Effects of shrub encroachment on soil organic carbon in global grasslands

Over the last century, shrub encroachment, i.e., the increase in abundance and dominance of adjacent or local shrubs, has been observed in grasslands worldwide1,2,3. Shrub encroachment as a naturally occurring land cover transformation often accompanies the degradation of grasslands4,5. The shift from grasslands to shrub-encroached grasslands (SEG), which is often irreversible, can result in various ecological consequences, such as changes to biodiversity, soil organic carbon (SOC), and the regional carbon balance6,7,8. Studies have found that shrub encroachment in grassland often changes the SOC content, but these studies have many uncertainties and have resulted in contradictory conclusions9,10. Some studies found that shrub encroachment increases SOC11,12,13. In contrast, shrub encroachment resulted in SOC decreases in some areas7,14,15 or caused negligible changes in others16,17. These contradictory results revealed a large regional variation in the impact of shrub encroachment on SOC. Recently, two meta-analyses of American and global datasets found that shrub encroachment would increase the SOC content at a global scale8,18. However, the geographical pattern of the SOC changes and their controlling factors remain unclear. Further exploration of the controlling factors affecting SOC changes in relation to shrub encroachment will help us understand the encroachment process and its ecological impact on the carbon cycle10,19. The process of shrub encroachment is affected directly or indirectly by climate, soil conditions and local shrub types3,20. These factors can also influence the effect on SOC content induced by shrub encroachment. Jackson et al.7 identified a negative relationship between precipitation and changes in SOC and nitrogen content (STN) when grasslands were invaded by woody vegetation, with drier sites gaining and wetter sites losing SOC. A meta-analysis by Barge et al.18 found that the relationship between precipitation and SOC increment in response to shrub encroachment was generally negative18, however, the modeling result showed an opposite relationship21. These inconsistent findings suggest the need for further assessments. Soil properties are perhaps one of the most important factors leading to the uncertainties observed previously22,23. Different components of SOC combined with differences in water content, pH and particle-size structure can contribute to the differences in the changes to SOC under shrub encroachment24. Soil texture interactions with other physical and chemical properties will affect SOC accumulation25. Thus, the soil texture determines the initial grassland SOC content, which also determines the potential SOC content following shrub invasion. Wheeler et al.55 found that in sandy loam soil, the accumulation of SOC as a result of shrub encroachment was approximately 23% higher than in clay loam soil with the same precipitation and temperature. Moreover, different “invasive” shrub species have different impacts on soils; most grasslands have been encroached by leguminous plants10, and their impacts on SOC compared with non-legumes are still unclear. Moreover, the different approaches and scales used-e.g., some studies compared shrub patches with grass patches in SEG, whereas others compared SEG and non-SEG, treating SEG as a whole—make it difficult to draw general conclusions20,26. Fortunately, increasing worldwide data now provide an opportunity to further evaluate the patterns of SOC change following shrub encroachment and their underlying mechanisms. We conducted a meta-analysis to explore the impact of shrub encroachment on SOC in relation to climate factors, soil properties, and shrub types. A total of 142 datasets from 41 publications across the world were collected. Using these data, we addressed the following two questions: (1) Does shrub encroachment increase the SOC content, and what is the geographical pattern of the SOC change? (2) How do climate, soil conditions and shrub types influence encroachment-induced SOC changes?