A novel allele of the goldfish chdB gene: Functional evaluation and evolutionary considerations

Abstract The twin tail of ornamental goldfish is known to be caused by a nonsense mutation in one chordin paralogue gene. Our previous molecular studies in goldfish revealed that the ancestral chordin gene was duplicated, creating the chdA and chdB genes, and the subsequent introduction of a stop codon allele in the chdA gene ( chdA E127X) caused the twin‐tail morphology. The chdA E127X allele was positively selected by breeders, and the allele was genetically fixed in the ornamental twin‐tail goldfish population. However, little is known about the evolutionary history of the chdB paralogue, begging the question: are there the functionally distinct alleles at the chdB locus, and if so, how did they evolve? To address these questions, we conducted molecular sequencing of the chdB gene from five different goldfish strains and discovered two alleles at the chdB gene locus; the two alleles are designated chdB 1 and chdB 2. The chdB 1 allele is the major allele and was found in all investigated goldfish strains, whereas the chdB 2 allele is minor, having only been found in one twin‐tail strain. Genetic analyses further suggested that these two alleles are functionally different with regard to survivability ( chdB 1 >  chdB 2). These results led us to presume that in contrast to the chdA locus, the chdB locus has tended to be eliminated from the population. We also discuss how the chdB 2 allele was retained in the goldfish population, despite its disadvantageous function. This study provides empirical evidence of the long‐term retention of a disadvantageous allele under domesticated conditions.

two paralogues allowed for stable genetic fixation of the phenotype. These paralogues are derived from a lineage-specific allotetraploidization (the genome duplication of a species hybrid) in the common ancestor of the goldfish and common carp (Cyprinus caripio; Luo et al., 2006;Xu et al., 2014), however, the chdA E127X allele has only been found in the goldfish lineage. The chdA E127X allele lacks three of four functionally significant cysteinerich domains, and consequently, its function is highly compromised in comparison with the wild-type chdA allele (chdA wt ). This reduction of function causes a ventralized embryonic phenotype, bilaterally shifted caudal fin primordia, and twin-tail morphology in the adult (Abe et al., 2014;Abe & Ota 2017). The twin-tail goldfish was documented in Chinese archives around the 1600s common era (CE), and the origin of goldfish breeding for ornamental purposes dates back to the Song dynasty (around 1000 CE; Chen, 1956;Smartt, 2001); thus, it is expected that the chdA E127X allele was selected for and genetically fixed over the course of 600 years, after which it was maintained in the twin-tail goldfish population. In other words, the chdA locus has exhibited a drastic change in the mutated allele frequency under domesticated conditions.
In contrast to the chdA locus, the chdB locus probably did not experience such a drastic change in allele frequency during the domestication of twin-tail goldfish, based on its expression patterns and expected function (Abe et al., 2014). The partially overlapping expression patterns of chdA and chdB genes in embryos suggests that compensation by the chdB gene may prevent overreduction of dorsal tissue and increase the survival rate of twin-tail goldfish without a chdA wt allele. Thus, if the loss-of-function mutations were to occur at the chdB locus in the twin-tail goldfish, the combined absence of functional chdA and chdB genes would be expected to lead to high lethality, similar to the reported phenotypes of dino/chordin zebrafish and chordin UT600 medaka (Fisher & Halpern, 1999;Takashima et al., 2007). Consequently, a chdB-mutated twin-tail goldfish would most likely be eliminated from the population (Fisher & Halpern, 1999;Oelgeschläger, Kuroda, Reversade, & De Robertis, 2003;Schulte-Merker et al., 1997;Takashima et al., 2007), leading us to presume that the chdB locus must be functionally conserved in ornamental twin-tail goldfish.
However, we still know little about the evolutionary processes acting on the chdB gene in goldfish, and we pose several questions, including whether the chdB gene has maintained its original function, whether any functionally different alleles exist at the chdB locus, and if such alleles exist in the goldfish population, how these alleles have become distributed during the domestication process. To address these questions, we probed for the presence/absence of functionally differentiated chdB alleles by examining the chdB locus in various goldfish strains and tested how allele variation influences the phenotype of ornamental goldfish. Based on the results of these experiments, we consider how these duplicated paralogues have evolved after the switch from natural to artificial selection.

| Goldfish strains
Five different goldfish strains, containing a total of nine subpopulations were used in experiments, including two different groups of the Butterfly tail and Heimutan strains, Orandanshishigashira (Oranda), Ryukin, and the single-tail common goldfish strains of Japan, Taiwan, and Mainland China. Butterfly tail and Heimutan strains were purchased from a local aquarium (Yu-Dian Corporation) in Yilan, Taiwan. Oranda and Ryukin strains were purchased from an aquarium fish breeder (SHUEN-SHIN Breeding Farm) in Toucheng, Taiwan. The single-tail common goldfish strains from Japan and Mainland China were imported by an aquarium in Taipei, Taiwan (Limpid Aqua/Aqua project Taiwan).

| Molecular cloning and phylogenetic analysis
Homologs of chdB genes were isolated from complementary DNA (cDNA) derived from embryos of the single-tail common goldfish and Oranda strain using polymerase chain reaction (PCR). Total RNA was extracted from gastrula-stage Oranda goldfish embryos using TRIzol Reagent (Ambion). Specific PCR primers were designed based on the sequence of previously isolated chdB genes (accession number: BAO51897). Amplified PCR fragments were isolated and purified and then ligated into a vector using the TOPO TA Cloning Kit, Dual Promoter (Invitrogen). The resulting vector was used to transform DH5α Escherichia coli. More than 10 clones were picked for sequencing. The sequences of the cDNA fragments were then used as backbones to obtain nearly complete sequences of the alleles by PCR with specific primers.
The isolated gene was identified by generating multiple amino acid alignments with known goldfish, orthologous and paralogous genes using CLUSTALW. The phylogenetic relationship of chdB 2 alleles and closely related chordin genes was reconstructed in a maximum likelihood tree using MEGA5. Branch lengths were estimated based on the nonsynonymous and synonymous substitutions using PAMLX (Xu & Yang, 2013 fertilization was performed according to methods detailed in our previous report (Tsai, Chang, Liu, Abe, & Ota, 2013

| Genotyping
PCR primers were designed to amplify a region containing four SNP sites and a Cla I restriction enzyme site. PCR fragments amplified by these specific primers were digested by Cla I, and separated on 2% agarose gels. Genotypes were determined on the basis of the resulting band patterns. The band patterns were confirmed by at least two independent trials of PCR and restriction enzyme digestion.

| Injection of messenger RNA
To generate constructs for transcription, the coding regions of chdB 1 or chdB 2 were amplified by PCR and cloned into the pCS2+vector (Rupp, Snider, & Weintraub, 1994). These constructs were first digested with NotI and then used as templates to synthesize capped messenger RNA (mRNA) with the mMESSAGE mMACHINE SP6 Kit, according to the manufacturer's instructions (Ambion). The synthesized mRNA transcripts were purified with Quick Spin Columns and resuspended in nuclease-free water. A microinjector (Eppendorf Femtojet; Eppendorf, Hamburg, Germany) was used to inject mRNA into the yolk of one to two cell-stage fertilized eggs in 4 nl of 0.2 M KCl. Phenol red (Sigma) was added as an indicator at a final concentration of 0.05%. In total, 100 pg of mRNA was injected into chdB 1/1 twin-tail goldfish embryos. The injected embryos were incubated at 24°C. Four independent rescue experiments were performed by injecting chdB 1 or chdB 2 mRNA into twin-tail goldfish embryos. Control embryo phenotypes were examined in all four experiments. All control embryos exhibited mutant phenotypes. Two days after injection, embryos were classified by morphological inspection into the following four categories: dorsalized, weakly ventralized, bifurcated fin fold, and severely ventralized. The categorization was based on our previous report (Abe et al., 2014).
To investigate the larval morphology of injected individuals, larvae were anesthetized with MS222 two days after injection and photographed live and/or after fixation with 4% paraformaldehyde (PFA). The fixed larvae were stained by alizarin red solution (0.02% alizarin red in 70% ethanol) and examined under a stereomicroscope (SZX16; Olympus, Tokyo, Japan).

| In situ hybridization
Digoxigenin-labeled antisense RNA probes were produced using PCR product templates and the T7 RNA polymerase Riboprobe Combination System (Promega), according to the manufacturer's instructions.
The probes were purified using mini Quick Spin RNA Columns (Roche, Germany). Primer sets for the PCR amplification of szlA and foxb1a fragments were from a previous report (Abe et al., 2014). PCR products corresponding to relevant portions of 5′-or 3′-untranslated regions were used to generate probes.
Whole-mount in situ hybridization was performed as previously described (Schulte-Merker, Ho, Herrmann, & Nusslein-Volhard, 1992) with minor modifications. Fish embryos were fixed with 4% paraformaldehyde in phosphate-buffered saline (PBS) overnight. Embryos were fixed and then dechorionated using fine forceps. After fixation and dechorionation, embryos were dehydrated with methanol. Dehydrated embryos were then rehydrated with phosphate-buffered saline, 0.1% Tween-20 (PBT) and fixed with 4% paraformaldehyde in PBS. Embryos were subsequently treated with proteinase K for 20 min, after which the samples were fixed again. Prehybridization and hybridization were performed at 65°C for a period that ranged between 1 hr and overnight.
The samples were washed sequentially two times with 50% formamide/ 2 × SSCT at 65°C for 30 min, 2 × SSCT at 65°C for 15 min, and two final washes with 0.2 × SSCT at 65°C for 30 min. The samples were then

| Morphological analyses of bifurcated anal and caudal fins
Goldfish larvae from Asb to Pr stage were anesthetized with MS222 (Sigma), and then fixed with 4% paraformaldehyde in PBS. After fixation, samples were washed in 70% ethanol, stained with alizarin red solution (0.02% alizarin red in 95% ethanol), and washed again in 70% ethanol to reduce background (Li, Chang, Liu, Abe, & Ota, 2015).
The number of the caudal fin rays were counted under a stereomicroscope (SZX16; Olympus). All goldfish specimens that were used for morphological analysis were genotyped at the chdB locus and categorized into three groups according to their allelic combination: chdB 1/1 , chdB 1/2 , and chdB 2/2 .

| Statistical analyses
All plotting and statistical tests in the morphological and genetic analyses were performed with the R statistical computing package of RStudio v0.98.1049.

| Molecular background of two chdB alleles in goldfish
Our molecular cloning of chdB genes revealed that two different alleles exist in the Oranda population. One allele was first found in our previous report (accession number: BAO51897), and the other is a newly discovered allele (LC382263); the two alleles were named implying that these two SNPs might influence to the function of chdB gene. In addition, one of four nonsynonymous sites in exon 11 is recognized by the ClaI restriction enzyme, allowing us to develop a ClaI-based genotyping method.
To investigate whether the chdB 1 allele is a major or minor allele in the ornamental goldfish population, we used PCR amplification and ClaI restriction enzyme digestion to genotype 102 goldfish, comprising five different strains and nine subpopulations (Figure 1a and Table 1). Among the genotyped fish, only those from the Oranda strain carried the chdB 2 allele ( Table 1). The allele frequency for chdB 2 was 0.214 in the Oranda strain and 0.044 across all investigated strains, suggesting that the chdB 2 allele is minor allele in the overall ornamental goldfish population.
3.2 | Functional analyses of allelic differences of chdB gene | 375 Takashima et al., 2007;De Robertis, 2006, 2009Langdon & Mullins, 2011). However, we could not detect any differences between chdB 1 and chdB 2 in the phenotypes of mRNA-injected embryos at this stage, presumably due to subtle or a lack of functional differences at the stages we examined.
Because we expected that differences between chdB 1and chdB 2 -mRNA-injected embryos may be most clearly detected in early-stage embryos, we probed the gene expression patterns of foxb1a (an embryonic neural marker expressed in dorsal cells) and szlA (an embryonic ventral marker) at two different gastrula stages (Figure 3).
On the other hand, the szlA gene showed markedly reduced expression in early-to-late gastrula-stage chdB 1and chdB 2 -mRNAinjected embryos compared with controls (Figure 3d-i). Importantly, the reduction in szlA gene expression was more severe in chdB 1 -mRNA-injected embryos than chdB 2 -mRNA-injected embryos at both early and late gastrula stages (Figure 3e,f,h,i).
In addition to visual inspection (Figure 3a-i), quantification of the szlA gene expression region was performed (Figure 3j,k). We measured the szlA gene expression area in chdB 1or chdB 2 -mRNAinjected goldfish embryos that were derived from two different clutches across multiple developmental stages, ranging from more than 20% blastopore closure to less than 80% blastopore closure ( Figure 3g). All of the chdB 1 -mRNA-injected embryos showed narrower szlA gene expression areas in comparison with controls ( Figure 3j,k), consistent with our previous report (Abe et al., 2014).
On the other hand, the chdB 2 -mRNA-injected embryos tended to exhibit expanded szlA gene expression areas in comparison with chdB 1 -mRNA-injected embryos, even though we injected the same amount of mRNA, and the twin-tail goldfish embryos for each experiment were derived from the same clutch (Figure 3j,k). Because the same trend was observed in multiple experiments after different orders of injection, the differences might reflect actual differences in the function of the two alleles (Figure 3j,k). These results suggested that chdB 1 and chdB 2 both function as dorsal organizers, however coding-region-localized SNPs may affect szlA gene expression patterns in early embryonic stages (Figures 1 and 3).

| Genotyping and morphological analyses in F2 segregants from chdB 1/2 parents
To further clarify whether chdB 1 and chdB 2 differentially influence the phenotypes of twin-tail goldfish, we crossed chdB 1/2 parents of the Oranda strain, and subsequently examined the genotypes of the progeny at the juvenile stage (Ar stage). At this stage, almost all of caudal fin rays are countable (Li et al., 2015). In total, 205 individuals were observed from three different crosses (Figure 4 and Table 2).
The numbers of progeny with chdB 1/1 , chdB 1/2 , and chdB 2/2 genotypes were 69, 101, and 35, respectively. Therefore, the ratio of genotypes Mendelian ratio (p < 0.01, the chi-square test; Table 2 and Figure 5a). Since all of the fish were kept under identical conditions, it is expected that the minor deviation from Mendelian ratios reflects viability differences between the genotypes. Thus, the allelic variants in chdB may influence phenotypic features related to lethality (e.g., early embryogenesis and/or physiological features). From the genotype frequency of Oranda F2 segregants, we also estimated the fitness of chdB genotypes relative to chdB 1/1 (Figure 5a). The relative fitness estimates for chdB 1/1 , chdB 1/2 , and chdB 2/2 were 1.000, 0.732, and 0.507, respectively, suggesting additive gene action (solid circles, Figure 5a).
On the basis of these relative fitness estimates, we calculated the theoretically expected change in the frequency of the chdB 2 ( Figure 5b). According to this analysis, the frequency of the chdB 2 allele is expected to be less than 0.001 at the 21st generation, even when the allele frequency of the initial population is 0.9 in a random mating model (Hartl & Clark, 1997;Hedrick, 2005). The domestication history of twin-tail goldfish spans more than 400 years, and twoto four-year-old goldfish are optimal for obtaining the next generation of offspring (Matsui, 1972). Thus, ornamental goldfish  Table S1); more than 90% of the segregants (192 individuals) had bifurcated caudal fins, which varied in their symmetricity ( Figure 6g,h,i). Most of the segregants had a highly symmetrical bifurcated caudal fin (Figure 6g), but a few of the fish exhibited an asymmetrically bifurcated caudal fin, with one lobe being larger than the other (Figure 6h,i). To distinguish whether these morphological variations were correlated with the chdB allele, we counted the number of caudal fin rays in the 205 individual segregants. Caudal fin rays were counted and categorized as "left-side bifurcated caudal fin rays" (cfrL), "right-side bifurcated caudal fin rays" (cfrR), and the "midline located nonbifurcated caudal fin rays" (cfrM; Figure 6f). The summations of cfrM and cfrL as well as cfrM and cfrR were calculated, and the larger sum was used as an index of the maximal number of caudal fin rays along the dorsal-ventral axis; this index is designated as the "maximum caudal fin number." The distribution of the maximum caudal fin number indices was divided into two phenotypic groups (Figure 7a). The F2 progenies of 33 individuals had an index less than 20, whereas 172 individuals had an index of more than 20 ( Figure 7a); we categorized those less than 20 as the "less caudal fin ray number" group and those more than 20 as the "normal caudal fin ray number" group ( Figure 7a).
The total numbers of fin rays in the normal caudal fin number group was almost equivalent with single-fin common goldfish at the equivalent stage (Li et al., 2015); single-fin common goldfish exhibit 30 caudal fin rays at the juvenile stage, whereas the F2 progeny in the normal caudal fin ray number group have an average of 29 fin rays (Figure 7a). After distinguishing the fish phenotypes according to this scheme, we could not detect any significant relationship between allelic combinations and the maximum caudal fin numbers (Supporting Information Table S2).
We also plotted left and right fin ray number of the 192 individuals with bifurcated caudal fins, as shown in Figure 7b.
Although there are variations, our analyses suggested that the

| DISCUSSION
Our molecular sequencing and developmental analyses of the chdB locus revealed the presence of two functionally differentiated alleles, chdB 1 and chdB 2 (Figure 1). Moreover, the genotyping of various goldfish populations suggested that the chdB 2 allele is a minor allele across goldfish populations. This low allele frequency of chdB 2 in the investigated goldfish populations is consistent with the additive reductions in viability for chdB 2 allele copies in Oranda lab strain progeny (Table 2 and Figures 4 and 5). On the basis of minor differences between chdB 1 and chdB 2 mRNA-microinjected chdA E127X/E127X embryos (Figures 2 and 3), it is reasonable to conclude that the low survival rates of chdB 1/2 and chdB 2/2 genotypes might be caused by deleterious events at later developmental stages.
In addition, our analyses do not show an obvious relationship between the genotype and caudal fin morphology (Figures 6 and 7; Supporting Information Tables S1 and S2). These results illustrate a major difference between selective pressures on the chdA and chdB loci. The chdA E127X allele, which almost completely lacks function, was positively selected during the domestication process, based on its twin-tail morphology phenotype. On the other hand, the chdB locus might have been eliminated from the population, because it does not seem to contribute to any preferred morphology. Thus, it is F I G U R E 3 Gene expression patterns of foxb1a and szl. Expression patterns of foxb1a in control (a), and chdB 1 -(b), or chdB 2-mRNA-injected twin-tail goldfish embryos (c). Dorsal views of gastrula-stage embryos (50% blastopore closure) are shown. (d-i) Expression patterns of szlA in early blastopore closure stages (d-f) and late blastopore closure stages (g-i). Embryos are shown from lateral view. The left (d,g), middle (e,h), and right (f,i) panels represent control, chdB 1and chdB 2 -mRNA-injected embryos, respectively. (j,k) Plots of the relationship between developmental stage and szl gene expression patterns in the first (j) and second (k) experiments. Injected mRNA groups are represented by circles (control), crossed circles (chdB 1 ), circles with X (chdB 1 ), triangles (chdB 2 ), and inverted triangles (chdB 2 ). The order of the mRNA injection is represented by prefix numbers in the key displayed on each graph ( | 379 reasonable to presume that the chdB 2 allele has been eliminated from most goldfish populations during the domestication process, as would be predicted by the random mating and selection model (Hartl & Clark, 1997;Hedrick, 2005; Figure 5b).
However, the random mating and selection model does not explain our finding that the chdB 2 allele is still present in goldfish populations at least 100 generations after its fixation approximately 400 years ago. Under the random mating model, the chdB 2 allele will be almost eliminated from the population by the 25th generation due to its disadvantageous nature, even if the allele were initially a major allele ( Figure 5B). These results lead us to pose the question of how the chdB 2 allele is still retained in the modern ornamental goldfish population.
First, stochastic factors should be examined to test whether these mechanisms explain the retention of the chdB 2 allele in the Oranda goldfish population. For example, there is a possibility that some ornamental goldfish subpopulations carry the chdB 2 allele at a very high frequency, but this subpopulation was absent from our investigated fish populations. In fact, the ornamental goldfish population tends to be divided into small segregated subpopulations due to convenient maintenance in small ponds and aquarium tanks (Matsui, 1972). Thus, genetic drift may have occurred to cause the fixation of the chdB 2 allele in a goldfish subpopulation (Hartl & Clark, 1997;Hedrick, 2005). The existence of such a subpopulation might explain the retention of the chdB 2 allele in the overall goldfish population. To examine this assumption, extensive genotyping of the chdB locus should be performed in a wide variety of C. auratus populations, including ornamental goldfish and wild-type crucian carp, which represent an outgroup of the ornamental goldfish population (Komiyama et al., 2009;Takada et al., 2010).
In addition, it should be considered that the chdB 2 allele could be advantageous on a certain genetic background and/or in certain environments. For example, subtle differences in the numbers of the left and right caudal fin rays between the three chdB genotypes may suggest that these differences may be less subtle under other circumstances ( Figure 7b). The twin-tail goldfish with the chdB 2/2 genotype showed a slightly lower proportion of individuals with extreme asymmetry in the twin-tail morphology (Figure 7b). Therefore, we may speculate a possibility that this enforcement of symmetry could be enhanced under the different genetic backgrounds or environmental conditions. In such a case, the retention of the chdB 2 allele would be explainable by positive selection of symmetric twin-tail morphology for the ornamental purposes, as discussed by Darwin (1868). Because it is known that genetic background and/or environments influence phenotypes in several animals (Gilbert & Epel, 2009;Waddington, 1957), it would be worthwhile to examine whether the expressivity of the chdB 2 allele could be altered by applying different genetic backgrounds and/or environments. Such an experiment would likely require the creation of hybrids between various different types of ornamental goldfish strains with different morphologies, and raising the hybrids under various environmental conditions (Smartt, 2001).
We also discuss several remaining problems to be addressed in future studies. Our current study suggests that the divergence and contrasting evolutionary processes observed for goldfish chdA and chdB paralogues are due to the relative contributions of these genes to the twin-tail morphology under domesticated conditions. On the other hand, our previous study reported a conservative evolutionary process for the chdA and chdB genes in common carp, based on their gene expression patterns. Moreover, an equivalent stop codon allele for chdA has not been found in our previous studies or publicly available genomic F I G U R E 5 Viability and theoretical chdB 2 allele frequency in ornamental goldfish. (a) Genotype frequency and viability of the segregants derived from chdB 1/2 Oranda goldfish parents. Light gray bars indicate the theoretical allele frequency based on Mendelian predictions (chdB 1/1 :chdB 1/2 :chdB 2/2 = 1:2:1). Solid circles indicate the observed genotype frequency in the segregant population. Open circles show the viability calculated from the observed genotype frequency. (b) The theoretical expectation of allele frequency across generations is based on the relative fitness of each genotype using the generation-by-generation algorithm (open circles with fine solid lines; Hartl & Clark, 1997). The initial allele frequencies are 0.9, 0.5, and 0.1. The observed allele frequency in the Oranda goldfish strain population is indicated by a dotted line, and the average allele frequency amongst all investigated goldfish populations is indicated by a dashed line data for the common carp (Abe et al., 2014Xu et al., 2014).
These major differences lead us to ask the question of why the chdA and chdB paralogues have evolved in such a different manner in the two closely related lineages of goldfish and common carp, which shared a genome duplication and were both domesticated. Furthermore, our analyses could not sufficiently explain why the branch length of the goldfish chdB gene is longer in comparison with some other chordin genes in the phylogenetic tree (Abe et al., 2014; Figure 1B). More specifically, our molecular developmental genetics led us to conclude that while the chdA gene highly contributed to morphological evolution, F I G U R E 7 The relationship between genotype and caudal fin morphology. The histogram shows the distribution of the sum of the caudal fin rays in the medial portion and left (or right) side (cfrM+max(cfrL, cfrR)). The distributions of caudal fin ray numbers are highly deviated from normal distributions (Shapiro-Wilk test, p < 0.001). (b) The relationship between left and right fin ray numbers. The open circles, triangles, and closed circles represent chdB 1/1 , chdB 1/2 , and chdB 2/2 , respectively. Spearman's correlation coefficients of chdB 1/1 , chdB 1/2 , and chdB 2/2 are 0.631, 0.580, and 0.678 (p < 0.01), respectively. The sample number and ratio of the genotypes are described in Supporting Information Table S1. cfrL: left-side bifurcated caudal fin rays; cfrM: midline located nonbifurcated caudal fin rays; cfrR: right-side bifurcated caudal fin rays the nucleotide sequence of chdB evolved faster than that of chdA, suggesting the conservative nature of chdA gene sequence. Several hypotheses can be raised to explain this contradiction between molecular developmental observations and molecular evolutionary processes (e.g., differences between the genes in pleiotropic functions or chromosomal locations); however, this remaining question will require a further detailed study to resolve. We expect that answering above question will provide helpful insights into the mechanisms underlying the retention or loss of paralogous genes and the relationship between molecular evolution and developmental biology in domesticated animals.