THWOTH wrote:Guy's got a good beard. I trust him.

The_Piper wrote:... I'm having trouble with where these behavioral changes are coming from in the future generations.

Spearthrower wrote:

The_Piper wrote:... I'm having trouble with where these behavioral changes are coming from in the future generations.

Not sure what you're referring to specifically, and it's just as likely I wouldn't have the answer, but I do think that an interesting example to consider is the flounder.

In its larval stage, it's an entirely typical bilaterally symmetrical open water swimmer, then as it enters adulthood, its body begins a bizarre process involving shifting its skeletal anatomy about as an eye migrates to the other side of its face, loses pigmentation on the soon-to-be underside of its body, and swims inefficiently at an intermediate angle for some time before settling down to become a bottom-feeder effectively laying on its side.

Even in the course of its lifetime, the flounder experiences behavioral changes that result from anatomical drivers, and what's particularly interesting is that experiments suggest that the flounder's brain is already wired up to be a flatfish from birth, so its behavioral development over the course of just one life is odder than can even be imagined.

My sense is that behavior - in the abstract - seems to us to be coordinated and natural, yet is in fact anything but so well determined. All organisms undergo dramatic behavioral development at an early age when their brains are most plastic and able to adapt. A puppy that loses a couple of legs can often adapt to running on two, for example, so the 'later behaviors' you mention aren't so much coming from somewhere past, as coming from what's there at that moment.

Kon et al, 2020 wrote:Although domesticated goldfish strains exhibit highly diversified phenotypes in morphology, the genetic basis underlying these phenotypes is poorly understood. Here, based on analysis of transposable elements in the allotetraploid goldfish genome, we found that its two subgenomes have evolved asymmetrically since a whole-genome duplication event in the ancestor of goldfish and common carp. We conducted whole-genome sequencing of 27 domesticated goldfish strains and wild goldfish. We identified more than 60 million genetic variations and established a population genetic structure of major goldfish strains. Genome-wide association studies and analysis of strain-specific variants revealed genetic loci associated with several goldfish phenotypes, including dorsal fin loss, long-tail, telescope-eye, albinism, and heart-shaped tail. Our results suggest that accumulated mutations in the asymmetrically evolved subgenomes led to generation of diverse phenotypes in the goldfish domestication history. This study is a key resource for understandingthe genetic basis of phenotypic diversity among goldfish strains.

Kon et al, 2020 wrote:Goldfish (Carassius auratus), which belong to the family Cyprinidae, are a species that is closely related to the silver crucian carp; it was domesticated from wild goldfish during the Chinese Song dynasty (960–1279) [1]. Mitochondrial DNA analysis suggests that the domesticated goldfish was derived from a Chinese lineage of wild goldfish distinct from other C. auratus sublineages [2–4]. During the 1,000-year breeding history of goldfish strains, a wide variety of coloration and body, fin, eye, hood, and scale morphologies of the strains were established, mainly in East Asia. This wide variety of goldfish strains has long fascinated many researchers [1, 5, 6]. Charles Darwin was also interested in goldfish phenotypes and described the morphological features observed in goldfish strains [7]. Currently, at least 180 variants and 70 genetically established strains are produced and maintained all over the world [8]. Many goldfish strains display characteristic phenotypes in hoods (epidermal thickening around the head), narial bouquets (hypertrophy of nasal septa), and caudal fins that have not been observed among mutant strains in other teleost models, including zebrafish and medaka [9], which indicates the unique mechanisms generating diverse phenotypes in goldfish. Furthermore, several goldfish strains display phenotypes similar to those of human diseases, including congenital glaucoma, skeletal abnormalities, and albinism [9]. Therefore, analyzing goldfish strains may lead to elucidation of the molecular mechanisms underlying vertebrate morphogenesis and certain human diseases.

Whole-genome duplication (WGD) is a doubling of the entire genome during the evolution of certain lineages and is hypothesized to provide a large amount of raw material for diversification and evolutionary innovations after generation of ohnologs (a pair of genes originating from WGD) [10]. Vertebrates are thought to have experienced two rounds (1R and 2R) of WGD approximately 530–560 million years ago [11, 12]. In addition, a teleost-specific third-round WGD (Ts3R) occurred in a common ancestor of teleost fish approximately 320–350 million years ago [13–15]. The common ancestor of goldfish and common carp underwent a fourth round of WGD (Cs4R, carp-specific WGD) approximately 8–14 million years ago [9, 16–19]. Cs4R was suggested to be an allotetraploidization event, which is doubling of a complete set of chromosomes following interspecific hybridization of diploid progenitors (2n = 50) [20]. It has been reported, for certain allopolyploid plant species, that the two subgenomes have evolved asymmetrically after the WGD event [21, 22]. One of the two subgenomes is often preserved to stay more similar to the ancestral state, whereas the other experiences more chromosomal rearrangement, gene loss, and changes in levels of gene expression [23, 24]. Asymmetric subgenome evolution contributes to the phenotypic diversity of domesticated plants [21, 22, 25, 26]. In vertebrates, asymmetric subgenome evolution has also been observed in the genome of Xenopus laevis [23] and common carp (Cyprinus carpio) [27]. However, whether asymmetric subgenome evolution can contribute to genetic and phenotypic diversity in allopolyploid vertebrates remains unclear. Domesticated goldfish strains are an ideal model to clarify this issue because of their wide phenotypic diversity in morphology, revealing the phenotypic potential of an allopolyploid vertebrate under domestication.

Kon et al, 2020 wrote:In the current study, we named duplicated genes after Cs4R ‘‘gene L’’ or ‘‘gene S’’ according to the localized subgenome; in addition, we distinguished genes that were duplicated after Ts3R as ‘‘gene a’’ and ‘‘gene b.’’ For example, we named the four paralogs of kcnk5 in the goldfish genome kcnk5aL, kcnk5aS, kcnk5bL, and kcnk5bS.

Kon et al, 2020 wrote:Using these data, we analyzed the population structure of the goldfish strains (Figures 2A–2D, S3Z, and S4A–S4E). In the admixture analysis, we found that the cross-validation error was lowest at the number of ancestries (K) = 3 (red, green, and blue; Figures 2B and S3Z). Next we classified the 48 goldfish individuals into three groups according to the proportion of the individual’s genome from inferred ancestral populations at K = 3. We named these groups to reflect the history of goldfish breeding: China (blue), Ranchu (green), and Edo (red) (Figures 2A, 2B, and S3Z’; Data S1C). Consistent with the admixture analysis, principal-component analysis (PCA) also separated the individuals into three groups (Figures S4A–S4C). Furthermore, the maximum-likelihood tree and the neighbor-joining tree of 48 individuals also supported this grouping (Figures 2C, S4D, and S4E). To search the regions of reduced heterogeneity and potential fixation in different lineages, we calculated fixation index (Fst) values in 40-kb windows sliding 10 kb at a time for the China, Ranchu, and Edo groups. We identified 1,069 regions (total 90 Mb), 1,128 regions (total 110 Mbp), and 2,096 regions (total 236 Mb) with high Fst values for the China, Ranchu, and Edo groups (Data S1F). For example, the regions from 7.1–8.3 Mb on LG31, 12.1–12.2 Mb on LG15, and 15.3–16.8 Mb on LG19 showed high Fst values, respectively (Figures S4F–S4H). These regions are considered to be areas of reduced heterogeneity and potential fixation among strains in each group. It is possible that these include regions of reduced heterogeneity because of a domestication bottleneck that was also observed in other domesticated animals [29].

The genomic regions with a high degree of fixation in goldfish strains may include genes that were positively selected during the history of goldfish domestication [30, 31]. To identify such loci, we calculated the Z-transformed pooled heterogeneity (ZHp) in 40-kb windows sliding 10 kb at a time (Figures S4I and S4J). In this analysis, we identified reduced heterogeneity regions containing 2,020 genes (Figure S4K; Data S1G). The Gene Ontology (GO) analysis identified significantly enriched GO terms for this gene set (Data S1H). Notably, the most enriched GO term was ‘‘negative regulation of signal transduction’’ (GO: 0009968, 2.36-fold enrichment, false discovery rate [FDR] value = 0.00248). We also identified the highly enriched GO term ‘‘embryonic organ development’’ (GO: 0048568, 1.95-fold enrichment, FDR value = 0.00829) for this gene group. These enrichments of GO terms in lower ZHp regions may have undergone purifying selection for developmental genes because of the genes’ key function in development.

Kon et al, 2020 wrote:The Twin-Tail Phenotype

Many goldfish strains, including Ryukin, Oranda, and Telescope-eye, possess a bifurcated caudal axial skeleton, presenting as the twin-tail phenotype (Figures 2D; Data S1B). A mutation in one of the chordin paralogs (chdS) has been reported to cause the twin-tail phenotype in goldfish [32]. However, whether the other chordin paralog (chdL), an ohnolog, contributes to the twin-tail phenotype remains unknown. To test whether our GWAS detected mutations in chdL and/or chdS, we conducted GWAS using the data of 21 twin-tail strains as cases and seven single-tail strains and wild goldfish as controls. Our GWAS of the twin-tail phenotypes revealed the highest association between the genomic variants on LG40 at base positions 19,923,396 and 19,923,744 and the twin-tail phenotype (p = 6.583 × 10-15) (Figures S5A–S5C). Combined with a synteny analysis using zebrafish and medaka genome sequences, we identified the chdS gene in this region. We observed no significant association between the locus chdL on LG15 and the twin-tail phenotype, indicating that chdL does not significantly contribute to the twin-tail phenotype, at least in the strains tested. This analysis showed that the mutation in chordin of twin-tail goldfish is on the non-dominant S subgenome and that whole-genome sequencing and GWAS analysis based on the subgenome structures are effective for identifying loci responsible for phenotypes in goldfish strains.

Kon et al, 2020 wrote:The Dorsal Fin Loss Phenotype

Several goldfish strains, including the Ranchu group, exhibit a dorsal fin loss phenotype (Figures 2D and 3A–3C). The mode of inheritance of this phenotype is most likely recessive [33]. To investigate the detailed skeletal structure of the dorsal fin, we performed Alcian blue and alizarin red staining of common goldfish with a normal dorsal fin and of Albino Ranchu goldfish, which displays the dorsal fin loss phenotype. Common goldfish (singletailed Red Wakin strain) retain the body form and fins of the wild goldfish; however, it has red coloration. We used common goldfish as the wild type for comparing phenotypes with other strains. We observed complete loss of dorsal fin rays and endoskeletal elements (distal and proximal radials) in Albino Ranchu goldfish (Figure 3D). We also examined the dorsal fin fold formation at the larval stage and found that most of the dorsal fin fold in Albino Ranchu goldfish was lost 3 days post fertilization (dpf) (Figure 3E).

To identify the loci associated with the dorsal fin loss phenotype, we conducted GWAS using eight strains, with the dorsal fin loss phenotype as cases and the other 19 strains and wild goldfish as controls. Our GWAS identified the highest association between a genomic variant at LG29 position 6,731,343 and the dorsal fin loss phenotype (p = 4.893e × 10-15) (Figure 3F). A 26-kb homozygous haplotype (6,728,745–6,755,075) containing three genes in all eight goldfish strains with the dorsal fin loss phenotype was identified around this position (Data S1J). Within this 26-kb region, we identified an Wnt co-receptor Lrp6S as a strong candidate responsible for the dorsal fin loss phenotype because Wnt signaling is essential for fin formation as well as fin regeneration in teleost fish [34–36]. It has also been reported that partial knockdown of lrp6 in Xenopus embryos affects formation of the fin fold at the larval stage [37]. We searched for mutations in lrp6S of goldfish with the dorsal fin loss phenotype and identified four amino acid substitutions; however, all of them were conserved or semi-conserved in zebrafish or medaka lrp6 orthologs (Figure S5D), suggesting that these lrp6S missense mutations were unlikely to be causative mutations of the dorsal fin loss phenotype. The fin fold of goldfish embryos starts forming at the 25% otic vesicle closure (OVC) stage [38]. We analyzed the expression level of lrp6S at the 25% OVC stage and discovered that lrp6S expression notably decreased in Albino Ranchu embryos (Figure 3G), suggesting that a mutation in the regulatory elements of lrp6S may cause the dorsal fin loss phenotype in goldfish. We identified a 313-bp deletion in intron 21 of lrp6S in goldfish with the dorsal fin loss phenotype (6,729,459–6,729,771) (Figure S5E). This intronic deletion might affect induction and proper production of lrp6S mRNA at embryonic stages.

Because a goldfish genome editing method has not yet been established, it is technically difficult to observe the lrp6 loss-offunction phenotype in goldfish. We used zebrafish, which belong to the carp family, to observe the effect of loss of function of lrp6. To investigate whether partial loss of lrp6 function affects fin formation in teleost fish, we performed CRISPR-Cas9-mediated gene disruption of lrp6 in zebrafish. We identified a partial defect of dorsal fin fold formation in 4 dpf zebrafish larvae injected with lrp6 guide RNA (gRNA) and Cas9 (Figures 3H and S5F). The ratio of unaffected dorsal fin fold length to body length was significantly reduced in these embryos (Figure 3I). Lrp6 knockout in mice or knockdown in zebrafish or Xenopus causes severe embryonic developmental defects [39–41]. The remaining lrp6S expression and/or expression of the ohnolog lrp6L seem to prevent lethality in goldfish with the dorsal fin loss phenotype. Mosaic biallelic gene disruption by CRISPR-Cas9-mediated genome editing occurs in F0 zebrafish larvae [42]. Mosaic gene disruption in zebrafish larvae is most likely to prevent lethality although knockdown of lrp6 causes severe embryonic defects in zebrafish [41].

To analyze whether partial inhibition of the Wnt pathway mediated by Lrp6 affects dorsal fin formation in adult teleost fish, we generated zebrafish that ectopically expressed the Wnt inhibitor Dkk1 after the dorsal fin fold formation stage. Dkk1 negatively regulates Lrp6-mediated Wnt signaling via direct interaction with the Wnt co-receptor Lrp6 in vertebrates [43–45]. We observed that zebrafish larvae of the Gal4 driver line expressed enhanced green fluorescent protein (EGFP) in the dorsal part of the trunk 72 h post fertilization (hpf) (Figures 3J, S5G, and S5H). We found that larvae expressing dkk1 under control of an upstream activating sequence (UAS) lost the dorsal part of the fin fold (Figures S5G and S5H), which was similar to the larval phenotype of Albino Ranchu goldfish (Figure 3E). Notably, at the adult stage, we found that the dorsal fin was lost in these lines (Figure 3K). These results suggest that the Lrp6-mediated Wnt pathway regulates dorsal fin formation in adult zebrafish. These results support the hypothesis that the causative mutation for the phenotype of dorsal fin loss is around the lrp6S gene locus on the S subgenome and that decreased lrp6S expression affects adult dorsal fin formation in goldfish.

Kon et al, 2020 wrote:The Long-Tail Phenotype

Goldfish strains with the long-tail phenotype exhibit notably elongated caudal fin lobes in single- and twin-tailed goldfish strains (Figures 4A–4E). Goldfish with the long-tail phenotype usually show elongation of all median and paired fins, suggesting that this mutation modulates the length of all fin types. The long-tail phenotype in goldfish is most likely to be dominant [33]. A GWAS was conducted using a dataset of 14 strains with the long-tail phenotype as cases and 13 strains without the long-tail phenotype and wild goldfish as controls. This analysis identified a notable association between a locus on LG45 at base position 15,015,973 and the long-tail phenotype (p = 3.358 × 10-15) (Figures S5I–S5K). In the vicinity of this position, all 14 tested strains with the long-tail phenotype carried at least one copy of a 46-kb long haplotype spanning four genes (14,942,902–14,988,949; Data S1K). In this region, we identified kcnk5bS as a strong candidate gene because mutations in its zebrafish ortholog (another longfin [alf]) have been reported to cause a proportionally larger caudal fin phenotype similar to the goldfish long-tail phenotype [46]. KCNK5 is a member of the two-pore domain potassium channel family that produces background (leak) K+ currents over a large range of membrane potentials [47, 48]. alf mutations increase the K+ conductance of the channel to cause hyperpolarization of cells, although it is unknown how the change in K+ channel conductance regulates fin ray length [46]. Our sequence analysis identified five amino acid substitutions or deletions in kcnk5bS in goldfish with the long-tail phenotype (Figure S6A). Interestingly, one of these mutations, V165E, was in the vicinity of the mutated amino acid W169L, reported in the zebrafish alf mutant (Figure 4F). To test whether mutations in goldfish kcnk5bS cause a K+ conductance change, we performed voltage-clamp recordings using Xenopus oocytes injected with cRNA of wildtype and mutant goldfish kcnk5bS. K+ conductance in oocytes injected with mutant kcnk5bS cRNA significantly increased compared with oocytes injected with wild-type kcnk5bS cRNA (Figures 4G and 4H), suggesting that the goldfish mutant kcnk5bS causes significant cell hyperpolarization. Homology modeling of the kcnk5bS structure revealed that the V165E mutation in the goldfish kcnk5bS was localized in the M3 transmembrane domain (Figures 4I, S6B, and S6C). Substitution of a hydrophobic amino acid (valine) with a hydrophilic amino acid (glutamic acid) possibly caused a critical alteration of channel gating, supporting the hypothesis that the kcnk5bS mutation on the S subgenome is a causal mutation of the goldfish long fin phenotype.

Kon et al, 2020 wrote:The Telescope-Eye Phenotype

Certain goldfish strains possess enlarged protuberant eyes, known as the telescope-eye phenotype (Figures 5A–5D). A previous genetic study of Telescope-eye goldfish identified this phenotype as recessive [33]. However, the gene locus or mutations causing the telescope-eye phenotype are unknown. To identify the locus associated with the telescope-eye phenotype in goldfish, we conducted a GWAS of goldfish strains using datasets of four Telescope-eye strains as cases and the other 23 strains and wild goldfish with normal eyes as controls. Our GWAS analysis revealed a mild association (p = 1.29 × 10-7) between a genomic variant on LG9 at base position 5,399,083 and the telescope-eye phenotype (Figures S6D and S6E). In a region relatively close to this locus, we identified lrp2aL as a candidate gene for the telescope-eye phenotype because Lrp2a mutations have been shown to cause an enlarged eye phenotype in adult zebrafish and mice that is similar to the goldfish telescope-eye phenotype [49, 50]. In mouse models, LRP2 plays the role of sonic hedgehog (shh) clearance receptor and regulates shh-induced cell proliferation at the retinal margin, resulting in the large eye phenotype [50]. A detailed sequence analysis of lrp2aL variants in strains with the telescope-eye phenotype identified two nonsense mutations in lrp2aL genes that were only present in the genome of goldfish with the telescope-eye phenotype and absent in other goldfish (Figures 5E and S6F). We also performed RNA sequencing (RNA-seq) analysis to compare the gene expression profile in the eyes of the common goldfish with that of the Black Telescope-eye goldfish. Intriguingly, we detected a notable lrp2aL transcript abnormality in the Black Telescope-eye strain, where the lrp2aL mRNA lacked half of its 3ʹ portion (Figures S6G and S6H). The breakpoint was located between exon 45 and exon 46. Analysis of this region revealed a 13-k base pairs (bp) insertion in intron 45 in the Black Telescope-eye strain, but this was absent in common goldfish. Nucleotide sequencing analysis revealed that the insertion encoded a TE that is a type of foamy-like endogenous retrovirus (Figures 5E and S6I–S6K). The de novo transcriptome assembly of the RNA-seq data identified that an aberrant lrp2aL mRNA expressed in the Black Telescope-eye strain possessed a premature stop codon (Figures S6G and S6H). Almost no normal lrp2aL mRNA was expressed in the Black Telescope-eye strain (<0.1% of that found in common goldfish), suggesting that the 13-kbp retrotransposon insertion was responsible for the loss of lrp2aL function in Telescope-eye goldfish. Overall, we identified three types of loss-of-function mutations in the lrp2aL gene in strains with the telescope-eye phenotype. These results support the hypothesis that loss-of-function mutation in lrp2aL on the L subgenome causes the telescopeeye phenotype in goldfish.

Kon et al, 2020 wrote:Albinism

Many goldfish strains with xanthic bodies lack melanophores in the skin and scales but retain black retinal pigment epithelia in the retina. This is because the pigment cells in the body surface are derived from neural crest cells, whereas retinal pigment epithelia originate from optic lobe neuroepithelial cells [51]. These goldfish have black pupils that reflect the black retinal pigment epithelia. However, in albino goldfish, the melanin loss occurs on the body surface and in the retina, resulting in pink pupils because of depigmentation of the retinal pigment epithelia (Figures 6A, 6B, S7B, and S7C) and loss of dark body coloration in juveniles.

Although a previous genetic study reported that albinism in goldfish is double recessive for two independently assorting autosomal loci, m/m and s/s [52], the mutated genes are unknown. To identify mutations associated with the albino phenotype, we searched the 172 genes with SSVs in Albino Celestial goldfish (Data S1L) for any overlap with the 143 genes reported as being related to body color in vertebrates (Data S1N) and found that only oca2 overlapped between these two groups. Interestingly, the critical mutations identified in both oca2 ohnologs (oca2L and oca2S) of Albino Celestial goldfish had frameshift indels, resulting in production of truncated proteins comprising 470 and 519 amino acids, respectively (Figures 6C, S7D, and S7E). We observed that none of the 26 non-albino strains possessed the frameshift mutation in oca2L or the homozygous frameshift mutation in oca2S. We also analyzed mutations of oca2 ohnologs in the Albino Telescope-eye, Albino Azuma-nishiki, Albino Comet, Albino Ranchu, and Albino Oranda. We found that all of these strains possessed the same homozygous mutations in both oca2 ohnologs, similar to Albino Celestial goldfish. Because oca2 encodes an anion transporter regulating the pH of the melanosome, and because loss-of-function mutations of oca2 in several species, including humans, mice, and zebrafish, are known to lead to albinism [53–55], these mutations in oca2 ohnologs on the L and S subgenomes are most likely to be the causative mutations of albinism in goldfish strains.

Kon et al, 2020 wrote:The Heart-Shaped Tail Phenotype

The heart-shaped caudal fin phenotype was only displayed in Bristol Shubunkin goldfish among the 27 strains analyzed in this study. To clarify the heart-shaped tail phenotype, we analyzed the fin ray length of the caudal fins (Figure S7F). Adult common goldfish have 10 and 9 principal caudal fin rays in upper and lower lobes, respectively [56] (Figure S7F). We measured the

length of the tenth caudal fin ray in the dorsal side of the caudal fin cleft (10 dfr), the second caudal fin ray in the dorsal fin lobe (2 dfr), and the sixth dorsal caudal fin ray (6 dfr). In most cases, 2 dfr was the longest ray in the caudal fin of the dorsal lobe. We observed that the ratio of 2 dfr length/body length in Shubunkin and Bristol Shubunkin goldfish was significantly larger compared with common goldfish (Figure S7G) (p = 0.0029, control n = 6; Shubunkin n = 4, two-sided Welch’s t test; p = 1.9 × 10-6, common goldfish n = 6, Bristol Shubunkin n = 4, two-sided Student’s t test). Notably, the ratio of 6 dfr/2 dfr in Bristol Shubunkin goldfish was significantly higher than that of Shubunkin goldfish (p = 0.0077; Shubunkin, n = 4; Bristol Shubunkin, n= 4; two-sided Student’s t test), suggesting that regulation of the caudal fin ray length proportion was affected in Bristol Shubunkin goldfish.

We identified 1,466 genes with SSVs for Bristol Shubunkin goldfish (Data S1L). To identify the candidate genes responsible for the heart-shaped tail phenotype, we searched for overlapping genes between genes reported previously to regulate fin shape and genes with SSVs in the Bristol Shubunkin strain. We found that the Bristol Shubunkin SSVs contained the rpz gene cluster. In zebrafish, a teleost-specific transmembrane protein, Rpz, regulates caudal fin length [57], although its molecular mechanism is still unelucidated. The zebrafish rpz gene cluster is reported to contain five rpz-like genes [57]. Our detailed analysis revealed seven zebrafish rpz-like genes in this cluster. In goldfish, we found two rpz gene clusters on LG16 (L subgenome) and LG41 (S subgenome). Each of them contains seven rpz-like genes similar to those in zebrafish. We observed that rpzS and rpz4S were relatively highly expressed in the caudal fin (Data S1O). We found a Bristol Shubunkin-specific missense mutation (p.Asp124Glu) in rpzS and two (Tyr133Phe and Phe319Ser) in rpz4S. Further analysis is needed to clarify whether the rpz gene cluster in the S subgenome contributes to the heart-shaped tail phenotype. Interestingly, a previous proteomics analysis reported asymmetric distribution of Rpz and Rpz5 in zebrafish caudal fins; these proteins are significantly enriched in the proximal region of the caudal fin in zebrafish [51]. A calcineurin-mediated mechanism is hypothesized to operate as a molecular switch between the position- associated isometric and allometric growth program in caudal fins [58]. Thus, the asymmetric distribution of Rpz-like proteins may contribute to fin shape formation by regulating the balance between the isometric and allometric growth programs. Our results support the hypothesis that mutations in the rpz gene cluster on the S subgenome affect the fin ray length proportion in caudal fins and contribute to the heart-shaped tail phenotype in Bristol Shubunkin goldfish.

Calilasseia wrote:
The first surprise the paper has in wait for those not aware of this fact, is that domesticated Goldfish are tetraploid organisms. The diploid ancestors produced offspring that had two copies of the genome, which then differentiated so that the tetraploid genome is asymmetric.