Abstract
The purpose of this project is to aid the work of the Plant Health and Seeds Inspectors (PHSI) and Forestry Commission (FC) Inspectors by improving their ability to detect a range of statutory/quarantine pests using the acoustic methods initially investigated in a previous Plant Health Division funded project PH0191, which established proof of principle. Increased globalisation of trade in plants and plant products has resulted in an increasing risk of introducing exotic pests that may threaten UK horticulture, forestry, amenity trees, and the environment. Species of long horn beetles (Anoplophora chinensis and A. glabripennis) are good examples; these have recently been intercepted on hardy nursery stock in several EU countries, including the UK, and are a threat to native trees and forestry. Development of novel acoustic detection technologies represents a significant advancement of the strategic investigation of insect-generated sounds worldwide. Recent advances arising from University of York/Central Science Laboratory joint work (project PH0191) provided underpinning knowledge of automated systems for detecting quarantine invertebrate pests from which a practical system can be developed and tested. It involved a feasibility study into the use of low-cost acoustic sensors and efficient pattern recognition to detect and/or identify quarantine pests within a range of plant materials. Results indicated that detection of the larvae of wood-boring species within hard plant materials was possible. Recognition procedures enabled sufficiently accurate identification to species level to enable the development of systems that could potentially be used by PHSI and FC Inspectors for location of infected material in the field. As in current procedures, specimens of suspect insects would then be returned to expert taxonomists for confirmatory identification. The development of an effective practical system also requires close cooperation with potential end-users (Defra PHSI; FC) to ensure their requirements are fully addressed. The proposed project will refine the techniques developed in the earlier project and produce two systems for use by the PHSI and FC. The first system, potentially based around handheld computers (possibly with Bluetooth communication from a sensor that the inspector can attach to the suspect plant material though the project will determine compatibility issues with existing or planned PHSI/FC equipment) will provide real-time determination of the presence of quarantine species, and where required, an initial identification. It is anticipated that use (even in conjunction with conventional methods) will increase the probability of detection by enabling a larger sample of plant material to be inspected non-destructively during a single visit. The second system will be capable of stand-alone operation and will act through a datalogging system, which can be used during transport of suspect material as well as at the point of import or at in-land points of destination. The latter system will be capable of being left unattended for extended periods while operating, if required. Again it will detect the presence of suspect insects and provide initial identification. In addition to its potential to improve sample size and detection rates, the second system may also improve detection of relatively inactive insects by increasing sampling time without the need for constant attendance of the equipment by an inspector. To ensure that practical considerations and constraints are taken into account, and to facilitate the development of systems that are fit for purpose, it is proposed that all work under this project will be undertaken in discussion with the PHSI/FC, and the Forestry Commission, and that PHSI inspectors will be involved at all stages but particularly during the field testing procedures. The involvement of the PHSI and FC has been discussed and agreed in advance of the submission of this proposal. Work undertaken under the programme would include: Final consideration of optimal sensor selection, including an investigation into the use of low-cost two and three axis single chip accelerometers as well as broad band piezoelectric sensors and microphones. In addition, further refinement of the species identification stages of the process, involving artificial neural networks and other pattern recognition approaches, will be undertaken in order to maximise efficiency of insect detection, correct species identification and to optimise rejection of non-insect sounds such as natural and man-made sounds. One specific approach will be to investigate the temporal biting patterns (e.g. number of bites per minute, or average inter-bite interval) to improve the discrimination of bites against random and non-random background noise and interference. The work conducted under the earlier feasibility study (project PH0191) made use of single bites for species identification; this will be extended to accumulation of identification over multiple bites to increase accuracy. It is anticipated that this project will conclude with the development of a portable acoustic detection devices trained for a range of named quarantine wood-boring species infesting hard plant materials. Field testing of the system will be undertaken in conjunction with end users and equipment developed during this testing programme will be provided to the PHSI and FC as appropriate for further validation (if required) and use.
United Kingdom