• Latest Version, 2024 Assessment
• Physical loss, 2018 assessment

Potential Physical Loss of Predicted Biogenic Habitats

There was 0.23% potential loss of horse mussel (Modiolus modiolus) reefs and 3.00% potential loss of eelgrass (Zostera marina) beds, caused by anthropogenic activities, in areas where probability of predicted suitable habitat presence was greater than or equal to 0.5. Therefore, the threshold was not achieved. 

Background

Benthic habitats are vital for the health of marine ecosystems. Horse mussel (Modiolus modiolus) reefs and eelgrass (Zostera marina) beds are benthic biogenic habitats that provide essential ecosystem services, e.g., nursery grounds, refuges, and food sources for many species. However, these habitats are fragile and sensitive to anthropogenic activities such as trawling (Holt and others, 1998; De la Torriente and others, 2022), aggregate extraction and aquaculture (Holt and others, 1998), which can cause irreversible habitat loss. 

This indicator measures the potential physical loss of assessed habitats due to selected anthropogenic activities using habitat suitability models. These models were used to overcome the limitations of data paucity in the location and extent of these important habitats to assess potential physical loss as a result of damaging pressures deriving from anthropogenic activities. Physical loss is defined as a permanent change to the seabed, which has lasted or is expected to last at least 12 years, or two UK Marine Strategy (UKMS) reporting cycles. 

For each assessed habitat, anthropogenic activities likely to cause potential physical loss were identified to obtain relevant spatial footprint data. The extent of overlap between the predicted suitable habitat and activity footprints were weighted by an activity-specific factor to estimate the amount of predicted suitable habitat potentially lost across and within UKMS sub-regions (Figure 1). 

Figure 1: UKMS region and sub-regions assessed in this report. Note that as neither habitat was predicted in the Celtic Seas (Seabed and Subsoil only) sub-region (purple area), this sub-region was excluded from assessment. 

Further information

The potential physical loss (PPL) indicator is used to assess progress against the target set for biogenic seafloor habitats in the UK Marine Strategy (UKMS) Part One, which requires that the area of selected habitat is stable or increasing. Data on location and extent of biogenic habitats are limited and the frequent mapping of habitat for UKMS assessment is unfeasible, therefore habitat suitability models (HSMs) were used as a proxy for habitat distribution. This assessment of the PPL indicator focusses on subtidal horse mussel (Modiolus modiolus) reefs and eelgrass (Zostera marina) beds because the environmental variables that influence presence and extent have been well studied (Tyler-Walters, 2007; Gormley and others, 2013; d’Avack and others, 2022; Bertelli and others, 2022; Castle and others, 2022), and / or are readily available, allowing areas of suitable habitat to be predicted. 

Modiolus modiolus, commonly known as horse mussels, are bivalve molluscs found subtidally around the British Isles, and can aggregate into extensive reefs where conditions are suitable, e.g., off the north and northwest coasts (Tyler-Walters, 2007). Horse mussel reefs are defined as: 

“Solid, massive structures which are created by accumulations of organisms … forming a substantial, discrete community or habitat which is very different from the surrounding seabed… it may to some degree be composed of sediments, stones and shells bound together by the organisms.” (Holt and others, 1998).  

Zostera marina is a type of seagrass, commonly called eelgrass, and is a marine angiosperm found in sublittoral and infralittoral zones, though this assessment focusses on sublittoral habitats only. Eelgrass is distributed patchily throughout the UK, primarily off the coasts of south and south-west England, Wales, Orkney, the Shetland Islands, and Scotland (d’Avack and others, 2022). Eelgrass forms dense ‘beds’ defined as where plant densities account for at least 5% cover of the seabed, though usually more than 30% cover (OSPAR, 2009). 

Additionally, both horse mussel reefs and eelgrass beds represent important food sources, nursery grounds and refuges for a wide variety of species (Heck Jr., Nadeau and Thomas, 1997; Zieman and Wetzel, 1990; Duarte and Chiscano, 1999; Sanderson and others, 2008; McDevitt-Irwin, Iacarella, and Baum, 2016; Nordland and others, 2016; Kent and others, 2016; Kent and others, 2017). However, both habitats are fragile and vulnerable to anthropogenic impacts (Magorrian and Service, 1998; Tyler-Walters, 2007; Cook and others, 2013; Jones and Unsworth, 2016; Elliott and others, 2017; d’Avack and others, 2022). 

Key pressures and impacts 

The direct impacts of anthropogenic activities represent the biggest threat to horse mussel reefs (OSPAR, 2008) and eelgrass beds (Elliott and others, 2017). These habitats are particularly sensitive to activities that cause physical change of habitat, suspension of sediments, smothering, abrasion and physical loss of habitat (d’Avack and others, 2022; Tyler-Walters, 2007).  

Physical loss of habitat is an extreme pressure on the marine seabed ecosystem. Habitat is lost if the substrate, morphology or topography is permanently altered. There are anthropogenic activities that cause this type of damage, for example offshore installations, port anchorage, dredging and dumping, and all types of constructions in or over the seabed. Physical loss of habitat is here defined as a permanent change to the seabed, which has lasted or is expected to last a period of 12 years (two UKMS reporting cycles) or more (European Commission, 2017). For this assessment, physical loss is termed potential physical loss (PPL), as estimates of loss are calculated from modelled habitat data and not directly observed. Loss may derive from both the physical footprint (PFP) of the activity and / or from the near-field footprint (NFFP), which is the indirect area of effect of anthropogenic pressures, e.g., the settling of suspended sediment and subsequent smothering of habitat outside the area of direct dredging. PPL may be caused by both permanent and transient anthropogenic activities. 

Measures taken to address impacts 

Horse mussel reefs and eelgrass beds are protected under international and national legislation and conventions. The following measures can be used to manage activities to reduce PPL if the indicator results are considered under the following mechanisms: 

• Environmental Impact Assessments and marine licensing. 

• Marine Spatial Planning (environmental considerations). 

• Environment, strategic Habitat Regulations and marine conservation Zones assessments. 

• The Fisheries Act 2020. 

• Measures to protect The Convention for the Protection of the Marine Environment of the North-East Atlantic (OSPAR) Threatened and Declining (T&D) habitats. 

• The establishment of an ecologically coherent network of Marine Protected Areas. 

Monitoring, assessment, and regional co-operation 

Areas that have been assessed 

The purpose of this assessment was to calculate the area of predicted suitable habitat for horse mussel reefs and eelgrass beds lost to anthropogenic activities. This assessment has been carried out on modelled habitat data at the scale of the UKMS region and sub-regions (Greater North Sea and the Celtic Seas; Figure 1). 

Assessment thresholds 

The indicator assessment threshold was no PPL in areas where the probability of the presence of suitable habitat was greater than or equal to 0.5 (i.e., greater than or equal to a 50% likelihood of the presence of suitable habitat). 

Thresholds for assessment of the PPL indicator in the UKMS 2018 part one were based on the UK target for biogenic seafloor for the UKMS Part One, described above. Therefore, if there was any potential loss of predicted suitable habitat for horse mussel reefs or eelgrass beds from the modelled baseline extent, the target would not be met. However, the current methodology to predict the distribution of assessed habitats incorporates the probability of occurrence of suitable habitat, which throughout the extent of the HSM is never zero. Therefore, the area of PPL could only be zero if activities did not intersect the HSM. The threshold of no PPL was therefore only applied to areas where the probability of the presence of suitable habitat was greater than or equal to 0.5. 

Assessment method

The methodology of the potential physical loss (PPL) indicator was conducted in five stages: 

1. The production of habitat suitability models (HSMs) to provide information on the extent and location of predicted suitable habitat for horse mussel (Modiolus modiolus) reefs and eelgrass (Zostera marina) beds. 

2. The identification of pressures that cause PPL of horse mussel reefs and eelgrass beds. 

3. The identification of activities that cause pressures that induce PPL. 

4. The acquisition of spatial activity footprint data on those activities that cause PPL and the production of layers (hereafter termed strata) weighted by the proportion of habitat lost due to each activity (PPL factor). 

5. The intersection of activity data and HSMs, weighted by the probability of the habitat occurring and the PPL factor. 

Habitat Suitability Models 

There is a paucity of data on the extent and distribution of horse mussel reefs and eelgrass beds, and the full-scale mapping of habitats only using monitoring or survey data every six years for the assessment of Good Environmental Status (GES) across the UK is not feasible. Therefore, HSMs were used in place of in-situ habitat data to predict suitable distribution of these biogenic habitats across UK waters. New HSMs, created using a different methodology from that used in the UKMS 2018 assessment, were produced to include more recent biological and environmental data to improve the prediction of habitat suitability. Methods broadly followed the methodology set out in Castle and others (2022), but incorporated updated environmental and biological data, and guidance from benthic experts, to produce HSMs for horse mussel reefs and eelgrass beds. 

Data preparation 

Environmental Predictor variables 

Habitat-specific environmental variables that influence the growth and survival of horse mussel reefs and eelgrass beds were used to predict the extent and location of suitable habitat. These predictor variables were selected based on the literature review performed by Castle and others (2022) (Table 1). Environmental data were sourced as raster layers in tagged image file format (tif) from respective databases and clipped to the UK Exclusive Economic Zone (EEZ) using the ‘extract raster by mask’ tool in ArcGIS v10.1 (ESRI, 2012). Each environmental layer raster was projected to Lambert azimuthal equal-area (LAEA) European Terrestrial Reference System (ETRS)-extended 1989 European Petroleum Survey Group (EPSG): 3035 and resampled to a common raster grid with a resolution of 300 m prior to input into the model. 

All environmental layers except for the ‘depth to seabed’ and ‘slope’ layers were input directly into the model. The ‘depth to seabed’ layer was created by sourcing the higher resolution Department for Environment, Food and Rural Affairs (Defra) Digital Elevation Model (DEM; Defra, 2020) which was supplemented with the European Marine Observation and Data network (EMODnet) Digital Bathymetry DTM (2022) raster data where the former was not available. The two raster layers were merged using the tool ‘Mosaic to New Raster (Data Management)’ in ArcGIS v10.1 (ESR, 2012), with the Defra DEM prioritised over the EMODnet Bathymetry raster. The resultant raster was resampled to a resolution of 300 m by 300 m using the ’Resample’ tool in ArcGIS v10.1. The ‘slope’ layer was derived directly from the resultant ‘depth to seabed’ layer created above using the Terrain Attribute Selection for Spatial Ecology (TASSE) v1.1 toolbox (Lecours and others, 2017) in ArcGIS, using the Horn (1981) method. 

Following the methodology in Castle and others (2022), bathymetry layers were restricted to specific depths for each habitat. Based on the occurrence of eelgrass beds within the presence dataset, depth in the HSM was restricted to 0 – 15 m (d’Avack and others, 2022). For the horse mussel HSM, depth was restricted to 0 – 248 m, based on the occurrence of horse mussel reefs and the deepest sample observed within the available presence dataset. Further, to align with Castle and others (2022), the model was also restricted to remove the Severn estuary, which is an area of high sediment loading. 

The HSMs were used to predict the probability of suitable habitat for each 300 m by 300 m grid cell where environmental values were available from every environmental predictor variable layer. However, in the absence of one or more layers, a prediction on the probability of suitable habitat was not made. Whilst the coverage of both models was limited by data paucity to some extent, the eelgrass beds HSM extent was further limited to 5 km from the coastline to match the wave fetch layer (Burrows, 2020). The Burrows (2020) wave fetch layer was introduced to reduce the probability of predicted suitable habitat for eelgrass beds in high energy environments and prevent shallow offshore areas unsuitable for eelgrass beds (such as Dogger Bank) being represented in the model at low probability, inflating the area of predicted suitable habitat. 

Table 1: List of predictor variables used for predictive modelling. The most recent available data were used where possible. 

Predictor variable	Source	Units	Original Spatial Resolution	Release Date	Data Collection Year	Horse mussel reefs	Eelgrass beds
Depth to seabed	Defra’s Marine Digital Elevation Model (DEM’ Defra, 2020), supplemented with EMODnet Digital Bathymetry DTM (2022) where the former was not available.	Metres (m)	Defra Marine DEM: 1 arc second    EMODnet Digital Bathymetry DTM: 1/16 arc minute	Defra Marine DEM: 2020    EMODnet Digital Bathymetry DTM: 2022	Defra Marine DEM: 1851 - 2020      EMODnet Bathymetry: 1815 - 2022	Yes	Yes
Slope of the seabed	Derived from the “Depth to the seabed” layer (above), using the Terrain Attribute Selection for Spatial Ecology (TASSE) toolbox in ArcGIS as recommended in Lecours and others (2017). Calculated by using the Horn (1981) method.	Degrees (o)	Same as “Depth to seabed” layer	Same as “Depth to seabed” layer	Same as “Depth to seabed” layer	Yes	No
Kinetic energy at the seabed due to waves	EMODnet Seabed  Habitats (mean of annual  90th percentile values over six years). EMODnet (2019).	Newtons per Square Metre   (N/m2)	North Sea and Celtic Seas inshore (<6 km from the coast) 100 to 300 m.    North Sea and Celtic Seas inshore (>6 km from the coast) 12.5 km.	2023	2000-2005	Yes	Yes
Kinetic energy at the seabed due to currents	EMODnet Seabed  Habitats (mean of annual  90th percentile values over six years). EMODnet (2015).	Newtons per Square Metre   (N/m2)	300m at the  coast and  combination of  1.8 km in the  North and Celtic  Sea and 10 km  in the North  East Atlantic	2018	2000-2005	Yes	Yes
Seabed substrate (categorical)	EMODnet Seabed  Habitats  Biogenic substrate in Europe, used in EUSeaMap (2023).	1 Mud/sandy mud, 2 Sand/muddy sand, 3 Mixed sediment, 4 Coarse sediment, 5 Rock, 6 Biogenic substrate	N/A - vector	2023	1991-2023	Yes	Yes
Mean of annual minima temperature at the seabed	Bio-ORACLE Minimum temperature derived from the mean depth of benthic layers. Assis and others, (2017).	Degrees Celsius (°C)	0.08°	2017	2000-2014	No	Yes
Maximum temperature at the seabed	Bio-ORACLE Maximum temperature derived from the mean depth of benthic layers. Assis and others, (2017).	Degrees Celsius (°C)	0.08°	2017	2000-2014	Yes	No
Minimum salinity	Bio-ORACLE Minimum salinity derived from the mean depth of benthic layers. Assis and others, (2017).	Practical Salinity Unit (PSU)	0.08°	2017	2000-2014	Yes	Yes
Wave fetch	Wave fetch GIS layers for the UK and Ireland at 200m scale (Burrows, 2020)	Log10 cells	200 m	2020	Modelled	No	Yes

To reduce overfitting of the HSMs, the correlation between the different environmental predictor variables was plotted; in the event of high correlation between layers, one of the correlated layers would be removed. For example, previous runs of the eelgrass bed model (Castle and others, 2022) contained photosynthetic active radiation (PAR) at the seabed as a predictor variable, but the PAR layer, once updated (PAR at the seabed, EMODnet, 2018), was found to be highly correlated with ‘depth to seabed’. The high positive correlation was explained by the fact that PAR at the seabed was derived from light at the surface and adjusted for depth using EMODnet Digital Bathymetry DTM (EMODnet, 2022), one of the layers used to create the ‘depth to seabed’ layer. Therefore, PAR at the seabed was removed and depth retained in the model for the following reasons: 

1. The resolution of the bathymetry layer was higher, allowing the resulting model to better fit the coastline; and 

2. PAR has high intra-annual variability, being affected by storms, run-off, algal blooms among others, whereas depth is consistent. 

Following the removal of highly correlated predictor variables, all environmental layers were processed so that the extent and resolution of each layer was consistent. All environmental layers were intersected using the ‘raster’ package (Hijmans, 2023), which was used to crop the ‘depth to seabed’ layer. All other environmental layers were resampled to the ‘depth to seabed’ layer to ensure the resolution and extent of each layer matched. Environmental layers were then combined into a raster stack for processing. 

Habitat occurrence response variables 

Suitable habitat occurrence is predicted within HSMs using habitat presence data, which can be supplemented with absence data to further refine the model. Due to the paucity of true absence information from survey data, samples indicating the presence of habitats not known to co-exist with the habitat being modelled (Table 2) were used as a proxy for the absence of the respective habitats, hereafter termed pseudo-absence data (a common practice in HSMs; Chefaoui and others, 2016; Castle and others, 2022; Charbonnel and others, 2023). The combinations of presence and pseudo-absence data are referred to as response variables. 

The following parameters were taken into consideration prior to running the model: 

• Presence / pseudo-absence data were not restricted by date – since temporal variability is inherent in all biological datasets, this maximised the data available to train and test the HSM. 

• Only habitat records were used for presence / pseudo-absence data - species data were excluded from the model; this provided the best estimate of the distribution and extent of habitats. 

• Point rather than polygon data were used to denote occurrence type - to ensure the highest accuracy of environmental conditions required to support the prediction of suitable habitat. The model only accepts biological response variables in point format, polygon data would therefore have to be converted into point data at a specific density per polygon, potentially introducing bias into the model, and affecting the weighting of presence against pseudo-absence. Additionally, effective duplicate removal of the data would be unfeasible.  

• Only subtidal habitat records were selected - eelgrass beds occur in both intertidal and subtidal regions; however, the focus of the indicator assessment only concerns subtidal habitats. 

• Taxonomic status – Zostera angustifolia records were considered as taxonomic synonyms of Zostera marina (d'Avack, 2022; Guiry & Guiry, 2023). 

The following datasets were chosen as sources of presence and pseudo-absence data: OSPAR threatened and/or declining habitats database (OSPAR, 2022), Natural England Marine Evidence Base (Natural England, 2021), Small Isles MPA survey (Greathead and others, 2023), Annex I Reef database (Duncan and others, 2022) and the Marine Recorder database (JNCC, 2022). For each database, a specific selection method was used to obtain high confidence occurrence (see Table 2 for further details). 

To avoid pseudo-replication, and introducing bias into the weighting of the model, duplicate data points were removed from all databases used for both presence and pseudo-absence data. 

Table 2: Summary of datasets and selection criteria used for the presence and pseudo-absence response variables. Please see footnotes for explanation of habitats and codes. FOCI = Features of Conservation Importance, HOCI = Habitat of Conservation Importance (NE and JNCC, 2010). 

Habitat	Source	Habitats indicating presence	Habitats indicating pseudo-absence
Eelgrass beds	Annex I Reefs Database 2022	N/A	No Annex I Reef habitats excluded.
	Marine Recorder August 2022	EUNIS 2007 habitat code:  A5.5331 1	All EUNIS 2007 habitat codes excluding:  A5.533 2  A5.5331 1
	OSPAR Threatened and/or declining habitats database 2022	OSPAR T&D habitat:  Zostera marina beds	All OSPAR T&D habitat types excluding:  Carbonate mounds  Intertidal mudflats  Intertidal Mytilus edulis beds on mixed and sandy sediments  Littoral chalk communities, Maerl beds  Seamounts  Zostera beds
Horse mussel reefs	Natural England Evidence Base 2021	FOCI name:  Horse mussel (Modiolus modiolus) beds	All FOCI codes excluding HOCI 9 3  AND  All FOCI names excluding those containing “modiolus”  AND  All Marine Habitat Classification for Britain and Ireland habitats and excluding:  LR 4  LS 5
	Marine Recorder August 2022	EUNIS 2007 habitat codes:  A5.621 6  A5.622 7  A5.623 8  A5.624 9	All EUNIS 2007 habitat codes excluding A5.621, A5.621 6  A5.622 7  A5.623 8  A5.624 9
	Small Isles MPA survey	Community code abbreviations:  MXMS 10  MXM 11  MXMA 12  MXMP 13	All community code abbreviations excluding:  MXMS 10  MXM 11  MXMA 12  MXMP 13
	OSPAR Threatened and/or declining habitats database 2022	OSPAR T&D habitat: Modiolus modiolus horse mussel bed	All OSPAR T&D habitat types excluding:  Carbonate mounds  Intertidal mudflats  Intertidal Mytilus edulis beds on mixed and sandy sediments  Littoral chalk communities  Seamounts  Modiolus modiolus horse mussel beds  Ostrea edulis beds  Oceanic ridges with hydrothermal vents/fields

1 - A5.5331 = Zostera marina/ angustifolia beds on lower shore or infralittoral clean or muddy sand
2 - A5.33 = Zostera beds in full salinity infralittoral sediments,
3 – HOCI 9 = Horse mussel (Modiolus modiolus) reefs
4 – LR = Littoral rock (and other hard substrata)
5 – LS = Littoral sediment
6 - A5.621 = Modiolus modiolus beds with hydroids and red seaweeds on tide-swept circalittoral mixed substrata)
7 - A5.622 = Modiolus modiolus beds on open coast circalittoral mixed sediment) 
8 - A5.623 = Modiolus modiolus beds with fine hydroids and large solitary ascidians on very sheltered circalittoral mixed substrata)
9 - A5.624 = Modiolus modiolus beds with Chlamys varia, sponges, hydroids and bryozoans on slightly tide-swept very sheltered circalittoral mixed substrata
10 – MXMS = Mixed sediment and cobbles, occasional boulders with few burrows and Faunal Turf with Arachnanthus sarsi and Modiolus modiolus
11 – MXM = Mixed sediment and cobbles, occasional boulders with few burrows and Faunal Turf and Modiolus modiolus
12 – MXMA = Mixed sediment and cobbles, occasional boulders with few burrows and Faunal Turf with Modiolus modiolus and Atrina fragilis
13 – MXMP = Mixed sediment and cobbles, occasional boulders with few burrows and Faunal Turf and Modiolus modiolus with Pachycerianthus multiplicatus 

Prior to producing the HSMs, all presence and pseudo-absence data were intersected with the extent of the respective models to expedite processing efforts, which subsequently reduced numbers of available habitat records. 

The cell IDs from grid cells of the raster stack were then assigned to each sample point based on location to facilitate the reduction of spatial autocorrelation. 

Reducing spatial autocorrelation 

To maximise efficiency of data collection, sampling is often conducted in relatively small areas, such as MPAs, or as part of environmental assessments for industrial purposes in a specified location. As such, both presence and pseudo-absence observations were frequently found to be clustered together within raster grid cells. 

To reduce the effect of the clustered data points, both observation types were reduced to a single random observation of presence or pseudo-absence per grid cell (by cell ID, described above) using the programming language R (R Core Team, 2019) in R Studio (RStudio Team, 2022) and the ‘raster’ package (Hijmans, 2023). The reduction in observation data ensures a greater degree of independence between observations, reducing the effects of spatial autocorrelation, and improving the accuracy of the model by not over-predicting the probability of habitat presence or absence. The selection of occurrence was determined by which observation type had the greater number of records per grid cell. However, as the number of records of pseudo-absence were greater than the number of presence records, pseudo-absence records were weighted to balance the selection of points between the two observation types. Weighted pseudo-absence was calculated using Equation 1. 

Equation 1: The calculation of weighted pseudo-absences per grid cell. 

Where the ratio between the total number of presence observations (P) and the total number of pseudo-absence observations (A) is multiplied by the number of pseudo-absence observations within the raster cell (A_n). 

One point was randomly selected from the observation type with the highest value (i.e., number of presence records or value of weighted absence) in R (R Core Team, 2019). In the event of a cell with equal values of presence to weighted absence, the observation type was marked as present. The number of records used as biological response data for the HSMs is shown in Table 3. 

Table 3: Number of observations used to produce the HSM, per input dataset for the response variables, and after the selection process used to reduce the effect of spatial clustering. The date range of the biological response records used in the HSM are also shown. 

Model	Response	Date Range	Number of initial data points	Number of data points used to produce HSM after selection process
Eelgrass beds	Presence	1976 – 2023	4,730	886
Eelgrass beds	Pseudo-Absence	1962 – 2023	157,200	29,618
Horse mussel reefs	Presence	1966 – 2021	1,471	473
Horse mussel reefs	Pseudo-Absence	1962 – 2023	175,818	29,440

Training and test data 

In model development, response data are used to both train and test the model. For each run of the HSMs, a random selection of 75% of the biological data were used to train the model and the remaining 25% used to test the model. 

Modelling 

Following Castle and others (2022), the JNCC Species Distribution Modelling (SDM) Framework (an open-source R package; JNCC, 2023; https://github.com/jncc/sdms) was used to produce models for horse mussel reefs and eelgrass beds. The JNCC SDM Framework includes Boosted Regression Tree, Support Vector Machine, General Additive Model, General Linear Model and Maximum Entropy, and Random Forest (RF) algorithms (Castle and others, 2022). The RF algorithm was used in the production of HSMs within this assessment to align with Castle and others (2022) and based on expert opinion from the package authors and previous pilot studies that identified it as the best performing model. The JNCC SDM framework uses an ensemble modelling approach, which takes the average probability of occurrence of habitat from a user-defined number of model runs, as the output. For this assessment, the number of iterations of each HSM was set to 50 to align with Castle and others (2022).  

The output of the JNCC SDM package using the RF algorithm creates a prediction across the extent of the raster stack of environmental predictor variables. As a result, probability is calculated regardless of whether the habitats assessed are present in-situ or not. Although the probability of suitable habitat may be high, it does not necessarily represent realised (occupied) habitat. Additionally, the model is limited to the specific environmental variables that are used as predictor variables to describe the occurrence of each habitat. Other biological and environmental factors which might limit extent, such as disease, competition, storm events, and anthropogenic activities are not accounted for when predicting habitat suitability.  

In modelling, the receiver operating characteristic (ROC) curve shows the performance of a model, plotting the False Positive Rate against the True Positive Rate. The area under the plotted ROC curve (AUC) is used to evaluate model performance, and measures from 0 to 1. An AUC score of 0 indicates that the model has performed poorly (100% of predictions are wrong), an AUC score of 1 indicates that the model has performed extremely well (Hanley & McNeil, 1982). The AUC scores for all 50 runs of the HSM for horse mussel reefs were between 0.914 and 0.972, with a mean of 0.949, denoting that the model performed well. For eelgrass beds HSM the AUC scores were between 0.877 and 0.961, with a mean of 0.924, similarly denoting good model performance. 

Important Environmental Variables 

Alongside the HSM, the JNCC SDM framework outputs environmental variables used to predict species or habitat distribution ranked in order of importance, based on the mean increase in node purity (IncNode purity). A higher mean IncNode purity indicates the higher importance of the variable in predicting probability of occurrence, allowing the comparison of variables within the model (note there is no maximum value). The variables identified as the most important to predict the presence of horse mussel reefs in this study were maximum temperature at the seabed, followed closely by seabed substrate and minimum salinity (Table 4). 

Table 4: Importance of all environmental variables in the horse mussel reef HSM produced for this assessment. 

Rank	Environmental Variable	Mean IncNode Purity
1	Maximum temperature at the seabed	23.81
2	Seabed substrate	22.06
3	Minimum salinity	20.72
4	Kinetic energy at the seabed due to currents	18.18
5	Kinetic energy at the seabed due to waves	17.23
6	Depth to seabed	12.37
7	Slope of the seabed	8.66

The variables identified as the most important to predict the presence of eelgrass beds in this study were wave fetch, followed by depth to seabed and minimum temperature at the seabed (Table 5). 

Table 5: Importance of all environmental variables in the eelgrass beds HSM produced for this assessment. 

Rank	Environmental Variable	Mean IncNode Purity
1	Wave fetch	18.17
2	Depth to seabed	15.73
3	Minimum temperature at the seabed	10.58
4	Minimum salinity	7.64
5	Current energy at the seabed	6.48
6	Wave energy at the seabed	6.18
7	Seabed substrate	3.18

Habitat Suitability Model raster to vector conversion 

HSMs produced by the JNCC SDM package were output in raster format, where the value for each grid cell indicated the probability of occurrence of the associated habitat. To provide statistics such as the area of suitable habitat and area of PPL at the UKMS region or sub-region scale, raster layers were imported into R version 3.6.1 (R Core Team, 2019) using the ‘raster’ package (Hijmans, 2023) and converted to vector format using the ‘rasterToPolygon’ function. This vectorised HSM was then exported as a shapefile to be intersected with the UKMS sub-region layer. 

Intersection with UKMS sub-regions 

Model outputs were intersected with the UKMS sub-regions layer for assessment at the relevant scale for the UKMS 2024 Part One. Exported HSM shapefiles were imported into ArcGIS v10.1 (ESRI, 2012), along with the UKMS sub-region layer, and intersected using the ‘Intersect’ tool. ‘Fix Geometry’ tool was run on the layer in QGIS (QGIS.org, 2020) to ensure all geometries were valid before the removal of transitional waters. 

Removal of Transitional Waters 

Benthic indicator assessments for the UKMS 2024 Part One do not consider transitional waters, therefore these were removed prior to subsequent indicator analysis. Layers used in the creation of the transitional waterbodies were sourced from the following: Water Framework Directive (WFD) Transitional and Coastal Waterbodies Cycle 2 (Environment Agency, 2014), Transitional water bodies (Scottish Environmental Protection Agency, 2010), Northern Ireland Transitional Waters 2009-2015 (Northern Ireland Environment Agency, 2016), WFD Transitional Waterbodies Cycle 3 (Natural Resources Wales, 2023). These layers were merged into a single feature layer, prior to being transformed into LAEA projected coordinate reference system. The transitional waters layer was then erased from the HSM in ArcGIS v10.1 (ESRI, 2012) to remove this from the assessment of PPL. The intersection of the HSM with UKMS sub-regions and the removal of transitional waters altered the boundaries of some grid cells and as a result grid cells are hereafter referred to as ‘polygons’ to reflect that not all polygons will be 300 m by 300 m cells. The ‘Fix Geometries’ tool was then run in QGIS v3.16 (QGIS.org, 2020) to ensure geometries were valid prior to exporting the layer as a shapefile. 

Area Calculation 

Area calculations were carried out on the HSM intersected with the UKMS sub-regions, and with transitional waters erased. The HSM was imported into R using the ‘sf’ package (Pebesma, 2018). The area of each polygon was calculated using the ‘st_area’ function from the ‘sf’ package (Pebesma, 2018 & 2023). This area was then multiplied by the associated probability of the habitat occurrence to obtain the area of predicted habitat per polygon in km². To obtain the total area of habitat predicted by the HSMs, the area of suitable habitat per polygon was summed. This approach to predict the area of suitable habitat, followed the methodology used by Castle and others (2022) and suggested by Calabrese and others (2014), and was adopted to prevent applying ad-hoc thresholds based on the probability of presence for HSMs. This calculation of predicted suitable habitat area may be summarised by Equation 2. 

Equation 2: Method of calculating the total potential area of suitable habitat per HSM. 

Where the area (A) was multiplied by the probability (P) for each polygon (i) and summed over all polygons (N) in the HSM. 

Final HSMs used in the assessment of PPL 

Horse mussel reefs HSM 

The JNCC SDM package used the RF algorithm to predict the probability of suitable habitat for horse mussel reefs across the UKMS region (Figure 2). Probability of suitable habitat for horse mussel reefs was high in areas such as the Western Channel and Celtic Sea; the Irish Sea; Strangford Lough; the waters around Northern Ireland and the Firth of Clyde; and the Small Isles and Orkney. 

Figure 2: Mean predictive values of habitat suitability for horse mussel reefs across the UKMS region with transitional waters removed. 

The area of suitable habitat for horse mussel reefs was calculated as 64.772.85 km² across the entire UKMS region (approximately 8.89% of the total area). Within UKMS sub-regions, the area of suitable habitat for horse mussel reefs was much higher in the Celtic Seas (41,179.61 km²) than the Greater North Sea (23,593.24 km²) (Table 6). However, due to the size difference between sub-regions, the area of predicted suitable habitat represented as a percentage of the total area of the sub-region was nearly identical (8.88% for the Celtic Seas and 8.91% for the Greater North Sea). 

Table 6: Area of horse mussel reef per UKMS sub-region with transitional waters removed. 

UKMS sub-region	Total area of UKMS assessment unit (km2)	Predicted area of suitable habitat for horse mussel reef (km2)	Percentage of UKMS assessment unit as predicted suitable habitat for horse mussel reef (%)
Celtic Seas	463,991.23	41,179.61	8.88
Greater North Sea	264,910.00	23,593.24	8.91
Total	728,901.23	64,772.85	8.89

Of note is where the model produced different results from those of previous studies, such as in the North Sea off the coast of Norfolk and in the Bristol Channel. Previous models of the habitat suitability of horse mussels have highlighted the predicted presence of reefs off the coast of Norfolk (Gormley and others, 2013; Castle and others, 2022). Horse mussel reefs are not known to occur in this area, which is known for silty and turbid conditions, and Sabellaria spinulosa reefs are known to occur in this location as the area provides conditions necessary to extract materials to build tubes/reefs (Holt and others, 1998). Horse mussel reefs suffer from smothering (Hutchinson and others, 2016), likely contributing to the absence of reefs in this area. The model predicts high suitability of predicted suitable habitat for horse mussel reefs around the Hebrides, Orkney, Strangford Lough and Anglesey (Figure 3), which is consistent with previous HSMs (Gormley and others, 2013; Castle and others, 2022). The improved modelling approach presented in this assessment has reduced the potential for incorrect predictions, for example the model does not predict the erroneous presence of habitat suitable for horse mussel reefs off the coast of Norfolk identified in previous studies (Figure 2). Additionally, the model also predicts presence of horse mussel reefs in the Bristol Channel (Figure 2); although this differs from previous HSMs (Gormley and others, 2013; Castle and others, 2022), other publications have highlighted that dense aggregations of juvenile horse mussels occur in the Bristol Channel, though do not survive to reach adulthood (Fletcher and others, 2012). 

Figure 3: Mean predictive values of habitat suitability for horse mussel reefs across the UKMS region, with transitional waters removed, for A- the Hebrides; B – Orkney; C – Strangford Lough; and D – North Wales / Anglesey. 

Eelgrass bed HSM 

The JNCC SDM package used the RF algorithm to predict the probability of suitable habitat for eelgrass beds across the UKMS region (Figure 4). Probability of suitable habitat for eelgrass beds was high in areas such as the Llŷn peninsula, Portsmouth, the Isles of Scilly and Islay (Figure 5). The prediction of the probability of suitable habitat for eelgrass beds was also high in areas such as Strangford Lough, Loch Ryan, the Hebrides, Orkney, Luce Bay, Moray Firth and Dornoch Firth. 

Figure 4: Mean predictive values of habitat suitability for eelgrass beds across the UKMS region with transitional waters removed. 

Figure 5: Mean predictive values of habitat suitability for eelgrass beds across the UKMS region, with transitional waters removed, for A- Islay; B – Anglesey and the Llŷn peninsular; C – the Isles of Scilly; and D – Portsmouth and the Isle of Wight. 

The area of suitable habitat for eelgrass beds was calculated as 2,277.54 km² across the entire UKMS region (approximately 0.31% of the total area of the UKMS region). Similar to the results of the horse mussel reef HSM, when split between UKMS sub-regions, the area of suitable habitat for eelgrass beds was much higher in the Celtic Seas (1,652.92 km²) than the Greater North Sea (624.63 km2) (Table 7). However, when the predicted area of suitable habitat was taken as a percentage of the area of each sub-region, values varied between sub-region more than the results of the horse mussel HSM (0.36% for the Celtic Seas, and 0.24% for the Greater North Sea). 

Table 7: Area of eelgrass beds per UKMS sub-region with transitional waters removed. 

UKMS sub-region	Total area of UKMS assessment unit (km2)	Predicted area of suitable habitat for eelgrass beds (km2)	Percentage of UKMS assessment unit as predicted suitable habitat for eelgrass beds (%)
Celtic Seas	463,991.23	1,652.92	0.36
Greater North Sea	264,910.00	624.63	0.24
Total	728,901.23	2,277.54	0.31

Activity Data 

Spatial layers of activities that contribute pressures leading to PPL were used in the indicator to identify the location and extent of overlap with the HSMs (Figure 6 and Figure 7). Where possible data layers were restricted to the years since data used for the last assessment e.g., 2017-2022 (i.e., to continue on from the date of the last assessment to latest available data).  

Figure 6: Distribution of assessed anthropogenic activity layers that contribute to potential physical loss of predicted suitable habitat for horse mussel reefs for the UKMS 2024 assessment. Coastal defence and land claim protection, Coastal docks ports and marinas and Recreational activities (anchorages only) - © British Crown Copyright, 2023. All rights reserved. Permission Number Defra012017.002. This product has been derived in part from material obtained from the UK Hydrographic Office with the permission of the UK Hydrographic Office, Her Majesty’s Stationery Office and the Keeper of Public Records. NOT TO BE USED FOR NAVIGATION. Towed bottom-contact fishing - ICES, 2021. OSPAR request on the production of spatial data layers of fishing intensity/pressure. In Report of the ICES Advisory Committee, 2021. ICES Advice 2021, sr.2021.11. https://doi.org/10.17895/ices.advice.8297 – CC BY 4.0. Aquaculture - © British Crown Copyright, 2023. All rights reserved. Permission Number Defra012017.002 This product has been derived in part from material obtained from the UK Hydrographic Office with the permission of the UK Hydrographic Office, Her Majesty’s Stationery Office and the Keeper of Public Records. NOT TO BE USED FOR NAVIGATION; and contains public sector information, licensed under the Open Government Licence v3.0, from Crown Estate Scotland. Extraction – sand and gravel - © The Crown Estate. Extraction – navigational dredging - © JNCC. This JNCC Extraction - navigational dredging data is licensed under the Open Government Licence v3.0 except where otherwise stated. Contains source data: CNC/NRW, DAERA, Marine Scotland, MMO. Dredge and spoil disposal - © Cefas. This Dredge and spoil disposal data is licensed under the Open Government Licence v3.0 except where otherwise stated. Renewable energy – wind, and Submarine Cables - © KIS-ORCA 2023 kis-orca.org. Renewable energy – tidal (not including cables) and Renewable energy – wave (not including cables) - Contains public sector information, licensed under the Open Government Licence v3.0, from Crown Estate Scotland; and Contains data provided by The Crown Estate that is protected by copyright and database rights (The Crown Estate Open Data Licence (GIS) - Version 1.1). Marine hydrocarbon extraction - Contains information provided by the Oil and Gas Authority and/or other third parties. Wrecks - © UKHO. This wrecks data is licensed under the Open Government Licence v3.0 except where otherwise stated. Submarine Pipelines - Contains information provided by the OGA. 

Figure 7: Distribution of assessed anthropogenic activity layers that contribute to potential physical loss of predicted suitable habitat for eelgrass beds for the UKMS 2024 assessment. Coastal defence and land claim protection, Coastal docks ports and marinas and Recreational activities (anchorages only) - © British Crown Copyright, 2023. All rights reserved. Permission Number Defra012017.002. This product has been derived in part from material obtained from the UK Hydrographic Office with the permission of the UK Hydrographic Office, Her Majesty’s Stationery Office and the Keeper of Public Records. NOT TO BE USED FOR NAVIGATION. Aquaculture - © British Crown Copyright, 2023. All rights reserved. Permission Number Defra012017.002 This product has been derived in part from material obtained from the UK Hydrographic Office with the permission of the UK Hydrographic Office, Her Majesty’s Stationery Office and the Keeper of Public Records. NOT TO BE USED FOR NAVIGATION; and contains public sector information, licensed under the Open Government Licence v3.0, from Crown Estate Scotland. Extraction – sand and gravel - © The Crown Estate. Extraction – navigational dredging - © JNCC. This JNCC Extraction - navigational dredging data is licensed under the Open Government Licence v3.0 except where otherwise stated. Contains source data: CNC/NRW, DAERA, Marine Scotland, MMO. Dredge and spoil disposal - © Cefas. This Dredge and spoil disposal data is licensed under the Open Government Licence v3.0 except where otherwise stated. Renewable energy – wind, and Submarine Cables - © KIS-ORCA 2023 kis-orca.org. Renewable energy – tidal (not including cables) and Renewable energy – wave (not including cables) - Contains public sector information, licensed under the Open Government Licence v3.0, from Crown Estate Scotland; and Contains data provided by The Crown Estate that is protected by copyright and database rights (The Crown Estate Open Data Licence (GIS) - Version 1.1). Marine hydrocarbon extraction - Contains information provided by the Oil and Gas Authority and/or other third parties. Wrecks - © UKHO. This wrecks data is licensed under the Open Government Licence v3.0 except where otherwise stated. Submarine Pipelines - Contains information provided by the OGA. 

Pressure Identification 

The MarLIN (Marine Life Identification Network) Marine Evidence-based Sensitivity Assessment (MarESA; Tyler-Walters and others, 2018) was used to identify pressures that horse mussel reefs and eelgrass beds are sensitive to. Pressures that contributed to PPL were selected where the habitat had resistance (intolerance) categorised as ‘no’ or ‘low’, and resilience (recovery) categorised as ‘very low’ and ‘low’. The combination of these levels of resistance and resilience result in ‘high’ and ‘very high’ levels of sensitivity. 

Activity Identification 

Since data on pressure footprints alone are not available, anthropogenic activities were identified from the JNCC Pressures-Activities Database (Robson and others, 2018) to match each pressure causing PPL, to activities that cause the relevant pressure for horse mussel reefs and eelgrass beds. Those activities considered not to generate enough pressure to cause PPL, or not relevant for the habitats assessed, were removed based on expert judgment; these are reflected in the confidence assessment. 

The extent and distribution of each activity considered to cause PPL was obtained in spatial format. These were sourced from relevant databases (Table 8) and transformed into LAEA projected coordinate reference system (same as the HSM) prior to analysis. 

Selections were made on individual data layers to remove records that were not relevant to physical loss (e.g., removing proposed sites in Renewable energy). The resulting activity layers were then dissolved to remove individual records and attributes. In cases where there was more than one source of data for an activity, these data were merged. 

Activity data were split into two footprint types to ensure that both the direct and indirect footprints of pressures causing PPL from the activity were captured. The footprints of activities that cause direct PPL were each termed as a physical footprint (PFP) and were obtained from all activities assessed. For point or polyline data, buffers were used to produce PFP layers (Table 9). Several activities also cause PPL outside of the footprint of activity through, for example, habitat change or sedimentation; this was termed near-field footprint (NFFP).  

The extent and distribution of NFFP pressures were calculated through the application of buffer values, using another buffer where relevant (Table 9), the size of which were taken from Strong and others (2018), obtained from scientific literature, calculations, and expert judgement. These buffer values were used to delimit the area around the PFP still thought to cause PPL, to prevent areas that might only be damaged by activities being considered. NFFP buffers were created in ArcMap v10.1 (ESRI, 2012) using the ‘Multiple ring buffer’ tool. Submarine cable operations activity layers included both communications and power, which had different values for buffer extents and so these were processed separately before merging into one activity layer. 

Finally, activity layers were imported into QGIS v3.16 (QGIS.org, 2020) and the ‘Fix Geometries’ tool was used to correct any invalid geometries. 

Table 8: Activity data used to represent pressures contributing to PPL for horse mussel reefs and eelgrass beds. Stratum number = unique ID referring to activity name and footprint type, see Table 9 for further details; PFP = physical footprint, NFFP = near-field footprint. 

Activity	Stratum	PFP/NFFP	Source	Temporal range	Represent PPL in horse mussel reefs?	Represent PPL in eelgrass beds?
Coastal defence & land claim protection	S1    S16	PFP     NFFP	Defra Marine Reference Data (UKHO)	2022	Y	Y
Coastal docks, ports & marinas	S2    S17	PFP     NFFP	Defra Marine Reference Data (UKHO)	2022	Y	Y
Towed bottom-contact fishing	S3    -	PFP    No NFFP	ICES Fishing Intensity	2016-2020	Y	N
Aquaculture	S4    S18	PFP     NFFP	Crown Estate Scotland  Defra Marine Reference Data (UKHO)	2022	Y	Y
Extraction – sand and gravel (active area only)	S5    S19	PFP     NFFP	The Crown Estate	2017-2020	Y	Y
Extraction – navigational dredging (capital & maintenance)	S6    S20	PFP     NFFP	JNCC Extraction pressure layer (Navigation only)	2020	Y	Y
Dredge & spoil disposal	S7    S21	PFP     NFFP	Cefas https://data.cefas.co.uk/view/407	Active/Open licences to 2022 (data downloaded 6/7/2023)	Y	Y
Cultural & heritage sites/structures (wrecks)	S8    -	PFP    No NFFP	UKHO	2022	Y	Y
Renewable energy – wave (not including cables)	S9    S22	PFP     NFFP	Crown Estate Scotland  The Crown Estate	2022/23	Y	Y
Renewable energy – wind (not including cables)	S10    S23	PFP     NFFP	KIS-ORCA	2022	Y	Y
Renewable energy - tidal (not including cables)	S11    S24	PFP    NFFP	Crown Estate Scotland  The Crown Estate		Y	N
Marine hydrocarbon extraction (not including pipelines)	S12    S25	PFP    NFFP	Oil and Gas Authority	2022	Y	Y
Recreational activities (anchorages only)	S13    -	PFP     No NFFP	Defra Marine Reference Data (UKHO)	2022	Y	Y
Submarine pipeline operations	S14    S26	PFP     NFFP	Oil and Gas Authority	2022	Y	Y
Submarine cable operations (communications and power)	S15    S27	PFP     NFFP	KIS-ORCA	2022	Y	Y

Potential Physical Loss indicator 

Activity layers and HSMs were then overlaid to determine the extent of overlap between the predicted distribution of each habitat and activity. Activity layers identified to contribute to PPL within the assessed habitat and the relevant HSM were imported into QGIS v3.16 (QGIS.org, 2020) and intersected using the ‘Intersection’ tool to identify areas of overlap between the activity and predicted suitable habitat. The ‘Fix Geometries’ tool was used to resolve any invalid geometries on the intersected layers prior to further processing. 

Intersected HSM and activity layers were imported into R version 3.6.1 (R Core Team, 2019) using the ‘sf’ package (Pebesma and others, 2018) and merged to produce a layer with all assessed anthropogenic activity layers intersected with the HSM. Area of suitable habitat per polygon in the HSMs had already been calculated in a previous step, based upon the probability of suitable habitat occurrence multiplied by the area of the polygon. Therefore, the raw area of the footprint of activity did not equate to the area of predicted habitat within the area of the footprint per cell. The area of predicted suitable habitat covered by activities was therefore calculated by multiplying the probability of the habitat suitability by the area of activity footprint. This method assumes homogenous distribution of habitat within each polygon of the HSM but was the most appropriate method to calculate the area of habitat within the activity footprint per polygon of the HSM. 

The response (in this case, change in extent) of horse mussel reefs and eelgrass beds varies between activities due to the different associated pressures. Therefore, the area of PPL caused by each activity was calculated by multiplying the area of predicted habitat within the activity footprint by an activity-specific PPL factor (derived from expert knowledge) to determine the area of predicted suitable habitat lost due to that activity. PPL factors were calculated on a scale of 0 to 1 (where 0 represented no physical loss of habitat, and 1 as complete loss of habitat) (Table 9). The area of predicted habitat lost due to each activity assessed was then estimated and summed to derive the total area of predicted habitat loss. 

Table 9: PPL factors for each activity used in the assessment of the PPL indicator. 

Stratum	Activity Name	PFP / NFFP	Buffer (m)	PPL factor
S1	Coastal defence & land claim protection	PFP	0	1.0000
S2	Coastal docks, ports & marinas	PFP	0	1.0000
S3	Towed bottom-contact fishing	PFP	0	1.0000
S4	Aquaculture	PFP	0	1.0000
S5	Extraction – sand and gravel (active area only)	PFP	0	1.0000
S6	Extraction – navigational dredging (capital & maintenance)	PFP	0	0.5000
S7	Dredge & spoil disposal	PFP	0	0.5000
S8	Cultural & heritage sites/structures (wrecks)	PFP	17.55	1.0000
S9	Renewable energy – wave (not including cables)	PFP	0	0.0009
S10	Renewable energy – wind (not including cables)	PFP	0	0.0010
S11	Renewable energy - tidal (not including cables)	PFP	0	0.0001
S12	Marine hydrocarbon extraction (not including pipelines)	PFP	20.71	1.0000
S13	Recreational activities (anchorages only)	PFP	20	0.0200
S14	Submarine pipeline operations	PFP	0.36	1.0000
S15	Submarine cable operations (communications and power)	PFP	0.03/0.11	0.0300
S16	Coastal defence & land claim protection	NFFP	9.69	1.0000
S17	Coastal docks, ports & marinas	NFFP	9.69	1.0000
S18	Aquaculture	NFFP	1100	1.0000
S19	Extraction – sand and gravel (active area only)	NFFP	1100	0.0500
S20	Extraction – navigational dredging (capital & maintenance)	NFFP	1100	0.0500
S21	Dredge & spoil disposal	NFFP	1100	0.0500
S22	Renewable energy – wave (not including cables)	NFFP	0	0.0500
S23	Renewable energy – wind (not including cables)	NFFP	0	0.0020
S24	Renewable energy - tidal (not including cables)	NFFP	0	0.0034
S25	Marine hydrocarbon extraction (not including pipelines)	NFFP	9.69	1.0000
S26	Submarine pipeline operations	NFFP	0.36	1.0000
S27	Submarine cable operations (communications and power)	NFFP	0.03/0.11	0.0300

Assessment threshold 

The method for calculating the threshold was updated to consider whether there has been any potential physical loss of predicted suitable habitat from polygons where suitable habitat is more likely to be present than not (probability of greater than or equal to 0.5). 

Probability of the presence of suitable habitat was never zero throughout the extent of both HSMs. As a result, a small amount of habitat would be calculated per cell, and the threshold of no PPL would be unachievable using the current methodology without completely removing activities from the extent of the HSMs. The threshold of no PPL was, therefore, only applied to areas where the probability of the presence of suitable habitat was greater than, or equal to 0.5 (i.e., greater than or equal to a 50% likelihood of the presence of suitable habitat) and the HSM was subset to these areas by multiplying the probability of the presence of suitable habitat by the area of the polygon, and then summed for each polygon in the HSM (HSM subset). Importantly, this 0.5 threshold relates only to the HSMs, and the indicator threshold is no PPL when those values overlap with activity data. Therefore, intersected HSM and activity layers were also subset to only include polygons with a probability of greater than, or equal to 0.5 (PPL subset) and the estimated area of PPL was calculated as a percentage of the area of predicted suitable habitat within the HSM subset. 

If the resultant area of PPL (calculated from the PPL subset) was greater than zero, the threshold was considered to have been exceeded. 

Results

Horse mussel reefs 

• The indicator threshold of no potential physical loss caused by anthropogenic activities, where the probability of suitable habitat is greater than or equal to 0.5, has not been met: 

• • UKMS region: 34 km2 (0.23% of potentially suitable habitat with probability ≥ 0.5) potentially lost. 

• • Greater North Sea sub-region: 5 km² (0.56% of potentially suitable habitat with probability ≥ 0.5) potentially lost. 

• • Celtic Seas sub-region: 29 km² (0.21% of potentially suitable habitat with probability ≥ 0.5) potentially lost. 

• Aquaculture (16.30 km²), followed by dredge and spoil disposal (10.70 km²), towed bottom-contact fishing (3.38 km²), and extraction – navigational dredging (1.25 km²) were the activities contributing the greatest area of PPL of horse mussel reefs in polygons with a probability of suitable habitat greater than or equal to 0.5, and therefore contributing to the failure to meet the indicator threshold. All other activities contributed an area of less than 1 km² per activity to PPL in polygons with a probability of suitable habitat greater than or equal to 0.5. 

Eelgrass beds 

• The indicator threshold of no potential physical loss caused by anthropogenic activities, where the probability of suitable habitat is greater than or equal to 0.5, has not been met: 

• • UKMS region: 26 km² (3.00% of potentially suitable habitat with probability ≥ 0.5) potentially lost. 

• • Greater North Sea sub-region: 9km² (5.80% of potentially suitable habitat with probability ≥ 0.5) potentially lost. 

• • Celtic Seas sub-region: 17 km² (2.37% of potentially suitable habitat with probability ≥ 0.5) potentially lost. 

• Aquaculture (14.40 km²) and extraction – navigation dredging (9.92 km²) were the activities contributing the greatest area of PPL of eelgrass beds in polygons with a probability of suitable habitat greater than or equal to 0.5, and therefore contributing to the failure to meet the indicator threshold. All other activities contributed an area of less than 1 km² per activity to PPL in polygons with a probability of greater than or equal to 0.5. 

Further information

UK Marine Strategy (UKMS) Part One 2024 

Horse mussel reefs 

UKMS region 

Across the entire UKMS region, 621.98 km² of the 64,772.85 km² (0.96%) of habitat predicted to be suitable for horse mussel reefs has potentially been lost due to the anthropogenic activities considered within this assessment across the entire UKMS region, in this analysis of the PPL indicator (Figure 8). 

Figure 8: Percentage of predicted suitable habitat for horse mussel reefs, potentially lost due to assessed anthropogenic activities for the UKMS 2024 assessment as a percentage of the total area of predicted suitable habitat, across the entire UKMS region. 

The estimation of the area of predicted suitable habitat potentially lost per activity indicates that towed bottom-contact fishing caused the most potential physical loss (PPL) of suitable habitat for horse mussel reefs across the UKMS region at 350.93 km² (0.54% of total area of horse mussel reef) (Figure 9, Table 10). Towed bottom-contact fishing was followed by dredge and spoil disposal (171.30 km², 0.26%), extraction – navigational dredging (49.54 km², 0.08%), aquaculture (30.21 km², 0.05%) and extraction – sand and gravel (11.23 km², 0.02%), as the most potentially impactful activities (Figure 9, Table 10). Every other activity assessed was found to cause less than 0.01% potential loss of predicted suitable habitat. 

Results were then subset by UKMS sub-region (Greater North Sea, and the Celtic Seas) for UKMS 2024 Part One reporting to account for spatial variability. 

Figure 9: Contribution from assessed anthropogenic activities in the UKMS 2024 assessment to potential physical loss of predicted suitable habitat for horse mussel reefs as a percentage of total area of predicted suitable habitat within the UKMS region. The activity ‘Fishing’ refers to towed bottom-contact fishing. 

Table 10: Assessed anthropogenic activities contributing to the potential physical loss of predicted suitable habitat for horse mussel reefs across the UKMS region for the UKMS 2024 assessment. Activity footprint = raw footprint area of activity; Probability adjusted activity area = area of predicted suitable habitat for horse mussel reefs covered by activity footprint; PPL factor adjusted area = area of potentially suitable habitat lost due to relevant anthropogenic activity; Percentage of suitable habitat area subject to PPL = PPL factor adjusted area as a percentage of the predicted suitable habitat area across the UKMS region. 

Activity Name	Activity footprint (km2)	Probability adjusted activity area (km2)	PPL factor adjusted area (km2)	Percentage of suitable habitat area subject to PPL  (%)
Towed bottom-contact fishing	3048.89	350.93	350.93	0.54
Dredge & spoil disposal	7256.24	592.41	171.3	0.26
Extraction – navigational dredging (capital & maintenance)	2206.28	201.71	49.54	0.08
Aquaculture	88.07	30.21	30.21	0.05
Extraction – sand and gravel (active area only)	1164.91	70.43	11.23	0.02
Submarine pipeline operations	33	2.72	2.72	<0.01
Recreational activities (anchorages only)	524.88	90.82	1.82	<0.01
Renewable energy – wind (not including cables)	5128.09	774.41	1.16	<0.01
Marine hydrocarbon extraction (not including pipelines)	13.48	1.03	1.03	<0.01
Cultural & heritage sites/structures (wrecks)	10.33	0.84	0.84	<0.01
Coastal defence & land claim protection	6.15	0.57	0.57	<0.01
Coastal docks, ports & marinas	3.17	0.49	0.49	<0.01
Renewable energy – wave (not including cables)	23.39	4.63	0.12	<0.01
Renewable energy - tidal (not including cables)	34.11	10.9	0.02	<0.01
Submarine cable operations (communications and power)	3.43	0.57	0.02	<0.01

Greater North Sea 

The total area of predicted suitable habitat for horse mussel reefs within the Greater North Sea sub-region was 23,593.24 km² of which 281.39 km² (1.19%) was potentially lost due to assessed anthropogenic activities (Figure 10). 

Figure 10: Percentage of predicted suitable habitat for horse mussel reefs, potentially lost to assessed anthropogenic activities for the UKMS 2024 assessment as a percentage of the total area of predicted suitable habitat, per UKMS sub-region. 

In contrast to the results at the scale of the UKMS region, the greatest area of PPL was caused by dredge and spoil disposal (150.33 km², 0.64% of suitable habitat for horse mussel reefs in the Greater North Sea) (Figure 11, Table 11). Towed bottom-contact fishing contributed to the second highest overall loss of horse mussel reefs (74.07 km², 0.31%), followed by extraction – navigational dredging (39.72 km², 0.17%), extraction – sand and gravel (8.63 km², 0.04%) and aquaculture (3.21 km², 0.01%) (Figure 11, Table 11). All other activities contributed less than or equal to 0.01% potential loss of habitat within the Greater North Sea UKMS sub-region. 

Table 11: Assessed anthropogenic activities contributing to the potential physical loss of predicted suitable habitat for horse mussel reefs within the Greater North Sea sub-region for the UKMS 2024 assessment. Activity footprint = raw footprint area of activity; Probability adjusted activity area = area of predicted suitable habitat for horse mussel reefs covered by activity footprint; PPL factor adjusted area = area of potentially suitable habitat lost due to relevant anthropogenic activity; Percentage of suitable habitat area subject to PPL = PPL factor adjusted area as a percentage of the predicted suitable habitat area within the Greater North Sea sub-region. 

Activity Name	Activity footprint (km2)	Probability adjusted activity area (km2)	PPL factor adjusted area (km2)	Percentage of suitable habitat area subject to PPL  (%)
Dredge & spoil disposal	6823.73	480.35	150.33	0.64
Towed bottom-contact fishing	757.33	74.07	74.07	0.31
Extraction – navigational dredging (capital & maintenance)	1877.11	149.68	39.72	0.17
Extraction – sand and gravel (active area only)	1036.58	50.32	8.63	0.04
Aquaculture	23.19	3.21	3.21	0.01
Submarine pipeline operations	29.3	1.87	1.87	0.01
Recreational activities (anchorages only)	355.03	62.5	1.25	0.01
Marine hydrocarbon extraction (not including pipelines)	12	0.64	0.64	<0.01
Renewable energy – wind (not including cables)	4185.01	389.84	0.58	<0.01
Cultural & heritage sites/structures (wrecks)	5.62	0.38	0.38	<0.01
Coastal docks, ports & marinas	2.81	0.37	0.37	<0.01
Coastal defence & land claim protection	5.18	0.33	0.33	<0.01
Submarine cable operations (communications and power)	1.37	0.13	0	<0.01
Renewable energy - tidal (not including cables)	0.16	0.12	0	<0.01

Figure 11: Contribution from assessed anthropogenic activities in the UKMS 2024 assessment to potential physical loss of predicted suitable habitat for horse mussel reefs as a percentage of total area of predicted suitable habitat, per UKMS sub-region. The activity ‘Fishing’ refers to towed bottom-contact fishing.

Celtic Seas 

The total predicted area of suitable habitat for horse mussel reefs within the Celtic Seas sub-region was 41,179.61 km², of which 340.40 km² (0.83%) was potentially lost due to assessed anthropogenic activities (Figure 10). 

Towed bottom-contact fishing was found to contribute the most to the PPL of horse mussel reefs in the Celtic Sea sub-region, with a predicted potential loss of 276.86 km² (0.67% of reef area) (Figure 11, Table 12). Towed bottom-contact fishing was followed by aquaculture (27.00 km², 0.07%), dredge and spoil disposal (20.97 km², 0.05%), extraction – navigational dredging (9.82 km², 0.02%), and extraction – sand and gravel (2.60 km², 0.01%) (Figure 11, Table 12 and Table 14). Every other activity assessed caused less than 0.01% of potential loss within the Celtic Seas UKMS sub-region. 

Table 12: Assessed anthropogenic activities contributing to the potential physical loss of predicted suitable habitat for horse mussel reefs within the Celtic Seas sub-region for the UKMS 2024 assessment. Activity footprint = raw footprint area of activity; Probability adjusted activity area = area of predicted suitable habitat for horse mussel reefs covered by activity footprint; PPL factor adjusted area = area of potentially suitable habitat lost due to relevant anthropogenic activity; Percentage of suitable habitat area subject to PPL = PPL factor adjusted area as a percentage of the predicted suitable habitat area within the Celtic Seas sub-region. 

Activity Name	Activity footprint (km2)	Probability adjusted activity area (km2)	PPL factor adjusted area (km2)	Percentage of suitable habitat area subject to PPL  (%)
Towed bottom-contact fishing	2291.56	276.86	276.86	0.67
Aquaculture	64.88	27	27	0.07
Dredge & spoil disposal	432.51	112.06	20.97	0.05
Extraction – navigational dredging (capital & maintenance)	329.17	52.04	9.82	0.02
Extraction – sand and gravel (active area only)	128.33	20.11	2.6	0.01
Submarine pipeline operations	3.69	0.84	0.84	<0.01
Renewable energy – wind (not including cables)	943.08	384.57	0.58	<0.01
Recreational activities (anchorages only)	169.85	28.31	0.57	<0.01
Cultural & heritage sites/structures (wrecks)	4.71	0.46	0.46	<0.01
Marine hydrocarbon extraction (not including pipelines)	1.47	0.39	0.39	<0.01
Coastal defence & land claim protection	0.97	0.24	0.24	<0.01
Coastal docks, ports & marinas	0.35	0.12	0.12	<0.01
Renewable energy – wave (not including cables)	23.39	4.63	0.12	<0.01
Renewable energy - tidal (not including cables)	33.96	10.79	0.02	<0.01
Submarine cable operations (communications and power)	2.06	0.44	0.01	<0.01

Horse mussel reefs UKMS 2024 Assessment Threshold 

The results below relate to the HSM subset (Figure 12) and the PPL subset (Figure 13). As the maps are at UK scale, smaller areas of PPL may be difficult to identify due to the resolution of the image.  

UKMS region 

Predicted suitable habitat in the HSM subset was predominantly located within the Irish Sea, Firth of Clyde, Strangford Lough, the Hebrides, Shetland, off the coast of John o’ Groats, Peterhead, Stonehaven and Bamburgh (Figure 12). The total area of predicted suitable habitat from the HSM subset for horse mussel reefs across the UKMS region was 14,916.98 km², which represents 23.03% of the area of predicted suitable habitat estimated from the unrestricted HSM (i.e., the HSM not restricted to probabilities greater than or equal to 0.5). 

Figure 12: Mean predictive values of habitat suitability for horse mussel reefs (from the habitat suitability map for both UKMS 2024 and the UKMS 2018 re-assessment) where probability was ≥ 0.5, across the UKMS region with transitional waters removed. 

PPL in polygons with a probability of predicted suitable habitat of greater than, or equal to, 0.5 was identified off the coast of Barrow-in-Furness, the coast of Northern Ireland, the Firth of Clyde, and Peterhead (Figure 13). The area of PPL calculated from the PPL subset was 34.02 km², 0.23% of the area of predicted suitable habitat calculated from the HSM subset. Therefore, the threshold of no PPL in grid cells with a probability of suitable habitat greater than, or equal to 0.5, was exceeded and the target was not met for the UKMS region.

Figure 13: Mean predictive values of habitat suitability for horse mussel reefs where probability was ≥ 0.5, across the UKMS region with transitional waters removed, restricted to where assessed anthropogenic activities that cause PPL occur within the UKMS 2024 assessment. 

The activities contributing the greatest area of PPL of horse mussel reefs in polygons with a probability of suitable habitat greater than or equal to 0.5, and therefore contributing to the failure to meet the target for the UKMS region for the UKMS 2024 assessment were: aquaculture (16.30 km²), followed by dredge and spoil disposal (10.70 km²), towed bottom-contact fishing (3.38 km²), and extraction – navigational dredging (1.25 km²). All other activities contributed an area of less than 1 km² per activity to PPL in polygons with a probability of greater than or equal to 0.5. 

Assessments at the UKMS sub-region scale were then analysed to determine if the threshold had been met for the Greater North Sea and the Celtic Seas. 

Greater North Sea 

The area of predicted suitable habitat from the HSM subset for horse mussel reefs (Figure 12) within the Greater North Sea sub-region totalled 946.92 km², 4.01% of the area of predicted suitable habitat within the sub-region estimated from the unrestricted HSM. The area of PPL of predicted suitable habitat from the Greater North Sea sub-region due to assessed anthropogenic activities in the PPL subset (Figure 13) was 5.29 km², 0.56% of predicted suitable habitat from the HSM subset. Therefore, the threshold of no PPL in grid cells with a probability of suitable habitat greater than, or equal to, 0.5 was exceeded and the target was not met for the Greater North Sea sub-region. 

The activities contributing the greatest area of PPL of horse mussel reefs in polygons with a probability of suitable habitat greater than or equal to 0.5, and therefore contributing to the failure to meet the target for the Greater North Sea sub-region for the UKMS 2024 assessment were: dredge and spoil disposal (2.74 km²), followed by aquaculture (2.24 km²). All other activities contributed an area of less than 1 km² per activity to PPL in polygons with a probability of greater than or equal to 0.5. 

Celtic Seas 

The area of predicted suitable habitat from the HSM subset for horse mussel reefs (Figure 12) within the Celtic Seas sub-region totalled 13,970.06 km², 33.92% of the area of predicted suitable habitat within the sub-region estimated from the unrestricted HSM. The area of PPL of predicted suitable habitat from the Celtic Seas sub-region due to assessed anthropogenic activities in the PPL subset (Figure 13) was 28.72 km², 0.21% of predicted suitable habitat from the HSM subset. Therefore, the threshold of no PPL in grid cells with a probability of suitable habitat greater than, or equal to, 0.5 was not met for the Celtic Seas sub-region. 

The activities contributing the greatest area of PPL of horse mussel reefs in polygons with a probability of suitable habitat greater than or equal to 0.5, and therefore contributing to the failure to meet the target for the Celtic Seas sub-region for the UKMS 2024 assessment were: aquaculture (14.10 km²), dredge and spoil disposal (7.99 km²), towed-bottom contact fishing (3.38 km²) and extraction – navigational dredging (1.16 km²). All other activities contributed an area of less than 1 km² per activity to PPL in polygons with a probability of greater than or equal to 0.5. 

Eelgrass beds 

UKMS region 

As previously stated, the primary output of the PPL indicator is an assessment of the estimated amount of predicted suitable habitat potentially lost as a result of assessed activities. Across the entire UKMS region, 43.85 km² of the 2277.54 km² (1.93%) of habitat predicted to be suitable for eelgrass beds has potentially been lost as a result of the aforementioned (see Activity Data) activities considered by this assessment (Figure 14). 

Figure 14: Percentage of predicted suitable habitat for eelgrass beds potentially lost to assessed anthropogenic activities for the UKMS 2024 assessment as a percentage of the total area of predicted suitable habitat, across the entire UKMS region. 

Aquaculture and extraction-navigational dredging were the two activities that caused the greatest PPL of predicted suitable habitat for eelgrass beds across the UKMS region at 18.79 km² and 18.73 km² (0.83% and 0.82% of total area of eelgrass beds), respectively (Figure 15 and Table 13). Aquaculture and extraction – navigational dredging, were followed by dredge and spoil disposal (3.50 km², 0.15%), coastal docks, ports and marinas (1.33 km², 0.06%), and coastal defence and land claim protection (0.75 km², 0.03%) as the activities causing the greatest PPL (Figure 15 and Table 13). All other activities assessed were found to cause less than or equal to 0.02% potential loss of predicted suitable habitat. 

Table 13: Assessed anthropogenic activities contributing to the potential physical loss of predicted suitable habitat for eelgrass beds across the UKMS region for the UKMS 2024 assessment. Activity footprint = raw footprint area of activity; Probability adjusted activity area = area of predicted suitable habitat for horse mussel reefs covered by activity footprint; PPL factor adjusted area = area of potentially suitable habitat lost due to relevant anthropogenic activity; Percentage of suitable habitat area subject to PPL = PPL factor adjusted area as a percentage of the predicted suitable habitat area across the UKMS region. 

Activity Name	Activity footprint (km2)	Probability adjusted activity area (km2)	PPL factor adjusted area (km2)	Percentage of suitable habitat area subject to PPL  (%)
Aquaculture	33.91	18.79	18.79	0.83
Extraction - navigational dredging (capital & maintenance)	447.43	116.51	18.73	0.82
Dredge & spoil disposal	234.16	36.97	3.5	0.15
Coastal docks, ports & marinas	2.49	1.33	1.33	0.06
Coastal defence & land claim protection	3.87	0.75	0.75	0.03
Recreational activities (anchorages only)	58.65	20.15	0.4	0.02
Cultural & heritage sites/structures (wrecks)	1.42	0.32	0.32	0.01
Submarine pipeline operations	0.3	0.02	0.02	<0.01
Renewable energy - wind (not including cables)	29.41	1.23	0	<0.01
Renewable energy - wave (not including cables)	0.46	0.05	0	<0.01
Submarine cable operations (communications and power)	0.11	0.04	0	<0.01
Marine hydrocarbon extraction (not including pipelines)	0.01	0	0	<0.01

Figure 15: Contribution from assessed anthropogenic activities in the UKMS 2024 assessment to potential physical loss of predicted suitable habitat for eelgrass beds as a percentage of total area of predicted suitable habitat within the UKMS region. 

Results were then assessed at the UKMS sub-region scale (Greater North Sea and the Celtic Seas) for UKMS 2024 Part One reporting to account for spatial variability. 

Greater North Sea sub-region 

The area of predicted suitable habitat for eelgrass beds within the Greater North Sea sub-region totalled 624.63 km², of which 19.31 km² (3.09% of bed area) was potentially lost due to assessed anthropogenic activities (PPL, Figure 16). 

Extraction - navigational dredging was found to contribute the greatest amount of PPL to eelgrass beds in the Greater North Sea sub-region, at 11.63 km² (1.86% of bed area) (Figure 17 and Table 14). By amount of PPL, extraction – navigation dredging was followed by dredge and spoil disposal (3.10 km², 0.50%), aquaculture (2.58 km², 0.41%), coastal docks, ports and marinas (1.18 km², 0.19%) and coastal defence and land claim protection (0.37 km², 0.06%) (Figure 17 and Table 14). 

Figure 16: Percentage of predicted suitable habitat for eelgrass beds potentially lost to assessed anthropogenic activities for the UKMS 2024 assessment as a percentage of the total area of predicted suitable habitat, per UKMS sub-region. 

Figure 17: Contribution from assessed anthropogenic activities in the UKMS 2024 assessment to potential physical loss of predicted suitable habitat for eelgrass beds as a percentage of total area of predicted suitable habitat, per UKMS sub-region. 

Table 14: Assessed anthropogenic activities contributing to the potential physical loss of predicted suitable habitat for eelgrass beds within the Greater North Sea sub-region for the UKMS 2024 assessment. Activity footprint = raw footprint area of activity; Probability adjusted activity area = area of predicted suitable habitat for horse mussel reefs covered by activity footprint; PPL factor adjusted area = area of potentially suitable habitat lost due to relevant anthropogenic activity; Percentage of suitable habitat area subject to PPL = PPL factor adjusted area as a percentage of the predicted suitable habitat area within the Greater North Sea sub-region. 

Activity Name	Activity footprint (km2)	Probability adjusted activity area (km2)	PPL factor adjusted area (km2)	Percentage of suitable habitat area subject to PPL  (%)
Extraction - navigational dredging (capital & maintenance)	265.98	61.47	11.63	1.86
Dredge & spoil disposal	203.91	30.78	3.10	0.50
Aquaculture	7.11	2.58	2.58	0.41
Coastal docks, ports & marinas	2.21	1.18	1.18	0.19
Coastal defence & land claim protection	3.12	0.37	0.37	0.06
Recreational activities (anchorages only)	42.69	11.53	0.23	0.04
Cultural & heritage sites/structures (wrecks)	1.07	0.19	0.19	0.03
Submarine pipeline operations	0.24	0.01	0.01	<0.01
Renewable energy - wind (not including cables)	29.41	1.23	0	<0.01
Marine hydrocarbon extraction (not including pipelines)	0.01	0	0	<0.01
Submarine cable operations (communications and power)	0.04	0.01	0	<0.01

Celtic Seas 

The predicted area of suitable habitat for eelgrass beds within the Celtic Seas sub-region totalled 1652.91 km², of which 24.55 km² (1.48% of bed area) was potentially lost due to assessed anthropogenic activities (Figure 16). 

As with the assessment at UKMS region scale, aquaculture was found to contribute the most to the PPL of eelgrass bed in the Celtic Sea sub-region, predicting a potential loss of 16.21 km² (0.98% of bed area) (Figure 17 and Table 15). By amount of PPL, aquaculture was followed by extraction – navigational dredging (7.10 km2, 0.43%), dredge and spoil disposal (0.40 km², 0.02%), coastal defence and land claim protection (0.38 km², 0.02%), and all other activities assessed caused equal to or less than 0.01% potential loss within the Celtic Seas UKMS sub-region (Figure 17 and Table 15). 

Table 15: Assessed anthropogenic activities contributing to the potential physical loss of predicted suitable habitat for eelgrass beds within the Celtic Seas sub-region for the UKMS 2024 assessment. Activity footprint = raw footprint area of activity; Probability adjusted activity area = area of predicted suitable habitat for horse mussel reefs covered by activity footprint; PPL factor adjusted area = area of potentially suitable habitat lost due to relevant anthropogenic activity; Percentage of suitable habitat area subject to PPL = PPL factor adjusted area as a percentage of the predicted suitable habitat area within the Celtic Seas sub-region. 

Activity Name	Activity footprint (km2)	Probability adjusted activity area (km2)	PPL factor adjusted area (km2)	Percentage of suitable habitat area subject to PPL  (%)
Aquaculture	26.79	16.21	16.21	0.98
Extraction - navigational dredging (capital & maintenance)	181.45	55.04	7.10	0.43
Dredge & spoil disposal	30.25	6.2	0.40	0.02
Coastal defence & land claim protection	0.74	0.38	0.38	0.02
Recreational activities (anchorages only)	15.96	8.62	0.17	0.01
Coastal docks, ports & marinas	0.28	0.15	0.15	0.01
Cultural & heritage sites/structures (wrecks)	0.35	0.13	0.13	0.01
Submarine pipeline operations	0.06	0.01	0.01	<0.01
Renewable energy - wave (not including cables)	0.46	0.05	0	<0.01
Submarine cable operations (communications and power)	0.07	0.03	0	<0.01

Eelgrass beds UKMS 2024 Assessment Threshold 

The results below relate to the HSM subset (Figure 18) and the PPL subset (Figure 19). As the maps are at UK scale, smaller areas of PPL may be difficult to identify due to the resolution of the image. 

UKMS region 

Predicted suitable habitat for eelgrass beds in the HSM subset was predominantly located within Loch Ryan and the Hebrides in Scotland, the Llŷn Peninsula and Carmarthen Bay in Wales, and Southampton, Portsmouth, and the Isle of Wight in England (Figure 18). The area of predicted suitable habitat from the HSM subset for eelgrass beds (Figure 18) across the UKMS region totalled 878.67 km², 38.58% of the area of predicted suitable habitat estimated from the unrestricted HSM. 

Figure 18: Mean predictive values of habitat suitability for eelgrass beds (from the habitat suitability map for both the UKMS 2024 and UKMS 2018 re-assessment) where probability was ≥ 0.5, across the UKMS region with transitional waters removed. 

PPL in polygons with a probability of predicted suitable habitat of greater than, or equal to, 0.5 was identified in the Solent and off the coast of Belfast (Figure 19). The area of PPL calculated from the PPL subset was 26.40 km², 3.00% of the area of predicted suitable habitat calculated from the HSM subset. Therefore, the threshold of no PPL in grid cells with a probability of suitable habitat greater than, or equal to 0.5, was exceeded and the target was not met for the UKMS region. 

Figure 19: Mean predictive values of habitat suitability for eelgrass beds where probability was ≥ 0.5, across the UKMS region with transitional waters removed, restricted to where assessed anthropogenic activities that cause PPL occur. 

The activities contributing the greatest area of PPL of eelgrass beds in polygons with a probability of suitable habitat greater than or equal to 0.5, and therefore contributing to the failure to meet the target for the UKMS region for the UKMS 2024 assessment, were predominantly aquaculture (14.40 km²) and extraction – navigational dredging (9.92 km²). All other activities contributed an area of less than 1 km² per activity to PPL in polygons with a probability of greater than or equal to 0.5. 

Assessments at the UKMS sub-region scale were then analysed to determine if the threshold had been met for the Greater North Sea and the Celtic Seas. 

Greater North Sea 

The area of predicted suitable habitat from the HSM subset for eelgrass beds (Figure 12) within the Greater North Sea sub-region totalled 162.86 km², 26.07% of the area of predicted suitable habitat within the sub-region estimated from the unrestricted HSM. The area of PPL of predicted suitable habitat from the Greater North Sea sub-region due to assessed anthropogenic activities in the PPL subset (Figure 13) was 9.44 km², 5.80% of predicted suitable habitat from the HSM subset. Therefore, the threshold of no PPL in grid cells with a probability of suitable habitat greater than, or equal to, 0.5 was exceeded and the target was not met for the Greater North Sea sub-region. 

The activity contributing the greatest area of PPL of eelgrass beds in polygons with a probability of suitable habitat greater than or equal to 0.5, and therefore contributing to the failure to meet the target for the Greater North Sea sub-region for the UKMS 2024 assessment was predominantly extraction – navigational dredging (7.29 km²). All other activities contributed an area of less than 1 km² per activity to PPL in polygons with a probability of greater than or equal to 0.5. 

Celtic Seas 

The area of predicted suitable habitat from the HSM subset for eelgrass beds (Figure 18) within the Celtic Seas sub-region totalled 715.86 km², 43.31% of the area of predicted suitable habitat within the sub-region estimated from the unrestricted HSM. The area of PPL of predicted suitable habitat from the Celtic Seas sub-region due to assessed anthropogenic activities in the PPL subset (Figure 19) was 16.96 km², 2.37% of predicted suitable habitat from the HSM subset. Therefore, the threshold of no PPL in grid cells with a probability of suitable habitat greater than, or equal to, 0.5 was exceeded and the target was not met for the Celtic Seas sub-region. 

The activities contributing the greatest area of PPL of eelgrass beds in polygons with a probability of suitable habitat greater than or equal to 0.5, and therefore contributing to the failure to meet the target for the Celtic Seas sub-region for the UKMS 2024 assessment were predominantly aquaculture (13.60 km²), followed by extraction – navigational dredging (2.63 km²). All other activities contributed an area of less than 1 km² per activity to PPL in polygons with a probability of greater than or equal to 0.5. 

UKMS Part One 2018 PPL re-assessment 

To facilitate direct comparison between the previous UKMS 2018 Part One PPL indicator results (Strong and others, 2018) and those presented above, it was necessary to re-assess the PPL indicator (using the current methodology) using updated HSMs from the UKMS 2024 assessment and the activity data used in the UKMS 2018 assessment. The comparison of results from the UKMS 2018 re-assessment with the original UKMS 2018 assessment for both horse mussel reefs and eelgrass beds is presented below followed by the comparison of results between the UKMS 2024 assessment and the updated UKMS 2018 assessment. 

Comparison of results from UKMS 2018 re-assessment to original UKMS 2018 assessment 

Horse mussel reefs 

UKMS region 

In the original UKMS 2018 assessment it was estimated that 0.47% (39.31 km2) of the horse mussel reefs in UK waters had been potentially lost due to assessed anthropogenic activities up to and including the 2010 – 2016 assessment period (Strong and others, 2018). On recalculation of the PPL indicator using activity data from the UKMS 2018 assessment with the updated HSMs used in the current assessment, it was estimated that 0.30% (195.77 km²) of horse mussel reefs were potentially lost due to anthropogenic activities (Figure 20). Therefore, the estimated values of PPL between the original and re-assessment of the UKMS 2018 assessment were different. Whilst area of predicted suitable habitat for horse mussel reefs in km² lost to PPL has increased, when expressed as a percentage of predicted suitable habitat, the value has decreased. 

The HSMs represent the only change in data between the original published, and re-assessment of the PPL indicator in the UKMS 2018 assessment. Therefore, the difference in PPL could be attributed to the differences between HSMs and associated estimated areas of predicted suitable habitat. The HSM used in the UKMS 2018 assessment, denoted predicted suitable habitat as either present or absent based on the intersection of two separate HSMs, created using a general additive model and random forest algorithm. Where present, the entire area of predicted suitable habitat (and therefore footprint area of any overlaid activity) was considered as 100% predicted suitable habitat. However, with the updated HSM used in this assessment, the presence of predicted suitable habitat within a polygon of the HSM is indicated by a probability, which is then multiplied by the area of the polygon to estimate the predicted area of suitable habitat. Therefore, there is a difference between the raw area of the polygon (or raw area of the footprint of activity) and the area of the habitat. The updated HSM presented in this assessment followed recommended approaches for area calculation (Calabrese and others, 2014; Castle and others, 2022). However, no cells across the extent of the HSM had a probability of suitable habitat of zero. Therefore, activities contributed to the PPL of predicted suitable habitat even in areas where the probability of occurrence is very low. When summed up across the entire extent of the HSM, the small amount of PPL from activities in areas with low probability of suitable habitat accumulated. In the HSMs used in the original 2018 assessment, the habitat would have been denoted as absent, and therefore there would be zero PPL. 

The calculation of total area of predicted suitable habitat for horse mussel reefs also explains differences between the estimates of suitable habitat area in the UKMS 2018 assessment (8,304 km²), which is smaller than the area of suitable habitat estimated by the updated HSMs used in the UKMS 2024 assessment (64,773 km²). Due to the larger extent of the updated HSMs there is a greater chance of intersect with activities (even when probability of suitable habitat is low) than with the HSM used in the original UKMS 2018 assessment. Any intersect between activity area and the HSM used in the UKMS 2018 assessment will therefore have a greater weighting on percentage of total habitat area than with the updated HSMs. This accounts for the larger percentage area of PPL in the original UKMS 2018 assessment than in the UKMS 2018 re-assessment. For these reasons, we cannot directly compare percentages of predicted habitat lost due to assessed activities between the original UKMS 2018 results and the re-assessment with the updated HSM. However, comparisons on the rank order of contribution of PPL per activity may be made. 

The main activities that contributed to PPL in the original published UKMS 2018 assessment, in decreasing order of the amount of PPL for horse mussel reefs were: dredge and spoil disposal, aquaculture, submarine cable operations, wrecks, and coastal development (Table 16). From the re-assessment of the indicator using updated HSMs and the original 2018 activity data, towed bottom-contact fishing accounted for the greatest amount of PPL across the entire UKMS assessment area, followed by dredge and spoil disposal, extraction - sand and gravel, aquaculture, and extraction – navigational dredging (Table 16 and Figure 21). 

Table 16: Rank order of assessed activities contributing the greatest area of PPL to potentially suitable habitat for horse mussel reefs, compared between the original UKMS 2018 results and the UKMS 2018 re-assessment, using the activity data from 2018 and the updated HSM presented in this assessment. Fishing* refers to towed bottom-contact fishing. 

	Published UKMS 2018 assessment	Re-assessment of UKMS 2018 with original activity data and updated HSM
Area of horse mussel reef	8,304 km2	64,772.85 km2
Area of PPL due to assessed anthropogenic activities	39.31 km2 (0.47%)	195.77 km2 (0.30%)
Rank order of assessed anthropogenic activities contributing to PPL
1	Dredge & spoil disposal	Fishing*
2	Aquaculture	Dredge & spoil disposal
3	Submarine cable operations	Extraction – sand and gravel (active area only)
4	Wrecks	Aquaculture
5	Coastal Development	Extraction – navigational dredging (capital & maintenance)

Dredge and spoil disposal and aquaculture activities are within the top five activities contributing towards PPL of suitable habitat for horse mussel reefs for both the original published 2018 assessment and the UKMS 2018 re-assessment (Table 16). However, contribution of PPL from submarine cable operations, wrecks, and coastal development decreased, and towed bottom-contact fishing and extraction activities increased in the UKMS 2018 re-assessment. The difference in the contribution of activities to PPL was attributed to the updated HSMs. For example, towed bottom-contact fishing was relatively widespread in comparison with submarine cables, wrecks and coastal development, and covered larger areas in which the probability of predicted suitable habitat for horse mussel reef occurrence may have been low but still contributed towards the total area of habitat. This area would not have been considered using the HSM from the UKMS 2018 assessment, where habitat was marked as either present or absent based on probability. Additionally, this will have reduced the percentage of habitat lost due to smaller scale activities, such as submarine cable operations, as the habitat would no longer be classed as 100% present as a result of using probability to calculate the area of habitat. 

Eelgrass beds 

UKMS region 

In the original UKMS 2018 assessment it was estimated that 2% (32.32 km²) of the eelgrass beds in UK waters had been potentially lost due to anthropogenic activities up to and including the 2010 - 2016 assessment period (Strong and others, 2018). On recalculation of the PPL indicator using activity data from the UKMS 2018 assessment with the updated HSMs used in the current assessment, it was estimated that 1.45% (33.06 km²) of eelgrass beds were potentially lost due to anthropogenic activities (Figure 25). Therefore, there was an increase in the area of PPL due to assessed anthropogenic activities between the original UKMS 2018 assessment and this re-assessment, but a decrease in area as expressed as a percentage of the total potentially suitable habitat as estimated from the HSMs. 

As with the re-assessment of the horse mussel reef HSM with UKMS 2018 activity data, the only thing to have changed is the HSM, therefore any change in PPL values must derive from the HSM and associated estimated areas of predicted suitable habitat. Area of PPL in km² in both the original UKMS 2018 assessment and the re-assessment are similar (~32-33 km²); this indicates a high level of correlation in eelgrass bed presence and extent between the two models. However, expressed as a percentage, the area of PPL decreases between the original UKMS 2018 assessment and this re-assessment. As previously stated, we cannot directly compare the contribution of PPL from each activity between the original UKMS 2018 results and the re-assessment with the updated HSM. However, comparisons on the rank order of PPL per activity may be made.  

The main activities that contributed PPL in the original published UKMS 2018 assessment, in decreasing order, for eelgrass beds were: aquaculture, extraction - navigational dredging, extraction - aggregates, dredge and spoil disposal, and coastal docks, ports, marinas (Table 17). From the re-assessment of the indicator using updated HSMs and the original 2018 activity data, aquaculture accounted for the greatest PPL of eelgrass beds across the entire UKMS region, followed by extraction - navigational dredging, coastal docks, ports and marinas, dredge and spoil disposal, and coastal defence and land claim protection (Table 17, Table 24, and Figure 26). 

Table 17: Rank order of assessed activities contributing the greatest area of PPL to potentially suitable habitat for eelgrass beds, compared between the original UKMS 2018 results and the UKMS 2018 re-assessment, using the activity data from 2018 and the updated HSM presented in this assessment. 

	Published UKMS 2018 indicator assessment	Re-assessment of UKMS 2018 with original activity data and updated HSM
Area of eelgrass bed	1,583 km2	2,277.54 km2
Area of PPL due to assessed anthropogenic activities	32.32 km2 (2%)	33.06 km2 (1.45%)
Rank order of assessed anthropogenic activities contributing to PPL
1	Aquaculture	Aquaculture
2	Extraction - navigational dredging	Extraction - navigational dredging
3	Extraction - aggregates	Coastal docks, ports, marinas
4	Dredge and Spoil disposal	Dredge and Spoil disposal
5	Coastal docks, ports, marinas	Coastal defence and land claim protection

Aquaculture, extraction - navigational dredging, dredge and spoil disposal, and coastal docks, ports and marinas were within the top five activities contributing towards the predicted potential loss of eelgrass beds for both the original published 2018 assessment and the re-assessment (Table 17 and Table 24). However, contribution of PPL due to extraction - aggregates decreased, and coastal defence and land claim protection increased in the UKMS 2018 re-assessment. The differences in the contribution of activities to PPL of predicted suitable habitat for eelgrass was attributed to the updated HSMs. Extraction – aggregates for example, predominantly occurs further than 5 km from the coastline, and therefore beyond the extent of the eelgrass beds HSM, thereby reducing the area of PPL from this activity. Coastal docks, ports, marinas; and coastal defence and land claim protection activities are primarily located along the coastline, over which the updated HSM predicts higher probability of suitable eelgrass bed habitat. When combined with a smaller extent of the HSM, this increases the proportion of area over which these activities cause PPL. 

UKMS 2024 vs. Updated UKMS 2018 Comparison 

Horse mussel reefs 

UKMS region 

The area of PPL due to assessed activities across the UKMS region in the re-assessment of the UKMS 2018 activity data was 195.77 km² (0.30% of reef area), and for the UKMS 2024 assessment was 621.98 km² (0.96% of reef area), over three times as much (Figure 8, and Figure 20). Therefore, the PPL of predicted suitable habitat for horse mussel reefs has continued and increased since the last UKMS assessment. 

Figure 20: Percentage of predicted suitable habitat for horse mussel reefs potentially lost to assessed anthropogenic activities for the UKMS 2018 re-assessment as a percentage of the total area of predicted suitable habitat, across the entire UKMS region. 

Comparisons of the contribution of assessed anthropogenic activities to PPL were also made between the UKMS 2018 re-assessment and the UKMS 2024 assessment. In both assessments, towed bottom-contact fishing contributed the greatest amount of PPL of predicted suitable habitat for horse mussel reefs across the UKMS region (UKMS 2018 re-assessment = 0.30%, UKMS 2024 = 0.54% of reef area) (Table 22, Table 10, Table 18, and Table 19). For both assessments, towed bottom-contact fishing was broadly followed by the same activities, although the ranked positions of extraction – sand and gravel and extraction navigational dredging switched (Table 19). Notably, the same anthropogenic activities caused the greatest PPL of predicted suitable habitat for horse mussel reefs over both assessments. 

Figure 21: Contribution from assessed anthropogenic activities in the UKMS 2018 re-assessment to potential physical loss of predicted suitable habitat for horse mussel reefs as a percentage of total area of predicted suitable habitat within the UKMS region. The activity ‘Fishing’ refers to towed bottom-contact fishing. 

Table 18: Assessed anthropogenic activities contributing to the potential physical loss of predicted suitable habitat for horse mussel reefs across the UKMS region for the UKMS 2018 re-assessment. Activity footprint = raw footprint area of activity; Probability adjusted activity area = area of predicted suitable habitat for horse mussel reefs covered by activity footprint; PPL factor adjusted area = area of potentially suitable habitat lost due to relevant anthropogenic activity; Percentage of suitable habitat area subject to PPL = PPL factor adjusted area as a percentage of the predicted suitable habitat area across the UKMS region. 

Activity Name	Activity footprint (km2)	Probability adjusted activity area (km2)	PPL factor adjusted area (km2)	Percentage of suitable habitat area subject to PPL  (%)
Towed bottom-contact fishing	1417.49	104.63	104.63	0.16
Dredge & spoil disposal	312.51	72.03	33.24	0.05
Extraction – sand and gravel (active area only)	1530.38	120.29	24.47	0.04
Aquaculture	51.13	23.74	23.74	0.04
Extraction – navigational dredging (capital & maintenance)	238.48	21.78	3.76	0.01
Submarine cable operations (communications and power)	337.52	49.88	1.5	<0.01
Recreational activities (anchorages only)	372.98	54.07	1.08	<0.01
Cultural & heritage sites/structures (wrecks)	6.76	1.03	1.03	<0.01
Marine hydrocarbon extraction (not including pipelines)	9.13	0.72	0.72	<0.01
Coastal docks, ports & marinas	3.29	0.56	0.56	<0.01
Coastal defence & land claim protection	5.83	0.52	0.52	<0.01
Submarine pipeline operations	3.98	0.39	0.39	<0.01
Renewable energy – wind (not including cables)	875.51	109.56	0.11	<0.01
Renewable energy – wave (not including cables)	100.9	13.31	0.01	<0.01
Renewable energy - tidal (not including cables)	14.41	4.68	0	<0.01

Table 19: Comparison of area of horse mussel reefs across the UKMS region subject to PPL from assessed anthropogenic activities and the rank order of activities contributing to PPL, between the re-assessment of the UKMS 2018 PPL indicator with original activity data and updated HSMs, and the UKMS 2024 assessment of the PPL indicator. Fishing* refers to towed bottom-contact fishing. 

Item		UKMS 2018 Re-assessment	UKMS 2024
Area of PPL of horse mussel reefs		195.77 km2 (0.30%)	621.98 km2 (0.96%)
Ranked activity by contribution to PPL	1st	Fishing* (0.16%)	Fishing* (0.54%)
	2nd	Dredge & spoil disposal (0.05%)	Dredge & spoil disposal (0.26%)
	3rd	Extraction – sand and gravel (0.04%)	Extraction – navigational dredging (0.08%)
	4th	Aquaculture (0.04%)	Aquaculture (0.05%)
	5th	Extraction – navigational dredging (0.01%)	Extraction – sand and gravel (0.02%)

Greater North Sea sub-region 

Within the Greater North Sea sub-region, re-assessment of the UKMS 2018 PPL indicator using the updated HSM and the original UKMS 2018 activity data estimated the area of PPL as 0.25% (58.97 km²) of the 23,593 km² of suitable habitat for horse mussel reefs (Figure 22). The area of PPL in the Greater North Sea sub-region in the UKMS 2018 re-assessment was less than a quarter of that in the UKMS 2024 assessment (281.39 km²). Therefore, PPL of predicted suitable habitat for horse mussel reefs from assessed anthropogenic activities has continued in the Greater North Sea sub-region and has also increased. 

Figure 22: Percentage of predicted suitable habitat for horse mussel reefs, potentially lost to assessed anthropogenic activities for the UKMS 2018 re-assessment as a percentage of the total area of predicted suitable habitat, per UKMS sub-region. 

From re-assessment of the UKMS 2018 PPL indicator with updated HSMs, the main activities contributing to PPL of predicted suitable habitat for horse mussel reefs in the Greater North Sea sub-region were: towed bottom-contact fishing (0.11% of reef area), followed by extraction – sand and gravel (0.07%), dredge and spoil disposal (0.04%), aquaculture (0.01%), and extraction – navigational dredging (0.01%) (Table 20, Table 21, and Figure 23). Ranked anthropogenic activities by contribution to PPL of horse mussel reefs from the UKMS 2018 re-assessment differed from the UKMS 2024 assessment. In the UKMS 2024 assessment, the area of PPL was greatest from dredge and spoil disposal (0.64% of reef area within the Greater North-Sea sub-region) (Table 11 and Figure 11). Towed bottom-contact fishing was second to dredge and spoil disposal in the UKMS 2024 assessment (0.31%) (Table 11 and Figure 11). The same five anthropogenic activities contributed the greatest area of PPL of predicted suitable habitat for horse mussel reefs for both assessments. 

Table 20: Comparison of area of horse mussel reefs within the Greater North Sea sub-region subject to PPL from assessed anthropogenic activities and the rank order of activities contributing to PPL, between the re-assessment of the UKMS 2018 PPL indicator with original activity data and updated HSMs, and the UKMS 2024 assessment of the PPL indicator. Fishing* refers to towed bottom-contact fishing. 

Item		UKMS 2018 Re-assessment	UKMS 2024
Area of PPL of horse mussel reefs		58.97 km2 (0.25%)	281.39 km2 (1.19%)
Ranked activity by contribution to PPL	1st	Fishing* (0.11%)	Dredge & spoil disposal (0.64%)
	2nd	Extraction – sand and gravel (0.07%)	Fishing* (0.31%)
	3rd	Dredge and spoil disposal (0.04%)	Extraction – navigational dredging (0.17%)
	4th	Aquaculture (0.01%)	Extraction – sand and gravel (0.04%)
	5th	Extraction – navigational dredging (0.01%)	Aquaculture (0.01%)

Figure 23: Contribution from assessed anthropogenic activities in the UKMS 2018 re-assessment to potential physical loss of predicted suitable habitat for horse mussel reefs as a percentage of total area of predicted suitable habitat, per UKMS sub-region. The activity ‘Fishing’ refers to towed bottom-contact fishing. 

Table 21:  Assessed anthropogenic activities contributing to the potential physical loss of predicted suitable habitat for horse mussel reefs within the Greater North Sea sub-region for the UKMS 2018 re-assessment. Activity footprint = raw footprint area of activity; Probability adjusted activity area = area of predicted suitable habitat for horse mussel reefs covered by activity footprint; PPL factor adjusted area = area of potentially suitable habitat lost due to relevant anthropogenic activity; Percentage of suitable habitat area subject to PPL = PPL factor adjusted area as a percentage of the predicted suitable habitat area within the Greater North Sea sub-region. 

Activity Name	Activity footprint (km2)	Probability adjusted activity area (km2)	PPL factor adjusted area (km2)	Percentage of suitable habitat area subject to PPL  (%)
Towed bottom-contact fishing	513.84	25.19	25.19	0.11
Extraction – sand and gravel (active area only)	1369.49	70.79	16.41	0.07
Dredge & spoil disposal	173.49	23.6	10.31	0.04
Aquaculture	5.46	2.39	2.39	0.01
Extraction – navigational dredging (capital & maintenance)	153.79	10.81	1.87	0.01
Recreational activities (anchorages only)	215.21	30.68	0.61	<0.01
Cultural & heritage sites/structures (wrecks)	4.44	0.52	0.52	<0.01
Marine hydrocarbon extraction (not including pipelines)	8.12	0.44	0.44	<0.01
Coastal docks, ports & marinas	2.82	0.39	0.39	<0.01
Coastal defence & land claim protection	4.94	0.32	0.32	<0.01
Submarine cable operations (communications and power)	111.14	9.79	0.29	<0.01
Submarine pipeline operations	3.11	0.2	0.2	<0.01
Renewable energy – wind (not including cables)	560.57	27.26	0.03	<0.01
Renewable energy - tidal (not including cables)	0.08	0.06	0	<0.01

Celtic Seas sub-region 

Within the Celtic Seas sub-region, re-assessment of the UKMS 2018 PPL indicator using the updated HSM and the original UKMS 2018 activity data estimated the area of PPL as 0.33% (136.80 km²) of the 41,180 km² of horse mussel reefs (Figure 22). Similar to results from the Greater North Sea sub-region, the area of predicted suitable habitat potentially lost in the Celtic Seas sub-region in the UKMS 2018 re-assessment was again less than that estimated from the UKMS 2024 assessment, 0.83% (340.60 km²) (Figure 10). Therefore, PPL of predicted suitable habitat for horse mussel reefs from assessed anthropogenic activities has continued in the Celtic Seas sub-region and has also increased. 

From re-assessment of the UKMS 2018 PPL indicator with updated HSMs, the main activities that contributed to PPL of predicted suitable habitat for horse mussel reefs in the Celtic Seas sub-region were: towed bottom-contact fishing (0.19% of reef area), dredge and spoil disposal (0.06%), aquaculture (0.05%), extraction -sand and gravel (0.02%) and extraction – navigational dredging (<0.01%) (Table 22, Table 23 and Figure 23). Ranked anthropogenic activities by contribution to PPL of predicted suitable habitat for horse mussel reefs from the UKMS 2018 re-assessment differed from that of the UKMS 2024 assessment. However, the anthropogenic activity contributing the most towards PPL of predicted suitable habitat in the Celtic Seas sub-region for both assessments was towed bottom-contact fishing (UKMS 2024 = 0.67%, UKMS 2018 = 0.18%) (Table 12, Figure 11, Table 22, Table 23 and Figure 23). As with the Greater North Sea regions, the same five anthropogenic activities were found to be the greatest contributors towards PPL of predicted suitable habitat, although in different rank orders between the two assessments. 

Table 22: Comparison of area of horse mussel reefs within the Celtic Seas sub-region subject to PPL from assessed anthropogenic activities and the rank order of activities contributing to PPL, between the re-assessment of the UKMS 2018 PPL indicator with original activity data and updated HSMs, and the UKMS 2024 assessment of the PPL indicator. Fishing* refers to towed bottom-contact fishing. 

Item		UKMS 2018 Re-assessment	UKMS 2024
Area of PPL of horse mussel reefs		136.80 km2 (0.33%)	340.60 km2 (0.83%)
Ranked activity by contribution to PPL	1st	Fishing* (0.19%)	Fishing* (0.67%)
	2nd	Dredge and spoil disposal (0.06%)	Aquaculture (0.07%)
	3rd	Aquaculture (0.05%)	Dredge & spoil disposal (0.05%)
	4th	Extraction – sand and gravel (0.01%)	Extraction – navigational dredging (0.02%)
	5th	Extraction – navigational dredging (<0.01%)	Extraction – sand and gravel (0.01%)

Table 23: Assessed anthropogenic activities contributing to the potential physical loss of predicted suitable habitat for horse mussel reefs within the Celtic Seas sub-region for the UKMS 2018 re-assessment. Activity footprint = raw footprint area of activity; Probability adjusted activity area = area of predicted suitable habitat for horse mussel reefs covered by activity footprint; PPL factor adjusted area = area of potentially suitable habitat lost due to relevant anthropogenic activity; Percentage of suitable habitat area subject to PPL = PPL factor adjusted area as a percentage of the predicted suitable habitat area within the Celtic Seas sub-region. 

Activity Name	Activity footprint (km2)	Probability adjusted activity area (km2)	PPL factor adjusted area (km2)	Percentage of suitable habitat area subject to PPL  (%)
Towed bottom-contact fishing	903.65	79.44	79.44	0.19
Dredge & spoil disposal	139.02	48.43	22.93	0.06
Aquaculture	45.66	21.35	21.35	0.05
Extraction – sand and gravel (active area only)	160.89	49.5	8.06	0.02
Extraction – navigational dredging (capital & maintenance)	84.68	10.97	1.9	<0.01
Submarine cable operations (communications and power)	226.39	40.09	1.2	<0.01
Cultural & heritage sites/structures (wrecks)	2.31	0.51	0.51	<0.01
Recreational activities (anchorages only)	157.77	23.39	0.47	<0.01
Marine hydrocarbon extraction (not including pipelines)	1	0.28	0.28	<0.01
Coastal defence & land claim protection	0.89	0.2	0.2	<0.01
Submarine pipeline operations	0.87	0.19	0.19	<0.01
Coastal docks, ports & marinas	0.47	0.18	0.18	<0.01
Renewable energy – wind (not including cables)	314.94	82.3	0.08	<0.01
Renewable energy – wave (not including cables)	100.9	13.31	0.01	<0.01
Renewable energy - tidal (not including cables)	14.34	4.63	0	<0.01

Horse mussel reefs assessment threshold – 2018 re-assessment 

The assessment threshold of the PPL indicator for the UKMS 2018 re-assessment followed that for the UKMS 2024 assessment, i.e., it was based on no PPL in polygons of the HSM with a probability of greater than or equal to 0.5. 

UKMS region 

As the same HSM was used for both assessment periods, predicted suitable habitat was located in the same locations and in the same amount (14,916.98 km²) as in the UKMS 2024 assessment (Figure 12). However, because activity data varied between assessment periods, areas of PPL from the PPL subset differed to that of the UKMS 2024 assessment. PPL was identified off the coast of Liverpool, Holyhead, Northern Ireland and Barrow-In-Furness (Figure 24). The area of PPL calculated from the PPL subset was 30.88 km², 0.21% of the area of predicted suitable habitat calculated from the HSM subset. Therefore, the threshold of no PPL in grid cells with a probability of suitable habitat greater than, or equal to, 0.5 was exceeded and the target was not met for the UKMS region. 

Figure 24: Mean predictive values of habitat suitability for horse mussel reefs where probability was ≥ 0.5, across the UKMS region with transitional waters removed, restricted to where assessed anthropogenic activities that cause PPL occur within the UKMS 2018 re-assessment. 

The activities contributing the greatest area of PPL of horse mussel reefs in polygons with a probability of suitable habitat greater than or equal to 0.5, and therefore contributing to the failure to meet the target for the UKMS region for the UKMS 2018 assessment were: aquaculture (14.70 km²), followed by dredge and spoil disposal (7.94 km²), and extraction – sand and gravel (6.56 km²). All other activities contributed an area of less than 1 km² per activity to PPL in polygons with a probability of greater than or equal to 0.5. 

Re-assessment of the UKMS 2018 data were made at the sub-region scale to determine if the threshold had been met for the Greater North Sea and the Celtic Seas. 

Greater North Sea 

The area of predicted suitable habitat for horse mussel reefs within the Greater North Sea sub-region estimated from the HSM subset (Figure 18) was 946.92 km², 4.01% of the area of predicted suitable habitat within the Greater North Sea predicted by the unrestricted HSM. The area of PPL of predicted suitable habitat from the Greater North Sea sub-region due to assessed anthropogenic activities in the PPL subset (Figure 24) was 2.72 km², 0.29% of predicted suitable habitat from the HSM subset. Therefore, the threshold of no PPL in grid cells with a probability of suitable habitat greater than, or equal to 0.5, was exceeded and the target was not met for the Greater North Sea sub-region. 

The activity contributing the greatest area of PPL of horse mussel reefs in polygons with a probability of suitable habitat greater than or equal to 0.5, and therefore contributing to the failure to meet the target for the Greater North Sea sub-region for the UKMS 2018 assessment was aquaculture (2.08 km²). All other activities contributed an area of less than 1 km² per activity to PPL in polygons with a probability of greater than or equal to 0.5. 

Celtic Seas 

The area of predicted suitable habitat for horse mussel reefs within the Celtic Seas sub-region estimated from the HSM subset (Figure 18) was 13,970.06 km², 33.92% of the area of predicted suitable habitat within the Celtic Seas predicted by the unrestricted HSM. The area of PPL of predicted suitable habitat from the Celtic Seas sub-region due to assessed anthropogenic activities in the PPL subset (Figure 24) was 28.16 km², 0.20% of predicted suitable habitat from the HSM subset. Therefore, the threshold of no PPL in grid cells with a probability of suitable habitat greater than, or equal to 0.5, was exceeded and the target was not met for the Celtic Seas sub-region. 

The activities contributing the greatest area of PPL of horse mussel reefs in polygons with a probability of suitable habitat greater than or equal to 0.5, and therefore contributing to the failure to meet the target for the Celtic Seas sub-region for the UKMS 2018 assessment were: aquaculture (12.60 km²), dredge and spoil disposal (7.46 km²), and extraction – sand and gravel (6.56 km²). All other activities contributed an area of less than 1 km² per activity to PPL in polygons with a probability of greater than or equal to 0.5. 

Eelgrass beds 

UKMS region 

The area of PPL due to assessed activities across the UKMS region in the re-assessment of the UKMS 2018 activity data was 33.06 km² (1.45% of bed area) (Figure 25), and for the UKMS 2024 assessment was 48.85 km² (1.93% of bed area) (Figure 14). Therefore, the potential physical loss of predicted suitable habitat for eelgrass beds has continued and increased since the last UKMS assessment. 

Figure 25: Percentage of predicted suitable habitat for eelgrass beds, potentially lost to assessed anthropogenic activities for the UKMS 2018 re-assessment as a percentage of the total area of predicted suitable habitat, across the entire UKMS region. 

Comparisons of the contribution of assessed anthropogenic activities to PPL were also made between the UKMS 2018 re-assessment and the UKMS 2024 assessment. In both assessments, aquaculture contributed the greatest amount of PPL of predicted suitable habitat for eelgrass beds across the UKMS region (UKMS 2018 = 0.80%, UKMS 2024 = 0.83% of bed area) (Figure 15, Table 13, Figure 26, Table 24, and Table 25). In both assessments, aquaculture was followed by the same activities, although dredge and spoil disposal increased in its contribution to PPL (Figure 15, Table 13, Figure 26, Table 24, and Table 25). Notably the same anthropogenic activities caused the greatest PPL of predicted suitable habitat for eelgrass beds over both assessments. 

Figure 26: Contribution from assessed anthropogenic activities in the UKMS 2018 re-assessment to potential physical loss of predicted suitable habitat for eelgrass beds as a percentage of total area of predicted suitable habitat within the UKMS region 

Table 24: Assessed anthropogenic activities contributing to the potential physical loss of predicted suitable habitat for eelgrass beds across the UKMS region for the UKMS 2018 re-assessment. Activity footprint = raw footprint area of activity; Probability adjusted activity area = area of predicted suitable habitat for horse mussel reefs covered by activity footprint; PPL factor adjusted area = area of potentially suitable habitat lost due to relevant anthropogenic activity; Percentage of suitable habitat area subject to PPL = PPL factor adjusted area as a percentage of the predicted suitable habitat area across the UKMS region. 

Activity Name	Activity area (km2)	Probability adjusted activity area (km2)	PPL factor adjusted area (km2)	Percentage of habitat area  (%)
Aquaculture	31.3	18.26	18.26	0.8
Extraction - navigational dredging (capital & maintenance)	150.88	46.84	10.67	0.47
Coastal docks, ports & marinas	2.32	1.18	1.18	0.05
Dredge & spoil disposal	14.89	2.75	1.14	0.05
Coastal defence & land claim protection	3.56	0.65	0.65	0.03
Extraction - sand and gravel (active area only)	12.62	1.92	0.53	0.02
Recreational activities (anchorages only)	55.99	19.4	0.39	0.02
Submarine cable operations (communications and power)	16.31	4.1	0.12	0.01
Cultural & heritage sites/structures (wrecks)	0.5	0.09	0.09	<0.01
Renewable energy - wave (not including cables)	5.28	0.94	0.02	<0.01
Renewable energy - wind (not including cables)	30.22	1.26	0	<0.01
Submarine pipeline operations	0.05	0	0	<0.01
Marine hydrocarbon extraction (not including pipelines)	0.01	0	0	<0.01

Table 25: Comparison of area of eelgrass beds across the UKMS region subject to PPL from assessed anthropogenic activities and the rank order of activities contributing to PPL, between the re-assessment of the UKMS 2018 PPL indicator with original activity data and updated HSMs, and the UKMS 2024 assessment of the PPL indicator.  

Item		UKMS 2018 Re-assessment	UKMS 2024
Area of PPL of eelgrass beds		33.06 km2 (1.45%)	48.85 km2 (1.93%)
Ranked activity by contribution to PPL	1st	Aquaculture (0.80%)	Aquaculture (0.83%)
	2nd	Extraction - navigational dredging (0.47%)	Extraction - navigational dredging (0.82%)
	3rd	Coastal docks, ports & marinas (0.05%)	Dredge and spoil disposal (0.15%)
	4th	Dredge and spoil disposal (0.05%)	Coastal docks, ports & marinas (0.06%)
	5th	Coastal defence and land claim protection (0.03%)	Coastal defence and land claim protection (0.03%)

Greater North Sea sub-region 

Within the Greater North Sea sub-region, reassessment of the UKMS 2018 PPL indicator using the updated HSM and the original UKMS 2018 activity data estimated the area of PPL as 1.96% (12.27 km²) of the 624.63 km² of suitable habitat for eelgrass beds (Figure 27). The area of PPL in the Greater North Sea sub-region in the UKMS 2018 re-assessment was over half of that calculated in the UKMS 2024 assessment (19.31 km²). Therefore, the PPL of predicted suitable habitat for eelgrass beds has continued and increased in the Greater North Sea sub-region since the last UKMS assessment. 

Figure 27: Percentage of predicted suitable habitat for eelgrass beds, potentially lost to assessed anthropogenic activities for the UKMS 2018 re-assessment as a percentage of the total area of predicted suitable habitat, per UKMS sub-region. 

Activities contributing the greatest PPL of predicted suitable habitat in the UKMS 2018 re-assessment in the Greater North Sea sub-region were: Extraction - navigational dredging (1.33% of bed area), followed by aquaculture (0.21%), dredge and spoil disposal (0.17%), coastal docks, ports and marinas (0.16%), and coastal defence and land claim protection (0.05%) (Table 26, Table 27, and Figure 28). Ranked anthropogenic activities by contribution to PPL of eelgrass beds from the UKMS 2018 re-assessment differed only slightly from the UKMS 2024 assessment. Aquaculture and dredge and spoil disposal were ranked second and third in the activities contributing the most PPL of suitable habitat area in the UKMS 2018 re-assessment (0.21% and 0.17% of suitable habitat area, respectively), but in the UKMS 2024 assessment this was reversed (dredge and spoil disposal – 0.50% and aquaculture – 0.41%). Again, the same five anthropogenic activities were found to contribute the greatest area of PPL of predicted suitable habitat for eelgrass beds for both assessments. 

Table 26: Comparison of area of eelgrass beds within the Greater North Sea sub-region subject to PPL from assessed anthropogenic activities and the rank order of activities contributing to PPL, between the re-assessment of the UKMS 2018 PPL indicator with original activity data and updated HSMs, and the UKMS 2024 assessment of the PPL indicator. 

Item		UKMS 2018 Re-assessment	UKMS 2024
Area of PPL of eelgrass beds		12.27 km2 (1.96%)	19.31 km2 (3.09%)
Ranked activity by contribution to PPL	1st	Extraction - navigational dredging (1.33%)	Extraction - navigational dredging (1.86%)
	2nd	Aquaculture (0.21%)	Dredge and spoil disposal (0.50%)
	3rd	Dredge and spoil disposal (0.17%)	Aquaculture (0.41%)
	4th	Coastal docks, ports, marinas (0.16%)	Coastal docks, ports, marinas (0.19%)
	5th	Coastal defence and land claim protection (0.05%)	Coastal defence and land claim protection (0.06%)

Table 27:  Assessed anthropogenic activities contributing to the potential physical loss of predicted suitable habitat for eelgrass beds within the Greater North Sea sub-region for the UKMS 2018 re-assessment. Activity footprint = raw footprint area of activity; Probability adjusted activity area = area of predicted suitable habitat for horse mussel reefs covered by activity footprint; PPL factor adjusted area = area of potentially suitable habitat lost due to relevant anthropogenic activity; Percentage of suitable habitat area subject to PPL = PPL factor adjusted area as a percentage of the predicted suitable habitat area within the Greater North Sea sub-region. 

Activity Name	Activity area (km2)	Probability adjusted activity area (km2)	PPL factor adjusted area (km2)	Percentage of habitat area  (%)
Extraction - navigational dredging (capital & maintenance)	93.28	33.84	8.34	1.33
Aquaculture	2.65	1.29	1.29	0.21
Dredge & spoil disposal	13.4	2.55	1.07	0.17
Coastal docks, ports & marinas	1.94	0.99	0.99	0.16
Coastal defence & land claim protection	2.87	0.3	0.3	0.05
Recreational activities (anchorages only)	40.31	10.95	0.22	0.04
Cultural & heritage sites/structures (wrecks)	0.29	0.03	0.03	0.01
Submarine cable operations (communications and power)	7.06	0.94	0.03	<0.01
Extraction - sand and gravel (active area only)	2.55	0.29	0.01	<0.01
Renewable energy - wind (not including cables)	30.22	1.26	0	<0.01
Submarine pipeline operations	0.04	0	0	<0.01
Marine hydrocarbon extraction (not including pipelines)	0.01	0	0	<0.01

Figure 28: Contribution from assessed anthropogenic activities in the UKMS 2018 re-assessment to potential physical loss of predicted suitable habitat for eelgrass beds as a percentage of total area of predicted suitable habitat, per UKMS sub-region. 

Celtic Seas sub-region 

Within the Celtic Seas sub-region, re-assessment of the UKMS 2018 indicator estimated the area of PPL as 20.79 km² (1.26% of predicted suitable habitat for eelgrass beds) (Figure 27). The area of predicted suitable habitat potentially lost in the Celtic Seas sub-region in the UKMS 2018 re-assessment was less than that in the UKMS 2024 assessment (24.55 km², 1.48% of predicted suitable habitat). Therefore, PPL of predicted suitable habitat for eelgrass beds from assessed anthropogenic activities has continued in the Celtic Seas sub-region and has also increased. 

From re-assessment of the UKMS 2018 PPL indicator with updated HSMs, the main activities that contributed to PPL of predicted suitable habitat for eelgrass beds in the Celtic Seas sub-region were: aquaculture (1.03% of bed area), extraction - navigational dredging (0.14%), extraction - sand and gravel (0.03%), coastal defence and land claim protection (0.02%) and coastal docks, ports and marina (0.01%) (Figure 28, Table 28, and Table 29). Ranked anthropogenic activities by contribution to PPL of predicted suitable habitat for eelgrass beds from the UKMS 2018 re-analysis differs from that of the UKMS 2024 assessment. However, the anthropogenic activity contributing the most towards PPL of predicted suitable habitat in the Celtic Seas sub-region for both assessments was aquaculture (UKMS 2024 = 0.98%, UKMS 2018 = 1.03%). The five anthropogenic activities contributing the most PPL in the Celtic Seas differed, with extraction - sand and gravel and coastal dock, ports, marinas absent in the top five activities contributing to PPL in the UKMS 2024 assessment. 

Table 28: Comparison of area of eelgrass beds within the Celtic Seas sub-region subject to PPL from assessed anthropogenic activities and the rank order of activities contributing to PPL between the re-assessment of the UKMS 2018 PPL indicator with original activity data and updated HSMs, and the UKMS 2024 assessment of the PPL indicator.  

Celtic Seas		UKMS 2018 Re-assessment	UKMS 2024
Area of PPL of eelgrass beds		20.79 km2 (1.26%)	24.55 km2 (1.48%)
Ranked activity by contribution to PPL	1st	Aquaculture (1.03%)	Aquaculture (0.98%)
	2nd	Extraction - navigational dredging (0.14%)	Extraction - navigational dredging (0.43%)
	3rd	Extraction – sand and gravel (0.03%)	Dredge and spoil disposal (0.02%)
	4th	Coastal defence and land claim protection (0.02%)	Coastal defence and land claim protection (0.02%)
	5th	Coastal docks, ports, marinas (0.01%)	Recreational activities (anchorages only) (0.01%)

Table 29: Assessed anthropogenic activities contributing to the potential physical loss of predicted suitable habitat for eelgrass beds within the Celtic Seas sub-region for the UKMS 2018 re-assessment. Activity footprint = raw footprint area of activity; Probability adjusted activity area = area of predicted suitable habitat for horse mussel reefs covered by activity footprint; PPL factor adjusted area = area of potentially suitable habitat lost due to relevant anthropogenic activity; Percentage of suitable habitat area subject to PPL = PPL factor adjusted area as a percentage of the predicted suitable habitat area within the Celtic Seas sub-region. 

Activity Name	Activity area (km2)	Probability adjusted activity area (km2)	PPL factor adjusted area (km2)	Percentage of habitat area  (%)
Aquaculture	28.66	16.97	16.97	1.03
Extraction - navigational dredging (capital & maintenance)	57.6	13	2.34	0.14
Extraction - sand and gravel (active area only)	10.07	1.63	0.52	0.03
Coastal defence & land claim protection	0.69	0.35	0.35	0.02
Coastal docks, ports & marinas	0.39	0.19	0.19	0.01
Recreational activities (anchorages only)	15.68	8.45	0.17	0.01
Submarine cable operations (communications and power)	9.25	3.16	0.09	0.01
Dredge & spoil disposal	1.48	0.2	0.07	<0.01
Cultural & heritage sites/structures (wrecks)	0.21	0.06	0.06	<0.01
Renewable energy - wave (not including cables)	5.28	0.94	0.02	<0.01
Submarine pipeline operations	0.01	0	0	<0.01

Eelgrass bed assessment threshold – 2018 re-assessment 

UKMS region 

The same HSM was used for both the UKMS 2024 assessment and UKMS 2018 re-assessment, therefore the location of predicted suitable habitat for eelgrass beds in the HSM subset was the same (Figure 19), as was the estimated area of predicted suitable habitat for eelgrass beds from the HSM subset across the UKMS region, 878.67 km². 

However, as activity data varied between assessments, areas of PPL from the PPL subset differed to that of the UKMS 2024 assessment. PPL was identified in the Solent, off the coast of Belfast and off the coast of Bournemouth (Figure 29). The area of PPL calculated from the PPL subset was 23.31 km², 2.65% of the area of predicted suitable habitat calculated from the HSM subset. Therefore, the threshold of no PPL in grid cells with a probability of suitable habitat greater than, or equal to 0.5, was exceeded and the target was not met for the UKMS region. 

Re-assessment of the UKMS 2018 data was made at the sub-region scale to determine if the threshold had been met for the Greater North Sea and the Celtic Seas. 

The activities contributing the greatest area of PPL of eelgrass beds in polygons with a probability of suitable habitat greater than or equal to 0.5 and therefore contributing to the failure to meet the target for the UKMS region for the UKMS 2018 assessment were predominantly aquaculture (14.90 km²), and extraction – navigation dredging (6.97 km²). All other activities contributed an area of less than 1 km² per activity to PPL in polygons with a probability of greater than or equal to 0.5. 

Figure 29: Mean predictive values of habitat suitability for eelgrass beds where probability was ≥ 0.5, across the UKMS region with transitional waters removed, restricted to where assessed anthropogenic activities that cause PPL occur within the UKMS 2018 re-assessment. 

Greater North Sea 

The area of predicted suitable habitat for eelgrass beds within the Greater North Sea sub-region was 162.86 km² (Figure 18 and Figure 12), 26.07% of the area of predicted suitable habitat within the Greater North Sea estimated by the unrestricted HSM. The area of PPL of predicted suitable habitat from the Greater North Sea sub-region due to assessed anthropogenic activities in the PPL subset (Figure 29) was 7.97 km², 4.89% of predicted suitable habitat from the HSM subset. Therefore, the threshold of no PPL in grid cells with a probability of suitable habitat greater than or equal to 0.5, was exceeded and the target was not met for the Greater North Sea sub-region. 

The activity contributing the greatest area of PPL of eelgrass beds in polygons with a probability of suitable habitat greater than or equal to 0.5 and therefore contributing to the failure to meet the target for the Greater North Sea sub-region for the UKMS 2018 assessment, was predominantly extraction – navigational dredging (6.50 km²). All other activities contributed an area of less than 1 km² per activity to PPL in polygons with a probability of greater than or equal to 0.5. 

Celtic Seas 

The area of predicted suitable habitat for eelgrass beds from the HSM subset for the Celtic Seas was 715.82 km², 43.31% of the area of predicted suitable habitat within the sub-region predicted by the unrestricted HSM. The area of PPL of predicted suitable habitat from the Celtic Seas sub-region due to assessed anthropogenic activities in the PPL subset (Figure 29) was 15.34 km², 2.14% of predicted suitable habitat from the HSM subset. Therefore, the threshold of no PPL in grid cells with a probability of suitable habitat greater than or equal to 0.5 was exceeded and the target was not met for the Celtic Seas sub-region. 

The activity contributing the greatest area of PPL of eelgrass beds in polygons with a probability of suitable habitat greater than or equal to 0.5 and therefore contributing to the failure to meet the target for the Celtic Seas sub-region for the UKMS 2024 assessment was predominantly aquaculture (14.30 km²). All other activities contributed an area of less than 1 km² per activity to PPL in polygons with a probability of greater than or equal to 0.5. 

Conclusions

The UK target for biogenic seafloor habitats in the UK Marine Strategy Part One has not been met for either horse mussel (Modiolus modiolus) reefs or eelgrass (Zostera marina) beds. Across the UKMS region, in areas where the probability of predicted suitable habitat is greater than or equal to 0.5, 0.23% of predicted suitable habitat for horse mussel reefs and 3.00% for eelgrass beds has potentially been lost due to the overlap with anthropogenic activities. 

An assessment of the activities that contributed the greatest area of potential physical loss and to the failure to meet the indicator threshold for horse mussel reefs and eelgrass beds was also carried out. For horse mussel reefs across the UKMS region, aquaculture accounted for 16.30 km² and dredge and spoil disposal 10.70 km² of the 34.02 km² of the potential area of reefs lost. For eelgrass beds across the UKMS region, aquaculture accounted for 14.40 km² and extraction – navigational dredging 9.92 km² of the 26.40 km² of the potential area of beds lost. 

Potential Physical Loss for horse mussel reefs and eelgrass beds has continued and increased since the last UK Marine Strategy assessment. 

Further information

A paucity of data on the extent and distribution of both subtidal horse mussel (Modiolus modiolus) reefs and eelgrass (Zostera marina) beds, and the impracticality of full-scale monitoring of both habitats, prevented the use of in-situ habitat maps in this assessment. Therefore, habitat suitability models (HSMs) using the random forest algorithm were produced to predict the extent and distribution of both horse mussel reefs and eelgrass beds to fill this data gap. Despite the good performance and accuracy of the models (≥ 0.90 Area Under the Curve, AUC, score), the HSMs are based on a set of environmental variables, and their shared extent is used to predict the suitability of habitat. Therefore, importantly, the HSMs represent a prediction of habitat suitability only and are not necessarily representative of in-situ, or realised, habitat. It is because of this prediction of suitable habitat that physical loss in this assessment is termed potential physical loss (PPL). 

Additionally, the HSMs do not incorporate impacts from anthropogenic activities, climate change (e.g., increase in storm events, ocean acidification), or other biological or environmental influences (e.g., disease, competition, turbidity) which may limit the distribution of either habitat. For example, the horse mussel reefs HSM predicts the presence of a small area of suitable habitat within the Bristol Channel, which is supported by observations of dense aggregations of juveniles in previous literature; however, they are not thought to reach adulthood (Fletcher and others, 2012). The model may also predict historical suitable habitat extent for horse mussel reefs and eelgrass beds in areas where they are not present currently, or in areas where they may be present in the future. For example, eelgrass bed presence may have been predicted in areas where the seagrass wasting disease (Labyrinthula zosterae) decimated extent in the 1920s and 1930s (Butcher, 1934; Muelhstein, 1989) and where the habitat type is no longer present. Therefore, the HSMs may overestimate the distribution and extent of assessed habitats. 

Conversely, due to historic declines, the HSMs might also underpredict the distribution and extent of suitable habitat. For example, eelgrass beds are thought to have been subject to decline (e.g., from mining activities) prior to, and also since, the extent loss due to wasting disease in the early 20^th century (Jones and Unsworth, 2016; Unsworth and others, 2017; Green and others, 2021). All eelgrass bed presence records have been collected since the decline in extent, thus values of environmental variables used do not incorporate environmental conditions experienced prior to this decline. As a result, the extent of suitable habitat may be underpredicted, due to the restriction of environmental conditions, and therefore may impact the calculation of PPL. 

The indicator threshold (no PPL in cells containing a probability of suitable habitat of greater than or equal to 0.5) was exceeded for all combinations of habitat and UKMS region or sub-region. Therefore, the UK target was not met. Activities that contributed the greatest area of PPL of potentially suitable habitat for horse mussel reefs were aquaculture and dredge and spoil disposal, and for eelgrass beds were aquaculture and navigational dredging. These activities align with the previous assessment (Strong and others, 2018), which identified navigational dredging, dredge and spoil disposal, aquaculture, and recreational activity as the main sources of loss for horse mussel reefs; and aquaculture, navigational dredging, dredging and spoil disposal, and coastal development as the main sources of loss for eelgrass beds. 

Regardless of probability of predicted suitable habitat, results from the UKMS 2024 assessment estimate that 621.98 km² of the 64,772.85 km² (0.96%) of predicted suitable habitat for horse mussel reefs and 43.85 km² of the 2250.54 km² (1.93%) of predicted suitable habitat for eelgrass beds has potentially been lost within the UKMS region due to assessed anthropogenic activities. PPL of predicted suitable habitat for both habitats was greater in all assessment units in the UKMS 2024 assessment than the re-assessment of the UKMS 2018 activity data with the updated HSMs. Therefore, PPL of predicted suitable habitat for both horse mussel reefs and eelgrass beds has continued to occur and may have increased from 2017 to present. For the 2024 assessment, activity data were restricted to the years 2017 to present. However, temporal information was not available for all activities, meaning that it may be considered twice, in this assessment and the 2018 re-assessment. This could lead to an apparent increase in PPL between assessment periods because the 2024 assessment may include data prior to 2017 and therefore PPL may have been overestimated. 

Assessment of the contribution to PPL per assessed anthropogenic activity was also conducted. The activities that contributed towards the greatest PPL for horse mussel reefs within the UKMS 2024 Part One were towed bottom-contact fishing; dredge and spoil disposal; extraction – navigational dredging; aquaculture; and extraction – sand and gravel. These activities remained consistently in the top five activities contributing to PPL regardless of the scale at which the assessment was undertaken (by entire UKMS region or per sub-region), albeit in different orders of contribution. Towed bottom contact fishing was identified to be the greatest contributor to PPL at the UKMS region and Celtic Sea sub-region scale, whereas dredge and spoil disposal was the greatest contributor of PPL in the Greater North Sea sub-region. Aquaculture; extraction – navigational dredging; dredge and spoil disposal; and coastal docks, ports and marinas contributed the most to PPL for eelgrass beds and were consistently within the top four activities across the UKMS region and within sub-regions. Aquaculture contributed the most PPL of eelgrass beds for the UKMS region and Celtic Seas sub-region, whilst extraction – navigational dredging contributed the most PPL in the Greater North Sea sub-region. 

The values of PPL presented in this report may be an underestimation, as data were not available for all relevant activities known to cause PPL (for example, inshore fisheries). Alternatively, in some cases PPL may have been overestimated because probability of suitable habitat was never zero (therefore habitat was present in every polygon) throughout the entire extent of each HSM, and as a result small areas of habitat in areas with low probability of suitable habitat will have been subject to PPL. 

Knowledge gaps

Incorporation of true-absence, and additional presence data from monitoring programmes will help to improve the prediction of the habitat suitability models. Additionally, the inclusion of further predictor variables (e.g., climatic variables and accurate anthropogenic data) will improve the confidence and accuracy of suitable habitat prediction and associated loss calculations. 

The paucity of activity data, e.g., inshore fisheries, may underestimate potential physical loss. The exclusion of pressures causing habitat damage, but not potential physical loss was based on expert judgement, which should be re-evaluated in future assessments. Additionally, activity intensity information could help determine when cumulative habitat damage becomes habitat loss. 

Further information

Habitat Suitability Models 

Biological and environmental data considerations 

Data availability 

Due to the paucity of data on the extent and distribution of both horse mussel (Modiolus modiolus) reefs and eelgrass (Zostera marina) beds, habitat suitability models (HSMs) were produced to estimate the distribution of predicted suitable habitat. However, due to the lack of data on the distribution of assessed habitats, the availability of habitat occurrence response variable data was also limited. For instance, only 1,471 presence records for horse mussel reefs and 4,730 presence records for eelgrass beds were collated, which were then reduced to 473 and 886 respectively after the selection process to determine presence / pseudo-absence per grid cell (Table 3). 

Additionally, the paucity of true absence of habitat records meant that pseudo-absence data had to be used in place of absence data. The use of pseudo-absence data is common in HSMs, and the best available habitat occurrence datasets for the UK were used for this occurrence type. However, the use of true absence data would mitigate the risk of classifying habitats as pseudo-absence that have understudied relationships with those assessed in this report and would also improve confidence in the models. 

Future work, firstly, could look at increasing survey efforts for the assessed habitats across the UK, including both presence and true absence, which could be incorporated into the HSMs to provide more evidence of habitat occurrence and environmental range. Secondly, when incorporating habitat occurrence response variables into future HSMs, balancing the number of presence and pseudo-absence records to ensure that the model better predicts where the habitat is present rather than where it is absent should be considered.  

Environmental data on some factors known to influence assessed habitat distribution and persistence were either unavailable or unsuitable for the current assessment. For example, horse mussel beds are known to be impacted by burial events (Hutchison and others, 2016), however there is currently no UK-wide layer available for sedimentation rates. For eelgrass beds, UK-wide nutrient data (in particular, phosphate and nitrate) incorporating the dynamic variabilities in concentrations would be a valuable addition to the HSM. Future work to assess the availability and applicability of data on environmental variables that influence habitat distribution would assist in improving model accuracy. 

The HSMs presented in this assessment do not account for the impacts of climate change on predicted habitat suitability. Although temperature at the seafloor is included as an environmental predictor variable in both HSMs, future HSMs could incorporate environmental variables directly linked to climate change to improve their predictive power. Eelgrass, for example, is sensitive to elevated temperatures (Shields, Moore and Parrish, 2018; Scalpone and others, 2020; Sawall, Ito and Pansch, 2021), sedimentation (Mills and Fonseca, 2003), light stress (Wong, Vercaemer, and Griffiths, 2021), and storm events (d’Avack and others, 2022), all of which are likely to increase in frequency due to climate change. The inclusion of environmental predictor variable layers related to climate change pressures may ensure that impacts that have already occurred as a result of changing climate are incorporated into the prediction of suitable habitat for both horse mussel reefs and eelgrass beds, improving the predictive power of the models. 

Temporal resolution 

Historical presence and pseudo-absence records were used in the production of HSMs for both horse mussel reefs and eelgrass beds to maximise the available data to feed into the models. Since the predictor variables (environmental data) were sourced from the most recent and up to date data sources, the environmental data assigned to historical presence or pseudo-absence records may not be representative of the environmental conditions at the time of sampling. This is of particular importance when considering the effects of climate change and anthropogenic activities. This may affect the range of environmental conditions ultimately used for predicting areas of suitable habitat for horse mussel reefs and eelgrass beds, resulting in a shift in distribution of most suitable predicted habitat. Further iterations of the HSMs could consider only using temporally aligned biological and environmental variables, although this would drastically reduce the available biological records unless more extensive surveying efforts were undertaken. 

Spatial resolution 

The production of HSMs for predicted suitable habitat for horse mussel reefs and eelgrass beds used biological and environmental variables at different spatial resolutions. Biological data was available as point records, whereas environmental data layers were resampled to a 300 m by 300 m grid from their original resolution (Table 1). As a result, the environmental conditions assigned to presence and pseudo-absence records may not be representative of the conditions where the sample was taken. In future, finer resolution environmental layers may be sourced to reduce the disparity in spatial resolution, however the inclusion of such layers may be limited by computer processing power. The use of a higher resolution environmental raster for a specific case study area with intricate coastlines such as the Hebrides or Orkney, if available, may provide useful information on the effect of this type of variability on the HSM output. 

Additionally, detail on the fine-scale resolution, and therefore distribution, of presence and pseudo-absence records was reduced when the location of points was aggregated up to the 300 m by 300 m grid cell. Presence and pseudo-absence data were further simplified by calculating the proportion of presence to weighted pseudo-absence records per grid cell, thereby categorising the whole cell by one occurrence type. In future, finer resolution grid cells obtained from the environmental variables may mitigate the reduction in resolution of the biological sample points, though this may be restricted by computer processing power. 

Connectivity dynamics 

The current HSMs do not capture the dynamics of sources and sinks, in terms of patterns of larval dispersal and genetic connectivity for horse mussels and propagule dispersal for eelgrass. Although the HSMs predict areas of potentially suitable habitat in terms of environment conditions, a population may not be able to persist in that location if connectivity and self-recruitment are limited. 

In simulations of horse mussel larval dispersal, Gormley and others (2015) identified that connectivity between populations of horse mussel in Northern Ireland and those in North Pen Llŷn or Isle of Man is unlikely; however, extant connectivity between populations in the Irish Sea was identified. Certainly, population dynamics need to be examined on a location-by-location basis. Model simulations show that the west coast of Scotland populations are larval sources for Orkney horse mussel beds, but within the Orkney populations there is a weak connection and self-recruitment prevails; the Shetland populations, instead, are well connected to each other (Millar and others, 2019). 

Homogenous distribution of area potentially suitable for habitat 

The HSMs used in this assessment provide the probability of predicted habitat suitable for either horse mussel reefs or eelgrass beds. The methodology used to calculate the area of predicted suitable habitat in this assessment multiplies the area of each grid cell by the probability. As such, the area of suitable habitat calculated per cell is assumed to have a homogenous spread across the extent of the cell, whether the habitat area calculated is 90,000 m² or 90 m². The effect of this distribution is particularly important during the intersection of activity data with the HSM, where intersected polygons will also be assumed to have homogenous distribution of predicted habitat, which is unlikely. Therefore, within this assessment, fine-scale (at resolutions greater than 300 m) calculation of potential physical loss (PPL) is not possible. 

Incorporation of ground-truth habitat maps 

The data paucity in relation to the distribution and extent of horse mussel reefs and eelgrass beds necessitates the use of HSMs for assessment with the PPL indicator. To refine the models, where available, the observed distributions (mapping of realised habitat) could be incorporated to improve the PPL indicator assessment. The creation of this combined habitat map could be achieved in a similar way to the EUNIS Combined Map (Matear, Pinder, and Lillis, 2019), which uses habitat maps collected from survey, supplemented by modelled broad-scale habitat data in locations where habitat maps derived from survey are unavailable. The assessed habitats may be designated features of Marine Protected Areas (MPAs) in the UK MPA network, and further survey work within these sites would help to supplement this effort. 

Spatial blocking 

Due to the nature of surveying, habitat occurrence records were often in proximity to each other, and possibly autocorrelated. The model may therefore perform well in cells near sample data, but poorer in cells further away. The implementation of spatial blocking to address spatial autocorrelation may improve the predictive accuracy of the HSM and is considered best practice when partitioning data because it improves the spatial independence of the training and test data (Valavi and others, 2019). The use of spatial blocking may help to improve the predictive performance of the HSMs used in this assessment leading to more accurate predictions of habitat presence and PPL. 

Expansion of habitat extent 

It was not possible to measure any increases in habitat extent to assess against the UKMS Part One (HM Government, 2012) target for biogenic seafloor habitats, which requires the area of habitat to be stable or increasing and not smaller than the baseline value. The HSMs were developed specifically for use in this assessment and, therefore, only a reduction in habitat extent could be measured i.e., PPL from the total area of habitat. A decrease in PPL between assessment periods (although not observed) might imply an increase in habitat extent, however this might not necessarily be observed in-situ. Seagrasses, for example, may take 100 years to recover naturally, without human intervention (Reynolds and others, 2016), which would be unlikely to be observed between two assessment periods. However, with active seagrass restoration programmes using seeds, recovery may occur in as little as ten years (Reynolds and others, 2016). The results of this study may contribute to the identification of suitable locations for seagrass restoration efforts in the absence of activities that cause habitat loss and where conditions are suitable, provided other limiting factors are considered. 

Activity data 

Identification of activities that cause PPL 

The Marine Evidence Sensitivity Assessment (MarESA) sensitivity matrices were used to identify pressures relevant to horse mussel reefs and eelgrass beds. Pressures that only cause habitat damage and not physical loss, under realistic levels of activity, were identified using expert judgement. Likewise, expert judgement was used when querying the JNCC pressures-activities database (PAD) to identify and exclude pressures that are either unlikely to occur within the assessed habitats or occur at such an intensity to cause physical loss. In future, the sensitivity matrices could be updated to include intolerance / recovery code combinations equating to physical loss to facilitate the objective selection of pressures. Similarly, the JNCC PAD could be developed to facilitate the identity of pressures that occur within habitats and the types of impact. 

Data Paucity 

Activity footprint data 

Not all activities that generate PPL of predicted suitable habitat were available to be included in this assessment of the PPL indicator. Data on the presence of inshore fisheries (i.e., fishing conducted by vessels less than 12 m in length) are not presently available for use, thereby underestimating PPL due to fishing in coastal waters. However, with the introduction of inshore Vessel Monitoring (I-VMS), future assessments of the PPL indicator will be able to incorporate these data to better understand the impact of inshore fishing on the PPL of predicted suitable habitat. 

Data on rock or concrete mattress deposits to protect and stabilise submarine infrastructure (e.g., pipelines or drilling platforms) were also not available. These ‘dumps’ are commonly used to resolve freespans, prevent buckling of pipelines, and to protect structures from fishing gear, high currents and unstable seabed. Where the safety risk is too high during decommissioning and recovery of oil and gas infrastructure, and to a lesser degree subsea cables, dumps may also be utilised as a preventative measure on exposed infrastructure, rather than removal of the structure. Where utilised, rock deposits are not removed and thus represent a permanent feature of the seabed. In some cases, rock deposits can amount to hundreds of thousands of tonnes over large areas, therefore contributing to PPL of assessed habitats which are sensitive to this smothering. However, it should be noted that in some cases, applications for rock deposits for the stabilisation of drilling rigs are made pre-emptively in the event of it being required by location conditions, but it is not known if they are employed. Consents issued by the regulator indicate the location of rock deposits and material amount, however data on rock deposits are not available in spatial format and consequently have not been included in this assessment. Therefore, the PPL of assessed habitats are likely to be underestimated. 

Finally, the calculation of PPL is dependent on the underlying activity data. If activity data sources and the data contained within to represent each activity could be standardised, this would facilitate comparison of PPL between assessments. 

Activity intensity data 

Many of the sources of activity data included in this assessment indicated the presence of activities only. The inclusion of activity intensity within datasets would assist in identifying when cumulative effects of habitat damage would cause habitat loss. Further work on activities with measurable intensity information is required to understand when activity levels cause an identifiable and ecologically relevant level of damage to the habitat to the point that it results in physical loss. Resultant footprint information could be used with higher levels of confidence. 

In this assessment, the level for which intensity information has been categorised into classes causing physical loss is subjective. As with the sourcing of activity data and information contained within, future work on the calibration and standardisation of intensity data to facilitate comparisons between assessments would be required. 

Temporal restriction of activity data 

Activity data were restricted to the years 2017 to present where temporal data was available, though where missing this was not always possible. As a result, the PPL of habitats from 2017 to the present within this assessment may include data prior to 2017 and therefore be overestimated. The inclusion of temporal information in all activity datasets would ensure that PPL could be refined to specific time periods in future assessments. 

References

d'Avack, E.A.S., Tyler-Walters, H., Wilding, C.M. & Garrard, S.L., 2022. Zostera (Zostera) marina beds on lower shore or infralittoral clean or muddy sand. In Tyler-Walters H. and Hiscock K. (eds) Marine Life Information Network: Biology and Sensitivity Key Information Reviews, [on-line]. Plymouth: Marine Biological Association of the United Kingdom. [cited 18-07-2023]. Available from:  https://www.marlin.ac.uk/habitats/detail/257https://www.marlin.ac.uk/habitats/detail/257/zostera_zostera_marina_beds_on_lower_shore_or_infralittoral_clean_or_muddy_sand 

Bertelli, C.M., Stokes, H.J., Bull, J.C. & Unsworth R.K.F., 2022. The use of habitat suitability modelling for seagrass: A review. Front. Mar. Sci. 9:997831. doi: 10.3389/fmars.2022.997831 

Burrows, M., 2020. Wave fetch GIS layers for the UK and Ireland at 200m scale. figshare. Dataset. https://doi.org/10.6084/m9.figshare.12029682.v1 

Butcher, R.W., 1934. Zostera. Report on the present condition of eel grass on the coasts of England, based on a survey during august to October, 1933. ICES J. Mar. Sci. 9, 49–65. doi: 10.1093/icesjms/9.1.49 

Calabrese, J.M., Certain, G., Kraan, C. & Dormann, C.F., 2014. Stacking species distribution models and adjusting bias by linking them to macroecological models. Global Ecology and Biogeography, 23, 99–112. 

Castle, L., Pinder, J., Lillis, H. & Manca, E., 2022. Using an ensemble modelling approach to predict suitable habitat for Zostera marina beds, Modiolus modiolus beds and Sabellaria spinulosa reefs in UK waters. JNCC Report No. 718, JNCC, Peterborough, ISSN 0963-8091. https://hub.jncc.gov.uk/assets/c0b64152-83d5-4164-9a03-794744c9cb82 

Charbonnel, A., Lambert, P., Lassalle, G., Quinton, E., Guisan, A., Mas, L., Paquignon, G., Lecomte, M. & Acolas, M., 2023. Developing species distribution models for critically endangered species using participatory data: The European sturgeon marine habitat suitability. Estuarine, Coastal and Shelf Science, 280.  

Chefaoui, R.M., Assis, J., Duarte, C.M. & Serrão, E.A., 2016. Large-Scale Prediction of Seagrass Distribution Integrating Landscape Metrics and Environmental Factors: The Case of Cymodocea nodosa (Mediterranean–Atlantic). Estuaries and Coasts, 39, 123–137. 

Cook, R., Fariñas-Franco, J.M., Gell, F.R., Holt, R.H.F., Holt, T., Lindenbaum, C., Porter, J.S., Seed, R., Skates, L.R., Stringell, T.B. & Sanderson, W.G., 2013. The Substantial First Impact of Bottom Fishing on Rare Biodiversity Hotspots: A Dilemma for Evidence-Based Conservation. PLoS ONE 8(8): e69904. https://doi.org/10.1371/journal.pone.0069904 

Department for Environment Food and Rural Affairs. (2020) Defra's Marine Digital Elevation Model (DEM) - 1 arc second. 

Duarte, C.M. & Chiscano, C.L., 1999. Seagrass biomass and production: a reassessment. Aquat Bot. 65:159–174. 

Duncan, G., Pinder, J., Lillis, H. & Allen, H., (2022). Method for creating version 8 of the UK Composite Map of Annex I Reefs. Joint Nature Conservation Committee, Peterborough. 

Elliott, M., Burdon, D., Atkins, J.P., Borja, A., Cormier, R., De Jonge, V.N. & Turner, R.K., 2017. “And DPSIR begat DAPSI(W)R(M)!” – A unifying framework for marine environmental management. Marine Pollution Bulletin 118: 27–40. 

EMODnet, 2018. Light (PAR) at the seabed - Europe-wide. Available from: https://emodnet.ec.europa.eu/geoviewer/ 

EMODnet 2019. Kinetic energy at the seabed due to waves - Celtic, North Sea (mean of annual 90th percentile). Available from: https://emodnet.ec.europa.eu/geoviewer/ 

EMODnet 2015. Kinetic energy at the seabed due to currents in the Atlantic. Available from: https://emodnet.ec.europa.eu/geoviewer/ 

EMODnet, 2022. EMODnet Digital Bathymetry (DTM 2022). Available from: https://emodnet.ec.europa.eu/geoviewer/ 

Environment Agency, 2014. WFD Transitional and Coastal Waterbodies Cycle 2. Available from: https://www.data.gov.uk/dataset/3a75ec5f-a361-475c-80e3-52d93bbc5dbe/wfd-transitional-and-coastal-waterbodies-cycle-2 

ESRI, 2012. ArcGIS Release 10.1. Redlands, CA. 

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’ C/2016/5542, Official Journal of the European Union L 268, 1.10.2016, pages 113–117 (viewed on 27 November 2018) 

Fletcher, S., Saunders, J., Herbert, R., Roberts, C. & Dawson, K., 2012. Description of the ecosystem services provided by broad-scale habitats and features of conservation importance that are likely to be protected by Marine Protected Areas in the Marine Conservation Zone Project area. Natural England Commissioned Reports, Number 088. 

Gormley, K.S., Porter, J.S., Bell, M.C., Hull, A.D. & Sanderson, W.G., 2013. Predictive habitat modelling as a tool to assess the change in distribution and extent of an OSPAR priority habitat under an increased ocean temperature scenario: consequences for marine protected area networks and management. PLoS ONE, 8(7), e68263. 

Gormley, K., Mackenzie, C., Robins, P., Coscia, I., Cassidy, A., James, J., Hull, A., Piertney, S., Sanderson, W. & Porter, J., 2015. Connectivity and Dispersal Patterns of Protected Biogenic Reefs: Implications for the Conservation of Modiolus modiolus (L.) in the Irish Sea. PLoS ONE 10(12): e0143337. doi:10.1371/journal.pone.0143337. 

Greathead, C., Boschen-Rose, R.E., Langton, R., Clarke, J., Wright, P.J. & Boulcott, P., 2023. Priority marine feature surveys within the Small Isles MPA and surrounding waters. Main Report. Scottish Marine and Freshwater Science Vol 14 No 1, 52p. DOI: 10.7489/1614-1 

Green, A.E., Unsworth, R.K.F., Chadwick, M.A. & Jones, P.J.S., 2021. Historical Analysis Exposes Catastrophic Seagrass Loss for the United Kingdom. Front. Plant Sci. 12:629962. doi: 10.3389/fpls.2021.629962 

Guiry, M.D. & Guiry, G.M., 2023. AlgaeBase. World-wide electronic publication, National University of Ireland, Galway (taxonomic information republished from AlgaeBase with permission of M.D. Guiry). Zostera angustifolia (Hornemann) Reichenbach, 1845. Accessed through: World Register of Marine Species at: https://www.marinespecies.org/aphia.php?p=taxdetails&id=182740 

Hanley, J.A. & McNeil, B.J., 1982. The meaning and use of the area under a receiver operating characteristic (ROC) curve. Radiology, 143(1), pp.29-36. 

Heck, K., Nadeau, D. & Thomas, R., 1997. The Nursery Role of Seagrass Beds. Gulf of Mexico Science 15 (1). Retrieved from https://aquila.usm.edu/goms/vol15/iss1/8 

Hijmans R (2023). _raster: Geographic Data Analysis and Modeling. R package version 3.6-20, https://CRAN.R-project.org/package=raster 

Holt, T.J., Rees, E.I., Hawkins, S.J. & Seed, R., 1998. Biogenic reefs (Volume IX). An overview of dynamic and sensitivity characteristics for conservation management of marine SACs. Scottish Association for Marine Science (UK Marine SACs Project), 174 pp. 

Horn, B.K.P., 1981. Hill shading and the reflectance map. Proceedings of the IEEE 69(1), 14– 47. 

Hutchison, Z.L., Hendrick, V.J., Burrows, M.T., Wilson, B. & Last, K.S., 2016. Buried alive: the behavioural response of the mussels, Modiolus modiolus and Mytilus edulis to sudden burial by sediment. PloS ONE, 11(3). 

HM Government (2012) ‘Marine Strategy Part One: UK Initial Assessment and Good Environmental Status’ 

Joint Nature Conservation Committee, 2021. JNCC SDM Framework. R package version 0.1.0. https://github.com/jncc/sdms 

Joint Nature Conservation Committee, 2022. Data derived from UK Marine Recorder (All) snapshot (v20220804), available from https://jncc.gov.uk/our-work/marine-recorder/ - downloaded on 24/04/2023. Data licenses, access and use limitations as described per survey as contained within the database. 

Jones, B. & Unsworth, R., 2016. The perilous state of seagrass in the British Isles. R. Soc. Open sci. 3: 150596. http://dx.doi.org/10.1098/rsos.150596 

Kent, F.E., Gray, M.J., Last, K.S. & Sanderson, W.G., 2016. Horse mussel reef ecosystem services: evidence for a whelk nursery habitat supporting a shellfishery. Int. J. Biodivers. Sci. Ecosyst. Serv. Manag. 12, 172–180. 

Kent, F.E., Last, K.S., Harries, D.B. & Sanderson, W.G. 2017. In situ biodeposition measurements on a Modiolus modiolus (horse mussel) reef provide insights into ecosystem services. Estuar. Coast. Shelf. 84, 151–157. 

Lecours, V., Devillers, R., Simms, A.E., Lucieer, V.L. & Brown, C.J., 2017. Towards a framework for terrain attribute selection in environmental studies. Environmental modelling & software, 89, pp.19-30. 

Magorrian, B.H. & 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: 354-359. 

Matear, L., Pinder, J. & Lillis, H. 2019. Method for updating the UK full-coverage EUNIS level 3 seabed habitat map integrating fine- and broad-scale maps. https://hub.jncc.gov.uk/assets/538e8eaa-83cc-40cf-bc5a-ef599a716b79 

McDevitt-Irwin, J., Iacarella, J. & Baum, J., 2016. Reassessing the nursery role of seagrass habitats from temperate to tropical regions: A meta-analysis. Marine Ecology Progress Series. 557. 10.3354/meps11848. 

Millar, H., O’Hara Murray, R., Gallego, A., Gormley, K. & Kent, F., 2019. Connectivity of selected Priority Marine Features within and outwith the Scottish MPA network. Scottish Natural Heritage Research Report No. 1048. 

Mills, K.E. & Fonseca, M.S., 2003. Mortality and productivity of eelgrass Zostera marina under conditions of experimental burial with two sediment types. Marine Ecology Progress Series, 255, pp.127-134. 

Muehlstein, L.K., 1989. Perspectives on the wasting disease of eelgrass Zostera marina. Dis. Aquat. Org. 7, 211–221. 

Natural England and the Joint Nature Conservation Committee (2010). The Marine Conservation Zone Project: Ecological Network Guidance. Sheffield and Peterborough, UK. Available here: https://data.jncc.gov.uk/data/94f961af-0bfc-4787-92d7-0c3bcf0fd083/MCZ-Ecological-Network-Guidance-2010.pdf 

Natural England, 2021. Marine Habitats and Species Open Data (England), version June 2021. Available at: https://naturalengland-defra.opendata.arcgis.com/maps/25867d305dc545509e4d87420e296e12/about (Accessed 09/12/2021). 

Natural Resources Wales, 2023. Water Framework Directive (WFD) Regulations Cycle 3. Available from: https://datamap.gov.wales/layers/geonode:nrw_wfd_transitional_c3_baseline_classification 

Nordlund, L.M., Koch, E.W., Barbier, E.B., & Creed, J.C., 2016. Seagrass ecosystem services and their variability across genera and geographical regions. PLoS ONE, 11, e0163091. 

Northern Ireland Environment Agency, 2016. Northern Ireland Transitional Waters 2009-2015. Available from: https://www.data.gov.uk/dataset/42e0d654-e973-4c45-83a4-c19c57b33a90/northern-ireland-transitional-waters-2009-2015 

OSPAR Commission, 2008. Case Reports for the OSPAR List of Threatened and/or Declining Species and Habitats, 239-242, https://www.ospar.org/site/assets/files/44271/horse_mussel_beds.pdf 

OSPAR, 2009. Background document for Zostera beds, Seagrass beds. Biodiversity Series. OSPAR Commission. http://www.ospar.org/documents?d=7190 

OSPAR, 2022. OSPAR Threatened and/or declining habitats database. Accessed 18/05/2023. Available from: https://emodnet.ec.europa.eu/geoviewer/ 

11.000,51.600&layerIds=801&baseLayerId=-3&activeFilters= 

Pebesma, E., 2018. Simple Features for R: Standardized Support for Spatial Vector Data. The R Journal 10 (1), 439-446, https://doi.org/10.32614/RJ-2018-009 

Pebesma, E. & Bivand, R., 2023. Spatial Data Science: With Applications in R (1st ed.). Chapman and Hall/CRC. https://doi.org/10.1201/9780429459016 

QGIS.org, 2020. QGIS Geographic Information System. QGIS Association. http://www.qgis.org 

R Core Team, 2019. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. URL https://www.R-project.org/ 

RStudio Team, 2022. RStudio: Integrated Development Environment for R. RStudio, PBC, Boston, MA URL: http://www.rstudio.com/ 

Reynolds, L.K., Waycott, M., McGlathery, K.J. & Orth, R.J., 2016. Ecosystem services returned through seagrass restoration. Restoration Ecology, 24(5), pp.583-588. 

Robinson, L.K., Proudfoot, B., Rooper, C.N. & Bertram, D.F., 2021. Comparison of spatial distribution models to predict subtidal burying habitat of the forage fish Ammodytes personatus in the Strait of Georgia, British Columbia, Canada. Aquatic Conserv: Mar Freshw Ecosyst.;31:2855–2869 

Robson, L.M., Fincham, J., Peckett, F.J., Frost, N., Jackson, C., Carter, A.J. & Matear, L. 2018. UK Marine Pressures-Activities Database “PAD”: Methods Report. JNCC Report No. 624, JNCC, Peterborough, ISSN 0963-8091. 

Sanderson, W., Holt, R.H.F., Kay, L., Ramsay, K., Perrins, J., McMath, A.J. & Rees, E.I.S., 2008. Small-scale variation within a Modiolus modiolus (Mollusca: Bivalvia) reef in the Irish Sea. II. Epifauna recorded by divers and cameras', Journal of the Marine Biological Association of the United Kingdom, vol. 88, no. 1, pp. 143-149. https://doi.org/10.1017/S0025315408000040 

Sawall, Y., Ito, M. and Pansch, C., 2021. Chronically elevated sea surface temperatures revealed high susceptibility of the eelgrass Zostera marina to winter and spring warming. Limnol Oceanogr, 66: 4112-4124. https://doi.org/10.1002/lno.11947 

Scalpone, C.R., Jarvis, J.C., Vasslides, J.M., Testa, J.M. & Ganju, N.K., 2020. Simulated estuary-wide response of seagrass (Zostera marina) to future scenarios of temperature and sea level. Frontiers in Marine Science, 7, p.539946. 

Scottish Environmental Protection Agency, 2010. Transitional water bodies. Available from: https://www.data.gov.uk/dataset/480a311c-8345-47f3-9c54-20025da0d68a/transitional-water-bodies 

Shields, E.C., Moore, K.A. & Parrish D.B., 2018. Adaptations by Zostera marina Dominated Seagrass Meadows in Response to Water Quality and Climate Forcing Diversity 10, no. 4: 125. https://doi.org/10.3390/d10040125 

Strong, J., Vina-Herbon, C., Doria, L., Carter, A., Edwards, D., Lillis, H., Parry, M., Robson, L., Singleton, G., Young, M., Mackie, T., Boulcott, P. & Robison, K., 2018. Potential physical loss of predicted seafloor habitats. UK Marine Online Assessment Tool, available at: https://moat.cefas.co.uk/previous-assessments/2018-assessment/biodiversity-food-webs-and-marine-protected-areas/benthic-habitats/physical-loss/ 

De la Torriente Diez, A., González-Irusta, J.M., Serrano, A., Aguilar, R., Sánchez, F., Blanco, M. & Punzón, A., 2022. Spatial assessment of benthic habitats vulnerability to bottom fishing in a Mediterranean seamount, Marine Policy, Volume 135. https://doi.org/10.1016/j.marpol.2021.104850. 

Tyberghein, L., Verbruggen, H., Pauly, K., Troupin, C., Mineur, F., & De Clerck, O., 2012. Bio-ORACLE: A global environmental dataset for marine species distribution modelling. Global Ecology and Biogeography, 21, 272–281. 

Assis, J., Tyberghein, L., Bosh, S., Verbruggen, H., Serrão, E. A., & De Clerck, O., 2017. Bio-ORACLE v2.0: Extending marine data layers for bioclimatic modelling. Global Ecology and Biogeography. 

Tyler-Walters, H., 2007. Modiolus modiolus Horse mussel. In Tyler-Walters H. Marine Life Information Network: Biology and Sensitivity Key Information Reviews, [on-line]. Plymouth: Marine Biological Association of the United Kingdom. [cited 24-08-2023]. Available from: https://www.marlin.ac.uk/species/detail/1532 

Tyler-Walters, H., Tillin, H.M., d’Avack, E.A.S., Perry, F. & Stamp, T., 2018. Marine Evidence based Sensitivity Assessment (MarESA) – A Guide. Plymouth: Marine Biological Association. 

Unsworth, R.K.F., Williams, B., Jones, B.L. and Cullen-Unsworth, L.C., 2017. Rocking the Boat: Damage to Eelgrass by Swinging Boat Moorings. Front. Plant Sci. 8:1309. doi: 10.3389/fpls.2017.01309 

Valavi, R., Elith, J., Lahoz-Monfort, J. J. & Guillera-Arroita, G., 2019. Block CV: An r package for generating spatially or environmentally separated folds for k-fold cross-validation of species distribution models. Methods in Ecology and Evolution, 10(2), 225–232. 

Wong, M.C., Vercaemer, B.M. & Griffiths, G., 2021. Response and Recovery of Eelgrass (Zostera marina) to Chronic and Episodic Light Disturbance. Estuaries and Coasts 44, 312–324 (2021). https://doi.org/10.1007/s12237-020-00803-3 

Zieman J.C., & Wetzel R.G. 1990. Productivity in seagrasses: Methods and rates. In Phillips RC, McRoy CP, editors. Handbook of seagrass biology: An ecosystem perspective. New York: Garland STPM Press; pp 87–116. 

Authors

Lead authors 

Adam Smith, Jenny Booth, Laura Pettit and Cristina Vina-Herbon (JNCC) 

Supporting authors 

Anita Carter, Ashley Cordingley (JNCC) 

Contributors: 

Matt Service (AFBI) 

Tim Mackie (DAERA) 

Andrew Scarsbrook, Edward Wright (Defra) 

Graham Phillips (Environment Agency) 

Adam Britton, Helen Woods, Joe Kenworthy, Kirsty Woodcock, Lauren Molloy, Liam Matear, Marco Fusi, Sofie Voerman, Stefano Marra, Steven Duncombe-Smith (JNCC) 

Phil Boulcott, Rebecca Langton, Scott Gray (Marine Directorate of the Scottish Government) 

Eunice Pinn, James Dargie, Roddy MacMinn, Peter Webster (NatureScot) 

James Highfield (Natural England) 

Karen Robinson, Mike Camplin, Ben Wray, Harriet Lincoln, Harry Goudge, Charles Lindenbaum (NRW) 

Assessment metadata

Assessment Type
Point of contact email	marinestrategy@defra.gov.uk
Metadata date	Monday, January 1, 0001
Title
Resource abstract
Linkage
Conditions applying to access and use
Assessment Lineage
Dataset metadata
Dataset DOI	Please contact marinestrategy@defra.gov.uk

The Metadata are “data about the content, quality, condition, and other characteristics of data” (FGDC Content Standard for Digital Geospatial Metadata Workbook, Ver 2.0, May 1, 2000).

Metadata definitions

Assessment Lineage - description of data sets and method used to obtain the results of the assessment

Dataset – The datasets included in the assessment should be accessible, and reflect the exact copies or versions of the data used in the assessment. This means that if extracts from existing data were modified, filtered, or otherwise altered, then the modified data should be separately accessible, and described by metadata (acknowledging the originators of the raw data).

Dataset metadata – information on the data sources and characteristics of data sets used in the assessment (MEDIN and INSPIRE compliance).

Digital Object Identifier (DOI) – a persistent identifier to provide a link to a dataset (or other resource) on digital networks. Please note that persistent identifiers can be created/minted, even if a dataset is not directly available online.

Indicator assessment metadata – data and information about the content, quality, condition, and other characteristics of an indicator assessment.

MEDIN discovery metadata - a list of standardized information that accompanies a marine dataset and allows other people to find out what the dataset contains, where it was collected and how they can get hold of it.