Evaluating the performance of remote sensed rain-use efficiency as an indicator of ecosystem functioning in semi-arid ecosystems

• Published in: International journal of remote sensing 2018-05, Vol.39 (10), p.3344-3362
• Main Authors: Zhao, Yanchuang, Wang, Xinyuan, Vázquez-Jiménez, René
• Format: Article
• Language: English
• Subjects: Arid regions; Change detection; Correlation analysis; Correlation coefficient; Correlation coefficients; Desertification; Detection; Ecological monitoring; Ecology; Ecosystems; Electromagnetic interference; Environmental changes; Growing season; Indicators; Nitrogen; Normalized difference vegetative index; Nutrient cycles; Phosphorus; Phosphorus cycle; Precipitation; Primary production; Rain; Rainfall; Remote monitoring; Remote sensing; Vegetation; Vegetation index
• ISSN: 0143-1161; 1366-5901
• DOI: 10.1080/01431161.2018.1439598

Monitoring ecological functioning is a significant step towards detecting changes in ecosystem attributes that could be linked to desertification processes in drylands. The remote sensing proxies of ecological functioning, attract substantial attention due to its advantage on large spatial and increasingly long temporal scales. Remote sensed Vegetation Indices (VIs) have been proposed as the approach to plant productivity to be indicators of ecosystem functioning in local drylands. However, VIs are easily affected by rainfall, a limiting source in arid and semi-arid areas. Therefore, they may not be suitable indicators of ecosystem functioning when applied at large scales, with different rainfall regimes. To overcome the influence of precipitation, the performance of the remote sensing Rain-Use Efficiency (RUE, defined as aboveground net primary production divided by rainfall) was evaluated in 78 global drylands (of which 74 are located in semi-arid areas), as an indicator of multiple ecosystem functions, quantified by ecological multifunctionality index (EMI, integrated by carbon, nitrogen and phosphorus cycles). The correlation analysis showed that during the growing season, the linear relationships of summed EVI (Enhanced Vegetation Index) and RUE with EMI are both significant positive. However, RUE explained more variation (about 44%) in EMI than summed EVI (about 32%) did. The results obtained by partial correlation analysis by controlling the rainfall showed that correlation coefficient between summed EVI and EMI, increased about 20%, while correlation coefficient between RUE and EMI increased very slightly (about 3%). Similar results were also found by using the Normalized Difference Vegetation Index (NDVI). These facts indicated that both remote sensed VIs and RUE could be indicators of ecological multifunctioning. However, RUE was better due to its robustness to rainfall. Also, we must take care that the core assumptions related to the RUE should be fulfilled before using it as an indicator. The relationships between RUE and nutrient cycles showed that in comparison to phosphorus cycle, the carbon and nitrogen cycles had an apparent higher weight in determining the relationship between RUE and EMI. Our findings support the use of remote sensed RUE to monitor ecosystem functioning which could be linked with alternative dryland states and early detection of desertification in drylands.