Conference Agenda

Vegetation plays a vital role in the ecological functioning of the Earth's ecosystems, and it is essential to quantify the response of plant biodiversity to climate change. Trait-based plant ecology links vegetation functioning to climate change drivers and quantify dimensions of functional diversity through field-based and remote sensing techniques. However, field-based and remote sensing approaches to depict landscape-based traits and diversity often exhibit methodological mismatches that must be addressed to deepen our understanding of how functional diversity varies across scales. We aim to identify conceptual similarities and dissimilarities between remote sensing and field ecology in the study of plant functional diversity. We conducted research weaving, a combination of bibliometric and systematic mapping, to identify key concepts and topics of plant trait diversity, knowledge gaps, and conceptual mismatches from the perspectives of both disciplines. We found evidence that trait-based research is strongly biased geographically, being dominated by countries in the northern hemisphere, and considerably more papers published in ecology than in remote sensing. Our topic model identified seven key concepts in the literature, reflecting the level of organization and the ecosystem of interest. We further identified large differences in spatial and biological resolution between disciplines, with field-based ecology sampling smaller areas (resolution and extent), and using leaf-level trait data to estimate well-defined functional diversity indices (e.g., functional dispersion, evenness, divergence, CWM), based on an extensive list of traits. In contrast, remote sensing assesses functional diversity with spatial resolutions on hundreds of square meters - pixel size - and proxies for functional diversity estimations rather than specific indices. These proxies are mainly related to a few traits (e.g. LMA, pigments, height, or nutrient content). We recommend a standardized approach in functional diversity research, as similar traits, spatial resolutions, and functional diversity indices, across both disciplines to improve comparability and integration between them.

Nearly three decades ago, the Spectral Variation Hypothesis (SVH) was brought to life, proposing that spatial variability in the reflectance (i.e., spectral diversity) of vegetated surfaces relates to plant species richness. Particularly, the accessibility and capability of multispectral satellite data have fueled enthusiasm for the SVH, given its potential to enable straightforward biodiversity estimation from space. However, recent studies have raised significant issues regarding the validity of the SVH on large spatial scales. Spectral differences observed between species at the leaf level do not easily translate to landscape-level assessments. This challenges the effectiveness of spectral diversity for large-scale biodiversity mapping and monitoring, suggesting it may be time to let the SVH rest in peace as a one-size-fits-all, easily applicable solution. Yet even as we bury the SVH, some useful ‘SVH zombies’ emerge, providing valuable insights on biodiversity in specific contexts. Drawing on our experience, we will present some of these 'SVH zombies' that, while moving away from the initial simplicity of the SVH, leverage tailored large scales spectral diversity implementations. Examples include the data fusion of different sensors, object-based approaches, the incorporation of temporal information, sub-pixel classification, uncertainty quantification, and the combination of spectral diversity with other biodiversity-relevant predictors.

The capacity of remote sensing to track radiation-related ecosystem functions has significantly improved in the last decade. Furthermore, remote sensing has more recently emerged as a potential biodiversity monitoring tool, thereby bringing new perspectives for studying biodiversity-ecosystem function relationships from space. To properly exploit these opportunities, we still need to improve our understanding of several methodological questions related to the capability of remote sensing to capture the different aspects of plant diversity, particularly functional diversity, and determine the best approaches to connect these estimates with the ecosystem functions to which remote sensing is sensitive to.

To explore these questions in a controlled environment, we have developed BOSSE, a “Biodiversity Observing System Simulation Experiment” that simulates dynamic vegetation scenes featuring multiple species sensitive to meteorological conditions. BOSSE simulates vegetation traits and radiation-related functions (photosynthesis, transpiration, etc…) together with spectral signals related both to the biophysical properties and the physiological state of vegetation (sun-induced chlorophyll fluorescence, land surface temperature, and photochemical reflectance index). Remote sensing imagery can be simulated for specific missions and at different spectral and temporal resolutions.

Using BOSSE, we explore the capability of remote sensing to disentangle biodiversity-ecosystem function relationships from plant diversity and ecosystem function estimates based only on remote sensing data, whereas the simulated vegetation properties and functions are used as a benchmark. We expect BOSSE to improve the interpretation of some pioneering studies, as well as increase the robustness of future analyses based on remote sensing imagery and eddy covariance data.

Biodiversity affects ecosystem functioning by regulating the biogeochemical exchange of carbon, water, energy, and nutrients within and between ecosystems. However, large-scale, systematic measurements of plant biodiversity are still lacking, and the effects of biodiversity on measured biogeochemical processes are understudied.

We leveraged fine-scale remote sensing data from Sentinel-2 to estimate biodiversity at 148 sites across the globe. At these sites, measured eddy covariance fluxes of carbon, water, and energy can be used to compute ecosystem functional properties. To assess the effect of biodiversity on the biogeochemical functioning of the ecosystems, we related remotely-sensed biodiversity (Rao Q) to the derived ecosystem functions, including ecosystem multifunctionality.

Rao Q computed from near-infrared reflectance of vegetation (NIRv) was a major predictor of single ecosystem functional properties and multifunctionality, highlighting the mostly positive effects of biodiversity on the functioning of ecosystems. Rao Q was generally more important than climate and comparable to the structural components of the ecosystem in predicting ecosystem functions and multifunctionality. In addition, Rao Q was more important than traditional biodiversity indices of taxonomic diversity measured at a subset of sites in North America where systematic plant species surveys were available. This reinforces the idea that structural and functional diversity, rather than species identity per se, are key aspects in the worldwide functioning of natural ecosystems.

In summary, we provide strong evidence for significant positive effects of a biodiversity-proxy derived from earth observations on single ecosystem functions and ecosystem multifunctionality. The positive biodiversity effects are robust to the inclusion of most major meteorological and structural parameters that might drive ecosystem functioning or confound the biodiversity-ecosystem functioning relationship. Considering recent and future advances in remote sensing of both diversity and ecosystem functions, our study paves the way to continuous spatiotemporal assessments of the biodiversity-ecosystem functioning relationship at the landscape, regional, and global scales.

The horizontal and vertical structure of forests, both in terms of canopy architecture and the distribution of traits, are indicators and drivers of the connections between biodiversity and ecosystem function. More structurally heterogeneous forests are often more productive, and they should generate more niche space for a wider array of organisms. Here we explore connections between forest 3-D structure, carbon uptake, and biodiversity at the University of Michigan Biological Station (UMBS). Over the past ~100 years, several large-scale disturbance experiments have taken place at UMBS, making it an ideal site for exploring connections between structural heterogeneity, biodiversity, and ecosystem function. In August 2019, the National Ecological Observatory Network’s Airborne Observation Platform (NEON AOP) collected hyperspectral imagery and lidar data over UMBS. Simultaneously, field data were collected to train the AOP data to generate site-wide maps of the vertical distribution of leaf area density (LAD, m2 m-3), top of canopy leaf traits (leaf mass per area (LMA, g m-2), leaf carbon (%), and leaf nitrogen (%)), and spectral diversity. We then used these maps, both individually and combined through an unsupervised clustering approach, to assess connections with other patterns of biodiversity and ecosystem function.

We found that the AOP-derived maps successfully identified different disturbance regimes across the landscape, though more subtle disturbances at smaller spatial scales were more difficult to detect. Correlations between remotely sensed metrics of 3D structure and disturbance intensity varied, but overall were able to explain variation in forest age with 87% accuracy. Overall, this work demonstrates the importance of both horizontal and vertical structure in understanding spatial ecosystem processes and connections between biodiversity and ecosystem function at the landscape scale.

Tropical forest canopies represent the biosphere’s most significant and concentrated atmospheric interface for carbon, water and energy. Here, we present a pantropical analysis that maps the diversity of tropical forest tree canopy functional traits and functional diversity at high spatial resolution. We combine field-collected data from more than 1800 vegetation plots and tree traits and merge these with satellite remote sensing, terrain, climate and soil data to predict variation across 13 tree morphological, structural and chemical functional traits, using these to compute and map the functional diversity of tropical forests. This reveals that the tropical Americas, Africa and Asia tend to occupy different portions of the total functional trait space available across tropical forests. The functional trait analysis across continents shows that tropical American forests have 40% greater functional richness than tropical African and Asian forests. Our predictions represent the first ground-based and remotely enabled global analysis of how tropical forest canopies vary across space.

The forests of the Congo Basin are a unique biodiversity hotspot. They provide multiple ecosystem services, from being a significant carbon sink to regulating the water cycle and regional climate. Additionally, they offer invaluable resources for subsistence, serve global economic demands, and possess cultural and recreational value. However, various environmental and anthropogenic drivers are exerting considerable pressure on these ecosystems, threatening the sustainability of these services. Beyond deforestation, these pressures may lead to dramatic changes in forest tree functional composition, with potential deleterious feedback on carbon and water cycles. Despite their importance, the forests of the Congo Basin remain largely understudied, and our understanding of such subtle compositional changes is modest at best.

The CoForFunc project, funded by the European Biodiversa+ program, aims to advance research towards biome-scale monitoring of the Congo Basin forest's functional composition. A primary focus is on characterizing tree phenology, interpreting these phenological behaviours mechanistically, and upscaling them to ecosystem functional properties (EFPs) at the scale of the Congo Basin. This effort relies on coordinated campaigns of ground measurements, drone surveys, and satellite remote sensing.

This presentation will discuss the strategies we have adopted concerning remote sensing, where cloud cover, atmospheric and directional effects present considerable challenges. We are exploring three different approaches: (1) increasing the availability of cloud-free imagery by leveraging the sub-daily revisit capacity of geostationary satellites to enhance Sentinel-2 BRDF corrections necessary for mapping subtle phenological shifts; (2) investigating the capacity of passive (SMOS) and active (Sentinel-1) microwave data, which are largely insensitive to cloud cover, to provide information on variations in canopy water content associated with phenology and drought response strategies; (3) exploring the use of sun-induced chlorophyll fluorescence (SIF) retrieved from TROPOMI on Sentinel-5P, which is less sensitive to clouds than traditional optical indices, to assess variations in structure and physiology.