A fish-specific genome duplication (also known as the 3R duplication) occurred in an ancestor of the teleost lineage around 300–350 million years ago 3. This event, which endowed teleosts with additional new genes, has been hypothesized to be at least partly responsible for their biodiversity and species richness 245. Not all genes that emerged from the duplication are still present, however. In fact, the majority of duplicated genes (about 70–90%) have since been degraded and/or lost (a process termed nonfunctionalization). But because this massive post-duplication gene loss followed different routes, different teleost lineages now have different complements of paralogous genes derived from the original genome duplication. This process is called divergent resolution 45. Empirical support for divergent resolution between teleost lineages that diverged very early comes from a recent comparative genome-wide analysis of paralog loss in zebrafish and the green spotted pufferfish 6.

In many cases where both copies have been maintained in a genome, the functions of the ancestral gene are now distributed among the duplicates – a process called subfunctionalization. Given that retention of duplication-derived gene copies also followed different routes and that subfunctionalization can be neutral and stochastic, the partitioning of gene functions can also occur lineage-specifically. Finally, it is possible that one of the duplicates continues to fulfill the ancestral functions while the other acquires a completely new function (neofunctionalization). Differential functional evolution between teleost lineages has so far been shown for zebrafish, stickleback and medaka 4.

Together, the fish-specific genome duplication and the divergent resolution, subfunctionalization and neofunctionalization that followed it created a large evolutionary playground within teleost genomes. The duplication-diversification hypothesis predicts that gene and genome duplication and subsequent reciprocal gene loss and/or differential paralog evolution in divergent populations leads to genomic incompatibilities between isolated populations and, consequently, to postzygotic isolation and speciation. That is how the fish-specific genome duplication might have facilitated the radiation of teleosts 45.

Besides the overall impact of gene and genome duplication on reproductive isolation and thus on speciation, neo-functionalization of a duplicated gene copy can lead to the origination of a key evolutionary innovation that enables a group to radiate, for example in a new environment. In two new articles, one in BMC Evolutionary Biology 7 and the other in Molecular Biology and Evolution 8, Finn and colleagues examine an example of a cluster of genes that emerged by duplication and that apparently has enabled a whole group of fishes to diversify.

Finn and Kristoffersen had already in earlier studies 1 reconstructed the evolution of the vitellogenin (vtg) gene family in teleost fishes. Vitellogenins are yolk proteins synthesized in the liver and deposited in the maturing oocyte. Finn and Kristoffersen 1 suggested that neo-functionalization of the vtgAa gene in acanthomorphs, the most species-rich group of teleosts (comprising about 16,000 species, 78% of which are marine), was an important step towards adapting to a new spawning strategy in the marine realm. Proteolysis of the VtgAa yolk protein leads to an increase in the levels of free amino acids in the maturing oocyte and causes water influx. In this way, the hydrated eggs are protected against leakage of water into the hyperosmolar marine environment, so that the eggs float on the water surface. This is an important adaptation that makes pelagic ('floating') spawning strategies possible.

The initial phylogenetic analysis of teleost vitellogenins 1 suggested that the three vtg genes in acanthomorphs, vtgAa, vtgAb and vtgC, evolved through a progressive series of gene duplications and subsequent gene losses, involving the fish-specific genome duplication and the two earlier rounds of whole genome duplication in vertebrates (called 1R and 2R), and also an acanthomorph-specific duplication of the vtgA gene that generated the vtgAa and vtgAb duplicates. According to this scenario, lineage-specific neofunctionalization of the newly arising vtgAa paralog in acanthomorphs facilitated their conquest of the marine ecosystem from their original habitats in freshwaters.

New data presented by the same group in BMC Evolutionary Biology 7, as well as an earlier article by Babin 9, take the location of vitellogenin genes in vertebrate genomes into account and turn the duplication history of teleost vtg genes upside down. In acanthomorphs, vtgAa, vtgAb, and vtgC are located close to each other on the same chromosome. This is consistent with the arrangement of vitellogenin genes in other teleosts and in more distantly related vertebrate lineages, such as frog and chicken 79. The most parsimonious explanation for this arrangement is thus that a vitellogenin gene cluster consisting of three genes (Vtg1, called vtgC in fish, Vtg2, called vtgAb in fish, and Vtg3, called vtgAa in fish) was already present in the last common ancestor of fish and tetrapods about 450 million years ago (Figure 1). An ancestral vitellogenin gene (proto Vtg) was duplicated, giving rise to Vtg1 and Vtg2/3. The latter gene was then duplicated in tandem, generating Vtg2 and Vtg3 (Figure 1a). In the fish lineage, two vitellogenin gene clusters were present after the fish-specific genome duplication, but one of them degenerated so that this round of genome duplication did not increase the number of functional vtg genes.

In theory, phylogenetic reconstruction of the vitellogenin gene or protein family should reveal these three ancestral gene duplications. However, published vitellogenin phylogenies 17810 consistently suggest that the different vertebrate Vtg2 and Vtg3 genes have been generated in parallel but independently through lineage-specific tandem duplications (Figure 1b). One explanation for the failure of phylogenies to reconstruct the common duplication of the Vtg2/3 precursor could be that gene conversion has occurred between Vtg2 and Vtg3, keeping them alike. The new results by Finn et al. 7 and Babin 9 therefore illustrate how important it is to include synteny data for the correct inference of gene family evolution.

The evolutionary significance of vitellogenins is further substantiated by the high frequency of true lineage-specific duplication events in teleost fishes. In acanthomorphs, Vtg2/vtgAa has been duplicated in medaka, whereas Vtg3/vtgAb has multiple copies in marine labrids (wrasses). In the zebrafish, an ostariophysian, both Vtg3/vtgAb and Vtg2/vtgAa have been duplicated, the latter being present in as many as five copies 789. Nevertheless, acanthomorphs are special in their processing of the Vtg2/vtgAa protein and the exceptionally high expression of Vtg2/vtgAa in marine, pelagically spawning species 7. Although yolk proteolysis evolved before the divergence of Acanthomorpha and Otocephala (such as zebrafish and herring), it was not until the neofunctionalization of Vtg2/vtgAa in the acanthomorph lineage that highly hydrated marine pelagic eggs were made possible, thereby triggering the teleost radiation in the oceans. This happened at least 400 million years after the evolution of the Vtg2/vtgAa gene itself 78.

In another part of the vertebrate phylogeny, some lineages evolved that do not seem to have any use for yolk proteins such as vitellogenins: mammals have evolved placentation and lactation to nourish their offspring 10. It therefore does not come as a surprise that all three vitellogenin genes have been lost from the evolutionary lineage leading to the placental mammals and marsupials. Only the egg-laying monotremes have retained a single functional Vtg gene (Figure 2) 10. The evolution of vitellogenins in vertebrates nicely demonstrates an association between gene duplication and functional need. It also shows that adaptively very important genes underlying key evolutionary innovations can lose their relevance once a new innovation arises, with the consequence that such genes can vanish entirely from a genome. 'Use it or lose it' is the motto, or – in the context of genome evolution – duplicate it or delete it. An intriguing question remains: were there functional necessities of reproduction that were associated with the duplications of the vertebrate proto Vtg gene in the first place? The answer might, once more, be found in the oceans, where ancestral vertebrates used to spawn.