覆盖于土壤表面的苔藓层是森林生态系统的显著特征和重要组成部分，对于调节大气与土壤之间的水分及能量传输和提高土壤抗冲性并进一步维持土壤结构均有着重要意义，然而目前苔藓层的结构特征研究较少，苔藓层对土壤结构和水分保持的影响也尚不明晰。因此，本文以祁连山青海云杉林的苔藓层及其下的土壤为研究对象，通过野外采集原状苔藓层和土柱，并结合CT（computed tomography, 电子计算机断层成像）扫描、图像解译和土壤水分特征曲线测定等方法，分析苔藓层的结构特征、土壤的结构特征和水分保持特征，揭示苔藓层覆盖对土壤结构和水分保持的影响。主要研究结果如下：

（1） 研究区的苔藓种类以山羽藓为主，薄苔藓层和厚苔藓层的平均厚度分别为3.07 cm和7.10 cm，平均体积密度分别为0.086和0.141 cm³ cm^-3，并呈显著差异；苔藓层结构疏松多孔，体积密度随厚度的增加而增大，底层结构较表层而言更为紧密。苔藓层具有强大的持水能力和截留降水能力，自然持水率为156%-235%，最大持水率为810%-935%，苔藓层的截留量随降雨量的增加而增加，在降雨量为10 mm之后稳定在1.5 mm左右，而苔藓层的截留率则随降雨量的增加而减小。

（2） 苔藓层覆盖下的土壤大孔隙的连通性较好，薄苔藓覆盖的土壤大孔隙在整个剖面范围（0-25 cm）内均有分布，空间分布更均匀，横向分布的孔隙较多，而无苔藓和厚苔藓覆盖下的土壤大孔隙在深层分布较少；薄苔藓下的土壤大孔隙的长度密度约是无苔藓的2.7倍，并呈显著差异。研究区的土壤大孔隙在各个等效直径区间内分布均匀，但集中分布在表面积> 1000 mm²和体积> 100 mm³的范围内。薄苔藓覆盖下土壤中的石砾含量最高（0.015 mm³ mm^-3），分别是无苔藓和厚苔藓的5.8倍和3.5倍，但薄苔藓下的石砾大小不均，样点之间的差异也较大，表明薄苔藓覆盖下土壤剖面中的石砾的空间分布不均匀。

（3） Van Genuchten模型在拟合土壤水分特征曲线方面有较好的预测性能（R²>0.986），苔藓层覆盖下不同土层的土壤水分特征曲线的差异较小，其保水性能也优于无苔藓覆盖，其中薄苔藓覆盖下土壤的有效水最大含量最高（0.17 cm³ cm^-3）。苔藓层覆盖下的土壤含水量的样点之间和土层之间的差异均小于无苔藓覆盖，苔藓层的覆盖作用可以从横向和纵向两个维度同时减小土壤水分的空间差异；苔藓层覆盖下土壤的有效水含量也高于无苔藓覆盖，说明苔藓层的覆盖作用对提高土壤的有效水含量具有积极作用。

（4） 祁连山青海云杉林的苔藓层对土壤结构和水分保持具有重要影响，苔藓层指标与土壤结构和水分保持特征呈现显著相关。苔藓层覆盖下的青海云杉根系和土壤剖面中的石砾可影响土壤大孔隙的形成和三维形态特征，进而影响土壤的水分保持性能，而苔藓层覆盖下较高的土壤有机质含量对于土壤的保水性能也有显著正向影响，因此苔藓层覆盖下土壤的有效水最大含量和实际含量均高于无苔藓覆盖，苔藓层对于青海云杉林土壤的水分保持乃至水源涵养具有积极的生态水文效应。

Covered over the soil surface, moss layer is one of the significant components of forest ecosystem, and plays an important role in regulating the water and energy transfer between the atmosphere and soil, improving soil anti-erodibility and thus soil structure. However, the structure of moss layer, and the soil structure and water retention under the moss layer are still poorly investigated. The influence of moss layer on soil structure and water retention is also far from being fully understood. Therefore, we conducted the study in a Picea crassifolia forest ecosystem covered with moss layer on the soil surface in Qilian Mountain, northwest China. We combined with field sampling, CT (computed tomography) scanning, image interpretation, and soil water retention curves to reveal the structure of moss layer and soil, and the characteristics of soil water retention. We also discussed the effects of moss layer on soil structure and water retention in the Picea crassifolia forest ecosystem. The main results are as follows:

(1) The dominant species of bryophytes were Abietinella abietina in the study area. The average thicknesses of thin and thick moss layer were 3.07 cm and 7.10 cm, respectively. The volume densities of thin and thick moss layer were 0.086 and 0.141 cm³ cm^-3, respectively, and significant differences were found. The moss layer had a strong capacity of water holding and precipitation intercepting. The natural water-holding capacity of the moss layer was 156%–235%, and the maximum water-holding capacity was 810%–935%. The interception amount of moss layer increased with the increase of rainfall, and stabilized at about 1.5 mm after the rainfall reached 10 mm, while the interception percentage of moss layer decreased with the increase of rainfall amount.

(2) Soil macropores under moss coverage had a good connectivity. The soil macropores in the thin moss-covered plots were distributed in the whole soil profile (0–25 cm), and they were more uniformly distributed and horizontally oriented, while the soil macropores in the other two plots were much less at the deep soil layer. The length density of soil macropores in the thin moss-covered plots was about 2.7 times higher than that in the non-moss-covered plots, between which significant differences were found. The soil macropores were evenly distributed in all the equivalent diameter intervals, while concentrated in the surface area > 1000 mm² range and volume > 100 mm³ range. The highest volume density of rock fragment (0.015 mm³ mm^-3) was discovered in the thin moss-covered plots, 5.8 times and 3.5 times larger than that in the non-moss-covered and thick-moss-covered plots, respectively. There were some much larger rock fragments in the thin moss-covered plots, which indicated that the size distribution in the soils under the thin moss was not uniform, while the average volume of rock fragments in the other two plots was smaller and the distribution was more homogeneous.

(3) Van Genuchten (VG) model had a good prediction performance in fitting soil water retention curves (R²>0.986). The soil water retention under the cover of moss layer was better than that under no cover of moss layer. The plant-available water-holding capacity in the thin moss-covered plots was the highest (0.17 cm³ cm^-3). The differences of measured soil water content between replicates and soil layers under the cover of moss layer were much smaller than those under no cover of moss layer. The moss cover could reduce the spatial heterogeneity of soil water both horizontally and vertically. The plant-available water under the cover of moss layer was also higher than that under no cover of moss layer, suggesting that the cover of moss layer improved the plant-available water content.

(4) The roots of Picea crassifolia and rock fragment in the soil profile under the cover of moss layer had an important positive effect on the formation and three-dimensional morphology of soil macropores, which directly affected soil water retention. In addition, the higher soil organic matter content under moss coverage also had a significant effect on soil water retention. Therefore, the plant-available water-holding capacity and content of the mossy covered plots were higher than those of non-mossy plots. The moss coverage had positive ecohydrological effects on soil water retention in Picea crassifolia forest ecosystem in Qilian Mountain.