Reviews in Fisheries Science, 13:205–230, 2005 Copyright © Taylor & Francis Inc. ISSN: 1064-1262 print DOI: 10.1080/10641260500326842
A Risk Assessment Based Approach for the Management of Whirling Disease JERRI L. BARTHOLOMEW,1 BILLIE L. KERANS,2 RONALD P. HEDRICK,3 STUART C. MACDIARMID,4 AND JAMES R. WINTON5 1
Department of Microbiology and Center for Fish Disease Research, Oregon State University, Corvallis, Oregon, USA 2 Department of Ecology, Montana State University, Bozeman, Montana, USA 3 Department of Medicine and Epidemiology, University of California, Davis, California, USA 4 Programme Development Group, New Zealand Food Safety Authority, Wellington, New Zealand 5 Western Fisheries Research Center, U.S. Geological Survey, Seattle, Washington, USA This article began as an exercise during the 9th Annual Whirling Disease Symposium, “Whirling Disease: Managing the Risk” held in Seattle, Washington, in 2003. The exercise was designed to address the needs of fishery managers to more accurately assess the various factors associated with the risks of the introduction and the establishment of the myxosporean parasite, Myxobolus cerebralis, and the development of whirling disease in salmonids. Here we introduce concepts used in risk assessment, review aspects of whirling disease relevant to risk assessment, present a working model for a whirling disease risk assessment, and work through examples of how this model might be used to estimate risks and identify actions to reduce these risks. While this approach was designed to address factors controlling the introduction and establishment of M. cerebralis and the development of whirling disease in salmonids, the concepts are highly relevant to the management of other infectious diseases of aquatic animals. Keywords Myxobolus cerebralis, risk assessment, salmon, trout, parasite, disease management,
Introduction Myxobolus cerebralis, the cause of salmonid whirling disease, is present in many regions of the U.S. and its distribution continues to expand (Bartholomew and Reno, 2002). In some areas, the effect of the parasite on wild and cultured trout populations appears limited; however, in certain waters of the intermountain U.S., declines in rainbow trout (Oncorhynchus mykiss) and cutthroat trout (Oncorhynchus clarki) populations continue, and whirling disease remains a serious management concern (Nehring and Walker, 1996; Vincent, 1996). Address correspondence to Jerri L. Bartholomew, Department of Microbiology and Center for Fish Disease Research, 220 Nash Hall, Oregon State University, Corvallis, OR 97331, USA. E-mail: [email protected]
J. L. Bartholomew et al.
This has resulted in research to understand the ecological relationships between the parasite, its two hosts (salmonid and aquatic oligochaete), and the environment, with the goal of developing or identifying a means to prevent the spread of M. cerebralis or reduce its effects where already present. As a result of this research, we can now begin to determine the extent to which various factors influence the introduction and spread of the parasite and the development of whirling disease in salmonids. The interactions among these factors and the potential outcomes can be evaluated using the principles of risk analysis, which allows researchers and managers to integrate data from disparate sources and evaluate alternative management scenarios. Risk analysis, as a formal discipline, has been taken up only relatively recently by those working in the animal health field (MacDiarmid, 1997); however, several useful texts are available to assist the newcomer (Vose, 1997, 2000; Rodgers, 2000; Murray, 2002). This article introduces concepts used in risk assessment, which is a component of risk analysis, reviews aspects of whirling disease relevant to risk assessment, presents a working model for a whirling disease risk assessment, and works through examples of how this model might be used by managers to estimate risks and to identify the most effective actions to reduce these risks. There have been several examples of the application of risk methodology to aquatic animal diseases. Those relevant to the introduction of aquatic pathogens include work by Paisley (2001) for the monogenetic trematode Gyrodactylus salaris that has negatively impacted populations of Atlantic salmon in Norway. For whirling disease, Bruneau (2001) reported a quantitative risk assessment for the introduction of M. cerebralis into Alberta, Canada, via the importation of live salmonids. Another study by Hiner and Moffitt (2002) created a model to test habitat variables as predictors of whirling disease severity. Here, we introduce the concepts of risk assessment as a means to systematically identify and evaluate critical areas for the management of whirling disease. In addition, we illustrate how these concepts can be used to evaluate options that will allow managers to focus on the most important pathways of introduction and the most cost-effective methods for controlling the parasite if introduced. In the future, it is expected that new information will improve our ability to assess factors associated with the risk of the introduction and establishment of the parasite. As a result, this exercise also highlights the ability of a risk assessment to identify areas where more knowledge is needed to strengthen future risk analyses.
Risk Analysis Risk analysis is a tool intended to provide decision–makers with an objective, repeatable, and documented assessment of the risks posed by a particular course of action (MacDiarmid, 1997). Risk analysis is intended to answer the following questions:
r What can go wrong?—the hazard identification, r How likely is it to go wrong and what would be the consequences of it going wrong?— the risk assessment, and
r What can be done to reduce either the likelihood or the consequences of it going wrong?—risk management. Working through this process is an important aspect of decision-making. As an example, when facing the choice between two surveillance programs, a manager can choose between a highly sensitive but expensive test or a less sensitive but cheaper assay (MacDiarmid and Hellstr¨om, 1988). In this case, the “What can go wrong?” is that the surveillance program
Whirling Disease Risk Assessment
might fail to detect an infection. “How likely is it to go wrong?” and “What would be the consequences?” are answered by the risk assessment. A risk analysis is, in effect, a type of map (MacDiarmid, 2001). For example, having identified a hazard (e.g., a disease agent) in one country or region, a risk assessment may model the various pathways by which that hazard could travel from infected hosts in that country or region into susceptible hosts in another country or region (Vose, 1997; MacDiarmid et al., 2001; MacDiarmid and Pharo, 2003). Finally, risk management protocols are formulated based on the risk assessment. These protocols are designed to reduce the likelihood of the unwanted event occurring, or the magnitude of its consequences.
Hazard Identification Because the International Aquatic Animal Health Code of the Office International des Epizooties (OIE, 2003) is focused on trade, hazard identification is defined as “the process of identifying any pathogenic agents which could potentially be introduced in the commodity considered for importation.” However, risk analysis is equally applicable to other areas of decision-making, such as those affecting disease surveillance or control programs, and so hazard identification may be considered as the step of identifying what it is that might go wrong in whatever activity is being considered. The present discussion is concerned with a single hazard, the parasite M. cerebralis.
Risk Assessment Risk assessment is “the evaluation of the likelihood and the biological and economic consequences of entry, establishment, or spread of a hazard within the territory of an importing country” (OIE, 2003). However, the same assessment processes apply if one were considering the likelihood of M. cerebralis being introduced or escaping detection under different surveillance strategies (Thorburn, 1996; MacDiarmid, 2001; Bruneau, 2001; Williams and Moffitt, 2001). In this risk assessment, we focus on the biological consequences (e.g., disease in fish) and the likely pathways related to disease. Risk assessment may be qualitative, in which the likelihood of the outcome, or the magnitude of the consequences, is classified on a descriptive scale as described below (Moutou et al., 2001; Nowak et al., 2003):
r Extreme—the event is almost certain to occur (p > 0.9 or there is more than a nine in 10 chance)
r High—the event would be expected to occur (0.9 > p > 0.5 or there is more than a one in two chance but less than a nine in ten chance)
r Moderate—there is less than an even chance of the event occurring (0.5 > p > 0.1 or there is less than a one in two chance, but more than a one in ten chance)
r Low—the event is unlikely to occur (0.1 > p > 0.01 or there is less then a one in ten chance, but more than a one in 100 chance)
r Very low—the event would occur rarely (0.01 > p > 0.0001 or there is less than a one in a hundred chance but more than a one in ten thousand chance)
r Extremely low—the event would occur very rarely (0.0001 > p > 0.000001 or there is less than a one in ten thousand chance, but more than a one in a million chance)
r Negligible—chance of event occurring is so small that it can be ignored in practical terms (p > 0.000001 or there is less than a one in a million chance)
J. L. Bartholomew et al.
or it may be quantitative. In quantitative risk assessments, the likelihood is expressed in terms such as “one disease introduction in 100 years of trade” or “failure to correctly identify one diseased aquaculture establishment out of 100.” Both qualitative and quantitative approaches to risk assessment are valid and, in fact, every risk assessment must first be carried out qualitatively (Vose, 2001). Only if further insight is required is it necessary to attempt to quantify the risk. Indeed, as North (1995) puts it, quantitative “. . . risk analysis is best used to develop insights, and not to develop numerical results which might mistakenly be considered to be highly precise.” According to the OIE Code (OIE, 2003), risk assessment consists of four steps:
r r r r
release assessment exposure assessment consequence assessment risk estimation
Release Assessment. A release assessment (OIE, 2003) consists of describing the pathway(s) necessary for an activity to “release” (that is, introduce) M. cerebralis into a particular environment, and estimating the likelihood of that complete process occurring. Examples of the inputs required in a release assessment for M. cerebralis have been reviewed (Bartholomew and Reno, 2002; Hiner and Moffitt, 2002) and are summarized in Figure 1. Introduction of M. cerebralis, as with most pathogens, is typically accidental, although it often occurs with intentional movements of fish. If disinfected eggs are imported, there is a very low risk of introducing the parasitic, as vertical transmission does not occur. In the U.S., introductions of the parasite have most commonly occurred with transfers of subclinically
Figure 1. Schematic diagram demonstrating both direct and indirect pathways for the introduction, or release, of Myxobolus cerebralis.
Whirling Disease Risk Assessment
Table 1 Relationship between the life stage of the fish and the likelihood of introducing Myxobolus cerebralis with its movement Life stage Eggs∗ Fry, alevins Juveniles, adults
Source of parasite/parasite stage
Likelihood of detection
No Yes/immature parasite stages Yes/spores
NA Low to high Moderate to high
∗ Assumes disinfection and no transfer of water or material that might carry the parasite. NA = not applicable.
infected fish (Hoffman, 1970, 1990; Hedrick et al., 1998; Bartholomew and Reno, 2002). Not only is this the most common pathway for the introduction of M. cerebralis, but it is also the pathway over which managers have the greatest potential control. The risks associated with legal fish transfers are a result of not detecting the pathogen in the population. Bruneau (2001) showed that level of this risk was related to the prevalence of the pathogen in the population and the proportion of fish sampled as well as to the performance of the diagnostic test used. Pathogen levels in a population are in turn affected by species susceptibility (reviewed by MacConnell and Vincent, 2002). Susceptible species like rainbow trout are likely to carry larger numbers of parasites per infected fish (higher dose) and the prevalence of infection is likely to be greater (higher frequency) than for more resistant species. Similarly, the ability to detect the pathogen is affected by the age or life stage of the fish as well as by the sensitivity of the diagnostic test. For example, fry and alevins are likely to carry immature parasitic stages that may be difficult to detect, depending on the diagnostic method used (Table 1). Several diagnostic methods are available for detecting pathogen presence (reviewed by Andree et al., 2002; MacConnell, 2003). Bruneau (2001) quantitatively demonstrated the variability in uncertainty that occurs with the use of two of these methods: pepsin-trypsin digest and polymerase chain reaction (PCR). The importance of data quality will be further examined later in this article. Other factors affecting the risk of introduction with the movement of hatchery fish involve the nature of the source facility, its location and the history of the watershed. The planned movement of a hatchery stock that is determined to be M. cerebralis-free, but located within an endemic area, should be considered a greater risk than movement of fish from a hatchery in a watershed that extensive sampling has shown to be M. cerebralis-free. However, pathogen-free facilities certainly exist in watersheds where wild fish are infected, and therefore other factors such as hatchery water source, water treatment, and type of rearing unit (concrete, earthen) must also be considered. There is also potential for the introduction of M. cerebralis through less direct pathways that are under human control. Commercial routes for introduction of myxospores (the spore stages found in fish tissues) include sale of infected fish products for human consumption (Wolf and Markiw, 1982; Hoffman, 1990; El-Matbouli and Hoffman, 1991) and of infected trout as fish bait (D. J. Money, Wyoming Department of Game and Fish, Laramie, WY, personal communication), although the frequency of either practice is not well documented. Certain restoration activities (e.g., salmon carcass planting) may also constitute a high risk for introduction of myxospores. Introduction of both myxospores and actinospores (stage produced by the oligochaete host) might occur with recreational activities such as boating and angling, or other activities that could result in the transfer of mud or water from endemic to unaffected waters.
J. L. Bartholomew et al.
Introduction of the M. cerebralis actinospore via infected worms sold commercially has not been documented, although the movement of actinosporean stages of other myxozoan parasites with oligochaetes sold for ornamental fish food has been reported (Lowers and Bartholomew, 2003). The role of alternate hosts in the introduction and movement of invasive species is recognized and control measures often target these pathways. Potential introduction pathways for the oligochaete host, Tubifex tubifex, might include mud moved between drainages with boat trailers, equipment, and by commercial or restoration activities. Other routes for “human-assisted” introduction certainly exist and there are numerous anecdotes of illegal fish transfers that might account for certain unexplained introductions. The relative risk of introduction by each of these pathways is difficult to quantify, but the likelihood of introduction by indirect routes can be linked to geographic proximity to endemic waters (Schisler and Bergersen, 2002) and is likely to vary for different assessment scenarios. Introduction may also occur by natural dispersal of parasitic stages. In the wild, these pathways are difficult to control. Dispersal of myxospore and actinospore stages represent a potential introduction pathway downstream within a system, in drainages that are naturally connected, or when structures link otherwise unconnected waters (e.g., canals, drainage ditches). Although myxospore stages are physically adapted to settle out of the water column to increase the probability of encountering T. tubifex, the actinospore is adapted to remain suspended in the water column. This characteristic increases its potential for downstream dispersal, although this advantage may be offset by the shorter lifespan of the actinospore (Markiw, 1992; El-Matbouli et al., 1999). Natural dispersal may also occur with infected wild fish, and for large interconnected systems the life histories of salmonids in the watershed may affect how rapidly M. cerebralis is disseminated. In systems where fish have a resident life history, dispersal by this route may be limited. Fluvial and adfluvial life histories introduce greater potential for upstream or downstream movement because these fish migrate between spawning tributaries and a mainstem river or lake. Anadromous salmon may present the greatest potential for dispersal both within and between drainages. However, in addition to life history, the susceptibility of the species is also important. Species like coho salmon (Oncorhynchus kisutch), although anadromous, present a very low risk because of their low susceptibility to M. cerebralis. Viable myxospores have also been demonstrated to pass through the digestive tracts of piscivorous birds and fish (Taylor and Lott, 1978; El-Matbouli and Hoffman, 1991), although the role of avian vectors in disseminating M. cerebralis is unclear. Exposure Assessment. An exposure assessment (OIE, 2003) consists of describing the biological pathway(s) necessary to “expose” aquatic animals to M. cerebralis given that it has been introduced into an environment. In this case, the exposure assessment estimates the likelihood that M. cerebralis can establish, proliferate, and spread beyond the location of the introduction. Examples of the inputs required in an exposure assessment for M. cerebralis have been reviewed by Kerans and Zale (2002) and are summarized in Figure 2. Following the introduction of M. cerebralis to a new area, establishment and proliferation can only occur if the parasite encounters an appropriate host. The likelihood of this occurrence depends upon the magnitude and frequency of parasite introduction, as well as on the population density and susceptiblity of encountered hosts, the spatial and temporal overlap of the introduced parasite and the susceptible hosts, and the abiotic and biotic environment into which the parasite is introduced. Both the number of introductions and the parasite load can influence the establishment and proliferation of the parasite. Although little experimental information is available, there
Whirling Disease Risk Assessment
Figure 2. Schematic diagram depicting examples of inputs for assessing the likelihood that Myxobolus cerebralis establishes following introduction.
appears to be a positive relationship between the level of stocking of infected salmonids in lakes and the subsequent prevalence of infection and myxospore concentration of resident salmonids (Nehring and Thompson, 2000). In addition, if M. cerebralis is introduced through infected fish or through dispersal of myxospores within systems, it is likely that the parasite will survive long enough to encounter the oligochaete host because myxospores can survive for substantial periods of time outside the fish host (Halliday, 1976; El-Matbouli and Hoffman, 1991). Moreover, the likelihood that infectious myxospores will encounter a susceptible oligochaete will increase as the number of myxospores introduced increases. Although little is known about the numbers of myxospores required for M. cerebralis to establish, repeated introductions of large numbers of infected fish will increase the likelihood that M. cerebralis will become established. If M. cerebralis is introduced into the system through infected oligochaetes or through the dispersal of actinospores, then the parasite must survive long enough to encounter a susceptible salmonid. Although actinospores are more fragile and have a shorter lifespan than myxospores, they are adapted for downstream dispersal in the water column where they are more likely to encounter a fish host. Tubifex tubifex can live for almost 2 years, infections are persistent with recurring bouts of actinospore release, and infection appears to cause little worm mortality (Gilbert and Granath, 2001; Granath and Gilbert, 2002). Thus infected worms have the potential to release actinospores long after they are introduced. Population density and host susceptibility also influence the establishment and proliferation of M. cerebralis. The frequency of the encounter between parasite and host is dependent on host population density and proliferation of the parasite is dependent on host susceptibility. The likelihood that a water body contains the oligochaete host is high because T. tubifex is extremely tolerant of a wide range of environmental conditions and is found in almost every type of aquatic habitat (Brinkhurst and Jamieson, 1971). However, the abundance of T. tubifex can vary widely (from hundreds to hundreds of thousands of worms/m2 ). In streams, T. tubifex can be found in silty, backwater areas, in pools, and in low velocity side
J. L. Bartholomew et al.
channels, and is less likely to be abundant in rocky areas with high water velocities in the main channel (Lazim and Learner, 1987; Juget and Lafont, 1994; Krueger et al., in press). Genetically distinct strains of T. tubifex have been identified that differ dramatically in their susceptibility to M. cerebralis and in their ability to produce actinospores (Stevens et al., 2001; Beauchamp et al., 2002; Kerans et al., 2004). Many salmonid species are susceptible to the parasite (Hedrick et al., 1998; MacConnell and Vincent, 2002) and although not all species manifest signs of disease (see consequence assessment) or are of commercial or recreational interest, many would allow establishment of the parasite in a system if they came in contact with actinospores. In addition, parasite proliferation differs among fish host species. For example, Hedrick et al. (1999) showed that brown trout (Salmo trutta) produced fewer myxospores at a given actinospore dose than did rainbow trout. Similar results for Chinook salmon (Oncorhynchus tshawytscha) were demonstrated by Sollid et al. (2003). Thus, M. cerebralis is more likely to become established and proliferate in systems with large populations of T. tubifex that are dominated by highly susceptible worm strains and that contain large populations of highly susceptible salmonids. Abiotic environmental conditions, in particular, water temperature, sediment type, and water flows influence parasitic success either directly or through their influence on host ecology. Water temperature influences development of the parasite in both hosts (Halliday, 1973; Markiw, 1986; Blazer et al., 2003) and the survival and the timing of actinospore release (El-Matbouli et al., 1999). In Montana, Vincent (2002) indicated that severe whirling disease infections among rainbow trout in field exposures were associated with water temperatures between 12–16◦ C. The effect of temperature on actinospore levels in the water is less clear as high concentrations were found during winter months at locations in Colorado even when water temperatures were 0–5◦ C (R.B. Nehring, Colorado Division of Wildlife, Montrose, CO, personal communication). Thus, significant exposure to infective stages can occur over a rather broad range of water temperatures that overlap those needed for the successful natural reproduction of several species of trout and salmon. The effects of flow are important as well. Mean annual river flow was inversely correlated with infection severity in rainbow trout from the Madison River of Montana (E. R. Vincent, Montana Department of Fish, Wildlife and Parks, 1400 S. 19th Bozeman, MT 59715, personal communication). In that same river, Krueger et al. (in press) reported that side channels dominated by fine sediments and low flows contributed more to the parasite’s ability to infect sentinel rainbow trout than did side channels with the opposite characteristics. Conversely, the scouring effects of high flows on steep gradient streams may be one reason why the detrimental effects of whirling disease have not been reported from coastal streams in California (Modin, 1998). The establishment and proliferation of M. cerebralis are also dependent on conditions that maintain the continuity of the parasite life cycle and characteristics of the biotic community in which the hosts and parasite reside. For example, spatial and temporal overlaps between susceptible salmonids and viable actinospores are critical to the persistence of the parasite in natural systems (Downing et al., 2002). Close proximity of both hosts increases the probability of transmission from the worm to the fish and reduces the duration that the fragile actinospore is drifting and vulnerable. Variation in the species, subspecies or strains of salmonids or oligochaetes within the biotic community is important. Aquatic systems in which whirling disease may become endemic can contain multiple species of salmonids of varying resistance as well as multiple genotypes of T. tubifex of differing ability to produce the parasite. Thus, the composition of the strains and species of host may influence the success of the establishment of M. cerebralis in an aquatic system as well as the severity of infection in a particular target
Whirling Disease Risk Assessment
strain or species. For example, brown trout are reported to contribute to parasite success in systems in which they occur (Nehring et al., 2002), although they show few signs of disease. Consequently, all susceptible species should be considered when carrying out a risk assessment as each may provide different opportunities for management. In addition, other organisms may interact with the hosts or parasite in a complex aquatic environment. One example might be the elimination of viable myxospores from sediments following ingestion by resistant species of oligochaetes. Consequence Assessment. A consequence assessment consists of identifying the potential biological, environmental, and economic effects of a disease (OIE, 2003). A risk assessment for M. cerebralis might consider such consequences as production losses, facility closures, surveillance and control costs, effects on native species, etc. Reviews by Hedrick and El-Matbouli (2002) and MacConnell and Vincent (2002) are useful in conducting a consequence assessment. The inputs for making a consequence assessment are summarized in Figure 3. The consequences resulting from the successful introduction and establishment of M. cerebralis are dependent on a complex interaction among many variables involving not only the parasite but factors associated with the two hosts and the environmental conditions present (Travis and Hueston, 2001). The interactions of these variables have been often illustrated with either simple or more complex diagrams (Snieszko, 1976; Hedrick, 1998). Whirling disease provides a particularly interesting example of how these variables may interact, as radically different outcomes have been documented in watersheds where the
Figure 3. Schematic diagram depicting examples of inputs for assessing the likelihood that Myxobolus cerebralis establishment will result in disease.
J. L. Bartholomew et al.
pathogen and susceptible hosts are known to co-exist. In certain cases, there are no apparent effects on populations of susceptible hosts (Modin, 1998) while in others, catastrophic declines have been documented (Nehring and Walker, 1996; Vincent, 1996). Factors associated with the pathogen involve at a minimum the virulence, or the quality of being able to cause disease, and the dose of the pathogen the host(s) is apt to encounter. If a pathogen has low virulence, the impacts on a given population of fish will be relatively low and disease may only result, if at all, when the host is exposed to large numbers of the parasite. In contrast, a pathogen that is highly virulent for a host may have immediate impacts on the population by causing disease and mortality. In a laboratory challenge using geographically distinct isolates of M. cerebralis, there was no evidence for differences in virulence (T. McDowell, Department of Medicine and Epidemiology, University of California, Davis, CA, personal communication). Field observations also suggest that few qualitative differences in virulence exist between geographically diverse isolates of the parasite. To date, genetic approaches have demonstrated little variation between isolates of M. cerebralis (Andree et al., 1999; Whipps et al., 2004; A. Colwell, Western Fisheries Research Center, Seattle, WA, personal communication) and this may explain in part why virulence is similar for parasites from geographically diverse locations. In the absence of virulence differences, perhaps the single most important factor determining whether disease will manifest in a susceptible fish host population is the parasite dose. Laboratory exposures of young rainbow trout with increasing doses of actinospores result in a correspondingly greater severity of whirling disease (Markiw, 1992; Hedrick et al., 1999; Ryce, 2003). Studies on the upper Colorado River have also demonstrated that high concentrations of actinospores were associated with more severe whirling disease in wild populations of rainbow trout (Thompson and Nehring, 2000; Thompson et al., 2002). The size, age, and species of the salmonid host are key factors affecting the susceptibility of salmonids (MacConnell and Vincent, 2002). In both wild and cultured populations of salmonids, whirling disease has been most severe when the most susceptible size or age (e.g., less than 9 weeks or 756 degree days for rainbow trout; Ryce, 2003) and species of salmonids (e.g., rainbow trout) are present in areas with high levels of waterborne actinospores. When these conditions overlap, losses of year classes of wild rainbow trout (Nehring and Walker, 1996; Vincent, 1996) or outbreaks of whirling disease in young hatchery reared salmonids can occur (Hoffman, 1990). In affected wild populations of salmonids, sport fisheries can be significantly altered where prized rainbow or cutthroat trout are replaced by brown trout or other more resistant trout species. In waters where more naturally resistant salmonids are present [e.g., grayling (Thymallus thymallus) or bull trout (Salvelinus confluentus)], the consequences of whirling disease may be minimal (Hedrick et al., 1999; Bartholomew et al., 2003). Numerous environmental factors influence the interactions between the host and the pathogen. Of those variables examined, water temperature and flow are the most noteworthy. Water temperature affects numerous physiological processes, including the immune response in the fish host as well as development and reproduction of the parasite. Early studies by Halliday (1973) and later by El-Matbouli et al. (1995) demonstrated the effects of water temperature on the progress of infection in rainbow trout. At lower water temperatures (e.g., 7◦ C) both the appearance of trophozoites and then mature myxospores are delayed compared to temperatures of 12 and 17◦ C. Risk Estimation. The process of risk estimation integrates the results of the release assessment, exposure assessment, and consequence assessment to produce overall measures
Whirling Disease Risk Assessment
of risks associated with the hazards identified at the outset. Regardless of whether one is developing a qualitative or quantitative risk assessment model, there are a number of important steps that must be performed in a systematic manner (Vose, 1997, 2000); initially, the analysts must: 1. state the question to be answered clearly and explicitly, and 2. draw a scenario tree. From the outset, it is essential to have a clear understanding of the question to be answered. The process of defining the question is known as “scoping the outcome” and, if the outcome is poorly scoped, problems will arise in interpreting and communicating the results (Murray, 2002). The process of defining the question and the difficulties encountered in this stage of aquatic risk analysis are discussed by Travis and Hueston (2001). For M. cerebralis, defining the question will require careful consideration of the range of acceptable consequences. For example, in nonendemic areas, introduction of the parasite may be an unacceptable consequence, while in endemic waters, unacceptable consequences might include an increase in clinical disease or population level effects.
Factors Affecting the Quality of a Risk Assessment Carrying out a risk assessment requires working with the best information available at that time. In the course of working through the assessment, information gaps will be identified that require further research or more detailed analysis. As research supplies answers to the questions identified, the quality of the risk assessment can be expected to improve and the conclusions may be either supported or modified as a result of the improved analysis. Information Quality The quality of data on which it is based is an important component of a risk assessment. Where applicable, the precision, accuracy, and sensitivity of the methods used to collect the data should be known and included in the analysis as they allow the user to estimate the uncertainties in the process. Understanding the individual uncertainties and how they combine to produce the overall level of uncertainty is one of the most important aspects in a risk assessment. For example, in assessing the risk for fish movements it is important to evaluate the quality of the assay (sensitivity, specificity, reliability, and reproducibility) used to assess infection status (Thorburn, 1996; Bruneau, 2001; Williams and Moffitt, 2001; OIE, 2003). An assay that is 95% efficient will still miss infections in 1 of 20 fish, even when the animals are examined at a 100% sampling effort. For statistically based sampling, this false negative rate is multiplied further by the uncertainty introduced by the confidence limits imposed by the sample size. When a single, statistically based sample is collected from a hatchery of unknown health status, the magnitude of this error may become unacceptably large. In stocks sampled and assayed for infection status, highly infected animals should be relatively easy to detect. In contrast, a stock in which only a few fish are infected at levels below the detection threshold of the assay may represent a substantial risk of unintentional introduction of the parasite if the fish are moved to a new location. Calculations for the probability of failure to detect disease under different levels of infection prevalence, test sensitivity and sampling effort are presented in Table 2.
Sample size, S
Test sensitivity, Se Prevalence, P
30 60 100 300 500 1000
2% 5.83 * 10−1 3.39 * 10−1 1.65 * 10−1 4.48 * 10−3 1.21 * 10−4 1.40 * 10−8
1% 7.63 * 10−1 5.83 * 10−1 4.06 * 10−1 6.70 * 10−2 1.10 * 10−2 1.18 * 10−4
2.59 * 10−1 6.72 * 10−2 1.11 * 10−2 1.35 * 10−6 1.61 * 10−10 2.33 * 10−20
7.43 * 10−1 5.52 * 10−1 3.71 * 10−1 5.11 * 10−2 7.00 * 10−3 4.78 * 10−5
5.52 * 10−1 3.05 * 10−1 1.38 * 10−1 2.61 * 10−3 4.90 * 10−5 2.28 * 10−9
2.26 * 10−1 5.13 * 10−2 7.07 * 10−3 3.48 * 10−7 1.68 * 10−11 2.48 * 10−22
Table 2 Probability of failing to detect a fish population as infected under conditions of different samples sizes, test sensitivities, and disease prevalence. Here, the fish population is 100,000, the number of diseased fish (D) is defined as D = P*N, and the probability of failure to detect disease as (1-(S*Se/N))∧ D (an approximation where S N)
Whirling Disease Risk Assessment
Data Gaps Managers are often required to make decisions in the absence of important facts. Such data gaps may significantly affect the quality of a risk assessment, increasing its uncertainly and in some cases making it nearly valueless. However, one of the benefits of a riskbased management approach is that important data gaps may be revealed in the course of developing a risk assessment. For example, knowledge of the infection status of wild or free-ranging trout in a watershed may be critical to conducting a meaningful assessment of the consequences of the introduction of a M. cerebralis. Often, there will not be sufficient time to provide such data, and managers will be required to conduct a risk assessment using available knowledge. Consequently, when critical gaps are identified, it is important for managers to communicate these needs to appropriate funding agencies so that work can begin to produce the required information.
Examples of How to use a Risk Assessment Model to Answer Specific Questions The following examples include a discussion of some of the factors used in constructing risk assessments for two differing scenarios that might be addressed by fisheries managers. Naturally, many other scenarios are possible, but the basic steps are similar. Example 1. The Planned Movement of Hatchery Fish into a Disease-Free Area Description of the Problem. The intentional movement of hatchery-reared salmonids between watersheds is thought to be one of the most important sources of introduction of M. cerebralis into areas formerly believed to be free of the parasite. In many cases, this appears to be the only (or at least the most likely) method by which the pathogen could have been introduced. The scenario in this example involves a state hatchery located in the intermountain west, rearing rainbow trout for stocking as yearlings in other watersheds within the state. The hatchery has an open water supply and M. cerebralis-susceptible naturally reproducing trout populations are present in the watershed above and below the facility, as well as within the various receiving waters. One private fishing pond is located in the upper reaches of the watershed on which the state facility is located. The whirling disease status of this pond is undefined, the site is known to receive juvenile trout on at least an annual basis, and no regulations are present requiring inspections of fish moved into that facility or for reporting such movements. Currently, there is no disease history for any of the resident native fish species in any of the watersheds. At the state hatchery, trout eggs are obtained annually from off-site brood stock with a good disease inspection history. The eggs are disinfected before shipping and upon receipt and are hatched and reared on well water (12◦ C) until 8 wks post-hatch. After 8 weeks, the fry are reared in concrete raceways on river water. Fish at the state hatchery have no history of clinical whirling disease, and pre-release fish have been tested annually for the past five years for the presence of M. cerebralis by collection of a single, 60-fish sample consisting of fish from each raceway on river water for six months. No evidence of the parasite was found when the samples were assayed using pepsin-trypsin digest. Statement of the Question. The question here is “what is the likelihood for introduction of M. cerebralis into resident trout populations in receiving waters as a result of
J. L. Bartholomew et al.
Figure 4. Scenario tree for assessing the probability that infected fish would be released from the hatchery described in Example 1.
stocking fish from this state hatchery?” This could be depicted in a scenario tree as the probability of actinospores entering the hatchery times the probability that infection establishes in hatchery fish times the probability that the infection is not detected (Figure 4). Release Assessment. In this scenario, there are actually two introductions to consider—the introduction of the parasite into the hatchery, and, if that occurs and results in infection of hatchery fish, the transfer of infected fish to other watersheds. Because the state hatchery only imports disinfected eggs into the facility from sources with a good inspection history, introduction to the hatchery will most likely occur as a result of M. cerebralis myxospores or actinospores entering in the water supply. As there is no available information on the private pond, it would be prudent to assume that at some time infected fish will be imported into the pond and that the parasite life cycle could become established upstream of the hatchery. Likewise, the absence of information on infection status of naturally reproducing populations above the hatchery should be viewed as an important source of uncertainty. Because conditions in the hatchery (concrete raceways that are maintained free of sediment) would not support large numbers of T. tubifex, the likelihood of the parasite life cycle establishing in the hatchery is low, even if actinospores enter the facility and manage to infect susceptible fish. Fry are stocked into raceways at a size when they are susceptible to infection, but when they have developed some resistance to clinical disease and are likely to develop subclinical infections. As a result, infection prevalence within the hatchery is likely to be low and there is a high probability that infection would remain undetected using current diagnostic methods (pepsin-trypsin digest). Another factor that is important in assessing the risk of the introduction of the parasite into state waters receiving plantings of hatchery fish, is the number of time the introductions are to be performed. Risk from a single failure of a diagnostic assay is multiplied by the
Whirling Disease Risk Assessment
Table 3 Probability of failure of disease protection measures with multiple importations of fish Screening sensitivity Number of importations
1 2 3 4 5 6 7 8 9 10
0.10 0.19 0.27 0.34 0.41 0.47 0.52 0.57 0.61 0.65
0.05 0.10 0.14 0.19 0.23 0.26 0.30 0.34 0.37 0.40
0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.10
number of iterations, thus, the overall risk can increase rapidly to unacceptable levels, especially with less sensitive assays (Table 3). Exposure Assessment. In this scenario, receiving waters contain naturally reproducing populations of susceptible trout and it is likely that populations of susceptible strains of T. tubifex occur as well. Blocking the transfer of the parasite into receiving waters may be the only control option that can prevent M. cerebralis from becoming endemic in many of the receiving watersheds. However, the physicochemical characteristics of the receiving waters vary somewhat, thus, T. tubifex abundances probably vary. Thus, if the parasite is introduced through the movement of infected trout, there is a good possibility that it will become established in some receiving waters but not in others. Consequence Assessment. Once established in a natural system, there is little likelihood of eradicating M. cerebralis. In this scenario, the direct biological consequences for trout in the receiving watersheds could range from low to extreme and may take years to become apparent, depending on the characteristics of individual watersheds. If the parasite becomes established in a receiving watershed where sufficient numbers of young rainbow trout, or other highly susceptible fish species, and high densities of susceptible strains of T. tubifex are present, the levels of whirling disease in the system can be expected to result in significant changes in the abundance of year classes of susceptible species and changes in the population structure. The indirect consequences of the introduction and establishment of the parasite in this scenario will be significant. The introduction of the pathogen alone may be sufficient to cause substantial damage to management programs and may limit options for the hatchery or for the watersheds in question. Also important will be losses of revenue associated with declines in angling and adverse publicity for agencies involved. Risk Estimation. The likelihood that rainbow trout transferred from a facility on an unprotected water supply will lead to establishment in receiving watersheds is moderate to high if no management actions are taken (Table 4). The biological consequences of introduction may range from low to high and will vary between watersheds; however, management
J. L. Bartholomew et al.
Table 4 Qualitative estimates of the risks of different management options on the likelihood of transfer of Myxobolus cerebralis into a whirling-disease-free area via the movement of hatchery fish as outlined in scenario one. The single year and cumulative risks over a 10-year period are estimated Actions
None Risk—Single Year Introduction of parasite into hatchery Infection in hatchery fish Transfer of infected fish to uninfected watershed Cumulative risk— 10 years
Confirm pond and resident fish Increase testing negative. Restrict Change hatchery at hatchery imports to pond water supply to before transfer and watershed pathogen-free source
consequences of disseminating the parasite are significant. It should be obvious that the risk increases as introductions are repeated over a longer period and various management options can significantly affect these risks (Table 4). Risk Management. In this scenario, areas for management action are preventing the infection of fish in the hatchery and improved diagnostics for detection of infection in hatchery fish, thus preventing the movement of infected fish outside the drainage. There is an important area of uncertainty in the inspection and sampling methods used to assess infection status of the population to be moved and a major data gap regarding knowledge of the infection status of fish residing in the water supply for the hatchery. Increasing the level of sampling effort and using more sensitive detection methods will decrease, but not eliminate, the likelihood of transferring infected fish. However, if the prevalence of the parasite is low, it is likely that the transfer of infected fish from the hatchery will occur at some time. Taking steps to ensure the parasite-free status of the hatchery water supply would include testing susceptible species of resident fish and T. tubifex in the watershed, especially upstream of the hatchery. The history and infection status of the fishing pond should also be determined and regulations implemented requiring only importation of M. cerebralis-free fish. Because the greatest reduction in transfer risk would result from a secure water source, a high priority should be placed on alternatives to a surface water supply. In Table 4, the risks of introduction and establishment in the hatchery, and transfer to new watersheds are qualitatively estimated for different management actions. This analysis serves to highlight the reduction in overall risk achieved by addressing the most important pathways of introduction.
Whirling Disease Risk Assessment
Example 2. Assessment of Risk for a Native Stream as a Result of Fish Passage Description of the Problem. A series of three hydroelectric dams on the midsection of the Deschutes River, Oregon, have prevented the passage of migrating fish for about thirty years. It has been proposed to reintroduce sockeye (Oncorhynchus nerka) and Chinook salmon and summer steelhead (Oncorhynchus mykiss; anadromous rainbow trout) to the upper portion of this river. However, monitoring of adult steelhead entering the Deschutes River below the dams during the 1980’s demonstrated the presence of M. cerebralis in some of these fish (Engelking, 2002). The Deschutes River is a tributary of the mid-Columbia River basin, and M. cerebralis has not been detected in resident fish populations. The parasite is endemic in areas of the Columbia River basin in Washington, Oregon and Idaho. Adult salmonids entering the Deschutes River are of three origins. Two hatcheries in the Deschutes River basin produce spring Chinook salmon and summer steelhead that are released below the dams, referred to as hatchery salmonids. Salmonids from other river basins with markings indicating they were raised to smolt stage at hatcheries outside of the Deschutes River system are termed out of basin or stray fish. Unmarked salmonids (fish without clipped fins) are presumed of natural or wild origin (Deschutes River or unknown source). Statement of the Question. The specific question in this scenario is “What is the potential for adverse impacts of whirling disease on populations of resident salmonids as a result of passing adult salmon above the hydroelectric barriers.” Several options for passage were proposed, and for this exercise, three alternative questions are scoped: 1. What is the risk associated with no passage? 2. What is the risk with selective passage of Deschutes River hatchery salmonids? 3. What is the risk of unrestricted passage of all adult salmonids? The scenario trees created to illustrate the probability of the risks related to these passage options are presented in Figures 5a–c. The identification of critical information gaps identified in these figures directed research needs for each of the assessments. Release Assessment. The probability of introducing M. cerebralis above the barriers is the probability of introduction as a result of the fish passage option selected times the probability of introduction by other routes [(R1,2 = 1 − (1 − R1)(1 − R2); Figure 5a)]. To determine the likelihood of introducing M. cerebralis as a result of passage of each of the three groups of anadromous fish species identified for passage (hatchery, wild, stray), the prevalence of infection among adult salmon captured at fish traps was determined during a five-year survey (Engelking, 2002). Examination of over 600 adult salmon did not result in detection of M. cerebralis in any fish of known Deschutes River hatchery origin. Out of basin salmon (stray hatchery fish) were infected at the following prevalence: steelhead trout −18% and Chinook salmon −8%. Although the apparent prevalence of infection in sockeye salmon was high (25%), this represented only one of four fish tested. Of the unmarked (wild) fish, 5% of the Chinook salmon and 8% of the steelhead had M. cerebralis spores (Engelking, 2002). Based on the rate of straying of these species (straying = movement into nonnatal watersheds), an assumption was made that the infected unmarked fish originated in other watersheds. This assumption was supported by assay of resident fish in the lower Deschutes River in an
J. L. Bartholomew et al.
Figure 5. Scenario trees for assessing the probability that population impacts from Myxobolus cerebralis would occur in Example 2: (a) diagrams the scenario tree for assessing the probability that introduction of M. cerebralis would occur as a result of the three passage options for adult salmon, (b) diagrams questions that must be considered to determine the probability of establishment and proliferation, and (c) diagrams the probability that if established in host populations, that disease impacts would occur. The blocks that are shaded indicate information gaps addressed in the study that contributed to the risk assessment. (Continued)
Whirling Disease Risk Assessment
Figure 5. (Continued).
attempt to detect M. cerebralis and thus establish its presence below the dams. Assay of over 1700 yearling salmonids (Engelking, 2002) and multiple years of sentinel exposure data (Bartholomew, unpublished data) have not confirmed establishment of the parasite in the lower river. Thus, the likelihood of introducing M. cerebralis as a result of passing hatchery fish originating from the Deschutes River is low, but higher than the risk of no passage. Sources of risk in this case include human error in making the judgment of fish origin and the possibility that in the future the parasite may become established in the lower Deschutes River and result in exposure above the dams. The risk of introducing M. cerebralis with passage of stray hatchery fish is extreme, as infected adult fish were detected annually. Wild fish from the Deschutes River may currently represent a low risk; however, they cannot be distinguished from out of basin fish. Pathways other than anadromous fish passage that might result in introduction of M. cerebralis were also considered. Although there is potential for introduction by birds and mammals, the geographic isolation of the upper Deschutes River watershed from areas where whirling disease is endemic put these routes as a low risk. Likewise, the potential for human transport of the parasite was considered low because of the distance to recreational and sportfishing sites in the endemic area. The risk of introduction as a result of importation of infected fish for private ponds is unknown, and for this reason the risk for the no passage option was not negligible (Table 5). Exposure Assessment. The likelihood that if introduced, M. cerebralis would become established and proliferate is dependent on the density of susceptible salmonids and oligochaetes, their spatial and temporal overlap, and the environment (Figure 5b). Susceptible salmonid species are known to be present in the upper Deschutes River watershed. Presence and susceptibility of T. tubifex was unknown and this information gap was addressed through field surveys and laboratory studies. Over a five-year period, sediment samples from more
J. L. Bartholomew et al.
Table 5 Qualitative assessment of risks associated with each of the fish passage options for the Deschutes River, Oregon No passage Probability—Single Event Introduction Establishment Disease Disease risk over 30 years
Very low Very low Very low Very low to low
Selective passage Unrestricted passage Low Low Very low Low
Extreme Moderate Low to moderate Moderate to high
than 100 sites were collected, and T. tubifex was identified from 14 sites. Thus its presence above the barriers could be considered patchy, or discontinuous. Laboratory challenge studies conducted on several of these populations demonstrated the strains present could support the parasite life cycle with subsequent release of actinosporean stages (Bartholomew unpublished data). The sporadic occurrence of susceptible strains of T. tubifex in the upper watershed suggests that if M. cerebralis were introduced into the watershed, establishment could occur. Environmental conditions, at least in some areas of the upper Deschutes River basin, would be expected to support proliferation of the parasite. Consequence Assessment. The likelihood that adverse effects, or disease, would result from establishment of M. cerebralis is dependent on the susceptibility of the salmonid species, the parasite dose and temperatures that favor parasite proliferation in the host (Figure 5c). Controlled laboratory challenges demonstrated that of the resident species, redband rainbow trout were the most susceptible. Steelhead and kokanee salmon (landlocked sockeye salmon) both displayed a lower level of susceptibility, but clinical disease signs developed in both species (Sollid et al., 2002). Chinook salmon were susceptible when exposed at one week post-hatch, but resistance to clinical disease developed by 3 weeks post-hatch (Sollid et al., 2003). Bull trout did not show clinical disease signs, although they became infected (Hedrick et al., 1999; Bartholomew et al., 2003). When the life histories of each species are considered, the data indicate that risk would be highest for rainbow trout. These fish would also be more likely to be at a susceptible life stage when exposure might occur and during a period when water temperatures are sufficient to allow parasite proliferation. Risk Estimation. The likelihood of introducing M. cerebralis with passage of only fish that are identified as Deschutes River origin is low; however, introduction becomes almost certain if all fish are passed (Table 5). Conditions for establishment exist in a number of locations and some of these coincide with spawning areas for rainbow trout. The consequences of introduction would be greatest for resident rainbow trout populations as a result of their high susceptibility, locations of spawning, and emergence during late spring when parasite release would be expected to occur. When either of the reintroduction options are examined on a 30 year basis, the effects are compounded (Table 5). Although the probabilities are not quantitative, examining the scenarios in this manner provides an estimation of the magnitude of the risk between the fish reintroduction scenarios. Risk Management. Disease impacts were only one of the many factors examined when making decisions on fish passage in the Deschutes River (other factors included genetics,
Whirling Disease Risk Assessment
reestablishment and connectivity of populations, spawning and rearing habitat availability and the physical means by which passage and downstream migration would be accomplished). In the context of whirling disease, passage of adult hatchery fish of Deschutes River origin presents a relatively low risk for introducing M. cerebralis above the dams in comparison with unrestricted adult salmon passage. However, this would change if M. cerebralis were to become established in the lower Deschutes River, as outmigrant hatchery juvenile salmonids transiting the infected area could acquire low levels of infection. A portion of the population intended for passage should continue to be monitored for infection, and yearly sampling of resident rainbow trout in the lower river should be continued. If the infection status of that population changes, the risks associated with passage of both hatchery and naturally produced fish would be increased.
Intervention and Management Options An important outcome of a risk assessment is the identification of pathways that can be interrupted to prevent introduction and establishment of M. cerebralis, or managed to avoid development of whirling disease. The most desirable option, where possible, is prevention. Exclusion methods should be based on evaluation of the introduction pathways (release assessment) and will likely include inspection of fish. Because early detection is important for any effective control measure, site-specific surveys should be designed for locations at high risk (e.g., rainbow trout hatcheries that receive transfers of fish) or are of high conservation value (e.g., Yellowstone National Park). Managers should also support development and application of the most sensitive diagnostic methods available. Although it entails some expense, prevention is almost always the most cost-efficient option when compared with eradication and control programs. In most states, detection of M. cerebralis in a hatchery population results in immediate regulatory measures that can range from severely restricted movements of fish to complete depopulation and sanitation procedures (Hoffman, 1990). In nearly all regulatory actions, the economic costs to both public and private aquaculture can be substantial including facility closures. In hatcheries, one of the most successful approaches to preventing introduction has been protection of water supplies and many facilities have made major improvements in this area. A release assessment might identify other pathways of concern. For example, irrigation canals connecting watersheds, for which a simple intervention methods (e.g., a fish barrier) could be designed to prevent or reduce the likelihood of introduction. Evaluation of hatchery operations could reveal pathways of introduction that could be eliminated as well as areas needing increased disease surveillance or application of better diagnostic methods. If prevention fails, then the remaining options are eradication, containment, control, or no action. If detection occurs early, eradication may be feasible under certain conditions, most likely in culture facilities. These programs have been successful when the water supply can be isolated, contained, and made free of either oligochaetes or salmonid fish. Eradication may also be successful when an introduction has been made into a small, closed body of water from which all individuals can be removed. In contrast, once established in open waters containing populations of reproducing salmonid fish, eradication attempts have largely failed (Yoder, 1972; Modin, 1998). In these cases management procedures have been aimed at reducing infection levels by reducing point sources of actinospores (e.g., effluents from known infected culture facilities, water impoundments, etc.) or myxospores (e.g., reducing or eliminating stocking of infected fish) and by containing further spread. For these interventions, monitoring will be required to
J. L. Bartholomew et al.
ensure the success of the action and that other introduction pathways are not overlooked. Monitoring tools such as water filtration for actinospore stages of M. cerebralis and sediment sampling for oligochaetes have been effective in identifying point sources of infectivity for whirling disease (Nehring et al., 2002). With such sites identified, management approaches such as sand filtration or channel modification, which might effectively reduce the infection pressure on salmonids, can be considered. In areas where M. cerebralis has successfully become introduced and established, control options will depend on regulatory situations, fishery management objectives, and the disposition of affected fish stocks. Hatcheries where control of the water supply is not possible may rear more resistant salmonid species thus allowing continued production, particularly if fish are not used for stocking but instead are harvested for food. Utilization of more resistant strains of rainbow trout may also allow a fishery to exist, particularly in areas where indigenous salmonid populations will not be impacted by restocking efforts aimed at rebuilding lost sport fishery stocks (e.g., tail water fisheries for rainbow trout). One strain of rainbow trout, of hatchery origin in Germany, has been demonstrated to be significantly more resistant to whirling disease than other tested stocks in North America (Hedrick et al., 2003) and the resistance of other domestic strains is being examined. Through selective breeding these strains may offer alternatives to currently used hatchery strains that would gradually reduce parasite levels by reducing the number of spores returned to the system. In areas where there are naturally reproducing populations of wild trout, exploiting life history traits that might aid in escaping severe infections have been viewed as a means to ameliorate the adverse effects of whirling disease (Downing et al., 2002). When areas of a stream with decreased infectivity are utilized by certain strains/races of rainbow or cutthroat trout for spawning or for early rearing, the impacts of whirling disease can be substantially reduced. Thus for some rivers, key management options may lie in physical modifications to the watershed that result in decreased habitat for T. tubifex in critical spawning areas. Other options under study include management of flows on impounded systems to reduce densities of T. tubifex or favor the replacement of susceptible strains of this host with resistant strains and other species of oligochaetes.
Acknowledgments The 9th Annual Whirling Disease Symposium held in Seattle, WA in February 2003 was entitled “Whirling Disease: Managing the Risk” and contained sessions in which managers and experts in risk management and whirling disease research worked together to draft an initial outline of the pathways for the introduction and establishment of M. cerebralis that were used as the basis for this work. This effort would not have been possible without the support and encouragement of Dave Kumlien, Susan Higgins and Wanda McCarthy of the Whirling Disease Foundation. We would also like to thank Dr. J. C. Wilson who supplied information to illustrate the hatchery risk assessment example, Dr. H. M. Engelking for providing information for the Deschutes River risk assessment example, and Dr. David Vose for his constructive suggestions.
References Andree, K. B., M. El-Matbouli, and R. P. Hedrick. Comparison of 18S and ITS-1 rDNA sequences of specific geographic isolates of Myxobolus cerebralis. Int. J. Parasitol., 29: 771–775 (1999). Andree, K. B., R. P. Hedrick, and E. MacConnell. A review of the approaches to detect Myxobolus cerebralis, the cause of salmonid whirling disease. pp. 197–212. In: Whirling Disease: Reviews
Whirling Disease Risk Assessment
and Current Topics. American Fisheries Society, Symposium 29. (Bartholomew, J. L. and J. C. Wilson, Eds.). Bethesda, MD: American Fisheries Society (2002). Bartholomew, J. L., and P. W. Reno. The history and dissemination of whirling disease. pp. 3–24. In: Whirling Disease: Reviews and Current Topics. American Fisheries Society, Symposium 29. (Bartholomew, J. L., and J. C. Wilson, Eds.). Bethesda, MD: American Fisheries Society (2002). Bartholomew, J. L., H. V. Lorz, S. A. Sollid, and D. G. Stevens. Susceptibility of juvenile and yearling bull trout to Myxobolus cerebralis, and effects of sustained parasite challenges. J. Aquat. Anim. Health, 15: 248–255 (2003). Beauchamp, K. A., M. Gay, G. O. Kelley, M. El-Matbouli, R. D. Kathman, R. B. Nehring, and R. P. Hedrick. Prevalence and susceptibility of infection to Myxobolus cerebralis, and genetic difference among populations of Tubifex tubifex. Dis. Aquat. Org., 51: 113–121 (2002). Blazer, V. S., T. B. Waldrop, W. B. Schill, C. L. Densmore, and D. Smith. Effects of water temperature and substrate type on spore production and release in Eastern Tubifex tubifex worms infected with Myxobolus cerebralis. J. Parasitol., 89: 21–26 (2003). Brinkhurst, R. O., and B. G. M. Jamieson. Aquatic oligochaeta of the world. Toronto: University of Toronto (1971). Bruneau, N. A quantitative risk assessment for the introduction of Myxobolus cerebralis to Alberta, Canada, through the importation of live farmed salmonids. pp. 41–50. In: Risk Analysis in Aquatic Animal Health. (Rogers, C.J., Ed.). Paris: Office International des Epizooties (2001). Downing, D. C., T. E. McMahon, B. L. Kerans, and E. R. Vincent. Relation of spawning and rearing life history of rainbow trout and susceptibility to Myxobolus cerebralis infection in the Madison River, Montana. J. Aquat. Anim. Health, 14: 191–203 (2002). El-Matbouli, M., and R. W. Hoffman. Effects of freezing, aging, and passage through the alimentary canal of predatory animals on the viability of Myxobolus cerebralis spores. J. Aquat. Anim. Health, 3: 260–262 (1991). El-Matbouli, M., R. W. Hoffman, and C. Mandok. Light and electron microscopic observations on the route of the triactinomyxon-sporoplasm of Myxobolus cerebralis from epidermis into rainbow trout (Oncorhynchus mykiss). J. Fish Biol., 46: 919–935 (1995). El-Matbouli, M., T. S. McDowell, D. B. Antonio, K. B. Andree, and R. P. Hedrick. Effect of water temperature on the development, release and survival of the triactinomyxon stage of Myxobolus cerebralis in its oligochaete host. Int. J. Parasitol., 29: 627–641 (1999). Engelking, H. M. Potential for introduction of Myxobolus cerebralis into the Deschutes River watershed in central Oregon from adult anadromous salmonids. pp. 25–32. In: Whirling Disease: Reviews and Current Topics. American Fisheries Society, Symposium 29. (Bartholomew, J. L., and J. C. Wilson, Eds.). Bethesda, MD: American Fisheries Society (2002). Gilbert, M. A., and W. O. Granath Jr. Persistent infection of Myxobolus cerebralis, the causative agent of salmonid whirling disease, in Tubifex tubifex. J. Parasitol., 87: 101–107 (2001). Granath, W. O., and M. A. Gilbert. The role of Tubifex tubifex (Annelida: Oligochaeta: Tubificidae) in the transmission of Myxobolus cerebralis (Myxozoa: Myxosporea: Myxobolidae). pp. 79– 85. In: Whirling Disease: Reviews and Current Topics. American Fisheries Society, Symposium 29. (Bartholomew, J. L., and J. C. Wilson, Eds.). Bethesda, MD: American Fisheries Society (2002). Halliday, M. M. Studies on Myxosoma cerebralis, a parasite of salmonids. II. The development and pathology Myxosoma cerebralis, in experimentally infected rainbow trout (Salmo gairdneri) fry reared at different water temperatures. Nordisk Veterinaermedicin, 25: 349–358 (1973). Halliday, M. M. The biology of Myxosoma cerebralis: the causative organism of whirling disease. J. Fish Biol., 9: 339–357 (1976). Hedrick, R. P. Relationship of the host, pathogen, and environment: implications for diseases of cultured and wild fish populations. J. Aquat. Anim. Health, 10: 107–111 (1998). Hedrick, R. P., M. El-Matbouli, M. A. Adkison, and E. MacConnell. Whirling disease: Re-emergence in wild trout. Immunol. Rev., 166: 365–376 (1998). Hedrick, R. P., T. S. McDowell, M. Gay, G. D. Marty, M. P. Georgiadis, and E. MacConnell. Comparative susceptibility of rainbow trout Oncorhynchus mykiss and brown trout Salmo trutta to
J. L. Bartholomew et al.
Myxobolus cerebralis, the cause of salmonid whirling disease. Dis. Aquat. Org., 37: 173–183 (1999). Hedrick, R. P., and M. El-Matbouli. Recent advances with taxonomy, life cycle and development of Myxobolus cerebralis in the fish and oligochaete hosts. pp. 45–49. In: Whirling Disease: Reviews and Current Topics. American Fisheries Society, Symposium 29. (Bartholomew, J. L., and J. C. Wilson, Eds.). Bethesda, MD: American Fisheries Society (2002). Hedrick, R. P., T. S. McDowell, M. Gay, G. D. Marty, G. T. Fosgate, K. Mukkatira, K. Mykelbust, and M. El-Matbouli. Susceptibility of two strains of rainbow trout (one with a suspected resistance to whirling disease) to Myxobolus cerebralis infection. Dis. Aquat. Org., 55: 37–44 (2003). Hiner, M., and C. M. Moffitt. Modeling Myxobolus cerebralis infections in trout: associations with habitat variables. pp. 167–180. In: Whirling Disease: Reviews and Current Topics. American Fisheries Society, Symposium 29. (Bartholomew, J. L., and J. C. Wilson, Eds.). Bethesda, MD: American Fisheries Society (2002). Hoffman, G. L. Intercontinental and transcontinental dissemination and transfaunation of fish parasites with emphasis on whirling disease (Myxosoma cerebralis) and its effects on fish. pp. 69–81. In: Symposium on Diseases of Fisheries and Shellfishes, American Fisheries Society Special Publication 5. (Snieszko, S. F. Ed.). Bethesda, MD: American Fisheries Society (1970). Hoffman, G. L. Myxobolus cerebralis, a worldwide cause of salmonid whirling disease. J. Aquat. Anim. Health, 2: 30–37 (1990). Juget, J., and M. Lafont. Theoretical habitat templets, species traits, and species richness: aquatic oligochaetes in the Upper Rhone River and its floodplain. Freshwater Biol., 31: 327–340 (1994). Kerans, B. L., C. Rasmussen, R. Stevens, A. E. L. Colwell, and J. R. Winton. Differential propagation of the metazoan parasite Myxobolus cerebralis by Limnodrilus hoffmeisteri, Ilyodrilus templetoni, and genetically distinct strains of Tubifex tubifex. J. Parasitol., 90: 1366–1373 (2004). Kerans, B. L., and A. V. Zale. The ecology of Myxobolus cerebralis. pp. 145–166. In: Whirling Disease: Reviews and Current Topics. American Fisheries Society, Symposium 29. (Bartholomew, J. L., and J. C. Wilson, Eds.). Bethesda, MD: American Fisheries Society (2002). Krueger, R. C., B. L. Kerans, C. Rasmussen, and E. R. Vincent. Correlations among environmental features, Myxobolus cerebralis infection prevalence in oligochaetes, and salmonid infection risk in the Madison River, Montana. Ecol. Appl. (in press). Lazim, M. N., and M. A. Learner. The influence of sediment composition and leaf litter on the distribution of tubificid worms (Oligochaeta). Oecologia, 72: 131–136 (1987). Lowers, J. M., and J. L. Bartholomew. Detection of Myxozoan Parasites in Oligochaetes Imported as Food for Ornamental Fish. J. Parasitol., 89: 84–91 (2003). MacConnell, E. Whirling disease of salmonids. In: Suggested Procedures for the Detection and Identification of Certain Finfish and Shellfish Pathogens. Blue Book 5th Edition. Bethesda, MD: Fish Health Section, American Fisheries Society (2003). MacConnell, E., and E. R. Vincent. The effects of Myxobolus cerebralis on the salmonid host. Pages 95–107. In: Whirling Disease: Reviews and Current Topics. American Fisheries Society, Symposium 29. (Bartholomew, J. L., and J. C. Wilson, Eds.). Bethesda, MD: American Fisheries Society (2002). MacDiarmid, S. C. Risk analysis, international trade, and animal health. pp. 377–387. In: Fundamentals of Risk Analysis and Risk Management. (Molak, V. Ed). Boca Raton, FL: CRC Lewis (1997). MacDiarmid, S. C. Risk analysis in aquatic animal health. pp. 1–6. In: Risk Analysis in Aquatic Animal Health. (Rogers, C.J., Ed.). Paris: Office International des Epizooties (2001). MacDiarmid, S. C., and J. S. Hellstr¨om. Surveillance for brucellosis using a skin test of low sensitivity. In: Proceedings of the 5th International Symposium on Veterinary Epidemiology and Economics, Copenhagen. ACTA Veterinaria Scandinavica, 84: 209–211 (1988). MacDiarmid, S. C., H. J. Pharo, and N. J. Murray. A quantitative assessment of the risk of introduction of IBD virus through importation of chicken meat. pp. 289–300. In: II. International Symposium on Infectious Bursal Disease and Chicken Infectious Anaemia. Rauischholzhausen, Germany. Institut fur Geflugelkrankheiten, Justus Liebig University, Giessen (2001).
Whirling Disease Risk Assessment
MacDiarmid, S. C., and H. J. Pharo. Risk analysis: assessment, management and communication. Revue Scientifique et Technique de l’Office International des Epizooties, 22: 397–408 (2003). Markiw, M. E. Salmonid whirling disease: dynamics of experimental production of the infective stage–the triactinomyxon spore. Can. J. Fish Aquat. Sci., 43: 521–526 (1986). Markiw, M. E. Experimentally induced whirling disease. I. Dose response of fry and adults of rainbow trout exposed to the actinospore stage of Myxobolus cerebralis. J. Aquat. Anim. Health, 4: 40–43 (1992). Modin, J. Whirling disease in California: A review of its history, distribution, and impacts, 1965–1997. J. Aquat. Anim. Health, 10: 132–142 (1998). Moutou, F., B. Dufour, and Y. Ivanov. A qualitative assessment of the risk of introducing foot and mouth disease into Russia and Europe from Armenia and Azerbaigan. Revue Scientifique et Technique de l’Office International des Epizooties, 20: 723–30 (2001). Murray, N. Import Risk Analysis: Animals and Animal Products. Wellington, New Zealand, New Zealand Ministry of Agriculture and Forestry (2002). Nehring, R. B., and P. G. Walker. Whirling disease in the wild: the new reality in the intermountain west. Fisheries, 21: 28–32 (1996). Nehring, R. B., and K. G. Thompson. Evaluating the potential relationships among triactinomyxons of Myxobolus cerebralis, myxospore concentrations in brown trout, and stocking history in standing waters. pp. 33–34. In: Proceedings of the 6th Annual Whirling Disease Symposium, Coeur d’Alene, Idaho: Whirling Disease Foundation (2000). Nehring, R. B., K. G. Thompson, D. L. Schuler, and T. M. James. Using sediment core samples to examine the spatial distribution of Myxobolus cerebralis actinospore production in Windy Gap reservoir, Colorado. North Am. J. Fish. Management, 23: 376–384 (2002). North, D. W. Limitations, definitions, principles and methods of risk analysis. Revue Scientifique et Technique de l’Office International des Epizooties, 14: 913–923 (1995). Nowak, B., K. Rough, D. Ellis, M. Crane, A. Cameron, and S. Clarke. Aquafin CRC-Southern Bluefin tuna Aquaculture Subprogram: A risk assessment of factors influencing the health of southern bluefin tuna. Tasmanian Aquaculture & Fisheries Institute, University of Tasmania (2003). OIE. International Aquatic Animal Health Code, 6th ed. Paris: Office International des Epizooties (2003). Paisley, L. G. A Monte Carlo simulation model for assessing the risk of introduction of Gyrodactylus salaris to the Tana River, Norway: a second scenario. pp. 185–192. In: Risk Analysis in Aquatic Animal Health. (Rogers, C. J., Ed.). Paris: Office International des Epizooties (2001). Rodgers, C. J. Risk Analysis in Aquatic Animal Health. Paris: Office International des Epizooties (2000). Ryce, E. K. N. Factors affecting the resistance of juvenile rainbow trout to whirling disease. Ph.D. Thesis. Bozeman, MT: Montana State University (2003). Schisler, G. J., and E. P. Bergersen. Evaluation of risk of high elevation Colorado waters to the establishment of Myxobolus cerebralis. pp. 33–42. In: Whirling Disease: Reviews and Current Topics. American Fisheries Society, Symposium 29. (Bartholomew, J. L., and J. C. Wilson, Eds.). Bethesda, MD: American Fisheries Society (2002). Snieszko, S. F. Foreword. pp. 5–6. In: Environmental stress and fish diseases-Book 5. (Snieszko, S. F., and H. R. Axelrod, Eds.). Neptune City, NJ: TFH publications (1976). Sollid, S. A., H. V. Lorz, D. G. Stevens, and J. L. Bartholomew. Relative susceptibility of selected Deschutes River, Oregon, salmonid species to experimentally induced infection by Myxobolus cerebralis. pp. 117–124. In: Whirling Disease: Reviews and Current Topics. American Fisheries Society, Symposium 29. (Bartholomew, J. L., and J. C. Wilson, Eds.). Bethesda, MD: American Fisheries Society (2002). Sollid, S. A., H. V. Lorz, D. G. Stevens, and J. L. Bartholomew. Age-dependent Susceptibility of chinook salmon to Myxobolus cerebralis and effects of sustained parasite challenges. J. Aquat. Anim. Health, 15: 136–146 (2003).
J. L. Bartholomew et al.
Stevens, R., B. L. Kerans, J. C. Lemmon, and C. Rasmussen. The effects of Myxobolus cerebralis myxospore dose on triactinomyxon production and biology of Tubifex tubifex from two geographic regions. J. Parasitol., 87: 315–321 (2001). Taylor, R. L., and M. Lott. Transmission of salmonid whirling disease by birds fed trout infected with Myxosoma cerebralis. J. Protozool., 25: 105–106 (1978). Thompson, K. G., and R. B. Nehring. A simple technique used to filter and quantify the actinospore of Myxobolus cerebralis and determine its seasonal abundance in the Colorado River. J. Aquat. Anim. Health, 12: 316–323 (2000). Thompson, K. G., R. B. Nehring, D. C. Bowden, and T. Wygant. Response of rainbow trout Oncorhynchus mykiss to exposure to Myxobolus cerebralis above and below a point source of infectivity in the upper Colorado River. Dis. Aquat. Org., 49: 171–178 (2002). Thorburn, M. A. Apparent prevalence of fish pathogens in asymptomatic salmonid populations and its effect on misclassifying population infection status. J. Aquat. Anim. Health, 8: 271–277 (1996). Travis, D., and W. Hueston. Factors contributing to uncertainty in aquatic risk analysis. pp. 27–35. In: Risk Analysis in Aquatic Animal Health. (Rogers, C. J., Ed.). Paris: Office International des Epizooties (2001). Vincent, E. R. Whirling disease and wild trout: the Montana experience. Fisheries, 21: 32–34 (1996). Vincent, E. R. Relative susceptibility of various salmonids to whirling disease with and emphasis on rainbow and cutthroat trout. pp. 109–115. In: Whirling Disease: Reviews and Current Topics. American Fisheries Society, Symposium 29. (Bartholomew, J. L., and J. C. Wilson, Eds.). Bethesda, MD: American Fisheries Society (2002). Vose, D. J. Risk analysis in relation to the importation and exportation of animal products. Revue Scientifique et Technique de l’Office International des Epizooties, 16: 45–56 (1997). Vose, D. J. Risk Analysis: A Quantitative Guide. Chichester: John Wiley & Sons, Ltd (2000). Vose, D. J. Qualitative versus quantitative risk analysis and modelling. pp. 19–25. In: Risk Analysis in Aquatic Animal Health. (Rogers, C. J., Ed.). Paris: Office International des Epizooties (2001). Whipps, C. M., M. El-Matbouli, R. P. Hedrick, V. Blazer, and M. L. Kent. Myxobolus cerebralis internal transcribed spacer 1 (ITS-1) sequences support recent spread of the parasite to North America and within Europe. Dis. Aquat. Org., 60: 105–108 (2004). Williams, C. J., and C. M. Moffitt. A critique of methods of sampling and reporting pathogens in populations of fish. J. Aquat. Anim. Health, 13: 300–309 (2001). Wolf, K., and M. E. Markiw. Myxosoma cerebralis: inactivation of spores by hot smoking of infected trout. Can. J. Fish. Aquat. Sci., 39: 926–928 (1982). Yoder, W. G. The spread of Myxosoma cerebralis into natural trout populations in Michigan. Prog. Fish-Cult., 43: 103–106 (1972).