INTRODUCTION
Generalist parasites can infect a wide range of hosts with varying severities; some hosts can be infected but not support the reproduction of the parasite, others can support limited reproduction, whilst, in some hosts the parasite can maximize its reproductive output (Holmes, Reference Holmes1979). The potential host range of a parasite is dictated by physiological, behavioural and ecological attributes of the host that determine the ability of a particular parasite to infect and complete its life cycle (Solter and Maddox, Reference Solter and Maddox1998). A genetic basis for potential host suitability is suggested by parasites that are more likely to infect hosts phylogenetically close to their existing ones (Poulin, Reference Poulin2007). Due to existing barriers to dispersal, observed host ranges of parasites often represent only a subset of potential hosts (Perlman and Jaenike, Reference Perlman and Jaenike2003). The process of host translocation in new geographical areas or range expansion by an existing host, allows increased opportunities for parasites to expand their range of potential hosts (Poulin, Reference Poulin2007).
The rosette agent Sphaerothecum destruens is a multi-host parasite, which experimental studies have shown to be able to infect a number of salmonid and cyprinid species at varying levels (Arkush et al. Reference Arkush, Frasca and Hedrick1998; Andreou et al. Reference Andreou, Arkush, Guégan and Gozlan2012). The discovery of S. destruens associated with an invasive reservoir host, the topmouth gudgeon Pseudorasbora parva (Temminck & Schlegel, 1846), increased the parasite's known and potential species range (Gozlan et al. Reference Gozlan, St-Hilaire, Feist, Martin and Kent2005, Reference Gozlan, Whipps, Andreou and Arkush2009). The parasite has now been found in established wild populations of a range of fish species in Europe including several of high IUCN (International Union for Conservation of Nature) conservation status (Andreou et al. Reference Andreou, Hussey, Griffiths and Gozlan2011; Ercan et al. Reference Ercan, Andreou, Sana, Öntaş, Baba, Top, Karakus, Tarkan and Gozlan2015). In experimental studies, infection with S. destruens has also confirmed a high level of mortalities in a range of cyprinid species (Gozlan et al. Reference Gozlan, St-Hilaire, Feist, Martin and Kent2005; Andreou et al. Reference Andreou, Arkush, Guégan and Gozlan2012); proving to be a valuable tool for determining the potential host range of and providing important epidemiological information to model disease emergence (Alshorbaji et al. Reference Alshorbaji, Gozlan, Roche, Britton and Andreou2015).
As a generalist pathogen, S. destruens can infect a range of host species (n = 14 up to date) and due to its close association with the invasive P. parva, which acts as a healthy reservoir host of S. destruens and has rapidly invaded a wide range of ecosystems ranging from Eurasia to the north Africa (Gozlan et al. Reference Gozlan, Andreou, Asaeda, Beyer, Bouhadad, Burnard, Caiola, Cakic, Djikanovic, Esmaeili, Falka, Golicher, Harka, Jeney, Kováč, Musil, Nocita, Povz, Poulet, Virbickas, Wolter, Tarkan, Tricarico, Trichkova, Verreycken, Witkowski, Zhang, Zweimueller and Britton2010), a risk assessment of S. destruens’ potential risk of emergence needs to be established (Copp et al. Reference Copp, Vilizzi, Mumford, Godard, Fenwick and Gozlan2009). Here, we aim at (1) calculating the specificity index of S. destruens and (2) use all the available information on S. destruens and its hosts’ range to evaluate the risk associated with its introduction and make management and prevention recommendations to guide national management agencies and policy makers.
MATERIAL AND METHODS
Host specificity index for S. destruens
The specificity index (STD) proposed by Poulin and Mouillot (Reference Poulin and Mouillot2003) measures the average taxonomic distinctness of a parasite's host species. The specificity index was calculated for S. destruens by performing the following steps: (a) S. destruens host species were placed within a taxonomic hierarchy using the Linnean classification; (b) the number of steps taken in order to reach a taxon common to two host species were calculated for all possible species pairs; (c) the number of steps was averaged across all species pairs. Step lengths between each hierarchical level were given the equal value of one. The index was calculated using the formula by Poulin and Mouillot (Reference Poulin and Mouillot2003):
where s is the parasite's number of host species, the double summation is over the set (i = 1, …s; j = 1,…s, such that i < j), and ω ij is the number of taxonomic steps needed to reach a common taxonomic node between host species i and j. The maximum value that the index STD can reach is five (when using the five taxonomic levels of genus, family, order, class and phylum). The lowest value STD can reach is one and this occurs when all host species share the same genus. A measure of the taxonomic structure of the host species can be obtained by calculating the variance in taxonomic distinctness (Poulin and Mouillot, Reference Poulin and Mouillot2003):
where ɷ is the mean taxonomic distinctness or STD. The fish taxonomy proposed by Nelson (Reference Nelson1994) was used in calculating the STD for S. destruens.
Evaluating the risk posed by S. destruens
We have used an established framework to assess the risk posed by S. destruens as described in Williams et al. (Reference Williams, Britton and Turnbull2013). We considered whether it is possible to manage the spread of S. destruens, a criterion central to the OIE (World Organisation for Animal Health) definition underpinning the list of notifiable infectious diseases and determine the potential hazard posed by S. destruens using the established risk assessment for managing non-native parasites (Williams et al. Reference Williams, Britton and Turnbull2013). We used the case of S. destruens and its healthy host P. parva in England and Wales as a case study for calculating potential hazard. England and Wales were selected as the case study due to extensive protocols being in place to detect and control the spread of P. parva.
RESULTS
A tree representing the taxonomic hierarchy of S. destruens host species was constructed (Fig. 1) using the Linnean classification and the specificity index (STD) for S. destruens was calculated to be 3·21 with a variance in taxonomic distinctness (VarSTD) of 0·49; supporting the generalist nature of S. destruens.
Using the risk assessment developed by Williams et al. (Reference Williams, Britton and Turnbull2013) we calculated the potential hazard posed by S. destruens to freshwater fisheries in England and Wales with an overall score of 25, identifying this parasite as a high risk parasite. In evaluating the value and susceptibility of native resources, S. destruens scored as high risk due to the available evidence indicating that in farmed, semi-natural and controlled exposures S. destruens can cause mortality in a number of temperate freshwater species; Atlantic salmon Salmo salar (Linnaeus, 1758), and Chinook salmon Onchorhyncus tshawytscha (Walbaum)(both in farm conditions and in controlled exposures in the laboratory (Harrell et al. Reference Harrell, Elston, Scott and Wilkinson1986; Hedrick et al. Reference Hedrick, Friedman and Modin1989; Arkush et al. Reference Arkush, Frasca and Hedrick1998), cyprinid species Abramis brama (Linnaeus, 1758), Rutilus rutilus (Linnaeus, 1758) and Cyprinus carpio (Linnaeus, 1758) (controlled exposures, Andreou et al. (Reference Andreou, Arkush, Guégan and Gozlan2012), sunbleak Leucaspius delineatus (semi-natural experiments (Gozlan et al. Reference Gozlan, St-Hilaire, Feist, Martin and Kent2005) and controlled exposure (Paley et al. Reference Paley, Andreou, Bateman and Feist2012). The economic and ecological value of freshwater species found in the England and Wales that can be exposed to S. destruens (S. salar, A. brama, R. rutilus and C. carpio) is high as they are key angling species. Fishing rights for salmon robs had an estimated value of £128 million in England and Wales in 2001 and Inland recreational fisheries for the UK had an estimated value of £3 billion in 2001 with carp being the most stocked fish in coarse fisheries, followed by roach and bream being the 4th most stocked (Environment Agency, 2004). In addition, all species inhabit aquatic environments throughout Britain and are key components of lake and river communities. None of the before mentioned susceptible species are threatened in the UK, however, due to the high generalist nature of S. destruens we cannot exclude the possibility that additional species from different fish families could also be susceptible; hence only a certainty score of 2 for question 3 (Table 1).
England and Wales have been used as a case study and thus all questions are answered in relation to native population in England and Wales. The risk assessment follows the guidelines by Williams et al. (Reference Williams, Britton and Turnbull2013). Both authors filled in the assessment independently and both scores are presented with combined rationales. Scoring criteria are: 0 = very low or no; 1 = low; 2 = moderate; 3 = high; 4 = very high or yes. Certainty scores are provided for every answer, 1 = low, 2 = moderate, 3 = high. Scores were summed and an overall hazard score was calculated. The overall score was then translated to low (0–12 points), moderate (13–24 points) and high (25–36 points) disease risk to native populations. Overall certainty scores were translated as low (1–9 points), moderate (10–18 points) and high (19–27 points).
The colonisation potential of S. destruens was also evaluated as high, due to its close association with P. parva and its direct life-cycle and environmental transmission (through spores and zoospores). In addition the high fish movements of A. brama, R. rutilus and C. carpio increase the probability of spreading the parasite to new fish farms and fishing clubs. In 2002/03 the Environment Agency issued consents to stock in excess of 7·5 million fish of which 25% were carp and variants, 8% were for roach and 4% were for bream (Environment Agency, 2004). The close association of these still water bodies with rivers and streams increases the possibility of the parasite being introduced and becoming established in adjacent stream and river communities, through environmental transmission.
The potential of disease risk was scored as moderate to high due to the limited records of population declines linked to S. destruens in England and Wales. It is important here to note that evidence provided by the study of Ercan et al. (Reference Ercan, Andreou, Sana, Öntaş, Baba, Top, Karakus, Tarkan and Gozlan2015) and chronic mortality patterns observed in controlled exposures (Andreou et al. Reference Andreou, Arkush, Guégan and Gozlan2012) indicate that only long term monitoring of communities potentially in contact with S. destruens can detect such population declines. This uncertainty in the lack of evidence was reflected in the low certainty score for this section (mean = 1·8).
Following the hazard risk assessment, the potential management responses were investigated using module 3 from the risk assessment by Williams et al. (Reference Williams, Britton and Turnbull2013). This assessment accounts for the local/ national legislation and management practices that are already in place as well as the distribution and management policies of the reservoir host (P. parva) in the risk assessment area, as that will drive a great part of the risk. The rationale supporting the decision made at each step (Fig. 2) for the case study of England and Wales: (1) S. destruens is not currently covered by any legislation in the UK; (2) S. destruens can infect S. salar, A. brama, R. rutilus and C. carpio whose movement is not restricted under the national exotic fish legislation; (3) eradication of the parasite has been attempted by eradicating its healthy host P. parva using rotenone a process that is both ecologically and financially expensive and ineffective in removing S. destruens from adjacent communities; (4) the parasite can be detected using histology and molecular techniques; (5) fish movement restrictions would be the only effective method to prevent the spread of the parasite to naïve fisheries (however this would depend on the distribution of the parasite, which is currently not known). The recommended management option was to implement initial management measures to limit spread and assess their effectiveness and management of the parasite after these measures are implemented in a panel of experts (as described in module 2 in Williams et al. Reference Williams, Britton and Turnbull2013).
DISCUSSION
Sphaerothecum destruens as a multi-host parasite: implications for disease emergence
Host specificity and cellular tropism is one of the most important characteristics in a parasite's life cycle (Poulin, Reference Poulin2007). Host specificity can be quantified by enumerating the number of species a parasite can infect (Lymbery, Reference Lymbery1989). However, this does not provide information on the taxonomic distinctness of the species a parasite can infect. Here S. destruens’ observed host specificity is similar to helminths parasitic to Canadian freshwater fish (Poulin and Mouillot, Reference Poulin and Mouillot2003). The high variance around the STD index indicated that S. destruens is more likely to colonize new species and in doing so it is possible for the parasite to make bigger taxonomic jumps. Overall, S. destruens did not appear to be limited to a phylogenetically narrow host spectrum and the current data suggest that it is a true generalist. In addition, the lack of correlation between genetic distance and susceptibility (Andreou et al. Reference Andreou, Arkush, Guégan and Gozlan2012) suggests that susceptibility is not dictated by phylogenetic distance increasing the risk of infections to novel hosts cohabiting with species infected with S. destruens. It is possible that by exploiting a broader phylogenetic range of hosts, the parasite will use a number of locally available hosts and in doing so will maximize its survival and range expansion opportunities (Krasnov et al. Reference Krasnov, Khokhlova, Shenbrot and Poulin2008). This appears to be a key life history trait of S. destruens that could contribute to the parasite's persistence and one that is shared by other Rhinosporideacae members. With the exception of R. seeberi (which can infect species belonging to different classes), S. destruens is the only Rhinosporideacae member, which can infect species across families. Similar life strategies have been reported for other generalist parasites, notably Sarcocystis neurona, which is the cause of equine protozoal myeloencephalitis in horses (Elsheikha, Reference Elsheikha2009). Sphaerothecum destruens’ association with the highly invasive P. parva further increases the possibility for range expansion by this parasite and its generalist nature and the high mortalities it can cause in both salmonid and cyprinid species place it as a high risk parasite for freshwater biodiversity.
Management implications
Parasites such as S. destruens cause chronic mortalities, which are extremely difficult to detect in the wild (Gozlan, Reference Gozlan2012) in the absence of long term monitoring as evidenced in Ercan et al. (Reference Ercan, Andreou, Sana, Öntaş, Baba, Top, Karakus, Tarkan and Gozlan2015). Under the Habitats Directive, the health of riverine fish populations is assessed every 5 years, which does not allow the regular monitoring of fish populations to identify and respond to observed declines (European Topic Centre on Biological Diversity, Eionet). Although, bringing together the national fish population assessments in theory is possible, in many cases a combination of missing data or incompatible data (e.g. population sizes reported in different units) makes this exercise impossible. In such cases, the Eionet assessments of conservation status are determined as the proportion of the species in each country and then evaluated. With such approach, it is clear that the emergence of S. destruens associated with P. parva's invasion at a site or even catchment level would go undetected. A close look at the latest EU Eionet monitoring, revealed that among S. destruens susceptible hosts, only S. salar was included and that the great majority of reported data originated from regions such as Scandinavia or Scotland, where P. parva is absent.
A good example of the limitation of these types of large scale, weak and uncoordinated monitoring for the purpose of disease emergence, is characterized by the population crash in Europe of the sunbleak Leucaspius delineates, one of the most susceptible host to S. destruens. It is a species that since P. parva's introduction has become extinct in several European countries and in others has experienced severe population declines. However, it is still not recorded on the Eionet's conservation list, most likely due to data deficiency in Member States assessments.
A number of recommendations are made to policy makers both at the European level and a local level. The potential risk posed by S. destruens needs to be urgently re-evaluated in light of the results presented here and an extensive epidemiological survey should be performed by primarily focusing on aquaculture facilities and fisheries where P. parva have been reported. Pseudorasbora parva need to be screened for infection with S. destruens by following a specific sampling strategy involving the sampling of a minimum of 30 fish and the use of molecular techniques to detect the parasite using at least the kidney and liver as the organs of choice. Samples for both molecular analysis and histology should be collected for individual fish. Polymerase chain reaction (PCR) (nested or quantitative PCR (qPCR)) should be used as the first step of detection, followed by histological analysis of fish determined positive by PCR (described in Andreou et al. Reference Andreou, Arkush, Guégan and Gozlan2012). Where S. destruens is detected, wild populations in adjacent water bodies should also be tested for S. destruens. This will allow a more informed evaluation of the possibility that S. destruens spread can be controlled through fish movement restrictions and inform decisions on restricting fish movement.
Perspectives
The epidemiology of S. destruens in Europe needs to be further investigated, although the current close association of P. parva and S. destruens (Gozlan et al. Reference Gozlan, St-Hilaire, Feist, Martin and Kent2005; Spikmans et al. Reference Spikmans, van Tongeren, van Alen, van der Velde and Op den Camp2013; Ercan et al. Reference Ercan, Andreou, Sana, Öntaş, Baba, Top, Karakus, Tarkan and Gozlan2015) suggests that the parasite has been introduced to Europe via P. parva invasion. This should be complemented by an extensive review of the literature including technical reports on the health of wild populations that have been cohabited or are in adjacent connected water bodies to P. parva. The combination of the parasite's epidemiology as well as any population declines reported in the literature will better inform policy makers on the impact of the parasite as well as on management options. In addition, collection of S. destruens positive samples from the wild would further the characterization of S. destruens’ invasive status in Europe. The internal transcribed spacer (ITS) region can be used to determine geographic isolation between S. destruens populations (Gozlan et al. Reference Gozlan, Whipps, Andreou and Arkush2009). The existing predictions on host–parasite interactions of generalist parasites suggest that the local diversity of susceptible host community can influence their virulence (Woolhouse et al. Reference Woolhouse, Taylor and Haydon2001), which raises concerns for the conservation of fish diversity in Europe.
ACKNOWLEDGEMENT
We thank the reviewers for their helpful comments.
FINANCIAL SUPPORT
This work was funded by the Department for Environment, Food and Rural Affairs (DEFRA), contract FC1176 to DA and GRE. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.