bs_bs_banner
Environmental Microbiology (2014) 16(10), 3095–3114
doi:10.1111/1462-2920.12346
The paradox of marine heterotrophic nitrogen fixation: abundances of heterotrophic diazotrophs do not account for nitrogen fixation rates in the Eastern Tropical South Pacific
Kendra A. Turk-Kubo,1* Muskan Karamchandani,1 Douglas G. Capone2 and Jonathan P. Zehr1 1 Ocean Sciences Department, University of California, Santa Cruz, CA, USA. 2 Department of Biological Sciences, University of Southern California, CA, USA. Summary Results of recent modelling efforts imply denitrification-influenced waters, such as those in the Eastern Tropical South Pacific (ETSP), may support high rates of biological nitrogen fixation (BNF), yet little is known about the N2-fixing microbial community in this region. Our characterization of the ETSP diazotrophic community along a gradient from upwelling-influenced to oligotrophic waters did not detect cyanobacterial diazotrophs commonly found in other open ocean regions. Most of the nifH genes amplified by polymerase chain reaction (PCR) from DNA and RNA samples clustered with γ-proteobacterial nifH sequences, although a novel Trichodesmium phylotype was also recovered. Three quantitative PCR assays were developed to target γ-proteobacterial phylotypes, but all were found to be present at low abundances. An analysis of the expected BNF rates based on abundances and plausible cell-specific N2 fixation rates indicates that these γ-proteobacteria are unlikely to be responsible for previously reported BNF rates from corresponding samples. Therefore, the organisms responsible for the measured BNF rates remain poorly understood. Furthermore, there is little direct evidence, at this time, to support the hypothesis that heterotrophic N2 fixation contributes significantly to oceanic BNF rates based on our analysis of heterotrophic cell-specific N2 fixation rates required Received 14 July, 2013; accepted 23 November, 2013. *For correspondence. E-mail [email protected]; Tel. (+1) 831 459 3128; Fax (+1) 831 459 4882.
© 2013 Society for Applied Microbiology and John Wiley & Sons Ltd
to explain BNF rates reported in previously published studies. Introduction Biological nitrogen fixation (BNF) supplies a valuable source of reduced N to open ocean marine ecosystems (Karl et al., 2002) and is a metabolic capability restricted to a subset of micro-organisms within the bacteria and archaea. Although there are relatively few direct measurements of BNF rates in the South Pacific, rates in ultra-oligotrophic waters of the South Pacific Gyre (SPG) are among the lowest measured in oligotrophic marine waters (Raimbault and Garcia, 2008; Halm et al., 2012; Dekaezemacker et al., 2013). These low rates are assumed to be limited by the low dissolved Fe concentrations in surface waters because of the low influx of Fe from continental sources (Raimbault and Garcia, 2008; Sohm et al., 2011a). Nonetheless, in the centre of the SPG, BNF does appear to support a significant portion of new production, defined as phytoplankton production supported by allochthonous N sources (Dugdale and Goering, 1967; Raimbault and Garcia, 2008). A recent model-based estimate of the distribution of global marine BNF rates indicates that BNF may occur at significant rates in surface waters near denitrification zones [e.g. waters that are down-current of persistent oxygen minimum zones (OMZs; Deutsch et al., 2007)], suggesting that regions such as the Eastern Tropical South Pacific (ETSP), the eastern region of the SPG, may be a more significant source for fixed N than previously thought. Detectable rates of BNF have been reported in the Chilean upwelling zone, a coastal region in the eastern portion of the South Pacific, despite the high concentrations of nitrate in these waters that supports the majority of primary production (Raimbault and Garcia, 2008; Fernandez et al., 2011). BNF has also been reported in other upwelling regions in the presence of non-limiting N concentrations (Hamersley et al., 2011; Sohm et al., 2011b). This has led to the speculation that
3096 K. Turk-Kubo, M. Karamchandani, D. Capone and J. Zehr the regulation of marine BNF may not be as tightly coupled to the presence or absence of fixed N as once thought (Sohm et al., 2011a). The cyanobacterial diazotrophs that are generally abundant and active in other tropical and subtropical oligotrophic regions, such as filamentous Trichodesmium spp. (Capone et al., 1997; Church et al., 2008; Goebel et al., 2010), heterocyst-forming cyanobacterial symbionts of diatoms such as Richelia spp. (Church et al., 2005b; Foster et al., 2007), and unicellular cyanobacterial diazotrophs Crocosphaera spp. (UCYN-B) or group A (UCYN-A or Candidatus Atelocyanobacterium thalassa; Moisander et al., 2010), appear to be absent altogether or present at very low abundances in the eastern region of the South Pacific (Bonnet et al., 2008). High abundances (up to ∼3 × 105 nifH copies L−1) of an UCYN-A-like phylotype, as well as sporadic detection of their nifH transcripts, have been reported in the western region of the SPG (Halm et al., 2012). It is speculated that the sporadic detection of these cosmopolitan cyanobacterial diazotrophs in the SPG is a direct result of low dissolved Fe concentrations (Sohm et al., 2011a). However, recent studies in the Chilean upwelling zone (Farnelid et al., 2011; Fernandez et al., 2011) and in the Western SPG (Halm et al., 2012) have reported diverse diazotrophic communities dominated by organisms other than cyanobacteria using polymerase chain reaction (PCR)-based methods. The recovery of these sequences, as well as the sporadic quantification of high abundances of specific heterotrophic phylotypes in the Western SPG [e.g. abundances as high as ∼8 × 107 nifH copies L−1 for the Gamma 1 phylotype (Halm et al., 2012)], has led to the speculation that BNF is carried out by heterotrophic diazotrophs in the South Pacific. In the ETSP, there is evidence that nitrogen fixation is carried out by heterotrophic diazotrophs and regulated by the availability of dissolved carbon (Dekaezemacker et al., 2013). However, it remains an open question to what extent these heterotrophic communities contribute to BNF in the South Pacific or to the N budget of the oligotrophic marine environment in general. The main focus of this research effort was to characterize diazotrophic diversity along a cruise track in the ETSP that transited from coastal, upwelling-influenced to oligotrophic conditions (Fig. 1A) and to correlate diazotroph abundances to bulk in situ BNF rates measured in a parallel study (Dekaezemacker et al., 2013). We amplified and sequenced the nifH gene from DNA and RNA samples across this gradient from cruises occurring in two consecutive years, then used a quantitative PCR (qPCR)-based approach to quantify abundances of the dominant diazotrophs in this region. This enabled us to directly compare abundances of dominant diazotrophic
phylotypes to bulk BNF rates measured in complementary samples reported by Dekaezemacker and colleagues (2013). Results Diazotrophic diversity in the ETSP based on partial nifH sequences The biogeochemical conditions and BNF rates measured during cruises AT1561 (El Nino conditions; February– March 2010) and MV1104 (La Nina conditions; March– April 2011) are discussed in detail in a companion study of Dekaezemacker and colleagues (2013). Briefly, low but measurable rates of BNF were reported in the photic zone along this cruise track at all but one station on the southern transect. The highest measured BNF rates (ca. −1 0.8 nmol N L day−1) occurred during AT1561 at the southern transect Stn. 5 and were stimulated by the addition of nitrate and dissolved Fe (dFe), but not dFe alone. Relatively low BNF rates were measured along the southern transect during the following year (MV1104) despite conditions appearing more favourable for BNF, including higher relative concentrations of dFe. During both cruises, BNF in the high-nutrient, low-chlorophyll waters of the northern transect were low (between 0 and ca. −1 0.5 nmol N L day−1) and appeared to be limited by dFe (Dekaezemacker et al., 2013). The diversity of diazotrophs in the ETSP at the time of sampling was characterized by amplifying a fragment of the nifH gene from DNA and RNA extracts using a highly degenerate nested PCR assay (Zehr and McReynolds, 1989; Zani et al., 2000). Successful amplification of nifH sequences from both cruises was infrequent and successful in only 28 out of the ca. 100 samples screened, despite screening both 0.2–10 μm and > 10 μm size fractions from four depths from each of the 11 stations. All samples were tested for inhibition of the PCR reaction, and no inhibition was detected. In all, a total of 432 sequences were recovered from 28 samples and nifH was successfully amplified from both size fractions in some samples. A majority of the sequences were recovered from the 10°S transect during both cruises (Stns. 7, 9 and 11; Fig. 1A), where surface waters exhibited highnutrient, low-chlorophyll conditions because of equatorial upwelling (Dekaezemacker et al., 2013). Very few sequences were recovered from Stns. 1 and 3 from either cruise, and many of the sequences recovered from Stn. 1 were closely related to sequences amplified from sampling blanks taken from MV1104 milliQ DNA extracts (> 92% amino acid similarity; represented by ETSP_ DNA_43976A3, KF151439); thus, it is possible these sequences are contaminants (Fig. 1B and C). The contaminant phylotype, also recovered from AT1561 Stn. 7 samples, falls into nifH cluster 1K and is closely related
© 2013 Society for Applied Microbiology and John Wiley & Sons Ltd, Environmental Microbiology, 16, 3095–3114
N2-fixing potential of heterotrophs in the ETSP AT1561 (2010)
EQ
MV1104 (2011)
7
10S
B
Ocean Data View
A
10
9
8
87
3097
uncult. marine bact. (ADV51808.1) ETSP_DNA_43970A33 (KF151552)
Teredinibacter turnerae (YP_003073074.1) uncult. marine bact. (ABV00660.1)
69
Pseudomonas stutzeri (AEA83267.1)
74
11
ETSPcDNA_45274A8 (KF151844)
80
1G
uncult. marine bact. (ADO20625.1) ETSP_DNA_43598A59 (KF151495)
uncult. marine bact. (ADO20648.1)
6
90
ETSP2_50254A42 (KF151711) 96 UCYN−A (YP_0034241696.1)
Arica, Chile
20S 5
3
4
2
1 50
98 ETSP_DNA_43977A48 (KF151446) Trichodesmium erythraeum (YP_723617.1) 84
100W
90W
ETSPcDNA_44164A73 (KF151808) uncult. marine bact. (AEH77211.1)
76
80W
70W
1B
ETSP2_50283A4 (KF151468) uncult. marine bact. (AEA49666.1)
84
ETSP_DNA_44879A45 (KF151450)
No. nifH clones recovered from each station AT1561 (2010) 1
3
5
52
ETSP_DNA_43970A33 ETSP_cDNA_45274A8
7
1
2
26
MV1104 (2011) 9
11
38
77
10
29
3
7
9
21 1
6
ETSP_DNA_43598A59
5
38
11
mQ blank
C
96
98 85
15
1
ETSP_DNA_43131A49
2
ETSP_DNA_43976A3*
4
1
1 1
uncult. marine bact. (ADA57556.1)
1
ETSP_DNA_43131A3 (KF151417) uncult. marine bact. (ACI13876.1) 99
1
1
1
3
1
19
70
11
1
ETSP_DNA_43381A40
11
ETSP2_50253A6
2
12
2
33
99
54
ETSP_DNA_43977A37 (KF151587)
uncult. marine bact. (AEI72359.1) uncult. marine bact. (AAZ06737.1)
4
ETSP_DNA_43977A23
Σ
79
4
ETSP_DNA_44261A9
3
uncult. bact. (AAO67691.1)
12
ETSP_DNA_43977A37
2
ETSP2_50253A6 (KF151699)
61 ETSP_DNA_43962A3 (KF151523) ETSP_DNA_44261A9 (KF151610)
4
ETSP_DNA_43962A3
1A
Desulfovibrio sp. (ZP_06367518.1) uncult. marine bact. (AAL83734.1)
5
ETSP_DNA_43977A40 ETSP_DNA_43131A3
1
93 ETSP_DNA_43977A40 (KF151590)
1P
Geobacter sp. (YP_002538026.1) 11
ETSP2_50283A4 ETSP_DNA_44879A45
ETSP_DNA_43381A40 (KF151482)
uncult. marine bact. (ABQ50813.1)
93
2
1K
Bradyrhizobium sp. (BA167069.1) ETSP_DNA_43976A3* (KF151439)
uncult. marine bact. (ADO20599.1) 81
2 3 1
ETSP_DNA_43131A49 (KF151422)
53
1
ETSP2_50254A42 ETSP_cDNA_44164A73 ETSP_DNA_43977A48
93 Burkholderia ferrariae (ABO64209.1)
111
3
40
2
33
28
11
84 ETSP_DNA_43977A23 (KF151583)
0.1
Fig. 1. Station map from AT1561 and MV1104 cruise tracks, and nifH sequencing results from DNA and RNA samples. A. Parallel molecular sampling and BNF rate determination were conducted throughout the photic zone at stations along the displayed cruise track during both 2010 (AT1561) and 2011 (MV1104). Clone libraries for partial nifH genes were generated from odd-numbered stations, and qPCR analyses were conducted on all stations during AT1561 and odd-numbered stations between 3 and 11 during MV1104 (more details in Supporting information Table S1). B. Maximum likelihood tree of partial nifH amino acid sequences (91 positions) containing representative phylotypes from DNA and RNA samples. ETSP nifH amino acid sequences were clustered using CD-HIT at 92% amino acid identity to determine representative sequences. Branch lengths were determined using the JTT matrix-based model. The percentage of calculated trees in which sequences clustered together in the bootstrap test (1000 replicates) is shown next to the branch when greater than 50%. Phylotypes recovered from both DNA and RNA are marked with an open square; the phylotype recovered exclusively from RNA is marked with a closed square. The representative sequence marked with an asterisk (*) clusters with sequences recovered from milliQ water blanks thus is a possible contaminant. nifH cluster designations are at the right of the tree and use the convention established in Zehr and colleagues (2003a). C. Distribution of recovered sequences across stations, size fractions and cruises. Sequences represent clusters with > 92% amino acid similarity. mQ, milliQ sample collection blank.
(97% nucleotide similarity) to environmental sequences isolated from the abyssopelagic Sargasso Sea (e.g. DQ481260; Hewson et al., 2007), and in complementary DNA (cDNA) from both the North Pacific Subtropical Gyre (DQ105652; Zehr et al., 2007) and the Arabian Sea (e.g. JX064484; Bird and Wyman, 2012). Only three cyanobacterial phylotypes were recovered, represented by ETSP_DNA_43977A48 (KF151446), ETSP_cDNA_44164A73 (KF151808) and ETSP2_ 50283A4 (KF151468), which accounted for only 4% of the
total number of sequences. ETSP2_50283A4 is closely related to Trichodesmium erythraeum but was only obtained from Stn. 9 during MV1104 (Figs 1B and C, and 2). ETSP_cDNA_44164A73, recovered only from RNA samples at Stn. 7 during AT1561, is closely related to Candidatus A. thalassa, which has been detected in the eastern SPG using qPCR assays (Bonnet et al., 2008) and in metatranscriptomes (Thompson et al., 2012). As Candidatus A. thalassa was undetected in every sample based on qPCR analyses (discussed later), it had to be
© 2013 Society for Applied Microbiology and John Wiley & Sons Ltd, Environmental Microbiology, 16, 3095–3114
3098 K. Turk-Kubo, M. Karamchandani, D. Capone and J. Zehr
(0,3-6,0)a (0-1,2,0-1)b
91-95% nt similarity to Cand. A. thalassa
ETSPcDNA_44164A73 (KF151808) ETSPcDNA_44164A89 (KF151809) uncultured marine microorganism (EU187510) 99 uncultured marine microorganism (EU187529) UCYN-A Cand. A. thalassa (NC_013771) ETSPcDNA_44164A65 (KF151807) uncultured bacterium (AY974262) uncultured cyanobacterium (EU159545) unidentified marine bacterial clone HT1902 (AF299420) uncultured bacterium (HM210392) UCYN-A uncultured bacterium (HM210387) 92 uncultured bacterium (HM210391) uncultured bacterium (HM210399) 97 uncultured bacterium (HM210401) uncultured bacterium (HM210402) uncultured bacterium (HM210400) uncultured bacterium (HM210389) uncultured bacterium (HM210398) 63 uncultured bacterium (HQ611838) 98 UCYN-C (DQ273169) cyanobacterium endosymbiont of Rhopalodia gibba (AY728387) uncultured bacterium (EU594205) uncultured microorganism (EF174733) 96 uncultured nitrogen−fixing bacterium (JF826472) Myxosarcina sp. (U73133) 72 Xenococcus sp. PCC 7305 (U73135) 54 ETSP_DNA_44262A74 (KF151448) 66 75 ETSP_DNA_43977A48 (KF151446) uncultured marine bacterium (EU807709) 99 uncultured Richelia sp. (DQ225753) Cyanothece sp. PCC 7822 (NZ_ABVE01000002) 71 UCYN-B Crocosphaera watsonii WH 8501 (NZ_AADV02000024) uncultured marine bacterium (EU052677) uncultured marine bacterium (AY896403) 97 uncultured marine bacterium (AY896316) 99 Trichodesmium thiebautii (U23507) Trichodesmium erythraeum IMS101 (CP000393) ETSP2_50283A44 (KF151469) 72 ETSP2_50283A76 (KF151472) Trichodesmium 66 ETSP2_50283A28 (KF151466) 98 ETSP2_50283A36 (KF151467) ETSP2_50283A92 (KF151474) ETSP2_50283A52 (KF151470) 56 ETSP2_50283A60 (KF151471) ETSP2_50283A20 (KF151465) ETSP2_50283A12 (KF151464) ETSP2_50283A84 (KF151473) Nostoc punctiforme PCC 73102 (NC_010628) Anabaena variabilis ATCC 29413 (NC_007413)
55
'Nostoc azollae' 0708 (NZ_ACIR01000100) Synechococcus sp. PCC 7335 (NZ_DS989904) Cyanothece sp. PCC 7425 (NC_011884)
0.1
Fig. 2. Maximum likelihood tree of partial nifH nucleotide sequences affiliated with Cyanobacteria recovered from the ETSP and the SPG. Cyanobacterial nifH sequences recovered from this study (in bold) as well as the UCYN-A-like sequences recovered from the SPG (Halm et al., 2012), are displayed along with cultivated cyanobacterial representatives and their associated GenBank accession numbers (in parenthesis). The phylogenetic tree was built based on 291 nucleotide positions of the nifH gene and the Tamura-Nai model was used to determine branch lengths. The percentage of calculated trees in which sequences clustered together in the bootstrap test (1000 replicates) is shown next to the branch when greater than 50%. For the Trichodesmium-like phylotypes, the number of mismatches in the forward primer, probe, and reverse primer to the Church and colleagues (2005a) assay (A) and the Langlois and colleagues (2008) assay (B) are given in parenthesis.
present at Stn. 7 at abundances below detection limits of the qPCR assay or < 33 nifH copies L−1. ETSP_DNA_ 43977A48, clustered with group II unicellular cyanobacteria (e.g. Myxosarcina and Xenococcus; Figs 1B and 2) and has not previously been reported in this region of the ocean. The majority of sequences recovered from this study (75%) fall within nifH cluster 1G (Zehr et al., 2003a), a subcluster of nifH cluster I (Chien and Zinder, 1996), which includes cultivated isolates from the γ-proteobacteria including Pseudomonas stutzeri, Teridinibacter turnerae and Azotobacter vinelandii, as well as many environmentally derived uncultivated nifH
sequences from diverse environments. When clustered at 92% amino acid similarity, these 1G sequences resolve into only four phylotypes (Fig. 1B and C) with a majority represented by only two phylotypes, ETSPcDNA_ 45274A8 (KF151844) and ETSP_DNA_43970A33 (KF151552). However, three major distinct clusters are resolved from nucleic acid sequence analysis, which have been designated γETSP1–3 (Fig. 3). Both γETSP2 and γETSP3 are comprised of two subclusters (Fig. 3) and were recovered from both DNA and RNA samples during this study, while γETSP1 was solely recovered from DNA samples. Three new qPCR assays using Taqman® chemistry were designed to target each of these three clusters
© 2013 Society for Applied Microbiology and John Wiley & Sons Ltd, Environmental Microbiology, 16, 3095–3114
N2-fixing potential of heterotrophs in the ETSP
85
γETSP2a (72)
γETSP2
85 61 96
γETSP4 (14) 99
γETSP2b (22)
96
ETSP_DNA_43132A44 (KF151479) ETSP2_50332A75 (KF151756) uncultured bacterium (HM210386/ADO20625) 95 “Gamma 3” uncultured bacterium (HM210397/ADO20636) ETSP2_50358A18 (KF151762) Pseudomonas stutzeri DSM 4166 (CP002622) ETSP2_50254A42 (KF151711) 99 ETSP2_50254A75 (KF151719) uncultured bacterium (HM210409/ADO20648)
87
96
99
100
γETSP6 (8) incl. “Gamma 4” unclultured bacterium (HM210363)
ETSP_DNA_43966A69 (KF151531) ETSP_DNA_43966A91 (KF151548) ETSP_DNA_43977A38 (KF151588)
59
99
γETSP5 (20)
Teredinibacter turnerae T7901 (CP001614) ETSP_DNA_44879A41 (KF151656) uncultured bacterium (EF631902/ABV00660) 88
99
90
γETSP3b (18) γETSP3a (51)
γETSP3
ETSP_DNA_43369A67 (KF151481)
98
γETSP1 (104)
γETSP1
3099
Fig. 3. Maximum likelihood tree of partial nifH nucleotide γ-proteobacterial (1G) sequences from DNA and RNA samples collected during AT1561 and MV1104 cruises. Phylogenetic tree was built based on 291 nucleotide positions of the nifH gene and the Tamura-Nai model was used to determine branch lengths. The percentage of calculated trees in which sequences clustered together in the bootstrap test (1000 replicates) is shown next to the branch when greater than 50%. Phylotypes recovered from both DNA and RNA are marked with a shaded square. qPCR assays using Taqman chemistry were designed using Primer3 (primer3.sourceforge.net/) to specifically target the three most abundant clusters, γETSP1, γETSP2 and γETSP3. The numbers of sequences in each collapsed cluster are in parenthesis beside the cluster name. The placement of γ-proteobacterial phylotypes Gamma 3 and Gamma 4 described in Halm and colleagues (2012) are indicated in the tree. Sequences used as reference sequences in Fig. 1B have the protein accession number along with the nucleotide accession number in parenthesis.
incl. uncultured bacterium (HQ611628/ADV51808)
0.05
(Table 1), assuming that their relative abundances in clone libraries reflected their abundances in the environment. Sequences were also recovered from nifH clusters 1P, 1A, 1K, 2 and 3, but in sum, these sequences only account for 20% of the sequences recovered, and of these, almost half are closely related to the 1K sequence recovered from milliQ blanks (> 92% amino acid similarity; Fig. 1C). Although methodologies used in generating clone libraries were optimized to reduce contamination, the presence of a sequence alone does not guarantee that its source is from the sampled environment (see discussions in Turk et al., 2011; Bombar et al., 2013), even when sampling, extraction and PCR amplification blanks are included. ETSP samples may be particularly subject to contamination, as amplification of reagent contaminants has been shown to be responsible for higher percentages of clone libraries in environmental samples with low abundances of target organisms (Zehr et al., 2003b). Quantifying the abundance and nifH expression of diazotrophic phylotypes A subset of AT1561 DNA extracts (approximately four depths per station), as well as pooled DNA extracts from MV1104 were screened using qPCR assays targeting eight diazotrophs that have been described from other oligotrophic regions. Detailed in Supporting information Table S1, these assays targeted the uncultivated unicel-
lular cyanobacterial groups A (Candidatus A. thalassa, or UCYN-A; Church et al., 2005a), B (UCYN-B; Moisander et al., 2010), C (UCYN-C; Foster et al., 2007), Trichodesmium spp. (Church et al., 2005a), the diatomassociated heterocyst-forming groups het-1 (Richelia in Rhizosolenia; Church et al., 2005b), het-2 (Richelia in Hemiaulus, Foster et al., 2007), het-3 (Calothrix in Chaetocerous; Foster et al., 2007), as well as two proteobacterial groups γ-24774A11 and α-24809A06 (Moisander et al., 2008). DNA extracts from both cruises were negative for these phylotypes, with the exception of two MV1104 stations along the 10°S transect where Trichodesmium was present at low abundances (‘detected not quantified’ (DNQ) at Stn. 7, and 6.0 × 102 +/− 5.3 × 101 nifH copies L−1 at Stn. 9; Supporting information Table S1). It is important to note, however, that the Trichodesmium sequences recovered from MV1104 Stn. 9 have between three and four mismatches in the probe annealing region (Fig. 2); therefore, it is unclear whether low abundances of the targeted phylotype were sporadically present or whether these findings resulted from cross-reactivity with the relatively high abundances of the ETSP Trichodesmium phylotype. A smaller subset of cDNA generated from AT1561 RNA extracts screened for UCYN-A, Trichodesmium and γ-24774A11 confirmed that no nifH transcripts were present from these phylotypes. Based on clone libraries constructed from DNA and RNA extracts, dominated by 1G sequences assumed to be γ-proteobacteria, three new qPCR assays were
© 2013 Society for Applied Microbiology and John Wiley & Sons Ltd, Environmental Microbiology, 16, 3095–3114
γETSP3
γETSP2
Shaded cells indicate how well each assay targets the clusters closely related to each target sequence. Within each of these larger clusters, the number of sequences with 0, 1 or 2 mismatches (in bold) in the region targeted by each primer/probe are included in parenthesis.
GTTCTCACGAATGGGCATGG 0 (96), 1 (8) CATACGCACCCTCTTCTTCAAGG 0 (69), 1 (7) 0 (21), 1 (1) GATTACACCGCGACCAGCAC 0 (50), 1 (1) 0 (17), 1 (1) TGGGTGACGTGGTATGTGGTGGATTCG 0 (99), 5 (1) TTGGATGTGCCGGTCGCG GTG TTA 0 (69), 1 (4), 2 (3) 0 (22) GGGCTACGGCGACATCAAGTG CG 0 (50), 1 (1) 2 (18) GAGCCAGGTGTTGGTTGTGC 0 (104) AGGCACGGTTGAGGATCTCG 0 (75), 1 (1) 0 (21), 1 (1) TCATGGAAATGGCTGCTGAAG 0 (50), 1 (1) 1 (17), 2 (1) ETSP_DNA_43970A49 (KF151567) γETSP1 (104) ETSPcDNA_44943A46 (KF151819) γETSP2a (76) γETSP2b (22) ETSP_DNA_44879A48 (KF151661) γETSP3a (51) γETSP3b (18) γETSP1
Probe (5′-3′) Forward primer (5′-3′) Desired target (Genbank Accession No.) qPCR assay name
Table 1. New qPCR assays designed to target the most highly recovered γ-proteobacterial phylotypes from ETSP clone libraries.
Reverse primer (5′-3′)
3100 K. Turk-Kubo, M. Karamchandani, D. Capone and J. Zehr developed (Table 1). The γETSP1 assay targets a 1G cluster containing 104 sequences (24% of all sequences recovered) characterized exclusively from DNA samples. γETSP2 (98 sequences) and γETSP3 (69 sequences) assays target 1G clusters, which both contain sequences that were recovered from both DNA and RNA extracts (Fig. 3). Collectively these assays target over half of the phylotypes recovered in clone libraries. All cross-reactivity testing confirmed that each assay reliably targeted only the desired cluster. Despite being the most prevalent phylotype in clone libraries, γETSP1 was only sporadically detected in DNA extracts from the small size fraction during AT1561 along the 10°S transit, at abundances too low to quantify (Fig. 4A, Table S1), and was not detected in any RNA extracts. γETSP1 was not detected in DNA samples from MV1104. Although γETSP1 was not detected in sampling (milliQ) or DNA extraction blanks, and it has not been reported as a contaminant in other studies, the presence of γETSP1 in the clone libraries cannot be ruled out as a contaminant. The γETSP2 phylotype, detected in DNA extracts from AT1561 Stns. 3 and 5, as well as from MV1104 Stn. 5, appears to be associated with low-nutrient, lowchlorophyll conditions (Dekaezemacker et al., 2013). The highest measured abundance of this phylotype occurred at MV1104 Stn. 5 (80 m) but was still quite low at 4.2 × 102 nifH copies L−1 (Fig. 4A). This phylotype was also actively regulating the nif operon throughout the photic zone at Stn. 5 during AT1561, as evidenced by nifH expression quantified in both size fractions at 10, 140 and 150 m depths (Fig. 4B). Furthermore, nifH transcripts were quantified in both size fractions throughout the Stn. 5 AT1561 diel experiment with a peak of nifH transcription at noon, although the overall expression was relatively low (Fig. 4C). For each sample from Stn. 5 during AT1561 (whether depth or time), on average, 27% of the total γETSP2 nifH transcripts were recovered from the > 10 μm size fraction, which suggests that this phylotype may exist in a delicate association with larger particles or organisms that are disrupted upon filtering. γETSP3 was detected at low abundances in surface waters during AT1561 at Stns. 7 (DNQ) and 9 (4.2 × 102 nifH copies L−1; Fig. 4A) and was also detected throughout the photic zone at Stn. 10 (Supporting information Table S2). However, no γETSP3 nifH expression in either size-fraction was detected at any depth at Stns. 7, 9 or 11 in the noon samples, indicating that it was not actively expressing the nifH gene at the time of sampling. γETSP3 was not detected in any DNA samples from MV1104. There is an obvious discrepancy between the phylotypes that are dominant in clone libraries and the
© 2013 Society for Applied Microbiology and John Wiley & Sons Ltd, Environmental Microbiology, 16, 3095–3114
N2-fixing potential of heterotrophs in the ETSP
3101
Fig. 4. Photic zone abundances of gamma proteobacterial phylotypes γETSP1, γETSP2 and γETSP3 determined using qPCR at representative stations and γETSP2 nifH expression at Stn. 5 during AT1561. A. DNA samples recovered from the small size fraction (between 0.2 and 10 μm) were analysed for the abundances of ETSP-specific phylotypes. Many of the samples yielded values that were above detection limits, but beneath the limit of quantitation for these assays, thus determined to be DNQ. The grey box in each plot indicates this range of values between 13 and 75 nifH copies L−1. Sample collection blanks (shipboard milliQ water filtered according to sample filtration techniques) and DNA extraction blanks were negative for all three phylotypes. B. γETSP2 nifH expression depth profile throughout photic zone at Stn. 5 during AT1561. RNA samples recovered from both size fractions [between 0.2 and 10 μm (‘0.2 μm’) and greater than 10 μm (‘10 μm’)] at Stn. 5 were analysed for γETSP2. RNA samples were taken at local noon. No γETSP2 amplification was detected in controls for cDNA generation with no reverse transcriptase. C. γETSP2 diel nifH expression in surface waters at Stn. 5 during AT1561. Surface water, collected from a trace metal clean pump, was incubated in acid-cleaned cubitainers and subsampled at ∼6 h intervals over a 24-h period. No γETSP2 amplification was detected in controls for cDNA generation with no reverse transcriptase.
actual abundances determined using qPCR techniques, as evidenced by the relative numbers of γETSP1 sequences recovered from the AT1561 10°S samples (Fig. 1B) even though they were never present at abundances high enough to quantify (Fig. 4A). This result has also been previously reported in other studies using the
same primer set for marine environmental samples (Hewson et al., 2007; Turk et al., 2011; Farnelid et al., 2013). The nifH degenerate primer set used in this study has an amplification bias heavily weighted towards γ-proteobacteria (Turk et al., 2011) for reasons not wellunderstood. These findings underscore the importance of
© 2013 Society for Applied Microbiology and John Wiley & Sons Ltd, Environmental Microbiology, 16, 3095–3114
3102 K. Turk-Kubo, M. Karamchandani, D. Capone and J. Zehr cautious interpretation of relative abundances in clone libraries originating from PCR products. Comparison of nifH phylogenetic signatures within the South Pacific Ocean and between major ocean basins Use of nifH as a gene marker for the study of diazotroph diversity, abundance and distribution in diverse terrestrial and marine environments have been widespread (see reviews by Riemann et al., 2010, and Gaby and Buckley, 2011). By March 2013, the number of nifH sequences publically available in the GenBank nr/nt database, after curation using a pipeline for the import of these sequences into the ARB software environment (http://pmc.ucsc.edu/∼wwwzehr/research/database/), had grown to over 33 700 sequences (> 34 000 including sequences from this study), 10 073 of which were isolated from marine sources. In characterizing the nifH-based phylogenetic signature within the South Pacific and between different oceanic regions, a subset of the total reported marine sequences were chosen from studies that utilized molecular techniques similar to this study. In total, 3193 sequences (represented by 418 clusters at 97% nucleotide similarity) from 16 studies were selected that represented five major oceanic regions: the North Pacific, South Pacific, South China Sea, North Atlantic and Indian Ocean. An unweighted pairwise Unifrac P-test indicates that South Pacific diazotrophic communities are marginally significantly different (P ≤ 0.3; Fig. 5A) between the three existing studies (this study; Fernandez et al., 2011; and Halm et al., 2012), and an unweighted Unifrac lineagespecific analysis (LSA) identified three lineages that explain the potential differences (Fig. 5B). The LSA results indicated that cluster 1K sequences (primarily α- and β-proteobacteria) are over-represented in this study and in that of Fernandez and colleagues (2011), while being under-represented in Halm and colleagues (2012) (Fig. 5C). The other two lineages are 1P clusters (comprised primarily of β-proteobacteria) exclusively reported in Halm and colleagues (2012) originally reported as putative γ-proteobacteria and targeted by qPCR assays ‘Gamma 1’ and ‘Gamma 2’ (Fig. 5B and C). Although the pairwise unweighted Unifrac P-test conducted between all ocean basins resulted in a much higher P value (P ≤ 0.1; Fig. 5D), indicating that the difference among basins is of ‘suggestive’ significance, there were several nifH clusters that were either significantly overor under-represented in certain ocean basins based on the unweighted Unifrac LSA. Cluster 1B sequences (cyanobacteria) are over-represented in the North Atlantic and Pacific but are underrepresented in the South China Sea, Indian Ocean and the South Pacific (Fig. 5F). Cluster 1K and 1P sequences are over-represented in the South
Pacific, which is consistent with the LSA resulting from South Pacific studies described earlier. In the Indian Ocean, sequences within nifH cluster 1K are also overrepresented. The South China Sea is characterized not only by an under-representation of cyanobacterial sequences but also an over-representation of nifH cluster 1A sequences, which is comprised of non-sulphatereducing δ-proteobacterial genera. This subset of all marine-derived nifH sequences was also used to determine which heterotrophic phylotypes have been reported in multiple oceanic regions and studies, thus might be less likely to be contaminant sequences and may represent promising targets for future research. Of the 418 total clusters, 39 contain sequences reported from more than a single ocean basin, 28 of which do not cluster with Cyanobacterial nifH cluster 1B (Table 2). Many of the targets for both cyanobacterial and non-cyanobacterial qPCR assays that have been developed are contained within these clusters (Table 2 and Supporting information Table S2). Among the heterotrophic phylotypes reported, a majority are classified as nifH cluster 1G (primarily γ-proteobacteria) and cluster 1K (primarily α- and β-proteobacteria). Representatives are also present from nifH clusters 1E (Paenibacillus spp.), 1J (α-proteobacteria), as well as cluster 3 (e.g. Chlorobium spp. and Desulfovibrio spp. and others) (Table 2). Discussion South Pacific diazotroph communities It is evident that diverse populations of α,β-, γ-proteobacteria are present in the South Pacific, and thus far, there exists little evidence that cyanobacterial diazotrophs are abundant. Between this study and the study conducted by Halm and colleagues (2012), a total of seven new qPCR assays have been designed, chosen because of their frequent recovery in DNA- and RNAbased clone libraries, none of which target the same phylotypes. Although there is no overlap in sampling regions between these two studies, this finding is in contrast with the distribution of cyanobacterial diazotrophic phylotypes, where the same phylotype can be found distributed throughout entire ocean gyres [e.g. Trichodesmium in the Tropical North Atlantic (Goebel et al., 2010)]. Furthermore, very few similar sequence types (within 97% nucleotide similarity) were recovered from all sequencing efforts that have been conducted in the South Pacific thus far (this study; Fernandez et al., 2011; and Halm et al., 2012; Fig. 5B). Out of the four major groups identified by Halm and colleagues (2012), only the Gamma 4 phylotype was recovered in this study (γETSP6; Figs. 3 and 5B). Associated with nifH cluster 1G (Zehr et al., 2003a) and most
© 2013 Society for Applied Microbiology and John Wiley & Sons Ltd, Environmental Microbiology, 16, 3095–3114
N2-fixing potential of heterotrophs in the ETSP A
D This Study This Study
Halm et al., 2012 Fernandez et al., 2011
---
P ≤ 0.03
P ≤ 0.03
---
P ≤ 0.03
Halm et al., 2012 Fernandez et al., 2011
South Pacific
North Pacific
---
≤ 0.1 ---
South Pacific North Pacific South China Sea North Atlantic Indian Ocean
---
South North China Sea Atlantic ≤ 0.1 ≤ 0.1 ---
≤ 0.1 ≤ 0.1 ≤ 0.1 ---
3103
Indian Ocean ≤ 0.1 ≤ 0.1 ≤ 0.1 ≤ 0.1 ---
E
G2
G1
B
CI
II
76 54 32 1
43 21
A
3
1B
1B
1P
3
1A 1P
γ1
1G
1G
1K
1K
G3
G4
γ2
γ3
1 2 3 4
This Study Halm et al., 2012 Fernandez et al., 2011 highly significant (P < 0.001; Unifrac LSA) significant (0.01 < P < 0.05; Unifrac LSA)
C Halm et al., Fernandez et al., node This Study 2012 2011
P
1K
20 > 7
1 < 14
19 > 18
1.5E-07
1P.1
0 6
0 30 3 > 4 1P 19 >7 0>1 1A.1 03 0>2 16 > 5 0
Environmental Microbiology (2014) 16(10), 3095–3114
doi:10.1111/1462-2920.12346
The paradox of marine heterotrophic nitrogen fixation: abundances of heterotrophic diazotrophs do not account for nitrogen fixation rates in the Eastern Tropical South Pacific
Kendra A. Turk-Kubo,1* Muskan Karamchandani,1 Douglas G. Capone2 and Jonathan P. Zehr1 1 Ocean Sciences Department, University of California, Santa Cruz, CA, USA. 2 Department of Biological Sciences, University of Southern California, CA, USA. Summary Results of recent modelling efforts imply denitrification-influenced waters, such as those in the Eastern Tropical South Pacific (ETSP), may support high rates of biological nitrogen fixation (BNF), yet little is known about the N2-fixing microbial community in this region. Our characterization of the ETSP diazotrophic community along a gradient from upwelling-influenced to oligotrophic waters did not detect cyanobacterial diazotrophs commonly found in other open ocean regions. Most of the nifH genes amplified by polymerase chain reaction (PCR) from DNA and RNA samples clustered with γ-proteobacterial nifH sequences, although a novel Trichodesmium phylotype was also recovered. Three quantitative PCR assays were developed to target γ-proteobacterial phylotypes, but all were found to be present at low abundances. An analysis of the expected BNF rates based on abundances and plausible cell-specific N2 fixation rates indicates that these γ-proteobacteria are unlikely to be responsible for previously reported BNF rates from corresponding samples. Therefore, the organisms responsible for the measured BNF rates remain poorly understood. Furthermore, there is little direct evidence, at this time, to support the hypothesis that heterotrophic N2 fixation contributes significantly to oceanic BNF rates based on our analysis of heterotrophic cell-specific N2 fixation rates required Received 14 July, 2013; accepted 23 November, 2013. *For correspondence. E-mail [email protected]; Tel. (+1) 831 459 3128; Fax (+1) 831 459 4882.
© 2013 Society for Applied Microbiology and John Wiley & Sons Ltd
to explain BNF rates reported in previously published studies. Introduction Biological nitrogen fixation (BNF) supplies a valuable source of reduced N to open ocean marine ecosystems (Karl et al., 2002) and is a metabolic capability restricted to a subset of micro-organisms within the bacteria and archaea. Although there are relatively few direct measurements of BNF rates in the South Pacific, rates in ultra-oligotrophic waters of the South Pacific Gyre (SPG) are among the lowest measured in oligotrophic marine waters (Raimbault and Garcia, 2008; Halm et al., 2012; Dekaezemacker et al., 2013). These low rates are assumed to be limited by the low dissolved Fe concentrations in surface waters because of the low influx of Fe from continental sources (Raimbault and Garcia, 2008; Sohm et al., 2011a). Nonetheless, in the centre of the SPG, BNF does appear to support a significant portion of new production, defined as phytoplankton production supported by allochthonous N sources (Dugdale and Goering, 1967; Raimbault and Garcia, 2008). A recent model-based estimate of the distribution of global marine BNF rates indicates that BNF may occur at significant rates in surface waters near denitrification zones [e.g. waters that are down-current of persistent oxygen minimum zones (OMZs; Deutsch et al., 2007)], suggesting that regions such as the Eastern Tropical South Pacific (ETSP), the eastern region of the SPG, may be a more significant source for fixed N than previously thought. Detectable rates of BNF have been reported in the Chilean upwelling zone, a coastal region in the eastern portion of the South Pacific, despite the high concentrations of nitrate in these waters that supports the majority of primary production (Raimbault and Garcia, 2008; Fernandez et al., 2011). BNF has also been reported in other upwelling regions in the presence of non-limiting N concentrations (Hamersley et al., 2011; Sohm et al., 2011b). This has led to the speculation that
3096 K. Turk-Kubo, M. Karamchandani, D. Capone and J. Zehr the regulation of marine BNF may not be as tightly coupled to the presence or absence of fixed N as once thought (Sohm et al., 2011a). The cyanobacterial diazotrophs that are generally abundant and active in other tropical and subtropical oligotrophic regions, such as filamentous Trichodesmium spp. (Capone et al., 1997; Church et al., 2008; Goebel et al., 2010), heterocyst-forming cyanobacterial symbionts of diatoms such as Richelia spp. (Church et al., 2005b; Foster et al., 2007), and unicellular cyanobacterial diazotrophs Crocosphaera spp. (UCYN-B) or group A (UCYN-A or Candidatus Atelocyanobacterium thalassa; Moisander et al., 2010), appear to be absent altogether or present at very low abundances in the eastern region of the South Pacific (Bonnet et al., 2008). High abundances (up to ∼3 × 105 nifH copies L−1) of an UCYN-A-like phylotype, as well as sporadic detection of their nifH transcripts, have been reported in the western region of the SPG (Halm et al., 2012). It is speculated that the sporadic detection of these cosmopolitan cyanobacterial diazotrophs in the SPG is a direct result of low dissolved Fe concentrations (Sohm et al., 2011a). However, recent studies in the Chilean upwelling zone (Farnelid et al., 2011; Fernandez et al., 2011) and in the Western SPG (Halm et al., 2012) have reported diverse diazotrophic communities dominated by organisms other than cyanobacteria using polymerase chain reaction (PCR)-based methods. The recovery of these sequences, as well as the sporadic quantification of high abundances of specific heterotrophic phylotypes in the Western SPG [e.g. abundances as high as ∼8 × 107 nifH copies L−1 for the Gamma 1 phylotype (Halm et al., 2012)], has led to the speculation that BNF is carried out by heterotrophic diazotrophs in the South Pacific. In the ETSP, there is evidence that nitrogen fixation is carried out by heterotrophic diazotrophs and regulated by the availability of dissolved carbon (Dekaezemacker et al., 2013). However, it remains an open question to what extent these heterotrophic communities contribute to BNF in the South Pacific or to the N budget of the oligotrophic marine environment in general. The main focus of this research effort was to characterize diazotrophic diversity along a cruise track in the ETSP that transited from coastal, upwelling-influenced to oligotrophic conditions (Fig. 1A) and to correlate diazotroph abundances to bulk in situ BNF rates measured in a parallel study (Dekaezemacker et al., 2013). We amplified and sequenced the nifH gene from DNA and RNA samples across this gradient from cruises occurring in two consecutive years, then used a quantitative PCR (qPCR)-based approach to quantify abundances of the dominant diazotrophs in this region. This enabled us to directly compare abundances of dominant diazotrophic
phylotypes to bulk BNF rates measured in complementary samples reported by Dekaezemacker and colleagues (2013). Results Diazotrophic diversity in the ETSP based on partial nifH sequences The biogeochemical conditions and BNF rates measured during cruises AT1561 (El Nino conditions; February– March 2010) and MV1104 (La Nina conditions; March– April 2011) are discussed in detail in a companion study of Dekaezemacker and colleagues (2013). Briefly, low but measurable rates of BNF were reported in the photic zone along this cruise track at all but one station on the southern transect. The highest measured BNF rates (ca. −1 0.8 nmol N L day−1) occurred during AT1561 at the southern transect Stn. 5 and were stimulated by the addition of nitrate and dissolved Fe (dFe), but not dFe alone. Relatively low BNF rates were measured along the southern transect during the following year (MV1104) despite conditions appearing more favourable for BNF, including higher relative concentrations of dFe. During both cruises, BNF in the high-nutrient, low-chlorophyll waters of the northern transect were low (between 0 and ca. −1 0.5 nmol N L day−1) and appeared to be limited by dFe (Dekaezemacker et al., 2013). The diversity of diazotrophs in the ETSP at the time of sampling was characterized by amplifying a fragment of the nifH gene from DNA and RNA extracts using a highly degenerate nested PCR assay (Zehr and McReynolds, 1989; Zani et al., 2000). Successful amplification of nifH sequences from both cruises was infrequent and successful in only 28 out of the ca. 100 samples screened, despite screening both 0.2–10 μm and > 10 μm size fractions from four depths from each of the 11 stations. All samples were tested for inhibition of the PCR reaction, and no inhibition was detected. In all, a total of 432 sequences were recovered from 28 samples and nifH was successfully amplified from both size fractions in some samples. A majority of the sequences were recovered from the 10°S transect during both cruises (Stns. 7, 9 and 11; Fig. 1A), where surface waters exhibited highnutrient, low-chlorophyll conditions because of equatorial upwelling (Dekaezemacker et al., 2013). Very few sequences were recovered from Stns. 1 and 3 from either cruise, and many of the sequences recovered from Stn. 1 were closely related to sequences amplified from sampling blanks taken from MV1104 milliQ DNA extracts (> 92% amino acid similarity; represented by ETSP_ DNA_43976A3, KF151439); thus, it is possible these sequences are contaminants (Fig. 1B and C). The contaminant phylotype, also recovered from AT1561 Stn. 7 samples, falls into nifH cluster 1K and is closely related
© 2013 Society for Applied Microbiology and John Wiley & Sons Ltd, Environmental Microbiology, 16, 3095–3114
N2-fixing potential of heterotrophs in the ETSP AT1561 (2010)
EQ
MV1104 (2011)
7
10S
B
Ocean Data View
A
10
9
8
87
3097
uncult. marine bact. (ADV51808.1) ETSP_DNA_43970A33 (KF151552)
Teredinibacter turnerae (YP_003073074.1) uncult. marine bact. (ABV00660.1)
69
Pseudomonas stutzeri (AEA83267.1)
74
11
ETSPcDNA_45274A8 (KF151844)
80
1G
uncult. marine bact. (ADO20625.1) ETSP_DNA_43598A59 (KF151495)
uncult. marine bact. (ADO20648.1)
6
90
ETSP2_50254A42 (KF151711) 96 UCYN−A (YP_0034241696.1)
Arica, Chile
20S 5
3
4
2
1 50
98 ETSP_DNA_43977A48 (KF151446) Trichodesmium erythraeum (YP_723617.1) 84
100W
90W
ETSPcDNA_44164A73 (KF151808) uncult. marine bact. (AEH77211.1)
76
80W
70W
1B
ETSP2_50283A4 (KF151468) uncult. marine bact. (AEA49666.1)
84
ETSP_DNA_44879A45 (KF151450)
No. nifH clones recovered from each station AT1561 (2010) 1
3
5
52
ETSP_DNA_43970A33 ETSP_cDNA_45274A8
7
1
2
26
MV1104 (2011) 9
11
38
77
10
29
3
7
9
21 1
6
ETSP_DNA_43598A59
5
38
11
mQ blank
C
96
98 85
15
1
ETSP_DNA_43131A49
2
ETSP_DNA_43976A3*
4
1
1 1
uncult. marine bact. (ADA57556.1)
1
ETSP_DNA_43131A3 (KF151417) uncult. marine bact. (ACI13876.1) 99
1
1
1
3
1
19
70
11
1
ETSP_DNA_43381A40
11
ETSP2_50253A6
2
12
2
33
99
54
ETSP_DNA_43977A37 (KF151587)
uncult. marine bact. (AEI72359.1) uncult. marine bact. (AAZ06737.1)
4
ETSP_DNA_43977A23
Σ
79
4
ETSP_DNA_44261A9
3
uncult. bact. (AAO67691.1)
12
ETSP_DNA_43977A37
2
ETSP2_50253A6 (KF151699)
61 ETSP_DNA_43962A3 (KF151523) ETSP_DNA_44261A9 (KF151610)
4
ETSP_DNA_43962A3
1A
Desulfovibrio sp. (ZP_06367518.1) uncult. marine bact. (AAL83734.1)
5
ETSP_DNA_43977A40 ETSP_DNA_43131A3
1
93 ETSP_DNA_43977A40 (KF151590)
1P
Geobacter sp. (YP_002538026.1) 11
ETSP2_50283A4 ETSP_DNA_44879A45
ETSP_DNA_43381A40 (KF151482)
uncult. marine bact. (ABQ50813.1)
93
2
1K
Bradyrhizobium sp. (BA167069.1) ETSP_DNA_43976A3* (KF151439)
uncult. marine bact. (ADO20599.1) 81
2 3 1
ETSP_DNA_43131A49 (KF151422)
53
1
ETSP2_50254A42 ETSP_cDNA_44164A73 ETSP_DNA_43977A48
93 Burkholderia ferrariae (ABO64209.1)
111
3
40
2
33
28
11
84 ETSP_DNA_43977A23 (KF151583)
0.1
Fig. 1. Station map from AT1561 and MV1104 cruise tracks, and nifH sequencing results from DNA and RNA samples. A. Parallel molecular sampling and BNF rate determination were conducted throughout the photic zone at stations along the displayed cruise track during both 2010 (AT1561) and 2011 (MV1104). Clone libraries for partial nifH genes were generated from odd-numbered stations, and qPCR analyses were conducted on all stations during AT1561 and odd-numbered stations between 3 and 11 during MV1104 (more details in Supporting information Table S1). B. Maximum likelihood tree of partial nifH amino acid sequences (91 positions) containing representative phylotypes from DNA and RNA samples. ETSP nifH amino acid sequences were clustered using CD-HIT at 92% amino acid identity to determine representative sequences. Branch lengths were determined using the JTT matrix-based model. The percentage of calculated trees in which sequences clustered together in the bootstrap test (1000 replicates) is shown next to the branch when greater than 50%. Phylotypes recovered from both DNA and RNA are marked with an open square; the phylotype recovered exclusively from RNA is marked with a closed square. The representative sequence marked with an asterisk (*) clusters with sequences recovered from milliQ water blanks thus is a possible contaminant. nifH cluster designations are at the right of the tree and use the convention established in Zehr and colleagues (2003a). C. Distribution of recovered sequences across stations, size fractions and cruises. Sequences represent clusters with > 92% amino acid similarity. mQ, milliQ sample collection blank.
(97% nucleotide similarity) to environmental sequences isolated from the abyssopelagic Sargasso Sea (e.g. DQ481260; Hewson et al., 2007), and in complementary DNA (cDNA) from both the North Pacific Subtropical Gyre (DQ105652; Zehr et al., 2007) and the Arabian Sea (e.g. JX064484; Bird and Wyman, 2012). Only three cyanobacterial phylotypes were recovered, represented by ETSP_DNA_43977A48 (KF151446), ETSP_cDNA_44164A73 (KF151808) and ETSP2_ 50283A4 (KF151468), which accounted for only 4% of the
total number of sequences. ETSP2_50283A4 is closely related to Trichodesmium erythraeum but was only obtained from Stn. 9 during MV1104 (Figs 1B and C, and 2). ETSP_cDNA_44164A73, recovered only from RNA samples at Stn. 7 during AT1561, is closely related to Candidatus A. thalassa, which has been detected in the eastern SPG using qPCR assays (Bonnet et al., 2008) and in metatranscriptomes (Thompson et al., 2012). As Candidatus A. thalassa was undetected in every sample based on qPCR analyses (discussed later), it had to be
© 2013 Society for Applied Microbiology and John Wiley & Sons Ltd, Environmental Microbiology, 16, 3095–3114
3098 K. Turk-Kubo, M. Karamchandani, D. Capone and J. Zehr
(0,3-6,0)a (0-1,2,0-1)b
91-95% nt similarity to Cand. A. thalassa
ETSPcDNA_44164A73 (KF151808) ETSPcDNA_44164A89 (KF151809) uncultured marine microorganism (EU187510) 99 uncultured marine microorganism (EU187529) UCYN-A Cand. A. thalassa (NC_013771) ETSPcDNA_44164A65 (KF151807) uncultured bacterium (AY974262) uncultured cyanobacterium (EU159545) unidentified marine bacterial clone HT1902 (AF299420) uncultured bacterium (HM210392) UCYN-A uncultured bacterium (HM210387) 92 uncultured bacterium (HM210391) uncultured bacterium (HM210399) 97 uncultured bacterium (HM210401) uncultured bacterium (HM210402) uncultured bacterium (HM210400) uncultured bacterium (HM210389) uncultured bacterium (HM210398) 63 uncultured bacterium (HQ611838) 98 UCYN-C (DQ273169) cyanobacterium endosymbiont of Rhopalodia gibba (AY728387) uncultured bacterium (EU594205) uncultured microorganism (EF174733) 96 uncultured nitrogen−fixing bacterium (JF826472) Myxosarcina sp. (U73133) 72 Xenococcus sp. PCC 7305 (U73135) 54 ETSP_DNA_44262A74 (KF151448) 66 75 ETSP_DNA_43977A48 (KF151446) uncultured marine bacterium (EU807709) 99 uncultured Richelia sp. (DQ225753) Cyanothece sp. PCC 7822 (NZ_ABVE01000002) 71 UCYN-B Crocosphaera watsonii WH 8501 (NZ_AADV02000024) uncultured marine bacterium (EU052677) uncultured marine bacterium (AY896403) 97 uncultured marine bacterium (AY896316) 99 Trichodesmium thiebautii (U23507) Trichodesmium erythraeum IMS101 (CP000393) ETSP2_50283A44 (KF151469) 72 ETSP2_50283A76 (KF151472) Trichodesmium 66 ETSP2_50283A28 (KF151466) 98 ETSP2_50283A36 (KF151467) ETSP2_50283A92 (KF151474) ETSP2_50283A52 (KF151470) 56 ETSP2_50283A60 (KF151471) ETSP2_50283A20 (KF151465) ETSP2_50283A12 (KF151464) ETSP2_50283A84 (KF151473) Nostoc punctiforme PCC 73102 (NC_010628) Anabaena variabilis ATCC 29413 (NC_007413)
55
'Nostoc azollae' 0708 (NZ_ACIR01000100) Synechococcus sp. PCC 7335 (NZ_DS989904) Cyanothece sp. PCC 7425 (NC_011884)
0.1
Fig. 2. Maximum likelihood tree of partial nifH nucleotide sequences affiliated with Cyanobacteria recovered from the ETSP and the SPG. Cyanobacterial nifH sequences recovered from this study (in bold) as well as the UCYN-A-like sequences recovered from the SPG (Halm et al., 2012), are displayed along with cultivated cyanobacterial representatives and their associated GenBank accession numbers (in parenthesis). The phylogenetic tree was built based on 291 nucleotide positions of the nifH gene and the Tamura-Nai model was used to determine branch lengths. The percentage of calculated trees in which sequences clustered together in the bootstrap test (1000 replicates) is shown next to the branch when greater than 50%. For the Trichodesmium-like phylotypes, the number of mismatches in the forward primer, probe, and reverse primer to the Church and colleagues (2005a) assay (A) and the Langlois and colleagues (2008) assay (B) are given in parenthesis.
present at Stn. 7 at abundances below detection limits of the qPCR assay or < 33 nifH copies L−1. ETSP_DNA_ 43977A48, clustered with group II unicellular cyanobacteria (e.g. Myxosarcina and Xenococcus; Figs 1B and 2) and has not previously been reported in this region of the ocean. The majority of sequences recovered from this study (75%) fall within nifH cluster 1G (Zehr et al., 2003a), a subcluster of nifH cluster I (Chien and Zinder, 1996), which includes cultivated isolates from the γ-proteobacteria including Pseudomonas stutzeri, Teridinibacter turnerae and Azotobacter vinelandii, as well as many environmentally derived uncultivated nifH
sequences from diverse environments. When clustered at 92% amino acid similarity, these 1G sequences resolve into only four phylotypes (Fig. 1B and C) with a majority represented by only two phylotypes, ETSPcDNA_ 45274A8 (KF151844) and ETSP_DNA_43970A33 (KF151552). However, three major distinct clusters are resolved from nucleic acid sequence analysis, which have been designated γETSP1–3 (Fig. 3). Both γETSP2 and γETSP3 are comprised of two subclusters (Fig. 3) and were recovered from both DNA and RNA samples during this study, while γETSP1 was solely recovered from DNA samples. Three new qPCR assays using Taqman® chemistry were designed to target each of these three clusters
© 2013 Society for Applied Microbiology and John Wiley & Sons Ltd, Environmental Microbiology, 16, 3095–3114
N2-fixing potential of heterotrophs in the ETSP
85
γETSP2a (72)
γETSP2
85 61 96
γETSP4 (14) 99
γETSP2b (22)
96
ETSP_DNA_43132A44 (KF151479) ETSP2_50332A75 (KF151756) uncultured bacterium (HM210386/ADO20625) 95 “Gamma 3” uncultured bacterium (HM210397/ADO20636) ETSP2_50358A18 (KF151762) Pseudomonas stutzeri DSM 4166 (CP002622) ETSP2_50254A42 (KF151711) 99 ETSP2_50254A75 (KF151719) uncultured bacterium (HM210409/ADO20648)
87
96
99
100
γETSP6 (8) incl. “Gamma 4” unclultured bacterium (HM210363)
ETSP_DNA_43966A69 (KF151531) ETSP_DNA_43966A91 (KF151548) ETSP_DNA_43977A38 (KF151588)
59
99
γETSP5 (20)
Teredinibacter turnerae T7901 (CP001614) ETSP_DNA_44879A41 (KF151656) uncultured bacterium (EF631902/ABV00660) 88
99
90
γETSP3b (18) γETSP3a (51)
γETSP3
ETSP_DNA_43369A67 (KF151481)
98
γETSP1 (104)
γETSP1
3099
Fig. 3. Maximum likelihood tree of partial nifH nucleotide γ-proteobacterial (1G) sequences from DNA and RNA samples collected during AT1561 and MV1104 cruises. Phylogenetic tree was built based on 291 nucleotide positions of the nifH gene and the Tamura-Nai model was used to determine branch lengths. The percentage of calculated trees in which sequences clustered together in the bootstrap test (1000 replicates) is shown next to the branch when greater than 50%. Phylotypes recovered from both DNA and RNA are marked with a shaded square. qPCR assays using Taqman chemistry were designed using Primer3 (primer3.sourceforge.net/) to specifically target the three most abundant clusters, γETSP1, γETSP2 and γETSP3. The numbers of sequences in each collapsed cluster are in parenthesis beside the cluster name. The placement of γ-proteobacterial phylotypes Gamma 3 and Gamma 4 described in Halm and colleagues (2012) are indicated in the tree. Sequences used as reference sequences in Fig. 1B have the protein accession number along with the nucleotide accession number in parenthesis.
incl. uncultured bacterium (HQ611628/ADV51808)
0.05
(Table 1), assuming that their relative abundances in clone libraries reflected their abundances in the environment. Sequences were also recovered from nifH clusters 1P, 1A, 1K, 2 and 3, but in sum, these sequences only account for 20% of the sequences recovered, and of these, almost half are closely related to the 1K sequence recovered from milliQ blanks (> 92% amino acid similarity; Fig. 1C). Although methodologies used in generating clone libraries were optimized to reduce contamination, the presence of a sequence alone does not guarantee that its source is from the sampled environment (see discussions in Turk et al., 2011; Bombar et al., 2013), even when sampling, extraction and PCR amplification blanks are included. ETSP samples may be particularly subject to contamination, as amplification of reagent contaminants has been shown to be responsible for higher percentages of clone libraries in environmental samples with low abundances of target organisms (Zehr et al., 2003b). Quantifying the abundance and nifH expression of diazotrophic phylotypes A subset of AT1561 DNA extracts (approximately four depths per station), as well as pooled DNA extracts from MV1104 were screened using qPCR assays targeting eight diazotrophs that have been described from other oligotrophic regions. Detailed in Supporting information Table S1, these assays targeted the uncultivated unicel-
lular cyanobacterial groups A (Candidatus A. thalassa, or UCYN-A; Church et al., 2005a), B (UCYN-B; Moisander et al., 2010), C (UCYN-C; Foster et al., 2007), Trichodesmium spp. (Church et al., 2005a), the diatomassociated heterocyst-forming groups het-1 (Richelia in Rhizosolenia; Church et al., 2005b), het-2 (Richelia in Hemiaulus, Foster et al., 2007), het-3 (Calothrix in Chaetocerous; Foster et al., 2007), as well as two proteobacterial groups γ-24774A11 and α-24809A06 (Moisander et al., 2008). DNA extracts from both cruises were negative for these phylotypes, with the exception of two MV1104 stations along the 10°S transect where Trichodesmium was present at low abundances (‘detected not quantified’ (DNQ) at Stn. 7, and 6.0 × 102 +/− 5.3 × 101 nifH copies L−1 at Stn. 9; Supporting information Table S1). It is important to note, however, that the Trichodesmium sequences recovered from MV1104 Stn. 9 have between three and four mismatches in the probe annealing region (Fig. 2); therefore, it is unclear whether low abundances of the targeted phylotype were sporadically present or whether these findings resulted from cross-reactivity with the relatively high abundances of the ETSP Trichodesmium phylotype. A smaller subset of cDNA generated from AT1561 RNA extracts screened for UCYN-A, Trichodesmium and γ-24774A11 confirmed that no nifH transcripts were present from these phylotypes. Based on clone libraries constructed from DNA and RNA extracts, dominated by 1G sequences assumed to be γ-proteobacteria, three new qPCR assays were
© 2013 Society for Applied Microbiology and John Wiley & Sons Ltd, Environmental Microbiology, 16, 3095–3114
γETSP3
γETSP2
Shaded cells indicate how well each assay targets the clusters closely related to each target sequence. Within each of these larger clusters, the number of sequences with 0, 1 or 2 mismatches (in bold) in the region targeted by each primer/probe are included in parenthesis.
GTTCTCACGAATGGGCATGG 0 (96), 1 (8) CATACGCACCCTCTTCTTCAAGG 0 (69), 1 (7) 0 (21), 1 (1) GATTACACCGCGACCAGCAC 0 (50), 1 (1) 0 (17), 1 (1) TGGGTGACGTGGTATGTGGTGGATTCG 0 (99), 5 (1) TTGGATGTGCCGGTCGCG GTG TTA 0 (69), 1 (4), 2 (3) 0 (22) GGGCTACGGCGACATCAAGTG CG 0 (50), 1 (1) 2 (18) GAGCCAGGTGTTGGTTGTGC 0 (104) AGGCACGGTTGAGGATCTCG 0 (75), 1 (1) 0 (21), 1 (1) TCATGGAAATGGCTGCTGAAG 0 (50), 1 (1) 1 (17), 2 (1) ETSP_DNA_43970A49 (KF151567) γETSP1 (104) ETSPcDNA_44943A46 (KF151819) γETSP2a (76) γETSP2b (22) ETSP_DNA_44879A48 (KF151661) γETSP3a (51) γETSP3b (18) γETSP1
Probe (5′-3′) Forward primer (5′-3′) Desired target (Genbank Accession No.) qPCR assay name
Table 1. New qPCR assays designed to target the most highly recovered γ-proteobacterial phylotypes from ETSP clone libraries.
Reverse primer (5′-3′)
3100 K. Turk-Kubo, M. Karamchandani, D. Capone and J. Zehr developed (Table 1). The γETSP1 assay targets a 1G cluster containing 104 sequences (24% of all sequences recovered) characterized exclusively from DNA samples. γETSP2 (98 sequences) and γETSP3 (69 sequences) assays target 1G clusters, which both contain sequences that were recovered from both DNA and RNA extracts (Fig. 3). Collectively these assays target over half of the phylotypes recovered in clone libraries. All cross-reactivity testing confirmed that each assay reliably targeted only the desired cluster. Despite being the most prevalent phylotype in clone libraries, γETSP1 was only sporadically detected in DNA extracts from the small size fraction during AT1561 along the 10°S transit, at abundances too low to quantify (Fig. 4A, Table S1), and was not detected in any RNA extracts. γETSP1 was not detected in DNA samples from MV1104. Although γETSP1 was not detected in sampling (milliQ) or DNA extraction blanks, and it has not been reported as a contaminant in other studies, the presence of γETSP1 in the clone libraries cannot be ruled out as a contaminant. The γETSP2 phylotype, detected in DNA extracts from AT1561 Stns. 3 and 5, as well as from MV1104 Stn. 5, appears to be associated with low-nutrient, lowchlorophyll conditions (Dekaezemacker et al., 2013). The highest measured abundance of this phylotype occurred at MV1104 Stn. 5 (80 m) but was still quite low at 4.2 × 102 nifH copies L−1 (Fig. 4A). This phylotype was also actively regulating the nif operon throughout the photic zone at Stn. 5 during AT1561, as evidenced by nifH expression quantified in both size fractions at 10, 140 and 150 m depths (Fig. 4B). Furthermore, nifH transcripts were quantified in both size fractions throughout the Stn. 5 AT1561 diel experiment with a peak of nifH transcription at noon, although the overall expression was relatively low (Fig. 4C). For each sample from Stn. 5 during AT1561 (whether depth or time), on average, 27% of the total γETSP2 nifH transcripts were recovered from the > 10 μm size fraction, which suggests that this phylotype may exist in a delicate association with larger particles or organisms that are disrupted upon filtering. γETSP3 was detected at low abundances in surface waters during AT1561 at Stns. 7 (DNQ) and 9 (4.2 × 102 nifH copies L−1; Fig. 4A) and was also detected throughout the photic zone at Stn. 10 (Supporting information Table S2). However, no γETSP3 nifH expression in either size-fraction was detected at any depth at Stns. 7, 9 or 11 in the noon samples, indicating that it was not actively expressing the nifH gene at the time of sampling. γETSP3 was not detected in any DNA samples from MV1104. There is an obvious discrepancy between the phylotypes that are dominant in clone libraries and the
© 2013 Society for Applied Microbiology and John Wiley & Sons Ltd, Environmental Microbiology, 16, 3095–3114
N2-fixing potential of heterotrophs in the ETSP
3101
Fig. 4. Photic zone abundances of gamma proteobacterial phylotypes γETSP1, γETSP2 and γETSP3 determined using qPCR at representative stations and γETSP2 nifH expression at Stn. 5 during AT1561. A. DNA samples recovered from the small size fraction (between 0.2 and 10 μm) were analysed for the abundances of ETSP-specific phylotypes. Many of the samples yielded values that were above detection limits, but beneath the limit of quantitation for these assays, thus determined to be DNQ. The grey box in each plot indicates this range of values between 13 and 75 nifH copies L−1. Sample collection blanks (shipboard milliQ water filtered according to sample filtration techniques) and DNA extraction blanks were negative for all three phylotypes. B. γETSP2 nifH expression depth profile throughout photic zone at Stn. 5 during AT1561. RNA samples recovered from both size fractions [between 0.2 and 10 μm (‘0.2 μm’) and greater than 10 μm (‘10 μm’)] at Stn. 5 were analysed for γETSP2. RNA samples were taken at local noon. No γETSP2 amplification was detected in controls for cDNA generation with no reverse transcriptase. C. γETSP2 diel nifH expression in surface waters at Stn. 5 during AT1561. Surface water, collected from a trace metal clean pump, was incubated in acid-cleaned cubitainers and subsampled at ∼6 h intervals over a 24-h period. No γETSP2 amplification was detected in controls for cDNA generation with no reverse transcriptase.
actual abundances determined using qPCR techniques, as evidenced by the relative numbers of γETSP1 sequences recovered from the AT1561 10°S samples (Fig. 1B) even though they were never present at abundances high enough to quantify (Fig. 4A). This result has also been previously reported in other studies using the
same primer set for marine environmental samples (Hewson et al., 2007; Turk et al., 2011; Farnelid et al., 2013). The nifH degenerate primer set used in this study has an amplification bias heavily weighted towards γ-proteobacteria (Turk et al., 2011) for reasons not wellunderstood. These findings underscore the importance of
© 2013 Society for Applied Microbiology and John Wiley & Sons Ltd, Environmental Microbiology, 16, 3095–3114
3102 K. Turk-Kubo, M. Karamchandani, D. Capone and J. Zehr cautious interpretation of relative abundances in clone libraries originating from PCR products. Comparison of nifH phylogenetic signatures within the South Pacific Ocean and between major ocean basins Use of nifH as a gene marker for the study of diazotroph diversity, abundance and distribution in diverse terrestrial and marine environments have been widespread (see reviews by Riemann et al., 2010, and Gaby and Buckley, 2011). By March 2013, the number of nifH sequences publically available in the GenBank nr/nt database, after curation using a pipeline for the import of these sequences into the ARB software environment (http://pmc.ucsc.edu/∼wwwzehr/research/database/), had grown to over 33 700 sequences (> 34 000 including sequences from this study), 10 073 of which were isolated from marine sources. In characterizing the nifH-based phylogenetic signature within the South Pacific and between different oceanic regions, a subset of the total reported marine sequences were chosen from studies that utilized molecular techniques similar to this study. In total, 3193 sequences (represented by 418 clusters at 97% nucleotide similarity) from 16 studies were selected that represented five major oceanic regions: the North Pacific, South Pacific, South China Sea, North Atlantic and Indian Ocean. An unweighted pairwise Unifrac P-test indicates that South Pacific diazotrophic communities are marginally significantly different (P ≤ 0.3; Fig. 5A) between the three existing studies (this study; Fernandez et al., 2011; and Halm et al., 2012), and an unweighted Unifrac lineagespecific analysis (LSA) identified three lineages that explain the potential differences (Fig. 5B). The LSA results indicated that cluster 1K sequences (primarily α- and β-proteobacteria) are over-represented in this study and in that of Fernandez and colleagues (2011), while being under-represented in Halm and colleagues (2012) (Fig. 5C). The other two lineages are 1P clusters (comprised primarily of β-proteobacteria) exclusively reported in Halm and colleagues (2012) originally reported as putative γ-proteobacteria and targeted by qPCR assays ‘Gamma 1’ and ‘Gamma 2’ (Fig. 5B and C). Although the pairwise unweighted Unifrac P-test conducted between all ocean basins resulted in a much higher P value (P ≤ 0.1; Fig. 5D), indicating that the difference among basins is of ‘suggestive’ significance, there were several nifH clusters that were either significantly overor under-represented in certain ocean basins based on the unweighted Unifrac LSA. Cluster 1B sequences (cyanobacteria) are over-represented in the North Atlantic and Pacific but are underrepresented in the South China Sea, Indian Ocean and the South Pacific (Fig. 5F). Cluster 1K and 1P sequences are over-represented in the South
Pacific, which is consistent with the LSA resulting from South Pacific studies described earlier. In the Indian Ocean, sequences within nifH cluster 1K are also overrepresented. The South China Sea is characterized not only by an under-representation of cyanobacterial sequences but also an over-representation of nifH cluster 1A sequences, which is comprised of non-sulphatereducing δ-proteobacterial genera. This subset of all marine-derived nifH sequences was also used to determine which heterotrophic phylotypes have been reported in multiple oceanic regions and studies, thus might be less likely to be contaminant sequences and may represent promising targets for future research. Of the 418 total clusters, 39 contain sequences reported from more than a single ocean basin, 28 of which do not cluster with Cyanobacterial nifH cluster 1B (Table 2). Many of the targets for both cyanobacterial and non-cyanobacterial qPCR assays that have been developed are contained within these clusters (Table 2 and Supporting information Table S2). Among the heterotrophic phylotypes reported, a majority are classified as nifH cluster 1G (primarily γ-proteobacteria) and cluster 1K (primarily α- and β-proteobacteria). Representatives are also present from nifH clusters 1E (Paenibacillus spp.), 1J (α-proteobacteria), as well as cluster 3 (e.g. Chlorobium spp. and Desulfovibrio spp. and others) (Table 2). Discussion South Pacific diazotroph communities It is evident that diverse populations of α,β-, γ-proteobacteria are present in the South Pacific, and thus far, there exists little evidence that cyanobacterial diazotrophs are abundant. Between this study and the study conducted by Halm and colleagues (2012), a total of seven new qPCR assays have been designed, chosen because of their frequent recovery in DNA- and RNAbased clone libraries, none of which target the same phylotypes. Although there is no overlap in sampling regions between these two studies, this finding is in contrast with the distribution of cyanobacterial diazotrophic phylotypes, where the same phylotype can be found distributed throughout entire ocean gyres [e.g. Trichodesmium in the Tropical North Atlantic (Goebel et al., 2010)]. Furthermore, very few similar sequence types (within 97% nucleotide similarity) were recovered from all sequencing efforts that have been conducted in the South Pacific thus far (this study; Fernandez et al., 2011; and Halm et al., 2012; Fig. 5B). Out of the four major groups identified by Halm and colleagues (2012), only the Gamma 4 phylotype was recovered in this study (γETSP6; Figs. 3 and 5B). Associated with nifH cluster 1G (Zehr et al., 2003a) and most
© 2013 Society for Applied Microbiology and John Wiley & Sons Ltd, Environmental Microbiology, 16, 3095–3114
N2-fixing potential of heterotrophs in the ETSP A
D This Study This Study
Halm et al., 2012 Fernandez et al., 2011
---
P ≤ 0.03
P ≤ 0.03
---
P ≤ 0.03
Halm et al., 2012 Fernandez et al., 2011
South Pacific
North Pacific
---
≤ 0.1 ---
South Pacific North Pacific South China Sea North Atlantic Indian Ocean
---
South North China Sea Atlantic ≤ 0.1 ≤ 0.1 ---
≤ 0.1 ≤ 0.1 ≤ 0.1 ---
3103
Indian Ocean ≤ 0.1 ≤ 0.1 ≤ 0.1 ≤ 0.1 ---
E
G2
G1
B
CI
II
76 54 32 1
43 21
A
3
1B
1B
1P
3
1A 1P
γ1
1G
1G
1K
1K
G3
G4
γ2
γ3
1 2 3 4
This Study Halm et al., 2012 Fernandez et al., 2011 highly significant (P < 0.001; Unifrac LSA) significant (0.01 < P < 0.05; Unifrac LSA)
C Halm et al., Fernandez et al., node This Study 2012 2011
P
1K
20 > 7
1 < 14
19 > 18
1.5E-07
1P.1
0 6
0 30 3 > 4 1P 19 >7 0>1 1A.1 03 0>2 16 > 5 0
