The development of lotic-breeding Bufo torrenticola is described from zygote to completion of metamorphosis in captivity at 13±1°C. We delimit 56 developmental stages based on external features. We describe developmental stages so as to be comparable to common stages that are widely used for anurans. We also compare the larval development between B. torrenticola in lotic water and lentic water, and with lentic-breeding B. japonicus formosus in lentic water. Our results suggest tadpole mouth size in B. torrenticola is determined by genetic factors, but the tail muscle volume is determined by both genetic and environmental factors.
Introduction
In amphibians, morphological adaptation to a new habitat often involves speciation (Noble, 1931). In addition, the larval characters can show extensive environmental adaptation (Orton, 1953). Although such novel and adaptive larval characteristics have been described, their developmental bases are not demonstrated well in many cases. The Japanese stream toad, Bufo torrenticola Matsui, 1976a was described as distinct species, separating from the Japanese toad, B. japonicus. The Japanese stream toad lays eggs in running water, and the tadpoles inhabit streams in central Honshu, mainland Japan (Matsui, 1975). The larvae of B. torrenticola have wide mouth (especially when compared to lentic-breeding B. japonicus) which functions like a sucker and allows them to attach to stones in running water (Matsui, 1975). This wide mouth is also used for grazing algae on the stones in running water (Matsui, 1976b). The tail is muscular with a low fin, a likely adaptation to produce strong swimming force needed to counteract fast-flowing water (Matsui, 1975). These morphological adaptations to streams enable the separation of breeding habitat between sympatric populations of lotic-breeding B. torrenticola and lentic-breeding B. japonicus (Matsui, 1976b). However, the development of the adaptive morphology of B. torrenticola is not completely understood. Iwasawa and Saito (1989) compared the shape of mouth and body size at several larval stages between B. torrenticola and B. japonicus. However, Iwasawa and Saito (1989) made univariate comparisons between the two species, rather than employing bivariate comparisons which could take possible shape change into account. Further, the developmental stages of B. japonicus used by Ichikawa and Tahara (1966) that Iwasawa and Saito (1989) used were not standardized for comparison to the stages system developed by Gosner (1960), which is the commonly used developmental stages for frogs. Gosner (1960) provided generalized developmental stages in frogs, which enable comparison between species (McDiarmid and Altig, 2000). Gosner (1960) has accordingly been referred to by many anuran developmental studies (e.g. Iwasawa and Futagami, 1992; Shimizu and Ota, 2003; Wang et al., 2017). However, previous studies of developmental stages of Japanese toads did not refer to Gosner (1960), impeding comparisons between Japanese Bufo and other anurans. Here, we describe the complete development of lotic-breeding B. torrenticola for the first time so as to be comparable to the common anuran staging systems that are widely used. We also compare shape changes through development of B. torrenticola with related lentic-breeding species B. japonicus.
Materials and Methods
Breeding
We collected three pairs of adult Bufo torrenticola on 3 May 2020 from Kutsuki, Takashima City, Shiga Prefecture, Japan. The air temperature was 14.3°C and the water temperature was 12.3°C when we collected the pairs at 20:00 (flow velocity 0.48 m/sec). We brought these pairs into a laboratory (located in central Kyoto City 32.7 kilometers away from Kutsuki), and placed each pair into a separate plastic containers (62.4×43.2 cm in area, 31.6 cm in height, three containers in total), filled with enough tap water (23±1°C) as to cover the dorsal surface of the pairs' bodies. The water was Chlorine-free, made by holding tap water in a large tank for a few days prior to collection of amplectant pairs. After egg deposition on 4 May, we cut egg strings into ca. 10 cm lengths for allowing aspiration (Iwasawa, 1987b). Strings were kept in small plastic containers (25.7×17.3 cm in area, 4.3 cm in height) and petri dishes (8.8 cm in diameter) until they developed into tadpoles with spiracles. We then divided tadpoles into two groups, lotic water and lentic water, with group each consisting of 100 or fewer individuals. We kept each group in a container of the same size (29.2×19.0 cm in area, 17 cm in height) filled two-thirds with water. All containers were kept in an incubator (air temperature 14±1°C, water temperature 13±1°C). In one container, we attached a small electric pump (3.5×4.7 cm in area, 5.8 cm in height) to produce the lotic water like the natural habitat of the tadpole of B. torrenticola. In the other container without a pump there was no water current. After forelimbs appeared we reduced the water level and tilted the containers so that the metamorphs could move out of water (land: water=1: 1 in space). We replaced the water in the containers every second day (replacement water was dechlorinated as described above). We fed goldfish food (Hikari Co., Ltd.) to the tadpoles. We scattered one gram of food over the water surface in each container after every water change.
Developmental Stages
We distinguished and combined developmental stages following Gosner (1960) and Iwasawa (1987a), because the former is commonly used developmental staging system for frogs (McDiarmid and Altig, 2000), and the latter is the only previously available data on developmental stages of Japanese toads (Bufo). For the description of each stage we sampled eggs, embryos or tadpoles (usually five) at random. We chose Gosner (1960) or Iwasawa (1987a) for each stage of development and made complete developmental stages. Details of numbers of specimens examined are given in Table 1. Our new stages are italicized in this article, as “Stage”. We followed Iwasawa and Futagami (1992) and separated these stages into eight categories as follows: Cleavage-blastula stage (Stages 1–11); Gastrula (12–17); Neurula (18–23); Tail bud (24 and 25); External gill (26–34); Hindlimb bud (33–41); Hindlimb formation (42–50); and Metamorphosis (51–56).
Measurements
All measurements were taken to nearest 0.01 mm with a digital caliper under the stereoscopic microscope (character dimensions are shown in Fig. 1). After fixation in 5% formalin, we measured total length (TL) and observed the embryos and tadpoles at each stage with a stereomicroscope. From Stage 24, we measured body length (BL) at each stage. In addition, we used 211 specimens from Stage 40 for comparing across B. torrenticola in still water (hereafter, lentic B. torrenticola), B. torrenticola in flowing water (hereafter, lotic B. torrenticola), and B. japonicus formosus in still water. Details of the sample sizes of each stage per species are given in Table 3. From Stage 40 we measured oral width (OW), oral height (OH), tail length (TAL), maximum tail height (MTH), tail muscle height (TMH) and tail muscle width (TMW) as described by McDiarmid and Altig (2000), and maximum head width (MHW) and mid-tail muscle height (mTMH). We measured gill length (GL) and MHW at the stage with maximum gill size. We did not use BL but MHW for relative gill length comparison, because in the stage, ventral figure was not drawn in Iwasawa (1987a) and the border between body and tail was not shown in Matsui (1987). Drawings were prepared using camera-lucida. Voucher specimens are deposited at the Graduate School of Human and Environmental Studies, Kyoto University (KUHE). All procedures followed the Animal Experiment Guideline of Kyoto University and were approved by the institutional review committee of the Graduate School of Human and Environmental Studies of Kyoto University (approval nos. 30-A-7 and 20-A-5).
Comparison
To compare with a close relative (see Igawa et al. [2006] for their phylogenetic relationships), we grew eggs of Bufo japonicus formosus from Minamiechizen-cho, Fukui Prefecture (about 100 individuals) using the lentic water husbandary protocol as described above. We compared tadpoles across species and treatments after they reached Stage 40.
We compared ratios of OH/BL, OW/BL, TMW/TAL, TMW/TMH, TAL/TOL, MTH/ TAL, TMH/MTH, mTMH/MTH and mTMH/TMH among the lentic B. torrenticola, the lotic B. torrenticola, and B. japonicus formosus, by Kruskal-Wallis test and Steel-Dwass test (p<0.05 as significant) using R software (R Development Core Team, 2021).
In order to test for developmental changes in mouth shape we calculate OH/BL and OW/BL ratios in the lentic B. torrenticola at Stages 32 to 52 (equivalent to stages 31 to 42 in Ichikawa and Tahara [1966]), at which point the extensive change in the size of the mouth in the lentic B. torrenticola was observed (Iwasawa and Saito, 1989).
Table 1.
Stages of normal development of Bufo torrenticola. The initial number represents the frog's stage as defined by us, while the numbers in parentheses (G1–G46) and brackets (I1–I45) represent the corresponding developmental stages as defined by Gosner (1960) and Iwasawa (1987a), respectively. Total length (in mm) is given as x̄ ±SD, followed by ranges and sample sizes in parenthesis. Age (h:m) indicates the minimum age.
Table 1. (continued)
Table 1. (continued)
Results
Developmental stages
The complete normal development of Bufo torrenticola was characterized and divided up into a total of 56 developmental stages which we also illustrate (Table 1, Figs. 2, 3). Completion of metamorphosis took a minimum 128 days.
Stages 1 and 2 as we characterised them corresponded to those in Gosner (1960) but are not illustrated in Iwasawa (1987a). For Stages 9–25 and 27–31, we followed mainly stages 8–23 and 26–30 in Iwasawa (1987a) as this work was more detailed than Gosner (1960). Stages 3–8 corresponded to stages 2–7 in Iwasawa (1987a) and 3–8 in Gosner (1960).
We could not identify stage 24 in Iwasawa (1987a), which was defined by having 13 somites and showing muscular response against stimulation, but we observed a stage showing the muscular response and appearance of gill buds (shown in stage 25 in Iwasawa [1987a]) at the same timing. Thus, we could not separate stage 24 from stage 25 in Iwasawa (1987a) and combined these two stages as to be Stage 26.
Fig. 2.
Illustrations of eggs, embryos and larvae of Bufo torrenticola from Shiga Prefecture. Numerals at the bottom right of each drawing correspond to developmental stages defined in Table 1.
Fig. 3.
Illustrations of larvae of Bufo torrenticola from Shiga Prefecture. Numerals at the bottom right of each drawing correspond to developmental stages defined in Table 1.
In both this study and Iwasawa (1987a) hindlimb development began during opercular development.
We adopted Iwasawa's (1987a) stages 33 to 35 as Stages 33 to 35 because unlike Gosner (1960) but in concordance with Iwasawa (1987a) the papillae appearance and teeth development were observed as oral disc development. However, the timing of oral disc completion in Iwasawa (1987a) was different from our observations; we observed completion of oral disc development at Stage 35 before appearance of the spiracle, while Iwasawa (1987a) observed oral disc completion after the appearance of the spiracle.
The cloacal tail piece was lost at appearance of subarticular tubercles and inner and outer metatarsal tubercles in our observations (Stage 48), but lost after completion of development of subarticular tubercle and inner and outer metatarsal tubercles in Gosner's (1960) stage 41.
In hindlimb and toe development, we followed Gosner (1960) and defined Stages 33–50, because Gosner (1960) mentioned and defined some stages based on the subarticular tubercle and inner and outer metatarsal tubercles of both limbs, while Iwasawa (1987a) did not mention them.
In metamorphosis, we combined Gosner's (1960) and Iwasawa's (1987a) definitions to create Stages 51–55, because Gosner (1960) defined stages based on degree of mouth fissure, and Iwasawa (1987a) defined stages based on degeneration of the tail. At Stages 53 and 54 the skin covering the forelimbs tore and the external gills (which were drawn inside the body at Stage 34) became partly exposed again. However, in Bufo japonicus formosus at Stages 53 and 54 the external gills were re-exposed in only one out of eight individuals (Table 2, Fig. 4).
The developmental stages of embryos and larvae were identical between Bufo torrenticola and B. japonicus formosus, except for body color and relative size of some characters. The larvae of B. torrenticola had a darker body than B. j. formosus. We could not observe internal organs through dark skin in B. torrenticola, but could in B. j. formosus through its transparent skin.
Comparison
The median of GL/MHW of Bufo torrenticola at the Stage 31 (n=3) was 0.52. On the other hand, that of B. japonicus formosus was 0.71 at the same stage (data were measured from Iwasawa [1987a]), so the former has shorter gills than the latter, although a statistical test could not be applied.
At Stages 40, 42, 43, 45, and 46, OH/BL and OW/BL were significantly (p<0.05) larger in Bufo torrenticola (Table 3) (based on the Kruskal-Wallis test and Steel-Dwass test, implemented in the R software [R Development Core Team, 2021]). However, no significant difference in these ratios were detected between the lentic and lotic tadpoles of B. torrenticola. The larva of B. torrenticola had a mouth ratio against BL about two times larger than B. j. formosus from the Stage 40 until initiation of metamorphosis (Fig. 5).
There were no differences in TMW/TAL, TMW/TMH, TAL/TOL and MTH/TAL between species.
For mTMH/MTH the lotic Bufo torrenticola was significantly larger than B. japonicus formosus at Stages 40 and 43. The lentic B. torrenticola has larger mTMH/MTH than B. j. formosus at Stage 45. The lotic B. torrenticola has the largest mTMH/MTH, followed by the lentic B. torrenticola, and then B. j. formosus at Stage 46. In TMH/MTH, the lotic B. torrenticola was significantly larger than B. j. formosus at Stage 42. The lotic B. torrenticola has the largest TMH/MTH, followed by the lentic B. torrenticola, and then B. j. formosus at Stage 43. The lentic B. torrenticola has larger TMH/MTH than B. j. formosus at Stages 45 and 46. In MTH/TAL, B. j. formosus was larger than the lentic B. torrenticola at Stage 42. Bufo j. formosus has larger MTH/TAL than the lotic B. torrenticola at Stages 43 and 45. Bufo j. formosus has the largest MTH/TAL, followed by the lentic B. torrenticola, and then the lotic B. torrenticola at Stage 46 (Table 4 and Fig. 6). For OW/BL and OH/BL in the lentic B. torrenticola we found that these ratios were stable from Stages 32 to 50 and decreased from Stages 51 to 52 (Table 3 and Fig. 7). In summary, B. torrenticola had a more muscular tail and a smaller percentage of fins than B. j. formosus. However, the degree of the difference between the species varied between B. torrenticola raised with or without water current.
Table 2.
Major changes to the developmental stages chart.
Discussion
This study provides the first complete description and comparisons of the developmental stages of Bufo torrenticola. From Stage 31 this species has a wide oral disc, which must be adaptation to running water in streams and is a unique character for this species. The gills developed at the same stage as lentic-breeding B. japonicus formosus, however they do not increase in the length, probably because they live in oxygen rich flowing streams (Matsui, 1976b). The thick and low tail shown in Matsui (1975) was first documented through the detailed developmental stages and is also considered to be advantageous for swimming in streams (McDiarmid and Altig, 2000).
Table 3.
Median of OH/BL and OW/BL. We described the lentic Bufo torrenticola, the lotic B. torrenticola and B. japonicus formosus as A, B and C, respectively. Range is shown in parenthesis and number of samples is shown in bracket.
Fig. 5.
Developmental difference in values (%) of OH/BL (top) and OW/BL (bottom) in lentic Bufo torrenticola, lotic B. torrenticola and B. japonicus formosus from Stages 40 to 50.
Completion of metamorphosis took a minimum of 128 days at 13±1°C water temperature in our study. Under natural conditions complete metamorphosis occurs in about 150 days (Matsui and Maeda, 2018). Iwasawa and Saito (1989) reported that Bufo torrenticola took 53 days to reach stage 35 of Ichikawa and Tahara (1966) at 8°C water temperatures. In our study, it tooks 24–37 days to reach Stages 37–40 (equivalent to stage 35 of Ichikawa and Tahara [1966]) under conditions ca. 6°C warmer than Iwasawa and Saito (1989). The duration to completion of metamorphosis in our results seemed to be very long. However, the studies noted above did not observe until completion of metamorphosis, or kept embryos in very low temperature. In contrast, Iwasawa (1987a) reported that B. japonicus formosus from Niigata Prefecture took 58, 44 or 28 days to complete metamorphosis at 15, 18 or 20°C, respectively. Muto et al. (1968) showed that B. j. formosus took 37 days to complete of metamorphosis at 20°C. It has previously been documented that B. torrenticola has a longer larval life than B. j. formosus (Matsui, 1976a; Iwasawa and Saito 1989). Bufo torrenticola lays eggs in the stream with year-round stable cool temperature (Matsui and Maeda, 2018) and poor food availability (mainly algae). Conversely, B. j. formosus lays eggs in still water with a rich food supply, including some temporary water bodies that dry up relatively rapidly. We hypothesize that B. torrenticola does not metamorphose as rapidly as B. j. formosus because it occurs in a more stable but nutrient poor environment.
As the forelimbs protruded (Stages 53 and 54) we found internal gills were exposed in Bufo torrenticola, but not in most individuals of B. japonicus formosus, a difference linked to the larger skin tear in the protruding forelimb stages in B. torrenticola. We were also not able to find internal organs through the dark skin of B. torrenticola, but could do so in B. j. formosus through its transparent skin. In our observation, B. torrenticola seemed to have thicker skin than B. j. formosus, which might be advantageous for protecting the body in fast-flowing streams. Thick skin may be disadvantageous for oxygen intake through skin, the rich supply of dissolved oxygen in streams may compensate for this disadvantage.
Iwasawa and Saito (1989) reported that the mouth of B. torrenticola was larger than B. japonicus formosus at stages 35–38 as defined by Ichikawa and Tahara (1966). However, they did not consider an interspecific difference in body size into consideration. Here we demonstrate that relative mouth size (OH/BL and OW/BL) decreased through stages in B. j. formosus, but was stable in B. torrenticola. The larvae of B. torrenticola maintain a large mouth size relative to body size even in the stages close to metamorphosis, which could be advantageous in fast-flowing stream environments where the mouth functions as an adhesive organ (see similar cases in tropical lotic tadpoles: Inger, 1992).
Table 4.
Median of mTMH/MTH, TMH/MTH and MTH/TAL. We described the lentic Bufo torrenticola, the lotic B. torrenticola and B. japonicus formosus as A, B and C. Range is shown in parentheis and number of samples is shown in bracket.
Our results indicated that the larger oral disc in Bufo torrenticola than B. japonicus formosus is stable and possibly determined by genetic factors, however the tail of B. torrenticola became more muscular in individuals kept in water with a current than in still water, suggesting this phenotypic variation is linked to both genetic and environmental factors. The phenotypic plasticity of B. torrenticola larvae is possibly an adaptation to differences in water current of breeding streams. Although we did not keep larvae B. j. formosus in lotic conditions, we expect they cannot grow up well in such conditions, because they have no adaptive characters for lotic habitats like B. torrenticola. Matsui (1972) reported that larvae of B. bufo japonicus (presently B. j. japonicus or B. j. formosus) flushed into streams by flooding from nearby ponds were washed away by water currents.
Fig. 6.
Developmental difference in values (%) of mTMH/MTH (top), TMH/MTH (middle) and MTH/TAL (bottom) in lentic Bufo torrenticola, lotic B. torrenticola and B. japonicus formosus from Stages 40 to 50.
Fig. 7.
Developmental difference in values (%) of OH/BL (top) and OW/BL (bottom) in lentic Bufo torrenticola from Stages 32 to 52.
Asian species of Bufo consist of two clades, one contains B. bankorensis, B. japonicus, B. gargarizans (including lotic B. gargarizans andrewsi and three lentic subspecies), B. torrenticola, B. tuberculatus, and B. stejnegeri, and the other comprises B. aspinius, B. cryptotympanicus, and B. tuberospinius (Liu et al., 2000; Li et al., 2020). In the first clade, only B. g. andrewsi, B. stejnegeri, and B. torrenticola lay eggs in streams, while the remaining taxa use ponds (Liu, 1950; Matsui, 1986; Matsui and Maeda, 2018; Schmidt, 1931). The stream-breeding species in this clade do not form a monophyletic group (Fong et al., 2020). In the second clade, B. aspinius lays eggs in streams, and the breeding habit of B. cryptotympanicus and B. tuberospinius is unknown. However, tadpoles of B. tuberospinius are known to occur in streams and have an abdominal sucker, and metamorphs of B. tuberospinius and B. cryptotympanicus have indistinct tympanum (Liu and Hu, 1962; Rao and Yang, 1994; Yang et al., 1996). Anurans which live in and along streams have reduced or lost tympanic-columellar system (Duellman and Trueb, 1986). These lines of information suggest that all species of the second clade may be lotic-breeders, which were once treated as an independent genus Torrentophryne (Yang et al., 1996). It is obvious that the lotic-breeding and/or lentic-breeding has evolved multiple times in Asian Bufo. Fong et al. (2020) show that the lotic ecology of B. stejnegeri, B. torrenticola, and B. andrewsi evolved independently. However, if all toads in the second clade are the lotic-breeding, common ancestor of Asian species of Bufo could be assumed as the lotic-breeder. Unfortunately, their phylogenetic relationship has not been resolved with significant supports (Liu et al., 2000; Li et al., 2020). Detailed description of larval development of the other lotic-breeding Bufo might contribute to clarify evolution of lotic-breeding and the taxonomy of the genus.
Acknowledgments
We are grateful to Ibuki Fukuyama, Sally Kanamori, and Ryobu Fukuyama for assistance in field survey, Makoto Ito for facilitating our study and Sotaro Hara for providing tadpoles.
