Generalized linear models were used to analyze the data with the program R . Binomial errors and a chi-square test, or quasibinomial errors and an F test in cases where there was over dispersion, were used to analyze the effects of pesticide on the stage-specific survivor ship of eggs, larvae, pupae, and adults , the fertility of eggs, and gender ratio of emerging adults from the pupal bio-assay. Gaussian errors were used to assess the effect of pesticide on larval and pupal development time, and a negative binomial error distribution was used to analyze fecundity, with daily egg production for each female nested as a random effect to account for the repeated daily measurements from the same individuals. For all models, log likelihood ratio tests were used to assess the statistical significance of pesticide at α = 0.05. From these models, the pesticides with similar effects on stage survivor ship were grouped, and their toxicity was characterized according to the criteria developed by the International Organisation for Biological Control for classifying pesticide selectivity on beneficial insects in a laboratory bio-assay: < 30% impact is harmless, 30-79% impact is slightly harmful, 80-99% impact is moderately harmful, and an impact > 99% is harmful . Based on corrected adult female mortality of H. convergens alone, chlorantraniliprole would be categorized as slightly harmful and lambda-cyhalothrin as harmful,blueberry grow pot while all other pesticides would be categorized as harmless .

By adding sublethal effects on fecundity and fertility, the reduction coefficients generated some notably different results in comparison to corrected adult mortality. Novaluron was categorized as moderately harmful rather than harmless due primarily to the strong sublethal effects on fertility. Cyantraniliprole and chlorantraniliprole were categorized as slightly harmful rather than harmless due to a combination of effects on adult survivor ship and daily fecundity. Copper+mancozeb also had a much larger negative reduction coefficient due to the fact that treated adults had higher daily fecundity as well as survivor ship than controls. The intrinsic rate of increase was negative for lambda-cyhalothrin, suggesting that populations exposed to this insecticide would decline to local extinction. It was also reduced for chlorantraniliprole, cyantraniliprole, novaluron, and sulfur and estimated to cause significant delays in the recovery of H. convergens populations following exposure, ranging from 5 d to over a year. The estimates of the intrinsic rate of increase and delay in population growth showed a similar pattern of effects from pesticide exposure as seen for the reduction coefficients. The main difference observed was that although sulfur produced a negative reduction coefficient, it produced a reduced rate of increase and consequently a delay in population growth due to low larval survivor ship. In addition, exposure to copper+mancozeb and spinetoram lead to improved performance for a number of the individual life history parameters measured for H. convergens, and this was reflected in all four of the endpoint measurements . The effects of pesticide exposure were highly variable among each of the four life stages of H. convergens. Adults and larvae had lower rates of survivor ship than the pupal and egg life stages, and larvae were more affected than adults.

This is most likely due to the fact that as non-mobile and non-feeding life stages, eggs and pupae were only treated topically, whereas adults and larvae were treated via all three routes of exposure. Adult survivor ship was reduced by just two of the insecticides tested; lambda-cyhalothrin showed acute toxicity and exposure to chlorantraniliprole reduced survivor ship by 45%. This effect of chlorantraniliprole was unexpected given that numerous studies have demonstrated it to have little impact on beneficial insects through a variety of exposure routes . For example, it was harmless via oral, residual, and topical exposure to seven species of parasitoid wasps, including the two indicator species Aphidius rhopalosiphiand Trichogramma dendrolimi. Moreover, the weed control agent, Chrysochus auratus Fabricius , was not affected by chlorantraniliprole via topical exposure or ingestion of treated foliage . Contrary to this trend, however, both chlorantraniliprole and cyantraniliprole were toxic to immature and adult C. carnea and C. johnsoni . In contrast to adult survivor ship, survivor ship of H. convergens larvae was greatly reduced by all of the pesticides tested, with the exception of copper+mancozeb. Copper+mancozeb has also been demonstrated to be harmless for nymphs and adults of Deraeocoris brevis, and copper hydroxide alone was harmless for the predatory mite, Euseius victoriensis. Egg and pupal stages of H. convergens were more resilient to topical pesticide exposure for most of the pesticides tested. Similar observations of greater resilience of egg and pupal stages to pesticide exposure have been demonstrated for C. carnea , and for eggs of Stethorus punctum picipes Casey and Geocoris pallens Stål and pupae of Harmonia axyridis. However, survivor ship of both eggs and pupae of H. convergens were significantly reduced when exposed to lambda-cyhalothrin, and egg survivor ship was reduced by exposure to novaluron.

Although 82% of H. convergens pupae treated with novaluron were able to emerge as adults, nearly one third of the emerged adults could not be identified as male or female. Moreover, even where gender could be identified, the elytra were noticeably malformed compared to the control group. Thus, it appears that novaluron, an insect growth regulator, affected the success of development during pupation even though it did not prevent adult emergence. Similarly, few larvae hatched successfully from eggs of H. convergens that were treated with novaluron and lambda-cyhalothrin, and it is likely that the hatched larvae would not have survived if they had been monitored through the larval stage. Such larvae were frequently unable to crawl and/or could not right themselves, conditions that have been categorized as moribund and excluded from survivor ship in some other studies . For instance, when eggs of the predator Podisus maculiventriswere treated with novaluron, there was no significant effect on emergence, but the larvae that emerged from treated eggs were unable to molt . In addition, although egg hatch of H. convergens was only slightly reduced from controls following exposure to cyantraniliprole, the hatching larvae were often lethargic and may not have developed successfully to the pupal stage. As cyantraniliprole has a greater impact from oral exposure it is possible that larvae experienced a brief oral exposure when hatching from treated eggs. These observations highlight the importance of following the potential for carry forward effects of pesticide exposure on survivor ship between life stages of longer lived natural enemies. In view of these observations, egg hatch and adult emergence may not be the best measures of the effects of pesticide exposure on egg and pupal survivor ship because they do not take longer term viability into consideration. For instance, when eggs of Chrysoperla externawere treated with several conventional and biorational pesticides, egg hatch was not affected, but larval survivor ship declined significantly when allowed to develop for 48 h after hatch . Similarly, when Aphidius ervi Haliday mummies were exposed to deltamethrin there was no significant effect on adult emergence or mortality 48 h after emergence, but longevity was significantly reduced . In addition, the effects of pesticide exposure on egg hatch is frequently measured as a sublethal effect of treated adult females , but as for acute bio-assays, hatch rates may not reflect the subsequent viability of larvae. For example,square plastic pot hatch rates for eggs laid by G. occidentalis females exposed to copper+mancozeb were normal, but the larvae that did hatch experienced significantly greater mortality . Thus, endpoint measurements for the effects of pesticide exposure on survivor ship of non-mobile life stages in laboratory bio-assays could be improved by monitoring juveniles a minimum of 48 h after egg hatch and monitoring adults for several days after emergence from pupae. Effects of pesticide exposure on development time and gender ratio of H. convergens were minimal, but there were significant effects on fecundity and fertility. Daily fecundity was not affected by pesticide exposure with the exception of copper+mancozeb, which produced a 57% increase in daily fecundity compared to the control group, but had no effect on fertility. This observation for H. convergens differs from that for lacewings, which when exposed to copper+mancozeb, had reduced daily fecundity and fertility by 24% and 18.9%, respectively . However, pesticide-induced hormesis is not uncommon among arthropods , and while they have been observed in laboratorystudies with natural enemies, including the coccinellid Coleomegilla maculata de Geer , it has been argued that beneficial effects of pesticides on natural enemies are less likely to occur under field conditions . From a similar laboratory study, when treated with diazinon, Daphnia pulexexperienced a 41% increase in fertility compared to the control group .

In contrast, novaluron caused substantial reductions in fertility of H. convergens and thus has both lethal and sublethal effects on egg viability, via treated eggs or treated adult females, respectively. Given that novaluron inhibits cuticle formation, its effect on egg viability is to be expected. For instance, novaluron exposure also reduced fertility in other natural enemies, such as C. carnea, C. johnsoni, and D. brevis . For most of the pesticides tested, corrected acute adult mortality and reduction coefficients, that also included sublethal effects on reproduction, gave a similar ranking of pesticide effects on H. convergens. The greatest discrepancy was for novaluron, which changed from harmless for corrected acute adult mortality to moderately harmful for the reduction coefficient due to low fertility rates. Similarly, in testing the effects of six pesticides on C. carnea using laboratory bio-assays, both acute mortality and reductions coefficients that also incorporated reproductive effects yielded very similar results as endpoint measurements . However, spinosad and spinetoram were considered harmless or slightly harmful from acute mortality effects on Orius armatusunder semi-field conditions, but when reproductive effects were incorporated, the reduction coefficients were classified as moderately harmful . The importance of including reproductive effects into reduction coefficients for a particular pesticide may also vary between natural enemy species. For example, the reduction coefficient for H. convergens exposed to chlorantraniliprole was 65%, but was only 23% for Orius laevigatus. Moreover, the reduction coefficients for H. convergens from exposure to sulfur and spinetoram were negative, but for O. laevigatus were nearly 70% for sulfur and over 80% for spinosyns. When life table data was used to estimate the intrinsic rate of increase and expected delay in population growth of H. convergens following exposure to the pesticides, the results were largely consistent with the reduction coefficients, with the exception of sulfur. Sulfur generated improved performance for corrected adult mortality and the reduction coefficient, but caused a slight delay in population growth, due to the inclusion of the low survivor ship of treated larvae. Similarly, for the parasitoid B. nigricans, while the reduction coefficient for adults exposed to abamectin was classified as slightly harmful, the intrinsic rate of increase indicated that the population would become extinct . Under field conditions, it is possible for active life stages of natural enemies to be exposed to pesticides through topical exposure from direct spraying, residual exposure via treated surfaces, and/or oral exposure from contaminated prey, water droplets, or honey dew . Measuring toxicity from residual exposure is a common approach for testing the selectivity of pesticides for natural enemies, based on applications of pesticide residues to a variety of substrates from glass slides to plastic petri-dishes and plant foliage . While most studies have been based on fresh dried residues of pesticides, an equally important, but less commonly studied aspect of residues as a route of exposure for natural enemies is their persistence in the field . For example, Anagrus nilaparvataehad low mortality when exposed to 7 d aged residue of methamidophos, chlorpyrifos, or avermectin, whereas imidacloprid had a much longer residual toxicity, causing 80% mortality from 7 d aged residue . On the other hand, deltamethrin residues aged ranging from 1-14 d had no effect on the mortality of the parasitoid Diaeretiella rapae, suggesting this pesticide may be compatible with biological control efforts . Similarly, abamectin was harmless to Eretmocerus eremicus Rose and Zolnerowich regardless of residue age, but metaflumizon was slightly harmful for aged residues of 1-8 d and was not harmless until 15 d . Increasingly, pesticides that have a reduced risk to nontarget organisms are being developed and used.