Remotely-sensed Ecosystem Functional Types: monitoring functional diversity at the ecosystem level

PARUELO, JOSÉ M.; PEREZ CASORLA, BEATRIZ; ALCARAZ SEGURA, DOMINGO; JOBBÁGY, ESTEBAN; CABELLO, JAVIER; EPSTEIN, HOWARD; BAGNATO, CAMILO; CASTRO, ANTONIO; VALLEJOS, MARÍA

Leipzig

Workshop; GEO BON Open Science Conference & All Hands Meeting.; 2016

GEOBON

Essential Biodiversity Variables should inform on the status of the three dimensions recognised for biodiversity: composition, structure and function. Whereas composition and structure (from genes to ecosystems) have been traditionally used to assess biodiversity status, functional components of biodiversity, particularly at the ecosystem level, have been scarcely included. Satellite remote sensing can provide multiple descriptors of ecosystem function, though their relevance as essential biodiversity variables still needs to be assessed. Time-series of spectral data derived from satellite images can inform on key attributes of the dynamics of carbon, water, energy balance, disturbance regime or nutrient cycling. These ecosystem functional attributes can be integrated to identify Ecosystem Functional Types (EFTs), defined as groups of ecosystems with similar dynamics of matter and energy exchanges between the biota and the physical environment. Most popular EFTs used the three most informative metrics of the seasonal curves of spectral vegetation indices as surrogates of the most integrative descriptor of ecosystem functioning, the primary production dynamics: annual mean (estimator of primary production), seasonal coefficient of variation (descriptor of seasonality), and date of maximum (indicator of phenology). To search for simple metrics that could be used as a set of highly informative ecosystem functional attributes, we extended the analysis to the global scale across all terrestrial biomes and to other key dimensions of ecosystem functioning, i.e., albedo and surface temperature (related to the energy balance) and evapotranspiration (related to both the water cycle and the energy balance). The three first axes of a Principal Component Analysis run on the average seasonal dynamics of each variable and biome explained from 85% to 97% of variance. From more than 20 summary metrics analysed, the annual mean was highly correlated to the first axis (r2>0.9). The second and third axes were a combination of seasonality (correlation with standard deviation from 0.44 to 0.78) and phenology (r2=0.33), showing a contrast between solstices and equinoxes. Hence, these simple but biologically meaningful metrics represent most of the variance of multiple ecosystem functions across terrestrial biomes.