Plant-community structure and groundwater attributes were investigated in Ejina Delta in north-western China to understand spatial variability of groundwater depth and composition and their impacts on vegetation succession. Geostatistical methods and ordination analysis were performed to analyse the data. In addition, we tried to obtain vegetation successional series by using an approach of spatial sequences instead of temporal sequences. The findings of the present study were as follows: (1) the coefficient of variation for groundwater depth (GWD), salinity (SAL), total dissolved solids (TDS), electrical conductivity (EC), pH, Ca2þ , Mg2þ , K þ , Na þ , SO4 2 , HCO3 , NO3 , Cl and F ranged from 0.04 to 1.53; (2) GWD, Mg2þ , TDS, EC, Ca2þ , HCO3 , NO3 and pH showed strong spatial autocorrelation, whereas K þ and SAL showed moderate spatial autocorrelation; (3) canonical correspondence analysis revealed that groundwater heterogeneity, especially GWD, followed by pH, SAL, TDS, EC and HCO3 , had an important impact on vegetation succession, and thus showed a prevalence of groundwater attributes-based niche differentiation among plant communities; and (4) there were two vegetation successional processes (drought and salinisation) in the lower Heihe River Basin, and salinisation processes increased with drought processes. Our results indicated that high spatial variability of groundwater attributes contributes to promoting maintenance of species and landscape diversity in the lower Heihe River Basin.