Molecular Ecology (2015) 24, 656–672

doi: 10.1111/mec.13050

Diet strongly influences the gut microbiota of surgeonfishes S O U M I Y A K E , D A V I D K A M A N D A N G U G I and U L R I C H S T I N G L Red Sea Research Center, 4700 King Abdullah University of Science and Technology (KAUST), Thuwal 23955-6900, Saudi Arabia

Abstract Intestinal tracts are among the most densely populated microbial ecosystems. Gut microbiota and their influence on the host have been well characterized in terrestrial vertebrates but much less so in fish. This is especially true for coral reef fishes, which are among the most abundant groups of vertebrates on earth. Surgeonfishes (family: Acanthuridae) are part of a large and diverse family of reef fish that display a wide range of feeding behaviours, which in turn has a strong impact on the reef ecology. Here, we studied the composition of the gut microbiota of nine surgeonfish and three nonsurgeonfish species from the Red Sea. High-throughput pyrosequencing results showed that members of the phylum Firmicutes, especially of the genus Epulopiscium, were dominant in the gut microbiota of seven surgeonfishes. Even so, there were large inter- and intraspecies differences in the diversity of surgeonfish microbiota. Replicates of the same host species shared only a small number of operational taxonomic units (OTUs), although these accounted for most of the sequences. There was a statistically significant correlation between the phylogeny of the host and their gut microbiota, but the two were not completely congruent. Notably, the gut microbiota of three nonsurgeonfish species clustered with some surgeonfish species. The microbiota of the macro- and microalgavores was distinct, while the microbiota of the others (carnivores, omnivores and detritivores) seemed to be transient and dynamic. Despite some anomalies, both host phylogeny and diet were important drivers for the intestinal microbial community structure of surgeonfishes from the Red Sea. Keywords: 16S rRNA genes, 454 pyrotags, Epulopiscium, gut microbiota, surgeonfish Received 13 March 2014; revision received 8 December 2014; accepted 17 December 2014

Introduction The microbial community in the gastrointestinal tract of a host organism is critical to the development, diseases and immune responses of the host (Macdonald & Monteleone 2005; Round & Mazmanian 2009; Turnbaugh et al. 2009; Wu et al. 2011). Characterization of the gut microbiota and their ecological function is relatively well advanced in humans, ruminants and arthropods, with various studies indicating that its composition is strongly influenced by the host’s phylogeny and diet (Cazemier et al. 1997; Ley et al. 2008b; Qin et al. 2010). For instance, Ley et al. (2008a) showed that the structure Correspondence: Ulrich Stingl, Fax: +966-2-8020152; E-mail: [email protected]

of intestinal microbial communities of higher terrestrial mammals is correlated with the host’s phylogeny at the order level (Primates, Carnivora and Artiodactyla) and to both diet and digestion mechanisms (herbivore foregut fermenters, hindgut fermenters, folivores, frugivores and carnivores). In contrast, similar research on fish gut microbiota lags behind that of terrestrial mammals, and the few existing studies primarily focus on describing the abundance, diversity and phylogeny of the microbes (Huber et al. 2004; Clements et al. 2007; Kim et al. 2007; Ward et al. 2009; Smriga et al. 2010) using conventional techniques (cultivation or clone libraries) at suboptimal sequencing depths (see Nayak (2010); Sullam et al. (2012) for summaries of pertinent studies). As highlighted by Clements et al. (2014), considerable research © 2014 John Wiley & Sons Ltd

S U R G E O N F I S H G U T M I C R O B I O T A 657 has been carried out on the preingestive processes of fish (e.g. grazing rates and feeding behaviours) but not on the postingestive processes (e.g. the fate of food and the microbes responsible for digestion). Nevertheless, next-generation high-throughput DNA sequencing is beginning to shed new light on the complexity of prokaryotic diversity in some commercially viable fishes (Star et al. 2013; Xing et al. 2013; Carda-Dieguez et al. 2014; Zarkasi et al. 2014). The apparent lack of research on the gut microbiota is especially true for coral reef fishes. In general, coral reefs are hotspots of biodiversity that are dynamic, even on the levels of invertebrates (Gibson et al. 2011) and microorganisms (Rohwer et al. 2002). This is also true for reef fishes, which constitute one of the most diverse assemblages of vertebrates that are extremely dynamic due to medium abiotic disturbances at both global and local scale (e.g. Syms & Jones 2000). However, despite their unparalleled species diversity and population density (Roberts et al. 2002), very little is known about their intestinal tract microbiology. Of all coral reef fishes, members of the Acanthuridae family (surgeonfishes, tangs and unicornfishes) are among the most abundant, with over 80 described species that exhibit a wide range of feeding behaviours (Jones 1968; Fishelson et al. 1987). Most studies on these particular reef fishes focus on their physiology and ecology, mainly on their feeding behaviours (Robertson & Gaines 1986; Montgomery et al. 1989; Montgomery & Galzin 1993; Purcell & Bellwood 1993; Choat et al. 2002, 2004; Green & Bellwood 2009; Fishelson & Delarea 2014), with less focus on characterizing their microbiome (Macdonald & Monteleone 2005; Round & Mazmanian 2009; Turnbaugh et al. 2009; Wu et al. 2011). To address this shortcoming and thus initiate discovery on a diverse group of vertebrates from marine biodiversity hotspots, we therefore investigated the composition of the gut microbiota of nine surgeonfish, two parrotfish and one rabbitfish species from the central Red Sea. The samples were spearfished from adjacent reefs within the central Red Sea, and the DNA was extracted from both host fish fins and the gut contents. CO1 gene was PCR-amplified and sequenced from the fin DNA, while V3-V6 region of the gut DNA was PCR-amplified and 454 pyrosequenced. The main objective of this work was to resolve the bacterial taxonomic composition in the intestinal tracts of these fishes, evaluate potential variability between and within each host fish species, and assess the degree of similarity between the gut microbiota of different fishes. To the best of our knowledge, this is the first study to use an extensive high-throughput sequencing on the gut bacterial community of a family of coral reef fish. Given general roles of gut bacteria in vertebrate development © 2014 John Wiley & Sons Ltd

and immunity, and given the importance of digestive and nutritional contributions in fishes, our findings on the ecological and evolutionary correlates of gut communities provide important insights into symbiotic dynamics of a functionally important group of fishes in the coral reef ecosystems.

Materials and methods Sample collection Fish samples were collected by spearfishing from two adjacent reefs in the central Red Sea, Saut reef (N19.893747, E40.154029) and Brown reef (N19.88762, E40.15665), from 1:00 pm on the day of sampling and all within one hour to minimize possible time-dependent effects in gut community profiles. Naso unicornis could not be found in the two reefs and was subsequently collected the following day, at the same time, in the nearby Manila Bay reef (N19.874758, E40.10335), with extra Acanthurus sohal samples collected to be used as controls. Surgeonfishes were the primary targets, but three other coral reef species (two parrotfishes and one rabbitfish) were also sampled for comparison. The fish species sampled include Acanthurus gahhm, Acanthurus nigrofuscus, Acanthurus sohal, Ctenochaetus striatus, Naso elegans, Naso hexacanthus, Naso unicornis, Zebrasoma desjardinii, Zebrasoma xanthurum, Chlorurus sordidus, Scarus niger and Siganus stellatus. At least three replicates per species were collected within the given time range. The sampling variables, namely location, date of collection and fish size (length and weight) (Table S1, Supporting information), did not significantly affect the alphadiversity values of the gut microbiota (Kruskal–Wallis, P > 0.05 for all combinations). Additionally, analysis of similarity (ANOSIM, with 999 iterations) between gut communities of these variables revealed no significant effects of these variables on the gut community similarity (0.23 ≤ r ≤ 0.05, P > 0.05), confirming that the data set could be used together without considering the influence of these sampling variables. Notably, gut microbiota of A. sohal replicates from different reefs did not show significant difference.

Identification and phylogenetic reconstruction of fishes Fin clippings of reef fishes were stored in 75% ethanol at 20 °C until the DNA was extracted using a QIAGEN blood and tissue kit (QIAGEN, Limburg, Netherlands) following the manufacturer’s protocol. The extracted DNA was PCR-amplified using cytochrome c oxidase subunit I (CO1) primers designed for the universal barcoding of all fishes (Ivanova et al. 2007) along with a negative control. The PCR primer sequences and

658 S . M I Y A K E , D . K . N G U G I and U . S T I N G L PCR conditions are listed in Table S2 (Supporting information). PCR products were cleaned up using Illustra ExoStar (GE Healthcare Life Sciences, NJ, USA) following the standard protocol, and bidirectionally sequenced (using the same PCR primers) on an Applied Biosystems 3730xl DNA Analyzer at the Bioscience Core Lab of King Abdullah University of Science and Technology (KAUST). Raw sequences were trimmed with SEQUENCHER 4.9 (Gene Codes Corporation) and aligned using CLUSTALW (v1.2.0; Larkin et al. 2007). The preliminary species assignment by morphological characteristics (colour, body shape and fin count) against the reference text (Randall 2001) was confirmed by BLASTN analysis of the generated CO1 sequences against the nucleotide collection (nr/nt) database from the National Center for Biotechnology Information (NCBI; Altschul et al. 1990). Phylogenetic trees of the aligned CO1 sequences were subsequently constructed using both maximum-likelihood (RAXML 8.0.0; Stamatakis 2014) and Bayesian inference (MRBAYES; Ronquist et al. 2012) algorithms. The General Time Reversal (GTR) was selected as the optimal substitution model by JMODELTEST 2 (Darriba et al. 2012) for both trees. RAXML was run using a gamma distribution with four discrete categories and an estimated alpha parameter with 1000 bootstrap replicates, while MRBAYES was run twice with four chains for a million iterations, sampling every 100. The first 25% of the trees were discarded as ‘burn-in’, and the remaining trees were summarized as a 50% majority rule consensus tree.

MiniPrep kit (ZYMO Research, CA, USA) following the manufacturer’s protocol. A blank sample was also processed as a negative control during the method optimization of the DNA extraction, producing negative result from the PCR with primers used for pyrosequencing without the adaptors and barcodes (B343F and B1099rc). A fragment of the 16S rRNA gene targeting the V3V6 region was PCR-amplified using the primer pair B343F (forward) and B1099rc (reverse) (Liu et al. 2007) with FLX Titanium adaptors (Lib-L) and a domain-specific primer fused with a two-nucleotide linker and sample-specific 8-bp barcode for multiplexing (HSPLpurified; Sigma-Aldrich Life Science, MO, USA). The PCR primer sequences and PCR conditions are listed in Table S2 (Supporting information). One positive (genomic DNA from E. coli) and one negative control (sterile deionized water) were also PCR-amplified along with the samples. The samples and the positive control both produced positive bands of approximately 700 bp in length, while no amplification was obtained with the negative control upon visualization after gel electrophoresis. Fifty-nine samples (excluding the controls) were PCR-amplified in triplicates and purified using the Zymoclean gel DNA recovery kit (ZYMO Research, CA, USA). The purified amplicons were quantified with Quant-iT PicoGreen (Life technologies, CA, USA) and pooled at equimolar concentrations prior to sequencing on a Roche 454 GS FLX Titanium platform in our Bioscience Core Laboratory.

Gut DNA extraction and pyrosequencing

Sequence processing

From each fish, the intestinal tract was dissected using a sterile scalpel within one hour of sampling. Fishes with intestinal tracts punctured by the speargun were removed from the analysis. The gut contents from segments posterior to the stomach (spanning the midgut and hindgut regions) were squeezed out and stored in 75% ethanol in 1:3 ratios (w/v) at 20 °C until the DNA extraction. Great care was taken to ensure that no part of the gut wall was mixed with the gut contents to avoid contamination. Similar approach has been taken by previous studies (e.g. Clements et al. 2007). From fish species without a clear stomach (i.e. parrotfish), the anterior 25% of the intestinal tract was removed, and the gut contents of the remaining 75% were stored. Two grams of homogenized gut contents per sample was centrifuged (15 000 g, 3 mins) and washed three times with phosphate-buffered saline solution (PBS) to remove the residual ethanol. The coarse gut contents were homogenized using a three-minute bead-beating procedure at 30 Hz. DNA was then extracted using the ZR Soil Microbe DNA

Sequence cleanup and classification were conducted using software packages implemented in MOTHUR (version 1.31.0; Schloss et al. 2009) based on the protocol by Schloss et al. (2011). The raw reads were de-noised using the raw flowgram files via MOTHUR’s implementation of PyroNoise with minimum and maximum flows set at 320 and 800 bp (Quince et al. 2009). The reads were subsequently trimmed to remove the barcodes and primer sequences as well as any reads with (i) ambiguous bases, (ii) homopolymers of more than eight bp and (iii) lengths less than 200 bp. The trimmed reads were then aligned based on the SILVA SSUREF database (v102), and the chimeric sequences were removed using the UCHIME algorithm with self-reference (Edgar et al. 2011). The remaining reads were classified based on the Greengenes reference taxonomy (May 2013 release) with 80% confidence, and results are presented at the phylum and genus levels. Sequences from chloroplasts, mitochondria and nonbacterial origin were removed as contaminants from the classification and the downstream analysis. © 2014 John Wiley & Sons Ltd

S U R G E O N F I S H G U T M I C R O B I O T A 667 Table 3 The ten most abundant OTUs from each of the three clusters of surgeonfish gut microbial communities. The OTUs were classified according to the top BLASTN hit. Their relative abundance (in %) and the origin of the most closely related sequence – classified as coming from fish gut, gut (from other organims) or nongut environment – are presented OTU #

Top

BLASTN

hit

Cluster 1 001 Epulopiscium from Naso lituratus (unicornfish) gut (AY844985.1) (98) 044 Clostridiales from packing materials of crab shell column treating acid mine drainage (KF581577.1) (96) 010 Lachnospiraceae from Naso tonganus gut (HM630231.1) (99) 025 Lachnospiraceae from Naso tonganus gut (HM630231.1) (97) 057 Ruminococcaceae from rat faeces (GU958465.1) (92) 050 Akkermansia from Naso tonganus gut (HM630258.1) (99) 086 Clostridiales from Meles meles (European bager) gut (JN806175.1) (93) 061 Oscillospira from rat feces (AB702848.1) (96) 031 Cetobacterium from Ctenopharyngodon idellus (grass carp) gut (JN033193.1) (97) 143 Erysipelotrichaceae from Naso tonganus gut (HM630253.1) (99) Cluster 2 003 Clostridiales from Stylophora pistillata coral in the Red Sea (KC668954.1) (97) 002 Epulopiscium from Naso lituratus gut (AY844988.1) (97) 012 Clostridiaceae from human faeces (JX851940.1) (98) 005 Cetobacterium from Danio rerio (zebrafish) gut (DQ815035.1) (98) 021 Epulopiscium from Acanthurus nigricans gut (FJ653967.1) (99) 010 Lachnospiraceae from Naso tonganus gut (HM630231.1) (99) 001 Epulopiscium from Naso lituratus gut (AY844985.1) (98) 028 Lachnospiraceae from Naso tonganus gut (HM630231.1) (98) 041 Lachnospiraceae from Naso tonganus gut (HM630224.1) (97) 009 Epulopiscium from Naso tonganus gut (HM630230.1) (98) Cluster 3 004 Rhodobacteraceae from Montastraea faveolata coral with white plague disease (FJ202799.1) (99) 007 Peptostreptococcaceae from beef cattle gut (JX096000.1) (100) 005 Cetobacterium from Danio rerio (zebrafish) gut (DQ815035.1) (98) 020 Rhodobacteraceae from Porites compressa coral in Hawaii (FJ930373.1) (99) 003 Clostridiales from Stylophora pistillata coral in the Red Sea (KC668954.1) (100) 009 Epulopiscium from Naso tonganus gut (HM630230.1) (98) 026 Rhodobacteraceae from biofilms in the Great Barrier Reef (JF261987.1) (99) 029 Vibrionales from Pomacanthus sexstriatus (angelfish) gut (EU884928.1) (99) 091 Synechococcus from the surface sea water in the Red Sea (AY172831.1) (100) 032 Fusobacteriaceae from Siganus canaliculatus (rabbitfish) gut (HG326493.1) (98)

The intraspecies core gut microbiota of coral reef fishes Considering that all samples were captured from the same locality (with neither date nor location of sampling significantly influencing the gut microbiota), the fraction of shared OTUs among replicates was surprisingly low, particularly in A. gahhm, A. sohal, N. elegans, Z. xanthurum and Ch. sordidus (1.0–5.7%). These percentages are similar to the shared gut microbiota of wild and domesticated zebrafishes sampled from three geographically distinct locations (Roeselers et al. 2011). In other freshwater fishes, the proportions of core gut microbiota were significantly larger than observed in this study, such as the case of rainbow trout (Oncorhynchus mykiss) reared under aquaculture conditions, with approximately 50% of the OTUs shared between repli© 2014 John Wiley & Sons Ltd

%

Origin

84.2 1.0 0.7 0.7 0.4 0.3 0.2 0.2 0.2 0.2

Fish gut Nongut Fish gut Fish gut Gut Fish gut Gut Gut Fish gut Fish gut

9.0 5.3 2.9 2.0 1.7 1.5 1.4 1.2 1.0 1.0

Nongut Fish gut Gut Fish gut Fish gut Fish gut Fish gut Fish gut Gut Fish gut

1.7 1.3 0.8 0.7 0.5 0.5 0.4 0.4 0.4 0.4

Nongut Gut Fish gut Nongut Nongut Fish gut Nongut Fish gut Nongut Fish gut

cates on different diets, which accounted for approximately 82% of the sequences (Wong et al. 2013). The difference between freshwater fishes previously studied and the coral reef fishes in this study can be attributed to differences in food resources, the physiochemical conditions in their external environments, or perhaps due to the fact that the fishes used in other studies were in aquaculture. Despite the low counts of shared OTUs among surgeonfishes, these shared OTUs accounted for the majority of retrieved sequences in eight of the 12 fish species studied. This relatively small but dominant core gut microbiota was not restricted to surgeonfish species with high abundances of Epulopiscium (e.g. N. elegans and N. unicornis). Unfortunately, we cannot deduce from this study the extent to which these core (and

668 S . M I Y A K E , D . K . N G U G I and U . S T I N G L unique) gut populations are responsible for the breakdown of food matter, a topic that needs to be addressed by further functional studies. Conversely, the minimal shared OTUs among replicates, along with high alpha diversity, may be biased by the 454 pyrosequencing. There were two possible margins of errors, one of the sequencing platforms and another of the general problem associated with short sequencing reads (e.g. She et al. 2004; Reeder & Knight 2009; Tedersoo et al. 2010; Turnbaugh et al. 2010; Haas et al. 2011). As a result, many of the closely related OTUs in the study may have been clustered as independent OTUs, which in turn artificially increased the alpha-diversity values and decreased the number of shared OTUs among the replicates.

Influence of phylogeny and diet on the gut microbiota Although the host’s phylogeny and the similarities between gut microbial community compositions in the studied fishes were not completely congruent, there was a statistically significant association between them. This is in line with available literature, where Roeselers et al. (2011) found complete congruence of the gut microbiota between host phylogeny at subspecies level, while Ley et al. (2008a) observed similar trend in terrestrial mammals at the order level. Despite these studies targeting different organisms at different phylogenetic level, there seems to be an overall consensus that host’s phylogeny indeed plays an important role in the gut microbiota of vertebrates. This is a little surprising for reef fishes, which have limited interactions with the parents and thus are likely to acquire majority of their enteric bacteria postpartum from the environment. The distinction of intestinal bacterial communities according to the host’s diet has been reported previously in terrestrial vertebrates (for instance, Claesson et al. 2012; Ley et al. 2008a), while human enterotypes have been linked to the long-term dietary patterns (Wu et al. 2011). In fish, despite the lack of microbial characterization studies, much work has been done to infer the microbial activity, or to associate the nutritional ecology of the fish to the host’s diet. For instance, Mountfort et al. (2002) detected the microbial-driven fermentation in the hindgut of three temperate marine herbivores by measuring the production rates of shortchain fatty acids (SCFA). Although the study did not identify which microbes were responsible for SCFA production, the rates were comparable to those of terrestrial vertebrates, highlighting the importance of microbial activity in the food digestion mechanisms of these fishes. Additionally, nutritional components and the fermentation potential (inferred from the SCFA concentrations) of guts have been distinguished according

to their feeding types in several studies (e.g. Choat & Clements 1998; Choat et al. 2002, 2004; Crossman et al. 2005). For instance, concentrations of various SCFAs were significantly different in macro- and microalgavores, which was attributed to fundamental differences in the food digestion mechanisms (Choat & Clements 1998). While the former seems to rely largely on the hindgut fermentation, the latter undergoes acid lysis of the ingesta in the anterior tract and some fermentation in the hindgut. This study further directly confirmed that the association between the host’s diet and their gut microbiota is also true for selected surgeonfishes from the central Red Sea. Importantly, the study highlights the need for detailed classification of the diet of coral reef fishes in conjunction with the microbial characterization. Many studies on terrestrial organisms have highlighted the difference in the gut bacterial community of herbivores, carnivores and omnivores (Ley et al. 2008a; Muegge et al. 2011; Sullam et al. 2012), but the analysis here illustrated that further in-depth analysis of host diet may be necessary for some organisms with complex dietary habit, such as coral reef fishes. Admittedly, variations and plasticity in the diets of some species (e.g. A. gahhm) may require even more comprehensive diet analyses, on individual basis, to complement the gut microbiota characterization studies. Nevertheless, despite large differences in the composition, we found an unequivocal correlation between the diet of the host and its gut microbiota, confirming that the diet is fundamentally important for the gut communities in the fishes tested. Interestingly, there was substantial variation between replicates when the community structure (i.e. weighted UniFrac) was considered rather than the community composition. This suggests that the variations in the abundance of bacteria are substantially different between replicates in some species, sometimes to the extent that the diet-driven clusters could no longer be observed. The inclusion of A. nigrofuscus among other nonherbivorous fishes in cluster 3 is likely to be specific to the central Red Sea. Although previous studies have indicated that they are microalgal feeders similar to fishes in cluster 2 (Fishelson et al. 1987; Choat et al. 2002; Schuhmacher et al. 2009), their gut contained large amount of coarse reef debris/sediments, indicating that they are likely to be ingesting reef debris along with the microalgae (Table S1, Supporting Information). This phenomenon might be specific to A. nigrofuscus from this specific sampling region, where the habitat is dominated by the endemic and highly territorial A. sohal. They are much larger in size (> 30 cm) and display aggressive behaviour towards A. nigrofuscus, which may restrict them to feeding on low-quality algae that © 2014 John Wiley & Sons Ltd

© 2014 John Wiley & Sons Ltd

C

Naso

§§

3

3

3

4

3

11

8

5

3

8

4

11 260

5408

15 442

9348

7604

43 413

35 504

21 304

12 906

24 152

6769

17 187

Total

97  3 97  1

1803  57 3753  701

97  1

97  1

2337  510 5147  2117

99  1 98  1

2535  1198

98  1

99  1

92  7

97  1

3947  1147

4438  1989

4261  1858

4302  5316

3019  1039

93  4

94  5

4297  4496 1692  177

Mean  SD

coverage (%)

Good’s

Mean  SD

No. of sequences

278

131

249

142

117

61

128

86

345

140

225

331

Total§

158  30

103  32

103  31

101  15

74  35

30  18

71  23

46  13

205  101

83  21

175  55

180  90

Subsampled¶

Observed OTUs



478

266

492

218

228

133

231

167

609

300

469

590

Total§

Chao 1

290  42

195  105

235  65

178  18

144  58

69  38

134  33

89  19

478  188

171  42

387  196

413  246

Subsampled¶

609

330

808

276

289

235

318

258

807

430

753

856

Total§

ACE

392  82

239  142

378  89

258  27

206  88

127  63

193  70

139  34

763  210

250  71

638  425

666  419

Subsampled¶

3.92

3.09

2.48

2.89

1.84

0.50

2.87

2.03

3.38

2.42

3.91

3.57

Total§

Shannon





3.82  0.50

3.05  0.44

2.40  0.18

2.85  0.53

1.81  0.96

0.48  0.38

2.83  0.48

2.00  0.54

3.30  1.14

2.38  0.34

3.86  0.34

3.47  0.76

Subsampled¶

Diversity estimates

Diet category based on Choat et al. (2002). Mi, microalgavore; Ma, macroalgavore; D, detritivore; O, omnivore; C, carnivore. Richness and diversity estimates are OTU-based at 97% sequence identity level. § Average value of nonsubsampled data set. ¶ Subsampled to the smallest sampling effort (1083 reads) per replicate, given as mean  standard deviation. †† % OTUs shared amongst replicates of the same species. ‡‡ % of the sequences falling into the core OTUs. §§ Conventionally classified as a microalgavore but may be a detritivore in the central Red Sea.



O

Siganus

stellatus

D

Scarus niger

sordidus

Chlorurus

D

Ma

hexacanthus

Ma

Naso unicornis

Mi

Mi

D

Mi

Naso elegans

xanthurum

Zebrasoma

desjardinii

Zebrasoma

striatus

Ctenochaetus

sohal

Acanthurus

nigrofuscus

Acanthurus

Mi/D

O

Acanthurus

4

replicates

category†

Species

gahhm

No. of

Diet

Richness estimates

0.05

0.11

0.26

0.16

0.45

0.85

0.11

0.24

0.20

0.22

0.04

0.12

Total§

Simpson

0.05  0.03

0.11  0.05

0.26  0.10

0.16  0.09

0.45  0.25

0.85  0.12

0.06  0.11

0.24  0.12

0.20  0.14

0.22  0.09

0.04  0.01

0.12  0.10

Subsampledk

19.2

5.4

5.7

8.8

6.2

1.2

2.8

3.7

4.7

1.8

4.5

1.0

OTUs (%)

Shared ††

Core microbiota

84.8

49.1

28.2

73.6

72.4

94.9

60.4

70.2

73.7

74.1

40.6

12.8

(%)

‡‡

Shared seqs

Table 1 Overview of fish samples, their diet category and retrieved 16S rRNA gene sequence data including alpha-diversity estimates and percentage of shared microbiota

S U R G E O N F I S H G U T M I C R O B I O T A 661

670 S . M I Y A K E , D . K . N G U G I and U . S T I N G L Clements KD, Sutton DC, Choat JH (1989) Occurrence and characteristics of unusual protistan symbionts from surgeonfishes (Acanthuridae) of the Great Barrier-Reef, Australia. Marine Biology, 102, 403–412. Clements KD, Pasch IBY, Moran D, Turner SJ (2007) Clostridia dominate 16S rRNA gene libraries prepared from the hindgut of temperate marine herbivorous fishes. Marine Biology, 150, 1431–1440. Clements KD, Angert ER, Montgomery WL, Choat JH (2014) Intestinal microbiota in fishes: what’s known and what’s not. Molecular Ecology, 23, 1891–1898. Crossman DJ, Choat JH, Clements KD (2005) Nutritional ecology of nominally herbivorous fishes on coral reefs. Marine Ecology Progress Series, 296, 129–142. Curson ARJ, Todd JD, Sullivan MJ, Johnston AWB (2011) Catabolism of dimethylsulphoniopropionate: microorganisms, enzymes and genes. Nature Reviews Microbiology, 9, 849–859. Darriba D, Taboada GL, Doallo R, Posada D (2012) JMODELTEST 2: more models, new heuristics and parallel computing. Nature Methods, 9, 772. Edgar RC, Haas BJ, Clemente JC, Quince C, Knight R (2011) UCHIME improves sensitivity and speed of chimera detection. Bioinformatics, 27, 2194–2200. Fishelson L, Delarea Y (2014) Comparison of the oral cavity architecture in surgeonfishes (Acanthuridae, Teleostei), with emphasis on the taste buds and jaw “retention plates”. Environmental Biology of Fishes, 97, 173–185. Fishelson L, Montgomery LW, Myrberg AH (1987) Biology of surgeonfish Acanthurus-Nigrofuscus with emphasis on changeover in diet and annual gonadal cycles. Marine Ecology Progress Series, 39, 37–47. Flint JF, Drzymalski D, Montgomery WL, Southam G, Angert ER (2005) Nocturnal production of endospores in natural populations of Epulopiscium-like surgeonfish symbionts. Journal of Bacteriology, 187, 7460–7470. Gibson R, Atkinson R, Gordon J, Smith I, Hughes D (2011) Coral-associated invertebrates: diversity, ecological importance and vulnerability to disturbance. Oceanography and marine biology: an annual review, 49, 43–104. Green AL, Bellwood DR (2009) Monitoring functional groups of herbivorous reef fishes as indicators of coral reef resilience - A practical guide for coral reef managers in the Asia Pacific region IUCN, Gland, Switzerland. Haas BJ, Gevers D, Earl AM et al. (2011) Chimeric 16S rRNA sequence formation and detection in Sanger and 454-pyrosequenced PCR amplicons. Genome Research, 21, 494–504. Holmes I, Harris K, Quince C (2012) Dirichlet multinomial mixtures: generative models for microbial metagenomics. PLoS ONE, 7, e30126. Huber I, Spanggaard B, Appel KF et al. (2004) Phylogenetic analysis and in situ identification of the intestinal microbial community of rainbow trout (Oncorhynchus mykiss, Walbaum). Journal of Applied Microbiology, 96, 117–132. Huse SM, Welch DM, Morrison HG, Sogin ML (2010) Ironing out the wrinkles in the rare biosphere through improved OTU clustering. Environmental Microbiology, 12, 1889–1898. Ivanova NV, Zemlak TS, Hanner RH, Hebert PDN (2007) Universal primer cocktails for fish DNA barcoding. Molecular Ecology Notes, 7, 544–548.

Jones RS (1968) Ecological relationships in Hawaiian and Johnston Island Acanthuridae (surgeonfishes). Micronesica, 4, 309–361. Kim DH, Brunt J, Austin B (2007) Microbial diversity of intestinal contents and mucus in rainbow trout (Oncorhynchus mykiss). Journal of Applied Microbiology, 102, 1654–1664. Klanten SO, Herwerden LL, Choat JH, Blair D (2004) Patterns of lineage diversification in the genus Naso (Acanthuridae). Molecular Phylogenetics and Evolution, 32, 221–235. Larkin MA, Blackshields G, Brown NP et al. (2007) CLUSTAL W and CLUSTAL X version 2.0. Bioinformatics, 23, 2947–2948. Ley RE, Hamady M, Lozupone C et al. (2008a) Evolution of mammals and their gut microbes. Science, 320, 1647–1651. Ley RE, Lozupone CA, Hamady M, Knight R, Gordon JI (2008b) Worlds within worlds: evolution of the vertebrate gut microbiota. Nature Reviews Microbiology, 6, 776–788. Liu Z, Lozupone C, Hamady M, Bushman FD, Knight R (2007) Short pyrosequencing reads suffice for accurate microbial community analysis. Nucleic Acids Research, 35, e120. Lozupone C, Knight R (2005) UNIFRAC: a new phylogenetic method for comparing microbial communities. Applied and Environment Microbiology, 71, 8228–8235. Macdonald TT, Monteleone G (2005) Immunity, inflammation, and allergy in the gut. Science, 307, 1920–1925. Mendell JE, Clements KD, Choat JH, Angert ER (2008) Extreme polyploidy in a large bacterium. Proceedings of the National Academy of Sciences, USA, 105, 6730–6734. Moitinho-Silva L, Bayer K, Cannistraci CV et al. (2013) Specificity and transcriptional activity of microbiota associated with low and high microbial abundance sponges from the Red Sea. Molecular Ecology, 23, 1348–1363. Montgomery WL, Galzin R (1993) Seasonality in Gonads, fat deposits and condition of tropical surgeonfishes (Teleostei, Acanthuridae). Marine Biology, 115, 529–536. Montgomery WL, Myrberg AA, Fishelson L (1989) Feeding ecology of surgeonfishes (Acanthuridae) in the Northern Red-Sea, with particular reference to Acanthurus-Nigrofuscus (Forsskal). Journal of Experimental Marine Biology and Ecology, 132, 179–207. Mountfort DO, Campbell J, Clements KD (2002) Hindgut fermentation in three species of marine herbivorous fish. Applied and Environment Microbiology, 68, 1374–1380. Muegge BD, Kuczynski J, Knights D et al. (2011) Diet drives convergence in gut microbiome functions across mammalian phylogeny and within humans. Science, 332, 970–974. Nayak SK (2010) Role of gastrointestinal microbiota in fish. Aquaculture Research, 41, 1553–1573. Ngugi DK, Antunes A, Brune A, Stingl U (2012) Biogeography of pelagic bacterioplankton across an antagonistic temperature-salinity gradient in the Red Sea. Molecular Ecology, 21, 388–405. Purcell SW, Bellwood DR (1993) A functional-analysis of food procurement in 2 surgeonfish species, Acanthurus-Nigrofuscus and Ctenochaetus-Striatus (Acanthuridae). Environmental Biology of Fishes, 37, 139–159. Qin JJ, Li RQ, Raes J et al. (2010) A human gut microbial gene catalogue established by metagenomic sequencing. Nature, 464, 59–U70. Quince C, Lanzen A, Curtis TP et al. (2009) Accurate determination of microbial diversity from 454 pyrosequencing data. Nature Methods, 6, 639–641.

© 2014 John Wiley & Sons Ltd

S U R G E O N F I S H G U T M I C R O B I O T A 663

© 2014 John Wiley & Sons Ltd

Comparisons of gut communities The hierarchical cluster analysis, based on the traditional Jaccard dissimilarity index of the subsampled data set, grouped the gut microbiota into three speciesspecific clades (Fig. 3). The first cluster consisted of all replicates of N. elegans and N. unicornis; the second cluster of A. sohal, Z. desjardinii and Z. xanthurum; and the last cluster of A. gahhm, A. nigrofuscus, Ct. striatus and N. hexacanthus, as well the three nonsurgeonfish species (Ch. sordidus, Sc. niger and Si. stellatus). While replicates of most species formed separate and independent species-specific subclusters within these three clusters (e.g. N. hexacanthus and Si. stellatus), replicates of some species formed interspersed subclusters; for instance, Z. desjardinii and Z. xanthurum together formed a subcluster that could not be statistically differentiated (AMOVA, F = 1.44, P = 0.015). Both the Dirichlet Multinomial Mixtures (Holmes et al. 2012) (Fig. S3, Supporting information) and the k-means methods verified

Trophic level

Diet category

Clusters 0.05

Nele07 Nele05 Nele09 Nele02 Nele10 Nele06 Nele11 Nele01 Nele03 Nele04 Nele08 Nuni02 Nuni01 Nuni03 Asoh01 Asoh06 Asoh04 Asoh05 Asoh03 Asoh07 Asoh02 Asoh08 Zxan01 Zxan02 Zxan04 Zxan05 Zxan03 Zxan06 Zdes01 Zdes02 Zdes03 Zxan08 Zxan07 Zdes04 Zdes05 Sste01 Sste02 Sste03 Anig01 Anig04 Anig02 Anig03 Agah04 Agah01 Agah03 Cstr01 Cstr02 Cstr03 Csor03 Csor02 Csor01 Snig02 Snig01 Agah02 Snig03 Nhex01 Nhex02 Nhex03 Nhex04

Species

abundant taxa in that species (Table S3, Supporting information). For instance, surgeonfishes generally shared many Clostridia OTUs – especially Epulopiscium or closely related unclassified genus of Lachnospiraceae – among replicates, with the exception of Ct. striatus who shared mostly unclassified Rhodobacteraceae. Additionally, only a few OTUs were shared in each genus of surgeonfish, with individuals of Zebrasoma sharing six OTUs, and both Acanthurus and Naso species sharing only a single OTU among the individuals. Similarly, the fish species categorized by the diet as described by Choat et al. (2002) did not share a large core gut microbiota. The macroalgavores, which consisted of 14 individuals from two Naso species, shared four OTUs; the microalgavores with 25 individuals from four species shared a single OTU; the omnivores with seven individuals from two species shared two OTUs; and the detritivores with nine individuals from three species shared three OTUs. The carnivores shared 28 OTUs, but they consisted of only four individuals from a single species of surgeonfish.

Species A. nigrofuscus A. sohal A. gahhm Ct. striatus Surgeonfish Z. desjardinii Z. xanthurum N. elegans N. unicornis N. hexacanthus Ch. sordidus Parrotfish Sc. niger Si. stellatus Rabbitfish Diet Category Macroscopic algavores Microscopic algavores Detritus/sediment feeders Omnivores Zooplankton feeders Trophic Level 2.00 2.62 2.70 2.76 3.00

Fig. 3 Hierarchical clustering of individual gut community composition using the Unweighted Pair Group Method with Arithmetic Mean (UPGMA). The similarities between communities were compared using a Jaccard similarity coefficient based on observed richness. Diet category and trophic level of the host fish were adapted from Choat et al. (2002) and FishBase (www.fishbase.org), respectively. Pentagonal boxes are used for all nonsurgeonfishes, while square, triangular and circular boxes are used to denote species, diet category and tropic level, respectively, for surgeonfishes. The colour in each box represents the classification.

664 S . M I Y A K E , D . K . N G U G I and U . S T I N G L that three clusters was the optimal number of clusters in the data set. The Mantel test between the distance metrics of the host CO1 gene and the Jaccard dissimilarity index showed a significant overall correlation between the host’s phylogeny and the gut microbiota (r = 0.642, P < 0.001). However, a comparison of the hierarchical dendrogram and the host’s phylogeny (Figs 1 and 3) revealed that the two were not completely congruent. Although some sister species within the genera Naso (N. elegans and N. unicornis) and Zebrasoma (Z. desjardinii and Z. xanthurum) shared similar gut microbiota, this was not the case for the sister species of Acanthurus, nor for N. hexacanthus, whose gut microbiota were distant from the rest of the Naso species. More strikingly, both parrotfish and rabbitfish are phylogenetically distinct from surgeonfishes, but their gut microbiota grouped together with some surgeonfish species (A. gahhm, A. nigrofuscus, Ct. striatus and N. hexacanthus) in cluster 3 (Fig. 3). The effect of host phylogeny on the gut microbiota community composition was also confirmed by the analysis of similarity (ANOSIM), where the unweighted UniFrac distance matrix was significantly different among host genera (r = 0.62, P < 0.001). Observing the individual interactions, there were significant associations between Acanthurus–Naso, Acanthurus–Zebrasoma, Chlorurus–Naso, Ctenochaetus–Naso, Ctenochaetus–Zebrasoma and Naso–Zebrasoma (see Table S4, Supporting information, for the full summary). In addition, the unweighted UniFrac distance matrix was significantly different among host species as well (ANOSIM, r = 0.86, P < 0.001). Individual interactions that showed significant differences were as follows: A. gahhm–N. elegans, A. nigrofuscus–A. sohal, A. nigrofuscus–N. elegans, A. sohal–N. elegans, A. sohal–N. hexacanthus, A. sohal–Z. desjardinii, A. sohal–Z. xanthurum, N. elegans–N. hexacanthus, N. elegans–Si. stellatus and N. elegans–Z. xanthurum (Table S5, Supporting information). The three potential clusters of gut microbial community could be more clearly visualized using principal coordinates analysis (PCoA) of the unweighted UniFrac distance matrix (Fig. 4). The first two coordinates of PCoA, which accounted for 8.6% and 7.1% of the variance, showed that the fish clustered into three groups according to the diet categories described by Choat et al. (2002). Clusters 1 and 2 were composed of herbivores, but distinguished by the different types of algae they consume. Species in cluster 1 (N. elegans and N. unicornis) are reef browsers that feed primarily on macroalgae (brown fucoid algae), while species in cluster 2 (Z. desjardinii, Z. xanthurum and A. sohal) feed on smaller microalgae including filamentous and turfing

red and green algae (Montgomery et al. 1989; Choat et al. 2002; Green & Bellwood 2009). Cluster 3 is more complex and consisted of a planktivore (N. hexacanthus), omnivores (A. gahhm and Si. stellatus), detritivores (Ct. striatus, Ch. sordidus and Sc. niger) as well as a single herbivorous species (A. nigrofuscus) (Figs 3 and 4). All seven surgeonfish species in the third cluster appear to ingest large amounts of nonalgal organic matter (e.g. sandy sediments and coral debris) as indicated by their gut colour and texture (S1, Supporting Information) as well as by the dominance of bacterial OTUs often associated with nongut environments (Tables 3). The inclusion of A. nigrofuscus in cluster 3 was unexpected given that this species is known to have a similar diet as fishes in cluster 2 (Fishelson et al. 1987; Montgomery et al. 1989). The ANOSIM confirmed that unweighted UniFrac distance matrix was significantly different between the diet categories (r = 0.74, P < 0.001; see Table 2 for the full summary). There was a statistically significant difference between the gut microbiota of fishes with different diets, except between carnivores, omnivores and detritivores, who constituted the third cluster. Notably, neither the trophic level nor the simple diet classification often used for terrestrial gut microbiota studies (carnivores, omnivores and herbivores) was significantly correlated with the unweighted UniFrac distance matrix (ANOSIM, r = 0.02, P > 0.01). Interestingly, the weighted UniFrac, which also includes the abundance of bacteria in the community, showed a similar clustering pattern but with more variability between the individuals (Fig. S4, Supporting information), which is consistent with the large variability observed in taxonomic composition at the genus level. Furthermore, the abundant OTUs in each cluster differed considerably, revealing the fundamental differences between the three clusters (Table 3). Cluster 1 was distinct from clusters 2 and 3 in that the species in cluster 1 were dominated by a single OTU assigned as Epulopiscium, which accounted for 85% of the sequences. In contrast, clusters 2 and 3 had lower relative abundances, with the ten most common OTUs ranging from 1.1 to 9.2%, and 0.44 and 1.76%, respectively. The BLASTN search for the closest known relatives of these abundant OTUs showed that most OTUs from clusters 1 and 2 were closely related to sequences found in the guts of other fishes, especially the closely related surgeonfish species, Naso lituratus and Naso tonganus from Flint et al. (2005) and Mendell et al. (GenBank No. HM630231). In contrast, the abundant OTUs in cluster 3 were dominated by organisms associated with nongut environments, particularly corals, which were likely to be ingested by many of these fishes.

© 2014 John Wiley & Sons Ltd

S U R G E O N F I S H G U T M I C R O B I O T A 665 (a) Species

Diet category

PC2 - Per cent variation explained 7.1%

PC2 - Per cent variation explained 7.1%

(b)

PC1 - Per cent variation explained 8.6%

PC1 - Per cent variation explained 8.6%

(c) Trophic level

Surgeonfish

PC2 - Per cent variation explained 7.1%

b. Diet category a. Species Macroscopic algavores A. nigrofuscus Microscopic algavores A. sohal Detritus/sediment feeders A. gahhm Omnivores Ct. striatus Zooplankton feeders Z. desjardinii Z. xanthurum N. elegans c. Trophic level N. unicornis 2.00 2.62 N. hexacanthus 2.70 Ch. sordidus Parrotfish 2.76 Sc. niger 3.00 Si. stellatus Rabbitfish

PC1 - Per cent variation explained 8.6%

Fig. 4 Principal coordinates analysis (PCoA) of fish gut bacterial community compositions based on the subsampled unweighted UniFrac distance matrix. The individual samples are colour-coordinated according to a) species, b) diet category (Choat et al. 2002) and c) trophic level (FishBase; www.fishbase.org). PC1 and PC2 explained 8.6% and 7.1% of the total variance, respectively. Pentagonal boxes are used for all nonsurgeonfishes, while square, triangular and circular boxes are used to denote species, diet category and tropic level, respectively, for surgeonfishes.

Discussion Key players in surgeonfish gut microbiota Firmicutes (as well as Bacteroidetes) are reported to be the dominant phylum in gut microbiota of most terres© 2014 John Wiley & Sons Ltd

trial vertebrates (Ley et al. 2008b), while Proteobacteria are reported to be the most abundant phylum in many freshwater or some marine fishes (Sullam et al. 2012). In marine fish, the dominance of Firmicutes (and specifically Clostridia) in the hindgut of three oceanic fish species (Kyphosus sydneyanus, Odax pullus and Aplodactylus

666 S . M I Y A K E , D . K . N G U G I and U . S T I N G L Table 2 ANOSIM of unweighted UniFrac distance against the host diet category Interactions

r

P value

All Carnivore–detritivore Carnivore–macroalgavore Carnivore–microalgavore Carnivore–omnivore Detritivore–macroalgavore Detritivore–microalgavore Detritivore–omnivore Macroalgavore–microalgavore Macroalgavore–omnivore Microalgavore–omnivore

0.74 0.56 0.80 0.61 0.15 0.97 0.81 0.30 0.72 0.87 0.71

Diet strongly influences the gut microbiota of surgeonfishes.

Intestinal tracts are among the most densely populated microbial ecosystems. Gut microbiota and their influence on the host have been well characteriz...
830KB Sizes 2 Downloads 7 Views