Molecular analyses reveal high levels of eukaryotic richness associated with

Dans le document New insights into the diversity of deep-sea benthic foraminifera (Page 96-114)


3.2 Molecular analyses reveal high levels of eukaryotic richness associated with

Béatrice Lecroq1,Andrew John Gooday2, Tomas Cedhagen3, Anna Sabbatini4 and Jan Pawlowski1

1Department of Zoology and Animal Biology, University of Geneva, Switzerland.

2National Oceanographic Centre, European Way, Southampton SO14 3ZH, UK.

3Department of Marine Ecology, University of Aarhus, 8200 Aarhus N, Denmark.

4Department of Marine Sciences, Polytechnic University of Marche, 60131 Ancona, Italy

Published in: Marine Biodiversity (2009) 39: 45-55

doi: 10.1007/s12526-009-0006-7


Komokiaceans are testate agglutinated protists extremely diverse and abundant in the deep sea. About 40 species are described and share the same main morphological feature: a test consisting of narrow branching tubules forming a complex system. In some species, the interstices between the tubules are filled by sediment creating a mudball structure. Because of their unusual and sometimes featureless appearance, komokiaceans were frequently ignored or overlooked until they formal description in 1977. The most recent taxonomy places the Komokiacea within the Foraminifera based on general morphological features. To examine their taxonomic position at the molecular level we analysed the SSU rDNA sequences of two species, Normanina conferta and Septuma ocotillo, obtained either with specific foraminiferal or universal eukaryotic primers. Many different sequences resulted from this investigation but none of them could clearly be attributed to komokiaceans. Although our study failed to confirm univocally that Komokiacea are foraminifera, it revealed a huge eukaryotic richness associated with these organisms, comparable to the richness in the overall surrounding sediment. These observations suggest strongly that komokiaceans, and probably many other large testate protists, provide a habitat structure for a large spectrum of eukaryotes, significantly contributing to maintaining the biodiversity of micro- and meiofaunal communities in the deep sea.


The ocean floor is a remote environment, characterised by the absence of solar light, high pressures and food limitation, in which the microbial fauna and small eukaryotes play an important role in food webs and carbon cycling (Vickerman 1992; Moodley et al. 2002).

Within the meiofaunal size fraction (small eukaryotes), most of the biomass is made up by nematodes and foraminifera. At abyssal depths, the latter group is dominated by agglutinated species, which are very diverse and include many undescribed taxa. The superfamily Komokiacea (Tendal and Hessler 1977) is one of the most widely distributed deep-sea foraminiferal groups and also the most controversial one. Prior to their classification as Foraminifera, placements within the sponges and xenophyophores were proposed (Hessler and Jumars 1974). Komokiaceans (informally termed ‘komoki’) are particularly common in

oligotrophic, abyssal regions, but sometimes have broad bathymetric ranges (Schröder et al.

1989). They have a delicate, flexible test consisting of a complex system of fine branching tubules, sometimes subdivided by septae with foramina. In some forms, the interstices between the tubules are filled with mud to create a mudball structure (Shires et al. 1994). The tubules contain stercomata (waste pellets) and diffuse cytoplasm that does not stain readily with rose Bengal, making the recognition of individuals that were alive at the time of collection difficult. In at least some species, the cytoplasm appears to be multinucleate (Tendal and Hessler 1977). The fragility of komokiacean tests impedes their preservation as fossils; indeed, the existence of fossil komokiaceans has never been confirmed (Gooday et al.

2007a). Moreover, because of their peculiar structure and wide variety of tests morphologies, komoki are difficult to recognise as living organisms and in the past have often been overlooked (Tendal and Hessler 1977).

Although they possess some typical attributes of agglutinated Foraminifera, including a test wall comprising an organic layer overlain by a veneer of attached particles, the classification of komokiaceans within the Foraminifera is controversial. In particular, some typical foraminiferal features, such as granuloreticulate pseudopodia, have not yet been observed. This has led some authors to conclude that they represent a separate taxon, close to the foraminifera (Kamenskaya 2000). In order to explore the taxonomic position of komokiaceans, we analysed partial SSU rDNA sequences obtained from several specimens of two species from the Weddell Sea and North Atlantic: Septuma ocotillo Tendal and Hessler, 1977 and Normanina conferta (Norman 1878) (Gooday et al. 2007b). Our study failed to identify sequences that were derived univocally from komokiaceans. On the other hand, it revealed a spectacular eukaryotic richness associated with these overlooked organisms. This richness, as well as the possible reasons that impede the identification of genuine komokiacean rDNA sequences, are discussed.

Material and Methods

Sediment sampling and DNA extraction

Samples were collected during tree consecutive expeditions: “ANDEEP II”, (RV Polarstern cruise ANT XXII/3, Weddell Sea, Antarctica, 2002), RRS Charles Darwin cruise 158 (North Atlantic, 2003) and “ANDEEP III” (RV Polarstern cruise ANT XIX/4, Weddell Sea, Antarctica, 2005). One hundred thirteen komokiacean specimens (or fragments) of two species: Septuma ocotillo (71) and Normanina conferta (42) were analysed during this study.

Samples were collected with boxcorer, multicorer, Epibenthos sledge and Agassiz trawl at the depths and coordinates indicated in Appendix E, Table E1. In the case of corers, first top 1 cm layer of sediment was sliced off and sieved on 500 µm, 300 µm and 125 µm meshes. Some specimens were collected by elutriation (Sanders, Hessler and Hampson 1965); others were sorted by hand under a binocular microscope (mostly from the 300-500 µm fraction). Isolated specimens were either fixed directly in guanidine DNA extraction buffer (Pawlowski 2000) or stored frozen at -80°C. Some of them were cut into 2 or 3 pieces in order to compare sequences derived from different parts from the same individual. At stations where komoki were found and successfully extracted, 14 frozen sediment samples of about 1 ml were extracted with the FastDNA SPIN Kit for Soil (QBIOgene).

DNA amplification, cloning and sequencing

Partial SSU rDNA were amplified by PCR with a set of foraminiferal specific primers:

s14F3 (5’-ACGCA(AC)GTGTGAAACTTG-3’); s14F1 AAGGGCACCACAAGACGC-3’); s17 CGGTCACGTTCGTTGC-3’) and a set of universal eukaryotic primers s12.2


ATCA-3’). In the case of amplification with foraminiferal primers, the first PCR obtained with primers pair s14F3/s17 was reamplified using nested primer s14F1. Other foraminiferal and eukaryotic primers were also tried but did not provide successful amplifications.

Amplified products were purified using the High Pure PCR Purification kit (Roche) and cloned into the sequencing vector “pGEM®-T Easy (Promega)” and replicated in DH5α Competent Cells. After purification, 2 to 5 clones for each sample were sequenced from both directions for cross-checking. Finally, a total of 191 clones (46 for S. ocotillo, 77 for N.

conferta and 68 for the sediment samples) were sequenced. GenBank accession numbers are indicated in Appendix E, Table E2.

Phylogenetic analysis

Sequences were aligned manually using Seaview software (Galtier et al. 1996). Our database of 871 foraminiferal sequences was used to align the sequences obtained with foraminiferal specific primers. The 60 foraminiferal sequences most closely related to those obtained from komoki were retained in final analysis. A database of 505 eukaryotic sequences was used to analyse the sequences obtained with universal s12.2 – sB primers. Poorly aligned positions and divergent regions were eliminated and the alignments of foraminiferal and eukaryotic sequences were analyses separately. To increase the resolution of the eukaryotic phylogenetic analyses, sequences were analysed in four different sets, corresponding to the supergroups of eukaryotes recently defined (Keeling et al. 2005). The first set corresponds to

“unikonts” (Cavalier-Smith 2002) and includes Amoebozoa, Metazoa, Fungi and other Opisthokonta. The second set consists of various “bikonts”: Apusozoa, Euglenozoa, Plantae, Glaucophyta, Haptophyta and Cryptophyta. The third set is composed of Alveolates and Stramenopiles. Finally the fourth set represents the Rhizaria. The phylogenetic affiliation of each obtained rDNA sequence was initially investigated by NCBI BLAST search (Altschul et al. 1990) in order to include their closest relative sequence from GenBank (Bilofsky 1986).

Trees were built according to the Maximum Likelihood (ML) method using PhyML program (Guindon and Gascuel 2003) with the GTR + I + G model suggested by Mr Modeltest (Posada and Crandall 2001) with 10,000 replicates for bootstrap analysis. Additionally, bayesian phylogenetic analyses were performed on the same dataset with the same model using MrBayes program (Huelsenbeck and Ronquist 2001). All analyses were performed on the freely available Bioportal (, accessed 2008). In order to sum up all information from this phylogenetic analysis, we built pie charts of richness for S.

ocotillo, N. conferta and sediment samples respectively. The total number of Operational Taxonomic Units (OTU) found in each of the 12 major eukaryotic groups appearing in the phylogenetic trees was counted and reported in a pie chart.

Results and Discussion


In order to identify a “characteristic” komokiacean SSU rDNA sequences, extractions of Normanina conferta and Septuma ocotillo were amplified first with foraminiferal specific primers. Several combinations of these primers were tested in nested PCR with different annealing temperatures and different numbers of cycles for the first and second amplification.

Most of these combinations did not give any PCR product, suggesting that komoki do not have a “typical” foraminiferal sequence (if any komokiacean DNA has indeed been extract from those samples). Only the set of primers s14F3-s17 reamplified with s14F1-s17 led to a successful amplification in the case of 25% and 50% of S. ocotillo and N. conferta samples, respectively.

In total, 23 foraminiferal sequences were obtained for both species (15 for S. ocotillo and 8 for N. conferta). Their phylogenetic analysis shows that they do not group together but are widely spread between different foraminiferal taxa (Fig. 3.2.1). Some of them, such as S5930-82, which is identical to the sequence of Marginopora vertebralis from tropical shallow waters, are certainly a contamination by foreign foraminiferal DNA. Others are similar to the sequences of foraminifera that could be present in the area where the komoki were sampled, but which belong to different well-defined taxonomic groups. For instance, the sequences S5132-12 and N5976-102 are close to the calcareous rotaliid Oridorsalis umbonatus, while the sequence S5928-11 is close to calcareous miliolid Cornuspira antarctica. As these species are morphologically very different from the komokiaceans, the only explanations for their presence in our DNA extractions are that they represent foraminiferal inhabitants of the komokiacean tests, or the species living in their vicinity.

Most of foraminiferal sequences obtained here can be convincingly attributed to inhabitants of komokiaceans. However, for some sequences, a real komokiacean origin cannot be completely eliminated. In theory, we should obtain a separate clade consisting of two distinct subclades, one grouping sequences from S. ocotillo and the other grouping sequences from N. conferta. In our tree (Fig. 3.2.1) the best candidates (which are supposed to be distinct lineages) are the clade formed by N3502-52 plus N3474-52 and the isolated sequence N3509-32.

Figure 3.2.1. SSU rDNA maximum likelihood phylogenetic tree of foraminifera showing the position of the sequences obtained from N. conferta (in red) and S. ocotillo (in blue) in amplification using foraminiferal specific primers. Only support values higher than 60% are indicated.

However, the sequences N3502-52 and N3502-A originate from the same specimen but do not branch together, which suggests that one or other (if not both) represents an unknown foraminiferal lineage associated with komoki rather than the komoki themselves.

There is only one case where sequences from S. ocotillo and N. conferta branch together.

S5008-62 and N3502-A form a separate clade of two genetically extremely close sequences.

However, sequences that are so similar are unlikely to originate from two morphologically very distinct genera.

Other eukaryotes

One hundred and ninety-one sequences were obtained with the “universal” eukaryotic primers: 46 from S. ocotillo, 77 from N. conferta and 68 from sediment samples. The analysis of these sequences was performed separately for four large groups that correspond roughly to the eukaryotic supergroups as defined by Keeling et al. (2005). Five groups of sequences and one clone were not related to any well defined taxonomic group present in the GenBank database and therefore have been labelled “undetermined lineage” or “undeterminated clone”

with the name of the group to which they refer in literature when it was available.

Fig. 3.2.2 presents a partial SSU rDNA tree of Unikonta, including 38 new sequences.

Most of these sequences are found among Fungi. Many of them could result from contamination from different sources, including fungal spores in deep-sea sediments (Damare et al. 2006) and because almost all of them have a sister group among known fungal taxa, it is unlikely that they represent komokiaceans. On the other hand none of those fungal sequences was found in the sediment samples which might suggest that some fungi could somehow benefit from komokiacean structures or from komokiacean metabolites. Two sequences of N.

conferta and one of S. ocotillo branching within Metazoa could be some small annelids or other invertebrates living inside komokiacean tubes, but the examined fragment is too short to identify them safely.

Figure 3.2.2. Phylogenetic position of partial SSU rDNA sequences obtained from komokiacean and sediment samples grouping within Unikonta. The sequences from N. conferta are in pink, those from S. ocotillo are in blue and those from sediment samples are in green. Environmental sequences from GenBank are indicated with the original name of the clone and information on the sampling location. Except for outgroups, the eukaryotic sequences which did not branch with any new sequences were discarded from the trees. The black circles reflect the support of the nodes for both ML and Bayesian trees: 100% for the big circles and more than 50% for the small ones.

Fig. 3.2.3 presents various groups of Bikonta including Apusozoa, Euglenozoa, Plantae, Glaucophyta, Centrohelida, Haptophyta and Cryptophyta. Many of new sequences branch among Euglenozoa. Some of them form a very important clade distantly related to Euglena. It is composed of 7 N. conferta and 10 S. ocotillo sequences but without segregation between the two genera. For this reason it seems unlikely that this clade represents authentic komokiaceans. Except for this clade, we did not find other Euglenozoa among sequences derived from komoki, although many of them were present in the sediment samples. A few new sequences branch among Plantae and Haptophyta. As these are photosynthetic groups, our sequences certainly originated from organisms living in the surface waters. Presumably, these were conveyed rapidly to the seafloor on sinking particles (Thiel et al. 1989). Some of the sequences form two distinctive clades: “undetermined lineage 2” related in the literature to the clade “OLI11011” (Edgcomb et al. 2002) and “undetermined lineage 3” related to the clade “DH148-5-EKD18” (Takishita et al. 2007). One of these clades (“undetermined lineage 2”) branches as sister group to Apusozoa, but with no bootstrap support. The other (“undetermined lineage 3”) form a distinctive lineage separated from other groups by a very long stem branch. In this clade, we found 9 sequences from N. conferta and 1 from S. ocotillo branching together with six environmental sequences. As those “environmental” sequences also originated from the deep-sea samples, this clade could potentially represent komokiaceans, yet the strong divergence between N. conferta sequences and the presence of only one S. ocotillo sequence does not strongly support this hypothesis.

Fig. 3.3.4 shows a tree of Alveolates and Stramenopiles with very diverse new sequences. Most of them branch within the diatoms and probably represents organisms that have settled from the water column. Several other sequences were derived from ciliates or dinoflagellates. Interestingly, very few komokiacean sequences branch together with sequences from the sediment suggesting that different eukaryotes inhabit the komokiacean tests and the sediment. One lineage, “undetermined lineage 4”, was found to be related in the literature to the clade “NAI.2” (Not et al. 2007) and contains one S. ocotillo sequence.

However, its position within the dinoflagellates is rather well supported excluding the possibility that represents new komokiacean lineage.

Figure 3.2.3. Phylogenetic position of partial SSU rDNA sequences obtained from komokiacean and sediment samples grouping within Bikonta, except Alveolates, Stramenopiles and Rhizaria (see legend Fig. 3.2.2 for details).

Figure 3.2.4. Phylogenetic position of partial SSU rDNA sequences obtained from komokiacean and sediment samples grouping within Alveolates and Stramenopiles (see legend Fig. 3.2.2 for details).

Finally, Fig. 3.2.5 presents a tree of Rhizaria that also reveals a striking diversity of the new komokiacean sequences. Many of these sequences, including 19 from sediment samples, 13 from S. ocotillo and 3 from N. conferta, are located among the core Cercozoa.

Haplosporidia mainly include sequences from komoki and probably correspond to parasites living inside their tests. One sequence from the sediment (SED833) also represented a sister group of Foraminifera without any relatives and was therefore labelled “undetermined rhizarian clone”. New sequences were not found within Foraminifera because the eukaryotic primers that we used do not match the foraminiferal rDNA. A very distinct clade (“undetermined lineage 5”), composed of 10 sequences from the sediment, 1 from N. conferta and 1 from S. ocotillo, branched as a sister group to the polycystine radiolarians represented by Collozoum inerme. Other four sequences from the sediment appeared as sister group of the taxopodid Stycholonche zanclea.

Why it is so difficult to identify the true komokiacean rDNA sequences?

In previous studies, we have recovered DNA from foraminifera obtained at abyssal depths down to >6000 m (Gooday et al. 2004). However, in the present study, we were unable to determine which, if any, of the numerous sequences obtained from two komokiacean species represented the actual komokiaceans. There are several possible explanations for this lack of success.

First, the majority of examined specimens could be dead and their “well preserved”

agglutinated tests difficult to distinguish from living specimens. Although many individuals were filled with stercomata, these waste pellets probably persist intact for a considerable time after death. As is typical for abyssal foraminifera, we did not observe any signs of cellular activity, such as the pseudopodial movement or the cytoplasmic flux. Indeed, the cytoplasm of komokiaceans is diffuse and difficult to observe. Therefore, we cannot be sure that the DNA was actually extracted from the living komokiacean specimens.

Second, our specific foraminiferal primers may not fit to the komokiacean rDNA sequences. These primers were designed based on the specific region in the foraminiferal SSU rDNA containing unique 3-nucleotides insertion found exclusively in this group (Pawlowski 2000).

Figure 3.2.5. Phylogenetic position of partial SSU rDNA sequences obtained from komokiacean and sediment samples grouping within Rhizaria (see legend Fig. 3.2.2 for details).

Using our specific foraminiferal primers, we have successfully amplified the majority of foraminiferal taxa (Bowser et al. 2005). It would be very surprising if the komokiaceans are Foraminifera but do not recognise the specific foraminiferal SSU rDNA insertion, and at the same time do not amplify with universal eukaryotic primers.

Third, the komokiacean rRNA genes could be represented by a very low number of copies. In this case, the rDNA of other organisms coexisting in the sample could be amplified much more easily than the authentic komokiacean genes. The lack of clear evidence for authentic komokiacean rDNA sequence might result from a bias of amplification promoting majority copies of komoki inhabitants.

Komokiaceans as hot spots of eukaryotic diversity

The taxonomic composition of eukaryotes found in S. ocotillo, N. conferta and in the sediment samples is presented in Fig. 3.2.6. Twelve major eukaryotic groups identified in this study are represented in very different proportions. Each slice of pie chart reflects the number of OTU found in one of those major groups.

Both komokiacean and sediment samples yield an amazing diversity of major taxonomic groups and constituent OTUs. Another study based on sediment from a deep-sea methane cold seep reported similar results described by the authors as “an unexpected high diversity of microbial eukaryotes at various taxonomic levels” (Takishita et al. 2007). It is remarkable to find this level of diversity in 14 ml of sediment (resulting from the 14 extractions) and even more surprising to obtain a similar richness from a few (17 S. ocotillo and 23 N. conferta) millimeter-sized organisms, in some cases only fragments of specimens, with a total volume smaller than 1 ml. From this observation we can postulate that komoki concentrate the diversity inside or outside their branching tests. Fungi represent most of the eukaryotic richness in samples from N. conferta, with 28 different OTU recognised. The Euglenozoa (10 OTU) was the most diverse group in S. ocotillo, while the Cercozoa was by far the richest group (19 OTU) in the sediment samples.

Most of the new komokiacean or sediment sequences consisted of clearly distinct OTUs; in only a very few cases did two sequences represent the same phylotype. By sequencing more

Most of the new komokiacean or sediment sequences consisted of clearly distinct OTUs; in only a very few cases did two sequences represent the same phylotype. By sequencing more

Dans le document New insights into the diversity of deep-sea benthic foraminifera (Page 96-114)