Essential Biodiversity Variables

The theoretical potential for remote sensing to contribute to the measurement of Essential Biodiversity Variables (EBVs) to help track progress towards the 2020 Aichi targets of the CBD is recognised.  There are, however, few practical examples of remote-sensing data of relevance to field conservation also being used to make informative assessments of candidate EBVs’. Here we describe the use of a global dataset on forest-cover change and a tool for the rapid assessment of land-cover change in measuring progress in sites and species conservation towards Aichi Targets 11 and 12. For Target 11, we used two openly available and readily accessible tools to quantify land-cover change on a suite of Important Bird and Biodiversity Areas (IBAs). These data could be integrated to form an ‘effective intactness’ index which could compliment the statistics on coverage of protected areas. First we used global data on forest loss (Hansen et al. 2013) to quantify habitat loss on over 7000 important bird and biodiversity areas. Furthermore, we developed a dedicated system to quantify changes in all land cover types from freely available medium spatial resolution data using the tool which allows users to share classifications, so contributing to a collaborative data pool. From Hansen et al. (2013) data we also quantified forest loss within the geographic ranges of 11,186 species of amphibians, birds and mammals. Of these, some 500 species (878 if Data Deficient species are included) were found to be at greater risk of extinction than previous assessments have indicated. These data contribute directly to the Red List Index. Both site and species data have direct relevance to conservation priority setting, and have already been used to inform conservation-resource allocation. They indicate the potential for using such data both for EBVs and for informing the decisions of conservation practitioners in the field.

Essential Biodiversity Variable, Natural Capital, Biodiversity Indicator and Ecosystem Service are four concepts that underpin the most popular frameworks currently used to coordinate and structure biodiversity monitoring efforts worldwide. Satellite Remote Sensing (SRS) has considerable potential to inform these initiatives. To date, however, discussions on the role of SRS in supporting these frameworks have mostly evolved independently; tend to be led by different groups; sometimes target slightly different scales; are likely to reach different audiences; and are impacted by the level of confusion that exists around the exact definition of what these frameworks are and what they aim to achieve. Here I provide a brief overview of the role of SRS to date in informing these frameworks. Through a case study focused on the Sahara Desert ecosystem, I also show and discuss the relative applicability of SRS-based monitoring and metrics to each of these frameworks. This work highlights how more dialogue is required within the biodiversity monitoring community for SRS to reach its full potential in conservation. In particular, agreement on what is needed in priority given the realm of what is possible will be of paramount importance to developing SRS-based products that are used by policy makers and international conventions.

Finding ecological proxies of species diversity is important for developing effective conservation plans of natural areas at various spatial scales.
Remote sensing represents one of the most powerful approaches for measuring ecological heterogeneity at a number of spatial and temporal scales. So far, some progress has been made to promote remote sensing based measures of ecological variability as a direct proxy of biodiversity. On the one hand, they are powerful since they can explicitly detect the variability of ecosystem properties over space and time, while some pitfalls related to spatial and spectral resolutions and to measurement approaches should be faced.
In this study, we will summarize the power of heterogeneity-based variables to estimate biodiversity under the Essential Biodiversity Variables' umbrella to strengthen the knowledge of ecological changes over space and time by remote sensing. We will extend on the proposal of a new RS EBV candidate based on spatio-temporal scale analysis in order to improve global standardized biodiversity measures.

Global biodiversity loss is escalating as a consequence of rapid land cover change, intensification of land use and climate change. The remote sensing of Essential Biodiversity Variables (EBV) concept aims to develop a list of biodiversity variables that can be globally monitored over decades. But there is a need to agree on which biodiversity metrics to track. Through GEOBON sponsored meeting, 50+ variables were suggested for tracking biodiversity, with 10 variables shortlisted. However, much work remains to build consensus and confirm these lists in both the ecological and biodiversity communities. Interestingly, the expert group identified some Essential Climate Variables (ECVs) as also having potential to monitor biodiversity. A key issue is that biodiversity is hard to define – ecologists have many ideas on how biodiversity could be measured. Using remote sensing, monitoring of EBVs is a particular challenge, partly because of the ambiguous ecological definition. There are promising attempts to map a subset of essential biodiversity variables, using a variety of remote sensing techniques including hyperspectral, hypertemporal and high spatial resolution imagery from the Landsat and more recently ESA Sentinel series. However, as EBV variables may be either continuous, or categorized into threshold-based variables, implementing RS-EBVs causes mismatch in the definition of remote sensing and ecological units. Firstly there is a problem with the semantic definition of continuous surfaces as objects – for example what is a forest, or degradation? Land cover surfaces, even in highly human modified ‘patchwork’ landscapes are continuous; the continuity is particularly obvious at very fine (meters) or very coarse (kms) scales. In other words, the definition of a class (e.g. forest), or land cover, or ecosystem, or habitat, requires pre-classification of remotely sensed data, and if the definition (e.g. of a forest) changes then the area of the pre-classified object also changes. A solution to this conundrum are standards – though even with decades of effort, we cannot agree on global classification standards as there is much variation in nature as well as in people’s perceptions of what is a class. When using remote sensing, continuous essential biodiversity variables (e.g. vegetation height, crown cover, crown complexity, LAI, productivity, phenology etc) can be objectively derived, forming systematic data sets at a global scale, though calibrating continuous variables to absolute values remains challenging. Categorical products, such as forest or other land cover classes, may be then generated from continuous variables using thresholds. The categorical (discreet) variables are of increasing complexity and often encompass more abstract biodiversity concepts. Examples of continuous and categorical EBVs derived from Sentinel and other remote sensing systems will demonstrate that both types of essential biodiversity variables provide extremely useful information. It is concluded that continuous biodiversity variables products are most suited to directly derive from remote sensing as a first step in the processing chain categorical variables. 

Plants underpin food chains in nearly all ecosystems and play a vital role in the provisioning of ecosystem goods and services to society. However human pressures are causing rapid and widespread decline in the world’s plant diversity with consequent influence on the capacity of ecosystems to provide the goods and services on which humanity ultimately depends. Monitoring and reporting on changes in ecosystem function and health is essential for evaluating and prioritising global biodiversity conservation efforts.

The Group on Earth Observation Biodiversity Observation Network (GEO BON) is in the process of reforming the acquisition, coordination and delivery of biodiversity information and services to end users, in particular to decision-makers. Underpinning this initiative is the GEO BON’s effort to develop a candidate set of Essential Biodiversity Variables (EBVs), which provide guidance to observation systems as to what and how to measure key aspects of biodiversity status and trends. 

The RS4EBV project is a wholly novel project that aims to advance the conceptual development of EBVs concerning ecosystem structure and function. It is mapping a suite of plant functional traits over three pilot sites in Europe, representing three very different habitats – managed to natural grasslands, temperate forest and coastal saltmarsh- directly from Sentinel-2 imagery and transforming these variables into a higher-level, ecologically- meaningful estimate of Functional Diversity (FD); the value, range and relative abundance of plant traits in an ecosystem. The first phase of the project has relied on Sentinel-2 surrogate datasets, namely RapidEye for leaf chlorophyll retrieval and SPOT-5 for phenological metrics, owing to the presence of the red edge band and the 5-day revisit time of each sensor respectively. The second phase of the project will apply the algorithms and methods developed in phase 1 to Sentinel-2 imagery. The plant traits that can be directly retrieved from imagery are the so-called ‘direct’ EBVs while the FD, requiring a predictive model supported by in-situ and other trait data, is referred to as the ‘indirect’ EBV. In order to build the phenomenological relationships which drive this predictive model, a series of correlation tests are being carried out in order to characterise the relationship between the direct, satellite-derived EBVs and FD calculated from ground data alone.

The vision of this project is to initially pilot the methodology  to model FD for a set of controlled study sites while exploring the possibility to up-scale the method to monitor FD on a continental to global scale. In future, spatiotemporally consistent observations of FD will provide a ready indicator of ecosystem status and trends and greatly assist in global conservation efforts in support of the Convention on Biological Diversity.