Schut Antonius G T, Ivits Eva, Conijn Jacob G, Ten Brink Ben, Fensholt Rasmus
Plant Production Systems Group, Wageningen University, Droevendaalsesteeg 1, 6708 PB Wageningen, The Netherlands; PBL Assessment Agency for the Environment, Antonie van Leeuwenhoeklaan 9, 3721 MA Bilthoven, The Netherlands.
Joint Research Centre, Via Enrico Fermi 2749, I - 21027 Ispra, Italy.
PLoS One. 2015 Oct 14;10(10):e0138013. doi: 10.1371/journal.pone.0138013. eCollection 2015.
Detailed understanding of a possible decoupling between climatic drivers of plant productivity and the response of ecosystems vegetation is required. We compared trends in six NDVI metrics (1982-2010) derived from the GIMMS3g dataset with modelled biomass productivity and assessed uncertainty in trend estimates. Annual total biomass weight (TBW) was calculated with the LINPAC model. Trends were determined using a simple linear regression, a Thiel-Sen medium slope and a piecewise regression (PWR) with two segments. Values of NDVI metrics were related to Net Primary Production (MODIS-NPP) and TBW per biome and land-use type. The simple linear and Thiel-Sen trends did not differ much whereas PWR increased the fraction of explained variation, depending on the NDVI metric considered. A positive trend in TBW indicating more favorable climatic conditions was found for 24% of pixels on land, and for 5% a negative trend. A decoupled trend, indicating positive TBW trends and monotonic negative or segmented and negative NDVI trends, was observed for 17-36% of all productive areas depending on the NDVI metric used. For only 1-2% of all pixels in productive areas, a diverging and greening trend was found despite a strong negative trend in TBW. The choice of NDVI metric used strongly affected outcomes on regional scales and differences in the fraction of explained variation in MODIS-NPP between biomes were large, and a combination of NDVI metrics is recommended for global studies. We have found an increasing difference between trends in climatic drivers and observed NDVI for large parts of the globe. Our findings suggest that future scenarios must consider impacts of constraints on plant growth such as extremes in weather and nutrient availability to predict changes in NPP and CO2 sequestration capacity.
需要深入了解植物生产力的气候驱动因素与生态系统植被响应之间可能存在的脱钩现象。我们将来自GIMMS3g数据集的六个归一化植被指数（NDVI）指标（1982 - 2010年）的趋势与模拟生物量生产力进行了比较，并评估了趋势估计中的不确定性。年度总生物量重量（TBW）使用LINPAC模型计算。趋势通过简单线性回归、蒂尔 - 森中值斜率和具有两个分段的分段回归（PWR）来确定。NDVI指标的值与每个生物群落和土地利用类型的净初级生产力（MODIS - NPP）和TBW相关。简单线性趋势和蒂尔 - 森趋势差异不大，而PWR增加了解释变异的比例，这取决于所考虑的NDVI指标。在陆地上，24%的像素发现TBW呈正趋势，表明气候条件更有利，5%呈负趋势。根据所使用的NDVI指标，在所有生产区域的17 - 36%观察到脱钩趋势，即TBW呈正趋势，而NDVI呈单调负趋势或分段负趋势。在生产区域中，只有1 - 2%的像素尽管TBW呈强烈负趋势，但仍发现了发散和绿化趋势。所使用的NDVI指标的选择对区域尺度的结果有很大影响，生物群落之间MODIS - NPP解释变异比例的差异很大，因此建议在全球研究中结合使用NDVI指标。我们发现全球大部分地区气候驱动因素的趋势与观测到的NDVI之间的差异在增加。我们的研究结果表明，未来情景必须考虑天气极端情况和养分可用性等对植物生长的限制影响，以预测净初级生产力和二氧化碳封存能力的变化。
