Discussion
Selective elimination of candidate pathogens using a variety of antibiotics in controlled experiments can aid in the identification of specific roles of these organisms with respect to disease causation. This study shows that the specific form(s) of WS sampled in this study were the result of a polymicrobial disease associated with a number of potential pathogens (15 bacteria and 7 ciliates identified in this study) that were consistently associated with all samples of disease and absent from healthy samples. This result supports previous studies on WS (Sussman
et al.
2008; Luna
et al.
2010; Sweet & Bythell
2012) as well as others such as white plague (Denner
et al.
2003; Pantos
et al.
2003; Bythell
et al.
2004; Barash
et al.
2005; Efrony
et al.
2009; Atad
et al.
2012) and black band disease (Cooney
et al.
2002; Frias‐Lopez
et al.
2003) which show multiple specific associates of any given diseased state. The combined elimination of three bacteria and one ciliate ribosomal type from this group of potential pathogens resulted in immediate and complete cessation of disease lesion progression, strongly indicating that one or a consortium of more than one (possibly all 4) of these candidate pathogens can be considered the ‘primary’ pathogen of WS in this instance. Only one of these candidate pathogens has been implicated in WS before, the ciliate
Philaster lucinda (Sweet & Bythell
2012), which remains the only specific pathogen associated with the disease in all cases studied to date. However, as we have recently concluded for WBD in Caribbean acroporas,
Philaster lucinda is more likely to be a ‘secondary’ pathogen contributing to pathogenesis and not a primary pathogen as the lesion continued to progress even though the ciliate was reduced to undetectable levels in the metronidazole treatment.
To date, the histophagous ciliate,
Philaster lucinda, has not been detected in association with healthy corals in the natural environment, yet has been shown to be consistently present in the neighbouring reef environment, particularly associated with numerous marine algal species (Sweet
et al.
2013b). It is therefore possible that they remain in these nearby reservoirs until the coral becomes compromised. With this in mind and the fact that two of the antibiotics utilized caused the lesion progression to cessate, the primary pathogen, at least in this instance, appears to be one or a combination of up to three of the bacterial agents. In the absence of the histophagous ciliate, these bacteria cause a degree of tissue necrosis that promotes lesion progression. In the natural disease state, these compromised tissues are apparently consumed by ciliates before necrosis occurs. However, without fine‐scale temporal sampling at the initiation of infection, it cannot be determined whether prior physical disruption by ciliates, or some other primary agent, may allow these specific bacterial agents to infect the tissues, and they may actually still be secondary agents, in a complicated disease process.
In this study, transmission electron microscopy showed the appearance of stress associated with the coral's symbiotic algae in progressive lesions but not in healthy tissues or those treatments which successfully stopped the disease. The symbiotic algae are a vital part of the coral holobiont and dysfunctional or damaged symbionts appear to be part of the aetiology of this disease. Damage to the symbionts may represent an additional stress to the coral and potentially provide a vicious cycle, further reducing host defences. This result, in which the algae are shown to be affected by the disease, is again similar to our recent study on WBD in the Caribbean acroporids.
The general absence of a significant population of bacteria and a lack of bacterial‐induced necrosis at the disease lesion interface of WS and similar ‘white diseases’ (Ainsworth
et al.
2007; Work & Aeby
2011; Sweet & Bythell
2012) have been paradoxical to the finding of multiple bacterial causal agents for the disease by challenge experiments and culture‐independent studies (Ben‐Haim & Rosenberg
2002; Denner
et al.
2003; Barash
et al.
2005; Gil‐Agudelo
et al.
2006; Pantos & Bythell
2006; Luna
et al.
2010; Kline & Vollmer
2011; Sweet & Bythell
2012). This finding was also reflected in the present study. The altered histopathology that occurred in the metronidazole treatment may be explained by the lack of ciliates associated with this treatment, which would otherwise have consumed the tissues before necrosis could be detected. In the metronidazole‐treated corals, we were also able to locate numerous rod‐shaped bacteria associated with these fragmented tissues. This may explain why previous studies have failed to find bacteria associated with WS lesions, as infected tissues are rapidly consumed by the ciliates.
One further finding of note is the presence of
Bacteriovorax in healthy tissues but absent in diseased samples.
Bacteriovorax are predatory bacteria whose presence in host organisms is thought to shape the natural microbial community through trophic interactions (Thurber
et al.
2009; Chen
et al.
2012). Particularly,
Bacteriovorax are known to prey on Gram‐negative bacteria such as members of the genus
Vibrio. Its dynamics may therefore be of importance in maintaining a healthy population of microbes, and its subsequent absence in diseased tissues may then allow for pathogenic bacteria to infect (Chen
et al.
2012).
Although detection of viruses was not a focus of this study, two main types were observed to be associated with the samples. One, circular, electron dense cored capsid VLP was consistently found in all sample types and appeared to be associated with the coral's symbiotic algae. Similar VLPs have previously been reported in
Acropora hyacinthus (Pollock
et al.
2014) and found in both healthy and diseased tissues. In contrast, the occurrence of a unique, larger VLP only present in metronidazole‐treated samples warrants further investigation. While this result is not sufficient to imply a role in disease causation, an increasing number of studies are showing the presence of viruses associated with disease lesions (Lohr
et al.
2007; Lawrence
et al.
2014a,
b; Pollock
et al.
2014; Yvan
et al.
2014) and their role as a potential pathogen requires further study. However, interestingly, the study by Lawrence
et al. (
2014b) also highlighted the presence of ciliates within the diseased tissues of corals showing signs of
Porites tissue loss (PorTL), a disease that fits within the generic description of WS disease.
Significant questions remain over the similarity between WS in the Indo‐Pacific and WBD in the Caribbean and between geographically and temporally separated WS cases. All these diseases are associated with the specific ciliate
Philaster lucinda, and it remains the only agent with direct evidence for involvement in pathogenesis as it has been observed to consume intact coral tissues at the lesion interface (Sweet & Bythell
2012). However, we can conclude that the ciliate is a secondary pathogen due to its responses to antibiotic treatment. The diseases are also associated with a specific bacterial consortium that is absent in healthy coral and present in every case of the disease observed at any particular site. Thus, there is no distinction between WBD in the Caribbean and the equally variable geographical presentation of WS in the Pacific, either in associated microbial communities, histopathology or visible disease signs. The question then remains as to whether the different white syndromes and WBD consist of separate types of disease or are a single disease. It is possible that they are clearly separate diseases, caused by specific bacteria, but that the secondary, specific ciliate pathogen and other nonspecific pathogens cause disease signs that are indistinguishable. The fact that none of the specific agents associated with WS in Fiji in this study were previously detected in corals showing signs of WS in the GBR or Solomon Islands (Sweet & Bythell
2012) might suggest that they are separate diseases associated with specific pathogens. However, it is also possible that the bacterial infections are nonspecific and opportunistic (Lesser
et al.
2007) and that any of a wide range of potential pathogens can act to cause initial infection and weaken host defences to allow the ciliate pathogens to invade. However, the fact that the bacterial consortium associated with the disease is so consistent between cases (sometimes separated by several km) and that none of these agents can be detected in healthy coral, even with deep sequencing, suggests that WS is not a completely nonspecific infection. Whatever the case, these diseases are certainly polymicrobial diseases, not only with bacteria and ciliates playing different roles in pathogenesis, but with multiple, specific bacterial associates consistently associated with the disease at a particular location and time. Further work is needed to determine whether this is simply a result of the ‘beta diversity’ associated with coral reef environments (whatever specific bacterial pathogens are available in the environment at any one time cause the disease) or whether specific disease consortia are associated with a number of specific diseases. It is also possible that the disease consortia are continually changing, possibly in response to changes in the host defences. This latter hypothesis differs from the hologenome theory of coral disease evolution (Rosenberg
et al.
2007) in that changes in microbial consortia rather than individual pathogen selection and evolution would be required to enable coral diseases to persist in the environment.
