Drought and water-use efficiency are dominant environmental factors affecting greenness in the Yellow River Basin, China

Abstract

Revegetation is accelerating globally because of its benefits in terms of ecosystem restoration, desertification prevention, and warming mitigation. The Yellow River Basin (YRB), as an ecological barrier in northern China, has implemented revegetation projects (such as the 'Grain for Green' program) for over two decades. However, a consensus on whether a significant change in greenness has been achieved and to what extent have environmental factors contributed to this change, as well as their importance ranking, is lacking. Leaf area index (LAI) is a critical indicator for estimating global greenness and projecting the dynamics of climate change. Herein, we apply four methods (Geodetector, random forest, multiple linear regression, and structural equation models) to explore the contribution of different environmental factors to greenness using the LAI in the YRB. We found that greenness has been increasing (greening over 67.22% (p < 0.05; 47.7%) of the YRB) with great spatial heterogeneity in the entire basin since 2000. Specifically, the greening process differed with elevation and slope. Temperature vegetation dryness index (TVDI) and water-use efficiency (WUE) dominated the greening; however, the three subregions evaluated revealed differing performance. In the upstream region, LAI increased by 0.031 y^-1. The primary positive factors of greening change were WUE and the annual highest value of daily minimum temperature; the negative factors were TVDI and the highest number of consecutive days when precipitation <1 mm. In the midstream region, LAI increased by 0.025 y^-1; greenness was mainly affected by the negative interaction of TVDI and the positive interaction of WUE. Annual maximum consecutive 5-day precipitation and annual count when daily minimum temperature < 0 °C had a great indirect impact on greenness, mainly through TVDI and WUE. In the downstream region, LAI increased by 0.045 y^-1, and the main driving factors were the annual lowest value of daily minimum temperature with a negative influence and the annual lowest value of daily maximum temperature with a positive influence. In addition, we found that the effect of the interaction of any two driving factors on greenness was greater than or equal to the single effect of a driving factor. This study concludes that drought and WUE are important predictors to evaluate the greenness in arid and semi-arid regions. We emphasise that the selection and assessment of greenness factors should follow a scientific and rigorous process rather than experience, and increased attention should be paid to the interaction of multiple factors. Furthermore, the perspective of system analysis will deepen our understanding of vegetation change in a vulnerable ecosystem.

Keywords: Attribution detection; Greenness; Random forest regression; Structural equation model; Yellow River basin.