This work will explore the relationship between biodiversity and ecosystem function at different scales by combining satellite data (primarily Landsat and Sentinel-2) with in situ data and models. It will utilize several remote sensing-based Essential Biodiversity Variables (EBVs), specifically, the Ecosystem Structure EBV “Ecosystem Distribution” (from which habitat fragmentation and connectivity can be derived), and the Ecosystem Function EBV “Primary Productivity”, both of which are based on remotely sensed data. Understanding this relationship is essential for several SDGs and is critical for ecosystem management and accurate forecasting models, but also advances the application of satellite data for monitoring and managing ecosystem function and the services ecosystems provide to humans. The result of this study, which will be led by the Quebec Centre for Biodiversity Science in close collaboration with GEO BON’s Working Groups and Biodiversity Observation Networks, will be captured in a peer-reviewed open-access publication and in freely available software, workflows and methodology.
Update 2021-08-31
This reflects some of the early activity steps that will be built upon as the work progresses:
Update 2022-01-27
Update 2023-01-09
Over the past two decades various field-based observations studied how biodiversity affects ecosystem functioning (i.e., the so-called biodiversity-ecosystem functioning relationship, or BEF). Understanding of that relationship is limited but is crucial to understanding the role that biodiversity plays in the services that ecosystems provide to humans. The work explored here, which is ongoing, looked at one aspect of that relationship that is important in evaluating it using remote sensing data, i.e., measurement scale (pixel size). Recent studies have highlighted the scale dependence of the BEF relationship and demonstrated that the rate at which functions change with biodiversity varies as spatial and temporal measurement scales are increased.
Experiments were carried out for small geographical areas (between 1 to 200 m2) to see how the BEF relationship varied with different measurement grain/pixel sizes; this was done by applying the Spectral Variation Hypothesis (SVH) on MODIS remote sensing data for an area between 45N and 51N covering the province of Quebec (Canada). These vast ecosystems are shaped by different seasonal dynamics and by large disturbance regimes, urban and agricultural expansion and forest harvesting. Because of its large geographic area and its biotic and abiotic heterogeneity, this study region offers a relevant setting to explore the variations in the BEF slope with scale.
Remote sensing data was used for the month of June, the starting month of summer season and when deciduous tree species become lush green due to leaf flushing activity in the spring. The hypothesis that remote sensing-derived spectral heterogeneity (a proxy for biodiversity) results in a nonlinear BEF slope relationship was tested at varying scales for varying gross primary productivity (GPP--a key ecosystem function) in different ecoregions. A steeper slope implies a greater dependency of function/GPP on biodiversity. Using measures of both alpha and beta species diversity the relationship with GPP was computed at different scales by varying window size (25, 50, 75, 100, 125, 150, 175, 200, and 225 pixels) at a resolution of 500 × 500 m2 per pixel. The relationship between spectral diversity and function was evaluated by applying a two-phase exponential decay model and, at each scale, the slope of the BEF relationship was determined to assess how measurement scale affects that relationship. The results consistently showed nonlinear trends for both alpha and beta diversity, confirming that function depended on biodiversity, however, the strength of that relationship depended on the scale of measurement and there was continuous decrease in BEF slope as the spatial grains increased.
Background: Theory and models predict the strength of the relationship between biodiversity and ecosystem functioning (BEF) to vary with spatial scale. In this study, we are testing this understanding by assessing how changes in tree diversity mediate primary production at different spatial grains. We are applying spectral variation methods to MODIS remote sensing data to estimate tree diversity (alpha and beta) at these different spatial grains. We corroborate these estimates of diversity with in situ measures of tree diversity taken from thousands of permanent plots distributed across southern Quebec. Next, we quantify the slope of the correlation (regression estimate) between these spectral diversity estimates (Rao’s Q and Shannon indices) with measures of primary production (NDVI and EVI) at those scales. We obtain different slope estimates at different spatial grains. We then assess the nonlinear functional form describing how BEF slope values change across the different scales and explain this as turnover in the distribution of plant diversity and primary production across geographic gradients in topography, soils and climate. We expect that remote sensing derived estimates of spectral variation will be a powerful way to quickly assess how biodiversity change at different scales may alter ecosystem production across different ecoregions around the world.