Abstract: In order to understand the composition and structure of the roadside slope vegetation in Sichuan Basin, we use TWINSPAN, which is a wildly used approach of quantitative classification, to analyze the data from the survey of the slope vegetation along 14 roads, which include railways, express ways and highways. We also use canonical correspondence analysis (CCA) to assess the impact of environment variables on the species composition and structure of the vegetation, such as the road age, the altitude, the gradient, the aspect, the soil thickness and the weathering degree. The results show that the roadside slope vegetation of Sichuan Basin is divided into 8 associations. The ordination diagram can well reflect the adaptability of these associations to specific environmental conditions. Monte Carlo permutation test shows that the correlation between species and environmental variables is extremely significant. In quantitative environmental variables, explanatory contribution of road age, altitude, degree of weathering, gradient, aspect and soil thickness are 17.45%, 14.62%, 14.53%,13.11%,11.78% and11.60%, respectively. The former four are rarely affected by random error, and the significance level is 1% (P<0.01). In terms of aspect, the significance level is 5% (P<0.05). Explanatory ability of soil thickness is largely affected by random error, and it showed no significant difference (P>0.05).

Key words: roadside slope, vegetation survey, quantitative classification, ordination, TWINSPAN, CCA