Deep-sea demersal fish species richness in the Porcupine Seabight, NE Atlantic Ocean: global and regional patterns


Imants G. Priede, Oceanlab, University of Aberdeen, Newburgh, Aberdeenshire AB41 6AA, UK.


The ichthyofauna of ocean margin regions is characterised by a succession of different species occurring at different depths. This study was aimed at determining whether the resultant pattern of species richness with depth is a consequence of local factors in a given region or whether it simply reflects the global pattern of fish species distribution in the oceans. Along the ocean margin of the temperate NE Atlantic Ocean in the Porcupine Seabight and Abyssal Plain region, 48°–53°N, a total of 108 demersal fish species were identified from 187 trawls at depths from 240 to 4865 m. Fitting of species accumulation curves predicted an asymptote of 120, indicating that the fauna is 90% described. Baited cameras detected 22 scavenging species with a predicted asymptote of 24 species. Scavenging species represented a constant 22.7% (SD 3.5%) of the total species richness throughout the depth range studied. Species richness per trawl varied between a maximum of 16 at 1600 m and 4 on the abyssal plain > 4000 m with no significant influence of sea floor slope (a measure of topographic heterogeneity). Total species richness was 48 at 1600 m and 10 on the abyssal plain. There is a clear transition between slope species above 3000 m and abyssal species below. The depth at which peak species richness occurs (1100–2000 m) coincides with the depth of the permanent thermocline, presence of Mediterranean overflow water (MOW), seasonally strong currents, resuspension of particulate matter, high biomass of benthic filter feeders and pelagic biomass impinging on the slope. We suggest that these factors increase habitat and resource heterogeneity, thus supporting a wider range of fish species. The local pattern of species richness was compared with the global distribution of maximum depths of marine fish species from FishBase. Globally all three Classes of fishes, Agnatha, Chondrichthyes and Osteichthyes, showed a logarithmic decrease in species with depth, with the deepest observed species in each class occurring at 3003 m, 4156 m and 8370 m, respectively. In contrast, the local distribution of species maximum depths is idiosyncratic with a mean of 16.6 species maxima per 500 m at 1000–3000 m depth followed by three species per 500 m at 3500–4000 m and 11 species per 500 m at 5000 m. It is concluded that global patterns of species richness, as a source of recruitment, exert a weak influence on local patterns of species richness. Rather, global species richness is the sum of numerous regional and local patterns, each determined by characteristic environmental conditions.


Bottom-living or demersal fishes are a conspicuous component of the fauna of deep-sea ocean margins around the North Atlantic Ocean. Distinct patterns of species diversity have been observed in different regions (Haedrich & Merrett 1988; Koslow 1993; Merrett & Haedrich 1997) and the question arises regarding the role of environmental heterogeneity in generating those disparate patterns. Levin et al. (2001) proposed a balance between global, regional (landscape) and local (1–10 m2 patch) scales to explain patterns of deep-sea species diversity. In this paper we compare the patterns of global diversity of fishes with data for demersal fishes of the Porcupine Seabight and Porcupine Abyssal Plain region of the NE Atlantic as sampled by otter trawl and baited cameras.

Globally it is estimated that there are 19,082–20,518 (95% confidence limits) species of marine fish in existence, of which 15,716 have been described (Mora et al. 2008). Of these, the demersal (bottom-living fishes) and bathydemersal (demersal fish living deeper than 1000 m) groups with 8240 species described worldwide are the most speciose compared with 3235 pelagic, bathypelagic and benthopelagic species and 4241 reef-associated species. In the North Atlantic basin, Merrett (1994) recognised 589 pelagic species and 505 demersal species, 18.2% and 6.1% of the global fauna, respectively. The lower percentage of the global total indicates that the demersal ichthyofauna is regionally more specialised than the pelagic fauna, making them excellent candidates for a study on regional environmental heterogeneity and its influence on species richness. The global fauna can be regarded as a reservoir from which the fauna in a given locality may have been recruited. In this paper we consider some aspects of global diversity of fishes and then focus on the demersal fishes of the Porcupine Seabight. This area has been extensively studied through the use of trawl sampling (Merrett et al. 1991a,b) and baited cameras (Priede et al. 1994). In this paper we include new data leading to a cumulative total of 187 trawls between 1977 and 2002 and results from 33 baited camera lander deployments.

The specific questions addressed are: (i) Does the species richness of demersal fish follow similar depth patterns on regional and global scales? (ii) If regional patterns differ from the global pattern, is it possible to identify regional drivers of these differences?

Material and Methods

Global data

Archive data for depth of occurrence of marine and brackish water fish (excluding anadromous and catadromous species) were abstracted from global datasets available on FishBase (Froese & Pauly 2008).Records of maximum depth and maximum total body length were accepted for 68 species of Agnatha (jawless fishes) including the addition of a new record for the deepest occurring Agnatha species (Yeh & Drazen 2009), 669 species of Chondrichthyes (cartilagenous fishes), and 8691 species of Actinopterygii (ray-finned bony fishes). Dubious records, where sampling gear traversed a wide depth range and depth of capture was uncertain, were rejected by checking original references for the deepest species (Priede et al. 2006). This gave a total of 9428 species records corresponding to 60% of the global ichthyofauna hitherto described. We assume this is an unbiased representative sample. Some maximum depths may be underestimated because not all collections are represented in FishBase. Scatter plots of species maximum length and maximum depth were made and least square linear regressions were fitted to investigate size–depth trends. For further analysis, numbers of species maximum depths of occurrence per 500 m depth bin (0–500, 501–1000, etc.) were assembled. Linear least squares regressions were fitted to log10 of the species number versus depth.

Regional data – Porcupine Seabight – Abyssal Plain


A 45-foot (13.7-m) semi-balloon otter trawl (OTSB; Marinovich Trawl Co., Biloxi, MS, USA) was fished on a single warp (Merrett et al. 1991b). Nominal spread of the mouth of the trawl (width of seabed sampled) was 8.6 m and haul duration was varied between a bottom contact time of 30 min at the shallowest stations to 3 h on the abyssal plain, with a tow speed of 2–2.5 knots. This approximately equalised the sizes of catches across the depth range. The mean area of sea floor swept by the trawl was 59,978 ± 32,923 m2 (SD, n = 187). This small trawl was effective at catching a wide range of species, but because of the limited herding action and low towing speed, large and highly mobile species such as sharks and black scabbardfish were poorly represented in the catch (Merrett et al. 1991a; Gordon et al. 1996). Substantial overlaps in the personnel present on the 20 trawl cruises ensured consistency of techniques and fish identification. Voucher specimens of fish whose identification was uncertain were retained and verified with reference to museum collections. Data from a total of 187 trawls at depths from 240 to 4865 m included in this analysis were taken in two periods: period 1, 1977–1989 (127 trawls), and period 2, 1997–2002 (60 trawls). All trawls were conducted in the area of the Porcupine Seabight and adjacent Porcupine Abyssal Plain (Fig. 1). Cruises were in different seasons to avoid biasing of sampling in relation to any fish migrations that might occur (Priede et al. 2003) (Table 1). As an index of environmental heterogeneity for each trawl sample, the slope of the sea floor was determined by measuring the maximum slope at right angles to the contour lines, on large-scale charts of the area (Hunter & Kenyon 1984). Horizontal distances between 100-m and 200-m depth intervals, depending on slope, were measured.

Figure 1.

 Sample locations in the NE Atlantic Ocean. Upper panel – The NE Atlantic Ocean with the red square indicating the study are shown in detail below. Lower Panel – The Porcupine Seabight and Porcupine Abyssal Plain. Black contour line – 200 m depth. Colour depth scale transitions are at 1000, 2000, 3000 and 4000 m depth. Red symbols – trawl locations. Yellow symbols –baited lander locations.

Table 1.   List of sampling trawl sampling cruises in the Porcupine Seabight and Porcupine Abyssal Plain study area with dates of the first and last trawl on each voyage. Cruise reference numbers: C prefix –RRS Challenger, D prefix –RRS Discovery.
cruise of 1st trawldate of last trawlyearno. trawls
D889 November13 November19772
D927April24 April19788
C5051 June9 June197914
C5061 July8 July19796
C50713 October21 October19798
C50830 July9 August198020
C5097 November10 November19807
C5101 May10 May198118
C51216 September29 September19817
C51317 February24 February198213
C51424 March1 April19828
C51619 July21 July19824
C51825 September30 September19838
C5245 December5 December19861
D1855 September12 September19893
C1349 August16 August19979
D25018 September7 October20007
D25215 April27 April200110
D25516 August01 September200114
D26011 March22 March20028
D26630 September19 October200212

Baited camera

During 1989–2004 sampling was supplemented with the use of baited cameras to identify active scavenging species (Priede et al. 1994; Smith et al. 1997; Heger et al. 2007; C. Henriques, unpublished data, Oceanlab, University of Aberdeen). These were deployed using a free-fall lander technique (Priede & Bagley 2000). A piece of bait (usually mackerel weighing c. 0.5 kg) was deployed on the sea floor within the field of view of a downward-looking video or time-lapse stills camera (film or digital) and images of fish approaching, consuming and departing from the bait were recorded for up to 12 h. The system was recovered by acoustic command from the ship and images were analysed. Species were identified by reference to standard texts and comparison with voucher specimens captured in trawls.

Species accumulation curves

To estimate the total number of demersal fish species in the sample area the rate of discovery of new species as sample number increased was examined. Averaged species accumulation curves were calculated for the trawl and lander data by adopting a randomisation procedure, in which the sample order was randomised 100 times and the mean species richness determined for each value of n (n = trawls 1–187 or lander deployments n = 1–33) to produce a smoothed species accumulation curve (for details see Colwell & Coddington 1994; Gotelli & Colwell 2001).

This avoided variations in the slopes of the curve resulting from heterogeneity or sample variation that was evident when data were plotted simply in chronological order. The analyses were performed using the ‘Vegan’ package (Oksanen et al. 2008) in the r statistical programming environment (R Development Core Team 2007).

Statistical analysis

Statistical models were developed to investigate the determinants of demersal fish species richness at the Porcupine Seabight and Abyssal Plain. We used a generalized additive model (GAM) with a negative binomial (NB) distribution and a log link (Hardin & Hilbe 2007; Zuur et al. 2007), and we incorporated log swept area of the trawls as an offset variable (McCullagh & Nelder 1989; Zuur et al. 2009) to investigate the relationship between fish species richness and trawling depth (mean water depth of the trawl whilst on the seabed). GAM was used because initial data exploration revealed non-linear relationships and the negative binomial distribution was applied because there was overdispersion. Quasi-Poisson GLM was considered, but not used as these models do not allow the use of model selection tools such as AIC (Akaike Information Criteria, McCullagh & Nelder 1989). The analysis was performed using the ‘MGCV’ (Wood 2008) and ‘MASS’ (Venables & Ripley 1999) packages in the R statistical programming environment (R Development Core Team 2008). The initial model included the explanatory variables, ‘Period’ (Period 1 = 1977–1989, Period 2 = 1997–2002) as a nominal variable, and ‘Slope’ of the seabed (%) and ‘Depth’ (m) were fitted as smoothers. The negative binomial regression model can be summarised as:






where k is the dispersion parameter and Si the swept area of trawl i. The model assumes that log species richness Ri has a negative binomial distribution with the expectation μi, where the variance is larger than the mean. The notation f1,1(Depthi) and f1,2(Depthi) means that the smoothing was based on one Series only and thus each Series had its own smoother for depth. This smoother was fitted using the ‘by’ command in the GAM function (MGCV package). Model selection was applied by sequential removal of non-significant explanatory variables and verified based on AIC.


Global analysis

On a global scale, there is a general logarithmic decrease with depth in the numbers of demersal fish species (Fig. 2): Agnatha: y = −0.000482x + 1.708606, (R2 = 0.98, P < 0.001); Actinopterygii: y = −0.00040x + 3.34672 (R2 = 0.91 P < 0.001); and Chondrichthyes: y = −0.00077x + 2.99690 (R2 = 0.99 P < 0.001) where y is log10 of the number of species maximum depths within a 500-m depth stratum, and x is the depth in metres. The Agnatha and Actinopterygii decrease two- to threefold per 1000 m but Chondrichthyes decrease sixfold per 1000 m. Agnatha and Chondrichthyes become essentially extinct at the bathyal-abyssal boundary at 3000 m, with only the Actinopterygii able to colonise the abyssal (3000–6000 m) and hadal (> 6000 m) regions of the oceans. The maximum depth of occurrence of Agnatha is 3003 m, with the hagfish Eptatretus carlhubbsi recorded at that depth from the Hawaiian region of the North Pacific Ocean (Yeh & Drazen 2009), whilst the maximum depth for Chondrichthyes is 4156 m, represented by the ray (Rajella bigelowi). The deepest bony fish (Actinopterygii) is an ophidioid (Abyssobrotula galatheae), which was trawled from 8370 m.

Figure 2.

 Number of species maximum depths per 500 m depth increment. Green symbols – Agnatha (jawless fishes) global data, Blue symbols – Chondrichthyes (Cartilagenous fishes) global data, Red symbols – Actinopterygii (Ray-finned bony fishes) global data, Grey symbols – Demersal fish from the Porcupine Seabight and Abyssal Plain (all classes pooled). Note the idiosyncratic pattern compared with logarithmic decrease with depth in the global data.

Globally, the greatest size variation occurs at the shallowest depths and variation decreases with increasing depth (Fig. 3). In addition, there is no significant change in the maximum length with depth in any of the three groups of fish (P > 0.1); the mean maximum total length is 51.0 cm (± 21.7 SD ) for Agnatha, 119.2 cm (± 114.9 SD ) for Chondrichthyes and 32.3 cm (± 39.1 SD ) for Actinopterygii. New size and depth records for hagfish E. carlhubbsi (130–140 cm total length at 3000 m depth) imply a bigger-deeper trend in Agnatha but it is premature on the basis of this single observation by Yeh & Drazen (2009) to argue that this trend is globally significant.

Figure 3.

 Global dataset, maximum lengths (cm) and maximum reported depth (m) of occurrence of fish species. Yellow symbols – Agnatha, Blue symbols – Chondrichthyes, Red symbols, Actinopterygii. TL, total length.

Regional analysis – demersal fish diversity at the Porcupine Seabight – Abyssal Plain

A total of 108 demersal fish species from 39 families were identified from trawls (from 240 to 4865 m) in the study area (Table 2), of which 54 species (half of the total currently described) were caught in the first 12 trawls between 1977 and 1979. The asymptote of the species accumulation curve, 120 species (Fig. 4), indicates that 90% of demersal fish species that can be caught using bottom trawls in the study area have been discovered. For the baited cameras a total of 22 different species were recognized, of which 15 were fully identified (Table 3). The species accumulation curve for bait-attending species reaches its asymptote at 24 species, indicating that 92% have been sampled. All species observed by the baited camera were also captured in trawls, except for the shark Centroscyllium fabricii.

Table 2.   List of species captured in the trawls at the Porcupine Seabight and Porcupine Abyssal Plain between 1977 and 2002. The depth ranges of the individual species are based on maximum and minimum depth of capture by trawl in the present study. Also included are the species ranks in accordance with minimum depth of occurrence; these provide the species key to Fig. 7. English common names are from FishBase (Froese & Pauly 2008).
OrderFamilyspeciesauthoritycommon namedepth range (m)rank
MyxiniformesMyxinidaeMyxine iosFernholm, 1981White-headed hagfish1110–154059
ChimaeriformesRhinochimaeridaeHarriotta raleighanaGoode & Bean, 1895Narrownose chimaera1527–190082
Rhinochimaera atlanticaHolt & Byrne, 1909Spearnose chimaera1360–153369
ChimaeridaeChimaera monstrosaLinnaeus, 1758Rabbitfish695–154134
Hydrolagus affinis(de Brito Capello, 1868)Small-eyed rabbitfish1720–239785
Hydrolagus mirabilis(Collett, 1904)Large-eyed rabbitfish695–205835
LamniformesScyliorhinidaeGaleus melastomusRafinesque, 1810Blackmouth catshark506–84825
Galeus murinus(Collett, 1904)Mouse catshark1024–102457
Apristurus laurussonii(Saemundsson, 1922)Iceland catshark750–144843
SqualiformesCentrophoridaeCentrophorus squamosus(Bonnaterre, 1788)Leafscale gulper shark986–98652
Deania calcea(Lowe, 1839)Birdbeak dogfish740–74040
EtmopteridaeEtmopterus princepsCollett, 1904Great lantern shark1533–172083
Etmopterus spinax(Linnaeus, 1758)Velvet belly lantern shark407–179015
SomniosidaeCentroscymnus coelolepisBarbosa du Bocage & de Brito Capello, 1864Portuguese dogfish1027–172058
Scymnodon ringensBarbosa du Bocage & de Brito Capello, 1864Knifetooth dogfish706–98236
DalatiidaeDalatias licha(Bonnaterre, 1788)Kitefin shark545–136029
HexanchiformesHexanchidaeHexanchus griseus(Bonnaterre, 1788)Bluntnose sixgill shark763–96044
RajiformesRajidaeBathyraja richardsoni(Garrick, 1961)Richardson’s ray1987–253091
Dipturus nidarosiensis(Storm, 1881)Norwegian skate1263–131266
Leucoraja circularis(Couch, 1838)Sandy ray695–69532
Neoraja caerulea(Stehman, 1976)Blue ray1390–139072
Rajella bigelowi(Stehmann, 1978)Bigelow’s ray1000–411853
Rajella fyllae(Lütken, 1887)Round ray808–143046
ElopiformesHalosauridaeHalosauropsis macrochir(Günther, 1878)Abyssal halosaur1440–348575
Halosaurus johnsonianusVaillant, 1888Halosaur1379–144871
NotacanthidaeNotacanthus bonaparteRisso, 1840Shortfin spiny eel470–250418
Notacanthus chemnitziiBloch, 1788Spiny eel685–250031
Polyacanthonotus challengeri(Vaillant, 1888)Longnose tapirfish2410–204094
Polyacanthonotus rissoanus(De Filippi & Verany, 1857)Smallmouth spiny eel740–250041
AnguilliformesSynaphobranchidaeHistiobranchus bathybius(Günther, 1877)Deepwater arrowtooth eel1790–484287
Ilyophis arxRobins, 1976No common name1789–190086
Ilyophis blacheiSaldanha & Merrett, 1982No common name1284–178967
Ilyophis brunneusGilbert, 1891Muddy arrowtooth eel1533–287584
Synaphobranchus kaupiiJohnson, 1862Kaup’s arrowtooth eel407–250016
NettastomatidaeVenefica proboscidea(Vaillant, 1888)Whipsnout sorcerer1527–152781
ArgentiniformesArgentinidaeArgentina silus(Ascanius, 1775)Greater argentine265–6958
Argentina sphyraenaLinnaeus, 1758Argentine/Silver smelt/Lesser argentine240–3301
AlepocephalidaeAlepocephalus agassiziiGoode & Bean, 1883Dusky slickhead1368–256770
Alepocephalus australisBarnard, 1923Small scaled brown slickhead1500–257278
Alepocephalus bairdiiGoode & Bean, 1879Baird’s smooth-head706–250038
Alepocephalus productusGill, 1883Smalleye smooth-head1462–273777
Alepocephalus rostratusRisso, 1820Risso’s smooth-head706–250439
Bathylaco nigricansGoode & Bean, 1896Black warrior2292–229295
Bathytroctes microlepisGünther, 1878Smallscale smooth-head1845–484089
Bathytroctes macrolepisGünther, 1887Koefoed’s smooth-head2500–484098
Bathytroctes michaelsarsi(Koefoed, 1927)Michael Sars smooth-head2410–422296
Conocara macropterum(Vaillant, 1888)Longfin smooth-head1430–267073
Conocara murrayi(Koefoed, 1927)Murray’s smooth-head1440–287574
Conocara salmoneum(Gill & Townsend, 1897)Salmon smooth-head3639–4842100
Mirognathus normaniParr, 1951Norman’s smooth-head4020105
Narcetes stomias(Gilbert, 1890)Blackhead salmon1867–273790
Rinoctes nasutus(Koefoed, 1927)Abyssal smooth-head3639–4842101
Rouleina attrita(Vaillant, 1888)Softskin smooth-head1312–192768
AulopiformesBathysauridaeBathysaurus feroxGoode & Bean, 1883Deepsea lizardfish1519–363980
Bathysaurus mollisGünther, 1878Highfin lizardfish2645–424599
IpnopidaeBathypterois dubiusVaillant, 1888Spiderfish1016–243455
Bathypterois longipesGünther, 1878Abyssal spiderfish4118–4787106
GadiformesMacrouridaeCaelorinchus caelorhincus(Risso, 1810)Hollowsnout grenadier407–131214
Caelorinchus labiatus(Koehler, 1896)Spearsnouted grenadier472–190019
Coryphaenoides armatus(Hector, 1875)Abyssal grenadier2016–486593
Coryphaenoides brevibarbis(Goode & Bean, 1896)Shortbeard grenadier1845–129288
Coryphaenoides carapinusGoode & Bean, 1883Carapine grenadier1500–348579
Coryphaenoides guentheri(Vaillant, 1888)Günther’s grenadier1200–287562
Coryphaenoides leptolepisGünther, 1877Ghostly grenadier1993–486592
Coryphaenoides mediterraneus(Giglioli, 1893)Mediterranean grenadier743–264542
Coryphaenoides profundicolus(Nybelin, 1957)Deepwater grenadier3639–4865102
Coryphaenoides rupestrisGunnerus, 1765Roundnose grenadier706–193237
Echinomacrurus mollisRoule, 1916 4810–4810107
Malacocephalus laevis(Lowe, 1843)Softhead grenadier240–6852
Nezumia aequalis(Günther, 1878)Common Atlantic grenadier472 –205820
Pseudonezumia flagellicauda(Koefoed, 1927)No common name2486–308997
Trachyrincus murrayiGünther, 1887Roughnose grenadier1205–160064
Trachyrincus scabrus(Rafinesque, 1810)Roughsnout grenadier506–154126
Trachyscorpia cristulata echinata(Koehler, 1896)Spiny scorpionfish763–114445
MoridaeAntimora rostrata(Günther, 1878)Blue antimora853–297048
Guttigadus latifrons(Holt & Byrne, 1908)No common name985–186751
Halargyreus johnsoniiGünther, 1862Slender codling545–190030
Lepidion eques(Günther, 1887)North Atlantic codling506–242027
Mora moro(Risso, 1810)Common mora500–131223
MerlucciidaeMerluccius merluccius(Linnaeus, 1758)European hake247–7856
PhycidaePhycis blennoides(Brünnich, 1768)Greater forkbeard265–114410
GadidaeGaidropsarus macrophthalmus (Günther, 1867)Bigeye rockling265–207856
Gaidropsarus argentatus(Reinhardt, 1837)Arctic rockling102410
Molva dypterygia(Pennant, 1784)Blue ling490–146221
Molva macrophthalma(Rafinesque, 1810)Spanish ling470–69517
OphidiiformesCarapidaeEchiodon drummondiiThompson, 1837No common name407–54513
OphidiidaeSpectrunculus grandis(Günther, 1877)Pudgy cuskeel853–429849
Holcomycteronus squamosus(Roule, 1916)No common name4812108
BythitidaeCataetyx alleni(Byrne, 1906)No common name1205–120563
Cataetyx laticepsKoefoed, 1927No common name1144–250060
LophiiformesLophiidaeLophius piscatoriusLinnaeus, 1758Angler265–8089
BeryciformesTrachichthyidaeHoplostethus atlanticusCollett, 1889Orange roughy960–167750
Hoplostethus mediterraneus mediterraneusCuvier, 1829Mediterranean slimehead527–143028
BerycidaeBeryx decadactylusCuvier, 1829Alfonsino50624
ZeiformesOreosomatidaeNeocyttus helgae(Holt & Byrne, 1908)False boarfish146276
ScorpaeniformesSebastidaeHelicolenus dactylopterus dactylopterus(Delaroche, 1809)Blackbelly rosefish247–8537
PsychrolutidaeCottunculus thomsonii(Günther, 1882)Pallid sculpin1016–154154
LiparidaeParaliparis hystrixMerrett, 1983No common name69533
PerciformesEpigonidaeEpigonus telescopus(Risso, 1810)Bulls-eye490–256722
ZoarcidaeLycenchelys alba(Vaillant, 1888)No common name3995103
Lycodes terraenovaeCollett, 1896No common name1217–246265
Pachycara bulbiceps(Garman, 1899)Snubnose eelpout3995104
Pachycara crassiceps(Roule, 1916)No common name1144–250461
TrichiuridaeAphanopus carboLowe, 1839Black scabbardfish853–199347
PleuronectiformesScophthalmidaeLepidorhombus boscii(Risso, 1810)Fourspotted megrim247–7855
Lepidorhombus whiffiagonis(Walbaum, 1792)Megrim247–5454
PleuronectidaeGlyptocephalus cynoglossus(Linnaeus, 1758)Witch293–50612
SoleidaeMicrochirus variegatus(Donovan, 1808)Thickback sole247–5063
Figure 4.

 Species accumulation curves for demersal fish in the Porcupine Seabight and Abyssal Plain. Solid line, trawl data; dashed line, baited camera data.

Table 3.   List of species observed using baited camera lander deployments at the Porcupine Seabight and Porcupine Abyssal Plain between 1989 and 2004.
MyxiniformesMyxinidaeMyxine sp.
ChimaeriferiformesChimaeridaeHydrolagus affinis
LamniformesScyliorhinidaeGaleus melastomus
Galeus murinus
SqualiformesCentrophoridaeCentrophorus squamosus or Deania calcea
EtmopteridaeCentroscyllium fabricii
SomniosidaeCentroscymnus coelolepis
HexanchiformesHexanchidaeHexanchus griseus
ElopiformesHalosauridaeHalosauropsis macrochir
AnguilliformesSynaphobranchidaeHistiobranchus bathybius
Synaphobranchus kaupii
GadiformesMacrouridaeCoryphaenoides armatus
Trachyrincus sp.
MoridaeAntimora rostrata
Lepidion eques
Mora moro
PhycidaePhycis blennoides
OphidiiformesOphidiidaeSpectrunculus sp.
ScorpaeniformesLiparidaeParaliparus sp.

Plotting the distribution of maximum depths of occurrence by 500-m increments in Fig. 2 shows an idiosyncratic trend compared with the global dataset. Only one species (appearing as a zero in the logarithmic plot), Argentina sphyrea, had a maximum depth at less than 500 m; all other species captured in the 10 trawls at less than 500 m were also found deeper than 500 m There is no significant change from 1000 to 3000 m with a mean of 16.6 species per depth bin, then a decrease at 3500 m (three species) and subsequent increase onto the abyssal plain, with 12 species at the maximum depth of sampling. The maximum depth at which representatives of the Agnatha were trawled at the Porcupine Seabight is 1540 m (Myxine ios) and 4118 m for the Chondrichthyes (Rajella bigelowi); for the Actinopterygii 11 species were recorded at maximum depth (> 4800 m) on the Abyssal Plain (Histiobranchus bathybius (Synaphobranchidae), Bathytroctes microlepis, Bathytroctes macrolepis, Conocara salmoneum, Rinoctes nasutus (Alepocephalidae), Bathypterois longipes (Ipnopidae), Coryphaenoides armatus, Coryphaenoides leptolepis, Coryphaenoides profundicolus, Echinomacrurus mollis (Macrouridae), Holcomycteronus squamosus (Ophidiidae).

The Porcupine area is highly variable in terms of the topography of the seabed (Fig. 5). The slopes on which trawling was carried out varied between 0.1 and 13.5% with a mean of 2.05% (SD 1.66%) down to 4000 m and significantly lower slopes (P > 0.0001) at greater depth on the abyssal plains (mean of 0.66%, SD 0.44%). Despite the apparently strong variations in seabed slope, the final GAM describing changes in fish species richness, rejected slope as a significant explanatory variable and only included depth as a smoothing term (edf = 8.069, F-value = 40.65, P < 0.0001). The results show that the explained deviance was 0.721, indicating that 72% of the observed variation in demersal fish species richness is explained by depth. The fitted curve (Fig. 6) indicates that demersal fish species richness is greatest at depths around 1600 m, with a predicted value of 16 species per trawl, and decreases strongly with increasing depth towards the abyssal plain.

Figure 5.

 Sea floor slope (local maximum gradient, %) at each trawl location (mean depth).

Figure 6.

 Trends in demersal fish species richness (numbers of species per trawl) as a function of depth. Solid line, smoother for species richness; dashed lines, 95% confidence limits. The curve fitting included an offset to compensate for the effect of variation in trawl swept area with depth.

Plotting the ranges between minimum and maximum depth of occurrence in the trawls shows that the maximum number of overlapping ranges occurs at 1600 m, with 48 species potentially present at that depth (Fig. 7). The overall pattern is very similar to the species richness per trawl plot (Fig. 6), indicating that on average trawl samples capture approximately one third of the species present at all depths.

Figure 7.

 Depth ranges of demersal fishes in the Porcupine Seabight and Porcupine Abyssal Plain area of the NE Atlantic Ocean. (A) Depth ranges between minimum and maximum trawl captures in this study. Black bars, non-scavenging species; red bars, scavenging species identified by appearance in baited camera images. Species are ranked in accordance with minimum depth of occurrence, reference and rank numbers are given in Table 2. (B) Total number of species within 200-m depth strata. Summing across the data in panel (A) assuming all species ranges are continuous. Black line and symbols, all species; red line and symbols, scavenging species only.

The baited camera attracted a total 22 species compared with 108 species captured by trawl, indicating that 20.37% of species are bait-attending types or putative scavengers. Comparing the predicted asymptotic species numbers of trawls (120) versus camera (24) this is exactly 20%. At any given depth the percentage of putative scavengers is remarkably constant, with a mean of 22.7% (± 3.5% SD) so that the scavenger curve in Fig. 7B follows the trawl species curve.


The fish fauna living in a given region is a subset of the global fauna. Thus, before effects of local heterogeneity on biodiversity can be understood it is useful to consider the global trends in distribution of three main classes of fishes which may be independent of local conditions. Globally, the Agnatha (Myxiniformes) Chondrichthyes and Osteichthyes all show a maximum in species number at the shallowest depths (Fig. 2). Variation in size is also greatest at shallower depths but this trend is less pronounced in the Agnatha. These global trends can be the result of at least four possible factors: (i) global patterns of environmental heterogeneity, (ii) decrease in food supply with depth, (iii) evolutionary history of colonisation of the deep sea by fishes, (iv) the deepest parts of the ocean remain insufficiently studied to reveal the high diversity residing there. Mora et al. (2008) show that the bathydemersal species inventory is the least well known of all fish faunas; however, the estimated number of species remaining to be discovered in this zone is insufficient to invalidate the trends observed in Fig. 2. We therefore assume that the existing species inventory accurately reflects global patterns.

High species number and size variation at shallow depths are probably strongly influenced by the heterogeneity of shelf, coastal and estuarine habitats with scope for endemism around islands, reefs, bays and along stretches of coastline. There is also evidence of high diversity around seamounts reflecting topographic and hydrographic heterogeneity (Fock et al. 2002). Generally, in deeper environments there is greater geographic interconnectedness between habitats and the abyssal plains harbour cosmopolitan demersal species that occur throughout all the world’s oceans (e.g. the macrourid Coryphaenoides armatus, King & Priede 2008). At the deepest extremities of the ocean, in the hadal zone at over 6000 m depth, the contiguous abyss subdivides into 22 separate trenches or basins (Angel 1982) isolated to varying degrees from one another, in which 47–56% of species, mostly invertebrates, are endemic to trench systems (Vinogradova 1997). The presence of endemic hadal fishes (Jamieson et al. 2009) with different species in different trenches probably explains the apparent increase in species number at maximum depth in Fig. 2, which deviates from the general regression line. This is similar to the general decrease in terrestrial species richness with increasing altitude on individual mountains but with endemism at summits, resulting in high global species richness at high altitudes (Väre et al. 2003). It is important to note that global and regional species richness trends can therefore be divergent. There is a clear decrease in species richness at extreme depths or altitudes within a locality, but globally, there is an increase in richness at the extremes resulting from summing of all the locally endemic species.

A major driver of both biomass and biodiversity trends in the deep sea may be the attenuation of the food supply from the surface. Most measurements of biomass in relation to depth, either pelagic (Angel & Baker 1982) or benthic (Lampitt et al. 1986), show a logarithmic decrease; approximately 10-fold per 2000 m. It is therefore not surprising that species richness declines with depth as less and less biomass can be supported, particularly as mean body size does not decrease with depth. Rex (1973) shows that biodiversity of gastropods is low in the abyss and attributes this to extremely low productivity in this region but points out that biodiversity can be sustained in such environments by the adoption of small body size. Indeed Rosenzweig (1995) shows that the relationship of diversity versus productivity along a gradient is often unimodal with maximum species richness at intermediate productivity levels.

Within fish species, bigger-deeper trends are often observed (Merrett et al. 1991b), suggesting that dwarfism in the deep sea is not applicable to fishes. The Chondrichthyes are remarkable for their relatively rapid decrease in species richness with depth, resulting in the absence of sharks in the abyss. Priede et al. (2006) argued that one factor may the relatively high energy requirement for Chondrichthyes in maintaining neutral buoyancy by sequestering large volumes of oil compared with the low cost of air bladder buoyancy in the Actinopterygii. However, here we show that the Agnatha, which do not have swim-bladders, have a similar rate of decrease with depth as the Actinopterygii. In addition to the depth trends, regional differences in surface productivity can result in differences in food supply to the deep sea floor and regionally distinct ichthyofaunal assemblages (Merrett 1987, 1992).

The possibility of a relationship between body size and biodiversity has been widely discussed (McClain & Boyer 2009) so it is interesting to note that in all three classes of fishes, there is no trend in body size with depth; mean body length is constant. However, McClain & Boyer (2009) show that generally for metazoa, species richness is highly correlated with body size variation and they argue that greater body size range may allow for greater niche differentiation. This agrees with the observation here that the highest species richness occurs at the shallowest depths where the greatest size variation is present.

As well as the present-day environment, the patterns observed in Figs 2 and 3 are probably also influenced by the history of colonisation of the deep sea that has occurred during the last 70 million years and since establishment of the modern thermohaline circulation that transports oxygen-rich water to the deep sea. Andriashev (1953) proposed two phases of colonisation, an ancient group including the macrourids, and a more recent secondary deep-water group, such as perciformes largely derived from shore-living taxa. However, Howes (1991) argues that in the gadoids (Superorder Paracanthopterygii, Nelson 2006), including ophidioids, lophiformes and macrourids there has been repeated divergence from ancestral forms into abyssal- and shelf-dwelling forms as the modern ocean basin evolved.

According to the species accumulation curves, 90% of the ichthyofauna of the study area has been detectable by trawls and baited cameras (Fig. 4). Mora et al. (2008) show that globally, bathydemersal species inventories are 56% complete and demersal species 81% complete so the Porcupine Seabight area is relatively well sampled. The small trawl that was used is not very effective at catching larger mobile species so that the shark Centroscyllium fabricii which was detected by the baited camera was only caught by a much larger bottom trawl towed at 4 knots (Merrett et al. 1991a). Full characterisation of a fauna is best carried out using a variety of fishing gear, with different species selectivity characteristics. Merrett et al. (1991a) compared results for three different kinds of trawl. We chose the OTSB to obtain unbiased estimates of biodiversity at different depths because it is the only gear capable of being deployed at all depths. In a long-term study such as this it is possible that the ichthyofauna may change over time and, indeed, Bailey et al. (2009) have detected changes in relative abundance of different species during the time course of this sampling programme. However, our statistical analysis showed that there was no change in species composition between sampling periods, so the present analysis is not affected by this observation.

The baited camera selects those species that use olfactory foraging and are attracted to the odour of baits. Coryphaenoides rupestris, for example, has large eyes and a brain with a well-developed optic tectum but only a very small olfactory region (Priede et al. 1999) and is therefore never captured in baited camera images or baited fishing gear such as long lines or traps. However, Coryphaenoides armatus, with large olfactory and reduced optic regions of the brain, is a well-known ubiquitous scavenger at baits (King & Priede 2008). Species appearing at baits can be regarded as scavengers, but they may not be obligate scavengers as some species are often observed to feed on other animals such as amphipods attracted to bait rather than on the bait itself. Collins et al. (2005) showed that the scavenging species (identified as those that attend baited cameras) in the Porcupine Seabight have a significant bigger-deeper trend in body size. They explained this by showing the advantages of large body size for survival during the intervals between rare feeding opportunities at greater depths. This resulted in a slight, but not significant, increase in biomass (kg·km−2) of the scavenging species with depth, despite a decrease in abundance with depth. This contrasted with non-scavengers, which showed no significant trend in body size with depth. The observation that scavengers constitute an almost constant proportion (c. 20%) of the total number of species throughout the depth range (Fig. 7B) is very interesting and suggests that this may represent an optimal size for this functional component of the fish assemblage. There is the possibility, however, that cryptic rare scavenging species were not recognised in the photographs, e.g. the three Ilyophis species of synaphobranchid eels were not detected at baits, but small numbers (< 30 total in all samples) were logged in trawls. We cannot exclude the possibility that species richness at the landers was underestimated by inability to discriminate these rarities in images.

Examining the Porcupine Seabight maximum depths data in Fig. 2, it is evident that the regional data do not follow the global pattern. The Porcupine Seabight data show a more or less constant number of species per depth stratum until 3000 m, where there is a discontinuity between continental slope fauna above and abyssal fauna below. In contrast to the more localised species on the slopes, the abyssal plain species are relatively cosmopolitan, resulting in convergence towards the global fitted line at maximum depth. Koslow (1993) points out that shallower-living (upper and mid-slope) species tend to be restricted to only one side of the Atlantic, whereas many deeper-living species occur on both sides of the ocean. Contrary to the general trend of decrease in species number with depth in the global data set, in the Porcupine area (grey line in Fig. 2) there is an increase in numbers towards 5000 m. Six of the 11 deepest fishes recorded there in the 4500–5000 m depth bin have global maximum depths of occurrence exceeding 5000 m (Froese & Pauly, 2008) but in the Porcupine Abyssal plain they are restricted to the maximum sea floor depth of less than 5000 m resulting in a regional cluster of species at that depth. To a limited extent, global patterns may drive local species richness trends, e.g. Agnatha and Chondrichthyes are absent from the deeper stations because globally, members of these class are unable to survive at abyssal depths. However, we conclude that there is little evidence that the patterns of local species richness with depth in demersal species is a reflection of the global trend, i.e. local niches automatically filled by recruitment from a global pool of characteristic species at each depth stratum. Rather the converse is true; the global trend is the sum of numerous idiosyncratic local faunas that cumulatively generate the distributions seen in Figs 2 and 3.

Local phenomena therefore must determine patterns of species richness. In this paper we use slope as a proxy for topographical heterogeneity; however, the slopes recorded at the trawl stations (Fig. 5) are not fully representative of the range of topography occurring in the study area, as trawling must avoid the most precipitous slopes and rough terrain. Within the range sampled, it is clear that slope does not influence species richness in the trawls.

The most striking result in Fig. 6 is the elevated species richness between 800 and 2500 m with a significant peak at 1500–1600 m. This is reflected in the cumulative presentation in Fig. 7B. Such a pattern could be the result of the mid-domain effect (MDE) where a peak in species richness is observed resulting from random assembly of species within a defined bathymetric zone (Colwell & Lees 2000). Kendall & Haedrich (2006) tested this hypothesis for different regions of the North Atlantic Ocean and found that observed patterns did not match the random assembly null model. The diversity maximum is probably a result of a combination of physical factors and patterns of distribution of pelagic and benthic prey. There are a number of physical factors acting at this depth. For example, 1400 m is the base of the permanent thermocline (Rice et al. 1991) where the temperature reaches 4 °C and below which the temperature remains relatively stable. Flach et al. (1998) found peak current velocities at this depth, reaching 35 cm·s−1 in autumn to winter and resulting in resuspension of particulate matter and creating conditions in which filter feeders occurred in high abundance and biomass. Depths between 800 and 1600 m in the Seabight are dominated by Mediterranean Overflow Water (MOW) with Labrador Sea Water at 1600–1800 m and North Atlantic Deep Water (NADW) below (Howell et al. 2002).

We hypothesise that, in this depth zone, species find distinct niches in different water masses, temperatures and current regimes, resulting in greater species richness than would otherwise occur in a section of continental margin with uniform oceanographic conditions. Furthermore, the presence of sessile filter feeders indicates the possibility for survival of filter-feeding fishes in these localities. The sessile filter feeders themselves may form reefs, creating habitats for specialised fishes (Ross & Quattrini 2007) and further enhancing biodiversity.

The base of the photic zone and interaction with mesopelagic fauna impinging on the slope adds a further dimension of environmental heterogeneity at mid-slope depths. The macrourid Coryphaenoides rupestris is a commercially exploited fish that is most abundant between 800 and 1800 m depth with a peak at 1500 m, so it might be regarded as an archetypal mid-slope species occurring at the zone of peak diversity. It is a visual feeder probably predating in the twilight zone of the mesopelagic as well as exploiting benthic fauna. Mauchline & Gordon (1991) show that benthopelagic fish including C. rupestris feed on epipelagic and mesopelagic fauna that impinge on the continental slope during the day-time at the maximum depth of their diel vertical migratory pattern. They point out that in the Rockall Trough, a region north of the Porcupine Seabight area, maximum demersal fish abundance coincides with the depth of greatest impingement of pelagic fauna on the slope at 1200–1300 m depth. In the Porcupine Seabight area, Gillibrand et al. (2007) discovered a seasonal maximum of pelagic bioluminescence at 1200–1800 m within the MOW layer, indicative of high pelagic biomass at these depths. In addition to pelagic prey, demersal fishes also forage on benthic fauna. Howell et al. (2002) linked the distribution of asteroids to physical oceanography of the Porcupine Seabight and found a peak of diversity at 1800 m, but, in contrast to the demersal fishes, there was an increase in biodiversity again below 4000 m. For bivalves, Olabarria (2005) found a peak of diversity at 1600 m followed by a decrease to 2700 m and maximum diversity at 4100 m.

Peak diversity of fishes at c. 1600 m coincides with a peak in biomass and diversity of several taxa of benthic invertebrates and is probably reinforced by enhanced access to pelagic prey at this depth. However, the analysis of fish biodiversity in relation to heterogeneity is weakened by the coarse scale of trawl sampling, with a mean haul area of 60,000 m−2 and 2–10 km of linear extent. Using such means it is not possible to address the small-scale dynamics discussed by Levin et al. (2001). Ross & Quattrini (2007) surveyed fish fauna associated with coral banks of the NW Atlantic down to 783 m using the Johnson-Sea-link manned submersible and argued that using conventional otter trawls it is possible to define neither fish communities nor habitat relationships. A full understanding of habitat heterogeneity effects on demersal fishes of the NE Atlantic must await the application of new methods capable of high-resolution mapping of habitat utilisation. Uiblein et al. (2002) and Lorance et al. (2002) have observed Synaphobranchus kaupii and Hoplostethus atlanticus on the slopes of the Bay of Biscay by manned submersible and were able to show how their behaviour and abundance were associated with particular habitat features. Unfortunately, surveying sufficient area while retaining fine-scale spatial resolution is challenging and submersibles may also affect the behaviour of the focal animals. As a result, much remains uncertain about both the diversity patterns of deep-water fishes and the factors which cause them (Trenkel et al. 2004a,b).


This work was supported by a series of NERC grants to the principal investigators including NE/C512961/1. The results of the early joint SAMS and IOS surveys were digitised with support from EU MAST Contract MAS2-CT920033 1993–95 and this analysis is supported by EU FP7 Projects CoralFish, HERMES and HERMIONE. We thank Alain F. Zuur for advice on statistics and Camila Henriques for permission to use unpublished data from her PhD thesis. We thank the ships’ companies of the RRS Challenger and RRS Discovery.