Stable-Yield Rice Variety Screening Based on Seed-Setting Rate Thresholds and Machine Learning: A Case Study of the Luoding Regional Trial
1. Luoding Agricultural Development Center, Luoding 527200, China
2. Rice Research Institute, Guangdong Academy of Agricultural Sciences, Guangzhou 510640, China
| Abstract: | [Objective] To develop a machine-learning-assisted stable-yield screening framework for double-cropping rice regions, providing a data-driven auxiliary reference for variety certification and agricultural extension. [Method] Proposing a dual-track screening framework that integrates rule-based thresholds with machine learning, based on 230 valid trial materials spanning 3 years and 6 cropping seasons (2022-2024) at the Luoding regional trial site, core agronomic traits including yield, seed-setting rate, filled grains per panicle, 1000-grain weight, plant height, and growth duration were obtained through field investigation and laboratory testing. A dual-track framework was constructed: Track I used a rule-based threshold (seed-setting rate >=68% and yield increase over control >=0%) for preliminary screening, with K-nearest neighbor (KNN, K=5) binary classification as a consistency check; Track II employed logistic regression (LR) three-class classification for yield prediction, evaluated by repeated cross-validation. [Result] The 2023 late season, affected by typhoon-induced pest outbreaks, experienced a 52.6% yield reduction. The rule-based threshold screened 24 stable-yield materials (detection rate 72.7%), with a mean seed-setting rate of 77.3%, and KNN classification agreed with the rule at 72.7% (this is an internal consistency check between rule and classifier, not an independent prediction, since seed-setting rate is both a rule condition and a model input feature). Spearman rank correlation between seed-setting rate and yield was rho=0.411 (P<0.001). LR achieved repeated cross-validation accuracy of 58.9%+/-6.5% (about 1.8x the 33.3% random baseline); KNN binary classification accuracy was 66.1%+/-6.4%. The above key correlation factors remained significant after Bonferroni correction (alpha'=0.05/6 approx 0.0083), indicating robustness to multiple comparisons. [Conclusion] The dual-track framework integrating rule-based screening with data-driven classification can assist in identifying stable-yield materials under typhoon-impacted conditions, providing a low-cost technical solution for Luoding and ecologically similar regions; cross-regional adaptability awaits multi-site validation. |
| Keywords: | Rice; Variety Screening; Yield Stability; Seed-setting Rate; Machine Learning |
| DOI: | 10.57237/j.jaf.2026.01.002 |
View:
Downloads:
Received: 27 July 2026
Accepted: 13 August 2026
Published Online: 7 September 2026
Funding
| 1. | 省级乡村振兴战略专项 (种业振兴行动) 项目——农作物品种试验 (No. 2024-445381-69010000-0001) |
References
| [1] | 梁中尧, 李润东, 李金玲, 等. 罗定市环境友好型有机水稻生产实践分析 [J]. 农业技术与装备, 2022(2): 55-57. |
| [2] | 智琳然, 孙菊英, 柯瑷, 等. 基于AMMI模型和GGE双标图的水稻区域试验综合分析 [J]. 中国种业, 2025(2): 89-94. https://doi.org/10.19462/j.cnki.zgzy.20241115001 |
| [3] | 杨婷婷, 崔金凤, 吴文革, 杨安中. 21个杂交水稻品种主要农艺性状与产量的灰色关联度分析 [J]. 安徽科技学院学报, 2015, 29(3): 1-5. |
| [4] | 张春磊, 李颜娥, 丁煜, 等. 基于深度学习技术的水稻环境因素产量预测 [J]. 电子技术应用, 2024, 50(4): 81-86. https://doi.org/10.16157/j.issn.0258-7998.234657 |
| [5] | 陈嘉铧, 吴浩荣, 甘锦洪, 等. 改进的BP神经网络在水稻产量预测中的应用 [J]. 智能计算机与应用, 2025, 15(11): 170-177. https://doi.org/10.20169/j.issn.2095-2163.251128 |
| [6] | 胡林. 基于深度学习的田间水稻稻穗识别与产量预测研究 [D]. 武汉: 华中农业大学, 2024. https://doi.org/10.27158/d.cnki.ghznu.2024.000596 |
| [7] | 周曙东, 朱红根. 气候变化对中国南方水稻产量的经济影响及其适应策略 [J]. 中国人口·资源与环境, 2010, 20(10): 152-157. https://doi.org/10.3969/j.issn.1002-2104.2010.10.026 |
| [8] | 陈风波, 张梨靖. 热带气旋对南方水稻产量影响的实证检验——基于广东水稻种植户的调研数据分析 [J]. 广东农业科学, 2024, 51(9): 54-65. https://doi.org/10.16768/j.issn.1004-874X.2024.09.005 |
| [9] | 曾恺莹, 赖明建, 田瑞钧, 等. 广东省水稻品种区试中对照品种的特点与趋势 [J]. 广东农业科学, 2023, 50(12): 83-95. https://doi.org/10.16768/j.issn.1004-874X.2023.12.008 |
| [10] | 黄道强, 周少川, 李宏, 等. 优质稻新品种五山丝苗的选育及利用 [J]. 广东农业科学, 2011, 38(9): 15-16. https://doi.org/10.16768/j.issn.1004-874x.2011.09.043 |
| [11] | 林莉, 黄超, 文西强. 余庆县优质水稻品种的筛选 [J]. 农技服务, 2024, 41(10): 35-39. |
| [12] | 邓婉蓉. 安顺地区粳型水稻品种引种比较试验 [J]. 农技服务, 2024, 41(8): 28-33. |
| [13] | Dheer P, Yadav P, Katiyar P K. Comparative accuracy of different machine learning classifiers for characterizing varieties of pulse crops [J]. Journal of Food Legumes, 2019, 32(1): 53-56. https://doi.org/10.59797/jfl.v32i1.706 |
| [14] | Fu J, Jian Y, Wang X, et al. Extreme rainfall reduces one-twelfth of China's rice yield over the last two decades[J]. Nature Food, 2023, 4: 416-426. https://doi.org/10.1038/s43016-023-00753-6 |
| [15] | Kamilaris A, Prenafeta-Boldú FX. Deep learning in agriculture: A survey [J]. Computers and Electronics in Agriculture, 2018, 147: 70-90. https://doi.org/10.1016/j.compag.2018.02.016 |
Join the Editorial Board
We invite active, qualified and high profile scientists and researchers to join as Editorial Board Members.
Join UsBecome a Reviewer
Scholars with a strong interest in reviewing are invited to join the reviewer panel to ensure the quality of the research to be published.
Join Us