© Goldfish, by ぱたごん [CC BY-SA 3.0], via Wikimedia Commons.
Michael Behe opens his new book Darwin Devolves with the story of the polar bear. It's big and distinctive like a Darwin champion new species, but it's really a variety of brown bear that "evolved" to survive in arctic cold (in fact, it can hybridize with Alaskan brown bears). How did it do that? Behe shows that genes for regulating fat and for metabolizing cholesterol became broken or blunted, and this had a side effect of keeping the bears warm in cold climates, changing their coat color, while permitting them to survive on fatty diets of seals. Darwin's mechanism did not create anything new; it broke things, but in the case of the polar bear, it worked out.

A Similar Story

A similar story can be told about goldfish, it turns out. Goldfish underwent whole-genome duplication (WGD) events after they diverged from carp and zebrafish. According to evolutionary theory, this provided goldfish with extraordinary opportunities for advancement, because now there were two copies of each gene to evolve. One copy, called an "ohnolog" (as a hat tip to Susumi Ohno's 1970 idea of evolution through gene duplication), could maintain the old functions of a gene. The other copy would be free to undergo evolutionary change. explains:
[T]he goldfish (and its cousin the common carp) went through a "whole genome duplication" after evolutionarily "splitting off" from zebrafish. Now, having four copies of every gene instead of two allows for one copy to change and evolve without harming the fish. This can result in lost genes or new functions for genes. This is a natural complement to "knockout" laboratory studies.

Add in the common carp genome with its own ornamental varieties (known as koi), and there are plenty of avenues of comparison to provide researchers with a window into how genes change during evolution. [Emphasis added.]
So how did the genes change? Chen et al. tell what they found about the goldfish genome in their paper in Science Advances, "De novo assembly of the goldfish (Carassius auratus) genome and the evolution of genes after whole-genome duplication." There are four things a gene can do if it is no longer alone:
  • Both copies can be expressed.
  • Non-functionalization (non-F): One copy can go silent and not be expressed.
  • Sub-functionalization (sub-F): It can take on one of the functions the gene formerly had.
  • Neo-functionalization (neo-F): It can evolve a new function.
The first two responses involve loss. But what about neo-functionalization? That sounds like gain. It sounds like some new function emerges out of the idle code of the gene copy. Is that what they found?

What Was Gained?

The authors mention "neo-F" 27 times, but readers will look in vain for the key evolutionary words innovation or novel, as in some new, novel function arising that did not exist before. The word gain appears 21 times, but 16 of those appear in the ambiguous form "gain/loss." So which is it? The paper is filled with jargon and charts, but they obscure the question of what really was gained, if anything. The closest hint of a gain is that an existing gene got switched on in a cell type where it was inactive before:
One ohnolog of the gene scube3 gained new expression in heart, while the other scube3 copy maintained the same expression pattern as that in zebrafish, i.e., neo-F.
It seems they were most interested in writing statistics about which genes got turned on or off (i.e., which genes were "expressed"). At one point, they say, "We did not distinguish between gain and loss." The last paragraph of their discussion says:
Several other features of genome sequence evolution affect how gene pairs diverge in expression over time. Key factors include divergence of the primary genomic sequence through base substitution, gain/loss of exons, and gain/loss of CNEs [conserved noncoding elements], all of which affect gene expression in different ways. Gain/loss of exons is the most important mutation correlated with non-F, neo-F, and sub-F. This process is one that has been proposed to be a critical evolutionary phenomenon that drives vertebrate diversity, and the goldfish-carp speciation is a useful case to explore this evolutionary process.
It sounds like, in the end, they are only repeating the evolutionary dogma that gene duplication gives Darwinism a chance to tinker and create novelty. Neo-functionalization "has been proposed to be a critical evolutionary phenomenon" that drives evolution. It would be "a useful case to explore this evolutionary process." Wouldn't they have highlighted a new gene with some new function if they had found one?

In Search of Natural Selection

Let's look for natural selection. The word "selection" appears only 3 times in the text, but those refer to "purifying selection" (keeping things the same), "strong selection to maintain dosage balance" (keeping things stable), or "negative selection" (preventing changes). There was no mention of "positive selection" that would indicate something new or improved had arisen. Even the word adapt does not appear in the text, except in the references. The closest paragraph to novelty in reference to "neo-F" shows retention of existing functions:
In general, the retained duplicate genes maintained an overall expression that correlates closely with zebrafish and to each other. However, accelerated expression divergence of goldfish genes began at the carp WGD, which was also observed in the common carp, in zebrafish after the teleost WGD, and in Atlantic salmon and rainbow trout after the salmonid-specific WGD. Dosage compensation appears to be a major driver of goldfish duplicate gene retention after the carp WGD, which introduced a strong negative selection against loss of either duplicated gene, particularly genes involved in metabolic processes and protein complex formation. Gene expression divergence after carp WGD follows the usual paths of non-F and neo-F (either partially or totally) and, to a lesser extent, sub-F. Those genes under partial non-F may become completely inactive (specialization) and finally lost such as in the vertebrate 2R WGD and teleost 3R WGD. Goldfish ohnologs escaped from non-F more often through neo-F rather than sub-F, which was also observed in mouse-zebrafish comparisons (49) and salmon but not in X. laevis. Neo-F favored the retention of GO [Gene Ontology] terms "kinase" and "G protein-coupled receptor." It is easy to imagine how genes directly involved in cell-to-cell communication could create interesting evolutionary changes in body form by altering where and when the signaling molecule is expressed. Although sub-F may not be a dominant outcome of goldfish genes, at least in the short time after the carp WGD, we found a few ohnologs adopting an obvious sub-F, e.g., pde4ca and ogn. ogn had also subfunctionalized after the teleost WGD, suggesting that there may be evolutionary hotspots for particular genes to neo- or subfunctionalize.
"It would be easy to imagine," in short, that gene copies "may" neo-functionalize. Science is supposed to proceed by demonstration, not by imagination. Even so, they are only imagining how the changed expression of existing genes could affect body form. Thus, goldfish are smaller than carp. Most of the known varieties of goldfish have arrived by human breeding, which is intelligent design.

Overwhelming Loss

What the researchers did find in abundance, however, is information loss. Once genomes of zebrafish and goldfish were available, loss was evident:
The two subgenomes in goldfish retained extensive synteny and collinearity between goldfish and zebrafish. However, genes were lost quickly after the carp whole-genome duplication, and the expression of 30% of the retained duplicated gene diverged substantially across seven tissues sampled. Loss of sequence identity and/or exons determined the divergence of the expression levels across all tissues, while loss of conserved noncoding elements determined expression variance between different tissues. This assembly provides an important resource for comparative genomics and understanding the causes of goldfish variants.
While evidence for evolutionary gain was largely absent in their paper, the word loss appears 74 times, and lost 27 more times! Which process appears to have predominated?

Behe Vindicated

So how are goldfish like polar bears? They evolved primarily by loss. Genes became lost or inactive, or expression levels changed. Nothing new evolved to make a polar bear or a goldfish into some wonderful, innovative new creature. Goldfish are, essentially, broken carp that get along with different expressions of the same genetic information.

Speaking of polar bears, news from University of Massachusetts at Amherst supports Behe's theory, too. "Study Reveals New Genomic Roots of Ecological Adaptation in Polar Bear Evolution" the headline reads; UMass Amherst genomicist explores how natural selection shaped gene copy numbers." Watch for any new, novel, innovative gain in genetic information:
Gibbons points to two of the interesting findings. Of the genes annotated as olfactory receptors, 88 percent had lower copy numbers in polar bears, compared with brown bears and black bears. He explains, "First, there is less to smell in the Arctic. The polar bears mainly have to hone in on two things — seals and mates. They aren't looking for berries, grasses, herbs, roots and bulbs, like the brown bear."

Polar bears also were found to possess fewer copies of the gene AMY1B than brown bears. AMY1B encodes salivary amylase, the enzyme that jump-starts the digestion of starch when animals chew plant-based food. "Human populations with a high-starch diet have more copies of this gene in their genome than human populations with a lower-starch diet," Gibbons says. "We found the same thing with bears. If you think about their diets, it makes sense."

The new research concludes that analyzing copy number variants is an important tool when investigating evolutionary changes driven by natural selection.

"Evolution acts on different types of genetic variants to do the same thing," Gibbons says. "Now that we have the technology to detect CNVs, the consensus is that this type of mutation should be examined, along with the traditional methods for detecting parts of the genome that are shaped by natural selection."
Copy number variations do not add information. They just change the expression levels of existing information. By breaking or blunting existing information, polar bears get by in the white, cold arctic where the only thing to eat is a seal or fish. If that is what Darwin meant by natural selection, goldfish and polar bears will never evolve human brains.