Ecosystem functioning of protected and altered Mediterranean environments: A remote sensing classification in Doñana, Spain

Abstract

Spatial heterogeneity in ecosystem functioning is a key component of ecological variability requiring special attention in the context of global change. A large history of human use has produced high physiognomic heterogeneity in Mediterranean ecosystems. However, the consequences for ecosystem functioning remain insufficiently understood. We analyzed spectral indicators of matter and energy fluxes in the land surface to classify the functional ecosystem heterogeneity in a Mediterranean region covering different management histories and protection types. We specifically analyzed the spatial variability in seasonal and annual patterns in the Normalized Difference Vegetation Index (NDVI), surface temperature (Ts) and albedo from five Landsat ETM images. Then we classified numerically this variability into ecosystem functional types (EFTs) and explored their seasonal dynamics in terms of photosynthetic radiation absorption and evapotranspiration. We identified eight main EFTs with ecologically relevant differences including contrasting dynamics in fPAR seasonality, great variation in incoming radiation reflection and differing evapotranspiration rates, particularly during the water-limitation period. Functional variability in natural vegetation mostly consisted in dissimilar annual rates of NDVI and albedo, whereas differences in seasonality were more evident in transformed areas. Similarly, the spatial distribution of EFTs was partly associated to protection, with two EFTs exclusive of protected areas and comparatively higher functional diversity in humanized areas. Landform effects on water availability in protected areas and human activities under different ecological settings were seemingly responsible for the large functional diversity of the region. We advocate for the explicit incorporation of multifunctional ecosystem heterogeneity in ecosystem manage- ment and monitoring designs.