Revealing the spatial characteristics of rice heat exposure in Japan through panicle temperature analysis

Abstract

Elevated temperatures during the flowing stage contribute to heat-induced spikelet sterility in rice, posing a major threat to production considering climate change projections. Developing effective strategies for stable rice production through breeding and crop management is critical; however, our understanding of regional, seasonal, and long-term trends in rice heat exposure remains limited. Previous studies on spikelet sterility revealed that panicle temperature, estimated using a micrometeorological model and common meteorological factors, serves as a reliable indicator of rice heat exposure. In this study, we employed this model to identify the differences between panicle and air temperatures (DPAT) and their causes over the past 45 years in Japan. A gridded daily meteorological dataset covering Japan was interpolated at an hourly time step and used as input data of the micrometeorology model for estimating panicle temperatures during flowering. Statistical analysis of the resulting data revealed an increasing trend in the frequency of rice panicle heat exposure over time across many locations in Japan. During heat-receptive periods, panicle temperature generally exceeded air temperature, indicating the inadequacy of relying solely on air temperature to gauge rice heat stress. DPAT values showed substantial inter-regional variations in both mean values (from -0.5 to 3.0) and seasonality. Through machine learning and statistical methods, the relationship between DPAT and meteorological factors was characterized, delineating the effects of the meteorological factors on regional and seasonal DPAT variations. Focusing on major high-risk regions, we show that mitigation strategies should be adapted to consider regional characteristics and avoid high DPAT conditions during rice heading periods.

Plant phenotyping relevance

水稲穂温という植物状態を微気象モデルで推定し、推定手法を用いた長期・地域比較と機械学習解析が研究の中心であるため、計算型フェノタイピングの応用として含める。

Code and data availability

The paper describes model-based panicle temperature estimation (IM2PACT) using gridded meteorological data, but no author-generated phenotype dataset, analysis code, model checkpoints, or data deposit is mentioned. The only public URL cited (MLIT Land Use Mesh) is a generic third-party land-use dataset used as an input

No evidence-backed public reproduction asset is currently recorded.