Red, Gold and Green: Microbial Contribution of Rhodophyta and Other Algae to Green Turtle (Chelonia mydas) Gut Microbiome

The fitness of the endangered green sea turtle (Chelonia mydas) may be strongly affected by its gut microbiome, as microbes play important roles in host nutrition and health. This study aimed at establishing environmental microbial baselines that can be used to assess turtle health under altered future conditions. We characterized the microbiome associated with the gastrointestinal tract of green turtles from Guinea Bissau in different life stages and associated with their food items, using 16S rRNA metabarcoding. We found that the most abundant (% relative abundance) bacterial phyla across the gastrointestinal sections were Proteobacteria (68.1 ± 13.9% “amplicon sequence variants”, ASVs), Bacteroidetes (15.1 ± 10.1%) and Firmicutes (14.7 ± 21.7%). Additionally, we found the presence of two red algae bacterial indicator ASVs (the Alphaproteobacteria Brucella pinnipedialis with 75 ± 0% and a Gammaproteobacteria identified as methanotrophic endosymbiont of Bathymodiolus, with <1%) in cloacal compartments, along with six bacterial ASVs shared only between cloacal and local environmental red algae samples. We corroborate previous results demonstrating that green turtles fed on red algae (but, to a lower extent, also seagrass and brown algae), thus, acquiring microbial components that potentially aid them digest these food items. This study is a foundation for better understanding the microbial composition of sea turtle digestive tracts.


Introduction
Gut microbiomes offer essential health benefits to their vertebrate hosts [1] as vital contributors to host digestion and utilization of complex food particles and to the proliferation of their intestinal epithelium within gastrointestinal (GI) tracts [2,3]. Assessing the microbial communities present in the GI tract of vertebrates is becoming fundamental to increase knowledge on their health and biology [4] by offering an insight into its functions and dysfunctions [5] and, hence, aiding managers in the establishment of conservation and protection measures [6,7].

Study Sites
All samples used in this study came from two distinct locations in Guinea-Bissau, West Africa. Esophageal and cloacal swabs were collected from juveniles foraging in coastal waters around the islands of Unhocomo and Unhocomozinho (11 • 31 N-16 • 40 W, Figure 1a), in the westernmost limit of the Bijagós Archipelago, Guinea-Bissau, where important foraging grounds for juvenile green turtles exist [34,42]. Intestine and stomach samples were collected from hatchlings found dead at the nesting beach of Poilão Island (10 • 52 N-15 • 43 W, Figure 1b), which hosts one of the world's largest green turtle nesting populations, producing over one million hatchlings every year [43]. Local potential food items (macroalgae and seagrass species) were collected in the nearshore waters around Unhocomo, Unhocomozinho and Poilão Islands.

Green Turtle Sampling
Foraging turtles were captured with an entanglement net (800 m long, 4 m deep, 20 cm mesh size), deployed from a pirogue operated by local Bijagós fishermen, by enclosing their foraging site for periods of one hour (during which the net was constantly scanned for turtles). Seven juvenile green turtles (with sizes ranging from 36.5 cm to 48.1 cm of Sampling took place at the end of the wet season, which spans between May and November, peaking in August. We do not have the specific environmental conditions at the time but, overall, sea surface temperature average was 27.3 • C (ranging from 25.1 • C to 29.5 • C, [34]) and rainfall averages around 2000 mm per year [42].

Green Turtle Sampling
Foraging turtles were captured with an entanglement net (800 m long, 4 m deep, 20 cm mesh size), deployed from a pirogue operated by local Bijagós fishermen, by enclosing their foraging site for periods of one hour (during which the net was constantly scanned for turtles). Seven juvenile green turtles (with sizes ranging from 36.5 cm to 48.1 cm of curved carapace length, with an average of 40.6 cm ± 4.5 cm SD) captured on 27 October 2019, were sampled for this study. For more information on the captured individuals, see Table  S1. Once captured, the individuals were securely positioned aboard the research vessel anchored next to the net and covered with a wet towel for processing and sampling, with the plastron on a flat surface and the front flippers held close to the body. For the esophageal sampling, the mouth of the turtle was kept open and the esophageal entrance sampled as deep as possible (making a gentle swirl 2-3 times with a sterile swab of 20 cm long) while avoiding any other parts of the mucosa. For cloacal samples, the swab was inserted 2 cm into the cloaca (2-3 swirls; prior cleaning of the cloacal opening with chlorhexidine and gauze). To reduce stress to the individuals, their handling was kept to the lowest (a maximum of 15 min per turtle). Swabs were kept in individual centrifuge tubes filled with 96% EtOH which were maintained at room temperature for two weeks, until stored at 2-5 • C for a week followed by 20 • C storage at the CCMAR laboratory. Dissections of the digestive tubes were performed on dead hatchlings found on the beach in Poilão Island. Stomachal and intestinal chambers were sampled with the swabbing method and samples were kept in microcentrifuge tubes with 96% EtOH. These samples were taken with two purposes: (1) to be used as methodological control, as the individuals from which they were taken did not ingest any food item nor had they been in contact with seawater [44,45]; (2) to investigate what the microbiome of the green sea turtles and the different compartments of their GI tract are composed of. New sterile latex gloves were used for each sample, each individual and each GI section.

Macrophyte Sampling
Green turtle food items were collected in the same abovementioned period by free diving using latex gloves, and samples were transported in Ziploc bags filled with seawater. Algal species Sargassum sp. (n = 1), Caulerpa sp. (leaf, n = 2, and stolon, n = 2), Hypnea sp. (n = 1), Ulva sp. (n = 1), Padina sp. (n = 1), Colpomenia sp. (n = 1), Dyctiota sp. (n = 1) and Rhodophyta sp. (n = 1), and the seagrass Halodule wrightii (leaf, n = 2, and rhizome, n = 2) were sampled at Unhocomo. Samples consisted of a piece of 1 × 1 cm of each specimen (manipulation was done with gloves and EtOH was used to disinfect the materials used), which was placed into a microcentrifuge tube filled with buffer (DNA/RNA Shield from Zymo; 2 mL) and kept in dark conditions. Larger algal specimens were dried in the sun and stored in an herbarium for further taxonomic identification.

DNA Extraction and Analysis of Metabarcoding Data
Nucleic acid extraction was performed with the QIAGEN DNeasy PowerBiolfilm kit, following manufacturer's instructions, to which the following steps were added: samples were placed individually in microcentrifuge tubes with 100 µL of Milli-Q water, for washing off EtOH and salt, before being pelleted at maximum speed for 3 min and transferred to the bead tube. Bead tubes were incubated at 65 • C for 30 min and then homogenized in a TissueLyzer for 3 min at 18,000 rpm. DNA extracts were sent on dry-ice to MR DNA (mrdnalab.com, accessed on 1 February 2021; Shallowater, TX, USA) and a 465-bp fragment of the 16S rRNA gene including V5-V7 variable regions, was amplified using the HotStar-Taq Plus Master Mix Kit (Qiagen, Redwood City, CA, USA) and the primer set 799F-1193R Microorganisms 2022, 10, 1988 5 of 18 (forward: 5 -AMCVGGATTAGATACCCBG-3 ; reverse: 5 -ACGTCATCCCCACCTTCC-3 ) which avoided chloroplast cross-amplification [46]. The cycling profile was as follows: 95 • C for 5 min, followed by 30 cycles of 95 • C for 30 s, 53 • C for 40 s and 72 • C for 1 min, after which a final elongation step at 72 • C for 10 min was performed. PCR products were checked in 2% agarose gels and samples were multiplexed using unique dual indices and were pooled together in equal proportions based on their molecular weight and DNA concentrations. Pooled samples were purified using calibrated Ampure XP beads and were then used to prepare an Illumina DNA library.
Samples were sequenced in an Illumina MiSeq platform with 2 × 300-bp pairedend approach and QIIME2 pipeline (Version 2019.7, https://qiime2.org, accessed on 1 February 2021) was used to analyze the demultiplexed reads as single nucleotide variants, as described by Bolyen et al. [47]. Using FASTqProcessor, forward and reverse primers were removed (http://www.mrdnalab.com/mrdnafreesoftware/fastq-processor.html, accessed on 1 February 2021) followed by quality control with the removal of chimeric sequences using the DADA2 algorithm (qiime dada2 denoise-paired; quality phred > 20; trimming = 170) [48], which was also used to group the sequences based on 100% sequence similarity, generating representative sequences, subsequently referred to as amplicon sequence variants (ASVs). For taxonomic assignment, a Naïve-Bayes classifier was trained on the SILVA v132 99% database [49]. To assign taxonomy, the trained classifier was applied to the representative sequences (qiime feature-classifier classify-sklearn). Plastid-derived sequence reads and singletons were removed from the dataset, which was then rarefied to an even sequencing depth of 8000 reads (minimum number of reads per sample).
The evaluation of the contribution of the microbiome associated with different food items to the gut microbiota of green turtles was addressed with three commonly used statistical approaches. Identification of unique and shared/cosmopolitan ASVs across the different sampling groups was based on present/absence data and accomplished with the 'VennDiagram package' [56]. Differential abundance analysis was performed on nonrarefied data using the 'DESeq2 package' [57,58] to recognize specific microbial ASVs causing significant community composition changes amongst pairwise comparisons of the different sampling groups. Lastly, to identify ASV diagnostics of particular sampling groups, an indicator species analysis, using the 'indicspecies package' [59,60], was applied.
Before the mentioned analyses, the food items were pooled into "food item groups" (from now onwards) according to higher taxonomic levels (Division: green (Chlorophyta) and red (Rhodophyta) algae; Class: brown algae (Phaeophyceae) and seagrass) in order to circumvent low replication. Diversity estimates are given as average number of ASVs ± SD (i.e., ASV richness), unless otherwise specified. For values based on one data point only, no variation is given.

Results
A total of 35 samples were sequenced, including cloaca and esophagus of seven juveniles (cloaca swabs = 7, esophagus swabs = 7), intestines and stomach from five hatchlings (intestine samples = 3, stomach samples = 3), as well as all of the diverse macrophytes (n = 15). From these, a total of 4,485,114 good quality reads (non-chimeric, Table S2) (i.e., phred > 20) corresponding to 18.6% of the initial number of reads were generated (Table S2).
Samples had an average of 128,146 ± 70,916 reads, which clustered into 2311 unique ASVs across all samples combined (2303 ASVs after removal of chloroplast and mitochondria and 2108 ASVs after rarefaction and removal of singletons). One esophagus sample yielded a low number of reads (<8000) and was removed from all subsequent analyses.
PERMANOVA pairwise comparisons (after Bonferroni correction) across all of the different sample types (turtle GI compartments and food item groups) revealed that significant differences in the microbial community structure only happened between juvenile samples (esophagus and cloaca) and green algae, while the rest of the GI compartments and food item groups had no significant differences in their microbial community compositions (Table S3).
Microbial community composition varied across all sample types; nevertheless, there were two bacterial phyla present in all samples, which happened to be the most abundant ones: Proteobacteria (67.7 ± 10.3%) and Bacteroidetes (14.4 ± 5.9%) ( Figure S8). A single (undetermined) family was present across all sample types and belonged to the Gammaproteobacteria class, being the third most abundant family (14.2 ± 13.8%), followed by Rhodobacteraceae (10.8 ± 9.8%), which were only absent from esophageal samples. compartments (PERMDISP, F(3,15) = 20.71, p = 1.37 × 10 −5 , Figure S6). On the other hand, microbial communities did not appear to be significantly structured according to the food item group (PERMANOVA, F(3,11) = 1.26, p = 0.08, Figure S7) and there was no difference in the multivariate dispersion between them (PERMDISP, F(3,11) = 0.88, p = 0.48). Moreover, PERMANOVA pairwise comparisons (after Bonferroni correction) revealed no significant differences in community structure amongst the different food groups (p > 0.05 for all comparisons). PERMANOVA pairwise comparisons (after Bonferroni correction) across all of the different sample types (turtle GI compartments and food item groups) revealed that significant differences in the microbial community structure only happened between juvenile samples (esophagus and cloaca) and green algae, while the rest of the GI compartments and food item groups had no significant differences in their microbial community compositions (Table S3).
Microbial community composition varied across all sample types; nevertheless, there were two bacterial phyla present in all samples, which happened to be the most abundant ones: Proteobacteria (67.7 ± 10.3%) and Bacteroidetes (14.4 ± 5.9%) ( Figure S8). A single (undetermined) family was present across all sample types and belonged to the Gammaproteobacteria class, being the third most abundant family (14.2 ± 13.8%), followed by Rhodobacteraceae (10.8 ± 9.8%), which were only absent from esophageal samples.
More specifically, microbial families across green turtle GI compartments showed different compositions and relative abundances ( Figure 3A) but three phyla were dominant in all of the digestive sections: Proteobacteria (68.1 ± 13.9%), Bacteroidetes (15.1 ± 10.1%) and Firmicutes (14.7 ± 21.7%). Gammaproteobacteria_undet was the dominant More specifically, microbial families across green turtle GI compartments showed different compositions and relative abundances ( Figure 3A) but three phyla were dominant in all of the digestive sections: Proteobacteria (68.1 ± 13.9%), Bacteroidetes (15.1 ± 10.1%) and Firmicutes (14.7 ± 21.7%). Gammaproteobacteria_undet was the dominant family, present across all GI sections (56.8 ± 13.4%), either as first or second in order of abundance. It was the most abundant family in juvenile samples (38.95 ± 17.41%), and the second one in hatchlings (7.48 ± 10.38%). The family Enterobacteriaceae was present across all of the green turtle GI compartments, except in the intestines of hatchlings, and Rhodobacteraceae and Flavobacteriaceae families were only absent in the esophagus ( Figure 3A).
Only two phyla were present within all food item groups, Proteobacteria (85.6 ± 13.4%), and Bacteroidetes (7.6 ± 4.1%) ( Figure 3B), while four families showed up in all of the food item groups in high abundance, except in green algae: Gammaproteobacteria_undet (22.3 ± 18.5%), Rhodobacteraceae (18.5 ± 15.1%), Alteromonadaceae (8.6 ± 14.9%) and Flavobacteriaceae (7.3 ± 7.6%) ( Figure 3B). Similar to the green turtle GI compartments, the Gammaprotecobacteria_undet family was present across all food item groups, being the second dominant one in red algae (2.5 ± 8.2%), while Rhodobacteraceae was the principal one in this food item group (5.4 ± 10.4%). Green algae were the food item group that presented the most distinct community composition ( Figure 3B), showing the four above mentioned families, but in low abundance (<4%), as compared to the other food item groups.
Diagnostic ASVs for the different sample types were assessed by the INDVAL analysis. There were 24 indicator bacterial ASVs in the esophagus, six in the stomach and three in the intestines in green turtle GI compartments,. Besides, there was one common indicator bacterial ASV (Family: Gammaproteobacteria_undet.) for esophagus and stomach together ( Figure S9). From all food item groups, only two of them were signalized as having indicators, with a total of 14 indicator ASVs: eight in brown algae and six in red algae ( Figure S10). The INDVAL analysis, run for all of the sample types together (i.e., all food item groups and turtle GI compartments, Figure 4), gave the same result as the analysis family, present across all GI sections (56.8 ± 13.4%), either as first or second in order of abundance. It was the most abundant family in juvenile samples (38.95 ± 17.41%), and the second one in hatchlings (7.48 ± 10.38%). The family Enterobacteriaceae was present across all of the green turtle GI compartments, except in the intestines of hatchlings, and Rhodobacteraceae and Flavobacteriaceae families were only absent in the esophagus ( Figure 3A). indicator bacterial ASV (Family: Gammaproteobacteria_undet.) for esophagus and stomach together ( Figure S9). From all food item groups, only two of them were signalized as having indicators, with a total of 14 indicator ASVs: eight in brown algae and six in red algae ( Figure S10). The INDVAL analysis, run for all of the sample types together (i.e., all food item groups and turtle GI compartments, Figure 4), gave the same result as the analysis done only on the turtle compartments, except for the absence of the indicator ASV for stomach and esophagus together ( Figure S9).  Microbial families causing significant abundance differences between the distinct sample types were identified with the differential abundance analysis (DESeq2). Each green turtle GI tract compartment presented its own uniquely abundant families, while almost all food item groups (except red algae) showed a common dominant bacterial phylum, Planctomycetes, responsible for most statistical differences (see Table S4 for more details). Implementation of DESeq2 between the esophagus and the rest of the sample types revealed that Cardiobacteriaceae and Thiotrichaceae bacterial families were the ones displaying most significant differences in the microbial community composition, while Nannocystaceae and Lachnospiraceea families were the ones causing the differences when comparing cloacal and intestinal samples, respectively, to the rest of the sample types. Stomachs (of hatchlings) had significant differences in abundance in comparison to the rest of the groups, due to the high abundance of Enterobacteriaceae and Peptostreptococcaceae families. Red algae showed a high presence of Gammaproteobacteria_undet and Flavobacteriaceae families.

Discussion
Despite the extensive information that is now available on the trophic ecology of green turtles (i.e., [13,40,41,61,62]), insights on their gut microbial community composition are still scarce (but see [9,63,64]). Here, we explored the microbial community composition from four GI tract compartments of green sea turtles (Tables S5 and S6) from Guinea-Bissau, and from their potential food items (Table S7) using 16S rRNA metabarcoding, while, at the same time, establishing microbial baselines for the different green turtle GI compartments and the diverse food item groups. Since juvenile green turtles from Unhocomo and Unhocomozinho, in Guinea-Bissau, feed mainly on red algae [35], we further investigated the influence of the microbial composition of this food item on that of the host gut.

Turtle GI Microbial Baselines
Our findings reflected the already known change in gut microbial community composition taking place when green turtle hatchlings start to feed, as they experience a strong environmental influence from their diet and from seawater on their bacterial communities [65]. Moreover, Gammaproteobacteria_undet seemed to be an important undetermined bacterial family within the Gammaproteobacteria class in the gut of the green turtle. It occurred as the dominant microbial family across juvenile samples and the second most abundant one in hatchlings and, this bacterial class, has also been found in abundance in stranded green turtles from the Great Barrier Reef [8].
The indicator ASVs found in hatchlings are potentially part of the core microbiome of green turtles, as hatchlings have not yet had any contact with food nor seawater [44,45]. In the same line, the seven ASVs shared across all of the GI compartments together (hatchlings plus juveniles) were identified as bacterial families that do not often occur in seawater, which brought us to also consider them as potential members of the core microbiome of the turtles and possibly, maternally inherited.
The scarce 13 ASVs shared across all sample types corroborated the visual NMDS results where food item groups overlapped with turtle compartments, and were identified as microbial families typically found in the marine environment (Marinicella_undet; [66], Erythrobacter_undet; [67], Lewinella_undet; [68]) leading to the conclusion that these ASVs present across all sample types came from microbial families typically associated with seawater.
The esophagus was the GI compartment that presented the most homogeneous microbial composition, showing the greatest amount of indicator ASVs (Table S5), many of which corresponded to bacteria commonly found in the marine environment (Gammaproteobac-teria_undet, Bacteroidia_undet, Mariniflilum_undet, Corynebacterium1_undet, Bdellovib-rio_undet, Nannocystaceae_undet, Hyphomicrobiaceae_undet, Desulfatiferula_berrensis, Muricauda_undet [27,[69][70][71]. Since sea turtles regularly ingest bacteria associated with seawater, the variety of microbes residing in the mucosa of their esophagus can be altered by transient bacterial communities coming from the external environment [65]. As the esophagus is one of the first GI sections to be in contact with the marine environment, it serves as an entrance spot for (gut) microbes, hosting the highest diversity, which becomes reduced as these microbes move through the GI tract [65].
Hindgut microbial communities dominated by Proteobacteria, Firmicutes, Bacteroidetes and Verrucomicrobia, similar to our results, have also been found in other marine herbivorous organisms, such as in the rabbitfish, Siganus fuscescens [15]. Proteobacteria, Firmicutes and Bacteroidetes are also the dominant bacterial phyla in the gut of green sea turtles from different parts of the world [8][9][10][11]20,72]. Post rehabilitation green turtles from Australia presented a dominance of Proteobacteria followed by Bacteroidetes and Firmicutes [10]. Gut resident microbes from juvenile green turtles from coastal areas of Brazil have shown a codominance between Bacteroidetes and Firmicutes, and a high abundance of Proteobacteria [11].
Other sea turtle species have also been found to harbor the same dominant digestive microbiome phyla members that are typical of herbivore gut microbiota [20], such as the loggerhead (Caretta caretta, [11]), the olive ridley (Lepidochelys olivacea, [73]) and the hawksbill sea turtle (Eretmochelys imbricata, [74]), suggesting that despite being mostly carnivorous (feeding on sponges, soft corals and cnidarians), the loggerhead and the olive ridley might have the ability to digest algae and seagrass [75,76]. The occurrence of Proteobacteria in the GI tract of vertebrates has been suggested as an indication of disease and dysbiosis [77], but also as a possible response to stress associated with anthropogenic factors [11]. Bacteria within the phylum Bacteroidetes are common members of gut microbiota in many vertebrates, including healthy sea turtles [10,78]. Firmicutes are amongst the main gut microorganisms of sea turtles [78] and their presence across all GI compartments might be explained by their ability to break down plant-derived polysaccharides, such as cellulose, hemicellulose and xylan [10]. The Bacteroidetes and Firmicutes use nutrients from the food items consumed by the turtles [79,80], starting this activity in the caecum, while little digestion and absorption take place in the stomach [2]. Accordingly, we found that members of the phylum Firmicutes, especially the Lachnospiraceae family, had a high abundant presence in the intestines of hatchlings. Corroborating our findings, Ahasan et al. [65] demonstrated that Firmicutes are significantly dominant in green turtle stomachs and large intestines, which was further supported by analyses on fecal samples [8,9,11].

Microbial Contribution of Food Items to Green Turtle GI Microbiome
Since juvenile green turtles from Unhocomo and Unhocomozinho foraging grounds, in Guinea-Bissau, eat red algae [35], and our results revealed that microbial communities of red algae were not significantly different from the ones in the green turtle GI tract compartments, we further focused on the microbiome of this food item group to evaluate its potential contribution to the digestive microbiome of the green turtles. Nonetheless, as the microbial communities of brown algae and seagrass were neither significantly different from the ones in the green turtle GI tract compartments, we did not discard the possibility of these food item groups influencing the gut microbiome as well, as these green turtles also feed on seagrass, and, in a lower degree, on brown algae [35].
Bacterial phyla, such as Firmicutes, Planctomycetes, Verrucomicrobia and Cyanobacteria, along with Bacteroidetes and Proteobacteria [25,81], have been recognized as frequent colonizers of red algae [82], which was also supported by our findings (Table S7). One undetermined Gammaproteobacteria microbial family was the second most abundant one in this algae food item group, and it was one of the families causing significant microbial community composition differences in the DESeq2 analysis; some of the bacterial indicator ASVs of red algae belonged to this undetermined family as well. Members of this same bacterial class have been isolated from red algae species, such as Amphiroa anceps and Corallina officinalis [83].
In contrast with the 14 microbial ASVs indicators of the food item groups, there were no indicator ASVs of food items when this analysis was run for food item groups and green turtle GI compartment samples together. This showed that these ASVs were present in the turtle microbiome probably as a result of green turtles eating those food items. From the six red algae indicator ASVs in the first analysis, only two of them (Alphaproteobac-teria_undet, NCBI BLAST: Brucella pinnipedialis; and Gammaprotebacteria_undet, NCBI BLAST: Methanotrophic bacterial endosymbiont of Bathymodiolus sp.) were found in green turtle GI compartments, in both cloaca and stomach ( Figure 5A). While the indicator ASV identified as Alphaproteobacteria_undet had a 75 ± 0% abundance in cloacal samples and 25% in the stomach, the Gammaprotebacteria_undet indicator ASV abundance in both compartments was very low (<1%).
The presence of these ASV microbial indicators of red algae in the cloaca ( Figure 5A), as well as of shared bacterial ASVs between cloaca and red algae ( Figure 5B), was most likely due to their acquisition through the ingestion of red algae [35]. In fact, two of the green turtle individuals that presented the two red algae indicators (Animal J6 and J7, Figure 5A) along with animals J2, J5, J6 and J7 that shared microbial ASVs (in their cloaca) with red algae (Figure 5B), were the same individuals for which a parallel study on diet analysis identified red algae as the main diet items [35].
(Alphaproteobacteria_undet, NCBI BLAST: Brucella pinnipedialis; and Gammaprotebacteria_undet, NCBI BLAST: Methanotrophic bacterial endosymbiont of Bathymodiolus sp.) were found in green turtle GI compartments, in both cloaca and stomach ( Figure 5A). While the indicator ASV identified as Alphaproteobacteria_undet had a 75 ± 0% abundance in cloacal samples and 25% in the stomach, the Gammaprotebacteria_undet indicator ASV abundance in both compartments was very low (<1%). The presence of these ASV microbial indicators of red algae in the cloaca ( Figure 5A), as well as of shared bacterial ASVs between cloaca and red algae ( Figure 5B), was most likely due to their acquisition through the ingestion of red algae [35]. In fact, two of the green turtle individuals that presented the two red algae indicators (Animal J6 and J7, Figure 5A) along with animals J2, J5, J6 and J7 that shared microbial ASVs (in their cloaca) From the six microbial ASVs shared between green turtle GI compartments and red algae, all were present in cloacal samples (in different abundances and individuals), and two ASVs were found in one hatchling (Animal H2, intestines, Figure 5B). The same hatchling individual harbored a red algae indicator ASV in its stomach, which could hypothetically be due to contamination, but the possibility of vertical transmission should not be discarded [84]. As not all of the cloacal samples that presented shared ASVs with red algae ( Figure 5B) and red algal indicator ASVs ( Figure 5A) belonged to individuals found to have eaten red algae [35], we suggest the possibility that, in addition to recent acquisition from the diet [65], red algae indicators may actually be integrated in the core microbiome of the turtles and passed on via vertical transmission. This is an open research question that needs to be addressed in the future for marine turtles, as for freshwater turtles, such as the yellow-spotted Amazon River turtle (Podocnemis unifilis), the gut microbiome of which is known to be partially determined by vertical transmission [85].
Apart from cellulose and galactans, the cell walls of red algae have high protein content [86,87]. Some species of Alphaproteobacteria have the ability to hydrolize cellulose and lignin [88,89], while members of Gammaproteobacteria are capable of hydrocarbon utilization [30]. Marine Bacteroidetes are degraders of particulate matter, of proteins principally, as they can live attached to particles and have the capacity to degrade polymers, including peptidases, glycoside hydrolases, glycosyl transferases and adhesion proteins [90]. Particularly, Flavobacteriaceae bacteria are capable of degrading the main components of red algae cell walls [86], complex carbohydrates and proteins [90][91][92], due to the diversity of genes encoding their polymer-degrading enzymes [93]. Presence of the described taxa (Alphaproteobacteria, Gammaproteobacteria, Bacteroidetes, Flavobacteriaceae) in the cloaca of green turtles, if acquired by vertical transmission after initial incorporation from red algae microbiome, could indicate that they are important contributors to the digestion and degradation of the turtle food items, especially of red algae. Further research addressing functions other than assistance to digestion by the microbiome of food items could give more information on the health status and immune systems of the sea turtles [5].

Limitations and Recommendations
Taxonomic assignment of the microbial communities associated with the GI tract of the green turtles and of their potential food items was done using the best available reference library (SILVA, [49], see methods section). Nevertheless, the microbial identification could improve with updated microbial databases. Such a high proportion of unclassified reads, as found in other marine vertebrates [21,94], implies a significant presence of novel/undescribed bacteria, suggesting an important role for future research in the identification of unclassified bacterial within the GI tract of green turtles [8]. Moreover, we do not know if the sampled green turtle juveniles were feeding or not at the time of capture, and, thus, the analyzed microbial communities could be underrepresented relative to what would be expected in feeding animals [72]. On another line, cloacal samples may be unreliable indicators of digestive system microbiome due to contamination of external surfaces by seawater bacteria [20]. Additionally, we note that the replication used to characterize the microbiome of the different food item groups was relatively low and could have missed some of their rarest members. Further investigation, considering temporal and seasonal variations and during feeding events in a larger sample size, is suggested, to determine a more precise digestive baseline for green turtle microbial composition and for the evaluation of the role of diet on the gut microbiome. Finally, the real function of the taxa here revealed must be investigated in the future using transcriptomics on the active microbial fraction.

Conclusions
In line with our initial objectives, through the implementation of 16S rRNA metabarcoding, we revealed that green turtles experience a shift in gut microbial community composition when developing from hatchlings that just emerged from the nest, which likely include only core microbiome, to juveniles foraging in coastal waters, which, apart from the core microbiome, include transient microbial communities likely influenced by their environment. We suggest that the red algal microbiome present in the cloaca could potentially be part of the core microbiome of the turtles, but we cannot discard the possibility that it is a transient microbiota controlled by the diet. Our study represents the first detailed research of green turtle gut microbial communities in green turtles from Guinea Bissau. We hope to establish a baseline for a better understanding of the microbial composition of green sea turtle GI tracts and their feeding behavior, guiding the way to the future use of microbes as indicators for the quality of foraging grounds.
Supplementary Materials: The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/microorganisms10101988/s1, Table S1: Summary information of the foraging green turtles sampled for this study; Table S2: Summary table of samples collected from green turtles and from putative food items; Figure S1: Alpha-diversity boxplots for the four turtle GI compartments; Figure S2: Alpha-diversity boxplots for the food diet items; Figure S3: Venn diagram exploring specific and ubiquitous ASVs between the four GI turtle compartments; Figure S4: Venn diagram exploring specific and ubiquitous ASVs between the four food item groups; Figure S5: Venn diagram exploring specific and ubiquitous ASVs between the four food item groups and the four turtle compartments together; Figure S6: NMDS plot based on Bray-Curtis dissimilarity of the microbial community structure of green turtles GI compartments; Figure S7: NMDS plot based on Bray-Curtis dissimilarity of the microbial community structure of the green turtles food item groups; Table S3: PERMANOVA pairwise comparisons across all of the different sample types; Figure S8: Relative abundance of the microbial community composition at the family level across all of the sample types of this study; Figure S9: Indicator bacterial ASVs for the turtle GI compartments; Figure  S10: Indicator bacterial ASVs for the food item groups; Table S4: Differential Abundance in ASVs expression analysis (DESeq2) summary results; Table S5: Microbial baselines for juvenile samples of green turtle; Table S6: Microbial baselines for hatchling samples of green turtle; Table S7: Microbial baselines for diet items.