Oases in arid regions of northwestern China are main sites for agriculture and human settlement. Conservation of these important oasis habitats involves establishing shrub shelterbelts around the oases. We investigated how different types of shrub shelterbelts affect the diversity of ground beetles. We sampled carabid and tenebrionid assemblages with pitfall traps in spring and summer and measured key environmental variables in four habitats of the arid region: planted stands of pure and mixed plantations of two long‐lived shrubs (Haloxylon ammodendron and Tamarix ramosissima) and natural grassland (NG) from which planted systems were converted. Beetle abundance and species richness reduced significantly in the pure plantations but the mixed plantation had similar abundance and higher richness in spring and lower richness in summer relative to NG. We found differences in response between functional groups of beetles. The pure plantations resulted in reduced abundance and richness of herbivores and detritivores and increased predator abundance in spring and reduced predator abundance in summer. The mixed plantation resulted in increased detritivore richness in spring and reduced detritivore richness in summer as well as increased herbivore abundance in spring and reduced herbivore and predator abundance and richness in summer. The beetle community was largely determined by shrub cover and herbaceous plant diversity. We conclude that the pure shrub plantations reduce beetle diversity, whereas the mixed shrub plantation has less adverse effect on beetle diversity. Shrub mixtures can therefore be a practical measure to minimise the negative impacts of shrub shelterbelts on beetles.

Humans influence desert ecosystem structure and function by transforming native vegetation into agricultural and non-agricultural land. However, how different land uses in desert grasslands affect soil phosphorus (P) fractions with varying lability by creating distinct soil physicochemical and microbial properties remains poorly understood. To address this question, a field study was conducted in the arid region of northwestern China, where sites were selected for having an area containing natural desert grassland and three adjacent differently managed land uses converted from desert grassland: 28-year-old rainfed shrub (Haloxylon ammodendron) plantation, 35-year-old irrigated tree (Populus gansuensis) plantation, and 33-to 40-year-old irrigated and fertilized croplands. We collected data from four land-use types for concentrations of soil inorganic and organic P (P i and P o ) fractions of decreasing lability (labile P: resin-and NaHCO 3 -extractable P; moderately labile P: NaOH-extractable P; recalcitrant P: HCl-extractable P and residual P) and edaphic variables as predictors. The relative contributions of individual predictors to variation in P fractions were evaluated using a boosted regression tree analysis. We found significant increases in (1) all P fractions in croplands (77%-418%), with the largest increase in resin-P; (2) all P fractions except resin-P in tree plantations (52%-367%), with the largest increase in labile P o ; and (3) NaHCO 3 -P and HCl-P i in shrub plantations (14%-134%), with the largest increase in labile P i , compared with natural desert grasslands. Edaphic properties dominantly controlled the changes of different P fractions, explaining 51%-93% of the variation in P fractions, but the dominant influential factors differed across P fractions. We conclude that different land uses in desert grassland soil resulted in significant differences in the fractional composition and availability of soil P by affecting edaphic properties differently, informing that soil P cycling can be regulated by altering land-use types and management levels. Highlights• Assessing the impact of different land uses in desert grasslands on soil P fractions.• Different land-use types showed distinct patterns of soil P dynamics and availability.