PREDICTING PEANUT MATURITY WITH MAGNETIC-RESONANCE

Knowledge of peanut (Arachis hypogaea) maturity is crucial in harvest timing for minimizing aflatoxin and maximizing harvest yield. Low reso lution pulse nuclear magnetic resonance (NMR) was explored as an alter native to current maturity evaluation methods which are based on pod c olor as determined using the hull-scrape approach. For the 1992 throug h 1995 seasons, peanuts (cv. Florunner) were sampled weekly over a 3 t o 5 week period. Nearly 200 kernels per week were analyzed by hull-scr ape, gravimetric and NMR methods. The NMR data consisted of the Free I nduction Decay peak as observed at 20 mu s (FIDPK herein), FIDPK + 20 mu s (FID40) and spin-echo at 2000 mu s (ECHO). The FIDPK and the FID4 0 each strongly increased nonlinearly with maturity class as did ECHO; but to a lesser extent. Data from 1992-94 were processed to select ra ndomly an equal number of peanuts in each of six maturity levels. This data set was then divided into a discriminant classifier model buildi ng set (2/3) and a validation set (1/3). Chi square values based on pr edicted versus observed maturity distributions exceeded the P less tha n or equal to 0.05 value of 12.8; however the days to harvest from the classifier validation data set were nearly identical To that estimate d by the hull-scrape method. The maturity prediction model based solel y on the above mentioned NMR parameters predicted identical days to ha rvest as obtained from corresponding hull scrape data for a validation sample from the 1995 season.