Risk determination for non-target arthropods from outdoor space applications of mosquito adulticides Authors: Robert K D Peterson, Taylor E Kennedy, Jackson R Strand Published: 2026 ABSTRACT AbstractOutdoor space applications of ultra-low-volume (ULV) insecticides are widely used for adult mosquito management, but public concerns persist about their potential effects on non-target arthropods. Therefore, we conducted a systematic review and analysis to evaluate evidence from global peer-reviewed operational studies, field studies simulating operational studies, and modeled ecological risk assessments (ERA). A total of 38 articles met inclusion criteria, representing 43 distinct studies across multiple arthropod taxa and 10 insecticidal active ingredients. Study types comprised 25 field experiments, nine operational applications, and nine ERAs. We recorded the presence or absence of acute mortality effects on non-target arthropods and analyzed associations using Fisher’s exact test and Firth logistic regression. Operational studies—representing large-scale, real-world mosquito control applications—had no deleterious effects across all taxa and insecticides. In contrast, field studies documented effects in 15 of 25 cases, although results were equivocal, with 10 studies finding no effects despite conditions favoring direct exposure (eg caged insects placed in the direct aerosol path). The ERAs showed mixed outcomes, most likely reflecting their intentionally conservative design. Study type was strongly associated with effect detection (P < 0.001), with operational applications significantly less likely to report effects. Overall, evidence suggests that risks to non-target arthropod populations from operational ULV mosquito adulticide applications are low, primarily due to reduced exposure from nighttime treatments, aerosol droplet characteristics (size and volume), rapid dispersal, and rapid breakdown. Increased emphasis on operational-scale research is needed to strengthen the understanding of risks for non-target arthropods. Introduction Mosquito management by mosquito control districts within municipalities, counties, states, and other local government entities often involves using outdoor space treatments of ultra-low-volume (ULV) insecticides applied at night to reduce populations of adult mosquitoes. These treatments typically only occur when nuisance or pathogen surveillance thresholds are exceeded (AMCA 2021). Whether applied by truck or aircraft, the ULV insecticide forms an aerosol of 8- to 60-µm droplets that disperse through the air and impinge on adult flying mosquitoes (Lofgren et al. 1973, Schleier et al. 2012). Despite decades of use with documented benefits (Carney et al. 2008, Breidenbaugh et al. 2009, Macedo et al. 2010, Peterson 2010, Haley 2012, Stoddard 2018, Holcomb et al. 2021), the public is often concerned about the human health and ecological risks posed by outdoor space applications of ULV insecticides for mosquito management (Thier 2001, Roche 2002, Mendoza et al. 2023, 2025, Tai et al. 2025). More recently, concerns have been raised about the risks to beneficial insects like pollinators and other non-target arthropods (Rochlin et al. 2022, Barascou et al. 2025, Crowder et al. 2025). Because many insecticides are highly toxic to non-target arthropods, using toxicity to make conclusions about risk is of little value. Similarly, using the environmental presence of the insecticide as a surrogate for risk is also of limited value because it is only one aspect of risk. To understand and make decisions about risk to non-target arthropods, we must use the common definition of risk as the joint probability of toxicity and exposure (Kaplan et al. 1981, NRC 1983). This relationship is often referred to more mathematically as a function of toxicity multiplied by exposure (NRC 1983, 1996), and explains why hazard alone is an insufficient metric for environmental risk (Peterson 2025). Therefore, a valuable study is to examine the current knowledge of outdoor space applications of ULV adult mosquito insecticides and risk to non-target arthropods. The insecticides currently used to manage flying adult mosquitoes are broad-spectrum and highly toxic to most insects. In addition, these insecticides are applied over large areas and designed to stay aloft as an aerosol mist so that the insecticide droplets impinge on and kill flying adult mosquitoes. Consequently, in this study, we examine the primary scientific literature to answer several questions about the effects of outdoor space applications of ULV adult mosquito insecticides on non-target arthropods. Materials and Methods We adhered to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) process (Page et al. 2021) to identify, screen, and include journal articles in our analysis (Fig. 1). We conducted a global search of the Web of Science, PubMed, and Zoological Record for primary scientific literature for the key terms: adulticide, mosquitocide, insecticide, non-target insect, and non-target arthropod. A total of 19,783 records were identified. We removed 13,322 duplicate records, leaving 6,461 records to screen for eligibility. We then excluded records if they fit the following criteria: non-mosquito insecticide, mosquito larvicide, and non-outdoor space applications of insecticides for adult mosquitoes. This screening produced 40 articles for analysis. The papers were further screened for including necessary information such as application method, application timing, insecticide active ingredient, and application rates. Only two articles were identified as not having enough information, leaving 38 articles included in our review. PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) flow diagram of the review performed in this study. We examined the journal articles and tallied them for the following: study type (field, operational, ecological risk assessment [ERA]), application method (aerial or ground), taxa, insecticide active ingredient, and non-target taxa (Supplementary Table S1). Operational studies included those conducted by mosquito control districts as part of routine large-scale applications across numerous hectares. Field studies typically involved experiments conducted along a linear path of tens to a few hundred meters that simulate operational studies. In most cases, caged insects were used along the linear path in direct contact of the insecticide aerosol and caged insects placed upwind of the aerosol were used as the controls (non-treatments). The effects typically measured were short-term reductions (several hours to several days) in the numbers of collected non-target arthropods compared to arthropod numbers pre-treatment or in control cages. Ecological risk assessments either modeled potential insecticide exposures to arthropods based on application rate or used actual exposures to estimate risks. Because the study methods and effect sizes differed greatly across articles, we assessed if there were effects on taxa based on multiple lines of evidence. We did so because risk to non-target arthropods is much more nuanced than simple independent assessments of statistically significant reductions in individual taxa, especially in operational-scale studies. Consequently, for each study, we employed a weight of evidence approach for each study to assess if there were effects. For example, Davis and Peterson (2008) concluded, “no overall treatment effects were observed despite a potentially deleterious effect on amphipods on sample date 1 in 2004. During the same study, 1 of 54 responses had a significant overall treatment effect for sticky-card samples. Many of the responses for stickycard samples suggested significant time effects and time 3 treatment effects. Three response variables were associated with fewer individuals present in the insecticide-treated plots in a multivariate analysis. For the multiple-spray study conducted in 2005 and 2006, 6 of the response variables collected via sticky cards exhibited significant overall treatment effects, but none was associated with fewer individuals in the insecticide-treated plots. None of the responses collected using sweep-net sampling suggested overall treatment effects. Time and time × treatment effects were prevalent in 2005, but no discernable pattern was evident. In general, nearly all the responses evaluated for either study indicated few, if any, deleterious effects from insecticide application.” Consequently, we scored this study as “no effect” even though there were a few statistically significant effects within the study (as is common with studies of this type, Supplementary Table S1). We used this weight of evidence approach across all study types because most studies examined effects on multiple taxa. As such, and arguably as a limitation of our study, we relied on the interpretations of effect from the study authors as evidenced in the text in their respective papers (see article abstracts in Supplementary Table S1). To answer our primary question, “Are there differences between study type and effects on non-target arthropods?” we first conducted a simple vote tally of effect vs. no effect per the description above (Table 1). We then analyzed in R Studio version 2025.9.2.418 (Posit Team 2025) whether the likelihood of observing an effect differed among study types (field, operational, ERA) using both a nonparametric test and a logistic regression approach. Because the dataset contained relatively few studies in some categories and complete separation was present (operational studies reported no effects), we applied Fisher’s exact test to evaluate the contingency table and used Firth logistic regression to estimate model coefficients under separation. Fisher’s exact test is well-suited to sparse data because it calculates the exact probability of the observed distribution under the null hypothesis, while Firth regression provides penalized likelihood estimates that remain stable even when one category contains only zeros. These analytical choices were made specifically in response to the structure of the operational data, in which the complete absence of detected effects precludes standard logistic regression and warrants methods robust to zero-cell conditions. Table 1.Summary of non-target arthropod effects by article, study type, application type, insecticide active ingredient, and taxaArticleStudy typeApplication typeActive ingredientTaxaEffect/No Effect byTaxa common nameTaxonActive ingredient Bargar (2012) ERAAerialNaledLepidopteraButterfliesEffectEffect Bargar and Hladik (2024) ERAGroundPermethrinHesperiidaeSkipper butterfliesEffectEffect Davis et al. (2007) ERAGroundD-phenothrinDaphniidaeWaterfleasNo effectNo effect Davis et al. (2007) ERAGroundMalathionDaphniidaeWaterfleasNo effectNo effect Davis et al. (2007) ERAGroundNaledDaphniidaeWaterfleasNo effectNo effect Davis et al. (2007) ERAGroundPermethrinDaphniidaeWaterfleasNo effectNo effect Davis et al. (2007) ERAGroundPyrethrinsDaphniidaeWaterfleasNo effectNo effect Davis et al. (2007) ERAGroundResmethrinDaphniidaeWaterfleasNo effectNo effect Davis et al. (2007) ERAGroundD-phenothrinGammaridaeScudsNo effect Davis et al. (2007) ERAGroundMalathionGammaridaeScudsNo effect Davis et al. (2007) ERAGroundNaledGammaridaeScudsNo effect Davis et al. (2007) ERAGroundPermethrinGammaridaeScudsNo effect Davis et al. (2007) ERAGroundPyrethrinsGammaridaeScudsNo effect Davis et al. (2007) ERAGroundResmethrinGammaridaeScudsNo effect Hoang et al. (2011) ERAAerialNaledLepidopteraButterfliesEffectEffect Hoang et al. (2011) ERAAerialPermethrinLepidopteraButterfliesEffectEffect Hoang and Rand (2015a )ERAAerialNaledLepidopteraButterfliesEffectEffect Hoang and Rand (2015a )ERAAerialPermethrinLepidopteraButterfliesEffectEffect Hoang and Rand (2015b )ERAAerialNaledLepidopteraButterfliesEffectEffect Hoang and Rand (2015b )ERAAerialPermethrinLepidopteraButterfliesEffectEffect Schleier and Peterson (2010) ERAGroundNaledGryllidaeCricketsNo effectNo effect Schleier and Peterson (2010) ERAGroundPermethrinGryllidaeCricketsNo effectNo effect Schleier and Peterson (2013) ERAGroundPermethrinMultiple TaxaNo effectNo effect Schleier et al. (2008) ERAAerialPyrethrinsDaphniidaeWaterfleasNo effectNo effect Bargar et al. (2020) FieldAerialNaledNymphalidaeBrush-footed butterfliesEffectEffect Bargar et al. (2020) FieldAerialNaledPieridaeWhite/yellow butterfliesEffect Bargar and Jiang (2024) FieldGroundPermethrinNymphalidaeBrush-footed butterfliesEffectEffect Boyce et al. (2007) FieldAerialPyrethrinsMultiple TaxaNo effectNo effect Caron (1979) FieldGroundMalathionApidaeHoney and bumble beesEffectEffect Caron (1979) FieldGroundNaledApidaeHoney and bumble beesEffectEffect Caron (1979) FieldGroundPyrethrinsApidaeHoney and bumble beesEffectEffect Chaskopoulou et al. (2014) FieldAerialD-phenothrinApidaeHoney and bumble beesNo effectNo effect Chaskopoulou et al. (2014) FieldAerialDeltamethrinApidaeHoney and bumble beesNo effectNo effect Davis and Peterson (2008) FieldGroundD-phenothrinMultiple TaxaNo effectNo effect Davis and Peterson (2008) FieldGroundPermethrinMultiple TaxaNo effectNo effect Giordano et al. (2020) FieldGroundDeltamethrinNymphalidaeBrush-footed butterfliesEffectEffect Hester et al. (2001) FieldGroundMalathionApidaeHoney and bumble beesEffectEffect Jensen et al. (1999) FieldGroundMalathionMultiple TaxaNo effectNo effect Jensen et al. (1999) FieldGroundPermethrinMultiple TaxaNo effectNo effect Jensen et al. (1999) FieldGroundPyrethrinsMultiple TaxaNo effectNo effect Jensen et al. (1999) FieldGroundPrallethrinMultiple TaxaNo effectNo effect Kim et al. (2022) FieldGroundMalathionMultiple TaxaEffectEffect Kwan et al. (2009) FieldAerialPyrethrinsMultiple TaxaNo effectNo effect McGregor et al. (2021) FieldGroundDeltamethrinApidaeHoney and bumble beesEffectEffect Milam et al. (2000) FieldAerialPermethrinDaphniidaeWaterfleasEffectEffect Milam et al. (2000) FieldGroundPermethrinDaphniidaeWaterfleasEffectEffect Oberhauser et al. (2009) FieldGroundResmethrinNymphalidaeBrush-footed butterfliesEffectEffect Peterson et al. (2016) FieldGroundPermethrinCoccinellidaeLady beetlesEffectEffect Piccolomini et al. (2018) FieldGroundEtofenproxMegachilidaeLeafcutting beesNo effectNo effect Pokhrel et al. (2018) FieldGroundD-phenothrinApidaeHoney and bumble beesNo effectNo effect Pokhrel et al. (2018) FieldGroundDeltamethrinApidaeHoney and bumble beesNo effectNo effect Pokhrel et al. (2018) FieldGroundResmethrinApidaeHoney and bumble beesNo effectNo effect Rinkevich et al. (2017) FieldGroundD-phenothrinApidaeHoney and bumble beesNo effectNo effect Rinkevich et al. (2017) FieldGroundMalathionApidaeHoney and bumble beesNo effectNo effect Rinkevich et al. (2017) FieldGroundPermethrinApidaeHoney and bumble beesEffectEffect Rinkevich et al. (2017) FieldGroundPrallethrinApidaeHoney and bumble beesEffectEffect Rinkevich et al. (2017) FieldGroundResmethrinApidaeHoney and bumble beesNo effectNo effect Sanchez-Arroyo et al. (2022) FieldGroundPermethrinApidaeHoney and bumble beesNo effectNo effect Schleier and Peterson (2010) FieldGroundNaledGryllidaeCricketsNo effectNo effect Schleier and Peterson (2010) FieldGroundPermethrinGryllidaeCricketsNo effectNo effect Tietze et al. (1996) FieldGroundMalathionGryllidaeCricketsEffectEffect Wurzel et al. (2020) FieldGroundPermethrinMultiple TaxaEffectEffect Zhong et al. (2010) FieldAerialNaledLycaenidaeGossamer-winged butterfliesEffectEffect Zhong et al. (2003) FieldAerialNaledApidaeHoney and bumble beesEffectEffect Zhong et al. (2004) FieldAerialNaledApidaeHoney and bumble beesEffectEffect Barascou et al. (2025) OperationalAerialNaledApidaeHoney and bumble beesNo effectNo effect Barascou et al. (2025) OperationalGroundChlorpyrifosApidaeHoney and bumble beesNo effectNo effect Boyce et al. (2007) OperationalAerialPyrethrinsMultiple TaxaNo effectNo effect Breidenbaugh and de Szalay (2010) OperationalAerialNaledMultiple TaxaNo effectNo effect Caron (1979) OperationalGroundNaledApidaeHoney and bumble beesNo effectNo effect Caron (1979) OperationalGroundMalathionApidaeHoney and bumble beesNo effectNo effect Caron (1979) OperationalGroundPyrethrinsApidaeHoney and bumble beesNo effectNo effect Chaskopoulou et al. (2014) OperationalAerialD-phenothrinApidaeHoney and bumble beesNo effectNo effect Chaskopoulou et al. (2014) OperationalAerialDeltamethrinApidaeHoney and bumble beesNo effectNo effect Chaskopoulou et al. (2014) OperationalAerialD-phenothrinChrysopidaeLacewingsNo effect Chaskopoulou et al. (2014) OperationalAerialDeltamethrinChrysopidaeLacewingsNo effect Chaskopoulou et al. (2014) OperationalAerialD-phenothrinCoccinellidaeLady beetlesNo effect Chaskopoulou et al. (2014) OperationalAerialDeltamethrinCoccinellidaeLady beetlesNo effect Crowder et al. (2025) OperationalAerialNaledApidaeHoney and bumble beesNo effectNo effect Hart et al. (2024) OperationalGroundDeltamethrinMultiple TaxaNo effectNo effect Hart et al. (2024) OperationalGroundPermethrinMultiple TaxaNo effectNo effect Hart et al. (2024) OperationalGroundD-phenothrinMultiple TaxaNo effectNo effect Hart et al. (2024) OperationalGroundPrallethrinMultiple TaxaNo effectNo effect Kwan et al. (2009) OperationalAerialPyrethrinsMultiple TaxaNo effectNo effect Kwan et al. (2009) OperationalGroundPyrethrinsMultiple TaxaNo effectNo effect Rochlin et al. (2022) OperationalAerialNaledMultiple TaxaNo effectNo effectEffects in red text had statistically significant effects; no effects in blue text did not. ERA = Ecological Risk Assessment. Full references for article ID are in Table S1 (S1). Summary of non-target arthropod effects by article, study type, application type, insecticide active ingredient, and taxa Effects in red text had statistically significant effects; no effects in blue text did not. ERA = Ecological Risk Assessment. Full references for article ID are in Table S1 (S1). Results The 38 records (peer-reviewed journal articles) included 25 field studies, nine operational studies, and nine risk assessments (Table 1, S1). The total number of studies, 43, exceeds 38 journal papers because some papers include multiple study types. The application method included 25 ground (58%) and 20 aerial (47%) studies (Table 1, Fig. 2). Hierarchy plots of journal articles. Outdoor space applications of mosquito adulticides are summarized by application type, study type, insecticide active ingredient, and non-target arthropod taxon. Red text indicates statistically significant effects; blue text indicates no effects. (A) = ground applications; (B) = aerial applications. Numbers after active ingredients are journal articles identifications from Supplementary Table S1. Eleven studies (26%) were on bees (10 Apis mellifera, two Bombus spp., and one Megachile rotundata), 11 (26%) were on multiple taxa, 10 (23%) were on Lepidoptera (butterflies), three (7%) were on aquatic crustaceans (daphnids and gammarids), two (5%) were on coccinellids, two (5%) were on gryllids, and one (2%) was on chrysopids (Supplementary Table S1, Fig. 2). Ten insecticide active ingredients were evaluated over all studies: Organophosphates (chlorpyrifos, malathion, naled), Pyrethroids/Pyrethrins (d-phenothrin, deltamethrin, etofenprox, permethrin, prallethrin, pyrethrins, resmethrin). The simple vote tally revealed no effects for all 9 operational studies, regardless of application method (Table 1). Ecological risk assessments had five effects and four no effects for the 9 studies. For the 25 field studies, 15 had effects and 10 had no effects (Table 1). Fisher’s exact test revealed a significant association between study type and detection of non‑target effects (P = 0.00036). Operational studies showed a strong and significant negative association with effect detection (log‑odds = –3.56, 95% CI: –8.48 to –1.24, P = 0.00082). This result reflects the complete absence of reported effects in operational studies, in contrast to both field and ERA studies. Field studies did not differ significantly from ERA studies (log‑odds = 0.54, 95% CI: –0.63 to 1.8, P = 0.37). Application type was a significant predictor: ground-based applications had lower odds of detecting non‑target effects compared to aerial applications (log‑odds = –1.37, 95% CI: –2.68 to –0.19, P = 0.022). Together, these analyses demonstrate that study type is strongly associated with whether non‑target effects are reported, with operational studies uniquely not detecting effects. The regression results further highlight application method as an additional predictor, although less influential than study type. The strength of these associations is evident both in the contingency table and in the penalized regression estimates, underscoring the robustness of the finding despite small sample sizes and potential separation issues. Discussion Despite several limitations, which are discussed below, a few patterns emerged. Arguably the most important result is that the operational-level studies did not show deleterious effects on non-target arthropods at the population level. Although only nine published articles, the operational studies represent actual applications by mosquito management districts. These applications would yield the most relevant and realistic indication of deleterious effects on non-target arthropods because districts treat large areas. The consistent absence of effects across operational studies most likely reflects fundamental physicochemical and exposure factors of ULV applications rather than methodological insensitivity. Operational applications incorporate multiple exposure reduction factors reflected in label-application requirements: applications occur at night when most non-target arthropods are inactive and sheltered within foliage or other resting sites; the 8- to 60-µm droplet size is engineered to remain airborne only long enough to intercept flying mosquitoes before rapid dispersal and dilution reduce concentrations below critical exposure thresholds; and the aerosol dissipates before reaching most resting arthropods. The uniformity of null results across seven aerial and three ground-based operational studies conducted in different regions and using different equipment further suggest that these findings reflect consistent real-world exposure dynamics rather than artifacts of study design or other forms of bias. The results from the field studies are intriguing and important. Because the results from the field and operational studies were significantly different, this difference suggests either field studies are not effective surrogates for operational-level applications or operational studies are biased against showing deleterious effects on non-target arthropods when they are occurring. Although not unequivocal, the field studies revealed effects. These effects are likely because many of these field studies have experimental designs that include relatively small linear-path applications of ULV insecticides and caged insects above ground level directly in the path of the aerosol (Table 1). These types of applications do not seem to adequately simulate operational applications, in part because they do not provide vegetative shielding from plant canopy structures. Therefore, we believe it is likely that field studies with non-target arthropods are not designed or suited to assess if operational treatments by mosquito management districts result in significant effects on populations of non-target arthropods. Rather, they are designed to assess if broad-spectrum insecticides used in adult mosquito management are toxic to non-target arthropods (Fritz et al. 2014). However, we argue that is not the question because it is well established that broad-spectrum insecticides have the inherent ability to kill arthropods. In short, we believe that researchers (including current author RKDP) have used incorrect experimental designs to try to answer the relevant question. Interestingly, though, the field studies have equivocal results. Although field studies were significantly more likely to detect effects than operational studies, 10 of the 26 studies reported no effects. When examining taxa only across all field studies, 16 taxa were affected and 16 were not. This result shows that even caged insects or insects placed directly in the aerosol path did not uniformly experience statistically significant mortality. This is likely because of the low ULV application rate, small droplet sizes, and the lack of arthropod access to the aerosol as it passed through the study area. Peterson et al. (2016) examined the determinants of non-target insect mortality to a ground-based ULV mosquito insecticide using Hippodamia convergens as a surrogate non-target insect and found that acute mortality was largely caused by aspects of the insect’s position with respect to the insecticide aerosol. Specifically, acute mortality was determined by the insect’s ability to directly intercept the droplets and the insect’s height during the outdoor space application (Peterson et al. 2016). In addition, caging non-target insects in field studies is often not the main objective. Rather, caging adult mosquitoes to assess mortality or insecticide resistance is often the primary objective of the study. The ERA studies also show equivocal results. Five studies showed effects and 4 did not. The results from risk assessments need to be carefully interpreted because they tend to deliberately overestimate effect and exposure as part of their tiered structure for decision-making (NRC 1983, Peterson 2025). In a sense, they are reasonable worst-case assessments of risk. The tiered approach to risk assessment refers to the recursive process whereby when risk assessment shows concerns, the assessment is refined or mitigation options are implemented (SETAC 1994, USEPA 1998). Lower tiers are characterized relatively low cost and the use of overestimates of effect and exposure to screen out negligible risks. There are limitations to our study and to the studies included in our analysis. In addition to the diverse study types of operational, field, and ecological risk assessments, the studies use diverse methods across diverse arthropod taxa to assess deleterious effects. It should be noted that most of the studies only quantify acute mortality and risk, so potential sublethal effects such as behavior, reproduction, and chronic impairment were not measured. In addition, although the number of operational studies is limited, the consistency of null results across methodologically heterogeneous studies and the use of analytical methods robust to sparse data and complete separation argue against a power-only explanation. Although field studies documented acute mortality in 16 taxa, individual insect mortality does not equate to population-level risk. Mortality from ULV applications represents a localized, non-recurring perturbation that must be evaluated within the context of natural population dynamics. In most cases, ULV-induced mortality most likely functions as compensatory mortality, substituting for deaths that would have occurred from other causes rather than adding to baseline mortality rates (Peterson et al. 2009). Species with high intrinsic population growth rates typically can experience acute, localized, and non-repeated mortality events without population-level consequences (MacArthur and Wilson 1967, Pianka 1970, Breidenbaugh and de Szalay 2010, McMullen et al. 2017, Rochlin et al. 2022, Crowder et al. 2025). This framework is perhaps best illustrated by managed honey bees (A. mellifera), the subject of 10 studies in our review. Despite honey bees’ high individual susceptibility to pyrethroids, population-level risk is effectively managed by primarily applying insecticides at times that avoid periods of peak flight activity. If this simple mitigation strategy maintains population viability for one of the most economically important and intensively studied pollinators, the framework for assessing tolerable mortality thresholds arguably provides a sound basis for evaluating risk across other taxa where acute mortality was observed. Our analysis of 38 journal articles suggests that risks to bees seem low and can be mitigated by adjusting application timing to avoid direct interception of ULV droplets during flight. Risks to large insects seem low provided they do not directly intercept the droplets. Mortality to small, night-active insects was demonstrable, but mortality to individuals and risk to populations are different phenomena. Deleterious effects on populations were not the focus of most studies, but those studies that examined longer-term effects did not observe effects (eg Breidenbaugh and de Szalay 2010, Rochlin et al. 2022). The lack of effects on populations has also been observed for mosquito larvicides (Jensen et al. 1999, Lawler et al. 1999, Lawler 2017). Reduced risk to populations, even in the presence of individual mortality, is likely attributable to Integrated Pest Management (IPM) approaches that target applications to areas where mosquito thresholds have been reached. In addition, there may have been compensatory mortality in control areas, ie the mortality that would have been caused by ULV insecticide application if control areas were treated was replaced by other causes (Peterson et al. 2009). Risks to butterflies need more study and clarification. It is unclear whether the results reflect increased susceptibility to butterfly species or methodological issues in the field studies (Oberhauser et al. 2009, Zhong et al. 2010, Bargar et al. 2020, Giordano et al. 2020, Bargar and Jiang 2024) and ecological risk assessments (Hoang et al. 2011, Bargar 2012, Hoang and Rand 2015a, Bargar and Hladik 2024). As mentioned above, risk assessments tend to overestimate effect and exposure because they are often worst case as part of their tiered structure for decision-making (NRC 1983, Peterson 2025). Risk to non-target arthropods from outdoor space applications of mosquito adulticides seems to be relatively low most likely because exposure is greatly reduced by (1) night applications and the lack of many non-target arthropods in the direct path of the aerosol, (2) ULV droplet size and volume of insecticides, and (3) rapid dispersal, dilution, and degradation of the aerosolized insecticide. Because of their conservative nature, ecological risk assessments can be used to screen negligible risks, and, if estimated risks are unacceptable, to understand what additional data are needed to make the risk assessment more realistic or to mitigate risks (Peterson 2025). Therefore, ecological risk assessments should be considered as representing reasonable worst-case scenarios. These conservative approaches are used by regulatory agencies as they register and periodically review existing registrations for insecticides for adult mosquito management. Therefore, the results of risk assessments that are not conducted by regulatory agencies should be evaluated with caution. Smaller-scale field studies that attempt to simulate operational applications should be avoided if the research goal is to determine the effect on non-target arthropods from operational-level applications by mosquito management districts or other institutions. These smaller-scale field studies should also be thought of as worst-case scenarios. However, these studies may be informative because if they do not result in reductions in the numbers of caged or deliberately placed insects, they might obviate the need for assessments at operational-scale levels. When considering smaller-scale field studies, operational studies, and risk assessments, the results from this study arguably suggest that the regulatory risk assessment process works to manage non-target arthropod risks. We believe this study reveals that operational-level applications of adult mosquito insecticides should be the sin qua none for understanding deleterious effects on non-target arthropod populations. Although it is more challenging to plan and conduct, operational studies should be used whenever possible rather than smaller-scale field studies because these are the methods used by mosquito management districts. To date, only nine operational articles are published in the peer-reviewed literature. There needs to be more operational studies to support or reject our findings, as well as to serve as a robust foundation to assess risks. These studies should use standard before-and-after approaches with multiple sampling techniques for different arthropod taxa (Breidenbaugh and de Szalay 2010, Rochlin et al. 2022, Hart et al. 2024). They also should incorporate standard measurement and reporting metrics, such as abundance, taxa diversity, rare-taxa counts, post-application assessment intervals, and application methods. Supplementary Material