Comment on a perspective: Molecular detections of new agents in finfish—Interpreting biological significance for fish health management

Emerging infectious diseases in wildlife populations—and fish in particular—have drawn increasing attention in recent decades (Daszak et al. 2000; Alexander and McNutt 2010; Miller et al. 2014; Tompkins et al. 2015; Krkošek 2017), especially as climate change and biodiversity loss present a shifting context that will almost certainly affect disease dynamics across taxa (Daszak et al. 2000; Miller et al. 2014). Management of infectious disease is especially important when populations of conservation concern are potentially impacted. In this context, healthy and transparent discussion and debate about how scientists, veterinary professionals, conservationists, policymakers, and regulators approach emerging infectious diseases in wildlife are much needed.

Primarily, we are concerned that the perspective offered by Meyers and Hickey (2022) does not adequately consider precautionary management or the conservation status of wildlife populations. Additionally, although Meyers and Hickey recognize some of the potential power of molecular technologies (e.g., next-generation DNA sequencing and quantitative polymerase chain reaction [qPCR]) in finfish disease research, they make several statements and present misconceptions that require clarification, including instances in which relevant details of our own work (Mordecai et al. 2019) are misrepresented or inappropriately discounted.

Drawing on consultation with a confidential group of fish health professionals and administrators, Meyers and Hickey propose that upon molecular detection of a novel infectious agent in fish or shellfish (or a known agent in a new host or setting), a decision pathway should be followed to avoid wasting resources prior to any changes to regulatory policy. Their proposed pathway suggests that an infectious agent should be detected with multiple assays, should be shown to replicate in its host, should be associated with clinical disease and mortality, and should fulfill Koch's postulates of disease causality. In theory, this series of scientific confirmations, which are to be addressed prior to enacting changes in surveillance or regulation, would inform a conceptually valid decision-making pathway. While we welcome guidelines to facilitate management of emerging infectious disease (World Organisation for Animal Health 2021), we argue that the criteria proposed by Meyers and Hickey fail to capture the complex reality of managing emerging infectious diseases in wildlife populations, particularly those of conservation concern, and are not suitable in practice.

Here lies perhaps the most salient point of what we see as our disagreement with Meyers and Hickey: while the decision pathway that they propose seems (to us) to err on the side of avoiding false positives (where infectious agent risks are truly low and changes to policy or management would be overly cautious, thus wasting resources), we are concerned that the proposed pathway is likely to result in false negatives (where early indications of harm fail to meet a threshold, but policy changes are not implemented in time, causing undue risks to wildlife). To our way of thinking, avoiding such false negatives aligns well with the precautionary approach (Persson 2016) that is widely deployed in conservation-focused management and even government policy (Punt 2008; Fisheries and Oceans Canada 2013). Indeed, we would contend that the fields of endangered species conservation and human health, as examples, are already built around this approach—not to wait until the evidence is definitive but the opportunity to act has passed—even if it is imperfectly applied in practice. The international response to the global COVID-19 pandemic provides a recent concrete example in which drastic policy changes resulted largely from epidemiological information about a virus (severe acute respiratory syndrome coronavirus 2 [SARS-CoV-2]) that did not fulfill Koch's postulates of cause and effect until well into the pandemic, when a human challenge trial was finally carried out (Killingley et al. 2022). From a public health perspective, measures to curtail infections were justified but might not have been sanctioned under a strictly applied human health analog to the pathway proposed by Meyers and Hickey.

Translation of scientific study into medical or management action is sometimes referred to as the “Valley of Death” because crossing this gap is time consuming and expensive (Butler 2008; Peters et al. 2019). We argue that rather than a universal predetermined pathway, a case-by-case approach that is responsive to multiple factors, including potential risks to wildlife or human health, would be much more prudent (e.g., Alexander and McNutt 2010). Indeed, the diversity of events, circumstances, and conditions that influence wildlife health outcomes necessitates a pragmatic approach wherein cause-and-effect relationships can provide guidance, but ultimately (as Meyers and Hickey also recognize) some degree of experience and opinion is needed to assess whether sufficient evidence exists to enact management measures (Stephen 2022). Meyers and Hickey also acknowledge that evidence takes time to gather and, in the meantime, interim actions on a case-by-case basis may be required. However, from our perspective, the apparent requirement in their proposed decision pathway for demonstrable significant clinical disease and mortality to occur in laboratory challenge studies before action is taken in these cases fails to adequately acknowledge relevant ecological complexity (as discussed below). Further, while formal satisfaction of disease causation (e.g., via laboratory challenge trials) is useful, a decision tree (such as that proposed by Meyers and Hickey) can overlook the complex reality of disease within an ecological setting, something that is difficult to replicate in a laboratory (Stephen 2022). We suggest that this is especially important pertaining to the application of the precautionary principle when managing wild populations at risk of local extirpation or extinction. Below, we describe four key themes from Meyers and Hickey (2022) that call into question the utility and transparency of their suggested approach. Since the example of piscine orthoreovirus (PRV) was used as a case study by Meyers and Hickey, we include a synopsis of why we believe their conclusions that PRV is nonpathogenic and poses a low risk to Pacific salmon Oncorhynchus spp. are unfounded (see Appendix A.1). Below, we attempt to provide balance to the perspectives and evidence discussed by Meyers and Hickey rather than providing a comprehensive review, and we invite readers to consider Meyers and Hickey's points in concert with our own.

Meyers and Hickey (2022) note that in some instances, upon discovery of a newly identified infectious agent, results are publicized prior to evaluation of the biological importance of the discovery. They discuss some of the associated risks (e.g., waste of resources). Our perspective is that transparency and open data sharing are integral to the modern scientific process and that the reporting of initial results need not preclude appropriate scientific follow-up (e.g., as suggested by Meyers and Hickey). Aside from the loss of public trust that failing to disclose new agents might incur, decisions to not share research findings come at a cost to the advancement of science. That advancement can, in turn, benefit fish and human health. New discoveries in fish contribute to our overall understanding of viruses, including families of viruses that cause serious disease in humans. For example, cutthroat trout virus (CTV) is related to hepatitis E virus (HEV; a human pathogen), and its discovery was seen as a “promising tool to elucidate aspects of the replication cycle of Hepeviridae in general, and HEV in particular” (von Nordheim et al. 2016).

Science is iterative, and timely publication of initial results enables progress in the understanding and refinement of ideas. For instance, when CTV was first discovered, it was initially not associated with disease in fish (Batts et al. 2011), but a new strain of the virus (CTV-2) discovered in farmed salmon populations in British Columbia (BC), Canada, exhibiting a molecular indicator of viral disease (Miller et al. 2017) has since been associated with brain infection and (in certain cases) elevated levels in dying Atlantic Salmon Salmo salar on BC farms (Mordecai et al. 2020; Bateman et al. 2021). A controlled challenge study examining the impact of infection on Pacific salmon found that CTV-2 resulted in persistent infections and was significantly associated with mild lesions, but the study failed to demonstrate mortality (Long et al. 2021). Importantly, this study did not include assessment of brain tissue, an important site of infection (Mordecai et al. 2020; Lhomme et al. 2021). Moreover, this provides a pertinent example of how a single challenge study, although well meaning, may miss the complexities of disease etiology, potentially resulting in a false conclusion of minimal impact on the host. We argue that additional evaluation of the potential for this virus to impact physiological performance and disease is warranted.

Scientific study is an iterative process and relies on the sharing of results and data so that other researchers can scrutinize, attempt to replicate, and build upon findings. With the availability of modern molecular tools, the rate of infectious agent discovery far exceeds the capacity of researchers in a relatively small field (e.g., fish health) to determine biological importance in all possible hosts. According to the precautionary principle, policy needs to be proactive: anticipating risk and responding to emerging science (United Nations Economic Commission for Europe 1990). Some decision makers, such as those from Fisheries and Oceans Canada (DFO), are legally bound to act in a precautionary manner with respect to the protection of wild fish, and they must not use “the absence of adequate scientific information as a reason to postpone [action] or fail to take action to avoid serious harm” (Fisheries and Oceans Canada 2009). This logical fallacy of basing an argument on an absence of evidence is often used to argue for evidence of absence of an impact for instances in which no specific studies on the issue have been conducted. For example, the bacterium Tenacibaculum maritimum has been demonstrated to cause disease in marine fish around the world (Santos et al. 2022), including Pacific salmonids in California (Chen et al. 1995), Chile (Sandoval 2020; Valdes et al. 2021), and Alaska (Meyers et al. 2019). However, DFO fish health scientists in BC have argued that the disease must be demonstrated in BC salmon to be considered a risk (Canadian Science Advisory Secretariat 2020). This pathogen is widespread on Atlantic Salmon farms in BC, where it can cause acute oral ulcers (known as “mouth rot” disease) and death within days of ocean entry (Frisch et al. 2018; Canadian Science Advisory Secretariat 2020). In some instances, advice by DFO fish health researchers cites the low occurrence of mouth rot in Pacific salmon species as evidence of minimal risk (Canadian Science Advisory Secretariat 2020), despite a body of work linking the pathogen with disease in Pacific salmon species that manifests differently from the “mouth rot” manifestation, which is apparently unique to Atlantic Salmon (Chen et al. 1995; Meyers et al. 2019; Sandoval 2020; Valdes et al. 2021).

We are not advocating for the universal application of stringent precautions to all infectious agents, and we emphasize that absence of definitive evidence of harm (e.g., via satisfaction of Koch’s postulates) obviously does not itself justify conservation or management action. As Meyers and Hickey point out, “proving the negative”—showing that an infective agent is benign in every possible scenario—is unreasonable, and evidence gaps will persist even when a justifiable decision has been reached. Our point is that compelling correlational, ecological, or epidemiological findings, along with results from closely related systems, are sources of evidence that need to be taken into account when making policy decisions, especially where the risks to wildlife could be great.

Meyers and Hickey (2022) make a series of statements that we believe oversimplify the study of disease processes in wild animals, and they do not consider the conservation status of potentially affected populations. For example, they propose a requirement for a disease agent to have been associated with significant clinical disease and/or mortality in a laboratory setting before further management action is taken. This approach may be appropriate for managing an aquaculture facility for production purposes, but it does not capture the ecological complexity of factors that influence the survival of wild animals. The framework described by Meyers and Hickey calls for policy review only after the pathogenicity of emerging infectious agents has been determined via Koch's postulates. Two centuries after these postulates were formulated, our understanding of disease has improved substantially. Indeed, even Koch himself was aware of the limitation imposed by his postulates, as he was unable to fulfill them for the microbes responsible for cholera (found in healthy individuals) and leprosy (unculturable), which he correctly suspected resulted from disease-causing microbes (Fredricks and Relman 1996). Although Meyers and Hickey cite a more recent perspective of determining disease causality (Fredricks and Relman 1996), they fail to incorporate these updated views or acknowledge ecological complexity (Stephen 2022) in their decision pathway. In fact, Meyers and Hickey suggest that an infectious agent may be of low concern if it is detected in healthy or asymptomatic individuals—an argument that is at odds with the existence of asymptomatic spreaders, which is now common knowledge as a result of the recent COVID-19 pandemic (Mizumoto et al. 2020) but has been known since the time of Typhoid Mary (Marineli et al. 2013).

The decision pathway of Meyers and Hickey seems to imply that a lack of clinical signs of disease or mortality implies lower risk. While we agree on the importance of laboratory challenge studies to demonstrate cause-and-effect relationships between infectious agents and disease, it is fundamental that such studies are designed to detect ecologically relevant impacts of both acute and chronic agents. Reliance on a rapid onset of clinical signs of disease and/or demonstrable mortality in laboratory settings may, in fact, lead to false conclusions of a lack of cause-and-effect relationship with disease (Mordecai et al. 2023; see Appendix A.1).

We argue that an over-reliance on observable mortality has impeded progress in understanding the role of infectious disease in wild populations since hosts that are infected by pathogens are more vulnerable (e.g., to predation; Seppälä et al. 2004; Johnson et al. 2006; Miller et al. 2014; Furey et al. 2021). Sublethal impacts of disease that lead to host removal from wild populations reduce observed infection prevalence in those populations and further impede the ability to sample fish (or other animals) in the late stages of disease. It is for these reasons that, rather than relying on observations of clinical disease and mortality, our team and others exploit highly sensitive molecular approaches to address wildlife disease (e.g., Benton et al. 2015). We agree with Meyers and Hickey that these molecular techniques are important first steps in identifying pathogenic potential in free-ranging species. Tools applied by our group include molecular detection of novel or known agents (Miller et al. 2014, 2016; Mordecai et al. 2019, 2020), quantification of transcriptional activity associated with infection (e.g., activation of a viral disease response; Miller et al. 2017), and localization of agents within cells showing evidence of damage via in situ hybridization (Di Cicco et al. 2018; Mordecai et al. 2020). Large-scale pathogen surveillance of wild populations, holding studies, and acoustic tracking studies can further enable population-level modeling to provide key evidence of an infectious agent's impact (Miller et al. 2014; Bass et al. 2022).

Our disagreement with Meyers and Hickey arises because, to us, their recommended pathway does not adequately account for an epidemiological perspective or the interplay among the host, the environment, and the agent (Hanson 1988) to inform management policy changes. The suggested pathway also does not adequately incorporate the conservation status of potentially impacted populations or the known role that subclinical and asymptomatic individuals play in the transmission of infectious disease (Pantin-Jackwood and Swayne 2009; Marineli et al. 2013; Mizumoto et al. 2020). We acknowledge Meyers and Hickey's proposal that epidemiological information or “low evidence of disease” could be brought to bear via “a health risk assessment by multi-agency regional fish health professionals.” From our perspective, however, such a reliance on one type of practitioner can (although it need not) result in siloed thinking. Diverse perspectives (e.g., from quantitative ecologists or theoreticians) are valuable to avoid entrenched—and possibly inaccurate—thinking in any group. Further, existing risk assessment structures may not be up to the task as they are subject to manipulation by industrial or other non-scientific interests (e.g., the Canadian Science Advisory Secretariat process; Godwin et al. 2023).

In assessing whether infectious agents warrant management interventions, Meyers and Hickey (2022) seem to accept differing standards of evidence. For example, they disregarded a qPCR detection of Myxobolus cerebralis, despite confirmatory sequencing (Arsan et al. 2007). In contrast, in dating the presence of PRV in BC (which they argue predates salmon farming in the province and therefore poses low risk), they rely on weak detections from a single fish (Marty et al. 2015), with no published analysis of the confirmatory sequences, which were identical to more recently sequenced samples. This is despite their own guidelines, which dictate that the possibilities of false positives and contamination need to be considered.

Meyers and Hickey propose that no management action should be taken without fulfilling Koch's postulates and observing clinical disease and mortality. Meanwhile, Meyers and Hickey argue that science is never absolute, and they cite requests from the public and media that regulatory agencies provide “proof of the negative” (i.e., proof that an agent is zero risk). They state that demonstrating zero risk is an impossible task with the complexities of uncontrolled natural environments and diverse hosts. Although we by no means claim that every newly discovered infectious agent poses a risk to aquatic organisms, “consistency of association” that suggests the potential for impact (i.e., epidemiological evidence; Hill 1965) highly warrants (and, in some cases, legally requires; Morton v. Canada (Fisheries and Oceans) 2019) a precautionary approach, especially when considering species and populations of conservation concern.

Although viral families appear to often maintain tissue tropism over millions of years of evolution (Zhang et al. 2018), the taxonomic genus and the geographic origin of a virus are not predictive of its pathogenicity since viral pathogenicity is reliant on a multitude of factors (Rosenberg 2015). However, when confronted with the discovery of a novel aquareovirus in Chinook Salmon O. tshawytscha, Meyers and Hickey point out that related viruses in the northeast Pacific region are not considered pathogens, and they insinuate that any new variants of aquareoviruses are likely to have inconsequential impacts on their host. However, this suggestion—that taxonomic relatedness is suitable as a prediction of viral pathogenic potential—contradicts their contention that PRV-1a, which is closely related to the established PRV-1b pathogen, is benign. Importantly, some reoviruses infecting vertebrates cause disease, while others do not, and the closest relatives to the lineage of PRV in the eastern Pacific have all been shown to cause disease (Takano et al. 2016; Wessel et al. 2017, 2020; Vendramin et al. 2019).

Meyers and Hickey (2022) state that our RNA-sequencing-based (metatranscriptomic) viral discovery study (Mordecai et al. 2019) focused on mostly healthy fish. In fact, the majority of our sequences originated from “dead and moribund cultured Chinook Salmon” from farms and wild Chinook Salmon dying in a holding study in the Fraser River. All of these individuals tested negative for known viruses but exhibited a transcriptional viral disease signature (Miller et al. 2017) to identify fish in a virally induced immune response state that might (and did) harbor previously unidentified viruses.

Similarly, Meyers and Hickey erroneously state that the Pacific salmon nidovirus (PsNV) was detected only in apparently healthy fish. In fact, upon the discovery of PsNV and subsequent screening, dead and dying farmed Chinook Salmon had the highest proportion of positive detections out of any of the groups tested (Mordecai et al. 2019). Further, a more recent study localized PsNV to the gills, some of which showed disease lesions (Mordecai et al. 2020), offering preliminary evidence that this virus may be associated with respiratory disease, reminiscent of the disease caused by many of the mammalian coronaviruses.

While we agree that further study is needed to determine the role of newly discovered viruses in disease, Meyers and Hickey's framing of our viral discovery work is unfortunate.

Molecular tools for epidemiology are increasingly applied to understand disease processes in agriculture, aquaculture, and ecological settings (Benton et al. 2015). The COVID-19 pandemic changed the way in which these technologies are applied to human public health (Oude Munnink et al. 2021), and it is time for them to be similarly acknowledged by regulators in the context of emerging wildlife disease. We agree that a discussion of the limitations of molecular technologies and an associated series of guiding principles (see Appendix A.2)—effectively integrating veterinary, epidemiological, molecular, ecological, and other perspectives—could help policymakers to navigate these new streams of available evidence. We do not, however, consider the decision pathway for determining pathogenicity, as developed by Meyers and Hickey (2022), to be fit for purpose. Although the suggested approach likely has merit in animal husbandry, it is not suitable for the management, stewardship, or conservation of wild populations, which face a multitude of interacting and cumulative stressors (Stephen 2022).

The salient overarching dichotomy that we identified in the Introduction—whether to err on the side of precautionary wildlife management, naturally admitting a higher tolerance for false alarms, or to err on the side of reduced costs, awaiting certainty of harm before enacting major policy change that may come too late—itself represents a policy decision. Of course, such extremes exist on a continuum, and any decision about where wildlife conservation or management policy should fall on that continuum is not one that should be made by scientists, conservationists, or fish health professionals alone. The choice is societal, requiring judgment about the values society places on economic responsibility, food production, biodiversity, and myriad other factors. Ultimately, the availability of information and open, honest debate are critical to inform these important decisions.

Fisheries and Oceans Canada has a mandate to sustainably manage fisheries and aquaculture. The Pacific Salmon Foundation supports wild Pacific salmon in BC and the Yukon and has called for a move to closed-containment salmon aquaculture in BC to protect wild salmon from associated risks.

There were no ethical guidelines applicable to this article.