In recent decades, there has been an increasing number of reports of diseases in marine organisms, such as corals, mollusks, turtles, and mammals (Harvell et al., 1999, 2004). Infectious diseases have been acknowledged as a major factor of change in marine ecosystems. They can lead to a loss of keystone species and alter critical ecosystem processes (Harvell et al., 2002). However, their causal agents often remain unknown, thus slowing the development of management and remediation tools (Harvell et al., 2004). Together with scleractinian corals, crustose coralline algae (CCA) are major framework builders and carbonate producers on tropical reefs (Rasser and Riegl, 2002). Two disease symptoms have been recently investigated: the coralline white band syndrome (CWBS) and the coralline white patch disease (CWPD) (Quéré et al., 2015b). White band lesions are defined by bands that appear centrally or peripherally and advance on healthy tissue, whereas the patch disease shows distinct white patches expanding on healthy crusts (Quéré et al., 2015a). Both diseases result in tissue loss, which often leads to the death of the affected patch, with subsequent colonization by endophytic algae (Quéré et al., 2015b). Although these syndromes have been well-described in field surveys, standard histopathological techniques have failed to identify the potential infectious agents (Quéré et al., 2015a). Further molecular characterization of the microbial communities is needed to complete the diagnostic picture.

The microbes that are associated with macro-organisms play an important role in the host’s health and metabolism (Rosenberg and Zilber-Rosenberg, 2016). Microbial communities associated with CCA remain poorly understood (Webster et al., 2011; Sneed et al., 2015; Hester et al., 2016), but there are indications that they can induce settlement and metamorphosis of coral larvae (Negri et al., 2001). CCA bacterial communities can be altered by environmental changes such as temperature (Webster et al., 2011), pH (Webster et al., 2013), or disease (Kimes et al., 2010), but the identification of the causative pathogens associated with epizootics in CCA is still unknown. Recently, the development of “omic” studies has shown the limits of the Koch and Hill’s postulate “one microbe – one disease” and brought forward the pathobiome concept (Vayssier-Taussat et al., 2014), which has been proposed for the study of coral diseases (Sweet and Bulling, 2017). The pathobiome is defined as the pathogenic agent integrated within its biotic environment (Vayssier-Taussat et al., 2014). Pathogens interact with other microbes, and these complex interactions influence the outcome of the disease.

Here, we investigated both healthy and diseased CCA tissues to test the hypothesis that tissue lesions indicative of CWBS and CWPD are characterized by distinct pathobiomes.

A comparison of the bacterial community composition was conducted using the Bray–Curtis similarity index and MDS analysis. The MDS ordination of 16S tag sequencing data separated the N. mamillare samples into three main groups (Figure 2). One group corresponded to the CCA sampled from diseased CWPD tissues, a second group included CWBS diseased tissues and the third group had healthy tissues from both control and diseased individuals. There was only one exception with one diseased CWPD sample similar to a healthy community. There was more variability within CWBS disease tissue communities than within CWPD communities (Figure 2). In summary, our results show that bacterial communities associated with CWBS were significantly different from the ones found in CWPD diseased tissues (Table 1) and they were different from control tissues (ANOSIM, p < 0.01), demonstrating that the two syndromes corresponded to different pathobiomes. The pathobiome diversity in patch diseased samples was significantly lower than the diversity of the bacterial community in healthy tissues (Figure 3). This finding, together with the observed unchanged diversity of the band syndrome community, is in marked contrast to the patterns that are commonly observed for disease associated microbial communities. Higher diversity has been observed in diseased sponge tissues (Blanquer et al., 2016), and in corals, the yellow band or white plague diseases are associated with an increased number of bacterial taxa (Closek et al., 2014; Roder et al., 2014). These studies have hypothesized that a diverse array of opportunistic bacteria from the surrounding environment colonizes the compromised coral tissues. In the case of CCA, the lower diversity of the CWPD diseased tissues suggests the occurrence of a few specialized pathogens.

Diseased CWPD tissues were characterized by a higher proportion of Bacteroidetes and Gammaproteobacteria (White’s non-parametric t-test between control and diseased tissues, p < 0.05) (Figure 4). At a higher taxonomic resolution, the CWPD pathobiomes showed a significantly higher abundance of a Rickettsiales (OTU 20), similar to an uncultured bacterium isolated from Porites coral mucus (White’s non-parametric t-test, Table 1 and Figure 5), but without match to any cultured microorganisms. OTU 20 represented up to 13% of the sequences in diseased PD tissues. The CWPD pathobiome also showed a significantly higher abundance of a Vibrio sp. OTU (OTU 665, Figure 5), which was similar to Vibrio tubiashii (Table 1), a known pathogen of mollusks often associated with larval vibriosis (Beaz-Hidalgo et al., 2010). This OTU was observed in all the CWPD diseased tissues (Figure 5) and represented up to 2% of the sequences in diseased CWPD tissues. High abundances of Vibrio were previously found in another CCA species, P. pachydermum (21% of sequences), but without correlation to CCA diseases (Sneed et al., 2015). In our study, the vibrio OTU was present in three of the five healthy tissues of the CWPD individuals but was never detected in control individuals (Figure 5). This vibrio could thus be the putative agent causing CWPD.

Vibrios are ubiquitous and abundant in the aquatic environment (Thompson et al., 2004). Host endogen factors such as reproduction could play a crucial role in triggering vibriosis (Travers et al., 2009). However, the growth, behavior (Garren et al., 2016) and virulence (Kimes et al., 2012) of pathogenic vibrios in the natural environment are largely dictated by the temperature. The abundance of vibrios has therefore been proposed to represent a microbial barometer of climate change (Baker-Austin et al., 2017). The association between vibriosis and temperature supports our hypothesis of putative pathogenicity of vibrios because CWPD has recently been described as being likely to be temperature-dependent in Curaçao (Quéré et al., 2015b). Together, these results strengthen the hypothesis that CWPD is a temperature dependent disease.

The CWBS pathobiome was dominated by OTUs that belong to the Alphaproteobacteria class (Figure 4), most of which were Rhodobacterales (Table 1). The three OTUs, OTU 23, OTU 172 and OTU 239, represented up to 13, 6, and 1.8%, respectively, of the sequences obtained from diseased CWBS tissues (Figure 5). We did not identify any potential pathogens in the CWBS community, such as cyanobacteria known to be a causative agent of the black band disease in corals (Casamatta et al., 2012), but Rhodobacterales OTUs had close similarity (98%) to sequences found in corals affected by microbial originating white plague or black band disease (Frias-Lopez et al., 2002; Sunagawa et al., 2009). Thus, they could represent opportunist bacteria thriving on dying tissues. They may invade the depleted, diseased CCA cells, which have been shown to contain higher abundances of bioeroders by histological analysis, including boring sponges, helminths, and cyanobacteria (Quéré et al., 2015a). Such as in coral affected by white syndromes disease (Pollock et al., 2017), opportunist bacteria could play a role in pathogenesis and/or serve as a diagnostic criterion for disease differentiation. Further investigations of the pathobiome and/or viral component of the CWBS are required to reveal the causative agent of the disease.

The microbiome associated with control CCA might be specific to the species N. mamillare. Only 0.005% of the sequences found in N. mamillare were similar to those found in four different CCA species sampled in Belize, including Titanoderma textitprototypum and Hydrolithon boergesenii, which facilitate coral larval settlement, and P. solubile and P. pachydermum, which cause lower levels of coral settlement (Ritson-Williams et al., 2014). N. mamillare is not known to induce high coral settlements (Quéré et al., 2015b). The presence of a high proportion of cyanobacteria (average of 20% on healthy tissues, Figure 2) in N. mamillare, P. solubile, and P. pachydermum feeds the hypothesis that a high abundance of cyanobacteria in certain CCA species could inhibit coral settlements (Sneed et al., 2015). Benthic cyanobacteria, especially those in the order Oscillatoriales, are known to produce potent allelopathic compounds and inhibit coral settlements (Kuffner and Paul, 2004; Kuffner et al., 2006).

In summary, the two different CCA diseases appear to have different causes. The patch disease could be of bacterial origin as shown by the presence of Vibrio tubiashii, whereas the origin (i.e., bacterial, viral, or environmental stress) of band syndrome, characterized by opportunistic bacteria, remains unknown.

A-LM conceived and designed the experiments, analyzed the data and wrote the paper. PG contributed reagents/material/analysis tools, analyzed the data and wrote the paper. MN contributed materials and wrote the paper. GQ wrote the paper.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.