Impacts of different ecological restoration measures and environmental factors on water and soil conservation of the slope in the pure coniferous forest of the subtropical red soil area

The quantification results of the GeoDetector for the impact factors of the soil and water loss in the subtropical red soil area provide the scientific base for the local forest ecological restoration and the improvement of erosion models. This study explored the direct and interactive influences of different ecological restoration measures and environmental factors on water and soil conservation of the slope in forest by introducing the GeoDetector and using the forest survey data and the observation data of the runoff plots in Longmen Creek Watershed, Xinluo District, Longyan City, Fujian Province. The results showed that:(1) to compared with the middle and young age pure coniferous forest, the mixed conifer-broadleaf forest with the mixed ratio 7∶2 could cut down the runoff and sediment load by 46% and 76%, respectively which represented that the ecological restoration effect of this measure was excellent. The small amount of fertilization hardly worked on the "old tree" in the heavy erosion area.(2) The rank of the strong impact factors of the runoff of the slope was rainfall(q value was 0.5), soil bulk density, stand density, shrub coverage, tree height and forest mixed ratio(q values were around 0.08). The rank of the strong impact factors of the sediment loads was surface runoff(q value was 0.84), rainfall(q value was 0.2), stand density, soil bulk density, shrub coverage, soil moisture content, total biomass of shrub(q values were around 0.12).(3) There were complex interactions among the factors of soil and water conservation function. The enhance effect mode accounted for most of the total pairs of the interactions between two factors. The interaction between the weak factors(such as stand density, shrub coverage, and soil bulk density, etc.) and other factors could result in the nonlinear enhancement, and the influence was greater than 0.9. Therefore, they should be paid more attentions when we performed the forest restoration and model parameter optimization.