Abstract: Based on the field investigation and laboratory analysis, we analyzed the relationship between plant community distribution and soil properties in four habitats (mobile dune, semifixed dune, fixed dune, and grassland) in Horqin Sandy Land. We also examined the effects of key soil factors on the distribution of 24 plant communities by using three different methods: detrended correspondence analysis (DCA), canonical correspondence analysis (CCA) and detrended canonical correspondence analysis (DCCA). The results showed that the first axes of DCA, CCA and DCCA presented the consistent soil gradient that was related to the soil carbon (C), nitrogen (N), pH, EC, bulk density, silt and clay content, accounting for 33% of total variance of the speciessoil properties relationship. The soil gradient mainly determined the changes of communities’ niche which further affected the distribution pattern of plant community. The second axes of three methods could not express environmental gradient ideally and the correlations with these second axes and soil properties differed greatly. Second axis of DCA only had a significant positive relationship with fine sand content, while the second axis of CCA had a negative relationship with soil C:N ratio and fine sand content. Second axis of DCCA positively correlated with soil C:N ratio and fine sand, but negatively correlated with coarse sand content. Shannon index and Simpson index had significant correlations with the first two axes of DCA, CCA and DCCA, respectively, and the explanation of regression models of Shannon index was better than that of Simpson index. The first two axes of CCA explained 58.6% of speciesenvironment relationship, which was higher than DCA and DCCA did. Our results indicate that CCA performs better in interpreting speciesenvironment relationships than the other two methods.

Key words: ammonia-oxidizing bacteria (AOB), acidic forest soil, ammonia-oxidizing archaea (AOA), N cycling