On the relationship between environment and growth of sweet chestnut (Castanea sativa) in the Caucasus

Quantifying the environmental factors influencing tree growth dynamics is essential for predicting ecosystem responses, especially under global climate change. However, achieving comprehensive, long-term tree-growth monitoring across extensive regions can be resource-intensive. Ideally, dendrochronological measurements are complemented by models capable of efficiently estimating growth patterns, particularly in under-sampled regions. We applied a modeling approach combining generalized additive models (GAMs) and extensive dendrochronological data from 258 Sweet chestnut (Castanea sativa Miller) cores collected across Georgia and eastern Turkey. Although GAMs are widely used in ecological research, their application to modeling tree growth remains limited. Our models identified stand age, minimum temperature of coldest month, precipitation during the driest quarter, soil nitrogen content, and soil pH as significant predictors, explaining substantial variability in Ca. sativa growth rates. Younger stands (<50 years) in regions characterized by mild winter temperatures, moderate precipitation, acidic soils (pH 5.0–6.0), and elevated nitrogen content exhibited optimal growth conditions. Future scenario analyses (SSP126, SSP370, and SSP585) revealed regionally variable impacts, highlighting areas vulnerable to climate-induced stress or benefiting from warmer and drier conditions. Although the predictive validity of our model is restricted to the current distribution range of Ca. sativa, it provides a robust basis for estimating growth across the Caucasus ecoregion, particularly where detailed monitoring data is limited.量化影响树木生长动态的环境因子，对于预测生态系统响应至关重要，在全球气候变化背景下尤为如此。然而，在广袤区域开展全面且长期的树木生长监测往往会消耗大量资源。理想情况下，树木年代学测量（dendrochronological measurements）可辅以能够高效估算生长模式的模型，这在采样不足的区域尤为关键。我们采用了一种结合广义加性模型（generalized additive models, GAMs）与大量树木年代学数据的建模方法，这些数据采自格鲁吉亚及土耳其东部的258份欧洲板栗（Castanea sativa Miller）树芯样本。尽管广义加性模型在生态学研究中应用广泛，但将其用于树木生长建模的案例仍较为有限。我们的模型识别出林分年龄、最冷月最低气温、最干季降水量、土壤氮含量以及土壤pH值为重要预测因子，能够解释欧洲板栗生长速率的大量变异。林分年龄小于50年的林分，若处于冬季温和、降水适中、土壤呈酸性（pH值5.0~6.0）且氮含量较高的区域，则具备最优的生长条件。未来情景分析（SSP126、SSP370与SSP585）结果显示，气候变化的影响存在区域差异，部分区域将面临气候诱导的胁迫，而另一部分区域则可能从更暖更干燥的环境中获益。尽管我们的模型预测有效性仅局限于欧洲板栗当前的分布范围，但该模型为高加索生态区的树木生长估算提供了可靠基础，尤其是在详细监测数据匮乏的区域。