MAPPING AND ASSESSING ECOSYSTEM CONDITIONS OF SOUTH ...

MAPPING AND ASSESSING ECOSYSTEM CONDITIONS OF SOUTH AFRICAN KELP

FORESTSBY

Loyiso Victor DungaUniversity of Cape TownDepartment of Biological Sciences

Under supervision of: Dr. Sink Kerry; Dr. Blamey Laura; Dr. Rothman Mark Dr. Lück-Vogel Melanie and Dr. Bolton John

Introduction: What are kelps?

• Group of large brown algae or seaweeds from the Orders Laminariales

• Dominating 25% of world’s coastlines

• Spanning across 43% of the world’s ecoregions.

KELPS:Phyletically diverse, structurally complex and highly productivecomponents of cold-water rocky marine coastlines

Steneck et al. 2002

How are Kelp Forests Ecosystems (KFE) significant?

• Kelp beds / kelp forests

Terrestrial forests Kelp forests

Associated phyla Chordata, Arthropoda and Annelida Chordata. Arthropoda, Annelida, Echinodermata, Bryozoa, Cnidaria, Mollusca, Platyhelminthes, Brachiopoda, Porifera

Height 10 – 30 m in 20 - 30 yrs. 1 – 15 m in 1 – 3 yrs.

Longevity Centuries to millennia 25 years

KFE (structure producing kelps + myriad of associated biota

KFE (key ecosystem services with Ecological, Social and Economic significance)

Comparison terrestrial – Kelp forests

Ecological Social Economic

Biogenic habitat Human dependence on the resource (direct or indirectly)

Harvesting and processing

Dampening effect and detritus Coastal rocky shore protection Important food source

Harbours important species(Nursery, Shelter and food source

Intrinsic beauty Fisheries

Ecologic, Economic and Social KFE Services

(a) A kelp forest (Ecklonia maxima) in South Africa. (b) A kelp trawler in Norway. (c) A scientific diver undertaking kelp research in Australia.. Photographs: From (Filbee-Dexter et al. 2018)

Problem statement & research aims and objectives

Aim:To produce a national kelp forest map for South Africa

Objectives:1. Develop a S2 based approach for extraction of kelp forests as baseline for future monitoring2. Produce a kelp map for South African coast as a status-quo information

South African kelp forests are well studied, yet it has been more than a decade since they were mapped using low spectral and spatial resolution techniques. These mid-2000s maps by (Tarr (1993; Rand 2006, Anderson et al. 2007) after analysis were found to have substantial kelp beds missing. As significant components of biodiversity Kelp forest were not included in the 2011 National Biodiversity Assessment and are listed under the priority research areas.

To serve as a decision support tool in the operational coastal vegetation management context

Study area

Methods and Material• Sentinel 2 A satellite remote sensing imagery• Sen2Cor: TOA to BOA conversions • Three classification methods : Visual analysis and interpretation, (angular training polygons) supervised and

(cluster 2 pixel groups) unsupervised• ArcGIS software and QGIS : Raster imagery Pre-processing and post processing

SENTINEL 2AAcquisition

SEN2CORConversion

ArcGIS/QGIS

Training data

NDVI

Classifications

Mosaic Validation data

Analysis

NDVI MAPPre-processing

Processing Post-processing

Kelp beds

NW SpainSPOT 43 classification methodsResolution compatibilityCasal et al. 2011

Japan IKONOSSpectral signature Acute results Sagawa et al. 2012

West of South AfricaLandsat 4Multi-methods Under-estimationAnderson et al. 2007

West of South AfricaSentinel 2A3 classificationprogressDunga, 2018

Comparison of RS derived kelp maps

Satellite Bands Temporal

Resolution

Principle Spatial Resolution (m)

Sentinel 2A 13 10 days Push broom 10; 20; 60

Landsat 3-7 7 16 days Scanner 30; 60

SPOT-5 4 26 days Push broom 2.5; 10; 60

RS for Mapping kelp: WHY S2A?

Tar, 1993 (Hand drawn) S2A: 2017 (Raw image) S2A derived Kelp map

NDVI range of kelp non-kelp areas

Normalized Difference Vegetation Index = (NIR reflectance – Red reflectance) / (NIR reflectance + Red reflectance)

y = -0.087ln(x) + 0.34R² = 0.053

y = -0.0117x + 0.1935R² = 0.0415

-0.6

-0.4

-0.2

0

0.2

0.4

0.6

0.8

Kommetjie y = 0.0587x - 0.2479R² = 0.3958

y = 0.0147x - 0.0351R² = 0.0541

-0.4

-0.2

0

0.2

0.4

0.6

0.8

Cape Hangklip

y = 0.0297x - 0.283R² = 0.2357

y = 0.0042x - 0.237R² = 0.0071

-0.6

-0.4

-0.2

0

0.2

0.4

0.6

Robben Island

Kelp No Kelp Linear (Kelp) Linear (No Kelp)

y = -0.0009x + 0.3553R² = 0.004

y = -0.0002x - 0.3369R² = 0.0017

-0.6

-0.4

-0.2

0

0.2

0.4

0.6

0.8Dassen Island

Kelp No Kelp Linear (Kelp) Linear (No Kelp)

NDVI range of kelp non-kelp areas

Results

Drawn ConclusionsS2a capability to determine kelp is satisfactory (NDVI thresholds)

Objective 1:

• .Develop a S-2 based mapping system for kelp occurrence as baseline for future

monitoring

• Produce kelp map for the whole of South Africa coast as a status-quo information

• Kelp forest includes all surface emergent kelp and but excludes isolated beds that where

ground truth data is not available e.g. the case of the EC.

Three kelp ecosystem types (Namaqua, Cape and Agulhas) for the following chapter are

captured in the map

Where to from here?

• Pressure Mapping of three Kelp ecosystems (Namaqua; Southern Western Cape and Agulhas) Conditions.

• Based on identified Expert based pressures• Adopting Halpern et al 2008;2009 & Teck et al 2010

methods

ACKNOWLEDGEMENTS

Thank you all for listening!

Dr. Sink Kerry; Dr. Blamey Laura; Dr. Rothman Mark Dr. Lück-Vogel Melanie; Dr. Phil Desmet; Prof. Bolton John;