Mapping for ecological analysis

The ecological analysis mapping approach is based on terrestrial methods to look at individual species rather than habitat classes, which obscure detail and lead to unique classification systems being developed for every area studied.

The following have therefore to be taken in to account:

  • Sampling design considerations
  • Spatial Predictive Mapping
  • Supporting datasets

For details see:

  1. Guisan A, Zimmerman NE (2000) Predictive habitat distribution models in ecology. Ecological Modelling 135:147-186
  2. Austin MP (2002) Spatial prediction of species distribution: an interface between ecological theory and statistical modelling. Ecological Modelling 157:101-118

Back to top of page


Mental Connections:

Contemporary Research Approach - Mapping for ecological analysis

Spatial Predictive Mapping

The CRC benthic habitat mapping team chose to use a spatial predictive modelling method to map large areas based on statistical relationships between continuous, full coverage datasets and biological information gleaned from remote observations (video, photos) or physical sampling.

Error Assessment

All maps are models of reality, and by definition, models are simplified versions of complex systems. An understanding of the uncertainty, and the distribution of uncertainty is essential for use maps in reliable and responsible ways.

Errors creep into maps from:

  • Uncertainty in location information for field surveys
  • Equipment tolerances in field surveys and analyses
  • Mapping methods
  • Manual unit delineation

Unknown errors difficult to quantify, different for each person involved

  • Statistically defined map units

Global model error quantified, and distributed errors can be simulated spatially

  • Video and field interpretation errors or observer bias

Some errors are additive, while others are multiplicative. Keeping track of uncertainty in the process and using methods that present uncertainty estimates to the map user promote responsible map use.

Click here for a presentation on map error. Optimizing seagrass monitoring by assessing previous mapping uncertainty. K.W.Holmes, K. Van Niel, and G.A. Kendrick Coastal CRC, Schools of Plant Biology and Earth and Geographical Sciences. University of Western Australia. (2,513 KB)

See Seagrass species modelling and Point Addis case studies

  • Statistically defined habitats
  • Spatial Autocorrelation Assessment
  • Stratified nested spatial Sampling

Sampling designs must match the objectives of the statistical analysis planned, and for the organism or data type being studied. See Milestone 2003 on sampling, optimizing field surveys, sampling biodiversity

Spatial sampling designs for mapping the benthos:
a case study for seagrass species mapping, Owen Anchorage, Western Australia. Researchers: K. W. Holmesand, K. Van Niel. Sub-project leader: G.A. Kendrick. 1School of Plant Biology. 2School of Earth and Geographical Sciences. University of Western Australia, 35 Stirling Highway, Crawley, WA 6009 CRC for Coastal Zone Estuary and Waterway Management. Project CB: Benthic Biology and Habitat Mapping Milestone Report, December 2004. This paper discusses the application of classic (terrestrial) sampling approaches to benthic habitat mapping, and the development of a sampling design for generating maps of seagrass species over a 100-km2 area in Owen Anchorage, near Perth, Western Australia. Click here to download report (3MB)

Traditional maps based on Expert Knowledge

Traditional habitat maps consist of polygons defined by experts through interpretation of underwater video and hydroacoustic information, combined with diving experience and general knowledge of the area or similar coastal zones. This is a semi-quantitative method that produces a single map with map units representing hierarchical mixed classes of physical substrate and biological features.
See case study: Byron Bay

Below is an example from Marmion (WA):

Expert-drawn habitat map for test site in Marmion Marine Park (WA), based on Sidescan imagery (left) and information from video footage

Video coverage used for determining benthic classes at Marmion:

Back to top of page


Mental Connections are:

Traditional Management Approach - Mapping for management

  1. Methods for Video Analysis
  2. Cluster Analysis
  3. Video Interpretation
  4. Map production
  5. Expert Interpretation (Delphi Method)
  6. Local Knowledge
  7. Hand Digitizing
  8. Video point overlay
  9. Remote Sensing Methods for Multibeam Analysis
  10. Multibeam TPI
  11. Multibeam Backscatter
  12. Supervised/Unsupervised Classification
  13. Principal Components Analysis
  14. Types of data used

Remote sampling

Biological sampling

Remote sampling imagery

Bathymetry derived from hydroacoustic surveys
Seafloor Contours plotted over the Colour Shaded data at Point Addis.

Link to multibeam surveys.

Hydroacoustic backscatter:

Link to multibeam surveys.

Textural derivatives of hydroacoustics

Commonly used image processing and digital modelling methods for hydroacoustic data include grey-level co-occurrence matrices and terrain analysis techniques. Below are examples from the Marmion (WA) test site:

Back to top of page