The contribution of deep-sea macrohabitat heterogeneity to global nematode diversity

Authors


Ann Vanreusel, Marine Biology Research Group, Ghent University Krijgslaan, Ghent, Belgium.
E-mail: ann.vanreusel@ugent.be

Abstract

The great variety of geological and hydrological conditions in the deep sea generates many different habitats. Some are only recently explored, although their true extent and geographical coverage are still not fully established. Both continental margins and mid-oceanic seafloors are much more complex ecologically, geologically, chemically and hydrodynamically than originally thought. As a result, fundamental patterns of species distribution first observed and explained in the context of relatively monotonous slopes and abyssal plains must now be re-evaluated in the light of this newly recognized habitat heterogeneity. Based on a global database of nematode genus composition, collected as part of the Census of Marine Life, we show that macrohabitat heterogeneity contributes significantly to total deep-sea nematode diversity on a global scale. Different deep-sea settings harbour specific nematode assemblages. Some of them, like coral rubble zones or nodule areas, are very diverse habitats. Factors such as increased substrate complexity in the case of nodules and corals seem to facilitate the co-existence of a large number of genera with different modes of life, ranging from sediment dwelling to epifaunal. Furthermore, strong biochemical gradients in the case of vents or seeps are responsible for the success of particular genera, which are not prominent in more typical soft sediments. Many nematode deep-sea genera are cosmopolitan, inhabiting a variety of deep-sea habitats and oceans, whereas only 21% of all deep-sea genera recorded are restricted to a single habitat. In addition to habitat heterogeneity, regional differences are important in structuring nematode assemblages. For instance, seeps from different regions yield different genera that thrive on the sulphidic sediments. This study also shows that many areas and habitats remain highly under-sampled, affecting our ability to understand fully the contribution of habitat heterogeneity versus regional differences to global nematode diversity.

Problem

The deep-sea floor has long been considered to be a relatively homogeneous environment on a large scale, comprising vast areas of soft, well-oxygenated surface sediments. Mainly depth-related factors, such as food input, hydrodynamics and occasionally sediment composition, were assumed to be the main drivers of differences in benthic standing stock, biodiversity and community composition of the benthos (Grassle 1989; Gage & Tyler 1991). However, as a result of increasing exploration by means of bathymetric and visual mapping of habitats (Wefer et al. 2003), there is now a growing awareness of the true extent of habitat heterogeneity and associated biodiversity along continental margins and abyssal plains. Knowledge of the biological communities associated with particular, locally restricted habitats in the deep sea has increased significantly during the last decade, as has the understanding of how other interdependent variables such as substrate availability and type, biogeochemistry, nutrient input, productivity, hydrologic conditions and catastrophic events shape patterns of diversity on regional scales (Levin et al. 2001).

The increasing interest in particular deep-sea environments, such as cold seeps, hydrothermal vents, cold water corals, canyons and nodule areas, and the wider accessibility of ROV technology, have facilitated the direct sampling of these different habitats, which was often not possible using traditional remote coring techniques. Such studies have shown that they are occupied by benthic communities that are different from those living in surrounding areas of typical deep-sea floor (Wefer et al. 2003). However, the extent to which these special habitats contribute to the overall deep-sea biodiversity has never been investigated, as biodiversity studies focused on particular habitats were often restricted to comparisons between their biodiversity and that of the surrounding background environments on a local or occasionally regional scale (e.g. several papers from this volume). No comparisons have been made yet on a larger scale comprising different deep-sea habitats. This mainly reflects the lack of comprehensive databases required to determine if the high turnover between macrohabitats on these smaller scales also holds when data are compiled over ocean-basin or even global scales.

In this study, a large database containing quantitative data on nematode genus composition from different areas and habitats around the world was assembled, allowing a global comparison of nematode biodiversity to be made. This database was made possible through the global initiative, ‘The Census of Marine Life’, which aims to make a realistic estimation of currently known marine biodiversity by 2010, and to provide a better insight into the factors responsible for changes in biodiversity. Nematode data from several distinct deep-sea habitats, including soft sediments from different water depths, manganese nodules, coral, seamounts, cold seeps, hydrothermal vents, canyons and trenches, were included in this comparative analysis. Nematodes are among the most abundant and diverse benthic metazoan taxa. They are present from shallow water environments to the deep sea, and from oxygenated to anoxic, sulphidic sediments (Heip et al. 1985). They show a preference for soft sediment but also colonize hard substrates in close contact with deep-sea sediments, such as nodules and coral rubble. As nematode data at the species level are scarce, and the majority of deep-sea nematodes remain undescribed, we investigated patterns at the genus level. It has been shown that nematode community composition at the genus level reflects macro-ecological patterns (Vanaverbeke et al. 1997; Vanreusel et al. 2000; Fonseca & Soltwedel 2007) and thus provides an appropriate basis for comparisons of communities between habitats on a world-wide scale.

Based on this database of nematode genus assemblages collected within the Census of Marine Life projects CoMARGE and CeDAMar as well as the MarBEF European network of Excellence, several hypotheses can be put forward. Inevitably the compilation of various datasets collected for multiple purposes by different researchers includes a high degree of heterogeneity partly generated by differences in temporal and spatial scales of sampling. Furthermore the sampling design is highly unbalanced, leading to under-representation of different habitats and regions. Therefore caution is needed in the interpretation of the results taking into account the fragmented nature of the observations. With these restrictions in mind, the following three main testable hypotheses were identified: (i) habitat heterogeneity contributes significantly to the total deep-sea nematode diversity when integrated over large scales; (ii) different deep-sea habitats harbour specific nematode assemblages; and (iii) higher biodiversity is associated with particular deep-sea habitats.

Material and Methods

Data on nematode density and genus composition were obtained from 542 samples collected from the shelf to the hadal zone. To preserve the original composition and biodiversity estimates, data from replicate samples were kept separate and not pooled. Figure 1 shows all geographical areas (some including multiple samples) from which data were collected for this study. As the focus was on the deep sea, data obtained from shelf stations (<200 m) were only included if these were part of a bathymetric transect that covered a significant part of the continental slope. Samples were always collected quantitatively (using different types of corers) and treated with standardized extraction procedures to guarantee the most comparable data (Heip et al. 1985). Literature datasets that did not provide complete taxonomic lists were not included, as the analyses required full genus counts and densities, including the rare taxa. Detailed sample information is available on request.

Figure 1.

 World map showing location of the sampling areas classified according to macrohabitat.

Data analysis was performed using the statistical package PRIMER v6.0. nMDS was combined with SIMPER and ANOSIM to identify differences in genus composition between habitats. A Bonferroni correction was applied in the case of multiple pairwise comparisons and a significance level of 5% was used. Diversity indices were also calculated using the PRIMER v6.0 software. Genus richness was calculated as the total number of genera (Hill’s N0; Hill 1973). By analogy with the expected number of species (Hurlbert 1971), we calculated the expected number of genera for theoretical samples of 51 [EG (51)] and 100 [EG (100)] individuals. In the case of seamount samples, the number of individuals was lower than 50 and no EG (51) was calculated. Samples were classified into 10 different macrohabitats (also referred to as habitats throughout the text) (Table 1) based on the following criteria: substrate composition (homogeneous soft sediment versus presence of manganese nodules or large biogenic substrate such as coral rubble and mussels), water depth, topography (canyon, trench and seamounts) and biochemistry (oxygen, methane and H2S). Some macrohabitats were assumed to be more common than others, as indicated in Table 1. Also, the degree of connectivity between similar habitats differs as a function of their general distribution. Figure 2 shows some examples of visual habitat heterogeneity in the deep sea. The definition of the shelf, slope, abyssal plain and trench macrohabitats used here is rather arbitrary, being based on water depth and not considering differences in local or regional topography. For instance, the abyssal basins of the Mediterranean Sea are much shallower (3000–4000 m) than elsewhere, and the shelf of the Weddell Sea margin extends out to a depth of 500 m. However, all the slopes identified in this paper have soft sediments from the depth zone between 200 and 4100 m, and are from topographically regular settings, covered by well-oxygenated bottom waters, and lack any indication of nearby flows of reduced chemical compounds. Some macrohabitats are characterized by considerable patchiness and comprise different micro- (or sub-) habitats. For instance, seeps include both completely anoxic, sulphidic sediments and sediments that are well oxygenated at the surface but show an increase in sulphide concentration below the surface. Similarly, the coral samples include coral rubble and dead sponges as well as coralligeneous sediments. Temperature is not taken into account as a habitat characteristic, as the deep Mediterranean has much higher bottom temperatures than other oceans.

Table 1.   Basic criteria used to identify the 10 main macrohabitats.
HabitatSubstrateTopographyBiochemistryDepthConnectivityDistribution
ShelfSedimentsFlat <200 mHighCommon
SlopeSedimentsRegular  200–4100 mHighCommon
AbyssalSedimentsFlat 4100−6000 mHighCommon
NodulesMn nodules on sedimentsFlat >4100 mLowRare
CoralsCoral and other biogenic rubbleMounds  LowMedium
CanyonsSedimentsChannel and terraces  LowCommon
SeepsSoft sedimentsPockmarks or mud volcanoesSulphidic and methanic LowRare
Hydrothermal ventsSediments, Mussel bedsRidge or riseSulphidic LowRare
SeamountsSedimentsMounts  MediumMedium
TrenchesSediments  >6000 mExtreme lowRare
Figure 2.

 Overview of deep-sea habitat diversity. (A) Soft sediment in the Nazaré Canyon; (B) cold-water corals; (C) Beggiatoa mats at the Håkon Mosby Mud Volcano (© Ifremer Vicking 2006); (D) pingo colonized by siboglinid tube worms at Nyegga (© Ifremer Vicking 2006); (E,F) manganese nodule areas (© Ifremer Nodinaut 2004).

The number of samples per macrohabitat was unbalanced and ranged from three on seamounts to 355 from regular soft sediments along the slope (Table 2). Furthermore, the coral samples (NE Atlantic), the seamounts (NE Atlantic), the nodules (NE Pacific) and the trench samples (Atacama trench, SE Pacific) were all collected from within single regions, in contrast to samples from the slope, shelf, abyssal plains, seeps, canyons and hydrothermal vents, which covered different geographical regions. The slope sediments were geographically the best represented of all the macrohabitats and were distributed in many parts of the World Ocean, although the majority of these samples were collected from the Atlantic, including the Mediterranean Sea.

Table 2.   Total number of genera, number of habitat-restricted genera recorded, and number of samples analysed for the each of the 10 macrohabitats.
HabitatTotal no. of generaNo. of habitat- restricted generaNo. of samples analysed
Shelf210243
Slope32548355
Abyssal1431125
Nodules90914
Corals112222
Canyons130215
Seeps120126
Vents31236
Seamounts3303
Trenches2703
Total362  

Results

Differences in nematode community composition between habitats

A total of 362 genera was recorded from the 542 samples (Table 2). The majority of these genera (about 90%) were previously recorded from soft-bottomed, regular slope habitats, indicating that the additional habitat heterogeneity is only responsible for 10% of the total genus pool recorded from deep-sea environments. The proportion of genera restricted to a single habitat within the total number of genera found in that habitat was highest in regular soft slope sediments (15%), followed by the nodule area (10%), the abyssal plains (8%), and the hydrothermal vents (6%). In the remaining habitats the proportion of genera restricted to the habitat was <2%. Many of the dominant genera from soft-slope sediments were also represented in the other habitats, although in different proportions (Fig. 3; Table 3). The highly abundant genera Acantholaimus, Halalaimus and Thalassomonhystera, but also Desmodora, Desmoscolex and Theristus, are the main ones showing wide distributions that include most of the investigated habitats.

Figure 3.

 Average relative abundances (%) of genera present in more than eight macrohabitats and dominant (>5%) in at least one of the habitats.

Table 3.   Average relative abundances (%) per habitat of the genera dominantly responsible for the similarities within habitats and the dissimilarity between each macrohabitat and the slope habitat based on a SIMPER analysis.
 SlopeShelfAbyssNodulesCoralsCanyonsSeepsVentsSeamountsTrenches
Acantholaimus6.670.6514.4917.445.6211.574.320.540.5311.57
Halalaimus7.203.225.723.436.3710.063.080.082.157.77
Desmodora0.552.270.501.282.030.187.955.6618.190.36
Desmoscolex2.593.513.303.548.142.362.000.248.873.30
Thalassomonhystera9.652.2123.8019.472.5210.385.8052.39 24.85
Theristus1.991.104.9110.673.073.270.980.050.97 
Microlaimus2.484.137.911.791.307.874.46  8.84
Daptonema5.514.672.691.230.513.132.29  6.24
Ceramonema0.130.160.100.323.550.02  9.49 
Sabatieria8.7111.32  1.912.4012.972.37  
Anticoma0.22  0.674.44 0.239.05  
Richtersia0.270.88  0.230.40  10.88 
Epsilonema0.040.00  4.700.890.55 0.85 
Halomonhystera  0.06   24.0813.67  
Marisalbinema  0.014.83  0.02   

Multivariate analysis, on the other hand, suggested that different deep-sea habitats harboured significantly different nematode communities (Fig. 4) (ANOSIM: R = 0.39; P < 0.01). According to the MDS ordination based on nematode genus composition (%), samples collected at seeps, hydrothermal vents, coral rubble, seamounts and nodule areas differed in genus composition from the majority of soft sediment samples collected on the shelf, slope and abyssal plains (Fig. 4). Within these three regular soft sediment habitats, shelf samples plotted mainly on one side of the central cluster of slope samples, whereas the abyssal plains were grouped on the opposite side. Canyon and trench samples overlapped to a large extent with the slope samples. Nodule samples were clustered adjacent to the abyssal samples. The coral samples, as well as the seep, hydrothermal and seamount samples, were generally more separated from the central slope cluster, although samples from these specific habitats occasionally overlapped with slope samples in the MDS ordination. The pairwise comparison with Bonferroni correction (P < 0.05) showed that seeps (R = 0.367), hydrothermal vents (R = 0.759), corals (R = 0.336) and seamounts (R = 0.913) differed significantly in genus composition from the slope samples. Shelf communities also differed significantly from the slope communities (R = 0.426), whereas the communities from abyssal plains (R = −0.044), canyons (R = 0.095), nodules (R = 0.136) and trench samples (R = 0.197) were not significantly different from slope samples. All habitats also differed significantly from the abyssal plains (R > 0.377) except for the trench (R = 0.316) and slope samples (R = −0.044).

Figure 4.

 MDS ordination based on nematode genus percentage abundance using the Bray–Curtis similarity index, with symbols indicating the designated macrohabitats.

The average relative abundances of the dominant genera responsible for the similarity within each macrohabitat, as identified by a SIMPER analysis, are shown in Fig. 5. This list of genera (also shown in Table 3) overlapped largely with the main genera responsible for the dissimilarity between each of the habitats and the slope. In general, slopes were characterized by several dominant genera (e.g. Thalassomonhystera, Acantholaimus, Halalaimus, Daptonema and Sabatieria) that occurred in similar proportions. The genus Sabatieria, however, declined in abundance below 2000 m and was absent from the abyssal plains and trenches. From this analysis it was also clear that the average communities at abyssal plain, canyon and trench sites shared several dominant genera with the slope communities. The other habitats were more distinct both in the composition of the dominant genera and in their diversity in terms of evenness. The highest dissimilarity with slope communities was found in the seamount samples, which were characterized by high abundances of the genera Desmodora, Richtersia, Ceramonema and Desmoscolex, and a low diversity. In contrast to the slope samples, Thalassomonhystera, Sabatieria, Acantholaimus and Daptonema were uncommon. However, the seamount assemblages were not representative of general patterns because of the low number of samples (n = 3) and the restricted geographical coverage. The same was true for the trench habitat, which was represented by only three samples from the Atacama Trench.

Figure 5.

 Average relative abundances (%) of the main genera responsible for the similarity within a habitat and the dissimilarity between each habitat and the slope, as identified by SIMPER analysis.

Shelf and slope samples also differed in terms of the proportions of taxa. Thalassomonhystera, Acantholaimus and Halalaimus were abundant along the slope but found only occasionally on the shelf. Sabatieria was a dominant genus on the shelf and slope but, on average, less abundant along the slope compared with the shelf. Vent samples differed from slope samples in the increased dominance of the Monhysteridae (Thalassomonhystera and Halomonhystera), and the greater abundance of Anticoma and Desmodora, two genera that were rather rare on the slope. Seep samples differed from slope samples in the high dominance of Halomonhystera and Sabatieria; other typical slope genera, such as Acantholaimus, Thalassomonhystera and Halalaimus, were still present but reduced in abundance. Corals also showed a much reduced abundance of Thalassomonhystera and Sabatieria compared with soft sediments from similar depths, but were characterized by genera such as Desmoscolex and Epsilonema. However, the genera Acantholaimus and Halalaimus were still common. Some typical genera, such as Theristus and Marisalbinema, appeared in the nodule samples but were either less abundant or absent on slopes and in other abyssal samples.

Genus diversity per habitat

Sample diversity, expressed as the rarefaction index EG (51) (expected number of genera for 51 individuals), ranged from 1 to 33 over all habitats (Fig. 6). The highest values were recorded in the slope, shelf, nodule field and coral samples. However, whereas values from the shelf, and particularly from the slope, showed considerable variation, the coral and nodules estimates were always high (>15). Generic diversity was always low in the samples from the hydrothermal vents. The seeps exhibited a range of diversity values from very low to medium. This variation reflected the high degree of small-scale heterogeneity (patchiness) within seeps, which encompass (micro-) habitats ranging from highly-sulphidic sediments with low nematode diversity to well oxygenated surface sediments (e.g. in Siboglinidae tube worm fields) only influenced by seepage in deeper sediment layers and therefore characterized by higher nematode diversity. On average, diversity was lowest in the hydrothermal and seep samples.

Figure 6.

 Expected number of genera [EG(51)] per sample (black dots). Averages and standard deviations shown by the vertical bars with error bars (for number of samples per macrohabitat see Table 2).

Figure 7 shows the total diversity of pooled samples, combining each habitat respectively with the slope to illustrate the extent to which the different habitats contributed to overall slope diversity. As the number of genera will depend on the number of samples analysed within a habitat, diversity is also expressed as EG(100) (Fig. 7A,B). The abyssal plain and nodule habitats contributed particularly to the increased total genus richness of the slope (Fig. 7A). Except for the under-sampled seamounts and trenches, all habitats added to the total genus pool but to a lesser extent (see also Table 2). In terms of expected number of genera (Fig. 7B), the contribution of the abyssal plains and nodule areas became less pronounced due to the higher abundances of dominant taxa found in both these habitats. EG(100) values suggest that, except for the shelf samples, the coral habitat was mainly responsible for the increased diversity, as a result of greater evenness combined with the high number of genera present.

Figure 7.

 (A) Total genus richness. (B) Expected number of genera [(EG(100)] of the slope habitat and the slope combined with each of the other macrohabitats. All samples per macrohabitat are pooled.

Discussion

Methodological problems

Several studies have addressed the importance of habitat heterogeneity at local or regional scales but no previous attempt has been made to determine whether the high turnover between macrohabitats on these smaller scales also holds for larger scales. However, investigating ecological patterns on larger scales requires the compilation of large databases, thereby increasing the heterogeneity of the data involved. The interpretation of the analyses is therefore not without risk (Soetaert & Heip 1995). Data compiled for this study were obtained using a number of different sampling gears, from small box-corers (e.g.Muthumbi et al. 2004) to larger box-corers (e.g.Netto et al. 2005), multiple corers (e.g.Fonseca & Soltwedel 2007), ROV push cores (Van Gaever et al. 2010) or even mussel pots (Flint et al. 2006), for which sampling efficiency is known to vary especially for the surface sediment layers (Bett et al. 1994). Differences in sample processing (sieve size and extraction procedures), and the inherent small-scale and temporal variability, may have added some uncertainties to the comparison. Identification problems can occur, as several genera are differentiated by relatively small differences, possibly subject to personal interpretation. However, potential misidentifications of dominant genera were carefully checked by the different data-providers. As already indicated, the main limitation of the dataset is the unbalanced design in terms of sampling intensity within different habitats and regions. The slope is clearly over-represented compared with all other habitats both in terms of number of samples and geographical coverage. For these reasons, all comparisons between macrohabitats were focused on the slope; in other words, we investigated the extent to which macrohabitats differed in composition and diversity from those of typical slope sediments.

In general, the patterns observed in our analyses were robust across the dataset and the different habitats were represented by a multitude of characteristic genera. We are confident, therefore, that the approach used in this study is the only way to overcome the problems involved in conducting extensive sampling campaigns to detect large-scale patterns in deep-sea nematode communities.

Importance of habitat heterogeneity for deep-sea nematode biodiversity at different spatial scales

At the local scale (diversity per individual sample: Fig. 6), nematode diversity varied significantly within and between habitats. In some habitats, the coexistence of genera was always relatively high, especially in coral and nodule areas, two habitats characterized by an increased substrate complexity owing to the presence of coral rubble, sponge skeletons or manganese nodules on top of the soft sediments. These observations suggest that increased substrate heterogeneity plays an important role in structuring local nematode diversity and are in accordance with the small-scale habitat heterogeneity hypothesis (Bazzaz 1975). This hypothesis, proposed for terrestrial systems, assumes that structurally complex habitats provide more diverse ways for exploiting environmental resources, thereby increasing diversity.

In contrast, the coexistence of genera was occasionally very low in reduced habitats (e.g. seeps and hydrothermal vents), although some seep samples also showed high diversity. In reduced environments, harsh biochemical conditions led to reduced diversity, despite the high food availability. Some opportunistic genera take advantage of the increased organic load associated with seeps or vents and dominate these communities, whereas the more common deep-sea genera disappear. The high variability in diversity estimates within the seep habitat was due to differences in surface biochemical conditions between different seep microhabitats. Soft sediments along the slope also showed high variability in local diversity, from very genus-rich (33) to extremely poor (<5). The low values were often associated with oligotrophic areas with low densities, such as part of the Brazilian margin.

At the large scale, i.e. considering all samples from a given habitat as one (Fig. 7), it was the abyssal habitats which increased the genus richness the most when combined with the slope, contradicting the source sink hypothesis that the abyss only acts as a sink for typical bathyal species (Rex et al. 2005). This is in accordance with previous observations for abyssal copepods (Baguley et al. 2006). Corals increased the total slope diversity through increased evenness, whereas the nodules, an exclusively abyssal habitat, also increased total abyssal diversity. These results suggest that habitat heterogeneity plays an important role in maintaining the regional diversity of deep-sea environments by preserving taxa that are usually rare in soft sediments.

Habitat specific nematode assemblages

The most striking result emerging from the combination of all these independent datasets was that several nematode genera are cosmopolitan, inhabiting a variety of deep-sea habitats and oceans, whereas only a few genera are restricted to a single habitat. In fact, only a minority of genera (about 21% of the total genera) seem to be restricted to one particular habitat. Most of these were encountered in soft slope sediments, which may be partly explained by the higher number of samples collected in these settings (65% of all samples). All other habitats combined only contributed 10% of the genus richness. Most of these habitat-restricted genera were uncommon, suggesting that their absence from other habitats may also reflect (i) under-sampling, (ii) misidentifications or (iii) random colonization of the specific habitat. Most genera have the potential to colonize a variety of deep-sea substrates, although some that are dominant in one habitat are not found in others (e.g. Sabatieria is not found in the abyss) or become rather rare (Acantholaimus and Halalaimus in vents). The eurytopic, cosmopolitan character of most genera does not necessarily apply to species, as the few studies done at species level have shown that, while there may be some widespread nematode species, many are restricted in their distribution (Vermeeren et al. 2004; Ingels et al. 2006; Fonseca & Soltwedel 2007; Fonseca et al. 2007).

This analysis demonstrates that each habitat hosts certain nematode genera that are usually rare in ‘typical’ bathyal and abyssal sediments. This is mainly because such habitats have completely different sedimentary and biochemical characteristics compared to the adjacent sediments. For instance, the three-dimensional structure of deep-sea corals enhances the abundance of non-burrowing, interstitial or epifaunal forms such as epsilonematids, whereas the gravel sediments of the seamounts favour nematodes with coarsely ornamented cuticle, such as Ceramonema, Richtersia and Desmodora. Habitats rich in sulphide and hydrothermal vents had higher abundances of Terschellingia, Sabatieria and Halomonhystera, genera that are better known from organically enriched, shallow-water environments than from other deep-sea habitats.

Corals

The nematode communities associated with cold-water coral habitats included in this analysis were previously described by Raes & Vanreusel (2006) and Raes et al. (2008) from the Belgica Mound region of the Porcupine Seabight (NE Atlantic) at a depth of approximately 1000 m. Here, a series of seabed mounds occurs that support cold-water coral banks and their degradation zones; these zones originate from the progressive degradation of dead coral thickets until only small-sized coral debris remains. Samples were collected in sediment-clogged coral framework (Freiwald et al. 2002), a three-dimensionally complex habitat composed of (i) dead Lophelia pertusa (Linnaeus, 1758) thickets, (ii) glass sponges of the species Aphrocallistes bocagei (Scultze, 1886) and their skeletons, and (iii) sediment. It seems that the three-dimensional micro-structure of deep-sea coral fragments and sponges enhances the abundance of epifaunal nematodes, such as members of the Epsilonematidae and Draconematidae (for details see Raes & Vanreusel 2006 and Raes et al. 2008), which are unusual for ocean margins (Decraemer et al. 2001). Coral fragments and sponges are relatively unprotected on the ocean margin seabed and their associated fauna is therefore subject to stronger current activity, typical of areas with Lophelia reefs (White 2007). Taxa that are specially adapted to crawl on larger surfaces and to withstand this physical stress may have a competitive advantage in such habitats. Epsilonematidae and Draconematidae are characterized by the presence of unique locomotory structures. Most Epsilonematidae have ambulatory setae on the ventral side of their posterior body and Draconematidae have both cephalic and posterior adhesion tubes (Gourbault & Decraemer 1996; Decraemer et al. 1997). Together with the caudal glands, these structures enable the nematodes to attach themselves to a large substratum and/or crawl over its surface in a fashion that is similar to that of a geometrid caterpillar (Stauffer 1924; Lorenzen 1973). A comparable mode of locomotion was observed in Desmoscolex (Stauffer 1924), another dominant genus on coral fragments and sponge skeletons.

Seamounts

Interestingly, higher abundances of Desmoscolex, together with Desmodora, Richtersia and Ceramonema, were also observed on the seamounts included in this analysis (Great Meteor and Sedlo seamounts). In addition, members of the Epsilonematidae and Draconematidae were found here, although in low abundances. The Great Meteor seamount is characterized by coarse biogenic sediments composed of corals and mollusc shells, and by strong current activity (Gad 2004; Gad & Schminke 2004). These environmental conditions could be comparable to those in cold-water coral degradation zones as described above. Indeed, Gad (2004) stated that the nearest congeners of some Epsilonematidae species on the Great Meteor seamount are found in cold-water coral habitats along the North Atlantic continental margin. In addition to their distinct locomotory behaviour, the stout body shape together with the thick cuticle are additional morphological features that may bestow advantages for survival in such physically harsh environments. This comparison suggests that the intricate physical micro-structure of the substrate may be one of the most important factors structuring nematode assemblages. Unfortunately, little detailed information is available on the biology of the genera Desmodora, Richtersia and Ceramonema on the deep-sea floor.

Nodules

Polymetallic nodule deposits on the abyssal seafloor also represent a unique habitat type in which nematode assemblages inhabit both the hard nodule substratum (Mullineaux 1987; Veillette et al. 2007a,b), including the sediment accumulated in crevices on the nodule surface (Thiel et al. 1993), and the soft sediment that underlie the nodules and in which the nodules are partly submerged. Data from two nodule areas in the Clarion-Clipperton Fracture Zone (CCFZ) were analysed: the eastern area (CCFZ-E) (Radziejewska 2002) at depths of about 4300–4400 m and the central area (CCFZ-C) at depths of about 4950–5050 m (M. Miljutina unpublished observations). In both areas, samples were collected, using a multiple corer, from nodule-bearing and nodule-free patches. In the eastern area (CCFZ-E), Desmoscolex and Pareudesmoscolex were among the dominant groups, suggesting again that the presence of hard substrate favours genera with distinct locomotory behaviour. However, in the central area (CCFZ-C), the dominant genera were thread-like interstitial forms such as the Monhysteridae, Acantholaimus and Theristus, These genera were also common in soft sediments from around the World Ocean. Nevertheless, the analysis showed that 22 genera were unique for the nodulized seafloor; furthermore, none of these genera was common to the two CCFZ areas. The differences between these two areas were further accentuated by different dominant genera. In particular, Marisalbinema was one of the characteristic and dominant genera in the CCFZ-C. Also remarkable was the fact that the composition of the nematode fauna in the CCFZ-E differed significantly from all the other deep-sea samples included in this analysis. Owing to the dominance of Terschellingia, these samples showed the highest similarity with seep habitats (Nordic margin) and shelf samples. This observation suggests that the CCFZ-E environment is controlled by some factor(s) in addition to the presence of nodules and may not represent a typical nodule area. Therefore the CCFZ-E samples were not included in the MDS analysis. It is possible that the distinctly different nature of the CCFZ-E nematode fauna (low abundance of abyssal genera such as Acantholaimus and Thalassomonhystera and the high abundance of genera such as Terschellingia) is related to hydrothermal venting, the signature of which, in the form of elevated metal contents in the water column, has been reported from the area (Tkatchenko et al. 1997). Terschellingia dominates cold seep communities on the Nordic margin) (Van Gaever et al. 2009) and is also reported to be abundant in sulphidic, shallow-water habitats (Heip et al. 1985; Vranken et al. 1988). Apparently, nematodes of this genus are tolerant of harsh biochemical conditions that are often lethal to other meiofaunal organisms. In addition, nematode assemblages in the CCFZ-E area showed a distinct temporal shift in the suite of dominants, from Terschellingia in samples from the first (1995) campaign to Desmoscolecidae in the subsequent (1997) sampling programme. This was probably a response to a phytodetritus sedimentation event, the signature of which was detected in the sediment (Radziejewska 2002).

Seeps and hydrothermal vents

Relatively high abundances of certain nematode genera were occasionally observed at hydrothermal vents and more commonly at seeps. In particular, the Nordic cold seep was characterized by higher densities of Halomonhystera (Van Gaever et al. 2006) as well as by Terschellingia, although at lower densities (Van Gaever et al. 2009). The cold seep in the Gulf of Guinea was characterized by the dominance of Sabatieria (Van Gaever et al. in press). High densities of Thalassomonhystera, Halomonhystera and Anticoma were particularly characteristic of hydrothermal vents. Thalassomonhystera is a typical soft bottom deep-sea genus, but the other genera are mostly rare in deep-sea sediments and are known to attain high abundance and dominance in shallow waters (Heip et al. 1985). In particular, Sabatieria occurs at higher abundances along the shelf and upper slope but gradually disappears almost completely in well oxygenated soft sediments below 2000 m, corresponding with a decreasing flux of organic matter (Soetaert & Heip 1995; Vanaverbeke et al. 1997). There are different possible explanations for their presence in the reduced conditions of a seep environment; for example, their relatively larger body size may be an advantage for tolerating low oxygen availability (Jensen 1987). As already observed for other marine nematode genera (Ott et al. 2004), symbioses with sulphur-oxidizing chemoautotrophic bacteria are another adaptation for survival in seeps and hydrothermal vents. However, there is at present little evidence of symbiosis in deep-sea nematodes associated with reduced environments.

Some seep microhabitats, in particular the well oxygenated sediment underneath siboglinid tubeworm patches, are inhabited by a genus-rich nematode assemblage composed of genera similar to those of the slope sediments. Here, genera such as Acantholaimus, Halalaimus and Thalassomonhystera are present in high numbers. Cold seeps therefore harbour a wide variety of nematode assemblages.

Canyons

These large-scale geomorphological features disrupt the monotony of the seafloor and create another source of spatial heterogeneity in the deep sea. Canyon samples included in this analysis covered the Western Iberian Margin (Nazaré canyon) (Ingels et al. 2009), the Mediterranean Sea (Samaria canyon) (N. Lampadariou, unpublished observations) and the West African coast (Zaire canyon) (Van Gaever et al. in press). Canyons are normally characterized by an extraordinary topographic and hydrodynamic complexity, which is peculiar to each site and time scale (de Stigter et al. 2007). Highly active axes and the relatively undisturbed areas, such as the terraces beside the active channels, result in very contrasting environmental conditions (see also Ingels et al. 2009). We might expect that nematode assemblages would respond to the conditions prevailing in each sub-habitat and hence exhibit considerable variability. From the present study, it was indeed clear that heterogeneity in canyons is high, as illustrated by the low similarity value of 27.9%, reflecting their extreme environmental complexity that drives variability on various spatial and temporal scales (Canals et al. 2006; de Stigter et al. 2007). In particular, the highly active canyon axes and the more undisturbed terraces yield nematode communities that are very different in terms of their abundance, composition and diversity (Garcia et al. 2007; Ingels et al. 2009). This is consistent with the ANOSIM tests indicating that canyon assemblages differed significantly from those of the other habitats, except for the slope. Although there was a strong overlap between canyons and slope communities, nematode assemblages in canyons were characterized by a larger number of dominant genera such as Daptonema (4.5%), Paralongicyatholaimus (4.3%), Pomponema (3.5%), Dichromadora (3.5%), Elzalia (3.3%), Halalaimus (3.1%) and Acantholaimus (3.0%). This probably reflects the generally harsh canyon conditions, which lead to an increase in dominance and lower evenness. In contrast to sediments from areas of coral rubbles, sponges, seamounts and nodules, the sedimentary properties in canyons are more similar to those of soft, regular sediments.

Trenches

Only data from the Atacama trench (Gambi et al. 2003) were used in this analysis. This is an atypical hadal system, characterized by close proximity to the continent (c. 80 km) and a location directly beneath one of the largest upwelling regions (Peru-Chile upwelling system). This specific geographic setting imparts the characteristics of a eutrophic system, with extremely high concentrations of nutritionally rich organic matter (i.e. chlorophyll-a and proteins; Danovaro et al. 2002, 2003). It is therefore too early to attempt to establish a general pattern for nematodes in trenches. For example, nematode assemblage composition (not included here) has been analysed from only two other hadal systems (the Puerto Rico and the South Sandwich trenches). Although genus richness decreased significantly from the slope to hadal depths in all three trenches (Tietjen 1989; Gambi et al. 2003; Vanhove et al. 2004), genus composition varied significantly and each system was characterized by different dominant genera (Gambi et al. 2003). Studies carried out in the Venezuela basin and Puerto Rico trench suggest that the decrease in nematode biodiversity at hadal depths reflected, in addition to the reduced food availability, lower heterogeneity in sediment texture (Tietjen 1984, 1989). The more heterogeneous substrates at bathyal depths could be responsible for a higher number of microhabitats and hence an increase of nematode diversity (Tietjen 1984). The role of microhabitat heterogeneity is potentially important also in the Atacama trench, where the rather homogeneous sediments at hadal depths hosted approximately 40% fewer genera than at bathyal sites, where sediments were more heterogeneous (Gambi et al. 2003). The inaccessibility of hadal sediments makes fine-scale spatial studies, and a detailed analysis of microhabitat heterogeneity, difficult. Further studies are needed to clarify the influence of habitat heterogeneity on nematode biodiversity at hadal depths.

Abyssal

In comparison with the slope environment, the vast abyss also represents a peculiar habitat for the fauna (Rex et al. 2005; Smith et al. 2008). The abyss mainly differs from the other habitats considered here in having low current velocity, sediments consisting mainly of fine sand and clay, and habitat heterogeneity created by biogenic structures, such as the tests of giant protozoans and the burrows, mounds and tracks of megabenthos (for review see Smith et al. 2008). The abyss is normally characterized by a distinct macro- and mega-faunal community structure (Rex et al. 2005; Brandt et al. 2007; Smith et al. 2008). For nematodes, this is only partly true, as the ANOSIM did not show that the abyssal assemblages differed significantly from slope assemblages. However, the abyssal samples contained several (n = 16) additional genera not yet recorded from the slope. Nematode assemblages in this habitat are dominated by deposit-feeding genera. In most of the abyssal areas studied (Arctic Ocean, North Atlantic, Northeast tropical Atlantic, Southeast tropical Atlantic, Southern Atlantic, Northeast tropical Pacific), the dominant taxa are the Monhysteridae (including Thalassomonhystera, Monhystrella), Halalaimus and Acantholaimus. Apart from the unvarying dominance of monhysterids, the identity of sub-dominant abyssal nematode taxa seems to be related to the surface primary production. It has already been observed that higher fluxes of particulate-organic carbon promote changes in polychaetes and nematode assemblages in the equatorial Pacific (Smith et al. 1997; Lambshead et al. 2002). However, primary productivity is not the sole factor, as we also observed that areas characterized by similar primary production levels (Northeast tropical Atlantic and Northwest tropical Atlantic) showed different sets of dominant and subdominant nematode genera. In this case, other environmental factors may be involved.

Conclusions

It is apparent from this study that habitat heterogeneity in the deep sea is important for global nematode diversity. However, the question of the extent to which habitat heterogeneity contributes to global diversity has no single answer. It was confirmed by this analysis that many deep-sea nematode genera are cosmopolitan, inhabiting a variety of deep-sea habitats and oceans, whereas only 21% of all deep-sea genera recorded are restricted to a single habitat. Furthermore, the genera restricted to one habitat are never dominant or generally present in all samples within a habitat, suggesting that their presence or absence may be random rather than a selective colonization of particular habitats. On the other hand, different habitats, such as cold seeps, hydrothermal vents, cold water corals and nodule areas, do show typical nematode assemblages with dominant genera that are rare in other habitats. Factors such as increased substrate complexity in the case of nodules and corals, or strong biochemical gradients in the case of vents or seeps, seem to be responsible for the success of particular genera that are not prominent in ‘normal’ soft sediments. Furthermore, clear shifts in the relative proportions of the dominant genera were observed between soft-sediment habitats from the shelf to hadal depths. In this case we can conclude that different deep-sea habitats harbour specific nematode assemblages, but that few genera are restricted to one habitat. The soft sediments of the slope are responsible for more than 60% of all the habitat-restricted genera. However, it must be borne in mind that many other habitats, including nodule areas, corals, seamounts, canyons and trenches, remain under-sampled. In terms of local diversity, the nodule areas and coral rubble samples emerge as habitats where most genera co-exist in equal proportions. In both cases, the added complexity of the substrate facilitates the occurrence of sediment-dwelling as well as epifaunal taxa in the same environment.

Acknowledgements

The authors thank Census of Marine Life project CoMarge for financing the workshop held in Ghent where nearly all co-authors participated in compiling the database and in performing the analyses. This research was further supported by CENPES-PETROBRAS - BRASIL, and the BOF fund from Ghent University. The authors further acknowledge the support by the MarBEF Network of Excellence Marine Biodiversity and Ecosystem Functioning, which is funded by the Sustainable Development, Global Change and Ecosystems Programme of the European Community’s Sixth Framework Programme (Contract No. GOCE-CT-2003-505446). We thank the reviewers and editors for the useful comments on an earlier version of the manuscript.

Ancillary