Abstract: Precipitation and landscape pattern are two main factors affecting runoff process of the watershed. Understanding their runoff effect is of great significance to water resources management and ecological construction of watershed. Based on the data of precipitation, runoff, and land use from 1958 to 2020, we analyzed the characteristics of precipitation, landscape pattern and runoff in Lianshui watershed in red soil hilly area of southern Jiangxi Pro-vince, established the relationship between precipitation, landscape pattern and annual runoff, flood runoff, low runoff, respectively. The results showed that the annual precipitation, runoff and annual maximum one-day runoff in the watershed showed a non-significant downward trend during the study period, while the annual minimum one-day runoff showed a non-significant upward trend but with the largest inter-annual variation range. Forested land was the landscape type with the highest proportion in watershed, and other woodland had the most dramatic variation. At the landscape level, Shannon diversity index, Shannon evenness index, patch density and landscape shape index increased from 1.125, 0.541, 0.667 and 16.925 in 1980 to 1.348, 0.614, 0.731 and 18.172 in 2020, respectively, while the landscape contagion index decreased from 68.237 in 1980 to 64.293 in 2020. The overall landscape diversity, fragmentation degree and shape complexity of the watershed increased, the spatial distribution tended to be uniform, and the connectivity decreased. The correlation coefficients between precipitation and annual runoff, flood runoff, low runoff were 0.907, 0.594 and 0.558, respectively. At the class level, the reduction of cultivated land had a greater impact on annual runoff, flood runoff, and low runoff, while the overall change at the landscape level promoted a decrease in annual runoff and flood runoff and an increase in low runoff. The contribution rate of precipitation variation and landscape pattern evolution to the change of annual runoff, flood runoff ,and low runoff were 17.8%, 82.2% and 1.5%, 98.5% and -8.8%, 108.8%, respectively. Our results could provide theoretical refe-rence for landscape pattern allocation and comprehensive management of soil and water loss.

Key words: precipitation variation, landscape pattern evolution, annual runoff, flood runoff, low runoff.