The Use of Multiple Unmanned Aircraft Systems as a Swarm to Release Sterile Mexican Fruit Fly (Diptera: Tephritidae) Into South Texas Citrus Groves

Abstract

Mexican fruit fly Anastrepha ludens (Loew) (Diptera: Tephritidae) is a key economic pest of citrus and represents a quarantine issue along the United States and Mexico Border. In order to respond to this threat, the United States Department of Agriculture produces approximately 175 million sterile Mexican fruit fly pupae per week and releases approximately 150 million adult flies per week via conventional fixed wing aircraft. Unmanned aircraft systems (UAS) offer a novel means of releasing sterile insects aerially, can be deployed on short notice in rapid response scenarios, require a small footprint to operate, and offer an alternative means to releasing sterile insects to traditional manned aircraft. UAS, however, are currently limited in two key areas, range and payload capacity. Swarm technology, flying multiple UAS at once, may increase the utility of UAS by distributing payloads and release patterns across multiple UAS. In order to test the viability of swarm technology in the release of sterile insects we conducted multiple mark release recapture experiments over south Texas citrus groves during 2017, 2018, and 2019. The results of this study demonstrate improved release rates from 89.9% (n = 5) of flies released with ca. 0.64% recapture during 2018, to 98.2% (n = 6) released with ca. 0.74% recapture during 2019. These results demonstrate that swarm technology is a viable technique for increasing aerial release capacity and flexibility of sterile insect technique (SIT) programs.

The sterile insect technique (SIT) is a pest control tactic where large numbers of insects are reared, sterilized, and then released over a cropping system where they compete for mating opportunities with wild insects. The SIT has been shown to be a highly effective bio-rational control tactic for a number of insect pests. Deploying sterile insects over large areas is one of the most challenging aspects of SIT (Tan and Tan 2013), and numerous methods have been employed attempting to optimize releases (Dyck et al. 2006). Historically, small aircraft or helicopters have been the only option for quickly releasing SIT over large areas (Vargas et al. 1995). Aerial release of fruit flies is currently the largest SIT effort in USDA-APHIS (International Panel for Review of Fruit Fly Surveillance Programs 2006). More than 90% of these releases use conventional aircraft (Andress et al. 2013) for deploying sterile flies, with supplemental ground based releases from truck-mounted air blast equipment (USDA-APHIS 2014) and emergence buckets (Barry et al. 2002). Deploying conventional manned aircraft for sterile insect release requires a large footprint of support operations, accounting for approximately 40% of total operational costs associated with sterile fruit fly programs (Tan and Tan 2013). Ground based releases (trucks) provide uneven distributions of sterile insects, with released populations tending to cluster near release sites (Dowell et al. 2005). In some cases, releasing three times as many flies actually reduced recapture of fruit flies across the treatment area (Salvato et al. 2003).

The use of unmanned aircraft systems (UAS) in agriculture is relatively new; however, these highly flexible tools can support a diverse range of agricultural applications (Puri et al. 2017, Kulbacki et al. 2018). Swarm UAS enables a single Remote Pilot in Command to operate multiple unmanned aircraft, flying in formation from a single Ground Control Station. While a single UAS is limited by the amount of weight it can lift (payload capacity), the operation of multiple UAS, in a swarm, offers a means to increase payload capacity by spreading the load, so as to approximate the volume of insects released by conventional aircraft. Releases done by UAS also have the potential to increase accuracy over conventional aircraft (Tan and Tan 2013) because they can fly at much lower altitudes, decreasing the potential for released insects to drift off target.

Mexican fruit fly is a major pest of citrus across Central America, Mexico, and the southern United States (Robinson and Hooper 1989, Thomas 2004) impacting more than 60 commercial crops. Female Mexican fruit fly lay eggs in clutches of one to 23, with over 100 eggs laid per day (Baker 1944, Berrigan 1988, Berrigan et al. 1988), and 800 to 1,200 eggs over an average 20–40 d lifespan (Liedo et al. 1992, Mangan 2003). This invasive tephritid species represents a quarantine issue (Jang et al. 2015) along the United States and Mexico border (Ruiz-Arce et al. 2015). If left unchecked, establishment of Mexican fruit fly within the United States could cost upwards of $900 million in losses annually (USDA-APHIS 2008). Since 1984, the fruit fly program in Texas has used fixed wing aircraft to release sterile Mexican fruit fly as part of a suppression program (Holler et al. 1984, Nilhake et al. 1991). The damage and impact to trade caused by Mexican fruit fly led to the initiation of an eradication program in 2006 (USDA-APHIS 2008).

Efforts to eradicate Mexican fruit fly from both Mexican and United States sides of the Rio Grande began in earnest during the 2006–2007 growing season. In an effort to eradicate this economically destructive pest, USDA APHIS conducts weekly aerial releases of over 1,000 sterile Mexican fruit fly per hectare over 12,000 hectares of citrus groves throughout the Lower Rio Grande Valley of Texas and during Mexican fruit fly outbreaks in California (USDA-APHIS 2010). Releasing this many flies over such a large area requires considerable logistical efforts.

Occasionally, weather conditions or smoke from wildfires can interfere with release schedules, further complicating release logistics. When cloud ceilings are below minimum elevation for fixed wing aircraft, or when conventional aircraft are grounded due to inclement weather or Visual Flight Rules limitations (14 CFR § 91.155, 2017), scheduled insect releases cannot be conducted. In these situations, UAS are capable of flying safely at low altitudes and may complete scheduled flight missions to disperse sterile Mexican fruit fly, provided compliance with Visual Flight Rules (14 CFR § 107.51(c) minimum flight visibility and (d) minimum distance from clouds). As wildfires are predicted to increase in the future (Liu et al. 2010), in part due to our changing climate, we can expect more frequent visual limitation events. Swarm UAS technology offers sterile insect release programs, like the Mexican fruit fly program, with the ability to deploy multiple UAS, thus increasing release capacity. These swarmed UAS could be added to release programs as part of their routine operations or used to increase flexibility of deployment and augment existing release techniques or to respond to unforeseen weather events.

Insects have been released by UAS for a number of years. In 2007, radio controlled aircraft were developed by the Brazilian based Moscamed program to release sterile Ceratitis capitata (Diptera:Tephritidae) (Cavalcanti et al. 2008). While these systems provided an alternative to manned release platforms, they lacked the necessary technologies to operate autonomously (Tan and Tan 2013). By 2015, autonomous use of UAS was achieved, using pink bollworm (Pectinophora gossypiella (Saunders) (Lepidoptera: Gelechiidae)) as the model insect (Moses-Gonzales et al. 2015). These autonomous flights were conducted as part of the, now successful, Pink Bollworm Eradication Program (Henneberry and Naranjo 1998, Walters et al. 2009, Purdue 2018). Since then, autonomous UAS have been used to release sterile Lepidoptera (Howard 2018, Paterson et al. 2019) and sterile mosquitoes (Chung et al. 2018, Bouyer et al. 2020).

Being at the leading edge of research means that the tools needed to accomplish new tasks have not yet been developed. This article describes our efforts to adapt and improve available equipment and technology to accomplish swarm release of Mexican fruit fly. Over three seasons of field experiments, with as many release devices, we have designed and tested new equipment and improved release strategies to increase the percent of insects released and the precision of their application. Here, we describe our failures and successes with the hope that future researchers can access this knowledge and shorten the learning curve for future SIT UAS projects.

The objectives of this research were to 1) assess efficacy (percent release) of current release devices, 2) develop and test new release devices as needed, 3) model and field-test spacing between swarmed UAS needed for uniform insect distribution, and 4) measure percent recapture of Mexican fruit fly in monitoring traps.

Materials and Methods

Sterile Insects

Sterile Mexican fruit fly adults were obtained from the USDA-APHIS Mexican fruit fly mass-rearing facility located at Moore Air Base in Edinburg, Texas. The colony, known as the ‘Willacy strain’, was started from an infested grapefruit with 12 larvae, collected in Willacy county TX, in 2008 (Thomas et al. 2014). Insects are reared on an artificial meridic diet adjusted from Spishakoff and Hernandez-Davila (1968) for use in the implementation of the Mexican fruit fly Preventative Release Program in south Texas citrus.

Marking Insects

Color markers were necessary to distinguish released sterile flies from wild flies, as well as to distinguish between different releases within a UAS swarm. To mark insects a number of pupae (1.5 kg) were placed inside bags containing yellow (Saturn Yellow), orange (Fire Orange), or green (Signal Green) fluorescent dyes (DAY-GLO, Cleveland, OH). The pupae were gently shaken in the bag until the dye was uniformly coated over all the pupae (Arredondo et al. 2017). Following the marking process, pupae, while still in the bags, were sterilized in a cesium irradiator (Husman Isomedix, Inc., Whippany, NJ) set at 70 Gy (Hallman 2013). Sterilized pupae were placed into emergence trays and allowed to emerge following the standard Mexican fruit fly Rearing Facility procedures (FAO/IAEA 2016). During emergence, the ptilinum of adult Mexican fruit fly is coated with colored dye particles as shown in Fig. 1. The ptilinum is an eversible pouch on the fly’s head, above the base of the antenna, used to rupture the end of the puparium. After emergence the ptilinum is retracted back into the head along with the colored powder. This marking technique allows adult flies to be later identified under ultraviolet illumination.

Insect Handling

Four- to five-day-old marked insects were collected from the emergence and release facility chill room, set at 3.3 ± 1°C. Chilled, dyed insects were loaded into release containers and weighed. The number of flies per container was estimated, by weight, by taking the mean weight of five samples of 100 flies. This mean weight per fly (ca. 0.016 g), divided by weight per loaded release device, was used to estimate the number of flies per release. Containers with chilled flies were placed into temperature-controlled coolers set at 7.0 ± 1°C for transport to the test fields. Once in the field, containers were removed from transport coolers and mounted under each designated UAS for the release mission. After release, the containers were returned to the laboratory to estimate the number of flies still remaining inside the release container, by weight.

Unmanned Aircraft System

Over the course of this study, two types of UAS were used. In 2017 and 2018, we used three Spreading Wings S1000 octocopter airframes (SZ DJI Technology Co., Ltd., Shenzhen, China) modified with an open source flight controller (Pixhawk 2.1, Hex Technologies, Xiamen, Fujian, China). This custom configuration is named the Hermes V.1 UAS (M3 Consulting Group LLC., Dayton, OH) (Fig. 2A). In 2019, we used three custom built hexacopter airframes with Pixhawk 2.1 flight controllers and custom mounting brackets named the Hermes V.2 UAS (M3 Consulting Group LLC.) (Fig. 2B). Both the Hermes V.1 and V.2 are considered small UAS (<55 lbs.) (14 CFR § 107.3, 2017). Individual aircraft are identified by the last three alpha numeric characters of their FAA-UAS registration numbers. Table 1 contains registration numbers and further aircraft specifications.

Release Devices

Due to necessary innovation, distinct release mechanisms were developed and assessed in each year of these experiments. The first year featured a device we designed (M3 Consulting Group LLC.) to release codling moth (Cydia pomonella (L.) (Lepidoptera: Tortricidae)) which relied upon gravity and a dilating aperture to release insects (Fig. 3A). The second year we designed a release device that used paddlewheel design to expel insects (Fig. 3B). The third year we designed a conveyor belt mechanism underneath a straight walled insect compartment to meter and expel insects (Fig. 3C). Release device specifications are given in Table 2.

Swarmed Unmanned Aircraft Systems

The operation of multiple aircraft by a single person is prohibited under 14 CFR § Part 107.35 (2017). To overcome this prohibition, a waiver request was submitted to the Federal Aviation Administration (FAA) and approval was received to operate multiple UAS at three sites near Mission, Texas (14 CFR § 107.200, 2017; FAA 2017). The waiver was submitted in July 2017 and approval was granted in November 2017 for a turnaround time of 5 mo.

Multiple UAS were placed into a predefined formation, powered one at a time to establish a telemetry link and loaded with preprogrammed flight missions for the particular location in the formation (Fig. 2B). Once preflight inspections and individual flight mission loading was complete, the Remote Pilot in Command authorized all of the aircraft to be powered and for the insects to be loaded onto the aircraft.

To execute the swarm mission, the Remote Pilot in Command manually armed the three UAS used for the swarm and commanded mission start to launch the UAS automatically at predefined intervals. The UAS were launched according to their position in the formation. All missions were executed autonomously from takeoff, through release, and landing.

Field Sites

The same field sites where used in all 3 yr. Two fields within the citrus groves were selected near McCook, TX. Field #1 was located 9 km northeast of the Mexican Fruit Fly Rearing Facility on Moore Airbase, Edinburg, TX. Field #2 was located 8.5 km directly east of Moore Airbase. The two release fields were separated in a north to south direction by 5.4 km. Both fields were approximately 57 ha in size. Each field contained a mixture of ca. 80% Rio Red grapefruit, Citrus paradisi Macfadyen, and ca. 20% navel orange, Citrus sinensis (L.) Osbeck. Tree densities ranged from ca. 321 trees per hectare (ca. 130 trees per acre) for the 4 × 8 m spacing grapefruit, to ca. 375 trees per ha (ca. 152 trees per acre) for the 3 × 7 m spaced orange trees.

Releases

In 2017, releases were done with a single UAS flown in three passes for each citrus field. In 2018, releases were done with one flight mission of three UAS in a swarm for each of the two fields, with a total of ca. 108,000 flies released. In 2019, releases were also done with one flight mission of three UAS in a swarm for each of the two fields, with a total of ca. 120,000 flies released. Swarmed UAS were staged outside the citrus fields with the center UAS placed at row ‘0’ (Fig. 4), with additional UAS placed 10 m East and 10 m West of the central UAS. All UAS flew at 120 m above ground level, at 5 m/s groundspeed. Releases were conducted mid-morning between 9 and 11 am local time. Meteorological data for each release are given in Table 3. In all releases, devices were loaded with 500 g of sterile Mexican fruit fly. To release flies, UAS flew a single straight flight down the rows, releasing their cargo of flies at a right angle to the trap lines (A–F) (Fig. 4). After a mission had been accomplished, UAS automatically returned to the take-off point. Unreleased flies remaining in release devices were collected and weighed.

Trapping

Multi-lure traps (Better World Manufacturing Inc., Miami, FL) baited with a two-component (putrescine and ammonia acetate) lure (Scentry Biologicals, Inc., Billings, MT) and 300 ml of a 10% propylene glycol drowning solution were placed in a grid of 174 traps across each citrus field (Fig. 4). The grid consisted of six lines of traps labeled A–F (Fig. 4), at a right angle to the flight path, and 29 rows labeled 0–14 E and 0–14 W (Fig. 4). Row zero was considered the centerline for releases. Trapping data was collected at 1, 3, and 6 d after UAS release. Each UAS release device contained sterile flies marked with a unique color of dye. Captured flies were inspected under ultraviolet illumination to determine color of internal marker. The number of flies captured per individual trap was recorded by date.

Modeling Distribution

Dispersal parameters were determined by modeling (R Core Team 2019) the distribution of Mexican fruit fly perpendicular to the flight path with a normal curve (equation 1), where x is the distance from the center line, f(x) is the number of flies released, μ is the mean of the distribution, σ² is the variance of the distribution, and A is a conversion coefficient.

For analysis, all catch data were adjusted to the center line (0 m). The total µ is defined as the total width of discharge measured from leftmost to rightmost deposit where a minimum of one insect is recaptured (ASAE S327.4, 2012). Mean distance from the target centerline to the center of measured swath was calculated as mean distance off target (Table 4).

Results

In 2017, the gravity fed release devise did not effectively release flies during normal flight operations. In 2018, the paddle wheel release device released 72 ± 16% (n = 5) of the Mexican fruit fly load. In 2019, the conveyor belt design released 98 ± 3% (n = 6) of the Mexican fruit fly load (Fig. 5).

As we only planned to utilize release devices that proved to be more than 90% effective, we did not report swath width results from the gravity feed and paddle wheel devices. Trap capture data were concentrated toward the Northern end of the field (Fig. 6A and B). All releases were offset 70 ± 2 m east of their centerline flight path due to daily wind events (Table 4). In 2019, field #2 received a pesticide application between SIT release and data collection. The average total swath width of conveyor belt release device based on the model distributions was 196 ± 11 m (n = 3). For comparison, the total swath width of releases from manned aircraft is 268 m (FAO/IAEA 2016). Recapture rates for released flies in 2018 and 2019 were 0.64% and 0.74%, respectively.

Discussion

This article provides information on the research and development of new UAS technology that may enhance Mexican fruit fly SIT application. We discuss release devices that fail to perform and UASs that crash. We feel that it is important to share successes, as well as failures so that other researchers might learn from our mistakes and shorten the learning curves of similar technology applications. As other researchers attempt to bridge the technology gap in agriculture, there will likely be long learning curves. We hope that this publication can inform future researchers and expedite agricultural innovation.

In 2017, we began releasing Mexican fruit fly using a gravity feed release devise (Fig. 3A) mounted under the Hermes V.1 octocopter, and developing the ability to fly multiple UAS in a swarm. The gravity feed release device was developed for releasing codling moth SIT, and consistently meters out 2,000 moths per ha over 16 ha in under 5 min. This device works extremely well with codling moth as these insects pull in their legs during cold transport and release. Adult Mexican fruit fly behaves differently inside the release device, flies cling to the sides of the container as well as to one another, even in torpor. This behavior resulted in insects remaining in the release device in such high numbers (>50%) that it was determined that the device could not be used in future Mexican fruit fly releases. While we were able to dislodge the flies by shaking the aircraft in flight, manually forcing the UAS to pitch and roll was not a practical solution and so a new release device had to be developed. As we would not be able to use this device in future applications, the percent release and swath width data are not reported here. While the release device was not successful, we were able to successfully develop swarm flight of UAS in 2017.

In 2018, we designed a new release device that used a paddle wheel to eject Mexican fruit fly from the holding compartment (Fig. 3B). During bench top tests, some flies remained in the holding compartment, but it was believed that the vibrations during flight would dislodge most of the flies. The paddle wheel design was an improvement over the gravity fed device releasing 72 ± 16% (n = 5) (Fig. 5) of the Mexican fruit fly load, under normal flight conditions. The Hermes V.1 octocopters were flown in swarms of three in each of the two separate fields. During the releases, one UAS experienced a malfunction and crashed, reducing the replication to n = 5. While the paddle wheel device was an improvement over the gravity fed, device the slanted walls allowed 28% (over 30,000 on average) of the flies to remain inside the container and not be released. This forced us to redesign the release devise yet again.

In 2019, we designed and built a completely new release device and insect container to overcome the insects clinging behavior (Fig. 3C). Straight walls prevented the insects from clinging to the inside of the container, and a moving conveyer belt gently released insects at an even rate. Insects clinging to the conveyer belt were ejected, by gravity and wind, as the belt carried them to the underside of the release devise. The additional weight of the release device necessitated a UAS with more payload capacity. The Hermes V.2 was built to accommodate the new release device and up to 1 kg of flies. This final UAS and release device combination successfully completed all missions and released 98% of the flies. Release procedure, from start to finish took ca. 10 min.

Understanding the optimal spacing between UAS was a key objective of this research. In these releases, the swarmed UAS were spaced at only 10 m apart in their preprogrammed flight paths. However, this is not the most efficient spacing for evenly dispensing flies. By extending the lane-spacing further, a broader dispersal pattern can be achieved. One of our objectives was to determine the swath width of released flies so that we can optimize coverage over target orchards. Figure 7 illustrates the magnitude of catch and dispersive distance (total swath width) of released sterile flies using the conveyer belt release device. When designing a release strategy for a swarm of UAS, it is important to space aircraft close enough to not leave areas of the grove unprotected, and also not produce unnecessary overlap, and waste resources. Based on these data a spacing of 150 m (Fig. 7B) between aircraft would produce the most uniform insect distribution. Future research will explore coverage at these distances.

Accuracy of targeted application is a key objective of sterile insect release. Current UAS flight missions are flown at 120 m above ground level. While this altitude is lower than fixed wing piloted aircraft, it was still subject to some drift from prevailing winds. We saw that catch was pushed to the northern end of the block and blown off center by about 70 m on average (Fig. 6A and B). At the time of release, wind speeds at these sites were between 16 and 37 kph, with temperatures ranging from 10 to 28°C (Table 3). Temperature and wind speed, at time of release, will have impacts on accuracy of release and insect dispersal after release. Future missions can be adjusted to account for prevailing winds and potentially flown at lower altitudes to minimize drift. Adjustments to flight altitude will likely impact swath width. Finding the balance between accuracy of placement and total coverage area will require further research.

Distribution of flies was measured by the rate of recapture. The recapture rates of all these releases were extremely low, at less than 1% total recapture after 6 d. These low recapture rates are not completely unexpected and have been described in other trapping research (Miller et al. 2015, Adams et al. 2017). This is an important consideration for researchers to keep in mind when designing experiments. Understanding that catch of released insects will be less than 1% will help determine the amount of insects necessary to produce useful recapture data. Bait traps were used to estimate the distance insects dispersed after release. Sterile Mexican fruit fly are known to be more sedentary than wild-type flies, and dispersive distance did not increase from day 1 catch data to day 6 catch data. This makes uniform application of sterile Mexican fruit fly a critical objective of the program. Additional UAS flown at optimized heights and adjusted for wind speed and direction will increase precision of application and provide flexibility for SIT application.

Handling, transport, and release of SIT have been the target of much research (FAO/IAEA 2007). Using the Hermes V.2 UAS and conveyor belt release device, total handling time, from cooler to release, was ca. 10 min, compared to ca. 60 min (or more) in small aircraft. Insects released by UAS spend less time in release devices and experience less compaction because they are transported in smaller volumes. Storage conditions and compaction have been shown to directly impact survival of insects in SIT release programs (Chung et al. 2018). While direct measure of survivorship by release method is needed, we believe there are advantages to reducing impacts of mass storage and time in release devices. Future work should attempt to quantify these effects.

UAS provide an alternative means of releasing sterile insects into SIT treatment areas and can be used when small aircraft are unable to fly, such as when wildfire smoke or low cloud ceilings preclude piloted missions, or as a targeted rapid response tool in response to localized outbreaks. Swarmed UAS show significant promise in targeted applications of SIT, with multiple UAS able to distribute the payload and increase the swath width of insect applications. With appropriate adjustments in altitude and for prevailing winds, UAS have the ability to increase accuracy and improve coverage. UAS can be deployed as an effective aerial release platform for SIT programs, operating from a much smaller infrastructure footprint with lower overall costs. Shorter handling time and reduced compaction should improve the quality of delivered insects (FAO/IAEA/USDA 2003), which is critical for the success of SIT programs (Dyck et al. 2006).

Through the development of new tools and technology, we hope to provide the Mexican fruit fly SIT program with more release options and greater flexibility to respond to future challenges and augment the current eradication program. We hope that other researchers attempting to bridge the technology gap in agriculture will benefit from this description of our learning curve.

Disclaimer: The use of trade, firm, or corporation names in this publication is for the information and convenience of the reader. Such use does not constitute an official endorsement or approval by the United States Department of Agriculture of any product or service to the exclusion of others that may be suitable. The findings and conclusions in this publication are those of the author(s) and should not be construed to represent any official USDA or U.S. Government determination or policy.

Acknowledgments

We thank Daryl Hill for coordinating flights, Elma (Josie) Salinas, Jason Carlson, Kari Piña, Cynthia Garcia, Danny Martinez, Armando Loya, Eric Shah, Christian Muñoz, Corina Zamora, Ruth Galan, and Johnny Rodriguez for their time and effort in setting up, gathering data, and assisting with this test. We wish to acknowledge Dr. Michelle Walters who championed the development of UAS for SIT, and provided the guidance and vision for this project.

This material was made possible, in part, by a Cooperative Agreement (AP18PPQS&T0C032: ‘Use of Unmanned Aircraft System (UAS) Swarm Technology for Sterile Insect Technique Release, Survey, and Treatment’) from the United States Department of Agriculture’s Animal and Plant Health Inspection Service (APHIS). It may not necessarily express APHIS’ views.

References Cited