Comparative characterization of bacterial communities in geese fed all-grass or high-grain diets

Background Gut microbial composition is dependent on diet. Geese are herbivores and can digest crude fibre, but the relationship between composition of the microbiota and a fibre-rich diet in geese is not well understood. Results Here, caecal and faecal samples were collected simultaneously from all-grass-fed geese and high-grain-fed geese and the hypervariable V3–V4 regions of the bacterial 16S rRNA gene were sequenced. The results was identified that high-grass-fed geese possessed significantly higher alpha diversity both in caecum and faeces compared with that in all-grain-fed geese. In addition, the composition of dominant bacterium occurred remarkable shifting due to different diet patterns, Firmicutes were more abundant in all-grass-fed geese, whereas Bacteroidetes were abundant in high-grain-fed geese. Fusobacteria and Deferribacteres were obviously present in high-grain-fed geese and few in all-grass-fed geese. Most importantly, some specific microorgnisms such as Ruminococcaceae, Lachnospiraceae and Bacteroidaceae which may associated with cellulose-degrading that were characterized to show distinctly diverse between the two diet patterns. PICRUSt analysis revealed the metabolic pathways such as carbohydrate and amino acid metabolism were overrepresented in all-grass-fed geese. Conclusions In conclusion, Firmicutes and Bacteroidetes were identified abundantly when the geese was fed with all-grass feed and high-grain feed, respectively. And Ruminococcaceae, Lachnospiraceae and Bacteroidaceae were recognized as main cellulose-degrading bacteria in the geese. The functional profiles of gut microbiota revealed the dominant microbiota communities were involved mainly in the carbohydrate metabolism in all-grass-fed geese.


Introduction
The gut microbiota is the community of microorganisms that colonize the gut of a host organism. Microorganisms colonize the gut via a very complex process of interactions between the microorganisms and the host. The host provides a place for the survival and evolution of microorganisms, and in return, the microbiota benefits the host by providing essential nutrients, as well as stimulating growth performance [1,2]. The mammalian gut microbiota has been widely surveyed, and its diversity, structure, and function have been found to be mainly shaped by adaptation to diet [3][4][5]. Highly diverse cellulolytic obligate anaerobes, such as lineages within Bacteroidales, Fibrobacterales, and Spirochaetales colonize the gastrointestinal tracts of most herbivores, providing microbial fermentation to enhance nutrient absorption [6,7]. Previous research has shown that, the improvements in cellulose degradation would have a favourable impact on animal productivity in herbivores. Ruminococcus and Fibrobacter are important members of the rumen microbial community that enable the host to degrade and utilize fibrous plant materials efficiently as nutrients [8][9][10].
The goose (Anas cygnoides) is a commercially important food source that is widely cultivated in China [11]. Under intensive systems, geese are fed diets consisting of large proportions of grains to support rapid weight gain. Although these feeding practices may be helpful in enhancing cost-efficiency in the short term, it is unclear whether such intensive systems affect intestinal microbial structure or gastrointestinal health. After all, geese are basically herbivores under natural conditions.
It is well known that the goose has a strong ability to digest crude fibre [12][13][14]. However, the goose is genetically deficient in cellulose-digesting enzymes, and digests fibre mainly by microorganism fermentation in the caecum [15][16][17]. Therefore, it is important to study the diverse bacterial community to reveal the molecular mechanisms underlying this strong ability of the goose to digest crude fibre. Liu [15] analysed the goose caecal bacterial community by denaturing gradient gel electrophoresis (DGGE), and observed an increased diversity in caecal microbiota with an increase in dietary crude fibre levels. They further found that uncultured Clostridiaceae sp., uncultured Treponema sp., Cellulomonas sp., uncultured Bacteroides sp., and uncultured Eubacteriaceae bacterium were the major cellulosedegrading bacteria [15]. However, the DGGE approach only isolates small DNA fragments, and this might have resulted in the loss of some useful sequence information. Furthermore, this approach usually displays DNA fragments from dominant communities due to its low resolution. Recently, with the development of next-generation sequencing technologies, 16S rRNA gene amplicon deep sequencing has been applied widely to investigate the microbiota [1,18,19].
In the present study, we investigated the effects of different diets (all-grass diet and highgrain diet) on the microbiota of geese, using paired-end MiSeq sequencing to characterize the bacterial community of the caecum and faeces. These data may identify bacterial taxa that are associated with the digestion of crude fibre.

Ethics statement
The animal experiment was reviewed and approved by the Institutional Animal Care and Use Committee of Yangzhou University (approval number:151-2014). Procedures were performed in accordance with the Regulations for the Administration of Affairs Concerning Experimental Animals (Yangzhou University, China, 2012) and the Standards for the Administration of Experimental Practices (Jiangsu, China, 2008).

Animals and sample collection
A total of forty 35-day-old Yangzhou goslings weighing an average of 1404.05 g were selected from the breeding farm of the Yangzhou Goose Co. Ltd (Yangzhou, China). The goslings were randomly assigned to two groups. One group was fed by rotation grazing on a pasture of perennial ryegrass and Chinese trumpet creeper (220 g crude fibre/kg DM), and the other group was fed a high-grain diet consisting mainly of maize and soybean meal (100 g crude fibre/kg DM, produced by Changzhou Chia Tai Co., Ltd). During the experiment, geese were fed ad libitum with the grass diet or the high-grain diet. The experimental period lasted 5 weeks, the body weight was shown in S6 Table, the body weight of 70-day-old geese on the high-grain diet was higher than that of geese on the grass diet (P<0.05). The female geese from two groups were placed individually in a cage and their defecation observed from a distance. Once the faeces were excreted from the goose, the faece was selected immediately from the central portion of each cage to minimize contamination. After collection, samples were stored on ice until frozen at -80˚C.
Following this, the same geese were slaughtered immediately by anesthetizing them with sodium pentobarbital for intestinal sampling. The abdominal cavity was opened by a midline incision, and the intestinal tract was removed carefully. A segment of the caecum was tied at both ends and the contents collected on ice, and then stored at -80˚C until DNA extraction.

DNA extraction and PCR amplification of 16S rRNA gene
Microbial genomic DNA was extracted from each samples using the E.Z.N.A. 1 DNA Kit (Omega Bio-tek, Norcross, GA, U.S.) according to the manufacturer's protocols. DNA extraction was carried out using the bead-beating method, with a mini-bead beater. The frozen samples were placed into bead tubes, and then added 1ml Buffer of SLX Mlus to the samples, incubated at 70˚C for 10min. The samples were beaten for 90 s at 5000 rpm (maximum speed) in a Mini-Beadbeater TM (Biospec Products Inc., Bartlesville, OK). The solution was precipitated with ethanol, and the pellets were suspended in 50 μL Tris-EDTA buffer. The DNA concentration was determined using a NanoDrop spectrophotometer (NanoDrop, USA).

Illumina MiSeq sequencing
The amplicons were examined on a 2% (w/v) agarose gel. The band was extracted and purified with the AxyPrepDNA Gel (Axygen, USA) according to the manufacturer's instructions, and quantified using QuantiFluor™uantiFluorrding to the manufaamplicons were pooled in equimolar amounts and paired-end sequenced (2uenced (2end sIllumina MiSeq platform (Majorbio, China) according to the standard protocols.

Processing and bioinformatics analysing of sequencing data
Raw FASTQ files were de-multiplexed and quality-filtered using QIIME (version1.17) with the following criteria [21] (i) The 300-bp reads were truncated at any site that obtained an average quality score of <20 over a 50-bp sliding window, and truncated reads shorter than 50 bp were discarded. (ii) Only overlapping sequences longer than 10 bp were assembled according to their overlapped sequence. Reads that could not be merged were discarded. (iii) Reads with exact barcode matching, no more than two-nucleotide mismatch in primer matching were left, while reads containing ambiguous characters were removed. OTUs at or above the 97% similarity cutoff were clustered using UPARSE software (version 7.1, http://drive5.com/uparse) [22]. Taxonomic classification of OTUs was carried out using the Ribosomal Database Project (RDP) Classifier [23]. Rarefaction analysis based on MOTHUR (version 1.30.1) was conducted to reveal alpha diversity indices, including the Chao, ACE, and Shannon diversity indices [24]. Beta diversity analysis was performed using UniFrac to compare the results from principal component analysis (PCA) [25]. Linear discriminant analysis (LDA) coupled with effect size (LEfSe) was performed to identify bacterial taxa that were differentially represented between groups at the genus level or higher taxonomic levels [26], and only taxa meeting an LDA significance threshold ! 2 and the parameter of LEfSe for p valuevalue owere adopted. The functional profiles of microbial communities were predicted using PICRUSt [27]. The R packages 'heatmap' and 'vegan' were used for redundancy analysis (RDA) and plotting (including generation of the PCA and heat map figures) [28]. To compare bacterial communities of different geese, We downloaded and analysed the data of goose microbiota from NCBI (ERR011376). The sequences satisfied the inclusion criteria of sequence length>500 bp, and data sets containing>50 sequences.

Statistical analysis
Species richness and alpha diversity were analysed by two-way analysis of variance (ANOVA). Each individual OTU was analysed for effects of diet by ANOVA. The ANOVA analysis was performed using SPSS 17.0 software. P<0.05 was considered statistically significant. Differentially abundant OTUs were analysed using Metastats software [29].

General DNA sequencing observations
To characterize the bacterial lineages present in the faecal and caecal microbiota of all-grassfed geese and high-grain-fed geese, we performed Illumina sequencing of the V3gquehypervariable regions of 16S rRNA gene with the PE300 platform. We generated a dataset consisting of 495,646 filtered, high-quality, classifiable 16S rRNA gene sequences with a mean average (eanith a30,977±5,218 sequences per sample (S1 and S2 Tables). The sequence data file was submitted to the Short read archive at NCBI (Accession NO. SRR3711176). The quality-filtered sequences were clustered OTUs at the 97% similarity cut-off. The singletons were removed, and the OTUs including three sequences were retained. A total of 391,736 reads were assigned to 933 non-singleton OTUs at the 97% similarity cut-off, which were assigned into 149 taxa (genus level). The rarefaction curves tended to approach the saturation plateau (Fig 1). Good's coverage estimates indicated that a large part of the diversity in all samples had been captured, with the average coverage being 97.06%(S.D.1.47%).

Microbial richness and biodiversity
We then compared the microbial α-diversity in faecal and caecal samples between the allgrass-fed and high-grain-fed groups, including microbial richness estimated by the ACE and Chao1 indices, and diversity assessed by the Shannon and Simpson indices ( Table 1). At the 0.03 dissimilarity level, we found significant differences in richness between different diet-fed geese in both faecal and caecal samples (P<0.01 or P<0.05), with higher microbial richness in all-grass-fed geese. The caecum exhibited significantly high microbial diversity from the indices of Shannon and Simpson when compared with the faeces (P<0.05).
To compare community compositions across the different samples, The unweighted Uni-Frac metric was calculated to assess β-diversity by MOTHUR software. The results indicated that unweighted UniFrac was able to distinctly identify caecum microbiota from geese fed the two different diets (Fig 2(A)), and showed little differentiation in faecal samples (Fig 2(B)). Principal coordinate analysis revealed that principal component analysis (PCA) axis 1 accounted for 55.91% of the variation and PCA axis 2 for 14.11% of the variation in the caecum.

Differences in caecal bacterial communities between all-grass-fed geese and high-grain-fed geese
On comparing the caecal microbiota in geese receiving different diets, more than 80% of the sequences in all-grass-fed geese and high-grain-fed geese were found to belong to the two  most populated bacterial phyla, namely Bacteroidetes and Firmicutes (Fig 3). However, the phylum Firmicutes exhibited higher abundance in the caecal samples of all-grass-fed geese (56.44% versus 31.18%, P<0.05), whereas Bacteroidetes were more abundant in high-grainfed geese than in all-grass-fed geese (51.14% versus 34.69%, P>0.05). In addition, Fusobacteria (6.75%) and Deferribacteres (2.15%) was present in the high-grain-fed geese, while those were almost absent in all-grass-fed geese( Table 2). Interestingly, at the family level, the most highly abundance of the microbiota was Ruminococcaceae (Firmicutes) and Bacteroidaceae (Bacteroidetes) in all-grass-fed geese and high-grain-fed geese, respectively (Fig 4).  Comparative characterization of bacterial communities in geese fed all-grass or high-grain diets Although a little difference was seen at the phylum level, 75 OTUs were very significantly different in abundance between geese on different diets (P<0.01) when comparing individual OTUs by ANOVA. Of these, 54 were Firmicutes, 14 Bacteroidetes, 3 Proteobacteria, 3 Actinobacteria and 1 Tenericutes. The abundance of the family Lachnospiraceae from the phyla Firmicutes was significantly higher in all-grass-fed geese than high-grain-fed geese. The relative abundance of 195 OTUs was significant difference between all grass-fed geese and high grainfed geese (S3 Table).
In addition, a small number of OTUs were identified as making up the dominant community (the abundance accounting for over 1% of the total sequences at family level). It was clear that the 33 OTUs were shared in geese fed with different diets, which belonged mainly Bacteroidaceae and Ruminococcaceae families (S4 Table); and 3 OTUs and 8 OTUs were specific in grass-fed geese and high grain-fed geese, repectively (S4 Table).
To further identify the differences in bacterial taxa between all-grass-fed and high-grain-fed geese, we performed LEfSe as well as non-parametric factorial Kruskal-Wall Prevote sum-rank test to detect differences in abundance 18 and 16 taxa were overrepresented in grass-fed geese  (e.g. Megamonas, Anaerotruncus, Alistipes, P<0.05) and high-grain-fed geese (e.g. Fusobacteria, Fusobacterium, Butyricicoccus, P<0.05), respectively (Fig 5). We next used the PICRUSt tool to explore the functional profiles of the goose gut microbiota. The total OTUs were normalized by 16S rRNA gene copy number (http://picrust.github. io/picrust/tutorials/otu_picking.html) and their metagenomic functions predicted from the KEGG pathways. A total of 175 pathways were more abundant in all-grass-fed geese, while 117 pathways were significantly more abundant in high-grain-fed geese (P<0.01; S5 Table). Of these pathways, the greatest difference was observed among metabolic pathways between the two groups (Fig 6). For example, the KEGG Orthologies (KO) of 'carbohydrate metabolism', 'metabolism of cofactors and vitamins'o and 'amino acid metabolism' were more represented in all-grass-fed geese than in high-grain-fed geese (63 versus 39).

Discussion
The goose relies on the consumption of a certain amount of fibre that can be digested by the microbiota. Accumulating evidence emphasizes the tight interactions between the gut microbiota and the diet [30][31][32][33], but until now, no reports had explored the adaptive responses of Comparative characterization of bacterial communities in geese fed all-grass or high-grain diets the goose gut microbiota to grass feeding. In this study, we compared the characterization of bacterial communities between all-grass-fed geese and high-grain-fed geese. The goose microbiota was richer in OTUs and higher in alpha diversity when fed grass diet compared to grain diet. Previous studies have established that Shannon's diversity index varied from 3 to 6 in the chicken caecal microbiota [34,35], while varied from 7 to 9 in the caecal microbiota of rabbit [19], goat [33] and swine [36]. In this study, we found the average Shannon's diversity index was about 4.2. Our comparative analysis found that the goose gut microbial diversity was similar to that of the chicken, but relatively less than mammalian species. We also noticed that the microbial diversity in all-grass-fed geese was significantly higher than that of high-grain-fed geese. In theory, a high level of diversity provides "functional redundancy" that helps maintain the stability of intestinal microbiota after environmental stress [37]. From this point, the gut microbiota may be more stable in all-grass-fed geese than high-grain-fed geese. We also detected large variations within the faecal microbiota from the Shannon and Simpson indices across different individuals; these variations may be distinct from those between grass-fed and grain-fed samples by unweighted UniFrac analysis. Similar results have been observed in chickens [38]. Hence, the results further confirmed that faecal microbiota do not provide a complete indication of the caecal community structure. In human studies, it has been suggested that faecal samples do not accurately reflected the population of the gut microbiota [39,40]. In addition, Zhao et al have proved that the porcine microbial profile in feces was only the similar to those in the small intestine [41].
The goose is herbivorous; therefore, cellulose-degrading bacteria are particularly important for food degradation, especially when fed a high-cellulose diet. In the present work, regardless of diet, the dominant phyla appear to be Bacteroidetes and Firmicutes, which are typically enriched in other herbivores [42,43]. However, the ratio of Firmicutes to Bacteroidetes differs in grass-fed geese and grain-fed geese, and a higher proportion of Firmicutes and a lower proportion of Bacteroidetes emerged in the grass-fed geese. In humans, the proportion of Firmicutes to Bacteroidetes (the F/B ratio) decreased on a low-calorie diet [44]. In our study, the grass diet was a low-calorie diet compared to the high-grain diet, but the F/B ratio of goose showed the opposite trend from that of humans. This may be due to differences in co-evolution between the gut microbiota of the host (human or avian) and diet. Surprisingly, Fusobacteria and Deferribacteres were highly abundance in high-grain-fed geese, while few in grass- Comparative characterization of bacterial communities in geese fed all-grass or high-grain diets fed geese, which was further confirmed by LEfSe analyses. These results suggest that Fusobacteria and Deferribacteres microbiota have adapted to digest the high-grain diet under the intensive systems. At the family level, Ruminococcaceae (Firmicutes) and Bacteroides (Bacteroidetes) were present exclusively in all-grass-fed geese and high-grain-fed geese, respectively. As a family of cellulose-degrading bacteria, the Ruminococcaceae has been widely confirmed to digest cellulose in human and the other animals [45][46][47].
Furthermore, of the 75 OTUs that were found to show significant differences in abundance between the birds of two dietary treatment geese, the majority are classified as Firmicutes (54 OTUs) with some Bacteroidetes (14 OTUs). Different OTUs belonging to the family Lachnospiraceae were found to be significantly higher in grass-fed geese. Lachnospiraceae are known butyrate producers [48], and perhaps the increased butyrate production helps to stimulate gastric cells proliferation and increase nutrients absorption [49,50].
In the present study, the dominant bacterial community associated with the caecum in grain-fed geese(64.7%) was higher compared to grass-fed geese(47.3%), specifically in Bacteroidaceae which was more abundant in grain-fed geese (31.6%) than grass-fed geese (6.6%). The other microbiota community of goose from NCBI (ERR011376) also showed Bacteroidaceae was predominant bacterial community with an abundance of 43.03%. From the research above, Bacteroidetes may be considered as a dominant microbiota in goose. Zeng et al. (2015) studied the bacterial communities associated with body weight of Rex rabbits and found that the members of Bacteroidaceae were significantly enriched in high-weight rabbits [19]. In our study, we also measured the body weights of geese receiving different diets. The body weight of 70-day-old geese on the high-grain diet (2712.5±327.2g) was higher than that of geese on the grass diet (2630.0±309.3g)(S6 Table). Therefore, the members of Bacteroidaceae might contribute to the growth of animals.
With respect to metabolic pathways, a remarkable enrichment was observed in pathways relating to carbohydrate metabolism, metabolism of cofactors and vitamins, and amino acid metabolism in all-grass-fed geese. Due to the cellulose belonging to carbohydrate, those bacterial metabolic pathways might be more effective in cellulose digestion in the grass. PICRUSt provides important insight into bacterial community functions in the goose gut. Other omics approaches (e.g. transcriptomics and metabolomics) are desired to confirm these discoveries and improve our understanding of bacterial function in the goose gut.

Conclusions
In summary, we presented comparative characterization of bacterial communities in geese fed all-grass or high-grain diets, and found The diets had significant effects in shaping the microbial community. Bacterial profile was primarily predominated by Firmicutes in all-grass-fed geese and Bacteroidetes in high-grain-fed geese, respectively. In addition, some specific microorgnisms including the Ruminococcaceae, Lachnospiraceae and Bacteroidaceae were identified which may associated with cellulose-degrading. Furthermore, the presence of the dominant communities composed predominantly of the Lachnospiraceae and Rikenellaceae, which were involved mainly in the carbohydrate metabolism in all-grass-fed geese. Thus, the study may provide foundation for further investigated the role of specific microbes in efficiency digestion products research of goose.
Supporting information S1