224 Plant Disease / Vol. 83 No. 3(double-tree plots) per treatment. The ex-perimental unit was a tree, and the sametrees were used in both cycles.The HT treatment included removal ofsymptomatic shoots from the previouscycle (30 cm below the lowest diseasedshoot in a branch); nine fungicides sprays(copper oxychloride at 2.6 g a.i./liter [C],benomyl at 0.25 g a.i./liter [B], and man-cozeb at 4 g a.i./liter [M]) in the sequenceC(6)-B(2)-M applied singly in successionat fortnightly intervals from before flow-ering until fruit set; control of ants(Hymenoptera: Formicidae) (methyl para-thion at 8 g a.i./ha); and addition ofchicken manure (2.5 kg/tree once a year).The LT consisted of sanitary pruningsimilar to the previous treatment and twoalternated fungicide sprays (benomyl at0.25 g a.i./liter and mancozeb at 4 ga.i./liter) applied at fortnightly intervalsduring panicle development. The IM con-sisted of removing of diseased shoots (80cm below the lowest diseased shoot); fourfungicide sprays (copper oxychloride at 2.6g a.i./liter) applied at monthly intervalsduring the vegetative period and threesprays (captan at 1.5 g a.i./liter, benomyl at0.25 g a.i./liter, and mancozeb at 4 ga.i./liter) in succession applied at fort-nightly intervals from before floweringuntil fruit set; five applications of an acari-cide (sulfur 3.6 g a.i./liter) applied atmonthly intervals during the vegetativeperiod; and control of ants and addition ofchicken manure as before. For both, HTand IM, 3% potassium nitrate (KNO3) (18)was applied in water to the whole canopyto promote uniformity in flowering. Fruitpicking, done in April and May, and stan-dard cultural practices such as irrigation,fertilization, one general insecticide spray(malathion at 1.5 ml a.i./liter), and weedelimination were the same for all treat-ments.Disease assessment. For the purpose ofdisease assessment, 5 of 10 trees per treat-ment were selected in 1993-94. In 1994-95,one additional tree was evaluated. Fourbranches at the four cardinal points werelabeled per tree canopy. During each as-sessment, the total number of healthy anddiseased shoots (vegetative and floral) wascounted on each branch and averaged overthe four branches per tree. The diseaseprogress was determined as the accumu-lated proportion of diseased shoots per tree(Yic) corrected for host growth. At eachtime, i, Yic was calculated as: Yic = Yi/N, inwhich, Yi is the accumulated number ofdiseased shoots at time i; N is the totalnumber of shoots produced in the growingcycle. Evaluations were carried outmonthly in the vegetative stage (June toDecember) of mango growth and weeklyduring panicle development (January toFebruary).Temporal analyses. Disease progressdata were corrected initially for maximumdisease incidence (Ymax) (2,15), with arbi-trary values of 0.1, 0.5, and 0.7. Only oneYmax value, 0.7, was selected to correct allepidemic curves for purposes of compari-son. The apparent infection rate (r) wasestimated with the slope parameter of thelinearized forms of the monomolecular,Gompertz, and logistic models fitted withthe least square method of the GLM proce-dure of SAS (Release 6.03, SAS Institute,Cary, NC). The best model was selected byexamining the proportion of variance ex-plained (r2) and by plotting standardizedresidual versus predicted values (2). Afterthe best model for each epidemic was se-lected, the values of the slope parameterfrom the various models were transformedonto a standardized scale (rs) through theuse of the weighted mean absolute rate ofdisease incidence as the ρ-parameter of theRichard’s model (25) for the overall linearmodel that was selected most frequently.The epidemics also were characterizedwith the Weibull distribution functionmodified as a two-parameter model(20,26,33). The b parameter is related in-versely to apparent infection rate (r), andthe c parameter is related to the shape andslope of the density function (dy/dt perunit) for disease progress curve. The esti-mation of b and c was done empirically bymeans of the interactive process of nonlin-ear regression with the DUD method(SAS, Release 6.03, SAS Institute).Epidemics were also characterized bythe area under the disease progress curve(AUDPC) of malformation estimated withthe trapezoidal integration method.AUDPC was standardized for differentialepidemic duration as AUDPCS =AUDPC/Tt in the 1994-95 growth cycle,where Tt was length of duration of theepidemic (2). Other curve parameters in-cluded initial (YO) and final disease inci-dence (Yf).Effect of treatments. Analysis of vari-ance was performed on the r, YO, c, b–1, Yf,AUDPC, AUDPCS, and yield of mango foreach growing cycle. The Student-Newman-Fig. 1. Disease progress curves of malformation of mango cv. Haden in North Guerrero, México,during 1993-94 and 1994-95. HT = treatment with high technology; LT = treatment with low tradi-tional technology; and IM = integrated management. 1Jan to 4Jan = first to fourth week of January.
Plant Disease / March 1999 225Keuls (SNK) was used to separate treat-ment means, provided that the F value ofthe ANOVA was significant (31).Climatic data. Wind speed, air tem-perature, and relative humidity were meas-ured at the canopy level using an ane-mometer and hygrothermograph (both byWeather Measure Corporation, Sacra-mento, CA), respectively. Measurementswere recorded daily during 1 week permonth during the vegetative stage anddaily during panicle development.Fungus isolation and trapping ofspores. Fungus isolations were made from15 malformed shoots and 10 asymptomaticshoots from the experimental orchard.These shoots were kept in plastic bags at10°C until used, which usually was within48 h after shoot removal. Five to sevenpieces (4 to 6 mm) of tissue were takenfrom each of the malformed and asympto-matic shoots. The tissue was surface-disin-fested with sodium hypochlorite (0.5%) for5 min, rinsed three times in sterile distilledwater, and dried with paper toweling. Thetissue was aseptically transferred to 3.9%potato dextrose agar (PDA) (Difco Labo-ratories, Detroit, MI) in petri dishes andplaced under natural illumination at roomtemperature. Single-conidium transferswere made to PDA from developingFusarium colonies for identification pur-poses (16).A volumetric spore trap with a 7-dayrecord (9) was used to estimate conidialabundance in the air. The spore trap wasplaced at 60 m from the east side of theorchard border and 15 m to the south of theexperimental plots, at 2-m height in thecanopy. A clear tape on which spores weredeposited was cut into 39.5-mm sectionscorresponding to each 24-h period andmounted on microscope slides. Slides wereexamined at ×600 magnification in threetransects across each slide for counting ofspores. The mean number observed wascalculated. If fewer than five macroconidiaof Fusarium spp. were observed, an addi-tional three transects were counted and themean was calculated. Traps were operateddaily in 1 week per month during thevegetative stage and daily during flower-ing.Correlative studies. The variablesnumber of hours with relative humidity(RH) higher or equal to 60%, number ofhours with RH ≤ 40%, maximum (Tmax),and minimum (Tmin) daily temperature,average temperature per hour (Th), windspeed (m/s), and numbers of spores trappedwere regressed against the change of dis-ease incidence. Variables were recordedduring June to January in 1993-94 and Juneto February in 1994-95 growing cycle.RESULTSDisease assessment. The malformationdisease progress curves in 1993-94 and1994-95 varied greatly in shape amongtreatments, particularly in the first growthcycle; the final disease incidence (Yf) wasalso greater in 1993-94 (Fig. 1). The firstvisible symptoms on the vegetative shoots(YO) and the greatest incremental increaseof disease incidence occurred after fruitpicking in October and November (Fig. 2).A second, small incremental change oc-curred during full-bloom in the first (1993-94) and last (1994-95) 2 weeks of January.A third peak in disease change on vegeta-tive shoots was observed in March only inthe 1994-95 cycle. Correction of diseaseincidence for host growth was needed in 12of 14 and in 16 of 16 disease progresscurves in first and second growth cycles,respectively.Temporal analyses. Ten out of 14 epi-demics of 1993-94 were best described bythe Gompertz model, three by the mono-molecular, and one by the logistic. Overall,six epidemics were fitted by the best modelwith an r2≥ 0.90; the remaining epidemicshad an r2≥ 0.80 (Table 1). For comparisonpurposes, estimates of the rate of diseaseprogress for epidemics described by themonomolecular and logistic models weretransformed with the Richard’s procedure(25) to provide values of the rate parame-ters equivalent to those for the Gompertzmodel.All 16 epidemics of 1994-95 were bestdescribed by the monomolecular model.Nine epidemics were fitted with an r2≥0.73, and seven had an r2≥ 0.90 (Table 1).The rm was used for comparative purposes.Two IM repetitions that remained healthywere included with an rm of zero.The Weibull model adequately describedall the epidemics in both growing cycles(Table 1). In 1993-94, 10 out of 14 epi-demics were fitted with r2≥ 0.90. Theremaining epidemics had r2values of 0.83to 0.86. These epidemics in general had thelowest YO (0.021 to 0.030) and Yf (0.051 to0.247) values. In 1994-95, 10 out of 16epidemics were fitted with r2≥ 0.90. Theremaining epidemics had r2values of 0.84to 0.89. These epidemics had also the low-Fig. 2. Change of disease incidence of malformation of mango (MM) in vegetative and floral shootsof mango cv. Haden, number of macroconidia of Fusarium spp., maximum temperature (Tmax) andaverage temperature per hour (Th), number of hours with relative humidity (RH) higher than or equalto 60%, and wind speed. 1Jan to 4Jan = first to fourth week of January. North Guerrero, México,during 1993-94 and 1994-95.
226 Plant Disease / Vol. 83 No. 3est YO (0.010 to 0.069) and Yf (0.026 to0.155) values. The c and b parameter esti-mates were negatively correlated in bothgrowth cycles (r > –0.89, P = 0.05).Effect of treatments. In the 1993-94cycle, the HT treatment had the highestvalues of rs, YO, b–1, Yf, and AUDPC (P =0.05) followed by IM and LT. IM and LTwere not significantly different with re-spect to rs and YO but differed in values ofb–1, Yf, and AUDPC (P = 0.05) (Table 2).Values of the c parameter were not signifi-cantly different among treatments. Thehighest average yield (97 kg/tree) wasrecorded in IM (P = 0.05) (Table 2). Thisyield was 74 and 51% more than that ob-tained with the LT and HT treatments,respectively (Table 3). The significance ofthese yield differences was evident in thefinancial analysis (Table 3).In the 1994-95 cycle, the disease inci-dence was generally lower than in the pre-vious cycle (Figs. 1 and 2). HT had thegreatest values of rm, Yf, AUDPC, andAUDPCS followed by LT and IM. Valuesfor parameters associated with HT werestatistically different from those of LT andIM only for AUDPC and AUDPCS (P =0.05). LT and IM were different with re-spect to values for rm, YO, and AUDPCS (P= 0.05). Yield was lower in this cycle (6 to14 kg/tree) than in 1993-94, and no statis-tical differences were found among treat-ments (Tables 2 and 3).Fungus isolation. Fusarium sp. wasisolated consistently from diseased (86%)and asymptomatic (5%) vegetative andflowering mango shoots. These isolateswere identified as F. subglutinans (JeanJuba, Fusarium Research Center, PennState University). Other fungal genera(Pestalotia, Lasiodiplodia [=Botryodiplodia],and Aspergillus) were also recovered atlower frequencies.Correlative studies. The correlationamong change in malformation incidencein treatment HT, number of macroconidia,and climatic factors (Fig. 2) were exam-ined using a Pearson’s correlation matrix(Table 4). Change in incidence of malfor-mation was correlated with values for thenumber of macroconidia of Fusarium spp.(r = 0.90, P = 0.0001) and the wind speed(r = 0.83, P = 0.0001) obtained 4 monthsprior to the specific observation of diseaseincidence (=lag 4 months).Conidia were trapped most frequentlyduring July in the cycles 1993-94 and1994-95. Another peak in conidial numberwas reached in November (1993-94) andOctober (1994-95). A third peak, foundonly in 1994-95, was recorded in February.The number of conidia was correlatedpositively with wind (r = 0.812, P =0.0001) (Table 4). The largest number ofconidia was caught on the average betweenTable 3. Financial analysis of production of a 10-year orchard of mango cv. Haden under three technological management systems in Guerrero, MéxicoTreatmentsx Trees/ha Fruit kg/ha Fruit value ($)y Cost/ha ($) Net benefit/ha B/C B/C totalzCycle 1993-94HT 100 4,700 2,087 863 1,226 1.4 0.72LT 100 2,500 1,110 591 520 0.9 0.27IM 100 9,700 4,307 827 3,484 4.2 2.15Cycle 1994-95HT 100 1,400 617 714 –96 –0.13LT 100 600 265 489 –224 –0.46IM 100 1,000 441 684 –243 –0.36x Treatments: HT = high technology, LT = low traditional technology, and IM = integrated management.y Using $0.444 (1993-94) and $0.441/kg (1994-95) of fruit in average per cycle. The U.S. dollar exchange rates were 4.50 and 6.80 Mexican pesos in May1994 and May 1995, respectively.z Benefit/cost per two cycles (1993-94 and 1994-95).Table 2. Effect of treatments on parametersy of the curve of progress of the “malformation” and yield of the mango, cv. Haden, in the state of Guerrero forgrowth cycles in 1993-94 and 1994-95Treatmentsz r YO c b–1 Yf AUDPC AUDPCS Yield kg/treeCycle 1993-94HT 0.136 a 0.058 a 1.480 a 0.036 a 0.538 a 7.595 a … 47 bLT 0.020 b 0.026 b 1.300 a 0.005 c 0.070 c 0.904 c … 25 bIM 0.069 b 0.029 b 2.348 a 0.022 b 0.269 b 2.922 b … 97 aCycle 1994-95HT 0.0093 a 0.039 a … … 0.161 a 3.306 a 0.129 a 14 aLT 0.0082 a 0.041 a … … 0.105 ab 1.449 b 0.073 b 6 aIM 0.0033 b 0.013 b … … 0.051 b 0.797 b 0.030 c 10 ay r is the apparent infection rate (per unit week–1) standardized by Richard’s method to the Gompertz model (1993-94) and estimated directly with themonomolecular model (1994-95) (estimated by the slope of the line fitted to each epidemic); c and b are, respectively, the curve-shape and scale pa-rameters estimated by the Weibull model; YO and Yf are the initial and final disease incidences; AUDPC = area under disease progress curve (proportion-week), and AUDPCS = the AUDPC standardized by dividing AUDPC by time total duration of an epidemic in week.z Treatments: HT = high technology, LT = low traditional technology, and IM = integrated management. Multiple comparison of means by Student-New-man-Keuls test (P = 0.05).Table 1. Summary of the analysis of regression used to evaluate four models for the progress ofmalformation of mango, cycles 1993-94 and 1994-95Model Epidemicsw r2 MSEx Growth rateyCycle 1993-94Monomolecular 3 0.82-0.94 0.0001-0.0002 0.003-0.004Logistic 1 0.91 0.1865 0.151Gompertz 10 0.80-0.94 0.013-0.182 0.035-0.185Weibull 14 0.83-0.95 0.00002-0.007 1/22-1/4,052Cycle 1994-95zMonomolecular 16 0.73-0.94 0.0001-0.0023 0.003-0.013Logistic 0 … … …Gompertz 0 … … …Weibull 16 0.78-0.96 0.00001-0.0004 1/67.2-1/1,382.2w Number of epidemics best described by a particular model of 14 and 16 epidemics analyzed forcycles 1993-94 and 1994-95, respectively.x Mean square error of the estimate of the rate of apparent infection and r2 = coefficient of determi-nation.y Growth rate of apparent infection estimated as the slope of the monomolecular, logistic, and Gom-pertz linearized model forms and with the inverse of the Weibull scale parameter (b).z In this cycle, two epidemics in the integrated management treatment did not develop.
Plant Disease / March 1999 2270700 and 1100 h. The accumulated pro-portion of shoots with malformation wascorrelated negatively with Tmax (r = –0.681, P = 0.01), Th (r = –0.586, P = 0.04),and RH ≥ 60 (r = –0.82, P = 0.001), andcorrelated positively (r = 0.935, P =0.0001) with the wind speed (Table 4).DISCUSSIONOur goal in this research was to quantifyand examine the progress of malformationof mango and to compare the dynamics ofepidemic development in orchards withdifferent management tactics. Change inthe traditional technology of Mexicanmango growers, particularly pruning andthe use of flowering promoters (KNO3)(IM and HT), resulted in greater numbersof shoots produced and greater yield.However, increased shoot production alsorequires good management of MM to pre-vent high incidence levels due to the in-creased relative abundance of infectionsites. IM, a new and alternative manage-ment strategy proposed to control MM, ingeneral resulted in slower rates of epi-demic development, lower levels of initialand final disease incidence, and a lowerAUDPC in comparison with the other twomanagement strategies. Yield was clearlyhigher with IM than with the other strate-gies in the first cycle (1993-94). In thesecond production cycle, yield was lowerfor all treatments due to alternate bearing,a common phenomenon in mango. Thebenefit-cost ratio of 4.2 for IM was almostthree and four times higher than that ob-tained with HT and LT, respectively, in thehigher production year. Combining bothcycles, IM still had the higher benefit-costratio, i.e., 2.15 (Table 3). These resultsshow the efficacy of combining practices,such as pruning and burning of diseasedshoots, in reducing inoculum and allowingproduction of healthy vegetative shoots. Inaddition, the protection of these shootswith systemic and contact fungicides andthe control of mites and ants, which appar-ently are factors for spore disseminationand tissue wounding (6,24; D. H. Noriega-Cantú, unpublished data), contributed tothe reduction of MM. Trees in the LTtreatment did have low levels of diseasein the first and second years; however,this was attributed primarily to a lack ofvegetative and floral shoots. After the 2years of this study, the management ofMM using our IM approach appearspromising.In addition to examining the effects ofthe various management strategies, we alsoconsidered several factors related to epi-demic analysis in the mango malformationsystem and to the effects of environmentand inoculum availability on disease de-velopment. Several challenges, such as aneed for incorporating Ymax into the diseaseprogress models and for correcting for hostgrowth, suggested comparison amongtreatments with several alterations to curveparameters in order to generate consistentconclusions. Incidence of MM neverreached 100% in any of the treatments;thus a more appropriate value for Ymax wasselected to fit disease progress curves tolinear models (13). Only one Ymax was used(70%) to allow comparison among treat-ments (15). Also, the AUDPC was notused for the 1994-95 cycle due to differenttimes of epidemic duration. Rather,AUDPC standardized (AUDPCS) was usedinstead. This standardized parameterproved to be most appropriate for bothcycles. Another correction for diseasedilution due to host growth, i.e., thedivision of all values of number ofdiseased shoots by the final number ofshoots produced, was applied to all diseaseprogress curves. These corrections may beneeded for many pathosystems in tropicalperennial crops (2,12).The causal agent of MM is controver-sial; however, some studies support theinvolvement of an airborne pathogen(3,4,6,11,19,23). In this study, we isolatedF. subglutinans from diseased shoots, andthe inoculation tests have been positive forthis pathogen (D. H. Noriega-Cantú, un-published data), which agrees with previ-ous findings in Mexico and in many othermango-producing countries (3,4,19,23,32).However, F. oxysporum has also beenfound in some regions where mangos aregrown in México (4,6).The association between disease inci-dence and climatic variables reflected astrong dependence of disease developmenton microclimatic factors measured at thecanopy level. The cumulative disease inci-dence did not increase when the maximumdaily temperatures and the average tem-perature per hour increased and prevailedat levels greater than 33 and 25°C, respec-tively, usually from March to May. In In-dia, early-emerging flower buds were se-verely infected; whereas later buds escapedthe disease; this difference was empiricallyattributed to the relatively high temperatureduring panicle development (11).Even though F. subglutinans was con-sistently isolated from diseased shoots, itcannot be stated that the spores trapped inthe canopy were exclusively attributed tothis species. In general, the highest sporedensity was found during the rainy season,when wind speed (1.5 m/s) and relativehumidity (92 to 94%) were high and thetemperature was moderate (16 to 17.5°C).Similar results were reported in India (11).These researchers found high spore densityof F. subglutinans with min/max tempera-tures of 8/27°C and with humidity of 85%.In our study, the greatest number oftrapped airborne macroconidia of Fu-sarium spp. was characterized by morningperiodicity (0700 to 1100 h), when windspeed (2.8 m/s) and temperature (29°C)were high and humidity was relatively low(55%). Because conidial density wasstrongly correlated with wind speed, windmay play a major role in the dispersal ofthe causal organism of MM. Wind couldalso be more important in the liberation ofspores from dying or dry panicles thanfrom live, infected panicles.Integration of our results suggests thefollowing sequence of events for malfor-mation in a typical commercial orchard inthe North Guerrero region: (i) Vegetativeshoots emerge from the first (in mid-June)to the sixth (November) month after pick-ing (MAP) the fruit. In this stage, the api-cal meristems are colonized extensively byTable 4. Correlation coefficient, levels of significance, and number of observations in order to relate the incidence of malformation of mango with thedispersal of conidial and climatic variables. Growth cycles 1993-94 and 1994-95Variablesy Fspz Tmax Th RH ≥ 60 RH ≤ 40 WindCMM 0.9043 –0.403 –0.389 –0.064 –0.254 0.83360.0001 0.137 0.152 0.821 0.361 0.000115 15 15 15 15 15MM –0.681 –0.586 –0.820 –0.164 0.93450.010 0.035 0.001 0.593 0.000113 13 13 13 13Fsp –0.252 –0.243 0.029 –0.301 0.81200.346 0.365 0.916 0.258 0.000116 16 16 16 16y CMM is the change of the MM for the high technology treatment; MM is the accumulative proportion of disease; Fsp is the number of macroconidia ofFusarium spp.; Tmax is the maximum daily temperature; Tmin is the minimum daily temperature; Th is the average temperature per hour; RH ≥ 60 is thenumber of hours with relative humidity greater than or equal to 60%; RH ≤ 40 is the number of hours with relative humidity less than or equal to 40%;and wind speed is in m/s.z First number is the actual value for the correlation; second number is the significance level of the correlation or the probability of obtaining a greatervalue; and the third number is n, the number of observations in the correlation analysis.
228 Plant Disease / Vol. 83 No. 3F. subglutinans (first to third MAP). (ii)The first visible symptoms appear duringthe fifth (October) and sixth (November)MAP on emerging vegetative shoots (Fig.2). In this vegetative stage, disease inci-dence has the highest rate of incrementalincrease, with an incubation period from 2to 5 months. Usually these vegetativeshoots develop into diseased panicles,which are unproductive. (iii) Full bloomoccurs in the seventh (December) to eighth(January) MAP. In this period, a secondincremental increase in disease incidence isobserved (January) (Fig. 2). Panicles thatbecome diseased at this stage are also un-productive. (iv) A second vegetative flushoccurs during the eighth (January) to theninth (February) MAP, and a third incre-mental increase in disease incidence occurs(in mid-February to mid-March (Fig. 2).(v) Deformed vegetative and floral shootsremain in the tree until conditions of highhumidity, appropriate temperatures, andstrong winds promote dispersion of conidiaand extensive colonization of the new api-cal meristems after fruit harvest. Theemerging young shoots are susceptible toinfection, and the development of vegeta-tive and floral shoots is important to com-plete the inoculum production. However, itis unlikely that this accounts for all thevariation observed, and other sources ofseasonal variation are likely to be of im-portance. For example, changes in hostsusceptibility to infection between produc-tion cycles due to the influence of envi-ronmental and physiologic conditions ofthe host may be important. Because shootdevelopment appears to play a significantrole in the progress of MM, specific meas-ures of host development and its incorpo-ration into an epidemic model may beneeded in future studies. Managementinfluences shoot development and thusdetermines dynamics of the epidemic. En-vironmental factors, however, alsoinfluence changes of disease incidence,particularly through the inoculum disper-sal. Studies are underway to take into ac-count important environmental parametersand to consider host development moreextensively so that more efficient MMmanagement strategies can be evaluated,especially the IM strategy and other, evenbetter, strategies that may be proposed forthe North Guerrero region.ACKNOWLEDGMENTSWe thank the Castresana family for allowingus to use their orchards to conduct theseexperiments. We thank INIFAP, Campo Exptal.Iguala and R. Barajas B. for technicalassistance. We also thank CONACYT/Méxicofor the financial support they provided. Theidentification of our Fusarium isolates by JeanJuba at the Fusarium Research Center,Department of Plant Pathology, PennsylvaniaState University, is appreciated.LITERATURE CITED1. Bhatnagar, S. S., and Beniwal, S. P. S. 1977.Involvement of Fusarium oxysporum in cau-sation of mango malformation. Plant Dis.Rep. 61:894-898.2. Campbell, C. L., and Madden, L. V. 1990.Introduction to Plant Disease Epidemiology.John Wiley & Sons, New York.3. Chakrabarti, D. K., and Ghosal, S. 1989. Thedisease cycle of mango malformation inducedby Fusarium moniliforme var. subglutinansand the curative effects of mangiferin-metalchelates. J. Phytopathol. 125:238-246.4. Covarrubias, C. C. 1989. Pruebas de pato-genicidad de Fusarium sp. como agentecausal de la escoba de bruja del mangoMangifera indica L. Page 36 in: XVI Congr.Nal. Fitopatol. Montecillos, México.5. Covarrubias, R. A. 1980. Control de la“deformación” o “escoba de bruja” del mangoen México. Memorias del Simposium “La In-vestigación el Desarrollo Experimental y laDocencia en CONAFRUT durante 1979.”Tomo 3:795-806.6. Díaz, B. V. 1979. Etiología de la deformacióno “escoba de bruja” del mango en el estado deMorelos. M.C. thesis. Colegio de Postgradua-dos, Chapingo, México.7. Doreste, S. E. 1984. Información sobre eleriófido del mango, Eriophyes mangiferae(Sayed), en Venezuela. Rev. Facultad Agro-ciencia (Universidad Central de Venezuela)13:91-100.8. Fletchtman, C. H. W., Kimati, H., Madcalf, J.C., and Ferrer, J. 1970. Preliminary observa-tions on mango inflorescence malformationand the fungus, insects and mites, associatedwith it. Anais de E.S.A. “Luis de Queiroz”27:281-285.9. Gadoury, D. M., and MacHardy, W. E. 1983.A 7-day recording volumetric spore trap.Phytopathology 73:1526-1531.10. Kumar, J., and Beniwal, S. P. S. 1992. Role ofFusarium species in the etiology of mangomalformation. (Abstr.) Page 17 in: Int. MangoSymp., 4th, Miami.11. Kumar, J., Singh, U. S., and Beniwal, S. P. S.1993. Mango malformation: One hundredyears of research. Annu. Rev. Phytopathol.31:217-232.12. Kushalappa, A. C., and Ludwig, A. 1982.Calculation of apparent infection rate in plantdiseases: Development of a method to correctfor host growth. Phytopathology 72:1373-1377.13. Madden, L. V., and Campbell, C. L. 1990.Nonlinear disease progress curves. Pages 181-229 in: Epidemics of Plant Diseases. Mathe-matical Analysis and Modeling. J. Kranz, ed.Springer-Verlag, Berlin.14. Morales, E., and Rodríguez, H. 1961. Brevesanotaciones sobre una nueva plaga en árbolesde mango. México. Fitófilo 1:7-11.15. Neher, D. A., and Campbell, C. L. 1992.Underestimation of disease progress rateswith the logistic, monomolecular, and Gom-pertz models when maximum disease inten-sity is less than 100 percent. Phytopathology82:811-814.16. Nelson, P. E., Toussoun, T. A., and Marasas,W. F. O. 1983. Fusarium Species: An Illus-trated Manual for Identification. PennsylvaniaState University, University Park.17. Noriega, C. D., Rodríguez, J. A., Marbán-Mendoza, N., and de Zárate, L. G. 1988.Efecto de productos químicos sobre fitone-matodos asociados a la raíz y el ácaro E.mangiferae (Sayed) involucrado en la “escobade bruja” del mango (cv. Haden) en Iguala,Gro., México. Rev. Mex. Fitopatol. 6:61-72.18. Nuñez, E. R. 1988. Nitrato de amonio: Nuevaalternativa para adelantar la floración y co-secha del mango. SARH-INIFAP-CIFAP-COLIMA. Campo Experimental Tecomán,Colima. México. Desplegable para producto-res no. 4.19. Olivas, E. E., and Covarrubias, R. 1978.Identificación del agente causal de la defor-mación floral y vegetativa del mango enMéxico. Fruticultura Mexicana. CONAFRUTno. 1:13-16.20. Pennypacker, S. P., Knoble, H. D., Antle, C.E., and Madden, L. V. 1980. A flexible modelfor studying plant disease progression. Phyto-pathology 70:232-235.21. Pinkas, Y., and Gazit, S. 1992. Mango mal-formation-control strategies. (Abstr.) Page 22in: Int. Mango Symp., 4th, Miami.22. Ploetz, R. C. 1994. Part III. Mango. Pages 33-44 in: Compendium of Tropical Fruit Dis-eases. R. C. Ploetz, G. A. Zentmyer, W. T.Nishijima, K. G. Rohrbach, and H. D. Ohr,eds. American Phytopathological Society, St.Paul, MN.23. Ploetz, R. C., and Gregory, N. F. 1992. Mangomalformation in Florida: Distribution ofFusarium subglutinans in affected trees, andrelationship among strains within and amongdifferent orchards. Acta Hortic. 341:388-394.24. Raychaudiuri, S. P. 1992. Mango malforma-tion. (Abstr.) Page 126 in: Int. Mango Symp.,4th, Miami.25. Richards, F. J. 1959. A flexible growth func-tion for empirical use. J. Exp. Bot. 10:290-300.26. Rouse, D. I. 1985. Construction of temporalmodels: I. Disease progress of air-bornepathogens. Pages 11-28 in: MathematicalModelling of Crop Disease. A. C. Gilligan,ed. Academic Press, New York.27. Siddiqui, S., Sandooja, J. K., Mehta, N., andYamadagni, R. 1987. Biochemical changesduring malformation in mango cultivars as in-fluenced by various chemicals. Pesticides21:17-19.28. Singh, Z., and Dhillon, B. S. 1989. Presenceof malformin-like substances in malformedfloral tissues of mango. J. Phytopathol.125:117-123.29. Singh, Z., Dhillon, B. S., and Arora, C. L.1991. Nutrient levels in malformed andhealthy tissues of mango (Mangifera indicaL.). Plant Soil 133:9-15.30. Srivastava, R. P., and Butani, D. K. 1973. La“Malformation” DV Manguier. Division ofEntomology, IARI, New Delhi. Fruit 28:389-395.31. Steel, R. G. D., and Torrie, J. H. 1980. Princi-ples and Procedures of Statistics. 2nd ed.McGraw-Hill, New York.32. Summanwar, A. S., Raychaudhuri, S. P., andPathak, S. C. 1966. Association of fungusFusarium moniliforme Sheld. with the mal-formation in mango. Indian Phytopathol.19:227-228.33. Thal, W. M., Campbell, C. L., and Madden, L.V. 1984. Sensitivity of Weibull model pa-rameter estimates to variation in simulateddisease progression data. Phytopathology74:1425-1430.34. Varma, A., Raychaudhuri, S. P., Lale, V. C.,and Ram, A. 1971. Preliminary investigationson epidemiology and control of mango mal-formation. Proc. Indian Natl. Sci. Acad. Ser.37 B, No. 5:291-300.35. Vega, P. A., and Miranda, S. M. A. 1993.Distribución, incidencia y severidad de la es-coba de bruja del mango (Mangifera indicaL.) en el Valle de Apatzingán, Mich. Rev.Mex. Fitopatol. 11:1-4.