Predictive model for moisture content of Coffea liberica seeds subjected toambient air drying

Determination of seed moisture content (MC) is a destructive method according to the Department of Standards Malaysia and the International Seed Testing Association (ISTA) where the seeds were dried at 103±2℃ for 16±1 hrs in a convection oven to know the amount of water within seeds. The present work analysed the relationship between the weight and MC of Coffea liberica seeds so that the seeds with MC estimates could subsequently be applied as planting materials or for germplasm storage for future varietal improvement for the beverage sector. From the fully ripe fruits, single seeds, two seed units, three seed units, and four seed units were studied for suitability in developing a predictive model for MC. In each context, seeds of varying MCs for model building were obtained with air drying for 0, 2, 4, 6 and 8 days at the ambient temperature of 28±2℃ and relative humidity (RH) of 55±5% in the laboratory. Prior to desiccation treatments, the seed units were tagged and noted for fresh weight (FW), length and breadth. After desiccation treatments, weight after desiccation (DesW) was recorded and the actual MC of the seeds was determined with the oven-drying method. Plots of MC - desiccation period showed that ambient air-drying reduced the MC of seeds progressively and consistently. With the initial MC of 0.4 g H2O g-1 Wt on a wet weight (Wt) basis, air drying of the seeds in the laboratory for up to 8 days reduced their MC to 0.15 g H2O g-1 Wt. Subsequent correlation analysis revealed that MC had a positive relationship with DesW, but was negatively associated with (Loss of water), which is the difference between FW and DesW. Stepwise linear regression analysis implied that the MC of C. liberica seeds could best be estimated as MC = 0.185 + 0.118 FW – 0.412 (Loss of water) using two seed units. The model was statistically significant (P < 0.001) with appropriate R 2 and SEE values of 0.95 and 0.0187 g, respectively.