Predicting Specific Biogas Yield of Maize-Validation of Different Model Approaches

Abstract

A reliable tool for predicting the specific biogas yield (SBY) of maize is required for breeding purposes to support a more efficient biomass production and for a quality-based payment of maize-substrate supply. The objective of the current study was to validate the recently published prediction model by Rath et al. (Bioenergy Research 6:939–952) with an independent data set and to compare its predictive ability to the approaches of four previously reported models (Baserga, Keymer and Schilcher, Kaiser, Weisbach). The validation data set was based on a multisite field experiment, providing a large genotypic variation in maize chemical composition and SBY, which ranged between 612 and 826 lN kg−1 OM. Predicted and measured SBY were positively correlated (r = 0.15 to r = 0.48) for all approaches tested. The model by Rath et al. revealed the highest predictive ability and was the only approach that allowed the variability of SBY within the data set to be reflected. This could be attributed to an extensive and reliable calibration database covering the genotype × environment interaction, which allowed a thorough examination of the assumptions before conducting a multiple linear regression analysis. Even with this provision, sufficient samples per genotype are obviously required to predict satisfactorily the ranking of maize genotypes with respect to SBY. Using genotype averages over sites, instead of single-location values, is recommended to reduce the strong impact of weather conditions on the chemical composition, thereby enabling a more reliable estimate of the genotype SBY potential.