Algorithmic differentiation of linear mixed models with variance-covariance structures

Fernando Henrique Toledo Jose Crossa Juan Burgueño Keith Gardner Rosa Angela Pacheco Gil (2023, [Objeto de congreso])

CIENCIAS AGROPECUARIAS Y BIOTECNOLOGÍA MATHEMATICAL MODELS ALGORITHMS LINEAR MODELS

Big data, small explanatory and predictive power: Lessons from random forest modeling of on-farm yield variability and implications for data-driven agronomy

Martin van Ittersum (2023, [Artículo])

Context: Collection and analysis of large volumes of on-farm production data are widely seen as key to understanding yield variability among farmers and improving resource-use efficiency. Objective: The aim of this study was to assess the performance of statistical and machine learning methods to explain and predict crop yield across thousands of farmers’ fields in contrasting farming systems worldwide. Methods: A large database of 10,940 field-year combinations from three countries in different stages of agricultural intensification was analyzed. Random effects models were used to partition crop yield variability and random forest models were used to explain and predict crop yield within a cross-validation scheme with data re-sampling over space and time. Results: Yield variability in relative terms was smallest for wheat and barley in the Netherlands and for wheat in Ethiopia, intermediate for rice in the Philippines, and greatest for maize in Ethiopia. Random forest models comprising a total of 87 variables explained a maximum of 65 % of cereal yield variability in the Netherlands and less than 45 % of cereal yield variability in Ethiopia and in the Philippines. Crop management related variables were important to explain and predict cereal yields in Ethiopia, while predictive (i.e., known before the growing season) climatic variables and explanatory (i.e., known during or after the growing season) climatic variables were most important to explain and predict cereal yield variability in the Philippines and in the Netherlands, respectively. Finally, model cross-validation for regions or years not seen during model training reduced the R2 considerably for most crop x country combinations, while for wheat in the Netherlands this was model dependent. Conclusion: Big data from farmers’ fields is useful to explain on-farm yield variability to some extent, but not to predict it across time and space. Significance: The results call for moderate expectations towards big data and machine learning in agronomic studies, particularly for smallholder farms in the tropics where model performance was poorest independently of the variables considered and the cross-validation scheme used.

Model Accuracy Model Precision Linear Mixed Models CIENCIAS AGROPECUARIAS Y BIOTECNOLOGÍA MACHINE LEARNING SUSTAINABLE INTENSIFICATION BIG DATA YIELDS MODELS AGRONOMY

Optimizing nitrogen fertilizer and planting density levels for maize production under current climate conditions in Northwest Ethiopian midlands

Kindie Tesfaye Dereje Ademe Enyew Adgo (2023, [Artículo])

This study determined the most effective plating density (PD) and nitrogen (N) fertilizer rate for well-adapted BH540 medium-maturing maize cultivars for current climate condition in north west Ethiopia midlands. The Decision Support System for Agrotechnology Transfer (DSSAT)-Crop Environment Resource Synthesis (CERES)-Maize model has been utilized to determine the appropriate PD and N-fertilizer rate. An experimental study of PD (55,555, 62500, and 76,900 plants ha−1) and N (138, 207, and 276 kg N ha−1) levels was conducted for 3 years at 4 distinct sites. The DSSAT-CERES-Maize model was calibrated using climate data from 1987 to 2018, physicochemical soil profiling data (wilting point, field capacity, saturation, saturated hydraulic conductivity, root growth factor, bulk density, soil texture, organic carbon, total nitrogen; and soil pH), and agronomic management data from the experiment. After calibration, the DSSAT-CERES-Maize model was able to simulate the phenology and growth parameters of maize in the evaluation data set. The results from analysis of variance revealed that the maximum observed and simulated grain yield, biomass, and leaf area index were recorded from 276 kg N ha−1 and 76,900 plants ha−1 for the BH540 maize variety under the current climate condition. The application of 76,900 plants ha−1 combined with 276 kg N ha−1 significantly increased observed and simulated yield by 25% and 15%, respectively, compared with recommendation. Finally, future research on different N and PD levels in various agroecological zones with different varieties of mature maize types could be conducted for the current and future climate periods.

Maize Model Planting Density CIENCIAS AGROPECUARIAS Y BIOTECNOLOGÍA MAIZE MODELS SPACING NITROGEN FERTILIZERS YIELDS

Using an incomplete block design to allocate lines to environments improves sparse genome-based prediction in plant breeding

Osval Antonio Montesinos-Lopez ABELARDO MONTESINOS LOPEZ RICARDO ACOSTA DIAZ Rajeev Varshney Jose Crossa ALISON BENTLEY (2022, [Artículo])

Genomic selection (GS) is a predictive methodology that trains statistical machine-learning models with a reference population that is used to perform genome-enabled predictions of new lines. In plant breeding, it has the potential to increase the speed and reduce the cost of selection. However, to optimize resources, sparse testing methods have been proposed. A common approach is to guarantee a proportion of nonoverlapping and overlapping lines allocated randomly in locations, that is, lines appearing in some locations but not in all. In this study we propose using incomplete block designs (IBD), principally, for the allocation of lines to locations in such a way that not all lines are observed in all locations. We compare this allocation with a random allocation of lines to locations guaranteeing that the lines are allocated to

the same number of locations as under the IBD design. We implemented this benchmarking on several crop data sets under the Bayesian genomic best linear unbiased predictor (GBLUP) model, finding that allocation under the principle of IBD outperformed random allocation by between 1.4% and 26.5% across locations, traits, and data sets in terms of mean square error. Although a wide range of performance improvements were observed, our results provide evidence that using IBD for the allocation of lines to locations can help improve predictive performance compared with random allocation. This has the potential to be applied to large-scale plant breeding programs.

CIENCIAS AGROPECUARIAS Y BIOTECNOLOGÍA Bayes Theorem Genome Inflammatory Bowel Diseases Models, Genetic Plant Breeding

Monitoreo e Instalación visual de señales a un motor eléctrico de inducción de instalación trifásica de forma jaula de ardilla, mediante la implementación de tecnología industria 4.0

Luis Ricardo Uribe Dávila (2023, [Tesis de maestría])

Vivimos la industria 4.0, misma que no es nueva, ya que sus orígenes se remontan a finales de la década de los 2000, en Alemania. Un pilar de la industria 4.0 es el análisis de datos, conocido como Big Data. El conocer los datos de un proceso, de un estudio, ayuda en gran medida a predecir el comportamiento que tendrá el proceso o la máquina a estudiar en un periodo a corto o mediano plazo. En el presente proyecto se analizan los datos arrojados por un motor eléctrico de corriente alterna, del tipo inducción, jaula de ardilla. El motor está diseñado para trabajar de manera continua, sin embargo, el uso que se le da, es meramente educativo; es decir, no sobre pasa las 15 horas por semana de uso. Mediante la toma de datos de las tres fases de corriente RMS o corriente de valor eficaz que posee el motor eléctrico que se realizará con el microcontrolador Arduino UNO, se analizarán los mismos mediante el software de cómputo numérico MATLAB, ordenando los datos, descartando valores que no aporten información relevante para lograr la predicción de datos. Por último, se llevará a conocer este proyecto a la carrera mecatrónica, área sistemas de manufactura flexible y área automatización, con el fin de que puedan observar de una mejor manera la aplicación y funcionamiento de uno de los pilares de la actual industria 4.0.

We live in industry 4.0, which is not new, since its origins date back to the late 2000s, in Germany. One pillar of industry 4.0 is data analysis, known as Big Data. Knowing the data of a process, of a study, helps greatly to predict the behavior that the process or machine will have to study in a short- or medium-term period. This project analyzes the data released by an electric motor of alternating current, of the type induction, squirrel cage. The engine is designed to work continuously, however, the use given to it is merely educational, that is; only not over spends 15 hours per week of use. By taking data from the three phases of RMS current or effective value current of the electric motor that will be made with the Arduino UNO micro controller, they will be analyzed using MATLAB numerical computing software, ordering the data, discarding values that do not provide relevant information to achieve data prediction. Finally, this project will be presented to the mechatronics career, flexible manufacturing systems area and automation area, so that they can observe in a better way the application and operation of one of the pillars of the current industry 4.0.

Mantenimiento predictivo Regresión lineal Industria 4.0 Big data Corriente RMS Predictive maintenance Linear regression Industry 4.0 Big data RMS Current INGENIERÍA Y TECNOLOGÍA CIENCIAS TECNOLÓGICAS OTRAS ESPECIALIDADES TECNOLÓGICAS OTRAS OTRAS

Point placement of late vegetative stage nitrogen splits increase the productivity, N-use efficiency and profitability of tropical maize under decade long conservation agriculture

Hari Sankar Nayak C.M. Parihar Shankar Lal Jat ML JAT Ahmed Abdallah (2022, [Artículo])

Non-Linear Growth Model Nitrogen Remobilization Right Placement Precision Nitrogen Management CIENCIAS AGROPECUARIAS Y BIOTECNOLOGÍA GROWTH MODELS NITROGEN NUTRIENT MANAGEMENT

Alternative cropping and feeding options to enhance sustainability of mixed crop-livestock farms in Bangladesh

Timothy Joseph Krupnik Jeroen Groot (2024, [Artículo])

We investigated alternative cropping and feeding options for large (>10 cows), medium (5–10 cows) and small (≤4 cows) mixed crop – livestock farm types, to enhance economic and environmental performance in Jhenaidha and Meherpur districts – locations with increasing dairy production – in south western Bangladesh. Following focus group discussions with farmers on constraints and opportunities, we collected baseline data from one representative farm from each farm size class per district (six in total) to parameterize the whole-farm model FarmDESIGN. The six modelled farms were subjected to Pareto-based multi-objective (differential evolution algorithm) optimization to generate alternative dairy farm and fodder configurations. The objectives were to maximize farm profit, soil organic matter balance, and feed self-reliance, in addition to minimizing feed costs and soil nitrogen losses as indicators of sustainability. The cropped areas of the six baseline farms ranged from 0.6 to 4.0 ha and milk production per cow was between 1,640 and 3,560 kg year−1. Feed self-reliance was low (17%–57%) and soil N losses were high (74–342 kg ha−1 year−1). Subsequent trade-off analysis showed that increasing profit and soil organic matter balance was associated with higher risks of N losses. However, we found opportunities to improve economic and environmental performance simultaneously. Feed self-reliance could be increased by intensifying cropping and substituting fallow periods with appropriate fodder crops. For the farm type with the largest opportunity space and room to manoeuvre, we identified four strategies. Three strategies could be economically and environmentally benign, showing different opportunities for farm development with locally available resources.

Ruminant Feed Pareto-Based Optimization Farm Bioeconomic Model CIENCIAS AGROPECUARIAS Y BIOTECNOLOGÍA RUMINANT FEEDING BIOECONOMIC MODELS MIXED CROPPING FARMS LIVESTOCK

Clonal diploid and autopolyploid breeding strategies to harness heterosis: insights from stochastic simulation

Marlee Labroo Jeffrey Endelman Dorcus Gemenet Christian Werner Robert Gaynor GIOVANNY COVARRUBIAS-PAZARAN (2023, [Artículo])

Reciprocal Recurrent Selection Clonal Diploids CIENCIAS AGROPECUARIAS Y BIOTECNOLOGÍA DIPLOIDY BREEDING HETEROSIS STOCHASTIC MODELS

Improving hybrid rice breeding programs via stochastic simulations: number of parents, number of hybrids, tester update, and genomic prediction of hybrid performance

Roberto Fritsche-Neto Marlee Labroo (2024, [Artículo])

Genomic Prediction Reciprocal Recurrent Selection Heterotic Pools CIENCIAS AGROPECUARIAS Y BIOTECNOLOGÍA STOCHASTIC MODELS RICE HYBRIDS GENETIC IMPROVEMENT GENETIC GAIN BREEDING PROGRAMMES

Modeling the growth, yield and N dynamics of wheat for decoding the tillage and nitrogen nexus in 8-years long-term conservation agriculture based maize-wheat system

C.M. Parihar Dipaka Ranjan Sena Prakash Chand Ghasal Shankar Lal Jat Yashpal Singh Saharawat Mahesh Gathala Upendra Singh Hari Sankar Nayak (2024, [Artículo])

Context: Agricultural field experiments are costly and time-consuming, and their site-specific nature limits their ability to capture spatial and temporal variability. This hinders the transfer of crop management information across different locations, impeding effective agricultural decision-making. Further, accurate estimates of the benefits and risks of alternative crop and nutrient management options are crucial for effective decision-making in agriculture. Objective: The objective of this study was to utilize the Crop Environment Resource Synthesis CERES-Wheat model to simulate crop growth, yield, and nitrogen dynamics in a long-term conservation agriculture (CA) based wheat system. The study aimed to calibrate the model using data from a field experiment conducted during the 2019-20-2020-21 growing seasons and evaluation it with independent data from the year 2021–22. Method: Crop simulation models, such as the Crop Environment Resource Synthesis CERES-Wheat (DSSAT v 4.8), may provide valuable insights into crop growth and nitrogen dynamics, enabling decision makers to understand and manage production risk more effectively. Therefore, the present study employed the CERES-Wheat (DSSAT v 4.8) model and calibrated it using field data, including plant phenological phases, leaf area index, aboveground biomass, and grain yield from the 2019-20-2020-21 growing seasons. An independent dataset from the year 2021–22 was used for model evaluation. The model was used to investigate the relationship between growing degree days (GDD), temperature, nitrate and ammonical concentration in soil, and nitrogen uptake by the crop. Additionally, the study explored the impact of contrasting tillage practices and fertilizer nitrogen management options on wheat yields. The experimental site is situated at ICAR-Indian Agricultural Research Institute (IARI), New Delhi, representing Indian Trans-Gangetic Plains Zone (28o 40’N latitude, 77o 11’E longitude and an altitude of 228 m above sea level). The treatments consist of four nitrogen management options, viz., N0 (zero nitrogen), N150 (150 kg N ha−1 through urea), GS (Green seeker based urea application) and USG (urea super granules @150 kg N ha−1) in two contrasting tillage systems, i.e., CA-based zero tillage (ZT) and conventional tillage (CT). Result: The outcomes exhibited favorable agreement between the model’s simulations and the observed data for crop phenology (With less than 2 days variation in 50% onset of flowering), grain and biomass yield (Root mean square error; RMSE 336 kg ha−1 and 649 kg ha−1, respectively), and leaf area index (LAI) (RMSE 0.28 & normalized RMSE; nRMSE 6.69%). The model effectively captured the nitrate-N (NO3−-N) dynamics in the soil profile, exhibiting a remarkable concordance with observed data, as evident from its low RMSE = 12.39 kg ha−1 and nRMSE = 13.69%. Moreover, as it successfully simulated the N balance in the production system, the nitrate leaching and ammonia volatilization pattern as described by the model are highly useful to understand these critical phenomena under both conventional tillage (CT) and CA-based Zero Tillage (ZT) treatments. Conclusion: The study concludes that the DSSAT-CERES-Wheat model has significant potential to assess the impacts of tillage and nitrogen management practices on crop growth, yield, and soil nitrogen dynamics in the western Indo-Gangetic Plains (IGP) region. By providing reliable forecasts within the growing season, this modeling approach can facilitate better planning and more efficient resource management. Future implications: The successful implementation of the DSSAT-CERES-Wheat model in this study highlights its applicability in assessing crop performance and soil dynamics. Future research should focus on expanding the model’s capabilities by reducing its sensitivity to initial soil nitrogen levels to refine its predictions further. Moreover, the model’s integration with decision support systems and real-time data can enhance its usefulness in aiding agricultural decision-making and supporting sustainable crop management practices.

Nitrogen Dynamics Mechanistic Crop Growth Models Crop Simulation CIENCIAS AGROPECUARIAS Y BIOTECNOLOGÍA NITROGEN CONSERVATION AGRICULTURE WHEAT MAIZE CROP GROWTH RATE SIMULATION MODELS