Maybe this is somewhat of an ignorant question of me, and in all honesty I haven't tried very hard to find what I'm looking for1, but I keep hearing the argument that "information" can never be added to genetic code, from creationists / intelligent design proponents.

Although I believe this to be nonsense, I don't actually know of any scientific papers that talk about directly observed naturally occurring genetic insertion in the offspring of some organism.

In other words: is there any scientific study whereby they have extracted the DNA of, for instance, some bacterium, let that bacterium reproduce and extracted the DNA of this direct offspring, and then, when they compared the DNA they have indisputably concluded that genetic insertion has taken place in this reproduction?2

Some references to these types of claims (emphasis mine):

  1. answersingenesis.org (retrieved Sunday February 12th 2012 UTC):

    It would be better to say: "Mutations have been observed to destroy, delete or corrupt genetic information or to be neutral, but have not been observed to add information. [...]"

  2. evanwiggs.com (retrieved Sunday February 12th 2012 UTC):

    We don’t really know much about why the genes do this as we are still very weak in our knowledge of how our genome works. But none of these processes can add any data to the genome, they just move data around. I must add another point here: some evolutionists place recombination in this list, but recombination is sexual mixing and once again cannot add any data to the genome. Recombination just takes the genome and mixes what is there.

  3. comment by "codeman" on forandagainst.com (retrieved Sunday February 12th 2012 UTC):

    You make it sound like DNA can choose what it wants to do. Mutations are caused by damage being done to the DNA by ultra violet light, radiation, etc. You can't find one example of information being added to DNA.

  4. conservapedia.com quoting Dr. Grady S. McMurtry (retrieved Sunday February 12th 2012 UTC):

    Mutations do not build one upon another beneficially. Mutations do not create new organs; they only modify existing organs and structures. Mutations overwhelmingly lose information; they do not gain it; [...]

1) And I thought it would be nice to have this on this site for visitors / future reference.
2) But perhaps this doesn't even make sense and is just a display of my ignorance about the complex mechanisms of genetics, reproduction, etc.

  • According to the FAQ, Skeptics.SE is for researching the evidence behind the claims you hear or read. Please add a link to someone claimed that genetic information can never be added, so we can address that claim. Particularly, I want to see if they define and use the term "genetic insertion". (They don't mean [this](en.wikipedia.org/wiki/Insertion_(genetics)), do they?)
    – Oddthinking
    Feb 11, 2012 at 23:26
  • @Oddthinking: The claim was made in a discussion on a Dutch news site, so I'm not sure this would add any particular value here. I've heard the claim made on several other occasions as well though. The claim roughly goes that species only experience micro-evolution because information is never added to DNA but only deleted, and species therefor never evolve into other species but only ever go extinct (or something to that extent). They didn't use the term "genetic insertion"; it is the term I understand to be fit for the concept of "information" being added in strands of DNA. Feb 11, 2012 at 23:45
  • 2
    @Oddthinking: added references to claims made that information cannot be added to DNA. Feb 12, 2012 at 1:04
  • Great question, provoked some great answers, too. Feb 12, 2012 at 14:30
  • 1
    It should be noted that "information" is not purely a function of the number of base pairs. It's quite possible to code more proteins with fewer BPs and so on; similarly, additions can decrease functionality/information. Dec 10, 2012 at 23:04

5 Answers 5


tl;dr: There is evidence for not just mutations that allow organisms to do new things (which could be construed as microevolution), but for the appearance of completely new genes. Consequently, the claim "no new information can be added to DNA" is false.

The most common "gene insertions", horizontal gene transfer and gene insertion by retroviruses, have already been mentioned by @Wayfaring Stranger. Note that there are mitochondrial genes that have found their way into our DNA, but again, these are existing genes that have been moved around.

However, I have recently come across an article where they presented evidence that in maize (i.e. corn), transposable elements (DNA elements that can copy and insert themselves into other parts of the DNA, sometimes carrying additional stuff with them) had assembled a new, functional gene (functional being the important word here).

Here's an excerpt of the abstract (emphasis mine):

Transposons are major components of all eukaryotic genomes. Although traditionally regarded as causes of detrimental mutations, recent evidence suggests that transposons may play a role in host gene diversification and evolution. For example, host gene transduction by retroelements has been suggested to be both common and to have the potential to create new chimeric genes by the shuffling of existing sequences. We have previously shown that the maize (Zea mays subsp. mays) retrotransposon Bs1 has transduced sequences from three different host genes. Here, we provide evidence that these transduction events led to the generation of a chimeric new gene that is both transcribed and translated.

In other words, parts of existing genes have been shuffled around by transposons, and have resulted in a functional, new gene that is being transcribed and seems to have acquired a functional role in the plant. This isn't just a mutation. This is the assembly of a novel gene.

  • Transposable elements seem to be exactly the type of thing I was looking for, and indeed with "functional being the important word here". Great! Thank you. Feb 12, 2012 at 1:10

Perhaps one of the most famous papers related to this is:

As part of a two decade experiment, carefully watching the strains of a bacteria deal with a somewhat hostile (citrate-rich) environment, they found it had mutated the ability to thrive on citrate - something that that none of its ancestors could do.

No population evolved the capacity to exploit citrate for >30,000 generations, although each population tested billions of mutations. A citrate-using (Cit+) variant finally evolved in one population by 31,500 generations, causing an increase in population size and diversity. The long-delayed and unique evolution of this function might indicate the involvement of some extremely rare mutation. Alternately, it may involve an ordinary mutation, but one whose physical occurrence or phenotypic expression is contingent on prior mutations in that population.

Essentially, they showed evolution (not merely natural selection) occurring right there in the lab. A series of mutations occurred that gave the bacteria abilities its ancestors did not have. And because they had frozen specimens of previous generations, they could repeat it, and prove the original strain did not have these mutations.

This paper was greatly publicised when it was clumsily attacked by a creationist in what is now known as the Lenski Affair.

  • This is great information as well. Thank you! Feb 12, 2012 at 1:21
  • This is fantastic, however I fear that some might argue that information was not added because there might have been a net loss in information; i.e., the citrate-using ability might have come at the cost of other abilities. While it's not possible to enumerate every little thing E. coli is capable of and compare the strains, it should be possible to answer whether (as per the question) net additions to the genome occurred. I don't see this addressed in the abstract, do you know if the full paper goes into it? Dec 10, 2012 at 23:09

Precise study type you're looking for would be tougher to root out, but evidence of gene transfer between species is rampant. Commonplace gene insertions:

Retroviral DNA content of the mouse genome.

Horizontal gene transfer (Wikipedia, so generally readable)

Green sea slug is a plant-animal hybrid!

The slug steals chloroplasts from the algae it eats, but it does not incorporate the entire algae cells. Chloroplasts can’t perform the whole photosynthesis chemical reaction all by themselves, but rather need some genes/proteins from the host algal cell. This sea slug has incorporated the actual genes from the algae into its own cells

Google provides many more examples through: "cross species gene transfer"

  • Hey thanks. I was actually hoping for vertical gene insertion (at least: if my understanding of this concept is correct). I understand this would mean to be a genetic insertion as the result of a DNA copying "error", which is then inherited by the offspring. But I hadn't thought about horizontal gene transfer as a possibility (I had heard of it before, but simply forgot about it), so that's welcome information as well. Feb 11, 2012 at 23:15

The other answers have given a pretty exhaustive overview but let me add just some examples that can be observed in humans.

Up front: what you are asking for is a so-called copy-number variation (CNV), that is, a variation in the number of copies of a DNA segment from one individual to the next. CNVs actually encompass any variation in number – this includes deletion and insertion.

CNVs are heavily studied since they are the main source of genetic variation, and there are some monogenetic diseases known to be caused by a simple CNV, and there are interesting CNVs responsible for quantitative traits. I’ll give an example for each.

  • The elephant in the room: trisomy, as in trisomy 21, where a whole chromosome, or large parts of it, occur redundantly. This example is particularly interesting because first of all it’s a variation you can actually see under the microscope (by looking at the karyptype) and second of all, because it arises from one generation to the next in offsprings, so the copy number can be directly compared.

  • Higher copy number of segments of the CCL3L1 gene cause lower susceptibility to HIV.

  • There’s a correspondence between populations’ diets and their copy number of amylase genes. Amylase is an enzyme necessary for the break-down of long carbohydrate chains (starch) into sugars which can be metabolised by the body. Individuals from populations with high-starch diet (Europeans and Japanese who eat corn and rice) have more copies of the AMY1 gene (coding for amylase) than individuals from hunter-gatherer or fishing populations.

I want to conclude by saying that all this is well-known and not particularly new (although the amylase study is). Creationists who claim this hasn’t been observed and in particular Answers in Genesis are lying. They are not only ignorant, they are intentionally dishonest and deceptive. It’s as simple as that.


This is a particularly insidious canard by creationists, because it has the sound of plausibility to a layperson. I have my Facts Not Fantasy page set up where this canard is addressed head on as far as "information" and how it applies to genetics. As item 22 on this page says:

Now this is a particularly insidious brand of canard, because it relies upon the fact that the topic of information, and its rigorous analysis, is replete with misunderstanding. However, instead of seeking to clarify the misconceptions, creationist canards about information perpetuate those misconceptions for duplicitous apologetic purposes. A classic one being the misuse of the extant rigorous treatments of information, and the misapplication of different information treatments to different situations, either through ignorance, or wilful mendacity. For example, Claude Shannon provided a rigorous treatment of information, but a treatment that was strictly applicable to information transmission, and NOT applicable to information storage. Therefore, application of Shannon information to information storage in the genome is a misuse of Shannon's work. The correct information analysis to apply to storage is Kolmogorov's analysis, which erects an entirely different measure of information content that is intended strictly to be applicable to storage. Mixing and matching the two is a familiar bait-and-switch operation that propagandists for creationist doctrine are fond of.

However, the ultimate reason why creationist canards about information are canards, is simply this. Information is NOT a magic entity. It doesn't require magic to produce it. Ultimately, "information" is nothing more than the observational data that is extant about the current state of a system. That is IT. No magic needed. All that happens, in real world physical systems, is that different system states lead to different outcomes when the interactions within the system take place. Turing alighted upon this notion when he wrote his landmark paper on computable numbers, and used the resulting theory to establish that Hilbert's conjecture upon decidability in formal axiomatic systems was false. Of course, it's far easier to visualise the process at work, when one has an entity such as a Turing machine to analyse this - a Turing machine has precise, well-defined states, and precise, well-defined interactions that take place when the machine occupies a given state. But this is precisely what we have with DNA - a system that can exist in a number of well-defined states, whose states determine the nature of the interactions that occur during translation, and which result in different outcomes for different states. indeed, the DNA molecule plays a passive role in this: its function is simply to store the sequence of states that will result, ultimately, in the synthesis of a given protein, and is akin to the tape running through a Turing machine. The real hard work is actually performed by the ribosomes, which take that state data and use it to bolt together amino acids into chains to form proteins, which can be thought of as individual biological 'Turing machines' whose job is to perform, mechanically and mindlessly in accordance with the electrostatic and chemical interactions permitting this, the construction of a protein using the information arising from DNA as the template. Anyone who thinks magic is needed in all of this, once again, is in need of an education.

As for the canard that "mutations cannot produce new information", this is manifestly false. Not only does the above analysis explicitly permit this, the production of new information (in the form of new states occupied by DNA molecules) has been observed taking place in the real world and documented in the relevant scientific literature. If you can't be bothered reading any of this voluminous array of scientific papers, and understanding the contents thereof, before erecting this particularly moronic canard, then don't bother erecting the canard in the first place, because it will simply demonstrate that you are scientifically ignorant. Indeed, the extant literature not only covers scientific papers explicitly dealing with information content in the genome, such as Thomas D. Schneider's paper handily entitled Evolution And Biological Information to make your life that bit easier, but also papers on de novo gene origination, of which there are a good number, several of which I have presented here in the past in previous threads. The mere existence of these scientific papers, and the data that they document, blows tiresome canards about "information" out of the water with a nuclear depth charge. Post information canards at your peril after reading this.

Whilst dwelling on information, another creationist canard also needs to be dealt with here, namely the false conflation of information with ascribed meaning. Which can be demonstrated to be entirely false by reference to the following sequence of hexadecimal bytes in a computer's memory:

81 16 00 2A FF 00

To a computer with an 8086 processor, those bytes correspond to the following single machine language instruction:

ADC [2A00H], 00FFH

To a computer with a 6502 processor, those bytes correspond to the following machine language instruction sequence:


To a computer with a 6809 processor, those bytes correspond to the following machine language instruction sequence:

CMPA #$16 NEG $2AFF NEG ??

the ?? denoting the fact that for this processor, the byte sequence is incomplete, and two more bytes are needed to supply the address operand for the NEG instruction.

Now, we have three different ascribed meanings to one stream of bytes. Yet, none of these ascribed meanings influences either the Shannon information content, when that stream is transmitted from one computer to another, or the Kolmogorov information content when those bytes are stored in memory. Ascribed meaning is irrelevant to both rigorous information measures. As is to be expected, when one regards information content simply as observational data about the state of the system (in this case, the values of the stored bytes in memory). Indeed, it is entirely possible to regard ascribed meaning as nothing other than the particular interactions driven by the underlying data, once that data is being processed, which of course will differ from processor to processor. Which means that under such an analysis, even ascribed meaning, which creationists fallaciously conflate with information content, also requires no magical input. All that is required is the existence of a set of interactions that will produce different outcomes from the different observed states of the system (with the term 'observation' being used here sensu lato to mean any interaction that is capable of differentiating between the states of the system of interest).


You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .