高原鼢鼠（Eospalax baileyi）是青藏高原特有的小型穴居哺乳动物之一，它通过挖掘洞道和堆土活动改变草地生态系统的局部环境。目前有关高原鼢鼠对生态环境影响的研究主要集中于地上植被和地下生物量，较少涉及高原鼢鼠洞道对土壤水分的影响。本文在青海湖北岸的沙柳河流域利用探地雷达探测方法无损、高效地对高原鼢鼠地下洞道结构进行了量化研究，将模拟降雨和亮蓝染色方法相结合确定了降水后土壤水分的运移路径，最后借助时域反射仪观测方法研究了土壤水分的分布特征。主要研究结果如下：

（1）研究区的高原鼢鼠洞道平均深度为11.9 cm，洞道内坡度不大，最大深度为19.8 cm，最小深度为1.6 cm。洞道直径以5~15 cm为主，平均直径9.1 cm，较大的洞道推测为高原鼢鼠巢穴或洞道与巢穴连接处。对比高原鼢鼠地表鼠丘与地下洞道的分布可以发现，新鼠丘下方洞道保存较为完整，连续性较强，而旧鼠丘下方洞道较为破碎，长度较短。

（2）研究区发生频率较小的20 mm和30 mm降雨可以入渗至高原鼢鼠洞道所在深度，发生较为频繁的5 mm和10 mm降雨无法入渗至高原鼢鼠洞道所在深度。当降雨量达到30 mm时，高原鼢鼠洞道成为水分入渗的优势路径，可以小幅增加土壤水分入渗深度；当降雨量为20 mm时，高原鼢鼠洞道对土壤水分入渗具有明显的阻碍作用；当降雨量为10 mm和5 mm时，高原鼢鼠洞道对土壤水分入渗路径没有明显影响。

（3）从空间分布看，高原鼢鼠鼠丘样点的土壤储水量整体低于高原鼢鼠洞道和草地样点，0~10 cm土层高原鼢鼠洞道样点土壤储水量低于草地样点，50~80 cm土层高原鼢鼠洞道样点土壤储水量高于草地样点；总体而言，高原鼢鼠挖掘过程中的堆土行为减少了土壤储水量，洞道造成表层土壤水分降低，但有助于深层土壤水分储存。从时间分布看，降雨后高原鼢鼠洞道样点和草地样点浅层土壤含水量均有较为明显的变化，当发生较大降雨（17.2 mm）时，高原鼢鼠洞道样点0~30 cm土壤含水量变化百分比高于草地样点；当发生小降水（5.2 mm）时，高原鼢鼠洞道样点0~10 cm土壤含水量升高11.7%，草地样点10~30 cm土壤含水量升高3.1%，均为三种样点相应深度变化最明显的。

The Plateau Zokor (Eospalax baileyi) is one of the unique small burrowing mammals in the Qinghai Tibet Plateau. It changes the microenvironment of the grassland ecosystem by digging tunnels and stacking soil. Previous researches about the impact of plateau zokor on the ecological environment mainly focused on the aboveground vegetation and underground biomass, while the influence of plateau zokor’s tunnels on soil water was rare. In this paper, ground penetrating radar (GPR) method was used to quantify the underground tunnels of plateau zokor at Shaliu River basin on the north bank of Qinghai Lake. The migration paths of soil water after precipitation were determined by combining simulated rainfall with bright blue staining method. Finally, the distribution characteristics of soil water was researched by time domain reflectometry (TDR) method. The main results are as follows:

(1) The mean depth of plateau zokor’s tunnels in the research area was 11.9 cm, and the slope in the tunnels was not obvious, with maximum depth 19.8 cm and minimum depth 1.6 cm. The diameter of the tunnels was concentrated on 5-15 cm, with an average value 9.1 cm, and the bigger tunnel may be the nest or connection of the tunnels and nest of plateau zokor. Comparing the distribution of surface mounds and underground tunnels of plateau zokor, it can be found that the tunnels under new mounds was relatively intact and continuous, while the tunnels under old mounds was relatively broken and short.

(2) In the study area, 20 mm and 30 mm precipitation with low frequency could infiltrate into the depth of plateau zokor’s tunnels, while 5 mm and 10 mm precipitation with high frequency could not infiltrate into these depths. When the precipitation reaches 30 mm, the plateau zokor’s tunnels became the preferential paths of water infiltration, which could slightly increase the depths of soil water infiltration. When the precipitation is 20 mm, the plateau zokor’s tunnels had obvious blocking effect on soil water infiltration. When the precipitation is 10 mm and 5 mm, the plateau zokor’s tunnel had no remarkable effect on the infiltration paths of soil water.

(3) From the perspective of spatial distribution, the soil water storage at plateau zokor’s mounds were generally lower than these of plateau zokor’s tunnels and grassland sample points. It was lower at 0-10 cm of plateau zokor’s tunnels than that of grassland, and higher at 50-80 cm of plateau zokor’s tunnels than that of grassland. In general, the soil stacking behavior by plateau zokor reduced the soil water storage, and the tunnels led to the decrease of the surface soil water but was helpful to the deep soil water storage. From the perspective of temporal distribution, the soil water content had obvious change at the shallow layers of plateau zokor’s tunnels and grassland sample points. After heavy rainfall (17.2 mm), the variation of soil water content between 0-30 cm at plateau zokor’s tunnels was higher than that at grassland. After light rainfall (5.2 mm), the variation of soil water content was 11.7% between 0-10 cm at plateau zokor tunnels and 3.1% between 10-30 cm at grassland, and both of them were the most obvious variation among 3 types of sample points.