Multilayer perceptron-genetic algorithm as a promising tool for modeling cultivation substrate of Auricularia cornea Native to Iran

Abstract

Auricularia cornea Ehrenb (syn. A. polytricha) is a wood-decaying fungi known as black ear mushroom. Earlike gelatinous fruiting body distinguishes them from other fungi. Industrial wastes have the potential to be used as the basic substrate to produce mushrooms. Therefore, 16 substrate formulations were prepared from different ratios of beech (BS) and hornbeam sawdust (HS) supplemented with wheat (WB) and rice brans (RB). The pH and initial moisture content of substrate mixtures were adjusted to 6.5 and 70%, respectively. The comparison of in vitro growth characteristics of the fungal mycelia under the different temperatures (25, 28, and 30°C), and culture media [yeast extract agar (YEA), potato extract agar (PEA), malt extract agar (MEA), and also HS and BS extract agar media supplemented with maltose, dextrose, and fructose revealed that the highest mycelial growth rate (MGR; 7.5 mm/day) belonged to HS and BS extract agar media supplemented with three mentioned sugar at 28°C. In A. cornea spawn study, the substrate combination of BS (70%) + WB (30%) at 28°C and moisture contents of 75% displayed the highest mean MGR (9.3 mm/day) and lowest spawn run period (9.0 days). In the bag test, “BS (70%) + WB (30%)” was the best substrate displaying the shortest spawn run period (19.7 days), and the highest fresh sporophore yield (131.7 g/bag), biological efficiency (53.1%) and number of basidiocarp (9.0/bag) of A. cornea. Also, A. cornea cultivation was processed to model yield, biological efficiency (BE), spawn run period (SRP), days for pinhead formation (DPHF), days for the first harvest (DFFH), and total cultivation period (TCP) by multilayer perceptron-genetic algorithm (MLP-GA). MLP-GA (0.81–0.99) exhibited a higher predictive ability than stepwise regression (0.06–0.58). The forecasted values of the output variables were in good accordance with their observed ones corroborating the good competency of established MLP-GA models. MLP-GA modeling exhibited a powerful tool for forecasting and thus selecting the optimal substrate for maximum A. cornea production.