Supplementary information
1
2
Detailed Material and Methods
3
4
5
Flow cytometry sorting
In the present study a two step sorting procedure was applied to sort photosynthetic pico
6
and nanoeukaryotes (See main text). A nucleic acid stain, SYTO 13 (del Giorgio et al., 1996)
7
was used before the second sorting step. SYTO13 stains all cells, including heterotrophic
8
bacteria and eukaryotes and thus allows the discrimination of them from photosynthetic cells
9
which are SYTO stained but exhibit also the Chl-a fluorescence. In contrast without SYTO 13
10
heterotrophic cells are not visible in flow cytometry and can be sorted along with
11
photosynthetic cells.
12
13
14
DNA extraction and cell lysis
Genomic DNA was extracted from 20 phytoplankton strains isolated during the MALINA
15
cruise (Supplementary Table S1) as well as from 4 samples of sorted photosynthetic
16
picoeukaryotes (ES106, ES107, ES113 and ES114) and 1 sample of sorted photosynthetic
17
nanoeukaryote (ES116, Table 1). DNA was extracted from the environmental samples to
18
compare the microbial diversity recovered by PCR performed directly in lysed cells or after
19
DNA extraction.
20
For the cultures, 2 mL were collected during the stationary growth phase, centrifuged at
21
11 000 rpm for 10 min and 1.8 mL of supernatant removed. For the sorted populations DNA
22
was extracted from 10 (picoeukaryotes) or 2 (nanoeukaryotes) μL of sample. In both cases,
23
genomic DNA was then extracted using Qiagen Blood and Tissue kit. We applied a protocol
24
slightly different from that provided by supplier which yielded higher amounts of DNA. 200
25
μL ATL buffer (Qiagen Blood and Tissue kit) and 20 μL of 20 mg ml-1 lysozyme (Sigma1
26
Aldrich Chimie S.a.r.l. Lyon, France) were added to 200 μL of phytoplankton cultures
27
concentrated by centrifugation. Samples were incubated for 30 min at 37 °C on a shaking
28
incubator. Then, 200 μL of AL buffer (Qiagen Blood and Tissue kit), 75 μL of 20 mg ml-1
29
proteinase K (Sigma-Aldrich Chimie S.a.r.l. Lyon, France) and 20 μL glycogen (Applied
30
Biosystems, Foster city, USA) were added and samples were incubated at 56 °C for 30 min on
31
a shaking incubator. Proteinase K was then inactivated by incubating samples at 75 °C for 10
32
min. 200 μL of absolute ethanol (Fisher Bioblock Scientific, Illkirch, France) were added and
33
samples were transferred into filter columns (Qiagen Blood and Tissue kit) and centrifuged at
34
8 000 rpm for 1 min. Filtrate was then discarded and 500 μl buffer AW1 (Qiagen Blood and
35
Tissue kit) were added to the columns which were then centrifuged again using the same
36
conditions as above. Filtrate was discarded and 500 μL buffer AW2 (Qiagen Blood and
37
Tissue kit) were added to samples. Tubes were then centrifuged for 3 min at 14 000 rpm to
38
remove any trace of ethanol which may otherwise interfere with the following elution step.
39
DNA was then eluted from the filters by adding 100 μL buffer AE (Qiagen Blood and Tissue
40
kit), incubating samples for 3 min and centrifuging them at 14,000 rpm for 1 min.
41
For sorted natural samples which contained much fewer cells than the culture samples, we
42
used the volumes of reagents recommended by the supplier for a given number of cells
43
(Qiagen, Courtaboeuf, France). We added a variable volume of ATL buffer according to the
44
number of cells present. For example for sorts containing 500 cells, we added 4 μL of ATL.
45
Extraction was then carried out as described above but all the reagents were added
46
proportionally to the ATL buffer.
47
For both cultures and environmental samples the amount of DNA extracted was quantified
48
using a Nanodrop ND-1000 Spectrophotometer (Labtech International, France) and quality
49
was then checked on an agarose gel (1.5%).
2
50
For 59 picoeukaryote and 79 nanoeukaryote samples, PCR was performed in triplicate
51
directly on lysed cells: about 2 μL of material containing sorted cells, previously stored at -
52
80°C, were collected immediately after thawing and incubated at 95°C for 5 min.
53
54
3
55
56
Polymerase Chain Reaction (PCR)
One μL of genomic DNA, or 0.5 μL of lysed nanoplankton cells, or 1.5 μL of lysed
57
picoplankton cells were used as template. For the sorted samples these volumes corresponded
58
to at least 500 cells. Template was mixed with 0.5 μL of 10 μM of primer 63f (5’-ACGCTT-
59
GTC-TCA-AAG-ATT-A-3’) labelled with 6-carboxyfluorescein (6-FAM, Eurogentec), 0.5
60
μL of 10 μM of primer 1818r (5’-ACG-GAAACC-TTG-TTA-CGA-3’), 15 μL of HotStar
61
Taq Plus Master Mix Kit (Qiagen, Courtaboeuf, France) and 3 μL of coral load (Qiagen,
62
Courtaboeuf, France). PCR reactions were performed in triplicate with an initial incubation
63
step at 95 °C during 5 min, 35 amplification cycles (95 °C for 1 min, 57°C for 1 min 30 sec,
64
and 72 °C for 1 min 30 sec) and a final elongation step at 72°C for 10 min.
65
66
Multiple Displacement Amplification (MDA)
67
The 18S rRNA gene could not be correctly amplified from 1 sample of sorted
68
picoeukaryotes (ES152, Supplementary Table S2) and 11 samples of sorted nanoeukaryotes
69
(ES041, ES057, ES058, ES059, ES067, ES082, ES090, ES097, ES098, ES099 and ES125,
70
Supplementary Table S3). The DNA of these samples was amplified by Multiple
71
Displacement Amplification (MDA) using Repli-G Qiagen Midi kit (Qiagen, Courtaboeuf,
72
France) as described previously (Gonzalez et al., 2005; Lepère et al., 2011). We took special
73
care to keep the work environment sterile and DNA free. Plastic ware and distilled water was
74
cleaned from any trace of DNA by exposition under a UV lamp (at a distance from the lamp
75
not exceeding 20 cm) for 30 min. Briefly, 1.5 (picoplankton) or 0.5 (nanoplankton) μL of
76
sorted samples containing at least 500 cells were added to microtubes containing 2 volumes of
77
Phosphate Buffered Saline (PBS, Qiagen, Courtaboeuf, France). 3.5 volumes of buffer D2
78
(Qiagen, Courtaboeuf, France) were then added and samples were incubated for 10 min in the
79
ice. A master mix containing 0.5 μL Midi DNA polymerase, 14.5 μL Midi reaction buffer
80
(Qiagen, Courtaboeuf, France) and MQ water up to a final volume of 20 μL was prepared and
4
81
added to samples which were then incubated for 16 h at 30 °C. DNA polymerase was then
82
inactivated at 65 °C for 5 min. The 18S rRNA gene was then amplified from these samples as
83
described above (0.5 μL were used as template).
84
85
Identification of T-RFLP ribotypes
86
The experimental T-RFs from clones and phytoplankton strains isolated during the
87
MALINA cruise were compared with the T-RFs obtained from environmental samples for
88
species identification as described previously (Lepère et al., 2006): a ribotype associated with
89
a clone or a strain was assumed to be present in an environmental sample if the T-RFs
90
generated with the two main enzymes, MnlI [restriction site CCTC(N)7^N] and HhaI
91
(GCG^C) were detected in this sample. Consistent with a previous study (Vigil et al., 2009)
92
MnlI generally yielded higher numbers of T-RFs of a size evenly distributed between 100 and
93
500 bp, compared to HhaI (Supplementary Figure S7) and the results from MnlI were thus
94
used to infer ribotype distribution by calculating the proportion to the total peak area from
95
each T-RF. For ribotypes sharing the same MnlI-specific T-RFs the relative contribution of
96
HhaI-specific T-RFs was used to infer the ribotype composition. Both HhaI and MnlI do not
97
allow discriminating among the different Micromonas clades and between Micromonas and
98
Bathycoccus. We therefore used an additional endonuclease, Hpy188I (TCN^GA), for all the
99
samples where Mamiellales occurred, and the relative contribution of each of the three species
100
(M. pusilla, B. prasinos and M. squamata) to the total MnlI T-RF was inferred using the T-
101
RFs obtained with Hpy188I. This endonuclease allows also to distinguish between the
102
different Micromonas clades (Supplementary Table S7). In our T-RFLP data, we only
103
detected T-RFs specific of Micromonas clade B. Clade B includes Arctic Micromonas as
104
well as other strain from temperate waters (Lovejoy et al., 2007). However, sequences from
105
clade B which are different from the Arctic Micromonas have never been found in clone
5
106
libraries from the Arctic in the present work as well as in previous studies. Therefore this
107
suggests that all the ribotypes of the clade B found in this study are associated with the Arctic
108
Micromonas.
109
110
111
6
112
In silico T-RFs database
113
114
An in silico T-RF database has been constructed by writing a program using the language
115
Lazarus (http://www.lazarus.freepascal.org). A large (≈ 20 000) sequence database
116
comprising most of the sequences from eukaryotic microorganisms was downloaded from
117
GenBank in Mars 2011 Chimeric sequences were removed using KDNAtools (Guillou et al
118
2008). From this database, we selected all the sequences ≥ 500 bp which contained the 63f
119
primer and constructed a T-RF database for the endonucleases HhaI and MnlI. The
120
unidentified T-RFs were then tentatively identified by comparison with the in silico database.
121
The algorithm is shown below:
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
//S is an array of strings loaded from the file: S[0] corresponds to the ID
//of the first sequence from the database, S[1] is the first DNA sequence
//from the database; S[2] = ID of the second sequence from the database,
//S[3] = second DNA sequence from the database etc.
//S.Count is the number of elements in the array (number of sequences in
//the database
for I := 1 to S.Count div 2 - 1 do
begin
//temp contains the sequence string
temp:= S[I*2-1];
//Pos outputs the position of the primer 63f (GTCTCAAAGATT) within the
//sequence. Although the primer 63f amplifies most of the eukaryotes, a
//number of species have mismatches with its sequence. Therefore for the in
//silico T-RF database we selected a region of the primer 63f which is
//constant throughout the eukaryotes: GTCTCAAAGATT
primer:= Pos ('GTCTCAAAGATT', temp);
//PosEx outputs the position of HhaI cutting site (GCGC) within temp but
//after the primer sequence
hhai:= PosEx('GCGC', temp, primer) - primer +9;
//The real 63f primer is thus 6 bp longer than that used here, moreover
//HhaI cuts on the third nucleotide of GCGC, therefore 6+3=9 bp are to be
//added to the HhaI value found. The restriction site of HhaI has a reverse
//complement (GCGC) identical to itself.
mnlia:= PosEx('CCTC', temp, primer) - primer + 17;
//Similarly MnlI cuts 7 bp after the sequence CCTC, therefore 11 bp after
//the start of the cutting site (CCTC): 6+11=17. In this case the
7
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
//restriction site of MnlI difers from its reverse complement, therefore 2
//values (MnlIa and MnlIb) are to be found for the 2 restriction sites
(CCTC and GAGG). //Only the smaller of these values is then to be retained.
mnlib:= PosEx('GAGG', temp, primer) - primer;
//MnlI thus cuts also GAGG, 6 bp before the restriction site, therefore 6//6=0
if (hhai <= 0-primer-12+21) then
hhai:= 9999;
if (mnlia <= 0-primer-12+29) then
mnlia:= 9999;
if (mnlib <= 0-primer) then
mnlib:= 9999;
// If the sequence does not have a HhaIi (or MnlI) restriction site, then
//show 9999
if mnlia < mnlib then
mnli:= mnlia
else
mnli:= mnlib;
// Between the two MnlI restriction sites (CCTC and GAGG) find and retain
//the one forming a shorter fragment
Memo1.Lines.Add(S[I*2-2]+ ', ' + IntToStr(hhai) + ', ' +
IntToStr(mnli));
end;
//Show results.
8
192
Phylogenetic Analyses
193
Partial sequences were analysed using Bioedit software (Hall, 1999). From a total of 314
194
sequences, 11 chimeras were detected using KeyDNAtools (http://keydnatools.com, Guillou
195
et al. 2008) and removed from further analyses. Sequences were then aligned with clustalW2
196
(http://www.ebi.ac.uk/Tools/msa/clustalw2) and grouped into 47 Operational Taxonomic
197
Units (OTUs) based on 99.5 % similarity. We selected this similarity threshold because it
198
allows discriminating Arctic Micromonas from the other genotypes occurring within the clade
199
B of this genus; similarly 99.5 % similarity discriminates some close related genotypes which
200
show different T-RFLP patterns (e.g. Cylindrotheca closterium, Table 2). The almost full
201
length 18S rRNA gene was then sequenced for at least one clone per OTU using the primers
202
63f and 1818r (Lepère et al., 2011). Partial sequences obtained for each of the three primers
203
(63f, 528f, 1818r) were combined to obtain almost full length sequences. Sequences were
204
aligned with a number of reference sequences from GenBank including sequences from
205
MALINA strains for a total of 115 sequences using clustalW2
206
(www.ebi.ac.uk/Tools/msa/clustalw2). Poorly aligned and highly variable regions of the
207
alignment were manually removed and a neighbour-joining (Saitou and Nei, 1987)
208
phylogenetic tree was constructed from 1556 unambiguously aligned nucleotide positions
209
using Geneious software (www.geneious.com). Genetic distance was estimated using
210
Tamura-Nei model and bootstrap values were estimated from 1 000 replicates.
211
From our sorted samples we obtained 15 sequences belonging to Cercozoa, a division
212
which comprises heterotrophic and to a lesser extent photosynthetic (Chloroarachniophyta)
213
microorganisms. We analysed the phylogeny to assess whether these sequences are likely
214
associated to photosynthetic or heterotrophic microorganisms. Sequences were aligned with
215
reference sequences as described above for a total of 32 sequences comprising 1520 bp. The
216
evolutionary history was inferred by using the Maximum Likelihood (ML) method based on
9
217
the Data specific model (Nei and Kumar, 2000) and a ML phylogenetic tree (Figure S6) based
218
on 1000 replicates was constructed using Mega5 (Tamura et al., 2011). The tree was
219
branched with Massisteria marina as outgroup, according to previous studies (Ota and
220
Vaulot, 2011).
221
10
222
Statistical analyses
223
The Spearman rank correlation coefficient (ρ) and the Pearson’s product moment
224
correlation are considered to be good estimators to compare ecological communities
225
(Legendre and Legendre, 1998). We thus applied these coefficients to the T-RFLP inferred
226
microbial compositions of our samples in order to validate our methods and to interpret the
227
spatial variation of our communities. We used the Vegan package (Legendre and Legendre,
228
1998) of the R software (http://www.r-project.org).
229
These coefficients were first calculated to evaluate the variability within our replicates of
230
the T-RF composition found after PCR on cells and that recovered after PCR on DNA. The
231
two techniques were applied on sample ES116 and both HhaI and MnlI endonucleases were
232
used. The same coefficients were calculated to compare the microbial composition inferred
233
by cloning/sequencing and T-RFLP for the (8) samples sorted from Stations 320 and 390.
234
To interpret the nanoplankton community composition of Leg 2b with respect to the
235
environmental conditions, we also applied the Spearman rank correlation coefficient. The
236
environmental conditions used were temperature, salinity, nitrate and Chl-a concentrations as
237
well as total abundances of pico and nanoplankton. Similarly to Figures 5 and 6, the ribotypes
238
recovered more frequently within our T-RFLP chromatogram (Arctic Micromonas, C.
239
socialis, C. cf. neogracile, Mantoniella squamata, Pelagophyceae) are shown individually
240
whereas other ribotypes occurring less frequently have been grouped to higher taxonomic
241
levels (Alveolata, Chaetoceros spp., Chrysochromulina spp., Chrysophyceae,
242
Dictyochophyceae, Other diatoms, Pyramimonas spp.). The Spearman rank correlation
243
coefficient and the Pearson’s product moment correlation provided similar results and only ρ
244
values are used in the present paper.
11
245
The impact of unidentified T-RFs to the overall microbial diversity, and the T-RF richness
246
for surface and DCM waters are presented using box and whisker plots which have been
247
drawn using Sigmaplot (www.sigmaplot.com/products/sigmaplot/sigmaplot-details.php).
248
249
Methodological consideration
250
Comparison of PCR on cells vs. DNA
251
Many studies in microbial ecology are based on filtered samples. In such case PCRs
252
applied on extracted DNA rather than directly on cells. In our case, since cells are
253
concentrated by flow cytometry and sample volumes extremely small, we performed PCR
254
directly on cells. PCR directly on cells is likely to amplify preferentially naked cells or with a
255
thin cell wall rather than cells with strong cell walls. However even in the case of DNA
256
extraction, the genetic diversity recovered from environmental samples varies also according
257
to extraction efficiency as shown for soil, sediment (Carrigg et al 2007) and biofilm
258
associated bacteria (Ferrera et al., 2010).
259
PCR performed directly on cells vs. on extracted DNA was compared on four
260
picoeukaryote (ES106, ES107, ES113 and ES114) and one nanoeukaryote (ES116) samples,
261
all sorted from Station 540 (Table 1). Analysis was performed in triplicate and data were
262
analysed by T-RFLP as described in the Method Section of the main text. The four
263
picoplankton samples were found to be monospecific (Arctic Micromonas) using both
264
techniques, thus we only show results from the nanoeukaryote sample ES116 (Supplementary
265
Table S6, Supplementary Figure S5) analyzed with both HhaI and MnlI restriction enzymes in
266
triplicate.
267
PCR directly on sorted cells yielded 6 ribotypes using HhaI and 14 ribotypes using MnlI
268
whereas PCR on DNA yielded 14 ribotypes for both enzymes (Supplementary figure S5). A
269
highly significant ( ρ> 0.75, p < 0.01) correlation has been found between replicates for PCR
12
270
on cells using both enzymes (Supplementary Table S6). In contrast the correlation for PCR
271
on DNA was not significant indicative of higher variation between replicates.
272
Both HhaI and MnlI digests of PCR products from cells of ES116 revealed a dominance
273
of Chaetoceros cf. neogracile and Pelagophyceae (T-RFs at 397 and 411 for HhaI and 324
274
and 360 for MnlI, respectively, Supplementary Figure S5. See Table 2 for OTU-specific T-
275
RFs). In contrast digests obtained from PCR made on extracted DNA provided a more
276
confused pattern: the relative abundance of Pelagophyceae was highly variable, especially in
277
MnlI digests, Chaetoceros cf. neogracile was not detected at all in the MnlI chromatogram
278
(Supplementary Figure S5). Similarly T-RFs specific of Fragilariopsis/Cylindrotheca where
279
detected with both enzymes for PCR on cells (T-RFs at 389 and 340 bp, for HhaI and MnlI,
280
respectively) but only with HhaI for PCR on DNA.
281
Overall, PCR on DNA yielded a higher microbial diversity than that on cells but more
282
importantly a larger variability among replicates (Supplementary Figure S5, Supplementary
283
Table S6). Therefore performing PCR on cells rather than on DNA appears more appropriate
284
to compare a large number of samples as in the present study.
285
286
13
287
Multiple Displacement Amplification (MDA)
288
The microbial composition assessed before MDA may vary significantly from that found
289
after MDA (Lepère et al., 2011). Sample ES152 (Station 245, 45m), amplified after MDA,
290
was found to be monospecific (Arctic Micromonas) similarly to the other samples sorted from
291
the same stations (below and above, Figure 5). All the other samples analysed after MDA
292
except ES041 corresponded to subpopulations within nanoeukaryotes (In some cases we
293
sorted both the total population of nanoeukaryotes as well subpopulations forming well-
294
defined clusters on cytograms, Supplementary Table S3). In fact, seven of these samples
295
were not used to infer the nanoplankton composition because we had each time one sample
296
corresponding to the total nanoeukaryote population analyzed without MDA. The only
297
samples analysed after MDA and used to infer the nanoeukaryote composition (Figure 6) are
298
from the deep chlorophyll maximum (DCM) of stations 280 (ES041) and 345 (ES097, ES098
299
and ES099, Supplementary Table S3). Although caution is needed when interpreting data
300
from these samples, the results found for these stations after MDA were similar to those found
301
at the nearby stations where MDA was not applied: the DCM of Station 280 was
302
monospecific (C. socialis) and T-RFs of C. socialis dominated the DCM of the nearby coastal
303
stations 170 and 390. The DCM of Station 345 was found to contain the same groups as
304
those found at the nearby stations 220 and 320 (Figure 6).
305
306
14
307
Flow cytometry sorting
308
The two step sorting procedure applied in the present study for flow cytometry sorting
309
(FCS) is more efficient in capturing photosynthetic eukaryotes compared to previous FCS
310
approaches. Only 14 out 303 sequences recovered from our sorting samples might belong to
311
heterotrophic microorganisms. Thirteen of these sequences belong to Cercozoa: 5 sequences
312
group with heterotrophic Cercozoa, mainly from the genera Cryothecomonas, Ebria, and
313
Protaspis (Supplementary Figure S6) whereas the other 8 sequences form a clade which
314
branches next to that of Chlorarachniophyta (100 % bootstrap support) and might belong to a
315
new lineage from this taxonomic group. In contrast, putative heterotrophic microorganisms
316
comprised about one third of all the sequences recovered from the South East Pacific during a
317
previous study based on one step sorting (Shi et al., 2009). In some cases, contamination by
318
heterotrophs has been reported to be low even with a simple one step sorting (Cuvelier et al.,
319
2010). However the flow cytometer used in the latter study (Influx, Cytopeia) is different
320
from that used in the present study as well as in the work of Shi et al (2009) and Marie et al.
321
(2011) which is a BD FACSAria. It is likely that contamination is instrument dependent (for
322
example, it will depend on the triggering parameter).
323
15
324
T-RFLP identification
325
The T-RFs occurring in our environmental samples were mainly identified by comparison
326
with the experimental T-RFs database obtained from our clone library or our phytoplankton
327
strains. Some of the unidentified ribotypes were then tentatively identified using the in silico
328
T-RF database.
329
However, consistent with previous studies (Kaplan and Kitts, 2003), we found a
330
significant drift between the size of the T-RFs measured and that predicted in silico for our
331
clones (Supplementary Table S8). Although a model has been proposed to predict this drift
332
(Bukovska et al., 2010), there is still a non-negligible difference between the predicted drift
333
and the real drift (Figure 2 in Bukovska et al., 2010) such that we included a ± 4 bp variability
334
to each T-RF present in our in silico database. As a consequence some ribotypes found
335
within our environmental samples were related to two or more genetically distant species
336
from the in silico T-RFs database and could not be identified.
337
Some T-RFs could not be identified because they did not correspond to any of the T-RFs
338
obtained from both the experimental and the in silico database, for the nanoplankton (39 T-
339
RFs with MnlI and 14 with HhaI) and to a lesser extent picoplankton (8 T-RFs with MnlI and
340
11 with HhaI). The unidentified T-RFs are either associated with T-RFLP artefacts or with
341
unknown microbes.
342
Overall 25 over 52 (HhaI) and 47 over 85 (MnlI) T-RFs could not be identified for
343
nanoplankton. Picoplankton samples were far less diverse than nanoplankton and yielded 7
344
over 18 and 11 over 21 unidentified T-RFs for HhaI and MnlI, respectively. It should be
345
noted that all the picoplankton T-RFs (both identified and unidentified) which are not related
346
to Arctic Micromonas occurred only in 22 out of 59 samples. Moreover the contribution of
347
Arctic Micromonas to the total peak area exceeded 75 % for 11 of these 22 samples. T-RFs
348
not related to Arctic Micromonas accounted thus for a very minor proportion of the
16
349
picoplankton community and the unidentified peaks accounted for ≥ 20 % of the total peak
350
area only for 6 picoplankton samples (Supplementary Figure S8).
351
Almost half of the T-RFs detected could not be identified using both enzymes, for
352
nanoplankton but unidentified T-RFs accounted for a low proportion of the overall peak area
353
for most of the samples: most (75 %) of samples contained 0 to 2 and 0 to 4 unidentified T-
354
RFs out of 2 to 7 and 2 to 9 total T-RFs for HhaI and MnlI, respectively. The latter values
355
suggest a significant impact, in terms of richness, of unknown to total T-RFs for the enzyme
356
MnlI. However unidentified peaks contributed to less than 25 % of the total peak area in 91
357
% of the samples (Supplementary Figure S8).
358
359
360
361
362
363
17
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
References for the Supplemental material
Bukovska P, Jelinkova M, Hrselova H, Sykorova Z, Gryndler M (2010). Terminal restriction
fragment length measurement errors are affected mainly by fragment length, G plus C
nucleotide content and secondary structure melting point. Journal of Microbiological
Methods 82: 223-228.
Cuvelier ML, Allen AE, Monier A, McCrow JP, Messie M, Tringe SG et al (2010). Targeted
metagenomics and ecology of globally important uncultured eukaryotic phytoplankton.
Proceedings of the National Academy of Sciences of the United States of America 107:
14679-14684.
del Giorgio P, Bird DF, Prairie YT, Planas D (1996). Flow cytometric determination of
bacterial abundance in lake plankton with the green nucleic acid stain SYTO 13. Limnology
and Oceanography 41: 783-789.
Ferrera I, Massana R, Balague V, Pedros-Alio C, Sanchez O, Mas J (2010). Evaluation of
DNA extraction methods from complex phototrophic biofilms. Biofouling 26: 349-357.
Gonzalez JM, Portillo MC, Saiz-Jimenez C (2005). Multiple displacement amplification as a
pre-polymerase chain reaction (pre-PCR) to process difficult to amplify samples and low copy
number sequences from natural environments. Environmental Microbiology 7: 1024-1028.
Hall TA (1999). BioEdit: a user-friendly biological sequence alignement editor and analysis
program for Windows 95/98/NT. Nucleic Acid symposium Series 41: 95-98.
Kaplan CW, Kitts CL (2003). Variation between observed and true Terminal Restriction
Fragment length is dependent on true TRF length and purine content. Journal of
Microbiological Methods 54: 121-125.
Legendre P, Legendre L (1998). Numerical ecology, 20. Elsevier: New York.
Lepère C, Boucher D, Jardillier L, Domaizon I, Debroas D (2006). Succession and regulation
factors of small eukaryote community. composition in a lacustrine ecosystem (Lake pavin).
Applied and Environmental Microbiology 72: 2971-2981.
Lepère C, Demura M, Kawachi M, Romac S, Probert I, Vaulot D (2011). Whole Genome
Amplification (WGA) of marine photosynthetic eukaryote populations. Fems Microbiology
Ecology.
Lovejoy C, Vincent WF, Bonilla S, Roy S, Martineau MJ, Terrado R et al (2007).
Distribution, phylogeny, and growth of cold-adapted picoprasinophytes in arctic seas. Journal
of Phycology 43: 78-89.
Nei M, Kumar S (2000). Molecular Evolution and Phylogenetics. Oxford University Press:
New York.
Ota S, Vaulot D (2011). Lotharella reticulosa sp. nov.: A Highly Reticulated Network
Forming Chlorarachniophyte from the Mediterranean Sea. Protist.
18
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
Saitou N, Nei M (1987). The neighbor-joining method: a new method for reconstructing
phylogenetic trees. . Molecular Biology and Evolution 4: 406-425.
Shi XL, Marie D, Jardillier L, Scanlan DJ, Vaulot D (2009). Groups without cultured
representatives dominate eukaryotic picophytoplankton in the oligotrophic South East Pacific
Ocean. Plos One 4: e7657.
Tamura K, Peterson D, Peterson N, Stecher G, Nei M, Kumar S (2011). MEGA5: Molecular
Evolutionary Genetics Analysis Using Maximum Likelihood, Evolutionary Distance, and
Maximum Parsimony Methods. Molecular Biology and Evolution.
Vigil P, Countway PD, Rose J, Lonsdale DJ, Gobler CJ, Caron DA (2009). Rapid shifts in
dominant taxa among microbial eukaryotes in estuarine ecosystems. Aquatic Microbial
Ecology 54: 83-100.
19