Accurate prediction of soil respiration (Rs) under climate change requires a comprehensive understanding of the dominant factors. However, the combined effects of multiple biophysical factors on long-term Rs remain uncertain due to the paucity of direct observations. Here, we developed a new method termed PLSR-lag, which integrates time lags and nonlinearity between variables into the Projection to Latent Structures Regression (PLSR) model. We employed PLSR-lag to explore the impacts of biophysical variables on Rs using 10-year field observations. Our results demonstrated a more significant correlation between nocturnal Rs and both photosynthetically active radiation (PAR) and surface soil temperature (STP) during growing season. Soil surface temperature (STP) dominates Rs throughout the year, while PAR, leaf area index (LAI), and soil surface moisture (SWC) modulate Rs during the growing season only. PLSR-lag significantly enhanced the explanatory power of growing-season Rs by 13–19 % compared to the original PLSR method. These findings provide evidence that accounting for the delayed response of Rs to photosynthetic proxies (such as PAR), alongside the nonlinear response to soil moisture, would improve model accuracy. This improvement would also enhance our understanding of soil carbon cycle in response to climate and environmental changes.

中文翻译：

---

温带落叶林生长季和非生长季白天和夜间土壤呼吸的生物物理控制

气候变化下土壤呼吸（Rs）的准确预测需要全面了解主导因素。然而，由于缺乏直接观察，多种生物物理因素对长期 Rs 的综合影响仍然不确定。在这里，我们开发了一种称为 PLSR-lag 的新方法，它将变量之间的时间滞后和非线性集成到潜在结构回归投影 (PLSR) 模型中。我们利用 PLSR-lag 通过 10 年的现场观测来探索生物物理变量对 Rs 的影响。我们的结果表明，生长季节夜间 Rs 与光合有效辐射 (PAR) 和表面土壤温度 (STP) 之间存在更显着的相关性。土壤表面温度 (STP) 全年主导 Rs，而 PAR、叶面积指数 (LAI) 和土壤表面湿度 (SWC) 仅在生长季节调节 Rs。与原始 PLSR 方法相比，PLSR-lag 将生长季 Rs 的解释力显着增强了 13-19%。这些发现提供的证据表明，考虑 Rs 对光合代理（例如 PAR）的延迟响应以及对土壤湿度的非线性响应，将提高模型的准确性。这一改进还将增强我们对土壤碳循环应对气候和环境变化的理解。