Available online at www.sciencedirect.com

Biological Control 45 (2008) 386–395 www.elsevier.com/locate/ybcon

Predation by Podisus maculiventris (Hemiptera: Pentatomidae) on xylostella (: ) larvae parasitized by Cotesia plutellae (Hymenoptera: Braconidae) and its impact on

Nathan J. Herrick a,*, Stuart R. Reitz b, James E. Carpenter c, Charles W. O’Brien d,1

a Department of Entomology, Virginia Polytechnic Institute and State University, 216 Price Hall (0319), Blacksburg, VA 24061, USA b USDA-ARS-CMAVE, 6383 Mahan Drive, Tallahassee, FL 32308, USA c USDA-ARS-CPMRU, 2747 Davis Road, Tifton, GA 31794, USA d Formerly: CBC-CESTA, 105 Perry-Paige Building [South], A&M University, FL 32307, USA

Received 17 September 2007; accepted 20 February 2008 Available online 4 March 2008

Abstract

Biological control offers potentially effective suppression of the diamondback (DBM), Plutella xylostella, a serious pest of Bras- sica crops. Little is known of whether multiple natural enemies have additive, antagonistic, or synergistic effects on DBM populations. No-choice and choice tests were conducted to assess predation by Podisus maculiventris on DBM larvae parasitized by Cotesia plutellae and unparasitized larvae. In no-choice tests, P. maculiventris preyed on greater numbers of parasitized than unparasitized larvae and greater numbers of young larvae than old larvae. In choice tests with early third instar DBM, there was no difference in predation between parasitized or unparasitized larvae. However, in choice tests with older prey, P. maculiventris preyed on more parasitized than unparasitized larvae. Two field studies were conducted to test if this predator and have additive, antagonistic or synergistic effects on DBM populations and plant damage in cabbage ( var. capitata). In 2002, DBM populations were significantly lower in the presence of C. plutellae but not in the presence of P. maculiventris. There was not a significant interaction between the nat- ural enemies. Plant damage was reduced only with C. plutellae. In 2003, DBM populations were significantly lower in the presence of C. plutellae and P. maculiventris, although the combination of natural enemies did not lead to a non-additive interaction. Plant damage was unaffected by the presence of either natural enemy. Because of its greater predation on parasitized larvae, P. maculiventris could be an intraguild predator of C. plutellae. Yet, their overall combined effect in the field was additive rather than antagonistic. Ó 2008 Elsevier Inc. All rights reserved.

Keywords: Plutella xylostella; ; Podisus maculiventris; Spined-soldier bug; Cotesia plutellae; Intraguild predation; Predator–parasitoid interaction; Biological control

1. Introduction tions among natural enemies affect mortality in pest popu- lations because the outcome of these interactions can Interactions among multiple natural enemies of pests, significantly influence suppression of a pest and subsequent especially the interactions between predator and parasitoid plant damage. guilds, have generated considerable interest within the field Ecologists once considered that mortality from multiple of biological control. A primary interest lies in how interac- natural enemies is additive; however, it is now understood that synergistic or antagonistic interactions can transpire * Corresponding author. Fax: +1 540 231 9131. (Ferguson and Stiling, 1996; Sih et al., 1998; Swisher E-mail address: [email protected] (N.J. Herrick). et al., 1998). An additive association occurs when the 1 Retired. proportion of mortality caused by one natural enemy is

1049-9644/$ - see front matter Ó 2008 Elsevier Inc. All rights reserved. doi:10.1016/j.biocontrol.2008.02.008 N.J. Herrick et al. / Biological Control 45 (2008) 386–395 387 independent of the mortality induced by any other natural DBM (Vuorinen et al., 2004). Although previous research enemy. Resource partitioning, such as when natural ene- has shown that P. maculiventris has potential as a biologi- mies prefer different areas of a shared habitat, can result cal control agent for DBM (Ibrahim and Holopanin, 2004), in an additive association (Streams, 1987). The combined Mallampalli et al. (2002) and DeClerq et al. (2003) reported mortality from multiple natural enemies acting indepen- instances of intraguild predation (IGP) by P. maculiventris dently can be predicted from a multiplicative risk model on alternate predators. Mallampalli et al. (2002) found that based their individual effects (Soluk, 1993). Deviations P. maculiventris could prey on larvae of the predatory cocc- from these independent, additive effects would indicate inellid Coleomegilla maculata Lengi (Coleoptera: Coccinel- the natural enemies act antagonistically or synergistically. lidae), and their results indicate that interactions between Antagonism occurs when natural enemies interfere with these two predators on their extraguild prey were antago- each other and cause less mortality than the additive effect. nistic. DeClerq et al. (2003) found that IGP by P. maculiv- Antagonism results in greater pest populations, and conse- entris on Harmonia axyridis (Pallas) (Coleoptera: quently increases crop damage. Antagonism can result ) decreased significantly in the presence of from interactions such as intraguild predation (Pemberton the extraguild prey, Spodoptera littoralis (Boisduval) (Lep- and Willard, 1918; Lindgren and Wolfenbarger, 1976; idoptera: ), but not when aphids, Myzus persicae DeClerq et al., 2003; Finke and Denno, 2003). Alterna- (Sulzer) (Homoptera: Aphididae), served as the extraguild tively, synergism has the potential to produce the greatest prey. reduction in pest populations (Furlong and Groden, Similarly, Bilu and Coll (2007) studied IGP of the pred- 2001) and is desirable in pest management because total ator Coccinella undecimpunctata L. (Coleoptera: Coccinelli- pest mortality would be greater than the additive mortality dae) on the parasitoid Aphidius colemani Viereck of the multiple natural enemies (Rosenheim et al., 1993; (Hymenoptera: Braconidae) with M. persicae as the source Ferguson and Stiling, 1996; Cloutier and Jean, 1998; Losey of extraguild prey. They conclude that the presence of the and Denno, 1998, 1999; Colfer and Rosenheim, 2001). predator had a short-term negative impact on parasitism. The diamondback moth (DBM), Plutella xylostella (L.) However, after five days the combined activity of the pred- (Lepidoptera: Plutellidae), is one of the most serious pests ator and parasitoid resulted in the greatest suppression of of brassica crops [Brassica oleracea (L.)]. The solitary spe- aphids. They suggest that this response was because the cialist parasitoid Cotesia plutellae (Kurdjumov) (Hyme- predator did not show a preference for parasitized prey noptera: Braconidae) is one of the most important and instances of IGP were low due to low parasitism rates. of DBM (Talekar and Yang, 1991). Talekar If P. maculiventris preferentially preys on larvae parasitized and Yang (1991) found high levels of parasitism by C. plu- by C. plutellae, DBM populations and plant damage may tellae among brassica cultivars: 55%, 75%, 60%, and 37.5% be greater when C. plutellae is in the presence of P. macu- in cabbage, (B. chinensis L.), cauliflower, liventris than when C. plutellae acts alone. and , respectively. It has been released on numer- Our goal in these studies was to understand how DBM ous occasions throughout the world for DBM control populations and plant damage are affected by P. maculiven- (Lim, 1982; Mitchell et al., 1997, 1999). Parasitized individ- tris and C. plutellae alone and in combination. We studied uals remain on the host until egression of wasps, so both if P. maculiventris preferentially attacks larvae parasitized parasitized and unparasitized larvae are exposed to preda- by C. plutellae or unparasitized DBM larvae to determine tion. Yet, when DBM is parasitized by C. plutellae, its lar- whether such a preference has the potential to affect the val developmental time increases (Shi et al., 2002), which interaction between these natural enemies. We then con- could expose parasitized individuals to predation for ducted field cage studies to determine the effects that this longer than unparasitized individuals. Consequently, pre- predator and parasitoid have, alone and together, on dators may disproportionately consume parasitized larvae. DBM populations and consequent plant damage, which If so, populations of a parasitoid can be severely reduced is the most important measure from a crop management (Hassell and May, 1986). perspective. Podisus maculiventris (Say) (Hemiptera: Pentatomindae) is a generalist predator that preferentially feeds on lepidop- teran larvae, including larvae of DBM (Warren and Wallis, 2. Materials and methods 1971; McPherson, 1980; Muckenfuss, 1992; Wiedenmann et al., 1994; DeClerq et al., 1998; Westich and Hough- 2.1. maintenance Goldstein, 2001; Pell et al., 2008). Podisus maculiventris is found in a variety of agroecosystems and is common in A colony of DBM was established in 2001 from brassicas, especially when Lepidoptera populations are obtained from the USDA-ARS-CMPRU, Tifton, Georgia. high (McPherson, 1980; Culliney, 1986). Throughout much Larvae were maintained on a pinto bean-based artificial of its range, P. maculiventris is active during the same time diet (Carpenter and Bloem, 2002). Adults were held in of season as DBM (Ru and Workman, 1979; Aldrich et al., 30.5 30.5 30.5 cm cages and fed water through a water 1984; Herrick and Reitz, 2004), and, like C. plutellae,itis vial with a cotton wick. Tin foil (6 4 cm) was wrinkled, attracted to cabbage plants that have been damaged by dipped into cabbage juice (i.e. liquid extract obtained from 388 N.J. Herrick et al. / Biological Control 45 (2008) 386–395 three cabbage leaves blended in 1 L of water then strained) leaves in test arenas immediately after 6 h of exposure to and presented to adults for oviposition. parasitism, using a soft tipped paint brush. Unparasitized A colony of C. plutellae was established in 2001 with larvae for the 24 h experiments were placed in test arenas specimens obtained from Bio-Fac, Mathis, TX, and main- after 6 h of feeding on cabbage. All larvae were allowed tained on DBM larvae. Adult C. plutellae were provided to acclimate in test arenas without a predator for 18 h with honey and water. To maintain parasitoid populations, before predator introduction (i.e. making them 24 h old 200–500 second and third instar DBM were brushed onto post-parasitized-third instars or 24 h old post-third live host plant material daily and presented to C. plutellae instars). in a 122 76 76 cm screened cage containing 100–150 C. The 72 h experiments also were initiated with newly plutellae adults (50:50, male:female) for 24 h. All colonies ecdysed third instar DBM. Parasitized larvae for the 72 h and experiments were maintained in growth chambers at experiments were removed from the parasitism cage after 25 ± 1 °C, 50% RH with a 14:10 L:D photoperiod. 6 h and fed cabbage for 48 h, and then allowed to acclimate A colony of P. maculiventris was established in 2001 in test arenas for 18 h before predator introduction to from specimens obtained from the USDA-ARS-CMAVE, allow for larval and parasitoid development (thus equaling Gainesville, Florida. The colony was periodically infused 72 h old post-parasitized larvae). Unparasitized larvae were with F2 progeny of P. maculiventris collected at a site in placed on cabbage leaves in test arenas after 54 h of feeding Tallahassee, Florida, from 2001 to 2002. Podisus maculiv- on cabbage (i.e. 54 h development on cabbage plus 18 h entris were fed larvae of the yellow mealworm, Tenebrio acclimation equals 72 h old post-third instars). molitor L. (Rainbow Mealworms, Compton, CA). For no-choice feeding tests, P. maculiventris were offered DBM larvae that were: (1) 24 h post-third instars, 2.2. Prey selection trials (2) 24 h post-parasitized, (3) 72 h post-third instars, or (4) 72 h post-parasitized in a two-way factorial design, with Predation by P. maculiventris on DBM larvae was DBM larval age and parasitism status as the factors. assessed in a series of no-choice and choice experiments. Treatments were replicated 30 times, and each replicate These prey selection experiments were conducted in clear container contained 20 larvae of the appropriate treat- plastic containers measuring 19.5 14 10 cm, with lids ment. Experiments were conducted daily with five repli- containing a 154 cm2 hole covered with fine mesh screen. cates at one time. After the 24 h predator exposure, the Cabbage leaves (cv. ‘Constanza’), approximately predator was removed and the number of individuals 8 8 cm, were used as host plant material. Petioles were preyed on and still living was recorded. Larvae that were placed into 40 ml water vials (3.5 8.5 cm). Leaves, with still alive were placed on artificial diet and reared to - vials, were then set into the plastic containers on a card- tion to determine if they were parasitized or not. Because board stand 6 cm in height and at a 45° angle. This angle of the binomial nature of the response data (i.e. each simulated the natural posture of leaves and permitted the was either preyed on or not), the proportions of lar- leaves to be suspended off the bottom of arenas. All exper- vae preyed on relative to the total per replicate were ana- iments were conducted in growth chambers at 14:10 L:D, lyzed using a generalized linear model with a binomial 25 °C, and 68% RH. distribution and logit link function (Agresti, 1996; PROC were prepared in a similar manner for the no- GENMOD, SAS Institute, 2002). This analysis produces choice and choice experiments. Parasitized DBM larvae likelihood ratio statistics to determine the significance of were obtained by placing approximately 200 newly ecdysed larval age, parasitism status and their interaction on the third instars on cabbage leaves and positioned into the proportions of larvae preyed on. Treatment means were abovementioned screened cage containing 50–100 C. plutel- compared with the least squares means option (a = 0.05) lae adults (50:50, male:female) for 6 h. This duration of (LSMEANS, SAS Institute, 2002). exposure minimized superparasitism and resulted in In choice tests, P. maculiventris were offered a group of >95% of the larvae being parasitized (NJH, unpublished DBM larvae that consisted of either: (1) equal numbers of data). Unparasitized larvae were treated similarly but were 24 h post-third instars and 24 h post-parasitized or (2) not exposed to parasitism. equal numbers of 72 h post-third instars and 72 h post-par- The following experiments were named according to asitized. There were 30 replicates for each of the two age how long after parasitization the trial occurred, for exam- groups of DBM larvae, and data for the two age groups ple, the 24 h experiments all began 24 h post-parasitism were analyzed separately. Each replicate container con- and the 72 h experiments all began 72 h post-parasitism tained 10 unparasitized and 10 parasitized larvae. Because with same age unparasitized larvae in the appropriate treat- the parasitism treatments (parasitized and unparasitized) ments. Individual third instar P. maculiventris wereusedin were paired within each replicate, the replicate containers each experimental replicate, with DBM larvae exposed to a were considered blocks in a randomized complete block predator for 24 h. Each predator was used once. design. Experiments were conducted daily with five repli- The 24 h experiments were conducted with third instar cates at one time. DBM larvae that were 24 h post ecdysis. Parasitized larvae After the 24 h experimental exposure, the predator was for the 24 h experiments were placed directly on cabbage removed and the numbers of individuals preyed on and still N.J. Herrick et al. / Biological Control 45 (2008) 386–395 389 living were recorded. Larvae that were still alive were were released one week after the initial DBM release. Five placed on artificial diet and reared to pupation to deter- third instar P. maculiventris were released on each plant mine if they were parasitized or not. To account for the (n = 30 per cage) and 60 C. plutellae pupae (<24 h from random block component, the proportions of larvae eclosion, typical male: female in rearing is 1:1) were preyed upon were analyzed using a generalized linear released per cage in the appropriate treatments. We chose mixed model with parasitism status as a fixed main effect these densities to have approximately equal populations and replicate as a random block effect. The model was fit of predators and viable female parasitoids in the cages. with a binomial distribution and logit link function (Mad- To test the effectiveness of our field cages to prevent den et al., 2002; PROC GLIMMIX, SAS Institute, 2006). immigration of naturally-occurring DBM, we also maintained four cages designed identically to the treat- 2.3. Field cage interaction trials ment cages. Plants in these cages were sampled in the same fashion as treatment cages; these plant only cages Two field studies were conducted at the Florida Agri- were always sampled before the insect treatment cages. cultural and Mechanical University, Viticulture and At no point were DBM or damage by DBM found. Small Fruit Research Center, Tallahassee, Florida, from Cotesia plutellae and P. maculiventris were present 31 March to 17 June 2002 and from 10 March to 19 throughout the experiments and only in the appropriate June 2003. Four plots were tilled, covered with landscap- treatments. ing cloth (to limit alternate plant growth), and re-covered To estimate DBM populations, whole plant counts of with approximately 2.5 cm of topsoil (to simulate natural larvae and pupae were made weekly on each plant and conditions). Sixteen 1.8 1.8 1.8 m aluminum pipe summed to give a total for each cage per date. To account field cages covered with 32 32 mesh screen (159 holes for counts made on the same experimental units (i.e. cages) cm2) and a 1.8 m zippered door were erected on top of over time, a repeated measures ANOVA was performed on the plots in a randomized complete block design with the log(y + 1) transformed DBM counts per cage to deter- four blocks consisting of four cages. The distance mine the effects of the natural enemies on DBM popula- between cages was 1.0 m with 1.3 m between blocks of tions (Littell et al., 1998; PROC MIXED, SAS Institute, cages. 2002). The logarithm transformation reduced heterosced- In the 2002 growing season, cabbage, Brassica oleracea asticity and allowed us to test for an additive versus non- var. capitata cv. ‘Bonnie’s Hybrid’ was used as the host additive effect of the combined natural enemies (Sih plant. Brassica oleracea var. capitata cv. ‘Constanza’ was et al., 1998). The natural enemy treatments (with or with- used in the 2003 growing season. Prior to establishment out C. plutellae, and with or without P. maculiventris) in the field, all plants were grown from seed in a and their interaction were treated as fixed effects and block 4m2 2.4 m room with growth lamps at approximately was a random factor. Because measurements taken more 18 °C. Plants were watered with a 20:20:20 N:P:K mixture closely in time were more strongly correlated than those of fertilizer approximately twice a week. Six cabbage trans- more separated in time, we used a first-order auto-regres- plants, at the V4 stage (i.e. a transplant with approximately sive covariance structure for the model. Back transformed 6–8 leaves) (Andaloro et al., 1983), were planted in two data are presented. Graphs are plotted with connecting rows in each cage with approximately 30.5 cm between lines between data points (weeks) for ease of visual plants. This is similar to the arrangement that growers interpretation. use. Plants were allowed to establish for two weeks, A second analysis was conducted to determine if DBM watered with approximately 1000 L of water per plant populations in the combined natural enemy treatments per season, and given a 20:20:20 N:P:K mixture of fertilizer met the expectations of the multiplicative risk model. To once a month. do this analysis, the observed numbers of DBM in the indi- Four treatments were randomly assigned in a factorial vidual natural enemy treatments for each block were first arrangement to individual cages that were laid out in a ran- expressed as a proportion relative to their respective domized complete block design: (1) DBM, (2) DBM with DBM alone treatment. These values were then used to gen- P. maculiventris, (3) DBM with C. plutellae, and (4) erate a predicted number of DBM for the combined natu- DBM with P. maculiventris and C. plutellae. Treatments ral enemy treatment, using the following equation modified were laid out in a randomized block design to account from equation 2 presented in Soluk (1993): for the slope along the field where the experiments were conducted. Three weeks after transplanting, 12 DBM lar- P cm ¼ N d fðN d N cÞ=N d ðN d N pÞ=N d vae (three of each instar) were released per plant and 30 þ½ðN d N cÞ=N d x½ðN d N pÞ=N d g DBM pupae (50:50 male:female, <24 h from eclosion) were released per cage. One week later, 10 additional DBM lar- where Pcm is the predicted number of DBM in the com- vae (five each of second and third instars) were released per bined natural enemy treatment cage for a specific block plant. Releases were done this way to create a broad age on a specific sample date, Nd is the observed number of distribution of DBM and allow for continuously reproduc- DBM in the no natural enemy cage for that specific block ing populations in each cage. The natural enemies also and sample date, Nc is the observed number of DBM in the 390 N.J. Herrick et al. / Biological Control 45 (2008) 386–395 corresponding C. plutellae cage, and Np is the observed 3. Results number of DBM in the corresponding P. maculiventris cage. 3.1. Prey selection trials These predicted values were then compared with the observed values, using a randomized complete block, In no-choice tests, P. maculiventris preyed on greater repeated measures ANOVA, with a first-order auto-regres- numbers of parasitized than unparasitized larvae sive covariance structure, as described above. These analy- (v2 = 21.40; df = 1; P < 0.001) (Fig. 1). There was no inter- ses were conducted using transformed log(y + 1) values. If action between DBM age and treatment (v2 = 1.41; df = 1; the observed and predicted values do not differ, it would P = 0.24); significantly more parasitized than unparasitized suggest the natural enemies act independently. If the larvae were preyed on within each of the age groups (24 H: observed numbers of DBM exceed the predicted numbers, v2 = 6.74; df = 1; P = 0.0094; 72 H: v2 = 14.60; df = 1; it would suggest that the natural enemies act antagonisti- P = 0.0001). Overall, larval age did affect predation, with cally. If the observed numbers are less than the predicted P. maculiventris preying on more young larvae (24 h) than numbers, it would suggest that the natural enemies act syn- older larvae (72 h) (v2 = 19.63; df = 1; P < 0.0001; Fig. 1). ergistically (Sih et al., 1998). This difference held among both parasitized and unparasit- At the end of each growing season, the plants were ized larvae (parasitized: v2 = 6.05; df = 1; P = 0.0139; removed from the cages and nine leaves were randomly unparasitized: v2 = 13.59; df = 1; P = 0.0002; Fig. 1). sampled from each plant. Three leaves were sampled In choice tests with the younger larvae as prey (24 H), P. from the top (new growth), the middle (intermediate maculiventris did not show a significant prey preference growth), and the base (old growth) of each plant. Plant between parasitized and unparasitized DBM (F = 1.07; damage estimates were made by photocopying individual df = 1, 29; P = 0.31) (Fig. 2). However, in choice tests with leaves at high resolution and taking four random the older DBM larvae (72 h), predators preyed on parasit- 10.5 cm2 leaf disc samples per leaf. Leaf discs were ized larvae significantly more than on unparasitized larvae obtained by turning the photocopy upside down so dam- (F = 9.29; df = 1, 29; P = 0.0049; Fig. 2). Based on ad hoc aged areas were not visible. The outer leaf margin was visual observations, development of parasitized larvae was traced and the entire leaf was partitioned into four, slower than that of unparasitized larvae. After 72 h, most roughly equal sized main quadrants. Each of the four parasitized larvae remained third instars and were smaller main quadrants was divided into four additional equal than unparasitized larvae, which had molted to fourth sized quadrants. One sample was taken randomly from instars by the end of the trial. one of the four equal sized subquadrants with a 2 10.5 cm black ink stamp. Damaged areas within each 3.2. Field cage interaction trials disc sample were then traced to obtain definitive damage margins. The traced samples were scanned and analyzed In 2002, DBM populations were significantly lower in using Scion Image software (Scion, Frederick, MD) to the presence of C. plutellae (F = 5.21; df = 1, 9; obtain the total area damaged per sample, as described P = 0.048) (Fig. 3A); however, DBM populations were by O’Neal et al. (2002). Plant damage estimates were not affected by the presence of P. maculiventris (F = 0.19; analyzed as a randomized complete block-split plot design with natural enemy treatments as the whole plots 7 and plant growth stages as the subplots. Proportions of Unparasitized Parasitized damage relative to sample area were log(y + 1) trans- 6 formed (Sokal and Rohlf, 2001) before being analyzed by ANOVA (PROC MIXED, SAS Institute, 2002). Pair- 5 wise comparisons of growth stages were made using the least squares means option (a = 0.05) (LSMEANS, SAS 4 Institute, 2002). As with the DBM count data, a second analysis was 3 performed on the log(y + 1) plant damage data to deter- 2 mine if plant damage in the combined natural enemy Mean Larvae Preyed On treatments met the expectations of the multiplicative risk 1 model. Predicted values were generated for the combined natural enemy treatment using the observed damage from 0 the individual natural enemy treatments cages relative to 24 Hours 72 Hours their corresponding DBM only treatment cage. The actual Treatment and predicted log transformed values were then com- Fig. 1. Mean (± SEM) larvae preyed on by P. maculiventris in no-choice pared, using a randomized complete block, split plot tests of parasitized and unparasitized P. xylostella larvae at 24 h and 72 h ANOVA as described above for the initial plant damage post-parasitism. Stars indicate a significant difference in predation between analysis. unparasitized and parasitized larvae within each age group (a = 0.05). N.J. Herrick et al. / Biological Control 45 (2008) 386–395 391

7 no synergistic or antagonistic interactions. The test of the Unparasitized Parasitized multiplicative risk model for the combined natural enemies 6 also indicates that C. plutellae and P. maculiventris did not interact in a synergistic or antagonistic manner on DBM 5 populations, as there was no significant difference between observed and predicted DBM populations in the combined 4 natural enemy treatment (F = 0.05, df = 1, 3; P = 0.83). The significant date effect in the initial factorial ANOVA 3 (F = 7.58; df = 11, 33; P < 0.0001) reflects changes in DBM 2 populations over time. The date C. plutellae interaction

Mean Larvae Preyed On (F = 3.05; df = 11, 33; P = 0.0065, Fig. 3A) suggests that 1 C. plutellae were significantly more effective in mid to late season than early season. 0 Plant damage in 2002 was significantly lower in the two 24 Hours 72 Hours treatments where C. plutellae was present (F = 5.07; df = 1, Treatment 9; P = 0.05) (Fig. 4A). There was no effect of P. maculiven- Fig. 2. Mean (± SEM) larvae preyed on by P. maculiventris in choice tests tris on plant damage (F = 0.73; df = 1, 9; P = 0.4), nor was of parasitized and unparasitized P. xylostella larvae at 24 h and 72 h post- there a significant interaction between the natural enemies parasitism. Stars indicate a significant difference in predation between (F = 0.00; df = 1, 9; P = 0.95). There was a significant dif- unparasitized and parasitized larvae within each age group (a = 0.05). ference in plant damage among growth stages (F = 9.57; df = 2, 24; P = 0.0009) where the new growth received df = 1, 9; P = 0.67). There was no significant interaction the least amount of damage (Fig. 4A). As with the DBM between the two natural enemies on DBM populations populations, there was no evidence of antagonistic or syn- (F = 0.02; df = 1, 9; P = 0.88), which suggests there was ergistic interactions between C. plutellae and P. maculiven-

280 260 A C. plutellae 240 P. maculiventris 220 200 P. xylostella 180 160 C. plutellae and P. 140 maculiventris 120 100 80 60 40 20 0 3/24 3/31 4/7 4/14 4/21 4/28 5/5 5/12 5/19 5/26 6/2 6/9 6/16 6/23

140 B C. plutellae 120

Mean Larvae and Pupae P. maculiventris 100 P. xylostella

80 C. plutellae and P. maculiventris 60

40

20

0 2/24 3/3 3/103/17 3/24 3/31 4/7 4/14 4/21 4/28 5/5 5/12 5/19 5/26 6/2 6/9 6/16 6/23 6/30 Date

Fig. 3. Mean (± SEM) P. xylostella larvae and pupae under four treatments: (1) P. xylostella only, (2) P. xylostella with P. maculiventris, (3) P. xylostella with C. plutellae, and (4) P. xylostella with P. maculiventris and C. plutellae in spring 2002 (A) and spring 2003 (B) (arrow indicates natural enemy releases). 392 N.J. Herrick et al. / Biological Control 45 (2008) 386–395

12 A New Growth 10 Intermediate Growth

8 Old Growth

6

4

2

0 P. xylostella P. maculiventris C. plutellae P. maculiventris and C. plutellae

12 B New Growth 10 Intermediate Growth Old Growth Mean Percent Plant Damage 8

6

4

2

0 P. xylostella P.maculiventris C. plutellae P. maculiventris and C. plutellae Treatment

Fig. 4. Mean (± SEM) percent plant damage under four treatments: (1) P. xylostella only, (2) P. xylostella with P. maculiventris, (3) P. xylostella with C. plutellae, and (4) P. xylostella with P. maculiventris and C. plutellae in spring 2002 (A) and spring 2003 (B). tris on the amount of plant damage (test of the multiplica- produced by the released DBM in 2002 created less damage tive risk model: F = 0.01, df = 1, 3; P = 0.93). but higher DBM estimates. Also, there could have been an In 2003, DBM populations were significantly lower in effect of plant cultivar. Perhaps, cv. ‘Constanza’ is nutri- the presence of C. plutellae and in the presence of P. mac- tionally inferior for DBM and individuals required more uliventris (F = 9.89; df = 1, 9; P = 0.012 and F = 6.24; plant material to complete development. df = 1, 9; P = 0.034, respectively; Fig. 3B). Although Despite the fact that DBM populations were lower in DBM populations were lowered by both natural enemies, the presence of the natural enemies, plant damage in there was no significant interaction between the natural 2003 was not affected by natural enemies (C. plutellae: enemies (F = 1.53; df = 1, 9; P = 0.25). Analysis of vari- F = 0.05; df = 1, 9; P = 0.82; P. maculiventrus: F = 4.72; ance revealed a highly significant date effect (F = 40.82; df = 1, 9; P = 0.058; interaction: F = 0.63; df = 1, 9; df = 15, 45; P < 0.0001). The date C. plutellae interac- P = 0.45). Furthermore, the test of the multiple risk model tion (F = 2.19; df = 15, 45; P = 0.022, Fig. 3B) suggests showed that there was no evidence of antagonistic or syn- that C. plutellae were significantly more effective in mid ergistic interactions between C. plutellae and P. maculiven- to late season than early season. tris on the amount of plant damage (F = 0.72, df = 1, 3; Although in this season both natural enemies had signif- P = 0.46). There was a highly significant growth stage icant effects on DBM populations, the lack of a difference effect (F = 60.16; df = 2, 24; P < 0.0001) where all three between observed and predicted values in the test of the stages differed. The intermediate stage received the most multiplicative risk model indicated that the effects of C. damage, then old growth, followed by the new growth, plutellae and P. maculiventris were independent rather than which received the least damage (Fig. 4B). antagonistic or synergistic (F = 1.18; df = 1, 3; P = 0.36). Plant damage in 2003 was almost twice as much as in 4. Discussion 2002 although total seasonal DBM populations were nearly half those in 2002 (Figs. 3A, B and 4A, B). We sus- In our laboratory trials, larvae of DBM were readily pect that the greater numbers of young larvae (1st instar) preyed on by P. maculiventris, but there was greater preda- N.J. Herrick et al. / Biological Control 45 (2008) 386–395 393 tion on DBM parasitized by C. plutellae than on unparasit- (Mitchell et al., 1997; Waladde and Leutle, 2001; Guilloux ized DBM. Regardless of caterpillar age, P. maculiventris et al., 2003). Therefore, natural enemies that complement preyed more on parasitized larvae of DBM than on unpar- the action of C. plutellae would be desirable for pest asitized larvae, in no-choice trials. In choice trials, greater management. predation on parasitized larvae occurred, but only with Ibrahim and Holopainin (2004) have suggested that P. the older DBM larvae, which had been parasitized for three maculiventris can play an important role in the suppression days. Still, the results support the hypothesis that parasit- of DBM. Podisus maculiventris are attracted to cabbage ized larvae are more susceptible to predation than unpara- damaged by DBM (Vuorinen et al., 2004) an individual sitized larvae, meaning that C. plutellae populations may P. maculiventris nymph has the potential to consume over be negatively affected by P. maculiventris. 100 DBM larvae during development (Ibrahim and Holop- Factors contributing to greater predation on parasitized ainin, 2004). Our studies indicate that P. maculiventris had larvae are uncertain but may be related to parasitism effects no effect in the 2002 field trial and DBM suppression was on the DBM host. Many parasitoids are known to alter caused by C. plutellae alone. However, in 2003 we found their hosts’ physiology and development (Vinson and that P. maculiventris significantly reduced DBM popula- Iwantsch, 1980; Hurd, 1990; Pennacchio and Strand, tions, the year with the lower DBM populations. However, 2006). In the case of DBM, parasitism by C. plutellae pro- its presence did not enhance the control of DBM by C. plu- longs the larval development of its host (Shi et al., 2002), tellae or reduce plant damage. Therefore, their combined which probably confers a nutritional benefit to the parasit- effect appears to be additive despite evidence for differential oid. While such host regulation may have adaptive benefits predation on parasitized individuals by P. maculiventris. for the parasitoid, these changes also may facilitate preda- Most importantly, P. maculiventris does not disrupt con- tion on the host and therefore carry adaptive costs for the trol exerted by C. plutellae. parasitoid (Lafferty, 1992). Adamo (1997) has shown that The effects of generalist predators in an overall biologi- movement of Manduca sexta larvae is inhibited when par- cal control program depend on how the constituent species asitized by Cotesia congregata. Often, when larvae of DBM of a particular food web interact (Janssen et al., 1998). are attacked, they escape by dropping from the leaf on a Harvey and Eubanks (2005) found that Solenopsis invicta silk thread (Roux et al., 2005). This behavior may become Buren (Hymenoptera: Formicidae) is an intraguild preda- limited in larvae parasitized by C. plutellae as the parasit- tor of DBM parasitoids, but this intraguild predation is oid develops, which could make parasitized larvae more not disruptive to overall control because the ants did not vulnerable to predation. show differential predation for parasitized caterpillars. Prey size can be important in selection of prey by P. Despite the preference of P. maculiventris for parasitized maculiventris (Mukerji and LeRoux, 1969; DeClerq et al., larvae in our laboratory trials, the combined natural enemy 2003). DeClerq et al. (2003) that found that predation by treatment in our field studies did not produce any antago- P. maculiventris on H. axyridis is far greater on young lar- nistic interactions. This discrepancy could indicate that as vae than on older larvae and adults because young, smaller the spatial scale and complexity of the environment H. axyridis larvae are less able to defend themselves and increases, the intensity of intraguild predation by P. macu- lack the ability to escape the larger P. maculiventris.At liventris is reduced (C¸ akmak et al., 2006). Furthermore, 24 h after parasitization, C. plutellae are still in the Bilu and Coll (2007) point out that the intensity of intra- stage (Schuler et al., 2004), but their hosts still have guild predation can vary with predator–parasitoid ratio reduced feeding rates compared with unparasitized larvae and timing of the potential interaction. (Bae and Kim, 2004). Differences in prey apparency at this Inclusion of plant damage estimates from long-term tri- stage may not have been great enough to affect predation als is important for the study of multiple species interac- by P. maculiventris in choice trials. However, at 72 h tions and provides practical information to understand post-parasitism, parasitized DBM larvae are smaller than the effects of natural enemy combinations on crop damage. their unparasitized counterparts, and this size difference Furthermore, experiments that manipulate predator–para- may contribute to differential predation. sitoid ratios and timing of introduction of either predator The preferential selection for parasitized DBM led us to or parasitoid into a system over multiple consecutive sea- hypothesize that the presence of P. maculiventris could neg- sons will be important in future research that addresses atively impact the parasitoid C. plutellae and its ability to multiple natural enemy interactions. Understanding such suppress DBM. However, over the course of two field sea- phenomena will further facilitate understanding interac- sons, the effect that these natural enemies had on suppress- tions among natural enemies of DBM and other agricul- ing DBM populations and plant damage was variable. The tural pests, which will ultimately enhance biological presence of C. plutellae did significantly reduce DBM pop- control. ulations in both years; however, C. plutellae reduced dam- age in 2003 only, the year with the higher DBM Acknowledgments populations. Although C. plutellae can be an effective par- asitoid of DBM, its parasitism levels can vary significantly We thank Melany Combs, Ryan Stype, Ricardo Smith, over the course of a growing season and among seasons and Byron Franklin, Jr. (USDA-ARS-CMAVE, Tallahas- 394 N.J. Herrick et al. / Biological Control 45 (2008) 386–395 see, FL); Robert Caldwell, and Susan Drawdy (USDA- Finke, D.L., Denno, R.F., 2003. Intra-guild predation relaxes natural ARS-CMPRU, Tifton, GA) for their assistance with field enemy impacts on herbivore populations. Ecological Entomology 28, set-up, insect rearing, and/or data entry. We also thank 67–73. Furlong, M.J., Groden, E., 2001. Evaluation of synergistic interactions Buddy Maedgen, Jr., Biofac, for supplying insects. Thanks between the Colorado potato beetle (Coleoptera: Chrysomelidae) to Stephen Hight (USDA-ARS-CMAVE), Kenneth Bloem pathogen and the , imidacloprid, and (USDA-APHIS-PPQ-CPHST), Loke Kok (Virginia Poly- cyromazine. Journal of Economic Entomology 94, 344–356. technic Institute and State University), and Joe Funder- Guilloux, T., Monnerat, R., Castelo, B.M., Kirk, A., Bordat, D., 2003. burk () for their helpful reviews of Population dynamics of Plutella xylostella (Lep., Yponomeutidae) and its parasitoids in the region of Brasilia. Journal of Applied Entomol- this manuscript. This research was supported in part by ogy 127, 288–292. the USDA-Sustainable Agricultural, Research and Educa- Harvey, C.T., Eubanks, M.D., 2005. Intraguild predation of parasitoids tion Grant RD309-040/3581407 and by cooperative agree- by Solenopsis invicta: a non-disruptive interaction. Entomologia ment 58-6615-8-022 USDA-ARS and Florida A&M Experimentalis et Applicata 114, 127–135. University. This article reports the results of research only. Hassell, M.P., May, R.M., 1986. Generalist and specialist natural enemies in insect predator–prey interactions. Journal of Ecology 55, Mention of a proprietary product does not constitute an 923–940. endorsement by USDA for its use. Herrick, N.J., Reitz, S.R., 2004. Temporal occurrence Podisus maculiven- tris (Hemiptera: Heteroptera: Pentatomidae) in North Florida. Florida Entomologist 87, 587–590. References Hurd, H., 1990. Physiological and behavioral interactions between parasites and invertebrate hosts. Advances in Parasitology 29, 271– Adamo, S.A., 1997. How parasites alter the behavior of their insect hosts. 319. In: Beckage, N.E. (Ed.), Parasites and Pathogens: Effects on Host Ibrahim, M.A., Holopanin, J.K., 2004. Podisus maculiventris (Hemiptera: Hormones and Behavior. Chapman and Hall, , pp. 231–245. Pentatomidae), a potential biocontrol agent of Plutella xylostella is not Agresti, A., 1996. Introduction to Categorical Data Analysis. John Wiley repelled with limonene treatments on cabbage. In: Kirk, A.A., Bordat, and Sons, New York. D. (Eds.), Improving Biocontrol of Plutella xylostella, . Proceedings of Aldrich, J.R., Kochansky, J.P., Abrams, C.B., 1984. Attractant for a the International Symposium, Montpellier, France. CIRAD, pp. 200– beneficial insect and its parasitoids: of the predatory spined 203. soldier bug, Podisus maculiventris (Heteroptera: Pentatomidae). Envi- Janssen, A., Pallini, A., Venzon, M., Sabelis, M.W., 1998. Behaviour and ronmental Entomology 13, 1031–1036. indirect interactions in food webs of plant-inhabiting . Andaloro, J.T., Rose, K.B., Shelton, A.M., Hoy, C.W., Becker, R.F., Experimental and Applied Acarology 22, 497–521. 1983. Cabbage growth stages. New York’s Food and Life Sciences Lafferty, K.D., 1992. Foraging on prey that are modified by parasites. The Bulletin 101, 1–4. American Naturalist 140, 854–867. Bae, S., Kim, Y., 2004. Host physiological changes due to parasitism of a Lim, G.S., 1982. The Biology and Effects of Parasites on the Diamond- braconid wasp, Cotesia plutellae, on diamondback moth, Plutella back Moth, Plutella xylostella (L.). Ph.D. Dissertation. University of xylostella. Comparative Biochemistry and Physiology 138, 39–44. London, Berkshire, London. Bilu, E., Coll, M., 2007. The importance of intraguild interactions to the Lindgren, P.D., Wolfenbarger, D.A., 1976. Competition between Tricho- combined effect of a parasitoid and a predator on aphid population gramma pretiosum and Orius insidiosus for caged tobacco budworms suppression. BioControl 52, 753–763. on cotton treated with chlordimeform sprays. Environmental Ento- C¸ akmak, I., Janssen, A., Sabelis, M., 2006. Intraguild interactions mologist 5, 1049–1052. between the predatory mites Neoseiulus californicus and Phytoseiulus. Littell, R.C., Henry, P.R., Ammerman, C.B., 1998. Statistical analysis of Experimental and Applied Acarology 38, 33–46. repeated measures data using SAS procedures. Journal of Animal Carpenter, J.E., Bloem, S., 2002. Interaction between insect strain and Science 76, 1216–1231. artificial diet in diamondback moth development and reproduction. Losey, J.E., Denno, R.F., 1998. Positive predator–predator interactions Entomologia Experimentalis et Applicata 102, 283–294. enhanced predation rates and synergistic suppression of aphid popu- Cloutier, C., Jean, C., 1998. Synergism between natural enemies and lations. Ecology 79, 2143–2152. biopesticides: a test case using the stinkbug Perillus bioculatus Losey, J.E., Denno, R.F., 1999. Factors facilitating synergistic predation: (Hemiptera: Pentatomidae) and tenebrionis the central role of synchrony. Ecological Applications 9, 378–386. against Colorado potato beetle (Coleoptera: Chrysomelidae). Journal Madden, L.V., Turechek, W.W., Nita, M., 2002. Evaluation of general- of Economic Entomology 91, 1096–1108. ized linear mixed models for analyzing disease incidence data obtained Colfer, R.G., Rosenheim, J.A., 2001. Predation on immature parasitoids in designed experiments. Plant Disease 86, 316–325. and its impact on aphid suppression. Oecologia 126, 292–304. Mallampalli, N., Castellanos, I., Barbosa, P., 2002. Evidence of intraguild Culliney, T.W., 1986. Predation on adult Phyllotreta flea beetles by predation by Podisus maculiventris on ladybeetle, Coleomegilla mac- Podisus maculiventris (Hemiptera: Pentatomidae) and Nabicula ame- ulate: implications for biological control of Colorado potato beetle, ricolimbata (Hemiptera: Nabidae). Canadian Entomologist 118, 731– Leptinotarsa decemlineata. BioControl 47, 387–398. 732. McPherson, J.E., 1980. A list of the prey species of Podisus maculiv- DeClerq, P., Peeters, I., Vergauwe, G., Thas, O., 2003. Interaction entris (Hemiptera: Pentatomidae). Great Lakes Entomologist 13, 17– between Podisus maculiventris and Harmonia axyridis, two predators 24. used in augmentative biological control in greenhouse crops. BioCon- Mitchell, E.R., Tingle, F.C., Navasero-Ward, R.C., Kehat, M., 1997. trol 48, 39–55. Diamondback moth (Lepidoptera: Plutellidae): parasitism by Cotesia DeClerq, P., Merlevede, F., Mestdagh, I., Vandendurpel, K., Mohaghegh, plutellae (Hymenoptera: Braconidae) in cabbage. Florida Entomolo- J., Degheele, D., 1998. Predation on tomato looper Chrysodeixis gist 80, 477–489. chalcites (Esper) (Lep., Noctuidae) by Podisus maculiventris (Say) and Mitchell, E.R., Hu, G.Y., Okine, J.S., Carpenter, J.E., 1999. Parasitism of Podisus nigrispinus (Dallas) (Het., Pentatomidae). Journal of Applied diamondback moth (Lepidoptera: Plutellidae) larvae by Cotesia Entomology 127, 93–98. plutellae (Hymenoptera: Braconidae) and (Hyme- Ferguson, K.I., Stiling, P., 1996. Non-additive effects of multiple natural noptera: Ichneumonidae) in cabbage fields after inundative releases of enemies on aphid populations. Oecologia 108, 375–379. C. plutellae. Journal of Entomological Science 34, 101–112. N.J. Herrick et al. / Biological Control 45 (2008) 386–395 395

Muckenfuss, A.E., 1992. Assessment of Natural Mortality of Diamond- Shi, Z., Liu, S., Li, Y., 2002. Cotesia plutellae parasitizing Plutella back Moth on Collards in Coastal South Carolina Using Cages, xylostella: host-age dependent parasitism and its effect on host Insecticides, and Field Sampling. Ph.D. Dissertation. Clemson Uni- development and food consumption. BioControl 47, 499–511. versity, South Carolina. Sih, A., Englund, G., Wooster, D., 1998. Emergent impacts of multiple Mukerji, M.K., LeRoux, E.J., 1969. The effect of predator age on the predators on prey. Tree 13, 350–355. functional response of Podisus maculiventris to the prey size of Galleria Sokal, R.R., Rohlf, F.J., 2001. Biometry, third ed. WH Freeman and mellonella. Canadian Entomologist 101, 314–327. Company. O’Neal, M.E., Landis, D.A., Isaacs, R., 2002. An inexpensive method for Soluk, D.A., 1993. Multiple predator effects: predicting combined measuring leaf area and defoliation through digital image analysis. functional response of stream fish and invertebrate predators. Ecology Journal of Economic Entomology 95, 1190–1194. 74, 219–225. Pell, J.K., Baverstock, J., Roy, H.E., Ware, R.L., Majerus, M.E.N., 2008. Streams, F.A., 1987. Within-habitat spatial separation of two Notonecta Intraguild predation involving Harmonia axyridis: a review of current species: interactive vs. noninteractive resource partitioning. Ecology knowledge and future perspectives. BioControl 53, 147–168. 68, 935–945. Pemberton, C.E., Willard, H.F., 1918. Interrelations of fruit fly parasites Swisher, B.J., Soluk, D.A., Wahl II, D., 1998. Non-additive predation in in Hawaiian. Journal of Agricultural Research 12, 285–297. littoral habitats: influences of habitat complexity. Oikos 81, 30–37. Pennacchio, F., Strand, M.R., 2006. Evolution of developmental Talekar, N.S., Yang, J.C., 1991. Characteristics of parasitism of strategies in parasitic Hymenoptera. Annual Review of Entomology diamondback moth by two larval parasites. Entomophaga 36, 95–104. 51, 233–258. Vinson, S.B., Iwantsch, G.F., 1980. Host regulation by insect parasitoids. Rosenheim, J.A., Wilhoit, L.R., Armer, C.A., 1993. Influence of intraguild Quarterly Review of Biology 55, 143–165. predation among generalist insect predators on the suppression of an Vuorinen, T., Nerg, A.-M., Ibrahim, M.A., Reddy, G.V.P., Holopainen, herbivore population. Oecologia 96, 439–449. J.K., 2004. Emission of Plutella xylostella-induced compounds from

Ru, N., Workman, R.B., 1979. Seasonal abundance and parasites of the grown at elevated CO2 and orientation behavior of the imported cabbageworm, diamondback moth, and cabbage webworm natural enemies. Plant Physiology 135, 1984–1992. in Northeast Florida. Florida Entomologist 62, 68–69. Waladde, S.M., Leutle, M.F., 2001. Parasitism of Plutella xylostella Roux, O., Barren, J.V., Gers, C., Arvanitakis, L., Legal, L., 2005. (Lepidoptera: Plutellidae): field and laboratory observations. South Antennal structure and oviposition behavior of the Plutella xylostella African Journal of Plant and Soil 18, 32–37. specialist parasitoid: Cotesia plutellae. Microscopy Research and Warren, L.O., Wallis, G., 1971. Biology of the , Podisus Technique 68, 36–44. maculiventris (Hemiptera: Pentatomidae). Journal of the Georgia SAS Institute, 2002. SAS System for Windows, Release 9.1. SAS Institute, Entomological Society 6, 109–116. Cary, NC. Westich, R., Hough-Goldstein, J., 2001. Temperature and host plant SAS, 2006. SAS/STAT: The GLIMMIX Procedure. SAS Institute, Cary, effects on predatory stink bugs for augmentative biological control. NC. Biological Control 21, 160–167. Schuler, T.H., Denholm, I., Clark, S.J., Stewart, C.N., Poppy, G.M., Wiedenmann, R.N., Legaspi, J.C., O’Neil, R.J., 1994. Impact of prey 2004. Effects of Bt plants on the development and survival of the density and facultative plant feeding on the life history of the predator parasitoid Cotesia plutellae (Hymenoptera: Braconidae) in suscepti- Podisus maculiventris (Heteroptera: Pentatomidae). In: Alomar, O., ble and Bt-resistant larvae of the diamondback moth, Plutella Wiedenmann, R.N. (Eds.), Zoophytophagous Heteroptera: Implica- xylostella (Lepidoptera: Plutellidae). Journal of Insect Physiology 50, tions for Life History and Integrated Pest Management. Entomolog- 435–443. ical Society of America, Lanham, Maryland, pp. 95–118.