What Robot Fish Can Tell Us About Parallel Evolution

When housed in an aquarium with a swirling robotic school, what determines whether a fish will join the crowd?

bony plates
Top: The ninespine stickleback, Pungitus pungitus, is typical of the saltwater form. Bottom: A freshwater form of stickleback with fewer bony plates and fewer spines. Image based on drawings from the Queensland Government Fisheries

One of the most interesting large-scale patterns in evolution is parallelism. For example, flight has evolved many times, in parallel, from numerous non-flying organisms; many species of vertebrates that are not fish have evolved swimming, in parallel. One study discovered parallel evolution in body armor among freshwater stickleback fish from numerous saltwater ancestors.

Another interesting thing about evolution, which has only been appreciated in recent decades, is the fact that there is not a simple correspondence between genes and traits. Rarely does one gene determine one trait, and rarely does one trait vary because of one gene. There are dozens of examples of simple gene-trait relationships, many of which were discovered years ago. Because these relationships were relatively easy to find and describe, our textbooks are full of them and our thinking about genetics was for a long time based on them. But this is a little like basing our conception of how all vehicles work by deeply understanding the workings of a toy wagon. The mechanics and engineering of a little red wagon will not help us understand escalators, submarines, or Apollo lunar launch systems. We now think that most genes affect multiple traits and most traits are affected by multiple genes, and that it is all very complex.

A recent study looking at stickleback behavior seems to be an example one gene affecting multiple traits.

Sticklebacks are members of the Gasterosteidae family of fish, with species that live in salt and fresh water. The freshwater sticklebacks evolved from saltwater ancestors who were landlocked less than about 17,000 years ago at many locations across the Northern Hemisphere. For this reason, differences among freshwater and saltwater sticklebacks represent recent and rapid evolution among a well-known group of species and are thus especially interesting to scientists.

Saltwater sticklebacks have up to 36 bony plates associated with a smaller number of sharp spines. These plates and spines protect the fish from predators, but they are costly to produce and maintain. The bony plates require extra calcium, which is rare in some environments, and they restrict the body movements of the fish.

Freshwater sticklebacks tend to have fewer spines and bony plates. Some have a gap in the row of plates (this is called a “partial morph”) while others have only a few plates at the back end of the fish (“low morph”). Fresh water has less calcium than salt water, so this may be an adaptation to a limiting resource. Also, freshwater environments tend to have fewer predators than saltwater environments, so the protective features of the bony plates may be less important in fresh water; perhaps there was relaxed natural selection on this armor, and over time it was lost in many different populations in parallel.

In a 2005 study, scientists looked at a gene (Eda) that determines the growth of the bony plate and found that freshwater sticklebacks had a variant of the gene that caused fewer plates to form in those populations. The gene Eda probably serves a regulatory function, so it could determine one of a range of phenotypes from the fully armored saltwater version to the two lesser armored versions found in fresh water. A combination of genetic and population analysis led the researchers to discover that most freshwater sticklebacks in the Northern Hemisphere which exhibit a loss of bony plates do so because they all inherited a variant of Eda that is rare in the original saltwater populations. So the trait evolved in parallel in many lineages, all of which came from different saltwater populations, but it also evolved from a single pre-existing form of the gene. However, it was also found that one or more of the Northern Hemisphere sticklebacks with reduced bony plates got this trait from an entirely different genetic change.

This trait is thus an example of a feature determined by more than one gene, and an example of parallel evolution occurring by more than one means.

A second study just reported at a scientific meeting looks at what seems to be an entirely different question about stickleback evolution. Most sticklebacks form schools, which is a common adaptation among fish, following the principle that there is safety in numbers. But there is one population of freshwater sticklebacks that does not form schools. The sticklebacks of Paxton Lake, in British Columbia, Canada swim around alone most of the time. Rather than forming schools, they hide out in thick vegetation on the bottom of Lake Paxton.

The research team led by Anna Greenwood of the Fred Hutchinson Cancer Research Center in Seattle devised a machine to test for and measure schooling behavior in sticklebacks. This consists of a mobile-like cluster of fake fish which move together as a robotic school in a circle around a large aquarium. When fish from a schooling population of sticklebacks were placed in the water with this machine, they joined the fake fish and swam around with them. When fish from the non-schooling population were placed in the water with this machine, they did not school. These two populations are so closely related that they can interbreed. The researchers tested offspring of the schooling and non-schooling fish to see which behavior each fish would exhibit. As expected, some schooled, and some did not. Once the hybrid fish were sorted out, their genes were examined to see if there was a particular signature that went with schooling versus solitary swimming.

It turns out that the gene that seems to control schooling behavior in these fish is none other than Eda, the same gene that controls the number of bony plates.

So the sticklebacks not only give us a great example of how parallel evolution can arise, but also a great example of a gene affecting more than one trait. But how does that work? The fish that do not develop bony plates also do not develop a fully functioning lateral line. A lateral line is a sense organ many fish have that allows fish to detect movement elsewhere the water. Some predatory fish use the lateral line to find their prey, other fish use the lateral line to detect predators and thus avoid becoming prey, and schooling fish use the lateral line to keep track of the other fish in the school. Apparently, the sticklebacks with the poorly developed lateral lines can’t school because they can’t properly sense the other fish with whom they would need to coordinate their movements.


Colosimo, Pamela F., Kim E. Hosemann, Sarita Balabhadra, Guadalupe Villarreal, Jr., Mark Dickson, Jane Grimwood, Jeremy Schmutz, Richard M. Myers, Dolph Schluter, and David M. Kingsley. 2005. Widespread Parallel Evolution in Sticklebacks by Repeated Fixation of Ectodysplasin Alleles Science 25 March 2005: 307 (5717), 1928-1933.

Pennisi, Elizabeth. 2012. Robotic Fish Point to Schooling Gene. News and Analysis. Science 335(6066):276-277. DOI: 10.1126/science.335.6066.276-b

Get the latest Science stories in your inbox.