4. DISCUSSION
This study investigated the relationship among three factors (larval developmental stage, minute tubercle height, and position of the larvae inside the host mussel) (Table 1; Figures 4, 5), and three main results were found. Firstly, the minute tubercles were concentrated on EHR and WLP of the Rhodeus bitterling larvae, being more developed in the latter region than in the former. Second, during the formation, growth, peak, and abrupt reduction stages, larval development occurred in the interlamellar space of the demibranchs (U and M parts); larval migration to the suprabranchial cavity (L part) occurred only during the reduction stage, when the minute tubercles became shorter (Figures 5, 6). Thirdly, when the larvae migrated to the suprabranchial cavity, morphological and physiological changes regarding their locomotion ability were apparent; in fact, individuals that migrated to the suprabranchial cavity clearly developed faster than those that remained in the demibranches.
Two types of minute tubercles of R. pseudosericeus larvae were found: hemispheric and vestigial-shaped. Among the Rhodeus bitterlings, R. atremius , R. suigensis , R. ocellatus , and R. o. smithi have only hemispheric minute tubercles, whereas R. uyekiiand R. pseudosericeus have both hemispheric and vestigial-shaped tubercles (Suzuki & Hibiya, 1984a, 1984b; Suzuki & Jeon, 1988b; Suzuki et al., 1985). The minute tubercles are unique features in bitterlings, and as wing-like projections exist in Rhodeus bitterlings but not in Acheilognathus and Tanakia , they were used as taxonomic characteristics to differ among acheilognathinae genera (Kim, 1982, 1997). Suzuki & Hibiya (1984a, b) proposed the existence of three types of yolk projections in bitterlings, and Rhodeus was considered to have type-C projections, which was confirmed for R. pseudosericeus in the present study. Suzuki & Jeon (1987) reported that the type and morphology of these minute tubercles change over time and from species to species. In the present study, R. pseudosericeus larvae with two types of minute tubercle shapes similar to those of R. uyekii larvae were found; however, R. pseudosericeus larvae had a high concentration of hemispheric tubercles only on the EHR and WLP, whereas R. uyekii larvae developed these tubercles throughout most of their PR (Suzuki et al., 1985). The two species are very similar not only in the shape of the wing-like projection and the egg yolk during the development and disappearance stages, but also in the morphology of the adult fish; however, there were differences in the developmental area and height of the epidermis (Kin et al., 2006). Moreover, R. pseudosericeus eggs are not sticky and are laid in the interlamella space of the demibranches, whereas those of R. uyekii are sticky and laid in egg masses in the suprabranchial cavity (Kim et al., 2015). The reason for these similarities and differences cannot be determined based on the results of this study. Thus, in-depth studies on speciation based on ecological characteristics and specific factors are necessary (Mayr, 1969; Arai et al., 2001).
The hemispheric minute tubercles on the WLP were approximately twice as large as those in the EHR, and the direction of the minute tubercles was slightly inclined posteriorly, making it easy for them to fixate on the gills but difficult to be removed, like a harpoon. The WLP was the largest and most developed part of the entire larva, have also the largest surface area. The minute tubercles on the WLP began to develop shortly after hatching and began to shrink during the abrupt reduction stage. The hatched larvae that entered through the mussel’s exhalant siphon settled on the demibranches, growing in their interlamellar space, during which time the larvae’s widest surface area is the WLP (Song & Kwon, 1994). Mortality of bitterling larvae occurs by two main factors: premature ejection by the mussel and death in the mussel gill by asphyxiation or nutrient deficiency (Smith et al., 2000; Kitamura, 2005; Kawamura & Uehara, 2005). The minute tubercles are formed by large unicellular epidermal cells and are presumed to be polysaccharidal in nature; studies have shown that they perform an attachment function that enables them to attach to vegetation and submerged objects (Laale, 1980). The minute tubercles occur only in larvae with no swimming ability; when fins (and consequently, the larvae’s swimming ability) start to develop, the minute tubercles are abruptly reduced (Table 1; Figures 4, 5). The minute tubercles in Acheilognathus andTanakia bitterlings, which do not have a wing-like projection, develop most intensively in the foremost part of the head, and the form of the yolk projection is scaly or hilly, different from that ofRhodeus bitterlings (Fukuhara et al., 1982; Suzuki & Hibiya, 1985; Suzuki & Jeon, 1987, 1988a, 1988c, 1988d, 1989, 1990; Park et al., 2008). The development of larger and sharper minute tubercles inAcheilognathus and Tanakia larvae compared to those ofRhodeus larvae (20–40 µm vs. 3–15 µm) is an adaptation strategy that also prevents premature ejection and allows larvae to tightly fit in the interlamellar space of the hosts’ demibranchs (Suzuki & Hibiya, 1985; Kitamura, 2006b). Further research will be required to compare with migration inside mussels in Acheilognathus andTanakia larvae for investigating the role of minute tubercle as their types.
Mussels have one exhalant and one inhalant siphon. The bitterling’s ovipositor enters a mussel’s exhalant siphon, and eggs are placed in the suprabranchial cavity or interlamellar space of the demibranch (Wu, 1998). As the inhalant siphon is connected to the mantle cavity, when the mussel’s shell opens, the larvae would become exposed to the environment; therefore, bitterling spawning must occur in the exhalant siphon to increase larval survival (Tankersley & Dimock, 1993a). The interlamellar space of the demibranchs expands as the larvae grow and becomes a limiting factor. The larvae that remained in the interlamellar space for more than 11 days after hatching were found to have a slower development than those that migrated to the suprabranchial cavity. By migrating to the suprabranchial cavity, which is larger than the interlamellar space, ventilation rates can be increased, thus increasing oxygen supply and space (Davenport & Woolmington, 1982; Mills & Reynolds, 2002). Song & Kwon (1994) reported that A. yamtsutaelarvae return to the U part as they gain physical abilities over the developmental stages. A. signifier and R. sericeus larvae, in contrast, remain in the interlamellar space only during the initial developmental stages, and as their swimming ability increases, they migrate to the suprabranchial cavity, in the direction opposite to the exhalant siphon (Aldridge, 1997; Back & Song, 2005). A. rhombeuswas reported to initially remain in the suprabranchial cavity and then migrate in the direction opposite to the exhalant siphon (Kim et al., 2018). The bitterlings’ eggs inside the gills may compete with glochidia for oxygen and space (Smith et al., 2001; Kitamura, 2005). The migration of larvae from the interlamellar space of the demibranchs to the suprabranchial cavity may reduce intraspecific competition and lower larval mortality rate in the suprabranchial cavity by providing space for growth and increased oxygen supply (Kitamura, 2006b; Spence & Smith, 2013; Methling et al., 2018).
Many previous studies have reported that mussel gill structure and conditions such as size, water flow speed, and dissolved oxygen content vary among gill positions, sexes, and density of larvae (Tankersley & Dimock, 1993a, b; Aldridge, 1999; Mills & Reynolds, 2002, 2003; Smith et al., 2004; Kitamura, 2005, 2006a, b). No glochidia were found during the present study, so the sex of the mussels was unknown. However,R. pseudosericeus larvae were mainly found in the two outer demibranchs of the four gills. U. d. sinuolauts is known to brood glochidia only in the outer demibranchs, but as the spawning season is after May, no glochidia care was observed during this study. Aldridge (1997) and Mill and Reynolds (2003) reported that the bitterlings mainly use the inner demibranchs, which had more larvae than the outer demibranchs, because of four reasons: active choice, space availability, ovipositor accessibility, and ejection ability. Studies have reported that A. rhombeus, A. cyanostigma , and R. o. kurumeus eggs were found at a higher rate in inner demibranchs than in outer demibranchs, suggesting that it is these species’ choice to avoid competition for oxygen and space with glochidia of mussels that use the outer demibranchs as brood pouches (Kitamura, 2006b, 2006c; Kim et al., 2018). Tankersley (1992a) proposed that the total flow in gills during brooding would be approximately 16% and 4% of those in non-gravid and non-marsupial gills, respectively. Kitamura (2006c) reported that female bitterlings may have been more constrained in their spawning inside the inner demibranchs irrespective of mussel sex during group spawning. Moreover, Mills and Reynolds (2003) reported that when mussels brood larvae, bitterlings spawn in inner demibranchs, but that after the mussels release their larvae, the widened outer demibranchs can be used as spawning sites. Interestingly, when the spawning patterns of mussels in March and April (i.e., before mussels brood the larvae) and in May and June (after the brooding season) were analyzed, R. pseudosericeus was found to have higher spawning rates in outer demibranchs than in inner demibranchs (per. observation). For A. signifier , twice as many larvae were identified in inner demibranchs without brooding pouches compared to the outer demibranchs with brooding pouchs (Kim et al., 2014). Further studies are necessary to elucidate the selectivity of bitterlings regarding gill position and whether it is related to gill structure or active selectivity of bitterlings (Tankersley & Dimock, 1993b).
Bitterlings have a unique early life history. The bitterlings’ eggs can be classified into four types: bulb-like, pear-shape, spindly, and ovoid; moreover some eggs are sticky (Kim et al., 2006; Kim et al., 2011). They lay a small number of eggs, develop unique tissue structures called minute tubercles during the early stages of larval development, have a very fast hatching time, and are unique in laying eggs in mussels. However, this species, of which 60 types are known worldwide, evolved due to various factors such as type of maturation, development, spawning type, spawning position and larval migration in mussels, and host selection (Smith et al., 2004; Nelson, 2006). In conclusion, the present study, by examining the development of minute tubercles, the migration of larvae inside mussels, and the physiological characteristics of the larvae, provided direct and comprehensive evidence that minute tubercles are developed to prevent the premature ejection of larvae by their mussel hosts. Thus, this finding may enhance our understanding of the evolutional advantages of the development of the minute tubercles and migration of larvae inside mussel for better survival. In this present study, however, the investigation was limited to the determination of the main factors causing growth or reduction of minute tubercles development and advantages of migration of larvae. Therefore further physiological research will be required to determine the role physiological factors.