Compositional nutrient diagnosis and associated yield predictions in maize: A case study in the northern Guinea savanna of Nigeria

Abstract

Developing optimal strategies for nutrient management of soils and crops at a larger scale requires an understanding of nutrient limitations and imbalances. The availability of extensive data (n = 1,781) from 2‐yr nutrient omission trials in the most suitable agroecological zone for maize (Zea mays L.) in Nigeria (i.e., the northern Guinea savanna) provides an opportunity to assess nutrient limitations and imbalances using the concept of multi‐ratio compositional nutrient diagnosis (CND). We also compared and contrasted the use of linear regression models and bootstrap forest machine learning to predict maize yield based on nutrient concentration in ear leaves. The results showed that 35% of the experimental plots had low yields due to nutrient imbalances (hereafter referred to as low yield imbalanced [LYI]). These experimental plots were dominated by control plots (without any nutrients applied), plots without N fertilization, and plots without P fertilization. Using the control plot as the ultimate indicator of nutrient imbalance, the significantly limiting nutrients in order of decreasing frequency of deficiency were N, P, S, Ca > Cu, and B. Both linear regression and bootstrap forest machine learning models fairly predicted maize grain yield based on nutrient concentration in ear leaves only in the LYI group and when examining all data with an independent validation dataset. These results suggest that nutrient management strategies, especially through the site‐specific management approach, should consider S, Ca, Cu, and B in addition to the existing nutrients N, P, and K to improve nutrient balance and maize yield in the study area.

Research on Evaluation Methods of Black Soil Farmland Productivity Based on Field Block Scale

“…Machine learning models can effectively avoid this evaluation uncertainty caused by empirical knowledge and subjective judgement, and by learning from soil data, assessment models can be built more quickly and accurately. Zhou et al [14] used a combination of optical and radar remote sensing data to apply the SVM algorithm to build a Soil organic C (SOC) prediction model; Zou et al [15] collected historical soil data from southern China and combined multivariate linear model (MLM) and mixed effects regression model (MEM) for soil productivity assessment; Shehu et al [16] obtained 1781 sets of maize farmland data comparison in Northern Nigeria using linear regression models, as well as random forest machine learning to predict maize yields based on nutrient concentrations in spike leaves; Pan Y et al [17] provided an estimate of land productivity in the conterminous United States of America (CONUS) through machine learning algorithms using a data-driven approach to incorporate relationships from the data into the land productivity evaluation. However, challenges remain in terms of applicability and interpretability of machine learning models [18], including the requirement of datasets (e.g., combining large remote sensing datasets and larger historical datasets) and due to the black box nature of machine learning resulting in little insight into agricultural management.…”