Evaluating the LepiLED light trap and iNaturalist identification for detection of invasive pests in urban high-risk sites

LepiLED(R) light traps were deployed in 2021 at two urban locations, to test the efficacy of the traps, as well as ground vs. tree deployment, for collecting and detecting invasive moths. As well, we photographed resulting specimens, and uploaded the images to iNaturalist, to evaluate the power of citizen scientists and the "computer vision" algorithm behind the iNaturalist platform, for making identifications of resulting specimens. Final identifications were made by the authors, via dissection and/or DNA barcoding as required.

The LepiLED(R) traps are very effective at collecting moths and other nocturnal insects. There were differences in the species collected in ground and tree traps, but they were equally effective overall. Several invasive species were detected, including a new provincial record among the hundreds of species collected.

The iNaturalist computer vision algorithm was moderately effective at identifying macromoths, but poor for micromoths. Less than half of our observations were identified or confirmed by citizen scientists; it appears that the high number of observations overwhelmed the small number of volunteer identifiers on iNaturalist. Nevertheless, the new provincial record was first identified by an expert citizen scientist.

The LepiLED(R) traps are effective and easily deployed; the more limiting factor is expertise in identifying the resulting catches. As the computer vision algorithm improves and participation by humans increases, iNaturalist has good potential for assisting with this aspect of monitoring and identifying our moth fauna.