Land Surface Phenology as an indicator of performance of conservation policies like Natura2000

Considering the growing pressure on various ecosystems all around the world, the development of ecological indicators for assessing the health and condition of the ecosystems has become an integral part of environmental management and policy making efforts. Such indicators do not only simplify the complex ecosystem processes but can also act as a measure of performance of a conservation policy. The use of remote sensing in developing these indicators has been very limited due to a gap that exists between nature conservation agencies and remote sensing communities. The present research is an attempt at bridging this gap by testing the potential of land surface phenology (LSP) in developing indicators which can illustrate the performance of conservation policies like Natura2000. A total of 60 sites were selected across five largest biogeographical regions of Europe, out of which 30 sites were within the Natura2000 protected network while 30 sites were outside the protected network. Moderate Resolution Imaging Spectroradiometer (MODIS) Normalised Difference Vegetation Index (NDVI) data was used to derive phenological metrics for land cover class pastures, for within and outside the Natura2000 protection network for a time period between 2001-2015. Eight phenological parameters namely seasonal amplitude, peak value, large integral, small integral, start of growing season, end of growing season, length of growing season and peak of season were generated using the TIMESAT software and were compared for differences within and outside protected areas. A change in management practices after the establishment of Natura2000 was assumed to have an impact on the vegetation vigour of the pastures, thereby confirming the implementation of Natura2000. The results showed a consistent significant difference in phenological parameters such as start of growing season, length of growing season and large integral (integrated NDVI values), between sites within and outside Natura2000 network, across majority of the biogeographical regions. These differences in phenological parameters emerged as a result of differences in management practices within and outside protected areas because of their conservation status, thus confirming the effectiveness of Natura2000. Additionally, a trend analysis was performed on all the sites using the Mann-Kendall test to detect the presence of monotonic trends. Only 10% significant trends were observed for different parameters. Although a majority of the sites displayed a tendency for negative trends, not enough evidence was found to confirm that the protected and unprotected sites differed consistently in trends for any given phenological metric. A plausible reason for this could be the short time series and single pixel representation of each study site. The SPEI (Standardised Precipitation-Evapotranspiration Index) and FAPAR (Fraction of Absorbed Photosynthetically Active Radiation) drought indicator data was analysed to see if the 10% significant trends were drought-induced. However, only two sites in the Mediterranean region with significant trends were observed to coincide with the significant drought-indicator values suggesting that the sites were not largely influenced by drought. For related future studies, use of time series data for a longer period of time and study sites represented by more than a single pixel have been recommended in order to achieve better results from trend analysis.Ecosystem health assessment is a metaphor used to describe the state or condition of the ecosystem. To assess the health of an ecosystem, ecological indicators play an important role. They not only simplify the complex ecosystem processes but can also act as a measure of performance of a conservation policy. Ecosystem health can be characterized by the seasonal vegetation growth cycle (phenology) and vegetation productivity acquired through remote sensing. This satellite observed phenology is generally known as Land Surface Phenology (LSP). The use of remote sensing in developing these indicators has been very limited due to a gap that exists between nature conservation agencies and remote sensing communities. The present research is an attempt at bridging this gap by testing the potential of remote sensing data or LSP in developing indicators which can illustrate the performance of conservation policies like Natura2000. A total of 60 pasture sites were selected across five largest biogeographical regions of Europe, out of which 30 sites were within the Natura2000 protected network while 30 sites were outside the protected network. A change in management practices after the establishment of Natura2000 was assumed to have an impact on the vegetation vigour of the pastures, thereby confirming the implementation of Natura2000. To analyse the differences in vegetation vigour within and outside conserved areas, eight indicators of vegetation phenology were derived from remotely sensed Normalised Difference Vegetation Index (NDVI) data for the time period 2001-2015, using a software called TIMESAT. These indicators or phenological parameters were seasonal amplitude, peak value, large integral, small integral, start of growing season, end of growing season, length of growing season and peak of season. Statistical tests were done to compare these indicators within and outside the Natura2000 protected areas. The results showed a consistent significant difference in phenological parameters such as start of growing season, length of growing season and large integral (integrated NDVI values), between sites within and outside Natura2000 network, across majority of the biogeographical regions. These differences in phenological parameters emerged as a result of differences in management practices within and outside protected areas because of their conservation status, thus confirming the effectiveness of Natura2000. The results from trend analysis showed only 10% significant trends. Although a majority of the sites displayed a tendency for downward trends in various phenological indicators, however, not enough evidence was found to confirm that the protected and unprotected sites differed consistently in trends for any given phenological parameter. A plausible reason for this could be the short time series and single pixel representation of each study site. Additionally drought indicator data was analysed to see if the 10% significant trends were drought induced. However, no correlation was observed indicating that these downward trends were not impacted by drought. For related future studies, use of time series data for a longer period of time and study sites represented by more than a single pixel have been recommended in order to achieve better results from trend analysis