不同去趋势方法对基于Dendrometer数据的茎干水分动态分析的影响——以白扦为例

树木茎干半径变化记录仪(Dendrometer)监测的高精度数据不仅包括木质部的年内径向生长过程, 还包含由茎干水分的消耗和补充引起的可逆变化。然而, 不同的年内生长去趋势方法获得的茎干水分波动之间的差异性仍缺乏对比研究。基于芦芽山北坡针叶林下限白扦(Picea meyeri) 2015年生长季的茎干半径变化和环境因子的实时监测数据, 使用Gompertz生长模型(GPZ)、线性生长模型(LG)、零生长模型(ZG)、日值法(D)和茎干循环法(SC)模拟并去除茎干年内的生长趋势, 然后提取5种不同类型树木水分缺乏引起的茎干收缩(TWD_GPZ、TWD_LG、TWD_ZG、TWD_D和TWD_SC)以表征茎干水分亏缺, 并进一步对比分析了不同茎干水分亏缺序列对环境中水分状况的响应特征。研究发现: (1)不同去趋势方法计算的茎干水分亏缺的趋势和幅度有所差异, 可聚类为3组: TWD_LG和TWD_ZG、TWD_GPZ以及TWD_D和TWD_SC。同组或聚类距离接近的序列在生长季内每个月份都展现出显著的相关性。然而, TWD_LG、TWD_ZG和TWD_GPZ与TWD_D和TWD_SC在8月份相关性较弱。(2) TWD_D和TWD_SC与空气饱和水汽压差(VPD)的正相关关系比TWD_GPZ、TWD_LG和TWD_ZG更加稳定, 且具有更大的相关系数。5种茎干水分亏缺序列和土壤含水量(SWC)的关系在生长季内变化很大。(3)不同去趋势方法的茎干水分亏缺都随着水分胁迫程度(VPD/SWC)升高而显著增长。当胁迫程度较低时, TWD_SC对VPD/SWC的变化最为敏感(R² = 0.39, p < 0.001), 但是与TWD_ZG差别不大(R ² = 0.37, p < 0.001); 当胁迫程度较高时, TWD_ZG对VPD/SWC的敏感性最高(R ² = 0.59, p < 0.001)。综合对比来看, 零生长模型是比较适合研究区白扦生长季内茎干水分波动的去趋势方法, 其可为干旱胁迫条件下预测研究区树木的茎干水分动态及特征提供科学 依据。

Influence of different de-trending methods on stem water relations of Picea meyeri derived from Dendrometer measurements

Aims The high-precision data measured by the Dendrometer includes not only the stem radial growth process caused by the enlargement of xylem cells but also the reversible changes caused by the consumption and replenishment of the stem water. Our objectives were to assess the difference in stem water relations of Picea meyeri obtained by different de-trending methods and their responses to water conditions in soil and air. Methods Hourly stem radius variations in P. meyeri and corresponding environmental factors were monitored in the lower limit of the coniferous forest on the northern slope of Luya Mountain, northern Shanxi Province, China. Gompertz growth model (GPZ), linear growth model (LG), and zero growth model (ZG), daily approach (D), and stem cycle approach (SC) were used to fit the stem growth trend in the growing season of 2015. Then, the growth trend and extract five different types of tree water deficit-induced stem shrinkage (TWD_GPZ, TWD_LG, TWD_ZG, TWD_D, and TWD_SC) were removed to characterize the dynamic of stem water relations. Moving window correlation (31 days) and ordinary least square regression were further employed to analyze the responses of different types of stem water relations to soil and air moisture conditions. Important findings The results showed that: 1) stem water relations derived from different de-trending methods had contrasting trends and amplitude, which could be clustered into three groups: TWD_LG and TWD_ZG, TWD_GPZ, and TWD_D and TWD_SC. Each month, significant correlations between stem water relations in the same group or a lower clustering distance showed, however, TWD_LG, TWD_ZG, and TWD_GPZ had weaker correlations with TWD and TWD_SC in August. 2) TWD_D and TWD_SC had a closer and more stable positive relationship with vapor pressure deficit (VPD) than TWD_GPZ, TWD_LG, and TWD_ZG did. The responses of all types of stem water relations to soil water content (SWC) varied greatly during the growing season. 3) All stem water relations significantly increased as the water stress (VPD/SWC) intensified (p < 0.05). When the stress was low, TWD_SC was most sensitive to the changes of VPD/SWC (R² = 0.39, p < 0.001), whereas it was not much different from TWD_ZG (R² = 0.37, p < 0.001); when the stress was high, TWD_ZG showed the greatest sensitivity to VPD/SWC (R² = 0.59, p < 0.001). Our results suggested that the zero-growth model was more suitable to detrend the stem radius variations during the growing season, and provided a crucial reference for predicting the stem water status and dynamics, especially under drought stress.