Evaluating the effectiveness of participatory science dog teams to detect devitalized Spotted Lanternfly (Lycorma delicatula) egg masses

Abstract

The spotted lanternfly (Lycorma delicatula, SLF) is an invasive planthopper first detected in the United States in 2014, with initial sightings in Pennsylvania. SLF poses a serious threat to agriculture, particularly targeting grapevines, hops, and ornamental plants, resulting in substantial annual economic losses. Due to its life cycle, the early detection and removal of egg masses are the most effective strategies for preventing long-distance dispersal. However, visual detection by humans is time-consuming and inefficient. Detection dogs have demonstrated high accuracy in locating SLF egg masses and differentiating them from environmental distractors. Despite their effectiveness, the number of dogs available through governmental channels is insufficient to meet demand. This study evaluated whether community scientist dog-handler teams could meet standardized detection criteria using SLF egg masses. Teams from across the U.S. were recruited and trained using devitalized egg masses, with oversight provided by local trainers. Following a 3- to 6-month independent training period, team performance was assessed through an odor recognition test and a field trial. Dogs demonstrated a sensitivity of 82% in controlled testing and 58% in field conditions. These results provide proof of concept; community scientist dog teams could play a significant role in protecting their local environments and agriculture from invasive species.

Keywords: Community science; Conservation detection dogs; Detection dogs evaluations; Invasive species detection; Participatory science; Spotted lanternfly.

Figure 1. Participatory science dog team SLF detection study overview.

Participants were given three attempts to achieve an 80% passing score on the ORT, FE, and LE-ORT. Full details of the evaluations are listed in Experiments 1 and 2.

Figure 2. Picture of devitalized SLF egg training aid (SLF-training aid).

Constructed of folded and crimped stainless steel mesh with 0.5 g devitalized SLF eggs.

Figure 3. SLF ORT Set up.

Picture showing the typical ORT setup with camera locations, box spacing, and numbers in the array.

Figure 4. Typical of FE area.

The photograph shows a typical FE search area, including lumber and other objects.

Figure 5. Cumulative number of teams passing the ORT and FE.

The x-axis represents the attempt number for the ORT and FE. The y-axis represents the number of teams with a passing score (80% or greater). The numbers above each bar indicate the cumulative total number of LE-ORT attempts made. Teams had to complete the ORT before progressing to the FE.

Figure 6. ORT evaluation version scores.

The graph shows the score distribution across the four ORT versions, which differed in the number of blank trials: versions one and two had two blank trials, version three had three blank trials, and version four had one blank trial. The bar represents the mean score for that version, the bars represent the standard deviation, and the dots show the individual scores on each version.

Figure 7. Proportion of correct responses (accuracy) per trial.

The graph shows the mean and standard deviation of all versions for correct responses (true positives and correct rejections). Calculation provided in ‘Analysis’.

Figure 8. ORT allocation of incorrect responses.

The graph shows the mean percent allocation of responses across dogs to the distracting odors in both hot and blank trials.

Figure 9. Time searching for hot trials and blank trials.

The violin plot displays the median (dashed line), along with the 25th and 75th percentiles (dotted lines), showing the time in seconds spent searching during hot trials and blank trials, separated by pass and fail outcomes. The width of the plot at each point reflects the distribution density of the data at that Y-axis value (Time (s)) **** p < 0.0001.

Figure 10. Cumulative number of teams passing the LE-ORT.

The graph shows the progression of teams meeting the passing criteria across successive attempts for the ORT evaluation. The x-axis represents the attempt number, while the y-axis displays the cumulative number of teams passing each test. The numbers above each bar indicate the cumulative total number of LE-ORT attempts made.

Figure 11. Time searching for LE-ORT pass and fail trials.

The violin plot displays the median (dashed line) along with the 25th and 75th percentiles (dotted lines), showing the time in seconds spent searching during hot trials and blank trials, separated by pass and fail outcomes. The width of the plot at each point reflects the density of the data distribution at that Y-axis value (time in seconds).