Access this dataset on Dryad

This dataset supports the analyses reported in Katsumura et al. (2026), PNAS 123(13), e2534817123 (https://doi.org/10.1073/pnas.2534817123). It comprises morphometric measurements (standard length and gut length) of Japanese medaka (Oryzias latipes) from three origins — wild-caught individuals, laboratory-reared fish derived from wild populations, and CRISPR/Cas9-generated plxnb3 upstream-deleted (∆upPlxnb3) mutants — along with immunohistochemistry images of medaka gut tissue, genome-wide SNP genotypes obtained by RAD-seq, and geographic coordinates of the sampling regions. These resources underpin Fig. 1, Fig. 3, Fig. 4, Fig. 5, and Fig. S6 of the manuscript.

Description of the data and file structure

Morphometric and geographic data are provided as CSV files (comma-delimited, UTF-8 with BOM, a single header row). SNP data are provided in a standard VCF file. Immunohistochemistry images are organised in per-individual folders bundled into a single ZIP archive.

Population genetic grouping. Medaka populations are assigned to seven genetic subgroups following the structure reported in Katsumura et al. (2019, G3 9: 217–228): six Japanese subgroups — NJPN1 (derived Northern Japanese group: Kamikita, Yokote, Niigata, Kaga, Maizuru, Miyazu); NJPN2 (ancestral Northern Japanese group: Amino, Kumihama, Toyooka, Kinosaki, Hamasaka); SJPN1 (PO.NEH: Ichinoseki, Toyota, Mishima, Iwata, Sakura, Shingu); SJPN2 (Sanyo/Shikoku/Kinki: Ayabe, Tanabe, Okayama, Takamatsu, Kudamatsu, Misho); SJPN3 (San-in: Kasumi, Tsuma, Tottori, Hagi); SJPN4 (Northern and Southern Kyushu: Kusu, Arita, Hisayama, Fukue, Kazusa, Izumi, Hiwaki, Nago) — and KOR (KOR/CHN: Yongcheon, Maegok, Bugang, Sacheon, Shanghai) for populations from the Korean peninsula and eastern China.

Missing data. No missing values are present in the CSV files.

Files and variables

File: WildCapturedMedaka.csv

Standard length and gut length measurements of wild-caught medaka (and of their descendants reared at an outdoor facility for the transplantation experiment), used in Fig. 1 and related analyses. Each row corresponds to a single individual (n = 309).

• ID — Unique identifier for each specimen. The prefix encodes the sampling river (e.g., KMB = Kabe, EJ = Ejiri, SHD = Shihodo).
• Sex — Sex of the individual. Values: female, male.
• River — Name of the river or stream from which the lineage was captured in the wild (Kabe, Ejiri, Shihodo; all in Kagawa Prefecture, Japan).
• Place — Location at which the individual itself was sampled and measured. Two values occur:
  • Kagawa — individuals sampled directly from the wild in Kagawa Prefecture.
  • Kashiwa — individuals from the transplantation / outdoor breeding experiment (Fig. S5). In August 2015, wild medaka from the Kabe river were transported to the outdoor breeding facility on the Kashiwa campus of The University of Tokyo; the offspring bred there were sampled in September 2016 and February 2017.
• Sampling date — Date of sampling, formatted as YYYY/M/D (e.g., 2015/2/22). Range: 2014/1/25 – 2017/2/20.
• Year — Year of sampling, as an integer (2014–2017).
• Season — Season of sampling. Values: Summer, Winter.
• Standard length (mm) — Standard body length, in millimetres, measured from the most anterior part of the head to the end of the vertebral column on photographs using ImageJ.
• Gut length (mm) — Length of the isolated gut (from esophagus to anus), in millimetres, fixed in 4 % paraformaldehyde and measured on photographs using ImageJ.

File: WildDerivedLabMedaka.csv

Standard length and gut length measurements of laboratory-reared medaka derived from 40 wild populations, sampled in 2015 at the outdoor breeding facility of The University of Tokyo, and used in Fig. 1 and related analyses. Each row corresponds to a single individual (n = 341). The stocks are descendants of populations collected from each wild habitat in 1983 (Shima et al., 1985) and have been maintained by taking offspring every 1–2 years in independent tanks under common rearing conditions.

• ID — Unique identifier for each fish (prefix encodes the population of origin; e.g., YC = Yongcheon).
• Sex — Sex of the individual. Values: female, male.
• Population — Population of origin (locality name).
• Subgroup — Genetic subgroup to which the population belongs (NJPN1, NJPN2, SJPN1, SJPN2, SJPN3, SJPN4, KOR; see "Population genetic grouping" above).
• Sampling date — Date of sampling at the breeding facility, formatted as YYYY.M.D (e.g., 2015.8.25). All entries fall within 2015.
• Standard length (mm) — Standard body length, in millimetres.
• Gut length (mm) — Length of the isolated gut, in millimetres.

File: WildDerivedLabMedaka2021.csv

Standard length and gut length measurements from the common-garden experiment in 2021, used in Fig. S6. Medaka were sampled on 3 February 2021 (winter) and 30 August 2021 (summer) from available wild-derived laboratory stocks. Each row corresponds to a single individual (n = 167).

• ID — Unique identifier for each fish (e.g., MY-w01, where MY = Miyazu and w = winter sampling).
• Sex — Sex of the individual. Values: female, male.
• Population — Population of origin (locality name).
• Subgroup — Genetic subgroup (NJPN1, NJPN2, SJPN1, SJPN2).
• Sampling date — Date of sampling at the breeding facility, formatted as YYYY/M/D (all within 2021).
• Year — Year of sampling (all 2021).
• Season — Season of sampling. Values: Summer, Winter.
• Standard length (mm) — Standard body length, in millimetres.
• Gut length (mm) — Length of the isolated gut, in millimetres.

File: DeltaUpPlxnb3Medaka.csv

Standard length and gut length measurements of CRISPR/Cas9-generated medaka used in Fig. 3. The mutant allele (∆upPlxnb3) carries a 334-bp targeted deletion of the seasonally methylated region upstream of plxnb3; individuals are F3 progeny from heterozygous F2 × F2 mating, reared at 28 °C under a 14 h light / 10 h dark cycle. Each row corresponds to a single individual (n = 59).

• ID — Unique identifier (dP01–dP60; one of the 60 genotyped F3 individuals was excluded because the gut could not be isolated).
• Sex — Sex of the individual. Values: female, male.
• Day after hatch — Age at measurement, in days post-hatching (range: 94–122; mean ≈ 103 days).
• Genotype — Genotype at the plxnb3 upstream-deletion locus:
  • +/+ — Homozygous wild-type (reference allele on both chromosomes) (n = 25).
  • +/- — Heterozygous (one wild-type allele and one ∆upPlxnb3 allele) (n = 22).
  • -/- — Homozygous ∆upPlxnb3 (deletion on both chromosomes) (n = 12).
• Standard length (mm) — Standard body length, in millimetres.
• Gut length (mm) — Length of the isolated gut, in millimetres.

File: PopulationLocations_forAnnualTemperature.csv

Geographic coordinates and genetic subgroup assignments for the 35 sampling regions. Each row corresponds to one population/locality. These coordinates were used in Fig. 5C to relate gut-length variation to the climatic environment of each region. Annual mean air temperatures used in the manuscript were obtained from the Global Solar Atlas database (https://globalsolaratlas.info; CC BY 4.0 license); they are not included in this file to maintain compatibility with the Dryad CC0 license waiver.

• Population — Locality name (35 populations in total).
• Subgroup — Genetic subgroup to which the population is assigned (NJPN1, NJPN2, SJPN1, SJPN2, SJPN3, SJPN4; see "Population genetic grouping" above).
• Latitude — Latitude of the sampling site, in decimal degrees (WGS84). Range: 26.59°–40.77° N.
• Longitude — Longitude of the sampling site, in decimal degrees (WGS84). Range: 127.98°–141.26° E.

File: Gut_IHC_images.zip

Immunohistochemistry (IHC) image data of medaka gut tissue used in Fig. 3D. Wild-type (+/+) and homozygous ∆upPlxnb3 (-/-) medaka were compared (n = 5 individuals per genotype). Five-micrometre serial tissue sections were prepared and incubated with a mouse monoclonal anti-phosphorylated neurofilament H antibody (BioLegend, #smi-31, 1:1000) whose cross-reactivity for medaka has been confirmed (Uemura et al., 2015, PLoS Genet. 11, e1005065). An average of six images of intestinal villi per individual were captured at 100× objective magnification on a light microscope (Olympus BX63) equipped with a digital camera (Olympus DP74).

Archive structure. When extracted, the archive contains ten per-individual folders and an accompanying README.txt:

• drR_1 – drR_5 — Wild-type (+/+) individuals (n = 5). The prefix drR refers to the d-rR/Tokyo medaka line used as the genetic background for the CRISPR/Cas9 experiment (unedited siblings).
• dUpPlxnb3_1 – dUpPlxnb3_5 — Homozygous ∆upPlxnb3 (-/-) mutant individuals (n = 5).

Image files. Each folder contains 2–8 JPEG (.jpg) images of intestinal villi captured on the dates 2019-07-26 and 2020-03-12. File names follow the pattern YYYYMMDD_NNNN.jpg, where YYYYMMDD is the imaging date and NNNN is the camera-assigned sequential image number. A few file names include + or - between numbers (e.g., 20200312_2858+2859.jpg, 20200312_2889-2893.jpg); these denote composite / stitched images assembled from adjacent captures on the same specimen.

File: SNPsRADseq.vcf

Single-nucleotide polymorphism (SNP) genotypes obtained by RAD-seq and used in Fig. 4 (association scan) and Fig. 1C / Fig. 5 (phylogeny and molecular-evolution analyses). The file conforms to the Variant Call Format v4.2 specification (https://samtools.github.io/hts-specs/). Genotypes are imputed and phased with Beagle v4.1 (beagle.21Jan17.6cc.jar); this is reflected in the phased-genotype separator (|) in the GT field and the imputation-quality metrics in the INFO field (see below). The distributed file contains 63,265 SNPs across 24 autosomes (chromosomes 1–24).

Reference genome. Reads were aligned to the PacBio-assembled medaka reference Medaka-Hd-rR-pacbio_version2.2.4.fasta (http://utgenome.org/medaka_v2/#!Assembly.md).

Standard VCF fields.

• CHROM — Reference chromosome (autosomes 1–24 of the Hd-rR PacBio assembly v2.2.4).
• POS — 1-based position on the reference chromosome.
• ID — Variant identifier formatted as {LocusID}_{SNP_position_within_locus} as output by the Stacks populations module (e.g., 12700_17).
• REF — Reference allele.
• ALT — Alternative allele(s).
• QUAL — Phred-scaled variant quality score. Encoded as . (missing) because values are not recomputed after imputation.
• FILTER — Filter status. All retained sites are marked PASS.
• INFO — Variant-level annotations:
  • AF — Estimated ALT allele frequency (Float).
  • AR2 — Allelic R-squared: estimated squared correlation between the most probable REF dose and the true REF dose (Beagle imputation-accuracy metric; Float).
  • DR2 — Dosage R-squared: estimated squared correlation between the estimated REF dose [P(RA) + 2·P(RR)] and the true REF dose (Beagle imputation-accuracy metric; Float).
  • IMP — Flag indicating that the marker is imputed.
• FORMAT — Per-sample genotype specification:
  • GT — Phased genotype (e.g., 0|0, 0|1, 1|1).
  • DS — Estimated ALT dose [P(RA) + P(AA)] (Float, 0–2).
  • GP — Estimated genotype probabilities for the three possible genotypes (Float triple).

Sample columns. One column per individual. Sample IDs are formatted as {RegionPrefix}_{PopulationCode}{IndividualNumber} (e.g., EKOR_YC2 = individual YC2 from the Yongcheon population within the KOR genetic group). The PopulationCode portion corresponds to the ID prefix used in WildDerivedLabMedaka.csv (e.g., YC = Yongcheon). Population-to-subgroup (NJPN1/2, SJPN1–4, KOR) assignments can be looked up via PopulationLocations_forAnnualTemperature.csv or WildDerivedLabMedaka.csv.

Note on regional groupings used during preprocessing. The distributed file was produced by merging seven per-region VCFs with bcftools merge v1.2: 1_KOR, 2_Kyushu, 3_Sanin, 4_Sanyo-Shikoku, 5_Chubu-Kanto-KitaNihon, 6_Hybrid, and 7_NJPN. These regional labels are coarser geographic pre-groupings used only for the imputation workflow and are distinct from the finer-grained genetic subgroups (NJPN1/2, SJPN1–4, KOR) reported in the manuscript and the CSV files.

SNP calling and processing pipeline (exact parameters as reported in the manuscript):

1. Single-end reads (51 bp, HiSeq 2500) quality-trimmed with Cutadapt v1.12 (-m 50 -e 0.2).
2. Demultiplexed with Stacks process_radtags v1.44 (-c -r -t 44 -q -s 0 --barcode_dist_1 2).
3. Aligned to the reference with BWA aln/samse v0.7.15-r1140 (-n 0.06 -k 3); multimapped reads filtered with SAMtools v1.9 (-q 1 -F 4 -F 256 -F 2048).
4. SNPs called with the Stacks pipeline: pstacks -m 2 --model_type snp --alpha 0.05; cstacks -g -n 1; sstacks -g; rxstacks --lnl_filter --lnl_lim -10 --conf_filter --conf_lim 0.75 --prune_haplo --model_type bounded --bound_low 0 --bound_high 0.1; cstacks -g -n 1; sstacks -g; populations -r 0.5 -p 7 -m 6 -f p_value -a 0.0 --p_value_cutoff 0.1 --lnl_lim -10.
5. Genotype imputation and haplotype phasing performed per regional subset with Beagle v4.1 (beagle.21Jan17.6cc.jar; GT format).
6. The seven imputed/phased regional VCFs merged with bcftools merge v1.2 (htslib v1.2.1) / VCFtools into the distributed file (63,265 SNPs).

For downstream association analyses (Fig. 4A), a minor-allele-frequency filter (MAF > 0.05) was applied within the analyses, yielding 22,235 SNPs; this filter is not pre-applied to the distributed file, so all 63,265 SNPs are available.

Sharing / Access information

Links to other publicly accessible locations of the data associated with this study:

• Manuscript: Katsumura et al. (2026) PNAS 123(13), e2534817123. https://doi.org/10.1073/pnas.2534817123
• Raw sequencing data (DDBJ Sequence Read Archive):
  • DRA010581 — MBD-seq reads.
  • DRA010605 — RAD-seq reads (input to SNPsRADseq.vcf).
  • DRA015161, DRA015162 — tBS / WGBS reads.
  • DRA017652 — additional sequencing data.
• Mitochondrial DNA D-loop sequences (DDBJ): accession LC719344–LC719462.
• Wild-derived laboratory stocks (living biological material): available from NBRP Medaka (https://shigen.nig.ac.jp/medaka/).

Data were generated by the authors. Annual mean air temperature values used in the manuscript (Fig. 5C) were obtained from the Global Solar Atlas (https://globalsolaratlas.info; CC BY 4.0 license) and are not redistributed in this repository.

Code / Software

Analyses reported in the manuscript were conducted with the following software. Versions are as reported in the manuscript; only the tools directly relevant to reproducing analyses of the files in this repository are listed.

• R v3.5.2 (initial analyses) / v4.1.1 (later analyses). Key packages: brms (Bayesian GLMM / GLM via MCMC), lmerTest (tBS analysis), NSM3 (Steel–Dwass test), SNPRelate (PCA), ggplot2 (visualisation).
• ImageJ (Schneider et al. 2012, Nat. Methods 9: 671–675) — morphometric measurements on photographs and quantification of IHC staining.
• Cutadapt v1.12 — read trimming.
• Stacks v1.44 (process_radtags, pstacks, cstacks, sstacks, rxstacks, populations) — RAD-seq locus assembly and SNP calling.
• BWA v0.7.15-r1140 — read alignment.
• SAMtools v1.9 — BAM filtering.
• Beagle v4.1 (beagle.21Jan17.6cc.jar) — genotype imputation and haplotype phasing.
• bcftools v1.2 (htslib v1.2.1) — VCF merging.
• VCFtools — VCF handling.
• PLINK v1.90b3.46 — FST calculation and association testing with permutation.

Analysis scripts are not deposited in this repository.

Usage notes

• Microsoft Excel, R, Python (pandas), or any text editor can be used to open the CSV files.
• bcftools, VCFtools, or R (VariantAnnotation) are recommended for handling SNPsRADseq.vcf.
• Image files inside Gut_IHC_images.zip can be viewed with Fiji/ImageJ or any standard image viewer.