Biodiversity monitoring and mapping
Eunis habitat types
Model predictors
PredictorContribution

Climate

0%
0%
0%
0%
0%
0%
0%

Soil

0%
0%
0%
0%
0%
0%
0%
0%

Topography

0%
0%
0%

RS-enabled EBV

0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%

Anthropogenic

0%

Task 6.2 BIODIVERSITY: RS-EBVs for biodiversity monitoring

EBVs have been proposed as a layer between biodiversity observation and biodiversity indicators used in policy. More concretely, EBV classes – such as species traits, species populations, ecosystem functions as well as ecosystem structure – are being implemented by ecologists to identify global monitoring priorities (Pereira et al. 2013). However, the biodiversity community still lacks a global observing system that revolves around the monitoring of a set of agreed variables essential to the tracking of changes in biological diversity on Earth. Such a gap is worrying, as operational systems and the identification of priority biodiversity variables to be monitored are key to (i) coordinating globally consistent data collection across all dimensions of biodiversity, (ii) minimizing duplication of efforts so that scarce conservation funds are not wasted, and (iii) optimizing the allocation of the limited funds available for biodiversity monitoring worldwide (Pettorelli, 2016). There is an urgent need for remote sensing for EBVs to fill the spatial and temporal gaps between in situ observation data of biodiversity from the field. In other words, without remotely sensed systematic and continuous observations, a global framework for monitoring biodiversity cannot exist. Several RS-EBVs are anticipated to be derived from satellite remote sensing, because satellite remote sensing is the only methodology able to provide a global coverage and continuous measures across space at relatively high spatial and temporal resolutions.

Implementation of RS-EBVs in habitat modelling

Task 6.2.2 demonstrates the use of high resolution RS-EBVs for habitat monitoring in order to support the European Environment Agency (EEA) and it’s Topic Centre for Biological Diversity (ETC-BD). EEA and ETC-BD have special responsibilities with regard to European habitats, with specific emphasis on the reporting obligations towards to the Birds and Habitat Directives. The spatial identification of European habitats and related changes are a difficult task, and much effort is nowadays being put in the spatial identification of EUNIS habitat types. Remote sensing data could play a much larger role than it has now, and a good integration of the large amount of in-situ field observations (vegetation relevés) with high resolution RS-EBVs will be key. The integration with of high resolution RS-EBVs will be demonstrated for forest and heathland habitats. In the first place for their spatial identification, but in the second place for identification of changes in habitat quality. This means indicators for forest degradation and encroachment of heathlands.
This NextGEOSS pilot demonstrates the usage of Essential Biodiversity Variables (EBVs) for habitat distribution modelling (HDM). As input for the modelling observation data (derived from the EVA database) of about 200 EUNIS habitats is available. The index with habitat types follow the revised classification according to the Red list of European terrestrial habitats.The model can be executed using a selection of a maximum of 30 predictors (comprising 7 climate parameters, 7 soil parameters, and 13 RS-EBVs). All predictors have the same extent (Europe) and a resolution of 1x1km. For the modelling Maxent version 3.4.1 is used.

Run a distribution model

Take the following steps to run a distribution model in the cloud using Maxent:

  1. Select a EUNIS habitat type at the lowest level (3) for which a model needs to be created. Some information on the selected habitat type will be shown on the Description tab, and the spatial occurrence of the type on the tab In situ observations.
  2. Select a number of predictors of your own choice that need to be included in the model or select the default predictors.
  3. Select Run model to start the modelling. The process time takes about 2 minutes.
  4. After the modelling process is finished the results are shown on different tabs; Prediction map, where you choose between a fraction and a threshold map, Graphs with response curves for each selected predictor, and Result which is a download link to the package containing all Maxent output files. The contribution of each of the predictors to the modelling is shown on the left under Model predictors

The application keeps track of the most recent calculations of the last two days, which enables quick access to a maximum of 20 model runs.

Connect

Facebook Twitter Google+ Youtube Linkedin