Dissemination of Beauveria bassiana by Honey Bees (Hymenoptera: Apidae) for Control of Tarnished Plant Bug (Hemiptera: Miridae) on Canola

Transcription

1 BIOLOGICAL CONTROLÐMICROBIALS Dissemination of Beauveria bassiana by Honey Bees (Hymenoptera: Apidae) for Control of Tarnished Plant Bug (Hemiptera: Miridae) on Canola M. S. AL MAZRA AWI, 1 J. L. SHIPP, 2 A. B. BROADBENT, 3 AND P. G. KEVAN 4 Environ. Entomol. 35(6): 1569Ð1577 (2006) ABSTRACT Large screened cages trials were conducted to assess the potential of honey bees, Apis mellifera L., to vector Beauveria bassiana (Balsamo) to canola, Brassica napus L., against the tarnished plant bug, Lygus lineolaris (Palisot de Beauvois). The bees effectively vectored the inoculum from the hives to the crop. Conidia of B. bassiana was recovered from 100, 64Ð77, 70Ð82, and 47Ð83% of bees, ßowers, leaves, and L. lineolaris samples, respectively, collected on four sampling dates in 2002 and Mean mortalities of L. lineolaris collected from the Beauveria-treated cages in the Þeld were 56 and 48% compared with 9 and 10% in the controls on the Þrst and second sampling dates in 2002, respectively, and 22 and 45% in the treated cages in the Þeld compared with 15 and 22% in the controls on the Þrst and second sampling dates, respectively, in These results indicate that bees may provide a new novel means for applying B. bassiana to manage L. lineolaris on canola. The beneþts are better pollination, reduction in pest pressure of L. lineolaris, and reduced reliance on insecticides. KEY WORDS honey bees, Beauveria, canola, Lygus lineolaris, inoculum dispenser 1 Corresponding author: Department of Biotechnology, Al balqa Applied University, Assalt, Jordan ( m_mazrawi@ yahoo.com). 2 Agriculture and Agri-Food Canada, Greenhouse and Processing Crops Research Centre, Harrow, Ontario, Canada N0R 1G0. 3 Agriculture and Agri-Food Canada, Southern Crop Protection and Food Research Centre, London, Ontario, Canada N5V 4T3. 4 Department of Environmental Biology, University of Guelph, Guelph, Ontario, Canada N1G 2W1. The tarnished plant bug, Lygus lineolaris (Palisot de Beauvois) (Hemiptera: Miridae), is a serious pest of many Canadian crops including canola, Brassica napus L. and B. rapa L. (Broadbent et al. 2002). Adults and late-instar nymphs cause direct yield loss in canola by feeding on individual seeds through the silique walls and indirect loss by feeding on other plant tissues (Butts and Lamb 1990). Direct yield loss by Lygus bugs can range between 1Ð7% in Alberta (Butts and Lamb 1991) and 3Ð5% in Manitoba (Turnock et al. 1995). Current methods of L. lineolaris management rely on the application of chemicals (Noma and Strickler 1999). However, resistance to some chemicals has been reported in Þeld populations of L. lineolaris on cotton plants (Snodgrass 1996). Moreover, chemicals can have adverse effects on pollinators during bloom (Noma and Strickler 1999). A promising biological control agent against L. lineolaris is the entomopathogenic fungus, Beauveria bassiana (Balsamo) Vuillemin. Previous studies have found that this fungus is lethal to several species of Lygus bugs including L. lineolaris (Bidochka et al. 1993, Noma and Strickler 1999). In Arkansas, an isolate of the fungus (ARSEF 3769) isolated from naturally infected L. lineolaris was found to be highly infective to L. lineolaris nymphs and adults under laboratory and Þeld conditions (Steinkraus and Tugwell 1997). A commercial product, Mycotrol WP (Mycotech, Butte, MT), which is based on B. bassiana strain GHA, was evaluated against L. lineolaris on ßowering canola planted within a cotton Þeld applied alone and mixed with imidacloprid. The mixture resulted in 98.9% mortality 7 d after treatment, which was higher than that caused by imidacloprid, a recommended insecticide, alone (75%) or MycotrolWP alone (91.5%) (Steinkraus and Tugwell 1997). More recently, the susceptibilities of second-instar L. lineolaris to 32 fungal isolates from six genera including 18 isolates of B. bassiana were evaluated under laboratory conditions (Liu et al. 2002). Results showed that 10 of the 11 isolates that produced 90% mortality were B. bassiana (Liu et al. 2002). In the early 1990s, a novel technology was developed for the delivery of microbial biological control agents against plant pathogens. It required a special inoculum dispenser attached to hives of honey bees so that the honey bees, when exiting the hive, became coated with the control agent, which they dispersed to the ßowers. Since then, this technology has been evaluated successfully against many agricultural pests such as fungi, bacteria, and insects (Kevan et al. 2003, Al mazraõawi et al. 2006). Bee vectoring has several advantages over spraying as a means to deliver biological control agents to the ßowers of crops. Bees usually place the inoculum

2 1570 ENVIRONMENTAL ENTOMOLOGY Vol. 35, no. 6 precisely in or on the ßowers of the crop, thereby targeting the pests that feed on or inhabit the ßowers. This reduces the total amount of inoculum needed to treat a certain area and reduces exposure to nontarget organisms. Moreover, bees forage the crop almost daily, thus continually carrying the inoculum to newly opened ßowers. This eliminates the need for frequent spraying to protect the newly opened ßowers (Kevan et al. 2003). Given the potential of honey bees to vector microbial control agents and the susceptibility of L. lineolaris to B. bassiana, this study was initiated to investigate the ability of honey bees to deliver B. bassiana to canola ßowers targeting L. lineolaris. Developing a pollinator vector technology for the management of L. lineolaris populations on canola brings the beneþt of reducing pest populations and pesticide use while improving pollination of the crop. Materials and Methods This study was conducted over 2 yr (2002Ð2003), with the Þrst trials of bee vectoring of B. bassiana being made under controlled conditions in a greenhouse. Subsequent trials were made in the Þeld using large screened cages. Later, the study expanded to include both screened cages and open Þeld plots. Insect Cultures. Newly emerged adults of L. lineolaris were obtained from a laboratory culture maintained in the insectaries at Agriculture and Agri-Food Canada (AAFC), Southern Crop Protection and Food Research Center (SCPFRC), London, Ontario, Canada. This culture was established from Þeld-collected L. lineolaris adults on alfalfa at the SCPFRC research farm and kept in cages containing green beans, potato sprouts, and romaine lettuce at 24 C, 60% RH, and 16-h photoperiod. Nuclear honey bee colonies of the Buckfast strain were used in the greenhouse trials and in the Þeld trials. In the 2002 Þeld trials, nuclear hives consisted of 10 Langstroth frames of bees and brood. In the greenhouse and the 2003 Þeld trials, they consisted of Þve frames (Townsend House Apiculture Field Laboratory, University of Guelph). During the trials, the hives were supplied with 50% (wt:wt) sugar syrup and water outside the hive. Each honey bee hive had a modiþed entrance containing an inoculum dispenser (Fig. 1) similar to that used by Peng et al Greenhouse Trials. Three isolates of B. bassiana were evaluated in the bee vectoring studies in a greenhouse at the University of Guelph. The isolates were ARSEF 3769 (ARK), NY (NY; BB008, SCPFRC, London, Ontario, Canada), and the commercial isolate GHA (BotaniGard WP; Emerald BioAgriculture, Salt lake City, UT). All isolates were cultured on SabouraudÕs dextrose agar (SDA; Difco, Detroit, MI.) for 2 wk in the dark at 24 1 C. Conidia were harvested with a spatula and stored at 4 C until used. To estimate the colony forming units (CFUs) per unit weight of the harvested conidia of each isolate, six 0.1-g samples taken at random were each suspended in 100 ml sterile distilled water and 0.1% Tween 80 and agitated on a Fig. 1. The inoculum dispenser. rotary shaker at 125 rpm for 2 h. Three 0.1-ml aliquots of 10-fold serial dilutions of each suspension were spread on oatmeal agar petri plates amended with 550 g/ml Dodine, 400 g/ml penicillin G, 1,000 g/ml streptomycin sulfate, and 5 g/ml crystal violet (Beilharz et al. 1982). After incubating the plates for 4Ð5 d in darkness, CFUs were counted. To determine the viability of the conidia before each trial, six 0.01 g samples of conidia were suspended in 100 ml distilled water and 0.1% Tween 80. Two hundred microliters of the conidial suspension was added to 1 ml of SabouraudÕs dextrose broth amended with 1% yeast extract in a sterile test tube. The suspension was incubated at 24 1 C for 24 h. Four subsamples from each suspension consisting of 200 conidia were examined for germination using a hemacytometer under a compound microscope. Inocula from each strain were prepared by mixing the dry conidia with corn ßour (particle size, 45Ð90 m) to a concentration of CFU/g. Corn ßour (45Ð90 m) was previously found to increase the acquisition of B. bassiana conidia compared with corn ßour of larger particle sizes or other ßours such as wheat, oat, and barley ßours (Al mazraõawi 2004). Trials were conducted inside four screened cages (1.5 by 4 by 1.8 m height) placed inside a greenhouse compartment. Three groups of ßowering canola plants, variety DKL 35Ð25 (Monsanto Canada, Listowel, Ontario, Canada), were placed randomly inside each cage. Each group consisted of 10 canola plants planted in a 15-cm-diameter pot. Plants were infested with newly emerged L. lineolaris adults at a rate of Þve adults per plant. A nuclear honey bee colony with an inoculum dispenser was placed inside each cage. The dispensers were kept empty for 2 d to allow honey bees time to adjust to the presence of the dispenser before the vectoring trials were initiated. Then, dispensers in three hives were Þlled with formulations of the different B. bassiana isolates. The fourth hive served as control, and its dispenser was Þlled with noninoculated sterile corn ßour (particle size 45Ð90 m). In both the inoculated and noninoculated treatments, the dispensers were Þlled with 25 g of tested material. After 48 h of bee foraging, 10 L. lineolaris adults were collected from each group of plants by using an aspirator.

3 December 2006 AL MAZRAÕAWI ET AL.: BEE-VECTORED B. bassiana AGAINST TARNISHED PLANT BUG 1571 The L. lineolaris were placed in ventilated PVC cylinders (10 cm in diameter by 20 cm in height) lined with moist Þlter paper and incubated at 24 1 C and a 16-h photoperiod. Lettuce leaves and bean pods were provided as a food source to the caged L. lineolaris. Mortality was recorded daily for 7 d. Dead insects were removed and placed in petri plates lined with moist Þlter paper to encourage external growth of conidia. The experiment was replicated three times, each time with a new set of plants and freshly prepared inocula. Temperature in the greenhouse was controlled and monitored by an Argus Greenhouse Management System (Argus Control System, White Rock, British Columbia, Canada). Field Trials Based on the results from the greenhouse trials, the commercial strain of B. bassiana GHA was chosen for the Þeld trials. Viability of the conidia as well as CFU per unit weight of the commercial product was assessed as above. Inocula were prepared directly before the trials, which consisted of dry conidia of the GHA strain mixed with corn ßour of particle sizes ranging between 45 and 90 m toa concentration of CFU/g. Field trials were made in Monkton, Ontario, using nine insect-proof cages (1.8 by 4 by 1.8 m height) that were placed over a ßowering canola crop, Monsanto variety DKL 35Ð25. The natural infestation of L. lineolaris was augmented by releasing laboratory reared adults and Þfth-instar nymphs. A total of 300 adult and 200 nymphal L. lineolaris were released inside each cage. A nucleus hive of honey bees supplied with a dispenser was placed in the corner of each of six cages. The remaining three cages served as controls without bees. Dispensers in three hives were Þlled with 25 g of corn ßour inoculated with heat-killed conidia, whereas dispensers in the three remaining hives were Þlled with 25 g corn meal inoculated with living conidia of B. bassiana. During the experiment, the dispensers were reþlled at 48-h intervals. The experiment was arranged in a randomized complete block design with three replicates. The three treatments, honey bees and viable conidia, honey bees and inactivated conidia, and no bees, were randomized within the blocks. Average daily maximum and minimum temperatures and total precipitation were recorded during the experiment. To assess conidial dissemination, samples of L. lineolaris, honey bees, and canola ßowers were collected from the cages during peak ßowering and on two sampling dates. The Þrst sampling date occurred 2 d after Þlling the dispensers on 14 July and the second 4 d later. On each sampling date, 10 honey bees and 20 ßowers were collected from each cage treated with B. bassiana. The bees were collected directly after they left the hive through the dispenser. The third fully opened ßower from the top of the main raceme was collected from each of 20 different plants. Flowers adjacent to the hive were not sampled. To enumerate the viable conidia, honey bees and ßower samples were agitated individually in ßasks containing 100 ml sterilized distilled water and 0.1% Tween 80 on a rotary shaker for 2 h. Three 0.1-ml aliquots of 10-fold serial dilutions of each suspension were spread on oatmeal agar petri plates as above. After 4Ð5 d of incubating the plates in the dark, CFUs were counted. To assess L. lineolaris mortality, six subsamples each consisting of 10 adults were collected from all cages (60 per cage and total of 540) into separate plastic vials using an aspirator. One subsample from each cage was treated as the honey bees and ßower samples to enumerate the amount of viable spores on them. Each of the remaining Þve subsamples from each cage was treated as in the greenhouse trials above to assess mortality and mycosis on them. Field Trials Inocula for the 2003 trials were prepared as above in the 2002 trials using fresh batches of the commercial product of B. bassiana strain GHA (BotaniGard WP). The Þeld trials were carried out in Monkton by using the same experimental design as in 2002 but expanded to include two more Þelds. In addition to the nine screened cages placed in one of the Þelds, the experiment included three open Þeld plots of canola each one measuring 10 by 100 m. One of the plots was located in the same Þeld where the cages were placed, but the other two were each located in a different Þeld. The three Þelds were separated by 6 8 km. Two of the Þelds were planted with the variety DKL 30Ð55, whereas the third one was planted with the variety Ebony (Monsanto Canada). To assess B. bassiana dissemination, samples from the cages were collected on two dates as described above in the 2002 Þeld trials except that on the Þrst sampling date, three subsamples each consisting of eight L. lineolaris adults were collected from the treated cages instead of Þve subsamples of 10 adults. In addition, 10 canola leaves were collected from each treated cage in 2003 and processed as above to determine the CFUs of B. bassiana on them. Sampling from the open Þeld plots was carried out on the same dates as for the cages. Four hundred fully opened ßowers were collected randomly from each open Þeld plot. Flowers adjacent to the honey bee hive were not sampled. The collected ßowers were separated into subsamples of 20 and treated as above to determine the density of B. bassiana on them. L. lineolaris adults from the open Þeld plots were collected by 180 sweeps with an insect net of 38 cm diameter. Low population densities of L. lineolaris prevented the collection of sufþcient numbers of adults for assessment of Beauveria vectoring. No bees or leaves were collected from the open Þeld plots. Temperature and relative humidity were monitored inside the cages and in the open Þeld plots using shaded temperature/humidity sensors (Hycal Co., Elmonte, CA). The sensors were installed 50 cm above the ground in the middle of the plant canopy. Statistical Analysis. Field trials for the different years were analyzed separately. age mortality of L. lineolaris adults were arcsine square-root transformed in all trials. In the greenhouse trials, percentage mortality was analyzed as a randomized complete block design (PROC GLM; SAS Institute 1999), and means were compared by polynomial contrasts. age mortalities of 2002 and 2003 Þeld trials were

4 1572 ENVIRONMENTAL ENTOMOLOGY Vol. 35, no. 6 Fig. 2. age mortality ( SE) of L. lineolaris adults treated with three isolates of B. bassiana (ARSEF 3769 [ARK], NY [NY], and the commercial isolate [GHA]) delivered by honey bees to potted canola plants in greenhouse cage trials. analyzed using a mixed linear model in a split plot design with sampling dates as subplots and treatments as whole plots (PROC MIXED; SAS Institute 1999). Means were separated using polynomial contrasts by which the mortality in the Beauveria-treated cages were compared with the average mortality of both inactivated Beauveria and no Beauveria treatments. Another contrast was tested to compare mortality in the inactivated Beauveria with the no Beauveria treatment. Data on the estimated values of B. bassiana CFU were log(x 1) transformed and analyzed using repeated measurement procedure (PROC MIXED; SAS Institute 1999). For both CFU and mortality data and for all trials, means were compared only when significant differences were found by the F-test. The type I error rate (%) was set at 0.05 level for all tests. Mortality and CFU data were back-transformed to their original scales for presentation in tables and Þgures. Results Greenhouse Trials. All the Beauveria isolates resulted in L. lineolaris mortality that was signiþcantly greater than the control, but no signiþcant differences were found among the isolates (Fig. 2). The highest percentage mortality was achieved by the ARK isolate and ranged between 20 and 60%, with a mean of % (SE). age mortality achieved by the GHA isolate ranged between 20 and 50%, with a mean of %, whereas percentage mortality for the NY isolate ranged between 10 and 40%, with a mean of %. The higher mortalities of L. lineolaris collected from the infested cages were conþrmed by the percentage of mycosed cadavers of the total number of dead bugs. Mean percentages of mycosed bugs were , , and % for the ARK, GHA, and NY isolates, respectively. No mycosed cadavers were found among the bugs collected from the control cages. Field Trials All sampled bees on the two sampling dates contained B. bassiana spores (Table 1). The propagule density carried by the emerging bees ranged from to and to CFU/bee for the Þrst and the second sampling dates, respectively. Densities of CFUs per ßower were 0Ð and 0Ð for the Þrst and second sampling dates, respectively. The delivered inoculum was acquired by L. lineolaris feeding on the crop. Densities of CFUs per adult L. lineolaris were 0Ð for the Þrst sampling date and 0Ð for the second sampling date. No signiþcant differences in density of B. bassiana conidia on honey bees (F 1,8 0.13, P 0.71), canola ßowers (F 1,8 0.02, P 0.98), and L. lineolaris (F 1,8 0.04, P 0.95) were found between the Þrst and the second sampling dates (Table 1). Statistical analysis of the mortality of L. lineolaris showed signiþcant differences among the treatments (F 2, , P 0.01). However, effects of sampling dates (F 1, , P 0.8) and the interaction between sampling date and treatments (F 2,6 1.85, P 0.24) were not signiþcant (i.e., the treatment effects did not differ on the different sampling dates). Means testing of L. lineolaris mortality indicated that adults collected from canola blooms with bee-delivered living B. bassiana conidia had signiþcantly higher mortalities than did adults collected from heat-inactivated or non-b. bassiana canola blooms on the Þrst (F 1,4 88.5, P 0.01) and second sampling dates (F 1,4 58.4, P 0.01; Table 2). No signiþcant differences in mortality were found among L. lineolaris collected from cages supplied with bees and the inactivated B. bassiana and L. lineolaris collected from the non-b. bassiana treatments on the Þrst (F 1,4 0.24, P 0.65) and second sampling dates (F 1,4 0.69, P 0.45; Table 2). In the living B. bassiana treatment, most of the mortality occurred between days 3Ð6 after collection of the bugs. Table 1. Mean no. CFU/bee, flower, and L. lineolaris and mean percentage of honey bees, canola flowers, and L. lineolaris with detected densities of B. bassiana as collected from cages supplied with honey bees and B. bassiana in the 2002 field trials on two sampling dates Sample First sampling date Second sampling date Mean SE age SE Mean SE age SE Bees a a Flowers a a L. lineolaris a a Means within a row with the same letters are not signiþcantly different at 0.05 level using the F-test. N 10 for bees, ßowers, and L. lineolaris.

5 December 2006 AL MAZRAÕAWI ET AL.: BEE-VECTORED B. bassiana AGAINST TARNISHED PLANT BUG 1573 Table 2. Mortality and confirmed mycosis percentages of L. lineolaris on canola plants caged without honey bees, with honey bees and inactivated B. bassiana, and with honey bees and B. bassiana on two sampling dates in the 2002 field trials Treatments mortality SE First sampling date mycosed SE mortality SE Second sampling date mycosed SE L. lineolaris bees B. bassiana a a L. lineolaris bees heat-killed B. Bassiana b b 0 L. lineolaris only b b 0 Means within columns and within rows with different letters are signiþcantly different at 0.05 level using polynomial contrasts. N 3 for mortality and mycosis percentages. The higher mortalities in the living B. bassiana treatment were conþrmed by the number of mycosed insects. No mycosed insects were observed in the inactivated Beauveria and no Beauveria treatments on the Þrst and the second sampling dates (Table 2). Field Trials Similar results were obtained in the 2003 Þeld trials. All collected bees emerging from the hives that were equipped with the B. bassiana dispensers carried to and to CFU/bee for the Þrst and second sampling dates, respectively. Densities of inoculum detected on the ßowers ranged from 0 to and 0Ð CFU/ßower for the Þrst and second sampling dates, respectively. Densities of CFUs per leaf were 0Ð and 0Ð for the Þrst and second sampling dates, respectively. Mean incidences of L. lineolaris adults with detectable densities of B. bassiana ranged from 47 to 57% on the two sampling dates. Densities of CFU per L. lineolaris were 0Ð7,333 and 0Ð11,500 for the Þrst and second sampling dates, respectively. No signiþcant differences in the density of B. bassiana conidia on honey bees (F 1,8 5.23, P 0.08), canola ßowers (F 1,8 0.44, P 0.54), canola leaves (F 1,8 1.64, P 0.27), and L. lineolaris (F 1,8 4.2, P 0.11) were found between the Þrst and the second sampling dates (Table 3). Statistical analysis of the mortality data indicated signiþcant differences among the treatments (F 2,4 10.4, P 0.03). The interaction effects of sampling dates and treatments were not signiþcant (F 2,6 4.72, P 0.06), i.e., the treatment effect is similar on each sampling date. For the Þrst sampling date (Table 4), no signiþcant difference in mortality was found between the living B. bassiana treatment and the average of the inactivated B. bassiana and non-b. bassiana treatments at the 0.05 level (F 1,4 3.9 P 0.12). However, signiþcantly higher mortalities were observed (F 1,4 20.7, P 0.01) on the second sampling date when the mortality in the living B. bassiana treatment was compared with the average of the inactivated B. bassiana and non-b. bassiana treatments (Table 4). No significant differences were found between the inactivated B. bassiana treatment and non-b. bassiana treatment in both the Þrst sampling date (F 1,4 1.93, P 0.26) and second sampling date (F 1,4 3.6, P 0.13). Similar to the 2002 Þeld trials, the higher mortalities in the living B. bassiana treatment were conþrmed by the number of mycosed insects. A few mycosed insects were observed in the inactivated B. bassiana treatment in the second sampling date only. No mycosed L. lineolaris were observed in the non-b. bassiana treatment for both the Þrst and the second sampling dates (Table 4). In the open Þeld plots, low population densities of L. lineolaris prevented the collection of sufþcient samples to assess mortality. However, estimates of CFU/ groups of 20 ßowers collected from the open Þeld plots showed that honey bees vectored the fungus to canola ßowers. Means of CFU/20 ßowers ranged from 0 to (mean, 10,683 2,212) and 0 to (mean, 18,344 2,898) in the Þrst and second sampling dates, respectively. Frequency Distribution of Sampled Bees, Flowers, Leaves, and L. lineolaris. Statistical analysis for estimates of CFU in the 2002 and 2003 Þeld seasons indicated no signiþcant differences between years for honey bees (F 1,8 0.28, P 0.61) and ßowers (F 1,8 0.84, P 0.39) but not for L. lineolaris adults (F 1,8 10.4, P 0.01). Therefore, the data were pooled between the two Þeld seasons for honey bees and ßowers only. Table 3. Mean no. CFU/bee, flower, leaf, and L. lineolaris and mean percentage of honey bees, canola flowers, canola leaves, and L. lineolaris with detected densities of B. bassiana as collected from cages supplied with honey bees and B. bassiana in the 2003 field trials on two sampling dates Sample First sampling date Second sampling date Mean SE age SE Mean SE age SE Bees a a Flowers a a Leaves a a L. lineolaris a , a Means within a row with the same letters are not signiþcantly different at 0.05 level using the F-test. N 10 for bees, ßowers, and L. lineolaris.

6 1574 ENVIRONMENTAL ENTOMOLOGY Vol. 35, no. 6 Table 4. Mortality and confirmed mycosis percentages of L. lineolaris on canola plants caged without honey bees, with honey bees and inactivated B. bassiana, and with honey bees and B. bassiana on two sampling dates in the 2003 field trials Treatments mortality SE First sampling date mycosed SE mortality SE Second sampling date mycosed SE L. lineolaris bees B. bassiana a b L. lineolaris bees heat-killed B. Bassiana a a L. lineolaris only a a 0 Means within columns and within rows with different letters are signiþcantly different at 0.05 level using polynomial contrasts. N 3 for mortality and mycosis percentages. The frequency distribution of honey bees sampled as they left the hives equipped with B. bassiana dispensers in 2002 and 2003 Þeld trials is shown in Fig. 3. The highest frequency was observed with bees carrying to CFU/bee. Frequencies of bees carrying between to CFU/bee were 51%. Few bees carried CFU of inocula or within the ranges of to , and 14% of the sampled bees carried CFU/bee (Fig. 3). With regard to ßowers, B. bassiana was not detected on 28% of the sampled ßowers. The frequency of ßowers with and CFU/ßower was 39%. Few ßowers with inoculum densities of to were sampled from the cages (25Ð30%; Fig. 4). The frequency distribution of canola leaves with detectable amounts of bee vectored B. bassiana conidia in the 2003 Þeld trials showed that the highest frequency on leaves was observed with the range to CFU/leaf. B. bassiana was not detected on 25% of the sampled leaves. Few leaves with inocula within the ranges of to were sampled from the treated cages (Fig. 5). The frequency distribution of L. lineolaris collected from canola blooms infested with bee delivered B. bassiana conidia in 2002 showed that B. bassiana was not detected on 24% of the collected L. lineolaris. Moderately high frequencies occurred within the ranges 2,001Ð3,000, 3,001Ð4,000, and 5,001Ð6,000 CFU/L. lineolaris (Fig. 6). In the 2003 Þeld trials, the highest frequency of 44% occurred with L. lineolaris showing no detectable amounts of B. bassiana conidia. Frequencies within the ranges 1Ð1,000 and 1,001Ð2,000 were moderate. Low frequencies were observed for the rest of the ranges (Fig. 6). Discussion Results from this study indicate that honey bees effectively disseminated B. bassiana from the inoculum dispensers to canola and subsequently to L. lineolaris. In the greenhouse trials, L. lineolaris adults that were collected from plants placed in the cages with bee hives equipped with inoculum dispensers Þlled with B. bassiana conidia had higher mortalities compared with the control, regardless of the isolate. These results showed no signiþcant differences among the isolates, suggesting that virulence of the three isolates to L. lineolaris was similar. Thus, the commercial isolate GHA was chosen for the Þeld trial for practical reasons. In the 2002 and 2003 Þeld trials, all sampled bees acquired hundreds of thousands of B. bassiana conidia as they left their hives. The pooled mean CFU/ bee for the 2-yr Þeld study was equivalent to 0.5 mg inoculum/ bee. Yu and Sutton (1997) found that honey bees acquired CFU/bee, which was equivalent to 0.3 mg of the inoculum when they emerged from similar dispensers as vectors of Clonostachys (Gliocladium) rosea (roseum) Bainer (Hypomycetes) to raspberry ßowers. The bee-acquired conidia were deposited on the ßowers of the canola plants while honey bees foraged Fig. 3. Frequency distribution of sampled honey bees with detected densities of conidia of B. bassiana after emerging from hive-mounted dispensers Þlled with CFU of B. bassiana/g corn ßour in the 2002 and 2003 Þeld trials. Fig. 4. Frequency distribution of sampled canola ßowers with detected densities of conidia of B. bassiana from the cages containing honey bees and B. bassiana in the 2002 and 2003 Þeld trials.

7 December 2006 AL MAZRAÕAWI ET AL.: BEE-VECTORED B. bassiana AGAINST TARNISHED PLANT BUG 1575 Fig. 5. Frequency distribution of sampled canola leaves with detected densities of conidia of B. bassiana from the cages containing honey bees and B. bassiana in the 2003 Þeld trials. the ßowers for pollen and nectar. The pooled mean of CFU/ ßower for the 2-yr Þeld study was (Fig. 4). This indicates that the amount deposited on the ßowers constituted 6% of the amount of inocula acquired from the dispensers by the bees. A similar percentage was reported when honey bees were used as vectors of C. rosea to strawberry ßowers for the suppression of gray mold when 10% of the inocula were detected on the ßowers compared with the density detected on the honey bees (Peng et al. 1992). Conidia of B. bassiana were also detected on the leaves sampled in the 2003 Þeld trials. The pooled mean of CFU/leaf of the two sampling dates constituted 4.6% of the amount acquired by the bees (Fig. 5). The presence of inoculum on the leaves might be attributed to sedimentation of conidia from ßying bees and redistribution of conidia by rain, wind, and other organisms. This conclusion is also supported by the frequency distribution of the leaves with detectable amounts of conidia (Fig. 5). Moreover, observations of honey beeõs behavior after leaving the hives dusted with the inoculum showed that some honey bees ßew directly to the leaves. This would explain the larger densities of conidia found on some leaves. Fig. 6. Frequency distribution of sampled L. lineolaris adults with detected densities of conidia of B. bassiana from the cages containing honey bees and B. bassiana in the 2002 and 2003 Þeld trials. Conidia of B. bassiana were also detected on the target pest. This is conþrmed by the high percentage of L. lineolaris adults with detectable amounts of B. bassiana conidia in both sampling dates and in the 2-yr Þeld study (Tables 1 and 3). The pooled means of inoculum density per L. lineolaris for the 2002 and 2003 Þeld studies constituted 0.7 and 0.3% of the pooled mean density of conidia acquired by the bees, respectively. Those means also constituted 12 and 4% of pooled means of conidia detected on the ßowers in 2002 and 2003, respectively. Most of the measured L. lineolaris mortality occurred between days 3 and 6 after collection of the insects. It is difþcult to assess the time needed to kill the L. lineolaris in bee delivery systems because the exact time at which the collected insects came in contact with the fungus is not known, particularly on the second sampling date. L. lineolaris collected from cages not supplied with B. bassiana showed low mortality (9Ð22%) on the Þrst and second sampling dates for both 2002 and 2003 Þeld trials (Tables 2 and 4). This suggests that L. lineolaris adults were not stressed by the presence of the bees and that the carrier, Þne corn ßour, also had no adverse effect on L. lineolaris. Low population densities of L. lineolaris in 2003 Þeld trials hindered the efforts to evaluate their mortality in open Þeld plots. However, ßower samples from the open Þeld plots further conþrmed the efþciency of honey bees in vectoring the fungus to the crop. A large variability in conidial density was observed among canola ßowers (0 to CFU/ßower). This variability may be attributed to the foraging behavior of honey bees. Honey bees foraging on ßowering canola crops were found to visit one to two ßowers per cluster and move from one ßower cluster to another ßower cluster and from one plant to another without visiting all the ßowers in the same cluster or plant (Langridge and Goodman 1982, Mohr and Jay 1988). This behavior may explain the observation that not all the ßowers of the plant received B. bassiana inocula by foraging honey bees. Cresswell et al. (1995) studied the spatial dispersal of canola pollen by honey bees and bumble bees using particulate ßuorescent dust as a pollen analog. They found that most of the dust was deposited at the Þrst few ßowers visited by a bee and smaller amounts were deposited on subsequently visited ones. Similarly, a bee after emerging from the inoculum dispenser carrying B. bassiana conidia might deposit most of the conidia at the Þrst visited ßowers, whereas smaller amounts of conidia might be deposited on subsequent ßowers. Other factors that might have contributed to the variability of inocula on the ßowers include repeated visits to the same ßower by the bees and redistribution of the inocula by the bees, wind, rain, and movement of other organisms within and between the plants. Weather conditions such as temperature, light intensity, and wind during vectoring trials can also affect the efþciency of propagule dissemination. Average temperatures between 12 and 17 C and heavy rainfall were experienced during the foraging period before

8 1576 ENVIRONMENTAL ENTOMOLOGY Vol. 35, no. 6 collection of the samples on the Þrst sampling date in Those conditions might explain the lower mortality of L. lineolaris collected on the Þrst sampling date in 2003 Þeld trials (Table 4) and may explain the low propagule density detected on the L. lineolaris collected on the Þrst sampling date in 2003 (Table 3). When temperatures exceeded 20 C before sample collection during the second sampling date in 2003, the percentage of ßowers with conidia was higher and higher propagule densities were found on the L. lineolaris adults. Similarly, those conditions might explain the difference between the Þrst and second sampling dates in propagule density found on the ßowers collected from the open Þeld plots in These Þndings agree with earlier observations with C. rosea vectored by honey bees to strawberries and raspberries (Peng et al. 1992, Yu and Sutton 1997). Moreover, Abrol (1988) observed increased honey bee activity on B. campestris with increasing air temperature, which supports these conclusions. Although no signiþcant differences were detected between the two sampling dates for the 2-yr Þeld trials, there was a tendency for higher percentages of ßowers, leaves, and L. lineolaris adults with detectable amounts of the fungus in the second sampling date compared with the Þrst sampling date in 2002 and 2003 Þeld trials (Tables 1 and 3). This might be attributed to a carryover effect because the Þrst sampling date occurred 48 h after the dispensers were Þlled with the inoculum for the Þrst time, whereas the second sampling date occurred after 6 d of bee foraging with dispensers reþlled every 48 h. However, this tendency was not reßected in terms of higher mortality of L. lineolaris in the second sampling date, with the exception of the Þrst sampling date in 2003, which was explained previously. This may be explained by a possibility that some L. lineolaris adults were infected earlier in the vectoring process and killed by the fungus before collection on the second sampling date. Collectively, the data indicated that honey bees were efþcient vectors of B. bassiana to canola and that the vectored inocula infected the population of L. lineolaris. However, for this technology to be practical, B. bassiana should not be harmful to honey bees. Monitoring of the colonies during and after the trials showed no adverse effect on the bees. These observations agreed with previous reports that indicated the safety of B. bassiana to colonies of honey bees (Al mazraõawi 2004). Although Vandenberg (1990) reported reduced longevity of honey bees when exposed to B. bassiana in cages, this test might be misleading because the exposed workers were isolated from their queen, which rendered them more susceptible than usual to the entomopathogen (Goettel and Jaronski 1997). Exposure of whole hives to B. bassiana strain GHA resulted in 1% infection of the workers and no infection of the brood (Goettel and Jaronski 1997). Similar Þndings were reported by Alves et al. (1996), who showed that both B. bassiana and M. anisopliae caused high mortality to caged Africanized honey bees maintained at high temperatures, but they did not cause epizootic levels in Þeld hives even when applied at high levels. He concluded that B. bassiana and M. anisopliae can be used in microbial control programs without any risks to Africanized honey bee colonies (Alves et al. 1996). Previous studies have compared the efþcacy of the pollinator vector technology to chemical sprays (Peng et al. 1992). However, economical consideration for the application of the pollinator vector technology should take into consideration the reduction in insect populations and the improved pollination of the crop. Further studies are warranted for evaluating the impact of applying B. bassiana using the pollinator vector technology on canola yield compared with spraying applications. Many researchers have concluded that insect pollination of canola is necessary for high seed germination rate (Kevan and Eisikowitch 1990), high seed set, and yields (Langridge and Goodman 1975, Westcott and Nelson 2001). Honey bees are the preferred pollinators of canola (Free 1993). The management of Lygus bugs on canola currently relies on the application of broad spectrum chemicals. For example, an outbreak of Lygus bugs in Alberta in 1998 required the spraying of 1,400,000 ha with chemicals (Government of Alberta 2002). Such applications could harm honey bees and other pollinators foraging the crop during ßowering. Canola ßowers secrete large amounts of nectar and are very attractive to many pollinators such as native solitary bees belonging to the Andrenidae, Halictidae, and Megachilidae families (Westcott and Nelson 2001). The pollinator vector technology not only reduces pesticide applications but also improves pollination resulting in increased yield and quality of the crop. Acknowledgments We thank P. Kelly, G. Wilson, and M. Adjaloo for technical assistance. Emerald BioAgriculture, Salt Lake City, UT, provided B. bassiana and Monsanto Canada, Listowel, Ontario, Canada, provided the canola Þelds. Financial support for this study was provided by Al balqaõ Applied University, Assalt, Jordan, the Natural Sciences and Engieering Research Council, the Biocontrol Network, Canada, and the University of Guelph. References Cited Abrol, D. P Environmental factors inßuencing pollination activity of Apis mellifera on Brassica campestris. J. Ind. Inst. Sci. 68: 49Ð52. Al mazra awi, M. S Biological control of tarnished plant bug and western ßower thrips by Beauveria bassiana vectored by bee pollinators. PhD dissertation, University of Guelph, Ontario, Canada. Al mazra awi, M. S., J. L. Shipp, A. B. Broadbent, and P. G. Kevan Biological control of Lygus lineolaris (Hemiptera: Miridae) and Frankliniella occidentalis (Thysanoptera: Thripidae) by Bombus impatiens (Hymenoptera: Apidae) vectored Beauveria bassiana in greenhouse sweet pepper. Biol. Control 37: 89Ð97. Alves, S. B., L. C. Marchini, R. M. Pereira, and L. L. Baumgratz Effects of some insect pathogens on

9 December 2006 AL MAZRAÕAWI ET AL.: BEE-VECTORED B. bassiana AGAINST TARNISHED PLANT BUG 1577 the Africanized honey bees Apis mellifera L. (Hym., Apidae). J. Appl. Entomol. 120: 559Ð564. Beilharz, V. C., D. G. Parbery, and H. J. Swart Dodine: a selective agent for certain soil fungi. Trans. Br. Mycol. Soc. 79: 507Ð511. Bidochka, M. J., G. S. Miranpuri, and G. G. Khachatourians Pathogenicity of Beauveria bassiana (Balsamo) Vuillemin toward Lygus bug (Hem., Miridae). J. Appl. Entomol. 115: 313Ð317. Broadbent, A. B., P. G. Mason, S. Lachance, J. W. Whistlecraft, J. J. Soroka, and U. Kuhlmann Lygus spp., plant bugs (Hemiptera: Miridae), pp. 152Ð159. In P. G. Mason and J. T. Huber (eds.), Biological control programmes in Canada, 1981Ð2000. CABI Publishing, New York. Butts, R. A., and R. J. Lamb Injury to oilseed rape caused by mirid bugs (Lygus) (Heteroptera: Miridae) and its effect on seed production. Ann. Appl. Biol. 177: 253Ð266. Butts, R. A., and R. J. Lamb Pest status of Lygus bugs (Hemiptera: Miridae) in oilseed Brassica crop. J. Econ. Entomol. 84: 1591Ð1596. Cresswell, J. E., A. P. Bassom, S. A. Bell, S. J. Collens, and T. B. Kelly Predicted pollen dispersal by honeybees and three species of bumble-bees foraging on oilseed rape: a comparison of three models. Funct. Ecol. 9: 829Ð 841. Free, J. B Insect pollination of crops, 2nd ed. Academic, London, UK. Goettel, M. S., and S. T. Jaronski Safety and registration of microbial agents for control of grasshoppers and locusts. Mem. Entomol. Soc. Can. 171: 83Ð99. Government of Alberta Lygus bugs in canola ( all/agdex741). Kevan, P. G., and D. Eisikowitch The effect of insect pollination on canola (Brassica napus L. cv. OAC Triton) seed germination. Euphytica 45: 39Ð41. Kevan, P. G., M. S. Al-mazra awi, J. C. Sutton., L. Tam, G. Boland, A. B. Broadbent, S. V. Thomson, and G. J. Brewer Using pollinators to deliver biological control agents to crops, pp. 148Ð153. In R. A. Downer, J. C. Mueninghoff, and G. C. Volgas (eds.), Pesticide formulations and delivery systems: meeting the challenges of the current crop protection industry. American Society for Testing and Material, West Conshohocken, PA. Langridge, D. F., and R. D. Goodman A study on pollination of oilseed rape (Brassica campestris). Aust. J. Exp. Agric. 15: 285Ð288. Langridge, D. F., and R. D. Goodman Honey bee pollinators of oilseed rape, cultivar Midas. Aust. J. Exp. Agric. 22: 124Ð126. Liu, H., M. Skinner, B. L. Parker, and M. Brownbridge Pathogenicity of Beauveria bassiana, Metarhizium anisopliae (Deuteromycotina: Hyphomycetes), and other entomopathogenic fungi against Lygus lineolaris (Hemiptera: Miridae). J. Econ. Entomol. 95: 675Ð681. Mohr, N. A., and S. C. Jay Nectar and pollen collecting behavior of honey bees on canola (Brassica campestris L. and Brassica napus L.). J. Apic. Res. 27: 131Ð136. Noma, T., and K. Strickler Factors affecting Beauveria bassiana for control of Lygus bug (Hemiptera: Miridae) in alfalfa seed Þelds. J. Agric. Entomol. 16: 215Ð233. Peng, G., J. C. Sutton, and P. G. Kevan Effectiveness of honey bees for applying the biocontrol agent Gliocladium roseum to strawberry ßowers to suppress Botrytyis cinerea. Can. J. Plant Pathol. 14: 117Ð129. SAS Institute SAS/STAT userõs guide, version 8. SAS Institute, Cary, NC. Snodgrass, G. L Insect resistance in Þeld population of tarnished plant bug (Heteroptera: Miridae) in cotton in the Mississippi Delta. J. Econ. Entomol. 89: 783Ð790. Steinkraus, D. C., and N. P. Tugwell Beauveria bassiana (Deuteromycotina: Moniliales) effects on Lygus lineolaris (Hemiptera: Miridae). J. Econ. Entomol. 32: 79Ð90. Turnock, W. J., G. H. Gerber, B. H. Timlick, and R. J. Lamb Losses of canola seeds from feeding by Lygus species [Heteroptera: Miridae] in Manitoba. Can. J. Plant Sci. 75: 731Ð736. Vandenberg, J. D Safety of four entomopathogens for caged adult honey bees (Hymenoptera: Apidae). J. Econ. Entomol. 83: 755Ð759. Westcott, L., and D. Nelson Canola pollination: an update. Bee World 82: Yu, H., and J. C. Sutton Effectiveness of bumble bees and honey bees for delivering inoculum of Giocladium roseum to raspberry ßowers to control Botrytis cinerea. Biol. Control 10: 113Ð122. Received for publication 17 February 2006; accepted 11 September 2006.