Mounting evidence: near-slope seamounts are faunally indistinct from an adjacent bank


Kerry L. Howell, Marine Biology and Ecology Research Centre, Marine Institute, University of Plymouth, Drake Circus, Plymouth. PL4 8AA, UK.


Seamounts have been described as island habitats harbouring a unique fauna, and as biodiversity hotspots with high rates of endemism. However, recent research suggests that these generalisations are inappropriate and poorly supported, though appropriate on and off seamount comparative data are lacking. This study uses quantitative data derived from video analysis to compare epibenthic megafaunal community composition, diversity and potential species endemism on two seamounts and one bank in the Rockall Trough region of the NE Atlantic. Sample data were standardised for substratum type across all three features and as far as possible for depth and geomorphological variation. The results suggest that under similar environmental conditions, e.g. similar substratum, depth and geomorphology, there is little difference between the communities of the bank and seamounts in the Rockall Trough. Where differences are observed, the ‘guyot’ seamount is as different to the conical seamount as it is to the bank. The seamounts are no more or less diverse than the bank; endemism is low or non-existent. The results are discussed in the context of the common generalisations made concerning seamount communities and the implications for conservation and management of the deep sea.


Seamounts have been described as islands harbouring unique or characteristic fauna (Moore et al. 2003; Stocks 2004) and as biodiversity hotspots (Santillo & Johnston 2005) with high rates of endemism (Richer de Forge et al. 2000). However, this view of seamounts has recently been challenged by new research indicating that communities on seamounts are similar to those of comparable regions of the continental slope (McClain et al. 2009), with similar levels of species richness (O’Hara 2007), and no more, and in some cases less, diverse than other submarine features such as canyons (Schlacher et al. 2007), with low or negligible rates of endemism (Samadi et al. 2006; Hall-Spencer et al. 2007; O’Hara 2007). This view of seamounts is not entirely new. In their review, Wilson & Kaufmann (1987) concluded that the faunal composition of shallow seamounts <1000 m was dominated by regional species with an equal representation of cosmopolitan species, whereas deeper seamounts were dominated by the latter. Although these authors also suggested rates of endemism among invertebrates on seamounts as maximally 15.4%, they were careful to highlight the potential error in this figure as a result of under-sampling of the deep-sea environment, as was the most recent review of this topic (Stocks & Hart 2007).

Understanding the significance of submarine features like seamounts to regional and global biodiversity is important, particularly with regard to the way in which we use, manage, and conserve our marine environment. In the North East Atlantic, seamounts are defined under the Oslo-Paris Convention (OSPAR) as undersea mountains of volcanic origin, with a crest that rises more than 1000 m above the surrounding seafloor. Through the OSPAR Convention, seamounts (conforming to this strict definition) are afforded some measure of protection through being listed as a threatened and declining habitat for which OSPAR marine-protected areas may be established. However, other submarine features, such as banks, ridges and canyons are not listed, presumably because they are perceived as ‘less important’. The question is: is this view justified? Evidence so far from the North East Atlantic would suggest not, with highly biodiverse and fragile cold water coral reef systems identified from Porcupine Bank (Wheeler et al. 2005), Rockall Bank (Wienberg et al. 2008), and the continental slope (Foubert et al. 2005). However, appropriate comparisons between the benthic communities on and off seamounts (sensu stricto) within this region have not been made. Globally, there are few published data comparing directly the fauna of seamounts to that of other raised submarine features.

In his recent review of seamount research, McClain (2007) made 10 recommendations for further research needed to attempt to clarify the ecological nature of seamounts. One of these recommendations stipulated comparative studies on and off seamounts with quantitative sampling controlled for different substratum types and depth. The present study attempts to achieve just this objective. The European Continental Margin to the west of the British Isles is one of the best known regions of the deep sea in the world, and has been described as ‘the cradle of deep-sea biology’ (Gage 2001). Within this region lie three features identified as seamounts under the OSPAR definition, as well as numerous banks and hills. This area is therefore the ideal setting to investigate questions concerning the importance and relatedness of benthic communities of submarine features such as seamounts and banks.

The aim of this study was to investigate whether the epibenthic megafaunal communities on seamounts are significantly different to those off seamounts under similar environmental conditions. Our hypotheses were: (i) the benthic communities of seamounts are different from those of other topographic features (i.e. are unique), (ii) seamounts support higher levels of diversity than equivalent areas (i.e. are biodiversity hotspots), (iii) seamounts support higher numbers of unique species than equivalent areas (i.e. have high endemism).

Material and Methods

Study area

The benthic megafaunal communities of two near-slope seamount features (>1000 m tall, volcanic in origin) and one bank feature (sea-floor elevation not of volcanic origin) were investigated using video sampling techniques. All three features lie within or border the Rockall Trough in the North East Atlantic (Fig. 1).

Figure 1.

 (a) The study areas (rectangles identify the areas shown in b and c), (b) Rosemary Bank Seamount (RBS) and Anton Dohrn Seamount (ADS), (c) Hatton Bank (HB). • Video transects. Depth contours in 100- m isobaths down to 1000 m, thereafter in 500- m isobaths.

Anton Dohrn Seamount (ADS) (11°6’ W, 57°28’ N) is volcanic in origin and is estimated to have been formed in the Maastrichtian period, being approximately 70.6–65.5 million years or more old (O’Connor et al. 2000). It is a flat-topped seamount, or ‘guyot’, sub-circular in shape, and surrounded by a shallow moat, best developed to the NW (Jones et al. 1974, 1994; Jacobs 2006). The summit has a diameter of approximately 40 km and the highest point confirmed to date occurs at about 520 m below sea level. The flanks are steep, rising from about 2200 m to 850 m in the NW and 1200 m to the SE. Its total surface area is approximately 2795 km2. The summit is covered by thick sediment above the igneous core, with an area of exposed basalt in the centre of the summit. A few small parasitic cones have been observed on the flanks of the seamount on the NW side.

Hatton Bank (HB) (17°20’ W, 59° N) is part of the Rockall-Hatton Plateau, which itself is a large piece of continental crust that separated from the main European continental shelf around 100 million years ago. It is not of volcanic origin and thus is not recognised as a seamount under the OSPAR Convention, and is recognised as a bank under the IHO definition (International Hydrographic Organisation 2008). It forms a narrow arc stretching over 400 km and rises from the Rockall–Hatton Basin at 1100 m to <500 m below sea level at the summit. Its total surface area is approximately 19,833 km2. Seismic profiling undertaken by the British Geological Survey has shown that mound structures, which are acoustically similar to the biogenic mounds found on Rockall Bank and the Porcupine Seabight, are fairly common features on Hatton Bank (Long et al. 2006). In 2005, the presence of cold water coral reefs on these mound features was confirmed (Narayanaswamy et al. 2006).

Rosemary Bank Seamount (RBS) (10°9’ W, 59°19 N) is volcanic in origin and has been dated at approximately 71–69 million years old (Morton et al. 1995). It is broadly domed and elongate in shape with a summit diameter of 70 km. It rises from ∼1900 m (2300 m at the base of the moat) to a peak at 316 m below sea level (Pudsey et al. 2004). Its total surface area is approximately 4097 km2. In places the seamount flanks are steep (∼20°). On the summit of the bank, dozens of small parasitic cones occur, some of which are up to 150 m high and with a base of ∼300 m diameter. A narrow moat, up to 400 m deep, encircles Rosemary Bank Seamount (Pudsey et al. 2004).

Data collection

Collection of biological (video) data from eight stations on ADS and two stations on HB were undertaken over a 1-month period (August–September) in 2005 using the commercial research vessel Kommandor Jack. Collection of biological data from eight stations on RBS and two further stations on HB was undertaken over a 2-month period (August–October) in 2006 using the commercial research vessel M/V Franklin. Video sampling stations were selected using high resolution sidescan sonar (ADS) and high (RBS and HB 2006 stations) and low resolution (ADS and HB 2005 stations) multibeam bathymetry and backscatter data. Video tows were selected to cover a range of geomorphology, substratum types and depths at each feature (Table 1).

Table 1.   Feature sample data. Values in parentheses are: mean depth (standard deviation), depth range sampled (total depth).
 Anton Dohrn SeamountHatton BankRosemary Bank Seamount
  1. *Morphospecies occurrence data compared with morphospecies recorded from video analysis by Howell from George Bligh Bank, Rockall Bank and the Wyville-Thomson Ridge. Unpublished data. Sample positions of comparable video stations given in Narayanaswamy et al. 2006 and Howell et al. 2007.

total no. of video tows888
total number subsamples used646464
subsample depth
 mean depth581 m (65)755 m (132)534 m (190)
 depth range sampled523–725 m (202)502–931 m (429)333–888 m (555)
subsamples by substratum
 gravelly sand323232
 sandy gravel101010
subsamples by geomorphology
 total no. geomorphological types366
 flat seabed31611
 iceberg ploughmarks036
 ledge halo03112
 outcrop halo221624
 basalt outcrop1139
morphospecies present
 total number morphospecies305236
 total no. of unique morphospecies (morphospecies only observed at one feature)92510
 total number of morphospecies not currently known from other features in the region*030
 % Potential endemism060

The Seatronics drop frame camera system was deployed from the starboard side of the vessel. The system comprised an integrated DTS 6000 digital video telemetry system, which provided a real time video link to the surface, and a digital stills camera (5 mega pixel, Kongsberg OE14-208). In 2005, the video stream from the viewing screen of the digital stills camera provided video data; in 2006, separate video (Kongsberg 14-366) and stills cameras were used. Cameras were mounted at an oblique angle (video: 24º; stills: 22º from the horizontal) to the seabed to aid in morphospecies identification. Sensors monitored depth and altitude, and an Ultra Short Base Line (USBL) beacon provided accurate (to approximately 1 m) position data for the camera frame.

Video tows were between 175 and 1400 m long. For the majority of tows, vessel speed was approximately 0.5 knots (min 0.3 and max 0.7 knots), with most tows lasting between 0.5 and 1.5 h. The drop frame was towed in the water column between 1 and 3 m (dependent on substratum type and currents) above the seabed. At the beginning of each tow, starting from when the sea floor became visible, a 2–3-min period was allowed before sampling, to enable the camera to stabilise before commencing the transect. The fields of view of both the stills and video cameras were calibrated using a gridded quadrat of known dimensions. Calibrations were made for ‘on bottom’ (drop frame fully landed on the seabed) and at 1, 2 and 3 m above the seabed to aid in quantitative analysis and particle size discrimination of the substratum.

Video analysis

Identification of species from video is difficult and in many cases impossible without physical samples; this is particularly problematic when working in the deep sea where our knowledge of the fauna is more limited. In this study, distinct ‘morphospecies’ were defined, catalogued and used in subsequent video analysis. Morphospecies may correspond to species, genus, family or higher taxonomic levels depending on the group. The morphospecies catalogue is available from the author upon request. Only those morphospecies that could be consistently identified from up to 3 m off the bottom were included in the analysis. This criterion inevitably resulted in size-selective sampling, with only the conspicuous megafauna recorded. However, all forms of sampling are size-selective, and in the case of trawls and video, species-selective also (with some species more effectively caught within a net or more conspicuous in a video).

Videos were reviewed and megafaunal morphospecies were identified and quantified. Substratum type was visually classified along the video tow using a modified version of the Folk (1954) classification (see Long 2006 for details). The occurrence of each individual animal as well as each change in substratum type was linked to the navigational data from the USBL on the camera system such that the location of each individual/change in substratum type (X and Y coordinates to 1dp), were recorded along with the depth (m). To ensure that the morphospecies data recorded at each feature were not biased by the substratum types present, the data were processed in the following manner. For each tow the total length and length (m) of each continuous strip of single substratum class was measured in ARCGIS 9.2. The frequency distribution of substratum class lengths was inspected and a strip length of 20 m was chosen as standard subsample length to minimise data loss and avoid pooling of strips. Tows were then split into 20 m subsamples such that each 20 m subsample corresponded to a single substratum class within a tow. Subsamples of <20 m were discarded. The total number of subsamples of each substratum class for each feature was standardised to the smallest number available at a single feature such that each feature was represented by an equal number of subsamples of each substratum class. This was done to avoid sample bias, as the abundance of morphospecies is likely to be influenced by the substratum. Excess subsamples were removed across all tows from the tow end backward to retain subsamples that crossed the geomorphological target (see below); as a target is approached, crossed and moved away from during the course of a tow, removing subsamples from the end of each tow would ensure that similar samples were retained from the beginning of the tow. Thus, each feature had equal numbers of subsamples from each substratum, drawn from a set of eight video transects, contributing to the dataset (Table 1). Each subsample was assigned to one of six geomorphological types (Table 1) by reference to available acoustic data and using standard geomorphological terms, and the average depth within the subsample was calculated. Subsample metadata are provided in an electronic Appendix.

Statistical analysis

Multivariate analyses

Faunal data (excluding fish) were first analysed using PRIMER v.6 (Primer-E Ltd., Plymouth; Clarke & Warwick 2001) with the PERMANOVA add-on (Anderson et al. 2008). Subsample morphospecies abundance data were square-root transformed to reduce the impact of highly abundant species and a Bray–Curtis similarity matrix calculated. Because of the well known relationship between depth and assemblage composition (Carney et al. 1983; Howell 2010), assemblage data were analysed using a covariate PERMANOVA with sequential (Type I) sums of squares. Depth was employed as the covariate, and features and substratum types were fixed factors. Because of the haphazard distribution of geomorphologies amongst sites and substrata, geomorphology was a random factor nested within the feature × substratum type interaction term. A full factorial model with 10,000 permutations was employed using permutation of residuals under a reduced model. The effect of changing the order of addition of the fixed factors in the hierarchical model was examined by repeating the analysis, as suggested by Anderson et al. (2008), but no effect of altering the order of main effects was detected. A priori contrasts between the seamounts and Hatton Bank were also performed to test for the effect of on and off seamount differences in assemblage composition.

To validate the patterns observed using data as comparable as possible, a series of subsets of the data was analysed using one-way sequential SS PERMANOVAs accounting for residual depth effects before comparing the features. These subsets compared the features using data selected on the basis of comparable depth range, substratum type and geomorphology. The following analyses were possible: ADS versus HB – bedrock on basalt outcrops (10 subsamples), sandy gravel on outcrop halos (11 subsamples), gravelly sand on outcrop halos (21 subsamples); RBS versus HB – gravelly sand of ledge halos (14 subsamples); ADS versus RBS – sand on flat sea bed (17 subsamples).

Univariate diversity analyses

The effects of features, substratum type and geomorphology upon morphospecies diversity of the megafauna were analysed using ANOVA and Shannon–Wiener diversity indices. As a result of the haphazard distributions of geomorphologies amongst features and substrata, a mixed model ANOVA using the same structure as the PERMANOVA analysis described above was employed. Prior to conducting the main analysis the subset of subsamples from Hatton Bank was examined to confirm that year of sampling had no effect upon the diversity encountered (F1,54 = 0.94; P = 0.364), therefore all data were used in the subsequent analyses.

A full factorial sequential (Type I SS) model was conducted first to test for the effect of depth as a covariate and homogeneity of covariate relationships between combinations of factors with depth. No significant effect of depth upon diversity was detected (P > 0.2) and there was no heterogeneity of covariate slopes across factor combinations. The covariate term was therefore removed and a Type III SS mixed model was performed using feature and substratum type as fixed main effects and geomorphology as a random factor nested within both feature and substratum. Conformity to assumptions of ANOVA was determined by inspection of residuals in all instances; all residuals inspected were statistically normal.

Numbers of morphospecies

Finally, the number of morphospecies expected in a large number of subsamples from each feature was calculated using the nonparametric Chao1 estimator (see Foggo et al. 2003) using the software ESTIMATES (Colwell 2005) to perform 100 randomised resamplings; these estimates were compared with Mao Tau per-sample rarefaction estimates (Colwell et al. 2004) to determine the completeness of the sets of samples from each feature. The mean numbers of singleton morphospecies (those occurring in only a single sample) occurring in the randomised morphospecies accumulations were also plotted to provide an indication of the rate of encounter of rare morphospecies.


The distribution of sample effort (in terms of substratum type, geomorphology and depth range), total number of morphospecies identified, and total number of morphospecies unique to a feature are presented in Table 1.

Multivariate analyses

The results of PERMANOVA (see Table 2) and planned contrasts indicated that there was no statistical difference between the benthic assemblage composition on and off a seamount (Seamounts versus Bank Pseudo-F1.2,16 = 2.459, P = 0.064). Depth had a highly significant influence upon assemblage composition; however, after correcting for this effect, no remaining differences between features were found. After correcting for depth influences, substratum type did not explain a significant proportion of the remaining variation (much of the substratum-related variability being accounted for in the higher order interactions involving this term). However, geomorphology did account for a significant variance component of the model (P < 0.001), despite entering the model after fixed main effects and their interactions.

Table 2.   Results of Type I SS PERMANOVA of assemblage data from the subsamples from the three features.
depth × feature245590.5770.946
depth × substratum455940.6760.956
feature × substratum878250.9940.464
geomorphology (features, substratum)1677803.0410.001
depth × feature × substratum740681.5900.011
depth × geomorphology (feature, substratum)946631.8230.001

Analysis of the assemblages in the fully comparable geomorphological subsets of data showed a lack of consistent difference between the three features (ADS versus HB – bedrock on basalt outcrops P = 0.027, sandy gravel on outcrop halos P = 0.068, gravelly sand on outcrop halos P = 0.914; RBS versus HB – gravelly sand of ledge halos P < 0.001; ADS versus RBS – sand on flat sea bed P < 0.001), with differences between the two seamounts being as pronounced as those between the bank and the seamounts.

Univariate diversity analyses

Mixed model ANOVA indicated a highly significant effect of geomorphology nested within substratum type and feature, and a weak significant effect of substratum type consistent across features (see Table 3). However, the feature sampled did not significantly affect diversity in the subsamples, irrespective of the differing ranges of depths sampled.

Table 3.   Results of mixed model ANOVA of diversity data from the subsamples from the three features.
geomorphology (feature, substratum)160.9263.23<0.001
feature × substratum80.8960.950.505

Numbers of morphospecies

The two seamounts sampled supported fewer morphospecies compared with Hatton Bank (ADS = 30, RBS = 36, HB = 52), and Chao1 estimates indicate that the likely values for total morphospecies richness follow the same trend (Fig. 2a). The lack of convergence evident in the plots of observed morphospecies against Chao1 estimates indicate that none of the features was completely sampled with 64 subsamples. However, Fig. 2b indicates that the number of singleton morphospecies encountered was declining at Hatton Bank, stable at Anton Dohrn and still increasing at Rosemary Bank, meaning that the latter may represent a more under-sampled habitat and Hatton Bank the most completely sampled. Nonetheless, mean Chao1 estimates (±1 SD) of total expected morphospecies are still higher at Hatton bank (63.38 ± 8.09) than at Rosemary Bank (47.14 ± 8.23) or Anton Dohrn (37.50 ± 6.35) Seamounts.

Figure 2.

 (a) Mau Tau observed morphospecies accumulation curves (solid lines) and Chao1 estimates (dashed line) of total morphospecies richnesses at Anton Dohrn Seamount (black lines), Hatton Bank (blue lines) and Rosemary Bank Seamount (orange lines). (b) Mean numbers of singleton morphospecies occurring in the randomised morphospecies accumulations at the three features.


This study compares the epibenthic megafaunal communities of a ‘guyot’ type seamount, a conical seamount and a bank, using data that have been standardised (as far as possible) for depth, substratum and geomorphology. Although the conclusions that can be drawn from the analysis of only one example of each feature are necessarily limited, the paucity of published works making appropriate comparisons between such feature types, and the difficulty of obtaining strictly comparable datasets, makes the results valuable.

Are the benthic communities of seamounts unique?

The epibenthic megafaunal communities of the seamounts studied are not unique from those of a neighbouring bank, and the idea of a distinct ‘seamount’ fauna is called into question. PERMANOVA Pseudo-F values indicate that discrimination between features was relatively low, and on the whole similarity amongst the complete assemblages sampled was relatively high. However, the two seamounts were faunally most similar, with ADS and HB exhibiting the highest degree of faunal difference. These differences between features appear to be attributable to differences in geomorphology and the depth sampled, rather than presenting a strong difference between the assemblages on seamounts and the bank. These data suggest that where appropriate comparisons are made, e.g. similar depth, similar substratum and similar geomorphology, there is little difference between the communities of Hatton Bank and two seamounts in the same region.

These data support the finding of O’Hara (2007) based on ophiuroid assemblages from the South-west Pacific Ocean. He found that seamounts support a similar suite of species to that found on neighbouring seamounts or the continental margin when accounting for depth. Similarly, Lundsten et al. (2009) found the benthic communities of three seamounts off Central and Southern California to show a strong degree of overlap with those of the continental shelf and neighbouring seamounts. McClain et al. (2009) also found the megafauna of Davidson Seamount to consist of an assemblage of species that also occurred on adjacent continental margins.

Where appropriate comparisons could be made, for the small subsets of data that allowed standardisation between subsamples on features for depth, substratum and geomorphology, results from one-way sequential SS PERMANOVAs were inconsistent. For some combinations of factors, no significant differences were observed, whereas for others they were. This is most likely a result of the small number of subsamples that were strictly comparable. Where differences were observed, these data suggest that there is no greater distinction between communities of the seamounts and those of the bank studied than between seamounts. Although both AD and RBS are ‘seamounts’, AD is a ‘guyot’ type seamount, whereas RBS is more conical in shape. The suggestion then is that a guyot is as different to a conical seamount as it is to a bank; however, caution is needed in this interpretation because of the small subsets of data used and limited number of comparisons made.

Do seamounts support higher levels of diversity?

The seamounts under study supported lower levels of diversity than the bank under study, with ADS being the least diverse. The results of both PERMANOVA of assemblage composition and ANOVA of diversity data suggest that the observed differences in assemblage and diversity between features are also a result of differential distribution of substratum type and geomorphology rather than representing differences in the megafaunal species of the different features per se. The apparently lower levels of diversity on the seamounts may also be a result of the difference in total depth range sampled between the three features. ADS was the least diverse, but had the smallest depth range sampled (202 m; Table 1). Although RBS had the largest total depth range sampled (555 m; Table 1), the sampling over this range was uneven, with no samples taken between 519 and 851 m (Appendix 1), leading to an effective depth range of 223 m. Hatton Bank was the most diverse site but, given the uneven sampling on RBS, had the largest depth range sampled (429 m; Table 1).

These data suggest that where appropriate comparisons are made, e.g. similar depth, similar substratum and similar geomorphology, there is no distinction in the diversity of the seamounts or banks under study. This concurs with the findings of the most recent research on seamount fauna, which suggests that seamounts are no more species-rich than equivalent areas on the continental slope (O’Hara 2007), neighbouring oceanic islands (Hall-Spencer et al. 2007), or other submarine features (Schlacher et al. 2007). The findings contradict the notion that seamount features are biodiversity hotspots harbouring increased morphospecies richness (Samadi et al. 2006).

The observed lower levels of diversity on the seamounts may also be a result of species–area effects (Kohn & Walsh 1994). The two seamounts are significantly smaller in area than HB. MacArthur & Wilson (1967) propose that more species survive on larger ‘islands’ for two reasons: (i) populations on large islands are large enough to make extinctions less likely, and (ii) larger islands contain the specialist species of a greater number of habitat types. Kohn & Walsh (1994) demonstrated that island area contributes to species number both directly and indirectly, through habitat diversity, and that while the direct effects of area and habitats on species are roughly equal in magnitude, the total effect of area is almost twice that of habitats.

Lower levels of diversity on the seamounts would lend some support to the ‘Seamounts as Islands’ concept (sensuMcClain 2007), in which we might expect seamounts to exhibit lower species diversity in accordance with the patterns of diversity evident in oceanic or habitat islands (sensuMacArthur & Wilson 1967). However, as reviewed by McClain (2007), for seamounts to function as islands there would need to be some form of isolating mechanism, either by distance or through hydrographic barriers such as Taylor columns (Roden 1987). There is little evidence of the existence of Taylor columns at ADS or RBS (Due et al. 2006; Howe et al. 2006) and some evidence to suggest that the distances between features would not present a barrier to connectivity (Le Goff-Vitry et al. 2004); see below for further discussion.

Do seamounts support higher numbers of unique species?

The seamounts studied support lower numbers of unique morphospecies than the bank (Table 1). For both seamounts none of those morphospecies found as unique within the context of this study could be considered endemic to the seamount, having been reported from other features in the region (Howell, unpublished data), and only three morphospecies from Hatton Bank could be considered potential endemics, although these examples are likely to be a case of under-sampling rather than true endemism, as witnessed by the declining curve describing the number of singleton morphospecies at higher numbers of subsamples in the randomised morphospecies accumulations (Fig. 2b). However, it is important to remember that the estimates of potential endemism within this study are based on a very small number of total species for each feature (Table 1).

Seamounts have, since the review of Wilson & Kaufmann (1987), and later the influential paper of Richer de Forges et al. (2000), been regarded as features supporting high levels of endemism. These authors suggested levels of 15.5% (invertebrates) and 29–34% (macro- and megafauna), respectively. More recently, Rowden et al. (2002, 2003, 2004) have reported potential endemism from the seamounts around New Zealand of 15%, 5.5% and 17%, respectively. All of these authors were careful to state that these figures were potential endemics, highlighting the fact that incomplete knowledge of the regional faunal pool may have contributed to these apparently high levels. The Rockall Trough is one of the best known areas of the deep sea in the world. Our findings concur with more recent research (Samadi et al. 2006; Hall-Spencer et al. 2007; O’Hara 2007; Lundsten et al. 2009) that suggests high levels of endemism may not be the case for all seamounts. Potential endemism on Hatton Bank, although higher than that of the two study seamounts, is also low (6%), and may actually be non-existent.

The question of whether seamounts (and other isolated topographic highs, such as banks and hills) support high numbers of endemics is a difficult one given the conflicting evidence. In his recent review, McClain (2007) re-emphasised that given the severity of under-sampling of the enormous deep-sea ecosystem, nearly any new area of the deep sea investigated will reveal a substantial number of new species or potential endemics. He cited the work of Grassle & Maciolek (1992), who in their investigation of the more typical deep-sea mud floor, noted that 58% of the species present in their study were new to science (and thus potential endemics).

The lack of apparent endemism on the seamounts and banks of the Rockall Trough is not unexpected given the proximity of the European continental slope and the hydrographic structure of the region. It has been hypothesised that the apparently high rates of endemism observed on some seamounts may be a result of larval dispersal being limited by hydrological phenomena such as Taylor columns (Roden 1987), resulting in reproductive isolation and subsequent speciation. No such hydrological phenomena are known to occur at either Hatton Bank or Rosemary Bank Seamount (Due et al. 2006; Howe et al. 2006), although there is some evidence to suggest the presence of a Taylor column over Anton Dohrn Seamount (Booth 1988). In general, the water-mass structure and flow around the Rockall Trough would suggest there is connectivity, in terms of larval dispersal potential, between all of the topographic features of this region, including the continental slope. Studies of the population genetic structure of Lophelia pertusa from the European continental slope suggest gene flow is at least occurring along the continental margin (Le Goff-Vitry et al. 2004); however, no data are available on connectivity of seamount populations.

What are seamounts?

Our findings, together with those of O’Hara (2007) and Lundsten et al. (2009), highlight the fact that ‘seamounts’ cannot be considered a single habitat type, and that raised features of the sea bed in general need to be evaluated independently to assess their importance in the context of conservation efforts. These studies also serve to emphasise the overriding importance of depth as a driving factor of community structure in the deep sea, and to a degree, small-scale geomorphology (depth and geomorphology acting as surrogates for many other environmental factors). This has important implications for marine conservation efforts and in particular for habitat mapping for marine protected area (MPA) network design. Globally, ‘seamounts’ (under the strict geological definition used in this study) are being used as a broad-scale mapping unit (Davies et al. 2004; Harris 2007) on the basis that these features are biologically different to other submarine topographic highs, as they are biodiversity hotspots, centres of endemism and highly productive oases, and are, at least within a biogeographic region, biologically similar to one another. Our findings lead us to concur with O’Hara (2007) that ‘seamounts’ appear to be at the wrong scale to be considered a useful sea-floor habitat.

However, it is important not to cast aside seamounts as a potentially useful conservation consideration. Some seamounts (and other isolated oceanographic features) may yet prove to function as islands, although this may be the exception rather than the rule. Seamounts are thought to support higher biomasses of organisms, including commercial fish species (Clark 2001). This aspect of seamounts was not considered by us. Seamounts (as well as topographic highs) may act as stepping stones facilitating long-distance dispersal of species across oceans (Wilson & Kaufmann 1987), and play an important role in the connectivity of populations. In addition, it is possible that the relative meso-scale habitat heterogeneity that seamounts add to the broader landscape given their size and distribution may make them practical management units for conservation of a diverse range of habitat types. The seamounts in this region may not support distinct communities; however, they may have other important functional roles not provided by the neighbouring banks. For example, incidental evidence suggests that RBS may provide an important birthing and nursery ground for the commercially fished leafscale gulper shark Centrophorus squamosus (Defra 2007). Seamounts thus remain an important part of the deep-sea ecosystem, but must be considered so and used as a conservation or management ‘unit’ for the right reasons, not based on the assumption that they are in some way different or more important than other submarine features.


The authors would like to acknowledge with thanks the scientists, officers and crew of RV Kommandor Jack and MV Franklin, the staff at Geotek and Marin Mätteknik AB, and the wider project partners J. Davies, C. Marshall, H. Stewart, C. Jacobs, N. Golding, B. Narayanaswamy and D. Hughes. The manuscript was improved following helpful comments from three anonymous reviewers. The collection of data used was funded by the Department for Business, Enterprise and Regulatory Reform through Strategic Environmental Assessment 7 (formerly the Department for Trade and Industry) and the Department for Environment, Food and Rural Affairs through their advisors the Joint Nature Conservation Committee and the offshore Special Areas for Conservation programme. Data analysis was funded by a mini grant awarded to K.L.H. from the Census of Marine Life field project CenSeam (a global census of marine life on seamounts), a joint Research Councils of the UK fellowship awarded to K.L.H, and the University of Plymouth.