Integrating soil, plant and microclimate variables for optimizing resource use and yield of greenhouse grown pepper in a rainforest environment

Abstract

An experiment was conducted to evaluate microclimate, growth, yield and water use (evapotranspiration: ET) of peppers (Capsicum species: Bell and Habanero) under greenhouse condition. Pepper plants were drip fertigated (irrigation at 100 and 70% field capacity) and soluble nutrient formulation (0, 60 and 100 of recommended rates of N P K compound fertilizer). A mobile weather station was installed with sensor networks for monitoring microclimate variables (solar radiation, minimum and maximum temperatures, humidity, wind speed and photosynthetic active radiation: PAR) as well as methane (CH₄) and carbon dioxide (CO₂). Pepper evapotranspiration was determined using Penman-Monteith and Hargreaves equations. Agronomic parameters were taken on pepper plants (height and leaf development, number and weight of fruits). Correlation and regression analyses were conducted between some weather and pepper yield variables. A diagnostic algorithm was evaluated using python programming language for yield simulation. Data collected were subjected to statistical analysis using ANOVA test and e significant treatment means were separated at 5% level of probability. Results showed that the fertigation regimes significantly influenced the growth and yield of both habanero and bell peppers. Habanero performed best with F₂W₁ (100 kg N/ha + 70% Fc), while Bell pepper benefits from moderate irrigation F₁W₁ (60 kg N/ha + 70% Fc). Tailoring fertigation regimes will enhance productivity and resource efficiency of greenhouse cultivation. Habanero performed best with F₂W₁ (100 kg N/ha + 70% Fc), while Bell pepper benefits from moderate irrigation F₁W₁ (60 kg N/ha + 70% Fc). These findings confirm the critical role of regulated fertigation for optimizing pepper growth and yield in the greenhouse Maximum fertigation rates (100 kg ha^-1 NPK and 100% field capacity watering) produced best growth and yield variables of peppers followed by 100 kg ha^-1 and 70% Fc. High fertilizer rate combined with moderate watering (70% Fc) was optimal conditions for pepper under greenhouse condition. The time course of microclimate variables (temperature, humidity, PAR and wind speed) differed during the period of observation (March to June, 2024). Pepper water use (ET) was 4.6 mm day^-1 (Penman Monteith) and 5.1 mm day^-1 (Hargreave) while CH₄ and CO₂ were 29.4 and 8.7 ppm respectively. There were both positive and negative associations between pepper yield and water use and weather factors with correlation coefficients (R²) ranging from 0.90 (strong and positive) and 0.023 (weak and positive) and -0.62 (moderate negative) to -0.023 (very week negative). In particular, there were strong but negative correlations between temperature-related variables (maximum temperature (Tmax) and growing degree days (GDD) and fruit weight and water use (ET) and between humidity and Tmax and GDD. The Gradient Boosting Model predicted pepper yield in the greenhouse based on metrics of Mean Absolute Error (MAE: 18.34) and Root Mean Squared Error (RMSE: 24.01). Both MAE and RMSE were used to assess the predictive performance of the model. Information generated on weather, soil and plant can serve as inputs in the development of control system for improving crop yield and resource efficiency of greenhouse practice. Integration of sensor networks with machine learning algorithm offer opportunity for improving real-time decision-making for greenhouse crop production.