Ampleur des perturbations physiques causées aux habitats prédominants et spéciaux

The analysis has four components:

• A composite habitat map showing the extent and distribution of habitats (based on observational and modelled data), including the mapped extent of any relevant features (e.g. records and distribution of particular species and biotopes such as the European Nature Information System (EUNIS) Level 5 habitats or other biological characteristics). For this assessment a biotope is defined as the combination of an abiotic habitat and its associated community of species (Connor et al., 2004). All habitat data were combined at EUNIS level 3.
• Tables relating benthic habitat type to habitat sensitivity scores based on their resistance and resilience (i.e. recoverability) (Tillin et al., 2010; BioConsult, 2013; Tillin and Tyler-Walters, 2014). Sensitivity is assessed at species, biotope and EUNIS Level 3, depending on what habitat mapping information is available within each grid cell (c-square).
• Distribution and intensity of pressures causing physical damage. This analysis focused on surface and sub-surface abrasion caused by bottom trawling (for fishing from vessels >12 m only) within 0.05° grid cells (c-squares) (JNCC, 2011; ICES, 2015; Church et al., 2016).
• Distribution of levels of disturbance per habitat type per year. Calculation of disturbance is based on the intensity and duration of pressures and habitat sensitivity per pressure type. The pressures of abrasion (non-fisheries), siltation and selective extraction are not currently included in the assessment, but will be incorporated in future developments of the Indicator.

Data concerning the four components are combined using a step-wise approach to calculate the total area of different levels of disturbance, and a combination of those, across the region, per habitat type. The results are used to calculate variability in fishing intensity and trends in disturbance per year and for a six-year period.

The spatial assessment of this indicator is presented at EUNIS level 3, and has been prepared by combining the sensitivity and pressure data from habitats, biotopes and species within the EUNIS Level 3 habitat polygons. For this assessment the OSPAR regions have been sub-divided following biogeographic boundaries (OSPAR, 2016) into: Southern North Sea and Northern North Sea Southern Celtic Sea and Northern Celtic Sea English Channel and Bay of Biscay / Iberian Coast. For The Wider Atlantic, only a partial assessment has been possible owing to insufficient habitat and sensitivity data for the deep seafloor areas.

The methodology is based on a series of analytical steps that combine the distribution and intensity of physical damage pressures with the distribution and range of habitats and their sensitivities.

Step 1: Establishing Habitat Extent

An important component of this indicator is the production of a composite habitat map showing the extent and distribution of predominant and special habitats and their associated sensitivities. This map is produced using a combination of benthic survey data and modelled habitat maps (Figure b). As a basis for the assessment, a full coverage EUNIS Level 3 habitat map has been produced for the OSPAR Maritime Area, integrating maps from surveys and broad-scale models. The new EUNIS classification has not been used because it is still under development therefore version 2007-11 is used in this assessment.

The EUNIS habitats are mapped at different levels of detail, from Level 3 physical habitats to Level 6 biological communities and then aggregated to EUNIS Level 3 where more detail is provided. The majority of the habitat maps were obtained from the European Marine Observation and Data Network (EMODnet) Seafloor Habitats portal, including the broad-scale physical habitat map, EUSeaMap (EMODnet, 2010, 2016), and more detailed habitat maps created from survey data available through EMODnet, or as part of OSPAR habitat data calls. The information on the coverage and type of data were taken into account for the calculation of confidence maps (Step 6). The specification for a habitat map for the assessment of the Extent of Physical Damage indicator included the following conditions:

1. To contain information on the relevant EUNIS habitat/biotope type at any level between Levels 3 and 6;
2. To refer data on biotopes to Level 3 of the EUNIS habitat classification system;
3. To use the broad-scale modelled map, EUSeaMap at EUNIS Level 3 when higher resolution maps from surveys are not available;
4. To use the best available evidence on habitat data;
5. To cover the greatest possible area of the OSPAR Maritime Area; and
6. To contain no overlaps.

Mapping rules were established in order to decide objectively which of the overlapping datasets would be the sole occupant in the overlapping area. Where a EUNIS habitat map developed from survey data overlapped with a broad-scale predictive habitat map, a threshold confidence score of 58% was used as a simple rule for deciding whether to favour the habitat map from survey data. This threshold was based on the MESH protocol (Mapping European Seabed Habitats) (EMODnet, 2010). Within the MESH scoring system, for any map to have a score greater than 58%, the survey techniques must have used a combination of remote sensing and ground-truthing to derive the habitat types, hence physical and biological elements are included for its production. Therefore, 58% was deemed to be the lower threshold at which an overlapping survey map is considered to be of higher quality than the broad-scale predictive map. Pre-processing conditions and rules for the combining of data are available in the CEMP Guidelines (OSPAR, in prep).

Step 2: Assessing Habitat Sensitivity

The sensitivity of benthic habitats is determined based on a combination of the resilience (recoverability) and resistance (tolerance) of key structural, functional and characterising species of the habitat in relation to a defined intensity of each pressure (Tillin et al., 2010; BioConsult, 2013; Tillin and Tyler-Walters, 2014). Due to data limitations, the sensitivity scores are defined using a categorical scoring approach (Tillin et al., 2010). Sensitivity assessments for ecological groups have also been undertaken using Bray-Curtis cluster similarity analysis and Multidimensional Scaling, where resistance and resilience scores are assigned to groups of species with similar biological traits (e.g. burrowers) (Tillin and Tyler Walters, 2014).

The sensitivity map is created with three stages, using best available evidence where present (Figure e and Figure f):

• Stage one. Species records from survey data, that match a list of species assigned to a specific ecological group, are mapped using their maximum sensitivity value (based on the combination of resilience and resistance). Data are plotted as the intersection between the habitat polygon and a 0.05° grid.
• Stage two. If there is a high enough density of species recorded and agreement between the species and the underlying habitat they are within, then the sensitivity from the same species records used in stage one are used to assign a modal sensitivity to the surrounding habitat polygon. For example, if data coverage is sufficient, and the substrate and habitat type in the surrounding polygons are the same, then the same sensitivities are applied to these areas.
• Stage three. In order to act as a background map and to fill in areas not covered by the data collected from the first two stages, the habitat map outlined in Step 1 is used to assign EUNIS Level 3 benthic habitat sensitivities to the whole area. The sensitivities used in this step are often a range (from very low sensitivity to very high), in which case the maximum sensitivity is selected.

The maps are then combined geographically to show the highest confidence information across all regions.

Step 3: Assessing the Extent and Distribution of Physical Damage Pressures

For annual assessments the first task is to determine the relevant human activities causing physical pressures and their spatial and temporal extent. Bottom trawling is known to be affecting a large area of the seafloor (Dinmore et al., 2003; Eastwood et al., 2007; Foden et al., 2010, 2011; JNCC, 2011; Jennings et al., 2012) so the assessment method currently focuses on the corresponding pressures: surface abrasion (damage to seafloor surface features) and subsurface abrasion (penetration and / or disturbance of the substrate below the surface of the seafloor). Aggregated Vessel Monitoring System (VMS) fishing data are used to calculate the ‘swept area’ of a specific group of fishing gears (at the level of metier if this is available). Metier is a group of fishing operations targeting a similar (assemblage of) species, using similar gear, during the same period of the year and / or within the same area and which are characterised by a similar exploitation pattern.

The swept area is calculated using the parts of the fishing gear in contact with the seabed and is calculated on the width of fishing gear (in metres) multiplied by the average vessel speed (in knots) and the time fished. This calculation is undertaken on a cell-by-cell (grids or c-squares) basis per gear and per year, using data covering the six-year period 2010–2015 (see ICES Working Group Spatial Fisheries Data and CEMP Guidelines for Extent of Physical Damage to Predominant and Special Habitats (OSPAR, in prep) for the detailed method). Only the part of gear in contact with the seafloor is used for the analysis, and as a result all the gears have been classified according to type and/or metier group (Eigaard et al., 2015; Church et al., 2016). The swept area ratio (proportion of cell area swept per year; SAR) is then calculated by dividing the swept area by the grid cell area. The trawling effort is classified with an intensity scale ranging from ‘none’ to ‘very high’ (cell area swept more than 300% or three times per year). Separate Geographical Information System (GIS) layers are produced for surface abrasion and sub-surface abrasion. See Figure 1, Figure c, Figure d.

Assessments across a cycle (six years). To differentiate between cells with low and high SAR variability an analysis of variance is undertaken using the surface and sub-surface yearly results. This analysis allows the distinction between areas where fishing intensity seems to be consistent or at similar levels across years, and areas where fishing intensity levels fluctuates. A grid cell was considered to be ‘variable’ when the variance analysis showed a change in three or more categories of the classified SAR across all years.

In order to produce a layer showing the aggregated surface and sub-surface pressures that takes into account the variations in fishing pressures across years, a generic rule was used:

• For cells with low variability (i.e. under similar fishing pressure year on year) the mean of SAR across all years is calculated
• For cells with high variability (i.e. fishing pressure variable) the highest SAR value across all years is selected to define the pressure category because it represent the maximum level of exposure within the cycle. In this case, the maximum values, instead of mean were chosen to capture the highest level of damage caused by the initial impact of the first pass, which is regarded to be most damaging for benthic habitats. Due to the short timescales, it is not anticipated that the habitats will have recovered at the end of the assessment period, hence the approach captures the maximum level of damage per grid across all years. A change of three categories or more on the fishing pressure in the same grid across years is considered to be highly variable.

Step 4: Combining Pressure Intensity and Habitat Sensitivity

The degree of disturbance of a habitat is a prediction based on the spatial and temporal overlap of its sensitivity and exposure to a specific pressure. Sensitivity and pressure are combined via a matrix, producing ten categories of disturbance (0–9, where 0 is no disturbance and 9 is the maximum disturbance possible). The matrix was created from the result of previous studies that looked at the impacts of pressures on sensitive species and habitats when applied at different levels of intensities (Schroeder et al., 2008; BioConsult, 2013). The matrix is used to calculate the disturbance for each surface and sub-surface abrasion per year. These disturbance maps are then combined by selecting, for each of the cells, the highest disturbance category from the two pressure layers (Figure 2,Figure g,Figure m).

At present, other activities that could cause physical damage pressures, are not included. It is anticipated that due to the different nature of the pressures ‘selective extraction’, ‘abrasion’ and ‘changes in siltation’, separate disturbance matrices or algorithms will be required which will take into account the spatial distribution of pressures and the temporal effects. This information is not currently available and will be included in the next assessment cycle for an overall calculation of disturbance caused by physical damage.

Step 5: Disturbance Aggregation Method and Trend Analysis

Disturbance values across years are combined using the aggregated fishing pressure spatial layers, developed in Step 3. Results are used to calculate trend between years in those grid cells or c-squares identified as variable. This allows the variation of disturbance across years per habitat type to be assessed. The trend analyses are simple plots over the six-year period, rather than a linear regression, which was not possible due to the small number of years assessed (Figure l).

The disturbance categories were aggregated into two groups: disturbance categories 0–4 (representing lower levels of disturbance) and disturbance categories 5–9 (representing higher levels of disturbance). This scale is the result of a combination of the temporal variation of fishing activity and the range of sensitivity values.

Step 6: Confidence assessments 

These results show current distribution and extent of habitat sensitivity; the overlapping fishing pressure causing surface and subsurface abrasion; and the resulting habitat disturbance. These results allow the distinction to be made between seafloor habitats of varying sensitivity that are under pressure from these types of fishing activities.

Some areas under high levels of surface and especially sub-surface abrasion, show low habitat disturbance. This could be caused by sensitive features being replaced by opportunistic and less sensitive species.

At present there are limitations due to data availability and accessibility for assessing habitat extent and distribution and associated habitat sensitivity. Combining observational data with model results is a highly effective approach in the absence of detailed survey data.

Some areas have already lost sensitive species and biotopes due to past human activities, such occurrences cannot be assessed by this Indicator and this will result in a lower disturbance score in such areas. To address this, the results of applying this Indicator will be validated using the assessment results from other benthic condition indicators, such as the assessments on Condition of Benthic Habitat Defining Communities, including the development of reference conditions, as they become available.

The method used has been thoroughly tested and reviewed by national experts working within the OSPAR framework. It represents a realistic approach for assessing the distribution of physical disturbance across the OSPAR regions based on current knowledge and using all available evidence. However, it is important to note that the strength of any assessment is dependent on the quality of the input data and the approach to its analysis, and this will in turn dictate the power and utility of the resulting information. As it stands the method proposed gives a partial assessment of physical damage at an OSPAR regional scale focused only on physical abrasion as a result of bottom fishing.

It is also expected that disturbance matrices and the final algorithm will be calibrated in the future and, if required, modified using the results from site-scale condition indicators, not only for fishery information but also for other human activities. Use of this indicator in conjunction with other site condition indicators such as the Condition of Benthic Habitat Defining Communities indicator, which is based on the European Union Water Framework Directive (WFD) (2000/60/EC) multimetric indicators, as well as the typical species will help to improve the confidence and accuracy of the results on the distribution and levels of impacts. 

Although the indicator is particularly useful for assessing large sea areas where few data exist, it would benefit from access to more accurate, abundant and high‑quality data to improve confidence in the results. Key gaps are:

Data access:

Additional information from benthic species collected during surveys can be added to improve the quality of the results, and to improve the evidence available on the distribution of sensitivity information across the regions. This alongside access to the aggregated fisheries Vessel Monitoring System (VMS) data, especially data on metiers, being collected as part of the ongoing data calls is important to improve the accuracy of results without incurring additional costs.

Quantified impacts values:

Development of quantifiable pressure-impact relationship values are very important in order to incorporate information on other activities causing physical damage, and to assess additive effects in multi-pressure systems. Scale of fishing pressure and matrix underpinning disturbance values need to be evaluated using experimental and field studies to improve evidence on the pressure-impact-response relationships. Results from the assessment of the Condition of Benthic Habitat Communities will be used to validate and calibrate impacts calculated under this assessment.

Sensitivity and disturbance data:

The evidence base can be improved if more information on the resilience and resistance of biotopes, or habitats at European Nature Information System (EUNIS) Levels 4 to 6 is made available and / or further studies are undertaken. Sensitivity information used in the production of the surface abrasion and sub-surface abrasion sensitivity layers was developed based on studies, literature reviews and matrices from the United Kingdom, France and Germany, it is therefore mainly relevant for the Greater North Sea and the Celtic Seas. This gap is of particular importance for the Bay of Biscay and Iberian Coast region as it is expected that the characterising species and biotopes will be slightly different from those in the Greater North Sea and Celtic Seas. The confidence of the assessment results will benefit from an increase in the availability of more accurate habitat data. As part of the European Union co-funded EcApRHA project, new information has been tested to increase the qualitative aspects of the indicator, especially for the analyses of sensitivities and disturbance, using case studies in The Bay of Biscay and Iberian Coast. These results, alongside data and information generated by other European Union projects have been incorporated within the methodology for the next assessment cycle. A review of the sensitivity method should also be undertaken; in particular, the relevance of incorporating recoverability aspect into the assessment and when it may be more appropriate to factor it into the temporal trend analysis and to management actions (such as priorities for habitats with very long recovery times).

Calculation of an index per region and habitat type:

The final outputs of this model are levels of disturbance per habitat type across a region.These levels of disturbance are amalgamated into a ‘Physical Damage Index‘ value for each benthic habitat. However, during indicator development a formula for calculating this Index was tested and found not to deliver satisfactory results to capture temporal changes in fisheries pressures. The formula and results are not presented here as they are still under development and further testing, which will include the revision of sensitivity/pressure matrix to incorporate new data from fishing impact curves and condition indicators. Further understanding of the cumulative effects of historical damage from some human activities and in particular the effects that the continued loss of sensitive habitats could cause on the integrity and ecological function of benthic habitats needs to be investigated to allow the definition and development of reference values. Physical damage by human activities other than fisheries, will be calculated for the next assessment cycle.

Effects of other gear types:

Better understanding is needed of the effects of differing and innovative gear types (for example, SumWing trawls), which may result in a lower abrasion pressure.

Other OSPAR Regions:

At present this indicator is still being investigated for its application in Arctic Waters and the Wider Atlantic, mainly due to limited data availability within these regions. It is anticipated that data on habitat distribution and sensitivity will be included at a later stage, to allow the analysis to address the wider spatial scale and assess variations in disturbance across OSPAR regions more accurately.