Humans have domesticated a large number of animals throughout their history, some for food, some as companions and protectors. A few species — think animals like rabbits and guinea pigs — have shifted in part between these two categories and currently serve as both food and pets. But one species has completely left its past as a food source. And, in another rarity, it ultimately served not so much as a companion, but as a decoration.
We’re talking about goldfish here, and we’ve now looked at their genome. And it’s almost as weird as the fish themselves.
A nice kettle of fish
It’s worth pausing for a moment to consider how weird they are in terms of domestication. They started out as light-colored variants of a carp that is otherwise used entirely for aquaculture. We’ve completely removed them from the food chain and turned them into pets, but they’re not the kind of pets we interact with like a dog or cat, or even a guinea pig. For the most part, they just sit there and look decorative. And to make them even more decorative, we have grown many varieties that are much less functional than fish.
(I invite you all to come up with an example of a species I don’t think of that has had an equally unusual trajectory and let me know in the comments.)
There is also a bit of strange history here. Although we commonly refer to them as goldfish, most of what we have is not the true goldfish. After their domestication in China (and later the move from garden ponds to indoor tanks), gold colored fish were reserved for the emperor, so they are still quite rare. In the meantime, we’ve bred species with multiple tails, species without dorsal fins, and more.
That probably does some weird things to the fish, genetically. But it turns out they were pretty weird at first.
Even the process of reporting the genome turned out to be a little strange. It was first reported in May when a group described the genome of a goldfish and compared it to its ancestor, the common carp. But the analysis was quite minimalistic. Then, this week, a huge consortium dropped an analysis of not only one species of goldfish, but also 185 different species. Plus 16 different genomes from wild carp for comparison. Although the goldfish genome is only 1.8 billion base pairs long (1.8 Gibases), the raw sequence needed to do all this is 4.3 trillion bases. It’s an amazing effort.
But because another group had already published the data, the researchers published them in PNAS through a route that only allows them to go through informal peer review. There doesn’t seem to be anything problematic about the article that would cause peer review to fail, but publishers generally want new results, and this apparently wasn’t new enough.
Fish fish fish fish
Most animals have two sets of similar chromosomes, one each from their mother and father. In humans, there are 23 chromosomes and we have two of each, meaning we carry 46 each. In both the goldfish and the carp from which they are derived, there are 25 chromosomes, but each fish carries 100 – instead of two copies, they have four, or rather two sets of two. Apparently the lineage that spawned the carp is a hybrid of two closely related lineages (possibly separate but closely related species).
Consequently, unless some copies of the genes have been deleted or disabled by mutation, the fish should have four copies of them. But there are some specific cases where they don’t, such as DNA repair genes, where one set of copies has been eliminated. And in many tissues, one or the other set of genes is more active, but there is no clear and consistent pattern of which of the sets it is. So we’re not at the point where we really understand what’s going on with the fish’s four sets of genes, but the answer probably isn’t simple.
The fish have only recently been isolated and have undergone some pretty serious selection for unusual characteristics – just check out the images in Wikipedia’s list of goldfish species. Many of the genetic variants underlying these physical traits are likely recent and have been selected as the only variant present in the species. This creates a so-called “selective sweep” in which the variant, and any others that happened to be around when it emerged, are the only ones present in a population.
So the researchers checked the fish for selective sweeps and, unsurprisingly, found quite a few. The top 1 percent of possible sweeps contained nearly 1,000 genes in all. In zebrafish, a species not too distantly related to carp, 173 of these genes had been removed. Fish with these deletions had changes in characteristics such as pigmentation and body shape, which is exactly what you would expect given the differences between goldfish and most other carp. There were also some behavioral changes, but it’s important not to think of them in terms of canine behavior – with fish it’s more a matter of how they feed or their response to fragrances.
The researchers also looked closely at the genome of egg goldfish, which no longer make dorsal fins. They identified a total of nearly 400 genes associated with dorsal fin loss. Curiously, most of the variants came from one of the two ancestral genomes. And when the subset of that was knocked out in zebrafish (57 of them), a quarter had a marked change near the dorsal fin. Whether the rest have a subtler issue, contribute to the difference in some other way, or are simply fake is not clear.
While the current analysis is incomplete, the sheer number of strains and sequences means there’s enough data here to keep researchers busy for a long time. However, with up to four copies of each gene, it will never be easy to do genetics in these animals, so it will take significantly more work to understand what the sequence differences mean.
To some extent, that idea justifies most geneticists’ decision to focus on zebrafish, which may not have as many species as goldfish but do have simpler genetics. Still, the authors suggest those two species may be related. With an extra pair of copies of each gene, the goldfish may have tolerated many more mutations than the zebrafish could.
PNAS2020. DOI: 10.1073/pnas.2005545117 (About DOIs).