Ecological implications of removing a concrete gas platform in the North Sea

Artificial structures such as offshore oil and gas platforms can significantly alter local species communities. It has been argued that this effect should be considered during decisions over their removal during decommissioning. In the North Sea, leaving such structures in place is prohibited but derogations are allowed for large concrete installations. To assess removal options for one such installation, the Halfweg GBS (gravity-based structure) a concrete platform foundation off the Dutch coast, we studied the resident fouling macrofauna community. The faunal structure, biomass and trophic composition of the Halfweg was then compared with those from the surrounding seabed sediments, other local artificial structures and a natural rocky reef. In total, 65 macrofaunal species were observed on the concrete (52 species), steel legs (32) and surrounding rock dump (44) of Halfweg. Mean Simpson diversity per sample was highest on the rock dump (0.71) but concrete (0.54) and steel (0.60) of the GBS were lower than seabed (0.69). Ten of the species observed on the concrete were not reported on other substrates while 10 of the species were also observed in the surrounding seabed. The GBS structure was numerically dominated by Arthropoda which comprised 98% of the total abundance. Mean ash free dry weight (AFDW) was significantly higher (p<0.001) on the Halfweg substrates (204 g AFDW per m2) than in the surrounding seabed (65 g AFDW per m2). Over 94% of the biomass on Halfweg consisted of the plumose anemone Metridium senile. While common on other reefs, this species was absent from the surrounding seabed. Macrofaunal feeding mechanisms of the concrete and rock dump communities on the GBS were similar to those of nearby sediments, although these differed from those on the Halfweg steel legs. Therefore, the presence of Halfweg alters the local community feeding modes. Multivariate analysis revealed that taxonomic structure of the GBS and other artificial structures significantly differed from that of the sedimentary habitats. Low numbers of non-indigenous species on Halfweg indicated that the structure does not act as a stepping stone for species invasions. Our data show that the Halfweg structures significantly increase local biodiversity and biomass. Removal of the concrete and steel legs of the GBS (leaving the rock dump) will significantly reduce local macrofauna biodiversity. The long-term impact on macrofaunal biomass is low. Leaving the complete Halfweg structure in place will result in an enriched local macrofaunal biodiversity and feeding mode diversity.

In total, 65 macrofaunal species were observed on the concrete (52 species), steel legs (32) and 26 surrounding rock dump (44) of Halfweg. Mean Simpson diversity per sample was highest on the rock 27 dump (0.71) but concrete (0.54) and steel (0.60) of the GBS were lower than seabed (0.69). Ten of the 28 species observed on the concrete were not reported on other substrates while 10 of the species were 29 also observed in the surrounding seabed. The GBS structure was numerically dominated by Arthropoda 30 which comprised 98% of the total abundance. Mean ash free dry weight (AFDW) was significantly higher 31 (p<0.001) on the Halfweg substrates (204 g AFDW per m 2 ) than in the surrounding seabed (65 g AFDW 32 per m 2 ). Over 94% of the biomass on Halfweg consisted of the plumose anemone Metridium senile. 33 While common on other reefs, this species was absent from the surrounding seabed. Macrofaunal feeding 34 mechanisms of the concrete and rock dump communities on the GBS were similar to those of nearby 35 sediments, although these differed from those on the Halfweg steel legs. Therefore, the presence of 36 Halfweg alters the local community feeding modes. Multivariate analysis revealed that taxonomic 37 structure of the GBS and other artificial structures significantly differed from that of the sedimentary 38 habitats. Low numbers of non-indigenous species on Halfweg indicated that the structure does not act as 39 a stepping stone for species invasions. 40 Our data show that the Halfweg structures significantly increase local biodiversity and biomass. Removal 41 of the concrete and steel legs of the GBS (leaving the rock dump) will significantly reduce local 42 macrofauna biodiversity. The long-term impact on macrofaunal biomass is low. Leaving the complete 43 Halfweg structure in place will result in an enriched local macrofaunal biodiversity and feeding mode 44 diversity. 45

46
The presence of artificial structures such as oil and gas platform or wind turbine foundations in the 47 marine environment induces significant changes on the local species diversity (Dannheim et al., 2020;48 Fowler et al., 2018). These de facto artificial reefs attract a community of epifouling species (Goddard 49 and Love, 2010; Krone et al., 2013;Picken, 1986), fishes (Fujii, 2015;Love et al., 2003;Pradella et al., 50 2014) and mammals (Russell et al., 2014). Differences as well as similarities have been reported 51 between artificial and natural reefs (Coolen et al., 2018;Dannheim et al., 2020;Wilhelmsson and Malm, 52 2008). In general, the presence of these installations is considered to increase local biodiversity 53 (Dannheim et al., 2020) and population connectivity (Coolen et al., 2020;Henry et al., 2018; van der 54 Molen et al., 2018). The structures have been suggested to act as stepping stones which can influence 55 the distribution of native species (Friedlander et al., 2014) as well as non-native species (Yeo et al.,56 2010). Apart from their positive effects on biodiversity, artificial structures can have a positive impact on 57 marine food webs. Studies have shown that fish species are attracted towards such structures due to the 58 increased prey abundance, i.e. fouling fauna (Reubens et al., 2013(Reubens et al., , 2011. Furthermore, the introduction 59 of scour protection layers increases the local food web complexity, supporting a high diversity of trophic 60 levels (Mavraki et al., 2020a). The biodeposition processes of fouling organisms create organic-matter 61 rich soft sediments near the base of offshore wind foundations, which in turn increases the abundance 62 and species richness of the macrofaunal communities (Coates et al., 2014). Finally, fouling organisms 63 are responsible for a negligible reduction of the local primary producers (Mavraki et al., 2020b). All these 64 suggest that artificial structures could have beneficial effects on the local food web properties. The 65 ecological importance of these structures should, therefore, be considered when decisions are taken 66 regarding their removal as part of their decommissioning processes (Fowler et al., 2018). 67 Worldwide, a high number of oil and gas installations are scheduled to be decommissioned in the coming 68 years (Fowler et al., 2018). During decommissioning, structures can be removed and brought ashore for 69 scrapping or reuse (Schroeder and Love, 2004). In some cases, the structure foundations are left in 70 place (Bull and Love, 2019) or relocated to be used as artificial reefs (Picken et al., 2000). In the North 71 Sea region, the OSPAR decision on the Disposal of Disused Offshore Installations (OSPAR Commission, 72 1998), dictates that "The dumping, and the leaving wholly or partly in place, of disused offshore 73 installations within the maritime area is prohibited". However, derogations of this prohibition are allowed 74 if assessment by the relevant competent authority "shows that there are significant reasons why an 75 alternative disposal [..] is preferable". One of the exclusion structures considered in this decision is the 76 foundation of gravity based concrete installations. 77 In the Netherlands, the removal obligation for installations has been embedded in the Mining Act 78 (Kingdom of the Netherlands, 2020) which states that an unused mining installation is to be removed. 79 Currently, 160 oil and gas production installations are present within the Dutch North Sea (de Vrees, 80 2019). Most of these consist of steel jacket structures but a few Dutch installations are constructed using 81 a concrete gravity-based foundation. To date, none of these concrete structures have been removed 82 from the Dutch North Sea, and no derogations from the OSPAR decision have been proposed. In 2016, a 83 proposal to leave two jacket foundations in place for 15 years to gain insights into the ecological effects 84 of leaving structures in place after decommissioning (Maslin, 2016) was withdrawn for financial and 85 permitting reasons. Thus, data regarding the ecological implications of removing such structures as 86 opposed to leaving them in situ during decommissioning in this part of the North Sea is still lacking. 87 One of the few concrete foundations in Dutch waters is the Halfweg gravity-based structure (GBS). The 88 Halfweg gas production platform was built in 1995 and operated by Petrogas E&P Netherlands B.V., the 89 Netherlands (henceforth: Petrogas). The GBS was never used to store hydrocarbons and only functioned 90 as a foundation to the platform. Gas production on the platform ended in 2016, followed by its 91 decommissioning. Although the original intention was to remove the GBS together with its supporting 92 steel legs and topside, a collision by a gas tanker vessel in December 2017 damaged the legs, preventing 93 the lift of the whole structure at once (personal communication Alan Shand, Petrogas). Therefore, the 94 topside and legs were removed from the GBS in January 2019 and the concrete structure remained on 95 the seabed. Currently, Petrogas is evaluating options for the removal of the GBS and is considering four 96 options. Three of these options involve the complete removal of the GBS using different methods to lift 97 it, scattering approximately 50% of the surrounding rock dump in the area where the GBS is now, while 98 the fourth option consists of leaving the whole GBS structure in situ. 99 To provide empirical data to aid decisions regarding the decommissioning of the GBS, we conducted a 100 survey to acquire data to allow a comparison of the structural and functional (biomass and feeding  101  modes) characteristics of the macrofaunal assemblages of the GBS structure and its associated rock  102  dump with those of the surrounding sedimentary habitats. To place the ecological importance of the GBS  103  into a wider context, these macrofaunal attributes were also compared with those of other artificial  104  structures in the region including the steel piles of wind turbines and a natural rock reef. To aid this  105 comparison the four removal options were reduced to two scenarios, these were: 106 1. The GBS is fully removed and the surrounding rock dump is partly scattered across the area, 107 leaving the other part of the rock dump untouched; 108 2. The GBS is left in situ with the rock dump remaining unaltered. 109 Model representation of the concrete GBS as it currently lies on the seabed (rock dump not shown). 127

MATERIALS AND METHODS
Image provided by Petrogas E&P Netherlands. 128 129 It is placed on the seabed in a south-east to north-west orientation and it is surrounded by a rock dump 130 of different sizes (10-70 cm diameter; Figure 3) in a radius of approximately 15-20 m from the GBS and 131 a height of up to 6 m, covering a total seabed area of approximately 2,889 m 2 ( Figure S1). Since rock 132 dumps form complex structures it was assumed that the hard substrate area available for macrofauna 133 was larger than the area of seabed covered. The rock surface area available for macrofauna was 134 calculated by generating 1,000,000 rocks of random size between 10-70 cm length, width and height, 135 calculating the approximated surface area per rock (Graham et al., 1988), in comparison with the seabed 136 covered by that rock, assuming a vertically projected rectangle shaped area (length*width of each rock). 137 This ratio was averaged to 3.59 m 2 rock per m 2 seabed and multiplied by the 2,889 m 2 of seabed 138 covered. It was assumed that the top layer of rocks was fully available for macrofauna on all sides. 139 Unavailability of the surface on rocks touching each other, additional surface available on rocks below the 140 first layer as well as cover by sand of rocks on the edges of the rock dump area were not considered. The 141 estimated total surface area of rocks available to macrofauna was 10,389 m 2 . Most hard surface present 142 at Halfweg is composed of concrete and the rock dump, with a small steel surface available in the form of 143 the remains of the 1.7 m diameter legs, which are between 2 and 3.5 m in height above the GBS ( Figure  144 2). The hollow legs, which intrude 6 m into the GBS, are open on the top allowing their inner surface to 145 provide a potential substratum for macrofaunal colonisation. Total steel surface area available for marine 146 growth excluding the deeper, internal regions where limited water exchange results in oxygen-and 147 nutrient-depleted waters, was estimated to be 120 m 2 . 148 The spatial area of seabed occupied by the complete GBS structure is approximately 3,617 m 2 . The 149 surrounding seabed habitats within a 30 km radius of the structure are largely composed of sand with 150 patches of coarse substrate mostly located north of the GBS (EMODnet, 2020). Depth varies between 0 151 m to the east to 33 m in the west of the area (Figure 1). To the northeast the area borders the Wadden 152 Sea, which is enclosed between islands (e.g. Texel; Figure 1 Modifications to the samplers described therein were as follows. The putty knife was attached to the 173 airlift via a flexible ribbed hose (50 mm internal diameter). This was connected to a 48 mm (outer 174 diameter) stainless steel pipe to which the air inlet was connected. The flow of air was regulated by a 175 needle valve, which was connected to a scuba tank regulator by a 10-bar pressure hose. The stainless-176 steel pipe was connected to a straight 50 mm outer diameter pvc pipe ending in a 180⁰ bend on top, 177 made of 75 mm outer diameter PVC. This then ended into a sample net with a mesh size of 0.5 mm. The 178 total length of the airlift above the air inlet was 150 cm. The screw cap nets, which were replaced for 179 each sample, allowed an easy exchange during the dive. During a 50 minute dive, a two-person dive 180 team was able to collect up to 12 samples. 181 After collection, the samples were processed on board by depositing the collected macrofauna on a 212 182 µm mesh sieve, rinsing the nets with seawater to isolate all specimens. Ethanol (99%), measuring at 183 least twice the volume of fauna, was then added to each sample. Within 48 hr, the ethanol was drained 184 from each sample and replaced with fresh 99% ethanol, again with a volume of at least twice that of the 185 fauna. 186 All samples were processed in the benthic laboratories of Bureau Waardenburg b.v. and Wageningen 187 Marine Research. Each sample was sorted into major taxonomic groups, after which all specimens were 188 identified to the lowest possible taxonomic level, mainly to species level. Where more than 200 189 individuals of the same species were present in a sample, species counts were undertaken by 190 subsampling to a level where between 100 and 200 individuals of the species were left in a subsample. 191 After identification, the individuals from each non-colonial species from every sample with a wet weight 192 >0.01 g were ash free dry weighed (6 hr at 500°C after drying) using a Precisa Gravimetrics prepASH 193 340 series. Colonial species such as Bryozoa, Hydrozoa, Porifera and Tunicata were not weighed or 194 counted, but the area covered in a horizontal plane per species per sample was estimated to the nearest 195 cm 2 by flattening the species on grid paper.

Seabed data 203
To allow a comparison of the macrofaunal assemblages present on the Halfweg GBS with those of its 204 regional setting, seabed macrofaunal data were acquired from the were within range of 30 km from the Halfweg structure and were sampled using box corer and sieved on 211 a 1 mm mesh (no smaller mesh size was available) were included. We appreciate that differences in 212 sampling gear and mesh sizes used during sample processing with those from the GBS result in 213 difficulties over direct comparisons with the data. Thus, caution will be applied when these results are 214 compared with those of the GBS. The resulting dataset held 2,003 records from 118 samples. These 215 samples had been taken between 1991 and 2015, with yearly samples between 1991 and 2010 as well 216 as samples in 2012 and 2015. 217

Artificial & natural reef data 218
A set of published marine growth data from reefs in the Dutch part of the North Sea (Coolen et al., 2018) 219 was used to assess the uniqueness of the species on Halfweg on a larger scale. These data were acquired 220 from scraped samples from five oil and gas structures, a wind farm and a rocky natural reef in the 221 Borkum Reef Grounds. The locations are between 32 and 184 km distance from Halfweg. The distance 222 from Halfweg to some of these locations, in particular to Borkum (164 km) and the D15-A platform (184 223 km) is large, but it is the most proximate dataset available that includes geogenic reef formations as well 224 as the most detailed dataset on oil and gas platform fouling communities in the North Sea. The included 225 artificial structures include ages both lower and higher than the 25 yr Halfweg has been in place, with an 226 average age of 23 yr. No information on whether the installations had ever been treated with anti-fouling 227 coatings was available. Although marine growth removal for inspection could have had an impact on the 228 communities, previous analysis showed that whether an installation had been cleaned recently, had no 229 significant effect on the species richness and only accounted for 0.3% of the variation (Coolen et al.,  230 2018). Samples were taken by divers using a similar airlift as used for Halfweg (the platforms, rocky 231 reef) or a sampling net (wind farm). The dataset from 145 samples contained only species abundances 232 (individuals per m 2 or presence-only for colonial species) and no biomass data. 233

Data preparation 234
For data preparation and analysis, R version 3.6.1 (R Core Team, 2019) and Rstudio version 1.2.5001 235 (RStudio, 2019) were used. Prior to the analysis, all data were updated to include the most recent 236 species names as published on the World Register of Marine Species (WoRMS Editorial Board, 2019), 237 using the wormsbynames function from the worms package (Holstein, 2018). For the seabed data from 238 the 30 km radius around Halfweg, seabed depth was obtained from EMODnet bathymetry data 239 (EMODnet, 2019) using the extract function from the raster package (Hijmans, 2019). This resulted in a 240 mean depth of 22 m from a range of 9 to 33 m. Mean depth of the Halfweg data was 21 m from a range 241 of 17 to 24 m. 242 When samples included specimens that were not identified to species level, their abundance was added 243 to a species in the lowest common higher taxon or removed when more than one species was present in 244 the lowest common higher taxon in the same sample. Only individuals in samples with no species in a 245 common higher taxon were left at the higher level (Coolen et al., 2018(Coolen et al., , 2015a. Since macrofaunal 246 densities and weights in the seabed data were given per m 2 , all the marine growth data were converted 247 to values per m 2 . 248

Feeding traits 249
To provide a comparison of the feeding modes of the assemblages across the different datasets, the 250 numerical composition of each assemblage across major feeding modes was assessed using biological 251 traits. Each species was categorised across one or more of five feeding mode traits (suspension-feeders, 252 deposit-feeders, predators, scavengers, parasites) using a fuzzy-coding approach based on the traits 253 information used by Bolam et al. (2017Bolam et al. ( , 2016. Fuzzy-coding allows the multi-faceted feeding behaviour 254 of many species to be accounted for and overcomes the need to confine each species to a single mode of 255 feeding. The feeding mode composition of each assemblage was calculated based on the most abundant 256 species that, in total, accounted for >90% of the total abundance within each habitat. 257

Data analysis 258
For each sample, species richness, Simpson biodiversity index (Simpson, 1949)  Therefore, to compare total species richness among substrates, the mean total species richness was 264 calculated from subsets of the seabed data and of artificial structures data. Subsets of 39 samples were 265 randomly selected from each of these datasets. This process was repeated 10,000 times for both, and for 266 every repetition the subset was used to calculate total species richness in all samples as well as 267 extrapolated species richness based on the Chao estimate (Chao, 1987)

292
All data underlying the results presented here are available as online supplement S2. 293

294
In total, 65 species were observed on the Halfweg GBS and its associated rock dump. This included 52 295 species found on the concrete GBS, 44 species on the surrounding rock dump and 32 on the steel legs. 296 Based on the extrapolated species richness, it was predicted that a total of 83 ± 11 (standard error) 297 species are present on the combined structures of Halfweg, indicating that the survey approach resulted 298 in an under-sampling of between 7 and 29 species ( Most substrate types revealed species that were not observed on any of the other substrates (Table 1). 319 On the concrete of the GBS, 10 unique species were observed that were not observed on Borkum, other 320 structures and the seabed. Six of them were also observed on the rock dump and steel, showing that the 321 other four unique species on the concrete were not found on the rock dump, nor on steel. On rock dump, 322 nine unique species were observed, out of which three species were not found on the GBS. The steel legs 323 contained two additional unique species (out of four) that were not observed on the concrete and rocks. 324 In Euler diagram presenting overlap in species between substrates (ellipses) for the strongest relations. 342 Numbers present the total number of species per substrate (numbers in ellipse area without overlap) or 343 shared between substrates (in the area of overlap between ellipses The plumose anemone Metridium senile accounted for the high AFDW on concrete (183 (96%) ± 24 g 390 AFDW per m 2 ) and rock dump (245 (99%) ± 37 g AFDW per m 2 ) of the GBS (Table 3). M. senile was also 391 responsible for the high dominance of Cnidaria in most samples collected from the concrete and the rock 392 dump (Figure 7). This species was not observed on the sandy seabed. 393 The biomass of the surrounding seabed was dominated by Mollusca (50 g AFDW per m 2 ).    The total biomass on the concrete GBS was estimated to be 262 kg AFDW (based on the average weight  411  times the available substrate), 20 kg on the steel legs and 2,564 kg on the surrounding rock dump.  412 Therefore, the total weight of the macrofauna on the Halfweg substrates was 2,846 kg. The area of 413 seabed covered by the GBS by both hard substrates was estimated to be 3,617 m 2 which, based on the 414 mean total biomass per m 2 in the seabed, would hold a total macrofauna weight of 232 kg if Halfweg was 415 absent. 416

417
The feeding trait analysis indicated similarities between the concrete and rock dump of Halfweg and the 418 seabed, where deposit feeding was similar (rock and seabed) or higher (concrete) to suspension feeding 419 (Figure 8). On all other hard substrates, including the steel legs of Halfweg, suspension feeding was the 420 main feeding trait, followed by scavenging and predation, while deposit feeders were very few or even 421 absent (other steel structures). therefore, increased local species richness by 55 (37% compared to 118 seabed samples, 53% when 451 compared to 39 seabed samples). On the complete Halfweg structure, totalling 3,617 m 2 , the 452 macrofauna AFDW biomass was 12 times that of the comparable area of sandy seabed. This increase 453 results from a combination of a higher biomass per m 2 of area available to fouling species and from the 454 three-dimensional structural complexity of the concrete and rock dump increasing the available area 455 compared to a relatively flat sedimentary seabed (Calow, 1972 and wind turbine foundations), compared to an increased dominance of deposit feeders on the concrete 501 and rock on the GBS which was more in line with assemblages of the surrounding seabed. The 502 similarities between the concrete and the seabed, and the evident differences between Halfweg concrete 503 and the other artificial hard substrates may result from a number of factors. The similarities with the soft 504 seabed could be the result of the combination of different deposit feeding mechanisms, i.e. surface and 505 subsurface deposit feeders, into one general deposit feeding trait. Separating the deposit feeding 506 mechanisms into these two categories could have resulted in more differentiation between these two 507 habitats, since there is a lack of sub-surface deposit feeders on hard substrates and since the most 508 abundant arthropod on the seabed, the burrowing amphipod species Urothoe poseidonis, is a sub-surface 509 deposit feeder (Kröncke et al., 2013). The high relative abundance of deposit feeding organisms on the 510 Halfweg concrete compared to the other artificial structures could be the after-effect of the colonisation 511 of the jacket foundation that was on top of this concrete substrate for a prolonged period which was 512 removed eight months before sampling. Fouling organisms inhabiting vertical hard substrates such as oil 513 and gas platforms and offshore wind turbines produce biodeposits that sink on the seabed, causing very 514 local accumulation of organic matter (Coates et al., 2014 The data acquired during this study allow a simplistic comparison of the implications of the two options 548 on localised macrofaunal diversity to be undertaken. In the first scenario, all the fauna that is currently 549 present on the concrete GBS will be removed or scattered during removal, in essence resulting in the 550 removal of most of the species present on the concrete. An estimated 50% of the rock dump will be 551 placed elsewhere or dredged up (personal communication Gert de Raadt, Petrogas). When moved by 552 crane, some of the macrofauna may remain on the rock dump, although it is unlikely that the rocks will 553 be placed in an identical orientation, likely resulting in the burial or damaging of most fauna present on 554 the moved rock dump. Assuming 50% of the rock will remain untouched, the fauna present on these 555 rocks will mostly remain. On a longer term, the moved rock dump is likely to be colonised by a similar 556 community as is currently present, originating from the untouched rock dump, small hard substrates on 557 the seabed or other source locations where the current species originate from, e.g. by colonisation by 558 settling larvae or via migration of juveniles and adults (Coolen et al., 2020; Krone and Schröder, 2011; 559 Luttikhuizen et al., 2019). In this instance, species richness may recover to approximately the 44 species 560 that were observed on the rock dump in this study. This scenario will therefore result in the loss of 21 561 macrofaunal species that are currently present on the concrete and steel but not on the rock dump. 562 However, four of these 21 species have also been observed in the seabed, so the net loss of species 563 richness to the local area is predicted to be 17 species or more as the observed species richness is, as we 564 have demonstrated, not exhaustive. No unique species could be calculated for extrapolated species 565 richness but given that these numbers are higher than observed richness the net loss might be higher 566 than 17. Since after the removal of the GBS, the rock dump will be scattered across the area, the total 567 area of substrate available to fouling communities could remain approximately like what is presently 568 available on the GBS plus rock dump. Assuming this would be similarly colonised, this may result in a 569 total biomass comparable to the current biomass. Feeding mode diversity is likely to be reduced when 570 Halfweg GBS is removed, as the communities on the steel parts of the structure show different dominant 571 feeding traits than present on the seabed. This would lead to a shift towards suspension feeding, 572 reducing the functional evenness since species share a specific functional trait (Mason et al., 2005). 573 In the second scenario, the lack of management intervention will results in no species or biomass loss. 574 The GBS and the rock dump have been in place since 1996 and have possibly reached, although this was 575 not possible to assess in this study, a stable stage of succession (Oshurkov, 1992

580
The presence of Halfweg, including its different materials in the form of steel, concrete and rock, has a 581 clear effect by increasing local species richness as well as feeding mode diversity. Removal of the 582 gravity-based foundation of Halfweg will result in the loss of a significant number of species from the 583 local area and possibly a lowering of feeding mode diversity. Due to the scattering of the rock dump after 584 removal, local impact on macrofouling biomass is considered low. The option to leave the GBS in place as 585 it is will result in the highest number of species maintained in the local area. 586 Tjalling van der Wal (Wageningen Marine Research) were involved in our lab work or data-processing 592 and we thank them for their help. The work reported in this publication was funded by Petrogas E&P 593

ACKNOWLEDGEMENTS
Netherlands B.V.. We thank Alan Shand and Gert de Raadt at Petrogas for making this work possible. We 594 thank Ruud Schulte and Michiel Harings at EBN for their comments on our analysis and outcome. We 595 thank Prof. em. Han Lindeboom for commenting on our work and providing interesting viewpoints on the 596 topic. An earlier version of this text was reviewed by Oscar Bos (Wageningen Marine Research), we 597 thank him for his help improving the manuscript. We thank our anonymous reviewers for their 598 constructive comments on the earlier versions of this text. 599 600