三峡库区典型农林流域景观格局对径流和泥沙输出的影响

Effects of landscape patterns on runoff and sediment export from typical agroforestry watersheds in the Three Gorges Reservoir area,China

以三峡库区秭归县25个典型农林小流域作为研究对象,斑块类型水平上选取斑块面积比例指数(PLAND),景观水平上选取斑块丰富度密度(PRD)、双对数回归分维数(DLFD)、Shannon多样性指数(SHDI)、聚集度(CONT)等5类景观指数,基于遥感和GIS技术,应用AnnAGNPS模型模拟各流域径流和泥沙输出,FRAGSTATS软件计算流域景观格局指数,SPSS19.0分析景观指数与径流、泥沙的相关回归关系,从景观水平探讨景观格局特征对径流、泥沙输出的影响。结果表明:流域径流量与农坡地、农梯地、农林梯地、居民地等斑块面积比例指数以及CONT等景观指数显著正相关,林地、灌木地PLAND与径流量显著负相关;农坡地、农林梯地、居民地等斑块面积比例指数、CONT指数等与泥沙输出量显著正相关,而林地、灌木地PLAND、SHDI等与泥沙显著负相关。景观指数(因子)与径流和泥沙输出复相关系数R为0.856和0.962(均高度相关),复相关系数R均大于相对应的单因子相关系数。景观水平上,流域景观格局对径流和泥沙输出影响均显著,景观空间格局是景观功能多样性的反馈,是景观格局影响径流、泥沙输出等生态过程的又一例证。

The Three Gorges Reservoir area is one of China＇s the most ecologically sensitive areas. Transport and deposition of eroded material as well as substances dissolved in runoff and attached to soil particles lead to declines in water quality, eutrophication and negative impacts to agricultural lands and to waters of the Three Gorges Reservoir. Generally, land use changes, especially the conversion of forests to agriculture in watersheds, are thought to cause these problems. The conversion of cropland to forest as part of an extensive ecological afforestation engineering scheme has played a very important role in reducing runoff and sediment export in this area. Landscape pattern analysis plays an important role in landscape ecology, whose ultimate goal is to link spatial patterns of landscapes with ecological processes and detect status of processes using landscape pattern information. By analyzing thespatial arrangement of relevant factors related to landscapes, landscape pattern analysis can obtain information to predict dynamic patterns of ecological processes. A close relationship exists between land use structure and runoff and sediment export. Landscape patterns have a significant and complex effect on soil erosion and sediment delivery. In this study, a total of 25 typical agroforestry watersheds in Zigui County, Hubei, China, in the Three Gorges Reservoir area were selected to investigate the effects of watershed landscape patterns on runoff and sediment by using RS and GIS techniques. Watershed landscape pattern indices selected were percentage of landscape patch area （PLAND） at the patch-class level, patch richness density, double log fractal dimension （DLFD）, Shannon＇s diversity index （SHDI） and contagion （CONT） at the landscape level. Watershed landscape pattern indices were calculated by FRAGSTATS 4.1. Runoff and sediment exported from 25 watersheds in study area were simulated by the AnnAGNPS （Annualized AGricultural Non- Point Source） model, which is a continuous watershed-scale model that can simulate runoff and sediment in watersheds covering up to 300000 hm2. Correlation between pattern indices and runoff sediment was analyzed using SPSS 19.5 software. The results show a significant positive relationship exists between runoff and CONT and PLANDs of sloping farmlands, terraces, agroforestry terraces and residential land. A significant negative correlation was found between runoff and PLANDs of forestland and shrubland. Soil erosion correlation analysis also showed a significant positive relationship between sediment and DLFD, CONT and PLANDs of sloping farmlands, agroforestry terraces and residential land. A significant negative correlation was found between sediment and SHDI and PLANDs of forestland and shrubland. Multiple correlation coefficients between landscape indices and runoff and sediment were 0.856 and 0.962, respectively, both of which were higher than those of single-factor coefficients. There were linear regression relationships between the landscape indices and runoff and sediment. Landscape pattern significantly influences runoff and sediment delivery in watersheds at the landscape-level. Our results also present empirical evidence for the existence of direct relationship between the landscape patterns in study area and runoff and sediment processes as well as suggest landscape spatial patterns and functional diversity may be linked through feedback mechanisms.