Access the full text.
Sign up today, get DeepDyve free for 14 days.
Background: Deciphering the mechanisms governing population genetic divergence and local adaptation across heterogeneous environments is a central theme in marine ecology and conservation. While population divergence and ecological adaptive potential are classically viewed at the genetic level, it has recently been argued that their microbiomes may also contribute to population genetic divergence. We explored whether this might be plausible along the well-described environmental gradient of the Baltic Sea in two species of sand lance (Ammodytes tobianus and Hyperoplus lanceolatus). Specifically, we assessed both their population genetic and gut microbial composition variation and investigated not only which environmental parameters correlate with the observed variation, but whether host genome also correlates with microbiome variation. Results: We found a clear genetic structure separating the high-salinity North Sea from the low-salinity Baltic Sea sand lances. The observed genetic divergence was not simply a function of isolation by distance, but correlated with environmental parameters, such as salinity, sea surface temperature, and, in the case of A. tobianus, possibly water microbiota. Furthermore, we detected two distinct genetic groups in Baltic A. tobianus that might represent sympatric spawning types. Investigation of possible drivers of gut microbiome composition variation revealed that host species identity was significantly correlated with the microbial community composition of the gut. A potential influence of host genetic factors on gut microbiome composition was further confirmed by the results of a constrained analysis of principal coordinates. The host genetic component was among the parameters that best explain observed variation in gut microbiome composition. Conclusions: Our findings have relevance for the population structure of two commercial species but also provide insights into potentially relevant genomic and microbial factors with regards to sand lance adaptation across the North Sea–Baltic Sea environmental gradient. Furthermore, our findings support the hypothesis that host genetics may play a role in regulating the gut microbiome at both the interspecific and intraspecific levels. As sequencing costs continue to drop, we anticipate that future studies that include full genome and microbiome sequencing will be able to explore the full relationship and its potential adaptive implications for these species. Keywords: Microbiome, Holobiome, Local adaptive potential, Population genomics, Sand lance, Baltic Sea * Correspondence: firstname.lastname@example.org; email@example.com Natural History Museum of Denmark, Section for Evolutionary Genomics, University of Copenhagen, Øster Voldgade 5-7, 1350 Copenhagen, Denmark Full list of author information is available at the end of the article © The Author(s). 2018 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated. Fietz et al. Microbiome (2018) 6:82 Page 2 of 16 Background We explored the potential of a host genomic-microbial A major current focus within marine ecology and con- approach by assessing both the population genetic and servation is to improve our understanding of the mecha- gut microbial variation in two sand lance species along nisms governing population genetic divergence and local the well-described environmental gradient in the Baltic adaptation across heterogeneous environments. Al- Sea. The Baltic Sea is a semi-enclosed brackish water though gene flow may hamper local adaptation, genetic basin in Northern Europe, which changes from a nearly outlier loci across environmental gradients in several limnetic to an almost fully marine environment. Fish marine fishes imply possible local adaptation despite low species, such as herring [1, 3, 11], cod [2, 12], and three- overall levels of population genetic divergence [1–3]. In spined stickleback , which have low levels of gen- a time during which the planet’s oceans are expected to omic divergence at neutral genetic markers, have been undergo considerable changes in oxygen, temperature, shown to exhibit substantial levels of divergence at some and salinity levels , leading to extensive changes in loci (single nucleotide polymorphisms (SNPs)), which the conditions of available habitats, an organism’s ability accordingly were inferred to be linked with genomic re- to adapt swiftly to these changes will be vital for its sur- gions under selection. The divergence at outlying SNPs vival. It is therefore particularly important to understand is likely the result of adaptations in marine species to which processes drive the genetic divergence that may the brackish conditions in the Baltic Sea after its forma- be at the basis of ecological adaptation, yet our under- tion as a marine habitat ca. 8000 years before present standing of these mechanisms is rudimentary. Studying [14–16], although other causes, such as a secondary con- the genetic basis of ecological adaptation in natural pop- tact zone, cannot be ruled out . ulations is particularly difficult when population sizes Five species of sand lances (fishes of the family are limited, because in such cases, random effects dom- Ammodytidae) occur at high abundances in the North- inate over deterministic effects and will prevent the pos- east Atlantic and adjacent waters. Sand lances are known sibility of selection to act. In the marine realm, however, to be closely associated with specific soft substrates  there are many species whose populations have high and characterized by high levels of residency and short abundances and span different ecological conditions so dispersal ranges [19–21], which likely restrict long- that natural selection is expected to dominate over ran- distance gene flow in the absence of physical barriers. dom effects [5, 6]. Accordingly, a number of studies have Sand lances are keystone organisms as a prey for a large begun to study adaptive genetic variation across environ- number of marine birds, mammals, and other fish spe- mental gradients. However, while this growing body of cies [22–24]. Sand lances are also targeted by commer- research correlates outlier loci signatures with key envir- cial fisheries and thus represent a considerable economic onmental parameters, such as salinity and water resource . Although these characteristics render temperature [2, 3, 6, 7], very few studies have gone be- them both relevant to the study in the context of marine yond standard environmental parameters. management and as an interesting potential model or- Although the debate about adaptation to different envi- ganism for studying local adaptation, previous research ronments is classically viewed at the genetic level, it has re- has principally focused on defining sand lance species cently been argued that an organism’sassociated and populations using few genetic markers [26–30]. microbiome might also play a role . This argument is We generated genome-wide SNP data and gut micro- nested within the holobiome concept, which views an or- biome taxonomic composition data for two ecologically ganism as an entity encompassing not only its own but also and economically important sand lance species, Ammo- its microbial symbionts’ genetic information [9, 10]. Specif- dytes tobianus and Hyperoplus lanceolatus, along the ically regarding adaptation, Alberdi et al. argued that, given North Sea–Baltic Sea environmental gradient. We used (i) gut microbiome communities can have significant and the data to firstly estimate the levels of population differ- rapid phenotypic effects on their hosts and (ii) the relatively entiation across this gradient. Secondly, we tested if the short time frame of many environmental changes, micro- observed population genetic structure correlated with biome community changes may provide an important environmental factors, including salinity and sea surface mechanism for adaptation. While challenging to study temperature (SST), as well as the relative bacterial com- directly, the first steps in this direction can come from position in the water at sampling sites. Thirdly, we char- assessing the variation in host species’ microbial communi- acterized each species’ gut microbiome, its inter-specific ties against the host species genomic divergence and variation, and its relationship with both environmental environmental gradients. Although some studies are parameters as well as host genomic divergence. Lastly, considering host genomic-microbial relationships, and even in light of our findings, we discuss the future potential environmental-microbial relationships, few studies so far of how a hologenomic approach may add significantly to have attempted to take a full hologenomic approach (both our understanding of marine species’ ecological adaptive genomic and microbial) across environmental variation. potential. Fietz et al. Microbiome (2018) 6:82 Page 3 of 16 Methods the gut microbiome analysis were collected from the Sample collection frontal gut located directly behind the stomach and Sand lances were collected at multiple sites across the ending in a tight loop in the gut (hereafter referred to as environmental gradient from the brackish Inner Baltic gut). The external part of the gut was cleaned with Sea to the marine North Sea. Samples from A. tobianus sterile equipment to remove any tissue, and the gut wall and H. lanceolatus were collected during May through including contents was used as a sample. Gut samples September 2015 from 12 and 15 different sites, respect- were stored in 96% ethanol at − 20 C° until further ively (Fig. 1, Table 1). Individual fishes were collected processing. during commercial and research trawls or caught with near-shore seine nets. Individual muscle tissue samples Environmental data were collected upon capture (seine nets) or 1–6 h (com- We collected data on salinity, SST, and sea water mercial trawls) post-mortem. If no direct sampling was bacterial taxonomic composition to assess the correl- feasible, individuals were stored in 96% ethanol upon ation between genetic data and environmental variables capture and stored at − 20 C° until sub-sampling. Guts (Additional file 1: Table S1). Salinity and SST data were were collected under sterile conditions upon capture retrieved from www.smhi.se and www.ices.dk as the an- from a subset of individuals. The samples collected for nual average salinity during the period 2010 to 2014, as well as the annual, minimum, and maximum average SST during the same period. Data of the relative abun- dance of six major bacterial taxa in the water column near our sampling sites were obtained from Hu et al. . Although these data were collected in 2013, the long residence time of water in the Baltic basin (3– 30 years ) and identical sampling season imply that these data likely are broadly representative of the water at our sampling sites. Genotyping-by-sequencing (GBS) DNA extraction and sequencing library preparation Whole-cell genomic DNA was extracted using the KingFisher™ Duo Prime Purification System (Thermo Fisher Scientific Inc., Waltham, USA) following the manu- facturer’s protocol for the KingFisher™ Cell and Tissue DNA purification kit (Thermo Fisher Scientific Inc., Wal- tham, USA). The DNA concentration was estimated using a Qubit™ 2.0 fluorometer (Thermo Fisher Scientific Inc., Waltham, USA). The fragment size range of DNA extrac- tions was estimated for a subset of DNA extractions using an Agilent 4200 TapeStation™ (Agilent Inc.). DNA extrac- tions were subsequently cleaned using the ZR-96 Gen- omic DNA Clean & Concentrator™ (Zymo Research Inc., Orange, CA, USA) following the manufacturer’sprotocol. Population genomic data were generated from the DNA extracts following the GBS approach originally developed by Elshire et al.  at the Institute of Biotechnology com- mercial service (Cornell University, NY, USA) following Fig. 1 Sampling sites for A. tobianus (above) and H. lanceolatus (below). Pie charts represent genetic ancestry proportions per their standard pipeline . The genomic DNA extracts sampling site as estimated in ADMIXTURE v.1.3.0 for K =3 (A. tobianus) were digested with the DNA restriction endonuclease and K =2 (H. lanceolatus). Pie charts encircled in blue indicate sites EcoT221, which has a six-base pair (bp) recognition se- from which we included 16S data in addition to GBS data. Sampling quence, and fragments in the size range from 200 to sites for H. lanceolatus for which we had < 8 individuals are 380 bps were used as the basis for the GBS libraries. indicated as hollow circles. TE Texel, SA W-Sylt A, SB W-Sylt B, SHB SW-Hanstholm B, DB Doggerbanke, TB Tannisbugt, HR Horns Rev, SHA SW-Hanstholm A, NWH NW-Hanstholm, LA Læsø, EB Ebeltoft, GBS sequencing and SNP calling HB Hornbæk, HØ Helsingør, HK Halsskov, MB Musholm Bugt, KB GBS libraries were sequenced at the Institute of Biotech- Køge Bugt, FB Faxe Bugt, BH Bornholm, ÅL Åland, BӦ Bönan nology commercial service as single-end 64 bp using an Fietz et al. Microbiome (2018) 6:82 Page 4 of 16 Table 1 Overview of tissue samples for GBS analyses, gut samples for 16S amplicon analyses, and environmental parameters; proportions of major bacterial taxa in the water column are stated as percentage (%) of total composition per sample Fish Sampling Sampling GBS samples Gut samples Mean # of Min # of Max # of Annual av. Annual av. Annual min Annual max SST_Coeff_ AP VM BC BP AB GP species site date (n) (n) OTUs OTUs OTUs sal (PSU) SST (°C) SST (°C) SST (°C) var (C°) AT TE 10.07.15 8 33.47 9.58 − 1.5 21.7 23.2 SB 31.07.15 19 31.40 9.90 − 1.1 21.7 22.8 HR 05.–06.06.15 15 33.58 12.31 0.3 21.6 21.3 LA 29.08.15 27 23.77 9.50 − 1.2 21.9 23.1 27.0 1.6 24.9 2.9 3.3 14.8 EB 26.09.15 24 5 14.4 10 22 20.73 11.26 − 0.7 23.4 24.1 29.3 2.7 24.4 3.1 3.9 16.5 HB 03.07.15 29 19.44 11.17 − 0.8 22.6 23.4 29.3 2.7 24.4 3.1 3.9 16.5 HK 12.08.15 27 10 16.6 9 26 16.54 10.60 − 0.5 23.5 23.9 KB 14.07.15 26 9.89 9.90 − 0.2 24.6 24.9 9.9 20.3 13.3 2.9 15.6 2.6 FB 16.07.15 30 7 21.3 8 35 8.67 9.50 − 0.3 24.8 25.1 9.9 20.3 13.3 2.9 15.6 2.6 BH 16.07.15 18 7.27 8.48 0.3 22.1 21.8 9.9 27.7 9.5 11.9 10.8 5.0 ÅL 09.09.15 33 5.83 10.41 − 0.3 23.6 23.9 10.6 16.4 16.6 6.0 18.7 2.9 BӦ 18.09.15 30 9 40 8 55 5.35 8.18 − 0.2 23.4 23.6 10.1 15.1 17.7 10.0 18.7 4.3 HL TE 10.07.15 8 33.47 9.58 − 1.5 21.7 23.2 SA 03.07.15 17 31.40 9.90 − 1.1 21.7 22.8 HR 05.–06.06.15 28 33.58 12.31 0.3 21.6 21.3 SHB 23.08.15 1 SHA 25.–27.06.15. 14 10 17.1 7 34 33.90 14.23 1 20.4 19.4 NWH 30.04.15 17 33.18 10.08 − 0.4 23.2 23.6 DB 08.07.15 2 TB 24.08.15 1 LA 29.08.15 1 MB 01.07.15 2 HØ 09.09.15 29 9 11.7 5 31 14.29 10.51 − 0.5 22.3 22.7 FB 16.07.15 25 8.67 9.50 − 0.3 24.8 25.1 BH 16.07.15 25 7.27 8.48 0.3 22.1 21.8 ÅL 09.09.15 1 BӦ 18.09.15 1 For sampling site abbreviations, see Fig. 1 AT A. tobianus, HL H. lanceolatus, OTU operational taxonomic unit, AB Actinobacteria, AP Alphaproteobacteria, BC Bacteroidetes, BP Betaproteobacteria, GP Gammaproteobacteria, VM Verrucomicrobia Fietz et al. Microbiome (2018) 6:82 Page 5 of 16 Illumina HiSeq2000™ (Illumina Inc., San Diego, US). Library preparation and amplicon sequencing Subsequent analytical steps were conducted separately We employed a two-step PCR amplification approach for each sand lance species. Details of data quality filter- for microbial 16S library preparation. The V3-V4- ing and SNP calling are described in Additional file 1. regions of the bacterial 16S rRNA gene were amplified by PCR using the primers 341F (5′-CCTAYGGGRBG- CASCAG-3) and 806R (5′-GGACTACNNGGGTATC- Population genomic analyses TAAT-3) . A subsequent PCR amplification was We employed the AMOVA  implemented in GEN- performed with the Nextera™ XT index primers (Illu- ODIVE v.2.0 to estimate overall and pairwise levels of mina Inc., San Diego, US) in order to attach Illumina genetic divergence as F . We assumed an infinite alleles ST MiSeq™ sequence adapters and barcodes to each DNA model  and employed 999 permutations to estimate extract. Full details of 16S library preparation and se- the probability of homogeneity. In order to further as- quencing are listed in Additional file 1. In the following, sess population genetic structure, we conducted a princi- the term microbiome refers to the data obtained from pal component analysis (PCA) using the SMARTPCA the 16S-based libraries. program in the EIGENSOFT package . The datasets were reduced to ten eigenvectors, and the principal Data filtration and operational taxonomic unit (OTU) components 1 and 2 as well as 1 and 3 were plotted clustering using the Perl script PLOTEIG (EIGENSOFT package). In We performed data filtration and clustering in USEARCH order to infer ancestry among sand lances in various v.8.1.1861 . Details of data filtration and SNP calling areas, we used the model-based approach implemented can be found in Additional file 1. in the software ADMIXTURE v.1.3.0 . ADMIXTURE esti- mations were performed for values of K ranging from 2 Gut microbiome taxonomic composition to 14. Convergence was assumed when the log- −4 We employed the QIIME (Quantitative Insights into Mi- likelihood difference among iterations was < 10 .We crobial Ecology v.1.8.0 ) bioinformatic pipeline to es- employed the fivefold cross-validation approach to select timate the α diversity within the sampling sites as well the most probable estimate of K . The ADMIXTURE as the Shannon-Wiener  and Chao1 indices . analysis was undertaken both with and without remov- The differences in OTU frequencies among sampling ing loci that deviated significantly from the expected sites were assessed using an ANOVA and corrected for Hardy-Weinberg genotype frequencies (HWE) under multiple simultaneous comparisons using a step-down random mating. resampling algorithm described by Westfall and Young . We report all bacteria with P < 0.05 at the lowest Detection of outlier loci possible taxonomic level (genus being the lowest level). We applied three different Bayesian approaches to detect We further estimated β diversity among sampling sites SNPs deviating from neutral expectations and to assess by quantifying the degree of dissimilarities in micro- the degree of correlations with environmental parame- biome composition among sites by a principal coordin- ters. We employed the F -based approach implemented ST ate analysis (PCoA) in which we employed a weighted in BAYESCAN v.2.1  to identify outlier loci. In order UniFrac distance matrix to account for OTU abundance to test for associations between population genetic diver- and phylogenetic ancestry . We tested for significant gence and environmental parameters, we also employed differences in microbiome composition among sampling two approaches implemented in BAYESCENV v.1.1  sites at the family level using a non-parametric Kruskal- and BAYENV V.2 . Details of the estimations are listed Wallis H test  and applied the Benjamini-Hochberg in Additional file 1. In addition, allele frequencies were False Discovery Rate (FDR) correction to adjust for mul- plotted for outlier loci in order to assess the spatial tiple simultaneous tests . cline. Predictors of gut microbiome composition Microbial 16S profiling In order to determine which factors correlate with DNA extraction and purification changes in gut microbiome composition, we imple- Total-cell DNA was extracted from the gut samples mented various approaches in the R packages vegan  using the MoBio Power Soil kit™ (MoBio Laboratories and phyloseq . In the case of A. tobianus, we also in- Inc., Carlsbad, USA) following the manufacturer’s in- cluded the relative abundances of major bacterial taxa structions. DNA concentrations were quantified using a found in the Baltic water as independent parameters in Qubit™ 2.0 Fluorometer (Thermo Fisher Scientific Inc., addition to salinity and SST. Waltham, USA) and normalized to a final DNA concen- We employed a permutational multivariate analysis of tration at 10 ng/μL. variance (Permanova) in the R package vegan  using Fietz et al. Microbiome (2018) 6:82 Page 6 of 16 a distance matrix based on Bray-Curtis dissimilarity to maximum possible number of individuals in analyses. identify each variable that has a significant influence on Following data filtration, we were left with a final dataset our dataset variation. Results were then plotted by non- consisting of 4039 SNPs and an overall genotyping rate metric multidimensional scaling (NMDS) analysis. A of 0.976 for A. tobianus (n = 286) and with 4328 SNPs biologically plausible combination of significant environ- and an overall genotyping rate of 0.980 for H. lanceola- mental parameters based on knowledge of the system tus (n = 163). All analyses were conducted for both spe- was fitted onto our unconstrained NMDS ordination. cies, unless otherwise indicated. We then used a constrained analysis of principal coor- dinates (CAP) ordination implemented in the R package Population genomic divergence phyloseq  to assess which combination of independ- The global F was estimated at 0.012 (95% CI 0.011–0. ST ent variables explained the largest proportion of the vari- 013) in the case of A. tobianus, and at 0.017 (95% CI 0. ance in gut microbiome composition. To this end, we 015–0.018) in the case of H. lanceolatus. Pairwise F ST included those environmental parameters that proved estimates among samples of A. tobianus ranged from 0. significant in the Permanova described above. Variables 001 to 0.041; the highest estimate was observed between were added to the model in order of explanatory the most brackish and the marine sampling sites. The variance. In both the NMDS and CAP ordination, collin- degree of genetic divergence between the geographically earity among variables was accounted for by excluding intermediate sampling sites and between the brackish variables varying in a linear manner with variables and marine sampling sites, respectively, ranged from 0. already added to the model. 011 to 0.030 (Additional file 1: Table S2a). The overall pattern of genetic divergence was similar for H. lanceo- Assessing the impact of environment versus host genetics latus, with pairwise F estimates ranging from 0.002 to ST upon the gut microbiome composition 0.039. The highest degree of genetic divergence was ob- We employed an interspecific dataset including the gut served between the Inner Baltic Sea and the North Sea microbiome data collected from A. tobianus and H. lan- sampling sites (Additional file 1: Table S2b). ceolatus, as well as from a number of Baltic fish species The most probable number of genetic clusters was es- sampled in Køge Bugt during 2016 (listed in Add- timated at three and two for A. tobianus and H. lanceo- itional file 1: Figure S5). The inclusion of additional fish latus, respectively (Figs. 1 and 2, Additional file 1: Table species served as a reference to contrast the influence of S3). The outcome was identical with or without removal environmental factors with the influence of host genotype of SNPs deviating from HWE (to keep consistent with on gut microbiome composition. We tested influence of other studies that used the UNEAK pipeline, we con- the host species identity with a Permanova and a subse- tinue with data that included the HWE filtering step). quent CAP ordination with the R package phyloseq . The clustering based upon the PCA yielded the same We used microbial composition data collected along a outcome (Additional file 1: Figure S2). 2000-km transect in the Baltic Sea during 2013 reported Overall, we found that the clustering in the case of ei- by  to estimate the degree of correlation between the ther sand lance species corresponded well with environ- sand lance gut microbiome and the microbial composition mental “regimes.” Ammodytes tobianus from marine in the water column. We focused our analysis on the rela- sampling sites belonged to a different genetic cluster tive abundances per sample of non-normalized reads ob- than individuals from Baltic Sea sampling sites (K =3 in tained from six bacterial taxa at the phylum and class Fig. 2). In the Inner Baltic Sea, A. tobianus consisted of level, namely, Alphaproteobacteria, Verrucomicrobia, Bac- two genetically distinct clusters. Hyperoplus lanceolatus teroidetes, Betaproteobacteria, Actinobacteria, and Gam- shows a similar pattern: individuals in the North Sea maproteobacteria. We used a χ test of homogeneity sampling sites had pure marine ancestry, while individ- implemented in R  to assess the statistical significance uals at the two Inner Baltic sites had pure Baltic Sea an- of the observed differences in relative bacterial abundance cestry (K = 2 in Fig. 2). In the following, we will refer to among sampling sites. the different clusters as “populations.” Results Loci under putative local adaptation Genotyping-by-sequencing (GBS) Our analyses aimed at detecting outlier loci identified For A. tobianus, no individuals were genotyped at more numerous SNPs where the spatial change in allele than 20% of the overall identified loci, while no H. frequencies among sampling sites correlated with the lanceolatus were genotyped at more than 30% of loci gradient in the environmental parameters included in (Additional file 1: Figure S1). We excluded all individuals our analysis. A total of 43 and 72 SNPs were identified that had more than 90% missing data in either species to as outlier loci in all three of the approaches we applied keep filtering criteria consistent and to include the in A. tobianus and H. lanceolatus, respectively (Fig. 3a, Fietz et al. Microbiome (2018) 6:82 Page 7 of 16 Fig. 2 Plots of ancestral fractions from the ADMIXTURE cluster analysis for both sand lance species (top = A. tobianus; bottom = H. lanceolatus). Each vertical bar represents one individual, while the colors indicate the likelihood of this individual belonging to a particular ancestral population. K refers to the number of ancestral populations that were assumed to be present in the dataset. K = *indicates the most likely K. The asterisks in the sampling site heading of H. lanceolatus indicates fish from sampling sites with < 8 individuals that were removed for population-level analyses. Samples are sorted from the North Sea (left) to the Baltic Sea (right) sampling sites Additional file 1: Table S4). Of these, the allele frequen- which a total of 210 OTUs were detected. We identified cies at 22 and 29 SNPs in A. tobianus and H. lanceola- 107 different OTUs among 19 H. lanceolatus gut micro- tus, respectively, also changed along the environmental biome samples from two sampling sites (Table 1). The gradient, adding further support to the hypothesis that read depths per sample ranged between 1051 and 25,352 these loci might be subject to selection by factors co- in A. tobianus and between 1348 and 12,492 in H. lan- varying with the environment (Fig. 3b, Additional file 1: ceolatus (Additional file 1: Table S5). Both extraction Figure S3). Although we acknowledge that a clinal pat- and PCR negative control samples resulted in very low tern such as this could also be expected for a number of read coverage (≤ 315) suggesting that contamination was neutral loci, we propose that the allele frequency plots negligible and were hence excluded from the final presented here represent additional evidence in support dataset. of the findings of the outlier analyses. Most of these 22 outlier SNPs in A. tobianus were correlated with the Sand lance gut microbiome composition relative proportions of three major bacterial taxa found At phylum level, the gut microbiome composition of A. in the water (Actinobacteria, Alphaproteobacteria, and tobianus and H. lanceolatus was similar and varied only Gammaproteobacteria) as well as with the minimum slightly between H. lanceolatus and A. tobianus (Fig. 4a). and maximum SST (Additional file 1: Table S4a). The Proteobacteria was the only phylum present in all indi- allele frequencies at ten outlier SNPs correlated with viduals of both species. The phyla Tenericutes, Cyano- the change in salinity, while the allele frequency bacteria, Firmicutes, Actinobacteria, Bacteroidetes, and change in seven SNPs correlated significantly with the Spirochaetes were present in a large percentage of change in nearly all environmental parameters. In H. individuals in both species. The largest difference was lanceolatus, 12 of the 29 SNPs were correlated with observed in Verrucomicrobia, which was only present in all tested environmental parameters (salinity and SST) 5% of H. lanceolatus individuals but in 45% of the A. (Additional file 1:Table S4b). tobianus individuals. Alpha gut-microbiome diversity was higher overall in Microbial 16S profiling A. tobianus (Chao1 = 8–101; Shannon = 2.89–5.63) com- The final microbial dataset consisted of 31 A. tobianus pared to H. lanceolatus (Chao1 = 5–35.6; Shannon = 2. gut microbiome samples from four sampling sites among 21–4.96). Both the Shannon-Wiener and the Chao1 Fietz et al. Microbiome (2018) 6:82 Page 8 of 16 Fig. 3 a Venn diagram displaying outlier SNPs identified with different outlier detection software (A. tobianus =left, H. lanceolatus = right). b Allele frequency plots for a subset of four candidate SNPs for divergent selection, as well as sampling-site-specific values for annual average salinity andSST indices were higher in A. tobianus gut samples collected while for H. lanceolatus, the first axis explained 30. at brackish sampling sites compared to marine sites. No 5% of variation (Fig. 5). In A. tobianus,the levelof such difference between brackish and marine sites was variation among gut samples from the same sam- detected between H. lanceolatus gut samples (Fig. 4b). pling site was smaller than that among sites with the The relative abundance of four bacterial genera was sig- exception of Faxe Bugt where three data points nificantly different among A. tobianus sampling sites. In group with individuals from other sites. In H. lan- comparison, the relative abundance of seven bacterial ceolatus, the level of variation within and between genera was significantly different between the two H. the two sites was similar. lanceolatus sampling sites (P < 0.05) (Fig. 4c). Two of these genera belonged to the obligate phototrophic Predictors of gut microbiome composition Synechococcus. It is worth noting that only two of these We tested a range of environmental and host genetic genera (Synechococcus and Shewanella) overlapped be- factors to assess how well they explained the observed tween sand lance species. variance in sand lance gut microbiome composition. The In the PCoA, the first principal coordinate axis absence of microbial data from the water column at the explained 35.9% of variation in gut microbial com- Halsskov sampling site necessitated the exclusion of this munities between sampling sites for A. tobianus, site from the Permanova. We defined the ancestry Fietz et al. Microbiome (2018) 6:82 Page 9 of 16 microbiome composition in A. tobianus and H. lanceola- tus, respectively (P < 0.001) (Additional file 1: Table S6). The data was visualized in a two-dimensional NMDS or- dination to assess possible multivariate interaction of the gut microbiome composition with the environmental pa- rameters (Additional file 1: Figure S4). The main axis among which A. tobianus sampling sites were separated was characterized by differences in salinity, SST, and the relative abundances of four water bacterial taxa. In H. lanceolatus, the gut microbiome among sampling sites was not as clearly differentiated, the main axis being characterized by differences in SST, distance, and date, describing the same variation in opposite directions (Additional file 1: Figure S4, Table S6). We used CAP ordination to test for statistical signifi- cance in the multivariate interactions among environ- mental parameters and gut microbiome composition. In A. tobianus, the combination of variables accounting for a significant proportion of the variance in the gut micro- biome composition included SST, geographic distance from the westernmost sampling site, and the host gen- etic component Q (CAP1 = 26.7%) (Fig. 6). The one- dimensional ordination in H. lanceolatus explained 19. 8% of the variance (results not shown). Impact of environment versus host genetics on the gut microbiome composition We conducted a CAP ordination on a multi-specific dataset in order to extend our understanding of the po- tential influence of host species factors on gut micro- biome composition. This revealed that the host species identity was a key explanatory variable for gut micro- biome composition (Permanova P < 0.001; PC1 = 6.5%, PC2 = 5.5%) (Additional file 1: Figure S5). We then dis- played the gut microbiome composition of multiple spe- cies from three adjacent sampling sites to illustrate the different scales of similarity in gut microbiome compos- ition among species and sampling sites (Additional file 1: Figure S6). We therefore chose A. tobianus samples from two sites 131 km apart and samples from an out- Fig. 4 Composition, diversity, and differentiation of A. tobianus and H. lanceolatus gut microbial communities. a Bar plot depicting the group of fishes (sand goby (Pomatoschistus minutus), percentage of individuals with the occurrence of the bacterial phyla flounder (Platichthys flesus), stickleback (Gasterosteus found in A. tobianus and H. lanceolatus guts. b Alpha diversity aculeatus)) from a geographically intermediate sampling (Chao1 Index left, Shannon-Wiener Index right) increased significantly site (Køge Bugt). Our hypothesis was that gut micro- with decreasing salinity level (indicated as an arrow) in A. tobianus, biome composition would be more similar within than while no clear trend is observed in H. lanceolatus. c Heat map of OTUs that changed significantly in relative abundance (%) as a function of between host species, even if individuals of the same sampling site at genus level (A. tobianus and H. lanceolatus combined) species were sampled at different sites. With the excep- tion of a single individual (ZMUC P611035, from Faxe fraction Q of the North Sea group as identified by AD- Bugt), the gut microbiome composition was visibly less MIXTURE for the most likely K as the host genetic compo- variable among A. tobianus across sites than it was nent for either species. We included this host genetic among A. tobianus and other fish species from sites that component as an “environmental parameter” in our ana- are geographically closer to each other. lyses. The Permanova identified ten and three environ- Correlations between gut microbiome composition mental variables correlating significantly with the gut and relative abundance trends of bacterial taxa in the Fietz et al. Microbiome (2018) 6:82 Page 10 of 16 Fig. 5 Principal coordinate analysis (PCoA) of the dissimilarity between sand lance gut microbial communities in different sampling sites for A. tobianus (left) and H. lanceolatus (right). Color of the circles denotes the sampling site environment revealed significant changes in the relative in the environment, whereas the opposite trend was ob- abundances of some of the major bacterial taxa along served in Actinobacteria (Fig. 7). the North Sea–Baltic Sea environmental gradient (Fig. 7, Additional file 1: Table S7). While some bacterial taxa Discussion demonstrated identical abundance trends in gut and en- Our findings not only have relevance for the popula- vironment, others showed opposing relative abundance tion structure of two commercial species, but also trends. Specifically, trends were identical in Gammapro- provide insights into potentially relevant genomic teobacteria which became proportionately less dominant and microbial factors with regards to sand lance in both environmental (water) and gut samples as envir- adaptation across the North Sea–Baltic Sea environ- onmental salinity decreased (Fig. 7). In Alphaproteobac- mental gradient. Furthermore, our findings provide teria, relative abundance increased in guts and decreased evidence that sand lance gut microbial communities are influenced by both host genetics and environ- mental parameters. Isolation of Baltic Sea sand lances The general level of population genetic divergence ob- served in sand lance was similar to those previously ob- served in other marine fishes (e.g., [2, 54]). In both sand lance species, the highest degree of genetic divergence was observed between the sampling sites in the North Sea and the Inner Baltic Sea, similar to the population genetic divergence observed in, e.g., Atlantic cod , herring , flounder , and sprat . The propor- tional change in genetic divergence was highest between sampling sites located in the areas with the steepest change in environmental parameters, such as salinity. Similar qualitative changes in genetic divergence have been reported previously in several other Baltic Sea fish species (e.g., Fig. 2 in ). Sand lances in the North Sea tend to be resident and associated with specific habitat Fig. 6 Results of CAP analyses displaying the combination of environmental parameters explaining the largest amount of variation types . Consequently, most dispersal is likely to in gut bacterial communities in A. tobianus. The shape of symbols occur during the larval phase, but even this is thought to indicates the sampling site while the color of symbols indicates be limited [20, 21]. Assuming that these dispersal char- the salinity acteristics are shared with Baltic Sea sand lances, Fietz et al. Microbiome (2018) 6:82 Page 11 of 16 Fig. 7 Relative abundance (of non-normalized read numbers) of the six most abundant bacterial taxa at phylum and order level of A. tobianus guts (above) and in environmental (below) water samples  they match the genetic divergence between these investigated A. tobianus in the Southern Baltic Sea in two seas. the Gulf of Gdansk and based on vertebral counts assigned the individuals of this area to the autumn- Ammodytes tobianus in the Baltic Sea consists of two spawning component of the stock. While we did not find genetically differentiated populations significant differences in vertebral counts in our fishes Ammodytes tobianus in the Baltic Sea—mainly between that would support a hypothesis of different spawning Køge Bugt and Åland in the Western to Inner Baltic types (data not shown), we hypothesize that the two Sea—seems to belong to two genetically differentiated genetic A. tobianus Baltic populations that we detected populations. We suggest that this observation may be here may nonetheless represent two sympatric spawning attributed to either spatial or temporal segregation. types. In order to gain certainty about co-occurrence of Spatial segregation would occur if different breeding different spawning types in the Baltic Sea, future studies stocks left their spawning areas and mixed at sampling will have to sample individuals during the respective sites outside the spawning season. As we did not sample spawning season (autumn and spring) and investigate during the spawning season, we cannot exclude this op- the presence of ripe gonads in adult individuals. The tion. We find spatial segregation unlikely, however, given presence of different spawning types in the same habitat the sand lances’ residential behavior and limited disper- is also known from other Baltic species such as herring sal capacity [20, 21]. Another possible cause for the ob- [11, 56]. Our results highlight the power of using genetic served genetic pattern may be temporal segregation. markers as a tool to monitor the relative proportion of Temporal segregation would occur if different spawning the two spawning types in fishery landings. types were sympatric but reproduced at different times . Populations of A. tobianus may indeed consist of Sand lances along the North Sea–Baltic Sea two distinct but often sympatric spawning types, an au- environmental gradient display signatures tumn and a spring spawning contingent  that are consistent with local adaptation known to occur together, e.g., off the coast of West As a geologically young sea that has undergone extreme Ireland . These spawning types generally differ from changes in environmental conditions in the last 8000 years, each other in the mean number of vertebrates which is the Baltic Sea has served as a popular model to assess higher in the autumn group . As early as 1934, it divergence along an environmental gradient in marine was suggested that A. tobianus in the Baltic Sea consists organisms . Our study supports previous work hy- of two spawning types . More recently,  pothesizing that marine fish populations show signatures Fietz et al. Microbiome (2018) 6:82 Page 12 of 16 of potential local adaptation along the Baltic Sea–North basic level, the sand lance gut microbial community Sea gradient [2, 63, 64]. Given the relatively small number composition was comparable to that reported in other of SNPs, the lack of a reference genome, and the fact that fishes [68–70]. Members of the Proteobacteria were the we aimed to exercise great care to avoid over-interpreting most common phylum observed in the gut microbiome our results, we refrained from blasting outlier loci and in both sand lance species. Among other common mi- chose to instead focus on drawing conclusions from the crobial phyla were Actinobacteria, Tenericutes, and observed correlations and allele frequency plots. In both Cyanobacteria. Proteobacteria are one of the major phyla A. tobianus and H. lanceolatus, we found elevated levels in the guts of many other studied fish species (as are of genetic divergence at loci that correlated with SST and Actinobacteria, Tenericutes, and Cyanobacteria) and are salinity (we discuss the relationship with the microbial found in high abundance in fish guts [68–71]. The mi- communities in the water column below). Fishes have crobial taxa observed in the sand lance guts are taxa that evolved physiological mechanisms that differ fundamen- aid in nutrient absorption and homeostasis . For tally between high- and low-saline environments in order example, Proteobacteria help degrade and ferment to maintain an internal salinity at 9 PSU . Divergent complex sugars. Actinobacteria assist in maintaining selection, especially at loci that are associated with salinity host homeostasis and in inhibition of Gram-negative tolerance, may therefore be expected to promote repro- pathogens and lactic acid fermentation. Tenericutes aid ductive isolation between Baltic Sea and North Sea popu- nutrient processing . lations. In a study on Atlantic cod, Berg and colleagues Microbial diversity estimates in both sand lance spe- identified outlier SNPs in the cod genome where the cies varied substantially among individuals. The micro- population genetic divergence correlated with a salinity bial diversity was overall higher in A. tobianus compared gradient as well. The divergent SNPs were located within to H. lanceolatus. Although gut microbial diversity is genes that were known to be associated with osmoregula- likely correlated with diet [73, 74], we did not include tion and oocyte development. Accordingly, argued that dietary analyses in our study. The relationship between the divergent alleles were the result of selection to a low- the diet and gut microbial flora is not simple: some au- saline environment. The results of acclimation experi- thors found that a more diverse diet will increase gut ments and molecular phenotyping in tilapia (Oreochromis microbial diversity , whereas others have reported an mossambicus), a euryhaline cichlid, led  to infer that inverse relationship in some fishes . The more gen- osmoregulatory stress could affect gill development. Here, eral observation is that diversity of fish gut microbial we hypothesize that similar mechanisms related to physi- communities likely is influenced by multiple environ- ology, egg, and larval survival in low-saline waters likely mental variables. Our results suggest that the diversity of are the underlying causes of the elevated levels of popula- the gut microbiome increases with decreasing salinity in tion genetic divergence between sand lances in the Baltic A. tobianus. However, inclusion of diet analysis and add- Sea and North Sea. Ambient temperature is important in itional spatio-temporal replicates in future studies will affecting metabolic reactions especially of poikilotherm or- help us better understand the processes shaping this ganisms such as fishes. Allele frequencies in a set of pattern. genome-wide SNP loci, for example, showed parallel temperature-associated clines in Atlantic cod on either side of the Atlantic Ocean . The observation that the Microbial communities of the sand lance gut are not degree of population genetic divergence in multiple of the mere reflections of environmental microbial communities sand lance outlier loci was significantly associated with Numerous studies across a wide range of organisms, salinity and water temperature led us to the hypothesis from ants overfishes through humans [78, that environmental heterogeneity in these, or correlated, 79], have shown that the composition of the core gut factors may be an important driving force behind the microbiome is not a simple function of the environ- genetic divergence between the Baltic and North Sea ment but correlates with the host genetics as well. sand lance populations. We acknowledge that spatial The gut microbial composition may be a product of replicates will be needed to be able to exclude the host phylogenetic affinities and/or the host’secology, possibility that the observed pattern is purely a result both of which are likely to determine the mode of of thenatureofthe data. microbiome acquisition, i.e., by vertical transmission and/or from the environment . Gut microbiome composition and diversity We assessed the potential differences among fish spe- We incorporated data of gut microbiome composition cies and their gut microbial community composition in from a subset of specimen in each sand lance species to order to gain insights into possible drivers of the gut explore associations between host genomics and envir- microbiome composition. Our analysis indicated non- onment upon the gut microbiome composition. At a random differences in gut microbiome composition Fietz et al. Microbiome (2018) 6:82 Page 13 of 16 among different fish species. However, our analysis did gut microbiome alone and/or with additional environ- not account for variation in microbial communities mental parameters, ultimately fully linking the system among sampling sites. Consequently, we cannot attribute together. The combination of in situ work and common the observed differences to key and specific variables, garden experiments seems particularly promising for such as spatial heterogeneity, fish age, or prey. In order moving beyond the detection of correlations and to as- to account for this caveat, we visualized the gut micro- sess the causal roles of microbiome and environmental biome composition of A. tobianus from a subset of sites, factors in shaping the adaptive potential of wild verte- in addition to samples from other fish species from a brate species. However, taken together, our results sup- geographically intermediate site. As expected, the intra- port the notion that the gut microbial community specific gut microbial composition was more similar composition is, in part, a function of the host’s genomic within species compared to that among species background. (Additional file 1: Figure S6). Our results suggest that a potential influence of host genetic factors on gut micro- The potential of combining population genomics with biome composition depends on the bacterial taxon gut microbial data to gain insight to an organism’s under investigation. The host genetic component Q was ecological adaptive potential among the parameters that best explained observed vari- It has recently been suggested that in order to cope with ation in gut microbiome composition. On the other rapidly changing environmental conditions, organisms hand, the presence of obligate photoautotrophic bacteria may rely on high phenomic plasticity conferred by their (Synechococcus) suggests at least partial bacterial uptake gut bacterial symbionts . Changes at the genetic level from the environment. In our comparison of relative that bring about evolutionary change often need many abundance changes of major bacterial taxa between fish generations to be established in a population. For verte- guts and surrounding water along the North Sea– Baltic brates, with their slow reproductive strategies and long Sea environmental gradient, Gammaproteobacteria generation times, this can be insufficient for short-term showed identical relative abundance trends in guts and adaptation and survival. Phenotypic plasticity may hence water. This implies that the relative abundance of Gam- be an essential factor in determining how well verte- maproteobacteria in sand lance guts may be driven by brates adapt to fast environmental change . Among their relative abundance in the surrounding environ- host-associated microorganisms, the bacteria of the gut ment. In Alphaproteobacteria and Actinobacteria on the are thought to be the most influential symbiotic commu- other hand, gut-bacterial relative abundance trends were nity, affecting health, immunity, digestive metabolism, converse to those of the surrounding water, suggesting and consequently fitness of their hosts [80–82]. If indeed that relative abundances in these taxa might depend an organism’s adaptive potential may be enhanced more strongly on host species. through help of its gut microbiome, it stands much bet- Last, we expanded our environmental dataset in the ter chances of survival when faced with the need to population-genomic outlier analyses of A. tobianus to in- adapt swiftly. Given this putative central role of the gut clude—besides SST and salinity—the proportional abun- microbiome, knowledge of such communities is a vital dance of major microbial taxa in the gut microbial addition to population genomics in studies of local adap- samples. The relative abundance of Actinobacteria, tation [72, 83]. Recent research investigating the genetic Alphaproteobacteria, and Gammaproteobacteria corre- basis of microbes (the “who”) hypothesizes that bacteria lated significantly with nearly all detected outlier loci. might confer certain abilities or tolerances to their hosts. Actinobacteria are known to be part of the gut, mucosa, However, only very few studies have explicitly tested and skin in fishes, while Proteobacteria are thought to whether certain bacteria actually do confer such abilities be the dominant phylum in the guts of many fishes. or tolerances (the “how” and “why”). In fishes, Among the Proteobacteria, Gammaproteobacteria typic- microbiome research has mainly focused on lab-reared ally break down and ferment complex sugars and pro- [71, 85, 86], captured , or aquacultured fish , vide particularly important digestive roles . The while to date there is only a limited number of studies results of our outlier analysis may indicate that bacteria on wild populations and only on few economically im- adopted from the environment play important roles in a portant species such as salmon . From few studies population’s adaptability to its local habitat. involving wild populations, we know that inferences However, our study design prevented us from identify- about the gut microbiome made from captive animals ing functional aspects, e.g., how salinity and SST corre- may not always be transferred to their wild counterparts lated with the microbial community in the water column [70, 88]. In this sense, studies of wild non-model organ- in the Baltic. Controlled laboratory experiments would isms are invaluable to gain a better understanding of be needed in future studies to differentiate between cor- how the gut microbiome and host act together as an en- relations of environmental bacteria and an organism’s tity in facilitating rapid ecological adaptation. In this Fietz et al. Microbiome (2018) 6:82 Page 14 of 16 study, we for the first time use both population genetic Additional file and gut microbial data to make inferences about the Additional file 1: Supplementary Material. (DOCX 13909 kb) population genetic divergence of two fish species in their natural habitat. Our study may serve as a benchmark for Acknowledgements future work that aims to integrate population genomic We are most grateful to Morten Tange Olsen for helpful discussions and with gut microbial data to investigate an organism’s eco- comments on previous manuscript versions. Thanks to George A. Pacheco logical adaptive potential. for his assistance with the laboratory work. We would also like to thank the Cornell University Biotechnology Resource Center for its genotyping services, in particular Sharon E. Mitchell and all involved laboratory technicians. We Conclusions further thank Filipe J.G. Viera and Shyam Gopalakrishnan for useful discussions In this study, we take a new approach by using popula- and support with the data analyses. Thanks are also due to Henrik Carl, Michelle Svendson, Felipe Torquato, Fredrik Landfors, Tore Holm-Hansen, Kristian Vedel, tion genomic and 16S gut microbial data to shed light Lara Puetz, Chris Höhne, Andrea-Pil Holm, Hannah Jensen, Flora Laughier, Kay onto the population genetic divergence of two closely re- Panten, Mikkel Holger Strander Sinding, and Mads Holger Strander Sinding for lated non-model fish species along an environmental support in sample collection. Finally, we most kindly wish to thank the fishermen Hans Holger Strander Petersen, Bo Nielsen and crew, Lars, gradient. Three major findings emerge from our study: Kristian and Jens Olsen, and Kenneth Søbye for providing samples. First, the Baltic Sea harbors unique genetic populations of sand lances that are differentiated from the North Funding Sea. We discovered that genomic regions showed The authors acknowledge the Universities of Copenhagen and Groningen for providing a PhD scholarship to KF and the University of Copenhagen EU elevated divergence not only as a potential response to Bonus award to MTPG for funding the research costs. salinity- and SST-related natural selection, but that these regions also correlate with the relative bacterial compos- Availability of data and materials All genetic and environmental raw data are uploaded as supplementary material ition of the water. This could hint at a potential influ- or in ERDA: http://www.erda.dk/public/archives/YXJjaGl2ZS1NZmpvVU4=/ ence of environmental microbes on the adaptive genetic published-archive.html divergence of these marine fishes. Second, we confirmed Authors’ contributions that A. tobianus in the Baltic Sea exists as two genetic KF, CORH, PRM, and MTPG designed the research. KF, CORH, MK, and PRM stocks co-occurring in the same habitat. This sparks collected samples. KF and CORH performed the laboratory work. KF, CORH, interest in adapting future fishery management mea- TKN, MS, and MTL performed the analyses. KF drafted the manuscript. PJP, LHH, PRM, and MTPG provided funding. All authors commented on previous sures. Third, the gut microbial communities of sand versions of the manuscript and approved the final version. The work is part lances are not a mere reflection of environmental of KF’s PhD thesis which is supervised by MTPG and PJP. microbes, but rather the fishes seem to excerpt some degree of internal control and selection. Ethics approval and consent to participate Not applicable Recent insights into the extent of microbial influence on an organism’s well-being and fitness suggest that the Competing interests potential of individuals, populations, and species to gen- The authors declare that they have no competing interests. etically adapt to changing environments may not be gov- erned by the interactions of these entities with the Publisher’sNote Springer Nature remains neutral with regard to jurisdictional claims in published environment alone. Rather, bacteria seem to partake in maps and institutional affiliations. forming the adaptive potential of many organisms by composing part of their holobiome. In this sense, popu- Author details Natural History Museum of Denmark, Section for Evolutionary Genomics, lation genomic studies aimed at understanding a species’ University of Copenhagen, Øster Voldgade 5-7, 1350 Copenhagen, Denmark. local adaptive potential will miss part of the story when Marine Evolution and Conservation, Faculty of Science and Engineering, not considering bacterial communities. University of Groningen, Nijenborgh 7, 9747 AG Groningen, The Netherlands. Department of Environmental Science, Environmental Microbial Genomics We are only beginning to understand how the inter- Group, Aarhus University, Frederiksborgvej 399, 4000 Roskilde, Denmark. play among hosts, their microbes, and the environment NTNU University Museum, 7491 Trondheim, Norway. regulates ecological adaptation. One can imagine that in Received: 10 July 2017 Accepted: 26 March 2018 the future, as genomic technologies become cheaper and more readily applicable, our study may serve as a bench- mark to design yet more rigorous approaches to learn References what enables an organism to survive and adapt in a rap- 1. Guo BC, Li ZT, Merila J. Population genomic evidence for adaptive differentiation in the Baltic Sea herring. Mol Ecol. 2016;25:2833–52. idly changing environment. The combination of in situ 2. Berg PR, Jentoft S, Star B, Ring KH, Knutsen H, Lien S, Jakobsen KS, Andre C. work and common garden experiments seems particu- Adaptation to low salinity promotes genomic divergence in Atlantic cod larly promising for moving beyond the detection of cor- (Gadus morhua L.). Genome Biol Evol. 2015;7:1644–63. 3. Limborg MT, Helyar SJ, de Bruyn M, Taylor MI, Nielsen EE, Ogden R, relations and to assess the causal roles of bacterial and Carvalho GR, Bekkevold D, Consortium FPT. Environmental selection on environmental factors in shaping the adaptive potential transcriptome-derived SNPs in a high gene flow marine fish, the Atlantic of wild vertebrate species. herring (Clupea harengus). Mol Ecol. 2012;21:3686–703. Fietz et al. Microbiome (2018) 6:82 Page 15 of 16 4. IPCC. In: Pachauri RK, Meyer LA, editors. Climate change 2014: synthesis report. 28. Han Z, Yanagimoto T, Zhang Y, Gao T. Phylogeography study of Contribution of working groups I, II and III to the fifth assessment report of the Ammodytes personatus in Northwestern Pacific: Pleistocene isolation, intergovernmental panel on climate change. Geneva: The Intergovernmental temperature and current conducted secondary contact. PLoS One. 2012;7: Panel on Climate Change; 2014. p. 151. e37425. 5. Nielsen EE, Hemmer-Hansen J, Larsen PF, Bekkevold D. Population genomics 29. Warner T. En multidiciplinær undersøgelse af fire sameksisterende tobisarter of marine fishes: identifying adaptive variation in space and time. Mol Ecol. (Ammodytes tobianus, Ammodytes marinus, Hyperoplus lanceolatus og 2009;18:3128–50. Gymnammodytes semisquamatus) ved Horns Rev i Nordseøen. Copenhagen: 6. Martinez Barrio A, Lamichhaney S, Fan G, Rafati N, Pettersson M, Zhang H, University of Copenhagen, Natural History Museum of Denmark; 2011. 30. Orr JW, Wildes S, Kai Y, Raring N, Nakabo T, Katugin O, Guyon J. Systematics Dainat J, Ekman D, Hoppner M, Jern P, et al. The genetic basis for ecological of North Pacific sand lances of the genus Ammodytes based on molecular adaptation of the Atlantic herring revealed by genome sequencing. elife. 2016;5:1–32. and morphological evidence, with the description of a new species from Japan. Fish Bull. 2015;113:129–56. 7. Vilas R, Vandamme SG, Vera M, Souza C, Maes GE, Volckaert FAM, Martinez P. A genome scan for candidate genes involved in the adaptation of turbot 31. Hu YOO, Karlson B, Charvet S, Andersson AF. Diversity of Pico- to (Scophthalmus maximus). Mar Genomics. 2015;23:77–86. Mesoplankton along the 2000 km salinity gradient of the Baltic Sea. Front 8. Alberdi A, Aizpurua O, Bohmann K, Zepeda-Mendoza ML, Gilbert MTP. Do Microbiol. 2016;7:679. vertebrate gut metagenomes confer rapid ecological adaptation? Trends 32. Reissmann J, Burchard H, Feistel R, Hgen E, Lass HU, Mohrholz V. State-of- Ecol Evol. 2016;31:689–99. the-art review on vertical mixing in the Baltic Sea and consequences for 9. Rosenberg E, Zilber-Rosenberg I. Symbiosis and development: the eutrophication. Prog Oceanogr. 2009;82:47–80. hologenome concept. Birth Defects Res C Embryo Today. 2011;93:56–66. 33. Elshire RJ, Glaubitz JC, Sun Q, Poland JA, Kawamoto K, Buckler ES, Mitchell SE. A robust, simple genotyping-by-sequencing (GBS) approach for high 10. Bordenstein SR, Theis KR. Host biology in light of the microbiome: ten diversity species. PLoS One. 2011;6:e19379. principles of holobionts and hologenomes. PLoS Biol. 2015;13:e1002226. 34. Excoffier L, Smouse PE, Quattro JM. Analysis of molecular variance inferred 11. Lamichhaney S, Martinez Barrio A, Rafati N, Sundstrom G, Rubin CJ, Gilbert from metric distances among DNA haplotypes - application to human ER, Berglund J, Wetterbom A, Laikre L, Webster MT, et al. Population-scale mitochondrial-DNA restriction data. Genetics. 1992;131:479–91. sequencing reveals genetic differentiation due to local adaptation in Atlantic herring. Proc Natl Acad Sci U S A. 2012;109:19345–50. 35. Kimura M, Crow JF. The number of alleles that can be maintained in a finite 12. Nielsen EE, Hemmer-Hansen J, Poulsen NA, Loeschcke V, Moen T, Johansen population. Genetics. 1964;49:725–38. T, Mittelholzer C, Taranger G-L, Ogden R, Carvalho GR. Genomic signatures 36. Patterson N, Price AL, Reich D. Population structure and eigenanalysis. PLoS of local directional selection in a high gene flow marine organism; the Genet. 2006;2:2074–93. Atlantic cod (Gadus morhua). BMC Evol Biol. 2009;9:276. 37. Alexander DH, Novembre J, Lange K. Fast model-based estimation of 13. Guo BC, DeFaveri J, Sotelo G, Nair A, Merila J. Population genomic evidence ancestry in unrelated individuals. Genome Res. 2009;19:1655–64. for adaptive differentiation in Baltic Sea three-spined sticklebacks. BMC Biol. 38. Alexander DH, Lange K. Enhancements to the ADMIXTURE algorithm for 2015;13:19. individual ancestry estimation. BMC Bioinformatics. 2011;12:246. 39. Foll M, Gaggiotti O. A genome-scan method to identify selected loci 14. Bjorck S. The late Quaternary development of the Baltic Sea. In: Team A, appropriate for both dominant and codominant markers: a Bayesian editor. Assessment of climate change for the Baltic Sea Basin. Berlin Heidelberg: Springer Verlag; 2008. p. 398–407. perspective. Genetics. 2008;180:977–93. 15. Ojaveer E, Kalejs M. The impact of climate change on the adaptation of 40. de Villemereuil P, Gaggiotti OE. A new FST-based method to uncover local marine fish in the Baltic Sea. ICES J Mar Sci. 2005;62:1492–500. adaptation using environmental variables. Methods Ecol Evol. 2015;6:1248–58. 16. Emeis KC, Struck U, Blanz T, Kohly A, Voss M. Salinity changes in the central 41. Gunther T, Coop G. Robust identification of local adaptation from allele Baltic Sea (NW Europe) over the last 10 000 years. The Holocene. 2003;13: frequencies. Genetics. 2013;195:205-+. 411–21. 42. Hansen CHF, Krych L, Nielsen DS, Vogensen FK, Hansen LH, Sorensen SJ, 17. Gaggiotti OE, Bekkevold D, Jorgensen HBH, Foll M, Carvalho GR, Andre C, Buschard K, Hansen AK. Early life treatment with vancomycin propagates Ruzzante DE. Disentangling the effects of evolutionary, demographic, and Akkermansia muciniphila and reduces diabetes incidence in the NOD environmental factors influencing genetic structure of natural populations: mouse. Diabetologia. 2012;55:2285–94. Atlantic herring as a case study. Evolution. 2009;63:2939–51. 43. Edgar RC. Search and clustering orders of magnitude faster than BLAST. Bioinformatics. 2010;26:2460–1. 18. Nelson JS, Grande TC, Wilson MVH. Fishes of the world. 5th ed. New Jersey: Wiley; 2016. 44. Caporaso JG, Kuczynski J, Stombaugh J, Bittinger K, Bushman FD, Costello 19. Gauld JA. Movements of lesser sandeels (Ammodytes marinus Raitt) tagged EK, Fierer N, Pena AG, Goodrich JK, Gordon JI, et al. QIIME allows analysis of in the Northwestern North Sea. J Conseil. 1990;46:229–31. high-throughput community sequencing data. Nat Methods. 2010;7:335–6. 20. Proctor R, Wright PJ, Everitt A. Modelling the transport of larval sand eels on 45. Shannon CE. A mathematical theory of communication. Urbana: University the north-west European shelf. Fish Oceanogr. 1998;7:347–54. of Illinois Press; 1948. 21. Christensen A, Mosegaard H, Jensen H. Spatially resolved fish population 46. Chao A, Shen TJ. Nonparametric estimation of Shannon's index of diversity analysis for designing MPAs: influence on inside and neighbouring habitats. when there are unseen species in sample. Environ Ecol Stat. 2003;10:429–43. ICES J Mar Sci. 2009;66:56–63. 47. Westfall PH, Young SS. Resampling-based multiple testing. New York: Wiley; 1993. 22. Rindorf A, Wanless S, Harris MP. Effects of changes in sandeel availability on the reproductive output of seabirds. Mar Ecol Prog Ser. 2000;202:241–52. 48. Lozupone C, Lladser ME, Knights D, Stombaugh J, Knight R. UniFrac: an 23. Frederiksen M, Edwards M, Richardson AJ, Halliday NC, Wanless S. From effective distance metric for microbial community comparison. ISME J. 2011; plankton to top predators: bottom-up control of a marine food web across 5:169–72. four trophic levels. J Anim Ecol. 2006;75:1259–68. 49. Kruskal WH, Wallis WA. Use of ranks in one-criterion variance analysis. J Am 24. Furness RW. Management implications of interactions between fisheries and Stat Assoc. 1952;47:583–621. sandeel-dependent seabirds and seals in the North Sea. ICES J Mar Sci. 50. Benjamini Y, Hochberg Y. Controlling the false discovery rate—a practical 2002;59:261–9. and powerful approach to multiple testing. J R Stat Soc Ser B Methodol. 25. Anderson J, Carvalho N, Contini F, Virtanen J. In: Scientific TaECfFS, editor. 1995;57:289–300. The 2012 annual economic report on the EU fishing fleet (STECF-12-10). 51. Oksanen J, Blanchet FG, Friendly M, Kindt R, Legendre P, McGlinn D, Luxembourg: European Commission; 2012. Minchin PR, O'Hara RB, Simpson GL, Solymos P, et al: vegan: community ecology package., R package version 2.4–1; 2016. 26. Thiel R, Knebelsberger T. How reliably can northeast Atlantic sand lances of the 52. McMurdie PJ, Holmes S. phyloseq: an R package for reproducible interactive genera Ammodytes and Hyperoplus be distinguished? A comparative application of morphological and molecular methods. ZooKeys. 2016;617:139–64. analysis and graphics of microbiome census data. PLoS One. 2013;8:e61217. 27. Kim JK, Watson W, Hyde J, Lo N, Kim JY, Kim S, Kim YS. Molecular 53. R Development Core Team. R: a language and environment for statistical identification of Ammodytes (PISCES, Ammodytidae) larvae, with computing. Vienna: R Foundation for Statistical Computing; 2015. ontogenetic evidence on separating populations. Genes Genomics. 2010;32: 54. Limborg MT, Pedersen JS, Hemmer-Hansen J, Tomkiewicz J, Bekkevold D. 437–45. Genetic population structure of European sprat Sprattus sprattus: Fietz et al. Microbiome (2018) 6:82 Page 16 of 16 differentiation across a steep environmental gradient in a small pelagic fish. 78. Goodrich JK, Waters JL, Poole AC, Sutter JL, Koren O, Blekhman R, Beaumont Mar Ecol Prog Ser. 2009;379:213–24. M, Van Treuren W, Knight R, Bell JT, et al. Human genetics shape the gut 55. Florin AB, Hoglund J. Population structure of flounder (Platichthys flesus) in microbiome. Cell. 2014;159:789–99. the Baltic Sea: differences among demersal and pelagic spawners. Heredity. 79. Blekhman R, Goodrich JK, Huang K, Sun Q, Bukowski R, Bell JT, Spector TD, 2008;101:27–38. Keinan A, Ley RE, Gevers D, Clark AG. Host genetic variation impacts microbiome composition across human body sites. Genome Biol. 2015;16:191. 56. Jørgensen HBH, Hansen MM, Loeschcke V. Spring-spawning herring (Clupea harengus L.) in the southwestern Baltic Sea: do they form genetically 80. Tremaroli V, Backhed F. Functional interactions between the gut microbiota distinct spawning waves? ICES J Mar Sci. 2005;62:1065–75. and host metabolism. Nature. 2012;489:242–9. 81. Vatanen T, Kostic Aleksandar D, d’Hennezel E, Siljander H, Franzosa Eric A, 57. Kandler R. Untersuchungen über Fortpflanzung, Wachstum und Variabilität Yassour M, Kolde R, Vlamakis H, Arthur Timothy D, Hämäläinen A-M, et al. der Arten des Sandaals in Ost- und Nordsee, mit besonderer Variation in microbiome LPS immunogenicity contributes to autoimmunity Berücksichtigung von Ammodytes tobianus. L Kieler Meeres. 1941;5:45–139. in humans. Cell. 2016;165:842–53. 58. Oconnell M, Fives JM. The biology of the lesser sand-eel Ammodytes tobianus 82. Amato KR, Yeoman CJ, Kent A, Righini N, Carbonero F, Estrada A, Gaskins L in the Galway Bay area. Biol Environ Proc R Ir Acad. 1995;95B:87–98. HR, Stumpf RM, Yildirim S, Torralba M, et al. Habitat degradation impacts 59. Whitehead PJP, Bauchot M-L, Hureau J-C, Nielsen J, Tortonese E. Fishes of black howler monkey (Alouatta pigra) gastrointestinal microbiomes. ISME J. the North-eastern Atlantic and the Mediterranean. Paris: Unesco; 1986. 2013;7:1344–53. 60. Bahr K. Der kleine Sandaal (Ammodytes tobianus L.) der Ostsee. Zeitschr 83. Benson AK. The gut microbiome—an emerging complex trait. Nat Genet. Fisch u d Hilfswiss. 1934;33 2016;48:1301–2. 61. Wiecaszek B, Krzykawski S, Antoszek A. Meristic and morphometric 84. Kohl KD, Stengel A, Dearing MD. Inoculation of tannin-degrading bacteria characters of small sandeel, Ammodytes tonianus L. (Actinopterygii: into novel hosts increases performance on tannin-rich diets. Environ Ammodytidae), from the Gulf of Gdansk, Baltic Sea. Acta Ichthyol Piscat. Microbiol. 2016;18:1720–9. 2007;37:37–45. 85. Rawls JF, Mahowald MA, Ley RE, Gordon JI. Reciprocal gut microbiota 62. Johannesson K, Andre C. Life on the margin: genetic isolation and diversity transplants from zebrafish and mice to germ-free recipients reveal host loss in a peripheral marine ecosystem, the Baltic Sea. Mol Ecol. 2006;15: habitat selection. Cell. 2006;127:423–33. 2013–29. 86. Rawls JF, Samuel BS, Gordon JI. Gnotobiotic zebrafish reveal evolutionarily 63. Corander J, Majander KK, Cheng L, Merila J. High degree of cryptic conserved responses to the gut microbiota. Proc Natl Acad Sci U S A. 2004; population differentiation in the Baltic Sea herring Clupea harengus. Mol 101:4596–601. Ecol. 2013;22:2931–40. 87. Huber I, Spanggaard B, Appel KF, Rossen L, Nielsen T, Gram L. Phylogenetic 64. Jørgensen HBH, Pertoldi C, Hansen MM, Ruzzante DE, Loeschcke V. Genetic analysis and in situ identification of the intestinal microbial community of and environmental correlates of morphological variation in a marine fish: rainbow trout (Oncorhynchus mykiss, Walbaum). J Appl Microbiol. 2004;96: the case of Baltic Sea herring (Clupea harengus). Can J Fish Aquat Sci. 2008; 117–32. 65:389–400. 88. Eichmiller JJ, Hamilton MJ, Staley C, Sadowsky MJ, Sorensen PW. 65. Kultz D. Physiological mechanisms used by fish to cope with salinity stress. J Environment shapes the fecal microbiome of invasive carp species. Exp Biol. 2015;218:1907–14. Microbiome. 2016;4:44. 66. Kultz D, Li J, Gardell A, Sacchi R. Quantitative molecular phenotyping of gill remodeling in a cichlid fish responding to salinity stress. Mol Cell Proteomics. 2013;12:3962–75. 67. Bradbury IR, Hubert S, Higgins B, Borza T, Bowman S, Paterson IG, Snelgrove PVR, Morris CJ, Gregory RS, Hardie DC, et al. Parallel adaptive evolution of Atlantic cod on both sides of the Atlantic Ocean in response to temperature. Proc R Soc B Biol Sci. 2010;277:3725–34. 68. Givens CE, Ransom B, Bano N, Hollibaugh JT. Comparison of the gut microbiomes of 12 bony fish and 3 shark species. Mar Ecol Prog Ser. 2015; 518:209–23. 69. Llewellyn MS, McPGinnity P, Dionne M, Letourneau J, Thonier F, Carvalho GR, Creer S, Derome N. The biogeography of the Atlantic salmon (Salmo salar) gut microbiome. ISME J. 2016;10:1280–4. 70. Sullam KE, Rubin BER, Dalton CM, Kilham SS, Flecker AS, Russell JA. Divergence across diet, time and populations rules out parallel evolution in the gut microbiomes of Trinidadian guppies. ISME J. 2015;9:1508–22. 71. Di Maiuta N, Schwarzentruber P, Schenker M, Schoelkopf J. Microbial population dynamics in the faeces of wood-eating loricariid catfishes. Lett Appl Microbiol. 2013;56:401–7. 72. Colston TJ, Jackson CR. Microbiome evolution along divergent branches of the vertebrate tree of life: what is known and unknown. Mol Ecol. 2016;25: 3776–800. 73. Ley RE, Hamady M, Lozupone C, Turnbaugh PJ, Ramey RR, Bircher JS, Schlegel ML, Tucker TA, Schrenzel MD, Knight R, Gordon JI. Evolution of Submit your next manuscript to BioMed Central mammals and their gut microbes. Science. 2008;320:1647–51. 74. Bolnick DI, Snowberg LK, Hirsch PE, Lauber CL, Knight R, Caporaso JG, and we will help you at every step: Svanback R. Individuals’ diet diversity influences gut microbial diversity in • We accept pre-submission inquiries two freshwater fish (threespine stickleback and Eurasian perch). Ecol Lett. 2014;17:979–87. � Our selector tool helps you to ﬁnd the most relevant journal 75. Laparra JM, Sanz Y. Interactions of gut microbiota with functional food � We provide round the clock customer support components and nutraceuticals. Pharmacol Res. 2010;61:219–25. � Convenient online submission 76. Sanders JG, Powell S, Kronauer DJC, Vasconcelos HL, Frederickson ME, Pierce NE. Stability and phylogenetic correlation in gut microbiota: lessons from � Thorough peer review ants and apes. Mol Ecol. 2014;23:1268–83. � Inclusion in PubMed and all major indexing services 77. Li XM, Zhu YJ, Yan QY, Ringo E, Yang DG. Do the intestinal microbiotas � Maximum visibility for your research differ between paddlefish (Polyodon spathala) and bighead carp (Aristichthys nobilis) reared in the same pond? J Appl Microbiol. 2014;117: Submit your manuscript at 1245–52. www.biomedcentral.com/submit
Microbiome – Springer Journals
Published: May 2, 2018
Access the full text.
Sign up today, get DeepDyve free for 14 days.