WP 4: Data product creation | Emodnet Biology

WP 4: Data product creation

Lead: Deltares

+  U. Sheffield, ULg, VLIZ, MBA, Ifremer, SMHI, Cefas, NIOZ  

Objectives

The objective of WP4 is to create digital map layers allowing users to analyse changes in species abundance and extent over time and space. This work was initiated during the previous EMODnet phase, by the development and production of gridded map layers from one or more data sources showing the average abundance of several species per functional species group for different time windows (seasonal, annual or multi-annual as appropriate) using geospatial modeling. Now, we want to:

 

  • capitalize on the investments in EMODnet Biology phase II, in particular (a) the use of Data-Interpolating Variational Analysis (DIVA) gridding for spatio-temporal distributions of species, groups and indicators. DIVA allows the spatial interpolation of data (analysis) in an optimal way, comparable to optimal interpolation (b) the traits database, which can now be put to use (c) the compilation of historic data that may allow reconstructing long-term evolution of some selected groups, and to cover large parts of European seas in distributions for some groups (c) the databases on distributed presence-only data on many marine groups, that have not yet been put to use extensively.
  • capitalize on investments in other EMODnet lots, by intensifying the exchange of information across lots and using chemical, habitat, physical and human-use data sets as background landscape layers that can help improving the interpretation/prediction of occurrence of species groups, interpret species (group) sensitivity and relate indicators to human pressures.

 

Methodology & activities

 

It is proposed that WP5 on Uptake and outreach investigates with EMODnet regional offices, regional seas authorities, MSFD authorities, ICES, global initiatives such as GOOS, IOC and GEOBON where specific demands exist that can be fulfilled with gridded data products based on existing or newly acquired datasets in EMODnet Biology. An important tool to identify these needs, will be the organization of a workshop with invited experts from the regional sea commissions, the GOOS and BON community, and the different research networks (see WP Uptake and outreach for more information).

1. Based on the technical expertise of the consortium and input from other initiatives, we will develop more gridded maps of additional species, species groups and selected indicators based on species distributions. This is continuation of EMODnet PhaseII activities, using similar methodology (DIVA gridding, presence-absence data), but will be more demand-driven.

2. We will consult with the EMODnet lots chemistry, physics, habitats, geology, bathymetry, human activities on the construction of relevant background layers covering European seas. Therefore, we will contact other EMODnet lots in order to establish a comprehensive set of background layers useful for species distribution modeling in EMODnet biology. These layers will contain information on the seabed (for benthic datasets), water column chemistry, water column physics, human pressures. The intention is to inventory existing efforts and combine them into a set of useful rasterized layers that can form the basis for species distribution modeling. These should be targetted to allow testing of specific questions regarding how distributions are shaped by thermal, chemical and physical environments and modified by human pressures. We will construct a set of layers based on the available information (D4.2).

3. We will apply species distribution modeling based on the background layers, and biological information available within EMODnet biology. We will select a limited set of standard models to use for different data types, in order to produce comparable and standardized data products while maximizing the use of existing data. Modular structures of existing project for species distribution modelling provides a useful and reproducible framework we can further built on. We can apply species distribution modeling tools, either regression-based (e.g. generalized linear models, GAM) for presence-absence data, or based on machine-learning methods for presence-background (e.g. MAXent) and presence-only (e.g. envelope models) data. We’ll investigate Dynamic Linear Model Tools to analyse trends of marine time series and look at random forest models for upscaling. A comprehensive set of methods for different data types will be maintained and put in operation; emphasis will be on the selection of suitable data sets and production of relevant maps for species or species groups (D4.3: Portfolio of modelling tools and products for European marine species)

4.To make use of the traits database collected in EMODnet Phase II and traits information available in public databases, to abstract from the species level and produce maps of functional or response groups. In particular, trait groups that have significance as indicators in MSFD and other applied context will be selected. Special attention will be paid to statistical distribution models for the trait groups, as these are expected to be less variable than models for individual species. We will use traits information, as well as information on available species data, to select a number of cases where comprehensive spatio-temporal distribution maps of relevant species groups can be made. The priorities here will be linked to the demand, and could be used to the calculation of data products giving information on community indicators. We will use traits information, as well as information on available species data, to select a number of cases where comprehensive spatio-temporal distribution maps of relevant species groups can be made depending on the identified demands, in particular to derive indicators. A standardized methodology will be defined for these data products. We will investigate whether incorporation of environmental information (as in Task 2) improves the reliability of the indicators. Examples include using body size to calculate community-level indicators (e.g. the Large Fish Indicator), mapping size diversity, and matching species occurrences to environmental temperature or calcification to calculate Community Thermal Indices or impacts on Ocean Acidification on calcifying organisms. We will select and test two relevant cases, for which we will publish the results highlighting the added value of using trait information to improve our knowledge base on the biology of the oceans (D4.4: 3 Two Trait Use Cases).

Deliverables

D4.1: Atlas of data products of European Marine Life, free of charge and free of restrictions of use (M18)

D4.2: Set of relevant baselayers from EMODnet projects for environmental modelling (M12)

D4.3: Portfolio of modelling tools and products for European marine species       (M24)

D4.4: Two examples of application of trait based approaches -related Use Cases (M24)

 

  •