Comparison of the community structure of the marine molluscs of the “Banco D. João de Castro” seamount (Azores, Portugal) with that of typical inshore habitats on the Azores archipelago

Abstract

D. João de Castro Bank (DJC) is a shallow seamount with hydrothermal activity located between the islands of Terceira and São Miguel, Azores (Portugal). A survey of the molluscan fauna was conducted in this peculiar habitat. We found a smaller number of molluscan taxa and a different community structure compared to shore environments on the Azores. This coincides with impoverished environmental settings at this location. Possible reasons such as the degree of isolation (inshore vs. offshore locations), geological age, and special ecological conditions related to seamounts and hydrothermal vents are discussed.

Introduction

D. João de Castro Bank (38°13.3′N; 26°36.2′W) is an isolated seamount located in the archipelago of the Azores (northern Atlantic), between the islands Terceira and São Miguel (Fig. 1a), at 36 and 40 nautical miles, respectively, from these islands. Today, the upper part of the seamount has its base at 50 m depth and reaches 13 m depth at its top. Inside the volcanic cone (300 × 600 m), the depth is around 40 m (Fig. 1b). The shallower part of the seamount was formed by several submarine eruptions during 2 weeks in December 1720. These eruptions occurred at 100 m depth and were visible from São Miguel and Terceira. The ashes accumulated, forming a small island approximately 1 km across and 150 m high, which was completely destroyed by the winter storms 2 years later (Agostinho 1934, 1960; Weston 1964; Machado 1967).

Fig. 1

a Geographic location of D. João de Castro Bank; b topography of D. João de Castro Bank (^©ImagDOP and ISR/IST); c geographic location of Capelas (CAP) and Ribeira Quente (RQT) at São Miguel Island

Formations such as D. João de Castro Bank with exuberant hydrothermal activity at shallow depths are rare in the Atlantic Ocean (Santos et al. 1996). The gas is released within a small area (100 × 50 m) at the bottom (16–45 m depth) in the western part of the volcanic cone. This gas, mainly CO₂ (90%) with lesser H₂S, H₂, and CH₄ (Cardigos et al. 2005), has its source in a magmatic chamber located between 1 and 5 km under the seafloor (Machado and Lemos 1998). Recently, a second area with hydrothermal activity was discovered, in the depth range of 150 to more than 400 m, southeast of the shallowest part of the seamount (Pascoal et al. 2000). For a discussion on the chemical characterization of the fluids released by the vents at D. João de Castro Bank, see Cardigos et al (2005). The molluscan fauna at D. João de Castro Bank is still poorly known and has mainly been studied from a taxonomical point of view (Ávila 1997; Ávila et al. 2004). Nevertheless, seamounts provide opportunities to study the processes and patterns of dispersion and colonization in marine animals and contribute to the theories of biogeography and speciation (Shuto 1974; Scheltema 1995). The aim of this paper is to compare the molluscan community on this seamount with that of more typical inshore habitats on the Azores, and to relate possible differences to differences in the degree of isolation (inshore vs. offshore locations), geological age, type of substrate, depth, and the presence/absence of hydrothermal vents.

Methods

The molluscan fauna of D. João de Castro Bank (DJC), a shallow offshore bank with hydrothermal activity, was compared with two different inshore sites: Capelas (CAP), a rocky shore, densely covered by algae, without hydrothermal activity; and Ribeira Quente (RQT), a sandy area with hydrothermal activity (located on the north and south coast, respectively, of São Miguel; Table 1). In the Azores, there are no algae-covered inshore rocky sites with hydrothermal activity that could be compared with the DJC Bank.

Table 1 Characterization of the samples

Samples were obtained by SCUBA-diving. At each site, 25 × 25 cm quadrates covered with algae were scraped off with a chisel. A total of 21 quadrates covered by algae were collected, 4 at DJC, 5 at RQT, and 12 at CAP (Table 3). The attached algae and associated organisms were collected into a labeled cotton drawstring bag with fine mesh net. In the laboratory, the samples were sieved by pouring water through a 1 mm mesh sieve. The molluscs were then removed from the algae, sorted to species level under a stereomicroscope, counted, labelled, and preserved in 70% ethanol. Only live molluscs were counted. All samples were given a number and were deposited in the collection of the Department of Biology of the University of the Azores (DBUA). The samples from CAP are a transformed subset of those published by Ávila (2003). The minimum area for subtidal quantitative ecological studies on algae, in the Azores is 50 × 50 cm (Neto 1997), in contrast to 25 × 25 cm for molluscan studies (Azevedo 1991; Ávila 1998). As scraping is a destructive technique, Ávila (2003) used the molluscan data from the algae area, so for the present study, the original data of the 50 × 50 cm CAP samples were transformed into a 25 × 25 cm area by a simple division by the factor 4.

Data analysis

Data analysis was carried out using the software package PRIMER v5.2 (Clarke and Gorley 2001). Species diversity was calculated using species richness (S), and the species diversity indices of Shannon–Wiener (H′) (Pearson and Rosenberg 1978) and Pielou (J′) (Pearson and Rosenberg 1978; Warwick and Clarke 1993). Additionally, the total number of molluscs per sample (N) was used for comparisons between sites. The associations between samples were visualized using the UPGMA clustering technique (Clarke and Warwick 1994). Prior to the analysis, data were transformed by double square root in order to standardize the data and to avoid clustering that would be determined by the most abundant species (Clarke 1993). Transformed species abundance (n ^1/4 _i ) was used to generate triangular matrices of similarities using the Bray–Curtis similarity index (Field et al. 1982; Clarke 1993).

The ANOSIM randomization/permutation test (Warwick and Clarke 1993) was used to test for differences between sites or selected sets of sites. These sets of sites were selected “a priori”, based on their environmental conditions (e.g. sites with and without hydrothermal activity; inshore vs. offshore sites; sites with sand vs. rocky substrate). Species with a ratio higher than 1.4 (using SIMPER subroutine—similarity percentages analysis) were considered as mainly responsible for the dissimilarity between sites/sets of sites (Warwick et al 1990). Similarity percentages analysis examines the contribution of each species to the average Bray–Curtis dissimilarity, within and between groups of samples.

Diversity (species number) of DJC and CAP was compared using previous samples from both places: DJC—Ávila (1997), ASIMOV Expeditions in 1998, 1999 and 2000; Santos et al. (1996), Ávila et al. (2004); CAP—Ávila (2000c, 2003). The number of species was summed cumulatively (using species-area analysis implemented by PRIMER v 5.2; (Clarke and Gorley 2001) over all samples taken at those particular sites (irrespective of the technique employed) and plotted against the number of samples (Williamson 1990; Morri et al. 1999). A tendency line was added to the graph.

Modes of development

The mode of development of the caenogastropods was inferred by scanning electronic microscope pictures of the protoconch in the majority of the taxa, supplemented with data from literature when the species had a well-known life history. Two types of larval development were considered: non-planktotrophic (either lecithotrophic or direct development) and planktotrophic (i.e. with a free-swimming feeding larval stage) (Jablonski and Lutz 1983).

Results

A total of 2,360 molluscs with more than 1 mm were identified (337 at DJC, 219 at RQT, and 1,804 at CAP) (see Table 3). All taxa had already been described from the Azores (Ávila and Azevedo 1997; Ávila et al. 1998, 2000; Ávila 2000a, b). Forty-two taxa are known from the DJC Bank: 32 gastropods and 9 bivalves (some illustrated in Fig. 2), and 1 cephalopod (Ávila 1997; Ávila et al. 2004); no endemic vent species were found.

Fig. 2

a, e Alvania angioyi (van Aartsen 1982); a shell (DBUA 412/22-2), e protoconch (DBUA 335/23-3). b, fAlvania sleursi (Amati 1987); b shell (DBUA 446/34-1), f protoconch (DBUA 446/34-1). c, gManzonia unifasciata (Dautzenberg 1889); c shell (DBUA 662/39-2), g protoconch (DBUA 181/49-2). d, hRissoa guernei (Dautzenberg 1889); d shell (DBUA 662/10-2), h protoconch (DBUA 662/10-2). i, j, nAnachis avaroides (Nordsieck 1975); i shell (DBUA 785/56-1), j shell (DBUA 785/56-1), n protoconch (DBUA 785/63-2). k, oJujubinus pseudogravinae (Nordsieck 1973); k shell (DBUA 574/125-2), o protoconch (DBUA 350/14-9). l, pColumbella adansoni (Menke 1853); l shell (DBUA 570/59-1), p protoconch (DBUA 570/59-1). m, qNassarius incrassatus (Ström 1768); m shell (DBUA 766-F/71-3), q protoconch (DBUA 766-F/71-3). rBittium latreillii (Payraudeau 1826), shell (DBUA 701/13-1). sTricolia pullus azorica (Dautzenberg 1889), shell (DBUA 767/68-1). tGregariella semigranata (Reeve 1858), shell (DBUA 812/176-2). uParvicardium vroomi (van Aartsen, Menkhorst and Gittenberger 1984), shell (DBUA 355)

On average, species number and total number of specimens were lower at DJC and at sites with hydrothermal activity, when compared to CAP. The highest average diversity value (H′ = 1.67) was found at RQT, the other two sites had similar and lower diversity values (see Tables 2, 3).

Table 2 Diversity indices

Table 3 Molluscs collected at D. João de Castro Bank (DJC), Ribeira Quente (RQT), and Capelas (CAP); CAP samples from Ávila (2003), transformed by 1:4

For each location, the maximum densities of the most common species are summarized in Table 4. A clear dominance of bivalves was found at DJC. Gregariella semigranata (Reeve 1858), a small bivalve with a Macaronesian and Mediterranean distribution, was the most abundant mollusc, with an average density (± standard error) of 740 ± 260 m⁻² and a maximum value of 1,296 m⁻². This species and Crassadoma pusio (Poli 1795) were the only species that occurred in all four samples collected at DJC. Anachis avaroides (Nordsieck 1975) and Columbella adansoni (Menke 1853) were common in only some of the DJC samples, but their densities at DJC were higher than at RQT and CAP. Bittium latreillii (Payraudeau 1826) occurred at DJC, but always in very low densities (maximum 192 m⁻², average 6 ± 2 m⁻²), while it was the most abundant species at CAP (maximum density 11,324 m⁻², average 1,377 ± 288 m⁻²) regardless of depth, followed by Tricolia pullus azorica (Dautzenberg 1889), an endemic subspecies that occurred in all CAP samples. Nassarius incrassatus (Ström 1768) was the most common species at RQT. B. latreillii, Parvicardium vroomi (van Aartsen, Moolenbeek and Gittenberger 1984), Rissoa guernei (Dautzenberg 1889) and also Tricolia pullus azorica were present with low densities in most RQT samples.

Table 4 Maximum densities of the most common molluscs collected at DJC, RQT and CAP (ind m^{− 2})

Clustering analysis shows a clear separation between sites with and without hydrothermal activity, all DJC samples grouping together, as well as RQT samples. There is also a split on the CAP samples, related with depth (shallow samples versus deeper samples) (Fig. 3).

Fig. 3

Dendrogram of DJC, CAP, and RQT samples (abbreviations as in Table 3) (Bray–Curtis similarity index, UPGMA)

The pre-defined groups of replicates (see Table 5) were compared against random simulations. A total of 20,000 permutations were used with the subroutine ANOSIM. The test statistic R was low for the comparison between sites with rocky/sandy substrate. In the other comparisons (inshore/offshore sites; sites with/without hydrothermal activity) the differences were higher (Table 6).

Table 5 Pre-defined groups (with respective sampling sites) used for ANOSIM and SIMPER analysis

Table 6 ANOSIM results

SIMPER analysis indicated that G. semigranata (Gre) and C. pusio (Cra) are responsible for more than 50% of the similarity of the samples collected at DJC. At RQT, the typical species is N. incrassatus (Nas), which is responsible for almost 40% of the similarity between the replicates of this location. B. latreillii (Bit) and Tricolia pullus azorica (Tpa) characterize the samples from CAP, defining 45% of the similarity. Inshore sites and sites without hydrothermal activity are typified by B. latreillii and Tricolia pullus azorica, whereas offshore sites are best defined by the presence of G. semigranata. B. latreillii is associated with a rocky substrate covered by algae, and N. incrassatus with a sandy bottom (Table 7). The absence of Tricolia pullus azorica at offshore sites and the higher densities of G. semigranata at these sites define the main differences between inshore and offshore mollusc communities found. The presence of Tricolia pullus azorica also distinguishes sites with and without hydrothermal activity, being very abundant especially in the shallow waters at CAP. This species had low densities at RQT and was absent from the quantitative samples at DJC.

Table 7 SIMPER results

Despite the low number of samples collected at DJC, it is possible to infer, by extrapolation, that for a similar collecting effort, the number of species is expected to be higher at CAP than at DJC (Fig. 4).

Fig. 4

Cumulative number of mollusc species plotted against the number of samples from the sites CAP (filled triangle) and DJC (filled square)

Modes of development

Of the 29 caenogastropods reported for the DJC Bank (Ávila 1997; Ávila et al. 2004), 8 have a non-planktotrophic type of larval development, 17 species are planktotrophic and for 4 species the mode of development remained unknown (Table 8).

Table 8 Number of caenogastropod species and mode of development

Discussion

A pattern that appears from this and other studies (see Costa and Ávila 2001) is that the average numbers and densities of species are lower at hydrothermal vents than at similar sites without hydrothermal activity. At Kraternaya Bight (Kurile Islands), Tarasov et al. (1985, 1986, 1990) and Tarasov and Zhirmunskaya (1989) found that volcanic activity significantly changed the structure and specific composition of the communities. Comparable to these findings, we found striking changes in the relative densities of mollusc species in the presence of hydrothermal vents. In the Azores, in undisturbed environments (CAP), the most common littoral mollusc is B. latreillii (Ávila 2000c, 2003; Costa and Ávila 2001). This species, however, is not common at the shallow hydrothermal vents of RQT and DJC. The endemic Tricolia pullus azorica is common at RQT and in non-venting conditions (Ávila 2003), but only one operculum was found at DJC (Ávila et al. 2004). At CAP, a non-venting site, B. latreillii, T. pullus azorica, Ocinebrina aciculata (Lamarck 1822) define almost 50% of the similarity of the samples. Nassariid gastropods were reported as dominant forms at Milos Island (Aegean Sea) reaching densities higher than 200 m⁻² (Southward et al. 1997) and at Matupi Harbour (Rabaul Caldera, New Britain Island, Papua New Guinea) with densities of 13 m⁻² (Tarasov et al. 1999); both sites exhibit submarine hydrothermal activity. In the Azores, N. incrassatus is a common species in sandy and muddy habitats, being especially abundant inside harbours (personal observation). This species was found at RQT in high numbers (maximum density 256 m⁻²), but although present at DJC (Ávila et al. 2004) and at CAP, its density in these locations was very low. Its abundance at RQT is thus probably related to the sandy type of bottom rather than to the hydrothermal conditions.

Many species associated with hydrothermal areas have adaptations (Ballard 1977; Grassle 1985) that allow them to survive in these extreme environments under the form of chemical contamination (Grassle et al. 1979) and/or extremely high temperatures (Jollivet et al. 1995). The trophic structure of shallow-water hydrothermal ecosystems differs from that of deep-sea hydrothermal ecosystems, which are dominated by chemosynthetic communities (Kharlamenko et al. 1995). At shallow-water hydrothermal vents, bacteria constitute a more important food source than at typical coastal ecosystems (Tarasov and Zhirmunskaya 1989; Kharlamenko et al. 1995). Rock samples collected from around the vent outlets at DJC were partially covered with bacterial mats, mainly an attached form of Beggiatoa (Cardigos et al. 2005).

The dominance of bivalves in DJC might thus be related to their filter-feeding habit which allows these animals to take advantage of the additional primary production by chemosynthetic bacteria living nearby the vents. This relationship was also used to explain the high diversity and density of Serpuloidea (Annelida: Polychaeta), also filter-feeders, in venting conditions on Milos Island (Bianchi and Morri 2000). No specialized hydrothermal molluscs were found at DJC or RQT, similarly to what was reported from other locations with shallow-water vents (Tarasov et al. 1986, 1990, 1999; Kamenev et al. 1993; Dando et al. 1995). Nevertheless, physiological and ecotoxicological tests are needed to study the tolerance to the toxic compounds that are released by these vents, such as Cu, Pb, Ba, Fe, Mn, Co, and Cd (Cardigos et al. 2005).

Similar to other seamounts around the Azores, DJC has a predominance of molluscs with a planktotrophic mode of development (17 taxa or 58.6%). Although there is still a high number of Azorean caenogastropod species with unknown mode of development, preliminary results indicate that there is a slightly higher number of planktotrophic (33.9%) relative to non-planktotrophic caenogastropods (29.8%) (Table 8).

At Ormonde peak (the top of the eastern seamount of the Gorringe, located southwest of Portugal) Ávila and Malaquias (2003) investigated the mode of development of the caenogastropods located in the photic zone (down to 50 m depth) and found that from a total of 25 taxa, 11 species (44%) were non-planktotrophic, whereas 14 taxa (56%) were planktotrophic (Table 8). Leal (1991) did a similar study at Atol das Rocas (ROC) and Fernando de Noronha islands (NOR), located off the northeastern Brazilian coast, and at Trindade (TRI) and Martin Vaz Islands (MAR), located off the eastern Brazilian coast. We have modified his data by pooling his littoral and sublittoral species (down to 30–40 m depth) into a single category and by using only two modes of development: non-planktotrophic (with pooled lecithotrophic and direct development) and planktotrophic. At TRI there is an equilibrium between both modes of development, whereas at ROC, NOR and MAR there is a higher number of species with non-planktotrophic development (Table 8). This is especially evident at ROC, the island nearest to Brazil, located just 200 km off the coast, with 64% of non-planktotrophic and 32% of planktotrophic species.

The bias in favour of planktotrophic developers is probably related to the relatively early age of the DJC Bank (less than 300 years), its higher degree of isolation from source populations, and the small area (only 300 × 600 m above 50 m depth). For non-planktotrophic species such as Calliostoma sp., Jujubinus pseudogravinae, Alvania angioyi, Alvania sleursi, A. avaroides and Omalogyra atomus, and especially for those with direct development, the major problem is to reach this seamount. Once there, and provided the species’ ecological demands are met, gravid females may give rise to new successful populations (Leal 2000). Planktotrophic species may reach DJC Bank with higher frequencies than non-planktotrophic ones, but they will meet with greater difficulties in establishing self-sustained populations, because most larvae will get lost by drifting away from the small area of the seamount. As a consequence, these “pseudopopulations” (Bouchet and Taviani 1992) can persist only by a regular input of larvae from other source-locations, in the case of DJC from nearby islands of the Azores archipelago (e.g. Terceira and São Miguel).

The absence at DJC of species common on the intertidal zone of the Azorean shores (e.g. the littorinids: Littorina striata and Melarhaphe neritoides, the little skeneopsid: Skeneopsis planorbis, the anabathrid: Pisinna glabatra, and the rissoids: Cingula trifasciata and Alvania mediolittoralis) must be related to a lack of the physical and/or biotic conditions typical of littoral zones. The highest elevation of DJC is at around 13 m depth. Consequently, species that usually live in the upper 10 m can be expected to be absent from the DJC Bank. Ávila (2003) established the zonation pattern for the most common shallow marine molluscs of the Azores. From these data and the mollusc checklists of DJC (Ávila 1997; Ávila et al. 2004) it is evident that why many common littoral Azorean shore species are missing at DJC (e.g. the rissoids: R. guernei and Manzonia unifasciata, the trochid: Gibbula delgadensis).

Features inherent to offshore shallow seamounts, such as turbulent hydrodynamic conditions, both currents and waves, are likely to result in reduced diversity and abundance of species, as seen at DJC. On the other hand, this may explain the higher numbers of the two mentioned bivalves, one (Crassadoma) cemented to the substrate, and the other (Gregariella) attached byssally.