Abstract

Abstract
Background: Repeated application of chemical fertilizers has uplifted the productivity of agricultural crops but has deteriorated the soil quality in Malda District of West Bengal. There are different statistical models available, but ANN model is one of the few statistical modeling software which can predict the statistical correctness of any hypothesis based on already performed experiments. Here ANN have been successfully applied to model the effects of different nutrient sources such as organic manures, biofertilizers and chemical amendments on crop yield, biomass production, and soil fertility indices.
Methods: Different combinations (T1 to T10) of L. minor (L.) biomass were applied to the soil for cultivating Cicer arietinum L. (CV- IPL 220) with commercial inorganic fertilizer (negative control) and vermicompost (positive control). Effects were assessed based on different morphological and physiological aspects of C. arietinum after the 7th day and 14th day with SPSS v16 and v24 software.
Results: Different formulations showed variable impact on the morphological and biochemical parameters of C. arietinum L. (CV- IPL 220). Biochar from L. minor L. mixed Oscillatoria sancta (T5) showed 43.8% increase in R:S ratio over control plants of C. arietinum after the 7th day and 14th day. Similar observations were also made in case of SDW and RDW in comparison to the control sets. By training the network on experimental data, ANN model were applied to predict responses of C. arietinum to various fertilizer formulations, to validate the results as a multivariate model.
Conclusion: Comprehensive dataset were compiled to train and validate the ANN model to identify its efficacy in the prediction of growth promotion, minimizing experimental trials and aiming at minimum consumption of energy for maximum output.

Key words: Duckweeds, ANN, Malda, Lemna minor L. biofertilizers, Cicer arietinum (L.), Biochar, Shoot Dry weight (SDW), Root Dry Weight (RDW) and Root Length Density (RLD)