The upper and lower parts of the Plexiglas sandwich were solid and between the lower base and test arena a piece a piece of filter paper was placed to allow moisture exchange and to extend the life of the leaf during the bio-assay. Airflow through the test arena was provided through two holes drilled through the center cell layer directly opposite one another, with fine-mesh screening melted onto the interior of the test arena to prevent escape. The Plexiglas sandwich was held together with four binder clips positioned such that the airflow was not covered. Once dry, the leaves were placed on the filter paper in Munger cells and the respective thrips species was added. The lid was placed on the cell but leaving the cell arena exposed, so that once the thrips were added, the cells could be closed quickly. Female and late second-instar avocado thrips and citrusthrips were then placed on treated leaves of their respective host plants inside the Munger cell. Control leaves for both species were treated with a mixture of the same suspension ingredients minus the protein. Bioassays were conducted concurrently in the following manner for both species: adult female thrips were placed on leaves coated with activated or inactivated forms of both Cyt1A and Cry11A, immature thrips were also placed on leaves coated with activated or inactivated forms of both Cyt1A and Cry11A, and all combinations for adults and immature thrips were carried out along with the corresponding control cells. The Munger cells were closed and placed in an environmental chamber at 28ºC, 55% RH,raspberry container and long daylight conditions . Each cell was carefully removed daily and the filter paper doused with water to prevent leaf desiccation. The bio-assay was replicated on two separate dates .
A minimum of 10 individuals was placed into each Munger cell and thrips were checked daily for eight days to assess mortality. Post seven days, the integrity of the leaves was questionable and in all but one bio-assay, mortality was observed before seven days; thus data were analyzed using day 7 mortality. Six strains of B. bassiana were obtained from the USDA-ARS Western Integrated Cropping Systems Research Unit located in Shafter, CA. GHA is the commercially available strain found in the field formulation of B. bassiana, Mycotrol O and the greenhouse formulation BotaniGard ES, and each of the other five strains were obtained via isolation from soils in Kern County by USDA-ARS collaborators in 2000. They were stored at – 80ºC. Culture methods for the thrips experiments were similar to those described previously for Lygus hesperus Knight bio-assays and were conducted by collaborators from USDA-ARS, Shafter, CA. Briefly, isolates were grown on SDAY media, or Sabouraud’s dextrose agar plus yeast extract . The conidia were harvested from culture plates after 10–14 days incubation by scraping with a sterile rubber policeman into a 0.01% solution of Silwet L- 77 . The conidia were then enumerated with a hemocytometer. For preservation and storage, glycerol was added to the conidial suspension and stored in aliquots of 2 × 108 in a 2 ml solution at −80°C until needed for bio-assays. Conidial viability was assessed following incubation for 16 h in potato dextrose broth just prior to use in experiments. Viability was determined by adding a sample of approximately 107 conidia to 20 ml potato dextrose broth and incubating ca. 16 h in a rotary shaker at 28°C.
Conidia germination was examined under a compound microscope at 400× and scored as viable if the germ tube was at least twice the length of the conidium. Percentage viability was measured on 250 conidia of each isolate. All bio-assays were conducted on the basis of the number of viable conidia measured after thawing and the desired concentrations were formulated by serial dilution. The strain from Mycotrol was isolated and cultured exactly as above to eliminate possible effects of production methods and formulation ingredients on insecticidal activity. Glycerol was not removed prior to using the conidia in bio-assays. All six B. bassiana strains were suspended in 0.01% Silwet in a de-ionized water solution and evaluated on the same date at four concentrations for each thrips species. The control consisted of 0.01% Silwet in de-ionized water solution. Each of the 25 treatments was evaluated using five Munger cells , which contained a minimum of ten adult female thrips. These bio-assays were repeated on 10 dates with both species tested simultaneously on each date . Groups of thrips were anesthetized by exposure to CO2 for 15-30 sec, and each strain was administered to the dorsum of the abdomen of each knocked out thrips quickly and carefully in a 1µl drop with a Burkard Hand Microapplicator over filter paper. The droplet spread the length of the thrips immediately and the thrips was then deposited, still knocked out, onto the leaf tissue in the Munger cell. Once a minimum of 10 treated thrips were added, Munger cells were closed and sealed with binder clips and placed in an environmental chamber at 28ºC, 55% RH, and long daylight conditions . Each cell was checked daily for seven days to observe infection by the fungus. Each cell was carefully removed daily and the filter paper doused with water to prevent leaf desiccation. Individuals infected with B. bassiana were defined as those whose natural activity was retarded and/or showed arrestment and subsequently produced mycelia, which was confirmed post bio-assay.
Mortality caused by mycosis was confirmed on the basis of visual observation and then crushing individuals to reveal the presence of mycelial growth. When mycelial growth was not apparent, crushed individual thrips were placed on potato-dextrose agar plates for 5 days and then re-examined for the presence of mycelial growth. Data were analyzed after Abbott’s correction for control mortality using log-probit analysis with PROC PROBIT on SAS 9.2 and using the Raymond Statistics package . The purpose of the probit analysis was strictly for gross strain comparison. Probit analysis was used to estimate the LC50 and LC95 levels, confidence intervals, and χ 2 values for each strains. Lethal concentrations with overlapping 95% confidence intervals were not considered significantly different. The daily check data were analyzed as non-cumulative counts per day via the Survival Distribution Function on SAS 9.2 , where observation time represented the probability that the experimental unit from the population would have a lifetime exceeding that time with the variables strain and concentration. Assessments for each variable by species were done with Log-rank and Wilcoxon tests and multiple comparisons for the log-rank test were adjusted by using Tukey-Kramer method. Data were then plotted as estimates of the survivor function for the different strains separately for each species. Bacillus thuringiensis israelensis produces two groups of toxic proteins, the Cry and Cyt toxins that have different modes of action. In this investigation, results with Cyt1A and Cry11A were disappointing as both activated and inactivated forms of both proteins showed little effect against adult and second instar citrus thrips and avocado thrips. To our knowledge, there have been no reports of Bt endotoxins with activity against Thysanoptera, although Cyt1Aa was found to be toxic to the non-target species Chrysomela scripta Fabricius . Many hypothesize that because thrips feed with a punch and suck method, rather than direct chewing and mastication of leaf tissues, they do not receive toxic amounts of the Bt proteins . Alternatively, they may not possess the proper binding receptors for the Bt proteins tested to date and thus, no pore can be formed in the midgut lining and the Bt proteins are excreted . The literature indicates the latter hypothesis is more likely based on findings from life table parameters where development, fecundity,growing raspberries in container and adult longevity or relative abundance are not significantly different from thrips reared on Bt positive versus Bt negative corn, cotton, or potato plants. The aforementioned studies were not specifically looking at Bt effects on thrips nor were the Bti toxins tested here involved in previous studies involving thrips. The combinations of proteins used in this study were, to date, unique pairings with thrips. It is indeed possible that there are no Bt endotoxins currently available that cause mortality to Thysanoptera. The LC50 with strain GHA was 8.61 x 104 conidia/ ml and was two orders of magnitude lower than for the other five B. bassiana strains tested . GHA also gave the only statistically valid dose-response values in probit analysis, and provided the only data that fit the probit model. The other B. bassiana strains failed to provide a linear relationship based on their p-values , i.e. the probit regression lines were of poor quality, except for GHA. Therefore, data were evaluated based on line slopes as is commonly seen in the scientific literature with other biological agents where data lines are not straight and do not fit the model . Strains 1741ss, SFBb1, S44ss, and NI1ss showed a flat dose-response between concentrations, did not fit the model, and LC50’sranged from 2.7 x 106 – 9.6 x 108 . Assessment of Beauveria strain while adjusting for concentration, in both Logrank and Wilcoxon tests showed that strain and concentration had a highly significant effect on the infection rate. Multiple comparisons for the Logrank test to assess the strain effect while adjusting for the concentration differences showed that strains 1741ss, S44ss, 3769ss, and NI1ss infection rates were not distinct from one another.
Strain GHA and SFBb1 had infection rates different from each other as well, and GHA had the fastest infection rate and SFBb1 showed the slowest kill rate . The Survival Distribution Function analysis coupled with the probit analysis clearly shows that GHA would be the best strain choice for citrus thrips control. Results with avocado thrips. The LC50 for strain GHA was 2.2 x 106 conidia / ml and was similar to that obtained with the other five B. bassiana strains tested . Again, because a strong linear response was not observed, the performance between strains was rated based upon the LC50 and relative linearity of the response. Based on overlap of confidence intervals, there were no significant differences between any of the strain LC50’s or LC95’s . Assessment of Beauveria strains while adjusting for the concentration, using both Log-rank and Wilcoxon analysis showed that strain did not have an effect on the infection rate. The multiple comparisons for the Log-rank test to assess the strain effect while adjusting for the concentration differences showed infection rates for all 5 strains were not distinct from one another . The Survival Distribution Function analysis coupled with probit analysis indicated there was no one best strain to select for avocado thrips management. Citrus thrips were more susceptible to Beauveria than avocado thrips; citrus thrips LC values were much lower for the most active strain, GHA, indicating that significantly lower dosages of strain GHA were required to infect and kill citrus thrips compared with avocado thrips. The overall survival analysis results showed a similar pattern to the results of the probit analysis; GHA had the fastest infection rate and SFBb1 had the slowest rate . Infection rates for the other three strain’s fit in between the rates for GHA and SFBb1, and 1741ss, S44ss, 3769ss, and NI1ss infection rates were not separable. This low dosage association and having the fastest infection rate suggest GHA is the best candidate for field-testing among the strains examined. Except for the worst performing strain, SFBb1, the performance of all of the strains with avocado thrips were similar. The LC50 value for citrus thrips was 8.6 x 104 conidia/ml, which may suggest economical feasibility in some cases, e.g., for use on organic products. The maximum recommended field application rate is 5.0 x 1012 conidia/ha. Therefore, 8.6 x 1011 conidia/ha of GHA is needed based on the estimated LC50 of 86 conidia/µl and this amount is reasonable to obtain in a field setting. Conducting the same analysis for avocado thrips control using GHA, with an LC50 of 2.2 x 106 , 2.2 x 1013 conidia/ha would be required. This is 4.4 times greater than the standard field use rate of GHA. We hypothesize that differences in susceptibility between citrus and avocado thrips may be due to the different habitats in which they evolved.