Understanding biomass allocation patterns is critical for predicting plant adaptation strategies under climate change scenarios. At present, the theory of biomass allocation (optimal allocation and equidistant allocation) is still controversial, especially in the "fragile" desert steppe, which is especially sensitive to climate change. Therefore, we set up four treatments with varying precipitation gradients [natural precipitation reduced by 50 % (W-50 %), control (WCK), natural precipitation increased by 50 % (W+50 %), and natural precipitation increased by 100 % (W+100 %)] in the desert steppe of Inner Mongolia and applied them for six years before analyzing the trade-off relationship between aboveground and below-ground biomass, verifying the distribution theory, and considering the factors affecting change in biomass distribution. Our findings revealed that the aboveground biomass increased significantly with the increase of precipitation gradient, while the below-ground biomass was significantly different between years. Desert steppe plants allocated more biomass to the below-ground and followed the optimal allocation theory, W-50 % amplified the trade-off of biomass to the belowground, W+50 % and W+100 % treatments transferred the biomass aboveground instead, which claimed that both plant diversity and soil physicochemical properties regulate biomass allocation. At W-50 % treatment, plant growth was limited by soil water content. At W+50 % and W+100 % treatments, plants were limited by soil available nutrients. The "opportunistic" strategy of annual herbs explains why biomass was re-directed to the aboveground organs. Our results emphasize desert plants can adapt to precipitation change using high variation and optimal biomass allocation.