Physical loss - Marine online assessment tool

Up to 2016, 2% of the potential habitat for seagrass beds (Zostera marina), and 0.5% of the potential habitat for horse mussel reefs (Modiolus modiolus) has been lost because of activities that have taken place. Therefore, the UK target has not been met.

Background

UK target on potential physical loss of predicted seafloor habitats

This indicator is used to assess progress against the target set for biogenic seafloor habitats in the Marine Strategy Part One (HM Government, 2012), which requires the area of selected habitat to be stable or increasing and not smaller than the baseline value.

Key pressures and impacts

Aquaculture, navigational dredging, dredge and spoil disposal, coastal development, and recreational activity are the main causes of physical loss to horse mussel reefs (Modiolus modiolus) and seagrass beds (Zostera marina). Pressure from these activities results in direct substratum loss, smothering by suspended sediment, and cause changes to the characteristics of the habitat. This assessment focuses on horse mussel reefs and seagrass beds. Both habitats are very fragile and vulnerable to pressures from human activities. These habitats are an important food source, refuge, and nursery ground for many species.

Measures taken to address the impacts

Horse mussel beds and seagrass beds are protected under national and international legislation and conventions and the target is consistent with this protection. Activities can be managed to reduce potential physical loss through the following measures:

marine planning and marine licensing (marine spatial planning)
environment, strategic and habitat regulations assessments
the Common Fisheries Policy
measures to protect Threatened and Declining OSPAR habitats
the establishment of an ecologically coherent network of Marine Protected Areas

Monitoring, assessment and regional co-operation

Areas that have been assessed

Due to data limitations with regards to the actual location and extent of these biogenic habitats, the assessment has been carried out on modelled data at the Marine Strategy Framework Directive UK level.

Monitoring and assessment methods

This indicator measures the potential physical loss caused by human activities from locations where horse mussel reefs and seagrass beds are predicted to be. Data from current known horse mussel and seagrass distributions (presence and absence) were used to model and select areas of potentially suitable habitat elsewhere in the UK (Figures 1 and 2).

Figure 1. The predicted distribution of potential habitat for seagrass beds (Zostera marina) in UK waters.

Figure 2. The predicted distribution of potential horse mussel reef (Modiolus modiolus) in UK waters.

For each habitat, the human activities that are most likely to cause potential physical loss were identified based on existing available evidence. Data on the spatial location and duration of these activities during 2010 to 2016 were used to create a physical footprint relevant to each habitat. These footprints show the spatial variation in potential physical loss, ranging in value from 0 to 1, where a potential physical loss value of 1 is equal to the complete loss of the habitat. The area of potential physical loss includes the contribution from (1) the physical substitution of the substratum and (2) modification of the physico-chemical parameters beyond the ranges that sustain the habitat/typifying species for a minimum of 12 years. It was not always possible to restrict activity data to the period 2010 to 2016. The inclusion of historical information on human activity means that the potential physical loss calculated for each habitat will represent historical loss up to 2016.

The extent of overlap between the predicted distribution of each habitat and the relevant physical footprint is used to estimate the amount of habitat that has potentially been lost due to human activity prior to 2016.

Assessment thresholds

The thresholds are based on the assessment target above.

Regional co-operation

It has been proposed that this indicator could be used at the OSPAR scale in future assessments.

Assessment method

Matrices from the Marine Evidence-based Sensitivity Assessment of the Marine Life Information Network were used to identify the pressures from human activities causing an impact on each of the habitats. For each pressure identified, potential physical loss was considered to have occurred if the habitat was in the ‘high’ or ‘very high’ sensitivity category. Expert judgement was used to remove pressures that are believed not to cause irreversible loss of habitat within a 12-year period (potential physical loss is understood as a permanent change to the seabed which has lasted or is expected to last for a period of two reporting cycles (12 years) or more (European Commission, 2017)). The assessment by the Marine Life Information Network does not align pressures with human activities, so the Joint Nature Conservation Committee Pressures-Activities Database was used to identify the activities contributing to each potential physical loss pressure.

Spatial data was obtained for each activity, but it was not always possible to date every activity because temporal information was not available in each case. When it was possible to determine the date and duration of activities, only those occurring between 2010 and 2016 have been used. Those activities without a clear start date of the operations were used jointly for the calculation of the physical footprint. These activities were kept separate in the final calculations of the analysis to allow for the distinction from other activities which have a defined start date. As such, the resulting assessment of potential physical loss should be considered as a comparison between a baseline (based on the beginning of the reporting period, in this case, 2010) and a point up to assessment date.

The activity data were then converted into physical footprint and near-field footprint strata. Each stratum required a potential physical loss factor, a value that states the predicted amount of potential physical loss within a stratum (representing the extent response). The potential physical loss factors ranged from 0 to 1 where 0 indicates no potential physical loss within the stratum, and 1 equals complete loss of habitat. The potential physical loss factors are specific to each habitat being considered and were obtained either from (i) the scientific literature, (ii) calculations based on known dimensions of physical footprint and near-field footprint pressures for the activity, or (iii) estimated using expert judgement, potential physical loss factors were also used to convert activities, represented as site polygons, into an actual footprint of the activity inducing potential physical loss. Activities were also allocated into two temporal groups, namely those that are specific to 2010 to 2016 and those that include historical activity predating 2010. The contribution of each activity group to the overall decline could, therefore, be calculated.

It was not possible to represent the pressure generated by inshore fishing vessels (less than 12 meters in length) that are not covered by Vessel Monitoring Systems, as at present standardised data showing the distribution and intensity of inshore fishing vessels across the UK are not available. This also included activities such as scallop dredging which is known to cause the loss of biogenic habitats, horse mussel reefs in particular.

Species distribution modelling was used to establish the extent and distribution of potential habitat supporting seagrass beds and horse mussel reefs. The ensemble modelling method combined the use of a regression-based method (Generalised Additive Modelling) and computer-learning method (Random Forest) to predict the potential distribution of suitable habitat for both seagrass beds (approximately 1,500 km² of potential habitat) and horse mussel reefs (approximately 8,000 km² of potential habitat) within UK territorial waters. This distribution reflects the potential biogenic habitat using data from known horse mussel reef and seagrass bed locations as well as absence point data (individual horse mussel specimen records were not included as input reef locations) and was used here as a proxy for real occupied habitat. The presence-absence data was split 3:1 for model calibration and validation respectively.

The species distribution models used the regression-based Generalised Additive Modelling and Random Forest fitting tools available within the Marine Geospatial Ecology Tools (v 0.8) module for the Environmental Systems Research Institute ArcMap. The output of the ensemble approach was used to weight activity strata by the presence of the habitat of interest within each stratum. The overall performance of both ensemble models was considered to be good and overall accuracy was high.

The proposed method of stratified extrapolation to assess potential physical loss enables assessment of regional change in habitat extent in the absence of existing coverage. Calculations within the stratified extrapolation framework take account of (i) the area of each stratum within the regional assessment area, (ii) the potential physical loss factors associated with each stratum, and (iii) the area of potential habitat within each stratum. The stratified extrapolation framework reports (i) values of potential physical loss for the feature of interest within a specified assessment area; and (ii) a confidence assessment associated with the estimate of potential physical loss (Figure 3). The regional assessment areas are the assessments units used to calculate the spatial overlaps between the pressure/activity data and the habitat modelled outputs.

Figure 3. Overview of the method used to implement the stratified extrapolation framework and assess physical potential loss for potential habitat supporting seagrass (Zostera marina) beds and horse mussel (Modiolus modiolus) reefs.

Additional information can be found in Morris (2015) and in Strong (2015).

Results

Findings from the 2012 UK Initial Assessment

This indicator was not available as part of the Initial Assessment (HM Government, 2012).

Latest findings

Status assessment

According to the model, 2% of the potential habitat for seagrass beds (Zostera marina) in the UK has been lost because of activities taking place up to 2016, equating to 32 km²of lost habitat. A larger area of horse mussel reefs (Modiolus modiolus) has been lost because of human activities, totalling 39km², but this represents only 0.5% of potential habitat for horse mussel in UK waters. The assessment results show that the indicator target has not been achieved during this reporting period.

The main causes of potential physical loss on seagrass beds are aquaculture, navigational dredging, dredge and spoil disposal, and coastal development. Navigational dredging, dredge and spoil disposal, aquaculture, and recreational activity contributed the most to potential physical loss of horse mussel reefs (an example of the overlap between some activities and horse mussel habitats is shown in Figure 4).

Figure 4. Example areas of overlap between the predicted distribution of horse mussel reefs (Modiolus modiolus) and human activities that cause potential physical loss.

Due to the lack of available data, it was not possible to include all relevant activities in the calculations of potential physical loss. Of particular importance is the absence of spatial data on use of bottom contact gears by small fishing vessels with bottom contact gears. This absence of data may mean that the extent of habitat loss reported above has been underestimated. Confidence levels in the status assessment values, therefore, are low.

Although other pressures are known to impact on these habitats, and are currently of greater concern, many are reversible (within 12 years) or damaging and as such were not in the scope of the analysis of potential physical loss (no recovery within 12 years).

Trend assessment

The trend is unknown, the indicator was not considered as part of the 2012 Initial Assessment HM Government, 2012).

Further information

The stratified extrapolation framework was able to incorporate activity data, potential physical loss factors (obtained from expert judgement and the scientific literature) and habitat information to extrapolate potential physical loss values for the regional assessment area. The activity and habitat data required for this process are widely available and regularly updated.

Identification of pressures generating potential physical loss

The Marine Evidence-Based Sensitivity Assessment matrices suggested 6 pressures were relevant for seagrass beds (Zostera marina) and 13 pressures for horse mussel reefs (Modiolus modiolus). However, expert judgement was used to limit these to the pressures associated with potential physical loss (‘selected’ pressures). As such, substratum loss and smothering were selected for seagrass beds whilst substratum loss, smothering and changes in the water flow rate were selected for horse mussel reefs.

Identification of activities generating pressures

Activities generating the selected pressures were identified using the Joint Nature Conservation Committee Activity/Pressures Database and are shown in Table 1. Activities that were (i) unlikely to occur with the specific habitat, (ii) highly unlikely to occur at an intensity sufficient to generate potential physical loss, and/or (iii) highly transient, were excluded from the analysis (labelled as shortlisted activities).

Table 1. Activities selected to represent pressures generating potential physical loss. Activities associated with a high or very high sensitivity but not considered to act upon the potential habitat or be irreversible within 12 years are termed “shortlisted” and were excluded.

Activity	Seagrass *(Zostera marina)*	Horse mussel (Modiolus modiolus)
Aquaculture - finfish, macroalgae, and shellfish	Selected	Selected
Artificial reefs and other environmental structures	Shortlisted	Shortlisted
Coastal defence & land claim protection¹	Selected	Selected
Coastal docks, ports, and marinas	Selected	Selected
Coastal tourist sites	Shortlisted	Shortlisted
Cultural & heritage sites/structures	Selected	Selected
Dredge & spoil disposal	Selected	Selected
Extraction – navigational dredging	Selected	Selected
Extraction – rock/ mineral (coastal quarrying)
Extraction – sand, and gravel (aggregates)	Selected	Selected
Extraction – water (abstraction)		Shortlisted
Extraction of genetic resources	Shortlisted	Shortlisted
Fishing		Selected
Gas storage operations	Shortlisted	Shortlisted
Marine hydrocarbon extraction	Selected	Selected
Marine research activities		Shortlisted
Military activities		Shortlisted
Power stations - thermal effluent and nuclear discharge
Recreational activities (anchorages only)	Selected	Selected
Renewable energy - tidal (not including cables)	Selected	Selected
Renewable energy – wave (not including cables)	Selected	Selected
Renewable energy – wind (not including cables)	Selected	Selected
Sewage disposal
Shipping – general (at sea)		Shortlisted
Shipping – port operations		Shortlisted
Submarine cable operations	Selected	Selected
Submarine pipeline operations	Selected	Selected

Extrapolating potential physical loss for the potential supporting habitat of seagrass (Zostera marina) beds and horse mussel reefs (Modiolus modiolus)

The 13 activities inducing potential physical loss in seagrass habitat were allocated to 24 strata (Table 2). Similarly, 15 activities, relevant for potential physical loss in horse mussel habitat, were allocated to 27 strata (Table 3). The footprint, buffer values and the potential physical loss factors for the strata are also provided in Tables 2 and 3.

Table 2. potential physical loss factors and buffer values used for activity footprints and the adjustment of activities provided as site polygons for activities relevant for seagrass beds.

Stratum	Activity	Physical Footprint (PFP) / Near-Field Footprint (NFFP)	Buffer (m)	potential physical loss factor
1	Coastal defence & land claim protection	PFP	0	1.00
2	Coastal docks, ports & marinas	PFP	0	1.00
3	Aquaculture	PFP	0	1.00
4	Extraction – navigational dredging (capital & maintenance)	PFP	0	0.50
5	Extraction – sand, and gravel (active area only)	PFP	0	1.00
6	Dredge & spoil disposal	PFP	0	0.50
7	Cultural & heritage sites/structures (wrecks)	PFP	17.55	1.00
8	Marine hydrocarbon extraction (not including pipelines)	PFP	20.71	1.00
9	Submarine pipeline operations	PFP	0.36	1.00
10	Submarine cable operations (communications and power)	PFP	0.03/0.11	0.04
11	Renewable energy – wave (not including cables)	PFP	0	0.0009
12	Renewable energy – wind (not including cables)	PFP	0	0.0001
13	Recreational activities (anchorages only)	PFP	20	0.02
14	Coastal defence & land claim protection	NFFP	9.69	1.00
15	Coastal docks, ports & marinas	NFFP	9.69	1.00
16	Aquaculture	NFFP	20	1.00
17	Extraction – navigational dredging (capital & maintenance)	NFFP	1100	0.05
18	Extraction – sand, and gravel (active area only)	NFFP	1100	0.05
19	Dredge & spoil disposal	NFFP	1100	0.05
20	Marine hydrocarbon extraction (not including pipelines)	NFFP	9.69	1.00
21	Submarine pipeline operations	NFFP	0.36	1.00
22	Submarine cable operations (communications and power)	NFFP	0.03/0.11	0.03
23	Renewable energy – wave (not including cables)	NFFP	0	0.05
24	Renewable energy – wind (not including cables)	NFFP	0	0.002

Table 3. potential physical loss factors and buffer values used for activity footprints and the adjustment of activities provided as site polygons relevant for horse mussel reefs.

Stratum	Activity	Physical Footprint (PFP) / Near-Field Footprint (NFFP)	Buffer (m)	potential physical loss factor
1	Coastal defence & land claim protection	PFP	0	1.00
2	Coastal docks, ports & marinas	PFP	0	1.00
3	Fishing	PFP	0	1.00
4	Aquaculture	PFP	0	1.00
5	Extraction – sand, and gravel (active area only)	PFP	0	1.00
6	Extraction – navigational dredging (capital & maintenance)	PFP	0	0.50
7	Dredge & spoil disposal	PFP	0	0.50
8	Cultural & heritage sites/structures (wrecks)	PFP	17.55	1.00
9	Renewable energy – wave (not including cables)	PFP	0	0.0009
10	Renewable energy – wind (not including cables)	PFP	0	0.001
11	Renewable energy - tidal (not including cables)	PFP	0	0.0001
12	Marine hydrocarbon extraction (not including pipelines)	PFP	20.71	1.00
13	Recreational activities (anchorages only)	PFP	20	0.02
14	Submarine pipeline operations	PFP	0.36	1.00
15	Submarine cable operations (communications and power)	PFP	0.03/0.11	0.03
16	Coastal defence & land claim protection	NFFP	9.69	1.00
17	Coastal docks, ports & marinas	NFFP	9.69	1.00
18	Aquaculture	NFFP	1100	1.00
19	Extraction – sand, and gravel (active area only)	NFFP	1100	0.05
20	Extraction – navigational dredging (capital & maintenance)	NFFP	1100	0.05
21	Dredge & spoil disposal	NFFP	1100	0.05
22	Renewable energy – wave (not including cables)	NFFP	0	0.05
23	Renewable energy – wind (not including cables)	NFFP	0	0.002
24	Renewable energy - tidal (not including cables)	NFFP	0	0.0034
25	Marine hydrocarbon extraction (not including pipelines)	NFFP	9.69	1.00
26	Submarine pipeline operations	NFFP	0.36	1.00
27	Submarine cable operations (communications and power)	NFFP	0.03/0.11	0.03

Species distribution modelling

The predicted distribution of potential habitat supporting seagrass beds and horse mussel reefs, based on an ensemble approach, are shown in Figures 1 and 2. The results of this validation indicate high levels of sensitivity, specificity, and accuracy, as well as a good level of model agreement (kappa) for both features (Table 4). The ensemble distribution was used as the ultimate representation of the potential habitat and was used to clip those areas overlapping with activity strata.

Table 4. Validation of the ensemble model used to predict the distribution of potential habitat for seagrass and horse mussel. Model performance indices are also provided for the Generalised Additive Modelling (GAM) and Random Forest models that contributed to the ensemble.

Inputs	*Seagrass* *(Zostera marina)*	Horse mussel (Modiolus modiolus)
Calibration points/ alidation points (presence points)	278 / 75	312 / 201
Adjusted R-squared for GAM	0.995	0.896
Generalized (Approximate) Cross Validation for GAM	0.000	0.053
‘Our of Bag’ (OOB) estimate of error rate for Random Forest	0.73%	0.33%
True Positive Rate, known as Sensitivity	97.1%	99.0%
True Negative Rate, known as Specificity	88.5%	98.9%
Accuracy	89.3%	98.9%
False Positive Rate	11.5%	1.1%
False Negative Rate	2.9%	1.0%
Kappa	0.51	0.78

Weighting activity strata by the predicted distribution: seagrass beds

The distribution produced by the ensemble approach was used to weight activity strata. Table 5 provides the area of overlap between potential habitat of seagrass (Zostera marina) and each activity stratum. These areas were then adjusted according to the potential physical loss factor. Strata with small potential physical loss factors are typically associated with activities that have been provided as site polygons. Before potential physical loss factor adjustment, the activities with the greatest area of overlap were navigational dredging, sand and gravel extraction, and aquaculture.

Table 5. Area of overlap between the habitat extent for seagrass and activity strata.

Stratum	Activity	Physical Footprint (PFP) / Near-Field Footprint (NFFP)	Stratum area overlapping with predicted distribution (km²)	Stratum area adjusted by potential physical loss factor for the overlap area (km²)
1	Coastal defence and land claim protection	PFP	0.59	0.63
2	Coastal docks, ports and marinas	PFP	1.39	1.39
3	Aquaculture	PFP	12.79	12.79
4	Extraction – navigational dredging (capital and maintenance)	PFP	15.35	7.67
5	Extraction – sand, and gravel (active area only)	PFP	2.07	2.07
6	Dredge and spoil disposal	PFP	5.13	2.57
7	Cultural & heritage sites/structures (wrecks)	PFP	0.10	0.10
8	Marine hydrocarbon extraction (not including pipelines)	PFP	0.00	0.00
9	Submarine pipeline operations	PFP	0.00	0.00
10	Submarine cable operations (communications and power)	PFP	0.02	0.00
11	Renewable energy – wave (not including cables)	PFP	0.25	0.00
12	Renewable energy – wind (not including cables)	PFP	3.85	0.00
13	Recreational activities (anchorages only)	PFP	7.04	0.15
14	Coastal defence and land claim protection	NFFP	0.63	0.63
15	Coastal docks, ports and marinas	NFFP	0.13	0.13
16	Aquaculture	NFFP	1.35	1.35
17	Extraction – navigational dredging (capital and maintenance)	NFFP	31.89	1.43
18	Extraction – sand, and gravel (active area only)	NFFP	27.79	1.25
19	Dredge and spoil disposal	NFFP	1.33	0.06
20	Marine hydrocarbon extraction (not including pipelines)	NFFP	0.00	0.00
21	Submarine pipeline operations	NFFP	0.00	0.00
22	Submarine cable operations (communications and power)	NFFP	2.47	0.08
23	Renewable energy – wave (not including cables)	NFFP	0.25	0.01
24	Renewable energy – wind (not including cables)	NFFP	3.85	0.01

Weighting activity strata by the predicted distribution: horse mussel reefs

The distribution produced by the ensemble approach was used to weight activity strata. Table 6 provides the area of overlap between potential habitat of horse mussel reefs and each activity stratum. These areas were then adjusted according to the potential physical loss factor. Figure 4 provides example images of the overlap between the predicted distributions of both habitats with common anthropogenic activities inducing potential physical loss.

It should be noted that there are limitations with regards to the input data layers in shallow waters, which means that the accuracy of the model is reduced in many coastal areas. An upgrade of substrate information is planned in the near future, alongside additional model improvements should increase the confidence of the model outputs.

Table 6. Area of overlap between the potential habitat suitable for horse mussel reefs and activity strata.

Stratum	Activity	Physical Footprint (PFP) / Near-Field Footprint (NFFP)	Stratum area (km²)	potential physical loss factor adjusted area (km²)
1	Coastal defence and land claim protection	PFP	0.02	0.02
2	Coastal docks, ports & marinas	PFP	0.09	0.09
3	Fishing	PFP	0.00	0.00
4	Aquaculture	PFP	7.45	7.45
5	Extraction – sand and gravel (active area only)	PFP	0.00	0.00
6	Extraction – navigational dredging (capital & maintenance)	PFP	0.00	0.00
7	Dredge and spoil disposal	PFP	59.60	29.80
8	Cultural and heritage sites/structures (wrecks)	PFP	0.13	0.13
9	Renewable energy – wave (not including cables)	PFP	0.00	0.00
10	Renewable energy – wind (not including cables)	PFP	9.80	0.01
11	Renewable energy - tidal (not including cables)	PFP	5.19	0.00
12	Marine hydrocarbon extraction (not including pipelines)	PFP	0.00	0.00
13	Recreational activities (anchorages only)	PFP	0.40	0.01
14	Submarine pipeline operations	PFP	0.00	0.00
15	Submarine cable operations (communications and power)	PFP	0.04	0.00
16	Coastal defence and land claim protection	NFFP	0.00	0.00
17	Coastal docks, ports and marinas	NFFP	0.01	0.01
18	Aquaculture	NFFP	1.43	1.43
19	Extraction – sand and gravel (active area only)	NFFP	0.00	0.00
20	Extraction – navigational dredging (capital and maintenance)	NFFP	0.00	0.00
21	Dredge and spoil disposal	NFFP	2.81	0.13
22	Renewable energy – wave (not including cables)	NFFP	0.00	0.00
23	Renewable energy – wind (not including cables)	NFFP	9.80	0.02
24	Renewable energy - tidal (not including cables)	NFFP	5.19	0.02
25	Marine hydrocarbon extraction (not including pipelines)	NFFP	0.00	0.00
26	Submarine pipeline operations	NFFP	0.00	0.00
27	Submarine cable operations (communications and power)	NFFP	6.13	0.19

Summarising potential physical loss for predicted distribution of seagrass (Zostera marina) and horse mussel (Modiolus modiolus) within the regional assessment area

The ensemble model predicted a total area of 1,583 km² of seagrass beds (potential supporting habitat) and 8,304 km² of potential habitat suitable for horse mussel reefs within the regional assessment area. The conclusion of the Stratified Extrapolation framework suggests that 2% (32 km²) of the potential habitat of seagrass has been lost over the period covered by the activity data (partly historical to 2016) and 0.5% (39 km²) for horse mussel reefs (Table 7). The confidence associated with these values is low to moderate. The Stratified Extrapolation framework suggests that aquaculture, navigational dredging, sand and gravel extraction, dredge and spoil disposal, and coastal development were the main sources of potential physical loss, in order of severity, for seagrass. For horse mussel reef, dredge and spoil disposal, aquaculture, submarine cable operations, wrecks, and coastal development were the main sources of potential physical loss, in order of severity.

Table 7. potential physical loss summary for modelled seagrass beds and horse mussel reefs in UK territorial waters (up to 2016). The top 5 activities that contributed the most to the potential physical loss (PPL) of both habitats are also provided.

Item	Items or values	Confidence score (0-4)
Modelled area of the potential habitat for seagrass beds within the regional assessment area (km²) at baseline conditions (un-impacted)	1,583	2.4
Area of PPL within the potential habitat for seagrass beds (km²)	32 (17 km²historical – 2016 / 15 km² 2010 – 2016)
Modelled area of the potential habitat for seagrass beds within the regional assessment area (km²) in 2016	1,551
Percentage decline in the area of the potential habitat for seagrass beds within the regional assessment area (km²) up to 2016	2.0% (1.1 % historical – 2016 / 0. 9% 2010 – 2016)
Activities ranked by contribution to PPL (1st)	Aquaculture
Activities ranked by contribution to PPL (2nd)	Extraction – navigational dredging (capital and maintenance)
Activities ranked by contribution to PPL (3rd)	Extraction – sand and gravel (aggregates)
Activities ranked by contribution to PPL (4th)	Dredge and spoil disposal
Activities ranked by contribution to PPL (5th)	Coastal docks, ports and marinas
Modelled area of the potential habitat for horse mussel reefs within the regional assessment area (km²) at baseline conditions (un-impacted)	8,304	2.5
Area of PPL within the potential habitat for horse mussel reefs (km²)	39 (9 km²historical – 2016 / 30km² 2010 – 2016)
Modelled area of the potential habitat for horse mussel reefs within the RAA (km²) in 2016	8,265
Percentage decline in the area of the potential habitat for horse mussel reefs within the regional assessment area (km²) up to 2016	0.5% (0.1 % historical – 2016 / 0.4% 2010 – 2016)
Activities ranked by contribution to PPL (1st)	Dredge and spoil disposal
Activities ranked by contribution to PPL (2nd)	Aquaculture
Activities ranked by contribution to PPL (3rd)	Submarine cable operations
Activities ranked by contribution to PPL (4th)	Cultural and heritage sites/structures (such as wrecks, sculptures, foundations)
Activities ranked by contribution to PPL (5th)	Coastal docks, ports and marinas

Conclusions

The target has not been met for either seagrass beds or horse mussel reefs in UK waters. The extent and distribution of each habitat is neither stable nor increasing. This assessment is based on estimates of the extent of potential physical loss to each habitat caused by human activities up to and including the period 2010 to 2016. Potential physical loss is likely to be underestimated because, due to the lack of data, not all relevant activities were included in the assessment.

Mathematical modelling was used to estimate the potential distribution and extent of suitable habitat for horse mussel reefs and seagrass beds. However, it is also important to note that the predicted suitable habitat areas represent ‘potential’ habitat, which will differ from the ‘realised’ (occupied). The model could be improved if further data from realised habitat is included in future assessments.

Further information

An ensemble of Generalised Additive Modelling and Random Forest modelling was used to estimate the distribution of the features of interest. Species distribution modelling was used to establish the extent and distribution of potential habitat supporting seagrass beds and horse mussel reefs due to a lack of data allowing mapping of habitat extent. The overall performance of both ensemble models was considered to be good and overall accuracy was high. However, is also important to remember that the predicted suitable habitat areas represent potential habitat, which will differ from the realised (occupied) habitat.

Species may not be present in the potential habitat at all times due either to biological factors (disease, competition, predation, or inaccessibility) or anthropogenic activities (presence of structures or activities generating pressures and ultimately impacting the seabed). It is well documented that biological factors have resulted in the decline of habitat extent in the past. For example, it is estimated that 90% of the seagrass beds were lost in the UK during the 1920 and 1930s outbreak of the seagrass wasting disease (Labyrinthula zosterae) (Muehlstein and others, 1991). Other biological factors may also be contributing to the continued decline of horse mussel reefs in the UK (Strong and others, 2016). As such, the distributions produced are likely to over-estimate the realised habitat for these features. This may, in turn, bias the final potential physical loss values.

The extrapolated values indicate that the potential habitat supporting seagrass (Zostera marina) has declined by 2%. The decline of potential habitat for horse mussel was 0.5%. However, in terms of absolute area, the potential physical loss value was greater within horse mussel habitat. It is important to stress that it was not always possible to restrict activity data to the 2010 to 2016 assessment period. The inclusion of historical activity means that the potential physical loss calculated for the habitats will include loss predating 2010, which will subsequently over-estimate the total potential physical loss values within the time frame of this assessment (2010 to 2016).

Of the 13 activity classes relevant for seagrass (Zostera marina), aquaculture, navigational dredging, dredge and spoil disposal, and coastal development were the main sources of potential physical loss. At the regional level, Orth and others (2006) state that the environmental effects of excess nutrients or sediments are the most common and significant causes of seagrass decline, and result in small to very large areas of seagrass being lost. At the local scale, Orth and others (2006) suggest that dredging, hydrological change, eutrophication, and sediment deposition are the most important activities driving loss for temperate subtidal seagrass. The identification of dredging and sediment deposition concurs with results from the stratified extrapolation framework, which clearly identified nutrient-enriching activities such as aquaculture and dredging-related activities to be the greatest contributor to potential physical loss for seagrass.

Navigational dredging, dredge and spoil disposal, aquaculture, and recreational activity contributed the most to potential physical loss for horse mussel. Although other pressures are known to impact these habitats and are currently of greater concern, many are reversible (within 12 years) or damaging, and as such were not in the scope of this analysis of potential physical loss (no recovery within 12 years). For example, scallop dredges and trawls have been documented to degrade the structure of horse mussel reefs in UK waters (Magorrian, 1997; Magorrian and Service, 1998). This activity was considered to induce a significant and immediate decline in reef condition but not always the potential physical loss of the habitat. Strong and others (2016) states that the horse mussel reef in Strangford Lough has continued to decline despite the introduction of a ban on all mobile fishing gear within the Lough. They suggest that although anthropogenic activity may have initiated the decline in the condition of the reef in Strangford Lough, the eventual loss of occupied areas may well be related to biological processes such as high levels of starfish predation following the loss of the clumped reef structure, competition from other filter feeding species or the development of Allee effects within the local population of horse mussel.

Knowledge gaps

Lack of data from inshore fisheries is highly likely to cause an underestimation of potential physical loss. Additionally, information on activity intensity would be useful to calculate cumulative impacts that can turn habitat damage into habitat loss. Another important data gap relates to submarine infrastructure that is deposited for contingency measures such as rock dumps to protect pipelines or stabilise drilling platforms, which is currently not available as spatial layers.
Expert judgement was used to exclude pressures that caused habitat damage but not potential physical loss under realistic levels of activity. This subjectivity could be reduced if future iterations of the sensitivity matrices can contain intolerance/recovery code combinations equating to potential physical loss.
It is also recommended that greater use is made of observed distributions (mapping of realised habitat) to supplement and refine modelled outputs. Useful sources of information for this includes survey work undertaken to characterise and monitor features within the UK’s Marine Protected Areas network.

Further information

Use of existing sensitivity matrices highlighted the relevant pressures for each habitat, however, expert judgement was used to exclude pressures that caused habitat damage but not potential physical loss under realistic levels of activity. Equally, the Joint Nature Conservation Committee pressure/activity database matched the selected pressures with anthropogenic activities. Expert judgement was again used to exclude activities that were either unlikely to occur within specific habitats or unlikely to occur at a sufficient intensity to cause potential physical loss. This subjectivity could be reduced if future iterations of the sensitivity matrices can contain intolerance/recovery code combinations equating to potential physical loss such as the ultimate level of impact. Equally, it would be helpful if activity/pressure matrices could be filtered by pressures within specific habitats and that generate specific levels of impact.

Activity data underpins this, and potentially other, indicators. As such, efforts to regulate data sources, and the objects/activities within these sources are included to represent an activity, will be an important step toward standardisation. Equally, some activities with a measurable intensity (rather than a presence/absence pressure) need to be converted into footprints of specific damage or potential physical loss, such as what swept area ratios equate to an identifiable and ecologically relevant level of damage within a specific habitat. Arbitrary thresholds have been used to categorise intensity information, for example, Vessel Monitoring Systems Swept Area Ratios of 10 for mud and 20 for sand. The consistent conversion of activity intensity into damage and potential physical loss will require a process of calibration and standardisation to ensure consistent and compatible calculation of damage and potential physical loss between iterations of the assessment.

Currently, not all activities generating pressures that can determine potential physical loss are spatially recorded in Geographic Information System format. A notable example is represented by rock or concrete mattress deposits (dumps) that are used as contingency measures to resolve freespans, buckling of pipelines, or as a protection measure from fishing gear. Rock deposits are also used to stabilise drilling rigs in areas where currents are strong, or the seabed is unstable. These types of deposits have also been used in oil and gas decommissioning operations when comparative assessment showed high safety risks in recovering pipelines which were, therefore, rock-dumped where exposed, or as a preventative measure. To a lesser degree, this also applies to subsea cables. At times these rock deposits amount to many thousands or even hundreds of thousands of tonnes of material covering a considerable area which gets smothered. These are permanent deposits, will not be removed and, therefore, cause potential physical loss for habitats, which are sensitive to this type of pressure. In the case of stabilisation material for drilling rigs, application for these deposits is done pre-emptively in case local conditions require it, but it is not known if they actually take place.

While for these deposits there is a record held by the regulator in the form of consents issued indicating their location and amount, the fact that they are not yet available in a digitised Geographic Information System format represents a limitation to how accurate the estimate of potential physical loss can be. We can, therefore, infer that the area of potential physical loss assessed here for horse mussel (Modiolus modiolus) and seagrass (Zostera marina) might be slightly underestimated.

It is also recommended that greater use is made of observed distributions (mapping of realised habitat) to supplement and eventually replace modelled outputs. Useful sources of information for this includes the survey work undertaken to characterise and monitor features of interest within the UK’s Marine Protected Area network. Many of the features of interest contained within the UK’s Marine Protected Area network are largely mapped. Equally, the network is designed to protect a large proportion of the extent for many of the features, thereby providing a useful and potentially extensive data set.

New and more accurate modelled environment layers will also improve the distribution of habitats in inshore waters. Potential general improvements which can be investigated in the short term are:

Light Penetration: EUSeaMap doesn’t cover very shallow areas. Potential use of local modelled data.
Current: There is a possibility of some artefacts in the model due to islands. The current data is also of very low resolution which is not very appropriate for the model.

Potential site-specific improvements which can be investigated in the short term are:

Rathlin Island: areas of no substrate information in Marine Protected Areas (the model does not extend completely inshore). There are also areas of rock which overlap with the observed seagrass points which would result in no prediction of seagrass in the model.
Skerries and Causeway: Areas of ‘seabed’ which means that there was no survey data to feed into the model in these areas.

Waterfoot: The substrate is predicted to be not infralittoral in areas, with very low light levels. Investigate potential errors.

References

HM Government (2012) ‘Marine Strategy Part One: UK Initial Assessment and Good Environmental Status’ (viewed on 5 July 2018)

European Commission (2017) ‘Commission Implementing Decision (EU) 2016/1757 of 29 September 2016 on setting up the European Multidisciplinary Seafloor and Water Column Observatory — European Research Infrastructure Consortium (EMSO ERIC) (notified under document C(2016) 5542) (Text with EEA relevance)’ C/2016/5542, Official Journal of the European Union L 268, 1.10.2016, pages 113–117 (viewed on 27 November 2018)

Magorrian BH (1997) ‘The extent and temporal variation of disturbance to epibenthic communities in Strangford Lough, Northern Ireland’ Journal of the Marine Biological Association of the United Kingdom, 77: 1151-1164 (viewed on 26 November 2018)

Magorrian BH, Service M (1998) ‘Analysis of underwater visual data to identify the impact of physical disturbance on horse mussel (Modiolus modiolus) beds’ Marine Pollution Bulletin, 36(5): 354-359 (viewed on 26 November 2018)

Morris, E (2015) ‘Defining Annex I biogenic Modiolus modiolus reef habitat under the Habitats Directive’ Report of an inter-agency workshop, March 4th and 5th, 2014, JNCC Report Number 531. JNCC, Peterborough (viewed on 26 November 2018)

Muehlstein LK, Porter D, Short FT (1991) Labyrinthula zosterae sp. nov., the causative agent of wasting disease of eelgrass, Zostera marina. Mycologia, 83 (2): 180-191 (viewed on 26 November 2018)

Orth RJ, Carruthers TJ, Dennison WC, Duarte CM, Fourqurean JW, Heck KL, Hughes AR, Kendrick GA, Kenworthy WJ, Olyarnik S, Short FT (2006) ‘A global crisis for seagrass ecosystems’ Bioscience, 56(12): 987-996 (viewed on 26 November 2018)

Strong JA (2015) ‘Habitat area as an indicator of Good Environmental Status under the Marine Strategy Framework Directive: the identification of suitable habitat mapping methods with recommendations on best-practice for the reduction of uncertainty’ Defra Project ME5318 (viewed on 26 November 2018)

Strong JA, Service M, Moore H (2016) Estimating the historical distribution, abundance and ecological contribution of Modiolus modiolus in Strangford Lough, Northern Ireland’ in Biology and Environment: Proceedings of the Royal Irish Academy, Volume 116, Number 1, pages 1-16. Royal Irish Academy (viewed on 26 November 2018)

Acknowledgements

Assessment Metadata

Please contact marinestrategy@defra.gov.uk for metadata information

Recommended reference for this indicator assessment

James Strong¹, Cristina Vina-Herbon², Luca Doria², Anita Carter², Dan Edwards², Helen Lillis², Megan Parry², Laura Robson², Gemma Singleton^2,Mike Young³, Tim Mackie⁴, Phil Boulcott⁵ and Karen Robison⁶ 2018. Potential physical loss of predicted seafloor habitats. UK Marine Online Assessment Tool, available at: https://moat.cefas.co.uk/biodiversity-food-webs-and-marine-protected-areas/benthic-habitats/physical-loss/

¹University of Hull – Institute of Estuarine and Coastal Studies

²Joint Nature Conservation Committee

³Natural England

⁴Department of Environment, Agriculture & Rural Affairs, Northern Ireland

⁵Marine Scotland

⁶Natural Resources Wales