Skip to main content

The genetic basis of scale-loss phenotype in the rapid radiation of Takifugu fishes

Cite this dataset

Kikuchi, Kiyoshi et al. (2019). The genetic basis of scale-loss phenotype in the rapid radiation of Takifugu fishes [Dataset]. Dryad.


The genetic basis of scale-loss in Takifugu pufferfishes


Rapid radiation associated with phenotypic divergence and convergence provides an opportunity to study the genetic mechanisms of evolution. Here we investigate the genus Takifugu that has undergone explosive radiation relatively recently and contains a subset of closely-related species with a scale-loss phenotype. By using observations during development and genetic mapping approaches, we show that the scale-loss phenotype of two Takifugu species, T. pardalis Temminck & Schlegel and T. snyderi Abe, is largely controlled by an overlapping genomic segment (QTL). A search for candidate genes underlying the scale-loss phenotype revealed that the QTL region contains no known genes responsible for the evolution of scale-loss phenotype in other fishes. These results suggest that the genes used for the scale-loss phenotypes in the two Takifugu are likely the same, but the genes used for the similar phenotype in Takifugu and distantly related fishes are not the same. Meanwhile, Fgfrl1, a gene predicted to function in a pathway known to regulate bone/scale development was identified in the QTL region. Since Fgfr1a1, another memebr of the Fgf signaling pathway, has been implicated in scale loss/scale shape in fish distantly related to Takifugu, our results suggest that the convergence of the scale-loss phenotype may be constrained by signaling modules with conserved roles in scale development.


Hybrid Crosses

T. niphobles (scale-covered) were captured from Washizu port (Hamamatsu, Shizuoka: 34°43’ N, 137°33’ E), while T. pardalis and T. snyderi (both scale-uncovered) were obtained from Sagara port (Omaezaki, Shizuoka: 34°41’ N, 138°12’ E). All three species were transferred to Fisheries Laboratory, University of Tokyo. Interspecific progeny were produced by in vitro fertilization using male T. niphobles and a female from either of the two scale-uncovered species (Figure S1). F2 progeny of T. niphobles × T. pardalis (NP-F2) were obtained by crossing the NP-F1s (F1 hybrids produced from a male T. niphobles and a female T. pardalis). From the NP-F2s, a total of 109 F2 fish at 104–122 dph and 358 fish at 149 dph were used for genome- and chromosome-wide mapping, respectively. In parallel, F1 hybrids between a male T. niphobles and a female T. snyderi (NS-F1) were also generated. Next, the backcross (NS-BC) was done by crossing a male F1 hybrid with a female T. snyderi. A total of 87 fish at 42 dph were used for genome-wide mapping, whereas 203 fish at 42 dph and 222 fish at 110 dph were used for chromosome-wide mapping. In addition, F2 progenies of T. niphobles × T. snyderi (NS-F2) were obtained through crossing the F1 hybrids. From these, a total of 196 fish at 109–110 dph were used for chromosome-wide mapping.


The scale phenotype on the dissected skin samples was noted after alizarin red S staining was done. First, we evaluated scales as a binary trait where individuals with clearly stained scales and those without were considered scale-covered and scale-uncovered, respectively. Individuals with very faint staining were treated as missing data and excluded from the analysis based on the binary data. Next, we quantified the total scale number, the total area occupied by the scales, and the average area per scale (“size”) within a 10 × 10 mm area. To determine the total scale area, the 10 × 10 mm microscope image was converted into 100 × 100 pixels using Photoshop CS5 software (Adobe Inc., San Jose, USA), and the pixels on the scales were counted using the ImageJ software package. Finally, the data obtained for the total area of the scales was divided by the total number of scales within 100 × 100 pixels to calculate the size of a scale.

Usage notes

The uploaded files containg genotype and phenotype can be read using R/qtl.

R/qtl, QTL mapping in experimental crosses. Bioinformatics 2003, 19, 889–890.