Analysis of Resistance in Cabbage Varieties toDamage by Lepidoptera and Thysanoptera

A M SHELTON C W HOY R C NORTHM H DICKSON ANDJ BARNARD

Department of EntomologyNew York State Agricultural Experiment Station

Geneva New York 14456

J Econ Entomo 81(2) 634-640 (1988)ABSTRACT Differences in susceptibility to damage by Thrips tabaci Lindeman and thelepidopteran pest complex (Plutella xylostella (L) Artogeia rapae (L) and Trichoplusiani (Hiibner)) were documented and analyzed for fresh market and experimental cabbagevarieties Patterns of resistance to damage were different for Lepidoptera and T tabaci Thecommercial green varieties did not differ in susceptibility to damage by Lepidoptera butdid differ in susceptibility to damage by T tabaci The two experimental varieties and onecommercial red variety used for comparison were less susceptible to damage by Lepidopterabut varied in susceptibility to damage by T tabaci The best regression models for predictingdamage rating without including the effect of variety included one measure of seasonalpest total and one plant parameter head weight Further regression analysis showed thatin most cases the relationship of seasonal pest total to damage had the same slope for allvarieties but different intercepts This result indicates that at a given population densitythe models predict different damage ratings for the varieties implying that nonpreferenceis not the only resistance mechanism operating and that mechanisms involving tolerance orantibiosis were involved Use of regression models that test for varietal differences in therelationship of plant parameters and pest population densities to damage would be helpfulin designing good assays of resistance for selection of new breeding lines

KEY WORDS Insecta plant resistance cabbage damage prediction

AN IMPORTANTcomponent in the management ofagricultural pests should be the use of resistant orless susceptible varieties These are particularlyuseful when other control methods such as naturalor chemical control are not completely effectivein preventing economic injury Management ofpests on New York cabbage particularly onionthrips Thrips tabaci Lindeman and Lepidoptera(Plutella xylostella (L) Artogeia rapae (L) andTrichoplusia ni (Hubner)) can be difficult if re-liance is placed solely on insecticides or naturalcontrol

Several studies have evaluated plant resistanceto Lepidoptera in crucifers (Harrison amp Brubaker1943 Pimentel 1961 Radcliffe amp Chapman1965ab Radcliffe amp Chapman 1966ab Wolfen-barger 1967 Brett amp Sullivan 1974 Creighton etal 1975 Dickson amp Eckenrode 1975ab Dunn ampKempton 1976 Kim 1979 Latheef amp Irwin 1979Lin et al 1983) These studies vary considerablyin the methods used (eg evaluating plant damageor insect populations) in the range of varietiestested the severity of insect pressure and the re-sults for even a single variety Some previous stud-ies on cabbage resistance to Lepidoptera show that

Department of Horticultural SciencesNew YorkState Agri-cultural Experiment Station Geneva NY 14456

2 Computer ServicesNew YorkSlate Agricultural ExperimentStationGeneva NY 14456

red varieties generally have reduced oviposition byA rapae plant maturity may affect resistance andlines derived from the dark green glossy cauliflowerPI 234599 have shown high levels of resistance butthis resistance is partially linked to the glossy natureof the leaf (Dickson amp Wallace 1986)

Although onion thrips has been reported as apest of cabbage since the late 1800s (Sirrine amp Lowe1894) plant resistance has been only studied re-cently In a survey of varietal resistance to thripsin processing cabbage Shelton et al (1983) foundthat no commercial varieties were immune fromdamage but that substantial differences were foundin the severity and numbers of layers injured Re-cent work by Stoner amp Shelton (unpublished data)indicated that resistance is caused largely by dif-ferences in within-plant distribution of thrips Thuson susceptible and resistant varieties the thrips pop-ulations on the entire plant may be equal but theratio of thrips on the head to frame leaves will belowest in resistant varieties

From all of the published studies surveying therange of susceptibility of commercial and experi-mental cabbage lines to various foliage insect pestsa fundamental question arises-how can cabbageresistance to insects be assessed in small plot trialsunder natural infestations In small plot trials dif-ferences in ovipositional preferences may cause dif-ferences in larval counts and plant injury but in a

0022-0493880634-0640$02000 copy 1988 EntomologicalSocietyof America

April1988 SHELTONET AL RESISTANCEOF CABBAGETO LEPIDOPTERAAND T tabaci 635

nonchoice situation such as a commercial fieldsuch ovipositional differences may not occur (Can-telo amp Sanford 1984) Although some people in-volved in breeding plants resistant to insects haveovercome this limitation by artificially infestingplants this is not always possible Another way ofovercoming such limitations is to use data from asmall plot trial and then create a paradigm thatwould predict the amount of plant injury to eachvariety based on an equal number of pests beingpresent on all varieties Ultimately a model thatcould predict plant injury to each variety at a com-mon pest density would aid in designing good as-says of resistance for selection of new breedinglines

The goals of our study were to document sus-ceptibility of storage and fresh market cabbagevarieties for New Yorks major foliage insect pestsand construct statistical models that demonstratethat resistance to damage among cabbage varietiescan be explained as varietal differences in the re-lationship of insect densities and plant parameters

Materials and Methods

Twenty-four cabbage varieties were tested forsusceptibility to T tabaci and the lepidopteran pestcomplex at the Vegetable Research Farm NewYork State Agricultural Experiment Station Ge-neva These included 17 varieties of green cabbagethat are grown commercially for fresh market inNew York and several other cabbage types forcomparison one savoy cabbage two commerciallygrown varieties of red cabbage two green varietiesthat were developed more than 50 years ago andare no longer commonly grown commercially(Danish Ballhead and Early Jersey Wakefield)and two experimental glossy varieties (Geneva 8329and Geneva 8395 both of which had PI 234599 asa parent) Eighteen of the varieties were trans-planted on 10 June 1982 and 14 June 1983 sixshorter season varieties were transplanted on 30June 1982 and 25 June 1983 (King Cole and Su-perette) and 12 July 1982 and 7 July 1983 (DanishBallhead Sunup Special Golden Acre and Ear-ly Jersey Wakefield)

Staggered planting dates were intended to resultin all varieties reaching maturity at approximatelythe same time and to be subject to the same pestpressure (high pest populations do not normallybegin until late July) Varieties were transplantedinto blocks consisting of 24 rows (91 m) with 22plants of a single variety per row replicated fourtimes in a randomized complete block design Un-planted alleys (45 m) separated blocks Each rowwas split into two plots one of these chosen ran-domly received treatments of Bacillus thurin-giensis Berliner (Dipel 112 kg [AI]ha) twiceduring each season to suppress lepidopterous pop-ulations Applications were made with a backpacksprayer delivering 141Iitersha Treated plots wereevaluated only for T tabaci untreated plots sep-

arated by a I-m alley were evaluated only forLepidoptera This split plot design was used toeliminate possible interactions between defoliationby Lepidoptera and infestation by onion thripsLarvae and eggs of the three species of Lepidopterawere counted on four plants per plot in the 18earlier varieties planted on 22 and 29 July 17 Au-gust 2 September and 23 September in 1982 andon three plants per plot of all varieties on 19 July3 15 and 30 August and 14 and 27 September in1983 Lepidoptera were not counted on the sixvarieties planted later in 1982 because the plantsgrowth was retarded by poor weather conditionssoon after transplantation and the relationship ofthese counts to damage rating could not be com-pared with the 18 earlier planted varieties

At harvest (23 September 1982 and 27 Septem-ber 1983) damage by Lepidoptera in the un-sprayed plots was scored on a scale of 0 to 5 ac-cording to the amount of feeding damage on thehead and wrapper leaves (Greene et al 1969 mod-ified by Chalfant et al [1979]) Three heads perblock free of disease and other damage were takenat harvest from the treated plots and examined foronion thrips injury by splitting the head along thecore axis and peeling back each of the outer 10leaves on one half A scale of 0 (no injury) to 4(severe injury) was assigned to each half head Thisscale was based on the number of injured leavesand the extent of injury to each leaf (Shelton et al1983) In addition four heads from each of fivevarieties which showed distinct differences in thripsinjury in 1982 and heads of all varieties in 1983were dissected and washed with alcohol to collectand enumerate all stages of T tabaci present

Amount of damage (mean damage rating foreach of the four replicates) for both pest types atharvest in each year was analyzed by analysis ofvariance To examine the causes of plant damageat harvest a two-step analysis was performed Firstdata from all varieties were combined and regres-sion analyses were performed to determine the bestpredictors of damage ratings in addition to varietaleffects For Lepidoptera the predictors were sea-sonal totals per plant of each pest seasonal larvalunits per plant (Shelton et al 1982) (in this study1 larval unit was equivalent to 1 large T ni 15small T ni or large A rapae and 10 small A rapaeor any P xylostella with the small category equalto the first three instars) plant frame weight andplant head weight For T tabaci predictors werethe square root of the 1982 single count and the1983 seasonal total head weight and head diam-eter All possible regressions were performed fordamage rating predicted by the plant parametersand the appropriate pest totals and selection of thebest candidate regression models was based on Mal-lows C(p) statistic (Hocking 1976) R2 and residualmean square (RMS) The best of these predictorswere then used to describe the relationship of insectabundance and plant parameters to damage ratingfor both Lepidoptera and T tabaci Once this first

636 JOURNAL OF ECONOMIC ENTOMOLOGY Vol 81 no 2

1013

Fig 1 Visual ratings of damage to cabbage varietiesby Lepidoptera and T tabaci Geneva NY 1982 and1983 Mean damage ratings with the same letter abovethe bars are not significantly different at the P = 005level (Duncans [1951multiple range test)

cluded as a predictor in this analysis so we ex-pected R2 to be low Because the purpose of thisanalysis was to compare predictor variables R2 andthe RMS are used only to compare candidatemodels Larval units P xylostella and head weightwere the predictors most often included in the 1982candidate models (Table 1) Larval units aloneshould explain the variability in damage rating dueto P xylostella however inclusion of P xylostellaindicates that larval units do not adequately rep-resent the potential effects of P xylostella on dam-age rating This discrepancy between larval unitsand P xylostella was again found in the 1983 datain which P xylostella was included in all of thebest models and larval units in none (Table 1) Apossible explanation is that feeding damage causedby P xylostella tended to be numerous small ir-regularly shaped holes which would increase visualdamage ratings more than if the same area of feed-ing occurred at the edge of the leaf or in fewerlarger holes Head weight was also included in allof the 1983 candidate models (and frame weightin two of them) indicating that a measure of plantsize can help explain differences in damage ratingThe sign of the head weight parameter estimatesis positive indicating that as the heads get largerthey receive higher damage ratings However the

1082

zi=laquocrw

laquolaquoodgtlaquo

zi=laquocrw

ltlaquoodgtlaquo

step was completed the effect of variety on thatrelationship was evaluated by analysis of variance(ie the predictors selected in the first step wereused as covariates) Results from this step couldthen be used to predict whether or not given thesame insect abundance and plant size varietieswould have different damage ratings

Results and Discussion

Damage Ratings Observed damage ratings forLepidoptera were fairly consistent during the twoyears of the study (Fig 1) Differences in lepidop-teran damage rating were readily apparent withconsistently lower ratings observed for the exper-imental (glossy) varieties (Geneva 8329 and 8395)The red varieties (Red Head and Red Danish)had higher and nearly equal ratings in 1982 butdiffered in 1983 The green varieties had the high-est ratings and among them there were no sub-stantial differences in observed damage The dif-ferences in T tabaci damage ratings were morevariable between years although the eight mostdamaged varieties in 1982 were also the eight mostdamaged varieties in 1983 For thrips injury nopattern was readily apparent between red greenor glossy types within each type there was consid-erable variability Thrips damage to the older va-rieties Danish Ballhead and Early Jersey Wake-field was consistently low throughout the studyVarietal susceptibility to damage by Lepidopteraand T tabaci differed considerably Noteworthyexceptions were the experimental varieties and RedDanish which were only moderately or lightlydamaged by both orders of insects

Analysis of Lepidoptera Using the data fromall varieties on seasonal pest totals plant size mea-surements and damage ratings we first selectedregression models for predicting damage ratingfrom factors other than variety Selected regressionmodels for predicting damage were chosen on thebasis of fewest parameters highest R2 lowest re-sidual mean square (RMS) and Mallows C(p) sta-tistic closest to the number of parameters (p) (Table1) Mallows C(p) provides an indication of thenumber of parameters that should be included inthe model We selected as candidate models thosewith the smallest number of parameters (p) forwhich the C(p) statistic is less than or nearly equalto p For the 1982 data the C(p) statistic was neverequal to or less than p but it came close for someof the three parameter models In 1983 C(p) wasltp for 2- 3- and 4-parameter models This in-dicates that a single measure of pest populationdensity alone (for example larval units) does notgive the best model for predicting damage ratingand additional predictors should be included Thebest additional predictors can be selected by R2and RMS The R2 statistic an indication of theproportion of variation in the data explained bythe model is fairly low for all of the models How-ever the effects of variety had not yet been in-

April1988 SHELTON ET AL RESISTANCE OF CABBAGE TO LEPIDOPTERA AND T tabaci 637

Table 1 Selected regression models for predicting visual damage rating (see text) by Lepidoptera on fresh marketcabbage in Geneva NY in 1982 and 1983

Parameter estimates forNo Seasonal total ofpara- RZ RMS C(p) Head wt

meters Intercept P Frame wtA rapae xylostela T ni LU

19822 01196 1648 3461 3308 0045 01012 01193 1648 3487 3055 0098 02063 01364 1634 3677 3251 0048 0037 00743 01337 1639 3962 3061 0036 0075 01634 01483 1629 4407 2818 0032 0047 0071 0172

19832 0250 1270 1970 2857 0091 02683 0263 1262 2429 3042 0095 -0090 03103 0260 1268 2820 2618 0025 0109 02744 0272 1261 3341 2806 0024 0112 -0088 0315

Larval units (see text)

sign for the frame weight parameter estimates isnegative indicating that damage rating decreasesas frame size increases This could be because dam-age is less obvious on a large plant frame Basedon these results and all of the selection criteriacombined we included head weight and seasonaltotals for either P xylostella A rapae T ni orlarval units in regression models used to test forvarietal differences in the relationship of these vari-ables to damage rating Thus the next step was totest for differences in these relationships ie inslopes or intercepts of regression equations amongthe different varieties

The slope for seasonal pest total is an estimateof the increase in damage rating for a unit increasein seasonal pest total When the slope is the samefor each variety the damage is inflicted by pestsat a similar rate for all the varieties at the popu-lation densities experienced during the study Ifthe slopes are the same for both seasonal pest totaland head weight we can compare the varieties atany common seasonal pest total and head weightand expect the same differences in damage ratingsIf the slopes are different the predicted differencesin damage ratings among the varieties at a commonseasonal pest total and head weight will depend onthe common pest total and head weight chosen Ifthe interaction term between variety and seasonalpest total is not significant in the analysis of vari-ance we conclude that the slopes are not signifi-cantly different In this case a type II error con-cluding that the slopes are not significantly differentwhen they are is more serious than a type I errorthus we use a higher significance level than 005to reduce the probability of making a type II errorThe term for interaction between variety and sea-sonal pest total was not significant for any of themodels tested in 1982 (F 5 165 df = 17 18 P 010) or for the model containing P xylostella in1983 (F 5 154 df = 23 24 P gt 015) indicatingthat the slopes are not significantly different amongthe varieties tested Exceptions were models con-

taining larval units A rapae or T ni in 1983 (F 192 df = 23 24 P 5 006) which is an indicationthat the slopes of these regression equations maydiffer for different varieties when some of the mea-sures of seasonal pest total are used in some yearsThe term for interaction between variety and headweight was not significant in any of the models in1982 (F 5 121 df = 17 18 P gt 0347) or in themodel including P xylostella in 1983 (F 5 115df = 23 24 P gt 036) indicating that the increasein damage rating for a unit increase in head weightis the same for each variety Interaction betweenvariety and head weight was significant when Arapae T ni or larval units were included in modelsin 1983 (F 192 df = 23 24 P 5 0059) indi-cating that the relationship of head weight to dam-age rating may also differ at times among the va-rieties

For each model the effects of seasonal pest totalwere significant (F 6405 df = 1 52 P 5 00001in 1982 F 1285 df = 124 P 5 00015 in 1983)as were the effects of head weight (F 2940 df =1 52 P 5 00001 in 1982 F 1160 df = 1 24P 5 00001 in 1983) Damage by each of these pestpopulations over time on different parts of the plantmay not be best described by the total number oflarvae found during biweekly samples A bettermeasure of the impact of pest populations on dam-age rating (eg more frequent samples and a func-tion to calculate expected cumulative area of feed-ing over time) would improve the analysisHowever the terms for pest total and head weightexplained enough variation in damage rating tomerit their inclusion in the models as covariatesbefore testing for effects of variety

Testing for effects of variety in the analysis ofvariance after the variation due to seasonal pesttotal and head weight has been explained gives anindication of whether or not the intercept termsare different in regression equations for the differ-ent varieties The effects of variety on the interceptterms were significant (F 1206 df = 1752 P lt

638 JOURNAL OF ECONOMIC ENTOMOLOGY Vol 81 no 2

001 in 1982 F ~ 20 df = 23 24 P lt 005 in1983) for all of the models indicating that theintercepts are different among the varieties Formodels in which the slopes are not significantlydifferent the differences in intercept terms providean estimate of differences in damage ratings forany common seasonal pest total and head weightwithin the range of these variables present duringthe study Estimates of these intercept terms forthe best of the regression models damage ratingpredicted by P xylostella and head weight aregiven in Table 2 Estimates for the other regressionmodels were very similar to those in Table 2 Wespeculate that differences in the intercept termsamong the varieties is caused either by toleranceor antibiosis Tolerance could be the result of dif-ferences in location of feeding on these varietiesbecause the damage rating system gives higher val-ues to feeding on the head and wrapper leaves thanon the frame leaves Differences in appearance offeeding damage on the different varieties couldalso result in differences in intercepts because manysmall or irregular holes may receive higher damageratings than fewer large round holes or strip feed-ing at the edge of the leaf Antibiosis could resultin pests being present but not feeding or feedingless than on other varieties If the experimental orred varieties are removed from the analysis thevarietal effects are still significant (F ~ 359 df =15 46 in 1982 F ~ 283 df = 21 64 in 1983 bothP lt 001) but if they are both removed the dif-ferences are no longer significant (F S 082 df =13 40 P gt 063 in 1982 F s 203 df = 19 58P gt 005 for all but A rapae in 1983) Thus theimportant differences lie between these two groupsand the other varieties Actual differences in dam-age ratings could be because of either a differentcapacity for injury at a given pest infestation levelor to varietal differences in numbers of pests pres-ent In general differences in intercepts corre-sponded with differences in damage ratings Forexample the red and experimental varieties hadboth lower damage ratings and lower interceptsthan the other varieties

Analysis of seasonal egg total showed significantdifferences (P = 005) for each species except forA rapae in 1983 (F ~ 14 df = 17 54 in 1982F = 12 df = 23 72 in 1983 both P s 0047) Thered and experimental varieties were consistentlylower in seasonal egg totals except for T ni on redcabbages in 1983 Without the regression modelsthe lower damage ratings could be attributed tofewer eggs (ie nonpreference) However theregression analysis indicates that even with thesame number of larvae on the red or experimentallines the damage ratings would be lower (sug-gesting tolerance or antibiosis) For the experi-mental lines antibiosis seems most likely becausethese lines had a much higher ratio of eggs tosubsequent larvae than the other varieties Thissuggests antibiosis or lack of feeding by early instarsand explains the very low intercepts found in the

regression analysis and low damage ratings seen inthe tests Additional data (unpublished) indicatesthat at least for A rapae resistance is manifestedin the first instars inability to feed on glossy resis-tant varieties

Thrips tabaci Aswas the case with Lepidopterathrips numbers alone did not provide the best pre-dictor of damage rating when potential regressionmodels were compared for the 1983 data (Table3) The addition of some measure of head sizeeither head weight or head diameter improved R2by ca 6 above square root transformed seasonalthrips totals alone and brought the C(p) statisticclose to the number of parameters in the modelNeither measure of head size alone explained asmuch of the variation in the data as thrips numbersOur conclusion is that a measure of head size andthrips population density should be included inregression models before testing for effects of va-rieties For the five varieties on which thrips weresampled in 1982 neither head weight nor headdiameter explained a significant amount of varia-tion in the model and they were not included inthe analysis

The interaction between seasonal thrips total andvariety was not significant (F = 0803 df = 4 10P gt 02 in 1982 F = 174 df = 23 24 P = 0093in 1983) indicating a common positive slope forthe five varieties tested in 1982 and the 24 varietiestested in 1983 The interaction term between va-riety and head weight was not significant either(F = 115 df = 23 24 P = 037) for the 1983 dataVarietal effects were significant (F = 653 df = 414 in 1982 F = 838 df = 23 24 in 1983 bothP lt 001) indicating because the slopes were notdifferent the same differences in the thrips damagerating at any given seasonal total number of thrips(Table 2) Again we speculate that this could becaused either by tolerance or antibiosis One ex-ample of tolerance could be thrips moving andfeeding deeper into the head in more susceptiblevarieties because the damage rating depends onboth the severity and depth of feeding damageStoner amp Shelton (unpublished data) found thattotal numbers of thrips on certain susceptible va-rieties were the same as found on resistant varietiesbut a greater proportion of the thrips were in thehead on the susceptible varieties Another exampleof tolerance could be the way the plant respondsto thrips feeding Original feeding damage is vis-ible only under a microscope but it eventuallybecomes a rough bronzed edema This bronzingand roughening could occur to different degreesin different varieties

Differences in susceptibility to thrips damagecould not be attributed strictly to any obvious cat-egory (experimental red older varieties savoy orseed company lines) When any of these categoriesor a combination of them were restricted from theanalysis the slopes of regression lines for the re-maining varieties were still not significantly dif-ferent and differences still remained in the inter-

April1988 SHELTONET AL RESISTANCEOF CABBAGETO LEPIDOPTERAANDT tabaci 639

Table 2 Parameter estimates for regression models grouped by variety of visual damage rating predicted byseasonal pest total and cabbage hcad weight

Seasonal total of

P xylostella T tabaci

1982 (SE) 1983 (SE) 1982 (SE) 1983 (SE)

Common slope for pest 0019 (0023) -00532 (0034) 00666 (00237)Commoo slope for head wt 0183 (Olll) 01237 (00494) 0242 (0112)

Intercept for varietyHitoma 3272 (0450) 4685 (0368) 1360 (0350)SlIpprgrft1l 3878 (0388) 5004 (0365) 1761 (0341) 1499 (0420)Bartolo 3839 (0390) 4076 (0327) 1415 (0305)Excel 4053 (0429) 4983 (0373) 1978 (0354)Suplrdane 3858 (0435) 4239 (0430) 0766 (0429)Red Head 1919 (0443) 4127 (0420) 0889 (0458)Hinova 3948 (0436) 4222 (0401) 0939 (0389)Market Prizl 3754 (0378) 4641 (0445) 1355 (0401) 1120 (0658)Dcema Extra 3917 (0381) 4499 (0320) 1080 (0272)Grff1l Vinler 4348 (0411) 4231 (0353) 0675 (0299)Lilli bull Rock 3529 (0474) 4493 (0416) 0366 (0375)SUIlUP 4573 (0290) 0206 (0343) 0794 (0289)Houndup 3590 (0575) 4559 (0476) 0232 (0436)Cnbullva 8329 -0036 (0313) 1281 (0304) 0750 (0287)Hed Danish 2560 (0425) 2564 (0349) -0308 (0270)Titanic 90 3868 (0522) 4427 (0419) 1026 (0389) 0172 (0400)Sptcial Gold bulln Acre 5033 (0299) 0801 (0326)Falcon 3564 (0544) 4195 (0363) 0554 (0376) -00277 (0327)Chieftain Savoy 3826 (0396) 4305 (0366) -0044 (0257)King Cole 4351 (0349) 0373 (0333)Supbullrbulltt bull 5021 (0391) 0469 (0360)Danish Ballhead 4475 (0281) 0743 (0242)Early Jtfsey Wakefield 4443 (0288) -0133 (0242)Gbullneva 8395 0205 (0287) 0329 (0276) -0112 (0243)

cepts However the final damage rating dependson both the relationship of thrips numbers to dam-age and the number of thrips present Generallydifferences in damage ratings were consistent withdifferences in the intercept terms in the regressionmodel for the 24 varieties tested However in somecases similar damage ratings had different causesIn 1982 for example Sunup had a lower interceptfor the regression of damage rating on thrips num-bers than Titanic but the seasonal total of thripsin Sunup was great enough that the damage ratingwas similar to that of Titanic In 1983 the situ-ation was reversed for the two varieties Againthese differences in numbers of thrips present maynot appear in a nonchoice situation

Conclusions drawn from these data on cabbageresistance to Lepidoptera and T tabaci can beviewed in two different ways From a standpoint

of screening varieties we have documented vary-ing levels of resistance in fresh market cabbagevarieties to T tabaci and Lepidoptera and notedthe patterns of resistance were quite different forthe two different kinds of pests Varieties showinglittle damage caused by either may prove usefulto breeding programs Second this new approachused to compare the varieties regression modelsthat compare damage rating at a common pestpopulation and plant size provides a better indi-cation of expected differences in damage ratingsin commercial situations than if damage ratingsalone were compared

Acknowledgment

The authors acknowledge the help of Kimberly Stonerin reviewing the literature on crucifer resistance

Table 3 Candidate regression models for predicting visual damage rating by T tabaci on fresh market cabbage inGeneva NY 1983

No Parameter estimates forpara- R2 MSEb e(p) Seasonal total

meters Intercept thrips Head wt Head diam

1 04688 04717 140196 06133 012812 05215 04295 55088 -01873 01047 007392 05382 04145 21729 04301 00946 042053 05390 04182 40000 06394 00935 05278 -00236

a The best 2 and 3 parameter models according to R2 and MSE are included in the tableb Mean square errorC Mallows e(p) statistic (Hocking 1976)

640 JOURNAL OF ECONOMIC ENTOMOLOGY Vol 81 no 2

References Cited

Brett C H amp M J Sullivan 1974 The use of re-sistant varieties and other cultural practices for con-trol of insects on crucifers in North Carolina NCState Univ Agric Exp Stn Bull 449

Cantelo W W amp L L Sanford 1984 Insect pop-ulation response to mixed and uniform plantings ofresistant and susceptible plant material EnvironEntomol 13 1443-1445

Chalfant R B W H Denton D J Schuster amp R BWorkman 1979 Management of cabbage cater-pillars in Florida and Georgia by using visual damagethresholds J Econ Entomol 72 411-413

Creighton C S T L McFadden amp M L Robbins1975 Complementary influence of host plant re-sistance on microbial chemical control of cabbage cat-erpillars HortScience 10 487-488

Dickson M H amp C J Eckenrode 1975a Variationin Brassica oeracea resistance to cabbage looper andimported cabbage worm in the greenhouse and fieldJ Econ Entomol 68 757-760

1975b Breeding for resistance in cabbage and cau-liflower to cabbage looper imported cabbagewormand diamondback moth J Am Soc Hortic Sci 105782-785

Dickson M H amp D H Wallace 1986 Cabbagebreeding pp 395-432 In M J Bassett [ed] Breedingvegetable crops AVI Westport Conn

Dunn J A amp D P H Kempton 1976 Varietaldifferences in the susceptibility of Brussels sprouts tolepidopterous pests Ann Appl BioI 82 11-19

Greene G L W C Genung R B Workman amp E GKelsheimer 1969 Cabbage looper control in Flor-ida-a cooperative program J Econ Entomol 62798-800

Harrison P K amp R W Brubaker 1943 The relativeabundance of cabbage caterpillars on cole crops grownunder similar conditions J Econ Entomol 36 589-592

Hocking R R 1976 The analysis and selection ofvariables in linear regression Biometrics 32 1-50

Kim T H 1979 Brassica oeracea L resistance tocabbage caterpillars in New York State MS thesisCornell Univ Ithaca NY

Latheef M A amp R D Irwin 1979 Factors affectingoviposition of Pieris rapae on cabbage EnvironEntomol 8 606-609

Lin J C J Eckenrode amp M H Dickson 1983 Vari-ation in Brassica oeracea resistance to diamondbackmoth (Lepidoptera Plutellidae) J Econ Entomol76 1423-1427

Pimentel D 1961 An evaluation of insect resistancein broccoli Brussels sprouts cabbage collards andkale J Econ Entomol 54 156-158

Radcliffe E B amp R K Chapman 1965a The rel-ative resistance to insect attack of three cabbage va-rieties at different stages of plant maturity AnnEntomol Soc Am 58 897-902

1965b Seasonal shifts in the relative resistance to in-sect attack of eight commercial cabbage varietiesAnn Entomol Soc Am 58 892-897

1966a Plant resistance to insect attack in commercialcabbage varieties J Econ Entomol 59 116-120

1966b Varietal resistance to insect attack in variouscruciferous crops J Econ Entomol 59 120-125

Shelton A M J T Andaloro amp J Barnard 1982Effects of cabbage looper imported cabbagewormand diamondback moth on fresh market and pro-cessing cabbage J Econ Entomol 75 742-745

Shelton A M R F Becker amp J T Andaloro 1983Varietal resistance to onion thrips (ThysanopteraThripidae) in processing cabbage J Econ Entomol76 85-86

Sirrine F A amp V H Lowe 1894 Insects affectinglate cabbage notes on the stalk borer insecticidesNY Agric Exp Stn (Geneva) Bull 83 657-685

Wolfenbarger D A 1967 Variation in Brassica sppto cabbage looper infestations J Econ Entomol 60874-875

Received for publication 13 March 1987 accepted 21October 1987

April1988 SHELTONET AL RESISTANCEOF CABBAGETO LEPIDOPTERAAND T tabaci 635

nonchoice situation such as a commercial fieldsuch ovipositional differences may not occur (Can-telo amp Sanford 1984) Although some people in-volved in breeding plants resistant to insects haveovercome this limitation by artificially infestingplants this is not always possible Another way ofovercoming such limitations is to use data from asmall plot trial and then create a paradigm thatwould predict the amount of plant injury to eachvariety based on an equal number of pests beingpresent on all varieties Ultimately a model thatcould predict plant injury to each variety at a com-mon pest density would aid in designing good as-says of resistance for selection of new breedinglines

The goals of our study were to document sus-ceptibility of storage and fresh market cabbagevarieties for New Yorks major foliage insect pestsand construct statistical models that demonstratethat resistance to damage among cabbage varietiescan be explained as varietal differences in the re-lationship of insect densities and plant parameters

Materials and Methods

Twenty-four cabbage varieties were tested forsusceptibility to T tabaci and the lepidopteran pestcomplex at the Vegetable Research Farm NewYork State Agricultural Experiment Station Ge-neva These included 17 varieties of green cabbagethat are grown commercially for fresh market inNew York and several other cabbage types forcomparison one savoy cabbage two commerciallygrown varieties of red cabbage two green varietiesthat were developed more than 50 years ago andare no longer commonly grown commercially(Danish Ballhead and Early Jersey Wakefield)and two experimental glossy varieties (Geneva 8329and Geneva 8395 both of which had PI 234599 asa parent) Eighteen of the varieties were trans-planted on 10 June 1982 and 14 June 1983 sixshorter season varieties were transplanted on 30June 1982 and 25 June 1983 (King Cole and Su-perette) and 12 July 1982 and 7 July 1983 (DanishBallhead Sunup Special Golden Acre and Ear-ly Jersey Wakefield)

Staggered planting dates were intended to resultin all varieties reaching maturity at approximatelythe same time and to be subject to the same pestpressure (high pest populations do not normallybegin until late July) Varieties were transplantedinto blocks consisting of 24 rows (91 m) with 22plants of a single variety per row replicated fourtimes in a randomized complete block design Un-planted alleys (45 m) separated blocks Each rowwas split into two plots one of these chosen ran-domly received treatments of Bacillus thurin-giensis Berliner (Dipel 112 kg [AI]ha) twiceduring each season to suppress lepidopterous pop-ulations Applications were made with a backpacksprayer delivering 141Iitersha Treated plots wereevaluated only for T tabaci untreated plots sep-

arated by a I-m alley were evaluated only forLepidoptera This split plot design was used toeliminate possible interactions between defoliationby Lepidoptera and infestation by onion thripsLarvae and eggs of the three species of Lepidopterawere counted on four plants per plot in the 18earlier varieties planted on 22 and 29 July 17 Au-gust 2 September and 23 September in 1982 andon three plants per plot of all varieties on 19 July3 15 and 30 August and 14 and 27 September in1983 Lepidoptera were not counted on the sixvarieties planted later in 1982 because the plantsgrowth was retarded by poor weather conditionssoon after transplantation and the relationship ofthese counts to damage rating could not be com-pared with the 18 earlier planted varieties

At harvest (23 September 1982 and 27 Septem-ber 1983) damage by Lepidoptera in the un-sprayed plots was scored on a scale of 0 to 5 ac-cording to the amount of feeding damage on thehead and wrapper leaves (Greene et al 1969 mod-ified by Chalfant et al [1979]) Three heads perblock free of disease and other damage were takenat harvest from the treated plots and examined foronion thrips injury by splitting the head along thecore axis and peeling back each of the outer 10leaves on one half A scale of 0 (no injury) to 4(severe injury) was assigned to each half head Thisscale was based on the number of injured leavesand the extent of injury to each leaf (Shelton et al1983) In addition four heads from each of fivevarieties which showed distinct differences in thripsinjury in 1982 and heads of all varieties in 1983were dissected and washed with alcohol to collectand enumerate all stages of T tabaci present

Amount of damage (mean damage rating foreach of the four replicates) for both pest types atharvest in each year was analyzed by analysis ofvariance To examine the causes of plant damageat harvest a two-step analysis was performed Firstdata from all varieties were combined and regres-sion analyses were performed to determine the bestpredictors of damage ratings in addition to varietaleffects For Lepidoptera the predictors were sea-sonal totals per plant of each pest seasonal larvalunits per plant (Shelton et al 1982) (in this study1 larval unit was equivalent to 1 large T ni 15small T ni or large A rapae and 10 small A rapaeor any P xylostella with the small category equalto the first three instars) plant frame weight andplant head weight For T tabaci predictors werethe square root of the 1982 single count and the1983 seasonal total head weight and head diam-eter All possible regressions were performed fordamage rating predicted by the plant parametersand the appropriate pest totals and selection of thebest candidate regression models was based on Mal-lows C(p) statistic (Hocking 1976) R2 and residualmean square (RMS) The best of these predictorswere then used to describe the relationship of insectabundance and plant parameters to damage ratingfor both Lepidoptera and T tabaci Once this first

636 JOURNAL OF ECONOMIC ENTOMOLOGY Vol 81 no 2

1013

Fig 1 Visual ratings of damage to cabbage varietiesby Lepidoptera and T tabaci Geneva NY 1982 and1983 Mean damage ratings with the same letter abovethe bars are not significantly different at the P = 005level (Duncans [1951multiple range test)

cluded as a predictor in this analysis so we ex-pected R2 to be low Because the purpose of thisanalysis was to compare predictor variables R2 andthe RMS are used only to compare candidatemodels Larval units P xylostella and head weightwere the predictors most often included in the 1982candidate models (Table 1) Larval units aloneshould explain the variability in damage rating dueto P xylostella however inclusion of P xylostellaindicates that larval units do not adequately rep-resent the potential effects of P xylostella on dam-age rating This discrepancy between larval unitsand P xylostella was again found in the 1983 datain which P xylostella was included in all of thebest models and larval units in none (Table 1) Apossible explanation is that feeding damage causedby P xylostella tended to be numerous small ir-regularly shaped holes which would increase visualdamage ratings more than if the same area of feed-ing occurred at the edge of the leaf or in fewerlarger holes Head weight was also included in allof the 1983 candidate models (and frame weightin two of them) indicating that a measure of plantsize can help explain differences in damage ratingThe sign of the head weight parameter estimatesis positive indicating that as the heads get largerthey receive higher damage ratings However the

1082

zi=laquocrw

laquolaquoodgtlaquo

zi=laquocrw

ltlaquoodgtlaquo

step was completed the effect of variety on thatrelationship was evaluated by analysis of variance(ie the predictors selected in the first step wereused as covariates) Results from this step couldthen be used to predict whether or not given thesame insect abundance and plant size varietieswould have different damage ratings

Results and Discussion

Damage Ratings Observed damage ratings forLepidoptera were fairly consistent during the twoyears of the study (Fig 1) Differences in lepidop-teran damage rating were readily apparent withconsistently lower ratings observed for the exper-imental (glossy) varieties (Geneva 8329 and 8395)The red varieties (Red Head and Red Danish)had higher and nearly equal ratings in 1982 butdiffered in 1983 The green varieties had the high-est ratings and among them there were no sub-stantial differences in observed damage The dif-ferences in T tabaci damage ratings were morevariable between years although the eight mostdamaged varieties in 1982 were also the eight mostdamaged varieties in 1983 For thrips injury nopattern was readily apparent between red greenor glossy types within each type there was consid-erable variability Thrips damage to the older va-rieties Danish Ballhead and Early Jersey Wake-field was consistently low throughout the studyVarietal susceptibility to damage by Lepidopteraand T tabaci differed considerably Noteworthyexceptions were the experimental varieties and RedDanish which were only moderately or lightlydamaged by both orders of insects

Analysis of Lepidoptera Using the data fromall varieties on seasonal pest totals plant size mea-surements and damage ratings we first selectedregression models for predicting damage ratingfrom factors other than variety Selected regressionmodels for predicting damage were chosen on thebasis of fewest parameters highest R2 lowest re-sidual mean square (RMS) and Mallows C(p) sta-tistic closest to the number of parameters (p) (Table1) Mallows C(p) provides an indication of thenumber of parameters that should be included inthe model We selected as candidate models thosewith the smallest number of parameters (p) forwhich the C(p) statistic is less than or nearly equalto p For the 1982 data the C(p) statistic was neverequal to or less than p but it came close for someof the three parameter models In 1983 C(p) wasltp for 2- 3- and 4-parameter models This in-dicates that a single measure of pest populationdensity alone (for example larval units) does notgive the best model for predicting damage ratingand additional predictors should be included Thebest additional predictors can be selected by R2and RMS The R2 statistic an indication of theproportion of variation in the data explained bythe model is fairly low for all of the models How-ever the effects of variety had not yet been in-

April1988 SHELTON ET AL RESISTANCE OF CABBAGE TO LEPIDOPTERA AND T tabaci 637

Table 1 Selected regression models for predicting visual damage rating (see text) by Lepidoptera on fresh marketcabbage in Geneva NY in 1982 and 1983

Parameter estimates forNo Seasonal total ofpara- RZ RMS C(p) Head wt

meters Intercept P Frame wtA rapae xylostela T ni LU

19822 01196 1648 3461 3308 0045 01012 01193 1648 3487 3055 0098 02063 01364 1634 3677 3251 0048 0037 00743 01337 1639 3962 3061 0036 0075 01634 01483 1629 4407 2818 0032 0047 0071 0172

19832 0250 1270 1970 2857 0091 02683 0263 1262 2429 3042 0095 -0090 03103 0260 1268 2820 2618 0025 0109 02744 0272 1261 3341 2806 0024 0112 -0088 0315

Larval units (see text)

sign for the frame weight parameter estimates isnegative indicating that damage rating decreasesas frame size increases This could be because dam-age is less obvious on a large plant frame Basedon these results and all of the selection criteriacombined we included head weight and seasonaltotals for either P xylostella A rapae T ni orlarval units in regression models used to test forvarietal differences in the relationship of these vari-ables to damage rating Thus the next step was totest for differences in these relationships ie inslopes or intercepts of regression equations amongthe different varieties

The slope for seasonal pest total is an estimateof the increase in damage rating for a unit increasein seasonal pest total When the slope is the samefor each variety the damage is inflicted by pestsat a similar rate for all the varieties at the popu-lation densities experienced during the study Ifthe slopes are the same for both seasonal pest totaland head weight we can compare the varieties atany common seasonal pest total and head weightand expect the same differences in damage ratingsIf the slopes are different the predicted differencesin damage ratings among the varieties at a commonseasonal pest total and head weight will depend onthe common pest total and head weight chosen Ifthe interaction term between variety and seasonalpest total is not significant in the analysis of vari-ance we conclude that the slopes are not signifi-cantly different In this case a type II error con-cluding that the slopes are not significantly differentwhen they are is more serious than a type I errorthus we use a higher significance level than 005to reduce the probability of making a type II errorThe term for interaction between variety and sea-sonal pest total was not significant for any of themodels tested in 1982 (F 5 165 df = 17 18 P 010) or for the model containing P xylostella in1983 (F 5 154 df = 23 24 P gt 015) indicatingthat the slopes are not significantly different amongthe varieties tested Exceptions were models con-

taining larval units A rapae or T ni in 1983 (F 192 df = 23 24 P 5 006) which is an indicationthat the slopes of these regression equations maydiffer for different varieties when some of the mea-sures of seasonal pest total are used in some yearsThe term for interaction between variety and headweight was not significant in any of the models in1982 (F 5 121 df = 17 18 P gt 0347) or in themodel including P xylostella in 1983 (F 5 115df = 23 24 P gt 036) indicating that the increasein damage rating for a unit increase in head weightis the same for each variety Interaction betweenvariety and head weight was significant when Arapae T ni or larval units were included in modelsin 1983 (F 192 df = 23 24 P 5 0059) indi-cating that the relationship of head weight to dam-age rating may also differ at times among the va-rieties

For each model the effects of seasonal pest totalwere significant (F 6405 df = 1 52 P 5 00001in 1982 F 1285 df = 124 P 5 00015 in 1983)as were the effects of head weight (F 2940 df =1 52 P 5 00001 in 1982 F 1160 df = 1 24P 5 00001 in 1983) Damage by each of these pestpopulations over time on different parts of the plantmay not be best described by the total number oflarvae found during biweekly samples A bettermeasure of the impact of pest populations on dam-age rating (eg more frequent samples and a func-tion to calculate expected cumulative area of feed-ing over time) would improve the analysisHowever the terms for pest total and head weightexplained enough variation in damage rating tomerit their inclusion in the models as covariatesbefore testing for effects of variety

Testing for effects of variety in the analysis ofvariance after the variation due to seasonal pesttotal and head weight has been explained gives anindication of whether or not the intercept termsare different in regression equations for the differ-ent varieties The effects of variety on the interceptterms were significant (F 1206 df = 1752 P lt

638 JOURNAL OF ECONOMIC ENTOMOLOGY Vol 81 no 2

001 in 1982 F ~ 20 df = 23 24 P lt 005 in1983) for all of the models indicating that theintercepts are different among the varieties Formodels in which the slopes are not significantlydifferent the differences in intercept terms providean estimate of differences in damage ratings forany common seasonal pest total and head weightwithin the range of these variables present duringthe study Estimates of these intercept terms forthe best of the regression models damage ratingpredicted by P xylostella and head weight aregiven in Table 2 Estimates for the other regressionmodels were very similar to those in Table 2 Wespeculate that differences in the intercept termsamong the varieties is caused either by toleranceor antibiosis Tolerance could be the result of dif-ferences in location of feeding on these varietiesbecause the damage rating system gives higher val-ues to feeding on the head and wrapper leaves thanon the frame leaves Differences in appearance offeeding damage on the different varieties couldalso result in differences in intercepts because manysmall or irregular holes may receive higher damageratings than fewer large round holes or strip feed-ing at the edge of the leaf Antibiosis could resultin pests being present but not feeding or feedingless than on other varieties If the experimental orred varieties are removed from the analysis thevarietal effects are still significant (F ~ 359 df =15 46 in 1982 F ~ 283 df = 21 64 in 1983 bothP lt 001) but if they are both removed the dif-ferences are no longer significant (F S 082 df =13 40 P gt 063 in 1982 F s 203 df = 19 58P gt 005 for all but A rapae in 1983) Thus theimportant differences lie between these two groupsand the other varieties Actual differences in dam-age ratings could be because of either a differentcapacity for injury at a given pest infestation levelor to varietal differences in numbers of pests pres-ent In general differences in intercepts corre-sponded with differences in damage ratings Forexample the red and experimental varieties hadboth lower damage ratings and lower interceptsthan the other varieties

Analysis of seasonal egg total showed significantdifferences (P = 005) for each species except forA rapae in 1983 (F ~ 14 df = 17 54 in 1982F = 12 df = 23 72 in 1983 both P s 0047) Thered and experimental varieties were consistentlylower in seasonal egg totals except for T ni on redcabbages in 1983 Without the regression modelsthe lower damage ratings could be attributed tofewer eggs (ie nonpreference) However theregression analysis indicates that even with thesame number of larvae on the red or experimentallines the damage ratings would be lower (sug-gesting tolerance or antibiosis) For the experi-mental lines antibiosis seems most likely becausethese lines had a much higher ratio of eggs tosubsequent larvae than the other varieties Thissuggests antibiosis or lack of feeding by early instarsand explains the very low intercepts found in the

regression analysis and low damage ratings seen inthe tests Additional data (unpublished) indicatesthat at least for A rapae resistance is manifestedin the first instars inability to feed on glossy resis-tant varieties

Thrips tabaci Aswas the case with Lepidopterathrips numbers alone did not provide the best pre-dictor of damage rating when potential regressionmodels were compared for the 1983 data (Table3) The addition of some measure of head sizeeither head weight or head diameter improved R2by ca 6 above square root transformed seasonalthrips totals alone and brought the C(p) statisticclose to the number of parameters in the modelNeither measure of head size alone explained asmuch of the variation in the data as thrips numbersOur conclusion is that a measure of head size andthrips population density should be included inregression models before testing for effects of va-rieties For the five varieties on which thrips weresampled in 1982 neither head weight nor headdiameter explained a significant amount of varia-tion in the model and they were not included inthe analysis

The interaction between seasonal thrips total andvariety was not significant (F = 0803 df = 4 10P gt 02 in 1982 F = 174 df = 23 24 P = 0093in 1983) indicating a common positive slope forthe five varieties tested in 1982 and the 24 varietiestested in 1983 The interaction term between va-riety and head weight was not significant either(F = 115 df = 23 24 P = 037) for the 1983 dataVarietal effects were significant (F = 653 df = 414 in 1982 F = 838 df = 23 24 in 1983 bothP lt 001) indicating because the slopes were notdifferent the same differences in the thrips damagerating at any given seasonal total number of thrips(Table 2) Again we speculate that this could becaused either by tolerance or antibiosis One ex-ample of tolerance could be thrips moving andfeeding deeper into the head in more susceptiblevarieties because the damage rating depends onboth the severity and depth of feeding damageStoner amp Shelton (unpublished data) found thattotal numbers of thrips on certain susceptible va-rieties were the same as found on resistant varietiesbut a greater proportion of the thrips were in thehead on the susceptible varieties Another exampleof tolerance could be the way the plant respondsto thrips feeding Original feeding damage is vis-ible only under a microscope but it eventuallybecomes a rough bronzed edema This bronzingand roughening could occur to different degreesin different varieties

Differences in susceptibility to thrips damagecould not be attributed strictly to any obvious cat-egory (experimental red older varieties savoy orseed company lines) When any of these categoriesor a combination of them were restricted from theanalysis the slopes of regression lines for the re-maining varieties were still not significantly dif-ferent and differences still remained in the inter-

April1988 SHELTONET AL RESISTANCEOF CABBAGETO LEPIDOPTERAANDT tabaci 639

Table 2 Parameter estimates for regression models grouped by variety of visual damage rating predicted byseasonal pest total and cabbage hcad weight

Seasonal total of

P xylostella T tabaci

1982 (SE) 1983 (SE) 1982 (SE) 1983 (SE)

Common slope for pest 0019 (0023) -00532 (0034) 00666 (00237)Commoo slope for head wt 0183 (Olll) 01237 (00494) 0242 (0112)

Intercept for varietyHitoma 3272 (0450) 4685 (0368) 1360 (0350)SlIpprgrft1l 3878 (0388) 5004 (0365) 1761 (0341) 1499 (0420)Bartolo 3839 (0390) 4076 (0327) 1415 (0305)Excel 4053 (0429) 4983 (0373) 1978 (0354)Suplrdane 3858 (0435) 4239 (0430) 0766 (0429)Red Head 1919 (0443) 4127 (0420) 0889 (0458)Hinova 3948 (0436) 4222 (0401) 0939 (0389)Market Prizl 3754 (0378) 4641 (0445) 1355 (0401) 1120 (0658)Dcema Extra 3917 (0381) 4499 (0320) 1080 (0272)Grff1l Vinler 4348 (0411) 4231 (0353) 0675 (0299)Lilli bull Rock 3529 (0474) 4493 (0416) 0366 (0375)SUIlUP 4573 (0290) 0206 (0343) 0794 (0289)Houndup 3590 (0575) 4559 (0476) 0232 (0436)Cnbullva 8329 -0036 (0313) 1281 (0304) 0750 (0287)Hed Danish 2560 (0425) 2564 (0349) -0308 (0270)Titanic 90 3868 (0522) 4427 (0419) 1026 (0389) 0172 (0400)Sptcial Gold bulln Acre 5033 (0299) 0801 (0326)Falcon 3564 (0544) 4195 (0363) 0554 (0376) -00277 (0327)Chieftain Savoy 3826 (0396) 4305 (0366) -0044 (0257)King Cole 4351 (0349) 0373 (0333)Supbullrbulltt bull 5021 (0391) 0469 (0360)Danish Ballhead 4475 (0281) 0743 (0242)Early Jtfsey Wakefield 4443 (0288) -0133 (0242)Gbullneva 8395 0205 (0287) 0329 (0276) -0112 (0243)

cepts However the final damage rating dependson both the relationship of thrips numbers to dam-age and the number of thrips present Generallydifferences in damage ratings were consistent withdifferences in the intercept terms in the regressionmodel for the 24 varieties tested However in somecases similar damage ratings had different causesIn 1982 for example Sunup had a lower interceptfor the regression of damage rating on thrips num-bers than Titanic but the seasonal total of thripsin Sunup was great enough that the damage ratingwas similar to that of Titanic In 1983 the situ-ation was reversed for the two varieties Againthese differences in numbers of thrips present maynot appear in a nonchoice situation

Conclusions drawn from these data on cabbageresistance to Lepidoptera and T tabaci can beviewed in two different ways From a standpoint

of screening varieties we have documented vary-ing levels of resistance in fresh market cabbagevarieties to T tabaci and Lepidoptera and notedthe patterns of resistance were quite different forthe two different kinds of pests Varieties showinglittle damage caused by either may prove usefulto breeding programs Second this new approachused to compare the varieties regression modelsthat compare damage rating at a common pestpopulation and plant size provides a better indi-cation of expected differences in damage ratingsin commercial situations than if damage ratingsalone were compared

Acknowledgment

The authors acknowledge the help of Kimberly Stonerin reviewing the literature on crucifer resistance

Table 3 Candidate regression models for predicting visual damage rating by T tabaci on fresh market cabbage inGeneva NY 1983

No Parameter estimates forpara- R2 MSEb e(p) Seasonal total

meters Intercept thrips Head wt Head diam

1 04688 04717 140196 06133 012812 05215 04295 55088 -01873 01047 007392 05382 04145 21729 04301 00946 042053 05390 04182 40000 06394 00935 05278 -00236

a The best 2 and 3 parameter models according to R2 and MSE are included in the tableb Mean square errorC Mallows e(p) statistic (Hocking 1976)

640 JOURNAL OF ECONOMIC ENTOMOLOGY Vol 81 no 2

References Cited

Brett C H amp M J Sullivan 1974 The use of re-sistant varieties and other cultural practices for con-trol of insects on crucifers in North Carolina NCState Univ Agric Exp Stn Bull 449

Cantelo W W amp L L Sanford 1984 Insect pop-ulation response to mixed and uniform plantings ofresistant and susceptible plant material EnvironEntomol 13 1443-1445

Chalfant R B W H Denton D J Schuster amp R BWorkman 1979 Management of cabbage cater-pillars in Florida and Georgia by using visual damagethresholds J Econ Entomol 72 411-413

Creighton C S T L McFadden amp M L Robbins1975 Complementary influence of host plant re-sistance on microbial chemical control of cabbage cat-erpillars HortScience 10 487-488

Dickson M H amp C J Eckenrode 1975a Variationin Brassica oeracea resistance to cabbage looper andimported cabbage worm in the greenhouse and fieldJ Econ Entomol 68 757-760

1975b Breeding for resistance in cabbage and cau-liflower to cabbage looper imported cabbagewormand diamondback moth J Am Soc Hortic Sci 105782-785

Dickson M H amp D H Wallace 1986 Cabbagebreeding pp 395-432 In M J Bassett [ed] Breedingvegetable crops AVI Westport Conn

Dunn J A amp D P H Kempton 1976 Varietaldifferences in the susceptibility of Brussels sprouts tolepidopterous pests Ann Appl BioI 82 11-19

Greene G L W C Genung R B Workman amp E GKelsheimer 1969 Cabbage looper control in Flor-ida-a cooperative program J Econ Entomol 62798-800

Harrison P K amp R W Brubaker 1943 The relativeabundance of cabbage caterpillars on cole crops grownunder similar conditions J Econ Entomol 36 589-592

Hocking R R 1976 The analysis and selection ofvariables in linear regression Biometrics 32 1-50

Kim T H 1979 Brassica oeracea L resistance tocabbage caterpillars in New York State MS thesisCornell Univ Ithaca NY

Latheef M A amp R D Irwin 1979 Factors affectingoviposition of Pieris rapae on cabbage EnvironEntomol 8 606-609

Lin J C J Eckenrode amp M H Dickson 1983 Vari-ation in Brassica oeracea resistance to diamondbackmoth (Lepidoptera Plutellidae) J Econ Entomol76 1423-1427

Pimentel D 1961 An evaluation of insect resistancein broccoli Brussels sprouts cabbage collards andkale J Econ Entomol 54 156-158

Radcliffe E B amp R K Chapman 1965a The rel-ative resistance to insect attack of three cabbage va-rieties at different stages of plant maturity AnnEntomol Soc Am 58 897-902

1965b Seasonal shifts in the relative resistance to in-sect attack of eight commercial cabbage varietiesAnn Entomol Soc Am 58 892-897

1966a Plant resistance to insect attack in commercialcabbage varieties J Econ Entomol 59 116-120

1966b Varietal resistance to insect attack in variouscruciferous crops J Econ Entomol 59 120-125

Shelton A M J T Andaloro amp J Barnard 1982Effects of cabbage looper imported cabbagewormand diamondback moth on fresh market and pro-cessing cabbage J Econ Entomol 75 742-745

Shelton A M R F Becker amp J T Andaloro 1983Varietal resistance to onion thrips (ThysanopteraThripidae) in processing cabbage J Econ Entomol76 85-86

Sirrine F A amp V H Lowe 1894 Insects affectinglate cabbage notes on the stalk borer insecticidesNY Agric Exp Stn (Geneva) Bull 83 657-685

Wolfenbarger D A 1967 Variation in Brassica sppto cabbage looper infestations J Econ Entomol 60874-875

Received for publication 13 March 1987 accepted 21October 1987

636 JOURNAL OF ECONOMIC ENTOMOLOGY Vol 81 no 2

1013

Fig 1 Visual ratings of damage to cabbage varietiesby Lepidoptera and T tabaci Geneva NY 1982 and1983 Mean damage ratings with the same letter abovethe bars are not significantly different at the P = 005level (Duncans [1951multiple range test)

cluded as a predictor in this analysis so we ex-pected R2 to be low Because the purpose of thisanalysis was to compare predictor variables R2 andthe RMS are used only to compare candidatemodels Larval units P xylostella and head weightwere the predictors most often included in the 1982candidate models (Table 1) Larval units aloneshould explain the variability in damage rating dueto P xylostella however inclusion of P xylostellaindicates that larval units do not adequately rep-resent the potential effects of P xylostella on dam-age rating This discrepancy between larval unitsand P xylostella was again found in the 1983 datain which P xylostella was included in all of thebest models and larval units in none (Table 1) Apossible explanation is that feeding damage causedby P xylostella tended to be numerous small ir-regularly shaped holes which would increase visualdamage ratings more than if the same area of feed-ing occurred at the edge of the leaf or in fewerlarger holes Head weight was also included in allof the 1983 candidate models (and frame weightin two of them) indicating that a measure of plantsize can help explain differences in damage ratingThe sign of the head weight parameter estimatesis positive indicating that as the heads get largerthey receive higher damage ratings However the

1082

zi=laquocrw

laquolaquoodgtlaquo

zi=laquocrw

ltlaquoodgtlaquo

step was completed the effect of variety on thatrelationship was evaluated by analysis of variance(ie the predictors selected in the first step wereused as covariates) Results from this step couldthen be used to predict whether or not given thesame insect abundance and plant size varietieswould have different damage ratings

Results and Discussion

Damage Ratings Observed damage ratings forLepidoptera were fairly consistent during the twoyears of the study (Fig 1) Differences in lepidop-teran damage rating were readily apparent withconsistently lower ratings observed for the exper-imental (glossy) varieties (Geneva 8329 and 8395)The red varieties (Red Head and Red Danish)had higher and nearly equal ratings in 1982 butdiffered in 1983 The green varieties had the high-est ratings and among them there were no sub-stantial differences in observed damage The dif-ferences in T tabaci damage ratings were morevariable between years although the eight mostdamaged varieties in 1982 were also the eight mostdamaged varieties in 1983 For thrips injury nopattern was readily apparent between red greenor glossy types within each type there was consid-erable variability Thrips damage to the older va-rieties Danish Ballhead and Early Jersey Wake-field was consistently low throughout the studyVarietal susceptibility to damage by Lepidopteraand T tabaci differed considerably Noteworthyexceptions were the experimental varieties and RedDanish which were only moderately or lightlydamaged by both orders of insects

Analysis of Lepidoptera Using the data fromall varieties on seasonal pest totals plant size mea-surements and damage ratings we first selectedregression models for predicting damage ratingfrom factors other than variety Selected regressionmodels for predicting damage were chosen on thebasis of fewest parameters highest R2 lowest re-sidual mean square (RMS) and Mallows C(p) sta-tistic closest to the number of parameters (p) (Table1) Mallows C(p) provides an indication of thenumber of parameters that should be included inthe model We selected as candidate models thosewith the smallest number of parameters (p) forwhich the C(p) statistic is less than or nearly equalto p For the 1982 data the C(p) statistic was neverequal to or less than p but it came close for someof the three parameter models In 1983 C(p) wasltp for 2- 3- and 4-parameter models This in-dicates that a single measure of pest populationdensity alone (for example larval units) does notgive the best model for predicting damage ratingand additional predictors should be included Thebest additional predictors can be selected by R2and RMS The R2 statistic an indication of theproportion of variation in the data explained bythe model is fairly low for all of the models How-ever the effects of variety had not yet been in-

April1988 SHELTON ET AL RESISTANCE OF CABBAGE TO LEPIDOPTERA AND T tabaci 637

Table 1 Selected regression models for predicting visual damage rating (see text) by Lepidoptera on fresh marketcabbage in Geneva NY in 1982 and 1983

Parameter estimates forNo Seasonal total ofpara- RZ RMS C(p) Head wt

meters Intercept P Frame wtA rapae xylostela T ni LU

19822 01196 1648 3461 3308 0045 01012 01193 1648 3487 3055 0098 02063 01364 1634 3677 3251 0048 0037 00743 01337 1639 3962 3061 0036 0075 01634 01483 1629 4407 2818 0032 0047 0071 0172

19832 0250 1270 1970 2857 0091 02683 0263 1262 2429 3042 0095 -0090 03103 0260 1268 2820 2618 0025 0109 02744 0272 1261 3341 2806 0024 0112 -0088 0315

Larval units (see text)

sign for the frame weight parameter estimates isnegative indicating that damage rating decreasesas frame size increases This could be because dam-age is less obvious on a large plant frame Basedon these results and all of the selection criteriacombined we included head weight and seasonaltotals for either P xylostella A rapae T ni orlarval units in regression models used to test forvarietal differences in the relationship of these vari-ables to damage rating Thus the next step was totest for differences in these relationships ie inslopes or intercepts of regression equations amongthe different varieties

The slope for seasonal pest total is an estimateof the increase in damage rating for a unit increasein seasonal pest total When the slope is the samefor each variety the damage is inflicted by pestsat a similar rate for all the varieties at the popu-lation densities experienced during the study Ifthe slopes are the same for both seasonal pest totaland head weight we can compare the varieties atany common seasonal pest total and head weightand expect the same differences in damage ratingsIf the slopes are different the predicted differencesin damage ratings among the varieties at a commonseasonal pest total and head weight will depend onthe common pest total and head weight chosen Ifthe interaction term between variety and seasonalpest total is not significant in the analysis of vari-ance we conclude that the slopes are not signifi-cantly different In this case a type II error con-cluding that the slopes are not significantly differentwhen they are is more serious than a type I errorthus we use a higher significance level than 005to reduce the probability of making a type II errorThe term for interaction between variety and sea-sonal pest total was not significant for any of themodels tested in 1982 (F 5 165 df = 17 18 P 010) or for the model containing P xylostella in1983 (F 5 154 df = 23 24 P gt 015) indicatingthat the slopes are not significantly different amongthe varieties tested Exceptions were models con-

taining larval units A rapae or T ni in 1983 (F 192 df = 23 24 P 5 006) which is an indicationthat the slopes of these regression equations maydiffer for different varieties when some of the mea-sures of seasonal pest total are used in some yearsThe term for interaction between variety and headweight was not significant in any of the models in1982 (F 5 121 df = 17 18 P gt 0347) or in themodel including P xylostella in 1983 (F 5 115df = 23 24 P gt 036) indicating that the increasein damage rating for a unit increase in head weightis the same for each variety Interaction betweenvariety and head weight was significant when Arapae T ni or larval units were included in modelsin 1983 (F 192 df = 23 24 P 5 0059) indi-cating that the relationship of head weight to dam-age rating may also differ at times among the va-rieties

For each model the effects of seasonal pest totalwere significant (F 6405 df = 1 52 P 5 00001in 1982 F 1285 df = 124 P 5 00015 in 1983)as were the effects of head weight (F 2940 df =1 52 P 5 00001 in 1982 F 1160 df = 1 24P 5 00001 in 1983) Damage by each of these pestpopulations over time on different parts of the plantmay not be best described by the total number oflarvae found during biweekly samples A bettermeasure of the impact of pest populations on dam-age rating (eg more frequent samples and a func-tion to calculate expected cumulative area of feed-ing over time) would improve the analysisHowever the terms for pest total and head weightexplained enough variation in damage rating tomerit their inclusion in the models as covariatesbefore testing for effects of variety

Testing for effects of variety in the analysis ofvariance after the variation due to seasonal pesttotal and head weight has been explained gives anindication of whether or not the intercept termsare different in regression equations for the differ-ent varieties The effects of variety on the interceptterms were significant (F 1206 df = 1752 P lt

638 JOURNAL OF ECONOMIC ENTOMOLOGY Vol 81 no 2

001 in 1982 F ~ 20 df = 23 24 P lt 005 in1983) for all of the models indicating that theintercepts are different among the varieties Formodels in which the slopes are not significantlydifferent the differences in intercept terms providean estimate of differences in damage ratings forany common seasonal pest total and head weightwithin the range of these variables present duringthe study Estimates of these intercept terms forthe best of the regression models damage ratingpredicted by P xylostella and head weight aregiven in Table 2 Estimates for the other regressionmodels were very similar to those in Table 2 Wespeculate that differences in the intercept termsamong the varieties is caused either by toleranceor antibiosis Tolerance could be the result of dif-ferences in location of feeding on these varietiesbecause the damage rating system gives higher val-ues to feeding on the head and wrapper leaves thanon the frame leaves Differences in appearance offeeding damage on the different varieties couldalso result in differences in intercepts because manysmall or irregular holes may receive higher damageratings than fewer large round holes or strip feed-ing at the edge of the leaf Antibiosis could resultin pests being present but not feeding or feedingless than on other varieties If the experimental orred varieties are removed from the analysis thevarietal effects are still significant (F ~ 359 df =15 46 in 1982 F ~ 283 df = 21 64 in 1983 bothP lt 001) but if they are both removed the dif-ferences are no longer significant (F S 082 df =13 40 P gt 063 in 1982 F s 203 df = 19 58P gt 005 for all but A rapae in 1983) Thus theimportant differences lie between these two groupsand the other varieties Actual differences in dam-age ratings could be because of either a differentcapacity for injury at a given pest infestation levelor to varietal differences in numbers of pests pres-ent In general differences in intercepts corre-sponded with differences in damage ratings Forexample the red and experimental varieties hadboth lower damage ratings and lower interceptsthan the other varieties

Analysis of seasonal egg total showed significantdifferences (P = 005) for each species except forA rapae in 1983 (F ~ 14 df = 17 54 in 1982F = 12 df = 23 72 in 1983 both P s 0047) Thered and experimental varieties were consistentlylower in seasonal egg totals except for T ni on redcabbages in 1983 Without the regression modelsthe lower damage ratings could be attributed tofewer eggs (ie nonpreference) However theregression analysis indicates that even with thesame number of larvae on the red or experimentallines the damage ratings would be lower (sug-gesting tolerance or antibiosis) For the experi-mental lines antibiosis seems most likely becausethese lines had a much higher ratio of eggs tosubsequent larvae than the other varieties Thissuggests antibiosis or lack of feeding by early instarsand explains the very low intercepts found in the

regression analysis and low damage ratings seen inthe tests Additional data (unpublished) indicatesthat at least for A rapae resistance is manifestedin the first instars inability to feed on glossy resis-tant varieties

Thrips tabaci Aswas the case with Lepidopterathrips numbers alone did not provide the best pre-dictor of damage rating when potential regressionmodels were compared for the 1983 data (Table3) The addition of some measure of head sizeeither head weight or head diameter improved R2by ca 6 above square root transformed seasonalthrips totals alone and brought the C(p) statisticclose to the number of parameters in the modelNeither measure of head size alone explained asmuch of the variation in the data as thrips numbersOur conclusion is that a measure of head size andthrips population density should be included inregression models before testing for effects of va-rieties For the five varieties on which thrips weresampled in 1982 neither head weight nor headdiameter explained a significant amount of varia-tion in the model and they were not included inthe analysis

The interaction between seasonal thrips total andvariety was not significant (F = 0803 df = 4 10P gt 02 in 1982 F = 174 df = 23 24 P = 0093in 1983) indicating a common positive slope forthe five varieties tested in 1982 and the 24 varietiestested in 1983 The interaction term between va-riety and head weight was not significant either(F = 115 df = 23 24 P = 037) for the 1983 dataVarietal effects were significant (F = 653 df = 414 in 1982 F = 838 df = 23 24 in 1983 bothP lt 001) indicating because the slopes were notdifferent the same differences in the thrips damagerating at any given seasonal total number of thrips(Table 2) Again we speculate that this could becaused either by tolerance or antibiosis One ex-ample of tolerance could be thrips moving andfeeding deeper into the head in more susceptiblevarieties because the damage rating depends onboth the severity and depth of feeding damageStoner amp Shelton (unpublished data) found thattotal numbers of thrips on certain susceptible va-rieties were the same as found on resistant varietiesbut a greater proportion of the thrips were in thehead on the susceptible varieties Another exampleof tolerance could be the way the plant respondsto thrips feeding Original feeding damage is vis-ible only under a microscope but it eventuallybecomes a rough bronzed edema This bronzingand roughening could occur to different degreesin different varieties

Differences in susceptibility to thrips damagecould not be attributed strictly to any obvious cat-egory (experimental red older varieties savoy orseed company lines) When any of these categoriesor a combination of them were restricted from theanalysis the slopes of regression lines for the re-maining varieties were still not significantly dif-ferent and differences still remained in the inter-

April1988 SHELTONET AL RESISTANCEOF CABBAGETO LEPIDOPTERAANDT tabaci 639

Table 2 Parameter estimates for regression models grouped by variety of visual damage rating predicted byseasonal pest total and cabbage hcad weight

Seasonal total of

P xylostella T tabaci

1982 (SE) 1983 (SE) 1982 (SE) 1983 (SE)

Common slope for pest 0019 (0023) -00532 (0034) 00666 (00237)Commoo slope for head wt 0183 (Olll) 01237 (00494) 0242 (0112)

Intercept for varietyHitoma 3272 (0450) 4685 (0368) 1360 (0350)SlIpprgrft1l 3878 (0388) 5004 (0365) 1761 (0341) 1499 (0420)Bartolo 3839 (0390) 4076 (0327) 1415 (0305)Excel 4053 (0429) 4983 (0373) 1978 (0354)Suplrdane 3858 (0435) 4239 (0430) 0766 (0429)Red Head 1919 (0443) 4127 (0420) 0889 (0458)Hinova 3948 (0436) 4222 (0401) 0939 (0389)Market Prizl 3754 (0378) 4641 (0445) 1355 (0401) 1120 (0658)Dcema Extra 3917 (0381) 4499 (0320) 1080 (0272)Grff1l Vinler 4348 (0411) 4231 (0353) 0675 (0299)Lilli bull Rock 3529 (0474) 4493 (0416) 0366 (0375)SUIlUP 4573 (0290) 0206 (0343) 0794 (0289)Houndup 3590 (0575) 4559 (0476) 0232 (0436)Cnbullva 8329 -0036 (0313) 1281 (0304) 0750 (0287)Hed Danish 2560 (0425) 2564 (0349) -0308 (0270)Titanic 90 3868 (0522) 4427 (0419) 1026 (0389) 0172 (0400)Sptcial Gold bulln Acre 5033 (0299) 0801 (0326)Falcon 3564 (0544) 4195 (0363) 0554 (0376) -00277 (0327)Chieftain Savoy 3826 (0396) 4305 (0366) -0044 (0257)King Cole 4351 (0349) 0373 (0333)Supbullrbulltt bull 5021 (0391) 0469 (0360)Danish Ballhead 4475 (0281) 0743 (0242)Early Jtfsey Wakefield 4443 (0288) -0133 (0242)Gbullneva 8395 0205 (0287) 0329 (0276) -0112 (0243)

cepts However the final damage rating dependson both the relationship of thrips numbers to dam-age and the number of thrips present Generallydifferences in damage ratings were consistent withdifferences in the intercept terms in the regressionmodel for the 24 varieties tested However in somecases similar damage ratings had different causesIn 1982 for example Sunup had a lower interceptfor the regression of damage rating on thrips num-bers than Titanic but the seasonal total of thripsin Sunup was great enough that the damage ratingwas similar to that of Titanic In 1983 the situ-ation was reversed for the two varieties Againthese differences in numbers of thrips present maynot appear in a nonchoice situation

Conclusions drawn from these data on cabbageresistance to Lepidoptera and T tabaci can beviewed in two different ways From a standpoint

of screening varieties we have documented vary-ing levels of resistance in fresh market cabbagevarieties to T tabaci and Lepidoptera and notedthe patterns of resistance were quite different forthe two different kinds of pests Varieties showinglittle damage caused by either may prove usefulto breeding programs Second this new approachused to compare the varieties regression modelsthat compare damage rating at a common pestpopulation and plant size provides a better indi-cation of expected differences in damage ratingsin commercial situations than if damage ratingsalone were compared

Acknowledgment

The authors acknowledge the help of Kimberly Stonerin reviewing the literature on crucifer resistance

Table 3 Candidate regression models for predicting visual damage rating by T tabaci on fresh market cabbage inGeneva NY 1983

No Parameter estimates forpara- R2 MSEb e(p) Seasonal total

meters Intercept thrips Head wt Head diam

1 04688 04717 140196 06133 012812 05215 04295 55088 -01873 01047 007392 05382 04145 21729 04301 00946 042053 05390 04182 40000 06394 00935 05278 -00236

a The best 2 and 3 parameter models according to R2 and MSE are included in the tableb Mean square errorC Mallows e(p) statistic (Hocking 1976)

640 JOURNAL OF ECONOMIC ENTOMOLOGY Vol 81 no 2

References Cited

Brett C H amp M J Sullivan 1974 The use of re-sistant varieties and other cultural practices for con-trol of insects on crucifers in North Carolina NCState Univ Agric Exp Stn Bull 449

Cantelo W W amp L L Sanford 1984 Insect pop-ulation response to mixed and uniform plantings ofresistant and susceptible plant material EnvironEntomol 13 1443-1445

Chalfant R B W H Denton D J Schuster amp R BWorkman 1979 Management of cabbage cater-pillars in Florida and Georgia by using visual damagethresholds J Econ Entomol 72 411-413

Creighton C S T L McFadden amp M L Robbins1975 Complementary influence of host plant re-sistance on microbial chemical control of cabbage cat-erpillars HortScience 10 487-488

Dickson M H amp C J Eckenrode 1975a Variationin Brassica oeracea resistance to cabbage looper andimported cabbage worm in the greenhouse and fieldJ Econ Entomol 68 757-760

1975b Breeding for resistance in cabbage and cau-liflower to cabbage looper imported cabbagewormand diamondback moth J Am Soc Hortic Sci 105782-785

Dickson M H amp D H Wallace 1986 Cabbagebreeding pp 395-432 In M J Bassett [ed] Breedingvegetable crops AVI Westport Conn

Dunn J A amp D P H Kempton 1976 Varietaldifferences in the susceptibility of Brussels sprouts tolepidopterous pests Ann Appl BioI 82 11-19

Greene G L W C Genung R B Workman amp E GKelsheimer 1969 Cabbage looper control in Flor-ida-a cooperative program J Econ Entomol 62798-800

Harrison P K amp R W Brubaker 1943 The relativeabundance of cabbage caterpillars on cole crops grownunder similar conditions J Econ Entomol 36 589-592

Hocking R R 1976 The analysis and selection ofvariables in linear regression Biometrics 32 1-50

Kim T H 1979 Brassica oeracea L resistance tocabbage caterpillars in New York State MS thesisCornell Univ Ithaca NY

Latheef M A amp R D Irwin 1979 Factors affectingoviposition of Pieris rapae on cabbage EnvironEntomol 8 606-609

Lin J C J Eckenrode amp M H Dickson 1983 Vari-ation in Brassica oeracea resistance to diamondbackmoth (Lepidoptera Plutellidae) J Econ Entomol76 1423-1427

Pimentel D 1961 An evaluation of insect resistancein broccoli Brussels sprouts cabbage collards andkale J Econ Entomol 54 156-158

Radcliffe E B amp R K Chapman 1965a The rel-ative resistance to insect attack of three cabbage va-rieties at different stages of plant maturity AnnEntomol Soc Am 58 897-902

1965b Seasonal shifts in the relative resistance to in-sect attack of eight commercial cabbage varietiesAnn Entomol Soc Am 58 892-897

1966a Plant resistance to insect attack in commercialcabbage varieties J Econ Entomol 59 116-120

1966b Varietal resistance to insect attack in variouscruciferous crops J Econ Entomol 59 120-125

Shelton A M J T Andaloro amp J Barnard 1982Effects of cabbage looper imported cabbagewormand diamondback moth on fresh market and pro-cessing cabbage J Econ Entomol 75 742-745

Shelton A M R F Becker amp J T Andaloro 1983Varietal resistance to onion thrips (ThysanopteraThripidae) in processing cabbage J Econ Entomol76 85-86

Sirrine F A amp V H Lowe 1894 Insects affectinglate cabbage notes on the stalk borer insecticidesNY Agric Exp Stn (Geneva) Bull 83 657-685

Wolfenbarger D A 1967 Variation in Brassica sppto cabbage looper infestations J Econ Entomol 60874-875

Received for publication 13 March 1987 accepted 21October 1987

April1988 SHELTON ET AL RESISTANCE OF CABBAGE TO LEPIDOPTERA AND T tabaci 637

Table 1 Selected regression models for predicting visual damage rating (see text) by Lepidoptera on fresh marketcabbage in Geneva NY in 1982 and 1983

Parameter estimates forNo Seasonal total ofpara- RZ RMS C(p) Head wt

meters Intercept P Frame wtA rapae xylostela T ni LU

19822 01196 1648 3461 3308 0045 01012 01193 1648 3487 3055 0098 02063 01364 1634 3677 3251 0048 0037 00743 01337 1639 3962 3061 0036 0075 01634 01483 1629 4407 2818 0032 0047 0071 0172

19832 0250 1270 1970 2857 0091 02683 0263 1262 2429 3042 0095 -0090 03103 0260 1268 2820 2618 0025 0109 02744 0272 1261 3341 2806 0024 0112 -0088 0315

Larval units (see text)

sign for the frame weight parameter estimates isnegative indicating that damage rating decreasesas frame size increases This could be because dam-age is less obvious on a large plant frame Basedon these results and all of the selection criteriacombined we included head weight and seasonaltotals for either P xylostella A rapae T ni orlarval units in regression models used to test forvarietal differences in the relationship of these vari-ables to damage rating Thus the next step was totest for differences in these relationships ie inslopes or intercepts of regression equations amongthe different varieties

The slope for seasonal pest total is an estimateof the increase in damage rating for a unit increasein seasonal pest total When the slope is the samefor each variety the damage is inflicted by pestsat a similar rate for all the varieties at the popu-lation densities experienced during the study Ifthe slopes are the same for both seasonal pest totaland head weight we can compare the varieties atany common seasonal pest total and head weightand expect the same differences in damage ratingsIf the slopes are different the predicted differencesin damage ratings among the varieties at a commonseasonal pest total and head weight will depend onthe common pest total and head weight chosen Ifthe interaction term between variety and seasonalpest total is not significant in the analysis of vari-ance we conclude that the slopes are not signifi-cantly different In this case a type II error con-cluding that the slopes are not significantly differentwhen they are is more serious than a type I errorthus we use a higher significance level than 005to reduce the probability of making a type II errorThe term for interaction between variety and sea-sonal pest total was not significant for any of themodels tested in 1982 (F 5 165 df = 17 18 P 010) or for the model containing P xylostella in1983 (F 5 154 df = 23 24 P gt 015) indicatingthat the slopes are not significantly different amongthe varieties tested Exceptions were models con-

taining larval units A rapae or T ni in 1983 (F 192 df = 23 24 P 5 006) which is an indicationthat the slopes of these regression equations maydiffer for different varieties when some of the mea-sures of seasonal pest total are used in some yearsThe term for interaction between variety and headweight was not significant in any of the models in1982 (F 5 121 df = 17 18 P gt 0347) or in themodel including P xylostella in 1983 (F 5 115df = 23 24 P gt 036) indicating that the increasein damage rating for a unit increase in head weightis the same for each variety Interaction betweenvariety and head weight was significant when Arapae T ni or larval units were included in modelsin 1983 (F 192 df = 23 24 P 5 0059) indi-cating that the relationship of head weight to dam-age rating may also differ at times among the va-rieties

For each model the effects of seasonal pest totalwere significant (F 6405 df = 1 52 P 5 00001in 1982 F 1285 df = 124 P 5 00015 in 1983)as were the effects of head weight (F 2940 df =1 52 P 5 00001 in 1982 F 1160 df = 1 24P 5 00001 in 1983) Damage by each of these pestpopulations over time on different parts of the plantmay not be best described by the total number oflarvae found during biweekly samples A bettermeasure of the impact of pest populations on dam-age rating (eg more frequent samples and a func-tion to calculate expected cumulative area of feed-ing over time) would improve the analysisHowever the terms for pest total and head weightexplained enough variation in damage rating tomerit their inclusion in the models as covariatesbefore testing for effects of variety

Testing for effects of variety in the analysis ofvariance after the variation due to seasonal pesttotal and head weight has been explained gives anindication of whether or not the intercept termsare different in regression equations for the differ-ent varieties The effects of variety on the interceptterms were significant (F 1206 df = 1752 P lt

638 JOURNAL OF ECONOMIC ENTOMOLOGY Vol 81 no 2

001 in 1982 F ~ 20 df = 23 24 P lt 005 in1983) for all of the models indicating that theintercepts are different among the varieties Formodels in which the slopes are not significantlydifferent the differences in intercept terms providean estimate of differences in damage ratings forany common seasonal pest total and head weightwithin the range of these variables present duringthe study Estimates of these intercept terms forthe best of the regression models damage ratingpredicted by P xylostella and head weight aregiven in Table 2 Estimates for the other regressionmodels were very similar to those in Table 2 Wespeculate that differences in the intercept termsamong the varieties is caused either by toleranceor antibiosis Tolerance could be the result of dif-ferences in location of feeding on these varietiesbecause the damage rating system gives higher val-ues to feeding on the head and wrapper leaves thanon the frame leaves Differences in appearance offeeding damage on the different varieties couldalso result in differences in intercepts because manysmall or irregular holes may receive higher damageratings than fewer large round holes or strip feed-ing at the edge of the leaf Antibiosis could resultin pests being present but not feeding or feedingless than on other varieties If the experimental orred varieties are removed from the analysis thevarietal effects are still significant (F ~ 359 df =15 46 in 1982 F ~ 283 df = 21 64 in 1983 bothP lt 001) but if they are both removed the dif-ferences are no longer significant (F S 082 df =13 40 P gt 063 in 1982 F s 203 df = 19 58P gt 005 for all but A rapae in 1983) Thus theimportant differences lie between these two groupsand the other varieties Actual differences in dam-age ratings could be because of either a differentcapacity for injury at a given pest infestation levelor to varietal differences in numbers of pests pres-ent In general differences in intercepts corre-sponded with differences in damage ratings Forexample the red and experimental varieties hadboth lower damage ratings and lower interceptsthan the other varieties

Analysis of seasonal egg total showed significantdifferences (P = 005) for each species except forA rapae in 1983 (F ~ 14 df = 17 54 in 1982F = 12 df = 23 72 in 1983 both P s 0047) Thered and experimental varieties were consistentlylower in seasonal egg totals except for T ni on redcabbages in 1983 Without the regression modelsthe lower damage ratings could be attributed tofewer eggs (ie nonpreference) However theregression analysis indicates that even with thesame number of larvae on the red or experimentallines the damage ratings would be lower (sug-gesting tolerance or antibiosis) For the experi-mental lines antibiosis seems most likely becausethese lines had a much higher ratio of eggs tosubsequent larvae than the other varieties Thissuggests antibiosis or lack of feeding by early instarsand explains the very low intercepts found in the

regression analysis and low damage ratings seen inthe tests Additional data (unpublished) indicatesthat at least for A rapae resistance is manifestedin the first instars inability to feed on glossy resis-tant varieties

Thrips tabaci Aswas the case with Lepidopterathrips numbers alone did not provide the best pre-dictor of damage rating when potential regressionmodels were compared for the 1983 data (Table3) The addition of some measure of head sizeeither head weight or head diameter improved R2by ca 6 above square root transformed seasonalthrips totals alone and brought the C(p) statisticclose to the number of parameters in the modelNeither measure of head size alone explained asmuch of the variation in the data as thrips numbersOur conclusion is that a measure of head size andthrips population density should be included inregression models before testing for effects of va-rieties For the five varieties on which thrips weresampled in 1982 neither head weight nor headdiameter explained a significant amount of varia-tion in the model and they were not included inthe analysis

The interaction between seasonal thrips total andvariety was not significant (F = 0803 df = 4 10P gt 02 in 1982 F = 174 df = 23 24 P = 0093in 1983) indicating a common positive slope forthe five varieties tested in 1982 and the 24 varietiestested in 1983 The interaction term between va-riety and head weight was not significant either(F = 115 df = 23 24 P = 037) for the 1983 dataVarietal effects were significant (F = 653 df = 414 in 1982 F = 838 df = 23 24 in 1983 bothP lt 001) indicating because the slopes were notdifferent the same differences in the thrips damagerating at any given seasonal total number of thrips(Table 2) Again we speculate that this could becaused either by tolerance or antibiosis One ex-ample of tolerance could be thrips moving andfeeding deeper into the head in more susceptiblevarieties because the damage rating depends onboth the severity and depth of feeding damageStoner amp Shelton (unpublished data) found thattotal numbers of thrips on certain susceptible va-rieties were the same as found on resistant varietiesbut a greater proportion of the thrips were in thehead on the susceptible varieties Another exampleof tolerance could be the way the plant respondsto thrips feeding Original feeding damage is vis-ible only under a microscope but it eventuallybecomes a rough bronzed edema This bronzingand roughening could occur to different degreesin different varieties

Differences in susceptibility to thrips damagecould not be attributed strictly to any obvious cat-egory (experimental red older varieties savoy orseed company lines) When any of these categoriesor a combination of them were restricted from theanalysis the slopes of regression lines for the re-maining varieties were still not significantly dif-ferent and differences still remained in the inter-

April1988 SHELTONET AL RESISTANCEOF CABBAGETO LEPIDOPTERAANDT tabaci 639

Table 2 Parameter estimates for regression models grouped by variety of visual damage rating predicted byseasonal pest total and cabbage hcad weight

Seasonal total of

P xylostella T tabaci

1982 (SE) 1983 (SE) 1982 (SE) 1983 (SE)

Common slope for pest 0019 (0023) -00532 (0034) 00666 (00237)Commoo slope for head wt 0183 (Olll) 01237 (00494) 0242 (0112)

Intercept for varietyHitoma 3272 (0450) 4685 (0368) 1360 (0350)SlIpprgrft1l 3878 (0388) 5004 (0365) 1761 (0341) 1499 (0420)Bartolo 3839 (0390) 4076 (0327) 1415 (0305)Excel 4053 (0429) 4983 (0373) 1978 (0354)Suplrdane 3858 (0435) 4239 (0430) 0766 (0429)Red Head 1919 (0443) 4127 (0420) 0889 (0458)Hinova 3948 (0436) 4222 (0401) 0939 (0389)Market Prizl 3754 (0378) 4641 (0445) 1355 (0401) 1120 (0658)Dcema Extra 3917 (0381) 4499 (0320) 1080 (0272)Grff1l Vinler 4348 (0411) 4231 (0353) 0675 (0299)Lilli bull Rock 3529 (0474) 4493 (0416) 0366 (0375)SUIlUP 4573 (0290) 0206 (0343) 0794 (0289)Houndup 3590 (0575) 4559 (0476) 0232 (0436)Cnbullva 8329 -0036 (0313) 1281 (0304) 0750 (0287)Hed Danish 2560 (0425) 2564 (0349) -0308 (0270)Titanic 90 3868 (0522) 4427 (0419) 1026 (0389) 0172 (0400)Sptcial Gold bulln Acre 5033 (0299) 0801 (0326)Falcon 3564 (0544) 4195 (0363) 0554 (0376) -00277 (0327)Chieftain Savoy 3826 (0396) 4305 (0366) -0044 (0257)King Cole 4351 (0349) 0373 (0333)Supbullrbulltt bull 5021 (0391) 0469 (0360)Danish Ballhead 4475 (0281) 0743 (0242)Early Jtfsey Wakefield 4443 (0288) -0133 (0242)Gbullneva 8395 0205 (0287) 0329 (0276) -0112 (0243)

cepts However the final damage rating dependson both the relationship of thrips numbers to dam-age and the number of thrips present Generallydifferences in damage ratings were consistent withdifferences in the intercept terms in the regressionmodel for the 24 varieties tested However in somecases similar damage ratings had different causesIn 1982 for example Sunup had a lower interceptfor the regression of damage rating on thrips num-bers than Titanic but the seasonal total of thripsin Sunup was great enough that the damage ratingwas similar to that of Titanic In 1983 the situ-ation was reversed for the two varieties Againthese differences in numbers of thrips present maynot appear in a nonchoice situation

Conclusions drawn from these data on cabbageresistance to Lepidoptera and T tabaci can beviewed in two different ways From a standpoint

of screening varieties we have documented vary-ing levels of resistance in fresh market cabbagevarieties to T tabaci and Lepidoptera and notedthe patterns of resistance were quite different forthe two different kinds of pests Varieties showinglittle damage caused by either may prove usefulto breeding programs Second this new approachused to compare the varieties regression modelsthat compare damage rating at a common pestpopulation and plant size provides a better indi-cation of expected differences in damage ratingsin commercial situations than if damage ratingsalone were compared

Acknowledgment

The authors acknowledge the help of Kimberly Stonerin reviewing the literature on crucifer resistance

Table 3 Candidate regression models for predicting visual damage rating by T tabaci on fresh market cabbage inGeneva NY 1983

No Parameter estimates forpara- R2 MSEb e(p) Seasonal total

meters Intercept thrips Head wt Head diam

1 04688 04717 140196 06133 012812 05215 04295 55088 -01873 01047 007392 05382 04145 21729 04301 00946 042053 05390 04182 40000 06394 00935 05278 -00236

a The best 2 and 3 parameter models according to R2 and MSE are included in the tableb Mean square errorC Mallows e(p) statistic (Hocking 1976)

640 JOURNAL OF ECONOMIC ENTOMOLOGY Vol 81 no 2

References Cited

Brett C H amp M J Sullivan 1974 The use of re-sistant varieties and other cultural practices for con-trol of insects on crucifers in North Carolina NCState Univ Agric Exp Stn Bull 449

Cantelo W W amp L L Sanford 1984 Insect pop-ulation response to mixed and uniform plantings ofresistant and susceptible plant material EnvironEntomol 13 1443-1445

Chalfant R B W H Denton D J Schuster amp R BWorkman 1979 Management of cabbage cater-pillars in Florida and Georgia by using visual damagethresholds J Econ Entomol 72 411-413

Creighton C S T L McFadden amp M L Robbins1975 Complementary influence of host plant re-sistance on microbial chemical control of cabbage cat-erpillars HortScience 10 487-488

Dickson M H amp C J Eckenrode 1975a Variationin Brassica oeracea resistance to cabbage looper andimported cabbage worm in the greenhouse and fieldJ Econ Entomol 68 757-760

1975b Breeding for resistance in cabbage and cau-liflower to cabbage looper imported cabbagewormand diamondback moth J Am Soc Hortic Sci 105782-785

Dickson M H amp D H Wallace 1986 Cabbagebreeding pp 395-432 In M J Bassett [ed] Breedingvegetable crops AVI Westport Conn

Dunn J A amp D P H Kempton 1976 Varietaldifferences in the susceptibility of Brussels sprouts tolepidopterous pests Ann Appl BioI 82 11-19

Greene G L W C Genung R B Workman amp E GKelsheimer 1969 Cabbage looper control in Flor-ida-a cooperative program J Econ Entomol 62798-800

Harrison P K amp R W Brubaker 1943 The relativeabundance of cabbage caterpillars on cole crops grownunder similar conditions J Econ Entomol 36 589-592

Hocking R R 1976 The analysis and selection ofvariables in linear regression Biometrics 32 1-50

Kim T H 1979 Brassica oeracea L resistance tocabbage caterpillars in New York State MS thesisCornell Univ Ithaca NY

Latheef M A amp R D Irwin 1979 Factors affectingoviposition of Pieris rapae on cabbage EnvironEntomol 8 606-609

Lin J C J Eckenrode amp M H Dickson 1983 Vari-ation in Brassica oeracea resistance to diamondbackmoth (Lepidoptera Plutellidae) J Econ Entomol76 1423-1427

Pimentel D 1961 An evaluation of insect resistancein broccoli Brussels sprouts cabbage collards andkale J Econ Entomol 54 156-158

Radcliffe E B amp R K Chapman 1965a The rel-ative resistance to insect attack of three cabbage va-rieties at different stages of plant maturity AnnEntomol Soc Am 58 897-902

1965b Seasonal shifts in the relative resistance to in-sect attack of eight commercial cabbage varietiesAnn Entomol Soc Am 58 892-897

1966a Plant resistance to insect attack in commercialcabbage varieties J Econ Entomol 59 116-120

1966b Varietal resistance to insect attack in variouscruciferous crops J Econ Entomol 59 120-125

Shelton A M J T Andaloro amp J Barnard 1982Effects of cabbage looper imported cabbagewormand diamondback moth on fresh market and pro-cessing cabbage J Econ Entomol 75 742-745

Shelton A M R F Becker amp J T Andaloro 1983Varietal resistance to onion thrips (ThysanopteraThripidae) in processing cabbage J Econ Entomol76 85-86

Sirrine F A amp V H Lowe 1894 Insects affectinglate cabbage notes on the stalk borer insecticidesNY Agric Exp Stn (Geneva) Bull 83 657-685

Wolfenbarger D A 1967 Variation in Brassica sppto cabbage looper infestations J Econ Entomol 60874-875

Received for publication 13 March 1987 accepted 21October 1987

638 JOURNAL OF ECONOMIC ENTOMOLOGY Vol 81 no 2

001 in 1982 F ~ 20 df = 23 24 P lt 005 in1983) for all of the models indicating that theintercepts are different among the varieties Formodels in which the slopes are not significantlydifferent the differences in intercept terms providean estimate of differences in damage ratings forany common seasonal pest total and head weightwithin the range of these variables present duringthe study Estimates of these intercept terms forthe best of the regression models damage ratingpredicted by P xylostella and head weight aregiven in Table 2 Estimates for the other regressionmodels were very similar to those in Table 2 Wespeculate that differences in the intercept termsamong the varieties is caused either by toleranceor antibiosis Tolerance could be the result of dif-ferences in location of feeding on these varietiesbecause the damage rating system gives higher val-ues to feeding on the head and wrapper leaves thanon the frame leaves Differences in appearance offeeding damage on the different varieties couldalso result in differences in intercepts because manysmall or irregular holes may receive higher damageratings than fewer large round holes or strip feed-ing at the edge of the leaf Antibiosis could resultin pests being present but not feeding or feedingless than on other varieties If the experimental orred varieties are removed from the analysis thevarietal effects are still significant (F ~ 359 df =15 46 in 1982 F ~ 283 df = 21 64 in 1983 bothP lt 001) but if they are both removed the dif-ferences are no longer significant (F S 082 df =13 40 P gt 063 in 1982 F s 203 df = 19 58P gt 005 for all but A rapae in 1983) Thus theimportant differences lie between these two groupsand the other varieties Actual differences in dam-age ratings could be because of either a differentcapacity for injury at a given pest infestation levelor to varietal differences in numbers of pests pres-ent In general differences in intercepts corre-sponded with differences in damage ratings Forexample the red and experimental varieties hadboth lower damage ratings and lower interceptsthan the other varieties

Analysis of seasonal egg total showed significantdifferences (P = 005) for each species except forA rapae in 1983 (F ~ 14 df = 17 54 in 1982F = 12 df = 23 72 in 1983 both P s 0047) Thered and experimental varieties were consistentlylower in seasonal egg totals except for T ni on redcabbages in 1983 Without the regression modelsthe lower damage ratings could be attributed tofewer eggs (ie nonpreference) However theregression analysis indicates that even with thesame number of larvae on the red or experimentallines the damage ratings would be lower (sug-gesting tolerance or antibiosis) For the experi-mental lines antibiosis seems most likely becausethese lines had a much higher ratio of eggs tosubsequent larvae than the other varieties Thissuggests antibiosis or lack of feeding by early instarsand explains the very low intercepts found in the

regression analysis and low damage ratings seen inthe tests Additional data (unpublished) indicatesthat at least for A rapae resistance is manifestedin the first instars inability to feed on glossy resis-tant varieties

Thrips tabaci Aswas the case with Lepidopterathrips numbers alone did not provide the best pre-dictor of damage rating when potential regressionmodels were compared for the 1983 data (Table3) The addition of some measure of head sizeeither head weight or head diameter improved R2by ca 6 above square root transformed seasonalthrips totals alone and brought the C(p) statisticclose to the number of parameters in the modelNeither measure of head size alone explained asmuch of the variation in the data as thrips numbersOur conclusion is that a measure of head size andthrips population density should be included inregression models before testing for effects of va-rieties For the five varieties on which thrips weresampled in 1982 neither head weight nor headdiameter explained a significant amount of varia-tion in the model and they were not included inthe analysis

The interaction between seasonal thrips total andvariety was not significant (F = 0803 df = 4 10P gt 02 in 1982 F = 174 df = 23 24 P = 0093in 1983) indicating a common positive slope forthe five varieties tested in 1982 and the 24 varietiestested in 1983 The interaction term between va-riety and head weight was not significant either(F = 115 df = 23 24 P = 037) for the 1983 dataVarietal effects were significant (F = 653 df = 414 in 1982 F = 838 df = 23 24 in 1983 bothP lt 001) indicating because the slopes were notdifferent the same differences in the thrips damagerating at any given seasonal total number of thrips(Table 2) Again we speculate that this could becaused either by tolerance or antibiosis One ex-ample of tolerance could be thrips moving andfeeding deeper into the head in more susceptiblevarieties because the damage rating depends onboth the severity and depth of feeding damageStoner amp Shelton (unpublished data) found thattotal numbers of thrips on certain susceptible va-rieties were the same as found on resistant varietiesbut a greater proportion of the thrips were in thehead on the susceptible varieties Another exampleof tolerance could be the way the plant respondsto thrips feeding Original feeding damage is vis-ible only under a microscope but it eventuallybecomes a rough bronzed edema This bronzingand roughening could occur to different degreesin different varieties

Differences in susceptibility to thrips damagecould not be attributed strictly to any obvious cat-egory (experimental red older varieties savoy orseed company lines) When any of these categoriesor a combination of them were restricted from theanalysis the slopes of regression lines for the re-maining varieties were still not significantly dif-ferent and differences still remained in the inter-

April1988 SHELTONET AL RESISTANCEOF CABBAGETO LEPIDOPTERAANDT tabaci 639

Table 2 Parameter estimates for regression models grouped by variety of visual damage rating predicted byseasonal pest total and cabbage hcad weight

Seasonal total of

P xylostella T tabaci

1982 (SE) 1983 (SE) 1982 (SE) 1983 (SE)

Common slope for pest 0019 (0023) -00532 (0034) 00666 (00237)Commoo slope for head wt 0183 (Olll) 01237 (00494) 0242 (0112)

Intercept for varietyHitoma 3272 (0450) 4685 (0368) 1360 (0350)SlIpprgrft1l 3878 (0388) 5004 (0365) 1761 (0341) 1499 (0420)Bartolo 3839 (0390) 4076 (0327) 1415 (0305)Excel 4053 (0429) 4983 (0373) 1978 (0354)Suplrdane 3858 (0435) 4239 (0430) 0766 (0429)Red Head 1919 (0443) 4127 (0420) 0889 (0458)Hinova 3948 (0436) 4222 (0401) 0939 (0389)Market Prizl 3754 (0378) 4641 (0445) 1355 (0401) 1120 (0658)Dcema Extra 3917 (0381) 4499 (0320) 1080 (0272)Grff1l Vinler 4348 (0411) 4231 (0353) 0675 (0299)Lilli bull Rock 3529 (0474) 4493 (0416) 0366 (0375)SUIlUP 4573 (0290) 0206 (0343) 0794 (0289)Houndup 3590 (0575) 4559 (0476) 0232 (0436)Cnbullva 8329 -0036 (0313) 1281 (0304) 0750 (0287)Hed Danish 2560 (0425) 2564 (0349) -0308 (0270)Titanic 90 3868 (0522) 4427 (0419) 1026 (0389) 0172 (0400)Sptcial Gold bulln Acre 5033 (0299) 0801 (0326)Falcon 3564 (0544) 4195 (0363) 0554 (0376) -00277 (0327)Chieftain Savoy 3826 (0396) 4305 (0366) -0044 (0257)King Cole 4351 (0349) 0373 (0333)Supbullrbulltt bull 5021 (0391) 0469 (0360)Danish Ballhead 4475 (0281) 0743 (0242)Early Jtfsey Wakefield 4443 (0288) -0133 (0242)Gbullneva 8395 0205 (0287) 0329 (0276) -0112 (0243)

cepts However the final damage rating dependson both the relationship of thrips numbers to dam-age and the number of thrips present Generallydifferences in damage ratings were consistent withdifferences in the intercept terms in the regressionmodel for the 24 varieties tested However in somecases similar damage ratings had different causesIn 1982 for example Sunup had a lower interceptfor the regression of damage rating on thrips num-bers than Titanic but the seasonal total of thripsin Sunup was great enough that the damage ratingwas similar to that of Titanic In 1983 the situ-ation was reversed for the two varieties Againthese differences in numbers of thrips present maynot appear in a nonchoice situation

Conclusions drawn from these data on cabbageresistance to Lepidoptera and T tabaci can beviewed in two different ways From a standpoint

of screening varieties we have documented vary-ing levels of resistance in fresh market cabbagevarieties to T tabaci and Lepidoptera and notedthe patterns of resistance were quite different forthe two different kinds of pests Varieties showinglittle damage caused by either may prove usefulto breeding programs Second this new approachused to compare the varieties regression modelsthat compare damage rating at a common pestpopulation and plant size provides a better indi-cation of expected differences in damage ratingsin commercial situations than if damage ratingsalone were compared

Acknowledgment

The authors acknowledge the help of Kimberly Stonerin reviewing the literature on crucifer resistance

Table 3 Candidate regression models for predicting visual damage rating by T tabaci on fresh market cabbage inGeneva NY 1983

No Parameter estimates forpara- R2 MSEb e(p) Seasonal total

meters Intercept thrips Head wt Head diam

1 04688 04717 140196 06133 012812 05215 04295 55088 -01873 01047 007392 05382 04145 21729 04301 00946 042053 05390 04182 40000 06394 00935 05278 -00236

a The best 2 and 3 parameter models according to R2 and MSE are included in the tableb Mean square errorC Mallows e(p) statistic (Hocking 1976)

640 JOURNAL OF ECONOMIC ENTOMOLOGY Vol 81 no 2

References Cited

Brett C H amp M J Sullivan 1974 The use of re-sistant varieties and other cultural practices for con-trol of insects on crucifers in North Carolina NCState Univ Agric Exp Stn Bull 449

Cantelo W W amp L L Sanford 1984 Insect pop-ulation response to mixed and uniform plantings ofresistant and susceptible plant material EnvironEntomol 13 1443-1445

Chalfant R B W H Denton D J Schuster amp R BWorkman 1979 Management of cabbage cater-pillars in Florida and Georgia by using visual damagethresholds J Econ Entomol 72 411-413

Creighton C S T L McFadden amp M L Robbins1975 Complementary influence of host plant re-sistance on microbial chemical control of cabbage cat-erpillars HortScience 10 487-488

Dickson M H amp C J Eckenrode 1975a Variationin Brassica oeracea resistance to cabbage looper andimported cabbage worm in the greenhouse and fieldJ Econ Entomol 68 757-760

1975b Breeding for resistance in cabbage and cau-liflower to cabbage looper imported cabbagewormand diamondback moth J Am Soc Hortic Sci 105782-785

Dickson M H amp D H Wallace 1986 Cabbagebreeding pp 395-432 In M J Bassett [ed] Breedingvegetable crops AVI Westport Conn

Dunn J A amp D P H Kempton 1976 Varietaldifferences in the susceptibility of Brussels sprouts tolepidopterous pests Ann Appl BioI 82 11-19

Greene G L W C Genung R B Workman amp E GKelsheimer 1969 Cabbage looper control in Flor-ida-a cooperative program J Econ Entomol 62798-800

Harrison P K amp R W Brubaker 1943 The relativeabundance of cabbage caterpillars on cole crops grownunder similar conditions J Econ Entomol 36 589-592

Hocking R R 1976 The analysis and selection ofvariables in linear regression Biometrics 32 1-50

Kim T H 1979 Brassica oeracea L resistance tocabbage caterpillars in New York State MS thesisCornell Univ Ithaca NY

Latheef M A amp R D Irwin 1979 Factors affectingoviposition of Pieris rapae on cabbage EnvironEntomol 8 606-609

Lin J C J Eckenrode amp M H Dickson 1983 Vari-ation in Brassica oeracea resistance to diamondbackmoth (Lepidoptera Plutellidae) J Econ Entomol76 1423-1427

Pimentel D 1961 An evaluation of insect resistancein broccoli Brussels sprouts cabbage collards andkale J Econ Entomol 54 156-158

Radcliffe E B amp R K Chapman 1965a The rel-ative resistance to insect attack of three cabbage va-rieties at different stages of plant maturity AnnEntomol Soc Am 58 897-902

1965b Seasonal shifts in the relative resistance to in-sect attack of eight commercial cabbage varietiesAnn Entomol Soc Am 58 892-897

1966a Plant resistance to insect attack in commercialcabbage varieties J Econ Entomol 59 116-120

1966b Varietal resistance to insect attack in variouscruciferous crops J Econ Entomol 59 120-125

Shelton A M J T Andaloro amp J Barnard 1982Effects of cabbage looper imported cabbagewormand diamondback moth on fresh market and pro-cessing cabbage J Econ Entomol 75 742-745

Shelton A M R F Becker amp J T Andaloro 1983Varietal resistance to onion thrips (ThysanopteraThripidae) in processing cabbage J Econ Entomol76 85-86

Sirrine F A amp V H Lowe 1894 Insects affectinglate cabbage notes on the stalk borer insecticidesNY Agric Exp Stn (Geneva) Bull 83 657-685

Wolfenbarger D A 1967 Variation in Brassica sppto cabbage looper infestations J Econ Entomol 60874-875

Received for publication 13 March 1987 accepted 21October 1987

April1988 SHELTONET AL RESISTANCEOF CABBAGETO LEPIDOPTERAANDT tabaci 639

Table 2 Parameter estimates for regression models grouped by variety of visual damage rating predicted byseasonal pest total and cabbage hcad weight

Seasonal total of

P xylostella T tabaci

1982 (SE) 1983 (SE) 1982 (SE) 1983 (SE)

Common slope for pest 0019 (0023) -00532 (0034) 00666 (00237)Commoo slope for head wt 0183 (Olll) 01237 (00494) 0242 (0112)

Intercept for varietyHitoma 3272 (0450) 4685 (0368) 1360 (0350)SlIpprgrft1l 3878 (0388) 5004 (0365) 1761 (0341) 1499 (0420)Bartolo 3839 (0390) 4076 (0327) 1415 (0305)Excel 4053 (0429) 4983 (0373) 1978 (0354)Suplrdane 3858 (0435) 4239 (0430) 0766 (0429)Red Head 1919 (0443) 4127 (0420) 0889 (0458)Hinova 3948 (0436) 4222 (0401) 0939 (0389)Market Prizl 3754 (0378) 4641 (0445) 1355 (0401) 1120 (0658)Dcema Extra 3917 (0381) 4499 (0320) 1080 (0272)Grff1l Vinler 4348 (0411) 4231 (0353) 0675 (0299)Lilli bull Rock 3529 (0474) 4493 (0416) 0366 (0375)SUIlUP 4573 (0290) 0206 (0343) 0794 (0289)Houndup 3590 (0575) 4559 (0476) 0232 (0436)Cnbullva 8329 -0036 (0313) 1281 (0304) 0750 (0287)Hed Danish 2560 (0425) 2564 (0349) -0308 (0270)Titanic 90 3868 (0522) 4427 (0419) 1026 (0389) 0172 (0400)Sptcial Gold bulln Acre 5033 (0299) 0801 (0326)Falcon 3564 (0544) 4195 (0363) 0554 (0376) -00277 (0327)Chieftain Savoy 3826 (0396) 4305 (0366) -0044 (0257)King Cole 4351 (0349) 0373 (0333)Supbullrbulltt bull 5021 (0391) 0469 (0360)Danish Ballhead 4475 (0281) 0743 (0242)Early Jtfsey Wakefield 4443 (0288) -0133 (0242)Gbullneva 8395 0205 (0287) 0329 (0276) -0112 (0243)

cepts However the final damage rating dependson both the relationship of thrips numbers to dam-age and the number of thrips present Generallydifferences in damage ratings were consistent withdifferences in the intercept terms in the regressionmodel for the 24 varieties tested However in somecases similar damage ratings had different causesIn 1982 for example Sunup had a lower interceptfor the regression of damage rating on thrips num-bers than Titanic but the seasonal total of thripsin Sunup was great enough that the damage ratingwas similar to that of Titanic In 1983 the situ-ation was reversed for the two varieties Againthese differences in numbers of thrips present maynot appear in a nonchoice situation

Conclusions drawn from these data on cabbageresistance to Lepidoptera and T tabaci can beviewed in two different ways From a standpoint

of screening varieties we have documented vary-ing levels of resistance in fresh market cabbagevarieties to T tabaci and Lepidoptera and notedthe patterns of resistance were quite different forthe two different kinds of pests Varieties showinglittle damage caused by either may prove usefulto breeding programs Second this new approachused to compare the varieties regression modelsthat compare damage rating at a common pestpopulation and plant size provides a better indi-cation of expected differences in damage ratingsin commercial situations than if damage ratingsalone were compared

Acknowledgment

The authors acknowledge the help of Kimberly Stonerin reviewing the literature on crucifer resistance

Table 3 Candidate regression models for predicting visual damage rating by T tabaci on fresh market cabbage inGeneva NY 1983

No Parameter estimates forpara- R2 MSEb e(p) Seasonal total

meters Intercept thrips Head wt Head diam

1 04688 04717 140196 06133 012812 05215 04295 55088 -01873 01047 007392 05382 04145 21729 04301 00946 042053 05390 04182 40000 06394 00935 05278 -00236

a The best 2 and 3 parameter models according to R2 and MSE are included in the tableb Mean square errorC Mallows e(p) statistic (Hocking 1976)

640 JOURNAL OF ECONOMIC ENTOMOLOGY Vol 81 no 2

References Cited

Brett C H amp M J Sullivan 1974 The use of re-sistant varieties and other cultural practices for con-trol of insects on crucifers in North Carolina NCState Univ Agric Exp Stn Bull 449

Cantelo W W amp L L Sanford 1984 Insect pop-ulation response to mixed and uniform plantings ofresistant and susceptible plant material EnvironEntomol 13 1443-1445

Chalfant R B W H Denton D J Schuster amp R BWorkman 1979 Management of cabbage cater-pillars in Florida and Georgia by using visual damagethresholds J Econ Entomol 72 411-413

Creighton C S T L McFadden amp M L Robbins1975 Complementary influence of host plant re-sistance on microbial chemical control of cabbage cat-erpillars HortScience 10 487-488

Dickson M H amp C J Eckenrode 1975a Variationin Brassica oeracea resistance to cabbage looper andimported cabbage worm in the greenhouse and fieldJ Econ Entomol 68 757-760

1975b Breeding for resistance in cabbage and cau-liflower to cabbage looper imported cabbagewormand diamondback moth J Am Soc Hortic Sci 105782-785

Dickson M H amp D H Wallace 1986 Cabbagebreeding pp 395-432 In M J Bassett [ed] Breedingvegetable crops AVI Westport Conn

Dunn J A amp D P H Kempton 1976 Varietaldifferences in the susceptibility of Brussels sprouts tolepidopterous pests Ann Appl BioI 82 11-19

Greene G L W C Genung R B Workman amp E GKelsheimer 1969 Cabbage looper control in Flor-ida-a cooperative program J Econ Entomol 62798-800

Harrison P K amp R W Brubaker 1943 The relativeabundance of cabbage caterpillars on cole crops grownunder similar conditions J Econ Entomol 36 589-592

Hocking R R 1976 The analysis and selection ofvariables in linear regression Biometrics 32 1-50

Kim T H 1979 Brassica oeracea L resistance tocabbage caterpillars in New York State MS thesisCornell Univ Ithaca NY

Latheef M A amp R D Irwin 1979 Factors affectingoviposition of Pieris rapae on cabbage EnvironEntomol 8 606-609

Lin J C J Eckenrode amp M H Dickson 1983 Vari-ation in Brassica oeracea resistance to diamondbackmoth (Lepidoptera Plutellidae) J Econ Entomol76 1423-1427

Pimentel D 1961 An evaluation of insect resistancein broccoli Brussels sprouts cabbage collards andkale J Econ Entomol 54 156-158

Radcliffe E B amp R K Chapman 1965a The rel-ative resistance to insect attack of three cabbage va-rieties at different stages of plant maturity AnnEntomol Soc Am 58 897-902

1965b Seasonal shifts in the relative resistance to in-sect attack of eight commercial cabbage varietiesAnn Entomol Soc Am 58 892-897

1966a Plant resistance to insect attack in commercialcabbage varieties J Econ Entomol 59 116-120

1966b Varietal resistance to insect attack in variouscruciferous crops J Econ Entomol 59 120-125

Shelton A M J T Andaloro amp J Barnard 1982Effects of cabbage looper imported cabbagewormand diamondback moth on fresh market and pro-cessing cabbage J Econ Entomol 75 742-745

Shelton A M R F Becker amp J T Andaloro 1983Varietal resistance to onion thrips (ThysanopteraThripidae) in processing cabbage J Econ Entomol76 85-86

Sirrine F A amp V H Lowe 1894 Insects affectinglate cabbage notes on the stalk borer insecticidesNY Agric Exp Stn (Geneva) Bull 83 657-685

Wolfenbarger D A 1967 Variation in Brassica sppto cabbage looper infestations J Econ Entomol 60874-875

Received for publication 13 March 1987 accepted 21October 1987

640 JOURNAL OF ECONOMIC ENTOMOLOGY Vol 81 no 2

References Cited

Brett C H amp M J Sullivan 1974 The use of re-sistant varieties and other cultural practices for con-trol of insects on crucifers in North Carolina NCState Univ Agric Exp Stn Bull 449

Cantelo W W amp L L Sanford 1984 Insect pop-ulation response to mixed and uniform plantings ofresistant and susceptible plant material EnvironEntomol 13 1443-1445

Chalfant R B W H Denton D J Schuster amp R BWorkman 1979 Management of cabbage cater-pillars in Florida and Georgia by using visual damagethresholds J Econ Entomol 72 411-413

Creighton C S T L McFadden amp M L Robbins1975 Complementary influence of host plant re-sistance on microbial chemical control of cabbage cat-erpillars HortScience 10 487-488

Dickson M H amp C J Eckenrode 1975a Variationin Brassica oeracea resistance to cabbage looper andimported cabbage worm in the greenhouse and fieldJ Econ Entomol 68 757-760

1975b Breeding for resistance in cabbage and cau-liflower to cabbage looper imported cabbagewormand diamondback moth J Am Soc Hortic Sci 105782-785

Dickson M H amp D H Wallace 1986 Cabbagebreeding pp 395-432 In M J Bassett [ed] Breedingvegetable crops AVI Westport Conn

Dunn J A amp D P H Kempton 1976 Varietaldifferences in the susceptibility of Brussels sprouts tolepidopterous pests Ann Appl BioI 82 11-19

Greene G L W C Genung R B Workman amp E GKelsheimer 1969 Cabbage looper control in Flor-ida-a cooperative program J Econ Entomol 62798-800

Harrison P K amp R W Brubaker 1943 The relativeabundance of cabbage caterpillars on cole crops grownunder similar conditions J Econ Entomol 36 589-592

Hocking R R 1976 The analysis and selection ofvariables in linear regression Biometrics 32 1-50

Kim T H 1979 Brassica oeracea L resistance tocabbage caterpillars in New York State MS thesisCornell Univ Ithaca NY

Latheef M A amp R D Irwin 1979 Factors affectingoviposition of Pieris rapae on cabbage EnvironEntomol 8 606-609

Lin J C J Eckenrode amp M H Dickson 1983 Vari-ation in Brassica oeracea resistance to diamondbackmoth (Lepidoptera Plutellidae) J Econ Entomol76 1423-1427

Pimentel D 1961 An evaluation of insect resistancein broccoli Brussels sprouts cabbage collards andkale J Econ Entomol 54 156-158

Radcliffe E B amp R K Chapman 1965a The rel-ative resistance to insect attack of three cabbage va-rieties at different stages of plant maturity AnnEntomol Soc Am 58 897-902

1965b Seasonal shifts in the relative resistance to in-sect attack of eight commercial cabbage varietiesAnn Entomol Soc Am 58 892-897

1966a Plant resistance to insect attack in commercialcabbage varieties J Econ Entomol 59 116-120

1966b Varietal resistance to insect attack in variouscruciferous crops J Econ Entomol 59 120-125

Shelton A M J T Andaloro amp J Barnard 1982Effects of cabbage looper imported cabbagewormand diamondback moth on fresh market and pro-cessing cabbage J Econ Entomol 75 742-745

Shelton A M R F Becker amp J T Andaloro 1983Varietal resistance to onion thrips (ThysanopteraThripidae) in processing cabbage J Econ Entomol76 85-86

Sirrine F A amp V H Lowe 1894 Insects affectinglate cabbage notes on the stalk borer insecticidesNY Agric Exp Stn (Geneva) Bull 83 657-685

Wolfenbarger D A 1967 Variation in Brassica sppto cabbage looper infestations J Econ Entomol 60874-875

Received for publication 13 March 1987 accepted 21October 1987