Farmers are requesting in-depth knowledge about how to successfully integrate UAS (Unmanned Aerial Systems) into strawberry production to improve management practices. California produces 88% of U.S. strawberries, with an annual value of approximately $2.6 billion, and the California Strawberry Commission identifies spider mite management a key research priority. Little is known about the spatio-temporal dynamics of spider mites in strawberry, and consequently, about the optimal timing to release predatory mites, their natural enemies. In this multi-disciplinary project, we will demonstrate that airborne remote sensing can be used to detect and diagnose spider mite hotspots and therefore pinpoint when and where predatory mite releases are needed. This three-year project will lead to the following outcomes: 1) demonstrate relationships between spider mite presence and unique changes in leaf reflectance features acquired from strawberry plants; 2) a spatially optimized, reliable and practically feasible sampling plan for spider mites in strawberry fields; 3) hyperspectral airborne remote sensing-based characterization of spider mite hotspots to spatially optimize releases of natural enemies. Deployment of novel, labor-extensive, and precise airborne remote sensing technologies to monitor crop health and mitigate pest risks, as well as aerial predator distribution techniques, are highly compatible with existing strawberry management practices. Educational outreach involves hands-on workshops and lectures on spider mite sampling at grower and ag professional meetings throughout the California strawberry growing region. Growers outside this area will be reached through publications in trade journals and other grower media. Also, we will organize a panel discussion on the potential of UAS in strawberry management and beyond. Both the educational outreach and research outcomes are easily transferable to strawberry growing regions elsewhere, and could be expanded to include monitoring for other pests and pathogens. This project will enhance efficacy of biocontrol agents and reduce dependence on miticides, enhancing sustainability of spider mite management practices in strawberry.
The research component of this project is based on two fundamental hypotheses (both supported by preliminary data):
1) Spider mite outbreaks are most likely in strawberry plants under abiotic stress (imbalances or deficiencies of nutrient fertilization or inadequate irrigation), and abiotic stress is closely linked to descriptive leaf traits (toughness and micro- and macro-element composition).
2) Airborne remote sensing data of strawberry canopy can be used to detect and map descriptive leaf traits (toughness and micro- and macro-element composition), which are important indicators of risk of spider mite outbreaks.
The educational outreach component of this project is based on the hypotheses that:
1) sampling of strawberry fields can be optimized through development of reliable and practically feasible sampling plans.
2) use of predatory mites for control of spider mite outbreaks in strawberry can be significantly improved through better detection (either through remote sensing or sampling) of spatio-temporal aggregations of emerging spider mite outbreaks.
As part of this 3-year project, we will address the following numbered objectives:
Objective 1: Experimental characterization of the relationship between:
1) leaf micro- and macro-element composition and leaf reflectance.
2) spider mite-induced stress and leaf reflectance in project years 1 (May – July 2017), 2 (May – July 2018), and 3 (May-July 2019). Quantifiable outcome: A research article describing a model of the factorial relationships between leaf micro- and macro-element composition, leaf reflectance, and spider mite-induced stress. This objective will start in May 2017 and will be completed in April 2020. The first portion of an experimental study has been completed in project year 1, and the results are described in the section “Results and discussion.”
Objective 2: Characterization of the spatio-temporal distribution of spider mites and their natural enemies in project years 2 (May – July 2018) and 3 (May – June 2019). Quantifiable outcomes: 1) spatio-temporal data sets from 120 combinations of commercial strawberry field, sampling event, and growing season, 2) a research article describing a spatially optimized sampling plan for spider mites in strawberry fields. Moreover, sampling of spider mites in commercial strawberry fields will provide valuable insight into spatial trends in the distribution of emerging infestations and their spread. That is, we predict to document consistent “edge-effects” of spider mite infestations with highest spider mite densities near strawberry field edges. Quantitative information about prevailing spatial distribution patterns is critically important as part of optimizing both sampling/detection and releases of natural enemies. Although data collection was scheduled in 2017, we were unable to locate suitable field sites, but addressing this project objective is a major priority in the 2018 field season. This objective will be completed in April 2020.
Objective 3: Hyperspectral airborne remote sensing-based characterization of hotspots with emerging spider mite outbreaks in project years 1 (May – July 2017), 2 (May – July 2018), and 3 (May – June 2019). Quantifiable outcomes:
1) hyperspectral airborne remote sensing data sets from 120 combinations of commercial strawberry field, sampling event, and growing season.
2) a research article describing the use of hyperspectral airborne remote sensing to spatially optimize releases of natural enemies to control spider mite outbreaks in strawberry fields.
Field data collection has been initiated, and preliminary results are described in the section “Results and discussion.” This objective will be completed in April 2020.
Objective 4: Participatory (hands-on) educational outreach on how to improve spider mite sampling and interpretation of sampling data in project years 2 and 3. Quantifiable outcomes:
1) dissemination of project results on spider mite sampling to a minimum of 300 growers.
2) publication of project results on spider mite sampling on the UC IPM website (2). This objective will be addressed starting in May 2018 and be completed in April 2020.
Objective 5: Panel discussion on the potential of UAS in strawberry management and beyond in project year 3. Quantifiable outcomes:
1) we will organize a panel discussion with ca. 15 growers and ag professionals.
2) main discussion points from panel discussions will be summarized in a white paper to be presented to the California Strawberry Commission and other stakeholder organizations. This objective will be addressed starting in October 2019 and be completed in April 2020.
Objective 1. We hypothesize that strawberry canopy reflectance features vary significantly and uniquely in response to specific biotic stressors, such as spider mites. Remote sensing is a non-invasive, labor-extensive method to detect plant stress before changes are visible by eye, through assessment of leaf reflectance. Particular stressors, such as arthropod infestations, induce physiological plant responses, causing changes in the plants’ ability to perform photosynthesis, which leads to changes in leaf reflectance. Most previous studies involve only one arthropod species and one crop. However, in a realistic situation, multiple arthropod pests will be present in commercial crop fields. In order to optimize pest management and control, we aim to distinguish strawberry infestation with spider mites from strawberry infestation with other types of pests, using remote sensing.
Objective 3. We hypothesize that, in a commercial field, strawberry plant reflectance features vary significantly when attacked by spider mites and when healthy. In order to optimize pest management and control, we aim to characterize spider mite hotspots based on airborne remote sensing.
Objective 1. We hypothesize that strawberry canopy reflectance features change significantly and uniquely in response to specific biotic stressors, such as spider mites. To characterize the “uniqueness” of reflectance responses by strawberry plants to spider mites, we conducted an experimental greenhouse study, in which individually potted strawberry plants (Fig. 5a) were divided into four treatments: 1) non-infested control plants, 2) plants infested with spider mites, 3) plants infested with whiteflies, and 4) plants infested with beet armyworms. These different biotic stressors were chosen as they represent different arthropod pests with different feeding biology. We acquired data on leaf reflectance from the plants after 0 (baseline), 1, 3, 7, 10 and 14 days of infestation (DAI). We acquired data on photosynthesis, leaf element composition, and biomass from the plants after 0 (baseline) and 14 DAI.
Hyperspectral imaging data were acquired using a true push-broom hyperspectral camera with 116 spectral bands from 470-980nm (OCI Imager, OCI-UAV-D1000), which was mounted on a rail system inside a greenhouse (Fig. 5b). However, to eliminate spectral bands associated with high signal/noise ratio, we only included data in 106 spectral bands from 513-967 nm in the analyses. Fig. 5c shows the actual spatial resolution of the acquired hyperspectral imaging data.
A grounded hyperspectral spectrometer (SuperGamut imaging spectrometer 380-1020 nm wavelength range, here referred to as the “reference spectrometer”) was mounted in vertical position and with the lens 50 cm above a piece of matte white Teflon (referred to as white calibration board). Data from the reference spectrometer is used for continuous white calibration of leaf reflectance data. This project component is one of the most unique and important aspects of this project, because it enables high-precision calibration of leaf reflectance data, and therefore markedly higher likelihood of accurately detecting subtle reflectance responses to biotic stressors.
Data on photosynthesis, as well as the related factors stomatal conductance, transpiration rate, and intercellular CO2, were obtained with a LiCOR 6400XT Portable Photosynthesis System. Measurements were taken from the youngest fully expanded leaves. Also, we obtained plant aboveground fresh weight. Afterwards, plant material was dried in an oven at 60°C for four days, and dry weight measurements were taken. Dry leaf material was ground to a fine powder and shipped to a commercial laboratory for element composition analysis (N, S, P, K, Mg, Ca, Na, B, Zn, Mn, Fe, Cu, and Al).
Leaf reflectance data from 0 and 14 DAI were subjected to advanced data processing and filtering and we then classified data based on the following parameters: 1) all leaf reflectance data from 0 DAI were considered “control” (non-infested), and 2) leaf reflectance data from 14 DAI were assigned to their respective treatment. Details on processing and classification of reflectance data have been described extensively in studies published by PI Nansen (Nansen and Elliott 2016; Zhao et al. 2014; Nansen et al. 2013a; Zhao et al. 2013; Nansen 2012, 2011; Qi et al. 2011).
Objective 3. We identified pairs of commercial strawberry fields in the same strawberry growing area, of which one is healthy, and one is heavily infested with spider mites. It is important that the strawberry variety in both fields is the same, as our hyperspectral camera is sensitive enough to detect differences between varieties. We used the same hyperspectral camera and continuous white calibration as described under project results for objective 1. We developed a classification method to distinguish spider mite-infested from healthy strawberry plants, and we used the same processing and classification procedures as described extensively in studies published by PI Nansen (Nansen and Elliott 2016; Zhao et al. 2014; Nansen et al. 2013a; Zhao et al. 2013; Nansen 2012, 2011; Qi et al. 2011).
Objective 1. We acquired data on leaf reflectance from strawberry plants in individual pots after 0 (baseline), 1, 3, 7, 10 and 14 days of infestation (DAI) with spider mites, whiteflies, and beet armyworm, as well as control, non-infested plants. We acquired data on photosynthesis, leaf element composition, and biomass from the plants after 0 (baseline) and 14 DAI. Here, we only present data from 0 and 14 DAI.
At 14 DAI, spider mite-infested plants contained 159 ± 75.2 eggs (average ± SE) and 193 ± 52.3 nymphs and adults (mobiles). Whitefly-infested plants contained, on average, 27 ± 7.5 eggs and 26 ± 8.1 nymphs. In addition, it was confirmed that strawberry plants infested with beet armyworms (3 second-instar larvae) all experienced extensive damage. Thus, arthropod infestations were successful under the experimental conditions.
For three of the leaf elements (P, K, and Mg), there was a significant effect of biotic stress at 14 DAI (P-value < 0.05), while the remaining ten leaf elements (N, S, Ca, Na, B, Zn, Mn, Fe, Cu, and Al) did not show a significant response to biotic stressors (P-value > 0.05). Moreover, we found that significant differences in levels of magnesium (Fig. 1a), potassium (Fig. 1b) and phosphorus (Fig. 1c), were attributed to differences between plants subjected to beet armyworms and spider mites (different letters above bars indicate statistically significant differences). We observed a negative association between spider mite infestation and low potassium leaf content, which is consistent with published research by PI Nansen (Nansen et al. 2013b; West and Nansen 2014). In addition, published research by PI Nansen has shown that low potassium leaf content has been associated with higher likelihood of infestation by two aphid species (Lacoste et al. 2015; Severtson et al. 2016). Thus, we have provided initial evidence of strawberry plants showing partial element composition responses to different biotic stressors.
Average reflectance profiles from strawberry plants at 0 DAI (Fig. 2a) showed that there was negligible difference in leaf reflectance prior to exposure to biotic stressors. Fig. 2b shows that at 14 DAI, especially beet armyworms and spider mites caused considerable increase in leaf reflectance. Fig. 2c shows the same data as presented in Fig. 2a and 2b, except that the 14 DAI data were divided with the average reflectance profiles from 0 DAI. Thus, Fig. 2c shows the relative change over time in terms of average leaf reflectance, with: 1) values >1 indicating an increase in reflectance, 2) values < 1 indicating a decrease in reflectance, and values close to 1 indicating little or no change in reflectance. From Fig. 2c, it is seen that: 1) non-infested control plants had the least change in reflectance within the 14-day time period, 2) whitefly infestation appeared to have negligible effect on leaf reflectance (very similar average reflectance values as non-infested control plants), 3) beet armyworm infestation caused a marked increase in leaf reflectance, especially at 648 nm, and 4) spider mite infestation caused a modest increase in leaf reflectance in spectral bands from 580-665 nm and a similar response as beet armyworm infestation in spectral bands from 750-980 nm.
It is important to emphasize that the hyperspectral imaging data were acquired inside a greenhouse, which means that glass interfered with the obtained leaf reflectance data. It is therefore not possible to directly translate the obtained imaging data into commercial field conditions. However, the data are very important and encouraging as they demonstrate that it may be feasible to uniquely separate spider mite infestations from other biotic stressors of strawberry plants.
A total of 156 average reflectance profiles from potted strawberry plants were analyzed, and the results from the classification are presented in a confusion matrix (Table 1). In Table 1, the left column lists the actual treatments and the other columns lists how the average reflectance profiles from the different treatments were classified. Thus, all (100%) of average reflectance profiles from armyworm infested strawberry plants were classified as armyworm infested. In addition, 80% of average reflectance profiles from spider mite infested strawberry plants were classified as spider mite infested (5% were incorrectly classified as armyworm infested, and 5% were incorrectly classified as whitefly infested). Across the four treatments, the leaf reflectance based classification of plants was 85%, which is well within the acceptable accuracy range for these types of applications of reflectance based analyses.
Objective 3. As part of this project, we intend to demonstrate that airborne remote sensing can be used to detect hotspots with emerging spider mite outbreaks based on unique reflectance responses by strawberry plants to spider mite infestations. In project year 1, we conducted flight missions above non-infested strawberry fields and above hotspots with emerging spider mite outbreaks (Fig. 3a). Flying at an altitude of 30 m, we obtained hyperspectral imaging data as shown in Fig. 3b, in which the multiple pixels are acquired from individual strawberry leaves. As part of the data analysis, we established two categories of “soil” (with or without shadow), and representative pixels were selected from hyperspectral images of non-infested and infested hotspots. Fig. 3c shows the average reflectance profiles from the four categories (soil with or without shadow and non-infested and infested hotspots). In addition, the black curve indicates the difference between infested and non-infested strawberry plants (infested / non-infested), and it is seen that spider mite infestation causes a marked increase in leaf reflectance in spectral bands from 700-980 nm. In a study of spider mite infestations of maize plants, PI Nansen found very similar results; a spectral band near 740 nm was particularly indicative of spider mite infestation (Nansen et al. 2013b).
As an example of how the airborne hyperspectral data may be used, Fig. 4 shows two locations without (Fig. 4a) and with (Fig. 4b) ground-based confirmation of spider mite infestation of strawberry plants. The maps of hotspots were generated on the basis of processed and classified hyperspectral imaging data, and the maps clearly indicate that airborne remote sensing data may be able to detect spider mite infestations. We wish to point out that at the infested hotspot, 5-10 spider mites were found on most leaves, so this represented an unusually high infestation level. Considerably more field data during growing seasons and among growers are needed before we can comfortably argue that spider mite infestation hotspots can be detected consistently and with sufficient accuracy. However, the preliminary data are encouraging, and the field data are partially supported by the reflectance responses shown in the experimental greenhouse study (Fig. 2).
Lacoste C, Nansen C, Thompson S, Moir-Barnetson L, Mian A, McNee M, Flower KC (2015) Increased susceptibility to aphids of flowering wheat plants exposed to low temperatures. Environmental Entomology 44 (3):610-618.
Nansen C (2011) Robustness of analyses of imaging data. Optics Express 19:15173-15180.
Nansen C (2012) Use of variogram parameters in analysis of hyperspectral imaging data acquired from dual-stressed crop leaves. Remote Sensing 4 (1):180-193.
Nansen C, Elliott NC (2016) Remote sensing and reflectance profiling in entomology. Annual Review of Entomology 61:139-158.
Nansen C, Geremias LD, Xue Y, Huang F, Parra JR (2013a) Agricultural case studies of classification accuracy, spectral resolution, and model over-fitting. Applied Spectroscopy 67 (11):1332-1338.
Nansen C, Sidumo AJ, Martini X, Stefanova K, Roberts JD (2013b) Reflectance-based assessment of spider mite “bio-response” to maize leaves and plant potassium content in different irrigation regimes. Computers and Electronics in Agriculture 97:21-26.
Qi B, Zhao C, Youn E, Nansen C (2011) Use of weighting algorithms to improve traditional support vector machine based classifications of reflectance data. Optics Express 19:26816-26826.
Severtson D, Callow N, Flower K, Neuhaus A, Olejnik M, Nansen C (2016) Unmanned aerial vehicle canopy reflectance data detects potassium deficiency and green peach aphid susceptibility in canola. Precision Agriculture 17 (6):659-677.
West K, Nansen C (2014) Smart-use of fertilizers to manage spider mites (Acari: Tetrachynidae) and other arthropod pests. Plant Science Today 1 (3):161-164.
Zhao C, Qi B, Nansen C (2013) Use of local fuzzy variance to extract the scattered regions of spatial stray light influence in hyperspectral images. Optik 124:6696-6699.
Zhao C, Qi B, Youn E, Yin G, Nansen C (2014) Use of neighborhood unhomogeneity to detect the edge of hyperspectral spatial stray light region. Optik – International Journal for Light and Electron Optics 125 (13):3009-3012.
In the first year of this project, we presented preliminary results at two CAPCA ED (California Association of Pest Control Advisers Education) meetings in California, as well as at the Annual Santa Maria Strawberry and Vegetable Meeting, for an audience of growers, pest control advisers, industry professionals, and researchers.
In the second and third year of this project, we will demonstrate our remote sensing equipment and present tangible results of our study, including a revised spider mite sampling plan, at various meetings throughout the California strawberry growing area. To reach more growers, we will publish findings of our study in newsletters, electronic extension journals, trade journals, and other grower media that will reach several thousand people. Also, we will present our results on the UC IPM website, as well as on Twitter and the IPMinfo app.
We have contacted the Institutional Review Board (IRB) of UC Davis in order to have our outreach surveys approved, to assess growers’ willingness to adopt and ag professionals’ willingness to disseminate the recommended practices. We will start handing out surveys after educational activities as soon as the surveys are approved.
Educational & Outreach Activities
Elvira de Lange: Oral presentation at the Western Alfalfa Seed Growers Association, 49th Annual Winter Seed Conference, in San Antonio, TX, USA. Audience: Alfalfa growers and pest control advisers. Title: Unmanned Aerial System (UAS)-guided releases of predatory mites for management of spider mites in strawberry. January 28-30, 2018.
Christian Nansen: Oral presentation at the Annual Santa Maria Strawberry and Vegetable Meeting in Santa Maria, CA, USA. Audience: Strawberry growers, pest control advisers, industry professionals, and researchers. Title: Improving management of arthropod pests in strawberry production systems. November 28, 2017.
Elvira de Lange: Oral presentation at CAPCA ED (California Association of Pest Control Advisers Education) in Oxnard, CA, USA. Audience: Pest control advisers and industry professionals. Title: IPM in strawberries. June 28, 2017.
Elvira de Lange: Oral presentation at CAPCA ED in San Jose, CA, USA. Audience: Pest control advisers and industry professionals. Title: Integrated pest management in California strawberries: optimizing spray coverage for improved control of arthropod pests. May 3, 2017.
Elvira de Lange: Oral presentation at Entomology 2017, the Annual Meeting of the Entomological Society of America, in Denver, CO, USA. Audience: Researchers and industry professionals. Title: Detecting spider mite-induced stress in California strawberry using innovative remote sensing technology. November 5-8, 2017.
Elvira de Lange: Departmental Seminar at Rutgers University, NJ, USA. Audience: Members of the Department of Entomology at Rutgers University, NJ. Title: Drone-guided releases of predators and optimized pesticide spray coverage as strategies for sustainable pest management in strawberry. September 22, 2017.
Elvira de Lange: 1st Biotweeps Twitter Conference, online. Audience: Biologists in all disciplines from all over the globe. Title: Drone-guided releases of predators for sustainable pest management in strawberry. 28-30 June, 2017.
Elvira de Lange: 16th Annual Strawberry Production Meeting in Ventura County. Audience: Strawberry growers, pest control advisers, industry professionals, and researchers. September 7, 2017.
Researcher uses drones, predators against strawberry pests. Author: Tim Hearden. Capital Press, The West’s Ag Website. June 6, 2017. http://www.capitalpress.com/Research/20170606/researcher-uses-drones-predators-against-strawberry-pest
UC Davis scientist targets spider mites in strawberry fields. Author: Kathy Keatley Garvey. The Davis Enterprise. May 31, 2017. http://www.davisenterprise.com/local-news/uc-davis-scientist-targets-spider-mites-in-strawberry-fields/
How a UC Davis Scientist is targeting spider mites in strawberry fields. Author: Kathy Keatley Garvey. Davis Patch. May 30, 2017. https://patch.com/california/davis/how-uc-davis-scientist-targeting-spider-mites-strawberry-fields
Innovative research: How Elvira de Lange is targeting spider mites in strawberry fields. Author: Kathy Keatley Garvey. Entomology & Nematology News. May 24, 2017. http://ucanr.edu/blogs/blogcore/postdetail.cfm?postnum=24192
- Use of drones for sustainable agriculture
The project outcomes will improve efficacy and increase use of biological control organisms of spider mites, a devastating pest of strawberry in California and elsewhere. This will ultimately reduce the use of miticides, and therefore increase health and safety for growers and farm workers and reduce consumer exposure to pesticide residues. Also, the project outcomes will lead to reduced yield losses due to spider mites. With the use of our state-of-the-art airborne remote sensing system, we are demonstrating the feasibility of airborne remote sensing as a cost-saving approach to improved crop monitoring and early detection of crop stress. This project is based on grower-identified needs and experience, and most of the project activities are conducted in commercial strawberry fields. Through integration of participatory research and educational outreach, this project will enhance effectiveness of releases of natural enemies to control spider mite outbreaks, so it will promote an essential pillar of sustainable agriculture: biological control. This project is still in progress.