2.5.1 Sequence data
We obtained 122 DNA and 135 cDNA sequences, after removing four and five chimeras, respectively. Average neighbor clustering of DNA and cDNA sequences up to 10% di-vergence did not show large variation of the number of OTUs between the OTUs formed at 3% and 5% levels (Table 2.3: Supplementary information). The 3% divergence level was thus considered sufficient to balance PCR-driven and sequencing errors as well as intra-specific variation. The number of OTUs formed at 3% divergence was similar for the DNA and cDNA datasets with 35 and 36 OTUs, respectively. Once combined and clustered at 3% divergence, the total number of OTUs in both datasets composed of 257 sequences was 62, suggesting that most of sequences obtained from DNA and
cDNA amplifications were different.
2.5.2 Sequence identification
Taxonomic assignments gave similar results using either signature or full length se-quences. Both methods allowed the assignment of 93.8% and 79.8% of the sequences to the same first and second taxonomic levels, respectively. It was not possible to assign 39 out of the 52 cases to the second level using one of the two approaches. The remaining 13 cases of inconsistent assignments were due to conflicts that were checked manually. In general, the signature method performed better, with only 3, 5 and 40 cases of taxonomic conflicts at the three respective taxonomic levels against 23, 85 and 58 cases observed with the second method. In total, 76 out of 257 sequences (29.5%) were assigned to the species level, 169 sequences (65.7%) were assigned to the second taxonomic level, and 15 sequences remained unassigned.
The assigned sequences were clustered into 62 OTUs and evenly distributed among major groups (Table 2.2). The Textulariida (TEX) were the most diverse with 16 OTUs, while monothalamids (MON), Rotaliida (ROT) and undetermined foraminifera (UNK) accounted for 14 OTUs each. Robertinida (ROB) and planktonic Globigerinida (GLO) orders were only represented by three and one OTUs, respectively. Rotaliida and Textulariida were the most sequenced orders with 81.7% of all sequences (ROT:
155 sequences, TEX: 55 sequences). Interestingly, although the same number of OTUs was found for Rotaliida and monothalamids, the later were only constituted of 21 se-quences. Even more striking was the divergence level among UNK sequences, with 14 OTUs formed out of 15 sequences. The sequences assigned to GLO showed >98%
identity to the reference sequence of Neogloboquadrina pachyderma.
2.5.3 DNA versus cDNA datasets
The number of sequences and their taxonomic richness inferred from the DNA and cDNA datasets were similar (Fig. 2.1A). However, the sequences assigned to Rotali-ida were more abundant in the cDNA dataset, in which OTUs assigned to Elphidium, Epistominella and Uvigerina were exclusively detected. The number of textulariid OTUs in cDNA dataset accounted for half of the OTUs found in the DNA dataset,
Table 2.2: Taxonomic identification of environmental deep sea sediment Foraminifera sequences. The number of Operational Taxonomic Units (OTUs) is indicated for each major taxonomic group and sample and the number of sequences in these OTUs is indicated into brackets. The columns labelled “Total”, “OTUs” and “Sequences” cor-respond to the numbers and proportions after average neighbor clustering of combined DNA and RNA datasets at 3% divergence.
A2-6 A3-5 A7-6 B1-5 B4-4 B5-3 B7-2 Total OTUs (%) Sequences (%)
Sequences 40 50 47 26 39 36 19 257
OTUs 17 12 16 10 10 14 9 62
Taxonomic groups
Monothalamiids 3 (4) 2 (4) 6 (6) 1 (1) 1 (2) 3 (4) 0 14 (21) 22.58 8.17 Rotaliida 8 (27) 5 (38) 4 (21) 3 (15) 5 (25) 4 (18) 6 (11) 14 (155) 22.58 60.31 Textulariida 1 (1) 3 (5) 6 (16) 6 (7) 4 (12) 6 (13) 1 (1) 16 (55) 25.81 21.40
Robertinida 1 (1) 0 0 0 0 0 2 (7) 3 (8) 4.84 3.11
Globigerinida 0 0 1 (2) 1 (1) 0 0 0 1 (3) 1.61 1.17
Undetermined 6 (7) 3 (3) 2 (2) 2 (2) 0 1 (1) 0 14 (15) 22.58 5.84
with most of the OTUs assigned to an environmental textulariid clade. Interestingly, although similar proportion of monothalamids were found in the two datasets, not a single OTU was shared (Fig. 2.1B). The diversity of monothalamids was higher in the DNA dataset with eight sequences assigned to seven different clades, contrasting with the six monothalamid OTUs delineated from 14 cDNA sequences. Out of the seven OTUs stemming from DNA sequences, two corresponded to environmental clades (EN-FOR5 and ENFOR6) while three out of the six RNA-derived OTUs were identified to well described clades and genera: cladeA (Allogromia) and cladeBM (Bathysiphon and Micrometula). Notably, planktonic foraminiferal sequences were only detected in the DNA dataset while sequences belonging to the genus Robertina were found exclusively in the cDNA dataset. Only eight OTUs belonging to rotaliids and textulariids were shared between DNA and cDNA datasets (Fig. 2.1B). However, these eight OTUs were represented by 178 sequences, i.e. 69.2% of all sequences.
2.5.4 Bathymetric distribution
The comparison of foraminiferal sequences across two bathymetric transects from the continental margin to the deep Sea of Japan, displayed interesting distribution patterns (Fig. 2.2). One OTU found both in DNA and cDNA datasets and corresponding to an environmental rotaliid lineage was restricted to shallower samples (A2-6, A3-5, B7-2 and B5-3) with only one sequence found in the DNA of deeper B4-4 sample. Conversely,
Figure 2.1: Comparison of foraminiferal assemblages inferred from DNA and cDNA datasets.
(A) Proportions of assigned foraminiferal OTUs formed at 3% divergence. The first two tax-onomic levels are detailed. Hatched bars correspond to assignments to the third, species level (i.e. >97% similarity to a reference sequence). Numbers on top represent the total number of sequences and OTUs. (B) Venn diagram showing the observed number of unique and shared OTUs and sequences (italic) among DNA and cDNA datasets at 3% divergence. The proportion of sequences in each shared OTU is displayed both for DNA (inner ring, 86 shared sequences) and RNA (outer ring, 92 shared sequences). GLO: Globigerinida, MON: monotha-lamids; ROT: Rotaliida, ROB: Robertinida, TEX: Textulariida, UNK: Undetermined.
another abundantly sequenced rotaliid OTU was mainly restricted to deeper samples (A7-6, B1-5 and B4-4), except for a RNA sample from B5-3 station located close on the slope. Two eurybathymetric OTUs corresponded to environmental lineages of rotaliids and textulariids while no single OTU was shared between the three deepest samples
(A7-6, B1-5 and B4-4; >3000 m water depth) and the two shallowest samples (A2-6 and B7-2; <1000 m water depth) (Fig. 2.4: Supplementary information). Within each sample, more than 50% of the cDNA sequences were assigned to Rotaliida while this proportion was lower for DNA sequences, especially in the deep samples with 37% (A7-6) and 25% (B1-5). Three of these RNA-derived cDNA sequences were assigned to well described rotaliids. The sequences of genera Robertina and Epistominella were found in the shallowest samples of both transects, while those ofBolivina andElphidiumwere found in A2-6 sample. Remarkably, Rose Bengal stained cells ofEpistominella, Elphid-iumandBolivinawere also observed in the corresponding samples (Fig. 2.3). The OTU assigned to a planktonic species was found in the DNA dataset of the deepest station of each transect (samples A7-6 and B1-5). The majority (61.9%) of monothalamid se-quences were present in the cDNA dataset of the five shallower stations while most of the DNA monothalamid diversity was concentrated in the deep station of transect A, with five out of eight DNA OTUs (sample A7-6). In this station, the monothalamids were mainly represented by environmental clades.