Executive Summary

Partnering with USDA and local farmers in NY, NJ, and PA areas, the different trap experiments of capturing the Spotted Lanternfly (SLF) can help formulate guidelines of SLF control and future invasive organisms for local farms to mitigate the potential crop loss. The research plan, recommendation of traps, statistical data, and simulation are introduced in this report.

The increased sightings of SLF on local farms alerts the authorities to take immediate actions to resolve the emergence of the harm SLF can do to important food resources. Although there are many traps that can help to reduce the SLF invasion into local farms, there is limited research providing insight into the effectiveness of various traps which could lead to weak SLF invasion responses and the misallocation of resources that may present financial and crop yield damages. The research proposed in the report helps to determine the effectiveness of different traps for reducing the SLF invasion through both physical deterrents and traditional deterrents containment measures. The data in this research will be collected from participating USDA-approved farms who volunteer to test an assigned SLF containment method. Assignment is based on the farm’s characteristics as to best approach one of the methods including SLF trained dogs, predator consumption through chickens, BugBarrier tree trap, and Tree of Heaven insecticide systemic trap, as well as the control group with traditional insecticides as the performance benchmark. The study period will occur from May to October 2023 during the crop growing season that coincides with the high SLF activity period. Through a one way ANOVA, Tukey HSD test under our study’s conditions with simulations, we were also able to model potential outcomes for when the study is actually conducted in the coming year.

The research helps to set an effective SLF control plan for the local farmers to take immediate action to stop the invasion of SLF and minimize the crop loss, which solves the lack of SLF trap effectiveness studies and gives out the fast solution for SLF invasion based on physical deterrents and traditional deterrents methods. The research might not include all the possible traps for controlling the SLF, however, it can be a guideline for future SLF trap development and capture improvement.

As the SLF is continuously invading the local farms and expanding to new areas of the farms and lands, a fast and effective plan for immediate response to the invasion is needed. The research can provide actionable and quick solutions for this problem.

Key words: Spotted Lanternfly, Organism Invasion, SLF, SLF traps, SLF control, Infestation

Statement of the Problem

Spotted Lanternfly (SLF) is an invasive insect that was first detected in Pennsylvania in 2014 and is able to spread quickly through human activity, like transportation, due to the insects’ ability to lay eggs in various environments (Urban, 2020, p. 10). Since its detection, the SLF threatened many farms and natural ecosystems across North America; with at least 70 known species of hosts and an estimated 255 candidates, without increased intervention the SLF will easily travel westward toward the breadbasket states (Huron & Helmu, 2022, pp. 3-15). Researchers predict by 2033, SLF will be notably present towards California; moreso, though SLF’s preferred host is Tree of Heaven, agricultural produce like grapes, apples, almonds, and cherries are also at-risk (Jones et. al., 2022, pp. 1-3). Between 2014 and 2018, the insect has contributed to 25 millions acres of forest loss in the U.S. (Huron & Helmu, 2022, p.3). With containment efforts implemented in high infestation areas as late as this year, time is short to control the SLF before serious agricultural and economic consequences unfold. Although there are methods of reducing the SLF with traditional insecticides, there is little research focused on the cost of and effectiveness of applying alternative control methods as viewed by the farmers.

Research Questions, Hypotheses, and Effect Size

In this study, our team looks to answer the following:

  1. Do physical deterrents–introducing trained dogs and chickens–and traps–BugBarrier tree tape and tree systematic treatment–reduce SLF populations more than traditional insecticide use at tri-state areas farms producing grapes, apples, almonds, and cherries?

  2. Measuring perceived SLF population control, do farmers believe at least one of the other methods for SLF containment outperforms traditional insecticide?

  3. How do the participating farms’ crop yield compare to the traditional insecticide use when other SLF containment methods are substituted?

The first research question will measure the number of SLF within a farm, and the second research question will measure crop yield and a likert opinion survey. The following hypotheses frame our study’s direction:

  1. Research Question 1:

H0: Compared to traditional insecticide use, other methods of SLF containment will not be as effective in reducing the SLF populations found on farms.

HA: At least one of the other methods will reduce SLF counts on participating farms better than traditional insecticide use.

  1. Research Question 2:

H0: Compared to traditional insecticides use, farmers using other methods for SLF containment will not experience an improvement in crop yields.

HA: At least one of the other methods for SLF population control will be better than traditional insecticide use in terms of improving crop yields.

  1. Research Question 3:

H0: Compared to traditional insecticides use, farmers using other methods for SLF containment will not perceive an improvement in crop yields.

HA: At least one of the other methods for SLF population control will be better than traditional insecticide use in terms of improving perceived containment efforts.

In terms of ideal effect sizes, our team is looking for a moderate cohen’s f of 0.20 in terms of decreasing SLF activity by land coverage on participating farms. With the improvement of crop yields and farmers’ surveyed sentiments about SLF infestation, our team can expect to see success with effect sizes measured by Cohen’s f of 0.20 and 0.15, respectively. See the Sample Size, Effect Size, and Statistical Power under the Statistical Analysis Plan section for further details on the effect size.

Importance of the Study and Social Impact

Our study will help provide references for the USDA to formulate guidance on SLF control and other future invasive organisms. At present, the methods used by farms and the official recommendations for SLF control are proposed vertically, and there are few studies on the effectiveness of different methods. Our study horizontally compares several effective methods of SLF control, which can provide research evidence support for the organization to introduce effective and low-cost SLF control guidance in the future. On the other hand, it can also help the organization to respond quickly in official guidance when invasive species similar to SLF appear in the future.

Moreover, our study may not only provide research support for agricultural control of SLF, but also help prevent SLF from invading cities. For example, once SLF control actions are taken in the farms and rural areas around New York City, the city including Columbia campus, may be affected by SLF due to the trickle down effect. Through our study we can find a lower cost and effective SLF control method, which can save time and manpower research costs for urban prevention and rapid response to SLF.

Literature Review

Scholarly articles and research were pooled to assist in our study’s design and establish foundational understandings of SLF infestation in the US. Investigating current insect control practices in the agriculture industry, of the broad-spectrum insecticide classes–pyrethroid, neonicotinoid, and organophosphate–Leach et. al. in her 2019 study identified “the organophosphate chlorpyrifos provided 100% mortality of egg masses’’ and exposure among among nymph and adult stage SLF caused slowed activity; alternatively, “only neonicotinoid thiamethoxam and the pyrethroid bifenthrin… were able to offer residual mortality” with the nymph and adult stage SLF (pp. 4-5). Connecting such findings, Coyle et. al. propose using a “trap tree” management approach to contain the SLF; they noted that the Tree of Heaven was the preferred host of the insect, and assumed the tree’s presence could redirect SLF activity (2019, p. 5). Through their approach, systemic insecticide application would occur, involving coating the Tree of Heaven with insecticide while also separating the tree male and female variants due to their invasiveness.

As Leach et. al. note, the nymph and adult stage SLF threat not only impacts plants due to their feeding habits but also due to their honeydew liquid bi-product that can cause a build up of photosynthesis-inhibiting mold, especially on fruit plants (2019, p. 1). To mitigate on-tree SLF activity, Francese et. al. study the effectiveness of various bug traps; comparing tree tape brands, they found the highest catch rate with BugBarrier instead of Web-Cote for catching nymphs and early adult stages during the trapping period (2020, pp. 273-274). However, compared to BugBarrier, a circle trunk trap “was significantly more effective at capturing… fourth-instar nymphs… (Francese et. al., 2020, pp. 273-274). Considering our own research project, while all SLF are targeted, containing the population in early stages may promote long-lasting management.

Continuing with natural predators, a study by Virginia Tech discovered that trained dogs could be beneficial detection tools for targeting SLF. In her proof-of-concept study, Decker trained six pet dogs to detect the SLF through a training phase and series of five tests. In the five tests, the mean sensitivity of six dogs was nearly 80% and the mean positive predictive power was around 67% (Decker, 2021 p. 35). The results showed that pet dogs’ scent detection capability is unparalleled. Considering the strong odor of SLF honeydew bi-product, a dog’s detection only strengthens. Interestingly, with odor training Essler et. al. noted that trained dogs can also find live or dead SLF egg masses while ignoring relevant study controls with high sensitivity and specificity, such as tree bark, equipment, and plastics (2021, pp. 1-7). Based on these findings, the potential of dogs as detection tools to combat the spread of invasive species or agricultural threats can be expanded. In our research, comparing untrained chickens to trained dogs can provide greater insight into the value of training efforts and predator choice by a farm.

Considering much of the Northeastern US is currently battling the SLF infestation, our study’s goal is to assist in early quarantining and prevention of SLF migration throughout the rest of the country. Without effective preventative action, Jones et. al. estimate SLF migration could reach California and the West coast as early as 2030, and they also used models to estimate the probability of when SLF populations will impact major crop production; among these, grapes, almonds, apples, and cherries possess the highest economic value with approximately 6 billion, 5.5 billion, 3.5 billion, and 1 billion, USD respectively (Jones, 2022, pp. 2-3). According to the USDA, there were a total of 52,700 farms in Pennsylvania, 33,400 in New York and 9,900 in New Jersey as of 2021. This information will be used in gathering participants for our study and illustrate which farms should immediately consider SLF mitigation efforts.

Research Plan

Sample Size and Selection

We plan to gather participating farms through the USDA, as the department oversees farming practices and our study should adhere to such oversight. Once those farms combatting SLF are recruited as potential participants, they will be asked to complete a quality control survey (see Appendix A) that will determine their participation and random assignment to a treatment or control for which they likely qualify. For this experiment, we will be involving the high at-risk, high economic value grapes, almonds, apples or cherries farms, specifically in New Jersey, New York, and Pennsylvania. Because we leave much of the sample selection and participation approval to the USDA, we may pool a large number of farms, therefore to minimize the scale of the study, we will have a total of 50 farms–10 in each study group.

Independent Variables

The research design is structured by comparing SLF control methods. Participating farms will be treated to either of the five methods of containment: introducing trained dogs, chicken, BugBarrier tree trap, Tree of Heaven insecticide systemic trap, and the control group of traditional insecticide use. The control group will use organophosphate insecticide as it can provide 100% egg mass mortality and is already part of many farming practices, cited in the literature review.

Building from the Sample Size and Selection subsection, for the trap approach with systemic insecticide application, participating farms will include those with Trees of Heaven on or near the farmland. If at least 10 farms contain a sizable number of Trees of Heaven, covering about 20% of the surrounding crop production area, this will be sufficient to assess the treatment variant. For all other treatment variants and the control group, random assignment will occur based on the Appendix A survey. For the dog treatment group, we would like there to be 2 dogs or more per farm for SLF training. For the chicken treatment group, 4 chickens or more would be needed.

Dependent Variables

To measure the effectiveness of this study and the application of various SLF population treatment methods, we will use the count of SLF on participating farms–SLF count is gathered as described in the Data Collection section–and through survey, see Appendix B, farmers’ sentiments are measured through a likert scale and self-reported crop yield figures. Crop yield measured by the amount of crop harvested per unit of land area, usually acres.

Procedures

The experiment will be conducted between May and October in 2023 as the growing and harvest season of the subject crops are during this time. This section will outline the general procedures that will be followed during the study.

For the control group, we will have farmers incorporate organophosphate chlorpyrifos-based insecticides into their current practices if the farm’s primary insecticide product does not contain this compound. For the period of the study, these farmers do not need to do much more than this application. We anticipate wherever there is high insect activity, which includes SLF, the farmer will apply as usual.

With the BugBarrier Tape, we will instruct farmers to wrap trees that are near SLF and insect activity. Tape placement and replacement will occur at the farmer’s discretion, as we want to see how the farmer will approach this treatment application in their own practice. In the participant survey, found in Appendix B, we will ask these farmers about the frequency of tape use to also measure cost; a 10 ft roll of BugBarrier Tape can cost as much as $70 (“Bug barrier,” n.d.).

The second treatment is also a trapping method but exploits another behavior pattern of SLF. From the research, we know the Tree of Heaven is the preferred host of SLF; by removing and destroying most female Trees of Heaven and leaving only a few male trees as trap trees, we can then inject the same organophosphate chlorpyrifos insecticide into these trees. This method targets SLF when they ingest the chemical compound that is secreted with the tree’s sap. Additional Trees of Heaven will not be introduced to farmland due to their invasiveness. Application of the systemic trap will occur twice a week.

The third treatment involves training dogs to detect SLF and their eggs as a means of prevention. During the overwintering period, adults of lanternflies will disappear based on the temperature, and dogs can search out live egg masses, which is an opportunity for quarantine and management of the spread of lanternflies (Essler et al., 2021 p. 5). Under the ideal study conditions with 2 dogs, or more if available, per farm for SLF detection, once trained, those dogs will be released every other day in the morning at sunrise for three hours to search for SLF hotspots. Sunrise time was selected in consideration of the temperature behavior of SLF. We will follow the same training methodology as Collins, Essler et. al., used in their study, and we will do a total of 10 training sessions in 2 weeks on the farms before the study period, with continued but less rigorous training afterwards with instructions for the farmers to follow. From that study, dogs were odor trained with live SLF eggs masses and dead egg masses that were ensured deceased according to the USDA standards, frozen at -80 degrees celsius for 96 hours, and stored in a refrigerator between 0 to 4 degrees celsius until use for training; note, the training used the Japanese flowering cherry as a control scent for dogs (Essler et. al., 2021 pp. 5-6). Once SLF hotspots are detected by the trained dogs, we will leave it up to the discretion of the farmers to contain or eliminate the SLF hotspot.

The last treatment, a physical deterrent, is replacing dogs with chickens considering they spontaneously feed on SLF or their egg masses. We will involve 4, or more if available, chickens being released in the farming area with barriers set up so they do not escape from the farm area. Because most of the crops are not grown at ground level, the chicken should not pose a threat to the crop, with the exception of grapes, depending on the raising technique. Release frequency will follow the same as the trained dog treatment, every other day at sunrise for three hours.

Brief Schedule

Our 10-month project will be ideally scheduled including two main parts. 4 months are set up for our project preparation, and the rest of 6 months would be for research monitoring and performance review. We plan to finish our preparation on time, which is in our control. When the project enters the practical operation stage, although we hope to complete it on time, our study may also be slightly adjusted due to weather and other factors. If there is any adjustment, all details will be completely recorded for the final performance review.

knitr::include_graphics("/Users/jacobbrandt/Desktop/Gantt chart.png")

Data Collection

In the data collection process, the SLF population data is collected in a suitable weather condition between May and October 2023 since a cold temperature (e.g., below -12.72 °C) would cause a high mortality rate of the SLF population (Lee et al., 2019, p. 591). The time frame precludes the bias factor, weather, which also decreases the population of SLF. Then, we adopt a harmonic radar tracking method to assess the spatial population of SLF before and after the treatment application. Our radar approach sends a light microwave signal to the harmonic tagged sample, which the stimulus signal will radiate back to and be recorded in our receiving system. The lightweight radar also provides length, direction, and straightness information of tracking routes from the migration of SLF. This advanced tracking system is noticeably improved to capture the size of small insects without impacting their survival abilities under field conditions (Jung et al., 2016, p.47). For our study, we will use the SLF activity by land coverage measured by the radar as a proxy for the SLF population. Data collection for research question 2 taken by field survey amongst participating farmers in the tri-state area during the same period, see Appendix B. Research question 3 will be asked at the end of the study period. Both data collection methods for research questions 1 and 2 will be done biweekly to track progress. The questions in the survey include asking farmers’ perception of initial and post-experiment crop yields and the response be implemented through a likert scale. By collecting and analyzing the relevant data from our radar approach and field survey, the result can tangibly provide the research outcomes.

Data Securities

As the research is cooperated with USDA, hence the collected data securities will follow the cyber security regulations and be stored under the computing system of the U.S. Department of Agriculture. The information in this research would be maintained and protected in USDA’s system and network to prevent third parties or unrelated staff from attacking or utilizing the data without authorization. Apart from avoiding potential cyber threats, the data will also be regularly taken a complete assessment and authorization test through the risk management framework (RMF), which could protect data from vulnerabilities and incompleteness in the long-term. In addition, approval and certification of every trap use will be monitored by USDA to ensure the data collected from the research is valid. Training in trap placement and usage will be available for all experimental farms to improve trap placement effectiveness, and precise the usage of all traps to avoid potential environmental harm from affecting other important species in experimental farm areas. All the records of the experiment will be kept confidential and in our cyber security system, including information of participants.

Statistical Analysis Plan

To analyze our data, we will need to use three one way ANOVA tests per measure of the dependentment variables. We will use R to run a preliminary estimate to establish expectations.

Sample Size, Effect Size, and Statistical Power

When considering the time, scale, and monetary costs required to perform this study, we knew there would be a challenge in measuring the statistical performance of our analysis. To combat the scaling issue, we opted for 10 farms per participation group.

For our dependent variables, considering data will be collected biweekly for 6 months, by the end of the study, we will have gathered 12 data points per farm, totalling to 120 data points across all farms.

#SLF Land Coverage Statistical Power
#pwr.anova.test(k = 5,
#               n = 12 * 10,
#               f = 0.20,
#               sig.level = 0.05)

Under the one way ANOVA test with our studies conditions, we can approach a statistical power of 0.9856568 if we look for a moderate effect size of 0.2 for Cohen’s f variable. This means there is a 98.6% probability that our null hypothesis can be rejected when it is false, to say that at least one of the SLF treatment methods will outperform the standard insecticide application control group for reducing the land coverage activity of the SLF, which is a proxy we will use for the population of SLF of a farm.

Due to crop yield being only measured once in this study following the harvest season with only one measure provided per farm, confirming our study may find it difficult to confirm a relationship between SLF population containment efforts and potential crop yield improvements. This is because 10 data points per group produces a statistical power of 0.1579543, if we seek a moderate effect size of 0.2. Seeking a larger effect size, like 0.4, improves statistical power to 0.5540384.

#Crop Yield Statistical Power
#pwr.anova.test(k = 5,
#               n = 1 * 10,
#               f = 0.20,
#               sig.level = 0.05)

With the opinion survey, measured on the likert 1-5 scale, we are seeking an effect size moderately-small effect size with Cohen’s f-variable set to 0.15. This produces a 0.8491183 statistical power, meaning the likelihood of rejecting the null hypothesis for any one of the survey questions is 84.9%, comparing the control group farms to the treatment groups farms.

#Opinion Survey Statistical Power
#pwr.anova.test(k = 5,
#               n = 10 * 12,
#               f = 0.15,
#               sig.level = 0.05)

Possible Recommendations

For Research Question 1, if the null hypothesis is not rejected, it would indicate that other SLF containment methods will not be as effective in reducing the SLF populations compared to traditional insecticide use. In this case, the best and most practical approach for farms would be to continue their pest control practices as normal. If the null hypothesis is rejected, then there would be at least one treatment method that is more effective than traditional insecticide use. Therefore, based on the data findings, the most effective treatment will be recommended and extended to more farms across the country as to prevent further spread of the SLF.

For Research Question 2, if the null hypothesis is not rejected, it would indicate that farmers cannot expect to experience an improvement in crop yields when using other SLF containment methods compared to traditional insecticide use. Specifically, using traditional insecticide would be the most practical method to control SLF. On the other hand, if the null hypothesis is rejected, it will conclude that there would be at least one of the other treatment methods to be more effective than the traditional insecticide method in terms of improving crop yields.

For Research Question 3, if the null hypothesis is not rejected, then it could be inferred that farmers do not perceive an improvement in either crop yield or SLF containment when using other SLF containment methods compared to traditional insecticide use. That is to say, even though maybe one method could be found to significantly reduce SLF populations on farms or that there is improvement in crop yield, maybe a farmer does not view or perceive this to be so. Hence, if the null hypothesis is rejected, it can be concluded that at least one of the other containment methods made farmers perceive there was an impact from applying those methods.

Limitations & Uncertainties

The study may contain limitations in sample selection and treatment groups. Firstly, the Tree of Heaven could limit the representation of lanternfly population size. Urban (2019) argues that even though the Tree of Heaven is the preferred host plant for lanternfly, its preference could be assemblage of trees anywhere (p. 11). Second, the treatment group such as dog and chicken deterrents could be costly for SLF control; for instance, dog training and the chicken method may be a double edged sword solution because while also reducing SLF populations, the chickens may contribute to crop yield loss if they are not monitored when near low-ground growing crops such as grapes. Across all treatment groups, because we left treatment execution up to the individual farmers, there may be a difference between the instructions we provided and the actual practice of those instructions; this limitation is purposeful in limiting researcher involvement with the participants. In terms of data collection, we also are presented with a limitation in terms of crop yield data points; while we have 120 potential data points for the other dependent variables, percent land coverage of SLF activity and likert scale opinion survey which are measured bi-weekly, due to seasonality of harvesting, we are only gathering the crop yield data once, providing one data point per farm. As exhibited in our statistical power test, this limits our capability to reject the null hypothesis, which we are comparing the treatment groups to the control.

Simulations

In the following simulations, we ran simulations on each dependent variable to provide an illustration of likely results that can help the design process, validate our models, and benchmark performance. To be able to conduct an analysis on the ANOVA test and to run the simulation analysis function, we used Tukey’s test for a multiple comparison procedure.

Assumptions in the Simulation Design

Scenario: No Effect on SLF Land Coverage

When simulating no effect of the treatments compared to the control in terms of which method is better for reducing SLF activity on farmlands, we assumed that the control group of farms would experience an average of 10% SLF land activity whereas the treatments will either provide equal performance or underperform the control group, we used 10% SLF land activity, again.

Scenario: Effect on SLF Land Coverage

In this scenario, continuing to use the 10% control benchmark, we built the simulated dataset assuming the mean of treatment groups SLF land coverage was 7%.

Scenario: No Effect on Crop Yield

According to the 2021 Crop Production report, published by the USDA in January 2022, the average crop yield during the year was 177 bushels per acre. In the simulated model, we entered this value as a proxy for what the control group’s crop yield may look like when we conduct our experiment. In the no effect scenario, we can assume that the treatment groups will perform either equal to or less than this amount, so we set the mean of our treatments to 177 in the construction of the simulated dataset.

Scenario: Effect on Crop Yield

Continuing from the no effect crop yield scenario, the treatment groups mean value for creation of the simulated dataset was changed to an improved performance of 220 bushels per acre.

Scenario: No Effect on Farmers’ Opinion

For the survey questions, because we have multiple questions being addressed here, we will focus on simulating data for Appendix B General Q3 regarding perceived improvement of crop yield. On a likert scale with a range of 1 to 5, we set the mean control to 3, and did the same for mean treatments in this scenario of no effect.

Scenario: Effect on Farmers’ Opinion

From the previou scenario, we change the mean treatment survey score to 4 instead of 3.

knitr::include_graphics("/Users/jacobbrandt/Desktop/Simulation.png")

Notes on simulation outcome: The values presented are the means of each comparison, not specifically tracking the control to treatment group. Therefore, while there may be differences in comparisons specific to the control, our output will not reflect this, hence we are not experiencing a significant mean effect size for any of the simulation outputs. However, any increase or decrease in any of the dependent variables when our study goes into practice is still a win because the participating farms would experience either an increase in crop yield or protection of crop yield, and change in SLF activity. Additionally, statistically for the simulation, numerically we have too few farms, but in practice 50 participating farms is a large number to conduct studies on. In short, from our simulation, we can see how our study’s real world implications may not be best suited in terms of statistical analysis. For this reason, we can refer back to the statistical power analysis for potential outcomes at the end of our actual study.

Conclusion

The research cooperates with USDA to estimate the methods to eliminate the population of spotted lanternfly in New Jersey, New York, and Pennsylvania between May and October in 2023. This article investigates if at least one of other methods (e.g. bug barrier tape, trap, trained dogs, and chickens) could reduce the population of SLF and further impact on improvements of crop yields and farmers’ perceiving experience could outperform the traditional insecticide usage based on one way ANOVA test. The first research outcome of SLF population shows at least one of the other deterrents could effectively decrease the population of SLF compared to the insecticide with statistically significant and strong power by 0.986. However, the outcome of crop yields presents no effect in our statistical analysis with power 0.16, which mainly results from the data of crop yields that can be collected once only during the short harvest season leading to insufficient sample size given. Likewise, the relationship between improvements of crop yields and reducing population of SLF is statistically insignificant. Similarly, in this study, the farmers do not feel better off on SLF population control when implementing at least one of the treatment methods and it is assumed that the traditional insecticide might be more practical for them in a likelihood. Therefore, even though other treatments have statistically and effectively reduced the population of SLF in tri-state farms, we still cannot conclude that the bug barrier tape, trap, trained dogs, and chickens contribute to a potential improvement of crop yields and also the improved experience and perceptions among farmers.

References

Bug barrier tree band kit, 10ft. (n.d.). TreeHelp. https://www.treehelp.com/products/bug-barrier-tree-band-kit-10ft?gclid=Cj0KCQiAveebBhD_ARIsAFaAvrGhr1GCrNnRN_dpZEfxlkHqIm9ubBrF_KbjK_sKxuR_vemnwascLkwaAhe0EALw_wcB Coyle, D. R., Chong, J. H., & Blaauw, B. A. (2019). Spotted Lanternfly Management in Nurseries, Orchards, Vineyards, and Natural Areas in South Carolina and Georgia. Clemson (SC): Clemson Cooperative Extension, Land-Grant Press by Clemson Extension. http://lgpress.clemson.edu/publication/spotted-lanternfly-management-in-nurseries-orchards-vineyards-and-natural-areas-in-south-carolina-and-georgia/

Decker, H. (2021). Citizen Science: Training Pet Dogs to Detect the Spotted Lanternfly (Doctoral dissertation, Virginia Tech). https://vtechworks.lib.vt.edu/handle/10919/105006

Essler, J. L., Kane, S. A., Collins, A., Ryder, K., DeAngelo, A., Kaynaroglu, P., & Otto, C. M. (2021). Egg masses as training aids for spotted lanternfly Lycorma delicatula detection dogs. PLOS ONE, 16(5). https://doi.org/10.1371/journal.pone.0250945

Francese, J. A., Cooperband, M. F., Murman, K. M., Cannon, S. L., Booth, E. G., Devine, S. M., & Wallace, M. S. (2020). Developing Traps for the Spotted Lanternfly, Lycorma delicatula (Hemiptera: Fulgoridae). Environmental Entomology. https://doi.org/10.1093/ee/nvz166

Harper, J. K., Stone, W., Kelsey, T. W., & Kime, L. F. (2019). Potential economic impact of the spotted lanternfly on agriculture and forestry in Pennsylvania. The Center for Rural Pennsylvania, Harrisburg, PA, 1-84. https://cpb-us-w2.wpmucdn.com/sites.udel.edu/dist/a/9656/files/2021/01/Harper-et-al-2019-Potential-Economic-Impact-of-the-Spotted-Lanternfly-on-Agriculture-and-Forestry-in-Pennsylvania.pdf

Huron, N., & Helmus, M. (2022). Predicting host associations of the invasive spotted lanternfly on trees across the USA. bioRxiv. https://www.biorxiv.org/content/10.1101/2022.09.12.507604v1.full

Jones, C., Skrip, M.M., Seliger, B. J., Jones, S., Wakie, T., Takeuchi, Y., Petras, V., Petrasova, A., & Meentemeyer, R. K. (2022). Spotted lanternfly predicted to establish in California by 2033 without preventative management. Communications Biology, 5, 558. https://doi.org/10.1038/s42003-022-03447-0

Jung, M., Kim, J., Kim, H. G., & Lee, D. H. (2016). Effect of harmonic radar tagging on Lycorma delicatula (Hemiptera: Fulgoridae) nymphal mobility and survivorship. Florida Entomologist, 47-51.

Leach, H., Biddinger, D. J., Krawczyk, G., Smyers, E., & Urban, J. M. (2019). Evaluation of insecticides for control of the spotted lanternfly, Lycorma delicatula, (Hemiptera: Fulgoridae), a new pest of fruit in the Northeastern U.S.. Crop Protection, 124. https://doi.org/10.1016/j.cropro.2019.05.027.

Lee, D. Y., Park, Y. L., and Leskey, T. C. (2019). A review of biology and management of Lycorma delicatula (Hemiptera: Fulgoridae), an emerging global invasive species. Journal of Asia-Pacific Entomology, 22(2), 589-596.

Murman, K., Setliff, G. P., Pugh, C. V., Toolan, M. J., Canlas, I., Cannon, S., … & Cooperband, M. F. (2020). Distribution, survival, and development of spotted lanternfly on host plants found in North America. Environmental entomology, 49(6), 1270-1281. https://academic.oup.com/ee/article/49/6/1270/5947504

Schneck, M. (2021). Chickens, praying mantises appear to be top predators on spotted lanternfly, study says. Pennlive. https://www.pennlive.com/life/2021/03/chickens-praying-mantises-appear-to-be-top-predators-on-spotted-lanternfly-study-says.html

Urban, J.M. (2020). Perspective: shedding light on spotted lanternfly impacts in the USA. Pest Manag Sci, 76, 10-17. https://doi.org/10.1002/ps.5619

United States Department of Agriculture - National Agricultural Statistics Service. (2022, January 12). Crop Production 2021. chrome-extension://efaidnbmnnnibpcajpcglclefindmkaj/https://downloads.usda.library.cornell.edu/usda-esmis/files/k3569432s/sn00c1252/g158cj98r/cropan22.pdf

USDA. (2021, February). Farms and Land in Farms 2021 Summary. chrome-extension://efaidnbmnnnibpcajpcglclefindmkaj/https://www.nass.usda.gov/Publications/Todays_Reports/reports/fnlo0222.pdf

Appendix A. USDA SLF Study Participation Survey

As you may be aware, spotted lanternflies (SLF) have rapidly spread across the Northeastern US and are wreaking damage to the surrounding wildlife and agriculture production. If their spread is not contained, SLF are estimated to overrun US farmland as soon as 2030. In cooperation with and the oversight of the US Department of Agriculture, our research team is seeking your participation in a study regarding this infestation. This is a preliminary survey to determine participation and the resource allocation and insect contain method that could be used on your farm if you choose to participate. Resources will be provided by the research team and the USDA as to best meet your needs and ensure smooth incorporation into your current farming practices.

  1. Have you noticed a sizable number of spotted lanternflies on your farm? Yes/No
  2. If you answered “Yes” to the previous question, do you believe these insects are posing harm to your crop yields? Yes/No
  3. Do you use organophosphate chlorpyrifos-based insecticides on your farm? Yes/No
  4. What is the primary insecticide used at your farm? __
  5. For categorization purposes, which of the following crops does your farm produce? Grapes/Almonds/Apples/Cherries
  6. Does your farm have Trees of Heaven growing either on or near the crop production area? Yes/No
  7. Do you have 2 dogs on your farm that you are open to training for the detection and control of SLF? Yes/No
  8. If you answered “Yes” to the previous question, how many trainable dogs do you have for the purpose of detecting and controlling SLF? __
  9. Do you have chickens on your farm that you are open to using for the detection and control of SLF? Yes/No
  10. If you answered “Yes” to the previous question, how many chickens do you have for the purpose of detecting and controlling SLF? __
  11. If you are chosen to participate, do you agree to commit to study being conducted during the May and October growing and harvest season in 2023? Yes/No

Appendix B. Biweekly Participant Survey Questions

Questions to all participating farms

The following will be answered with a likert scale from 1-strongly disagree to 5-strongly agree

  1. During the past two weeks, there was a sizable decrease in insect and spotted lanternfly activity.
  2. During the past two weeks, it was relatively easy to adopt these containment methods to my current practices.
  3. During the past two weeks, adopting these containment methods helped protect or improve my farm’s potential crop yield.
  4. Continuing these containment efforts is worth pursuing even after the study period.

Control Group: Traditional Insecticide

  1. Location of farm: NJ/NY/PA
  2. During the last two weeks, how often did you spray insecticide on or near your crop production area? __

Treatment Group: BugBarrier Tape

  1. Location of farm: NJ/NY/PA
  2. How many BugBarrier Tape bands did you use over the observation period? __

Treatment Group: Trained Dog Detection

  1. Location of farm: NJ/NY/PA
  2. What method for eliminating the spotted lanternfly did you use? Insecticide/Fire/Smoke/Physical offense/Other: __

Treatment Group: Chicken Predator

  1. Location of farm: NJ/NY/PA
  2. Have you noticed any loss of crop due to chicken release? Yes/No/Not Really
  3. Have you noticed chickens feeding on spotted lanternflies? Yes/No/Not Really

Final survey questions

  1. What was your 2023 crop yield at the end of this study period? __