Journal of Sea Research 100 (2015) 46–61
Contents lists available at ScienceDirect
Journal of Sea Research journal homepage: www.elsevier.com/locate/seares
Habitat mapping as a tool for conservation and sustainable use of marine resources: Some perspectives from the MAREANO Programme, Norway L. Buhl-Mortensen a,⁎, P. Buhl-Mortensen a, M.J.F. Dolan b, G. Gonzalez-Mirelis a a b
Benthic Habitat Research Group, Institute of Marine Research, PO Box 1870 Nordnes, N-5817 Bergen, Norway Geological Survey of Norway, PO Box 6315 Sluppen, N-7491 Trondheim, Norway
a r t i c l e
i n f o
Article history: Received 5 February 2014 Received in revised form 18 September 2014 Accepted 18 October 2014 Available online 17 December 2014 Keywords: Marine landscape Habitat heterogeneity Sampling scale Biotope identification Health status Seafloor integrity
a b s t r a c t One of the main goals of marine spatial management is to promote a sustainable use of marine resources without putting biodiversity and habitats at risk. Environmental status assessments of benthic habitats have traditionally been conducted on soft bottom infauna communities. These communities represent only a limited part of the total diversity of seabed environments. Large and habitat forming organisms that are particularly vulnerable to physical disturbance of the seabed are in general associated with mixed or hard substrates. Together with mobile benthos these large organisms have been poorly mapped with the traditional approach. In accordance with a demand for information on all aspects of benthic habitats there is an increasing interest in a broader mapping approach for assessments of the distribution and status of benthos. This has led to an increased demand for a broader mapping approach for assessments of the distribution and status of benthos to include as many habitats as possible. We present mapping strategies based on literature together with perspectives from Norway's MAREANO Mapping Programme (www.mareano.no). A first and important step is to acquire high resolution topographical information from multibeam echosounder surveys. Building upon this baseline data set, video inspection along transects can provide information about how megafauna and surface sediments relate to local and broad scale topographical features. Classification of habitats/biotopes can be carried out through the analysis of the megafauna composition and its relation to environmental variables such as depth, quantitative terrain descriptors, and substrates. Marine landscapes have several definitions but generally refer to the major features delimiting broad-scale habitats. Each landscape will include different substrates that can be subdivided into smaller biotopes with specific fauna composition, functionality and production. For these habitats and biotopes information from all sampling gears can then be used to describe a more complete community composition. Experience from recent seafloor mapping indicates that a broad approach is required to support evidence-based policy and management of benthic species, communities and habitats. The approach presented can be used to identify biologically valuable areas and assess health status for bottom habitats/biotopes in a broad set of marine © 2014 Elsevier B.V. All rights reserved.
1. Introduction One of the main goals of marine spatial management is to promote a sustainable use of marine resources while not putting marine biodiversity and habitats at risk. Objectives for marine biodiversity and habitats are stated in the Biodiversity Convention, Habitat Directive and the Marine Strategy Framework Directive (EC, 2008a; EEC, 1992; UN, 1992), and they affirm that no species or habitats should be lost, and that the integrity of the sea floor should not be compromised by human activities. The ICES Working Group of Marine Habitat Mapping (WGMHM) reviewed existing definitions of Habitat (ICES, 2005) and
⁎ Corresponding author. E-mail address:
[email protected] (L. Buhl-Mortensen).
http://dx.doi.org/10.1016/j.seares.2014.10.014 1385-1101/© 2014 Elsevier B.V. All rights reserved.
developed the following working definition of the term: a recognizable space which can be distinguished by its abiotic characteristics and associated biological assemblage, operating at particular spatial and temporal scales (ICES, 2005). To make marine spatial plans (MSP) and decisions that can reach these objectives requires knowledge of the composition and distribution of benthic communities, the characteristics of a natural and healthy state, and the effects of different human activities (e.g. EC, 2008b; epbrs, 2013; Steltzenmüller et al., 2013). It has been estimated that only 5–10% of the seafloor is mapped at a comparable resolution to similar studies on land (Wright and Heyman, 2008). Furthermore, marine ecosystems are poorly described compared to their terrestrial counterparts. On land the proportion of unknown habitats has been estimated as 17 % whilst for the marine realm it has been estimated as 40 % (EC, 2007). In recommendations from the European Platform for Biodiversity Research Strategy (epbrs, 2013) it
L. Buhl-Mortensen et al. / Journal of Sea Research 100 (2015) 46–61
was emphasized that “a sound reporting based on scientific methods and knowledge is of major importance” and that it is recognized that “research is needed to substantially advance our knowledge of marine habitats and species in support of evidence-based policy and its implementation”. The ability to reach national and international management goals depends, to a large degree, on detailed knowledge of the benthic environment and ecosystem including its state of health and signs of human impact. In perspective of the present knowledge these management goals are ambitious and require a better understanding of bottom fauna environment relations. The management goals do not relate to any particular faunal group or size class of the benthic community, and it is also well known that terms like habitat, biodiversity and ecosystem are concepts that depend upon the sampling methods (Buhl-Mortensen et al., 2012a; Costello, 2009). This in turn demands for a comprehensive mapping strategy. We have summarized some specific research needs and knowledge gaps relevant to habitat mapping in Table 1. MAREANO is the Norwegian multidisciplinary seabed mapping programme, funded by the Central Government of Norway and designed to provide scientific information required to support Norway's obligations to national and international policy objectives pertaining to the conservation and sustainable use of its marine resources. The goal is to obtain information that can be used as a scientific basis for the regulation of human activities such as the petroleum industry and fisheries (Buhl-Mortensen et al., 2015). The provision of information on benthic habitat forms is an important component of MAREANO and the main focus of this paper. Information about benthic habitats and biological communities are important for the implementation of ecosystem-based management of the sea and in assessing the consequences of human activities, and assessment has to have a clear link to the management objectives (Buhl-Mortensen et al., 2012b; Steltzenmüller et al., 2013). Since it began in 2005, MAREANO has made good progress but has also encountered many of the technical, logistic and conceptual problems inherent in this type of work. Many solutions have been found and the programme has developed a pragmatic approach to seabed mapping suited to the particular environment of the Norwegian waters. Traditional assessment of the environmental status of benthic habitats/biotopes has often been based on infauna, sampled mainly on soft substrates where this fauna dominates (using grabs ~ 0.1 m2) (Brown et al., 2011; Gray and Elliott, 2009). This approach only documents a small part of the benthic substratum and fauna diversity.
47
Natural complexity occurs at several spatial and temporal scales (Fortin and Dale, 2005) and faunal responses to this complexity differ among organism groups (e.g. Buhl-Mortensen et al., 2012a). Depending on depth, hydrography, and local landscape complexity the substratum can be homogeneous but seasonally variable in fauna composition. It may also exhibit varied topography and sediment properties yet have a stable faunal composition over time (Buhl-Mortensen et al., 2010; Gray and Elliott, 2009). Capturing information on all these factors demands for a diverse approach to surveying and sampling, and is made most effective using multiple, complementary sampling gears. This is particularly important when the mapped outputs are intended to provide information that can support multiple management decisions related to marine biodiversity and habitats. Using acoustic technologies such as ship-borne multibeam echosounder it is possible to acquire high-resolution, full-coverage imagery of the seafloor over extensive areas. The resulting topographic data and geophysical attributes of the terrain, which are derived from their acoustic properties, form an excellent basis for geomorphic and geological classifications of the sea floor when interpreted together with supporting data from video or bottom samples. These physical classifications of the seabed in turn form an essential component in benthic habitat mapping that integrates biological properties and has been widely used (Brown et al., 2011 and references therein). Visual inspection using underwater video camera not only provides detailed information on the general features visible in multibeam data, they provide an excellent basis for biological investigation across a range of bottom types. Sessile benthic organisms are useful for habitat characterization because substrate is critical in determining their aggregated performance. Since they are static, it is these organisms that are most indicative of environmental conditions of the adjacent seafloor (Kostylev et al., 2001). Classification of seabed habitats and biotopes can therefore be based on the composition of benthic megafauna and substrates analyzed from video records of the sea floor (Buhl-Mortensen et al., 2009, 2015; Mortensen et al., 2009; Gonzalez-Mirelis et al., 2009). The spatial scale of variability is crucial for choosing an adequate sampling design involving number and density of sampling stations, gear and area covered by a particular sampler. The sampling design imposes a filter with its own specific temporal and spatial resolutions (Bradshaw and Fortin, 2000). In addition the substratum sets limitations on what type of sampling gear can be used and thus the assessability of the whole range of species diversity. Furthermore, contrasting
Table 1 A list of knowledge gaps and research needs that are relevant for marine mapping for biodiversity and habitat management. Information gathered from epbrs and OSPAR documents.
Mapping Research needs
Knowledge gaps
Species–environment relationships Research needs
Knowledge gaps
Monitoring Research needs
Knowledge gaps
Classification and description
Variability (spatial and temporal)
Habitat/biotopes in different marine environments (from shallow to deep sea, soft to hard bottom), based on species compositions and substrates. Definition of habitats/biotopes is lacking for most marine ecosystems. In particular for mixed and hard bottoms offshore.
Compare species composition between habitat/biotopes including infauna, epifauna, and hyperfauna
Environmental characteristics for habitats/biotopes.
Understand natural variation in environment and biodiversity in order to design optimal monitoring programs. Range of natural variability in distribution and abundance of most species and communities is poorly understood.
Distribution of marine organisms in relation to environment is unknown for many marine ecosystems and related organism groups. Define GES for habitats/biotopes based on abundances, biomass and morphological health indicators to representative organisms or an index that mirrors the health status. Relevant indicators of GES are not available for most habitats/biotopes and landscape elements. Habitat/biotope and species specific effects of pressures from different human activities are lacking.
Natural range in abundance within the same habitat/biotope is lacking.
Long term monitoring from the same sites that can provide environmental status for established habitats/biotopes.
Develop relevant monitoring programmes suitable for different habitat/biotopes.
48
L. Buhl-Mortensen et al. / Journal of Sea Research 100 (2015) 46–61
water bodies can explain fauna changes that are not supported by visible changes in the environment. The distribution of habitat/ biotope classes can reflect different environmental settings in marine landscapes (e.g. banks and troughs on the shelf). These are often the major features comprising habitats with different substrates that can be subdivided into biotopes with specific fauna composition, functionality and production. Marine habitat mapping has been defined as “Plotting the distribution and extent of habitats to create a map with complete coverage of the seabed showing distinct boundaries separating adjacent habitats” (MESH, 2008). In this context, the term “habitat” is defined as “both the physical and environmental conditions that support a particular biological community together with community itself” (MESH, 2008). Since most habitat classification schemes do not distinguish between ‘habitat’ and ‘biotope’ we combine these terms as ‘habitat/biotope’ in this paper (for a review of this terminology see ICES, 2005). Furthermore, we here use the term “mapping” to cover the whole process from survey planning via sampling through to geographic representation. There are several approaches to seascape and habitat/biotope classification that are used in marine mapping. For example Greene et al. (1999) provide a classification scheme for deep seafloor habitats where the issue of the scale is dealt with in a hierarchy of classes. A hierarchical classification system also forms the basis of the European Nature Information System (EUNIS) (Davies et al., 2004). Both classification systems take into account the biological components of the habitat classes. However, whereas the Greene et al. (1999) classification scheme uses the biological components as modifiers of geological and geomorphological features at an intermediate level (macro and meso habitats) the EUNIS classification emphasizes taxonomic composition at the lower levels (levels 8–10). The main differences are the scale and application of ecological knowledge when generating the habitat map. Mapping “habitat” normally requires knowledge of the benthic biota and ecosystem details that are not required for physical (landscape) approaches. The habitat approach that offers most successful predictions involves integration of biota and a statistical approach to habitats. Here we will present a mapping approach that is based on literature in perspectives from the experience drawn from the MAREANO mapping programme (www.mareano.no). The presented approach can: identify biologically valuable areas, assess human impact, and health status for marine bottom habitats/biotopes in a broad set of marine landscapes.
2.1. Methods
2. MAREANO — as an example
Fig. 1. provides an example of how information is used for survey planning. In addition to bathymetry and backscatter data pre-cruise analysis includes the production of an unsupervised classification map (ISOCLUSTER). This map provides the estimated distribution of seabed areas that differs with respect to topography and sediment type and for areas adjacent to earlier mapping sites this indicates the distribution of different habitat types. At sampling station fauna is collected using grab, epibenthic sled and beamtrawl. Whether the whole suite of gears can be used depends on the substratum, and the decision on which gears to deploy is based on the video information from the locality. In sedimentation areas with soft sediment a multicorer is also deployed providing data that is used to map the occurrence of pollutants at different depths in the sediment, reflecting the environmental status at different points in time (Boitsov et al., 2011, 2013).
The mapping conducted by MAREANO is designed to cover all parts and scales of the macro-benthic community equally well including diversity and productivity with no organism group having a specific priority (Buhl-Mortensen et al., 2015). By using a variety of complementary sampling gears to ensure that a broad set of benthic organisms on all types of seabed is sampled, MAREANO can offer insight into the species-diversity, biomass and production of benthic communities within various biotopes. This is a unique level of biological information for a mapping programme. As pointed out by Brown et al. (2011) all these gears provide detailed information on the small area of seafloor that they sample and it is difficult to derive an accurate representation of the broader spatial configuration of the seafloor biophysical characteristics. Depending on the part of the benthic fauna that is studied the number and size of habitats/biotopes will differ and there is not one approach that can provide a solution that is universally applicable for all the benthic fauna (e.g. Buhl-Mortensen et al., 2012a, 2012b; Pitcher et al., 2012). One way to overcome this problem is to let megafauna dictate the scale and number of biotopes in an area and then fill in data from other organisms sampled within the biotopes (Buhl-Mortensen et al., 2012a).
The MAREANO mapping strategy revolves around two major sets of data, one comprising a number of full-coverage, spatial data sets, and another one comprising detailed records of abundance of benthic megafauna for a number of sites (Buhl-Mortensen et al., 2015). This strategy is based on a diverse approach to seabed survey and mapping using multiple, complementary sampling techniques and various data integration methods, in accordance with the requirement that outputs are to provide information useful for the management of marine biodiversity and habitats. Mapping in an area usually starts with a multibeam survey conducted by the Norwegian Hydrographic Service which provides bathymetric maps and backscatter. The information is used to provide high resolution bathymetry maps. In addition this information is vital for the planning of the proceeding sampling surveys acquiring geological and biological information in the same area. By the end of 2013, a total of 18 sampling cruises were completed, equalling 360 days at sea. Each cruise covers between 5000 and 10,000 km2 depending on depth. The mean density of observation sites is 10 video stations per 1000 km2, and at two of this sites sampling is conducted. It takes ~ 24 h to complete a video transect and sampling at a depth larger than 1000 m which is three to four times longer than at 300 m. The sampling sites are not evenly distributed within the mapping area, but are assigned to ensure sufficiently a representation of different depth strata, marine landscapes, landscape features and sharp environmental gradients indicated in the multibeam data. The decision on where to position video and sampling stations aim to document as many habitat types as possible and is loosely based on a random stratified sampling strategy. A few stations are also specifically targeted to features of special scientific interest. Results from previous mapping by MAREANO now allows for testing of what is the most parsimonious sampling design to be able to document the patchy of habitats in different landscape elements. Survey planning makes full use of multibeam bathymetry and backscatter data, and video stations are positioned to cover: • Topographic variation and gradients between major geomorphic features (e.g. banks, troughs, and canyon walls) • Variation in sediment type (indicated by multibeam backscatter) • Achieve good geographic coverage • Document features of special scientific interest.
2.2. Video registration and sampling The seabed is inspected with the video platform CAMPOD. This platform is a tripod equipped with a high definition color video camera (Sony HDC-X300) tilted forward at an angle of 45°. It also has a standard definition video camera for navigation purposes, lights (2 × 400 W HMI), depth sensor, CTD, current meter, turbidity sensor, and altimeter.
L. Buhl-Mortensen et al. / Journal of Sea Research 100 (2015) 46–61
49
Fig. 1. Example data used for planning MAREANO survey stations. (A) Location map at Skjoldryggen in the Norwegian Sea, surveyed by MAREANO in 2013. (B) Multibeam bathymetry shown as color shaded relief. (C) Multibeam backscatter map indicating variation in sediment properties. (D) ISOCLUSTER map based on unsupervised classification of bathymetry, backscatter and derived quantitative terrain variables. The colors indicate physically different areas based on these variables — similar colors indicate areas with similar physical characteristics while contrasting colors indicate physically different areas. Planned video transects indicated with white lines, green circles indicate full sampling stations.
The videos are recorded on hard-drives on board the vessel. Video transects are 700 m long, as this length was sufficient for comparing the fauna composition between sites and for documenting the diversity of megafauna (animals N 5 cm). Each transect starts with the video platform parked on the seafloor recording an overview and zooming in on details within an area of approximately 6 m2. After inspection of the starting point, the video platform “CAMPOD” is towed behind the survey vessel at a speed of 0.7 knots and controlled by a winch operator providing a near-constant altitude of 1.5 m above the seabed. To provide the best view the video is towed uphill in steep terrain. At the end of the video transect the video platform is parked at the sea bed and the area is inspected the same way as at the starting point. Geopositioning for the video data is provided by a hydroacoustic positioning system (Simrad HIPAP and Eiva Navipac software) with a transponder mounted on “CAMPOD”, giving a position accurate to 2% of water depth. Navigational data (date, UTC time, positions and depth) are recorded automatically at 10-second intervals using the software CampodLogger made at IMR. This software is also used to take systematically field notes of fauna, bottom types, signs of fishing impact, occurrence of litter and local geological seabed features during video recording (Fig. 2.). In the laboratory a detailed video analysis is undertaken where all organisms are identified to the lowest possible taxon and counted, or quantified as % seabed coverage following the method described by Mortensen and Buhl-Mortensen (2005) using another custom made software, VideoNavigator (IMR). Abundance data (the number of organisms counted divided by the area observed) were standardized as number of individuals per 100 m2 (Dolan et al., 2009; Mortensen et al., 2009). On the sampling stations a van Veen grab (0.25 m2) is used for the documentation of infauna; epifauna is sampled by beam trawl (mesh size 5 mm, 2 m opening width, 5 min hauls, see Bergman et al., 2009 for gear description). Hyperfauna is sampled with a modified Rothlisberg & Pearcy — epibenthic sled (mesh size 0.5 mm, 1 m opening width, 10 min hauls, see Buhl-Jensen, 1986 for a description). The
sampled organisms are identified to the species level in the lab and are counted and weighed for biomass registration (Buhl-Mortensen et al., 2012a, 2015). Information on geology and geology and geomorphology is an important basis for habitat mapping. Geological mapping in MAREANO makes full use of multibeam bathymetry and backscatter data, which are typically available to at least 5 m resolution on the continental shelf. Data are interpreted with respect to surficial bottom sediments following the methods described by Bellec et al. (2009) to produce maps of sediment grain size, sedimentary environment (erosion and deposition areas), and sediment formation. In making their interpretation geologists use video and other supporting data (e.g. TOPAS penetrating sonar) to classify backscatter decibel values in backscatter mosaics, and to interpret sediment variation with respect to topography. This expert driven process allows the geological processes occurring in a study area to be interpreted, and consolidated in map form at the appropriate level of detail for the 1:100 000 maps produced by MAREANO. Broad scale geomorphic features are also identified by MAREANO and represented as landscape maps. These maps are based on terrain analysis of bathymetry data as detailed by Elvenes (2013) and are in accordance with landscape level classification under the Norwegian Nature Type classification system Halvorsen et al. (2009). In addition to their utility as a delimiter for geomorphic feature types bathymetry-derived terrain variables e.g. slope and curvature (see summaries by Dolan et al., 2012; Wilson et al., 2007) may also serve as proxies to more direct effects that influence the distribution of benthic fauna (see below). 2.3. Classification and modeling of habitats/biotopes MAREANO performs two types of biotope modeling: 1) fullcoverage mapping where the whole mapping area (i.e., every pixel) is assigned to the most likely biotope type; and 2) targeted mapping of
50
L. Buhl-Mortensen et al. / Journal of Sea Research 100 (2015) 46–61
Fig. 2. A) Screen shot from the software (CampodLogger ver. 2.0.39; IMR) developed for real-time annotation of seabed observations during video recording. The software allows for selecting generic bottom types from a drop-down menu, entering taxa names and comments manually. The lower frame of the software window displays the bottom profile as the ship and the video-platforms move forward. B) Screen-shot from the software (Video Navigator; IMR), developed for post-cruise analysis of video records. The panels allow for selecting taxa from a drop-down menu, and slide bars for recording the composition of substrate types.
selected biotopes identifying the parts of the mapping area covered by this habitat. 1. Mapping is underpinned by a biotope classification scheme developed specifically for the area in question using classification of epibenthic megafaunal data from video transects. 2. Mapping is performed for a set of habitats or biotopes characterized by sessile epibenthic megafauna being particularly vulnerable, or sensitive to physical impact such as bottom fisheries or sedimentation. Both types of mapping are performed using species distribution modeling, which in turn is based on the relationships found between occurrence of megafauna and environmental characteristics obtained from full coverage data sets. These models use information on the characteristics and distribution of biological communities (based on video documentation) and combine it with physical characteristics of the seabed identified by terrain analysis and geological interpretation. MAREANO has refined the methods used for the spatial prediction of biotopes over the years since the first map for Tromsøflaket was produced in 2008, testing
various methods for classification and modeling. Currently prediction of biotope distribution is performed using maximum entropy distribution modeling (Maxent: Phillips et al., 2004), while modeling of vulnerable habitats has employed methods based on ‘Conditional Inference Forests’ (Hothorn et al., 2006). The workflow for biotope modeling is summarized in Fig. 3 (see also Buhl-Mortensen et al., 2009, 2015; Dolan et al., 2009; Elvenes et al., 2014; Mortensen et al., 2009). For biotope classification, Detrended Correspondence Analysis (DCA) is used to identify groups of samples (video sequences of 100–200 m length) with a high similarity of fauna composition. Clustering of these points allows the classification of biotopes. By re-plotting these samples in geographic space using GIS we are able to see the geographic distribution of biotope classified sample points. Moreover, this step allows extraction of physical characteristics for each sample point from the full coverage quantitative terrain variables described above, as well as sediment classes and landscape type. Together these provide a large number of physical seabed descriptors which potentially can serve as predictor variables which allow the model to predict from point observations to a full coverage map. To avoid using predictors that are strongly intercorrelated, or indicators for the same forcing factors, and to avoid overfitting of the
L. Buhl-Mortensen et al. / Journal of Sea Research 100 (2015) 46–61
51
Fig. 3. The general three steps used in the identification and prediction of biotopes by MAREANO. 1) Collecting information on fauna and environment for identification of fauna groups and related environmental descriptors. 2) Using the fauna environmental relation for model construction. 3. Produce biotope maps based on environmental information and model.
Fig. 4. Predicted biotopes in the MAREANO mapping areas Nordland VII and Troms II. The different biotopes are indicated with different colors and are numbered corresponding to Table 2. Modified from Elvenes et al. (2014).
52
L. Buhl-Mortensen et al. / Journal of Sea Research 100 (2015) 46–61
models, we use forward selection with Monte Carlo permutation using CANOCO for Windows 4.52 (ter Braak and Smilauer, 2002) to select the most suitable (continuous) predictor variables from all those available, while internal routines in Maxent are used to assess the importance of categorical variables. In the model the different biotopes are characterized by composition of megafauna species documented by video, substrata, depths, landscapes and terrain characteristics (Fig. 4 and Table 2). The total species diversity of the identified biotopes is described by the content of the bottom samples (Buhl-Mortensen et al., 2012a). Predictive modeling of vulnerable biotopes is presently under development and preliminary maps have been published on www.mareano.no (see Fig. 5). The vulnerable biotopes represent a modification of habitats listed by OSPAR as threatened and/or declining. Densities of colonies of the species known to contribute to these habitats are summarized for the video samples and maps are produced following a similar workflow to that used in full-coverage mapping for each separate biotope that is modeled. For a single biotope (e.g., hard bottom coral gardens), a subset of the video data set is extracted containing only samples where the habitat-defining species (e.g., Paragorgia arborea, Paramuricea sp. etc.) occur. Then this subset is further subdivided into training data and testing data, and a Conditional Inference Forest is fit using the training data (and tested using the testing data). The model is then used to predict the abundance of the habitatdefining species and these predictions are plotted in geographic space producing a continuous map layer. To simplify the output, this continuous layer, which depicts the variation in density of all the habitatdefining species across the modelled area, is then reclassified into a binary layer. In order to do this, a threshold value for density is required; pixels where the predicted density for the species are above that value will get classified as having the biotope present, whereas the pixels where the predicted density is lower than that will remain blank (or classified as absent). The threshold values were chosen taking into account two main factors: first, the distribution of observed density values from the field data, and second, the impact of different threshold values had on the total amount of area classified as present. 2.4. Methodology evolution in MAREANO The MAREANO Programme strives to provide the data required to meet Norway's marine spatial planning needs as well as being as cost-efficient as possible. The methods used by the programme have therefore been reviewed, and revised accordingly over the years. During the first MAREANO sampling cruise on Tromsøflaket video records were acquired from 1 km long video transects as described above, from 48 locations. The video records were analyzed in detail initially using sequences of 30 s long (average length 12 m), as described by BuhlMortensen et al. (2009). To standardize the sample size, the 30 s sequences were pooled into distances of 50, 200 m, and 1 km (whole transect). Following initial analyses, it was decided that the 200-m distance segments provided the most appropriate level of data for
habitat/biotope mapping. This is similar to the strategy suggested by Orpin and Kostylev (2006), who suggest that “data should be collected at the highest practical resolution but be reduced to a resolution meaningful for statistical analysis, in accordance with the total sample population”. Cumulative species curves were generated from the 48 video transects at Tromsøflaket (Fig. 6). The curves indicate that the video transects of 700 m length could be used without compromising the comparability between the locations and the standard length of the transects was reduced from 1000 to 700 m. The European standard for visual seabed surveys (EN, 16260:2012) states that “the total length of transects should be determined by the aims of the mapping”. If the aim is a representative description of species diversity of observed flora and fauna (large macrofauna and flora and megafauna), the total length of transects should be at least 500 m. MAREANO's sampling strategy has to fit the objectives of several aspects of the programme, just one of which is habitat mapping. With this in mind, a review of the sampling design is currently being carried out internally so as to optimize the number of sampling stations per unit area, while making sure that the whole range of all environmental gradients are being appropriately sampled. We are exploring two major aspects: Firstly, we are looking into how to reduce spatial autocorrelation between sampling stations, which is done by means of building rank-correlograms. Secondly, we are testing the potential use of automated algorithms (e.g. Generalized Random Tessellation strategy) for laying out sampling sites given the environmental heterogeneity of a mapping area. The methods for adapting the sampling effort based on broad-scale environmental variability are also being investigated, something that may enable MAREANO to move away from the pragmatically set guideline of 10 sampling stations per 1000 km2 and allow for more sampling in complex areas and less in heterogeneous areas. 2.4.1. Bathymetry data Access to high-quality multibeam echosounder data (bathymetry and backscatter) has been central to most of MAREANO's mapping activities, but in order to maximize the cost-effectiveness of future mapping and ensure timely delivery of scientific information, seabed mappers worldwide may increasingly need to look to existing bathymetry data as a basis for thematic maps. Elvenes et al. (2014) examines the potential of compiled single-beam bathymetry data (from the commercial company Olex) for sediment and biotope mapping. They simulate a mapping scenario where full coverage multibeam data are not available, but where existing bathymetry data sets are supplemented by limited multibeam data to provide the basis for thematic map interpretation and modeling. Results of sediment interpretation from the compiled single beam bathymetry data set suggest that production of sediment grain size distribution maps is feasible at a 1:250 000 scale or coarser, depending on the quality of available data. Biotope modeling made use of full-coverage predictor variables based on (i) multibeam data, and (ii) compiled single beam data supplemented by limited multibeam data. Using the same response variable (biotope point
Table 2 Characteristics of biotopes in the MAREANO mapping areas Nordland VII and Troms II. The biotopes are arranged in order of decreasing mean depth and their numbers correspond to the numbering in Fig. 4. Biotope # Landscape-element
Sediment
2 3 1 8 9 10 6 4 5 7
Mixed 2114 Mixed 1390 Soft 1389 Gravelly 747 Mud 290 Sandy gravel/coral reef 263 Gravelly 237 Sandy mud 221 Sandy gravel 164 Gravel 76
Lower slope/abbyssal plain Canyon/steep slope Mid slope Upper slope Shelf trough Shelf trough Shelf plain Shelf trough Bank slope Shallow bank
Mean depth (m) Slope Moderate Steep Steep Steep Level Moderate Moderate Moderate Moderate Level
Typical taxa Rhizocrinus/Bathycrinus, Elpidia, Hymenaster, Kolga, Caulophacus Chondrocladia, Lucernaria, Pycnogonida, Umbellula, Ophiopleura Nemertini pink, Actiniaria small pink, Hexactinellida bush, Lycodes sp., Bythocaris Gorgonocephalus, Crossaster, Paragorgia, Gersemia, Drifa Kophobelemnon, Stichopus, Pandalidae, Virgularia, Steletta Lophelia, Acesta, Axinella, Primnoa, Protanthea Phakellia, Craniella, Geodia, Stryphnus, Mycale Asteronyx, Funiculina, Ditrupa, Flabellum, Pteraster Pteraster, Ceramaster, Hippasteria, Sebastes, Spatangus Gorgonacea, Filograna, Tunicata white, Lithothamnion, Serpulidae
L. Buhl-Mortensen et al. / Journal of Sea Research 100 (2015) 46–61
53
Fig. 5. Predicted distribution of vulnerable biotopes in the MAREANO areas Nordland VII, Troms II and III, Tromsøflaket and Eggakanten (for geographic names see Fig. 4).
observations obtained from video data), the performance of the respective models was assessed. Biotope distribution maps based on the two data sets were visually similar, and performance statistics also indicated that there was little difference between the models, providing a comparable level of information for regional management purposes. However, while the results suggest that using compiled bathymetry data with limited multibeam is viable as a basis for regional mapping, it lacks the details for local description of biotopes and sediments. Backscatter data and the better feature resolution provided by multibeam data remain of great value for these purposes.
coverage information about the seabed “climate” as predictor variables (e.g. temperature, salinity and currents, including estimates for their variability, maximum and minimum values). The production of fullcoverage maps of oceanographic conditions is now underway, with the first biotope maps incorporating oceanographic data due for publication in 2014. These data will be an important prerequisite for the production of “harmonized” biotope maps across larger areas of the MAREANO mapping area.
2.4.2. Combining predictive biotope maps based on separate analyses The classes of video samples based on similarities in species composition are challenging to compare between areas modeled separately. As new areas are surveyed and new video samples are added to the database, biotope classes need to be revised. The current strategy in MAREANO is to perform new ordination analyses of such combined data sets, leading to new classifications. The tests of this for the MAREANO survey areas have proved useful with only small changes of clear classes identified by the prior separate ordinations (Fig. 7). However, when the geographic areas increase the “biological signals” resulting from regional “climatic” gradients become increasingly important. Across larger biogeographic regions oceanographic gradients become less or varyingly correlated with predictor variables derived from the terrain, that may have served as an adequate proxy within a certain ‘climatic setting’. Across areas spanning thousands of square kilometers spatial prediction based on backscatter, depth, terrain and sediment classes will have less explanatory ability and the model will be poor. To address this problem, the best solution is to include full-
What information is needed by management for marine spatial planning (MSP) and what implications might this have for the approach taken to seabed mapping? In accordance with the procedure for evaluation of spatially managed areas suggested by Steltzenmüller et al. (2013) the information required must correspond to management goals and operational objectives which we will argue that this in turn must have bearings on the chosen mapping strategy. Requirements of the European Marine Strategy Framework Directive (MSFD) (EC, 2008a) encompass the distribution and composition of bottom fauna on all scales as well as human pressures and related impacts. The goal for biodiversity management (MSFD, descriptor 1) is that “biological diversity is maintained. The quality and occurrence of habitats and the distribution and abundance of species are in line with prevailing physiographic, geographic and climatic conditions. Must address four ecological levels: ecosystem, landscape, habitat/community and species.” (EC, 2008a). Good environmental status (GES) is achieved if there is no further loss of the diversity of genes, species and habitats/ communities at ecologically relevant scales and when deteriorated
2.5. Results of particular relevance to area based management
54
L. Buhl-Mortensen et al. / Journal of Sea Research 100 (2015) 46–61
Fig. 6. Cumulative number of taxa versus distance along 48 video transects recorded at Tromsøflaket. Solid line indicate the mean cumulative number of taxa for the same distances for all 48 video transects. This relationship was used as a background for MAREANO to reduce the distance of each video transect from 1000 to 700 m.
components are restored to “target levels”. This agrees well with the definition of biodiversity of the Convention on Biological Diversity “the variability among living organisms ……… and the ecological complexes of which they are part; this includes diversity within species, between species and of ecosystems” (UN, 1992). Another management goal in the MSFD relates to “sea-floor integrity” stating that it should be “at a level that ensures that the structure and functions of the ecosystems are safeguarded and benthic ecosystems, in particular, are not adversely affected.” According to the MSFD this means that diversity and productivity are maintained and the uses do not cause serious adverse impacts to the natural ecosystem structure and functioning in both space and time, and recovery should be rapid and secure if a use ceases (EC, 2008a). These goals, related to biodiversity and maintenance of resilience, pose strong demands on the information needed from the mapping of marine benthos. To fully comply with the information requirements of MSFD countries would have to include the distribution of biotopes defined by community/environment relations and their natural variability to define the “target level” for GES and should provide information on the physical structure and biotic composition of the sea floor including functional groups, productivity, and diversity in space and time. The Directive requires an assessment of GES in 2012 and every six years thereafter (EC, 2008a). This would seem very ambitious, but once countries have conducted comprehensive baseline mapping (i.e. mapping that can provide the needed information to establish GES) assessment can be limited to a few representative sites and involve the status of representative indicator organisms (Gray and Elliott, 2009; Hoey et al., 2010; Katsanevakis et al., 2012). However, our knowledge of the extent, geographical range and ecological functioning of benthic habitats is still extremely poor due to limitations posed by conventional seabed survey methods. The variety of data sets produced within the MAREANO Programme provides an opportunity to generate knowledge relevant for different parts of the benthic community and therefore can be linked to a set of management goals. Information on the distribution of threatened and endangered habitats, the number of observed trawl marks on the seafloor and trawling activity provides essential information for the identification of habitats at risk and conservation plans (Fig. 8). The new visual information has been used to define new habitat and biotope entities for the varied marine areas off northern Norway. The classification of seabed locations based on multivariate analyses has provided details about taxonomic composition of “OSPAR habitats” and suggests the division of certain habitats/biotopes, such as deepsea sponge aggregations and coral gardens (Buhl-Mortensen et al.,
2013b). This information has also been used by OSPAR in the development of more relevant threatened and endangered habitat categories and to identify their health status. Several of the mapped biotopes are characterized by species typical for habitats defined by OSPAR as threatened and/or declining (OSPAR Commission, 2008). Their distribution has been used by the Norwegian Environment Agency to assess environmental value together with other valuable, vulnerable or threatened biological resources (fish, mammals, and birds). In addition the video recording has provided detailed information on physical impact from fisheries on the bottom substratum (Fig. 8). This has highlighted the need for the protection of specific vulnerable areas but also the importance of upgrading sea maps with information of these habitats e.g. the position of coral reefs. The availability of the compiled information for the public and politicians on the MAREANO website has been a success judging from the large number of visitors from many user groups and the questions from users related to content. The huge amount of information from video inspections and samples collected by MAREANO has been used in several national and international projects. In the national project for the development of a Norwegian nature type (biotope) classification system (NiN) input on marine biotopes identified by MAREANO has been important (Halvorsen et al., 2009). As a Barents Sea case study in the EU-project MESMA (Buhl-Mortensen et al., 2012b) effects of new knowledge gained from MAREANO on marine spatial management was tested using a framework for the monitoring and evaluation of spatially managed areas developed by MESMA (Steltzenmüller et al., 2013). The results from this test highlighted the importance of detailed baseline mapping and relevant indicators for informed management decision findings that have inspired the development of marine management plans in Norway. Information on the distribution impact of vulnerable habitats and signs of fisheries has provided valuable inputs to a report on fisheries effects on benthos (Buhl-Mortensen et al., 2013a), and to the EU project BENTHIS (Benthic Ecosystem Fisheries Impact Study). Furthermore, the input from MAREANO has been provided for the development of MSFD indicator in the EU project STAGES having the specific aims to improve the scientific knowledge base to support the implementation of the MSFD. 3. Mapping strategy Marine biologists know that the natural characteristics of biodiversity, which occur at a variety of scales on the same substratum, change according to the biogeographic regions as a result of oceanographic
L. Buhl-Mortensen et al. / Journal of Sea Research 100 (2015) 46–61
55
Fig. 7. Comparison of predicted biotopes in the MAREANO mapping areas from Nordland VI in the south to Eggakanten in the north. The areas Nordland VI, Nordland VII/Troms III, and Eggakanten/Tromsøflaket have been modeled separately based on multivariate community analyses of megafauna data sets for the three areas separately.
differences (e.g. Buhl-Mortensen et al., 2010; Levin et al., 2001; Pitcher et al., 2012). Useful categories of habitats/biotopes and their health status cannot be established without a thorough analysis of the relation between species distribution and all relevant environmental variables. In relation to “sea-floor integrity” (MSFD descriptor 6) there is a need for effective links to management responses, associated with particular pressures (or multiple pressures). This requires knowledge of the category and the magnitude of pressure that each human activity poses on a specific biotope and species, and their sensitivity to different magnitudes of specific pressures and their accumulated effect (Steltzenmüller et al., 2013). Unless the typical species composition is known for specific environments in defined bioregions it will not be possible to detect any response to pressures with the necessary power. Physical damage may be documented by various techniques (e.g. observations of tracks in the sediment or damaged organisms in situ, or the proportion of damaged organisms in bottom samples), and guidelines for this should be outlined. Expected natural levels for indicators (e.g. diversity index, biotope composition or species density) will vary between habitats/biotopes and bioregions. In this perspective there is no way around a thorough baseline mapping of marine benthos with the management of biodiversity and sea-floor integrity in mind. The workload and the cost of baseline mapping involved will be dictated by the spatial and temporal variability of the environment and community in the management area. For example, in intertidal areas natural variability often means the presence of many different communities that succeed each other, or occur in patches, across the same environmental background. In subtidal areas where the seabed is subject to high energy from waves or strong currents very few species or habitat types might be present. As depth increases, so does environmental stability, and changes in community composition over time will be relatively small (Buhl-Mortensen et al., 2010; Carney, 2005; Levin et al., 2001). What then is the relevant mapping strategy that satisfies today's major management goals for the marine benthic ecosystem? In the sections that follow we review potential mapping options
with focus on subtidal mapping of benthos on various substrates at landscape and habitat/biotope scale. 4. Review of habitat mapping methodology The specific methods used to derive habitat maps from seabed mapping data vary considerably. A recent review by Brown et al. (2011) suggests that studies generally can be categorized into one of three over-arching strategies: Abiotic surrogate mapping; Assemble first, predict later, top down (unsupervised classification); and Predict first, assemble later, bottom up (supervised classification) (see Fig. 9). Abiotic surrogates are based on unsupervised classification and limited or no ground validation and would not be relevant for biodiversity management. Environmental variables are often used as indirect surrogates for mapping biodiversity because species survey data are scant. However, environmental variables that are measured on arbitrary scales are unlikely to have simple, direct relationships with biological patterns. Instead, biodiversity may respond nonlinearly to the interactions of environmental variables. In a recent study by Pitcher et al. (2012) the role of the environment in the driving patterns of biodiversity composition in large marine regions was investigated in Australian waters. They found that important predictors (e.g. depth, salinity, temperature, sediment composition and current) differed among the regions and biota. The shapes of responses along gradients also differed and were nonlinear, often with thresholds indicative of step changes in composition. Nevertheless, the need for background information in relation to MSP has initiated a large scale compilation of existing abiotic seafloor information with EUSeaMap (Mapping European seabed habitats) representing a large scale EU initiative. It is arguable if the resulting large scale seabed habitat maps produced are relevant to the management goals of not loosing habitats and species and can be indicative of areas in need of protection. We suggest that a more biologically relevant
56
L. Buhl-Mortensen et al. / Journal of Sea Research 100 (2015) 46–61
Fig. 8. Information gained from mapping that is of special relevance to area based management. From top left to bottom right: trawling activity from VMS data, distribution of vulnerable habitats and number of trawl marks observed per 100 m of seabed documented with video.
approach is needed to integrate quantitative and continuous biological responses into mapping based on environmental data. The second strategy enforces a top-down approach, whereby the environmental data and in situ biological and geological data are analyzed before they are combined (an unsupervised classification strategy). Species presence can then be extrapolated based on where there is geographical concurrence between the data sets. Community mapping is done in a comparable way, where the ground-truthing data are usually organized into classes using (multivariate) statistical methods. Habitat (biotope) classes are then compared against the segmented environmental data set in a comparable way to the single species analysis. In this way the classes can be extrapolated on the basis of the segmented environmental data. This strategy can involve use of sediment categories as entities for fauna where information is lacking. However, there is a substantial risk involved in using abiotic proxies for biological entities where information is lacking (Stevensen and Connoly, 2004). The third strategy, predict first, assemble later (supervised classification) adopts a bottom up approach whereby the in situ biological and geological ground-truthing data are used to inform the organization and segmentation of the environmental data (a supervised classification strategy), and can be applied from a single species or a community stand point. The discrimination between second and third strategies is often not as clear-cut as suggested by Brown et al. (2011). The methodology used by MAREANO (see section on MAREANO as an example) and in the habitat mapping strategy described by Pitcher et al. (2012) and Ellis et al. (2012) seems to land between these two strategies. The background for full coverage habitat/biotope maps are models based on in situ biological and geological ground-truthing and the community/environment relation observed. Pitcher et al. (2012)
found that the regional-scale environmental variables predicted an average of 13–35% of the variation in species abundance distributions and the results indicate a limited scope for extrapolating bio-physical relationships beyond the region of source data sets. The comparable prediction strength is 76% in local areas mapped by MAREANO. 5. Review of sampling options A concern of every mapping programme should be to maximize efficiency in sampling, as this is a very costly activity. To be efficient, a sampling design needs to (1) define explicitly the spatial and temporal domains of expression of the processes under study (e.g., the relationships between environmental factors and species distribution); (2) determine that the spatial and temporal resolutions of the sampling design are able to capture the process under study; and (3) ensure that the spatial and statistical analyses are appropriate for the data type. These three steps interact with each other and must be considered as a whole before going out to do any sampling (Fortin and Dale, 2005). A sampling design can be defined on the basis of its extent (which is the extent of the study region), the lag (the distance between the samples) and the grain (size of the each sampling unit) (Legendre and Legendre, 1998). While the extent might be fixed, the lag can be modified by sampling areas with more or less samples per unit area. Grain can be modified only for certain sampling gear types, for example, for video data. Besides species–area relationships, the size of each sampling unit (e.g., a transect) must be related with the minimum mapping unit, which should be equal to the size of the smallest homogeneous area at a given scale. At most scales, heterogeneity is readily detected in the spatial distribution of organisms, meaning that the probability of occurrence
L. Buhl-Mortensen et al. / Journal of Sea Research 100 (2015) 46–61
57
Fig. 9. Three basic strategies for the production of benthic habitat maps. From Ferrier and Guisan (2006) and Brown et al. (2011).
of the species under study is not evenly distributed. The spatial structure in species distributions can be caused by biotic processes (e.g., predation and competition), the spatial variation of environmental drivers such as sediment composition, temperature, and historical events (Legendre and Fortin, 1989; Segurado et al., 2006). The environmental factors structuring the communities and biotopes operate simultaneously at different spatial scales. This can be exemplified with the hydrodynamic forces which at a broad scale may affect the transport of larvae, at an intermediate scale influence the food concentration, and at a small scale local topography may induce current patterns controlling the distribution of suitable settlement substrates (Doyle, 1975; Pineda, 2000). Fauna mobility and size affect their “ambit” (action range, see Jumars, 1975) and thus their response to habitat heterogeneity at a different scale, the ambit of a species may change during their life history. The species with larger ambit encounter species and structures in closer approximation to their proportion and thus their environment is coarser ‘grained’ (MacArthur and Wilson, 1967) compared with species with small ambit. Thus, fauna components vary in their response to habitat diversity due to their “ambit” (e.g. Buhl-Mortensen, et al., 2010, 2012a; Klitgaard-Kristensen and Buhl-Mortensen, 1999). In addition, the physical nature of the habitat will dictate what gears can be used and thus how well the benthic community can be documented. Furthermore, faunal groups do not respond identically to the same pressure or parameter (e.g. trawling pressure, temperature change), so different ‘indicator organisms’ and different multivariate indices of community health/condition will likely exhibit different responses. Broad-scale habitat or biotope classification is a useful background for selecting subareas for more detailed study to reveal finer spatial patterns of biology, surface geology and topography. More in-depth analyses of the seabed substrates and their associated fauna together
with a wider set of environmental data (currents, bottom temperature, surface primary production, etc.) may reveal clearer patterns that can better define marine landscape elements. Several studies have shown that thorough analyses of video results with a finer spatial scale combined with information from multibeam bathymetry enables the prediction of habitats at a finer scale with full areal coverage (Dolan et al., 2009; Gonzalez-Mirelis and Lindegarth, 2012; Holmes et al., 2008; Ierodiaconou et al., 2007, 2011; Mortensen et al., 2009; Rattray et al., 2009). Such analyses are more suitable for providing background for management decisions, and represent one fundamental outcome from the MAREANO Mapping Programme. The concept of marine landscapes is a broad-scale classification of the marine environment. It is a broad term used in different description systems both at an ecosystem-level and in reference to purely geomorphological characterizations. The concept of marine landscape was first developed for Canadian waters by Roff and Taylor (2000). They developed a classification system based on environmental factors such as water temperature, depth/light penetration, substratum type, exposure and slope. They termed the classes ‘seascapes’. Currently, the term ‘marine landscapes’ is commonly used (see Golding et al., 2004). This level represents an intermediate scale between regional seas and habitats. A marine landscape should have consistent physical and ecological characteristics and provide a practical scale related to the management of human activities such as fishing and hydrocarbon exploration. MAREANO uses a geomorphological classification based on bathymetry to describe landscapes. This classification is part of the Norwegian classification system, NiN (Nature-types in Norway) (Halvorsen et al., 2009). Some examples of entities at different levels and scale are: large marine ecosystems (Norwegian Sea, Barents Sea); marine landscapes in MAREANO (fjord, strandflat, continental shelf plain, smooth continental slope, marine canyon, marine valley,
58
L. Buhl-Mortensen et al. / Journal of Sea Research 100 (2015) 46–61
and deep sea plain (Fig. 10)); vulnerable biotopes in MAREANO (coral reefs, soft bottom sponge communities, hard bottom sponge communities, glass sponge populations, sea pen bottom, Umbellula populations, soft-bottom coral forests, and hard-bottom coral forests (Fig. 5)). Fauna composition within smaller spatial units can be used for the classification at the biotope level as described above. Commonly, video observations are analyzed from the playback of video on land after the cruises. Different bottom types have different inhabitants, and are not equally well sampled with the same equipment. Smaller infauna species (benthic organisms b 2 cm living in the sediment) dominate the muddy substratum and are easily documented using grab, box corer or multicorer. However mixed sediments are common within the various landscapes and biotopes found on the continental shelves and slopes. As the content of larger grains and gravel increases, or higher clay contents consolidate the sediment it becomes increasingly difficult to retrieve grab samples. As the hardness of the substratum increases epifauna (organisms living on top of the sediment) becomes more dominant. The epifauna provides stronger taxonomical signatures and gives more background for defining communities and biotopes than infauna. Independent of sediment composition different parts of the bottom fauna requires specific sampling gear to be properly documented. Large organisms (megafauna; N 2 cm) are mainly sampled using beam trawl and densities and patchiness is only available from the video surveys. Small and mobile crustaceans (hyperfauna) can only be documented using epibenthic sled (Bergman et al., 2009). Thus, if we want to provide a good and reliable picture of GES related to biodiversity and seafloor integrity many approaches for documenting bottom fauna is required if we are not to lose information on species and functionality of the bottom community. In order to survey long-lived and large benthic organisms visual inspection is crucial. Video-transects along the seabed documents the distribution of bottom types and megafauna, and can provide information of human impact (e.g. litter and trawl tracks). Grab samples normally cover 0.1 m2 and document the occurrence of smaller organisms typically dominated by polychaets living in soft sediment. Epibenthic sled hauls cover ~ 300–400 m2 and provides occurrence and composition of the crustacean-fauna living in the uppermost part of the sediment or swimming just above the bottom (e.g. shrimps, mysids, amphipods) while beam trawl covers 500–800 m2 and collects macro- and megafauna (N 2 cm) living in the upper part of the sediment. There are of course sampling biases associated with these gears, because it is not possible to sample with an epibenthic sled in rugged terrain and thus hyperbenthos is not well documented in this kind of habitat. An inherent problem with the data on the three fauna groups is that it is based on different sampling gears that, even though they are normally used in studies of these fauna groups, do not discriminate between epifauna and infauna. In addition the amount of bottom habitat sampled differs substantially between a trawl haul and a grab sample. These are obstacles that are hard to overcome. In their study of the role of environmental variables in shaping seabed composition Pitcher et al. (2012) found that, for all regions, the trawl-sampled species were predicted better than those sampled with smaller devices. This can be explained by a better match between the resolution of environmental descriptors and the large area sampled by trawls.
environmental variation while having a sound statistical basis are to be preferred. For example Clements et al. (2009) advocated the use of optimal allocation analysis whereby the number of samples is calculated taking into account costs, as well as variability of the environment. While this can help determine the number of samples required to map an area, it does not provide any information that can be used to determine the distribution of the sampling stations, or the size (and potentially shape) of each single sample. The methods such as rankcorrelograms based on spatial autocorrelation, semivariograms, fractal dimension and hierarchical analysis of variance can be used to determine the size of patches of epibenthic megafauna and the distances at which samples are independent (Underwood and Chapman, 1996), or in other words, to detect spatial structure. This reflects the scale at which organisms interact with the environment, and with one another. Gonzalez-Mirelis et al. (2009) measured fine-scale patterns of spatial structure in assemblages of epibenthic sublittoral megafauna off the Swedish west-coast using spatial rank correlograms. They revealed the presence of patches b20 m in all survey sites. The authors concluded that any area of the seafloor approximately 38 m across was likely to be sufficiently homogeneous so as to consider the variation within it largely as noise or unnecessary spatial detail, in the regional context. They also indicate that for epibenthic sublittoral megafauna samples of 20–80 m would efficiently capture the relationships between environmental factors and biota, and appropriately describe the system. Much smaller samples (at least one order of magnitude difference) will be more affected by stochastic processes, and much larger samples are likely to smooth out faunal discontinuities and depict, instead, regional trends. Effective mapping and monitoring of benthic communities including habitat-forming organisms and their associated fauna require that the standard methodology (grab, 0.1 m2) useful only for small sized macrobenthos in a muddy environment is supplemented and/or replaced by gears that can provide information from more complex bottom sediments and for fauna groups that are mobile and/or larger. Hard, topographically complex habitats are known from visual inspection to have biotic compositions that differ from sedimentary habitats (Kostylev et al., 2001; Pitcher et al., 2007). To assess the abundance and health status of long lived and habitat forming organisms high resolution visual documentation is needed. It is not enough to know that the habitat forming organism is present. Its health status and the associated fauna must also be investigated. Large and long lived organisms that are often particularly vulnerable to human activities may prove to be a good indicator of the status of the other organisms in the same habitat. 6.2. Seabed integrity Habitat maps provide the baseline for monitoring the integrity of the seabed. As long as the habitat definitions cover whole regions appropriately, this indicator can be implemented at a regional scale. Whether this indicator is operational or do not depend on the accuracy of the habitat definitions, for biogenic habitats without clear boundaries, this represents a true problem. For example, for coral gardens it can be difficult to outline the boundary of the habitat because quantitative data on colony density (colonies per area) is needed for habitat definition. For reef forming organisms like Lophelia pertusa the outline of the reef can in many cases be identified from the mulitibeam echosounder data.
6. Discussion
6.3. Single-taxon indicators for environmental status of habitats/biotopes
6.1. Sample size and sampling design
There is a great potential for video surveys to provide information on size-frequency distribution of bivalve or other sensitive/indicator. This information on sessile mega-fauna can be informative for some human pressures however, size-frequency distributions depend on more than human pressures. The mean sizes of Mytilus for instance,
There is increased interest among the benthic habitat mapping community in the optimization of sampling effort and increasing cost effectiveness. The methods which capture as much as possible the
L. Buhl-Mortensen et al. / Journal of Sea Research 100 (2015) 46–61
59
Fig. 10. Examples of MAREANO maps showing seabed classification at different levels from the area offshore Lofoten and Vesterålen. A) Marine landscapes showing broad scale geomorphic features following the definitions from Nature types in Norway (NiN) and described by Elvenes (2013). B) Sediment grain size map (1:100000 scale). C) Predicted biotope distribution (50 m raster map based on Maxent modeling of ordination classes from video analysis of benthic megafauna pooled into 200 m sequences).
will vary with natural environmental gradients, such as salinity. It is therefore crucial that baseline values are defined within relevant environmental regimes. There is no fixed colony density that represents the threshold value for a coral garden habitat (as well as for many other biogenic habitats). It is therefore important that the threshold values used in each case are reported. In cases where direct evidence of physical disturbance (breakage) of colonies are not clear, the abundance and size class distribution of colonial organisms (e.g. corals and sponges) can be monitored and used as an indicator of environmental status. 7. Conclusion • There is a lack of knowledge on marine habitats/biotopes and management entities (e.g. biotopes/habitats) on all levels need to be identified, thus reliable abiotic proxies to these has not been established for most areas. Furthermore there is not one unified scale, pattern or hierarchical system that fits marine benthic communities in general. • Comprehensive mapping is important to fill knowledge gaps. Technology has improved and mapping of bathymetry, sediment properties and visual documentation of sediment and larger fauna is getting easier. • Experience from MAREANO shows that the bottom up approach using megafauna defined biotopes as the main management entity is promising. To document: diversity, production, and functionality of all benthic fauna groups biotopes are sampled with specialized gears. • Video surveys of megafauna in the identified biotopes are likely sufficient to indicate if changes have occurred and if further sampling
is needed to document changes on all fauna levels. The focus on easily observed larger fauna is also relevant for increased knowledge of pressure specific response and resilience of the benthos.
Acknowledgments The MAREANO Programme is supported by the Norwegian Ministry for the Environment and the Ministry of Trade, Industry and Fisheries. Thanks to Sigrid Elvenes (NGU) for assistance with preparation of some of the figures. Thanks to reviewers whose comments have helped improve this manuscript. References Bellec, V.K., Dolan, M.F.J., Bøe, R., Thorsnes, T., Rise, L., Buhl-Mortensen, L., Buhl-Mortensen, P., 2009. Sediment distribution and seabed processes in the Troms II area — offshore North Norway. Nor. J. Geol. 89, 29–40. Bergman, M.J.N., Birchenough, S.N.R., Borja, Á., Boyd, S.E., Brown, C.J., Buhl-Mortensen, L., Callaway, R., Connor, D.W., Cooper, K.M., Davieas, J., De Boois, I., Gilkinson, K.D., Gordon Jr., D.C., Hillewaert, H., Kautski, H., Kluyver, M., De, Kröncke, I., Limpenny, D.S., Meadows, W.J., Parra, S., Pennington, S.E., Rachor, E., Rees, H.L., Reiss, H., Rumohr, H., Schratzberger, M., Smith, S., Tunberg, B.G., Van Dalfsen, J.A., Ware, S., Watling, L., 2009. Guidelines for the study of the epibenthos of subtidal environments. ICES Techniques in Marine Environmental Sciences. No. 42 (88 pp.). Boitsov, S., Petrova, V., Jensen, H.K.B., Kursheva, A., Litvinenko, I., Chen, Y., Klungsøyr, J., 2011. Petroleum-related hydrocarbons in deep and subsurface sediments from South-western Barents Sea. Mar. Environ. Res. 71, 357–368 (2011). Boitsov, S., Petrova, V., Jensen, H.K.B., Kusheva, A., Litvinenko, I., Klungsøyr, J., 2013. Sources of polycyclic hydrocarbons in marine sediments from southern and northern areas of the Norwegian continental shelf. Mar. Environ. Res. 87–88, 73–84.
60
L. Buhl-Mortensen et al. / Journal of Sea Research 100 (2015) 46–61
Bradshaw, G., Fortin, M.-J., 2000. Landscape heterogeneity effects on scaling and monitoring large areas using remote sensing data. Geogr. Inf. Sci. 6, 61–68. Brown, C.J., Smith, S.J., Lawton, P., Anderson, J.T., 2011. Benthic habitat mapping: a review of progress towards improved understanding of the spatial ecology of the seafloor using acoustic techniques. Estuar. Coast. Shelf Sci. 92, 502–520. Buhl-Jensen, L., 1986. The benthic amphipod fauna of the west-Norwegian continental shelf compared with the fauna of five adjacent fjords. Sarsia 71, 193–208. Buhl-Mortensen, P., Dolan, M., Buhl-Mortensen, L., 2009. Prediction of benthic biotopes on a Norwegian offshore bank using a combination of multivariate analysis and GIS classification. ICES J. Mar. Sci. 2009. http://dx.doi.org/10.1093/icesjms/fsp200. Buhl-Mortensen, L., Vanreusel, A., Gooday, A.J., Levin, L.A., Priede, I.G., Buhl-Mortensen, P., Gheerardyn, H., King, N.J., Raes, M., 2010. Biological structures as a source of habitat heterogeneity and biodiversity on the deep ocean margins. Mar. Ecol. 31 (1), 21–50. Buhl-Mortensen, L., Buhl-Mortensen, P., Dolan, M.F.J., Dannheim, J., Bellec, V., Holte, B., 2012a. Habitat complexity and bottom fauna composition at different scales on the continental shelf and slope of northern Norway. Hydrobiologia 685, 191–219. Buhl-Mortensen, L., Olsen, E., Røttingen, I., Buhl-Mortensen, P., Hoel, A.H., Lid Ringheim, S., Grøsvik, B.E., Nygaard Holen, S., Gonzalez-Mirelis, G., Skulstad, E.M., 2012b. Application of the MESMA Framework. Case Study: The Barents Sea, MESMA Report (138 pp.). Buhl-Mortensen, L., Aglen, A., Breen, M., Buhl-Mortensen, P., Ervik, A., Husa, V., Løkkeborg, S., Røttingen, I., Stockhausen, H.H., 2013a. Impacts of fisheries and aquaculture on sediments and benthic fauna: suggestions for new management approaches. Fisken og Havet 2 (69 pp.). Buhl-Mortensen, P., Buhl-Mortensen, L., Guillaumont, G., Skjoldal, H.R., 2013b. Deepwater megafauna habitats — report from a workshop on the improvement of the definitions of habitats on the OSPAR List, Bergen: 20–21 October 2011. Report for OSPAR (22 pp.). Buhl-Mortensen, L., Buhl-Mortensen, P., Dolan, M.F.J., Holte, B., 2015. The MAREANO programme – a full coverage mapping of the Norwegian off-shore benthic environment and fauna. J. Marine Research. 1, 4–17. Carney, R.S., 2005. Zoonation of deep biota on continental margins. Oceanogr. Mar. Biol. Annu. Rev. 43, 211–278. Clements, A.J., Strong, J.A., Flanagan, C., Service, M., 2009. Objective stratification and sampling-effort allocation of ground-truthing in benthic-mapping surveys. ICES J. Mar. Sci. 67, 628–637. Costello, M.J., 2009. Distinguishing marine habitat classification concepts for ecological data management. Mar. Ecol. Prog. Ser. 397, 253–268. Davies, C.E., Moss, D., Hill, M.O., 2004. EUNIS Habitat Classification Revised 2004. Report to the European Topic Centre on Nature Protection and Biodiversity, European Environment Agency (307 pp. (available online at http://eunis.eea.eu.int/eunis/ habitats.jsp)). Dolan, M.F.J., Buhl-Mortensen, P., Thorsnes, T., Buhl-Mortensen, L, Bellec, V.K., Bøe, R., 2009. Developing seabed nature-type maps offshore Norway: initial results from the MAREANO programme. Nor. J. Geol. Vol. 89, 17–28 1 & 2, side. Dolan, M.F.J., Thorsnes, T., Leth, J., Alhamdani, Z., Guinan, J., Van Lancker, V., 2012. Terrain characterization from bathymetry data at various resolutions in European waters — experiences and recommendations. NGU Report 2012.045. Geological Survey of Norway, Trondheim, Norway (44 pp.). Doyle, R.W., 1975. Settlement of planktonic larvae: a theory of habitat selection in varying environments. Am. Nat. 109, 113–126. EC, 2007. Data Completeness, quality and coherence. Habitats Directive Article 17 Technical Report (2001-2006) http://bd.eionet.europa.eu/activities/Reporting/Article_17/ Reports_2007/index_html. EC, 2008a. Directive of the European Parliament and the Council establishing a framework for community action in the field of marine environmental policy (Marine Strategy Framework Directive). European Commission. Directive 2008/56/ EC, OJ L 164. EC, 2008b. Roadmap for Maritime Spatial Planning: Achieving Common Principles in the EU. COM (2008) 791 Final. EEC, 1992. Habitats Directive 92/43/EEC, OJ L 206. Ellis, N., Smith, S.J., Pitcher, C.R., 2012. Gradient forests: calculating importance gradients on physical predictors. Ecology 93, 156–168 (http://www.esajournals.org/doi/abs/10. 1890/11-0252.1). Elvenes, S., 2013. Landscape mapping in MAREANO. NGU Report Number 2013.035. Geological Survey of Norway, Trondheim. Elvenes, S., Dolan, M.F.J., Buhl-Mortensen, P., Bellec, V.K., 2014. An evaluation of compiled single-beam bathymetry data as a basis for regional sediment and biotope mapping. ICES J. Mar. Sci. 71 (4), 867–881. EN 16260:2012, 2012. Water Quality — Visual Seabed Surveys Using Remotely Operated and/or Towed Observation Gear for Collection of Environmental Data. epbrs, 2013. Recommendations of the Meeting of the European Platform for Biodiversity Research Strategy Held under the Irish Presidency of the EU Dublin, Ireland, 15–17th of May, 2013. Ferrier, S., Guisan, A., 2006. Spatial modelling of biodiversity at the community level. J. Appl. Ecol. 43, 393–404. Fortin, M.-J., Dale, M., 2005. Spatial Analysis: A Guide for Ecologists. Cambridge University Press. Golding, N., Vincent, M.A., Connor, D.W., 2004. Irish Sea Pilot — Report on the Development of a Marine Landscape Classification for the Irish Sea. (30 pp. JNCC. (www.jncc.gov.uk/irishseapilot)). Gonzalez-Mirelis, G., Lindegarth, M., 2012. Predicting the distribution of out-of-reach biotopes with decision trees in a Swedish marine protected area. Ecol. Appl. 22 (8), 2248–2264. Gonzalez-Mirelis, G., Bergström, P., Lundälv, T., Jonsson, L., Lindegarth, M., 2009. Mapping the benthos: spatial patterns of seabed-dwelling megafauna in a Swedish Fjord, as derived from opportunistic video data. Mar. Biodivers. 39, 291–302.
Gray, J.S., Elliott, M., 2009. Ecology of Marine Sediments: From Science to Management. Oxford University Press, UK (225 pp.). Greene, H.G., Yoklavich, M.M., Starr, R.M., O'Connell, V.M., Wakefield, W.W., Sullivan, D.E., McRea Jr., J.E., Cailliet, G.M., 1999. A classification scheme for deep seafloor habitats. Oceanol. Acta 22, 663–678. Halvorsen, R., Andersen, T., Blom, H.H., Elvebakk, A., Elven, R., Erikstad, L., Gaarder, G., Moen, A., Mortensen, P.B., Norderhaug, A., Nygaard, K., Thorsnes, T., Ødegaard, F., 2009. Naturtyper i Norge — Teoretisk grunnlag, prinsipper for inndeling og definisjoner. Naturtyper i Norge versjon 1.0 Artikkel 1, pp. 1–210. Hoey, G.V., Borja, A., Birchenough, S., Buhl-Mortensen, L., Degraer, S., Fleischer, D., Kerckhof, F., Magni, P., Muxika, I., Reiss, H., Schröder, A., Zettler, M., 2010. The use of benthic indicators in Europe: from the Water Framework Directive to the Marine Strategy Framework Directive. Mar. Pollut. Bull. 60 (12), 2187–2196. Holmes, K., Niel, K.V., Radford, B., Kendrick, G., Grove, S., 2008. Modelling distribution of marine benthos from hydroacoustics and underwater video. Cont. Shelf Res. 28, 1800–1810. Hothorn, T., Hornik, K., Zeileis, A., 2006. Unbiased recursive partitioning: a conditional inference framework. J. Comput. Graph. Stat. 15, 651–674. ICES, 2005. Report of the Working Group on Habitat Mapping (WGMHM), 5–8 April. ICES CM 2005/E:05, Bremerhaven, Germany (87 pp.). Ierodiaconou, D., Burq, S., Reston, M., Laurenson, L., 2007. Marine benthic habitat mapping using multibeam data, georeferenced video and image classification techniques in Victoria, Australia. J. Spat. Sci. 52, 93–104. Ierodiaconou, D., Monk, J., Rattray, A., Laurenson, L., Versace, V., 2011. Comparison of automated classification techniques for predicting benthic biological communities using hydroacoustics and video observations. Cont. Shelf Res. 31 (Supplement 1), S28–S38. Jumars, P.A., 1975. Environmental grain and polychaete species diversity in a bathyal benthic community. Mar. Biol. 30, 253–266. Katsanevakis, S., Weber, A., Pipitone, C., Leopold, M., Cronin, M., Scheidat, M., Doyle, T.K., Buhl-Mortensen, L., Buhl-Mortensen, P., D'Anna, G., de Boois, I., Dalpadado, P., Damalas, D., Fiorentino, F., Garofalo, G., Giacalone, V.M., Hawley, K.L., Issaris, Y., Jansen, J., Knight, C.M., Knittweis, L., Kröncke, I., Mirto, S., Muxika, I., Reiss, H., Skjoldal, H.-R., Vöge, S., 2012. Monitoring marine populations and communities: review of methods and tools dealing with imperfect detectability. Aquat. Biol. 16, 31–52. Klitgaard-Kristensen, D., Buhl-Mortensen, L., 1999. Benthic foraminifera along an offshore-fjord gradient: a comparison with amphipods and mollusks. J. Nat. Hist. 33, 317–350. Kostylev, V., Todd, B., Fader, G., Courtney, R., Cameron, G., Pickrill, R., 2001. Benthic habitat mapping on the Scotian Shelf based on multibeam bathymetry, surficial geology and seafloor photographs. Mar. Ecol. Prog. Ser. 219, 121–137. Legendre, P., Fortin, M.-J., 1989. Spatial pattern and ecological analysis. Vegetatio 80, 107–138. Legendre, P., Legendre, L., 1998. Numerical Ecology. 2nd English edition. Elsevier. Levin, L.A., Etter, R.J., Rex, M.A., Gooday, A.J., Smith, C.R., Pineda, J., Stuart, C.T., Hessler, R.R., Pawson, D., 2001. Environmental influences on regional deep-sea species diversity. Annu. Rev. Ecol. Syst. 132, 51–93. MacArthur, R.H., Wilson, E.O., 1967. The Theory of Island Biogeography. Princeton University Press, Princeton (224 pp.). MESH, 2008. Mapping European Seabed Habitats. www.searchmesh.net. Mortensen, P.B., Buhl-Mortensen, L., 2005. Deep-water corals and their habitats in the gully, a submarine canyon off Atlantic Canada. In: In Freiwald, A., Roberts, J.M. (Eds.), Cold-water Corals and Ecosystems. Springer-Verlag, Berlin, pp. 227–257. Mortensen, P., Buhl-Mortensen, L., Dolan, M., Dannheim, J., Kröger, K., 2009. Megafaunal diversity associated with marine landscapes of northern Norway: a preliminary assessment. Nor. J. Geol. 89 (1 & 2), 163–171. Orpin, A.R., Kostylev, V.E., 2006. Towards a statistically valid method of textural sea floor characterization of benthic habitats. Mar. Geol. 225, 209–222. OSPAR Commission, 2008. Case Reports for the OSPAR List of Threatened and/or Declining Species and Habitats (261 pages). Phillips, S.J., Dudı´k, M., Schapire, R.E., 2004. A Maximum Entropy Approach to Species Distribution Modeling. Proceedings of the 21st International Conference on Machine Learning. Banff, Alberta, Canada, pp. 655–662. Pineda, J., 2000. Linking larval settlement to larval transport: assumptions, potentials, and pitfalls. Oceanography of the Eastern Pacific I, pp. 84–105 (2000). Pitcher, C.R., Doherty, P.J., Arnold, P., Hooper, J.N.A., Gribble, N., Bartlett, C., et al., 2007. Seabed biodiversity on the continental shelf of the Great Barrier Reef World Heritage Area. AIMS⁄CSIRO⁄QM⁄QDPI Final Report to CRC Reef Research (320 pp.). Pitcher, C.R., Lawton, P., Ellis, N., Smith, S.J., Incze, L.S., Wei, C.-L., Greenlaw, M.E., Wolff, N.H., Sameoto, J., Snelgrove, P.V.R., 2012. Exploring the role of environmental variables in shaping patterns of seabed biodiversity composition in regional-scale ecosystems. J. Appl. Ecol. 49, 670–679. Rattray, A., Ierodiaconou, D., Laurenson, L., Burq, S., Reston, M., 2009. Hydro-acoustic remote sensing of benthic biological communities on the shallow southeast Australian continental shelf. Estuar. Coast. Shelf Sci. 84, 237–245. Roff, J.C., Taylor, M.E., 2000. Viewpoint; national frameworks for marine conservation a hierarchical geophysical approach. Aquat. Conserv. Mar. Freshwat. Ecosyst. 10, 209–223. Segurado, P., Araújo, M.B., Kunin, W.B., 2006. Consequences of spatial autocorrelation for niched-based models. J. Appl. Ecol. 43, 433–444. Steltzenmüller, W., Breen, P., Stamford, T., Thomsen, F., Badalamenti, F., Borja, A., BuhlMortensen, L., Carlstøm, J., D'Anna, G., Dankers, N., Degraer, D., Dujin, M., Fiorentino, F., Galparsoro, I., Giakoumi, S., Gristina, M., Johnson, K., Jones, P.J.S.,
L. Buhl-Mortensen et al. / Journal of Sea Research 100 (2015) 46–61 Katsanevakis, S., Knittweism, L., Kyriazi, Z., Pipitone, C., Piwowarczyk, J., Rabaut, M., Sørensen, T.K., van Dalfsen, J., Vassilopoulou, V., Vega Fernández, T., Vincx, M., Voge, S., Weber, A., Wijkmark, Jak.R., Qiu, W., Hofstede, R., 2013. Monitoring and evaluating of spatially managed areas: a generic framework for implementation of ecosystem based marine management and its application. Mar. Pollut. Bull. 37, 149–164. Stevensen, T., Connoly, R.M., 2004. Testing the utility of abiotic surrogates for marine habitat mapping at scales relevant to management. Biol. Conserv. 119, 351–362. ter Braak, C.J.F., Smilauer, P., 2002. CANOCO Reference Manual and CanoDraw for Windows User's Guide: Software for Canonical Community Ordination (Version 4.5). Microcomputer Power, Ithaca, NY, USA.
61
UN, 1992. Convention on Biological Diversity, June 5,1992, art. 1, 31 I.L.M. 818. Underwood, A.J., Chapman, M.G., 1996. Scales of spatial patterns of distribution of intertidal invertebrates. Oecologia 107, 212–224. Wilson, M.F.J., O'Connell, B., Brown, C., Guinan, J.C., Grehan, A.J., 2007. Multiscale terrain analysis of multibeam bathymetry data for habitat mapping on the continental slope. Mar. Geod. 30, 3–35. Wright, D., Heyman, W., 2008. Introduction to the special issue: marine and coastal GIS for geomorphology, habitat mapping, and marine reserves. Mar. Geod. 31, 223–230.