Current status and methodology used in Estonia for marine habitat - - PowerPoint PPT Presentation

Current status and methodology used in Estonia for marine habitat mapping

Georg Martin & Kristjan Herkül

Estonian Marine Institute University of Tartu

Pu Purpose se of large ge scale le mappin ing/i g/inv nven entor tories es

• Protection and management of nature values
• Sustainable use of living and non-living marine

resources

• Basis for spatial planning of marine areas for

minimisation of possible conflicts between different types of marine uses

• International obligations (e.g. EU Marine Strategy

Directive, HELCOM Baltic Sea Action Plan)

What are the most t importa rtant nt marine ne nature ure value ues? s?

• Species (distribution pattern and dynamics)
• Communities (e.g. Communities having ecological

significant function)

• Habitats
• Single objects (interesting and unique geological

formations)

• Marine landscapes

Estonian sea area

• Salinity: 3-7 PSU
• Duration of ice-

cover: up to 90 days/year

• Largest depths:

up to 140 m

• Complicated

bottom morphology

• Diversity of

coastal types Area: 36481 km2

Territorial sea: 25 200 km 2 EEZ: 11 300 km 2

Current situation with benthic inventories in Estonian coastal sea

First large-scale complex inventories started in 2005. (EU Life project “Marine Protected Areas of Eastern Baltic Sea) Project duration 2005-2009. Currently inventories are carried out in the framework of different projects having two main objectives: 1. Development of Natura 2000 network in Estonian coastal areas: ESTMAR (EEA Grants), Gretagrund, Krassgrund, Paljassaare inventories (Elf, KIK), Nõva-Osmussaare SCA (EU Life) 2. EIA studies for larger technical development projects (offshore windparks, construction and reconstruction of harbours and bridges, establishemnt of new fishfarm and sand and gravel mining areas etc.)

6 project areas Fieldworks 2006-2007 Habitat/fish/bird inventories

Classiffication system of benthic habitats developed in framework of EU Life project “Marine Protected Areas in the Eastern Baltic Sea”

EBHAB Classification is based on physical and biological features:

Exposure Substrate type (quality) Light avaialability (photic zone) Biological communities

Alltogehther 25 classification units for Eastern Baltic Sea (18 for Estonian waters)

Habitat classification

• EBHAB (Eastern Baltic marine benthic HABitats)

SHELTERED MODERATELY EXPOSED HARD

• 1. Fucus
• 2. Mussels & barnacles
• 3. No dominance
• 8. Fucus
• 9. Furcellaria
• 10. Mussels & barnacles
• 11. No dominance in photic zone
• 12. No dominance in aphotic zone

SOFT

• 4. Vascular plants (excl. Zostera)
• 5. Charophytes
• 6. Mussels & barnacles
• 7. No dominance
• 13. Zostera marina
• 14. Vascular plants (excl. Zostera)
• 15. Charophytes
• 16. Furcellaria
• 17. Mussels & barnacles
• 18. No dominance

• EU Habitats Directive Annex I marine habitat types

– Sandbanks which are slightly covered by sea water all the time (1110) – Estuaries (1130) – Mudflats and sandflats not covered by seawater at low tide (1140) – Coastal lagoons (1150) – Large shallow inlets and bays (1160) – Reefs (1170)

Mapping based on field inventory

• Establishment of the sampling grid
• Field works
• Drop camera
• ROV
• Grab samplers
• SCUBA
• Analysis of field data
• visual data (dektop analysis): percentage cover of

species/groups, sediment type

• biomass data: biomass, abundance of benthic species
• Interpolation: point data  coverage data
• Overlay analysis, cell-wise classification

Benthos sampling

• Drop camera
• ROV
• SCUBA diving
• Bottom grab samplers

Establishment of sampling grid Interpolation Benthos sampling Overlay analysis Habitat maps

>9600 km2

Predictive modelling approach

Probability of occurrence of a key species

STATISTICAL MODEL Response variable: Biological point data Predictive variables: GIS- nvironmental data layers of e Prediction: GIS-layer of biological response variable

· · ·

Variable contributions Variable response curves Model validation

Predictive modelling method

Input data Biological point data (response)

• Coverage estimates from video and scuba
• Biomass samples
• Coverage, biomass  presence/absence (binomial models

perform better) GIS-layers of environmental data (predictors)

• bathymetric: depth, seabed slope, aspect
• wave exposure
• salinity
• temperature
• sediment
• currents

Predictive modelling method

Statistical models

• generalized additive models (GAM)
• boosted regression trees (BRT)
• random forests (RF)
• Training data: 70-90% of biological point data
• Validation data: 10-30% of biological point data
• Validation method: ROC-test

Examples of modelling input data – species distribution

Available environmantal layers:

• depth
• Slope

declination

• exposure
• sediment
• salinity
• temperature
• currents
• O2

Results of modelling of species: bladder wrack, Furcellaria lumbricalis, Charophytes, vascular plants

Results of modelling of species: eelgrass, Mytilus, Balanus, infauna-Bivalvs

Areas where probabilty of

• ccurence of characetristic

species for habitat type „sandbanks“ is higher than 0,5 and dominance of sandy susbtrate Areas where probability of

• ccurence of characteristic

species for habitat type „Reefs“ is higher than 0,5 and hard substrate

Multibeam sonar (in use from 2013)

• Reson SeaBat 7101-Flow
• 511 equidistant beams
• swath coverage 150°
• frequency 240 kHz
• depth range 0.5–200 m

Unsupervised classification Habitat maps Sonar measurements Processing of sonar data

Depth Backscatter Slope Rugosity

Ground truthing

• f classes

Modeling Sonar measurements Processing of sonar data

Depth Backscatter Slope Rugosity

Benthos sampling Species distribution maps

Future development

Integrate different data sources and models to improve accuracy

• f bethic habitat maps

Biological point data Remote sensing data Physical & chemical environment data Acoustic data Statistical methods Benthic habitat maps with high spatial resolution and accuracy

Remote sensing Data collection In situ sampling Other environmental data

• bathymetric data
• hydrodydamic data: wave

exposure, currents

• salinity, currents, temperature etc.

Modelling

• Filling in gaps between

ecological point data

• Predicting biological and

seabed substrate data based on remotely sensed and other environmental data

• Methods:

– interpolation – statistical models – machine learning Map production Distribution of:

• key species
• habitats
• species

richness

• HELCOM Underwater Biotope and habitat classification

(HELCOM HUB) – to be implemented in coming mapping projects

Area covered by mapping (2015): 12446 km2