Machine learning reveals drivers of yield sustainability in five decades of continuous rice cropping

The long-term sustainability of intensive rice systems under climate change is a critical challenge for global food security. Here, we use machine learning techniques to assess the impact of climate change, genotype, and nutrient management on rice yield in the world's longest-running continuous cro...

Full description

Bibliographic Details
Main Authors: Yamaguchi, Tomoaki, Angeles, Olivyn, Iizumi, Toshichika, Dobermann, Achim, Katsura, Keisuke, Saito, Kazuki
Format: Journal Article
Language:Inglés
Published: Elsevier 2025
Subjects:
Online Access:https://hdl.handle.net/10568/176855
_version_ 1855528244131921920
author Yamaguchi, Tomoaki
Angeles, Olivyn
Iizumi, Toshichika
Dobermann, Achim
Katsura, Keisuke
Saito, Kazuki
author_browse Angeles, Olivyn
Dobermann, Achim
Iizumi, Toshichika
Katsura, Keisuke
Saito, Kazuki
Yamaguchi, Tomoaki
author_facet Yamaguchi, Tomoaki
Angeles, Olivyn
Iizumi, Toshichika
Dobermann, Achim
Katsura, Keisuke
Saito, Kazuki
author_sort Yamaguchi, Tomoaki
collection Repository of Agricultural Research Outputs (CGSpace)
description The long-term sustainability of intensive rice systems under climate change is a critical challenge for global food security. Here, we use machine learning techniques to assess the impact of climate change, genotype, and nutrient management on rice yield in the world's longest-running continuous cropping experiment (LTCCE) at the International Rice Research Institute (IRRI) in the Philippines. In the experiment, three to six rice genotypes were cultivated from 1968 to 2017 in three annual cropping seasons—dry, early wet, and late wet seasons—with four nitrogen (N) fertilizer treatments. These genotypes were changed regularly to utilize the best high-yielding, disease- and insect-resistant varieties available at a given time. Our analysis showed that nitrogen application, varietal replacement, solar radiation, and seasonal temperature patterns were major determinants of yield variation. While nitrogen and solar radiation consistently improved yield irrespective of seasons, temperature effects were season-specific. In the dry season, lower temperatures during reproductive and ripening stages were beneficial. In the early wet season, yield gains were observed under higher vegetative-stage temperatures. Enhanced nitrogen mineralization and improved early rice growth may be contributing factors. The late wet season was constrained by low radiation, high disease pressure, and declining N response with prolonged varietal use. These findings demonstrate the value of combining long-term yield data with weather information to assess sustainability in intensive rice systems under increasing climatic and biotic pressures. They also illustrate the need for seasonally tailored and integrated crop, nutrient, and pest management practices, including more frequent variety replacement and rotating varieties between seasons. Breeding dry season varieties with reduced respiration losses and wet season varieties with improved tolerance to humid, low-radiation conditions can play a crucial role in enhancing seasonal adaptation and overall productivity.
format Journal Article
id CGSpace176855
institution CGIAR Consortium
language Inglés
publishDate 2025
publishDateRange 2025
publishDateSort 2025
publisher Elsevier
publisherStr Elsevier
record_format dspace
spelling CGSpace1768552025-10-26T13:02:45Z Machine learning reveals drivers of yield sustainability in five decades of continuous rice cropping Yamaguchi, Tomoaki Angeles, Olivyn Iizumi, Toshichika Dobermann, Achim Katsura, Keisuke Saito, Kazuki oryza sativa intensive farming climate change nitrogen fertilizers nutrient management genotype-environment yields varieties long-term experiments The long-term sustainability of intensive rice systems under climate change is a critical challenge for global food security. Here, we use machine learning techniques to assess the impact of climate change, genotype, and nutrient management on rice yield in the world's longest-running continuous cropping experiment (LTCCE) at the International Rice Research Institute (IRRI) in the Philippines. In the experiment, three to six rice genotypes were cultivated from 1968 to 2017 in three annual cropping seasons—dry, early wet, and late wet seasons—with four nitrogen (N) fertilizer treatments. These genotypes were changed regularly to utilize the best high-yielding, disease- and insect-resistant varieties available at a given time. Our analysis showed that nitrogen application, varietal replacement, solar radiation, and seasonal temperature patterns were major determinants of yield variation. While nitrogen and solar radiation consistently improved yield irrespective of seasons, temperature effects were season-specific. In the dry season, lower temperatures during reproductive and ripening stages were beneficial. In the early wet season, yield gains were observed under higher vegetative-stage temperatures. Enhanced nitrogen mineralization and improved early rice growth may be contributing factors. The late wet season was constrained by low radiation, high disease pressure, and declining N response with prolonged varietal use. These findings demonstrate the value of combining long-term yield data with weather information to assess sustainability in intensive rice systems under increasing climatic and biotic pressures. They also illustrate the need for seasonally tailored and integrated crop, nutrient, and pest management practices, including more frequent variety replacement and rotating varieties between seasons. Breeding dry season varieties with reduced respiration losses and wet season varieties with improved tolerance to humid, low-radiation conditions can play a crucial role in enhancing seasonal adaptation and overall productivity. 2025-11 2025-10-07T06:42:02Z 2025-10-07T06:42:02Z Journal Article https://hdl.handle.net/10568/176855 en Limited Access Elsevier Yamaguchi, Tomoaki, Olivyn Angeles, Toshichika Iizumi, Achim Dobermann, Keisuke Katsura, and Kazuki Saito. "Machine learning reveals drivers of yield sustainability in five decades of continuous rice cropping." Field Crops Research 333 (2025): 110114.
spellingShingle oryza sativa
intensive farming
climate change
nitrogen fertilizers
nutrient management
genotype-environment
yields
varieties
long-term experiments
Yamaguchi, Tomoaki
Angeles, Olivyn
Iizumi, Toshichika
Dobermann, Achim
Katsura, Keisuke
Saito, Kazuki
Machine learning reveals drivers of yield sustainability in five decades of continuous rice cropping
title Machine learning reveals drivers of yield sustainability in five decades of continuous rice cropping
title_full Machine learning reveals drivers of yield sustainability in five decades of continuous rice cropping
title_fullStr Machine learning reveals drivers of yield sustainability in five decades of continuous rice cropping
title_full_unstemmed Machine learning reveals drivers of yield sustainability in five decades of continuous rice cropping
title_short Machine learning reveals drivers of yield sustainability in five decades of continuous rice cropping
title_sort machine learning reveals drivers of yield sustainability in five decades of continuous rice cropping
topic oryza sativa
intensive farming
climate change
nitrogen fertilizers
nutrient management
genotype-environment
yields
varieties
long-term experiments
url https://hdl.handle.net/10568/176855
work_keys_str_mv AT yamaguchitomoaki machinelearningrevealsdriversofyieldsustainabilityinfivedecadesofcontinuousricecropping
AT angelesolivyn machinelearningrevealsdriversofyieldsustainabilityinfivedecadesofcontinuousricecropping
AT iizumitoshichika machinelearningrevealsdriversofyieldsustainabilityinfivedecadesofcontinuousricecropping
AT dobermannachim machinelearningrevealsdriversofyieldsustainabilityinfivedecadesofcontinuousricecropping
AT katsurakeisuke machinelearningrevealsdriversofyieldsustainabilityinfivedecadesofcontinuousricecropping
AT saitokazuki machinelearningrevealsdriversofyieldsustainabilityinfivedecadesofcontinuousricecropping