| 英文摘要: | Changes in the phenology of vegetation activity may accelerate or dampen rates of climate change by altering energy exchanges between the land surface and the atmosphere1, 2 and can threaten species with synchronized life cycles3, 4, 5. Current knowledge of long-term changes in vegetation activity is regional6, 7, 8, or restricted to highly integrated measures of change such as net primary productivity9, 10, 11, 12, 13, which mask details that are relevant for Earth system dynamics. Such details can be revealed by measuring changes in the phenology of vegetation activity. Here we undertake a comprehensive global assessment of changes in vegetation phenology. We show that the phenology of vegetation activity changed severely (by more than 2 standard deviations in one or more dimensions of phenological change) on 54% of the global land surface between 1981 and 2012. Our analysis confirms previously detected changes in the boreal and northern temperate regions6, 7, 8. The adverse consequences of these northern phenological shifts for land-surface–climate feedbacks1, ecosystems4 and species3 are well known. Our study reveals equally severe phenological changes in the southern hemisphere, where consequences for the energy budget and the likelihood of phenological mismatches are unknown. Our analysis provides a sensitive and direct measurement of ecosystem functioning, making it useful both for monitoring change and for testing the reliability of early warning signals of change14. Recent climate change has shifted species distributions15, 16 and leaf phenology17, 18 around the world, leading to mismatches in previously synchronized phenological cycles3, 4. Such mismatches greatly increase the risk of extinction for affected species, and ongoing climatic and phenological change is expected to further increase this risk5. Despite documenting and predicting effects of climate change on many organisms, these previous studies do not provide an easy way of inferring how widespread such changes are or where they are most severe. In addition to being a symptom of climate change, vegetation change also feeds back to the climate system by forcing rates of energy exchange between the land surface and the atmosphere. Changes in the vigour and timing of vegetation activity can therefore accelerate or slow down rates of climate change1. Yet, the extent to which changes in vegetation phenology will impact the climate system by modifying albedo, transpiration, partitioning between latent and sensible heat in the atmosphere, and cloud formation, has been identified as a major source of uncertainty in climate change projections2, 19. To quantify changes in global vegetation activity, previous studies have used remotely sensed data to quantify changes in primary productivity9, 10, 11, 12. These studies have indicated an overall increase in net primary productivity (NPP) during the 1980s and 1990s (ref. 9), whereas evidence for a decrease during the 2000s (ref. 11) has been debated20. Although estimating NPP is important for describing carbon sequestration, it is a highly integrated metric that masks important details of the nature of change. For example, it provides no information on the likelihood of phenological mismatches and limited information on consequences for the land-surface energy budget. Consequently, constant NPP does not guarantee that vegetation is not responding to changing climates and increased atmospheric CO2 in ways that affect the functioning of the Earth system. To quantify intra-annual shifts in the timing and vigour of vegetation activity, remotely sensed absorption of photosynthetically active radiation by the land surface can be used to directly infer photosynthetic activity. We present a global analysis of change in the seasonal pattern of photosynthetically active radiation absorbed by the land surface as measured by the normalized difference vegetation index (NDVI). We improve on previous analyses that have used NDVI to infer phenological change in two important ways. First, previous studies on long-term changes in leaf phenology have been regional6, 7, 8. We analyse the GIMMS3g data, a global record from 1981 to 2012, at 0.083° and 15-day resolution. Second, a problem that has prevented global analyses of phenology is that the information content of phenological metrics is not universal. For example, the onset of the growing season is an informative metric in deciduous forests, but less useful in evergreen forests. We use an improved method to estimate 21 ecologically interpretable metrics of the phenological cycle from the data (Fig. 1) and evaluate the magnitude of change within 83 phenologically similar zones (hereafter called phenomes, shown in Supplementary Fig. 1) to account for the fact that the information content of these metrics differs between ecosystem types. These phenomes were identified using a cluster analysis of the phenological data. To obtain a spatially comparable measure of change, the change per pixel was scaled by the variance of change for the phenome to which it was assigned. Change is therefore reported in standard deviations (s.d.), which can be interpreted as a measure of the severity of change.
| http://www.nature.com/nclimate/journal/v5/n4/full/nclimate2533.html
|
