Hey guys. Now we know about habitat mapping techniques, and today I am going to talk about using our data and making the final habitat map. OK. Let’s get to the point.
Habitat maps are the final result of integrating continuous coverage of physical properties of seafloor with the observation of communities present at discrete sections. It is worth mentioning that the final product is a prediction of the distribution of habitats the best habitat mapping approach is the one which integrates the remote-sensing data providing a full coverage of seabed and ground-truthing data that validate the sensed data. For remote sensing data, the best technique is acoustic remote sensing. However, with this technique, we cannot directly define the habitat. They provide the physical attribution of the seafloor, using the backscatter strength images; except for the regions that the seafloor is formed by biotas such as biogenic reefs or mussel/oyster beds (MESH).
The final map is a prediction of the distribution of seabed habitats, with the complete coverage environmental data acting as a proxy for the habitat data.
- Environmental data layer
Seafloor geology along with morphological characteristics and overlying water column attribute has a huge impact on benthic biology. This environmental data layer mainly provides spatial information of these parameters which are collected using different methods. Each method has its benefits and drawbacks. The new approach, using MBES; records information related to bathymetry and backscatter strength and other valuable seafloor information can be extracted from both bathymetry (slop, aspect, terrain variability) and backscatter (hardness, roughness, acoustic class) and the resolution of these system varies from ten centimeters to ten meters.
Another important factor that has a dominant influence on biological characteristics of the seafloor are oceanography parameters. Spatial oceanographic data are multi-sourced. They can be gathered at continuous coverage from satellite (SST or chlorophyll content that can be used as an indication of productivity) or interpolated from point sample measurements (Temperature, salinity, oxygen) or continuous coverage by models and in comparison to acoustic data(MBES), they have the resolution of tens of meters to tens of kilometers which in turn introduce a limit on the scale of habitat mapping and the habitat maps which use oceanography parameter tend to be conducted at border scales.
2. Utilization of environmental data sets (High-resolution acoustic data)
Environmental data should be mange in a way that can be integrated with in-situ data efficiently. Although habitat patterns can be continuous or discrete sections, however, in the majority of cases high-resolution acoustic data sets are divided into “spatial unites” before integration of habitat information. These units are called segment (C.J Brown et al., 2011).
- Backscatter analysis approaches
The most widely used from MBES data in habitat mapping is acoustic backscatter data which identify different region from acoustic backscatter strength and these acoustic classes are then linked to the specific seafloor habitat attributes from the ground-truthing dataset. MBES system covers a wide range of area with high resolution in comparison to SBES that suffers from discrete data between survey lines, however, segmenting the backscatter from SBEC are much easier than MBES. In the field of habitat mapping MBES are becoming the first choice due to being capable of gathering both bathymetry and backscatter data simultaneously and using motion reference system to adjust the vessel pitch, heave and roll when positioning the data relative to the seafloor. Different image-based approach methods in the segmentation of MBES backscatter have been used such as neural network technique, Bayesian decision rules, textural analysis based on grey level co-occurrence matrices but there is no widely accepted approach to segment the backscatter data. Another approach to extract quantitative information from MBES backscatter data is a signal-based method. In this method, the variation of backscatter strength with the angle of incident (which is an intrinsic property of seafloor) can be used for acoustic seafloor characterization by extracting several parameters from successive sonar pins. Then to link the acoustic backscatter observation to the seafloor properties, the average angular response is compared to mathematical models. This approach shows the best approach in the field of habitat mapping however it should be tested over a range of benthic marine ecosystems.
energy and dark-toned areas show low backscatter energy (J.Shaw et al., 2012)
- Bathymetric analysis approaches
Bathymetry can be used to segment an area in which identify distinctive biological characteristic because benthic species show a tendency to stay in certain depth and topographic conditions. We can use other features of bathymetry such as slope, orientation, and curvature to segment the seafloor (C.J Brown et al., 2011).
3. Utilization of environmental data sets: oceanographic data
We can predict the distribution of biological characteristic using the properties and conditions of underlying water column which are important in supplying food, nutrient, and gametes. According to the low resolution of oceanographic data (from ten meters to ten kilometers), there is a disparity between the resolution of acoustic data and oceanographic data to provide habitat maps.
4. Ground-truthing (adding the benthic biological information)
To measure the biological feature of the seafloor the best approach is in situ ground-truthing which can be linked to environmental layer and the type of in situ method depend on many factors (the purpose of the survey, survey platform and etc.) and it has a profound effect on the final product. Some methods are grab sampling, trawls, underwater photographs or video, each of which can be applied to a specific region or for comprehensive data collection, applied two of them together(C.J Brown et al., 2011). There should be some sample video recording for training site as ground-truthing and some other sites to validate the final result (A.Micallef et al., 2011).Two strategies are used to identify the distribution of a species. The first method is fully quantitative which is based on the abundance, the number of individuals of each species or the biomass (the total weight) of each species and is widely used in the grab and trawl sampling. In the second method, semi-quantitative; they score the abundance of each species according to their amount using a predefined table like SACFOR.
5. Correlate biological and physical patterns
A number of methods are available to examine the correlation between biological and physical patterns. Two commonly used computer packages with a variety of multivariate methods are PRIMER and CANOCO. These are based on a multivariate statistical analysis which calculates the correlation between the parameters.
6. Final product
The lack of high-resolution spatial environmental data has led to finding some difficulties in benthic habitat mapping. Therefore, most of the marine habitat mapping studies use simple statistical models to determine associations between ground-truthing and environmental data sets.
- abiotic surrogates (unsupervised classification – limited or no ground validation)
It is mostly used in broad scale such as continental shelf region but there are studies used this strategy for smaller scales, however, output maps in this scale are not a good reference for prediction of species patterns and useful for management objectives.
- Assemble first, predict later (unsupervised classification)
This is the most common strategy used for producing single species maps, community maps, or maps of generalized habitat classes based on observed geological/biological seafloor characteristics. This strategy uses a top-down approach and includes combining organized data of environment and biology/geology. In the case of single species habitat mapping, presence/absence of an important species is modeled considering environmental situations which are usually a simple statistical assessment of the correlation between the data sets. In community mapping ground-truthing data are organized into classes then the data sets are modeled.
- Predict first, assemble later (supervised classification)
The third strategy takes a bottom-up approach which uses biological/geological ground-truthing data to organize environmental data. In the case of single species mapping the presence of an important species is modeled as a function of the environmental predictors. Theoretically, we can combine single species habitat maps to produce a community distribution map. However, commonly for producing community maps, data are organized into classes then use to perform some form of supervised classification on the environmental data sets to segment the continuous coverage variables.
Over the past decade, there has been much attention in the field of benthic habitat mapping and rapid improvement in our ability to map seafloor habitat. With the advancement in acoustic survey tools and using three aforementioned strategies, we can provide valuable benthic habitat maps and insight for decision makers to achieve a sustainable marine environment.
References:
A.Micallef, T.P.L.Bas, V.A.I.Huvenne, P.Blondel, V.Huhnerbach, A.Deidun. (2012). A multi-method approach for benthic habitat mapping of shallow coastal areas with high -resolution multibeam data. Continental Shelf Research, 14-26.
C.J.Brown, S.J.Smith, P.Lawton, J.T.Anderson. (2011). Benthic Habitat Mapping: A review of progress towards an improved understanding of the spatial ecology of the seafloor using acoustic techniques. estuarine, Coastal and Shelf Science, 205-520.
J.shaw, B.J.Todd, M.Z. Li. (2012). Seascapes Of Bay of Fundy. Nova Scotia/New Brunswick: Geological Survey of Canada.
Mapping European Seabed Habitat (MESH).
Marine habitat mapping 