20 years of landscape dynamics within the world's largest multinational network of protected areas.

The world's largest network of protected areas (PAs), Natura 2000, is facing different types of disturbances and pressures, however, it still remains unclear the impact they have on the conservation status of sites. Remote sensing big data analysis and satellite data were used to quantify dynamics of the dominant land cover category, landscape structure, and vegetation greenness, as indicators of conservation status, as well as drivers of change, between 2000 and 2018, within each Natura 2000 protected area, across the entire European Union. Our results show that the majority of sites are 'favourable' on natural land cover range and areas, but heading to 'unfavourable' status regarding the landscape structure, while an alarmingly high number of sites experience both net loss of the dominant land cover type and degradation of landscape structure, labeled consequently as having an 'unfavourable' conservation status. The results also showed high differences between biogeographic regions and countries, with an extremely low number of sites suffering dramatic changes to other dominant land cover types, mainly among grasslands. Mediterranean region showed a high net forest increase (mainly extension of existing forests) as well as insignificant changes of landscape fragmentation and diversity (predominantly in Greece, Spain and, Italy), related to the intensification of forest planting, and to a high loss of grassland area and cropland (land abandonment). High net forest gain, but increasing landscape fragmentation, was observed in the Continental region (mainly in Bulgaria, Poland, Germany and, Italy), suggesting that forest developed in numerous new smaller patches, due to the development of invasive species through natural processes (agricultural land abandonment) and natural system modifications. The Alpine region also showed a low positive net forest change, but with significant dynamics of gains due to reducing of agricultural activities and human disturbances, and losses due to natural catastrophes such as natural fires, storms, avalanches or landslides. Contrarily, the Boreal and Atlantic regions recorded considerable net forest loss during the analyzed period, caused mainly by the occurrence of natural catastrophes, natural biotic and abiotic processes (erosion, parasitism, diseases), and the increase of forestry clearance. These results show the high potential of moderate resolution remote sensing big data in assessing PAs, even more as higher spatial and temporal resolution satellite data are continuously emerging.