About the project

Project team

David Crowder

Assistant Professor of Entomology
Washington State University

Crowder conducts research focusing on insect ecology and pest management within agricultural systems. He has worked on pests of potato cropping systems since 2010, focusing on biological control and landscape ecology. Crowder is leading the development of the interpolation models for the website.

Carrie Wohleb

Associate Professor of Extension
Regional Vegetable Crops Specialist
Washington State University

Wohleb is an extension specialist focusing on integrated pest management within vegetable crops of Washington, including potatoes. She leads the insect monitoring network in Washington State, where her team had sampled 40-50 sites for insect pests every week since 2007. Wohleb also maintains an e-mail list serve that is distributed weekly to potato growers and industry members to keep them updated over the course of each growing season.

Vince Jones

Professor of Entomology
Washington State University

Jones conducts research focusing on insect population dynamics and pest management, particularly biological control, within agricultural systems. He is the leader of the WSU Decision Aid System (DAS), which provides the tree fruit industry with up-to-date information on pest problems in tree fruit crops of Washington State. Jones and his team are leading development of the decision support website for potato crops, which is modeled after the tree fruit DAS system.

Stefano Borghi

Web Developer
Washington State University

Stefano is a computer programmer and web developer. He has taken the lead on development of the Decision Aid website, particularly integration of the interpolation figures into mapping software.

Development of the Potato Pest Mapping System

Figure1

Figure 1

Location of fields sampled for pests during 2016.

Potato growers must manage a suite of pests, including insects, diseases, and weeds. Sampling networks, where pests are monitored regularly across a broad region, allow growers to visualize “hot spots” of high pest activity and anticipate mobile pest populations. In Washington, Wohleb and her team conduct weekly monitoring of about 50 potato fields throughout the Columbia Basin for pests including aphids, beet leafhoppers, potato tuberworms, and potato psyllids (Figure 1). These data are sent to growers and other members of the potato industry through weekly e-mail updates.

Beginning in 2013, our team began to develop a decision-support tool using the data from the potato sampling network. Using geographical information systems (GIS) technology, we generated weekly predictions of pest densities based on the sampling data using a technique called “interpolation” (Figure 2). Briefly, to predict the density of a given pest at any location throughout the Columbia Basin, the GIS program calculates a weighted average of the number of insects from nearby monitoring sites; sampling sites that are closer to the location being estimated are given higher weight because they are assumed to have more similarity in insect densities. This produces a continuous prediction of insect densities throughout the region, where areas of similar pest density are grouped together and shown on our maps using the same color coding (Figure 2).

One benefit of this approach is that growers could get targeted predictions on expected pest densities in their specific field(s), regardless of where they were located in the state (Figure 2b). This is in contrast to the previous output of the sampling network, where only the densities at sampled locations were shown (Figure 2a). Moreover, by showing interpolated surfaces rather than densities from particular fields, growers involved in the network remain anonymous. In 2017 we extended our mapping approach into Oregon and Idaho.

Our predictive models are based on over 8 years of data, and we conducted extensive validation of our approach. This was done by comparing predicted pest densities from the interpolations with actual pest densities observed through sampling. In the validation stage, we used a portion of the sampling data to develop the interpolation models, and the rest to validate the models. We have demonstrated that our simple GIS approach produces approximately 70% congruence between predictions and observed pest densities. This is a high degree of precision, considering that many factors (such as insecticide use) can influence pest densities, and our predictions are based solely on monitoring data from 50 fields. However, it should be noted that the insect maps should be used as a guideline for what to expect, and not a definitive count of the number of insects in any given field at any given point in time. Variability in many factors, including grower management strategies, could cause predictions to vary from observed counts.

Figure2

Figure 2

(A) Example of the output of the potato sampling network prior to 2013 (showing aphid pests) and (B) the output that was developed as part of our project. These interpolations in panel B show pest predictions throughout the region based on data from the sampling network (panel A). The various colors are based on major action thresholds for each pest species. Blue = pest not present; green = pest at low density; yellow = pest at medium density; red = pest at high density. Locations with the same color are expected to have similar densities.

Once the models were validated, we began to publish these “heat maps” of insect densities in 2014 through the e-mail alerts. When maps across multiple weeks are plotted, it allows users to visualize how mobile insect pest populations are moving across, and developing within, the landscape; this information can be used to better anticipate when insect pests will arrive in certain regions and guide management decisions. In 2017, working with the team of Dr. Vince Jones and Stefano Borghi at the Washington Tree Fruit Research Center, we integrated these heat maps into a Google Earth mapping framework. This will allow individuals to zoom into particular areas of the interpolation maps to view predictions at smaller scales. In the future, we will code this tool for both desktop and mobile devices. Making the tool accessible on mobile devices will improve access and usage over time.

Overall, we will continue to refine our model approaches as we gather more data on each insect pest. In the coming years we also hope to link the pest maps with management recommendations. Thus, when a grower zooms in to see the predictions for his/her field, they will also receive management tips based on the expected number of insects. Our hope is that as the tool develops, growers can use it to better anticipate pest populations, lower their need for insecticide sprays, and increase their profits.

Questions about the pest mapping project can be directed to David Crowder (dcrowder@wsu.edu) or Carrie Wohleb (cwohleb@wsu.edu).