Character Flaws?

By Vanessa Schipani Character Flaws?

By | March 1, 2011

Character Flaws?

The West African forest gecko, which Fujita and Leaché determined to be at least four distinct species
Piotr Naskrecki / Minden pictures

Imposing distinct separations on the fluid process of evolution inevitably breeds disagreement on what exactly constitutes a species. On Christmas Eve, 1856, Charles Darwin wrote in a letter to Joseph Dalton Hooker, botanist and friend: “It is really laughable to see what different ideas are prominent in various naturalists’ minds, when they speak of ‘species.’…It all comes, I believe, from trying to define the undefinable.”

Delimiting species has challenged biologists since the 18th century, when Linnaeus first sorted and named organisms using only physical characteristics. It is only since the advent of molecular-sequencing techniques that species have become less confounding to researchers, no longer so able to obscure their true identities once their DNA is laid bare. Two or more look-alike organisms previously classified as one “cryptic” species can be split into separate species due to differences in their genetic code.

The International Commission on Zoological Nomenclature (ICZN), the administrative body responsible for logging and validating the names of newly discovered or described species, requires that every new species name be accompanied by a description that includes at least one physical or behavioral feature, called a character. The ICZN’s set of rules and guidelines, the International Code of Zoological Nomenclature, accepts characters such as scale counts, colors, or short DNA sequences that distinguish one species from another.

But character-based descriptions, even ones using short DNA sequences, can be subjective, herpetologists Matthew Fujita and Adam Leaché argue. Some cryptic species defied character-based identification for centuries; and two individuals could be designated as separate species by analyzing one short DNA sequence, but as the same species by examining another sequence, says Leaché.

“It is really laughable to see what different ideas are prominent in various naturalists’ minds, when they speak of ‘species.’”

Fujita, a Harvard postdoc, and Leaché, of the University of Washington in Seattle, say that ICZN rules need to do away with strict character-based descriptions. They’ve devised a way to distinguish species using an approach called Bayesian species delimitation, which, they argue, is exceptionally objective. The method, derived from a statistical inference that is commonly used to construct branching patterns when making phylogenetic trees, determines the probability, on a scale of 0 to 1, of a hypothesis—in this case, the number of species with a given set of genetic loci. It can also be used to update previously calculated probabilities by adding newly acquired data to the bunch and seeing how multiple loci sort.

Fujita and Leaché applied this method using six genetic loci to delimit four species of gecko, which were originally thought to be one. In their analysis, four populations of West African forest geckos had a probability of 1, strongly suggesting that they were four distinct species spread across areas of Guinean and Congolese rainforests.

And Bayesian species delimitation has proved precise when applied to sorting out other species. In a PNAS paper published in May 2010, Ziheng Yang and Bruce Rannala, researchers at the Chinese Academy of Sciences in Beijing and the inventors of Bayesian delimitation, tested their method on Homo sapiens. They were not only able to distinguish different races, but also grouped humans as one distinct species, using 20 autosomal loci, each about 20,000 base pairs long.

But some scientists, like Aaron Bauer, a herpetological systematist at Villanova University in Pennsylvania, think that Fujita and Leaché’s approach is flawed because it fails to include a character to distinguish the four geckos, as per ICZN requirements. Bauer maintains that characters—whether physical, behavioral, or genetic—are still needed to distinguish one species from another. Fujita and Leaché published only information about the geographic distribution of their geckos and failed to distinguish any single physical or genetic characteristic separating them.

“Geographic location can’t be [considered] a character because it’s not intrinsic,” says Ellinor Michel, a taxonomist and the executive secretary of the ICZN. “You can’t inherit geography.”

Leaché and Fujita plan to urge the ICZN to evolve and change its rules. “The only thing that unites all species is DNA,” argues Leaché, “so it seems like genomics is the most accurate and stable way to establish a definition of species.”

But for now, Michel and the ICZN aren’t budging. In and of itself, “taxonomy is a work in progress,” she says. “You’ll delimit a species and someone will come along and do a better job” down the road. And in order to build upon each other’s work, “we need these rules,” she adds.

“[Bayesian delimitation] is a method with promise,” concedes Bauer, “but because it’s still new, it needs further exploration.”


Avatar of: Andrew Brower

Andrew Brower

Posts: 12

March 4, 2011

I laughed out loud when Bayesian phylogenetic inference was described as "exceptionally objective." The fundamental premise of Bayesianism is based upon prior subjective beliefs. \n\nI was also tickled by the notion that "it is only since the advent of molecular-sequencing techniques that species have become less confounding to researchers, no longer so able to obscure their true identities once their DNA is laid bare." I would say that even the very study in question shows that exactly the opposite is true.\n\nFinally, last I heard, Ziheng Yang was a professor at University College, London and Bruce Rannala was at UC Davis.
Avatar of: Mike Noren

Mike Noren

Posts: 10

March 5, 2011

The argument is confused. Bayesian phylogenetic analyses are effectively simulations, where observed molecular data is run through a model to come up with an estimate of how evolution has progressed. The resulting probability estimates of the clades are not estimates of how likely the clades are to be _true_, but of how likely the clade is to be found by the analysis _given the model and the data_.\n\nHence, these researchers are proposing to recognize "species" which are allegedly indistinguishable from other "species" except in their model-and-data-dependent simulations.\n
Avatar of: Bob Grant

Bob Grant

Posts: 22

April 9, 2011

Hi Andrew,\n\nThanks for reading. You are absolutely correct that Drs. Yang and Rannala have labs at University College London and UC Davis, respectively. Both researchers do, however, maintain appointments at the Chinese Academy of Sciences, and in fact both were in Beijing when they collaborated on the Bayesian species delimitation work referenced in the story.\n\nCheers,\n\nBob Grant -- Associate Editor, The Scientist

Popular Now

  1. Scientists Activate Predatory Instinct in Mice
  2. Superbug Resistant to Every Antibiotic in the U.S. Killed Nevada Woman
  3. Next Generation: Mobile Microscope Detects DNA Sequences
  4. Tenure Under Attack in Two More States
    The Nutshell Tenure Under Attack in Two More States

    Proposed legislation would eliminate academic tenure at public universities in Iowa and Missouri, echoing a move that has already gutted such permanent posts in Wisconsin.