Interspecific variation of prevalence by Scaphanocephalus (Platyhelminthes: Trematoda: Heterophyidae) metacercariae in parrotfishes (Labridae: Scarini) from an Okinawan coral reef

Metacercarial cysts of the parasite Scaphanocephalus (Platyhelminthes: Trematoda: Heterophyidae) are frequently found on the pectoral fins and skin of parrotfishes (Labridae: Scarini) inhabiting Okinawan coral reefs in southern Japan. The prevalence of metacercarial cysts in 30 parrotfish species was investigated and compared through a market survey. Although parasite prevalence differed between parrotfishes, all species examined are considered to be suitable hosts. Prevalence was high in Scarus chameleon (38.5%, n = 13), S. rubroviolaceus (33.4%, 2797), S. ghobban (26.6%, 6441), and several other species that share, in part, common feeding habits. Conversely, prevalence was low in S. prasiognathos (0.4%, 1842), Bolbometopon muricatum (0.4%, 270), and Hipposcarus longiceps (0.1%, 8512) which have different feeding habits. Despite a few exceptions, feeding ecology and other indirect behaviors are considered to affect the prevalence of metacercarial cysts in parrotfishes. Taxonomic affiliation and nocturnal mucous cocoon usage are not considered to affect parasite prevalence.


Introduction
Parrotfishes (Labridae: Scarini) are an important component of coral reef ecosystems, and are harvested by commercial fisheries in the Okinawa region of southern Japan (Kishimoto, 1984;Shimose et al., 2019). The cysts (metacercariae) of Scaphanocephalus adamsi (Platyhelminthes, Trematoda, Heterophyidae) are frequently found on the pectoral fins and skin of commercially harvested parrotfishes in Okinawa (Iwata, 1997). These cysts, which are visible to the naked eye ( Fig. 1), markedly reduce the commercial value of parrotfishes (Shimose et al., 2019). To improve the effective utilization of these fishery resources, data on the occurrence patterns of this parasite are useful for assessing the negative impact that these parasites have on parrotfish consumption, and to optimize the management of parrotfish stocks.
The genus Scaphanocephalus contains three species --S. expansus, S. australis, and S. adamsi--and a previous study on metacercarial cysts found on parrotfishes in Okinawa (i.e., Scarus sordidus and S. rubroviolaceus) identified them as S. adamsi (Iwata, 1997). A recent study also identified the metacercarial cysts on some parrotfishes as S. adamsi, and all cysts infecting parrotfishes in Okinawa are considered to belong to the same species (Katahira et al., unpublished data). Metacercarial cyst infection rate (i.e., prevalence) appears to vary among parrotfish species, but host specificity has not yet been evaluated. Coastal snails, marine fishes, and piscivorous raptors are thought to be the first intermediate, the second intermediate, and the final hosts of Scaphanocephalus parasites, respectively (Iwata, 1997). However, the process of this parasite infection event is not fully understood. In this study, parasite prevalence in the second intermediate host, parrotfishes, was assessed by means of a fish market survey as this permitted a large number of samples to be examined. Interspecific differences in parasite prevalence were evaluated quantitatively, and ecological factors influencing the prevalence in parrotfishes are discussed based on a review of the literature. The findings are expected to contribute to a more comprehensive understanding of species interactions in coral reef ecosystems.  (Shimose et al., 2019). The market survey was conducted two or three days a week between November 2011 and October 2017 (six years), and comprised 740 days in total. The number of fish recorded did not fluctuate markedly among years or seasons. Parrotfishes being auctioned were identified to species, body weight was estimated, and parasite prevalence was assessed. The auction unit comprised a box containing one or more fishes. Basically, similarly sized individuals are placed into the same box for auction. Therefore, individual fish body size (assessed as in weight in kilograms) was estimated based on the weight of the auction box divided by the number of fish in the box. Parasite infection on fins and skin was checked by staff of the fisheries cooperative, and infected fish were placed into a same box before the start of the auction. Infected parrotfishes were occasionally found among uninfected parrotfishes, but the first author correctly recorded prevalence in these cases.

Estimation of parasite prevalence
The parasite prevalence (number of fish infected/number of fish recorded; Bush et al., 1997) of individuals can be explained by a logistic model with the component of each parrotfish species being the only explanatory variable. The components to which each parrotfish species belonged was modeled using a finite mixture distribution model (Leisch, 2004). This model was used to classify the parasite prevalence data for each parrotfish species. The Expectation-Maximization algorithm was then used to identify the components. This finite mixture regression model, which was implemented to assess parasite prevalence, was a mixture of binomial distributions with s components, where the likelihood of the ith observation is given by: Here, H denotes the mixture probability, j denotes the component number, πdenotes the mixing proportion (∑ = = π 1 j s j 1 ), Bi denotes the binomial distribution, α denotes all parameters, p i denotes the parasite prevalence, and θ j denotes the estimated probability of parasite prevalence for the jth component. The likelihood was given by the product of Eq. (1) for all observations. The Bayesian information criterion (BIC) for the finite mixture models was compared with 2-7 components. For the best fit of the component models, the estimated probability of parasite prevalence for each component was investigated and the components were ranked by the estimated probability. Analyses were performed using R ver. 1.0.136 (R Development Core Team, 2008) and the flexmix library ver. 2.3.13 (Leisch, 2004) was used for finite mixture regression analysis.

Detection of ecological factors affecting parasite prevalence
To evaluate the ecological factors that could potentially affect parasite prevalence, the following information was collected for each parrotfish species from the literature, current research by the authors, and interviews with fishermen: 1) extent of scarring on corals or coral reefs as a result of feeding (scarring observed, scarring occasionally observed, and scarring not observed; Bellwood and Choat, 1990); 2) diet (epilithic algae, algae with live coral, or algae with sand surface; Bellwood and Choat, 1990); 3) potential angling target as an adult (Joh, 2001;Konishi and Nakabo, 2007; WEB-sakanazukan, 2020; personal observations); 4) use of mucous cocoons at night (interviews with fishermen).

Parasite prevalence in parrotfishes
A total of 59,475 individual parrotfishes comprising 30 species in six genera were recorded (Table 1). The number of fishes recorded did not fluctuate markedly among years or seasons. However, the number of individuals of particular species ranged from 11,553 Chlorurus microrhinos individuals to only one Ch. capistratoides individual. Body weights also differed among species, ranging from a mean ( ± SD) of 0.33 ( ± 0.03) kg in Scarus spinus to 1.88 ( ± 3.51) kg in Bolbometopon muricatum. Scaphanocephalus adamsi infection was confirmed in all parrotfish species except for Ch. bleekeri, but this was considered to be due to the small sample size (n = 3). Two parrotfish species with a parasite prevalence of 0% (n = 3, Ch. bleekeri) and 100% (n = 1, Ch. capistratoides) were excluded from subsequent analysis.

Ecological factors influencing parasite prevalence
In the assessment of the effect of feeding behavior on parasite prevalence, "high" prevalence was observed in species in which scarring was "occasional and "no". In the assessment of the effect of diet on parasite prevalence, parasite prevalence was "high" in parrotfish species that fed on "epilithic algae with sand surface" and "low" or "no" in species that fed on "epilithic algae with live coral". Target angling species showed higher parasite prevalence than no target species. Mucous cocoon production did not appear to affect parasite prevalence.

Discussion
Although the prevalence of the metacercarial cysts differed markedly among host species, almost all species were infected by the parasite. A previous qualitative investigation of 119 species in 49 families of reef fishes from Okinawa showed that only eight species in four families (Aulostomidae, Mullidae, Labridae, and Tetraodontidae) were infected by S. adamsi (Iwata, 1997). Their results imply that the parasite exhibits weak host specificity; i.e., the parasite can infect some host fish species but prefers specific host species. The present study also showed the existence of host specificity in this parasite, even among hosts within the same tribe Scarini. This specificity may be related to host taxonomy (i.e., physical/physiological tolerance to the parasite; Loerch et al., 2015), ecological traits of host species (e.g., habitat preference, feeding ecology; Sikkel et al., 2009;Elmer et al., 2019), or factors related to taxonomy and feeding ecology (Sikkel et al., 2018).
The parrotfishes examined in the present study belonged to six genera (Bellwood, 2001). Of the six genera, Calotomus (1 sp. in this study) was the outermost group (Streelman et al., 2002) and was highly infected (Fig. 2). Conversely, the genera Bolbometopon (1 sp.), Cetoscarus (1 sp.) and Hipposcarus (1 sp.) were rarely infected. The genera Scarus and Chlorurus are considered to be sister groups (Streelman et al., 2002) and are highly divergent among parrotfishes (Bellwood, 1994); parasite prevalence in these genera differed markedly among species (Table 2). These results suggest that parasite prevalence in parrotfishes was not affected by phylogenetic relatedness (Streelman et al., 2002) or feeding mode among host species (feeding mode linked to genus; Bellwood, 1994). In addition, parasite prevalence was also found to be independent of host size.
A variety of ecological factors can affect parasite prevalence in parrotfishes. These factors act as encounter filters that determine whether the infective cercariae that are released from the first intermediate hosts successfully reach the second intermediate fish hosts (Combes, 1991;Poulin, 2007). Possible filters include the diurnal and nocturnal habitats used by the host parrotfishes, and whether or not they construct a mucous cocoon when they sleep at night. The habitat preferences of parrotfishes have been studied extensively, with most species inhabiting the shallow areas, around reefs, lagoons and dropoffs (Bellwood, 2001;Katoh, 2016); however, no clear relationships between habitat utilization and parasite prevalence in parrotfish was detected.
Constructing mucous cocoons to sleep at night is a well-known parrotfish behavior (e.g., Casimir, 1971;Kishimoto, 1984). The cocoons have been shown to protect the parrotfish against attacks by gnathiid isopods, which feed on fish blood (Grutter et al., 2011). Other possible functions of the cocoon include avoiding predation by moray eels (Winn and Bardach, 1959) and/or to protect the parrotfishes from infection by Scaphanocephalus parasites. According to fishermen on Ishigaki Island who target parrotfishes at night, Calotomus, Hipposcarus, and some Scarus species do not make cocoons, while all Chlorurus species and some Scarus species do (Table 2). In this study, the parrotfishes that made cocoons, such as S. festivus (14.3%) and C. sordidus (13.2%), were found to be moderately infected, whereas parrotfishes that did not make cocoons, such as S. prasiognathos (0.4%) and H. longiceps (0.1%), were almost never infected. These results suggest that the cocoons do not play an important role in protecting parrotfish from Scaphanocephalus parasites. Given that they are typically short-lived and highly susceptible to environmental perturbations (Pietrock and Marcogliese, 2003), the free-living trematode cercariae typically have a temporal pattern to be released from the first intermediate host (Combes et al., 1994). It is therefore the timing of making cocoons at night may not match the timing of release of cercariae.
As an alternative scenario for the mode of parasite infection, the range of feeding habits exhibited by parrotfish could be considered. The diet (Bellwood and Choat, 1990) and utilization of a particular target species for recreational angling (Joh, 2001;Konishi and Nakabo, 2007) are known for a variety of parrotfish with different feeding habits (Table 2). Parrotfishes in genera Bolbometopon, Cetoscarus, Chlorurus, Hipposcarus and Scarus generally feed on epilithic algae, and some species also feed on live corals and ingest sand surface (Bellwood and Choat, 1990). The diet of three highly infected parrotfishes, S. chameleon (38.5%), S. psittacus (25.0%), and S. schlegeli (22.5%), consists of epilithic algae with sand surface (Bellwood and Choat, 1990). The feeding behavior of two other highly infected parrotfishes, S. rubroviolaceus (33.4%) and S. ghobban (26.6%), can be characterized as "scarring occasionally observed" on corals (Bellwood and Choat, 1990). This scarring is referred to as being occasional because these species have a more generalist diet. On the other hand, the diet of parrotfishes that were only rarely or almost never infected, B. muricatum (0.4%), S. rivulatus (1.5%), Ce. bicolor (2.0%), and C. microrhinos (2.2%), consists of epilithic algae with live coral (Bellwood and Choat, 1990). These species also tend to occur in relatively shallow, wave-exposed locations (Bellwood et al., 2003;Fox and Bellwood, 2007) in areas with relatively clean algal turfs (Purcell and Bellwood, 2001).
These findings imply that feeding ecology appears to be most strongly associated with parasite prevalence in parrotfishes. For example, the parrotfish species that tend to cause occasional scarring in corals, ingest the sand surface, and are caught by angling using crustacean baits, tend to be infected more. Those species that feed on live corals and are not caught by anglers tend to be less infected. However, metacercarial infection is thought to occur via external routes, such as through penetration of the fins or skin of the host, and not through ingestion of prey items (Cribb, 2005). Therefore, the relationship between feeding ecology and parasite prevalence appears to be indirect. It also appears that parrotfish feeding behavior in habitats where the first intermediate host (e.g., snails) occurs (Iwata, 1997), or in which the Table 2 Rank group by parasite prevalence, taxonomic (based on genus) and ecological information of 28 major (n > 5) parrotfishes (Scarini spp.) recorded at Yaeyama fish auction site. Feeding mode is from Bellwood (1994). Scarring and diet are from Bellwood and Choat (1990); E: epilithic algae, E (LC): E with live corals, E(S): E with sand surface. Angling target is from Joh (2001), Konishi and Nakabo (2007), WEB-sakanazukan (2020), and personal observation. Mucous cocoon making behavior is based on the interviews with fishers in Yaeyama. *Prevalence and rank are reversed because of sample size. pelagic stage of Scaphanocephalus occurs, are the primary factors affecting parasite prevalence in parrotfishes. At this stage of development, the parasite may occur at high densities and have a localized distribution compared to other areas. For example, if the parasite is concentrated near the sand surface or in the habitats of the krill-like crustaceans that are preyed upon by the targeted parrotfishes, then the interspecific differences in parasite prevalence among parrotfish species could be explained. Similarly, it is possible that the parasite is rare near the live corals that are preyed upon by some parrotfish species. The prevalence of the congeneric Scaphanocephalus expansus is known to vary as a function of habitat characteristics (i.e., depth and area), even in the same host species (e.g., Acanthurus tractus, Elmer et al., 2019). In future studies, parasite prevalence in parrotfishes should therefore also be examined among different habitats.
In conclusion, the prevalence of Scaphanocephalus in parrotfishes is considered to be linked to the feeding ecology of the host, and not to taxonomic affiliation or cocoon-making behavior. The life history and infection process for Scaphanocephalus parasites are still poorly understood. Ecological traits of the host fish species and detailed studies of their feeding behavior will contribute to a more comprehensive understanding of parasite-host relationships in coral reef ecosystems.

Declaration of competing interest
On behalf of all authors, the corresponding author states that there is no conflict of interest.