Volatiles from Russet Burbank potatoes inoculated with Erwinia carotovora subsp. carotovora, E. carotovora subsp. atroseptica, Pythium ultimum, Phytophthora infestans, or Fusarium sambucinum were monitored by sampling the head space 3, 4, and 5 days after inoculation, using a solid phase microextraction (SPME) fiber to trap and gas chromatography with flame ionization detector (GC-FID) to fingerprint volatiles. Noninoculated (NON) potatoes served as the control. Volatile fingerprints varied among diseases. Within a disease, the fingerprints varied with time since inoculation and among blocks. In general, more volatiles were observed on the fourth and fifth day after inoculation than on the third day. The amount of volatile compounds produced (peak area) within a disease group increased with incubation time; however, the variation among blocks was much higher. The amount of volatiles produced, in general, was associated with disease severity. Disease-specific volatiles were observed. The F. sambucinum chromatogram had two unique peaks at retention time (RT) = 14.1 and 17.3 min. P. infestans produced few peaks and the profile was quite similar to NON. In contrast, E. carotovora subsp. carotovora, E. carotovora subsp. atroseptica, and Pythium ultimum produced many peaks, and the P. ultimum was different from the bacteria, in that the chromatogram peaks at RT = 4.04 and 8.76 min were absent. Instead, it produced a distinct peak at RT = 1.71 min. E. carotovora subsp. carotovora and E. carotovora subsp. atroseptica couldn't be discriminated based on unique peaks; however, they varied in concentration of volatiles produced. E. carotovora subsp. carotovora produced more of RT = 2.0 min and less of RT = 2.3 and 2.44 min than E. carotovora subsp. atroseptica. A back-propagation network (using neural networks) was developed to classify volatile profiles into six disease-groups. Cross-validation classification probabilities were NON = 71, E. carotovora subsp. carotovora = 71, E. carotovora subsp. atroseptica = 71, P. ultimum = 67, Phytophthora infestans = 46, and F. sambucinum = 75%.
Keywords: disease detection; disease monitoring; electronic nose; neural network; postharvest pathology.