Koninklijk Nederlands Instituut voor Zeeonderzoek WP 5 Creation

Koninklijk Nederlands Instituut voor Zeeonderzoek WP 5: Creation of gridded abundance data products Peter. Herman@nioz. nl

Objectives § Implement DIVA methodology to produce statistically optimized gridded map layers. § Make gridded maps of 3 species per group in appropriate time window § Estimate the accuracy of the gridding procedure by comparison with validation data. § Produce indications of the precision of the result based on the distribution of the basic data § Produce spatial maps (data products) relevant for MSFD Descriptor 2 (non-indigenous species). § Produce spatial maps of quality indicators for MSFD, if available and feasible

Activities. I. Implement DIVA § Ulg will make DIVA implementation available § Use INSPIRE and EEA grid convention (100 m, 1 km, 100 km) § Datasets: § Available: Calanus, benthos North Sea, mammals N. Sea, N. Atlantic, Bay of Biscay § Extend to other sets in other areas dependent on availability

Activities. II. Incorporate environmental covariables § Benthos: grain size, bedforms, depth; Pelagos: water column chemistry, currents § Datasets: § Co-variables: benthic environment (grainsize, maps); pelagic? ? § Presences: mammal observations. Presence/absence: benthic datasets N Sea; Calanus

Activities. III. Convert presence-only into presence/pseudoabsence datasets. § Develop algorithm based on knowledge of sampling programme within projects/campaigns (based on EMODNET I) § Check methodology with presence/absence datasets § Datasets: § Benthos Oosterschelde: full estimation of percentile distributions § Benthos N. Sea; Calanus: presence/absence § Apply to mammals, fish, . .

Activities. IV. Estimate accuracy and precision of gridding § Methodological development within DIVA. State of affairs? § Use to define minimal data requirements for gridding § Test dataset: benthos Oosterschelde § Apply to § Extensive datasets allowing estimation of sampling variability § Sparse datasets

Activities. V. Make invasion movies § Test dataset: Mnemiopsis Black Sea § Determine minimum time step for reliable gridding § Make sequential maps illustrating invasion § Apply to other invasions. Which?

Activities. VI. Produce gridded maps of everything available § Guided by MSFD needs § Check on quality / statistical distribution of indicators § Based on data available within different EMODNET branches

Work programme § Workshop I. Design general setup. Determine minimum required data distribution. Define data format for abundance data and environmental cofactors (all partners). § Implementation of DIVA methodology (ULg, NIOZ, VLIZ) § Inventory of availability of environmental data; setup of data flow model (NIOZ) § Workshop II. Preparation of biological data for production of gridded data products (all partners) § Implementation of suitable indicators for the production of maps (all partners, EU, lead NIOZ) § Production of data products (lead ULg) § Feedback and validation of the data products by data providers (all partners, lead VLIZ)

Playground dataset: Oosterschelde/Westerschelde benthos § Thousands of samples per system over 10 years § Non-linear multi-quantile regressions to estimate probability distributions as function of environmental variables § Univariate case:

abundance Univariate: estimate abundance quantiles from sediment grain size (d 50) D 50 sediment

Comparison Oosterschelde-Westerschelde

Multivariate: currents / depth 0. 975 quantile as function of environment Similar graphs for other quantiles

Validated Reconstructions Habitat suitability

High-resolution sampling from distributions