HORTICULTURAL ENTOMOLOGY

Intentional Coverage Gaps Reduce Cost of Mating Disruption forPhyllocnistis citrella (Lepidoptera: Gracillariidae) in Citrus

S. L. LAPOINTE,1,2 L. L. STELINSKI,3 C. P. KEATHLEY,4 AND A. MAFRA-NETO5

J. Econ. Entomol. 107(2): 718Ð726 (2014); DOI: http://dx.doi.org/10.1603/EC13388

ABSTRACT The leafminer, Phyllocnistis citrella Stainton (Lepidoptera: Gracillariidae), is a globalpest of citrus and contributes to the incidence and severity of citrus bacterial canker. SPLAT CLM(ISCA Technologies, Riverside, CA) is an emulsiÞed wax that provides sustained release of (Z,Z,E)-7,11,13-hexadecatrienal, the major component of P. citrella sex pheromone. Trials in commercialorchards demonstrated that SPLAT CLM applied to plots of varying width resulted in disruption oftrap catch of male P. citrella within treated rows and across untreated rows adjacent to treated rows.SPLAT CLM applied to plots of constant width (10 rows) disrupted trap catch across an untreatedgap as the square of the width of the gap. Similarly, the ability of the pheromone source in treatedrows to disrupt trap catch across untreated gaps of constant size declined as the square of the widthof adjacent treated areas. A coverage pattern of 4 rows skipped for every 10 treated rows resulted ina 4% reduction of trap shutdown, and reduced the product and application costs by 29%. MiningincidencebyP. citrella in treated rowswas reducedby53%comparedwithuntreatedareas. Intentionalcoverage gaps can signiÞcantly reduce the cost of mating disruption. Commercial lures for P. citrellaused in this study were highly potent with respect to attracting males. Each lure was �103 times asattractive as an individual P. citrella female. Disruption of trap catch using commercial lures mayunderestimate actual mating disruption achieved in the Þeld.

KEY WORDS sex pheromone, (Z,Z,E)-7,11,13-hexadecatrienal, citrus leafminer, SPLAT, Lepidoptera

Recent progress in Þeld trials (Lapointe et al. 2009,Stelinski et al. 2010) has propelled commercializationof SPLAT CLM (ISCA Technologies, Riverside, CA),a mating disruptant product for the leafminer Phyl-locnistis citrella Stainton (Lepidoptera: Gracillarii-dae), a pest of citrus crops throughout the world(Heppner 1993). Larval feeding by this species re-duces photosynthetic capacity of leaves and increasesthe susceptibility of leaves to citrus bacterial canker,Xanthomonas axonopodis pv. citri (Gottwald et al.2007, Hall et al. 2010). Successful mating disruption ofP. citrella in commercial citruswasdemonstrated(Ste-linski et al. 2008) using SPLAT CLM, an emulsiÞedwax formulation providing sustained release of an off-ratio blend of P. citrella sex pheromone over severalweeks (Lapointe and Stelinski 2011). This is the Þrstuse of mating disruption in Florida citrus; adoption bygrowers will depend on product and application costsas well as on efÞcacy.

Synthesis of the major component of the P. citrellasex pheromone, (Z,Z,E)-7,11,13-hexadecatrienal, in-

volves a highly pyrophoric reagent (Leal et al. 2006,Moreira et al. 2006) that limits the scale-up potentialand increases the cost above what may be acceptableto citrus growers. We previously conducted a limitedÞeld experiment that demonstrated the possibility ofreducing the amount of SPLAT CLM applied by in-corporating intentional coverage gaps (Lapointe andStelinski 2011) similar to what has been suggested formating disruption of the gypsy moth, Lymanria dispar(L.) (Tcheslavskaia et al. 2005). Here, we report amore complete set of experiments and analyses thatdescribe the effect on trap catch disruption of differ-ently sized coverage gaps and treatment blocks incommercial orchards typical of citrus operations inFlorida.

Trap catchdisruption (trap shutdown) is oftenusedas a proxymeasurement formatingdisruptionbecauseof thedifÞcultyof assessing themating status of femaleinsects in the Þeld. Typically, the number of maleinsects caught in traps baited with an attractive pher-omone blend in areas treated with pheromone to dis-rupt mating is compared with trap catch in untreatedareas under the assumption that the ability of phero-mone lures to attract males is equivalent to that ofunmated females. Disruption of trap catch is, there-fore, indirect evidence of mating disruption. In prac-tice, pheromone lures are loaded with many times theamount of pheromone released by individual females(Witzgall et al. 2008). Synthetic lures that are farmore

1 USDAÐARS, United States Horticultural Research Laboratory,2001 South Rock Rd., Fort Pierce, FL 34945.

2 Corresponding author, e-mail: stephen.lapointe@ars.usda.gov.3 Entomology and Nematology Department, Citrus Research and

Education Center, University of Florida, 700 Experiment Station Rd.,Lake Alfred, FL 33850.

4 USDAÐARS, 2001 South Rock Rd., Fort Pierce, FL 34945.5 ISCA Technologies, Inc., 1230 Spring St., Riverside, CA 92507.

potent than authentic females may underestimate theimpact of a pheromone treatment on mating disrup-tion. Measures of pheromone-mediated disruptionmay include: disruption of male moth catch in trapsbaited with synthetic pheromone or unmated females(Stelinski et al. 2005), quantiÞcation of feral mothmating (Knight and Light 2013), direct observation ofmoth response to pheromone sources in the Þeld(Witzgall et al. 1999, Stelinski et al. 2004), and mostimportantly, assessment of crop damage (Gut et al.2004, Knight et al. 2012). Of these, trap shutdown hasbeen widely used as a benchmark for estimating mat-ing disruption because of the ease of using this tech-nique (Gut et al. 2004). In addition to assessing trapshutdown as an estimate of mating disruption, weassesseddamage causedbyP. citrellabyevaluating thenumber of mines per shoot from plots treated withSPLAT CLM.

We assume that the cost savings achieved by nottreating a proportion of a crop is a nearly linear func-tion of the gap size. However, it has been suggested(Lapointe and Stelinski 2011) and we hypothesizehere that the loss of disruption resulting from incor-poration of a repeated treatment gap pattern into agrove is related to the square of the width of thetreatment gap. As a result, it should be possible toincorporate treatment gaps with signiÞcant savingsand only minor reductions in the effectiveness of mat-ing disruption. Such an approach is of particular in-terest in the case of P. citrella because the high cost ofsynthesis of the pheromone is the major limitation toadoption of a mating disruption strategy for this spe-cies. To prove this point, we conducted Þeld trials thatvaried the width of treatment gaps, and in a separateexperiment, we varied the width of treated blocks toidentify appropriate parameters for an economicallyadvantageous coverage pattern for pheromone de-ployment that maintains effective disruption as esti-mated by trap shutdown.

Determining the relative attractiveness of monitor-ing lures versus authentic females is a required mile-stone to fully understand trap catch disruption data astheyrelate tomatingdisruption.Wealsoconductedanexperiment to determine the relative attraction ofmaleP. citrella to syntheticpheromone luresof variousdosages as compared with unmated females.

Materials and Methods

SPLAT CLM Application. SPLAT containing 0.15%(Z,Z,E)-7,11,13-hexadecatrienal (SPLAT CLM, ISCATechnologies, Riverside, CA) was applied using anapplicator designed and fabricated in collaborationwith International Fly Masters (Fort Pierce, FL) aspreviously described (Lapointe and Stelinski 2011).The applicator consisted of a pair of computer-con-trolled variable speed peristaltic pumps that drew theSPLAT compound from reservoirs suspended abovethe pumps and delivered it to nozzles located aboveblowers on either side of a frame mounted to a tractorby a three-point hitch. Dollops of SPLAT CLM fell bytheir own weight into the airßow from the blowers,

and air velocity was only sufÞcient to loft the dollopsinto the tree canopy while avoiding disruption of thedollops. The product was applied as �1 g dollops at anominal rateof500g/hawith four replicates.Here, thenominal rate refers to the number of grams of SPLATCLM applied to treated areas, while the effective rateis the amount of SPLAT CLM averaged over treatedand untreated areas within a given treatment. Fieldtrials were conducted during 2011 at TRB Groves (27�01� N, 81� 46� W) in northern Charlotte County, FL,and Emerald Grove (27� 28� N, 80� 38�W), courtesy ofThe Packers of Indian River, located in northwesternSt. Lucie County, FL.

Effect ofVaryingWidthofTreatmentGaps.AtTRBGroves, SPLAT CLM was applied to Duncan grape-fruit in 10-row blocks. Tree spacing was 7.3 by 3.7 mbetween and within rows, respectively; rows were�185 m in length. The area of each individual plot was�1.5 ha. Each row in all plots contained 54 trees. Plotswere replicated four times. SPLAT CLM was appliedat a nominal rate of 500 g/ha. Treated blocks wereseparated by gaps of varying width, either 4, 8, or 10rows so that eachuntreatedgapwasßankedby treatedblocks of 10 rows each (10Ð4Ð10, 10Ð8Ð10, and 10Ð10Ð10 treatedÐuntreatedÐtreated rows, respectively;see schematic design in Fig. 2). To assess trap catchdisruption, twoPheroconVI traps (Trece, Adair,OK),each baited with a rubber septum loaded with 0.1 mgof (Z,Z,E)-7,11,13-hexadecatrienal and 0.033 mg of(Z,Z)-7,11-hexadecadienal (IT203, ISCAlure-Citrella,ISCA Technologies, Riverside, CA), were placed inthe center rows of treated and untreated plots on the18th and 36th trees in rows of 54 trees. Sticky trapliners were replaced and counted weekly for a periodof 10Ð11 wk after application of SPLAT CLM. Meanweekly trap catch disruption was calculated as a per-centage of the trap catch in baited traps placed inuntreated rows �115 m from treated rows. Data werecomparedby analysis of variance (ANOVA)andÞttedto a response surface using Design-Expert software(Stat-Ease Inc., Minneapolis, MN). The experimentwas repeated three times during 2011 by reapplyingSPLATCLMto the sameplots on 22April, 13 July, and27 September.

To estimate the effect of treatment gaps on trapcatch disruption in a hypothetically large grovetreated with the corresponding pattern (e.g., 10Ð4Ð10), weighted means of treated and untreated areaswere combined to produce an expected trap catchdisruption calculated as the sum of the experimentallydetermined trap catch disruption in treated plots mul-tiplied by the proportion of treated rows (number oftreated rows by the total number of rows) and theexperimentally determined trap catch disruption inuntreated plots multiplied by the proportion of un-treated rows (number of untreated rows by the totalnumber of rows). For this exercise, trap shutdownover the entire untreated gap was assumed to beequivalent to the mean experimental trap shutdownobserved in the central row of the untreated plots.

April 2014 LAPOINTE ET AL.: MATING DISRUPTION OF P. citrella 719

Effect of Alternating Treated and Untreated Dou-ble-Row Beds. An additional coverage design (desig-nated “zebra”) was tested during the same period ona production block of red grapefruit on Swingle root-stock (U.S. Department of AgricultureÐAgriculturalResearch Service [USDAÐARS]) at the TRB locationdescribed above using the same SPLAT CLM formu-lation and batch. The coverage pattern consisted ofalternating double-row beds. Tree spacing was 7.3 by3.7 m between and within rows, respectively; rowswere �360 m in length. The area of each individualplot was �0.84 ha. Each row in all plots contained 100trees. Plots were replicated three times. SPLAT CLMwas applied at a nominal rate of 500 g/ha to beds (tworows) separated by untreated beds. Replicated plotsconsisted of six treated beds each (see schematic de-sign in Fig. 1). Two traps baited with attractant lureswere located within each treated and untreated bedfor a total of 22 traps in each plot. The experiment wasrepeated three times during 2011 by reapplyingSPLATCLMto the sameplots on 22April, 13 July, and27 September. Trap catch disruption was calculated asdescribed above based on catch of traps located inadjacent untreated red grapefruit.

Effect of Varying Width of Treated Area. Rows oftrees within four production blocks (�183 by 1550 mor 28 ha each) of actively managed, mature Flamegrapefruit on Swingle rootstock located at EmeraldGroveweredivided into treatment plots, controls, andbuffer areas to avoid interactions between treatments.Trees were hedged and topped at between 4 and 5 mindouble-rowbeds spaced 3.8mbetween treeswithinrows (48 trees per row) and 7.6 m between rows.

Treatments were arranged in a randomized block de-sign with four blocks and one replication of eachtreatment per block. Treatments consisted of varyingthe number of rows treated with SPLAT CLM from2 to 10, while maintaining a constant number of un-treated rows (10) ßanked by treated areas of equalwidth (see schematic design in Fig. 5). Treatmentswere designated 2Ð10Ð2, 4Ð10Ð4, 8Ð10Ð8, and 10Ð10Ð10 indicating the number of treatedÐuntreatedÐtreated rows. Each treatment was replicated fourtimes; six blocks of 10 rows each were included aspositive (treatedwith SPLATCLM)and5blocks of 10rows eachwere included as negative (untreated) con-trols. All plots were separated by 10 untreated rows(buffer rows).Toassessdisruptionof themalesÕ abilityto orient to an attractive pheromone blend, traps(Pherocon VI, Trece, Adair, OK) baited with a lure(ISCAlure-Citrella)were deployed in the center rowsof treated and untreated plots. Two traps were de-ployed in the center rows of each of the two SPLATCLM-treated areas within each treatment on approx-imately the 16th and 32nd trees with the 48-tree rows.Three trapswere deployed at equal intervals along thecenter row of each of the untreated 10-row areas(gaps) within treatments on approximately the 12th,24th, and 36th trees within the 48-tree rows. Trapliners were replaced approximately weekly and lureswere replaced every 6 wk. Trap catch was expressedas the number of male P. citrella per trap per day. Theexperiment was repeated three times during the 2011growing season with SPLAT CLM applications occur-ring on 14 April, 10 June, and 21 September.

Leafminer infestation was evaluated in untreatedplots and in treated plots that were 2, 4, 8, and 10 rowswide. Evaluations were conducted in rows within un-treated plots that were �100m from any tree treatedwith SLATCLM. In theuntreatedplots andwithin thetreated and untreated areas of each of the experimen-tal plots, we collected 10 new shoots from themiddletwo rows, along the center three-Þfths of eachrow. The number of active mines on each of 10shoots (ßush) was counted and the mean number ofmines per shoot was calculated. Data were analyzedby one-wayANOVA followedby orthogonal contrasts(StudentÕs t-test) to compare the average number ofmines per shoot in the untreated versus treated plotsand then in treated rows versus untreated rows withintreated plots. Data were square-root (x � 1) trans-formed before ANOVA to improve homogeneity ofvariance. The number of mines per ßush was reportedas untransformed means (�SEM).

Relative Potency of PheromoneLures.Natural rub-ber septa loaded with varying amounts of the attrac-tive 3:1 blend of (Z,Z,E)-7,11,13-hexadecatrienal:(Z,Z)-7,11-hexadecadienal were prepared by ISCATechnologies (Riverside, CA) anddeployed in Phero-conVI trapswithadhesive liners incitrus groves at twolocations (St. Lucie and Polk counties, FL) along withtraps baited with virgin female P. citrella. The loadingrate of pheromone was varied in Þve log steps from0.00013 to 1.3 mg per lure. Commercially soldISCAlure-Citrella lures were loaded with 0.133 mg

Fig. 1. Seasonal mean � SEM trap catch disruptionwithin grapefruit plots treated with SPLAT CLM as depictedin the schematic diagram. Columns represent the trap catchdisruption as measured by pheromone-baited traps locatedwithin the four untreated exterior rows (U, Ext, n � 288), thefour treated exterior rows (T, Ext, n � 288), the six untreatedinterior rows (U, Int, n � 432), and the eight treated interiorrows(T, Int,n�576).Meanswerecalculatedas apercentageof the trap catch disruption of the grand mean of all traps.Charlotte County, FL, 2011.

720 JOURNAL OF ECONOMIC ENTOMOLOGY Vol. 107, no. 2

pheromone per lure. Lures were placed in traps anddeployed in a completely randomizeddesignover 18din JulyÐAugust 2011 in blocks of commercial citrus inSt. Lucie County, FL. Unmated female P. citrella wererearedbycollectingcitrus leaveswithmines fromÞeldlocations and allowing the adults to emerge in glasspetri dishes provided with moist Þlter paper in anenvironmental chamber maintained at a photoperiodof 12:12 (L:D)h and 30�Cduring photophase and 26�Cduring scotophase. Adult females were transported totheÞeld andplaced in cageswithin 18hof emergence.Cages consisted of screened plastic enclosures con-taining sections of dentalwickmoistenedwith a dilutesucrose solution. The adhesive liners of traps baitedwith lures and females were replaced daily. Trapswere numbered and assigned randomly to locationswithin the citrus orchard; locations were rerandom-ized and traps were relocated daily to minimize vari-ance because of variation in P. citrella population den-sity between trap sites. All traps were separated by�60 m. Four replicates of seven treatments were de-ployed: rubber septa loaded at Þve rates with a 3:1blend of (Z,Z,E)-7,11,13-hexadecatrienal:(Z,Z)-7,11-hexadecadienal, a control trap (baited with a blankrubber septum), and caged female(s). Availability offemales ßuctuated during the trial, so traps werebaited with 1Ð6 individuals per trap; data were ex-pressed as the number of males caught per female andcompared with a regression of trap catch on phero-mone loading rate.

Statistical Analysis. The application experimentswere laid out in a randomized block design with fourreplications. Weekly counts of trap liners from plotstreated with SPLAT CLM were expressed as a per-centage of the catch from traps in untreated plots(trap catch disruption). When appropriate, the angu-lar (arsine) transformation of the percentage trapcatch disruption was applied before analysis byANOVA (Kasuya 2004). Untransformed means arereported. A three-dimensional response surface to vi-sually demonstrate attenuation of trap catch disrup-tion over 11 wk after application of SPLAT CLM ineach of the treatments (gap width) was generated inDesign-Expert (v7.0.3; State-Ease Inc., Minneapolis,MN). Post hoc partitioning of the sum of squares togenerate orthogonal comparisons of seasonalmeans oftreatments was carried out using the least squaremeans contrast option in JMP (v. 10.0, SAS InstituteInc., Cary, NC). The Bonferroni procedure was usedto adjust � for determination of signiÞcant differenceswhen conducting multiple post hoc comparisons. At-tenuation of trap catch disruption was plotted as themean trap catch disruption as a function of increasingtreated area width or increasing gap width and Þttedto a quadratic curve (Lapointe and Stelinski 2011).

Questions regarding the relative strength of trapcatchdisruption as affectedby locationwithin theplotwere addressed in the zebra experiment of alternatingtreated and untreated rows. Mean trap catch disrup-tion was compared by ANOVA using a 2 by 2 factorialdesign to compare the main effects of location (inte-

rior or exterior rows) and treatment (treated or un-treated with SPLAT CLM).

To estimate the relative attraction of unmated fe-males, the log of the number of male P. citrella cap-tured in traps baited with synthetic pheromone wereregressed against the log pheromone-loading rate ofthe lures. The number of males or female in trapsbaited with unmated females was Þt to the regressionline to estimate the relative pheromone released byindividual females and as a rough estimate of therelative strength of the commercial lures.

Results

Effect of Varying Width of Treatment Gaps. Meantrap shutdown was calculated as the number of maleP. citrella captured in traps placed in the center row ofuntreated plots of varying widths ßanked by 10 rowstreated with SPLAT CLM, divided by the number ofmales captured in traps located in untreated controlplots (n � 5 plots). The correlation between trapshutdown and gap width was signiÞcant (� � 0.006)on each of the eight dates when trap liners werecollected through 9 wk after the April and June ap-plications of SPLAT CLM, at which point trap shut-down in the treated control plots fell below 50% (Fig.2). There was a strong and anticipated effect of timeafter application on trap catch disruption (P 0.0001)but there was no interaction between time and gapwidth (P � 0.39). There was a signiÞcant effect of gapwidth on the mean difference (loss) of trap catchdisruption compared with the positive control (10treated rows) for each of the three application dates(Þrst application: F � 12.59; df � 2, 21; P � 0.0003;second application: F � 15.70; df � 2, 24; P 0.0001;third application:F� 11.64; df� 2, 24;P� 0.001).Datafor the Þrst application are presented in Table 1. Toillustrate, trap catch disruption at 7 d after the Þrstapplication was Þtted to a quadratic expression (Fig.3), resulting in a strong correlation between increas-ing gap widths and declining trap shutdown. Seasonalmeans (�SEM, n � 8 dates) were also compared bysetting trap shutdown in the positive controls to 100%and plotting the weighted mean trap shutdown of thethree gap treatments as the loss of shutdown com-pared with positive controls for the three applications(Fig. 4). The reduction of trap shutdown comparedwith the positive controls in plots that included 4skipped rows (30.5 m) compared with the trap shut-down in the center rows of adjacent plots of 10 treatedrowswas�4%(y� 0.64x2 0.05x� 0.0005,wherey�trap shutdown and x � percentage of skipped area).A gap of 4 rows for every 10 treated rows correspondsto a 29% reduction in SPLAT CLM required to treat ahypothetically large area treated with a repeating pat-tern. The quadratic Þt of loss of trap catch disruptionwith increasing gap width (Figs. 3 and 4) suggests thatthe ability of pheromone applications in plots of con-stant width (10 rows) to disrupt trap catch across anuntreated gap declined with the square of the widthof the gap.

April 2014 LAPOINTE ET AL.: MATING DISRUPTION OF P. citrella 721

Effect of Alternating Treated and Untreated Dou-ble-Row Beds. Overall, the treatment consisting ofalternating treated and untreated two-row beds per-formed less well compared with the other treatments(Table 1). However, the nature of the design andplacement of traps differed from the other treatmentsbecause of the absence of central locations for un-treated and treated areas. Mean trap catch disruptionvaried according to the location of traps within theplot of six treated beds alternating with untreatedbeds. There was no interaction between location andtreatment (F � 2.15; df � 1, 1583; P � 0.14), so maineffects were examined. Treated rows had signiÞcantlygreater (F � 9.83; d.f. � 1, 1583; P � 0.0018) trap catchdisruption expressed as a proportion of the total trapcatch disruption in the treatment (104.4 � 2.1%) com-pared with the untreated rows (94.7 � 2.3%). Trapcatch disruption as a percentage of total trap disrup-tion in the treatment in interior rows (106.1 � 1.9%)was signiÞcantly greater (F�27.5795; df�1, 1583;P0.0001) compared with that in the exterior rows(89.3 � 2.5%).

Effect ofVaryingWidthofTreatedArea. Inspection(Fig. 5) of the seasonal means (�SEM, n � 26 datesover three applications when trap catch disruptionexceeded 50%) suggested the orthogonal contrastspresented in Table 2. There were signiÞcant differ-ences (P 0.0001, ANOVA) in the number of male P.citrella captured in baited traps for all three applica-tions and themeanof the three applications (Table 3).The Þrst contrast (C1) demonstrated that the numberofmaleP. citrella captured in thepositive control plots(treated) did not differ from the number captured inthe treated rows of the plots that consisted of 4, 8, or10 treated rows on any of the three dates (Table 3).The second contrast (C2) demonstrated a signiÞcantdifference between the number of males captured inthe treated rows of plots that consisted of two treatedrows and the number of males captured in the treatedrows of the remaining treatments for all three appli-cations (Table 3). The third contrast (C3) demon-strated that the trap catch in untreated gap rows in theplots that consisted of 2 treated rows and 10 untreatedrows (2Ð10Ð2) was signiÞcantly different from the

Fig. 2. A three-dimensional response surface plot of trap catch disruption within treated plots with varying coverage gapsover 11 wk after the application of SPLAT CLM at a nominal rate of 500 g/ha. Red points are response values above thepredicted value and pink points are response values below the predicted value.

Table 1. Trap catch disruption (percentage of trap catch in untreated plots) after application of SPLAT CLM calculated as the weightedmeans of treated and untreated rows in citrus plots with varying width of untreated gaps

Width(m)

Days after applicationa

Grand meanb Mean decreasec

7 14 21 28 39 48 55 62

0 98 96 94 88 75 66 64 70 81 � 5.030.5 97 94 88 84 68 60 59 69 77 � 5.3 4.2 � 0.9a61.0 94 91 87 80 59 55 51 56 72 � 6.4 9.8 � 1.6ab76.2 91 87 80 71 58 52 50 49 67 � 6.1 14.0 � 1.6bc15.25 83 73 81 67 60 51 29 44 61 � 6.7 20.4 � 2.7c

a Entries for each date are calculated as weighted means of the mean trap catch (n � 16) in untreated gaps and in treated plots (n � 24)divided by the trap catch in untreated control plots (n � 64).

b Grand mean of dates when mean trap catch disruption in control exceeded 50%, n � 8.c Means followed by the same letter do not differ by TukeyÕs honest signiÞcant difference (ANOVA: F � 15.11; df � 3, 31; P 0.0001).Gap widths consisted of 0 (positive control), 4, 8, or 10 gap rows ßanked by 10 treated rows (gaps of 0, 30.5, 61.0, or 76.2 m, respectively)

or alternating two-row beds of treated and untreated rows (gap of 15.25m). The mean decrease was calculated as mean (n � 8) differencein trap catch disruption between gap treatments and treated control plots

722 JOURNAL OF ECONOMIC ENTOMOLOGY Vol. 107, no. 2

trap catch in the untreated rows of the remainingtreatments for the second application (no. 2), mar-ginally so after the Þrst application, and not signiÞcant

after the third application and for the mean of threeapplications (Table 3).

The mean (�SEM, n � 4) number of P. citrellamines per shoot differed across treatments (F8, 31 �4.4; P � 0.001) ranging from a high of 15.1 � 2.9 inuntreated plots to 9.4 � 0.9, 9.8 � 1.5, 7.1 � 1.1, and6.5 � 1.3 for untreated rows surrounded by 4, 8, 16, or20 treated rows, respectively (Fig. 6). The mean (�

Fig. 3. Effect of increasing width of untreated gap onmean trap catch disruption (�SEM, n � 8 traps) expressedas the weighted mean of treated and untreated rows 7 d afterapplicationofSPLATCLM.Widthofgapwasvariedbetweenblocks of 10 treated rows (76 by 183 m) as shown in theschematic design (not to scale and not representative of plotarrangement). Traps were placed in the center rows oftreated blocks (0 gap) and untreated blocks. CharlotteCounty, FL, 2011.

Fig. 4. Effect of increasing gap width (percentage of total area) on mean (�SEM, n � 8 approximately weekly samplingdates) trap catch disruption (weighted mean of treated � untreated rows) over 62 d after application of SPLAT CLM,replicated on three dates. Gap widths were 0, 30.5, 61, and 76 m. Gaps were ßanked by 1.4 ha blocks of 10 treated rows (76by 183m) on either side as shown in the schematic design in Fig. 2. Traps were placed in the center rows of treated blocks(0 gap) and untreated blocks. Charlotte County, FL, 2011.

Fig. 5. Seasonal mean � SEM trap catch disruption over26 sampling dates from 14 April through 23 November, 2011,in grapefruit plots that received SPLAT CLM at a nominalrate of 500 g/ha. Plots consisted of a constant number (10)of untreated rows (gap) ßanked by a varying number (0, 2,4, 8, and 10) of treated rows (7.6 by 183 m) as shown in theschematic design (not to scale and not representative of plotarrangement). Traps were placed in the center rows oftreated and untreated blocks. St. Lucie County, FL, 2011.

April 2014 LAPOINTE ET AL.: MATING DISRUPTION OF P. citrella 723

SEM, n � 4) number of mines in rows treated withSPLATCLMranged from6.7� 2.0, 7.1� 0.7, 6.1� 0.8,and 5.5 � 0.7 (n � 8) for 2, 4, 8, and 10 treated rows,respectively. There were 53% fewer mines per shootin areas that were treated or surrounded by treatedareas (7.1 � 0.4, n � 36) compared with the untreatedcontrol (contrast, t � 4.8;P 0.0001). Therewere 24%fewer mines per shoot (t � 2.1; P 0.05) in treatedrows (6.2 � 0.5; n � 20) compared with adjacentuntreated rows (8.2 � 0.6; n � 16).

Relative Potency of PheromoneLures.Thenumberof male P. citrella caught in traps baited with variablyloaded lures or unmated females was expressed as thenumber of males per trap per day. The data for thevariably loaded lures were plotted as the log of

the daily catch and Þt to an exponential model (Fig.7). The capture of males in traps baited with a singleunmated female was equivalent to the rate of capturepredicted by the regression in a trap baited with 145ng of synthetic pheromone, approximately three or-ders of magnitude less than the standard pheromonelure.

Discussion

Trap shutdown is a commonly used proxy for mea-surement of mating disruption (Gut et al. 2004). Trapshutdown measures disruption of attraction to a syn-thetic pheromone blend released at rates attractive tomale moths caused by the deployment of one or moresyntheticpheromonecomponentswithin a crop.Field

Table 2. Orthogonal contrasts applied to treatments in anexperiment to compare the width of citrus plots treated with SPLATCLM on P. citrella trap shutdown within treated and flanking un-treated rows

Treatmenta TÐCÐSb C1c C2d C3e

� control T 3 1 02Ð10Ð2 S 0 0 32Ð10Ð2 T 0 4 04Ð10Ð4 S 0 0 14Ð10Ð4 T 1 1 08Ð10Ð8 S 0 0 18Ð10Ð8 T 1 1 010Ð10Ð10 S 0 0 110Ð10Ð10 T 1 1 0

a Treatments (number of treatedÐuntreatedÐtreated rows).b T, treated; C, untreated control; S, untreated (skipped).c Contrast 1: trap shutdown in the positive control (treated) versus

that in treated rows of plots consisting of more than two rows treated.d Contrast 2: trap shutdown in the treated rows of plots consisting

of two rows treated versus that in treated rows of plots consisting ofmore than two rows treated.

e Contrast 3: trap shutdown in skipped rows of plots with twotreated rows versus that in skipped rows of plots consisting of morethan two rows treated.

Table 3. Results of the orthogonal contrasts presented in Table2 comparing trap catch of male P. citrella in SPLAT CLM-treatedand untreated citrus plots with varying number of treated rows

Statistic C1 C2 C3

Application no. 1 (April)a

T value 0.02 4.09 2.18P (two-tailed) 0.99 0.0001* 0.03P (one-tailed) 0.49 0.0001* 0.015*

Application no. 2 (June)b

T value 0.007 6.92 2.79P (two-tailed) 0.99 0.0001* 0.006*

P (one-tailed) 0.49 0.0001* 0.003*

Application no. 3 (Sept.)c

T value 0.37 4.62 0.051P (two-tailed) 0.71 0.0001* 0.96P (one-tailed) 0.35 0.0001* 0.48

Mean of 3 applicationsd

T value 0.20 6.29 1.58P (two-tailed) 0.84 0.0001* 0.12P (one-tailed) 0.42 0.0001* 0.06

* SigniÞcant at � � 0.017 (Bonferroni correction).a Application no. 1: F9, 124 � 11.79; P 0.0001.b Application no. 2: F9, 124 � 12.60; P 0.0001.c Application no. 3: F9, 124 � 7.91; P 0.0001.Mean of three applications: F9, 124 � 12.73; P 0.0001.

Fig. 6. Mean (�SEM) number of mines per shoot inuntreated control plots, within treated areas of varyingwidth(number of treated rows), or untreated areas of 10 rowsßanked by treated areas of corresponding width. St. LucieCounty, FL, 2011.

Fig. 7. Effect of amount of the attractive 3:1 triene:dienepheromone blend loaded onto rubber septa on mean dailytrap catch of male P. citrella. The black circle represents themean number of male P. citrella caught in traps baited withIT203 lures and the black square (not included in regression)represents the mean number of males per female caught intraps baited with live virgin females and Þt to the regressionline to estimate relative pheromone emission of female P.citrella. St. Lucie County, FL, 2011.

724 JOURNAL OF ECONOMIC ENTOMOLOGY Vol. 107, no. 2

trials in which the effect of proportionality of a two-component pheromone blend on trap shutdown wasinvestigated for P. citrella in citrus demonstrated thattrap shutdown increased with the proportion of thetriene (Z,Z,E)-7,11,13-hexadecatrienal, one of twoprincipal pheromone components of P. citrella(Lapointe et al. 2009). As a result, the triene has beenchosen as the sole active ingredient for use in SPLATCLMcommercial formulation formatingdisruptionofP. citrella. The use of a single synthetic compound alsosimpliÞes manufacturing. However, synthesis of thetriene is costly and involves use of a highly pyrophoricreagent, tert-butyllithium. The relatively high cost ofsynthesis of the triene may limit adoption of matingdisruption as a cost-effective tool for citrus leafminercontrol unless ways to reduce cost are identiÞed(Lapointe et al. 2011). The purpose of the studiespresented here was to explore the potential for costsavings by varying the coverage of SPLAT CLM incitrus groves by leaving a proportion of “gap” rowsuntreated. The savings to be realized by this approachinclude the cost of pheromone and dispenser matrix(SPLAT) as well as labor, fuel, and other costs asso-ciated with the applicator machinery. Lapointe andStelinski (2011) reported that coverage gaps of 20 or33% of tree rows (every Þfth row or every Þfth andsixth rows, respectively) did not signiÞcantly reducethe trap shutdown obtained over the combined areawith machine-applied SPLAT CLM in grapefruit.

To design a coverage pattern for citrus, decisionsmust be made regarding the number of consecutiverows to treat and the number to leave untreated.Lapointe et al. (2011) showed that trap shutdown wasgreatest when pheromone dispensers were located ineach tree within a grove compared with levels ofaggregation of dispensers at the same rate of phero-mone per unit area. Therefore, we decided to conducttwo experiments: one where the number of untreatedrows was varied between blocks of 10 rows of citrustrees treated with SPLAT CLM at the nominal rate of500 g/ha (1 g triene per hectare), and a second trialwhere thenumberof untreated rowswas constant andthe number of rows (also treated at the same nominalrate) varied.

In the trial of varying gap width, there was a goodquadratic Þt of trap shutdown, indicating an exponen-tial decline in disruption with increasing gap width,similar to that reported andpredictedbyLapointe andStelinski (2011). A gap of 4 rows between blocks of 10treated rows (equal to a 28.6% reduction in amount ofSPLATCLM)resulted in amean loss of trap shutdownof 4% at 7 d after application (Fig. 3) and �5% for aseason-long average (Fig. 4) comparedwith trap shut-down in the treated control block. It should be notedthat this reduction was the trap shutdown observed inthe center row of the untreated area, and thereforerepresents the disruption that occurred at the farthestpoint from a pheromone source. The estimate of 4%reduction in disruption therefore overestimates thedisruption that likely occurred over the entire un-treated area. Regardless, the trade-off of a maximumestimated loss of efÞcacy of 4% in return for a 28.6%

savings in product and associated costs make a strongeconomic argument for this pattern of coverage.

The results of varying the width of treated blockssuggest that a minimum of four treated rows are nec-essary to achieve trap shutdown in both the treatedanduntreated rows(Fig. 5).Treatedwidthsof 4, 8, and10 rows resulted in similar levels of trap catch disrup-tion in both treated and untreated areas. This resultmayexplain thepoorer-than-expectedperformanceofthe 2/2 treatment consisting of 2 treated and 2 un-treated rows compared with the 10/10 treatment con-sisting of 10 treated and 10 untreated rows that resultsin the same application rate per unit area (Table 1).

All of the coverage patterns tested may be expectedto perform better when applied to larger areas. Thismaybeparticularly true in the case of the zebra designin comparison with the other designs tested. Addi-tional trials may be necessary to identify an optimumcombination of gap and treated widths. Multiple so-lutions may exist or be appropriate for speciÞc situa-tions. However, based on these results, a conservativerecommendation can be made to reduce SPLAT CLMapplication by 30%withminimal loss in trap shutdown(and presumably mating disruption) by skipping 4rows for every 10 rows treated at the nominal rate of500 g/ha equal to an effective rate of 350 g/ha, therebyreducing the cost of mating disruption for P. citrella.

Previous experience with ISCAlure-Citrella phero-mone lures (IT203) demonstrated that the lures werehighly potentwith respect to attractingmale P. citrella(Lapointe and Leal 2007). Our results demonstratedthat the IT203 lure is �103 times as attractive as anindividual P. citrella female. This suggests that we maybe underestimating the disruption achieved by usingdisruption of such powerful attractant lures as an es-timate of actual mating disruption. Stelinski et al.(2010) were able to document reduced incidence ofmines of P. citrella for several weeks after the appli-cation of SPLAT CLM to small (0.25 ha) plots. Here,we have demonstrated reductions in the number ofmines at a larger scale within both treated rows anduntreated gap rows ßanked by treated rows. It seemsreasonable to assume that measures of trap catch dis-ruption are an accurate gauge of mining damage forthis species. Our results demonstrate that coveragepatterns that incorporate intentional gaps can signif-icantly reduce the cost of pheromone deploymentwhile maintaining effective trap catch disruption andreducing the incidence of mining by P. citrella.

Acknowledgments

Larry Markle, Julia Morris, Matt Chellemi, Paul Robbinsand Rocco Alessandro (USDAÐARS, Ft. Pierce, FL), WendyMeyer, Siddharth Tiwari, and Ian Jackson (University ofFlorida,LakeAlfred,FL)provided technical assistance in theÞeld. Dave Robinson (International Fly Masters, Ft. Pierce,FL) provided prototype machinery and made the Þeld ap-plications. SigniÞcant funding was provided by the SpecialtyCrops Block Grant, USDA and the Florida Department ofAgriculture and Consumer Services. We are particularlygrateful to Tom Stopyra (The Packers of Indian River, FortPierce, FL) and David Kemeny (TRB Groves, Punta Gorda,

April 2014 LAPOINTE ET AL.: MATING DISRUPTION OF P. citrella 725

FL) for arranging access to those properties. Mention of atrademark or proprietary product does not constitute a guar-antee or warranty of the product by the USDA and does notimply its approval to the exclusion of other products thatmayalso be suitable.

References Cited

Gottwald, T. R., R. B. Bassanezi, L. Amorim, and A. Ber-gamin-Filho. 2007. Spatial pattern analysis of citrus can-ker-infected plantings in Sao Paulo, Brazil, and augmen-tation of infection elicited by the Asian leafminer.Phytopathology 97: 674Ð683.

Gut, L. J., L. L. Stelinski, D. R. Thomson, and J. R. Miller.2004. Behavior modifying chemicals: prospects and con-straints in IPM, pp. 73Ð121. InO.Koul, G. S.Dhaliwal, andG. Caperus (eds.), Integrated pest managementÑpoten-tial, constraints, and challenges. CABI, Wallingford,United Kingdom.

Hall, D. G., T. R. Gottwald, and C. H. Bock. 2010. Exacer-bation of citrus canker by citrus leafminer Phyllocnistiscitrella in Florida. Fla. Entomol. 93: 558Ð566.

Heppner, J. B. 1993. Citrus leafminer, Phyllocnistis citrella,in Florida. Trop. Lepidoptera 4: 49Ð64.

Kasuya, E. 2004. Angular transformationÐanother effect ofdifferent sample sizes. Ecol. Res. 19: 165Ð167.

Knight, A., and D. Light. 2013. Monitoring codling moth(Lepidoptera: Tortricidae) in orchards treated with pearester and sex pheromone combo dispensers. J. Appl. En-tomol. 137: 214Ð224.

Knight, A. L., L. L. Stelinski, V. Hebert, L. Gut, D. Light, andJ. Brunner. 2012. Evaluation of novel semiochemicaldispensers simultaneously releasing pear ester and sexpheromone for mating disruption of codling moth (Lep-idoptera: Tortricidae). J. Appl. Entomol. 136: 79Ð86.

Lapointe, S. L., and W. S. Leal. 2007. Describing seasonalphenology of the leafminer Phyllocnistis citrella (Lepi-doptera: Gracillariidae) with pheromone lures: control-ling for lure degradation. Fla. Entomol. 90: 710Ð714.

Lapointe, S. L., and L. L. Stelinski. 2011. An applicator forhigh viscosity semiochemical products and intentionaltreatment gaps for mating disruption of Phyllocnistis ci-trella. Entomol. Exp. et Appl. 141: 145Ð153.

Lapointe, S. L., L. L. Stelinksi, T. J. Evens, R. P. Niedz, D. G.Hall, and A. Mafra-Neto. 2009. Sensory imbalance asmechanism of mating disruption in the leafminer Phyl-locnistis citrella: elucidation by multivariate geometric

designs and response surface models. J. Chem. Ecol. 35:896Ð903.

Lapointe, S. L., L. L. Stelinksi, and R. D. Robinson. 2011. Anovel pheromone dispenser for mating disruption of theleafminer Phyllocnistis citrella (Lepidoptera: Gracillarii-dae). J. Econ. Entomol. 104: 540Ð547.

Leal, W. S., A. L. Parra-Pedrazzoli, A. A. Cosse, Y. Murata,J.M.S. Bento, and E. F. Vilela. 2006. IdentiÞcation, syn-thesis, andÞeld evaluationof the sexpheromone from thecitrus leafminer, Phyllocnistis citrella. J. Chem. Ecol. 32:155Ð168.

Moreira, J. A., S. McElfresh, and J. G. Millar. 2006. Identi-Þcation, synthesis, and Þeld testing of the sex pheromoneof the citrus leafminer, Phyllocnistis citrella. J. Chem.Ecol. 32: 169Ð194.

Stelinski, L. L., L. J. Gut, A. V. Pierzchala, and J. R. Miller.2004. Field observations quantifying attraction of fourtortricid moths to high-dosage pheromone dispensers inuntreated and pheromone-treated orchards. Entomol.Exp. et Appl. 113: 187Ð196.

Stelinski, L. L., L. J. Gut, R. E. Mallinger, D. Epstein, T. P.Reed, and J. R. Miller. 2005. Small plot trials document-ing effective mating disruption of Oriental fruit moth byusing high densities of wax-drop pheromone dispensers.J. Econ. Entomol. 98: 1267Ð1274.

Stelinski, L. L., J. R. Miller, and M. E. Rogers. 2008. Matingdisruption of citrus leafminer mediated by a non-com-petitive mechanism at a remarkably low pheromone re-lease rate. J. Chem. Ecol. 34: 1107Ð1113.

Stelinski, L. L., S. L. Lapointe, and W. L. Meyer. 2010.Season-long mating disruption of citrus leafminer, Phyl-locnistis citrella Stainton, with an emulsiÞed wax formu-lation of pheromone. J. Appl. Entomol. 134: 512Ð520.

Tcheslavskaia, K., C. Brewster, K. Thorpe, A. Sharov, D.Leonard, and A. Roberts. 2005. Effects of intentionalgaps in spray coverage on the efÞcacy of gypsy mothmating disruption. J. Appl. Entomol. 129: 475Ð480.

Witzgall, P., A.-C. Backman, M. Svensson, U. Koch, F. Rama,A. El-Sayed, J. Brauchli, H. Arn, M. Bengtsson, and J.Lofqvist. 1999. Behavioral observations of codlingmoth,Cydia pomonella, in orchards permeated with syntheticpheromone. Biocontrol 44: 211Ð237.

Witzgall, P., L. Stelinski, L. Gut, and D. Thomson. 2008.Codling moth management and chemical ecology. Annu.Rev. Entomol. 53: 503Ð522.

Received 6 September 2013; accepted 17 February 2014.

726 JOURNAL OF ECONOMIC ENTOMOLOGY Vol. 107, no. 2