Timely crop planting is a foundation for climate-resilient rice-wheat systems of the Eastern Gangetic Plains—a global food insecurity and poverty hotspot. We hypothesize that the capacity of individual farmers to plant on time varies considerably, shaped by multifaceted enabling factors and constraints that are poorly understood. To address this knowledge gap, two complementary datasets were used to characterize drivers and decision processes that govern the timing of rice planting in this region. The first dataset was a large agricultural management survey (rice-wheat: n = 15,245; of which rice: n = 7597) from a broad geographic region that was analyzed by machine learning methods. The second dataset was a discussion-based survey (n = 112) from a more limited geography that we analyzed with graph theory tools to elicit nuanced information on planting decisions. By combining insights from these methods, we show for the first time that differences in rice planting times are primarily shaped by ecosystem and climate factors while social factors play a prominent secondary role. Monsoon onset, surface and groundwater availability, and land type determine village-scale mean planting times whereas, for resource-constrained farmers who tend to plant later ceteris paribus, planting is further influenced by access to farm machinery, seed, fertilizer, and labor. Also, a critical threshold for economically efficient pumping appears at a groundwater depth of around 4.5 m; below this depth, farmers do not irrigate and delay planting. Without collective action to spread risk through synchronous timely planting, ecosystem factors such as threats posed by pests and wild animals may further deter early planting by individual farmers. Accordingly, we propose a three-pronged strategy that combines targeted strengthening of agricultural input chains, agroadvisory development, and coordinated rice planting and wildlife conservation to support climate-resilient agricultural development in the Eastern Gangetic Plains.

适时的作物种植是东恒河平原气候适应型稻麦系统的基础——这是全球粮食不安全和贫困的热点地区。我们假设个体农民按时种植的能力差异很大，受多方面促成因素和知之甚少的制约因素影响。为了解决这一知识差距，两个互补的数据集被用来描述控制该地区水稻种植时间的驱动因素和决策过程。第一个数据集是 来自广泛地理区域的大型农业管理调查（水稻-小麦：n  = 15,245；其中水稻：n = 7597），通过机器学习方法进行了分析。第二个数据集是基于讨论的调查（n = 112）来自我们使用图论工具分析的更有限的地理环境，以得出有关种植决策的细微差别信息。通过结合这些方法的见解，我们首次表明水稻种植时间的差异主要受生态系统和气候因素的影响，而社会因素则起着突出的次要作用。季风开始、地表水和地下水的可用性以及土地类型决定了村庄规模的平均种植时间，而对于资源有限的农民来说，在其他条件不变的情况下倾向于较晚种植，种植进一步受到农业机械、种子、肥料和劳动力的影响。此外，经济高效抽水的临界阈值出现在约 4.5 m 的地下水深度处；低于这个深度，农民不灌溉和延迟种植。如果不采取集体行动通过同步及时种植来分散风险，有害生物和野生动物构成的威胁等生态系统因素可能会进一步阻碍个体农民的早期种植。因此，我们提出了一个三管齐下的战略，结合有针对性地加强农业投入链、农业指导开发以及协调水稻种植和野生动物保护，以支持恒河平原东部气候适应型农业发展。