Distributed surveillance and decision support ecosystem for control of coconut rhinoceros beetle

Abstract

An introduced population of coconut rhinoceros beetle (CRB; Oryctes rhinoceros) was first discovered on the island of Oahu in late 2013. Adults of this invasive beetle feed on a variety of host plants, including palms, bananas, and sugar cane. As the name of the beetle implies, the preferred host is the coconut palm (Cocos nucifera). CRB feeding results in large boreholes near the crown, introducing a route for infection by a variety of pathogens. CRB damage can easily be recognized by characteristic cuts and notches on fronds as they grow out, and in cases of severe damage the tree can be completely defoliated and die. Surveillance efforts to delineate populations and identify spread poses one of the greatest challenges for control. Currently there are almost two thousand CRB panel traps installed throughout the island of Oahu, scaled back from over three thoUSAnd before populations spread to every significant part of the island and control efforts were reorganized to focus on ports of entry to prevent accidental export. Each of these traps is visually inspected at regular intervals (approximately once or twice per month, depending on location) by human operators. This manual trap checking is laborious, expensive with respect to personnel and transportation costs, and time-consuming. Intermittency of trap checks and associated manual data entry and processing to understand geographical distributions also result in delays responding to new outbreaks. Here, we have developed an energy-efficient, automated remote surveillance (CRB-Cam) and data management ecosystem, incorporating machine vision tools to facilitate the monitoring of CRB in highly distributed remote locations. Accurate object detection of beetles in traps, and associated coordinates of traps from on-board GNSS receivers are used to automatically generate maps of daily trap catches which are especially important for rapidly responding to new incipient populations before they can become established in an area. In preliminary field trials with systems uploading trap images hourly, adult CRB exhibited crepuscular behavior, with over 2/3 of observed trap catches occurring at night within 3 hours of civil twilight after sunset, catch rates trailing off gradually throughout the night, and fewer than 1% of catches occurring during daylight. Based on these observations and observations from earlier researchers that even with regularly changed pheromone lures panel traps catch only small percentages of CRB in a local area, we investigated the use of nocturnal artificial lighting to improve catch rates. In a field trial rotating programmable LED arrays programmed to illuminate with different color LEDs for 6 hours after sunset, and for 3 hours before sunrise, traps illuminated with ultraviolet (UV) caught significantly more CRB than those without illumination, and traps illuminated with all tested visible wavelengths resulted in significantly fewer CRB catches. Management of CRB relies primarily on effective green waste management to remove or destroy potential breeding sites in decaying organic matter, and pesticide applications in host trees to kill foraging adults. As a last element of this research we investigated the use of precision aerial application of pyrethroid based pesticides directly into crowns of palm trees as an alternative to or in rotation with the current practice of trunk injection of systemic neonicotinoid pesticides. Crown applications were shown to be effective at killing CRB and protecting the tree for a period of several months, and were also useful for identifying actively infested trees as CRB would emerge and fall to the ground to die. Data from these experiments were used to obtain crisis use exemption for using dilute formulations of Demon® Max Insecticide (25.3% cypermethrin) for controlling CRB on Maui, Hawaii, and Kauai, where it was used to treat an outbreak on Maui that thus far has putatively been eradicated.