One of the most outstanding issues in microbial ecology of the gastrointestinal (GI) tract is understanding how biological and physical factors influence gut microbiota and their hosts (Sullam et al., 2012). The GI tract is occupied by a complex and dynamic ecosystem of organisms composed of an enormous variety of aerobic, facultative anaerobic and obligate anaerobic microbes that interact with the host and with each other (Nayak, 2010). The role of microbiota on host health is increasingly being recognized, for example, the GI microbiota of fish contributes to the development of its host through xenobiotic metabolism, microbially-mediated digestion of food, essential nutrient supply including vitamins, amino acids and fatty acids, immunity, and resistance toward intestinal pathogens (Kesarcodi-Watson et al., 2008; Ringø et al., 2016).

Until recently, most studies examining the microbiota associated with marine organisms have employed culture-dependent methods (Forney et al., 2004). This approach is somewhat limited, given that the vast majority of microorganisms present in a natural environment cannot be cultured in vitro (Fjellheim et al., 2007). For example, in marine biomes the percentage of unculturable organisms is estimated to be higher than 97% (Rappé and Giovannoni, 2003). Culture-independent methods, such as the detection and sequencing of the microbial-derived 16S small subunit ribosomal RNA (rRNA) gene, have been developed to overcome this limitation, and have been applied toward the study of hatchery-associated bacterial populations of Atlantic cod (Gadus morhua, Brunvold et al., 2007; Reid et al., 2009; Bakke et al., 2013), abalone (Haliotis diversicolor, Zhao et al., 2012), Atlantic halibut (Hippoglossus hippoglossus L., Verner-Jeffreys et al., 2003; Jensen et al., 2004) and great scallop (Pecten maximus, Sandaa et al., 2003). However, the focus on PCR amplification of the 16S rRNA gene alone may provide a biased estimate of species abundance, given that “universal primers” for 16S PCR are not necessarily universal, i.e., not all species can be detected due to unknown sequence variation, and that biases associated with primer mismatch, and preferential amplification of the most abundant groups have been described (reviewed in Forney et al., 2004).

High throughput sequencing (HTS) methods are increasingly being applied to the characterization of microbial communities, leading to a more comprehensive appreciation of extant biodiversity (Mock and Kirkham, 2012). Although, HTS approaches still commonly rely on the amplification of one or two hypervariable regions of the 9 hypervariable regions (V1–V9) present in the 16S rRNA gene, its significant advantage over conventional 16S rRNA sequencing is the dense sampling of a given community increasing the likelihood of capturing lowly abundant species (Mock and Kirkham, 2012). Moreover, the sample throughput of HTS is significantly higher than traditional approaches, mediated by sample barcoding and multiplexing (the number of samples limited largely by the number of unique barcodes available), enabling study designs thousands of times more robust than other PCR-based techniques (Zarkasi et al., 2014; Huang et al., 2016).

The efficiency and sustainability of any mariculture system will likely be significantly influenced by the microbial composition of the species in question in their natural environment. Characterization of species composition, relative quantities, and the potential sources of the core intestinal microbiota commonly associated with feed and larvae at different stages of development is essential for viability and vitality of the organisms (Ringø and Birkbeck, 1999; Olafsen, 2001), and aid in the identification of possible microbial pathogens affecting larval mortality (Star et al., 2013). In cephalopod mollusks (octopus, squids and cuttlefishes), only three studies have analyzed the microbial diversity of Octopus species (de la Cruz-Leyva et al., 2011; Iehata et al., 2015, 2016). These studies have used Denaturing Gradient Gel Electrophoresis (DGGE) techniques to analyse cultured bacterial diversity sampled from adults and eggs of the Chilean Gould octopus, Octopus mimus (Gould, 1852), revealing differences between males and females in the microbial families present (mostly Vibrionaceae and Streptococcaceae) and their nutritional enzymatic activities (Iehata et al., 2015). In addition, a relationship between egg-associated bacterial diversity and egg health condition (dominated by Roseobacter) was also detected (Iehata et al., 2016). Vibrionaceae were the main bacterial group, identified using RNA transcripts of the 16S rRNA gene, from metabolically active bacterial flora of adult octopuses collected in Mexico (de la Cruz-Leyva et al., 2011). Considering the low throughput approaches used in these studies, it is likely that they only account for a fraction of bacterial diversity present within octopods or cephalopods in general. The application of genomic methods will significantly enhance the characterization of the cephalopod paralarvae microbial communities, and may in turn provide useful insight toward improving the aquaculture conditions of commercially important species such as the common octopus, Octopus vulgaris Cuvier, 1797.

In spite of the plethora of experiments to solve it (reviewed in Vidal et al., 2014), rearing O. vulgaris paralarvae in captivity is difficult and remains a significant hurdle that prevents viable aquaculture. Little is known about the ecology of wild O. vulgaris paralarvae and their unusual planktonic strategy in the open ocean, largely due to difficulties in obtaining specimens (Roura, 2013). It has been recently suggested that O. vulgaris paralarvae undertake a unique planktonic strategy, compared with that of other coastal cephalopods with planktonic stages (Roura et al., 2016). They hatch close to the coast with only three suckers per arm, and after <10–15 days, they are transported away from the continental shelf by coastal upwelling filaments, finishing their development in the open ocean. Remarkably, 58 O. vulgaris paralarvae containing more than three suckers per arm were collected in zooplankton samples off the NW Iberian Peninsula (42 specimens) and Morocco (16 specimens), with bottom depths ranging between 787 and 3,110 m (Roura, 2013). These paralarvae are the only specimens larger than three suckers per arm ever collected in the Eastern Atlantic (Rocha et al., 1999; González et al., 2005; Moreno et al., 2009; Otero et al., 2009; Roura et al., 2016). These rare samples therefore provide a unique opportunity to study the ontogenic changes of their microbial biota from the coast to the ocean, and to compare the natural microbiome against that found in aquaculture, with the aim to determine the importance of the GI microbiome on the health of captive paralarvae.

In this study, we have applied HTS to characterize the core gut microbiota of wild paralarvae collected in two different upwelling regions (NW Spain and W Morocco), and to identify the main microbial groups that differ between ecosystems. Furthermore, we have compared the GI microbiota of wild Octopus paralarvae against that of paralarvae reared with Artemia during 25 days in captivity. This enabled characterisation of the core gut microbiota of wild paralarvae and identification of bacterial groups that are not present in paralarvae reared in captivity, and to identify potential pathogens that may affect the health of paralarvae reared in captivity.

Planktonic samples were collected during the multidisciplinary project “Canaries-Iberian Marine Ecosystem Exchanges (CAIBEX)” (Figure 1, red frames), off the coasts of North-Western Iberian Peninsula (CAIBEX-I: July 7 to 24, Figure 1B) and Morocco (CAIBEX-III: August 16 to September 5, Figure 1C) in 2009. Mesozooplankton samples were collected day and night with two 750 mm diameter bongo nets equipped with 375 μm mesh and a mechanical flow-meter. Three double-oblique towings were carried out (at a ship speed of 2.5 knots) per station over the continental slope (>200m depth): (i) at the deep scattering layer (DSL: 500 m), (ii) at 100 m, and (iii) at the surface (0–5 m). Over the continental shelf (<200 m) only two double oblique towings were collected at 100 m (when sea-bottom was <100 m, otherwise 10 m above it) and at the surface (0–5 m). The bongo net was first lowered to the desired depth, towed for 30 min and subsequently hauled at 0.5 m s–1. The net was recovered, cleaned on board and placed back into the sea for the next towing. Plankton samples were fixed with 96% ethanol and stored at −20°C to facilitate DNA preservation. All cephalopod paralarvae were sorted from the zooplankton samples and stored individually in 70% ethanol at −20°C. In total, 134 O. vulgaris paralarvae were collected during CAIBEX-I (n = 99 specimens) and CAIBEX-III (n = 35 specimens). Of these, 41 paralarvae were chosen from CAIBEX-I (ranging from 3 to 5 suckers per arm) and 35 from CAIBEX-III (ranging from 3 to 15 suckers per arm) to study the ontogenic changes of the microbiota in the wild.

The microbiota of paralarvae reared in captivity with Artemia in 2012 at the facilities of the Spanish Institute of Oceanography in Vigo (IEO-Vigo), was also analyzed using five replicates at ages 0, 5, 10, 15, 20, and 25 days post hatchling (ranging from 3 to 5 suckers per arm). Paralarvae were anesthetized at the end of the study by immersing them in a 1.5% MgCl₂dissolved in seawater at room temperature (18–21°C) for 10 min, after which the MgCl₂ concentration was increased to 3.5% for 30 min to kill them. The procedures applied herein comply with Directive 2010/63/EU, in terms of minimizing the number of animals used and animal sacrificing method employed (Fiorito et al., 2015). This study was performed in accordance with corresponding Spanish guidelines and regulations (Ley 32/2007, November 7th) and was exempt from an ethics review process.

The euphausiid Nyctiphanes couchii, the crabs Pirimela denticulata and Pilumnus hirtellus, and the copepod Paraeuchaeta hebes, were sorted from the zooplankton samples collected near the coast of NW Spain and the microbiome of their gastrointestinal tract was analyzed. Crabs and krill are known prey of wild O. vulgaris paralarvae (Roura et al., 2012), whereas P. hebes has not been described as part of the Octopus diet. However, this copepod is an important member of the coastal zooplankton (Roura et al., 2013) and has been recently identified in the digestive tract of Alloteuthis media paralarvae (Olmos-Pérez et al., 2017), and therefore, has been included as a potentially informative bioindicator of the environmental microbiota present.

In this study, we present the first analysis of the GI microbiome of O. vulgaris paralarvae, characterizing both the complex microbial communities present in wild paralarvae and the ontogenic change in bacterial community composition based on diet and development in captivity. Paralarvae reared in captivity with Artemia showed a depletion of bacterial diversity, particularly after day 5 when almost half of the bacterial species present at day 0 were lost. In contrast, bacterial diversity increased in wild paralarvae as they developed in the ocean (Figure 4), likely due to the exposure of new bacterial communities via ingestion of a wide diversity of prey (Roura et al., 2012; Olmos-Pérez et al., 2017).

The number of bacterial sequences obtained per sample (average of 96,406 ± 35,302 SD) was almost 10 times the minimum sample depth needed to capture the structure of microbial communities (Caporaso et al., 2011). Only two samples had <10,000 sequences (833 and 8,574) and, despite their low depth, the main bacterial groups and their relative proportions were consistent with other samples. Despite using Greengenes, the most comprehensive microbial taxonomy database available (McDonald et al., 2012), we found a high percentage of unclassified sequences that may represent novel bacterial species present (between 20 and 60% per sample). High proportions of unclassified sequences have been described in other studies including mouse gut and anaerobic digester samples, where phylotypes unclassified at the genus level represented a greater proportion of the total community variation than classified OTUs, underscoring the need for greater diversity in existing reference databases (Werner et al., 2012). In our study, the percentage of unclassified reads explained up to 7.2% of the total variability found in the microbial communities. Interestingly, these unclassified OTUs were significantly more abundant in wild than captive paralarvae, indicating a high degree of novelty in the microbial species present in the digestive tract of wild paralarvae incorporated through the diet.

Consistent with this study, previous genomic studies have found that wild fish larvae have more diverse microflora than their captive relatives (e.g., Atlantic cod: Dhanasiri et al., 2011; olive flounder: Kim and Kim, 2013). This suggests that monospecific (Artemia) or even formulated diets are not as favorable as those diets encountered in nature, which seem to provide an important source of potentially beneficial microorganisms that might be exploited to supplement and diversify depleted microflora in captivity. The intestinal microbiota of a host can be classified as autochthonous (i.e., core bacteria in this study) or allochthonous bacteria (Ringø and Birkbeck, 1999). The autochthonous bacteria are those able to colonize the host's gut epithelial surface (microvilli), while the allochthonous bacteria are transient, associated with food or water, and cannot colonize except under abnormal conditions. Several studies have demonstrated that the endogenous microbiota is an important component of the mucosal barrier, representing the first line of defense against pathogens (Gómez and Balcázar, 2008). The diverse core bacteria (autochthonous) detected in recently hatched O. vulgaris and wild paralarvae was rapidly modified and substituted by two opportunistic bacterial families, Vibrionaceae followed by Mycoplasmataceae (Figure 4). The same succession of opportunistic bacteria was also detected in cod larvae reared in captivity (McIntosh et al., 2008). Both families are known pathogens affecting many larviculture systems, with the family Vibrionaceae often found in parasitic or mutualistic associations with the gut of marine animals, where they provide diverse metabolic capabilities (Thompson et al., 2004; Sullam et al., 2012; Zhao et al., 2012).

Although certain Vibrio species are beneficial for the host (Austin et al., 2005; Fjellheim et al., 2007), this opportunistic group is responsible for high mortalities in larviculture (Brunvold et al., 2007; Zhao et al., 2012). Indeed, studies have shown that Artemia are important vectors of pathogens (mostly Vibrionaceae) that colonize fish (e.g., McIntosh et al., 2008; Reid et al., 2009) and abalone larvae after first feeding (Zhao et al., 2012). Interestingly, Vibrionaceae was identified using a culture-dependent method and 16S rDNA clone library in wild adult specimens of O. mimus, however most of the cloned sequences belonged to the family Mycoplasmataceae (Iehata et al., 2015). They suggested that Mycoplasma might be a autochthonous member of the octopus GI bacterial community with an unknown function, as is has also been found within the GI tract of wild specimens of Norway lobster (Meziti et al., 2010) and Atlantic salmon (Star et al., 2013). Our results indicated that this genus is present in both, wild and captive paralarvae, but their abundance is markedly different (Table 5). However, we suggest that the Mycoplasma species observed in captive Octopus paralarvae are opportunistic and, together with Vibrio, are candidate pathogens that may be responsible for the high mortalities observed in Octopus larviculture. Mycoplasma has also been detected in farmed salmon sporadically, but when present, it dominated the GI tract communities (Zarkasi et al., 2014). The sporadic nature of Mycoplasma suggests host factors at play that may influence GI tract community structure and contribute to dynamic changes. The saprophytic nature of Mycoplasma, with a fermentative metabolism, and its increasing abundance in captive octopus paralarvae may be related with the presence of dead paralarvae and Artemia at the bottom of the tank, which provide optimal conditions for this opportunistic genus. More research is needed to accurately identify the different Mycoplasma and Vibrio strains in order to test this hypothesis.

In our study, the diversity of the GI microbiota found in recently hatched paralarvae in captivity (day 0) was unexpectedly high (Figure 4). Olafsen (2001) suggested that a dense, diverse but non-pathogenic egg epiflora may be a barrier against colony formation by pathogens. One possible explanation is that the diverse microbiota in captive hatchlings of Octopus might be derived from bacteria attached to the egg capsule. This suggestion is supported by the observed biodiversity of culturable epiflora associated with healthy eggs of O. mimus (Iehata et al., 2016). Bacterial diversity of healthy eggs was higher than that of infected eggs (i.e., eggs from the same female that changed color from whitish to yellow-brownish indicative of infection), which were dominated by pathogenic genera like Pseudoalteromonas, Vibrio, and Tenacibaculum. In our study, the initial diversity rapidly decreased when Octopus paralarvae started exogenous feeding on Artemia, and opportunistic bacteria colonized the GI tract. In contrast, a gradual increase in species richness was observed among wild paralarvae as they migrated from their coastal hatchling grounds to the oceanic realm (Figures 3, 4). This is the first time that this ontogenic change has been observed in O. vulgaris paralarvae and suggests a relationship between diversity of GI flora and paralarvae survival. This ontogenic change in the microbial community has also been observed in other marine organisms, including abalone (Zhao et al., 2012), sponges (Cao et al., 2012), white shrimp (Huang et al., 2016) and fish aquaculture (reviewed in Ringø et al., 2016), and has been suggested to be a natural process that likely plays a role in the correct development of the host's immune system and GI tract, preventing pathogens from colonization.

Marine larvae are in constant interaction with bacteria during their first feeding (Olafsen, 2001), and compared to wild conditions, intensively cultured larvae experience stress due to inappropriate feeding (Iglesias et al., 2007) and higher larval densities than in their oceanic environment (Roura et al., 2016). Furthermore, the high organic load associated with rearing conditions may enhance the proliferation of opportunistic pathogenic bacteria (Lauzon et al., 2010), which can be detrimental to the paralarvae and is one potential cause for the highly unpredictable growth and reduced survival that limits Octopus aquaculture. Our results clearly demonstrate that the gut flora of captive paralarvae was distinctly different from the “healthy” gut flora community of wild paralarvae (Figures 3, 4).

The bacterial families Comamonadaceae, Flavobacteriaceae, and Moraxellaceae were the most discriminating families enriched in the wild Octopus paralarvae core community, and could be a potential source of beneficial bacteria to test in captivity. This was the case of wild olive flounder (Kim and Kim, 2013), where wild fishes were an essential source of beneficial microbes that conferred resistance to pathogenic bacteria (Nayak, 2010). Bacterial composition in wild O. vulgaris (at the phylum level) was similar to carnivorous/herbivorous marine fishes (Sullam et al., 2012), with a composition hierarchy consisting of Proteobacteria>Actinobacteria>Bacteroidetes>Firmicutes. Interestingly, one of the core intestinal bacterial groups of wild Octopus paralarvae was the family Flavobacteriaceae (Table 4). Although this family was initially proposed to be exclusively found in herbivorous fishes (Sullam et al., 2012), this hypothesis was later rejected by a pyrosequencing study that found this group within the GI tract of wild Atlantic cod (Star et al., 2013).

Finally, it is remarkable the similarity of the microbial community found in wild zooplankton and that of the paralarvae growing near the coast (Figure 4). Although only four zooplankton species were analyzed in this study, the similarities observed support a close relationship between the microbial communities present in GI tract of the predator and that of its prey. Wild paralarvae continuously diversify their core gut microflora with a diverse diet (Roura et al., 2012; Olmos-Pérez et al., 2017), which provides a natural source of allochthonous bacteria. This diverse microbiota likely serve a variety of functions in the nutrition and health of the host by promoting nutrient supply, preventing the colonization of infectious agents, energy homeostasis and maintenance of normal mucosal immunity (Nayak, 2010). In summary, this study provides a comprehensive overview of the bacterial communities inhabiting the GI tract of O. vulgaris paralarvae, and reveals new research lines to challenge the current bottlenecks preventing sustainable octopus aquaculture.

AR, JS, and SD planned the work. AR captured the wild paralarvae and MN grew the paralarvae in captivity. AR, JS, and SD designed the genomic study. AR and SD performed the genomic analyses. AR, SD, JS, and MN wrote the manuscript. All authors have read and approved the content of the manuscript.