AN EARLY WARNING SYSTEM FOR DETECTION OF DISEASE IN STORED POTATOES
PTN SPENCER-PHILLIPS, BPJ deLACY COSTELLO, RJ EWEN, P EVANS, HE GUNSON and NM RATCLIFFE
Faculty of Applied Sciences, University of The West of England, Coldharbour Lane, Bristol, BS16 1QY, UK
Background and objectives
One of the major causes of losses of stored potato tubers is soft-rot from infection by Erwinia carotovora subsp. carotovora (ECC). Currently, visual inspection and temperature measurement are the most frequently used means of detecting infection. However, these methods are only effective at detecting late stages of disease development, at which point there is already significant potato damage. Our objective,was to develop a system to detect the disease at low levels of infection before significant loss had occurred. This was undertaken by monitoring and identifying the volatile organic compounds (VOCs) evolved during disease progression and using the data obtained to design and build a prototype sensory array aimed at differentiating between infected and uninfected potato tubers .
Materials and methods
A simplified model system was developed to emulate the conditions in which potato tubers were stored for a 4-week sampling period. The VOCs evolved were sampled using commercial chromatographic adsorbents and identified by GC-MS after using thermal desorption to release the captured volatiles. Commercial and in-house gas and vapour sensors were then evaluated against a number of the more significant volatiles from infected and uninfected tubers. A panel of these sensors was incorporated into a prototype sensory array and combined with suitable computing algorithms to determine the presence or absence of ECC infection in tubers incubated in the model system.
Results and conclusions
Volatile profiles of ECC-infected potato tubers showed a significant increase in the volume of low molecular weight volatiles as compared to uninfected tubers. No single volatile presented itself as a marker compound indicative of infection, but despite this the development of a sensor array capable of distinguishing between infected and uninfected tubers was possible. Rather than develop a selective sensor capable of detecting the very low levels of a specific compound associated with a particular disease, pattern recognition techniques and neural computing have been used in conjunction with the sensor array. The patterns generated from the sensor array are unique to the volatiles present. Consequently, an array may be trained against a set of calibration vapours before exposing them to real analyte odours. A range of sensors was evaluated against individual volatiles and infected tubers before a panel was selected. Development of a portable system based upon the sensors identified is now under way for testing in a commercial store.
Thanks are due to the British Potato Council for funding this project.
1. Delacy Costello BPJ, Ewen RJ, Ratcliffe NM, 1996. Diagnostics in Crop Production 65, 383-390.