Establishing protocols to apply repellents while hazing crop pests: importance of habitat, flock size, and time on blackbird (Icteridae) responses to a drone capable of spraying

Introduction

In crop production, profit loss owing to bird damage is prevalent across the globe, affecting a variety of agricultural sectors, including viticulture, horticulture, and other cropping systems (Conover 2002; Tracey et al. 2006). The nuisance bird differs depending on region, location, and/or crop, but blackbirds (Icteridae) are known to damage crops across North America (Shwiff et al. 2017). In sunflowers (Helianthus annuus), direct production losses reach US$3.5 million annually in North Dakota (Klosterman et al. 2013), whereas the additive impacts of indirect loss extend to US$18 million when considering local economies (i.e. impact of change in one sector on regional revenue and jobs) and the labor and time spent managing the issue (Ernst et al. 2019).

At the same time, blackbirds are a protected, native family of birds facing overall population declines of ~1% annually. Although red-winged blackbird (Agelaius phoeniceus) populations vary geographically, some regions, such as the northern Great Plains, are experiencing 1.5% annual population increases (Rosenberg et al. 2019; Sauer et al. 2019), along with significant conflicts with agriculture. Landscapes with wetlands dominated by cattails are preferred by blackbirds during the breeding season and as roosting habitat during fall molt and migration (Linz et al. 2003; Forcey et al. 2015). Blackbirds can travel 8–10 km from their night-time roost to forage; thus, sunflower fields with cattail marshes nearby or imbedded within the field suffer higher crop damage (Dolbeer 1990; Homan et al. 2005; Dolbeer and Linz 2016). In North Dakota, cattail roosts can host >1 million migrating blackbirds in the fall, causing significant damage to nearby crops (Clark et al. 2020).

Methods to mitigate damage caused by blackbirds are implemented at multiple spatial scales, but tools at the field scale are the most accessible and most frequently used by farmers (White 2021). However, current field-scale options (e.g. exclusion methods, frightening devices, habitat manipulation, and chemical repellents) are limited in terms of efficacy, range, feasibility, or cost (Sausse et al. 2021; Klug et al. 2023; Duttenhefner and Klug 2024). In situations of commercial agriculture, stationary field-based tools (e.g. propane cannons or scarecrows) are not sufficient to move large flocks out of large fields and birds are often tolerant of the tools. Thus, dynamic, mobile tools are needed and may require the integration of multiple tools to increase efficacy (Klug 2017).

Drones are dynamic, highly mobile, and show promise for use in human–wildlife conflicts (Hahn et al. 2017; Penny et al. 2019; Chilvers 2024). Managers of small-scale agriculture systems such as vineyards, rice paddies (Oryza spp.), and aquaculture ponds have deployed drones to detect and harass nuisance birds (Burr et al. 2019; Rhoades et al. 2019; Mohamed et al. 2020; Wang et al. 2020). In controlled aviary settings, drones elicit antipredator alert and escape responses by individual male, red-winged blackbirds (Wandrie et al. 2019; Egan et al. 2020). However, there is proven difficulty in drones producing a consistent abandonment response when hazing blackbird flocks in open agricultural landscapes (i.e. flock and field size impact success; Egan et al. 2023). Increasing the negative stimulus associated with drones, to affect a greater proportion of the flock, may increase abandonment rates. Suggestions have included emitting frightening acoustics (Wang et al. 2019), coordinating a fleet of swarming drones (Paranjape et al. 2018), designing drones to resemble predators (Egan et al. 2020), and deploying an avian repellent from a spraying drone (Ampatzidis et al. 2015; Duttenhefner 2023).

A spraying drone could be used to haze birds, while simultaneously applying an avian repellent to spot-treat areas being actively damaged by birds. Non-lethal repellents are classified as primary repellents if the chemical irritates peripheral chemical senses to initiate a reflexive withdraw (e.g. methyl anthranilate), and as secondary repellents if ingestion results in gastrointestinal irritation and learned avoidance (e.g. anthranilate; DeLiberto and Werner 2024). Traditionally, avian repellents registered for foliar application are applied once a season, on entire fields, by using an airplane or high-boy sprayer, which is expensive, not responsive to the timing of bird damage, and subject to limits imposed by crop structure (Werner et al. 2005; Kaiser et al. 2021). Delivery of an avian repellent via drone may increase direct contact with the chemical owing to real-time application and allows fields to be spot-treated multiple times, effectively reducing chemical costs (Duttenhefner et al. 2024). Although not tested in this study, the application of an avian repellent may increase hazing efficacy.

Prior to conducting studies on the efficacy of hazing blackbird flocks with a drone that also sprays an avian repellent, we need to establish best practices for approaching flocks and how perceptions of risk and responses to drones vary in relation to flock characteristics, habitat, environmental conditions, and drone hazing paths. Flock responses to a drone capable of spraying an avian repellent inform future approach protocols by (1) indicating whether the spraying drone can approach within range of the spray drift, and if within range, how close; and (2) identifying the baseline capability of the spraying drone to elicit abandonment and flock reductions without chemicals. We can minimize variation in future research by selecting one hazing approach (i.e. herding or chaos), understanding implications of avoiding cattail areas while spraying, and identifying ecological factors that confound tool efficacy. With this knowledge, we can establish protocols to effectively deliver repellents by using the same drones as used for hazing. The blackbird–sunflower conflict in North Dakota, USA, is a case study for evaluating flock responses to drones capable of spraying avian repellents (Klug 2017).

We evaluated the probability of being able to approach flocks within 80 m (i.e. range of potential spray drift; Duttenhefner 2023) and, if a close approach occurred, the flight initiation distance (FID) of free-ranging, blackbird flocks to a drone with spraying capabilities. FID is the measurable distance between a bird and an approaching object at the moment the bird decides to take flight (Cooper and Blumstein 2015), and is often used as a behavioral metric to determine how threatened birds are by an approaching object. Many factors influence FID, including habitat, group size, hunting pressure, approach angle, and speed (Fernández-Juricic et al. 2002; Cooper and Blumstein 2015; Egan et al. 2020; Lunn et al. 2022). The flock, habitat, landscape, and environmental variables that influence FID may differ from those influencing behavior during extended hazing. FID is a snapshot of perceived risk of an approaching threat, whereas extended hazing affects risk perception over time.

We evaluated variables influencing FID of blackbird flocks in sunflower–cattail habitat complexes when approached directly by a drone and the probability of abandonment and flock reduction when hazed by a drone. We predicted that larger flocks would react to the approaching drone at greater distances (Egan et al. 2023), given this trend was also seen in waterbirds when approached on foot and by drone (Laursen et al. 2005; Jarrett et al. 2020). We predicted that greater distances between the flock and drone launch site would increase FID, given longer exposure times, and the continual monitoring of an approaching threat is a cost in many bird species (Blumstein 2003; Vas et al. 2015). As shown by Egan et al. (2023), we predicted that smaller field sizes and flock sizes would increase flock abandonment, given the proximity and accessibility of refugia surrounding smaller fields and the behavior of birds in larger flocks (i.e. safety in numbers). We thought pushing the flock may increase abandonment of the habitat, given that repeatedly cutting through the flock may not direct birds out of the sunflower field. Regardless of abandonment, we expected an overall reduction in flock size as a result of drone hazing, as seen in other commodities (i.e. vineyards; Bhusal et al. 2018).

We expected blackbirds to respond to drones differently depending on environmental conditions. Temperature can affect bird behavior, with cold temperatures causing increased metabolic needs; thus, a stronger commitment to food, or heat inducing a stress response (Fernández-Juricic et al. 2002). Wind speed affects drone performance (e.g. speed), but birds may also choose to conserve energy by not flushing or flying during higher winds (Egan et al. 2020). High ambient light reduces detection rates of birds towards an approaching threat, owing to sunlight causing glare effects (Fernández-Juricic et al. 2012). Both habitat types provide refugia through structural safety, but activities (e.g. foraging in sunflower and loafing in cattail) could affect detection and commitment to a location, as seen in the antipredator responses of European starlings (Sturnus vulgaris) changing with grass height (Devereux et al. 2008). We predicted that flocks located in cattail would have a lower FID and lower abandonment, owing to its function as refugia and the vast area of cattail marshes on the landscape (Bansal et al. 2019). Feeding rates are typically highest in the morning and decline in the afternoon for blackbirds, as foraging needs have been met for the day (Hintz and Dyer 1970); thus, we expected a lower probability of abandonment in the morning than in the afternoon or evening. Last, we included Julian day because it could indicate food availability on the landscape (i.e. increased harvest as season progresses) or a progression towards migration and subsequent caloric needs (Cooper and Blumstein 2015).

Materials and methods

Study design

We conducted drone trials in commercial sunflower fields in North Dakota counties experiencing blackbird damage (Bottineau, Burleigh, Emmons, Kidder, Logan, and McHenry; Fig. 1). We targeted locations where blackbird flocks were either actively foraging on sunflower or roosting in cattails within or adjacent to commercial sunflower fields, from 4 September to 25 October in 2019 and 2020, between 07:30 hours and 19:00 hours. We conducted FID trials in both 2019 and 2020, but conducted only hazing trials in 2020.

Fig. 1.

(a) Trial locations in North Dakota, USA, where we approached and subsequently hazed mixed-species blackbird (Icteridae) flocks with a drone (DJI Agras MG-1P) in sunflower–cattail complexes where sunflower (Helianthus annuus) was being actively damaged, and cattail (Typha spp.) was used as daytime roosting habitat. We conducted 91 trials (2019: FID = 31; and 2020: combined FID and hazing = 60) by using the Agras from 4 September to 25 October (overlap occurs because of scale). (b) An example of a sunflower–cattail complex, where the dark brown and green, irregularly shaped wetlands are within and adjacent to the lighter brown agricultural field (Esri, Maxar, Earthstar Geographics, and the GIS User Community).

Large mixed-species flocks of blackbirds are mainly composed of red-winged blackbirds but also include yellow-headed blackbirds (Xanthocephalus xanthocephalus), common grackles (Quiscalus quiscula), brown-headed cowbirds (Molothrus ater), Brewer’s blackbirds (Euphagus cyanocephalus), and European starlings. Cattail marshes are often within or adjacent to sunflower fields in the Prairie Pothole Region of North Dakota, and are used as daytime loafing or refugia sites and night-time roosting sites (Fig. 1). Our study period coincided with the 8-week damage window (ray-petal drop of sunflower up until harvest) when blackbirds are undergoing fall molt and migration (Linz et al. 2017; Stonefish et al. 2021). We considered each blackbird flock as an independent experimental unit, given population turnover with incoming migrant birds and flock mixing at roosting sites resulting in no two flocks being the same over time (Linz et al. 1991). We did not mark flocks; thus, it is possible that individual blackbirds were approached multiple times. If we operated multiple trials at a single field, we allowed an average of 14 days (±9.1 s.d.) between trials in the same year. We conducted hazing trials at 30 fields (12 fields visited once, 11 fields visited twice, three fields visited three times, three fields visited four times, and one field visited five times). We conducted trials evaluating whether flight occurred within 80 m at 46 fields (16 fields visited once, 21 fields visited twice, four fields visited three times, four fields visited four times, and one field visited five times). Sunflower producers practice rotational agriculture; thus, no sunflower fields were repeated in 2019 (16 fields) and 2020 (30 fields). Egan et al. (2023) found that day did not explain significant variation in abandonment during drone hazing; thus, we did not include day as a random factor.

We used a precision agriculture spraying octocopter (DJI AGRAS MG-1P; DJI Shenzhen, China; hereafter, Agras) to approach and haze blackbird flocks (Fig. 2). We used smaller quadcopters (DJI Phantom 4 and DJI Mavic Air 2, DJI Shenzhen, China; Fig. 2) as the eye-in-the-sky drones to video blackbirds responding to the Agras. Wandrie et al. (2019) found no observable impact on flock behavior when flying a drone >60 m above ground level (AGL); thus, we flew the Phantom at 60 m AGL (2019) and the Mavic at 80 m AGL (2020) to increase the field of view. Both eye-in-the-sky drones captured 4K video, but the Mavic was 60 frames s⁻¹ and Phantom 30 frames s⁻¹.

Fig. 2.

We approached and subsequently hazed mixed-species flocks of blackbirds (Icteridae) with (a) the DJI Agras MG-1P, a drone with spraying capabilities, which has a diagonal length of 1500 mm, a max speed of 15 m s⁻¹, a tank that can hold 10 kg (approximately 9.5 L) of liquid, and has a maximum hovering time of 20 min with an empty tank (9 min with a full tank). We recorded the close approach and flight initiation distance of the Agras by using eye-in-the-sky drones, (b) the DJI Phantom 4, which has a diagonal length of 350 mm, a max speed of 20 m s⁻¹, and a max flight time of ~28 min, and (c) the DJI Mavic Air 2, which has a diagonal length of 302 mm, a max speed of 19 m s⁻¹, a max flight time of ~34 min.

We recorded the habitat where the flock was located (i.e. cattail or sunflower), time of day, Julian day, ambient light (μmol m⁻² s⁻¹) with a Li-Cor LI-250 Light Meter and LI-190SA Quantum Sensor (Lincoln Nebraska, USA), temperature (°C) and wind speed (m s⁻¹) with a Skymaster SM-28 weather meter (Speedtech Instruments, Great Falls, Virginia USA). We used Google Earth Pro (ver. 7.3.4.8248, image dates 2016–2020) to measure the area of sunflower and cattail marshes adjacent to or embedded in the crop fields to calculate the size of the sunflower–cattail complexes and the proportion of cattail (Fig. 1). The process for drone trials was (1) a pre-hazing observation period (15 min), (2) drone launch and initial approach (FID), (3) hazing period (10 min), (4) drone landing, and (5) post-hazing observation period (15 min).

Results

We approached 91 flocks (sunflower = 60, cattail = 31) to evaluate the probability of successful approach within the range of the potential spray drift (2019 = 31, 2020 = 60) and measured FID from the eye-in-the-sky drone video for 62 trials (2019 = 21, 2020 = 41) because of flocks taking flight beyond the potential spray drift (80 m) and sometimes outside of the field of view of the eye-in-the-sky. Of those 62 trials, we approached 36 flocks in sunflower and 26 in cattail adjacent to or within the sunflower field. The optimal model for probability of close approach (drone approached flock within 80 m) included habitat only (Tables 1, 2a, Fig. 3a). When we were able to closely approach flocks (≤80 m), average FID was 40 m ± 14.3 s.d. We combined years because we found no difference (t₆₀ = 0.10, P = 0.92) between FID in 2019 (39 m ± 2.2 s.e.) and 2020 (39 m ± 2.5 s.e.). The optimal model for FID included time of day, flock size, and distance to launch (Tables 1, 2b, Fig. 3b, c). When the Agras was able to closely approach the flock (≤80 m), we found that FID was shorter with smaller flocks, later in the day (Table 2b, Fig. 3b, c).

Fig. 3.

Results of trials using a drone with spraying capabilities (DJI Agras MG-1P) from 4 September to 25 October (2019–2020) in sunflower–cattail complexes where sunflower (Helianthus annuus) was being actively damaged by mixed-species flocks of blackbirds (Icteridae) in North Dakota, USA. Cattail (Typha spp.) is used as daytime roosting habitat by blackbirds. (a) The probability of flocks taking flight within 80 m of the approaching Agras when located in cattail or sunflower. Model-based estimates of flight initiation distance (FID) for flocks responding to the approach of the Agras as a function of (b) time of day and (c) flock size. We held other model covariates at mean values. The shaded area shows 95% confidence intervals.

We conducted 60 hazing trials in 2020 (chaotic = 30 flocks, herding = 30 flocks) with 36 flocks being initially located in sunflower and 24 flocks in cattail within or adjacent to the sunflower field. Hazing trials resulted in 52% of flocks abandoning the targeted habitat within the 10-min hazing period (n = 31), whereas 37% of the flocks partially abandoned (n = 22), and 12% remained (n = 7). During the post-hazing period, 81% of flocks that completely abandoned during the hazing period returned to the same habitat (n = 25). The optimal model for abandonment included time of day, flock size, and temperature (Tables 1, 2c, Fig. 4). The probability of flock abandonment early in the morning was ~12% and increased as the day progressed (Fig. 4a), whereas the probability of abandonment for small flocks started at 65%, and decreased as flock size increased (Fig. 4b).

Fig. 4.

Results of drone hazing trials from 4 September to 25 October 2020 in sunflower–cattail complexes where sunflower (Helianthus annuus) was being actively damaged by mixed-species flocks of blackbirds (Icteridae) in North Dakota, USA. Cattail (Typha spp.) is used as daytime roosting habitat by blackbirds. The probability of flocks abandoning the starting habitat (i.e. sunflower or cattails) after 10 min of drone hazing as a function of (a) time of day and (b) flock size. We held other model covariates at mean values. The shaded area shows 95% confidence intervals.

Flock size increased (n = 14), stayed the same (n = 9), and decreased (n = 37; 100% decline: n = 6) in response to hazing. Excluding instances where flock size increased, flocks were, on average, reduced by 50% ± 36.8 s.d. (n = 46). We get an average reduction of 38% ± 38.6 s.d. (n = 60) if we include flock size increases as 0% declines. If we include instances where flock sizes increased (e.g. 40–600 birds is a 1400% increase), the average change was a 13% ± 2.2 s.d. (n = 60) increase in birds. Thus, when evaluating percentage decline, we included flock size increases as 0% declines to bound the percentage decline between 0 and 100% and arcsine transforming the response variable. Excluding flock size increases, flocks hazed in cattail (n = 21) were reduced by 49% ± 36.6 s.d., whereas flocks in sunflower (n = 25) were reduced by 51% ± 37.7 s.d. Flocks were reduced by 53% ± 38.7 s.d. when operating the drone with a chaotic path (n = 22), whereas a herding path (n = 24) reduced flocks by 47% ± 35.6 s.d. The optimal model for flock reduction included a positive relationship with proportion of cattail (Tables 1, 2d).

Discussion

Flock perceptions of risk appeared highest in the morning (i.e. larger FIDs earlier in the day), but the probability of abandonment from hazing increased as the day progressed, likely owing to energetic needs being met and birds moving to night-time roosting locations. Teräväinen (2022) found that graylag geese (Anser anser) were more likely to move to wetlands when scared out of crop fields in the morning and speculated that it was because of the beginning of mid-day loafing. Foraging by blackbirds also peaks in the morning and declines throughout the afternoon (Hintz and Dyer 1970). Although blackbirds are harder to move in the morning, the integration of spraying with hazing should be tested in the morning to understand the impact of the additional negative stimulus for successful hazing, especially on large flocks.

We also saw that larger flocks responded to the approaching drone sooner, but probability of abandonment was reduced with larger flocks, likely owing to increased vigilance combined with safety in numbers and how birds in large flocks collectively behave to avoid predation. Overall, 10 min of hazing by the Agras resulted in 52% of flocks abandoning the hazed area. However, 81% of flocks returned within 15 min post-hazing. This propensity to return to a resource has been seen in other studies attempting to disperse birds, including gulls from rooftops (Pfeiffer et al. 2023), vultures from landfills (Pfeiffer et al. 2021), and graylag geese from crop fields (Teräväinen 2022). It is likely that the limited alternative habitat (i.e. productive foraging areas ≥ sunflower fields) within a reasonable distance contributed to the high rate of return, a prediction illustrated by Gill et al. (2001) when discussing the ability of species to avoid disturbance. Pfeiffer et al. (2021) found that latency to return was negatively correlated with drone speed for turkey vultures and attributed it to slower drones spending more time in the area, increasing exposure time. Exposure time, resulting from greater launch distances, influences escape behavior (>FIDs) of blackbirds foraging in sunflower fields (Egan et al. 2023) and waterbirds on wetlands (Ryckman et al. 2022). Greater launch distances also affect the ability to aggressively maneuver a hazing drone across large fields and limits the ability to accurately observe flocks (Duttenhefner et al. 2024).

Flight initiation distance in flocks is the response of the birds located on the nearest edge of the flock seeking safety and moving towards the center of the flock (Ballerini et al. 2008). The collective movement of birds on the outer edge towards the center is an antipredator strategy that results in murmurations (Sumpter 2010). Flock size influenced when the flock took flight, and larger flocks flushed earlier to the approaching drone. Sociability within wildlife aggregations have been shown to influence flight distances of guanacos (Lama guanicoe; Schroeder and Panebianco 2021). Our study also supports research showing larger flock sizes have an earlier response time, because more eyes are watching for threats (Lima and Dill 1990; Jarrett et al. 2020; Egan et al. 2023). Flock size is known to affect perceptions of safety (Cooper and Blumstein 2015). Although larger flocks may attract more raptors, the likelihood of an individual being attacked is reduced in large flocks (Cresswell 1994).

Although the individual risk of predation is reduced in larger flocks, that risk could still be highest at certain times of day when predators are also in search of food (Lima and Dill 1990). Blackbird flocks took flight sooner in response to an approaching drone earlier in the day. We observed blackbird flocks refusing to abandon locations early in the morning, and the probability of abandonment increasing throughout the day. When flocks are initiating feeding early in the morning and a threat approaches, it creates a critical foraging decision, i.e. either remain at the valuable food source or leave and travel to another location with an unknown outcome (Abdulwahab et al. 2019). The agricultural landscape of this study, large monoculture fields, does not provide abundant alternative refugia or food resources (Whittingham and Evans 2004), which may have led birds to decide that the drone was not a sufficient threat to cause field abandonment. The probability of abandonment increased later in the afternoon, which is after most damage had occurred, owing to avian foraging occurring in the morning, and birds likely being on their way to night-time roosts in the late afternoon (Hintz and Dyer 1970). Future drone research should target flocks in the morning, to determine the best methods to reduce damage during an important feeding window when flocks are committed to the location.

Drone hazing has an immediate short-term impact on bird behavior as evidenced by reduced flock sizes. Wang et al. (2019) also found that flock activity was reduced immediately following drone hazing of silvereyes (Zosterops lateralis) in an Australian vineyard; however, the birds were hunkered down in the vines and did not necessarily abandon the area. Bhusal et al. (2018) reported reduced flock sizes as a result of drone hazing and increased flock sizes on days where no hazing took place. Reduced activity does not necessarily mean that the birds are not present in the hazed area during the observation period, but instead are likely to be less mobile immediately after drone hazing. Thus, a reduction in activity does not necessarily correlate to a reduction in crop damage. Depending on the crop, disturbing a flock could cause them to target different plants within a field, creating more damage. For example, any damage to sweet corn kernels affects the sale of the whole cob; thus, a hazing strategy that limits damage and the number of plants affected would be necessary (Carlson et al. 2013). Comparing flock activity before and after drone exposure or comparing flock behavior in response to a natural predator versus a drone would provide foundational understanding of how drones affect behavior (Duttenhefner 2023).

We did not find that wind speed or Julian day affected flock behavioral responses. Ambient light was highly correlated with time of day and dropped from most models; thus, ambient light may have also contributed to a flock’s ability to detect a threat, affecting FID and abandonment. The drone was not flown in wind speeds higher than 7 m s⁻¹, which limited effects on drone speed and birds did not experience wind speeds that would have impeded flock movements. Wind speed may become more important when repellents are integrated on the hazing drone, given the impact on chemical spray drift. If we had not maintained a fairly constant drone speed, we likely would have seen shorter FIDs with an increased drone velocity similar to other mix-species bird flocks (Wilson et al. 2023). Julian day may not be impactful because of extreme climatological variation in the northern Great Plains during the trials, where direct weather measurements may be more informative.

Habitat can also have an impact on vigilance and alert to predators (Devereux et al. 2008; Jarrett et al. 2020). Studies evaluating FID have often looked at responses in open areas and distance to refugia habitat, such as nearby trees (Cooper and Blumstein 2015). Although birds can hide in both cattails and sunflower, cattails have denser leaf vegetation that could delay blackbird reactions by either reducing visibility or perceived risk, explaining why the drone could get closer to birds in cattail (Metcalfe 1984; Rodriguez-Prieto et al. 2008). Birds may perceive cattail as safer than sunflower, but detection of drones may also be delayed when birds are perched lower in the vegetation and occupying lower-lying parts of the field where cattail marshes are located. Although not statistically significant, flock reductions increased with the proportion of cattail, indicating that blackbirds were either harder to monitor in cattail or flocks sought refuge outside of the field complex on landscapes with abundant cattail. Avian repellents (e.g. methyl anthranilate) are applied to sunflower and not to areas with surface water. Although, we were able to closely approach flocks in cattail, future operations would need to cease spraying while moving birds out of cattail that is interspersed with sunflower.

Overall, 10 min of drone hazing in large sunflower fields was not likely to be sufficient to reduce crop damage, and future research should evaluate additional negative stimuli or the degree of continual hazing that is needed within a field to reduce yield loss. Smaller flocks are easier to move from a sunflower field, which highlights the importance of hazing early in the season prior to the formation of large flocks. Thus, early-season hazing, prior to sunflower maturity, may be beneficial to prevent establishment of foraging locations (Besser et al. 1979). Simultaneous damage estimates would benefit drone-hazing trials, because direct comparisons of tool efficacy would be possible. Drones are capable of aerially estimating damage in some scenarios (Barnas et al. 2019; Friesenhahn et al. 2023), but the structure of sunflower requires ground estimations of damage (Kaiser et al. 2021). Future research should evaluate increasing the duration or frequency of hazing, targeting the same fields over an entire season, or adding a negative stimulus, such as an avian repellent. If using a drone to apply an avian repellent, protocols should consider time of day, flock size, and habitat, but choice of flight path should mainly be concerned with optimizing spray drift to reach foraging blackbirds. There is promise for drone hazing in commercial agriculture, and additional research can use this study for a foundational understanding of the behavioral responses of blackbird flocks toward drones.