e late afternoon andth ebag swer e removed early inth enex tmorning .

Inoculation dates. The dates onwhic hth e experiments showni n Table 2t o 7wer e started arerespectively : 2Ma y and 29 Septem­ ber 1977,1 2April ,1 Januar y and 10Januar y 1978 and 2Ma y1977 .

Observations. The following components ofresistanc e havebee n recorded inth e leafdis ktests : 1. the firstda y of sporulation on adis k (FDS), 2. the latencyperio d (LP),whic h isth etim e inday s from inocula­ tionunti l 50percen t of all finally sporulating disks have comet o sporulation, 3. thenumbe r ofvisibl e lesionspe r disk (NLD), 4. thepercentag e ofdisk swit h lesions (PDL), 5. thepercentag e ofdisk s thatcom et o sporulation (PDS), 6. thepercentag e ofdisk swit h sporulating lesions as apercentag e ofth etota lnumbe r ofdisk swit h lesions (PDSDL= 100-PDS/PDL), 7. the sporeproductio npe r lesion ordis k (SP)determine d 10day s after LP is completed, 8. thepercentag e ofdisk swit hnecroti c lesions (PDN), and 9. thereactio n type (RT). When the first lesions appeared, recordswer e taken every second day onth enumbe r ofdisk swit h sporulating lesions,usin g ahan d lens.NL Dwa s determined atth eonse to f sporulation,whe n the indi­ vidual lesionswer ebes t recognizable. Iflesion soverlapped , an estimation ofth enumbe r of lesionswa smad ebase d onth e size of separate lesions.Symptom s ofinfectio n counted aslesion s included

19 tumefactions (groups of swollen spongy tissue cells,Rijo , 1972), flecks (tiny chlorotic spots)an d larger chlorotic areas,wit ho r without sporulation. Todetermin e SP,urediospore s were collectedb y meanso f asmal lvacuu m pump into aknow nvolum e ofwater . Spores were countedusin g ahaemocytomete r forhig h densities or droplets of0.02 5m lplace d onmicroscop e slides forlo w densities.R Twa s assessedb ymean s of a0 t o 9scal e (Eskes andToma-Braghini , 1981). Scalevalu e 0indicate s absence ofvisibl e symptoms,value s 1t o3 indicatevariatio nwithi n resistant reaction types (no sporulating lesions),value s 4t o 7 indicate heterogeneous reaction typeswit h increasing sporulation intensity and increasingpercentag e ofsporu ­ lating lesions,an dvalue s 8 and 9indicat e susceptible reaction typeswit h large lesionswit hmoderat e (8)t o intense (9)sporula ­ tion.Greenhous e and field observations included determination ofL P (numbero fday s from inoculation till 50%o fth e final number of sporulating lesions sporulated)an d lesion density (LD= numbe r of lesionspe r leafo r leaf area unit). Assessment ofnatura l infection inth e fieldwa smad eusin g a1 t o 5scale .Valu e 1indicate s ab­ sence ofsporulatin g lesions andvalue s 2t o 5indicat e increasing rust incidence and increasing reaction type.Fiel d scoreswer emad e annually, for Icatuplants ,an dtwic e ayea r (February and August), forKouillo uplants ,fro m 1976 through1981 .

Longevity of the disks. Using the standard inoculationmetho d the leafdisk s stayed green and apparently healthy formor e than three months.Th epercentag e ofdeteriorate d leafdisk swa s less than 1 percent forth ereporte d experiments.Bes tresult swer e obtained during themonth sOctobe r throughApril ,whe n thecoffe eplant s were actively growing.Disk s takeni na perio d ofnatura l leaf fallma y becomeyello w and senescentwithi n a fewweeks . Occasionally infections other thanthos e ofcoffe e leafrus toc ­ curred,mostl y duet ocoffe e leafmine r (Perileucoptera coffeella). When dirty leaveswer e used, agrayis h fungal growth developedsome ­ times atth e sideo fth e infection droplet,interferin gwit h the results. Inon e experiment adarkbrow nwe tro tdevelope d whichde ­ stroyed the disks.Thi swa sn o longer observed when autoclaving of theplasti c foamwa s adopted routinely. Inon ecase ,a n infestation ofmite swa s observed,whic h consumed freshlyproduce durediospores . Controlwa s obtained by applying an acaricide.Th eoccurrenc e of Verticillium hemileiae was frequently observed,bu tthi shyperpara - siteo fcoffe e rust appeared so late that itdi dno t interfere with theresults .

20 Statistics. Forth elea fdis ktes tth eplasti cboxe shav ebee ncon ­ sidereda sreplications .Generall y2 0disk shav ebee nuse dpe rrep ­ lication.T oelaborat eth eresult so fth elea fdisk stest sapplie d toth eKouillo ucultiva r (Tables8 an d9 )SPS S(Statistica lPackag e forth eSocia lSciences) ,ha sbee nused .

RESULTS

Evaluation of complete and/or major gene resistance. Great similar­ ity in RT was observed when comparing the results of leaf disk tests with the results of greenhouse inoculations (Table 1). The resis­ tance of gene S„4 was not always complete in the leaf disk test. This finding is in accordance with the relative instability of the S„4 gene under nursery conditions (Eskes, 1979). The precision of the leaf disk test in determining RT was esti­ mated in an experiment including seven plants of cv Kouillou and three rust races. Seven of the 21 combinations were compatible ones. The experiment was done in three replications, each containing 15 leaf disks. The LSDQ.OS value for RT was 1.10 for the comparison between combinations. If either 2 replications or no replications would have been used, the LSD0.os values would have been respec­ tively 1.34 and 1.90. This result suggests that, as far as qualitat­ ive differences in resistance are concerned, observations based on 15 leaf disks will provide a fair estimate.

Table 1. Variation in reaction type (RT), as observed on a 0 to 9 scale, in leaf disk and greenhouse tests of six coffee genotypes with compatible and incompati­ ble races of H. vastatrix. For the leaf disk test each figure is based on 15 to 30inoculate ddisk s

Coffee Resistance Leaf disktes t Greenhousetes t genotype geneinvolve d Compa­ Incompa­ Compa­ Incompa­ tible tible tible tible

CIFC87/ 1 8-9 0-2 7-9 0-2 CIFC32/ 1 V 8-9 1 8-9 0-2 CIFC33/ 1 sh - 0-1 - 1-2 CIFC110/ 5 S 4 8-9 2-5 8-9 2-3 IcatuH3851-4-4 0 5-8 1-2 5-8 1-2 KouillouC66-1 3 1 8 0-1 9 0

Tabel 1. Variatie in reactietype (RT), waargenomen op een schaal van 0 tot 9, bij bladschijf- en kastoetsen van zes koffiegenotypen met compatibele en incompa­ tibele fysio's van H. vastatrix. Bij de bladschijftoets is ieder getal gebaseerd op 15 à 30 geïnoculeerde bladschijven.

21 Factors that may influence the results of the leaf disk test. The effect of urediospore density and that of the addition of Tween 20 to the urediospore suspension were studied for cv Mundo Novo. The LP was slightly shorter at high inoculum densities, whereas NLD, PDS and PDN were linearly related to density (Table 2). Only the in­ crease in NLD was not completely linear, because at 400 mg 1~ over­ lapping of lesions occurred. Necrosis of sporulating lesions started about ten days after the onset of sporulation. The addition of Tween 20 reduced NLD and PDS (Table 2). This was unexpected, because germination of the urediospores can be enhanced by addition of Tween (Stahmann et al., 1976).

Table 2. Latency period (LP, in days), number of lesions per disk (NLD), percent­ age of disks with sporulating lesions (PDS), and percentage of disks with necrot­ ic lesions (PDN, observed 50 days after inoculation) of leaf disks of cv Mundo Novo inoculated with four urediospore densities. For the 200 mg 1 density the addition of Tween 20 was also tested. The germination percentage of the spores was 21. Each entry is based on 360 inoculated disks.

Urediospore density Parameters of infection mg 1 LP NLD PDS PDN

50 44 0.23 21 0 100 43 0.45 31 4 200 43 0.84 44 8 400 40 1.32 57 12

200 + 0.02% Tween 20 43 0.21 17

Tabel 2. Latentieperiode (LP, in dagen), aantal lesies per bladschijf (NLD), per­ centage bladschijven met sporulatie (PDS) en percentage bladschijven met necro- tische lesies (PDN, waargenomen 50 dagen na inoculatie) van bladschijven van cv Mundo Novo geïnoculeerd met vier urediosporedichtheden. Bij 200 mg/l werd ook de toevoeging van Tween 20 getoetst. Het percentage gekiemde sporen bedroeg 21. Ieder getal is gebaseerd op 360 geïnoculeerde bladschijven. The effect of the size of the leaf disk was studied for two cof­ fee genotypes (Table 3). The analysis of variance detected a signif­ icant effect of genotype on LP and PDS, but not of disk size and of interaction. The 1 cm diameter size is considered to be too small for routine experiments, because the 1 cm disks deteriorated rapidly after the onset of sporulation.

22 Table 3. Latency period (LP, in days) and percentage of disks with sporulating lesions (PDS) of leaf disks of different sizes of two coffee genotypes. Each entry is based on 60 inoculated disks.

Coffeegenotyp e Diseasescor e D Lameter 3f Comp onent of resistance inth efiel d d isk(cm ) LP PDS (1-5scale )

CvMund oNov o 4.5 1.0 24 78 1.5 27 72 2.0 27 59

CvKouillo uC68- 4 3.1 1.0 33 16 1.5 33 16 2.0 36 26

Tabel 3. Latentieperiode (LP, in dagen) en percentage bladschijven met lesies (PDS) van bladschijven van verschillende grootte van twee koffiegenotypen. Ieder getal berust op 60 geïnoculeerde bladschijven. The effect of leaf wetness period on germination of the uredio- spores, appressorium formation and infection was studied for three coffee genotypes (Table 4). The percentage of urediospore germina­ tion (PGU) was similar for the 9 and 24 h treatment. Appressorium formation (PAF) was near to zero after 9 h and had a mean value of 62 percent after 24 hours. No significant effect of genotype and of the genotype x treatment interaction was obtained for PGU and PAF.

Table 4. Percentage of germinated urediospores (PGU), percentage of appressoria in relation to PGU (PAF), and percentage of disks with sporulation (PDS) observed for three leaf wetness periods with three coffee genotypes. Entries are based on 10 (PGU, PAF) or 90 (PDS) inoculated disks.

Coffeegenotyp e Diseasescor e Wetne ssperio d inth efiel d 9 hours 24hour s 48 hours (0-5scale ) PGU PAF PDS PGU PAF PDS PDS

CvMund oNov o 4.5 17.2 0.4 1 17.4 60 71 65 IcatuH3851-2-29 1 4.8 17.5 0.4 11 18.1 64 86 86 CvKouillo uC69-1 4 2.0 15.9 0.4 0 18.1 63 21 16

Mean 16.9 0.4 4 17.9 62 59 56

Tabel 4. Percentage gekiemde urediosporen (PGU), percentage appressoria betrokken op PGU (PAF) en percentage bladschijven met sporulatie (PDS) waargenomen voor drie bladnatperiodes bij drie koffiegenotypen. De getallen berusten op 10 (PGU, PAF) of 90 (PDS) geïnoculeerde bladschijven. Germination and appressorium formation could not be determined after 48 hours, due to abundant growth of other fungi in the inoculation droplet. The PDS values were very low for the 9 hours treatment, in accordance with the PAF values observed for this treatment. For the

23 24 and 48 hours treatment, no significant differences were observed for PDS either between treatments or between genotypes. The standard inoculation method was compared to inoculation by spraying for two coffee genotypes at two urediospore densities (Table 5). About equal amounts of inoculum were applied per disk with both methods. The inoculation method did not significantly af­ fect LP. The results for NLD were less consistent with the spraying method than with the droplet method. For the Kouillou genotype lesion formation occurred mainly at the edge of the disk with the spraying method. These edge lesions showed an abnormal development. Therefore, the droplet method was preferred as the standard inocula­ tion method.

Table 5. Latency period (LP, in days), number of lesions per disk (NLD), and number of urediospores produced per lesion (SP, x 103) of two coffee genotypes inoculated in two different ways with two urediospore densities. The germination percentage of the urediospores was 24. Each entry is based on 166 inoculated disks.

Coffeegenotyp e Diseasescor e Urediospore Inoculationm e thod inth efiel d density Sp raying Droplets (0-5scale ) (ragl" 1) LP NLD SP LP NLD SP

CvMund oNov o 4.5 33 30 1.13 3.5 29 0.73 4.8 100 30 1.10 3.8 30 1.42 5.4 CvKouillo uC68 - 15 4.3 33 45 0.14 0.4 41 0.33 0.6 100 45 0.44 0.5 44 0.79 1.1

Tabel 5. Latentieperiode (LP, in dagen), aantal lesies per bladschijf (NLD) en aantal urediosporen geproduceerd per lesie (SP, x 103) van twee koffiegenotypen geïnoculeerd op twee manieren met twee urediosporedichtheden. Het kiempercentage van de urediosporen was 24. Ieder getal berust op 166 geïnoculeerde bladschijven. The effect of the hour of the day at which leaves were collected and of the time span between collection of the leaves and cutting of the disks was studied for two genotypes (Table 6). The experiment was carried out on a day with bright sunshine. The analysis of vari­ ance detected significant genotype and treatment effects on PDS, but the interaction was not significant. A significant increase in PDS occurred when leaves were collected later on the day (treatments 1 to 3). When some hours elapsed between the collection of the leaves and the cutting of the disks, a decrease in PDS occurred (treat­ ments 4 to 6 in comparison to 1 and 2). The difference between treatments 4 and 5, which represent the variation observed when the standard inoculation method is applied, was not significant. For latency period no significant effects were observed in this experiment.

24 Table 6. The percentage of disks with sporulation (PDS)o f leaves of two coffee genotypes collected at three different hours of the day and of which disks were cut at different time spans from collection. In all treatments disks were inocu­ lated at 17.30. The disease score in the field of cv Mundo Novo and cv Kouillou C70-11, was 4.3 and 1.8 respectively. Each entry is based on 80 inoculated disks. Different letters indicate significance of differences according to the LSDQ.OS value.

Treatment PDS Collection Cutting Coffe 5 genotype Num­ of leaves of disks Cv Mundo Cv Kouillou ber (hour) (hour) Novo C70-11 Mean

1 8.30 9.00 35 16 26b 2 13.30 14.00 48 29 38c 3 16.30 17.00 76 55 66d 4 8.30 14.00 23 8 15ab 5 8.30 17.00 18 9 14ab 6 13.30 17.00 14 13 13a

Mean 35 22 29

Analysis of variance: DF MS F P Genotype (G) 1 552.0 45.3 <0.01 Treatment (T) 5 698.3 57.3 <0.01 G x T 5 21.3 1.8 n.s Error 12 12.2 Total 24

Tabel 6. Het percentage bladschijven met sporulatie (PDS) van bladeren van twee koffiegenotypen geplukt in het veld op verschillende uren van de dag en waarvan schijven werden geponst op verschillende tijdstippen na het plukken. De ziekte- score in het veld van cv Mundo Novo en cv Kouillou C70-11 was 4.3, respectievelijk 1.8. Ieder getal berust op 80 geïnoculeerde schijven. Verschillende letters dui­ den op significantie van verschillen volgens de LSDo-ös waarde. Table 7 shows the effect of light intensities to which leaves of cv Mundo Novo were exposed in the field, before collection, and to which disks were exposed in the laboratory. LP was only slightly af­ fected by light intensity in the field, but far more infection de­ veloped on disks of sun exposed leaves than of shaded leaves. This result shows the importance of using leaves grown under similar light conditions for the leaf disk test. The light intensity treat­ ments in the laboratory did not affect LP, NLD, and PDS, but it did affect the longevity of the disks. At low light intensity in the laboratory, sporulation intensity was low and the infected disks soon became necrotic, sometimes even before sporulation began. Therefore, light intensities in the laboratory above a certain mini­ mum seem to be a requisite for success with the leaf disk method.

25 Table 7. Latency period (LP, in days), percentage of disks with sporulation (PDS), and percentage of disks with necrotic lesions (PDN, 50 days after inocu­ lation) of disks of cv Mundo Novo from shaded and sun-exposed leaves in the field placed at different light intensities (LI) in the laboratory (indirect daylight). Each entry is based on 400 inoculated disks.

LIo fleave s LIi nth elaborator y inth efiel d 100lu x 500lu x LP PDS PDN LP PDS PDN

Exposure toshad e 44 11 2 43 12 0 Exposuret osunl i ght 42 59 45 41 53 7

Tabel 7. Latentieperiode (LP, in dagen), percentage schijven met sporulatie (PDS) en percentage schijven met necrotische lesies (PDN, 50 dagen na inoculatie) van bladschijven van cv Mundo Novo verkregen van schaduw- en zonnebladeren in het veld geplaatst bij verschillende lichtintensiteiten (LI) in het laboratorium (in­ direct daglicht). Ieder getal is gebaseerd op 400 geïnoculeerde schijven. The efficacy of the leaf disk method in assessing incomplete resis­ tance. In three experiments the leaf disk test was compared to greenhouse or field inoculations. In the first experiment six leaves of each of eight plants of cv Kouillou were inoculated in the field. The opposite leaves were used for the leaf disk test (10 disks per leaf). The coefficient of correlation between the two inoculation methods was significant at P £ 0.05 for the number of lesions (r = 0.66) but not for the latency period (r = 0.06). In the second experiment, leaf disk and greenhouse tests were carried out on 25 seedlings of cv Mundo Novo. Two leaves per seedling were used for the leaf disk test (14 disks per plant) and the opposite leaves were inoculated in the greenhouse. The coefficients of correlation be­ tween the two methods were low but significant at P £ 0.05, both for latency period and for the number of lesions per leaf (0.43 and 0.56, respectively). In the third experiment the leaf disk test was applied to seedlings of 37 F3 progenies of the cross H7317 (Agaro C 1164-19 x cv Catuai), grown in the nursery. For these inoculations race XV (v4v5) has been used instead of race II (v5) because of the occurrence of S„4 in the F3 progenies. Fifty six leaf disks, ob­ tained from 20 randomly choosen leaves per progeny, were inoculated in February 1978. Two months later four leaves of each of 15 plants per progeny were used for the greenhouse inoculations. The coeffi­ cients of correlation between the results of the two methods were significant at P < 0.01 for latency period and for the number of lesions (0.80 and 0.66, respectively). In a fourth experiment the efficacy of the leaf disk method in assessing the level of disease in the field of 19 plants (genotypes) of cv Kouillou, was tested. The standard inoculation method was ap-

26 plied four times in different months of 1978 and 1979 (four series). For each series, 40 randomly chosen leaves were used per plant, in 4 or 5 replications with 20 disks per replication. Inoculation and incubation of the disks was done in an air-conditioned room at a temperature of 22 ± 2 °C. The mean PDS values of cv Mundo Novo were 35, 81, 19 and 86 res­ pectively for series 1 to 4. For the 19 Kouillou genotypes PDS varied from 1 to 69, 9 to 68, 1 to 55 and 16 to 100 in the four series. The mean level of disease in the field was 4.4 for cv Mundo Novo and varied from 1.6 to 5.2 for the Kouillou genotypes. All sim­ ple and multiple coefficients of linear correlation between the com­ ponents of resistance observed in the leaf disk test and the level of disease in the field were significant at P £ 0.05 (Table 8). The proportion of the variance for disease level in the field explained by the observed components, as assessed by the R2 values, varied

Table 8. Coefficients of simple and multiple linear correlation between compo­ nents of resistance, observed in four series of leaf disk tests, and the average disease score in the field of 19 plants of cv Kouillou. For each series 80 to 100 leaf disks per genotype were inoculated. The components of resistance ob­ served were the number of days from inoculation till first sporulation (FDS), latency period in days (LP), number of lesions per disk (NLD), percentage of disks with lesions (PDL), the percentage of disks with sporulation (PDS), and the percentage of disks with sporulation relative to PDL (PDSDL = 100 • PDS/PDL). Simple r is significant at P < 0.05 when r > 0.46.

Coefficiento f Component Series numberan d inoculât iondat e Coe fficients correlation ofresis ­ 1 2 3 4 for themean s tance December March May October of series 1978 1979 1979 1979 1t o4

Simpler FDS -.66 -.67 -.64 -.58 -.76 LP -.69 -.72 -.54 -.67 -.77 NLD .48 .55 .54 .71 .78 PDL .51 .61 .57 .60 .79 PDS .66 .70 .63 .69 .82 PDSDL .63 .55 .52 .53 .61

MultipleR .84 .78 .76 .81 .89 R2 .70 .61 .58 .66 .79

Tabel 8. Coëfficiënten van enkelvoudige en multipele correlatie tussen resisten­ tiecomponenten, waargenomen in vier series van bladschijftoetsen, en de gemiddel­ de ziektescore in het veld bij 19 planten van cv Kouillou. Voor ieder genotype werden per serie 80 à 100 bladschijven geïnoculeerd. De waargenomen resistentie­ componenten zijn: het aantal dagen vanaf inoculatie tot eerste sporulatie (FDS), latentieperiode in dagen (LP), het aantal lesies per bladschijf (NLD), het per­ centage bladschijven met lesies (PDL), het percentage bladschijven met sporulatie (PDS) en het percentage bladschijven met sporulatie betrokken op PDL (PDSDL = 100 • PDS/PDL). Enkelvoudige r waarden zijn significant bij P £ 0.05 als r > 0.46.

27 from 0.58 to 0.70 for the individual series. This proportion was 0.79 when the means of the four series were used for the calcula­ tions. All coefficients of correlation were higher for the means of the 4 series than for the individual series. The coefficients of correlations between the observed components were generally high, especially between FDS and LP and between NLD and PDL (Table 9). For the pooled analysis of variance over the four series a trans­ formation of the percentage data into the arcsine of their square roots was applied. The effects of series, genotypes, and interac­ tions were significant for all components. The coefficients of vari­ ation observed for FDS, LP, NLD, PDL, PDS and PDSDL were 17, 14, 41, 20, 24 and 23%, respectively.

Table 9. Coefficients of linear correlation between the means of six components of resistance observed in four series of leaf disk inoculations applied to 19 plants of cv Kouillou. For explanation of abbreviations see Table 8. Coefficients are significant at P < 0.01 when greater than 0.57.

Components of Components of resistance resistance LP NLD PDL PDS PDSDL

FDS .95 -.77 -.73 -.92 -.86 LP -.69 -.63 -.85 -.83 NLD .95 .87 .54 PDL .86 .53 PDS .87

Tabel 9. Coëfficiënten van lineaire correlatie tussen de gemiddelden van zes resistentiecomponenten waargenomen in vier series van bladschijftoetsen bij 19 planten van cv Kouillou. Voor verklaring van de gebruikte afkortingen zie tabel 8. De coëfficiënten zijn significant bij P < 0.01 als ze groter zijn dan 0.57.

DISCUSSION

Results. The leaf disk inoculation method appeared to be adequate in assessing complete and/or major gene resistance (Table 1). Obser­ vation of as few as 15 disks can give reliable results, if suffi­ ciently high infection percentages are attained in the susceptible control. At urediospore densities of 0.8 to 1.2 mg ml" infection of the control cv Mundo Novo was generally sufficiently high (PDS val­ ues of 70 to 100%). The use of leaves, harvested from twigs exposed to high light intensities, helped to assure high infection percent­ ages (Table 7) and is, therefore, recommended. For assessment of incomplete resistance the use of 2 to 4 repli­ cations, with 20 disks each, is recommended. Urediospore densities should be calibrated according to the percentage of germinated ure-

28 diospores, inorde r to obtain that level of infection atwhic hth e discriminative capacity ofth emetho d ishighes t (50t o 70%sporu - lating disks ofth e susceptible control). Inou r experiments the calibrationwa s done according toth e results shown inTabl e2 . The average number ofgerminate d urediospores needed toproduce , in average,on e lesiono ndisk s ofc vMund oNov owa s about 100,bu t variationbetwee n experiments was considerable.Extrem e average val­ ueswer e 52 (Table 5,drople tmethod )an d 193 (Table 2).Thi s com­ pareswel l to inoculations ofintac tleave s ofc vMund oNovo ,wher e this figure ordinarily variesbetwee n 100 and 300 (personal observa­ tions). Latency periods of leafdisk s aregenerall y about fivet ote n days shorter thanthos e of intact leaves on greenhouse or field plants.Whe n temperatures inth e laboratory were notcontrolle d the latencyperio d ofc vMund oNov o inlea fdisk svarie d from 24 (Table 3)t o4 4 (Table 1), attaining highestvalue s inth ewinter . When temperatures inth e laboratorywer e controlled, at2 2± 2°C , theobserve d variation inL Pbetwee n summer andwinte rwa s smaller (about2 3t o 30 days). For seedlings ofthi s cultivar,grow n inth e open, avariatio n from 31 days (summer)t o 54day s (winter)wa sre ­ ported (Moraes et al., 1976). Certain factorswer e shownt ogreatl y affectth e results ofth e leafdis k test (Tables 2t o 7), butn o important interactions between coffee genotypes and these factorswer e observed (Table 3t o 6). The absence of interaction indicates thatth e leafdis kmetho d canb e used to assess incomplete resistance butth evariabilit y ofth ere ­ sults dictates thatth emetho db e rigourously standardized.Th eex ­ perience gained so far suggests thatthi s alsohold s forother ,mor e laborious, inoculationmethods . Table 8show s the efficacy ofth e standardized leafdis kmetho d in assessing incomplete resistance of 19c vKouillo uplants .Betwee n 58 and 70%o fth evarianc e fordiseas e level inth e field couldb e explained by the results of individual tests,whic h is considered satisfactory. Amongth ecomponent s ofresistance ,PD Sgav eth ehigh ­ estcoefficient s of correlation (r).Accordin g toth e r2 values, this component alonecoul d explain4 0t o49 %o fth evarianc e in dis­ ease in field. Lowestbu t still significant rvalue swer e obtained forPDSDL ,whic h is an indicator of reaction type.Apparently , part ofth evariatio n for incomplete resistance among theKouillo uplant s isrelate d to reactiontype . Accurate observations oncomponent s ofresistanc e arelaborious . For routine applications ofth e leafdis k test,Eske s andToma -

29 Braghini (1981)propose d tous e assessment scales running from 0t o 9 for scoring lesion incidence and reactiontype .

Applications. The leafdis kmetho d seems suitable forvariou s types ofresearc h onth ecoffe e - H. vastatrix relationship, including race identification,biolog y ofth e rust, assessment ofcomplet e and incomplete resistance, and the study ofenvironmenta l andphysiolo ­ gical factors.Th ebreede r could useth emetho d as anearl y screen­ ingmetho d ofprogenie s grown inth e nursery oro f individual field grownplants .

Limitations. The limitations ofth e leafdis kmetho d mentioned hereafter possibly apply also to inoculations of intactleaves . Environmentma y influence incomplete resistance. InTabl e 8,th e highest coefficient ofvariatio nwa s observed forNLD , indicating thatresistanc e expressed by alo w lesiondensit y isparticularl y affectedb yenvironmen t and,therefore , isdifficul tt oassess . The leafdis kmetho d givesmor e reliable results thegreate r the differences inresistanc e are. Ifsmal l differences in resistance mustb emeasured , repetition ofth e experiment intim ewil lb enec ­ essary. Thetes tcanno tb e recommended to select for incompleteresis ­ tance ofindividua l nursery plants.Mor eplant spe r genotype are needed toobtai n satisfactory high coefficients ofcorrelation . Certain formso fincomplet e resistance mayno tb e observable in leafdisks ,a s isth ecas e forresistanc e related to leafretention , reported forsom ecoffe epopulation s (Eskese t al., 1979).

ACKNOWLEDGEMENTS

The researchwa s supported by the Food andAgricultura l Organization ofth eUnite dNation s (FAO), the InstitutoAgronômic o ofCampina s (IAC), SP,Brazil ,an d theAgricultura l University ofWageningen , TheNetherlands .Th e authorwishe s to thank all staffmember s and personnel ofth eGenetic sDepartmen t ofth e IAC forth e kind coopera­ tionreceived . Special thanks go toDrs .R.A . Robinson, L. Chiarappa andN.A .va nde rGraaf fo fFA O and toDr .Alcide sCarvalh o ofth e IAC forth e stimulus received.Professor s J.C. Zadoks and J. Sneep are acknowledged forcriticall y reviewing thetext .Mrs .Masak o Toma-Braghini,Mr .Jaa pHoogstrate n andMr .Andre sRiver a Fernandez are thanked for theirparticipatio n inth e experimentalwork .

30 SAMENVATTING

Het gebruik van bladschijfinoculaties voor het bepalen van resisten­ tie tegen koffieroest (Hemileia vastatrix).

De geschiktheid van inoculaties vanbladschijve nva n 1.8 cm diameter voor debepalin g van resistentie van koffie tegen Hemileia vasta­ trix, deveroorzake r van koffieroest,wer d nagegaan.Voo r hetbe ­ palenva nhe t reactietype vankoffieplanten ,me tcomplet e en/of monogene resistentie,blee k debladschijfmethod eresultate n opt e leveren dievergelijkbaa r warenme tdi eva n kasproeven.D egeschikt ­ heidva n demethod evoo rhe tbepale nva nonvolledig e resistentie werdbeproef d bij 19planten ,behorend e tothe t rasKouillo u van Coffea canephora, dievarieerde n inveldaantasting .Vie r inoculatie serieswerde n uitgevoerd invie rverschillend e maandenva nhe tjaar , waarbij zes resistentie-componenten werdenbepaald .Multipel ecorre ­ latievoo r degemiddelde nva nd evie r series toonde aanda t79 %va n devariati e inveldaantastin g teverklare n was door de waargenomen componenten ind ebladschijftoets .Voo r de individuele seriesvari ­ eerde ditpercentag e tussen de 58e n 70.D e zes componentenvertoon ­ den onderling een sterkemat eva n correlatie.He tpercentag eblad ­ schijvenme t sporulatieblee k demees tgeschikt e componentt e zijn voor het schattenva n onvolledige resistentie. Het aantal lesiespe rbladschij fwer d duidelijkbeïnvloe d door hetuu rva n deda gwaaro p debladere nwerde n geplukt end ehoeveel ­ heid lichtwaaraa nd ebladere nware nblootgestel d inhe tveld .D e grootteva n debladschijve n (1à 2 c m indiameter )e nd ebladnat - perioden a inoculatie (24e n4 8uur )bleke nhiero p geeneffec tt e hebben.D e inoculatiemethode,waarbi j druppelsva n0,02 5ml .werde n gebruikt,blee k meer consistente resultaten te geven dand emethod e waarbijhe t inoculum op debladschijve nwer d gespoten.E rwer d geen genotype xbehandelin g interactie waargenomenvoo rhe tuu rva n de dagwaaro p debladere nwerde n geplukt,d e grootteva n debladschij ­ ven, deinoculatiemethod e end eduu rva nd ebladnatperiode . Geconcludeerd wordtda td ebladschijfmethode ,mit s toegepast in gestandaardiseerde vorm, eenzee rbruikbaa r hulpmiddel kan zijnbi j deveredelin gva nkoffi e oproestresistenti e enbi jhe t onderzoek naar derelati e tussen koffie en H. vastatrix.

REFERENCES Atif,A.H .& Wilcoxson ,R.D. ,1975 .Response so fdetache dtissue so fadul twhea t plantst oPuccini agramini stritici .Phytopatholog y65 :318-321 .

31 Costa,W.M. ,Eskes ,A.B .& Ribeiro ,I.J.A. ,1978 .Avaliaça od onive ld eresistên - ciad ocafeeir oa H .vastatrix .Braganti a37 ,not ano .4 ,pag eXXII I- XXIX . Eskes,A.B. ,1979 .Instabilidad ed aresistênci aa Hemilei avastatri xconferid a pelogen eS„ 4e mcondiça oheterozigota .In :Resumo sd o7 °Congress oBrasileir o dePesquisa sCafeeiras ,Araxâ ,M.G. ,Brazil ,Dezembe r 1979,p .75-76 . Eskes,A.B. ,Toma-Braghini ,M. ,Kroon ,K .& Hoogstraten ,J. ,1979 .Parâmetro s paramedi ro gra ud esuscetibilidad e aHemilei avastatri xe mculti-vare se populaçoesd eCoffe aarabica .In :Resumo sd o7 °Congress oBrasileir od ePes ­ quisasCafeeiras ,Araxâ ,M.G. ,Brazil ,Decembe r 1979,p .70-71 . Eskes,A.B .& Toma-Braghini ,M. ,1981 .Assessmen tmethod s forresistanc et ocof ­ feelea frus t (Hemileiavastatri xBerk ,e tBr.) . PI.Prot .Bull .FA O29 : 56-66. Eskes,A.B. ,Toma-Braghini ,M .& Carvalho ,A. ,1981 .Teste sco mraça snova sd e H.vastatri xdiferenciada se mC .canephor a cvKouillo ue na spopulaçoe sd e Icatue Catimo rIn :Resumo sd o9 °Congress oBrasileir od ePesquisa sCafeeiras , SaoLourenço ,M.G. ,Brazil ,Novembe r 1981,p .195-198 . Hodgson,W.A. ,1961 .Laborator y testingo fth epotat ofo rpartia l resistancet o Phytophtora infestans.Am .Potat oJ .38 :259-260 . Mayne,W.W. ,1932 .Physiologi c specialisationo fHemilei avastatri xB .an dBr . Nature129 : 510. Monaco,L.C. ,1977 .Consequence so fth eintroductio no fcoffe elea frus tint o Brazil.Ann .N.Y .Acad .Sei .287 :57-71 . Moraes,S.A .de ,Sugimori ,M.H. ,Ribeiro ,I.J.A. ,Ortolani ,A.A .& Pedr oJr. , M.J.,1976 .Period od eincubaça od eHemilei avastatri xBerk ,e tBr .e mtrè s regioesd oestad od eSä oPaulo .Summ aPhytopathologic a 2:32-38 . Narasimhswamy,R.L. ,Nambiar ,N .& Sreenivasan ,M.S. ,1961 .Repor to nwor ko f testing raceso flea fdiseas e funguso ncoffe eselection sa tCoffe eResearc h Station,Balehonnur .India nCoff .25 :333-336 . Nutman,F.J .& Roberts ,F.M. ,1963 .Studie so nth ebiolog yo fHemilei avastatri x Berk,e tBr .Trans .Br .mycol .Soc .46 :27-48 . Rijo,L. ,1972 .Histopatholog yo fth ehypersensitiv e reactionT (tumefaction ) inducedo nCoffe aspp .b yHemilei avastatri xBerk ,e tBr .Agronomi a lusit .33 : 427-431. Saccas,A.M .& Charpentier ,J. ,1971 .L arouill ede scaféier sdu eà Hemilei a vastatrixBerk ,e tBr .Institu tFrançai sd uCaf ée td uCacau .Bulleti nno .10 , 123pp . Shain,L .& Cornelius ,P.L. ,1979 .Quantitativ e inoculationo feaster ncotton - wood leaftissu ewit hMelampsor amedusa eunde rcontrolle d conditions.Phyto ­ pathology69 :301-304 . Stahmann,M.A. ,Musemeci ,M.R .& Moraes ,W.B.C. ,1976 .Germinatio no fcoffe e rusturediospore san dthei rinhibitio nb ycinnami c acidderivatives .Phyto ­ pathology66 :765-769 . Verma,P.R .& Pétrie ,G.A. ,1978 .A detached-lea f culturetechniqu efo rth e studyo fwhit erus tdiseas eo fBrassic a species.Can .J .PI . Sei.58 :69-73 . Umaerus,V .& Lihnell ,D. ,1976 .A laborator ymetho d formeasurin gth edegre e ofattac kb yPhytophthor a infestans.Potat oRes .19 :91-107 . Ward,M.H. ,1882 .Researche so nth elif ehistor yo fHemilei avastatrix ,th e funguso fth eCoffe eLea fDisease .J .Linn .Soc .(Bot )19 :299-335 . Yarwood,CE. ,1946 .Detache d leafculture .Bot .Rev .12 :1-56 . Zadoks,J.C. ,1963 .A cas eo frac edifferentiatio no fbrow nrus to nmatur eplant s ofwheat .Neth .J .PI .Path .69 :145-147 .

ADDRESS

Rua Salesopolis194 ,Alt od aBarra ,1310 0Campinas ,S.P. ,Brazil .

32 3 THE EFFECT OF LIGHT INTENSITY ON INCOMPLETE RESISTANCE OF COFFEE TO HEMILEIA VASTATRIX

A.B. Eskes

Institute/ Agronômico of Campinas, Campinas, SP, Brazil

ABSTRACT

Resistance ofcoffe e to race IIo f Hemileia vastatrix was tested in different environments atligh t intensities (LI)fro m 17t o 100%o f total outdoor radiation.Nin e treatments,i nwhic h three levels of LIbefor e inoculationwer e combinedwit h three levels ofL I after inoculation,wer e applied to seedlings ofth e susceptible cvMund o Novo.Highe r LIbefor e inoculation induced asignifican t increase in lesion density,wherea s the oppositewa s observed fortreatment s after inoculation.Maximu m differences in lesion densitywer ethree ­ fold.Th e interactionbetwee npre -an dpost-inoculatio n treatments was also significant.Necrosi s oflesion soccurre dunde r extremely high LI after inoculation. Genotypes ofth e Icatupopulatio n and of Coffea canephora cv Kouillou,whic hvarie d indiseas e level inth e field,wer e tested in different environments,constan t LIbein g appliedbefor e and after inoculation. The resistance ofmos tgenotype s was higher atlo wL I than athig h LI,parallelin g the results obtained forth econtro l cv MundoNovo .Wit hc vKouillou , sporulating lesiondensity , latency period and reaction typewer e significantly affectedb y LI and genotype.Th e interaction effectwa s significant for sporulating lesion density and reaction type,mainl ybecaus e themos t resistent genotypewa s not affected, or affected inopposit e direction,b yLI . Environment affected the expression ofth e resistance gene S„4. Observations on asegregatin g F2populatio n indicated dominant gene action inth egreenhous e (lowLI )an d incomplete dominant tonearl y recessive gene action inth enurser y (high LI). This incomplete dominancewa s expressedb yheterogeneou s to susceptible reaction types ofheterozygot eplant s (S„4s„4), under highLI . Some ecological andbreedin g aspects ofth eobserve d effecto fL I onresistanc e to coffee leafrus t are discussed.

Additional keywords: complete resistance,majo r generesistance , temperature,heterogeneou s reaction type,component s ofresistance .

33 INTRODUCTION

The influence ofenvironmen t isa commo n aspecto fplan t disease (Colhoun, 1973). Knowledge ofho w environment affects disease iso f great importance toth eplan tbreeder ,wh omus t choose asuitabl e environment forresistanc e screening. Iti sals o essential fora basic understanding ofth ephysiolog y and ecology of host-pathogen relationships. Incomplete resistance isgenerall ymor e affectedb y environment thancomplete ,majo r generesistance .Polygenicall y inherited incom­ plete resistance,eve nwhe npartiall y race-specific, could provide a more durableprotectio n to diseases thancomplet emonogeni cresis ­ tance,bu tthi s idea is stillbein g discussed inth e literature (e.g. Simons,1972 ;Vanderplank , 1975;Parlevlie t and Zadoks, 1977). Durable resistance to orange coffee rusto rcoffe e leafrus t (Hemileia vastatrix Berk,e tBr. )i sconsidere d tob e ofgrea tvalu e forth eperennia l coffeecrop . Effects of light intensity (LI)o nresistanc e were reported fora number ofplan tdiseases ,e.g . yellow rusto fwhea t andbarle y (Bever,1934 ;Manners ,1950 ;Stubbs , 1967), latebligh to fpotat o (Umaerus, 1970;Schuman n andThurston , 1977)an dpowder ymilde w in oats (Jones,1975) . These reports refer tomajo r generesistanc e as well ast o 'fieldresistance 'o r 'adultplan tresistance' . Little isknow nabou t the influence ofenvironmen t on resistance tocoffe e leafrust .Accordin g tod'Oliveir a (1957), who selected forcoffe e rustresistance ,L I and temperature may affect symptom expression.Monac o et al. (1973)observe d inhibition of lesion growthwhe nplant swer e temporarily exposed to atemperatur e of 40 °C.Rodrigue s (1975)foun d treatments of4 5 °Cdurin g 1.5 ht o causepredispositio n to coffee rust.H ementione d thatthes etreat ­ ments resulted ina highe r susceptibility ofcompatibl e and ofsom e normally incompatible combinations. The studies reported inth epresen tpape rwer e undertaken follow­ ingobservation s inth e greenhouse andth e field,whic h indicated an effecto f shade onresistanc e to coffee leafrust .

MATERIALS AND METHODS

Definitions. Complete resistance isconsidere d here as any form of resistance inwhic h reproduction ofth epathoge n is completely inhibited. Incomplete resistance allows for atleas t somereproduc ­ tiono fth epathogen .Majo r gene resistance is related toon eo ra

34 fewgene s eacho fwhic hha s agrea teffec to nresistance .Majo r gene resistance isno talway s complete resistance (Parlevliet, 1979).

Location. All experimentswer e carried outa tth e experimental station ofth e InstitutoAgronômic o atCampinas ,S.P. ,Brazil .

Coffee genotypes. Oneyea rol dseedling so f Coffea arabica cvMund o Novo, cvCatua i and ofth e F2populatio n of crossH 7317 (AgaroC 1164-19 xcv .Catuai )wer euse d inexperiment s 1t o4 .C vMund oNov o and cv Catuai aregeneticall y related cultivars both susceptible to race IIo f H. vastatrix. TheAgar o C1164-19plan t ishomozygou s for the S„4resistanc e gene. Inexperiment s 5an d 6cutting s from field £1 plants ofC . canephora cvKouillo u and of Icatuwer e used. Icatu consists oftetraploi d plants of advancedbreedin g generations ofa n artificial hybridbetwee nC . canephora (2x= 22 )an d C. arabica (4x= 44). CvKouillo u iscommerciall y grownunde r thenam e of 'Conilon' inth e state ofEspirit o Santo,Brazil .Al l the Kouillou and Icatugenotype s showed atleas t some infection inth e field.

Rust material. Inal lexperiment s isolates ofrac e IIo f H. vasta­ trix, whichha s thev 5virulenc e factor,wer eused .Th e isolates weremaintaine d inth egreenhous e on susceptible coffeeplant s and sporeswer e stored inth e refrigerator at52 %relativ e humidity. Racial puritywa s regularly controlled by inoculation onth e coffee differentials for S„l, S„2, S„3 and S„4. Before each experiment,th e germinationpercentag e ofth eurediospore s indistille d waterwa s determined. Sporebatche swit h less than 10%germinatio nwer e not used.Accordin g toth e germinationpercentag e ofth e sporebatch , theurediospor e densities applied inth eexperiment s varied from 0.5 to 1.5 mgml" 1 (75x 103 to 225x 103 sporespe rml) .

Standard inoculation method. Uniform dilution ofth e urediospores indistille d waterwa s obtained by shaking and stirring the suspen­ siondurin g 10min . Inoculations were carried outb y spraying the suspension onth e abaxial surface ofth ecoffe e leaveswit ha Steula Ipain t sprayerpressurize d by asmal lpump .Th ewhol e leaf surfacewa s uniformly coveredwit h small droplets,a ta rat eo f about 0.3 mlpe r 10cm 2 leaf area.Th e coefficient ofvariatio n of the sporedeposit e asdetermine d onglas s slideswa s 23%.Th e number of inoculated leavespe rplan tvarie d from 3t o 5, forseedlings , and from 8t o3 6 forcuttings .Afte r inoculation theplant swer e placed ina humi d darkroo m at2 2± 2 °C for2 4t o4 8h .

35 Inoculation dates. Inoculations ofexperiment s 1an d 2wer e done in September, and ofexperimen t 3i nNovember , 1979. Inoculations of experiments 5an d 6wer e made inJanuar y andMarch ,1981 .

Light intensity treatments. Different LIwer e appliedb yplacin g theplant s inth e greenhouse, inth enursery , or intw o different screenhouses. Inthes e environments radiationwa s respectively17 , 66, 24an d43 %o ftota l radiation, asdetermine d with anEppli nra ­ diometero n abrigh t day,whe n the total radiationwa s 2200J.cm" 2. Average radiation inCampina s during the summermonth s is about 1900 to 2500, and during thewinte rmonth s 1000t o 1500J.cm" 2. Shadewa s providedb y chalk onth ewindo w panes ofth e greenhouse,b y wooden lathing inth enursery , andb yblac k nylon 'Sombrite'ne to fdiffer ­ entmes h sizes inth e screenhouses.Difference s inai r temperature betweenth e four environments didno texcee d 2 °C. LI treatments started 1.5 monthbefor e inoculation (exps 3,5 and 6)o r after inoculation (exps.2 and 3)an dwer eprolonge d till the end ofth e observationperio d (45t o 60 days after inoculation).

Observations. To determine the components ofresistance ,observa ­ tionswer emad e every two or three days after appearance ofth e first symptoms.Lesio n density (LD= numbe r of lesionspe r leaf area unit)wa s determined atth e onseto f sporulation and sporulating lesion density (SLD= numbe r of sporulating lesionspe r leafare a unit)a tth emomen t thatn o further increase inth enumbe r ofspo ­ rulating lesions occurred. Latency period (LP)i sth etim e inday s from inoculation till 50%o f all finally sporulating lesions have come to sporulation.Th e reactiontyp e (RT)wa s scored,betwee n 50 and 60day s after inoculation,usin g a0 t o 9scal e (Eskes andToma - Braghini, 1981). Scalevalu e 0indicate s absence ofvisibl e symp­ toms,value s 1t o3 variatio nwithi n resistant reactiontypes , values 4t o 7heterogeneou s reaction typeswit h increasig sporula­ tion intensity andpercentag e of sporulating lesions,an dvalue s8 and 9indicat e highly susceptible reaction types,wit h somevaria ­ tion insporulatin g intensity.R T scoreswer e applied to individual leaves ort o awhol eplan to rgenotype .Natura l infection inth e fieldwa s assessedb ymean s ofa 1t o 5scal e adapted from Costa and Ribeiro (1975). Value 1indicate s absence of sporulating lesions, values 2t o 5increasin g incidence of sporulating lesions,associa ­ tedwit h anincreas e inreactio n type.Thes e observations weremad e annually for Icatuplant s and every 6month s forKouillo uplants .

36 Datao nfiel d infection aremean so fthes e observations overth e period 1976t o 1981.

RESULTS

In the first three experiments the effect of LI on the susceptible coffee cultivars Mundo Novo and Catuai was investigated. Experi­ ment 1 showed that different LI, applied during some hours immedi­ ately before inoculation, did not significantly affect the lesion density and latency period of seedlings of cv Mundo Novo in the greenhouse (Table 1).

Table 1. Lesion density (LD= numbe r of lesion per leaf) and latency period (LP, in days) of C. arabica cv Mundo Novo seedlings, grown in the greenhouse, treated at different light intensities immediately before inoculation.

Treatment before inoculation LD LP

Darkness for 6h 17.2 38 Shade of the greenhouse 10.4 37 Full sunlight for 1h 8.9 37 Full sunlight for 6h 9.2 38

Tabel 1. Lesiedichtheid (LD = aantal lesies per blad) en latentieperiode (LP, in dagen) van zaailingen van het C. arabica ras Mundo Novo, opgegroeid in de kas, geplaatst bij verschillende lichtintensiteiten vlak voor inoculatie.

In experiment 2, different levels of LI after inoculation were applied to cv Catuai. The number of lesions per leaf and the per­ centage of necrotic lesions were affected significantly, but not so the latency period (Table 2).

Table 2. Infection of seedlings of C. arabica cv Catuai placed, after inoculation, at different light intensities, expressed as %o f total radiation. Different let­ ters indicate significant differences according to the Scheffé test at P = 0.05.

Parameterso finfectio n Light intensity(% ) 24 43 100

Numbero flesion spe rlea f 36a 22a 13b Percentageo fnecroti clesion s 30a 28a 94b Latencyperio d 43a 48a 41a

Tabel 2. Infectie van zaailingen van het C. arabica ras Catuai geplaatst, na ino­ culatie, bij verschillende lichtintensiteiten, uitgedrukt als %va n de totale instraling. Verschillende letters duiden significantie van verschillen aan voor iedere parameter, getoetst volgens de Scheffé test bij P =0.05 .

37 At 100%L Imos t lesionsbecam enecroti c before sporulationbegan . Only those lesions thatwer e notdirectl y exposed to sunlightcam e tonorma l sporulation. Therefore,th e 100%L I treatment afterinocu ­ lationwa s replacedb y the 66%L I treatment in followingexperi ­ ments.

The effects ofL Ibefor e aswel l as after inoculation were stud­ ied together inexperimen t 3 for cvMund oNovo .L I treatments were initiated 1.5 monthbefor e inoculation.Th e results (Table 3)showe d significantbu topposit e effects ofL Ibefor e and after inoculation onlesio n density and latencyperiod .Also ,th e interaction between treatments was significant. Ingenera l latencyperio dwa sno tgreat ­ ly affected and thedifference s could partlyb e explained by the observed negative correlationwit h lesion density (seeTabl e 3). In this experiment,thre e leaves ofeac h seedlingwer e inoculated.

Table3 .Lesio ndensit y(L D= numbe ro flesion spe rleaf )an dlatenc yperio d(LP , indays )o fC .arabic ac vMund oNov oseedling splace da tthre edifferen tligh t intensities(L I= percentag eo ftota lradiation )befor ean dafte rinoculation .

LD LP

LIbefor e LI after inoculation LI after inoculation inoculation 24 43 66 Mean 24 43 66 Mean

24 16 19 15 17 36 36 39 37 43 32 21 11 21 35 36 39 37 100 49 37 20 35 31 34 38 34

Mean 33 26 15 25 34 35 39 36

Analysiso fvariance : LD LP

Source DF MS F P< DF MS F P<

LIbefor e inoculation 2 12033 38.6 0.001 2 296.2 17.6 0.001 LIafte r inoculation 2 8741 28.0 0.001 2 595.5 35.5 0.001 Interaction 4 2011 6.5 0.001 4 42.9 2.6 0.05 Residual 357 312 341 16.8 Total 365 349

Tabel 3. Lesiedichtheid (LD = aantal lesies per blad) en latentieperiode (LP, in dagen) van zaailingen van het C. arabica ras Mundo Novo geplaatst bij drie ver­ schillende lichtintensiteiten (LI) vóór en nâ inoculatie. The youngest leaf was formed during the pre-inoculation treatments. For this leaf, the greatest difference in lesion density between the treatments was about a tenfold, which was much more than for the other two leaves, where these differences were about a fivefold

38 (second leaf) and a twofold (third leaf). The necrosis of lesions seen in experiment 2 was hardly noticed in experiment 3. In experiment 4, seedlings of an F2 population, segregating for resistance gene S„4, were inoculated in the nursery and in the greenhouse. In the greenhouse the expected IS : 3R ratio was ob­ tained (Table 4). However, in the nursery no clear distinction of the plants into S and R could be made. Many plants had an interme­ diate reaction type (MR), characterized by a mixture of sporulating and non-sporulating lesions. At 80 days after inoculation more plants were classified as S than at 50 days after inoculation. At both observation dates all plants of the susceptible parent cv Catuai (s„4s--4) were of the S type and all plants of the resistant parent Agaro (S„4S„4) were of the R type. For this reason, and be­ cause of the observed segregation ratio, it was concluded that the plants heterozygous for S„4 were incompletely resistant in the nur­ sery and completely resistant in the greenhouse. The difference be­ tween the two environments was probably due to differences in LI be­ cause air temperatures were very similar in both environments. This conclusion was further confirmed by results of the inoculation of two cuttings of the Fl plant of the Agaro x cv Catuai cross, which was heterozygous for S„4, at high and low LI (Table 5). A heteroge­ neous reaction type was observed under high LI and a completely re­ sistant reaction type under low LI.

Table 4. Segregation for resistance to race II (v5) of H. vastatrix among F2 seedlings derived from C. arabica cross H731 7 (S„4S„4 x s,,4s„4) in nursery and greenhouse environments. Resistance groups are indicated by: S (susceptible, sporulating lesions only), MR(moderatel y resistant, a mixture of resistant type lesions and sporulating lesions), and R (resistant, no sporulating lesions).

Environ­ Light Inocula­ Observation Number Frequencyo f ment intensity tion period (nro f ofseed ­ seedlings(% ) (%o ftota l date daysafte r lings radiation inoculation) observed S MR R

Nursery 66 8-2-79 50 197 28.4 33.5 38.1 8-2-79 80 192 57.8 14.6 27.6 2-5-79 64 409 26.2 36.2 37.7

Greenhouse 17 25-4-79 60 250 26.6 0 73.4

Tabel4 .U i tsplitsingvoo r resistentie tegenfysi oI I (v5)va nH .vaststri xbi j F2 zaailingen van C. arabica kruising H 7317 (SH4S„4 x s„4s„4) in de kwekerij en in de kas. Resistentiegroepen zijn aangeduid met: S (vatbare planten met alleen sporulerende lesies), MR (matig resistente planten met zowel lesies van het re­ sistente type als sporulerende lesies) en R (resistente planten zonder sporule­ rende lesies).

39 I o c a -H OJ 00 -vO m H to •H hJ <4H w 4J 4H O O d 4-1 B J4 OJ XI ai O 4J O 00 - (U •H 4JHH •H h-1 4-1 XI OJ to u o cor - m INOÛ en d d H OJ Cl o o to ai oj W -ri rH d in > 4-41J 3 O0 4-> •H d Ö 'H •o c •H M •H W OJ tH U OJ ^ OJ 4-1 4-> C aj en M X 4-1 01 es to 4-1 rH <4H ^ OJ 'H O C 3 CO CO r- -* ^o oo a\ oo 4-1 rH U •H U \^t ^ CO OJ <4H O OJ OJ PH O -M u OJ d EH -H vO OJ 0) <4H ö <« J3 O -ri 4-1 •H T3 •H 4-> 3 r-1 ^û 4-1 kl ^ U Ol 3 ^O n ^fr /—.-O CM a.i-1 ai OJ W -H c CJ OJ 4-1 a 4-1 o •H OJ o W > OO •H 4) a aj JJ M J3 o r-H 0) 4-1 XI OJ r-H TD •o Ö OJ (aD d OJ <4H -H U i—» •r-l O m 1—1 u-> O -J CM O >d- i—| M X 1 (0 (U M OJ CO CM CO CM co OJ 3 •H ^5 to w oo il OJ H fO -H •H 4-1 Q OJ HH N « M w l-H" •H XI d T3 U o T3 •OHJ S» fO äO T3 Ö 4-1 OO« > •H OJ to • •H Ö F—1 o 0J rH 4-1 "O V) 'H • H XI :cu o u B 3 O U •H 4-1 OJ OJ >»/—i4 e * O X! rH OJ -a a) O C U A3 OJ <-t > -H 3 •H > •O 3 c 4-1 2 TJ U o XI to >U /—-d- \ (U a d o •H 4-1 'H r^ r- CM co 35 •H ai to c W 'H T5 a co CM r- >U co CO OJ > :H OJ S a u ^o 3 m * co r- rQ U fO a OJ «H u £0 & OJ ^ EH 00 O u i—i « s « X o K E-t 4-1 &x

40 u 0) OJ tn Oi 3 o >>xl u c c 0)

00 Ol on \o oo r-- o> oo i i i i i i o o co ai a\^ 1-1 Tl d 0) es 1-4 4H 0) m C/l m o o C) Z co m oo m at r-H *J •H tn 3 o) 4-> Hl •ri m 4H O Hl r—1 O Pi m V) =i-l) H > o P5 rH d O O O 10 ai o o o u d O O O i-t <-—H % SI « OS m oo .O c n ni HHOl aw a+J~ 3 e O O O"! V—' 01 Z T1 o o .-« -H u c W i-l ai O O O 00 Ol U •I-I •H •H «HJ fflai H i V) «01 Hi 01 H H .M ^ OS H r-4 LD T3 01 o o co 01 i—1 3 >> o o o •a 01 .M d T3 o o o o> 0) •o 01 u •H •O ai S r-H ai d 3 00 •H u T1 Hl o i—1 O •> d (J 0) 0) Pi 01 m •H i—1 W /—d» 4-1 (0 CO \0 00 CN] r-t CO ^ w i -o d d •H •H w Q •H v 0) A3 —' J3 M i-l Pi- -C hJ 01 O c/l 3 u 13 Pi c •H VI >m> d o d a 01 d •a JS r-l D •H 0) •o •H d ft i-t U +J -H o ^3 c n) a u u *o •H d o Hi 01 d r-^ r-i r-i co * flï -H 01 •H / s O r-H 1-1*4 3 C/1 u i-> HI m rH 01 E •T-t rH d ai d o t-i *—*^- * auuoouou >u S w O H-I o

41 Inexperimen t 5,two-year-ol d cuttings of fieldplant s ofc v Kouillou and of Icatuwer e inoculated under three lightregimes .Th e cuttingswer e adapted to theenvironment s 1.5mont hprio r toinocu ­ lation.Th eresult s showed anincreas e in susceptibility within ­ creasedL I (Table 5). The relative increasewa s greater forth emor e resistantplant s than forth emor e susceptible ones.N o statistical analysiswa sperforme d because only one ortw o cuttingswer e used per treatment.Th edifference s indiseas e score ofth eplant s inth e field appeared tob ewel l related to thenumbe r of lesionspe r leaf in allthre e environments,bu tthi s relationwa s less clear for reactiontype . Experiment 6consiste d of inoculations ofcutting s from sixgeno ­ types ofc vKouillo u ando f seedlings ofc vMund oNov o inth egreen ­ house andnursery .Th e cuttingswer eplace d inthes e environments 6 weeksprio r to inoculation. The ageo fth e inoculated leaves varied from a fewweek s to afe wmonths .Th e results (Table 6) indicated asignifican t environment and genotype effecto n SLD,L P andRT .Mos tgenotype s were considerably more susceptible inth e nursery environment. Interactionbetwee n environment andgenotype , signicifant forSL D andRT ,wa smainl y due to thedifferen t response ofth e C67-7 genotype of cvKouillou .Th e reaction ofthi s genotype was characterizedb y flecks and large chlorotic areas,whic hmostl y becamenecroti c before sporulation started. The residual variation was large for SLD and relatively low forL P andRT ,a s indicated by the coefficients ofvariatio n (CV)o f these components (Table6) . However,th e CVvalue s forL P andR Tma yhav ebee n underestimated, becausemor e cuttingswer e used of the genotypes with relatively stable LP and RT (C68-15 and cvMund oNovo )tha no fth e genotypes with amor evariabl e reaction. Inthi s experiment thenumbe r of lesions ofc vMund oNov owa s abnormally low inrelatio n to the 'Kouillou' genotypes.A possibl e explanation forthi s unexpected result istha tth e seedlings ofc vMund oNov ower e transplanted 1.5 monthprio r to inoculation and thecutting s of cvKouillo u were not.

DISCUSSION

C. arabica cultivars. The susceptibility to coffee leafrus to fth e Brazilian cultivars Mundo Novo and Catuai appeared tob esignifi ­ cantly affectedb y LI. Treatments ofhig hL I during 6week sbefor e inoculation increased lesiondensit y ofc vMund oNov o threefold (Table 3)i ncompariso n to lowL I treatments.Thi s effectwa s not

42 foundwhe n high LIwa s applied only a fewhour s immediatelybefore inoculation (Table 1).Therefor e physiological differences which are inducedb y amor e long term adaptation of theplant s toth e environ­ mentsma yhav ebee n involved. High LI after inoculation had effects opposite tothos e ofhig h LIbefor e inoculation (Tables 2 and 3).Differen t treatments with an equal LIbefor e and after inoculationwer e applied to cvMund o Novo inexperiment s 3, 5an d 6. Inexperimen t 3n o significantdifferen ­ ceswer e observedbetwee n thesetreatment s (Table 3),bu t inexperi ­ ments 5an d 6mor e lesions developed inenvironment s with ahighe r LI thanwit h alowe r LI. This indicates thatth epre-inoculatio nef ­ fectwa smor e important thanth epost-inoculatio n effect inexperi ­ ments 5an d 6. Thebalanc e resulting from the twoopposin g effects isexpecte d tovar y according toth e intensity ofth e treatments applied.

The S„4 resistance gene. This resistance gene hasbee n reported to be adominan t gene (Rodrigues et al., 1975). Thepresen t results indicate thatth e action ofthi s genema y depend onenvironment . In environments with lowL I itwa s dominant indeed,bu t athig hL I incomplete dominancewa s observed (Tables 4 and 5).Th e incomplete dominancewa s expressed by aheterogeneou s reaction typewhich ,a s timeevolved ,becam e rather asusceptibl e reactiontype .

C. canephora cv Kouillou and Icatu. Theresistanc eo fmos tgeno ­ typeswa sbette r expressed inenvironment s with low LI. Genotypes with intermediate levels of resistance appeared tob emos t affected by environment and also displayed the greatest residual variation (Tables 5an d 6).Onl y one genotype ofc vKouillou ,wit h the highest level of incomplete resistance,wa s not affectedb y environment, and caused asignifican t environmentx genotype interaction forSL D and RT.

Relation between LI and temperature. Theobserve deffec to fL Ima y be confounded with aneffec to ftemperature .Althoug h the difference inaverag e air temperature between the environmentswa s not greater than2 °C,th e leaf temperatures inenvironment s with ahig hL Imus t havebee nhighe r than inenvironment s with low LI. Gomez and Jaramillo (1974)observe d thatth e average temperature of sunexpose d leaves ofcoffe ewa s 3t o 5 °Chighe r thantha t ofshade d leaves.I n one case,a differenc e of 10 °Cwa s observed. As reported byMonac o et al. (1973), lesion growtho f H. vastatrix was completely inhibited

43 when coffeeplant swer e treated at4 0 °Cdurin g4 h o n fivesucces ­ sive days. Leaftemperature s nearly ashig h as4 0 °Cma y occur in Campinas,Brazil ,durin g the summermonths ,whe n the air temperature frequently reaches 30 to 35 QC. Therefore,th e depressive effecto f highpost-inoculatio n LI on lesion development (Table2 an d 3)migh tb e explained by the effecto ftemperature . Hightemperatur e prior to inoculationma y induce ahighe r suscep­ tibility ofplant st o several rust fungi (Heath,1979 ;Vanderplank , 1978). Rodrigues (1975)reporte d an increased susceptibility of compatible and incompatible coffee/ff. vastatrix combinations when leaveswer e exposed to4 5 °Cdurin g 1.5 hprio r to inoculation. Therefore,th e increased susceptibility induced by high LIbefor e inoculation (Table 3)may , atleas t inpart ,b e due to atemperatur e effect.

Types of resistance involved. The coffee genotypes used inthi s studyha d either incomplete resistance,th e geneticbas e ofwhic h is unknown (Tables 1,2 ,3 , 5, and 6), ormajo r gene resistanceprovid ­ edb y S_4,whic h israce-specifi c (Table 4). The incomplete résis­ ta tance oftw oplant s of Icatushow n inTabl e 5 (H3851-2-43 7 and H 3851-2-513)i sknow nt ob e race-specific.Thi s resistance was lost by the appearance of ane w andmor e compatible race inth e Icatu population in1978 .Th e other Icatu andKouillo u genotypeswer e not affectedb y this race and they didno tchang e their level ofdiseas e inth e field during theyear s 1976 to 1981 (personal observations). Therefore, itappear s thatL Ima y affectvariou s forms ofresis ­ tance,whic hprobabl y depend ondifferen tgeneti cmechanisms .

Ecological considerations. Field resistance ofpotatoe s to late blight (Umaerus,1970 ;Schuman n and Thurston, 1977)an dpartia l adultplan tresistanc e ofoat s topowder ymilde w (Jones,1975 )ar e reducedwhe n lower LI are applied.Resistanc e ofwhea tt o yellow rust alsoma y decrease atlowe rLI ,althoug h Stubbs (1967)classi ­ fied somedifferential s asphotolabil e and others asphotostable . With coffee leafrus tth e tendency appears tob e opposite:a nin ­ crease inresistanc ewa s observed atlowe r LI. Itcoul db e argued thatthi s discrepancy iscause db y the different conditions under whichthes e species normally grow inth ewild .Bot h C. arabica and C. canephora are forest species,wherea spotatoe s and, certainly, cerealsnormall y grow inth e open.A s observed by Sylvain (1955) 'coffeerus ti srar eunde rnatura l forestcondition s inEthiopia , but incase swher e theuppe r storyo fth e forestha sbee n removed,

44 leafrus t and Cercospora coffeicola may spread quite rapidly'.Thi s is in agreementwit h thepresen t results showing thatcoffe eplant s athig hL I tend tobecom emor e susceptible. Itseem splausibl e that resistance ofplant s to diseaseswil lb e optimal under conditions representing thenatura l environment ofth especies . However, itshoul dno tb e concluded that shading of susceptible C. arabica cultivars would help to control coffee rust.Th e results ofTable s 2 and 3sho w thatpre -an dpost-inoculatio n LI treatments had opposite effects,thu s suggesting that intermediate levels ofL I mayb e optimal forrus tdevelopment .Thes e intermediate levelsma y predominate inth e leafcanop y ofcoffe e trees,whe nmediu m levels of shade are applied. Shading ofcoffe eplantation s will also change other environmental factors,amon gwhic h relativehumidity , and thesema y favour disease development instea d ofcontrollin git .

Practical implications for the coffee breeder. Theobserve dincom ­ plete dominance ofth e SH4resistanc e geneunde r nursery conditions mayhav epractica l implications for resistance screening and race identification. Coffee differentials heterozygous for S„4ar eavail ­ able fordetectio no frus t raceswit h complexvirulenc e (Rodrigues, et al., 1975). Care shouldb e taken inusin g thesedifferentials , because their resistance willno tb e complete inal lenvironments . The coffeebreede r interested inscreenin g forincomplet eresis ­ tancema yprefe r towor kunde r relatively low LI,whic hmak e this typeo fresistanc e more easily detectable. Similar LI shouldb e applied to allplant s inorde r to decrease theresidua l error ofex ­ periments.

Other environmental factors. Inexperiment s notreporte d here a seasonal influence onresistanc e to coffee leafrus twa s observed. For example,inoculation s ofcertai n Icatuplant swer emuc hmor e successful insumme r than inwinter . Inaddition , some races ofcof ­ fee leaf rustwer emor e difficult to growdurin gwinte r than insum ­ mer. Seasonal effectsma yb e explainedb y separateo rcombine def ­ fects ofLI ,temperature , relative humidity, and leaf age.Thes e factors deserve further investigation.

ACKNOWLEDGEMENTS

The researchwa s supportedb y theFoo d andAgricultura l Organization ofth eUnite dNation s (FAO), the InstitutoAgronômic o ofCampina s (IAC), S.P., Brazil,an dth eAgricultura l University,Wageningen ,

45 TheNetherlands .Th e author isgratefu l to all staffmember s and personnel ofth eGenetic s Department ofth e IAC forth ekin dcoop ­ eration received. Special thanks aredu et oDrs .Alcide s Carvalho (IAC), R.A.Robinson , L. Chiarappa andN.A .va n derGraaf f (FAO)fo r the stimulus received. The activeparticipatio n inth e experimental workb yMrs .Masak oToma-Braghini ,Mr .C .Kroo n and Mr. W.E.va n deWe g is acknowledged. The authorwishe s to thank Professors J.C. Zadoks and J. Sneep forcriticall y reviewing the text.

SAMENVATTING

Effect van lichtintensiteit op incomplete resistentie van koffie te­ gen Hemileia vastatrix.

Resistentie van koffie tegen fysio IIva n Hemileia vastatrix werd getoetst inmilieu sbi j lichtintensiteiten (LI),di evarieerde nva n 17to t 100%va n de totale instraling.Nege nbehandelingen , bestaande uitd e combinaties van drieniveau sva n LIvóó r inoculatie endri e nâ inoculatie,werde n toegepast op zaailingen vanhe tvatbar e Coffea arabica rasMund oNovo .Toenemend e LIvóó r inoculatieveroor ­ zaakte een significante toename in lesiedichtheid, terwijl hette ­ genovergestelde werdwaargenome n bij debehandelinge n na inoculatie. Maximaleverschille n inlesiedichthei d waren drievoudig.D e interac­ tie tussenbehandelinge nvóó r enn â inoculatie was ook significant. Bij extreem hoge LIn â inoculatie trad necroseva n de lesiesop . Genotypenva n de Icatupopulati e enva nhe tC . canephora ras Kouillou,me tverschillend e ziektescores inhe tveld ,werde nbe ­ proefd inverschillend e milieus,waarbi j eenconstant e LIvoo r enn a inoculatiewer d toegepast.D e resistentie van demeest e genotypen kwambete r totuitin gbi j lageL I danbi j hoge LI,wa too kwaargeno ­ menwer dvoo r hetcontrol e rasMund o Novo.Bi jhe tra sKouillo uwer ­ dend e dichtheid van sporulerende lesies,d e latentieperiode enhe t reactietype significantbeïnvloe d door LI engenotype .D e interactie tussen LI engenotyp ewa s ook significantvoo r dichtheid vansporu ­ lerende lesies envoo r reactietype,voornamelij k doordathe tmees t resistente genotypeniet ,o f ind e omgekeerde richting, beïnvloed werd door LI. De expressie vanhe tresistentiege n S„4blee k ook afhankelijk van hetmilieu .Waarneminge n aanee nuitsplitsend e F2-populatie duidden op eendominant e genwerking ind e kas (lageLI )e n een incompleet dominante,o fbijn a recessieve,genwerkin g ind e kwekerij (hogeLI) .

46 Deze incomplete dominantie uitte zichd.m.v . heterogene totvatbar e

reactietypes vanheterozygot eplante n (S„4s.T4)onde rhog eLI . ri ri Enkele ecologische en veredelingstechnische aspecten van de waar­ genomen invloed van LI worden besproken.

REFERENCES

Bever,W.M. ,1934 .Effec to fligh to nth edevelopmen to fth euredia lstag eo f Puccinia glumarum.Phytopatholog y 24:507-516 . Colhoun,J. , 1973.Effect so fenvironmenta lfactor so nplan tdisease .Ann .Rev . Phytopathol.11 :343-364 . Costa,W.M .& Ribeiro ,I.J.A. ,1975 .Resistênci a aH .vastatri xobservad an o caféIcatu .In :Resumo sd o3 °Congress oBrasileir od ePesquisa sCafeeiras , Curitiba,Pr. ,Brazil ,1975 .p .113 . D'Oliveira,B. ,1957 .A sferrugen sd ocafeeiro .Revt aCaf éport .Separat ano .3 . 61pp. Eskes,A.B .& Toma-Braghini ,M. ,1981 .Assessmen tmethod s forresistanc et o coffeelea frus t (Hemileiavastatri xBerk ,e tBr.) . PI.Prot .Bull .F.A.O .29 : 56-66. Gomez,G. ,L .& Jaramill oR. ,A. ,1974 .Temperatura se narbole sd ecaf éa lsol . Cenicafé 25:61-62 . Heath,M.C. ,1979 .Effect so fhea tshock ,actinomyci nD ,cycloheximid e andblas - ticidinS o nnonhos t interactionswit hrus tfungi .Physiol .PI .Path .15 : 211-218. Jones,I.T. ,1975 .Th epreconditionin geffec to fday-lengt han dligh tintensit y onadul tplan t resistance topowder ymilde w inoats .Ann .appl .Biol .80 : 301-309. Manners,J.G. , 1950.Studie so nth ephysiologi c specializationo fyello wrus t (Puccinia glumarum (Schm.)Erikss .an dHenn. )i nGrea tBritain .Ann .appl . Biol.37 :187-214 . Monaco,L.C. ,Ribeiro ,I.J.A. ,Tissel iF. ,0. ,Sugimori ,M.H .& Scali ,M.H ., 1973.Efeit od atemperatur a naesporulaça od aferrugem .In :Resumo s 1Congres ­ soBrasileir o sobrepraga se doença sd ocafeeiro ,Vitoria ,E.S .Brazil ,p .36 . Parlevliet,J.E. ,1979 .Component so fresistanc e thatreduc eth erat eo fepidemi c development.A .Rev .Phytopathol .17 :203-222 . Parlevliet,J.E .& Zadoks ,J.C. , 1977.Th eintegrate d concepto fdiseas eresis ­ tance;a ne wvie w includinghorizonta lan dvertica l resistance inplants . Euphytica 26:5-21 . RodriguesJr. ,C.J. , 1975.Th e coffeerusts :Hemileia.vastatri xB .an dBr .an d eflie H. coffeicola Maubl. and Rog. In:Proceeding s of 7 Colloque International sur la Chimie des Cafés,Hambour g 9-14 Juin, 1975. p. 401-408. Rodrigues Jr.,C.J. , Bettencourt, A.J. & Rijo, L., 1975. Races of the pathogen and resistance to coffee rust. A. Rev. Phytopathol. 13: 49-70. Schumann, G. & Thurston, H.D., 1977.ligh t intensity as a factor in field evalu­ ations of general resistance of potatoes to Phytophtora infestans. Phytopa­ thology 67: 1400-1402. Simons, M.D., 1972.Polygeni c resistance to plant diseases and its use in breed­ ing resistant cultivars. J. Environm. Qual. 1: 232-239. Stubbs, R.W., 1967. Influence of light intensity on the reactions of wheat and barley seedlings to Puccinia striiformis. Phytopathology 57:615-619 . Sylvain, P.G., 1955. Some observations on Coffea arabica L. in Ethiopia. Turrialba 5: 37-53. Umaerus, V., 1970. Studies on field resistance to Phytophtora infestans 5. Mecha­ nisms of resistance and applications to potato breeding. Z. PflZiicht.63 : 1-23. Vanderplank, J.E., 1975 . Principles of plant infection. Academic Press,Ne w York and London. 206pp. Vanderplank, J.E., 1978.Geneti c and molecular basis of plant pathogenesis. Springer Verlag, Berlin,Heidelberg , New York. 167pp.

47 ADDRESS

Rua Salesopolis 194,Alt o daBarra , 13100 Campinas,S.P. ,Brazil .

48 4 THE EFFECT OF LEAF AGE ON INCOMPLETE RESISTANCE OF COFFEE TO HEMILEIA VASTATRIX

A.B.Eske s andM . Toma-Braghini

InstitutoAgronômic oo fCampinas ,Campinas ,S.P. , Brazil

ABSTRACT

The influence of leafag e on incomplete resistance to race IIo f coffee leafrus t (Hemileia vastatrix), which isth emos t common race inBrazil ,wa s tested inth egreenhous e and inth e field.Thre e stages of leaf agewer e distinguished: 'young' leaveswhic hha d just fully expanded and still had ashin y appearance onth eda y ofinocu ­ lation, 'adult' leaves of 1t o4 month s old and 'old' leaves of 6t o 12month s oldwhic hwer e formed inth e foregoing growing season. Withth e susceptible Coffea arabica cvsMund oNov o andCatua i no importanteffec to f leaf age on latencyperio d and reaction typewa s observed. Lesiondensity , depending on inoculationmethod ,wa s affected to some extent,olde r leaves showing lowervalues .Wit h genotypes ofth eC . canephora cv.Kouillou ,whic hvarie d inleve l of incomplete resistance, adult leaves appeared tob emor e resistant thanyoun g andol d leaves.Thi s adultlea fresistanc ewa s expressed mainlyb y alowe r lesion density,bu t alsob y alonge r latency period and alowe r reactiontype . Within thehybri dpopulation s Icatu and Catimor, genotypeswit h race-specific incomplete resistance were tested andother s ofunknow n specificity. Resistance was expressed mainlyb y alo wreactio ntype . Generally, resistance decreased with increasing leafage ,whic hwa s occasionally associated with acomplet e change inreactio ntype . Contrary to cv.Kouillou , in Icatu andCatimo r nohighe rsuscepti ­ bility ofyoun g leaves thano f adult leaveswa s observed. The incomplete resistance torac e Io fth e differential CIFC H152/3,heterozygou s for S„4,wa s alsobette r expressed inyoun g leaves, older onesbecomin g gradually more susceptible. Iti sconclude d that for screening incomplete resistance to coffee leaf rust leaves of different ages shouldb e tested.

Additional keywords: Coffee leafrust ,coffe e differentials, durable resistance,heterogeneou s reactiontype ,component s ofresistance .

49 INTRODUCTION

Resistance topathogen s isofte naffecte d by the developmental stage ofth eplan to ro fth eplan torgans .Generally ,plant sbecom emor e resistantwhe n theymatur e (Dickinson and Crute,1974 ;Hooker ,1967 ; Hyde, 1977,Zadoks , 1961), butconflictin g results havebee n reported (Hooker, 1967). The increase inresistanc e duringmaturatio nma yb e followedb y adecreas e inresistanc ewit h further ageing (Jones and Hayes, 1971;Luttrel l et al., 1974;Menc e andPegg , 1971;Parlevliet , 1975, Umaerus, 1970). Padmanabhan and Ganguly (1954)observe d that resistance of ricet o Helminthosporium decreased with agebu t that resistance to riceblas t increased withage . Forperennia l crops the ageo fplan torgan s mayb emor e important thanth e age ofth ewhol eplant .Report s existo nth eeffec to f leaf ageo nresistanc e topathogen s (Sharmae t al., 1980,Weinhol d and English, 1964;William s and Kuc, 1969). The developmental stageo f coffeeberrie s appeared to influence resistance to coffeeberr y disease (Mulinge, 1970,va n derGraaff , 1981). Little isknow n about the influence of leaf age onresistanc e to orange coffee rusto r coffee leafrus t (Hemileia vastatrix Berk, et Br.).D'Oliveir a (1957), screening forresistanc e to coffee rust,mentione d that mature leaves aremor e resistant than succulentyoun gleaves . Results arereporte d here onth eeffec to f leaf ageo n incomplete resistance to coffee leafrus t atvariou s levels.

MATERIALSAN D METHODS

Details onresistanc e terminology, coffee and rustgenotype s used, standard inoculationmethod s and observations onnatura l and arti­ ficial infection aredescribe d by Eskes (1982).

Location. All experiments were carried outa tth e Experimental Station ofth e InstitutoAgronômic o ofCampina s (I.A.C.), S.P., Brazil.

Coffee genotypes. Thecoffe eplant suse dwer e one-year-old seed­ lings orcutting s from fieldplants .Th e age ofth e fieldplant swa s 10year swit h Icatu andwit h the Coffea arabica cvsMund oNov o and Catuai, 6year swit h Catimor, and about4 5year swit h the C. cane- phora cv.Kouillou . Icatu and Catimor consist of advanced breeding generations derived from crossesbetwee n C. arabica (4x= 44 )an d C. canephora (2x= 22 ) (Monaco,1977 ;Rodrigue s et al., 1975). Cvs

50 MundoNov o and Catuai are equally susceptible to the common race II of H.vastatrix. The genotypes used from cv.Kouillo u and fromth e Icatu andCatimo rpopulation s varied inleve lo f resistance.Th e cuttings ofth ecoffe e differentials used inexperimen t 6belonge d toth ecollectio n ofdifferential s ofth ePhytopatholog y Department ofth e I.A.C.Th e seedlings ofth eAgar o C1164-4 introduction,use d inth e sameexperiment ,wer e twoyear s old.

Rust materials. Unless stated otherwise,rac e IIo f H. vastatrix hasbee nemployed . Isolate 2,use d inexperimen t 5,wa s obtained in 1979 from an Icatu fieldplan twit h complete resistance to raceII .

Inoculation dates and method. Experiments 1t o 5wer e inoculated on 15Decembe r 1976;2 5February , 27Marc h and 24Marc h 1981,an d 10Novembe r 1980,respectively . Standard inoculationwa s doneb y spraying suspensions ofurediospore s onth e abaxial leafsurface , usingdensitie s of 0.5 to 1.5 mgurediospore s perml .Th enumbe r of spores inon em g isabou t 1.5 x io5. Inexperimen t IBdr yuredio ­ sporeswer e appliedwit h ascalpe l onth e abaxial leaf surface and spreadwit h asmal lbrush . Subsequently the sporeswer ewette db y sprayingdistille dwate r onth e leaf.Afte r inoculationth eplant s were always incubated ina humi d dark room at2 2 ±2 °Cdurin g 24t o 48hours .

Leaf age treatments. Coffee leaves canremai n onth eplan t formor e thanon eyear . InCampinas ,rapi dvegetativ e growth occurs during thewe tsumme r (Octobert o April), when4 to 8ne w leafpair sma y develop oneac hbranch .Growt h ceases during the drywinter .Natura l and also rust-induced leaf fall ishighes t atth eonse to fth edr y season (May, June), which coincideswit hharves ttime . Three stages of leaf agewer e considered. 'Young' leaveswer e just fully developed leaveswhic hwer e shiny and succulent. 'Adult' leavesvarie d in age from a fewweek s to afe wmonth s atth e day of inoculation andha d lostth eshin y appearance. 'Old'leaves , formed inth e foregoing growing season,varie d in age from 6t o 12months . A secondwa y of identifying leaf agewa sb ypositio n onth e branch.Lea fpositio n 1generall y indicated ayoun g leaf,highe r positions adulto r oldleaves .

Observations. Observations onth e components ofresistanc e were made every twoo r three days after appearance of firstsymptoms . Reactiontyp e (RT)wa s scoredbetwee n 50an d 60day s after inocula-

51 tionwhe nn o further increase insporulatio n occurred, using a0 t o 9 scale (Eskesan dToma-Braghini , 1981). Value 0indicate s absence ofvisibl e symptoms,value s 1t o 3variatio n within resistantreac ­ tiontypes ,4 to 7heterogeneou s reaction typeswit h increasing sporulation intensity andpercentag e of sporulating lesions,an d8 and 9susceptibl e reactiontype swit hmoderat e (8)t ohig h (9) sporulation intensity. Assessment ofnatura l infection inth e fieldwa sb y scoring ona 1 to 5scale .Valu e 1indicate s absence of symptoms andvalue s 2t o 5 increasing numbers ofsporulatin g lesions,associate d with anin ­ crease inreactio n type.Thes e observations weremad e annually with Icatuplant s and every 6month swit h cv.Kouillo uplants .Dat a on field infectionrepresen tmean s over theyear s 1976 to1981 .

Statistics. Forelaboratio n ofth eresults ,SPS S (Statistical Pack­ age forth e Social Sciences)ha sbee n applied. The Friedman test (Siegel, 1956)ha sbee nuse dwhe nth edistributio n of the data devi­ ated from thenorma l distribution.

RESULTS

Inexperimen t 1A and IB, the effecto f leafpositio n on thebranc h on susceptibility ofcv .Mund oNov owa s studied with two inoculation methods. Four or five leafpair s of sixbranche swer e inoculated. With the standard suspension inoculationmetho d only leaves of position 5showe d asignificantl y lower lesiondensit y (LD)tha nth e other leaves,wherea s latencyperio d (LP)wa s only slightly affected (Table 1). Resultswer e somewhat differentwhe n the inoculationwa s donewit hdr yurediospore s and subsequentwettin gb y spraying.Then , theyounges t leafha d thehighes t LDvalu e and LD declined morerap ­ idlywit h increasing leafpositio n thanwit h the suspension method. Thevariatio n inL Pwa s smallbu t significant, duet o the lowcoeffi ­ ciento fvariatio n (CV)fo rthi s component.Th e average CVvalue s forL D andL P inth e experimentswer e 88 and 6% , respectively. Inexperimen t 2th e effecto f leafag ewa s studiedwit h cv.Kouil ­ lou genotypes andcv .Mund oNov o under field conditions.O feac h genotype all leaves of fourbranche s were inoculated. All five genotypes ofcv .Kouillo u showed asignificantl y higher level of resistance inadul ttha n inol d leaves (Table 2). Differences with genotype C67-12wer e smallest and least significant.Wit h cv.Mund o Novon o significant differences were observed forth e resistance components.Wit hbot hcultivar sn o differencewa s observed between

52 leaves of similar age group but with different positions on the branches. The percentage of infected leaves that dropped from the branch before 70 days after inoculation varied from 0 to 26 and from 57 to 100 with adult and old leaves, respectively (Table 2).

Table 1. Lesion density (LD = number of lesions per leaf) and latency period (LP, in days) of leaves of cv. Mundo Novo at different positions on the branch inocu­ lated in the greenhouse by two methods.

Experiment Method of Component Leafpositio n* inoculation ofresistanc e

1A Spraying of a LD 27a2 33a 37a 29a llb suspension (stan­ dard method) LP 35a2 35a 35a 35a 36b

IB Applying dry ure- LD 38a 19ab 20ab 10b - diospores with sub­ sequent wetting LP 32a 34b 33ab 34b -

1 The age of leaves of position 1 to 4 increased from a few weeks to 3 months, respectively, while leaves of position 5 were about 7 months old on the inocula­ tion date. 2 Different letters indicate significant differences within rows according to the LSD 0.05 values.

Tabel 1. Lesiedichtheid (LD = aantal lesies per blad) en latentieperiode (LP, in dagen) van bladeren van cv. Mundo Novo met verschillende positie aan de tak geïnoculeerd in de kas op twee verschillende manieren. Leaf fall of symptomless leaves, from the same experiment, was about 0 and 20 percent for adult and old leaves, respectively. Therefore, the pronounced leaf fall of old leaves was apparently induced by the rust infections. Six leaves used in this experiment were considered as young leaves, because they still had a shiny appearance on the day of inoculation. Average lesion density of these leaves was about 2 to 4 times higher than of adult leaves, which is in agreement with the higher susceptibility of young leaves observed in the following experiment. In experiment 3, one-year-old cuttings of the same plants of cv. Kouillou as used in experiment 2 were inoculated in the greenhouse. The cuttings had grown vigorously during the two months prior to inoculation. Four to ten cuttings were inoculated per genotype, using one branch per cutting with three leaf pairs each. The leaves of the first or top pair were called young and of the second and third pairs were called adult. Generally, differences between leaf

53 £ ^ A3 « ,Û o. u «44 MD o r- r-. o CA 01 m o m OJ c 0) «-I J3 01 0) CO 4-> 00 /•C^U Ju f^ 3 oi 0) 0) W O 4-1 p. c 00 r- o. 01 •H r«H o CO •o 3 s CU 11 cj - r—1 o ai >M c o m •<]• ^o m oo CM 1) •H H ce S w 05 VJ cd o u £1 m •u G cd -o o O• * eu rH ni Q o •J -0 O CU S 00 55 I—1 00 u < cd oi 4-«> o u u OH c TJ (ü d >o 1) C -O 4^ 4^ cd 43 o* co Pu ta < O O O O CN1 4-> tj ftt» Nhlfl Ö P 1) -_/ ö CO O xi ce (0 r—1 c 43 4=1 43 co 43 43 cd 3 CO 01 o m o m oo ^ 3 l-J r-H O cd +J Ä r-4 M *"* ^^ cd cd cd co cd cd cd r—1 43 N tTi-T o m -J o co o o m r- i—i T3 W ^ •—I CU O O O r-i co »-H o O U I—1 c a ^r-i 1) H 'H UI S-J •O 05 3 43 U-) a x cd cd 43 cd Ö 4-1 O D ai o oo co I-I 1 O -H ai 4-1 43 m m -tf -tf -* -c*o E TD 0J 0J M S > Oi i—1 -H 0) 4-» c_> tH' CO +J 4J ti0) a - U /-N cd .c c o E a T3 /-N •o ai >CO O 00 • Q CU u eu S-l T3 14 -H m C hJ T3 a rH o i-H /•—s m w o 'H Ü3 d •H u QJ a. v o •o w •H i-t 01 0) o ^ -' /-a^ r-H 4-1 (0 00 4-1 0-1 U d CM •J O a» u ca co u hJ -] tn u ai w u ^-^ -H V ' s •o co 43 Ö 'H CL, CA d -' c T3 r-t w O 4->1 . Tj-rl M O UI •H •H u i-l .-H CN CO U C «4H +J rH d c o O 11 H ^ d 3 cd V •H > PH CU M 4-> W 4-1 O > CO aj 3 cd CU 1-J n-H O M W CU ^•H^ • d +J *o 3 Ö rH 4-t •H CN cd cd rH O H H mcM m d d ^ C/l i-t d 1 1 1 1 1 d < O 1—>>1 ^o O O I-- oo cd cd 43 -H o +J m 3 vO r*~ MO vO «43 O C_> O U CJ s CJ>>

54 u CU 'i-l CJ P,-H co CUXi O XJ C c/1 CO CU CU pH UI i—t m m i . M co O S r-t Ö CJ XI ai oo m a M^ï vj m oo r-t CU eu > ö w CU CO H Ui W jd CU co u> H o. s ra pO ra ra ra 3 01 O csi r-t r-H CO •H w M CS CU CU cu

m TJ c o e =s <5 S! B) f3 ff) o

rO rQ ra ra ra Il UU v£> Cv] ^ CO *«^ ra ra ra ra ra U> M cs^ HM •H CD m r- •O u TJ CU co ra ra „o ra rû rO XI 3 e« •H CU C -^ *uO O r-t CO LD U CJ -H w Ui •H •rt O TJ CU X! ö rH C CS 3 -H ui CU CU d *• Fi W CH (5 /^/—s O ra ra ra ra ra CU O CJ c ca ö Tl Ö t-H i-1 CO t~- U0 > -H OJ t-3 ai ^c/^i e ai \£> LT) U0 *U-t oo cn 01) cd CU CU co 'H CO --' UI >to . ui TJ cfl T3 M CA co ^ >> CU • e cu G dl rH Xi •H -H :* •H «uj XI u 0J d) W 3 C/l u) w •H r-H - CU o 4-1 CU CM r-t .—t eu a QJ U rl i-U u-t H yi > • -d ai C/l •i- l Cu TJ 3 •IUU> T) raI-I Ö •H r-H i-i CS] CO -* Ui O •H CU c/l U; Tl aj ^-* ai Fi CA ui Ui .-H cfl CU c/i 1* Ui C/l CU 3 Ui C j3 CU Cll CU CU Ui & O O ui CU •ui ui CU O ui •r4 dl Ui ui •H O. CU >CJ> C/1 3 XI ai c 4J Cll XS ai o >ctt •P ^-1 co tiCU r-H f-H ai Fi •P rH r-t O u> "O cd ot •H •M ö ai CO > > 1-1 d 3 01 •H |U) •H O It. CU UI Ui TJ Ö 4J a C CU Pu • 'H CU CO CO « 3 ü m cn cu cu r—t o m CU ai 3 >H •U •H XI r—I UI Ui r-t Ui >CJ 3 O ui OJ X! O O \ö o ^£> r-- oo Xi

55 ages within genotypes were notstatisticall y significant (Table3) , but thedifference s between themean s were significant, exceptfo r RT, which indicates a tendency. Thedifferenc e insusceptibilit ywa s most expressed by lesion density, which inaverag e was5 3 foryoun g leaves and2 1fo radul t ones. In experiment 4, leaves ofdifferen t positions onon eyea rol d seedlings ofCatimo r families were used. The families segregated for complete aswel l as incomplete resistance. Only seedlings with incomplete resistance were considered, 6t o1 2pe r family. Table 4 shows that theeffec t of leaf position onreactio n type varied according toth efamily . Themos t remarkable effect wasobserve d for the C2502 family, which showed nearly complete resistance inth e younger leaves whereas older leaves were normally susceptible.Th e difference inrespons e between families isprobabl y notdu et oa difference inag eo fth eleave s of similar positions, because growth rate ofth eseedling s wasabou t equal with all families. The control cv. Catuai didno tsho w a significant effect of leaf position on reaction type.

Table4 .Reactio ntyp eo fleave so fdifferen tag eo fseedling swit hincomplet e resistancet ocoffe elea frus tfo rfiv efamilie sderive d fromth ecv .Caturr ax Hybrido fTimo rcros s (Catimor)an do fcv .Catua iobserve di nth egreenhouse .

Origin Leafpositio n 1 of family 12 3 4 5 Mean

Catimor: ,2 UFV386-19 8.5a 8.3a 8.4a 8.7a 8.8a 8.5 C2502 4.0a 6.5 7.9C, 8.3C 8.4^ 7.0 UFV386-C798-A 5.2a 5.9a? 6.1a^ 6.6ab 6.7? 6.1 UFV386-C798-B 3.7a 4.3 4.3 4.7D 5.4° 4.5 C2501 2.9a 3.2a 3.6a 3.3a 3.0a 3.2

Cv.Catua i 8.5a 8.4a 8.5a 8.5a 8.9a 8.6

1 The age of the leaves of position 1 to 5 ranged from a few weeks to about 5 months on the day of inoculation, respectively. 2 Different letters indicate significant differences between leaf positions within families, according to the LSD 0.05 value.

Tabel 4. Reactietype van bladeren van verschillende leeftijd bij zaailingen van vijf families die afstammen van de kruising tussen cv. Caturra x Hybrid of Timor (Catimor) en van cv. Catuai, waargenomen in de kas. Experiment 5 was carried out in the field and included four coffee plants with differential reactions to two races of H. vasta- trix (Table 5). For each genotype/race combination all leaves of two

56 branches, growing in the shade, were inoculated. The resistance to race II of adult leaves of the genotypes H 4782-10-203 and C2012-19 was nearly complete, whereas old leaves were quite susceptible. Similar but smaller effects of leaf age were observed with other combinations. The differences in pathogenicity between race II and isolate 2 were more expressed in adult than in old leaves.

Table 5. Latency period (LP, in days), percentage of sporulating lesions (PSL) and reaction type (RT) of adult and old leaves for four coffee genotypes inocu­ lated in the field with H. vastatrix, race II and isolate 2.

Genotype Raceo r LP PSL RT isolate Leaf age1 Leaf age Leaf age Adult Old Adult Old Adult Old

Icatu: RaceI I 50 1 80 3 7 H4782-10-20 3 Isolate2 45 39 75 98 7 8

Icatu: RaceI I 39 39 75 100 8 9 H3851-2-68 9 Isolate2 36 35 94 100 9 9

Catimor: RaceI I 54 0 63 2 5 C2012-1 9 Isolate2 45 46 44 73 6 7

Cv.Mund oNov o RaceI I 38 35 97 98 9 9 Isolate2 48 38 59 92 6 8 1 The age of the adult leaves was 1 to 2 months and that of old leaves over 6 months on the day of inoculation.

Tabel 5. Latentieperiode (LP, in dagen), percentage sporulerende lesies (PSL) en reactietype (RT) van volwassen en oude bladeren bij vier koffiegenotypen geïnocu- leerd met twee isolaten van H. vastatrix in het veld (race II en isolate 2). Table 6 shows the results of inoculations, carried out in differ­ ent months and different years, of three differential clones with the resistance genes S„2 and S„4 and of seedling progeny of the £1 £1 Agaro C1164-4 introduction.Adul t leaves ofth e differential CIFC 110/5wer e not always susceptible when inoculated with the 'compatible' raceXV ;wit h one inoculationpredominantl y flecks and chlorotic lesions developed and sometimes early necrosis ofapparent ­ lysusceptibl e lesions occurred. These reactions wereno t observed withth eprogen y ofth eAgar o C1164-4 introduction,whic h isals o homozygous forth edominan t allele S„4. Therefore,th e unstable £1 adultlea fresistanc e ofth e 110/5 differential mustb e due toa resistance factor other than S„4. Table 6als o shows the relative susceptibility torac e Io fol d leaves inrelatio nt oyoun g leaves ofdifferentia l H152/3,whic h isheterozygou s for S„4. Resistance duet ogen eS H2wa sno t affectedb y leaf age or heterozygozity.

57 Table6 .Variatio n inreactio ntyp eobserve d onadul tan dol d leaveso fthre e coffeedifferential s ando fth eAgar oC1164- 4introduction ,inoculate d withtw o raceso fH .vastatri x inth egreenhouse .Eac hentr y isbase do n6 t o 12leaves .

Differential Genotype RaceI (v2V4 ) RaceX V (V2v4) orintroduc ­ tion Leaf age Leaf age Adult Old Adult Old

CIFC 32/1 S„2S„2s„4s„ 4 8-9 8-9 0-1 0-1 CIFCH152/ 3 S"2s"2S"4s" 4 2-4 4-7 0-1 0-1 CIFC 110/5 sîÎ2s"2Sfj4s" 4 2-3 2-3 4-8 7-9 AgaroC I164- 4 s|J2s[J2 SH4SH4 2-4 2-3 8-9 8-9

Tabel6 .Variati ei nreactietyp ewaargenome naa nvolwasse ne noud ebladere nva n driedifferentiërend ekoffieplante ne nva nd eAgar oC1164- 4introductie ,geïno - culeerdme ttwe efysio' sva nH .vastatri xi nd ekas .Iede rgegeve ni sgebaseer d op6 à 1 2bladeren . DISCUSSION

C. arabica cultivars. Leaf age orpositio n didno tgreatl y influ­ ence the susceptibility ofcv sMund oNov o andCatua i (Tables 1,2 , and 4).Onl y lesion densitywa s affected to someexten tb y leafpo ­ sition (Table 1).Th e decrease in lesion density of leaves atposi ­ tion4 and 5ma yb epartl y due to the effecto f light intensity (see below). Thetw o inoculationmethod s applied inexperimen t 1 appeared to interactwit h the effecto f leafposition .Thi s interaction might be related todifference s inlea f surface characteristics.Dr yure - diospores may adherebette r toth ewax y surface ofyoun g leaves than toth eharde r anddrie r surface ofolde r leaves.Besides ,wate r droplets coalesce and runnof fmor e easily fromyoun gtha n from olderleaves .

Other coffee genotypes. Tables 2 and 3sugges t theexistenc e of threephase s of susceptibility during thedevelopmen t of leaves of C. canephora cv.Kouillou .Adul t leaveswer e relativelymor eresis ­ tanttha nyoun g or old leaves.Result swit h Icatu and Catimorgeno ­ types (Tables 4 and 5)showe dhighe r resistance ofyoun g or adult leaves thano fol d leaves.Wit h some genotypes resistance was only expressed inyoun g leaves,olde r onesbein g normally susceptible,a combinationwhic hwa s never observed withth ecv .Kouillo ugenotypes .

Types of incomplete resistance involved. Completemonogeni cresis ­ tance isofte nrace-specifi c resistance, as is also the casewit h coffee leafrus t (Rodrigues et al., 1975). Tables 5an d 6sho who w certain types of incomplete resistance to coffee leafrus tma y also be race-specific.Nothin g isknow n aboutth e specificity ofth e resistance ofth e Catimor families shown inTabl e4 .

58 Within Catimor families quite distinctphenotype swer e observed with either complete resistance, intermediate resistance orhig h susceptibility. Therefore,thes e families could havemono -o roligo ­ genic resistance with incomplete dominance, asobserve dwit h theS„ 4 £1 resistance gene (Table6) . The geneticbas e ofth e resistance shownb ymos t genotypes ofcv . Kouillou seems complex. Some specificity ofresistanc e hasbee n observedwit h genotype 67-12,whic hwa smor e susceptible to arus t race detected inth eKouillo upopulatio n in198 0tha nt o raceII . The other genotypes havemaintaine d their resistance inth e field from 1976 to 1981.An y inference aboutth e durability ofthi sresis ­ tance is speculative.However ,th epecularitie s ofth eresistanc e of these genotypes ofcv .Kouillo u resemble those ofothe r reportedly stablepathosystem s (Jones andHayes ,1971 ;Parlevliet ,1975 , Umaerus, 1970). Therefore, the resistance ofcertai nKouillo u plants couldb e ofgrea tvalu e inobtainin g durable resistance to coffee rust.

Effect of leaf age on complete resistance. Mostresult sshow ni n thispape r relate to incomplete resistance,thoug h also genotypes with complete resistance were included inth e experiments.Complet e resistance iscommo n inth e Icatuan d Catimorpopulations ,bu ti s less frequent incv .Kouillo u (personal observations).A s fara sob ­ served, leaf agedoe sno tmuc h affectth e expression of complete resistance.However ,certai n genotypes with complete resistance showed symptoms onyoun g leaves (flecks and tumefactions),whil e adult leaveswer e symptomless.

Other factors influencing incomplete resistance. Asreporte dear ­ lierhig h light intensities maypredispos e coffeeplant s to coffee leafrus t (Eskes, 1982). Leaves ofdifferen tposition s onth ebranc h will receive different lightintensities .Therefor e the results of Tables 1,3 , and4 ma yb e confounded with aneffec to f lightinten ­ sity,thoug h shading of leaves was avoided asmuc h aspossibl e by suitable spacing ofth eplants . Seasonal effects onresistance ,whic h could notb e explained by differences inlea f age,hav ebee n observed with the genotypes of Icatumentione d inTabl e 5.Resistanc e ofthes e genotypes to race II washighe r inwinte r and spring than in summer and autumn.Th e same was observed forth e relative resistance ofcv .Mund oNov o to iso­ late 2.Th e resistance ofth egenotype s ofcv .Kouillo u inexperi ­ ment 2wa shighe r than inothe r field inoculations ofthes egeno -

59 types. Iti ssuppose d thatth e relatively hot and dryweather ,whic h prevailed during experiment 2,ma y have increased resistance ofth e planti nth e field.

Epidemiological aspects. Rustepidemic s incv .Kouillo u usually startearl y inth e seasonbu t subsequent development is slower than incv .Mund oNov o and Catuai (Cadena-Gomez andBuritica-Céspedes , 1980; personal observations).Thi s canb e explained, atleas ti n part,b y thedifferenc e ineffec to f leaf agebetwee nthes e culti- vars (Tables 1,2 ,3 , and4) . Thehig hpathogen-induce d leaf fall ofol d leaves alsoha s epi­ demiological connotations.Lea frus t inBrazi l has tooverwinte r on old leaveswhic h drop easily upon infection (Table 2),thereb y decreasing thecarry-ove r of inoculum toth enex tgrowin g season.

Breeding aspects. Screening forresistanc e to coffee leafrus ti s often doneb y inoculating young,recentl y expanded leaves (e.g. D'Oliveira, 1957). The results ofth epresen tpape r suggestthat , with certain coffee genotypes,thi s inoculationmetho dma y under-o r overestimate the real level ofresistanc e ofth e entireplant .Th e breederwh o isintereste d inmeasurin g incomplete resistance with precisionwil l havet o inoculate leaves ofdifferen t ages. The coffee differential heterozygous for S4 (Table 6)wa s incom­ pletely resistantwhe n inoculated with the 'incompatible' race I. Therefore,thi s differential shouldb euse dwit h care inrac eiden ­ tificationstudies .

ACKNOWLEDGEMENTS

The researchwa s supported by the Food andAgricultura l Organization ofth eUnite dNation s (FAO), the InstitutoAgronômic o ofCampina s (I.A.C.), S.P., Brazil and theAgricultura l University,Wageningen , theNetherlands .Th e authorswis h to thank all staffmember s and personnel ofth eGenetic s Departmento fth e I.A.C. for thekin d cooperation received. Special thanks aredu et oDrs .R.A . Robinson, L. Chiarappa andN.A .va n derGraaff ,o f the FAO,an d toDr .Alcide s Carvalho,o fth e I.A.C., forth e scientific stimulus received. The critical reviewing ofth e textb yprofessor s J.C. Zadoks and J. Sneep is acknowledged.

60 SAMENVATTING

Het effect van bladleeftijd op incomplete resistentie van koffie te­ gen Hemileia vastatrix.

Het effectva nbladleeftij d op incomplete resistentie van koffiete ­ gen fysio IIva n koffieroest (Hemileia vastatrix), het algemeen voorkomende fysio inBrazilië ,wer d getoetst ind eka se n inhe t veld.Dri e stadia ind ebladleeftij d werden onderscheiden: 'jong' blad datne tvolgroei d was enno g glansde opd e dagva n inoculatie, 'volwassen'bla d van 1to t4 maande n oud en 'oud'bla dva n 6to t 12maande n oud datgevorm dwa s inhe tvoorafgaand e groeiseizoen. Bijd evatbar e Coffea arabica rassenMund oNov o enCatua i werd geenbelangrijk e invloedwaargenome nva nbladleeftij d oplatentie - periode enreactietype .Lesiedichthei dwerd , afhankelijk vand e inoculatiemethode,we l enigszins beïnvloed. Bijvij fgenotype nva n hetC . canephora rasKouillou ,di evarieerde n innivea uva n incom­ plete resistentie,blee kvolwasse nbla dveela l resistenter dan jong enou dblad .Dez e resistentie uitte zichvoornamelij k door een lagere lesiedichtheid maar ook door een langere latenteperiod e en een lagerreactietype . Van dehybrid epopulatie s Icatue nCatimo rwerde n enkelegeno ­ typenme t fysiospecifieke incomplete resistentie getoetst alsmede genotypenwaarva n de specificiteitva n deresistenti e onbekendwas . Verschillen inresistenti e uitten zichvoornamelij k door verschillen inreactietype .Ove rhe t algemeenna m deresistenti e afbi jtoene ­ mendebladouderdom .Ander s danbi j de cv.Kouillo uwer dbi j Icatue n Catimor geengroter evatbaarhei d van jong tenopzicht eva n volwassen blad waargenomen. Bij sommige genotypen kwam resistentie zelfs alleenmaa r totuitin g inhe t jongeblad , terwijl hetvolwasse nbla d normaal vatbaarwas .Ee n soorgelijk effectva nbladleeftij d werd waargenomen bij inoculatie van dekoffiedifferentiee lCIF CH152/3 , dieheterozygoo t isvoo r hetresistentiege n S„4, methe tincompati ­ bele fysio I. Erword tgeconcludeer d datvoo rhe tbepale nva n incompleteresis ­ tentie tegenkoffieroes tbladere nva nverschillend e ouderdom getoetst dienen teworden .

REFERENCES Cadena-Gomez,G .& Buriticâ-Céspedes ,P. ,1980 .Expresió nd eresistenci ahori ­ zontala l aroy a(Hemilei avastatri xBerk .e tBr. )e nCoffe acanephor avarie - dadConilon .Cenicaf é31 :3-28 .

61 Dickinson,C.H .& Crute ,I.R. , 1974.Th einfluenc eo fseedlin gag ean ddevelop ­ mento nth einfectio no flettuc eb yBremi a lactucae.Ann .appl .Biol .76 : 47-61. D'Oliveira,B. ,1957 .A sferrugen sd ocafeeiro .Revt ad oCaf éport .Separat a no3 .1957 .6lpp . Eskes,A.B. ,1982 .Th eeffec to fligh tintensit yo nincomplet e resistanceo f coffeet oHemilei avastatrix .Neth .J . PI.Path .8 8 (inpress) . Eskes,A.B .an dToma-Braghini ,M. ,1981 .Assessmen tmethod s forresistanc et o coffeelea frus t (Hemileiavastatri xBerk ,e tBr.) . PI.Prot .Bull .FA O29 : 56-66. Graaff,N.A .va nder ,1981 .Selectio no fArabic a coffeetype sresistan tt ocof ­ feeberr ydiseas e inEthiopia .Thesis .Communication s 81-11o fth eAgricul ­ turalUniversit yWageningen ,Th eNetherlands .Il lpp . Hooker,A.L. ,1967 .Th egenetic san dexpressio no fresistanc e inplant st orust s ofth egenu sPuccinia .A .Rev .Phytopath .5 :163-182 . Hyde,P.M. , 1977.Th eeffec to flea fag eo ninfectio no fwhea t seedlingsb y Erysiphegramini so nsubsequen tcolon ydevelopment .Phytopath .Z .88 :299-305 . Jones,I.T .& Hayes ,J.D. , 1971.Th eeffec to fsowin gdat eo nadul tplan tresis ­ tancet oErysiph e graminisf .sp .avena ei noats .Ann .appl .Biol .68 :31-39 . Luttrell,E.S. ,Harris ,H.B .& Wells ,H.D. , 1974.Bipolari s leafbligh to f Panicum fasciculatum:Effect so fhos tag ean dphotoperio d onsusceptibility . Phytopathology 64:467-480 . Mence,M.J .& Pegg ,G.F. , 1971.Th ebiolog yo fPeronospor avicia eo npea :factor s affectingth esusceptibilit y ofplant s toloca linfectio nan d systemiccoloni ­ zation.Ann .appl .Biol .67 :297-308 . Monaco,I.e. ,1977 .Consequence s ofth eintroductio no fcoffe e rustint oBrazil . Ann.N.Y .Acad .Sei .287 :57-71 . Mulinge,S.K. , 1970.Developmen to fcoffe eberr ydiseas e inrelatio nt oth estag e ofth eberr ygrowth .Ann .appl .Biol .65 :269-276 . Padmanabhan,S.Y .& Ganguly ,D. , 1954.Relatio nbetwee nth eag eo fric eplant s and itssusceptibilit y toHelminthosporiu man dblas tdiseases .Proc .India n Acad.Sei .49 ,se cB, :44-50 . Parlevliet,E.J. , 1975.Partia l resistanceo fbarle yt olea frus tPuccini ahorde i I:Effec to fcultiva ran ddevelopmen tstag eo nlaten tperiod .Euphytic a24 : 21-27. Rodrigues,C.J. ,Bettencourt ,A.J .& Rijo ,I. ,1975 .Race so fth epathoge nan d resistancet ocoffe e rust.A .Rev .Phytopath .13 :49-70 . Sharma,J.K. ,Heather ,W.A .& Winer ,P. , 1980.Effec to flea fmaturit yan dshoo t ageo fclone so fPopulu sspecie so nsusceptibilit y toMelampsor alarici-popu - lina.Phytophatholog y 70:548-554 . Siegel,S. ,1956 .Nonparametri cstatistic s forth ebehavoria lsciences . McGraw-Hill.31 2pp . Umaerus,V. ,1970 .Studie so nfiel d resistancet oPhytophtor a infestans.5 . Mechanismso fresistanc e andapplicatio nt opotat obreeding .Z .PflZiicht .63 : 1-23. Weinhold,A.R .& English ,H. ,1964 .Significanc eo fmorphologica lbarrier san d osmoticpressur e inresistanc e ofmatur epeac hleave st opowder ymildew . Phytopathology 54:1409-1414 . Williams,E.B .& Kuc ,J. , 1969.Resistanc e inMalu st oVenturi a inaequalis.A . Rev.Phytopath .7 :223-246 . Zadoks,J.C. , 1961.Yello w rusto nwheat ,studie so nepidemiolog y andphysiologi c specialization.Tijdschr .PIZiekt .67 :69-256 .

ADDRESSES

A.B. Eskes : Rua Salesopolis 194,Alt o daBarra , 13100 Campinas, SP,Brazil . M. Toma-Braghini: Seçaod eGenetica , Instituto Agronômico of Campinas, CP 28,1310 0 Campinas,SP ,Brazil .

62 5 QUALITATIVE AND QUANTITATIVE VARIATION IN PATHOGENICITY OF RACES OF COFFEE LEAF RUST (HEMILEIA VASTATRIX) DETECTED IN THE STATE OF SAO PAULO, BRAZIL

A.B. Eskes

InstitutoAgronômic oo fCampina s (I.A.C.), C.P. 28,1310 0Campinas , SP, Brazil

ABSTRACT

Between 1977 and 1981,seve nqualitativel y distinctne w races of coffee leafrus t (Hemileia vastatrix) were detected inbreedin g plots inth e Stateo f SäoPaulo ,Brazil .Fou r races carry complex virulence againstth e resistance genes S„l, S„2, S„4an d S„5o f XI II XI £1 Coffea arabica. Three racesmatc hunidentifie d resistance geneso f C. canephora. Two ofthes ewer e isolated fromcv .Kouillo u and one from thehybri dpopulatio n Icatu.Pendin g further identification, these raceswer e indicated by thenumbe r ofthei r type cultures (Is. 2,1 0an d 11). Is.2 an d Is.1 0 showed extravirulenc e to some coffee genotypes anddecrease dvirulenc e toothe r coffeegenotypes , suggesting stabilizing selection. Three rust isolateswer e detected whichdiffere d quantitatively from the common rustrac e II.Is .1 wa smoderatel yvirulen tt oth e coffee differential for S„3 inth e laboratory but avirulent inth e greenhouse, indicating ahos tx pathoge nx environment interaction. Is. 3an d 12 showed levels ofvirulenc e intermediatebetwee n race II and Is.2 an d 10,respectively . The results show that incomplete resistance ofcoffe e to H. vasta­ trix, atvariou s levels,ca nb e race-specific.Th enatur e ofrac e formationo fcoffe e leafrus ti nBrazi l andbreedin g strategies for obtaining durable resistance are discussed.

Additional keywords: incomplete resistance, durable resistance, stabilizing selection.

INTRODUCTION

More than3 0 races of Hemileia vastatrix Berk,e tBr . (orange coffee rust, coffee leafrust )hav ebee n identified worldwide.The y are maintained atth e Coffee RustsResearc h Center (CIFC)i nOeiras,

63 Portugal (Rodrigues et al., 1975). Race IIha s thewides t distribu­ tion. Itwa s the firstt ob e detected onth eAmerica n continent,i n Brazil, 1970,a s identified atCIFC .Th e rustha sno w invaded most American coffeeproducin g countries (Wellman and Echandi, 1980). Nearly allcultivar s of Coffea arabica L. grown inthes e countries are susceptible torac e II.I nBrazil ,yiel d losses due to coffee leafrus thav ebee n estimated tob e 30%,i fn o controlmeasure s are taken (Monaco, 1977). Breedingeffort sbase d onth eresistanc e genes S„lt o SJJ4o f C. arabica (Bettencourt and Carvalho, 1968;Carvalh o et al., 1974)were ,t o agrea t deal, frustrated by the detection of the races III (v..c ) , I (v0 c) , andX V (v.c ) inth e Stateo fSä o Paulo,withi n fouryear s (Ribeiro et al., 1975). Presently the interest is focussed onresistanc e genes derived from Coffea canephora Pierre, available inadvance d breeding lines ofth ehybri d populations Catimoran d Icatu (Monaco, 1977). Thesepopulation s contain resistance to someo r all races kepta tCIF C (Rodrigues et al., 1975;Marque s andBettencourt , 1979). Susceptible plants ofthes epopulation s vary greatly inthei r level ofdiseas e inth e field (Costa andRibeiro , 1975). Incompletemino r gene resistancema y confer amor e durable protection toplan tpathogen s thancomplet e resistance related to major genes (Vanderplank, 1968;Parlevlie t and Zadoks, 1977). Durable resistance iso fgrea t importance, especially for aperennia l crop such ascoffee . To study thenatur e of incomplete resistance thevariabilit y in thepathoge nha s tob e taken into account. Inth epresen t research quantitative and qualitativevirulenc e ofcoffe e leafrus trace swa s studied.

MATERIALS AND METHODS

Definitions. 'Incomplete' resistance isher e defined as a form of resistance which doesno t fully inhibitreproductio n ofth epatho ­ gen. 'Race-specific' or 'specific' resistance canb epartl y or totally overcomeb y changes inth egenotyp e ofth epathogen . 'Major gene' resistance isrelate d toon e or a fewgenes ,eac hwit h agrea t effecto nresistance .Majo r gene resistance mayb e complete or incomplete (Parlevliet, 1979). Forth epathogen , theter m 'virulence' isuse dt o indicateth edegre eo f specificpathogenicity . 'Incomplete virulence' isuse d inanalog y to incomplete specific resistance.Th e term 'qualitative' isuse d here to indicate thatdifferen t genotypes showeasil y distinguishable phenotypes and 'quantitative'whe n differences between genotypes areno teasil y distinguishable.

64 Ä 'physiologicrace ' or 'race'contain s isolateswit h identical virulence spectrum. Inth etest s one isolate (Is.)pe r raceha sbee n used asa 'typeculture' .

Designation of virulence and resistance. The commonly adopted race designation for H. vastatrix hasbee nuse d (Rodrigues et al., 1975). Resistance genes ofcoffe e (S„)an dvirulenc e factors (v)o f H. vastatrix havebee npostulate d following the gene-for-gene theory (Noronha-Wagner andBettencourt , 1967). Racesmaintaine d atCIFC , Portugal, areidentifie d by theroma nnumeral s It oXXXII . Inth e presentpaper , raceswit hvirulenc e tounidentifie d resistance genes have notbee n given arac enumber , to avoid possible confusionwit h future racedesignatio n atCIFC .

Coffee genotypes. The differential genotypes of C. arabica (Tables1 and 2)belon g to the collection ofth eDepartmen t of Phytopathology atth e I.A.C.The y are cuttings from differentials maintained at CIFC inPortugal . C. arabica cv.Mund oNov owa s used as a susceptible check inmos texperiments .Cv . Ibaarê (LC1133-2 )i sderive d froma C. arabica accession from theHarra rprovinc e inEthiopia . Cv. Ibaarê ismor e susceptible inth e field thancv .Mund oNovo .Bot h cultivars carry theS^. 5resistanc e gene.Th e Icatu and Catimor populations jti consisto f advanced tetraploid breeding lines derived from an arti­ ficial (Icatu)an dnatura l (Catimor)hybri dbetwee n C. canephora (2x= 22 )an d C. arabica (4x= 44).C . canephora cv.Kouillo u is commercially growni nth e State ofEspirit o Santo,Brazil .Althoug h mostplant s ofcv .Kouillo u are susceptible to race II of H. vasta­ trix, greatvariatio n indiseas e level inth e field canb e observed betweenplants .Fiel dplant s usedwer e about 10year s old (Icatu, Catimor, C. arabica cvs)o r about4 5year s old (cv.Kouillou) . Cut­ tings from fieldplant swer e grownunde rmoderat e light conditions inth e greenhouse and used intest swhe n abouton eyea r old.

Rust genotypes. Single-lesion isolates from race II (v5), III

(v1 5), I (v? s) andX V (v4 5), obtained fromth eDepartmen t of Phytopathology ofth e I.A.C.,wer e used and inth e field isolates were takeneithe r fromman y lesions or single lesions. Single-lesion isolates can,wit hhig hprobability , be considered tob epur e as indicated by observations ofJ.G.J .Hoogstrate n (personal communica­ tion). Isolates 1,2 ,3 , 10,1 1 and 12wer e obtained from coffee genotypes which,unde r the conditions of isolation,wer e completely resistant to race II.Therefore ,the y canb e considered to represent races different from raceII .

65 Isolates andrace swer emaintaine d on susceptible plants inth e greenhouse, andurediospore s were stored in arefrigerato r at52 % relative humidity.Purit y ofrace s and isolateswa s checkedregular ­ lyb y inoculation ondifferentia l coffeeplants .

Inoculation methods. Inoculationswer e carried out inth e field, greenhouse or laboratory by applying suspensions ofurediospore s to the abaxial sideo fcoffe e leaves (Eskes, 1982a). The laboratory method consisted of inoculations of leafdisk s (1.8c m diameter), eachwit h a2 5 |jldrople t of asuspensio n ofurediospore s indis ­ tilledwate rwit h densities of 0.5 to 1.5 mg ml" .Th e reliability ofthi smetho d inassessin g resistance to coffee leafrus t is shown by Eskes (1982a). Thetemperatur e inth e laboratory was kepta t 22± 2°C . Thenumbe ro fdisk spe rtreatmen tuse d forth edeterminatio n of qualitative differences inresistanc e was 15t o 25 (Tables 1an d 4) and forquantitativ e differences 60 (Table 2), 120 (Table 3)an d4 5 (Table 5). Disks taken from the same leafwer e distributed over different treatments inorde r to decrease residual variance.Fo r field tests rust free leaves oftw obranche s wereused .Thes e tests were carried outwhe nth e level ofdiseas e inth e fieldwa s lowt o avoid contaminationb y natural infection. For greenhouse tests (Tables 2 and 6)betwee n 6 and 12 leaveswer e usedpe r treatment.

Inoculation dates. Results shownwer e obtained during 1978 to 1981 (Table 1), 1979-1980 (Table 2), November 1980 andMarc h 1981 (Table 3), March 1981 (Table 4), andMa y 1981 (Table5) .

Disease assessment. Reaction typeswer e scoredb y a0- 9 scale (Eskes andToma-Braghini , 1981). Scalevalu e 0indicate s absence of visible symptoms,value s 1t o3 variatio nwithi n resistant reaction types (small flecks and tumefactions to large chlorotic areaswith ­ out sporulation),value s4 to 7heterogeneou s reaction typeswit h increasing sporulation intensity andpercentag e of sporulating lesions, andvalue s 8an d 9indicat e susceptible reaction typeswit h moderate (8)t ohig h (9)sporulatio n intensity. Reaction types indicated byR (resistant)represen t scalevalue s 0t o 3,b yM R (moderately resistant)4 and 5,b yM S (moderately susceptible) 6an d 7, andb y S (susceptible)8 an d 9 (Table 1). Scoringwa s donewhe n no further development ofreactio n types occurred.Detaile d infor­ mation onho w resistance components were observed isgive nb y Eskes (1982a).

66 Statistics. Usewa smad eo fSPS S (Statisticalpackag e forth e Social Sciences). Scores forreactio n typeusin g the 0t o 9scal e were normally distributed ifn o extremeswer e involved.Therefore , linear regressionmodel swer e applied inth ecomputation s forreac ­ tion type shown inTabl e 5.

RESULTS

Races in relation to C. arabica. The sequence ofdetectio n of1 1 distinctrace s inth e Stateo fSä oPaulo ,betwee n197 1 and 1980,an d their qualitative reaction inlea fdis ktest s are shown inTabl e 1. All local C. arabica cultivars andmos tbreedin g lines of C. arabica arehomozygou s forth e S„5resistanc e allele.The y are susceptible torac e II (v,-),whic h arrived inth e State of SäoPaul o in1971 .

Till 1974,race s I (v2 5), III (v1 5) andX V (v4 5)wer e isolated from C. arabica cv.Mund oNov o (Ribeiro et al., 1975). From 1976t o 1981, the author assessed annually rust incidence inbreedin gplot s ofth eExperimenta l Station ofth e I.A.C.a tCampinas .Durin g this period,coffe eplant s carryingth eresistanc e genes S_l,S„ 2o rS„4 showed lowt omoderat e (S„4)o rmoderat e tohig h (S„l andS., 2) disease levels inth e field.Therefore ,race s I, III andX V were widely present inth ebreedin g plots. In1977 ,a smal l rust focuswa s found on aplan twit h genotype

SjjlSjjlS„2s H2 S 5S„5.Thi splan t stoodbetwee nplant s of genotypes S„ls„l s„2s„2 S„5S„5,whic hwer eheavil y attackedb y rust. Isolates ri H ri ri xi ri fromthi s rust focusbelonge d to raceXVI I (v. 25)" In I978'whe n rust incidence inth e fieldwa s low,th e size ofth e focus did not increasemuch . In1979 ,th e rustincidenc e inth e fieldwa s high and nearly allplant s ofgenotyp e SHls„lS H2s„2S H5S„5i n different breedingplots ,scattere d over theExperimenta l Station,becam e moderately orhighl y diseased.

In1979 ,rac eX (v,4 5)wa s isolated from SHlsHlS H4sH4S H5SH5 plants,whic h showed lowt omoderat e disease levels forth e first time. Inth e sameyea rmoderat e rustattack swer e found onplant swit h genotype S„2sH2 S„4s4 SH5SH5,bu tonl y race I (v2 5)coul db e isolated from theseplants .Th e resultsuggest s thatgenotype s heterozygous for S„4ar eno tcompletel y resistant inth e field to xi supposedly avirulent rustraces .A similar resultwa s obtained in thenurser y (Eskes,198 2 ).I n1980 ,th e rust incidence onS H2sH2

Su4s-,4S„5S„ 5plant s had increased considerably and raceXXI V n n ri xi (v, . J-)coul db e isolated. Inth e sameyea r this racewa s isolated by the Instituto Biologico inSä oPaul o (Fanucchi et al., 1980). 67 m m m r*- P- r- o> o> CJi CT\ O i—i -H i—i T—i i—i r- oo •*-! oo oo oo 0> C^ *• CO CJSCT\ CT i

td (0 CA « 1—1 1—t *J 1—t f—1 +J •u +-> «0 QJ O C0 (U O O O 0)

e V 'H V Hflf i-Hooco s aj 4-1 Qj 0>0>CJ\C7iC7iG\0,iO>^^Cy\ •p X O 4J

m m m J ot- -* >cN > i->t m> m > i n

« S g u o o r—1 ja r—1 ti 'H OJ 3 ö O m o « aäoäPSo5«caososDätno; u >U

l-H O m

uii/)(/3(ncßV](f]V]£(/}cf] J=> i

aspsoäcnttJcncnwo;«« 03 £ p. 1) C +J OJ o +J <+4 o. c ospiuoca«««««;«« s» OJ u oo

M u o J3 C flj «1 os •H Ei >

68 RaceXXII I (v._ .c )wa s isolated in198 0 from a C. arabica fieldplan twit hunidentifie d resistancegenotype .

.Races in relation to C. canephora. In197 9 and 1980,race swit h virulence to resistance factors of C. canephora were found forth e firsttim e inBrazil . In1979 ,som e formerly resistant Icatuplant s showed lowt omoderat e rust incidence. Is.2 (Table 1)wa s obtained fromplan tH3851-4-40 ,whic h in 1980 and 1981becam e asmuc h rusted as C. arabica cv.Mund oNovo . In1980 ,Is .2 cause dmoderat e tohig h levels ofdiseas e onabou t4 0 formerly resistant or moderately resistantplants, i na fiel dwit h about 1500 resistant Icatuplants . A fewo fthes e Icatuplant s had showndiseas ebefore ,i n 1975,bu t not in 1976 and 1977.The ybecam e diseased againbetwee n 1977an d 1980.Therefore , Is.2 issuppose d tohav ebee npresen t in Icatu in 1975 already. Itma yhav e disappeared temporarily due to frosti n thewinte r of 1975.Is .2 also overcomes the resistance of afe w Catimor genotypes (Table 3). Outo f 60genotype s tested, twower e more susceptible to Is.2 tha nt o raceII . Two races inrelatio n to genotypes ofC . canephora cv.Kouillo u were isolated in1980 . Is.1 1wa s obtained from genotype C67-1. This plantwa sheavil y rusted during thewhol e observationperiod , so Is. 11mus thav ebee npresen t atleas t from 1976 onward. Is.1 0wa s obtained from genotype C66-13,whic h in198 0 showed rust forth e firsttime ,b e ito non e leafonly . In198 2th e rustincidenc e on thisplan twa s still relatively low. Inth eobserve dpopulatio n of 65 'Kouillou' genotypes,nin ewer ehighl y resistant to race IIan d four outo fthes enin ewer e susceptible to either Is.1 0o r Is.11 .

Low virulence to S„3. About20 0 fieldplant s homozygous orhetero - ti zygous forth e S„3resistanc e allelewer eobserve d annually. On homozygous plants no rustwa s found during theobservatio nperiod . InJul y 1977,a fe wheterozygou s plants,showe d some rust lesions in the lower leafcanopy ,wit hmoderat e to intense sporulation. Small rust foci reappeared inth e following years,generall y onothe r plants, alsoheterozygou s for S„3, atth een d ofth egrowin g season (May,June )whe n the rustwa smor e severe.Betwee nyears ,som e variation innumbe r and size ofthes e foci occurred, depending on thegenera l disease level inth e field,bu tn o consistent increase was observed over theyears .Is .1 ,use d asth etyp e culture iso­ lated fromthes e foci,wa s tested twice in 1979 and 1980 inth e greenhouse andnurser y and gave similar results (Table 2). Inth e greenhouse tests, Is.1 wa s avirulent onclon eCIF C33/ 1 (S„3SH3

69 Table2 .Rang eo freactio ntype si nlaborator yan dgreenhous e testso fth edif ­ ferentialCIF C33/ 1an do fcv .Mund oNov owit hH .vastatrix ,rac eI I (v'jan d isolate 1(Is . 1).

Reactiontyp e (range)

Coffee Resistance Raceo r Laboratory Greenhouse genotype factors isolate Leaf Attached Attached disks leaves leaves

CIFC33/ 1 SH3SH3SH5SH5 RaceI I 0-1 0-2 0-1 Is.1 6-8 4-5 2-3

Cv.Mund oNov o s^Ss-iSS^S-Ji RaceI I 8-9 8-9 n n n n , T Is.1 8-9 8-9

Tabel2 .Variati e inreactietyp e inlaboratorium - enkastoetse nva nhe tdiffe ­ rentiërendegenotyp eCIF C33/ 1e nva ncv .Mund oNov ome tH .vastatrix ,fysi oI I (v) e nisolaa t 1(Is . 1).

S^BS,^), although its reaction type was higher than that of race II. In the laboratory tests, moderate to high virulence of Is. 1 on clone CIFC 33/1 was observed. On intact leaves, some well sporula- ting lesions developed among lesions of resistant types. Spores fror these lesions were reused for inoculations in the greenhouse, but again avirulence was observed on clone CIFC 33/1. On cv. Mundo Novo no difference between Is. 1 and race II was observed. In addition, isolates with virulence similar to Is. 1 were ob­ tained twice in leaf disk tests in low frequencies from rust cul­ tures belonging to race II. It is concluded that Is. 1 is a stable rust genotype, distinct from race II, with incomplete virulence to SH3. Is. 1 occurs appar­ ently rather frequently in the rust population of Campinas. Its presence can only be detected under specific test conditions, indi­ cating a host x pathogen x environment interaction.

Isolate 2. Is. 2 was isolated in the field from the Icatu genotype H3851-4-40, which was completely resistant to race II. Table 3 summarizes the differences in virulence between race II and Is. 2 observed in two field tests (1980, 1981) and in one laboratory test (1981). Differences in virulence between the rust isolates was somewhat greater in the laboratory than in the field but the same tendency was observed in both tests. Is. 2 was more virulent than race II on several Icatu genotypes and on a few Catimor genotypes (e.g. C2012-19, Table 3).Besides ,

70 Ol

Ö o> 00

a s M-H d d • O -H CO V) M <+4 m o o O O O ,O" ^ <ï il CO i-t w ^c/l >> fflS

oco m O 'H c* r-- m •H i-i d ti ai PH ai M •H n U 4-1 m co ci 4-> d

yi •o -o atii ai ö C o co co •—* u rH 4-1 «<0 ai S3 E

ö o O J2 •

ai CO oo u 1-1 CO O SS •o .o i—t 10 01 ai r-H •o o ai a Tl 4-1 4eJ »MO •H 01 O >o ^3 1 CO o 1 O 1 o^ 41 0) -* o ö z MH^ CM i-i O d 00 1 1 1 1 1 -o - .-t CO 4-1 •H 1 aj 3 • • co r~ oo O S3 C-J t-1 > o

71 Is. 2wa s lessvirulen t thanrac e IIo nC . arabica cv.Mund oNov o and avirulentt o some Catimor genotypes (e.g.C2477-2 ,Tabl e 3)t o whichrac e IIwa svirulent . Similar resultshav ebee n obtained in several greenhousetests . Is. 2 showed alowe rvirulenc e thanrac e II also onothe r C. arabica cultivars, likecv sCatuai , Ibaarê andMatari , as observed inth egreenhouse .Eve no nth ehighl y susceptible cv. Ibaarêsome ­ times alowe rreactio n typewa s obtained with Is.2 ,especiall y on young leaves. The differencesbetwee n race II and Is.2 wer e qualitative as well asquantitative .Difference s inreactio n typewer e associated with differences inlatenc yperiod , as shown inth e laboratory test (Table 3). The Icatugenotype s shown inTabl e 3ha d different levels ofresistanc e to race II,bu t allwer emor e susceptible to Is.2 . Apparently, theresistanc e factormatche d by Is.2 ca nb emor e or less effective depending onth e coffee genotype.Also , seasonal variation inreactio nwa s observed.Virulenc e was generally higher insumme r than inwinter .Furthermore , differences invirulenc e or resistance weremor e evident inyoun g leaves than inolde r leaves (Eskes andToma-Braghini , 1982). Iti sconclude d that Is.2 differ s from race II inthre easpects : 1)i tha s anextr avirulenc e factor tocertai n Icatu and Catimor genotypes, 2)i tlack s avirulenc e factor,presen t inrac e II,t o other Catimor genotypes,an d3 )i ti sslightl y lessvirulen t than race IIo n C. arabica cultivars.

Isolate 3. A racewit hvirulenc e intermediate between thato f race IIan d Is.2 wa s found inth e Icatupopulation . Is.3 , obtained fromH4782-10-203 ,showe d such intermediate virulence inman ytests . Ina laborator y test (Table4 )intermediat evirulenc e of Is.3 wa s observed inrelatio n to two Icatugenotypes . Inthi s test 6single - lesions isolates from Is.2 and 10single-lesion s isolates from Is. 3wer e used.N omarke d differences between single-lesion isolates within Is.2 and Is.3 wer e observed. Intermediate virulencewa s also observed for some single-lesion field isolates fromH3851-4-41 , a genotypewit hmoderat e resistance to race II.I na greenhous e test with cuttings from this genotype,th e field isolates showed three levels ofvirulence , similar to thato frac e II (lowvirulence) , Is. 3 (intermediate virulence)an d Is.2 (moderate virulence). Apparently, the three racesma y occur together inth e field onth e same genotype.

72 Table4 .Averag ereactio ntyp ei na lea fdis ktes to fthre eIcat ugenotype san d ofcv .Mund oNovo ,inoculate dwit hH .vastatrix ,rac eII ,an dsingle-lesio n isolatesfro mH4782-10-20 3(Is .3 )an dH3851-4-4 0(Is . 2).

Coffeegenotyp e Raceo risolat e

Race II Is. 3 Is. 2

Icatu: H3851-4-40 1.0 2.3 7.0 H4782-7-923 2.0 4.3 7.8 H4782-10-203 4.0 7.7 7.8 Cv.Mund onov o 9.0 8.5 7.5

Tabel4 .Gemiddel dreactietyp ei nd ebladschijftoet sva ndri eIcat ugenotype ne n cv.Mund oNovo ,geïnoculeer dme tfysi oI Iva nH .vastatrix ,e néénlesie-isolate n vanH4782-10-20 3(Is .3 )e nH3851-4-4 0(Is . 2).

Theresult s suggesttha t Is.3 represent s arac ewit h virulence intermediate between race IIan d Is.2 .Thi s racema y persist in future,beside s Is.2 , inth e Icatupopulation ,becaus e itha s the advantage ofbein gmor evirulen ttha n Is.2 o ncv .Mund o Novo (Table4) .

Isolates from C. canephora cv. Kouillou. In1980 ,rus tisolate s10 , 11an d 12wer e obtained inth e field from the 'Kouillou' genotypes C66-13, C67-1 and C68-11, respectively. Leaf disks from 29 Kouillou genotypes were inoculated with Is.10 ,Is .1 1 and Is.12 ,usin g three replications with 15disk s each.Tabl e 5show s the reaction type (RT)an dpercentag e ofdisk swit h sporulating lesions (PDS)o f 14representativ e genotypes ando fcv .Mund oNovo .PD S is aresis ­ tance component correlatingwit h the disease level inth e field (Eskes, 1982a). InTabl e 5th e 14genotype s havebee ndivide d into 5groups .On e Kouillou genotype (C70-12)wa s completely resistant to allisolates . The second group ofgenotype s (C69-5,C66-1 3 and C68-11)showe d susceptibility to Is.1 0an d Is.12 ,an d complete resistance to Is. 11an d race II.Genotype s ofth e third group (C67-12 to C69-15) weremor e susceptible to Is.1 1tha nt o race II,Is .1 0o r Is.12 . Inthi sgroup ,th edifference s inreactio nwer emor e quantitative thanqualitative . Is.1 0wa s often significantly lessvirulen t than race IIt ogenotype s ofth e third and fourth group. Inth e fifth group (C66-3,C66-1 2 and C70-14)difference sbetwee n rust isolates were small and rarely significant.Cv .Mund oNov o reacted equally to

73 Bl W n] (B m oo o CM O CM 00 CT> CM CM omco -J" ^ CM

D J= -H O. -P t/l d d w w .-H i-< rH 00 O, > TT t->>> T J «CL, HHfnn OH o Ö m ö d C 1 S-4-1 o 0) 0) O o &CH 4-» •U M O Q o o CO ö o «3 « « ,0 U U U *H hJ > 3ra rr ca# . c: *-u> ui 3 co S O « JBi JÜ n) CO CO X! ^^ 3 «o «o « o H^HO OD <* O w

o i/*i r- r~ co CM

0) CU > o r- o o -jn-j»

co o 4J T3 G WH raS u •H u • to cl CU o. > Û0 u u «OH •H u >U •o ca aj 3 (0 Si 4H a CO A3 U ] ra ra ra rQ ra „o Ä ra ra O »i a OH 11 • M au X raü o •H •H ca CU *>J> <4 H h )H e •^P o •U r-i CO CD U co 3 > 4-1 CO 3 • 11 O c HOB BS rH 0J « r* r-t « • -r* m r-4 lOHH i—i »-H r— »—« in 3 o 3 i i i i o o\ ^ 00 r- r^ ^ ui O0 CT\ CO ^o ^ o n « --N O o 'O vo ^ vu \£> v£3 r- o o u O O O O CJ o u U <_j CJ [K .-H Q P. « > »—I

74 all isolates. Is.1 2wa s oftenintermediat ebetwee nrac e IIan d Is. 10wit h respectt ogenotype s ofth e second, third and fourth group. Iti sconclude d thatcomplet e and incomplete race-specificresis ­ tance tocoffe e leafrus toccur s inth e 'Kouillou'populatio n which ismatche db y Is.1 0 and Is.11 ,respectively . The level of incom­ plete race-specific resistance matched by Is.1 1 (third group of genotypes, Table 5)varie dbetwee nKouillo u genotypes. Is.1 0wa s morevirulen t thanrac e IIt o some 'Kouillou'genotype s and less virulentt oothe rgenotypes , indicating thatth edisseminatio n of Is. 10 inth e 'Kouillou'populatio nma yb ehampere d to someextent . This is inaccordanc ewit h the slow increase of Is.1 0observe d in the fieldbetwee n 1979 and1982 .

DISCUSSION

Race formation in Brazil. Between 1976 and 1981,seve nqualitative ­ lydistinc t races ofcoffe e leafrus twer e found inth ebreedin g plotso fth e InstitutoAgronômic o atCampinas .Mos to fth ene w races appeared in197 9 and 1980,year s inwhic h disease level inth e field was high.Th e totalnumbe r ofrace s detected inBrazi l isno weleven . All races containvirulenc e toth e S„5resistanc e genetha ti s present inth eBrazilia n C. arabica cultivars.Eigh trace s contain, inaddition , single or combinedvirulenc e to the genes S„l, S„2an d S„4. Three racesmatc hye tunidentifie d resistance genes ofC . cane- phora. The type cultures ofthes e races (Is.2 , 10 and 11)hav ebee n sentt oCIFC , Portugal, for further identification. Therus tstrai nwhic h arrived inBrazi l in197 0wa s identified at

CIFC asrac e II (v5). This race isstil lb y farth emos t common race inBrazil .Apparently , the other 10race s detected inth e State of SäoPaul ohav e originated from race II.Mutatio n isgenerall yconsi ­ dered asth e driving force increatin gvariabilit y in asexually propagated fungi (Persone t al., 1976). Foryello w rusto fwheat , it hasbee n showntha t stepwisemutation s can explain the formation of races inth e fieldwit h increasing numbers ofvirulenc e factors (Stubbs, 1968,Stubbs , 1972). Similarly, the stepwise increase in virulence ofcoffe e leaf rusti nBrazi l (Table 1)coul dwel lb e explained by subsequentmutations . Themultipl e differences between race II and Is.2 and 10 (Tables 3 and 5)coul db ebette r explained by somatic recombination (Tinline andMa cNeill , 1969)tha nb ymutation .However ,the non emigh t also expectchange s invirulenc e inrelatio n toth e resistance genes S„l,

75 S„2 and S„4,whic h didno toccu rwit h the isolates (Table1) . Anotherpossibl e explanationmigh tb e the occurrence of interallelic interactions forvirulenc e in H. vastatrix. Three rust isolates (1,3 an d 12)hav e incomplete virulence to the corresponding coffee genotypes (Tables 2, 4 and 5).Thes e iso­ lates areconsidere d asgenotype s distinct from race II,bu tno ta s distinct races,becaus e differences invirulenc e were quantitative rather thanqualitative .Th e occurrence ofincomplet e virulence suggests that, inorde r to achieve completevirulence ,mor e thanon e change inth e genotypes of H. vastatrix mayb e necessary. Intermedi­ atevirulenc e indikaryoti c rust fungiha sbee n observed earlier (Watson andLuig , 1968;Statle r and Jones,1981 )an d canb emos t simply explained by assuming incomplete dominance ofth e avirulence allele. Is. 1i smoderatel y virulentt oth e differential 33/1 (S„3S„3 ri ri S„5S„5)i nth e laboratory, while race II isavirulent .Whe nteste d ti £1 atCIFC ,Portugal , Is.1 showe d anR reaction type (2-3)o nclon e 33/1 (C.J.Rodrigue s Jr.,persona l communication).Th e type culture ofrac eVI I (v_ _), maintained atCIFC , isno t completely virulent toclon e 33/1,givin g anM S reaction type,whil e other raceswit h thev „virulenc e factor givereactio ntype s So nthi s differential (Rodrigues et al., 1975). These observations suggestth e existence of four levels ofvirulenc e (R,MR ,M S and S)i nrelatio n toth eS„ 3 resistance gene.Th e existence ofmor e thanthre e levels ofviru ­ lencet omajo r resistance genes hasbee n reported forothe r rust fungi (Watsonan dLuig , 1968,Statle r and Jones, 1981). Schwartzbach (1979)reporte d fourvirulenc e levels ofth emonokaryoti c mildew fungus to theml- o resistance gene inbarley . The authors ascribed thiswid evariatio n invirulenc e to allelism forvirulenc e ort o non-allelic interactions inth epathogen . Thecomple xnatur e ofvirulenc e of H. vastatrix to the S„3resis ­ tance genemigh texplai nwh y this resistance gene is still effective inBrazil . In India,wher e the S„3 geneha sbee n applied incommer ­ cial cultivars from 1946 onwards,i tha s taken arelativel y long periodbefor e the rustdevelope d high levels ofvirulenc e onthes e cultivars (Narasimhaswamy, 1961;Visveswara , 1974). The future race formation ofcoffe e leafrus t inBrazi l might largely depend onth e selectionpressur e forvirulenc e as determined by theextensio n andmod e ofth eus e ofresistan tgene s inculti ­ vars. Besides,severa l factors inth e funguswhos e importance cannot be foreseenma ypla y arole .Th e frequency ofmutan t alleles for virulence isdetermine db y the relative fitness ofth e allele andb y

76 mutation rate,whic hbot hca nvar ybetwee n loci (Persone t al., 1976). Virulence may alsohav e acomple x geneticbas e (see above) and requiremultipl e changes inth e rustgenotype .Th emor emodifi ­ cations areneeded , the longer itwil l take forvirulenc e toappear , andth e greater the chancemigh tb e thatth e fitness ofth e strain isadversel y affected.

Stabilizing selection. Two ofth ene w rustrace s (Is.2 and10 , Tables 3 and 5)show , incompariso n torac e II,increase d virulence to somecoffe egenotype s anddecrease dvirulenc e toothe rgenotypes . This result indicates stabilizing selection, inth e sense that multiplication of Is.2 an d Is.1 0wil lb ehampere d to some extent inth e Icatuan dKouillo upopulation ,respectively . Thisha sbee n confirmed so far inth e field,wher e the dissemination of Is.2 an d Is. 10wa s relatively slow incompariso n to the otherne w rust races.Th ebasi s forthi s stabilizing effectcanno tb e known. It could be related to adirec teffec to fth evirulenc e (or avirulence) allele (Vanderplank, 1968), to interallelic interactions forviru ­ lence ort o anunbalanc e inth erus tgenotyp e createdb y somatic recombination ormutation .

Resistance breeding. New races ofcoffe e leafrus tappeare dwhe n onlyver y fewgenotype swit h the corresponding resistance genes were present inth ebreedin gplot s (e.g. Is.1 0an d Is.11 ,Tabl e1) . Resistance, applied incultivar splante d ata larg e scalewil l create amuc hhighe r selectionpressur e forvirulenc e than resistance present in smallbreedin g plots only.Thi s emphasizes thenee d for obtaining durable resistance to coffee leafrust .Th edurabilit y of resistance mayb e influenced by thetyp e ofresistanc e orb yth e management ofth eresistanc egenes . The results indicate thatth e resistance genes S„l, S„2 and S„4, used singly or incombination ,wil l notprovid e the coffee crop with a lastingprotectio n against coffee leafrust .Th eexperienc e with the S„3gen egaine d inBrazi l and India indicates thatthi s gene can bemor euseful ,especiall y incombination swit h other resistance genes. The quick appearance ofcomple x races from simple ones inBrazi l (Table 1)make s theefficienc y ofmultiline s incoffe e questionable. Amultilin e could facilitate, like thebreedin g plots in Campinas have apparently done,th ebuil d up ofcomple x racesb y stepwise increases invirulence .Besides ,th eperennia l coffee crop lacksth e flexibility of an annual crop needed formanagemen t of resistance

77 genes ina multiline .A multilin e could have some efficiency if cross-protectionbetwee nrace swit hdifferen tvirulenc e spectra is of importance under field conditions.N opertinen t information is available forcoffe e leafrust .Researc h inthi s direction isdesir ­ able. Thebes t approach forth eus e ofspecifi c resistance genes in coffeemigh tb e tocombin e several genes intoon e cultivar.Al l genes should giveresistanc e toth e rustpopulation . Genotypes with onlyon e ortw o genes shouldb e avoided insuc h acultiva r inorde r tomak e itmor e difficult forth e rustt obuil d upvirulenc egra ­ dually. Iti srealize d thatb y currentbreedin g methods sucha n approach isdifficul tt o achieve.Moder nmethod s ofmassiv evege ­ tativepropagatio n ofcoffe e (Söndahlan d Sharp,1977 ;Custers , 1980), mayb e ahel p inth e future toproduc e hybrid cultivars which containman y resistancegenes . Anotherbreedin g approach,whic hha sbee n initiated atth eI.A.C. , Campinas, Brazil isth e search fordurabl e resistance by selecting forincomplet e resistance withpolygeni c inheritance. So far,prom ­ isinggenotype s havebee n identified within the C. canephora cv. Kouillou andwithi n some C. arabica breeding lines (Eskes, 1981 ).Th epresen t results showed the occurence of gene-specific incomplete virulence incoffe e leaf rust and race-specific incomplete resistance incoffe e (Tables 2t o 5).Hence , selection for incom­ pleteresistanc ebase d on field observations alonewil l notb esuf ­ ficientt o indicate the durability ofth e resistance. Inaddition , information onth epolygeni c nature ofth e resistance is required (Parlevliet, 1979,Eskes , 1981a). Although this approach isno tver y simple, iti srecommende d that itshoul db e continued. Theresult s suggest still anotherbreedin g approach to durable resistance.Table s 3 and 5sho w thatth e extravirulenc e of Is.2 and Is.10 ,i ncompariso nt orac e II,wa s associatedwit h alos s in virulence tocertai n coffee genotypes. Inth e coffee literature a similar casewa s found.Th e type cultures,kep t atCIFC ,Portugal , of sevendistinc t rustrace s ofdifferen t originsbu t allwit h virulence towardsth eC . congensis 263/1differential , are only

moderatelyvirulen t to C. arabica differentials and lack the v5 virulence factorpresen t inmos tothe r rustrace s (Rodrigues et al., 1975). The aboveobservation s suggest adifficult y of H. vastatrix incombinin g certainvirulenc e factors.Possibly , thebreede r could take advantage ofthi sb y combining the corresponding resistance genes into one genotype.Fo r instance,th e resistance ofth e 263/1 differential couldb e introduced intoC . arabica cultivarswit hth e

78 S„5gene ,an d resistance of IcatuH3851-2-4 0 couldb e combined with that ofCatimo r 2477-2 (Table 3).I ncas e the rustovercome s both resistances, resultingvirulenc e canb e expected tob e low.Techni ­ cally, this approach consists of selection forth ecombinatio n of thetw o resistance factors inth ecros sprogeny ,whic h shouldb e relatively easybecaus eprobabl y major genes are involved. Iti srealize d thatthi s latter approach isne w andbase d on assumptions.However , some indications inliteratur e are in support ofsuc h anapproach .Durabl e orbroadl ybase d resistance hasbee n related to acombinatio n of specific resistance genes (Johnsonan d Taylor, 1976;Wolf e and Schwartzbach, 1978;Ah n andOu , 1982). Inter-allelic interactions forvirulenc e inth epathoge n could hamper completevirulenc e tocombination s of certain resistance genes. Recently theoccurrenc e ofdissociatio n ofvirulenc e genes in rust fungiha sbee n suggested (Vanderplank, 1982). This suggestion also leadst o the assumption thatcertai n combinations ofrace - specific resistance canb eusefu l inobtainin g durable resistance.

ACKNOWLEDGEMENTS

The research hasbee n supported by theFoo d andAgricultura lOrgani ­ zation ofth eUnite dNation s (F.A.O.), the InstitutoAgronômic o of Campinas (I.A.C.)an d theAgricultura l University ofWageningen ,Th e Netherlands. Thanks are due to allpersonne l and staffmember s of theGenetic s Department ofth e I.A.C.,wh o contributed to this research.Drs .Alcide s Carvalho (I.A.C.), N.A.va nde rGraaff , R.A. Robinson andL .Chiarapp a (F.A.O.)ar e thanked forth e stimulus received.Th e critical reviewing ofth etex tb y professors J.C. Zadoks and J. Sneep is acknowledged. Mrs.Masak oToma-Braghin i andMr .J.G.J .Hoogstrate n aret ob e thanked forthei r activeparti ­ cipation inth e experimentalwork .

SAMENVATTING

Kwalitatieve enkwantitatiev e variatie inpathogenitei tva n fysio's van koffieroest (Hemileia vastatrix) gevonden ind e staat SäoPaulo , Brazilië.

Tussen 1977e n 1981werde nzeve nkwalitatie fverschillend e nieuwe fysio'sva nkoffieroes t (Hemileia vastatrix) gevonden inveredelings - proefvelden ind e staatSä oPaulo ,Brazilië .Vie r fysio's bevatten complexevirulenti e tegen de resistentiegenen S„l, S„2, S„4e nS„ 5

79 van Coffea arabica. Drie fysio's neutralizeren nog ongeïdentificeerde resistentiegenen van C. canephora. Twee van deze fysio's werden van cv. Kouillou geïsoleerd en één van de hybride populatie Icatu. In afwachting van verdere identificatie zijn deze drie fysio's aangeduid met hun isolaatnummer (Is. 2, 10 en 11). Is. 2 en 10 vertoonden naast extra virulentie t.o.v. bepaalde koffiegenotypen verminderde virulentie t.o.v. andere genotypen, hetgeen kan wijzen op stabilise­ rende selectie. Tevens werden drie roestisolaten ontdekt die kwantitatief ver­ schilden van het gangbare fysio II. Is. 1 was matig virulent op de differentiërende kloon voor het S„3 resistentiegen in het laborato­ ry rium maar avirulent in de kas, wat duidt op een waardplant x fysio x milieu-interactie. De virulenties van Is. 3 en 12 lagen in tussen die van fysio II en Is. 2, respectievelijk fysio II en Is. 10. De resultaten tonen aan dat incomplete resistentie van koffie tegen koffieroest, van variërend niveau, fysio-specifiek kan zijn. De aard van het ontstaan van de nieuwe roestfysio's in Brazilië en veredelingsstrategieën voor het verkrijgen van duurzame resisten­ tie tegen koffieroest worden besproken.

REFERENCES

Ahn, S.W. & Ou, S.H., 1982. Quantitative resistance of rice to blast disease. Phytopathology 72: 279-282. Bettencourt, A.J. & Carvalho, A., 1968. Melhoramento visando a resistência do cafeeiro à ferrugem. Bragantia 27: 35-68. Carvalho, A., Fazuoli, L.C. & Monaco, L.C., 1974. Iarana, novo cultivar de Coffea arabica. In: Resumos 2° Congresso Brasileiro de Pesquisas Cafeeiras, Poços de Caldas, M.G., Brazil, p. 168. Costa, W.M. & Ribeiro, I.J.A., 1975. Resistência a H. vastatrix observada no café Icatu. In: Resumos 3° Congresso Brasileiro de Pesquisas Cafeeiras, Curitiba, Pr., Brazil, p. 113. Custers, J.B.M., 1980, Clonal propagation of Coffea arabica L. by nodal culture. In: Proceedings 9 Intern. Colloq. on Coffee, London, p. 589-596. D'Oliveira, B. As ferrugens do cafeeiro. Revta Café port. Separata no. 3. 61 pp. Eskes, A.B., 1981 . Expression of incomplete resistance to pathogens. In: Pro­ ceedings of NATO Advanced Studies Institute on Durable Resistance, Martina Franca, Italy,, October 1981. In press. Eskes, A.B., 1981 . Incomplete resistance to coffee leaf rust. In: Proceedings of NATO Advanced Studies Institute on Durable Resistance, Martina Franca, Italy, October 1981. In press. Eskes, A.B., 1982 . The use of leaf disk inoculations in assessing resistance to coffee leaf rust (Hemileia vastatrix). Neth. J. PI. Path. 88: 127-141. Eskes, A.B., 1982 . The effect of light intensity on incomplete resistance of coffee to Hemileia vastatrix. Neth. J. PI. Path. 88: 191-202. Eskes, A.B. & Toma-Braghini, M., 1981. Assessment methods for resistance to coffee leaf rust (Hemileia vastatrix Berk, et Br.). PI. Prot. Bull. F.A.O. 29: 56-66. Eskes, A.B. & Toma-Braghini, M., 1982. The effect of leaf age on incomplete resistance of coffee to Hemileia vastatrix. Neth. J. PI. Path. 88: 215-226. Fanucchi, M., Beretta, M.J.G., Martins, E.M.F., Moraes, W.B.C. & Figueiredo, M.B., 1980. Verificaçao da presença, no Estado de Sâo Paulo, da raça fisio- lógica de Hemileia vastatrix portadora dos genes de virulencia v„, v,, v_. In:

80 Resumos8 °Congress oBrasileir od aSociedad eBrasileir a deFitopatologia , Itaguai,R.J. ,Jul y 1980.p .131 . Johnson,R .& Taylor ,A.J. ,1976 .Geneti ccontro lo fdurabl eresistanc et oyello w rust (Pucciniastriiformis )i nth ewhea tcultiva rHybrid ed eBersee .Ann . appl.Biol .81 :385-391 . Marques,D.V . &Bettencourt ,A.J. ,1979 .Resistênci a àHemilei avastatri xnum a populaçaod eIcatu .Garci ad eOrta ,Sér .Est .Agron .6 : 19-24. Monaco,L.C. ,1977 .Consequence s ofth eintroductio no fcoffe elea frus tint o Brazil.Ann .N.Y .Acad .Sei .287 :57-71 . Narasimhaswamy,R.L. ,1961 .Coffe elea fdiseas e (Hemileia)i nIndia .India n Coff.25 :382-388 . Noronha-Wagner,M .& Bettencourt ,A.J. ,1967 .Geneti c studyo fth eresistanc eo f Coffea spp.t olea frus t 1.Identificatio nan dbehaviou ro ffou rfactor scon ­ ditioning diseasereactio ni nCoffe a arabicat otwelv ephysiologi c raceso f Hemileiavastatrix .Can .J .Bot .45 :2021-31 . Parlevliet,J.E. ,1979 .Component so fresistanc etha treduc eth e rateo fepidemi c development.A .Rev .Phytopath . 17:203-222 . Parlevliet,J.E .& Zadoks ,J.C. , 1977.Th eintegrate d concepto fdiseas eresis ­ tance;a ne wvie w includinghorizonta l andvertica lresistanc e inplants . Euphytica 26:5-21 . Person,C , Groth,J.V . &Mylyk ,O.M ., 1976 .Geneti c changei nhost-parasit e populations.A .Rev .Phytopath .14 :177-187 . Ribeiro,I.J.A. ,Sugimori ,M.H. ,Moraes ,S.A .& Monaco ,L.C. ,1975 .Raça sfisio - lógicasd eHemilei avastatri xBerk ,e tBr .n oEstad od eSä oPaulo .Summ a Phytopathologica 1:19-22 . RodriguesJr .C.J. ,Bettencourt ,A.J .& Rijo ,L. ,1975 .Race so fth epathoge nan d resistancet ocoffe erust .A .Rev .Phytopath . 13:49-69 . Schwartzbach,E. ,1979 .Respons et oselectio n forvirulenc eagains tth eml- o basedmilde w resistance inbarley ,no tfittin g thegene-for-gen ehypothesis . BarleyGen .Newsl .9 :85-88 . So'ndahl,M.R .& Sharp ,W.R ., 1977 .Hig hfrequenc y inductionso fsomati cembryo s inculture d leafexpiant so fCoffe a arabicaL .Z .Pflanzenphysiol .81 : 395-408. Statler,G.D .& Jones ,D.A. , 1981.Inheritanc eo fvirulenc ean duredia lcolo ran d size inPuccini a reconditatritici .Phytopatholog y 71:652-655 . Stubbs,R.W. ,1968 .Artificia lmutatio ni nth estud yo fth erelationshi pbetwee n raceso fyello w rusto fwheat .Proc .Eur .an dMed .Cerea lRust sConf. ,Lisbon , 1968,p .60-62 . Stubbs,R.W. , 1972.Th e international surveyo ffactor so fvirulenc eo fPuccini a striiformisWestend .Proc .Eur .an dMed .Cerea lRust sConf. ,Prague ,1972 , p.283-288 . Tinline,R.D .& Ma cNeill ,B.H. ,1969 .Parasexualit y inplan tpathogeni cfungi . A.Rev .Phytopath .7 :147-170 . Vanderplank,J.E. ,1968 .Diseas e resistance inplants .Academi cPress .20 5pp . Vanderplank,J.E. ,1982 .Host-pathoge ninteraction s inplan tdisease .Academi c Press.20 1pp . Visveswara,S. ,1974 .Periodicit y ofHemilei a inArabic a selectionS .795 .India n Coff.38 :49-51 . Watson,I.H .& Luig ,N.H. ,1968 .Progressiv e increase invirulenc e inPuccini a graminisf.sp . tritici.Phytopatholog y 58:70-73 . Wellman,F.L .& Echandi ,E. ,1981 .Th ecoffe erus tsituatio ni nLati nAmeric ai n 1980.Phytopatholog y 71:968-971 . Wolfe,M.S .& Schwartzbach ,E. ,1978 .Pattern so frac echange s inpowder ymildews . A.Rev .Phytopath .16 :159-180 .

ADDRESS

Rua Salesopolis 194,Alt o daBarra , 13100 Campinas,S.P. , Brazil.

81 6 VARIATION FOR INCOMPLETE RESISTANCE TO HEMILEIA VASTATRIX IN COFFEA ARABICA

A.B. Eskes1 andA . Carvalho

InstituteAgronômic o ofCampina s (I.A.C.), Campinas S.P., Brazil

INDEXWORD S Coffee LeafRust ,durabl e resistance,component s ofresistance , leaf retentionperiod , yield.

SUMMARY

Variation forincomplet e resistance tocoffe e leafrus twa s studied in Coffea arabica. Disease level inth e fieldwa s scoredb y a0 t o9 scale. Components ofresistanc e observed in laboratory andgreen ­ house testswer e latency period (LP),lesio n density (LD)an d leaf retentionperio d (LRP). LRP determines theduratio n of sporulation. Lines of 'MundoNovo ' and 'Catuai' showed relatively small but significant differences fordiseas e level inth e field. Heritibility was low (0.31)an dpar to fth evarianc e (34%)wa s explainable by a significant correlationbetwee n disease level andyieldin g capacity ofth e lines. Thehig h susceptibility of 'Ibaarê', incompariso n to other cul- tivars,wa sbes texplaine d by longer LRPvalue s for 'Ibaarê'.Als o its LPwa s generally shorter thantha to fothe rcultivars . Among coffee accessions fromEthiopi a greatvariatio nwa sob ­ served fordiseas e level inth e field.A significantpar t ofth e variance (35%)wa s explained by differences inyieldin g capacity. Variation forresistanc e components was observed betweenaccessions , tested inth egreenhous e and laboratory.However ,result s ofth e two testswer e inconsistent. Transgressive segregation for incomplete resistance was observed inF _population s ofth e crossbetwee nAgar o C1164-19 and 'Catuai', tested inth e laboratory andgreenhouse .Resistanc e was expressed by a longer LP,a lowe rLD ,a certai npercentag e ofnon-sporulatin g lesions and, in somepopulations ,b y early necrosis oflesions .

1 Present address:Ru a Salesopolis 194,Alt o daBarra , 13100 Campinas, S.P., Brazil

82 Prospects forbreedin g for incomplete resistance to H. vastatrix in C. arabica are discussed. Factorswhic hma yhampe r selection pro­ gress are:a )th epositiv e correlationbetwee nyiel d and disease level inth e field,b )th e relative small genetic variation forin ­ complete resistance among C. arabica cultivars andproductiv e breed­ ing lines,an d c)inconsistenc y ofresult sbetwee nresistanc etests .

INTRODUCTION

The centre oforigi n of Coffea arabica L., an autogamous tetraploid species, isEthiopia . Inthi s country coffee leafrus t (Hemileia vastatrix Berk, etBr. )i sendemi c but isn o serious diseaseproble m (Sylvain, 1955). Rust severity isgenerall y greater atlo w altitudes than athig h altitudes.Th ewa y C. arabica hasbee n distributed for cultivation has led tocoffe epopulation s with relatively limited geneticvariatio n inmos tcoffe eproducin g countries.Breedin gef ­ fortshav ebee n undertaken, including interspecific hybridization, to increasevariatio n (Monaco, 1977). InBrazil ,th etw omos tgrow ncultivar s are 'MundoNovo ' and 'Catuai'.The y derive fromvarieta l crosses andvariatio n ispresen t foryiel d and other characteristics between lines ofthes e cultivars (Carvalho, 1967,Carvalh o andMonaco , 1972). Bothcultivar s are sus­ ceptible to race IIo f H. vastatrix. Race II isth emos t common rust race and the firston e detected inBrazi l in 1970.Yiel d losses due tocoffe e leafrus t are estimated tob e about30 %(Monaco , 1977). In the Stateo f SäoPaulo ,rus tepidemic s develop yearly from December toJuly ,whe n themaximu m rust incidence isobserve d (Figueredo et al., 1974). Rustincidenc e declines during the dry season (May- September)mainl y duet orust - andenvironment-induce d leaf fall. Breeding forresistanc e to H. vastatrix, based onmajo r genes in C. arabica, hasno tbee n effective (Ribeiro et al., 1975,Eskes , 1983). Incomplete,polygenicall y inherited resistance mightb emor e durable thanresistanc e related tomajo r genes (Vanderplank,1968 , Parlevliet and Zadoks, 1977). Little isknow n aboutth e variation for incomplete resistance in C. arabica. Sylvain (1955)an dva nde r Graaff (1981)reporte d differences inth e level ofresistanc ebe ­ tween selections from different localities inEthiopia .Ribeir o et al. (1981)reporte d aconsistentl y higher disease incidence inth e field of 'Ibaarê'tha no f 'MundoNovo 'o r 'Catuai'i nBrazil .Eske s (1983)showe d thatth emajo r resistance genes S„3 and S„4ma y confer incomplete resistance inth e field to 'incompatible'rus traces .

83 Thepresen tpape r describes variation for incomplete resistance to H. vastatrix in C. arabica, unrelated toknow nmajo rgenes .

MATERIALS AND METHODS

Coffee genotypes inth e fieldwer e 4-15 year oldtree s of C. arabica accessions fromEthiopia ,cultivar s andbreedin g lines.Th eacces ­ sions are selections madeb y P.G. Sylvain (Sylvain, 1955)an db y an F.A.O. mission in196 4 (Meyer et al., 1968). CvsMund oNov o and Catuai consist ofadvance d generations derived from thevarieta l crosses 'BourbonVermelho ' x 'Sumatra' and 'MundoNovo 'x 'Caturra Amarelo', respectively. 'Caturra' is amutan twit h semi-dwarf growth, probablywit h a 'Bourbon'geneti cbackground . 'Amarelo' (yellow)an d 'Vermelho' (red)indicat e the color ofth e ripeberries ,whic h isa monogenic trait (Carvalho,1967 ,Carvalh o &Monaco , 1972). Symbolsuse d to indicate theorigi n of acoffe e genotype refer to the type ofselectio n (L= Line ,H =Hybrid )o r theplac e ofselec ­ tion (C= Campinas ,M =Mococa ,P =Pindorama ,R = Ribeirä o Preto) and are followedb y anumbe rwit h oneo rmor e dashes (generations). 'Ibaarê', isa selectio n (LC1133-2) from an accession (C1133)fro m theprovinc e ofHarra r inEthiopia . Itha s ahig hyieldin g capacity (Monaco, 1977), but isver y susceptible to rust (Ribeiro et al., 1981, Sylvain, 1955). The accessions C1164-4 and C1164-19 come from Agaro,Ethiopia .The y arehomozygou s dominant forth e S„4allele . All cultivars andbreedin g linesuse d arehomozygou s dominant for the S„5resistanc e allele (Bettencourt& Carvalho, 1968). Some of theaccession s fromEthiopi aposses s the S„lallel e (Meyere t al., 1968). Inal ltests ,compatibl e rustrace swer e employed with theviru ­ lence factorsv 5 (race II),v. .5 (race III)o r v„ 5 (race XV). Iso­ lates from these raceswer emaintaine d onplant s of 'Ibaarê' and 'MundoNovo ' inth e greenhouse.Purit y ofrace swa s checkedregu ­ larlyb y inoculation ondifferentia l coffeegenotypes . Inoculations shown inTabl e 1,2 and 6wer emad e inOctobe r1977 , August 1979,an dMa y 1979,respectively . Inoculationmethod s follow those specified byEske s (1982a). Ingreenhous e tests,1 0t o1 4 one-year-old seedlings wereuse dpe rprogeny , inoculating 4 leaves per seedling, except inth eexperimen t shown inTabl e 1,wher e6 two-year-old seedlingswer euse dpe rprogeny ,wit h 8 leaveseach . The disease level inth e fieldwa s assessedb y counting the number ofrus t lesionspe r leaf (rust incidence)o rwa s scoredb y

84 a 0t o 9scal e (disease score,DS ) (Eskes& Toma-Braghini , 1981). Value 0idicate s absence of sporulating lesions and 1t o 9increas ­ ing incidence of sporulating lesions.Component s of incomplete resistancewer e observed asdescribe d byEske s (1982a). Latency period (LP)i sth e time,i ndays ,betwee n inoculation and sporula­ tiono f50 %o fth e lesions.Lesio n density (LD)i sth enumbe r of lesionspe r leaf areaunit .Th e sporeproductio npe r lesion (SP)ha s beenmeasure d by diluting urediospores ina know nvolum e ofwate r and subsequent countingwit h ahaemocytometer .Th e durationo f sporulationwa s determined by theperiod , indays ,betwee n spo­ rulation of 50%o fth e lesions and leafabscissio n (leaf retention period, LRP).LR Pwa spositivel y correlated with LD.Therefore ,th e tablevalue s ofLR P shown arecorrecte d values obtained by analysis ofcovariance ,wit h LD asth ecovariate .Fo r statistical elaboration usewa smad e ofcompute rprogramme s from the Statistical Package for the Social Sciences (SPSS). For calculations ofL Dvalues, a squar e root transformationwa s applied to improvenormalit y ofth edata .

RESULTSAN D DISCUSSION

Lines of C. arabica cultivars. Eleven lines ofol d andne wBrazil ­ ian cultivars were tested inth e greenhouse (Table 1). These lines have similar levels ofdiseas e inth e field, except 'Ibaarê'whic h is substantially more affectedb y rust thanth e othercultivars . Table 1show s thatth e differencesbetwee n lines for latency period (LP)an d sporeproductio n (SP)wer e non-significant and for lesion density (LD)barel y significant atth e 5%level .Highes tsignifi ­ cancewa s observed forth e leafretentio nperio d iLRP). 'Ibaarê' showed a2 0 to 35day s longer LRP thanth e other cultivars.Th e coeffients ofvariatio n (C.V.)wer e extremely high forL D andSP .

Accessions from Ethiopia. In197 5 afiel d trialwa s established in Campinaswit h20 0 accessions selected by the F.A.O. mission in196 4 inEthiopi a (Meyere t al., 1968). Each accessionconsist s of seed progeny from individual trees.The y areplante d in6 replications with 3tree s each.Grea tvariatio n invegetativ evigour ,yield , growthhabit ,earlines s and adaptability isobserved . The average yield ofth eprogenie s is about30 %o fth eyiel d of 'MundoNovo 'o r 'Catuai'.Th e coffee leafrus tdiseas e levelvarie s greatly; some progenies showver y fewdiseas e symptoms others aremor e diseased than 'MundoNovo 'o r 'Catuai'.

85 Table 1.Component so fresistanc eo f1 1line so fC .arabic a cultivars inoculated withrus trac eI I

Ibaarê LC1133-2 39 1.0 3.3 97 Catuai H2077-2-5-81 40 1.4 3.4 60 MundoNov o LCP474-7 41 0.6 3.0 59 CaturraVermelh o - 42 0.4 2.8 73 BourbonAmarel o - 42 1.0 3.5 62 Erecta - 43 1.3 3.9 77 MundoNov o LCMP376-4-22 43 0.7 2.6 63 Catuai H2077-2-5-66 44 0.9 2.4 66 Arabica LC12-1-18-2 45 0.3 2.3 67 Sumatra - 45 0.5 1.7 68 Catuai H2077-2-5-99 46 0.6 2.0 63

Mean 42 0.8 2.8 68 C.V.(% ) 7 65 48 24 Fvalu eo flin eeffect s 2.0 2.2 1.1 3.7 Significance ofF P>.0 5 .05.05 P<.001

1 LP= latenc yperiod ,i ndays ,L D= lesio ndensity ,numbe ro flesion spe r cm2, SP= spor eproductio npe rlesion ,x 10 4,durin g 19day sfro mth eda yL Pwa scom ­ pleted,LR P= lea fretentio nperiod ,i ndays ,fro mth eda yL Pwa scomplete dtil l abscissiono fth eleaves .

Productivity appeared to influence the level of disease in the field. The coefficient of correlation between yield till 1980 and disease score in the field (0-9 scale) of the progenies in 1980 was significant (r = 0.59). Therefore, about 35% of the variance for disease level in this population can be explained by yield. Some progenies showed low productivity and a high disease level, but the opposite was not observed. Greenhouse tests were carried out with race III of H. vastatrix on 10 progenies with varying disease levels in the field (Table2) . Significant differences were observed for LP, LD and LRP. Only two progenies differed significantly from 'Mundo Novo' for LP and none for LD (Duncans test at P i 0.05). Five introductions had a signifi­ cantly longer LRP than 'Mundo Novo' at P i 0.05 (t-test). The coef­ ficients of correlation (r)betwee n the disease score in the field and the components of resistance and yield were low and non-signifi­ cant (r = -0.31, 0.15, 0.50 and 0.47, for LP, LD, LRP and yield, respectively). The coefficient of multiple correlation (R) was higher (0.72), but still not significant. The results indicate that significant variation for LP or LD observed in the greenhouse may not be correlated with the disease level in the field. Variation for LRP and yield appear to be better indicators for the disease

86 Table 2. Disease level, yield (kg fresh berries per tree), and components of resistance of 10 C. arabica introductions from Ethiopia and of 'Mundo Novo' inoculatedwit h rust raceII I thegreenhouse) . (vl,5) in

Field Progenytes ti nth e Introduction greenhouse or Disease score1 Yield cultivar (0-9 scale) (1978-81) LP LD LRP

LC2038 4.12 a 9.1 53 26 55 LC2181 4.1 a 6.0 54 16 57 LC2109 5.7 be 14.5 46 27 67 LC2063 6.32 be 12.9 46 29 62 LC2111 6.3 be 14.8 49 34 52 LC2066 6.5 be 10.3 50 38 56 LC2157 6.5 be 6.9 50 23 59 'MundoNovo ' 7.0 cd 21.2 50 26 53 LC2026 8.0 de 18.1 52 22 66 LC2025 8.4 e 13.7 49 25 68 LC2106 8.4 e 10.3 50 26 64

Mean 6.5 50 27 61 C.V. (W 14 9 58 20 Fvalu e 15.9 6.4 2.5 3.4 Significanceo f F (P) <.001 <.001 .007 .001 1 Average of two readings made in June 1980 and 1982. Different letters indicate significant differences according to the LSD. ». value. 2 These values may have been slightly underestimated due to the presence of the S„l gene in some plants of these lines. 3 LP = latency period, in days, LD = lesion density (number of lesions per leaf), LRP = leaf retention period, in days, from the day LP was completed till leaf abscission. level in the field. In December 1979, 65 genotypes were tested in the laboratory with race III. Fourty leaf disks, collected from field plants with rela­ tively good yield and/or low disease incidence, were inoculated. The suitability of the laboratory test in assessing incomplete resis­ tance is shown by Eskes (1982a). The components of resistance LP and LD, expressed as percentages in relation to the average values of 'Mundo Novo1 and 'Catuai' (= 100%) ranged from 90 - 130% and 20 to 180%, respectively. In May 1980, a greenhouse test was carried out on seedling progenies from the same 65 field plants. Ranges observed for LP and LD were similar to those in the laboratory test. However, the coefficients of correlation between the tests were low, insig­ nificant for LP (r = 0.12) and barely significant at the 5% level for LD (r = 0.26), indicating a genotype x test interaction. The coefficients of correlation between LP and LD were significant but not very high (-0.61 and -0.41 for the laboratory and greenhouse tests, respectively).

87 Based on the results of the resistance tests 10 accessions were selected and crossed with cv. Catuai. In the cross progenies trans­ gression for resistance might occur (see hereafter). Selections with good yielding capacity and increased incomplete resistance might be obtainable in these progenies.

Cvs Catuai and Mundo Novo. The Catuai cultivar consists of selected lines derived from a cross (H2077) between 'Mundo Novo' and 'Caturra Amarelo'. In May 1978, rust incidence was observed in the field with

18 F3, F4 and F,-lines , each with 6 replicates of single plants. For 17 lines, rust incidence varied from 3.3 to 6.5 lesions per leaf. Differences between these lines were not significant. One line (H2077-2-5-50) was significantly more diseased than the rest, with an average of 12.3 lesions per leaf. In June 1979, 1980 an 1981 the disease level was scored with 99 lines, derived from twelve different mother tree selections of 'Mundo Novo' and one of 'Catuai' (Table 4).Pe r line six replica­ tions of two plants each were observed. For individual years, the analysis of variance detected a significant line effect for disease score but coefficients of correlation between successive years were low or even negative (Table 3).Diseas e scores (DS)wer e positively

Table3 .Matri xo fcoefficient s oflinea rcorrelatio n (r)betwee naverag eyiel d (Y)an dcoffe elea frus tdiseas escor e (DS)o f8 8line so f 'MundoNovo 'an d1 1 lineso f 'Catuai'fo rthre eyear s (year1 = 1979,yea r2 = 1980 ,an dyea r3 = 1981). Significanceo fr = ± 0.2 0 (P= 0.05 )an d± 0.25 4( P= 0.01) .

Y +2 Y Totalyiel d DS DS Y +Y (1972-1981) 2 3 l Y2 Y3 3

DS .1.199 .56.5 6.6 7-.2 .671 -.2.431 .43 .47 DS, -.2-.288-.0 -.04 4.6 7-.5 .677 -.57 .40 _ Dsf _ .31-.5 .312 -.5.692 .69 .23 DS +2DS ,+ DS „ .44 .58 Y, -.1-.188 .41.4 1 .52 Y, -.7-.788 .34 ii -01

Y1 + 2Y2+ Y 3 .68 correlated with yields (Y),i n the same year, r values ranging from 0.67 to 0.69, and negatively correlated with yields in the foregoing or following year. Yield in 1980 was negatively correlated with yield in 1979 and 1981 (Table 3),illustratin g the effect of bien­ nial bearing of the coffee crop. Some lines were in a different biennial bearing pattern than others. To account for the differences

88 Table4 .Averag e coffeelea f rustdiseas eleve li nth efiel dan dtota laverag e yield (kgfres hberrie spe rtree )o fline sfro m 12mothe rtree so f 'MundoNovo ' (MN)an dfro m 1mothe rtre eo f 'Catuai'(C) .

Mother tree Numbero f riseas e score (1979-1981) selecti linesteste d on number 0- 9 Estimatednumbe r Yield sea! e oflesion s (1972-1981 ) perlea f

MN-474 14 6.8 a1 2.4 38.0 b MN-471 3 6.8 a 2.5 35.3 ab C-H2077 -2-5 11 7.0 a 2.8 35.3 a MN-502 9 7.3 b 3.6 42.2 de MN-467 10 7.3 be 3.7 39.8 be MN-500 8 7.4 be 3.8 39.2 be MN-501 3 7.4 bed 3.9 41.0 cde MN-480 4 7.5 bed 4.0 38.6 be MN-464 6 7.6 bed 4.3 40.5 cd MN-515 5 7.6 bede 4.5 41.9 cde MN-475 6 7.6 cde 4.5 44.4 e MN-388 12 7.7 de 4.6 43.6 e MN-376 8 7.9 e 5.6 44.4 e

Mean 7.4 3.9 40.3

1 Different letters indicatesignifican tdifference saccordin g toStudent s t-test atP i 0.01. inbearin gpattern , averageD S andY value swer e estimated by the formulas (DS1 +2-DS 2 +DS 3)/4 and (Y1 +2-Y 2 +Y 3)/4.Afte r apply­ ingthes e calculations,th e averageD Swa s still significantly cor­ related to average yield over 1979/81 and to total yield from 1972 to 1981 (r= 0.4 4 and 0.58, respectively) (Table3) . The analysis ofvarianc e forth e average disease score indicated significant differences between the 99 lines,wit h scalevalue s ranging from 6.2 to8.2 .Th e differences between the lines,arrange d permothe r tree selection, are shown inTabl e 4.Heritibility , as calculated from the line and residual mean squarevalues ,wa s 0.31. The significant correlation ofdiseas e scorewit h total yield (r=0.58 ,Tabl e 3)indicate s that (0.58)2 x 100= 34 %o fth evari ­ ance fordiseas e score isdu et o theyieldin g capacity ofth eline . Therefore, andbecaus e ofth erelativel y low geneticvariatio n in 'Catuai' and 'MundoNovo' ,selectio n for lowdiseas e scores within these cultivarsdoe sno t seem rewarding.

Crosses between 'Ibaarê' and other Brazilian cultivars. Table5 shows disease scores inth e field, observed inMarc h 1977,o f

•Ibaarê', 'MundoNovo' , 'Bourbon', 'Arabica', ando f fourF. ^an dF 2 populations from crossesbetwee n these cultivars. In1977 ,yiel d

89 of these coffee populations was low and no significant correlation between disease score and yield was observed. Table 5 shows that 'Mundo Novo' was least diseased and 'Ibaarê' most. The 'Ibaarê' plants showed intense defoliation on the date of observation. The average disease scores of the F. and F, populations were generally intermediate between both parents and no transgressive segregation was observed for any cross.

Table 5. Frequency distribution for coffee leaf rust disease score in the field (0-9 scale) in March 1977 on four cultivars of C. arabica and their F and F. progenies.

Parents Number of plants per class and Mean progenies 0123456789 score = ' Ibaarê' 2 13 8.9 p-1 = ' MundoNov o 8 6 1 3.5 2 — i Bourbon' 3 8 4 4.9 P 3 _ t Arabica' 2 11 2 1 6.1 P4 x P,) 2 6.0 ÏJ (P1 xP p 1 4 5 9 5 3 5 6.3 (P!x P,) 4 7.0 7 5 1' (P1 xP p 2 3 4 5 5 6.4 2 (Pl XP ) 1 1 7.5 xp£) 2 7 8 10 3 7.2 c2 (P1

(P2 x PJ 5 4 2 3.6 4 10 13 4 4.6 r2

90 ofth erespectiv emothe rplants .I nth egreenhous e test,th eparent s differed significantly foral lcomponent so fresistance .L Po f 'Ibaarê'wa s1 8day s shorter thano f 'MundoNovo' ,L Dwa smuc hhigh ­ eran dLR Pabou ttw otime s longer.I nth elaborator y test,signifi ­ cantdifference s were observed forLP ,no tfo rLD .L Po f 'Mundo Novo'wa s4. 5day s longer thano f 'Ibaarê',whic hi sa muc h smaller difference thani nth egreenhous e test.Fo rL Pan dL Dintermediat e valueswer e observed inth eF ,an dF _an dparenta lvalue s were recovered insom eF ,populations .Th elon gLR Po f 'Ibaarê' inherited ina mor e dominantway .Fo rth eF 3 populationsn oLR Pvalue s below 60day swer e observed.Th ecoefficient s ofcorrelatio n (r)betwee n

Table6 .Coffe elea frus tdiseas e scorei nth efiel dan dcomponent so fresistanc e inlaborator yan dgreenhous etest so f'Ibaarê ' (C1133-2-7), 'MundoNovo ' (CMP386-2-1),B C( P) ,B C(P,) ,an do f1 0F „fiel dplant san dthei rF ,progenie s (greenhouse test),inoculate awit hrus trac eI I(v_) .

Field Components ofresistance 2

Cultivaro r Diseasescore 1 Laboratory Greenhouse progeny (0-9 scale) LP LP LD LRP

P.('Ibaarê' ) 8.0 19.3 44 22 80 P,('Mund oNovo' ) 3.3 23.8 62 12 39 5.3 22.1 57 13 71 ?1 5.6 53 17 71 F2 - BC( P) - - 50 17 73 BC(P* ) - - 54 15 69 F -1z 3.7 23.9 62 6 61 F,-2 4.3 23.7 53 16 60 F,-3 4.3 21.6 48 18 67 F2"4 4.3 21.2 53 13 69 22.6 F2"5 4.3 50 23 67 21.2 F2"6 4.7 53 15 69 V7 5.1 22.8 52 11 69 F;-8 8.0 22.4 52 17 71 F,-9 8.0 21.7 45 24 77 F^-IO 8.5 20.3 52 14 84 Averageo ftw oreading smad ei nMarc h197 7an dMarc h197 8 LP latency period, in days, LD= lesion density (number of lesions per leaf), and LRP= leaf retention period, in days, measured as the time between the day that LP is completed and leaf abscission. disease score in the field and the components of resistance were -.65, -0.61, 0.47 and 0.78 for LP in the laboratory, and LP, LD and LRP in the greenhouse, respectively. Only the r value for LD was non-significant at the 5% level. Multiple R, calculated from LP, LD and LRP in the greenhouse was significant (0.84) at P i 0.01. The results indicate that LRP is the component which best explains the differences in disease level in the field. Consistently, 'Ibaarê' showed a much longer LRP than 'Mundo Novo' (see also Table 1).

91 Differences inL P appear tohav e some importance, although results variedbetwee n tests.L Dwa s the leastconsisten t component. Incon­ sistentresult s forL P and LDo f 'MundoNovo ' and 'Ibaarê'hav ebee n reported earlier (Moraes et al., 1976,Scal ie t al., 1974). The significantcorrelatio nbetwee n LRP and disease score inth e field canb e easilyunderstood .A long LRP increases thequantit y of inoculum atan ytim e ofth eyea r and iso fspecia l importance for thecarry-ove r of inoculum during the dry season,whe n leaf fall is highest.

Relative tolerance of 'Ibaarê'. 'Ibaarê'ha s ahig hyiel d potential inCampinas ,eve nhighe r than 'MundoNovo ' (Monaco, 1977). During the first four years ofrus toccurrenc e inCampina s (1971t o 1975), 'Ibaarê' could beclassifie d as 'tolerant'becaus e ityielde d well, notwithstanding itshig h susceptibility.However ,afte r 1975,yield s of 'Ibaarê'hav e dropped drastically inrelatio n to 'MundoNovo' . This canb e illustrated by the results oftw o comparable yieldex ­ periments carried out inCampina sbetwee n 1961 and 1970 (exp.1 )an d between197 1 and 1980 (exp.2) .I nth e absence ofrus t (exp.1) , 'Ibaarê'yielde d 159 and 119%o f 'MundoNovo ' during the first4 and last 6year s ofth eexperiment , respectively. Inth epresenc e of rust (exp.2 )thes e figureswer e 137 and45% ,respectively . The collapse ofth eproductivit y of 'Ibaarê' is apparently duet o the exhaustive effecto fcontinuou s high rustattack s inthi s cultivar.

Segregation for resistance in F, populations from the Agaro C1164-19 x 'Catuai' cross. Progenies from theAgar o C1164accessio n have shownlo w disease levels inth e field, despite ofth epresenc e of the compatible raceX V (v.5 ) inCampina s (Table 7).Th e low levels ofdiseas ema yb epartl y explained by the lowproductivit y ofthes e progenies but could also indicate ahig h level of incompleteresis ­ tance.Grea tvariatio n indiseas e level inth e fieldwa s observed for3 yea rol dF ,plant s ofth eC1164-1 9 x 'Catuai' cross. In1978 , 70F _population s were testedwit hrac eX V inth e laboratory,to ­ getherwit hbot hparents ,th e F„ and 'Ibaarê'.Figur e 1show s that transgressive segregation for latencyperio d (LP)toward s higher resistance andhighe r susceptibility occurred. Inth e sameyear , 37o fthes e 70F ,population swer e tested inth egreenhouse ,usin g 12 seedlings perpopulation .Transgressiv e segregationwa s again observed forLP ,an d also for sporulating lesion density (SLD). The coefficient ofcorrelatio n between LP observed inth e laboratory and inth e greenhouse was 0.80, significant atP i 0.01.

92 Table7 .Coffe elea frus tdiseas escor ei nth efiel dan dyield ,fro m197 2t o 1982, oftw oline so fth eAgar o introduction,tw oF 1 hybridsbetwee nAgar oan d Braziliancultivar san do fa susceptibl e controllin eo fC .arabica .

Lineo r Resistance Disease Yield hybrid genotype score1 (kgfres h (0-9 scale) berries)

AgaroLC1164-1 9 S„4S„4 3.1a2 3.6 11.3 AgaroLC1164- 4 SH4SH4 3-4h H7480 (C1164-19x 'MundoNovo' ) S 4s 4 4 8 20.4 H H - hb H7317 (C1164-19x 'Catuai') S 4S 4 4.9 20.3 H H C ControlH3437- 5 sH4sH4 8.0 17.0

1 Averageo ftw oobservation smad ei nJul y197 9an dJul y198 0 2 Differentletter s indicate significantdifference saccordin gt oth eTucke y testa tP i0.05 .

Figure1 .Var i riationfo rlatenc yperio d (indays )i nlea fdisk so f7 0F, . popula­ tions IrOrn thet enganjAgar oC1164-1 VJ±J.V+— J.97x " 'Catuai Kjai^ixai.'cross t-Luaa,o f,bot ui hparentsuuui i jfaituw,th eF» ) ,i-man _d io ~f, 'Ibaarê',inoculate dwit hrus trac eX V(v , ,). Abscissa:frequenc yo fF „popula ­ tions. '

cvlBAARi E P1.P2.F2r— . 15

o 10

3 L O LU CC 5 r \ : _r U O h m m m •M.*—.1 *--,,*• „I—A-, 20 25 30 35 40 45 LATENCY PERIOD

1 TheC1164-1 9paren tha sbee nrepresente db yit ssiste rlin eC1164-4 .

Teno fth emor e resistantF 3 populations were retested inth e greenhouse in1980 .I n1978 ,nin e ofthes epopulation s had showna significantly longer LP than 'Catuai' and 6a significantl y shorter SLD. In1980 ,th e differences with 'Catuai' for LP and SLDwer e much smaller,bein g significant onlywit h twoF 3populations .Th ecoeffi ­ cients of correlation between the two testswer e significant forSL D (0.69)bu t insignificant forL P (0.54). The analysis ofvarianc e showed interactionbetwee n tests and genotypes forL P (P< 0.01), butno t for SLD (P= 0.08). Therefore, the resistance ofth epopula ­ tions is,t o acertai n degree, affected by the testconditions .

93 Two other aspects ofresistanc e inth eF ,population s wereob ­ served.Man yF _population s showed aslightl y higherpercentag e of non-sporulating lesions (flecks,tumefactions )tha n 'Catuai'.Thi s percentage varied from 5t o 27i nth e 1980 test,bein g 7%fo r 'Catuai'.Also ,earl ynecrosi s of apparently susceptible-type le­ sionswa s observedwit h a fewpopulations ,reachin g amaximu m of31 % ofth e totalnumbe r of lesionswit h onepopulatio n inth e 1980test . Early necrosis oflesion s has alsobee n observed with othergeno ­ types fromth eAgar o accession (Eskes,198 3 ).Earl y necrosis ap­ peared tob e ahighl y unstable characteristic,bein gpronounce d in sometest s andnon-apparen t inothers . Themos tpromisin gF _population s havebee n transplanted toth e field for further selection and evaluation ofresistance . Itmigh t bepossibl e to findgenotype s inthes eF _population swit h high levels ofincomplet e resistance andgoo d yielding capacity.

GENERALBREEDIN G ASPECTS

Screening in the laboratory or greenhouse. Inth e laboratory and greenhouse tests significantvariatio n for latencyperio d(LP) , lesion density (LD)an d leafretentio nperio d (LRP)wa s observed in C. arabica. The coefficient ofvariatio n (C.V.)wa s relatively low forLP ,hig h forLR P and extremely high forLD . Interactions between genotypes and testswer e observed forL P and LD.Furthermore ,L P and LDwer eno talway s correlated with disease level inth e field. Therefore,selectio n fora lon gL P or alo wL D inth e laboratory or greenhouse mayb e of lowefficienc y inobtainin g lower levels of disease inth e field. Incomplete resistance tocoffe e leafrus t is affectedb y leafag e and light intensity (Eskes,198 2 ,Eske s andToma-Braghini , 1983). Also soilhumidit y may affectresistanc e (J.CG. Hoogstraten,per ­ sonal communication).Therefore , in screening forincomplet eresis ­ tance the test conditions before and after inoculation shouldb e as uniform aspossibl e for allplants . Heritabilities forL P andLD ,calculate d from thevariance s in parental andF ,population s ofth ecrosse sbetwee n 'Ibaarê' and 'MundoNovo ' andbetwee n 'Catuai' and 'AgaroC1164-19 'wer eextreme ­ ly low (0.41 and 0.26 forLP ,an d 0.00 and 0.24 forLD ,wit h the two crosses respectively).Hence , individual plant selection forL Po r LD in segregatingpopulation s inth egreenhous e isno texpecte d to be successful. Screening inth e greenhouse might onlyhav e somesuc ­ cess ifman y replications areuse dpe r genotype (clones orprogenie s

94 obtainedb y seifing). Iti sno tpossibl e to indicate one component ofresistanc e asth e most suitable selection criterion for all C. arabica populations. The importance ofeac h componentvarie d according to thepopulation . LRP appeared tob e important inth eprogenie s fromth e 'Ibaarê'x 'MundoNovo ' cross,wit h LP and LD successively asnex t inimpor ­ tance. Inth eprogenie s from theAgar o C1164-19 x 'Catuai' cross,L P and SLDwer e important components,bu t also earlynecrosi s ofle ­ sionsma yhav e importance. Ingeneral ,L P and LDwer ewel l correlated. With theprogenie s from 'Ibaarê'x 'MundoNovo ' (Table 6)th e coefficient ofcorrela ­

tionbetwee n LP and LDwa s -0.82. Withth eF 3 populations fromth e Agaro C1164-19 x 'Catuai' cross thisvalu ewa s -0.85. Since C.V. valueswer e lower forL Ptha n forLD ,i ti srecommendabl e tous eL P rather thanL D asa selectio n criterion. IfLR P ist ob euse d as aselectio n criterion,genotype s with eitherver y long orver y shortLRP' s shouldb e avoided.A longLR P mayb e related tohig h disease levels inth e field (Table 6). Short LRP'smigh tb e related to lowproductivity , as indicated by studies ofva n derVosse n and Browning (1978). These authors studied the effecto fethylen e on leaf abscission inC . arabica. They concluded thatth e ability to retain leaves under stress conditions mayb e vital for successful commercial cultivation ofC . arabica inKenya . Rust-induced leaf abscissionma yb e similar to ethylene-induced leaf abscission,hence ,shor tLRP' s inducedb y coffee leafrus tinfec ­ tionsmigh trelat e to alo wproductio n capacity ofth e coffeegeno ­ type.

Screening in the field. Heritabilities fordiseas e level inth e fieldwer e low fordifferen t lines from 'Catuai' and 'MundoNovo' . Higher heritabilities were found inth eF ,population s from crosses between 'Ibaarê' and other Brazilian coffee cultivars (Table5) , which isdu e toth e greater geneticvariatio n fordiseas e level in theseprogenies .Thi s indicates that individual tree selection for low disease level inth e field canb e successful,provide d thata certain amounto fgeneti cvariatio n ispresent . Disease level inth e fieldwa spositivel y correlatedwit h yield (Table 3). Therefore,whe n screening for lowdiseas e levels inth e field, theyieldin g capacity shouldb e taken into account.Fiel d observations shouldb e carried outdurin gmor e thanon eyea rt o account fordifference s inbearin gpatter nbetwee nth ecoffe etrees .

95 The relationship between disease level andyiel d inth e sameyea r isa know n feature ofrus tepidemic s inBrazi l (Monaco, 1977). In years ofhig hproductivit y thediseas e level isgenerall y higher than inyear s of lowproductivity . Iti s arelate dbu tne w aspect thatth eproductio n capacity of acoffe e linema y also significantly influence thediseas e level (Tables 3 and 4).However ,th ecoeffi ­ cients ofcorrelatio n areno tver y high and, therefore, selection of productive cultivars with increased levels ofresistanc e mightb e possible.Whethe r suchcultivar s canb e selectedwil l probably dependmor e onth egeneti c variation for incomplete resistance pres­ ent inth epopulatio n thano nth e relationship betweenyiel d and diseaselevel .

Conclusions. Breeding for increased levels of incomplete resistance tocoffe e leafrus tma yb e important forobtainin gdurabl eresis ­ tance. Thepresen tresearc h indicates someperspective s forselec ­ tion in C. arabica. Inprogenie s fromvarieta l crosses,suc ha s Agaro C1164-19 x 'Catuai',genotype sma yb e selectedwit h increased levels ofresistanc e and goodproductio n capacity. For selection success onth e long term, itseem s desirable to create avariabl e populationwit h goodproductio n capacity andwithou t thepresenc e of major genes.Th ecrosse s realized between the accessions fromEthi ­ opia and 'Catuai' are a first step inthi sdirection . Thepresen t research also indicates factorswhic hma y hamper selection success:a )th epositiv e correlationbetwee nyiel d and disease level inth e field,b )th e relatively lowvariatio n for incomplete resistance among C. arabica cultivars andproductiv e breeding lines,an d c)inconsistenc y ofresult sbetwee n resistance tests.

ACKNOWLEDGEMENTS

The researchwa s supported by the Food anAgricultur e Organization ofth eUnite d Nations (F.A.O.), the InstitutoAgronômic oo f Campinas (I.A.C.), S.P., Brazil and theAgricultura l University ofWagenin ­ gen, TheNetherlands .Th e authorswis h tothan k all staffmember s andpersonne l ofth eGenetic s Departmento f I.A.C. forth e kind cooperation received. Special thanks got oDrs .R.A .Robinson , L. Chiarappa, andN.A . van derGraaf f ofF.A.O . forth e stimulus received. Thecriticall y reviewing ofth e textb y professors J. Sneep and J.E.Parlevlie t is acknowledged.Mrs .Masak oToma - Braghini andMr .Jaa pHoogstrate n arethanke d forthei rparticipa -

96 tioni nth eexperimenta lwork .

REFERENCES

Bettencourt,A.J .& A .Carvalho ,1968 .Melhorament ovisand oa resistênci ad o cafeeiroà ferrugem .Braganti a 27:35-68 . Carvalho,A. ,1967 .'Espécie se variedades' .In :Manua ld ocafeicultor .Editor a daUniversidad e deSä oPaulo ,p .17-40 . Carvalho,A .& L.C .Monaco ,1972 .Transferênci ad ofato rCaturr apar ao cultiva r MundoNov od eCoffe aarabica .Bragenti a 31:379-399 . Eskes,A.B. ,198 2 .Th eus eo flea fdis kinoculation s inassessin gresistanc et o coffeelea frus t (Hemileiavastatrix) .Neth .J .PI .Path .88 :127-141 . Eskes,A.B. ,198 2 .Th eeffec to fligh tintensit yo nincomplet eresistanc eo f coffeet oHemilei avastatrix .Neth .J . PI.Path .88 :191-202 . Eskes,A.B. ,1983 .Qualitativ ean dquantitativ evariatio n inpathogenicit yo f raceso fcoffe elea frus t (Hemileiavastatrix )detecte d inth eStat eo fSä o Paulo,Brazil .Neth .J . PI.Path .89 ,i npress . Eskes,A.B .& M .Toma-Braghini ,1981 .Assessmen tmethod s forresistanc e tocoffe e leafrus t (Hemileiavastatri xBerk ,e tBr.) . PI.Prot .Bull .F.A.O . 29:56-66 . Eskes,A.B .& M .Toma-Braghini ,1982 .Th eeffec to flea fag eo nincomplet eresis ­ tanceo fcoffe et oHemilei avastatrix .Neth .J . PI.Path .88 :215-226 . Figueredo,P. ,A.P .d aSilveira ,P.R .Mariott o& C .Gerald oJr. ,1974 .Flutuaçâ o daferruge m docafeeir on oEstad od eSä oPaulo .In :Resumo s 2°Congress o Brasileirod ePesquisa sCafeeiros ,Poço sd eCaldas ,1974 .p .101-102 . Graaff,N.A .va nder ,1981 .Selectio no fArabic a coffeetype sresistan tt oCoffe e BerryDiseas e inEthiopia .Thesis .Communication s81-1 1o fth eAgricultura l Universityo fWageningen ,Th eNetherlands .Il lpp . Meyer,F.G. ,L.N .Fernie ,R.L .Narasimhaswamy ,L.C .Monac o& D.J . Greathead, 1968.F.A.O .Coffe eMissio nt oEthiopia .Foo dan dAgricultur eOrganizatio no f theUnite dNations ,Rome ,20 0pp . Monaco,L.C , 1977.Consequence s ofth eintroductio no fcoffe elea frus tint o Brazil.Ann .Ne wYor kAc .Sei .287 :57-71 . Moraes,S.A . de,M.H .Sugimor i& I.J.A .Ribeiro .Estud o sobreo desenvolviment o dasraça sII ,II Ie X Vd eHemilei avastatri xBerk .& Br .e mdiferente s cafeeiros suscetiveis.Ecossistem a 1:20-24 . Parlevliet,J.E .& J.C .Zadoks ,1977 .Th eintegrate d concepto fdiseas eresis ­ tance;a ne wvie w includinghorizonta lan dvertica l resistance inplants . Euphytica 26:5-21 . Ribeiro,I.J.A. ,M.H .Sugimori ,S.A .Morae s& L.C .Monaco ,1975 .Raça sfisiologi - casd eHemilei avastatri xBerk ,e tBr .n oEstad od eSä oPaulo .Summ aPhyto - pathologica 1:19-22 . Ribeiro,I.J.A. ,A .Bergami nFilh o& P.C.T .Carvalho ,1981 .Avaliacä od aresis ­ tênciahorizonta la Hemilei avastatri xBerk ,e tBr .e mcultivare sd e Coffeaarabic aL .e mcondicöe snaturai sd eepidemia .Summ aPhytopathologic a7 : 80-95. Scali,M.H. ,A .Carvalh o& L.C .Monaco ,1974 .Resistênci a horizontalà ferruge m (Hemileiavastatrix )e mcultivare sd eCoffea .In :Abstract so fXXV IAnnua l Meetingo fth eBrazilia nSociet yfo rth eProgres so fScience ,Jul y1974 , p.241 . Sylvain,P.G. ,1955 .Som eobservation s onCoffe aarabic aL .i nEthiopia . Turrialba5 :37-54 . Vanderplank,J.E. ,1968 .Diseas eResistanc e inPlants .Academi cPress .Ne wYork , London.20 6pp . Vossen,H.A.M ,va nde r& G .Browning ,1978 .Prospect so fselectin ggenotype so f Coffeaarabic aL .whic hd ono trequir etoni cspray so ffungicid e forincrease d leafretentio nan dyield .Keny aCoffe e43 :361-368 .

97 7 CHARACTERIZATION OF INCOMPLETE RESISTANCE TO HEMILEIA VASTATRIX IN COFFEA CANEPHORA CV. KOUILLOU

A.B. Eskes1

InstitutoAgronômic o ofCampinas ,S.P. ,Brazi l

INDEXWORD S Coffee LeafRust ,component s ofresistance ,heterogeneou s reaction type, durableresistance .

SUMMARY

Incomplete resistance to coffee leafrus t (Hemileia vastatrix) varied amonggenotype s of Coffea canephora cv.Kouillo u from near 'immunity't o ahighe r susceptibility thantha t of C. arabica cv. MundoNovo . In4 series oflaborator y tests resistance components andreactio n type (RT)wer e observed. Lowdiseas e levels inth e fieldwer emainl y associated with low,o rheterogeneous ,R T and/or a low lesiondensit y (LD).Amon g thecomponent s ofresistanc e sporu- lating lesion density (SLD)showe d thehighes t coefficiento fcorre ­ lationwit hdiseas e level inth e field (r= 0.83). Components of resistancewer e intercorrelated andwer e alsocorrelate dwit hRT . Interactionsbetwee n inoculation series and genotypes weresignifi ­ cant foral lparameter s ofresistance ,bu tespeciall y so forLD . A smallpar to fth eKouillo upopulatio n showed completeresis ­ tance torac e IIo f H. vastatrix. Two rustgenotype s isolated from 'Kouillou' interacted with somegenotype swit h complete or incom­ plete resistance.Th ecomplet e andrace-specifi c resistance ofon e genotypewa s governedb y adominan tgene . High levels ofincomplet e resistance appeared to inherit either monogenicallyo r ina mor e complexway .Fiv e genotypes with high levels ofincomplet e resistance,expresse d by alo wL D and arela ­ tivelyhig hRT ,wer e crossedwit h C. arabica cv. Catuai.Th etri - ploidF ,plant s showed avariabl ebu tsusceptibl eRT .Thes e crosses may havevalu e forobtainin g coffee cultivars with durableresis ­ tance tocoffe e leafrust .

1 Present address:Ru a Salesopolis 194,Alt o daBarra , 13100 Campinas, S.P., Brazil

98 INTRODUCTION

Durable resistance tocoffe e leafrus t (Hemileia vastatrix Berk,e t Br.)i sconsidere d as amai n objective inperennia l coffee crop breeding. 'Kouillou1 is acultiva r of Coffea canephora Pierre,whic h is adiploi d allogamous species. 'Kouillou1 isresponsibl e for50 % ofth e coffeeproductio n inth e State ofEspirit o Santo,Brazil . Individual genotypes of 'Kouillou' showwid evariatio n forresis ­ tance tocoffe e leafrust . Inth e State ofEspirit o Santo alltree s show some susceptibility to H. vastatrix butonl y aver y smallpar t ofth epopulatio nbecome s highly diseased. Chemical control ofcof ­ fee leafrus t in 'Kouillou' isreportedl y ineffective (Paulino, 1981). The incomplete resistance of 'Kouillou'migh tprovid e amor e durable type ofresistanc e thani sprovide d by complete ormajo r generesistance . Cadena G. andBuritic â C. (1980)studie d the epidemic development ofcoffe e leaf rustdurin g oneyea r ina populatio n of 'Kouillou' in Campinas,Brazil .Diseas e levels and leaf fall duet o the disease weremuc h lower in 'Kouillou'tha n in 'MundoNovo ' (C. arabica L.). Ingreenhous e tests,Scal ie t al. (1974)foun d alowe rnumbe r of lesionspe r leaf for 'Kouillou'tha n forC . arabica cvsCatua i and Ibaarê. Cadena G. (1978)detecte d significant differences inresis ­ tance components between 'Kouillou'genotype s tested inth e nursery and laboratory. Eskes (1982a)considere d alaborator y test,usin g coffee leafdisks ,suitabl e forassessin g resistance tocoffe e leaf rust.Th epresen t study further characterizes resistance of 'Kouil­ lou' inth e field and laboratory. Segregation forresistanc ewa s studied inF , populations inth enursery .

MATERIALS AND METHODS

'Incomplete'resistanc e isdefine d here asa form of resistance which doesno tcompletel y inhibitreproductio n ofth epathogen . Incomplete resistance mayb e governed bymino r genes,bu t alsob y major genes (Parlevliet, 1979). The genotypes of 'Kouillou'use d inthi s studybelon g to apopu ­ lation of 64trees ,raise d from seed introduced in193 3 fromCastel - lo,Espirit o Santo.Thi spopulatio n consists of 5families ,name d C66 toC70 .Tree s ofeac h family areprobabl y half-sibs.The y are part ofth ecoffe e collection ofth e InstituteAgronômic o of Campinas (I.A.C.). One 'unidentified' genotype (Table 2)refer s to anunname d 'Kouillou'tre epresen t inth e same collection. Sixtree s

99 of C. arabica cv.Mund oNov oplante d inth e same lota sth e 'Kouil- lou' genotypes havebee nuse d as acontrol . Unless stated otherwise, race IIo fS . vastatrix hasbee nused . Race II isth emos tcommo n race inBrazil . Inoculationmethod s and methods used formeasurin g resistance components are asdescribe d by Eskes (1982a and 1982b). Four laboratory tests,wit h leafdisk s of 23 'Kouillou1 genotypes,wer e carried out inDecembe r 1978,an d in March,Ma y and October 1979.Pe r genotype 80 to 120 leafdisk s from 40 randomly chosenmatur e leaves,distribute d over4 to 5replica ­ tions,wer euse d foreac htes t (series). Resistance componentsob ­ served are:latenc yperio d (LP= numbe r ofday s from inoculation to sporulation of50 %o fth e totalnumbe r of sporulating disks), lesiondensit y (LD= % ofdisk swit hvisibl e lesions), sporulating lesiondensit y (SLD= % ofdisk swit h sporulating lesions)an drela ­ tive sporulating lesion efficiency (RSLE= 100.SLD/LD) . A field testwa s carried out inMarc h 1981.I nthi s test,L Drefer s to the number ofvisibl e lesionspe r 10cm 2 leaf surface.Reactio n type (RT)wa s scoredb y a0 t o 9scal e (Eskes andToma-Braghini , 1981). Values 0t o3 indicate aresistan t RT,4 and 5a moderatel yresis ­ tantRT ,6 an d 7a moderatel y susceptible RT,an d 8an d 9a sus­ ceptible RT.RT' s4 to 7ar e generally heterogeneous,wit hresis ­ tant-type lesions and sporulating lesions occurring together. The disease level inth e fieldwa s scoredb y a1 t o 5scal e (disease score,DS) .Valu e 1indicate s absence ofsporulatin gle ­ sions,value s 2t o 5increasin g numbers of sporulating lesionsasso ­ ciated with increasing sporulating intensity ofth e lesions.Th e percentage of leaveswit h sporulating ornecroti c lesions hasbee n assessedb y sampling 200 leavespe rgenotype . For statistical elaboration ofth e results computerprogram s from the Statistical Package forth e Social Sciences (SPSS)hav ebee n used.Percentag e datahav ebee n transformed into the arcsine ofth e square rooto fth epercentages .

RESULTS

Field observations. Disease scores (natural infection)wer emad e 11 timesbetwee n 1976 and 1981 on 64 'Kouillou'genotypes .A continuous variation from complete resistance tohig h susceptibility wasob ­ served.Mos ttree s gave intermediate scores and afe wwer emor e severely affected than C. arabica cv.Mund oNovo .Th evarianc e for disease scoreswa shighes t forgenotype s with intermediatescores . The 95%confidenc e interval for averagediseas e scoresbetwee n 2.5

100 and3. 9wa s about0. 5scal e units. The epidemic development ofcoffe e leafrus t in 'Kouillou1 dif­ fered from thati nC . arabica. For the 'Kouillou'populatio n asa whole, epidemics started quite early inth e season,bu t subsequent developmentwa s slowertha n in C. arabica cv.Mund oNov o (Table1) . In 'Kouillou'th eoccurrenc e ofnecroti c lesions,beside ssporulat - ing ones,wa s commonly observed inth e field (Tables 1an d2) . Necrosis startedbefor e or after sporulation, causing small and largenecroti c lesions,respectively . Somegenotype s showeda remarkable increase inth e% ofnecroti c lesions during theseason , indicating thatthei r resistance was changing.Necrosi s of lesions in 'MundoNovo 'wa s only observed under extremely dryo rho tweathe r conditions. Yield affected slightly the level ofdiseas e ofKouillo ugeno ­ typeswit h incomplete resistance.A treewit hnormall y lowdiseas e levelswoul d becomemoderatel y diseased inyear s ofheav ybearing . However,th e coefficient ofcorrelatio nbetwee n average yield and disease scorewa s insignificant (r= -0.21).

Table 1.Averag epercentag eo fleave swit hsporulatin g rustlesion san dpercen ­ tageo fnecroti c lesionso f1 7tree so fC .canephor acv .Kouillo uan do f6 tree s ofC .arabic a cv.Mund oNovo , observe d inMarc han d May 1979an d1981 .

Leaveswit h sporulating Necrotic lesions(% ) Cultivars lesions

1979 1981 1981

March May March May March May

Kouillou 17 30 17 15 29 53 MundoNov o 20 57 11 39 0 0

Components of resistance and reaction type. Table2 show sresult s of field observations,o f laboratory tests (average of4 tests)an d ofa field test.Tw o genotypes showed complete resistance;C68-1 1 didno t show any disease symptoms and C67-3 showed only ahypersen ­ sitive type of lesions.Hig h levels ofincomplet e resistance were related to alow ,o rheterogeneous ,reactio n type (RT)and/o rt oa low lesion density (LD). Thedegre e ofheterogeneit y ofth ereac ­ tion isexpresse d by the relative sporulating lesion efficiency (RSLE).Tw o 'Kouillou'genotype s showed higher susceptibility than 'MundoNovo' .Th e 'unidentified' genotypewa s somuc hdisease d in the field thatthi s tree didno tproduc e any fruitdurin g theobser ­ vationperiod .

101 co ^ N o m -

I

O0M^o<ïvûMo^miNcriû0cn^

e > o o s& Ä o o CJ -o d o es X) 3 «a - s oocnNuirooocON^ocoioo •H l-J

3 W

00 r-( CT> 00 I-H a\ 0) (3 UI > i-H C CD 3 O ti \Û «3 O 4-» ^ Ci i—i eu -Hm« t/l 0e) H N 00 •P na y-i o 3 (X CT> CT* CJ W a «1 i-i fc U ^—- w o> 4-) s o d o 4-t -H *H •P >» o I fflaj r-t rH r—i 0> 1) CJ Î-I -H « O •-H "O T3 "O wo ca >,n d i-t 1U e s s ( CO (0 CJ CU >H o T3H T3 00 p, « Cfl M W aj d S OOy3SODOHMmHvCO^»ÛO d Ö G d d o e -H .»-> o o o •H -H A3 O P •H -H -H M-l -i-I « d +J -P +J * •- $-t Ö1H O •o >. 3 11 d 'H fl fl ffl ' OP WH *H P t> > > •i-t -i-t O CJ M M ^ • U W 4-1 00 ft « 0) CU CJ aj Ö « d w % «5 > CJ rH-^CS CJ dCJ i—t «H 00 rH d O E 3 -P « CJ -H O^O^ÛO>COOÛOCTiN'- OJ CJ OJ 4eJ -H \£> i >o vo \o r*- \o r^^^Dvûvor^vo^û p- > > P_ Q 9 W H O 3 U < < < t-3 I-J W 0=3 (13 e O Ui M o H CN « S"

102 The field test showed lower levels ofdiseas e ofmos t 'Kouillou' genotypes, incompariso n to 'MundoNovo' ,tha nexpecte d from thela ­ boratory tests or fromth e average disease scores inth e field.Th e dry andho tweathe rprevailin g during the field testma yhav ein ­ duced the relatively high resistance of 'Kouillou'. Rankcorrelatio n coefficients (r )wer e calculated betweenresis - s tance parameters of 20 'Kouillou' genotypes, with varying degrees of incomplete resistance (Table 3). In the laboratory tests, the sporu- lating lesion density (SLD) was best correlated to the disease score in the field (r = 0.83). The components of resistance were corre­ lated with each other. SLD was highly correlated with LD and RSLE. LD and RSLE were less correlated to each other. This shows that SLD is based on lesion number and reaction type. RSLE was also highly correlated with latency period (LP), indicating a relationship between LP and reaction type. The % of necrotic lesions in the field (%NL) was related to RSLE, a relationship which can be expected since both parameters are an indication for reaction type. Table 3. Rank correlation coefficients of Spearman (r ) between components of resistance observed in the laboratory and disease score (DS, 1-5 scale) and percentage of necrotic lesions (%NL) in the field of 20 plants of 'Kouillou', with varying degrees of incomplete resistance to coffee leaf rust. Significance of r s is + 0.44 at P = 0.05.

Componentso fresistance 1 Fieldobservation s

LD SLD RSLE DS %NL

LP -.49 -.70 -.74 -.68 .67 LD .89 .54 .80 -.17 SID .81 .83 -.38 RSLE .69 -.74

DS -.46 1 Forexplanatio no fabbreviation sse eTabl e3 Inth e laboratory testo fDecembe r 1978,sporulatio n of1 3 'Kouillou'genotype swa smeasure d by counting thenumbe r ofure - diospores produced per diskwit h sporulating lesions.Th e counts weremad e 10day s after the daytha t 50%o fth e disks sporulated. Between genotypes,spor eproductio nvarie d from 0.09 to 2.53 x10 4 per disk.A significant correlationwa s foundwit hRSL E (r= 0.81), indicating that sporeproductio n isnarrowl y related to reaction type.

103 Interaction between inoculation series and genotypes. Thefou r series oflea fdis k inoculationswer e carried outwit h4 o r 5repli ­ cations each.Th epoole d analysis ofvarianc e detected significant effects ofgenotypes , series ando f interactionbetwee n series and genotypes foral lresistanc e components ata leve l of significance lowertha n 0.005. The relativevarianc e duet o series (K|), geno­ 2 types (K ,), interactionbetwee n series and genotypes (K|x„)wa s estimated according to afixe dmode l analysis (Snedecor andCochran , 1967). Thevarianc e ratio ofK 2„/K2 was highest forLD ,interme ­ diate forSL D and relatively low forL P andRSL E (Table 4). There­ fore, LD appears tob e ales s reliable characteristic of resistance than isL P orRSLE .Hence ,selectio n fora lo w lesion density might be less successful than selection for alo w reactiontype .

Table4 .Estimatio no fth epercentag eo fvarianc edu et oserie s (K|),genotype s (ÏK),interactio nbetwee nserie san dgenotype s(K 2„ )an dexperimenta lerro r(a 2) calculatedfro m4 serie so flea fdis ktest sfo r2 Ö Kouillou'genotype swit h incompleteresistanc et ocoffe elea frust .

Variance component Componento f resistance1 orrati o LP ID SLD RSLE K I 9 52 38 6 "Ê 40 14 29 48 9 15 7 KSXG 15 a2 42 19 18 40 e v2 iv2 0.2 3 1.07 0.52 0.15 SXG'G Forexplanatio no fabbreviation sse eTabl e3 Specificity of resistance. Table 5show s representative results of alaborator y testwit h2 9 'Kouillou'genotype s inoculated with four rust isolates.Mor e detailed results ofthi s testhav ebee n published elsewhere (Eskes, 1983). Three 'Kouillou'genotype s with complete resistance torac e II (e.g.C68-11 , Table 5)wer e suscep­ tible to isolate 10 (Is.10) .Severa l genotypes,wit hvaryin g degrees of incomplete resistance torac e II (e.g.C67-1 , C69-7 and C69-15), weremor e susceptible to isolate 11 (Is.11 )tha nt o race II.Othe r genotypes,som e ofwhic hhav e lowdiseas e levels in the field (e.g.C69-1 4an d C70-14), didno t showhighe r susceptibil­ itywit hth e 'Kouillou' isolates thanwit h raceII .

104 Table 5. Average speculating lesion density (SLD = %o f leaf disks with sporulat- ing lesions) and reaction types (RT, 0 to 9 scale) of six 'Kouillou' genotypes and 'Mundo Novo' inoculated with coffee leaf rust race II, isolate 11 (Is. 11) and isolate 10 (Is. 10). Is. 10 and Is. 11 were obtained from 'Kouillou' C66-13 and C67-1, respectively.

SLD RT Coffee genotype RaceI I Is. 11 I s.1 0 RaceI I Is.1 1 Is.1 0

'Kouillou': a C68-11 o 80b a 0a 7.7b a ° hb a C67-1 62° 0 5.0 7-°h 2.0 C69-7 78e 22a 5.3a 6.7b 4.0a C69-15 60b 91e 13a 6.0b 6.7b 4.0a C69-14 71b 27a 24a 6.3a 6.0a 5.3a C70-14 24a 29a 22a 6.3a 6.3a 5.3a

'MundoNovo ' 62a 65a 59a 8.9a 8.8a 8.8a 1 Different letters indicate significant differences, between rust races within genotypes at LSD 0.05. Furthermore, Table 5 shows that some 'Kouillou' genotypes were more resistant to Is. 10 than to race II (e.g. C67-1, C69-7 and C69-15). This may indicate that Is. 10 and race II differ for more than one virulence gene or that interaction occurs between virulence genes in the rust genotype (Eskes, 1983). Segregation for resistance. Crosses between 'Kouillou' genotypes with different levels of resistance were made by controlled hand pollinations. The evaluation of resistance of two-year-old F, plants was done in a nursery test. Scoring was done by a 1 to 9 scale which expresses variation in reaction type and lesion density. Table 6 shows the resistance scores of plants from 9 F, populations. The populations have been grouped together when the following common parents have been used: C66-1, C68-11 and C69-9. The resistance of C66-1 is high but incomplete (Table 2). The bimodal segregation in the F,'s with C66-1 as a parent suggests mono- or oligogenic inheritance, although the results are not con­ clusive due to the low number of F, plants scored. The segregation in the F,'s with C68-11 as a parent indicate the presence of a domi­ nant major gene, in heterozygous condition, controlling complete resistance. This resistance is race-specific, it is overcome by iso­ late 10 (Table 5). Furthermore, continuous variation for incomplete resistance was observed in the F,'s from C68-11 suggesting also the presence of minor genes for resistance. The incomplete resistance of C70-11 and C70-14 appeared to be based on minor genes too, as con­ tinuous variation was observed in the F, populations derived from these genotypes. A few plants with complete resistance (classes 1, 2

105 Table 6. F plants of nine crosses between 'Kouillou' genotypes classified over 9 classes for coffee leaf rust resistance.

Crosso r Diseasescor e Numb sro f lants cultivar (1-5 scale) V ofparent si n Resistancec l ass1 the f leid 1 2 3 4 5 6 7 8 9

'Kouillou' C66-1x C67- 6 1.6 X 4.5 2 5 1 1 C66-1x C70-1 3 1.6 X 4.2 2 1 1 1 3 2

Total 2 3 6 1 4 3

C68-11x C67- 6 1.2 X 4.5 4 15 2 3 4 5 2 2 C68-11x C69- 9 1.2 X 3.5 3 14 1 4 7 4 4 1

Total 7 29 3 7 11 9 6 3

C70-11x C67- 6 1.8 X 4.5 1 1 5 3 5 9 3 C70-11x C69- 9 1.8 X 3.5 1 3 6 4 3

Total 1 1 1 8 9 9 12 3

C67-6x C69- 9 4.5 X 3.5 2 8 19 5 C68-4x C69- 9 3.1 X 3.5 1 2 3 5 16 7 1 C69-9x C70-1 4 3.5 X 2.1 2 9 7 8 8 4 1

'MundoNovo ' 4. 4 4 10

1 Value 1 indicates absence of disease symptômes, values 2 and 3 indicate pres­ ence of small flecks and large chlorotic areas without sporulation, respectively, and values 4 to 9 indicate increasing numbers of sporulating lesions per leaf area unit. Values 4 and 5 were often heterogeneous in reaction, showing sporu­ lating and non-sporulating lesions simultaneously. and 3, Table 6) were observed in the F, 's of these parents. These are possibly due to a very high level of resistance to infection and to an occasional outcross. The average level of resistance of the F, populations was fairly well related to the average disease scores of the parents in the field. This is clearly so with F,'s which have C69-9 as a common parent (Table 6).

Crosses between 'Kouillou' and 'Catuai'. With the aim of introduc­ ing minor genes for resistance to coffee leaf rust from 'Kouillou' into C. arabica cultivars, crosses were made in 1980 between 5 'Kouillou' genotypes, used as male parents, and 'Yellow Catuai'. The 'Kouillou' genotypes (C69-14, C70-9, C70-11, C70-12 and C70-14) were selected on the following criteria: 1) low levels of disease in the field despite of a quite susceptible reaction type, 2) absence

106 of interactionwit hth e rustisolate s shown inTabl e 6, 3)apparent ­ lycomple x inheritance ofresistance , and4 )fairl y good yielding capacity. C70-9 andC70-1 2wer e themos tresistan tgenotype s inth e field, showing only sporadically sporulating lesionsbu tn ohyper ­ sensitive typeo f lesions.Abou t25 0F. .seedling swer e raised in thenurser y in1981/1982 ,al l showing atriploi d appearance.Screen ­ ingo fth etriploid swa s done in198 2b y scoring reaction type and number oflesion spe rplan t resulting from natural infection. All triploids from C69-14, C70-11 and C70-14 showed susceptible reactiontypes ,bu tth enumbe r oflesion svarie d betweenplants . About80 %o fth e triploids from C70-9 and C70-12 also showed suscep­ tible reaction types.Th eres t showed arang e of intermediate,hete ­ rogeneous, reaction types and greatvariatio n forth enumbe r ofle ­ sionspe rplant .N oplant swer e foundwit h complete resistance.Thi s type of segregationwoul d exclude thepresenc e ofdominan tmajo rre ­ sistance genes inthes epopulations .

DISCUSSION

Some 'Kouillou'genotype s showed complete resistance toth e rust population ofCampinas .However , intht =Stat e ofEspirit o Santon o rust-free trees of 'Kouillou'ca nb e found (Paulino, 1981).There ­ fore, rust racesma yb epresen t inth e State ofEspirit o Santowhic h dono toccu r inCampinas . The susceptibility ofmos t 'Kouillou'genotype s toth e common race IIo f H. vastatrix hasbee n taken as anindicatio n thatit s resistance mightb e 'horizontal resistance' (CadenaG . and Buritica C, 1980). However,th epresen t results show thatrace-specifi c re­ sistance, atvariou s degrees,ca nb e found in 'Kouillou' (Table5) . Some genotypes appeared tohav emonogeni c dominant resistance (Table 6). Therefore,certainl y not all resistance in 'Kouillou' is durable resistance.Grea tcar e shouldb e taken inth e selectiono f 'Kouillou' genotypes tob e used inbreedin gprogramme s forresistance . Polygenic incomplete resistance toplan tpathogen s is considered tob emor e durable thanmonogeni c resistance ofth e hypersensitive type (e.g.Vanderplank , 1968,Simons ,1972 ,Parlevlie t andZadoks , 1977). Parlevliet (1979)suggeste d that selection forpolygeni c resistance to cereal rusts canb e achievedb y selecting genotypes with alo wdiseas e incidence,despit e ofa hig h (susceptible)reac ­ tiontype .Thi s socalle d 'partial'resistanc e canb emeasure db y the components ofresistanc e aslatenc yperiod , lesion density and sporulation intensity. Selectiono fgenotype s of 'Kouillou' for

107 breedingpurpose s has followedthes e criteria ofParlevliet .Th e five selected genotypes showed generally ahig h reaction type,bu t insom etest s (e.g.Tabl e 3, field test)intermediat e or lowreac ­ tiontype swer e scored.A clear distinctionbetwee n resistance of a higho r lowreactio ntyp ewa sno tpossible . Itseem s likely that polygenically inherited resistance to coffee leafrus tca nb eex ­ pressedb y alo wo rheterogeneou s reaction type,dependin g onth e testconditions . Thecomponent s ofresistanc e in 'Kouillou'wer e correlated with reactiontyp e (RT) (Table 3).A similarrelationshi p betweenresis ­ tance components andR Twa s found f.i.wit h slow leafrustin g wheat cultivars (Milusan dLine , 1980). Itseem s importanttha tR T isob ­ served instudie so n incomplete resistance,becaus e itma y givea cluet opossibl e resistancemechanisms . Incomplete resistance to H. vastatrix in 'Kouillou'varie s accor­ dingt olea fag ean d lightintensit y (Eskes,198 2 ,Eske s andToma - Braghini, 1982). These effectsma ypartl y explain the seasonal variation inresistance , the interactionbetwee n inoculation series andgenotype s (Table3 )an dth evariatio n inreactio ntype s observed withintrees .Effect s ofenvironmen t and development stageo fhos t tissues on resistance are common features forman y rust diseases (Hooker,1967 ,Sharm a et al, 1980,Milu s andLine ,1980 ,Stubbs , 1980). Brown and Sharp (1969)reporte d thatmino r genes forresis ­ tance toyello w rusto fwhea tvar y greatly ineffectivenes s depen­ dingo ntemperature . InKouillo u similar environment dependable minor genes forresistanc e to H. vastatrix mayb epresent . Minor geneswhic h are affected byplan t and environmental factors areno tcontinuousl y operative inth eresistanc eprocess .Hence , selectionpressur e forincrease d virulence inth epathoge n isno t continuous.Mutant s forvirulenc e may initially have alo w fitness (Parlevliet, 1981). Ifselectio npressur e isno tconstant ,suc h mutants are likely tob e less competitive thanothe r rustgenotypes . Therefore,th e influence ofplan t and environment factors onresis ­ tance asoccur s inKouillo uma y enhance the durability ofresis ­ tance. The crosses realized between 'Kouillou' and 'Catuai'ar e a first step to introducingmino r resistance genes from 'Kouillou' into C. arabica. Inth e firstbackcros s generation (BC-.)i twil lb e necessary towor kwit h agrea tnumbe r ofplant sbecaus e resistance willb e diluted. Selfing and intercrossing among selected BC^ genotypes shouldb e carried outt o allow foraccumulatio n ofth e resistance genes from 'Kouillou'.Th e success of such abreedin g

108 programme will depend on the number of resistance genes involved and on the expression of the 'Kouillou' resistance in the C. arabica genetic background.

ACKNOWLEDGEMENTS

The research was supported by the Food an Agriculture Organization of the United Nations (FAO), the Instituto Agronômico of Campinas (I.A.C.)/ S.P., Brazil, and the Agricultural University of Wagenin­ gen, The Netherlands. The author wishes to thank the staff members and personnel of the Genetics Department of the I.A.C. for the kind cooperation received. Special thanks go to Drs. R.A. Robinson, L. Chiarappa, N.A. van der Graaff (F.A.O.) and A. Carvalho (I.A.C.) for the stimulus received. The critical reviewing of the text by professors J. Sneep and J.E. Parlevliet is acknowledged. Mrs. Masako Toma-Braghini is thanked for her participation in the experimental work.

REFERENCES

Brown, J.F. & E.L. Sharp, 1969. Interaction of minor host genes for resistance to P. striiformis with changing temperature regimes. Phytopathology 59: 999-1001. Cadena G. , G. & P. Buriticâ C, 1980. Expression de resistencia horizontal a la roya (Hemileia vastatrix Berk, y Br.) en Coffea canephora variedad Conilon. Cenicafé 31: 3-27. Cadena G., G., 1978. Expression de resistancia horizontal a la roya (Hemileia vastatrix) en la variedad Conilon (Coffea canephora). M. Sc. thesis. Univer- sidad Nacional de Colombia. Bogota, Colombia. Eskes, A.B., 1982a. The use of leaf disk inoculations in assessing resistance to coffee leaf rust (Hemileia vastatrix). Neth. J. PI. Path. 88: 127-141. Eskes, A.B., 1982 . The effect of light intensity on incomplete resistance of coffee to Hemileia vastatrix. Neth. J. PI. Path. 88: 191-202. Eskes, A.B., 1983. Qualitative and quantitative variation in pathogenicity of races of coffee leaf rust (Hemileia vastatrix) detected in the State of Säo Paulo, Brazil. Neth. J. PI. Path. 89, in press. Eskes, A.B. & M. Toma-Braghini, 1981. Assessment methods for resistance to coffee leaf rust (Hemileia vastatrix Berk, et Br.). PI. Prot. Bull. F.A.O. 29: 56-66. Eskes, A.B. & M. Toma-Braghini, 1982. The effect of leaf age on incomplete resis­ tance of coffee to Hemileia vastatrix. Neth. J. PI. Path. 88: 215-226. Hooker, A.L., 1967. Th genetics and expression of resistance in plants to rusts of the genus Puccinia. Ann. Rev. Phytopathol. 5: 163-182. Milus, E.A. & R.F. Line, 1980. Characterization of resistance to leaf rust in Pacific Northwest wheats. Phytopathology 70: 167-172. Parlevliet, J.E., 1979. Components of resistance that reduce the rate of epidemic development. Ann. Rev. Phytopathol. 17: 203-222. Parlevliet, J.E., 1981. Stabilizing selection in crop pathosystems: an empty concept or a reality? Euphytica 30: 259-269. Parlevliet, J.E. & J.C. Zadoks, 1977. The integrated concept of disease resis­ tance; a new view including horizontal and vertical resistance in plants. Euphytica 26: 5-21. Paulino, A.J., 1981. Productividade de progenies de Guarini EP37 de Coffea cane­ phora no Espirito Santo. In: Resumos 9° Congresso Brasileiro de Pesquisas Cafeeiras. Säo Lourenço (MG), Brazil, October 1981. p 264-266. Scali, M.H., A. Carvalho & L.C. Monaco, 1974. Resistencia horizontal à ferrugem

109 (Hemileiavastatrix )e mcultivare sd eCoffea .In :Resumo sXXV IReuniä oAnua l Soc.Bras .Progress od aCiencia .Julho ,1974 ,p .241 . Sharma,J.K. ,W.A .Heathe r& P .Winer ,1980 .Effec to flea fmaturit y andshoo t ageo fclone so fPupulu sspecie so nsusceptibilit y toMelampsor alarici - populina.Phytopatholog y 70:548-554 . Simons,M.D ., 1972 .Polygeni c resistancet oplan tdisease san dit sus ei nbreedin g resistantcultivars .J .Envir .Qual . 1.:232-239 . Snedecor,G.W .& W.G .Cochran ,1967 .Statistica lMethods .Iow aStat eUniversit y Press,Ames ,Iowa ,USA .59 3pp . Stubbs,R.W. , 1980.Environmenta l resistanceo fwhea tt oyello w rust (Puccinia striiformisWestend . f.sp.tritici)• In:Proceeding sFift hEur .an dMed .Cerea l RustsConf. ,Bar ian dRome ,Italy .Jun e 1980.p .77-81 . Vanderplank,J.E. ,1968 .Diseas eResistanc e inPlants .Academi cPress .Ne wYork . 206pp .

110 8 CHARACTERIZATION OF INCOMPLETE RESISTANCE TO HEMILEIA VASTATRIX IN THE ICATU COFFEE POPULATION

A.B. Eskes1 andW.M .d a Costa

InstitutoAgronômic o ofCampina s (I.A.C.), Campinas,S.P . Brazil

INDEXWORD S Coffea spp.,Coffe e LeafRust ,heterogeneou s reaction type,durabl e resistance

SUMMARY

Incomplete resistance ofth ehybri d coffeepopulatio n Icatut o race IIo fcoffe e leafrus t (Hemileia vastatrix) was studied inth e field, greenhouse and laboratory.Th e resistance components observed were: latencyperio d (LP),lesio ndensit y (LD),sporulatin g lesion density (SLD)an drelativ e sporulation lesion efficiency (RSLE= 10 0 SLD/LD). RSLE is an indicator forth eheterogeneit y of thereaction . Disease score inth e fieldwa s highly correlatedwit hRSLE ,les s withL P and SLD,an dno twit hLD .L Pwa shighl y correlated with RSLE. A new rust race (Is.2) ,isolate d inth e field from Icatu in 1979, wasmor evirulen t thanrac e II on some resistant,moderatel y resistant ormoderatel y susceptible genotypes.Thi s indicates that incomplete resistance, atdifferen t levels,ca nb e race specific. Resistance was affected by leaf age and light intensity. Inherit­ ance studies suggesttha t incomplete resistance in Icatumigh tb e related tomajo r genes,th eeffectivenes s ofwhic hma y depend on gene dose and genetic background. Iti sconclude d that selection for incomplete resistance to coffee leafrus t in Icatuma yno t lead to durableresistance .

Present address:Ru a Salesopolis 194,Alt o daBarra , 13100 Campinas,S.P. ,Brazil .

Ill INTRODUCTION

The twomai n commercial coffee species are Coffea arabica L. (auto­ gamous, allotetraploid)an d C. canephora Pierre (allogamous, diploid). The first species isresponsibl e for80 %o fth eworl d coffeeproductio n and isgrowin gbes t intropica l highlands;th e second species isbes t adapted toth e tropical lowlands (Carvalho et al., 1969). In C. arabica fivemajo r genes,S„ lt o S„5, forresistanc e to coffee leafrus t (Hemileia vastatrix Berk,e tBr. )exist .Thes e genes haveno tbee n efficient incontrollin g coffee leafrust . InBrazil ,afte rte nyear s ofrus tpresence , only the S„3gen e is still effective (Eskes, 1983a). All combinations ofth e other 4 geneshav ebee nmatche d byvirulen t races.Mos tbreedin g efforts areno w focussedo ntw otetraploi d hybridpopulations ,calle d 'Icatu'an d 'Catimor',wic h arederive d from crossesbetwee n C. canephora and C. arabica. Thesepopulation s contain severalresis ­ tance genes from C. canephora, which differ from Sglt o SH5,an d provide resistance to someo r to allknow nrus trace s (Rodrigues and Bettencourt, 1975). Icatu showswid evariatio n for agronomic traits,resistanc e to coffee leafrust ,t o coffeeberr y disease andt ocoffe e rootkno t nematodes (Carvalho et al. 1976,Costa , 1978,Monaco , 1977). In relationt o thecoffe e leafrus tresistance ,Marque s and Bettencourt (1979)observe d inconsistency inreactio nbetwee n subsequentinocu ­ lations.Fiel d observations inBrazi l ofCost a andRibeir o (1975) indicated thatth emajorit y of Icatugenotype s ishighl y resistant to the commonrac e II (v,-)o f H. vastatrix, buttha t alsoman y genotypeswit h intermediate resistance arepresent .Cost ae t al. (1978)showe d ahig hcorrelatio nbetwee nresistanc e scores inth e greenhouse and field. The existence ofquantitativ evariatio n forresistanc e in Icatu may suggestth epresenc e ofmino r genes orpolygenes .Polygeni c resistance mayb emor e durable thanmonogeni c resistance (eg.Van - derplank, 1968, Simons,1972 ,Parlevliet , 1978). Thepresen twor k investigates characteristics of incomplete resistance in Icatu in order to assess itsvalu e inobtainin g durable resistance tocoffe e leaf rust.Th eter m 'incomplete resistance' isuse d to indicate all types ofresistanc e which allow for,a tleast ,som e reproduction of thepathogen .

112 MATERIALS AND METHODS

Icatu genotypes: Icatu isa tetraploi d coffeepopulatio n derived from acros sbetwee n a C. canephora genotype,wit h adouble d number ofchromosome s (2n= 4 x = 44),an d C. arabica cv.Bourbo nVermelh o (2n= 4 x =44) .Thi s crosswa smad e inth e 1950's atth e Instituto Agronômicoo fCampina s (I.A.C.), SP,Brazil .Th eF ,ha sbee nback - crossedmostl y to C. arabica cvsMund oNov o and CaturraVermelho . Progenies derived from thesebackcrosse s showgrea tvigour . Aneuploidy occurs ina numbe r ofprogenie s (Monacoe t al., 1974). The Icatu genotypes used inth epresen twor kbelon g to a field experiment,plante d in 1971,includin g 2400plant s derivedb y open pollination from someo fth ebes tBC 1 andBC 2plants .Code s used to indicate genotypes include hybrid numbers andplan tnumbers ,sepa ­ ratedb y dashes (generations).Th ehybri d numbersH 3849 andH 3851 are firstbackcrosse s ofth eF lwit h 'CaturraVermelho ' and 'Mundo Novo', respectively. H478 2 isa secon dbackcross , forwhic h 'Mundo Novo' hasbee nuse dtwice . Inoculation methods. Information onstandar d inoculationmethod s and on the suitability ofth e laboratory test in assessing resistance is givenb yEskes ,1982 a. For the greenhouse tests (Tables 1an d 2) two-year-old graftedplant swer e used.Thes e graftswer emad e on rootstocks with similar genotype and resistantt orac e IIo f H. vastatrix. The 24 Icatu genotypesuse dwer e selected because of good yielding capacity and greatvariatio n inresistance .Thre e plants ofC . arabica cv.Mund oNov ower euse d as acontrol .O feac h planttw obranches ,wit h 6leave s each,wer e inoculated. For the field test (Tables 1an d 2)thre ebranche swer euse d from the same 26 genotypeswit h 6leave spe rbranch .Onl ybranche s growing inth e shadewer e inoculated, to avoid interference withhig h insolation. For the laboratory test (Table 3),6 replicate swer e usedwit h 20 leafdisk s each.Fo r the field tests (Table 4),on ebranc hwit h 8 to 12leave swa s inoculated per genotype/isolate combination and forth e laboratory test (Table4 )1 5 leafdisk swer eused . Inocula­ tionswer e carried out inDecembe r 1976 (Tables 1an d 2),Decembe r 1977 (Table 3),Novembe r 1980 andMarc h 1981 (fieldtests ,Tabl e4) , March 1981 (laboratory test,Tabl e 4),an d January 1977 (Table5) . Exceptwhe nmentione d otherwise, for alltest s H. vastatrix race IIwa sused . Isolate ICT (Table 1)wa s obtained in197 6 from several Icatu fieldplants ,derivative s fromH 3851-2. Isolate 2 (Table4 )wa s obtained in 1979 fromH 3851-4-40,a genotyp e which is incompatible torac eII .

113 Evaluation of resistance. Reaction typewa s scoredb y a0 t o9 scale (Eskesan dToma-Braghini , 1981). Value 0indicate s absence of visible symptoms,1 t o 3variatio nwithi n aresistan t (R)reactio n type,4 to 7heterogeneou s reactiontype swit h increasing percentage ofsporulatin g lesions and increasing sporulation intensity.Value s 4 and 5ar e classified asmoderatel y resistant (MR),6 an d 7a s moderately susceptible (MS).Value s 8 and 9indicat e a susceptible (S)reactio n typewit hmoderat e (8)t ohig h (9)sporulatio n inten­ sity.Th e resistance components weremeasure d according toEske s (1982a). Latencyperio d (LP)i sth e time,i n days,betwee n inocula­ tionan d sporulation of 50%o fth e lesions thatcom et osporulate ; lesion density (LD)i sth enumbe r ofvisibl e lesions (resistant and susceptible types)pe r leaf areaunit ;sporulatin g lesion density (SLD)i sth enumbe r of sporulating lesionspe r leaf areaunit ;an d relative sporulating lesionefficienc y (RSLE)i sth epercentag e of sporulating lesions (100.SLD/LD).RSL E isa measur e ofth ehetero ­ geneity ofth ereaction . Inth e field thediseas e level hasbee n scored yearlyb y a1 t o5 scale (disease score). The scaleha sbee n adapted from a1 t o6 scaledescribe d by Costa (1978).Valu e 1indicate s absence ofsymp ­ toms orpresenc e ofonl y resistant-type lesions,2 t o 5 indicate increasing numbers of sporulating lesions associated with anin ­ crease in sporulation intensity.Value s 2 and 3generall y indicate heterogeneous reaction typeswit h slight (2)an dmoderat e (3)sporu ­ lation intensity. Furthermore inMa y 1977,th epercentag e of leaves with sporulating lesions and thepercentag e ofnecroti c lesionswa s observed by sampling 100 leavespe rgenotype .

RESULTS

Expression of resistance. Twenty four genotypes of Icatuwer e tested inth e field and greenhouse.Tabl e 1show s the results ofte n representative genotypes and ofth e control 'MundoNovo ' of C. ara­ bica. Inth e field, one genotype showed ahighe r disease score than 'MundoNovo' ,tw ower e similar to 'MundoNovo' , five showed interme­ diate or lowdiseas e levels and threewer e nearly completelyresis ­ tant. Increasing levels ofresistanc e were related to ahighe rper ­ centage ofnecroti c lesions inth e field. 'MundoNovo ' showed11 % necrotic lesions,whic h isprobabl y notdu et o aresistanc e reaction but to overheatingb y the sun (Eskes,198 2 ).Th e inoculation test inth e field resulted in sporulation on 5genotypes ;6 showe dresis ­ tanttyp elesions .

114 or-r-vooomr-.(r)0© o m os co m H

>» w 4-1 i-i ao •-*_ aj 3 00 +J ra

a m

01 en u •H s +raJ e T3 i—1 M ra Ö •H "O •H CA <^S cu c W ^-s o ra v d m o •H (ful t—o1

ca u 4-> ra a n cu d - o x P.-H e u O 4-1 u ra r-l a 4-> l-J TD 01 60 KI

S« u ra o3 CU i-l • ^•mcoooooLO'-HooLno O 4J O. 3 O •H XI 00 4-1 n o CM 4J CA O <-H •H o o .-H -H d &-—- CU 9 u •H OJ o c/l Xi CJ W 4-1 CU OJ OJ en 3 •rai O ON « CU CU u *J r-t a i—( XI -H CA •CaU £ 00 .-v ^ S d ä* •rl 3 <4-l tj 1^ •H •H >—' co d cu 00 aCd S •U •H w ( i—l ra co > d •o 1*4 VI r-l Ö >> o arai CU 3 O 3 IH 4J -H a U 'r4 2 • d O-H M > O VI d w oo ra CA CA OJ «CU p, CU o > H d r-l h-4 C*ï r-l ra eu cu •OBTJ M § d 3 d 3 d -H -H a O 4-> ^ -H ra OvO-J'Noa-J^OMO -1 CO rr uTÏ-^-^fOfOcnMHH i >> cu I X 4-1 CA 00 CM CU d ti CA CU TJ ra - n P, CU -H rH CU cu >. i TJ 4-1 CU •p >> ra i >* ra > CU CA 4J d i-i -H CA CU (N d o 3 4-1 4-> cu -H n ra «CU •\i—i\or-i—i H N rn M m c/l jino> i i ï •H O 'HH B M Cl I I CO CO I I I I t > O> O 55 i i i i i i i i i i-t HHMCJNHC^HiNH 1-4 CU O ïmooooco msf moo m 01 CA CA T3 Cl Tl 3oor*-r-r^cooooor»-oo ico^-J-^rococo^fco > XI 01 Cs4 Q hi « i-H Cl < O -I < -J a en Pa XI 3 S Hra

115 Inth egreenhouse , sporulation occurred on8 genotypes ,th e other two showed only resistanttyp e lesions. Iti sno texpecte d thatth e rootstock has influenced thereactio n ofth egraft s inth e greenhouse test,becaus e theresistanc e ofth e rootstocks has notpasse d to grafts from susceptible Icatugenotypes . Significantvariatio nbetwee n genotypes was observed foral l components ofresistanc e observed inth e field and greenhouse test (Table 1).Fo rth e analysis,eac h leafha sbee n taken asa repli ­ cate.Th e coefficient ofvariatio n (C.V.)wa s low for latency period (LP)i nth e field (8%)an d greenhouse (11%), buthig h for lesion density (LD)an d sporulating lesion density (SLD) (93an d 90%,res ­ pectively). TheC.V . forth e relative sporulating lesion efficiency (RSLE)wa s about 100%fo r intermediate RSLEvalue s (between 20an d 80), and lower forRSL Evalue s near to 0o r 100.A few genotypes with intermediate resistance showed significant differences inLD , SLD and RSLEbetwee ngraft s and alsobetwee nbranches ,withi ngrafts . However, ingenera l thevarianc e due to leaveswithi nbranche s was as greata sth evarianc e due tobranche s andgrafts . Table 2show s therelationshi p between resistance parameters for 18 Icatu genotypeswhic hha d atleas t some disease inth e field.

Table2 .Coefficient so fcorrelatio nbetwee nresistanc eparameter sobserve di n thefiel dan dgreenhous eo f1 8Icat ugenotype swit hincomplet eresistanc et oH ^ vastatrixrac eI I

Parameter Fieldobservations 1 Components ofresistance 1 Disease Necrotic Field Greenhouse score2 lesions(% ) LP LP LD SLD

Field:

Necrotic lesions(% ) -.75*3 LP -.71* .75*

Greenhouse:

LP -.80* .55* .65* LD .39 -.33 -.23 -.53* SLD .61* -.46 -.39 -.65* .93* RSLE .94* -.71* -.76* -.91* .48* .64*

1Fo rexplanatio no fabbreviation sse eTabl e1 2 Averagescor efo r197 5t o197 7 3 Anasteris k(* )indicate ssignificanc ea tP è0.0 5 For thecorrelation s withdiseas e score inth e field genotype H 3851-2-513 was excluded because of its interactionwit h the rust

116 isolate from Icatu (Is. ICT, Table 1). The highest coefficient of correlation was observed for RSLE (0.94), followed by LP (0.80) and SLD (0.61). The correlation between disease score in the field and LD was not significant. RSLE was correlated with the % necrotic lesions in the field, which can be expected since both are indica­ tors for reaction type. Furthermore, RSLE was significantly corre­ lated with LP and SLD, but not with LD. It is concluded that the variation for incomplete resistance of the Icatu genotypes is basi­ cally related to the variation in heterogeneity of the reaction type.

Effect of leaf age and light intensity. Table 3 shows the results of a laboratory test in which leaf disks were taken from 'young' and 'old' leaves growing in the shade or exposed to sunlight. Resistance was highest for young leaves growing in the shade, and lowest for old leaves exposed to sunlight. For the analysis of variance SLD was transformed into arcsin^SLD. Significant effects were detected of genotypes, leaf age (leaf position) and light intensity (shaded or

Table 3. Sporulating lesion density (SLD = % of leaf disks with sporulation) and reaction type (RT, 0 to 9 scale) for leaves from different positions on the branch and different light conditions in the field of 3 Icatu genotypes ino­ culated with H. vastatrix race II in the laboratory.

Leaf position1 Parameter 1,2 4,5 Genotype ofresistanc e ShadeSu n Shade Sun

H3851-4-4 1 SLD la" 8a 30b 43c RT 4.5 4.0 5.5 6.5

H3851-2-68 9 SLD 18a 39b 17a 67c RT 4.0 5.0 4.5 6.5

H 4782-13-152 SLD 49a 68b 67b 71b RT 5.5 7.0 6.5 7.5

Mean SLD 23a 38b 38b 60c RT 4.7 5.3 5.5 6.8 1 Leaf positions 1 and 2 indicate the youngest and second fullgrown leaf on the branch. Leaves of positions 4 and 5 were old leaves, grown in the foregoing growing season. 2 Different letters indicate differences for SLD between treatments within rows according to the LSD 0.01 value. sunexpose d leaves). Theeffect so f interactionbetwee n genotypes and leaf age andbetwee n genotypes and light intensitywer e also significant, althoughth e sametendenc ywa spresen t for allgeno -

117 types.Difference s inresistanc e were alsoexpresse d by differences inreactio n typewhic hwa s heterogeneous formos tcombinations .

Specificity of resistance. A first indication ofphysiologi c spec­ ialization of H. vastatrix inth e Icatupopulatio nwa s obtained in the greenhouse test,carrie d out inJanuary/Februar y 1977 (Table1) . The rust isolate from Icatu fieldplant s (Is.ICT )wa s slightly more pathogenic thanrac e II on four genotypes.However , attemptst o obtain ane w race from Is.IC T failed.Reinoculatio n ofth e4 geno ­ typeswit h spores of Is.IC Ttake n fromwel l sporulating lesionswa s done inth ewinte r of 1977.Almos t only resistant-type lesionswer e obtained,wherea s on 'MundoNovo ' 100%sporulatin g lesionswa sob ­ served.Thi s indicates thatth eresistanc e ofth e Icatu genotypes had changedbetwee n thetw o tests.Als o inothe r tests, Icatugeno ­ typeshav e shown amor e resistant reaction inth ewinte r than inth e summer. Between 1977 and 1980 about3 % formerly resistant Icatu genotypes became diseased inth e field. In197 9 ane w racewa s isolated from H 3851-4-40,a genotyp ewit h complete resistance to race IIo f H. vastatrix. Table 4 shows the average reactiontyp e of 6 Icatu genotypes inoculated twice inth e field and once inth e laboratory with race II and Is.2 .Th e 6genotype svarie d inresistanc e to race II, frommoderatel y susceptible tohighl y resistant,bu t all showed increased susceptibility to Is.2 .Therefore , intermediateresis ­ tance in Icatu canb e race-specific.Variatio n inreactio n typeswa s observed between tests.Th e reaction to Is.2 o fth emor e resistant genotypes (H3851-4-4 0 andH 3851-4-41)varie d frommoderatel y re­ sistantt o susceptible, indicating that additional resistance fac­ torsma yb epresent . 'MundoNovo 'ha s some resistance to Is.2 (Table 4). Thisha sbee nconfirme d inth egreenhouse ,wher emulti ­ plication of Is.2o n 'MundoNovo ' isdifficul t inth ewinte rmonths . Is. 2ha sbee n sentt oth e CoffeeRust sResearc h Center inOeiras , Portugal (C.I.F.C.)fo r further identification.

118 Table 4.Diseas e scorei nth efiel d (1t o5 scale )an daverag e reaction type (0 to9 scale )an drang efo rreactio n typeo f6 Icat ugenotype san do f'Mund oNovo ' inoculated twicei nth efiel dan donc ei nth elaborator ywit hH .vastatrix , race IIan disolat e 2 (Is. 2. obtained fromH 385 1 -4-40).

Reaction type Disease score Average Range Genotype 1975 1980 Race II Is. 2 RaceI I Is. 2.

Icatu: H3851-2-68 9 3 5 7.3 9.0 7.0-7.5 9.0 H 3851-2-513 4 5 7.0 7.7 7.0 7.0-9.0 H 4782-10-203 2 5 5.5 7.7 5.0-6.0 7.0-8.0 H 3849-7-104 2 5 3.5 8.0 2.0-5.0 7.5-8.5 H3851-4-4 1 2 4 3.0 7.0 2.0-4.0 5.0-8.0 H3851-4-4 0 1 4 2.5 6.5 1.0-3.0 5.5-8.0

'MundoNovo ' 5 5 9.0 7.7 9.0 6.5-9.0

Segregation for resistance in progenies from resistant genotypes. One-year-old seedlings,obtaine d by self-fertilization, fromresis ­ tant andmoderatel y resistantgenotype swer e tested inth egreen ­ house.Al l leaves ofeac h seedlingwer e inoculatedwit h race IIo f H. vastatrix. Segregation forresistanc e occurred inth e progenies of 7genotype s (Table 5).Withi neac h reactiontyp e (R,MR ,M S and S)a variatio n inth eexpressio n ofresistanc e existed.Th evaria ­ tion amongplant s ofth e sameprogen ywa s rather quantitative than qualitative.A fewresistant-typ e seedlings (R)showe d no disease symptoms. Othersvarie d inreactio n from tiny fleckst o large chlorotic areas,whic h oftenoccurre d together. Seedlingswit h intermediate reactiontype s (MR,MS )showed , ingeneral ,a hetero ­ geneous reactionwit h resistant and susceptible-type lesions occur­ ringo nth e same leafo r on the same seedling. Several seedlings showed aresistan treactio n onyoun g leavesbu twer e susceptible on old leaves,confirmin g the results ofTabl e 3.Geneti c analysis of the segregation ratio isdifficul tbecaus e ofth e quantitative variation inreactio n typeobserve d andbecaus e meiotic irregulari­ tiesma y occur insom e Icatuprogenie s (Monacoe t al. 1974).How ­ ever,polygeni c inheritance seemsunlikely ,becaus e the distribution ofth e seedlings over the resistance classeswa s generally notnor ­ mal. Itseem smor eplausibl e thatmono -o r oligogenic resistance is involved,whic hma yb e affected by gene dose,geneti c background, and/or environment.

119 Table 5. Percentage of leaves with sporulating lesions (% LSL) and reaction type (RT) in the field of 9 Icatu genotypes and RT of seedlings from these genotypes, obtained by self pollination, tested with H. vastatrix race II in the greenhouse.

Field1 Number of seedlings Genotype RT^ % LSL RT2 R MR MS S

H 4782-7-549 0 R 25 7 3 5 H 4782-7-918 1 MR 22 1 H 4782-13-206 2 MR 7 4 6 3 H 3851-2-363 3 MR 32 2 7 2 H 4782-7-724 3 MR 3 10 4 6 H 3849-14-3-22 5 MR 35 3 4 H 4782-7-896 6 MR 14 2 4 4 H 3849-7-104 8 MR 5 6 4 4 Control: H 4782-7-27 69 S 2 19

1 Observations made in May 1977. 2 'R' indicates presence of non-sporulating lesions, 'MR' a heterogeneous reaction type with a low % of sporulating lesions and a low sporulation in­ tensity, 'MS' a heterogeneous reaction type with a high % of sporulating lesions and intermediate sporulation intensity, 'S' a susceptible reaction type with sporulating lesions only. Quantitative variation, similar to what was observed in the Icatu progenies, is found in F_ populations segregating for the resis­ tance gene S„4. Depending on environment and leaf age this gene shows partial or complete dominance (Eskes, 1982 , Eskes and Toma- Braghini, 1983). In Icatu major genes with similar expression may be present. For instance the segregation in the progenies from H 4782-13-206, H 4782-7-724, and H 3849-7-104 suggests the presence of a major gene which, in heterozygous condition, gives an MR of MS reaction type. The resistance of H 3847-7-104 to race II is race- specific (Table 4), which is also an indication that a major gene is involved.

DISCUSSION

Incomplete resistance in Icatu is mainly characterized by the degree of heterogeneity of the reaction type (Table 2). Heterogeneous reac­ tion types have been related to race-specific major genes (Parle- vliet, 1979). The expression of such genes can be affected f.i. by temperature (Vanderplank, 1978), light intensity (Stubbs, 1967) and/ or development stage of the plant (Milus and Line, 1980). Equally, the heterogeneous reaction type of coffee leaf rust in Icatu might be related to race-specific genes which have a labile expression.

120 Marques andBettencour t (1979)reporte d inconsistent results between inoculation tests in Icatu,whic hmake s classification of Icatugenotype s inresistanc e groups difficult. Such inconsistency ispartl y explainable by thepresen t results,whic h indicate that resistance in Icatu canb e affected by light intensity and leaf age (Table3 )o rb y season.T o obtain consistent results in resistance testswit h Icatu itwil lb enecessar y toconfin e thetes t conditions asmuc h aspossible . Most studies on incomplete resistance toplan tpathogen s report only onresistanc e components,withou t observing reactiontype . Milus and Line (1980)showe d thatcomponent s of slow leaf rusting resistance inwhea twer e correlated toreactio n type.Simila rre ­ sultswer e observed in Icatu (Table 2).Thi s suggests thatther ema y ben obasi c difference inth emechanism s ofresistanc e expressed by resistance components orb yreactio ntype .Th e relationship between resistance components and reaction typeha sbee n discussed inmor e detailb y Eskes (1981). Theoccurrenc e ofrace-specifi c resistance atvaryin g levels (Table4 )indicate s that selection for incomplete resistance in Icatuma y not lead todurabl e resistance. Iti spossibl e thatdura ­ ble, polygenically inherited, incomplete resistance ispresen t in Icatu.However , itwil lb e extremely difficult to separate such resistance frommonogeni c incomplete resistance,unles s accurate inheritance studies aremade .Simila r difficulties are apparently facedwit hbreedin g for incomplete resistance to certain cereal rusts (Parlevliet, 1978 and 1979). Alternative breeding strategies fordurabl e resistance tocoffe e leafrus t in Icatu,base d onaccu ­ mulation of several resistance genes inon egenotyp e or oncombina ­ tiono fcertai n resistance factors from Icatuwit h others from Catimor, arediscusse d byEske s (1983) .

ACKNOWLEDGEMENTS

The researchwa s supported by theFoo d andAqricultur e Organization ofth eUnite dNation s (FAO), the InstitutoAqronômic oo f Campinas (I.A.C.), S.P., Brazil and theAgricultura l University ofWagenin ­ gen, theNetherlands .Th e authorswis h tothan k thepersonne l of theGenetic s Department ofth e I.A.C. forth ecooperatio n received. Special thanks go toDrs .R.A . Robinson,L . Chiarappa,N.A . vande r Graaff (F.A.0.)an dA . Carvalho (I.A.C.)fo rth e stimulus received. The critical reviewing ofth e textb yprofessor s J. Sneep and J.E.Parlevlie t is acknowledged.

121 REFERENCES

Carvalho,A. ,F.P .Ferwerda ,J.A .Frah m -Leliveld ,D.M . Medina,A.J.T .Mende s & L.C.Monaco ,1969 .Coffee .In :Outline so fperennia l cropbreedin g inth e tropics.H .Veenma n& Zone nNV ,Wageningen .Ed :F.P .Ferwerd a& F .Wit . p.189-241. Carvalho,A. ,L.C .Monac o& H.A.M ,va nde rVossen ,1976 .Caf éIcat ucom ofont ed e resistência aColletotrichu m coffeanum.Braganti a 35:343-347 . Costa,W.M. ,1978 .Relaça oentr egra ud eresistênci a aHemilei avastatri xe produtividade nocaf éIcatu .Braganti a 37:1-9 . Costa,W.M .& I.J.A .Ribeiro ,1975 .Resistênci a aH .vastatri x observadan ocaf é Icatu.In :Resumo sTerceir oCongress oBrasileir o dePesquisa sCafeeiras . Curitiba,Brazil ,Novembe r 1975.p.133 . Costa,W.M. ,A.B .Eske s& I.J.A .Ribeiro ,1978 .Avaliaça od onive ld eresistênci a docafeeir oa H .vastatrix .Braganti a 37 (notano .4) : XXIII-XXIX. Eskes,A.B. ,198 1Expressio no fincomplet e resistance topathogens .In :Proc . NATO/ASIconferenc eo nDurabl eResistance ,Martin aFranca ,Italy ,Octobe r 1981.I npress . Eskes,A.B. ,198 2 .Th eus eo flea fdis kinoculation s inassessin g resistancet o coffeelea frus t (Hemileiavastatrix) .Neth .J . PI.Path .88 :127-141 . Eskes,A.B. ,198 2 .Th eeffec to fligh tintensit yo nincomplet e resistanceo f coffeet oHemilei avastatrix .Neth .J . PI.Path .88 :191-202 . Eskes,A.B. ,198 3 .Qualitativ ean dquantitativ evariatio n inpathogenicit yo f raceso fcoffe elea frus t (Hemileiavastatrix )detecte d inth eStat eo fSa o Paulo,Brazil .Neth .J . PI.Path .89 :i npress . Eskes,A.B .& M .Toma-Braghini ,1981 .Assessmen tmethod sfo rresistanc et ocoffe e leafrus t (Hemileiavastatri xBerk ,e tBr.) . PI. Prot.Bull .FA O29 :56-66 . Marques,D.V .& A.J .Bettencourt ,1979 .Resistênci a àHemilei avastatri xnum a populaçaod eIcatu .Garci ad eOrta ,Sér .Est .Agron .Lisbo a6 : 19-24. Milus,E.A .& R.F .Line ,1980 .Characterizatio n ofresistanc e tolea frus ti n PacificNorthwes twheats .Phytopatholog y 70:167-172 . Monaco,L.C. ,1977 .Consequence so fth eintroductio no fcoffe erus tint oBrazil . Ann.N.Y .Ac .Sei .287 :57-71 . Monaco,L.C ,A .Carvalh o& L.C .Fazuoli ,1974 .Melhorament o docafeeiro ; germoplasma docaf éIcat ue se upotencia ln omelhoramento .In :Resumo sSegund o CongressoBrasileir o dePesquisa sCafeeiras ,Poço sd eCaldas ,MG ,Brazil , 1974.p.103 . Parleviet,J.E. ,1978 .Aspect s andproblem swit hhorizonta l resistance.Cro p Improv.5 :1-10 . Parleviet,J.E. ,1979 .Component so fresistanc etha treduc eth erat eo fepidemi c development.A .Rev .Phytopath . 17:203-222 . RodriquesJr. ,C.J. ,A.J .Bettencour t& L .Rijo ,1975 .Race so fth epathoge nan d resistancet ocoffe erust .A .Rev .Phytopath . 13:49-69 . Simons,M.D. ,1972 .Polygeni c resistancet oplan tdisease san dit sus ei nbreed ­ ingresistan tcultivars .J .Environm .Qual .1 :232-239 . Stubbs,R.W. ,1967 .Influenc eo fligh tintensit yo nth ereactio no fwhea tan d barleyseedling st oPuccini a striiformis.Phytopatholog y 57:615-619 . Vanderplank,J.E. ,1968 .Diseas e resistance inplants .Academi cPress ,Ne wYork . 205pp . Vanderplank,J.E. ,1978 .Geneti can dmolecula rbasi so fplan tpathogenesis . SpringerVerlag ,Berlin ,Heidelberg ,Ne wYork . 167pp .

122 GENERAL DISCUSSION

COFFEE RESEARCH

The scalespropose d for assessing thediseas e level of coffee leafrus t (Chapter 1)ma yb e ahel p tocoffe ebreeder s and epidem­ iologists.Th e 0t o 9scal e forreactio ntyp e isconsidere d tob e importantbecaus e itinclude sheterogeneou s reaction typeswhic h are commonwit hcoffe e leafrust . When screening forincomplet e resistance to coffee leafrus ti t isrecommende d tomeasur ebot h therus t incidence and the reaction type.Assessmen t ofresistanc e components,suc h aslatenc y period and sporulation intensity, isonl y recommended ifver y detailed observations need tob emad ebecause , ingeneral ,R T is correlated to latencyperio d and sporulation intensity. Thepredisposin g effecto fhig h light intensity onresistanc e to coffee leafrus t (Chapter 3)indicate s thatscreenin g for resistance shouldpreferabl y be doneunde r low lightintensity .Par to fth e variation for incomplete resistancema y notcom e toexpressio n under high lightintensities . The effecto f leaf age (Chapter4 )indicate s that leaves of different agenee d tob e inoculated for accurate assessment ofth e level ofresistance .Th e screeningmetho d applied atth e Coffee Rusts Research Center (C.I.F.C.)i nOeiras ,Portugal ,use s young tender leaves only.Th epresen t results indicate thatth e C.I.F.C. methodma yunder -o rover-estimat e the real level ofresistanc e of certaincoffe egenotypes . The laboratory inoculationmetho dprove d tob e aversatil e tool in assessing resistance to coffee leaf rust (Chapter 2). Itca nb e recommended e.g. forrac e identification, for resistance screening of fieldplant s andnurser ypopulations , forstudie s on quantitative host-pathogen interactions,an d for studies onth e effecto f envi­ ronment onresistance .A n advantage ofth emetho d istha tth e resi­ dual error inexperiment s canb e reduced to aminimu m by distribu­ tingdisk s from the same leafove r different treatments.Othe r advantages are thatth emetho d ischeap ,efficien t intim e and space, and that allnecessar y controls canb e easily incorporated intoon eexperiment . Till 1975,mos t information onresistanc e to coffee leafrus twa s related tomajo r genes forresistanc e present incoffe edifferen -

123 tialsemphasizin g thequalitativ e relationship betweenhos tan d pathogen.Th epresen t research gives information onth e quantitative relationshipbetwee ncoffe e and H. vastatrix. Iti sa n interesting findingtha t incomplete resistance to H. vastatrix, atvariou s levels, canb e race-specific.Thi s suggests that selection for incomplete resistance tocoffe e leafrus tma yno tnecessaril y lead todurabl e resistance.A bette r indication ofth e durability of resistance mayb e obtainedb y inheritance studies.Selectio n should be for incomplete resistance which inheritspolygenicall y (Parle- vliet, 1978b, Parlevliet and Zadoks, 1977). Somegenotype s ofC . arabica, Icatu, and 'Kouillou'wer emor e susceptible tocoffe e leafrus ttha nth eBrazilia n cultivarsMund o Novo and Catuai.Th e low level of resistance of 'MundoNovo 'an d 'Catuai' islikel y tob edurabl e resistance. Iti sexpresse db y a lower lesion density andb y arelativel y short leaf retention period. Till 1981,eleve n qualitatively distinct rustrace swer e detected inth e State ofSä oPaul o (Chapter 5). Mosto fthes e raceswer e found inth ecoffe ebreedin gplot s atCampinas .Thi s showstha tne w racesma y appear evenwhe n thecorrespondin g resistance factor is present inonl y afe w coffee trees.Monitorin g ofrac e formation in breedingplot s iso f importance forth eearl y detection ofthes e races and should therefore receive full attention. Some ofth ene w racesneutralize d theresistanc e ofhig h yielding breeding lines.Th e residual resistance of these linesvaried . In several cases the residual resistance ofthes e lineswa s consider­ ably lowertha ntha to f 'MundoNovo 'o r 'Catuai'.Thi s shows that the 'Vertifolia'effec t (Vanderplank, 1968)ma y occurwit hbreedin g for resistance tocoffe e leafrus t and emphasizes thenee d for durableresistance . Differentbreedin g strategies to acquire durable resistance are discussed inChapte r 5.Th eus eo f single genes incultivar s cannot be recommended.Th eus eo fmultiline sma y favour,a sobserve d inth e breedingplots ,th e formation ofcomple x races.Th e accumulation of severalne w resistance genes inon ecultiva r couldb evaluable ,bu t isdifficul t to realizewit hth e availablebreedin g techniques.Th e use ofne w technology forrapi dvegetativ e multiplicationma y in futurehel pt oproduc e hybrid cultivars carryinrrsevera lmajo r genes forresistance . Thepresen t results offer a firstevaluatio n of incompleteresis ­ tance forobtainin g durable resistance tocoffe e leafrust .Som e promising selections weremad e ofgenotype swhic h seem toposses s

124 polygenic incomplete resistance.A breedin gprogramm e based on crossesbetwee n 'Kouillou' genotypes andC . arabica cv.Catua i was started. Iti srealized ,however ,tha tselectio n forpolygeni c resistance isno ts oeasy .Progres s mayb e hampered by the following observations: a)incomplet e resistance canb e race-specific (Chapters 5an d8) , b)result sbetwee n resistance tests are sometimes inconsistent (Chapter 6), and c)diseas e level inth e field ispositivel y correlated with the yielding capacity (Chapter6) . Indications were obtained that still another approach fordurabl e resistance couldb eworthwhile .Tw o new races ofcoffe e leafrus t showed increased virulence to somecoffe e genotypes and decreased virulence to others (Chapter 5). This findingma ypoin tt o adiffi ­ culty for H. vastatrix tocombin e certainvirulenc e factors.There ­ fore, iti srecommende d to combine the resistance genes ofth ecor ­ responding coffee genotypes intoon e cultivar. Selection forthi s combined resistance mayb e relatively easybecaus e probably major genes are involved.

RELATIONSHIP TORESISTANC E CONCEPTS

Thepresen t results showparallel s with results reported from otherpathosystems .The y canb e explained with current resistance theories, asdiscusse dbelow .

The 'gene-for-gene' concept. According to the gene-for-gene concept there is foreac hresistanc e gene inth ehos t acorrespondin g gene forvirulenc e inth epathogen .Generally , the resistance (R)an dth e avirulence (A)allele s are dominant. Interactionsbetwee nhos tan d pathogen are supposedlybase d onrecognitio n betweenth edominan tR andA alleles,resultin g in incompatibility. Non-recognized alleles are expected tobehav e asthoug h theywer e asingl e recessive gene (Ellingboe, 1979). Followingthes epostulates ,tw o levels ofcompat ­ ibility canb e expected foreac hcorrespondin g R/A gene combination. Inth epast ,thes epostulate shav e created toomuc hth eimpres ­ siontha tresistanc e governed by the gene-for-gene relationship is exclusively aqualitativ e trait.Thi sgeneralizatio n appearst ob e based onth e facttha tR-gene s with agrea teffec to n resistance wereuse d inth estudie sunderlyin g thegene-for-gen ehypothesis . There ismuc h evidence thatquantitativ e resistance can alsob e governedb y thegene-for-gen e relationship.Thi s issupporte d in

125 partb y the research oncoffe e leafrust .Som eo fth e evidence is listedbelow . 1)Resistanc e and avirulence genesma yno tb e completely domi­ nant. Ifpartia l dominance occurswit hth eR-gen e and intermediate inheritancewit h theA-gene, the nu p to five levels of compatibility canb e expected between adiploi d hostan d adikaryoti c pathogen (Table1) .

Table1 .Level so fcompatibilit y( 0= completel yincompatible ,10 0= completel y compatible)assumin gintermediat einheritanc efo rth eavirulenc egen e(A ) and partialdominanc efo rth eresistanc egen e(R )i na gene-for-gen erelationship . Theeffect so fbot hvirulenc ean dresistanc eallele sar ethough tt ob eadditive .

Pathogen Host genotype genotype RR Rr rr

AA 0 20 100 Aa 50 70 100 aa 100 100 100

Incoffee ,resistanc e geneswer e foundwhic h showed intermediate inheritance orpartia l dominance (Chapters 3,4 and 8).Th e inter­ mediatevirulenc e foundwit h3 genotype s of H. vastatrix (Chapter 5) ismos t simply explainedb y assuming anintermediat ephenotyp e of theheterokaryo n carrying one avirulence and onevirulenc e allele. Several other examples inliteratur e indicate incomplete dominance ofresistanc e genes (e.g.Biffen , 1905)an d of avirulence genes (e.g. Statler andJones , 1981). 2)Man ymajo r genes forresistanc e areknow nwhic h dono tcom ­ pletely inhibitth e reproduction ofth epathogen .Eve nhos tdif ­ ferentials, chosenbecaus e of their qualitative reactions todif ­ ferentstrain s ofth epathogen , are oftenno tcompletel y resistant to avirulent races (e.g.Roelf s andMcVey , 1979). Furthermore,th e expression ofresistanc e genesma yb e influenced by test conditions such astemperature , light intensity andplan t age (Hooker,1967 , Lewellen and Sharp,1968 ,Stubbs ,1967 ,Vanderplank , 1978,Zadoks , 1961). Ina simila rway ,par to fth e race-specific resistance to coffee leafrus twa s affectedb y thetes tcondition s (Chapters 3,4 , 6 and 8). 3)Majo r genes for resistance mayb emodifie d by the genetic background ofth ehos t (Parlevliet andKuiper , 1977,Dyc k and Sam- barski,1968 ,Athwa l andWatson , 1954). Various levels of incomplete

126 resistance indifferen t coffee genotypeswer ematche d by one and the samene w race of H. vastatrix (Chapter 5), which also suggests an effecto fth e geneticbackgroun d onth eexpressio n ofmajo rgenes . Also,virulenc e genesmigh tb emodifie d by thegeneti cbackgroun d of thepathogen .Wid evariatio n invirulenc e toon emajo r resistance geneha sbee n observed,whic hwa s explained by allelism forviru ­ lenceo rb y interallelic interactions inth epathoge n (Schwartzbach, 1979, Statler and Jones,1981 ,Watso n and Luig, 1968). With coffee leafrus twid evariatio n invirulenc ewa s observed inrelatio nt o the S„3gen e (Chapter 5). Itseem s desirable to complement thecommo npostulate s forth e gene-for-gene hypothesis (Ellingboe,1981 )wit h the followingpostu ­ latewhic h describes quantitative effects ofR - andA-genes : 'Theexpression , i.e. themod e ofaction ,o fR - andA-gene s may be influenced by genedose ,geneti cbackground , and test conditions (environmental factors,conditio n ofth eplan ttissue ,an d the precisionwit hwhic h differences inresistance s aremeasured). '

The 'horizontal resistance' concept. This concept suggests that besides resistance governed by thegene-for-gen e relationship (ver­ tical resistance,VR )ther e is alsorace-non-specifi c resistance (horizontal resistance,HR) .H R isconsidere d tob e long lasting resistance (Vanderplank, 1968). Itsexpressio nwoul d generally be quantitative and its inheritancebase d onmino r genes (polygenes). Robinson (1976)ha s interpreted thepossibl e ecological valueo fH R andV R interm s ofth e eso-an d exodemic,respectively . Hisinter ­ pretation isbase d onth e supposition thatVR-gene s give acomplet e protection to allo-infection.Thi s supposition seems incorrectbe ­ causeV R canb e quantitative (seediscussio n gene-for-gene concept). TheH R conceptha s stimulated studies,suc h asth epresen tone , onth equantitativ e relationship betweenhos t andparasite .On e of themai ncriticism s to theH R conceptha sbee n thediscover y ofdif ­ ferential interactions betweenhos t andparasite s inpolygeni c sys­ tems (e.g.Parlevliet , 1979,Parlevlie t and Zadoks, 1977). Robinson (1979)argue s thatth e occurrence ofquantitativ e VR isn oproo f thatther e isonl y one kind ofresistance .Th emai n difficulty with theH R concept istha t the existence ofH R canneve rb e established experimentally. There canb e historical evidence onth e durability ofresistance ,bu tdurabl e resistance need notb e race-non-specific and canb ebase d onmajo r genes (Eenink, 1976,Johnso n andTaylor , 1976,Parlevlie t and Zadoks, 1977). Furthermore, race-non-specific resistance cantheoreticall y be overcomeb y genenon-specifi cpatho -

127 genicity or 'horizontalpathogenicity' . Therefore,th e definition of HR seems inadequate.A s alreadypointe d outb y Caldwell (1968)ther e is alson onee d forth eter mHR .Hence ,th e useo fthi s term should be avoided.

The 'durable resistance' concept. Although thewor d 'concept'ha s beenuse d inrelatio nt odurabl e resistance (Johnson, 1979), there is in factn o conceptbu tonl y adefinitio n ofth eter m durable resistance (long lasting resistance) and arecommendatio n tous e proven durable resistance inbreedin g programmes.Durabl e resistance should neverb e used asa ter m synonymous toHR . Withcoffe e somehistorica l evidence exists fordurabl eresis ­ tancet o H. vastatrix inC . arabica (inEthiopia )an d C. canephora (in Indonesia) (Eskes,1981 a). Iti sals o likely thatth e low level of incomplete resistance of 'MundoNovo' ,i ncompariso nt omor e sus­ ceptible coffee genotypes (Chapters 6an d 7), isdurabl e resistance. Thisresistanc e isexpresse db y arelativel y shortlea f retention period (LRP)an d low lesion density.Variatio n inLR Pmigh tb e largely race-non-specific.Histologica l evidence has indicated that resistance expressed by alo w lesion density isprobabl y based ona mechanism different fromhypersensitiv e cell death (Niks andKuiper , inpress) . Suchresistance , atlo w levels in 'MundoNovo ' and at high levels in 'Kouillou' (Chapter 7), may therefore be quitedura ­ ble.

The 'integrated resistance' concept and 'partial' resistance. Par- levlietan d Zadoks (1977)postulat e thatth e gene-for-gene relation­ ship applies alsot opolygeni c systems. Inthi s concept alltru e resistance genes havematchin gpathogenicit y genes.Hence ,race-non ­ specific true resistance orcultivar-non-specifi c pathogenicity wouldno toccur . Inha sbee n argued,however ,tha tther e is evidence forth eoccurrenc e ofnon-specifi c pathogenicity (Robinson,1976 , Eskes, 1981 ).Thi s canb e concluded for instance frompathogeni c variation occurring onhost swhic h dono tposses s any resistance genes, asreporte db yParlevlie t (1976 and 1978a)an db yNelso ne t al. (1970). Facultative parasites especially mayvar y in level of cultivar-non-specific pathogenicity. The occurrence ofcultivar-non - specific pathogenicity ishoweve r noproo f forth e occurrence of race-non-specific resistance. Polygenically inherited 'partial'resistanc e is likely tob e durable (Parlevliet and Zadoks,1977 ,Parlevliet , 1978b). Parlevl defines 'partial'resistanc e asresistanc e of asusceptibl e ora

128 high reaction typewhic h isexpresse d by resistance components like latencyperiod , lesiondensit y and sporeproduction .Wit h coffee leafrust , itha sbee n difficult to distinguishbetwee n 'partial' resistance andothe rtype s ofincomplet e resistance.Reactio ntyp e was affectedb y the testcondition s andb y leaf age (Chapters 3,4 , 7 and 8).Generally , reactiontyp ewa s related toth e resistance components. Some 'Kouillou'genotypes ,whic h resistance appeared to be complex, showedhig h levels of 'partial'resistanc e inmos t tests,bu t in afe wtest s lowo rheterogeneou s reactiontype swer e observed.Thi s suggests thathig h levels ofpolygeni c resistance might alsob e expressed by alo wo rhetergeneou s reactiontype .

'Horizontal resistance' based on defeated major genes for resis­ tance. Defeated major genes forresistanc e mayhav e a 'residual' effect (e.g.Marti n andEllingboe , 1976).Accordin gt oNelso n (1978), acombinatio n ofdefeate d major genes couldb e responsible for 'horizontal resistance'.A restrictiono fthi sconcep t appears tob e thatno t allmajo r genesma yhav e residual effects (Nasse t al., 1981)an d thatth e degree ofth e residual effectma y depend on thepathoge n genotype (e.g.Marti n andEllingboe , 1976). There ­ searchwit h coffee leafrus t indicates the same.N o apparent resi­ dual effectswer e observed forth egene s SH1, S„2 and S„4.Wit h other resistance genesvaryin g levels ofvirulenc ewer e observed for differentmatchin g rustgenotype s (Chapter 5)indicatin gtha tth e level ofresidua l resistance may depend onth e rustgenotype .Resid ­ ual resistancemigh tcontribut e todurabl e resistance only ifth e residual effect isbase d oninteraileli c interactions forvirulenc e oro n anegativ e effecto fvirulenc e genes onth e fitness ofth e pathogen.

Host-pathogen-environment interactions. Complexinteraction sbe ­ tweenhost ,pathoge n and environmenthav ebee n observed in several pathosystems (e.g.Eskes ,198 1 ).Suc h interactions may help to stabilizepathoge npopulation s (Chandrashekar andHeather , 1981), becausepathoge n racesma yhav e aselectiv e advantage under certain conditionsbu tn oo r anegativ e selective advantage under other conditions. Incomplete resistance isgenerall y more affectedb y environment orplan t factors than complete ormajo r gene resistance. Individual minor genes,contributin g topolygeni c incomplete resistance,migh t be affected in adifferen twa yb y environment orplan t factors.Thi s means thatno t allmino r genesma yb e effective under allconditions .

129 Then, selectionpressur e for increased virulence tomino r genes would notb e constant.Thi sma yhel p toexplai n durability ofpoly ­ genic incomplete resistances. The aboveconsideration s may also apply to thecoffee/ff . vasta­ trix relationship,becaus e incomplete resistance to H. vastatrix is considerably affectedb y thetes t conditions (Chapters 3 and 4).Th e effectiveness ofresistanc e to coffee leafrus tvarie d from leaft o leaf andeve nwithi n leaves,a s indicated by the frequent occurrence Ofheterogeneou s reaction types.Suc h avariabl e resistance reaction tocoffe e leafrus tmigh t compensate forth e relative shortage of geneticvariabilit y intim eo fth eperennia l coffeecrop .

Dissociation and association of virulence. Vanderplank (1982)sug ­ gested thatvirulenc e tends to either dissociate or associate.Wit h otherwords ,virulenc e toon eresistanc e genewoul dpromot eviru ­ lence or avirulence toothe r resistance genes.Stabilizin g selection (Vanderplank, 1968), in the context ofth ene whypothesis ,coul d result from dissociation ofvirulenc e factors inth e exodemic.Th e hypothesis isbase d onobserve d discrepancies between expected and found frequencies ofcombination s ofvirulenc e factors matching wheat stem rustresistanc e genes inAmeric a andCanada . Onepossibl e argument againstVanderplank 1s hypothesis istha t discrepancies between expected and foundvirulenc e frequencies might alsob e explained by absence ofrando m mating in thepathoge npopu ­ lation.Clearly ,mor e researchwil lb e needed alsowit h otherpatho - systems to comet o aconclusio n aboutth evalu e ofVanderplank' s hypothesis. A consequence ofdissociatio n ofvirulenc e factorswoul db e that thecombinatio n ofcertai nmajo r resistance genes canb e valuable forobtainin g durable resistance.Wit h resistance to coffee leaf rustther e is some evidence thatsuc h combinationsmigh tb e found. The strongest evidence comes from the rustrac e classification work done atth eCoffe eRust s Research Center (CIFC)i nPortuga l (Rodri- gues et al., 1975). At CIFC sevenrus t races ofdifferen t origins showvirulenc e toth e C. congensis 263/1 differential.Al l these races areonl ymoderatel y virulent to C. arabica differentials and lackvirulenc e toth e S„5resistanc e gene,wherea svirulenc e toS„ 5 iscommonl y found inothe r rust races. These results couldb eex ­ plainedb y assuming dissociation ofvirulenc e in H. vastatrix. Thepresen tresult s also suggest adifficult y for H. vastatrix to combine certainvirulenc e factors.Tw one w rustrace s showedin ­ creasedvirulenc e tosom egenotype s and loweredvirulence ,o raviru -

130 lence, toothers .Assumin g thatthes e races aroseb y single gene mutations from thewil d typerace ,the nthes e results could alsoin ­ dicate dissociation ofcertai nvirulenc e genes in H. vastatrix.

Breeding for durable resistance. Thebreedin g approach advised by the 'horizontal' resistance concept isclearl ymor e complicated than initially suggested. Firstly itma yb e questioned whether race-non­ specific resistance exists atall .Secondly , there isn owa yt o discern suchresistanc e inbreedin gprogrammes .Th ebes t suggestion wouldb e to select forpolygeni c resistance (Parlevliet, 1978 ).Th e strategybase d onselectio n inpopulation s derived from crosses between susceptible parents (Robinson, 1976)seem s suitable forth e accumulation ofmino r genes.Amon g others,Krupinsk y and Sharp (1979)observe d transgression towardshig hresistanc e inprogenie s from crossesbetwee n susceptible parents.On ewoul d expect such resistance tob e durable if additivemino r genes are involved. However,hig h levels of 'transgressive' resistance could alsob e based onmino r genes enhancing the expression ofrace-specifi c major genes. Researchwhic h distinguishes betweenbot hpossibilitie s seems desirable. Parlevliet (1978 )suggest s thatpolygeni c resistance against somecerea l rusts could be achievedb y selecting for incomplete resistance of ahigh , susceptible reaction type ('partial'resis ­ tance). With leafrus to fbarle y this strategy appears tob erela ­ tively easy,bu twit h otherpathosystem s the situationma yb emor e complicated (Parlevliet, 1979). There appears tob e adifficult y in clearly separating 'partial'resistanc e from other types ofincom ­ plete resistance (Eskes,198 1) . Breedingbase d onresistanc e thatha sprove d tob e long lasting (Johnson, 1979)ma y alsohav epractica l difficulties.Durabl eresis ­ tance canb ebase d on aneffectiv e combination ofresistanc e genes whichb y themselves dono tprovid e durable resistance. Suchcombina ­ tionsma yb e lost inth ebreedin gprocess .Base d onth e assumption thatdissociatio n ofvirulenc e may occur inth epathogen , Vanderplank (1982)suggeste d that asearc h shouldb e initiated for effective combinations ofresistanc e genes.A s forresistanc e to coffee leaf rust, itcoul db e thatther e are suchcombination s (Chapter5) . A final conclusion istha ti nbreedin g fordurabl e resistance the probabilities of success of individual strategies shouldb eweighte d againstth enee d forobtainin g such resistance.Then , for different pathosystems different approacheswil lb e recommended. Durableresis ­ tancet o Hemileia vastatrix iso f such importance forth ecoffe e crop

131 thatth ebreedin g approaches initiated asa resul to fth epresen t research deserve continuation, eventhoug h success cannotb e fully warrented.

132 REFERENCES

Athwal,S.D .& I.A .Watson ,1954 .Inheritanc ean dth egeneti c relationshipo f resistancepossesse db yth etw oKeny awheat st orace so fPuccini agraminis . Proc.Linn .Soc .N.S.W. ,79 :1-14 . Biffen,R.H. ,1905 .Mendel' s lawso finheritanc ean dwhea tbreeding .J .Agric . Sei. 1:4-48 . Caldwell,R.M. ,1968 .Breedin g forgenera land/o rspecifi cdiseas eresistance . Proc.3rd .Int .Whea tGen .Symp. ,Canberra ,Australia ,pp .263-272 . Carvalho,A. ,F.P .Ferwerda ,J.A .Frahm-Leliveld ,D.M .Medina ,A.J.T .Mende s& L.C.Monaco ,1969 .Coffee .In :Outline so fperennia l cropbreedin g inth e tropics.H .Veenma n& Zone nNV ,Wageningen .Ed .F.P .Ferwerd a& F .Wit .pp . 189-241. Chandrashekar,M .& W.A .Heather ,1981 .Temperatur e sensitivityo freaction so f Populusspp .t orace so fMelampsor a larici-populinaKleb .Phytopatholog y71 : 421-424. Dyck,P.L .& D.J . Samborski,1968 .Host-parasit e interactions involvingtw ogene s forlea frus tresistanc e inwheat .Proc .3r dInt .Whea tGen .Symp. ,Canberra , Australia,pp .245-250 . Eenink,A.H. ,1976 .Genetic so fhost-parasit e relationshipsan dunifor m anddif ­ ferential resistance.Neth .J .PI .Path .82 :133-145 . Ellingboe,A.H. ,1979 .Inheritanc eo fspecificity :th egene-for-gen ehypothesis . In:Recognitio nan d specificity inplan thost-parasit e interactions.Ed . J.M.Dal yan dI .Uritani .Un .Park .Press ,Baltimore ,pp .3-17 . Ellingboe,A.H. ,1981 .Changin g concepts inhost-pathoge ngenetics .Ann .Rev . Phytopath.19 :125-143 . Eskes,A.B. ,1981a .Incomplet e resistancet ocoffe elea frust .In :Proc .NAT O AdvancedStud yInstitut eo nDurabl eResistance .Martin aFranca ,Italy ,Octobe r 1981.I npress . Eskes,A.B. ,1981b .Expressio no fincomplet e resistance topathogens .In :Proc . NATOAdvance d StudyInstitut eo nDurabl eResistance .Martin aFranca ,Italy . October 1981.I npress . Hooker,A.L. ,1967 .Th egenetic san dexpressio no fresistanc e inplant st orust s ofth egenu sPuccinia .Ann .Rev .Phytopath .5 :163-182 . Johnson,R. , 1979.Th econcep to fDurabl eResistance .Phytopatholog y 69:198 . Johnson,R .& A.J .Taylor ,1976 .Geneti c controlo fdurabl e resistancet oyello w rust (Pucciniastriiformis )i nth ewhea tcultiva rHybrid ed eBersee .Ann . Appl.Biol .81 :385-391 . Krupinsky,J.M . &E.L .Sharp ,1979 .Reselectio n forimprove d resistanceo fwhea t tostrip erust .Phytopatholog y69 :400-404 . Lewellen,R.T .& E.L .Sharp ,1968 .Inheritanc eo fmino r reactiongen ecombina ­ tionsi nwhea tt oPuccini a striiformisa ttw otemperatur eprofiles .Can .J . Bot.46 :21-26 . Mariotto,P.R. , P.Figueredo ,A.P .d aSilveira ,C .Gerald oJunior ,H.V .d e Arruda,H .Lopes ,E.G .d eOliveira ,L.F.S .Buen oJunio r& N.L .d eOliveir a Filho,1979 .Estudo ssobr eo control equimic od aferruge md ocafeeir o(Hemi - leiavastatri xBerk .e tBr. )e seu sefeito sn aproduçao ,na scondiçôe sd o Estadod eSâ oPaulo .Biológic o45 :165-174 . Martin,T.J .& A.H .Ellingboe ,1976 .Difference sbetwee ncompatibl eparasite/hos t genotypes involvingth ePm 4locu so fwhea tan dth ecorrespondin g genesi nEry - siphegramini s f.sp. tritici.Phytopatholog y 66:1435-1438 . Monaco,L.C , 1977.Consequence so fth eintroductio no fcoffe elea frus tint o Brazil.Ann .Ne wYor kAc .Soc .287 :57-71 . Nass,H.A. ,W.L .Pedersen ,D.R . MacKenzie &R.R .Nelson ,1981 .Th eresidua lef ­ fecto fsom edefeate dpowder ymilde wresistanc egene si nisoline so fwinte r wheat.Phytopatholog y 71:1315-1318 . Nelson,R.R. , 1978.Genetic so fhorizonta lresistanc et oplan tdiseases .Ann . Rev.Phytopath .16 :359-378 . Nelson,R.R. ,MacKenzie ,D.R .& G.L .Scheifele ,1970 .Interactio no fgene sfo r pathogenicityan dvirulenc e inTrichometasphaeri a turcicawit hdifferen t numbero fgene sfo rvertica lresistanc ei nZe amays .Phytopatholog y60 : 1250-1254.

133 Niks,R.E .& H.J .Kuiper ,i npress .Histolog yo fth erelatio nbetwee nmino ran d majorgene sfo rresistanc e ofbarle yt olea frust .Phytopathology . Parlevliet,J.E. ,1976 .Evaluatio no fth econcep to fhorizonta l resistancei n thebarley/Puccini a hordeihost-pathoge nrelationship .Phytopatholog y66 : 494-497. Parlevliet,J.E. ,1978a .Furthe revidenc eo fpolygeni c inheritanceo fpartia l resistance inbarle yt olea frust ,Puccini ahordei .Euphytic a 27:369-379 . Parlevliet,J.E. ,1978b .Aspect san dproblem swit hhorizonta l resistance.Cro p Improv.5 :1-10 . Parlevliet,J.E. ,1979 .Component so fresistanc etha treduc eth erat eo fepidemi c development.Ann .Rev .Phytopath . 17:203-222 . Parlevliet,J.E .& H.J .Kuiper ,1977 .Resistanc eo fsom ebarle ycultivar st olea f rust,Puccini ahordei ;polygeni cpartia l resistancehidde nb ymonogeni chyper ­ sensitivity.Neth .J .PI .Path .83 :85-89 . Parlevliet,J.E .& J.C .Zadoks ,1977 .Th eintegrate d concepto fdiseas eresis ­ tance;a ne wvie wincludin ghorizonta l andvertica l resistance inplants . Euphytica 26:5-21 . Robinson,R.A. , 1976.Plan tPathosystems .Springer ,Berlin ,Heidelberg ,Ne wYork . Robinson,R.A. , 1979.Permanen tan d impermanentresistanc et ocro pparasites ;a re-examinationo fth epathosyste m conceptwit hspecia lreferenc et oric e blast.Z .Pflanzenziichtg .83 :1-39 . RodriguesJr. ,C.J. ,A.J .Bettencour t& L .Rijo ,1975 .Race so fth epathoge n andresistanc et ocoffe erust .Ann .Rev .Phytopath .14 :49-70 . Roelfs,A.P .& D.V .McVey ,1979 .Lo winfectio ntype sproduce db yPuccini agra - minisf.sp . tritician dwhea tline swit hdesignate d genesfo rresistance . Phytopathology 69:722-729 . Schwartzbach,E. ,1979 .Respons et oselectio nfo rvirulenc eagains tth eml- 0 basedmilde w resistance inbarley ,no tfittin gth egene-for-gen econcept . BarleyGenetic sNewslette r9 :85-88 . Simons,M.D. ,1972 .Polygeni c resistancet oplan tdisease san d itsus ei nbreed ­ ingresistan tcultivars .J .Environ .Qual .1 :232-239 . Statler,G.D .& D.A . Jones,1981 .Inheritanc eo fvirulenc ean duredia lcolo ran d sizei nPuccini arecondit atritici .Phytopatholog y 71:652-655 . Stubbs,R.W. ,1967 .Influenc eo fligh tintensit yo nth ereactio no fwhea tan d barleyseedling st oPuccini astriiformis •Phytopatholog y 57:615-619 . Vanderplank,J.E. ,1968 .Diseas e resistance inplants .Acad .Press .Ne wYork . Vanderplank,J.E. ,1978 .Geneti can dmolecula rbasi so fplan tpathogenisis . SpringerVerlag .Berlin ,Heidelberg ,Ne wYork . Vanderplank,J.E. ,1982 .Host-pathoge n interactions inplan tdisease .Acad . Press. Watson,I.H .& N.H .Luig ,1968 .Progressiv e increasei nvirulenc e inPuccini a graminis f.sp.tritici .Phytopatholog y 58:70-73 . Zadoks,J.C , 1961.Yello w rusto nwheat ,studie si nepidemiolog yan dphysiologi c specialization.Tijdschr .PI . Ziekten67 :69-256 .

134 SAMENVATTING

Onderzoek naar incomplete resistentie tegenveroorzaker s van planteziekten kanwaardevo l zijnvoo r hetverkrijge nva n duurzame resistentie.Bovendie n leverthe tee nbijdrag e totee nbete r begrip vanwaard-pathogee n relaties,omda td ekenni shieromtren tveela lbe ­ perkt isto t resistentie veroorzaakt door 'major'genen . Incomplete resistentie tegenkoffieroes t (Hemileia vastatrix Berk.e tBr. )i s bestudeerd bij driekoffiepopulaties : Coffea arabica L., C. cane- phora Pierre cv.Kouillo u end epopulati e 'Icatu'. Icatustam ta f van eenkruisin g tussen C. canephora en C. arabica. Methoden zijnontwikkel d omhe tnivea uva n resistentie tegen koffieroest tebepale n (Hoofdstuk 1). Voor hetbepale nva n deziek - tegraad inhe tveld , ind eka se n inhe t laboratorium werden4 geïl­ lustreerde schalenontworpen ,di eva n 0to t 9lopen .Teven sword t hetgebrui kva n een 0to t 9schaa lvoorgestel d voor hetbepale nva n hetreactietyp e van koffieroest.Dez e schaalbeva tee nbred e reeks vanheterogen e reactietypen dievaa kvoorkomen ,voora l inkoffiepo ­ pulaties metresistentiegene nva n C. canephora. De doelmatigheid van deze schalenword t aangetoond. Een laboratoriumtoets,waarbi j gebruik gemaaktword tva nblad ­ schijvenme tee ndiamete rva n 1,8 cm,blee k geschiktt ezij nvoo r hetbepale nva n incomplete encomplet e resistentie (Hoofdstuk2) . Bij 20 genotypenva n 'Kouillou',me tee nvariëren d niveauva n incom­ plete resistentie,wer d de laboratoriumtoets inviervou d toegepast. Metd e resistentiecomponenten gemeten indez etoetse nblee k 79%va n dewaargenome n variantie voorveldaantastin g van degenotype nt e verklaren te zijn.Onderzoeksresultate n worden gegeven die geleid hebben totd e standaardisatie van de laboratoriumtoets. Lichtintensiteit (LI)blee k de resistentie tegenkoffieroes tt e beïnvloeden (Hoofdstuk 3). Overhe t algemeenware nbladere n dieon ­ der eenlag e LI groeien resistenter danbladere n die onder eenhog e LI groeien.Ee nhog e LIvóó r inoculatieverhoogd e deaantastings ­ graad, terwijl eenhog e LIn ä inoculatie de aantastingsgraad ver­ laagde.Verschille n inresistenti e als gevolgva nd e verschillende LI-behandelingen uitten zichvoora l doorverschille n inlesiedicht - heid (LD)e n reactietype (RT).He t resistentiegen S„4wa svolledi g dominant ind eka s (lage LI), maarwa s onvolledig dominanto fzelf s bijna recessief ind ekwekeri j (hoge LI).E r kannie t geconcludeerd wordenui td eresultate n datbeschaduwin g van koffie-aanplanten automatisch zal leidento tee nverlagin gva n de ziektegraad inhe t veld.O fdi ta lda nnie tgebeur t zal ondermee r afhangenva n demat e

135 vanbeschaduwin g enva nhe tkoffiegenotype . InHoofdstu k 4word the teffec tva nd ebladleeftij d opresisten ­ tiebeschreven .Di teffec tvarieerd e per genotype.Bi j devatbar eC . arabica rassenwa s er slechts eengerin g effect. Fysio-specifieke incomplete resistentie in C. arabica eni nd e Icatupopulati e uitte zichvoornamelij k inhe t jongereblad ; ouderebladere nware nvat ­ baarder.Dez e resistentieverschillen waren gerelateerd aanverschil ­ len inRT . Inhe tra sKouillo uwerde n drie fasenva nbladleeftij d t.o.v.vatbaarhei d waargenomen:volwasse nbladere nware n resistenter danjong eo foud ebladeren .D e resistentieva nhe tvolwasse nbla d uitte zichvoornamelij k dooree n lagere LD,maa r ook enigszins door een lagerRT . InHoofdstu k 5word td e detectie en identificatie van 7roestfy - sio'sbeschreve n die gevonden zijn ind everedelingsproefvelde n in Campinas.E rwerde nkwalitatiev e enkwantitatiev everschille n in pathogeniteitwaargenomen .Vie r fysio's doorbreken combinaties van bekende resistentiegenen in C. arabica. Drie fysio's doorbreken de resistentieva nno gonbekend e genenui tC . canephora. Tevenswerde n 3 roestgenotypen gevondenme t intermediairevirulentie .D e mogelijke implicaties vand eresultate nvoo r deveredelin g op roestresistentie wordenbesproken . Ind eHoofdstukke n 6, 7e n8 word td e incomplete resistentiete ­ genkoffieroes tvan ,respectievelij k C. arabica, C. canephora cv. Kouillou en Icatu gekarakteriseerd. In C. arabica bleek genetische variatie aanwezig te zijnvoo r de periodeva nbladretenti e nadatd esporulati ebegonne n is (LRP). LRP bepaaltd eduu rva n sporulatie.Lang e LRPwaaarde nware ngecorre ­ leerdme tee nhog e ziektegraad inhe tveld . Ookwa s er significante variatie aanwezig in C. arabica voord e latentieperiode (LP)e nLD , maar de resultatenvoo r deze resistentie componentenware n enigszins inconsistent,voora l voor LD.D e correlatiecoëfficiënten tussenL P ofL De nd eziektegraa d inhe tvel dware n laag enove rhe t algemeen nietsignificant .Ee nhog eproduktiecapacitei t bleekpositie fgecor ­ releerdme tee nhog e ziektegraad inhe tveld .Di twer d waargenomen bij lijnenva n C. arabica rassen enbi j introducties uitEthiopië . De conclusie isda tselecti evoo rverhoogd e incomplete resistentie in C. arabica nieter gefficiën t zalzijn . Erwer d eengrot evariati evoo r incomplete resistentiewaargeno ­ mentusse ngenotype nva n 'Kouillou'.Ee n lage ziektegraad inhe t veldblee k gerelateerd tezij naa nee n laag,o fee nheterogee nR T en/of aanee n lageLD .Complet e en incomplete fysio-specifieke re­ sistentieswerde n aangetroffenbi j sommige genotypen.D e vererving

136 van de resistentie van sommige andere genotypenme tee nlag eziekte - graad inhe tvel dblee k complex.Dez e genotypenvertoonde n geenin ­ teractieme td eroestfysio' sdi egeïsoleer d werdenva n 'Kouillou'. Hun resistentie uittezic hdoo ree n lageL De nee nrelatie fhoo gRT . Daarom hebben dezegenotype nmogelij k waardebi jhe tverkrijge nva n duurzame resistentie tegen koffieroest.Ee nveredelingsprogramm a gebaseerd op kruisingen vandez egenotype nme tC . arabica werd gestart. De incomplete resistentie in Icatublee kvoora l gerelateerd aan RT.Veela l was de resistentiereactie heterogeen.D e resistentiecom­ ponenten,waaronde r LP,ware ngecorreleer d metRT . Incompleteresis ­ tentieva nverschillen d niveaublee k fysio-specifiek.D e conclusie isda t selectievoo r incomplete resistentie in Icatuwellich tnie t zal leidento the tverkrijge nva nduurzam e resistentie. Ind ealgemen e discussie,aa nhe tein dva n ditproefschrift , wordthe tbelan g vand everkrege n resultaten voor hetkoffieroest - onderzoekbediscussieer d enword te rverban d gelegdme t algemene resistentietheorieën.

137 RESUMO

Pesquisas sobreresistênci a incompleta contra agentes patogénicos podem tervalo rpar a aobtença o de resistência duradoura.Alé m disso, estaspesquisa s contribuiräopar au mmelho rconheciment od a releçaoentr epatógen oe hospedeiro , oquai ,muita svezes ,é restrit o aos genesmaiore sd e resistência.Nest a tese,a resistênci a incom­ pleta contra a ferrugemd o cafeeiro (Hemileia vastatrix Berk. et Br.)fo iestudad a em trèspopolaçoes : Coffea arabica L, C. canephora Pierre cv.Kouillo u e 'Icatu'. Icatué um apopulaça o de café hibrida derivada deu mcruzament oentr e C. arabica e C. canephora. Métodos foramdesenvolvido s para determinar ogra ud e resistência emcondiçoe sd e campo,estuf ae laboratório,co m us od eescala s de 10ponto s (Capitulono .1) .Cad avalo rnesta s escalas de 0a 9 recebeu umadescriça o e foi ilustrado com desenhosestilizados , indicando um aumento exponencial da incidência du ferrugem. Descreveu-se tambémum a escala de 0a 9 par a adeterminacä od o tipo dereacä od a ferrugem.Est a escala contemu m amploexpectr o detipo s dereacä oheterogenea , osquai s ocorremco m frequenciae mpopulaçôe s com resistência derivada de C. canephora. Autilidad eprätic ada s escalas de 0a 9 foidemonstrada . Desenvolveu-seu mmétod od elaboratóri opar amedi r resistência emdisco s de fôlhaco m diametro de 1,8 cm.Est emétod oprovo u ser util nadeterminacä od eresistênci a complétae incomplet a (Capitulo no. 2).O métod o foi aplicado quatroveze s emvint e genótipos de 'Kouillou',o squai s temu mnive lvariâve l de resistência. 0 coeficiente decorrelaçâ omûltipl a (R)entr e osvalore smédio s de seis componentes deresistência , observados no laboratório eo graud e ataquen o campo, foi altamente significativo (R= 0.89). Forammostrado s osresultado s daspesquisa s que levaram auni - formizar estemétodo . A intensidade da luz (LI)influenci a na resistência contra H. vastatrix (Capitulono .3) .Fôlha s docafeeir o quecresce m comL I baixa säoe m géraimai s resistentes doqu e fôlhas quecresce m com LI alta.L I alta antesd a inoculaçâo induziu um aumenton o graud e ataque,enquant o LI alta após ainoculaçâ otev eu m efeitooposto , provavelmente porque corna alt a luminosidade opatógen omorr e dentro da fôlha.O sdiferente s tratamentos deL I induziramvariacöe sn a densidade das lesöes (LD)e n otip o dereacä o (RT).O geneS H4 comportou-se comour ngen e dominante em condiçoes deestuf a (LI baixa), mas foiquas e recessivo em condiçoes deviveir o (LI alta). Os resultados näopermite m aconclusä o dequ e o sombreamento ira

138 diminuiro gra ud e ataque da ferrugem emplantacöe ssuscetiveis . Se istoocorr e dependerâ,entr e outres fatores, da intensidade da sombra ed o genótipo docafeeiro . No capitulo no.4 ,é descrit o oefeit o da idaded a fôlhasobr e a resistência.Est e efeito dépended o genótipo do cafeeiro.Co m cultivares suscetiveis de C. arabica oefeit o foipequeno .A resis ­ tência incompletae especific a em C. arabica e Icatusalientou-s e maisna s fólhasnova s do quena s fôlhasvelhas .Houvera m diferenças significativespar aRT .N o 'Kouillou'distinguiu-s e tres fases de suscetibilidade em relaçao àidad e da fôlha.Fôlha s adultas tiveram menos lesöese , asvezes ,u mR Tmai sbaix o do que fôlhas novas ou velhas. Entre 1976 e 1981,fora m detectadas sete raçasnova s de H. vasta- trix naEstaçâ oExperimenta l do InstitutoAgronômic oe mCampinas . Observaram-se diferenças qualitativas equantitativa s emvirulêneia . Quatro raças quebraram resistênciascomplexa sn o C. arabica, constituidas por combinaçôes dosgene s SR1, S„2e S„4.Trè s raças quebraram aresistênci a degene sdesconhecido s de C. canephora. Ainda foram detectados 3genôtipo s de H. vastatrix cornvirulênei a intermediâria. Säodiscutida s aspossivei s consequêneiasdo s resul- tadospar a omelhorament o visando resistência contra a ferrugem. Nos capitulos 6, 7e 8 , caracterizou-se aresistênci a incompleta contra a ferrugem encontrada em C. arabica, C. canephora cv.Kouil ­ lou e Icatu,respectivamente .E n C. arabica detectou-sevariaça o geneticapar a operiod o deretença o das fôlhas (LRP). LRPdétermin a o periodo deesporulaçao .Valore s altos de LRP tiveram relaçaoco m intensidades altos da ferrugem no campo.Detectou-s e tambémvariaça o significativa para operiod o de latência (LP)e par a LD.Porém , houve certa inconsistênciaco m estes componentes deresistência , especialmente para LD.O s coeficientes decorrelaçâ oentr e LPo uL D e ogra u de ataquen ocampo , forambaixo s ee m gérainä o significa­ tives.A produtividad e mostrou seru m fatord e importâneia.Obteve - secoeficiente s decorrelaçâ opositivo s entre aprodutividad e eo graud e ataque nocampo ,no s linhagens de cultivares Brasileiras e nas introduçôesd aEtiópia .Conclui-s e que aseleça opar aresis ­ tência incompleta em C. arabica näo terâum a altaeficiência . En 'Kouillou'observou-s e umagrand evariaça opar a resistência incompleta.Genôtipo s compouc o ataque da ferrugem no campo,mostra - ramu mbaix o RTo uR Theterogêneo , ou também uma LDbaixa .Algun s genôtipos cornresistênci a incompletamostrara m interaçôesdiferen - ciais com duas raçasnova s de H. vastatrix, isolados de 'Kouillou'. A herança daresistênci a incompleta de alguns outros genôtipos

139 parece sercomplexa .Este s genótiposnä omostrara m interaçaoco m as raçasnova s de H. vastatrix. Eies tinham em géraium aL Dbaix a eu m RTalto .Po r issoestima-s e queeste s genótipospode m tervalo r na obtençâo deresistênci a duradoura contra a ferrugem do cafeeiro.Fo i iniciado umprogram ad emelhorament obasead o emcruzamento sentr e genótipos de 'Kouillou'e C. arabica. A resistência incompleta em Icatu foi relacionada em grandepart e com RT.E m geralo R T eraheterogêneo .O s componentes deresis ­ tência,com oLP , foramcorrelacionado s com RT.Genótipo s com niveis variâveisd e resistência incompleta mostraram interaçao comum a raça novad e H. vastatrix, isolado de Icatu.Chegou-s e aconclusä o dequ e a seleçaopar a resistência incompleta no Icatu,provavelment enä o levara aum a resistência duradour. Nadiscussä o geral da tese,a importânci a dos resultadospar ao melhoramento do cafeeiro éavaliad a eo sresultado s säocomparado s com teórias sobreresistência .

140