Amylase and human evolution


I made a mistake that was quickly corrected by a correspondent. Yesterday, in writing about copy number variants in human genes, I used the example of the amylase gene on chromosome 1, which exists in variable numbers of copies in human populations, and my offhand remark was that the effect is “nothing that we can detect”, but that maybe people with extra copies would be “especially good at breaking down french fries”. Well, it turns out that we can detect this, that there was even a very cool study of this enzyme published last year, and that the ability to break down complex starches rapidly may have been a significant factor in human evolution.

So of course I have to tell you all about this now.

Amylase is a digestive enzyme that specifically helps break down complex sugars, like starch, into simpler sugars that can be processed by other enzymes or absorbed by the intestine. You are secreting this enzyme all the time into your saliva, so that as soon as you pop something made of starchy carbohydrates, it’s coated with these enzymes that get to work breaking it down. As it’s swallowed and enters the stomach and intestine, these enzymes are still busy rendering the starches down into something more manageable for your gut to handle. This is an important function: if you chew starchy foods made from rice, corn, or potatoes, saturating them with the enzyme, blood glucose levels afterwards are measurably higher than if you just swallow the food whole.

As I said yesterday, the amylase gene has another interesting attribute — it seems to be in a hotspot for duplication, and different people have different numbers of copies of the gene. If you just had one copy of the gene per chromosome, your cells would each have a grand total of two copies…but instead, we more typically have 5 to 7, with some people having only 2, and others having 15 or more.

There are a number of ways to count how many copies of the gene there are, but one of the more visually appealing ways is to use fiber FISH (fluorescent in situ hybridization). In this technique, intact DNA is extracted from cells and stretched out, then a probe for the amylase gene that is tagged with a fluorescent label is allowed to bind to the strand, and every place with an amylase gene lights up red. As you can see below, then you can directly count the genes: in this case, the individual at the top has 14 copies of the gene on two chromosomes, the one in the middle has six copies, and the very interesting fellow named Clint at the bottom has only two copies. Clint happens to be a chimpanzee.

High-resolution fiber FISH validation of AMY1 copy number
estimates. Red (~10 kb) and green (~8 kb) probes encompass the entire
AMY1 gene and a retrotransposon directly upstream of (and unique to)
AMY1, respectively. (a) Japanese individual GM18972 was estimated by
qPCR to have 14 (13.73 ± 0.93) diploid AMY1 gene copies, consistent with
fiber FISH results showing one allele with ten copies and the other with four
copies. (b) Biaka individual GM10472 was estimated by qPCR to have six
(6.11 ± 0.17) diploid AMY1 gene copies, consistent with fiber FISH results.
(c) The reference chimpanzee (Clint; S006006) was confirmed to have two
diploid AMY1 gene copies.

Chimpanzees don’t exhibit much copy number variation in the amylase gene, but humans do…how interesting. Could this variation be a reflection of something specific in human evolution? Why, yes.

This work by Perry and others went on to look for patterns in different human populations with different dietary historys, and discovered that there is a correlation: cultures with diets heavy in starch, agricultural populations such as Americans, Europeans, and Japanese, or hunter-gathers who live on many roots and tubers, have a higher average copy number than cultures that depend more on hunting and fishing.

Look at the distributions! Populations with little starch in their diets also have a relatively low copy number of 5.44 amylase genes per individual; we french fry eaters have a higher number of 6.72 amylase genes per individual. The difference is small, and the distributions also overlap significantly (note that some with high starch diets only have 2 copies, and some living on low starch diets have 13 copies), but the difference is measurable and significant. It implies that there may have been some selection for greater copy numbers in cultures with diets high in starchy plants.

Diet and AMY1 copy number variation. (a) Comparison of qPCR-estimated AMY1 diploid copy number frequency distributions for populations
with traditional diets that incorporate many starch-rich foods (high-starch)
and populations with traditional diets that include little or no starch (low-
starch). (b) Cumulative distribution plot of diploid AMY1 copy number for
each of the seven populations in the study.

We can’t entirely rule out drift as a possible cause of the difference; while we can see differences in the enzyme levels in saliva, low levels of the enzyme haven’t been shown to be directly deleterious to anyone. It is again implied: oral digestion of starches may be an important pathway for taking in energy during episodes of diarrhea, so it could be critical when individuals are experiencing disease-related stress.

It is very suggestive. The fact that it represents a distinct difference between other apes, such as chimpanzees, and ourselves also suggests that it may have had significant evolutionary consequences. Maybe we aren’t primarily the biggest-brained apes; we are the Apes Who Eat Roots As Well As Bananas. A core nutritional difference could have played a more significant factor in our early evolution than small differences in brain size, and may have been an enabler of brain expansion.

Perry GH, Dominy NJ, Claw KG, Lee AS, Fiegler H, Redon R, Werner J, Villanea FA, Mountain JL, Misra R, Carter NP, Lee C, Stone AC (2007) Diet and the evolution of human amylase gene copy number variation. Nat Genet. 39(10):1256-60.


  1. SC, OM says

    The differences across groups with different diets – on what of time scale do they develop? As the global diet appears to be becoming more homogenous, will this variation decrease, leading – it seems most likely – to higher enzyme levels generally? Would this potentially be a problem if humans in the future were forced to eat another kind of diet? And related to this, is there any evidence (perhaps you mentioned it, but I missed it) that copy numbers of this gene are related to the copy numbers of any other genes, or doesn’t it work like that?

  2. Charles says

    do you think that this data might indicate that the amylase gene could potentially somehow be linked to diabetes?

  3. Christopher Balambao says

    Thats what I respect most about ya PZ, someone shows you data that contradicts something you said you own up and change your view accordingly. Now if only the wakaloons would do the same.

  4. Ubi Dubium says

    Interesting. I wish we had the data to see if there was a correlation between the replication of the amylase gene and the introduction of cooked food into the early hominid diet. It strikes me that the ability to cook grains and starchy vegetables allowed the introduction of much more starch into our diets, and it may have been at that time that there began to be a direct benefit to additional amylase genes. Alas, I don’t hear that we’ve sequenced the Homo Erectus genome yet, let alone any of their predecessors.

  5. Schmeer says

    What do we know about other mammals that eat starchy foods? Don’t pigs eat a lot of roots? Do we see many multiple copies of an analogous gene in them?

  6. Kat says

    Very interesting. Is there any correlation between the number of Amylase copies and the risk for diabetes?

  7. says

    Whoa…hold on here, what the FUCK! Are you seriously saying that you were WRONG, that you’ve been CORRECTED by being shown EVIDENCE, and then actually WRITING UP THE NEW INFORMATION!

    TRUE CHRISTIANS READ THIS BLOG!!!!!oneelevenoneexclamationmark

    You were SUPPOSED to vehemently DENY the role of Amylase, SHOUTING THAT WE DON’T KNOW WHAT IT DOES!!! You must then refer to those who claim to know what it does as ELEEETTISST AMYLASIANS. You must start your OWN SECT! Write a TEXTBOOK, TEACH THE CONTROVERSY!!!!

    Oh, and get donations. Lots of donations.


    Really cool post, I always learn things here.

  8. LisaJ says

    That was awesome PZ. Thanks alot for that. The contrast between the two human examples and that of Clint the chimp is striking. I wonder which other genes we know of exhibit such copy number variation like this in humans, and what they can tell us about our own evolution. Anyone have any other examples?

  9. says

    Slightly OT but I feel the need to say this:

    This is a perfect example of the crucial difference between science and religion. A scientist is capable of saying “I made a mistake” – in fact, not only capable but quite willing when the need arises. And the result is that everybody learns, including the aforesaid scientist.

    How many times has a member of the religious fraternity (which includes cretinists and IDiots) say “I made a mistake”? Answer: never. None of them. Not even once. Result: a self-perpetuating conglomerate of intentionally ignorant slaves to the endlessly circulating rhetoric they keep swallowing down then throwing back up into their communal trough of dogma.

    I’m on a roll today. Must get that book finished…

  10. Charles says

    whoa kat, i had just asked the same thing! really interested to find out more information regarding the link to diabetes.

  11. Helioprogenus says

    Thanks for the update PZ. Those of us following this now have an even better understanding of these variable genes. Is there any guess as to how many there may actually be? Could they run into the hundreds? Out of roughly 20,000 genes, we’re more variable then previously thought.

    On a complete tangent, perhaps its blind ego, or a heaping amount of selfishness, but every time one of these studies occurs regarding human variation, I want to scrape some dna off my cheeks and send it to a lab for study. It makes the data all the more relevant when you can apply it to your own ancestry and heritage. Maybe it’s just a natural response to wondering how one stacks up to the rest of the population.

    There must be a great deal of research requiring people from various backgrounds submitting their DNA, and I’ve been looking. Eventually, I gave up and went to DNA heritage to trace my Y-chromosome and mtDNA results.

  12. Kat says


    lol, great minds think alike or something. I’m a diabetic myself so that was a really obvious question to me. I wonder how many copies I have. :)

  13. Nerd of Redhead says

    The amylase enzyme just helps to break down starch to glucose (dextrose). It has nothing to do with what happens once the glucose is in the bloodstream. Different proteins are involved in trasporting glucose from the blood stream to cells, where it is metabolized for energy. The cell transport proteins respond to insulin. Type I diabetes involves an autoimmune disease where the immune system takes out the pancreatic beta cells. Type II diabetes involves improper response to the insulin that is produced. I don’t see any overt connection, but there may be one.

  14. Ryan F Stello says

    Absolutely brilliant.

    And to echo Glen, we have yet another easy to digest (heh) example of a population developing an advance in a trait in unison.

    More please!

  15. DGKnipfer says

    @15; As a non-theist I have to admit that I have heard several Christians say to me that they were wrong and change their positions on many things. Just not on their faith. Many of them can be completely reasonable on almost any subject, except their faith. It is inappropriate of us to make such sweeping generalizations. Like anybody who is delusional, they can be perfectly reasonable as long as you don’t say or do anything that disrupts their world view. When you do, that’s when they become completely out of kilter.

  16. Charles says

    @Nerd of Redhead

    thanks a lot for the info. i’m sure i’m WAY behind here, but it’d be nice to see more gene research done on the proteins that actually transport glucose from the blood stream to the metabolizing cells. because those transport proteins are directly related to insulin it seems that there could be some kind of relationship there. but then again, it could be that it is simply the genetic indicator for the disease and nothing more or less.

    my brother was type II diabetic. lost his life to complications. i’ve read quite a bit into stem cell research and the potential for developing treatments and/or other possibilities that could arise from it in regards to diabetes, but haven’t explored copy number variants in human genes and that relationship (if there is one) with the disease so this struck me as a very interesting possibility. if there is a lot of data on this already, i’d love to read it if anyone can hunt it down.


  17. RamblinDude says

    “How many times has a member of the religious fraternity (which includes cretinists and IDiots) say “I made a mistake”?”

    Well, it depends. If you want to know what their lives were like before they became a Christian, before their souls were saved by the blood of Jesus and all their sins were forgiven, halleluiah, praise the lord, amen, well then they freely admit they made lots and lots and lots of mistakes. But that’s all before they realized that the bible is the literal, perfect word of God and therefore evolution is evil. Amen.

    Okay, I’m being mean. Actually, I have heard Christians admit to making mistakes. After all, only God (and his Holy Word) is perfect. And if He weren’t so full of love and forgiveness, he would send all mankind to eternal hellfire for making so many mistakes, halleluiah, amen.

  18. Didac says

    It seems clear that the AMY1 expansion has been selectively promoted in human populations. We have here an example of the first step in gene family expansion: the mere tandem duplication of copies. After that, we can have some specialization. One example of this process (a process that span more than 600 million of years) is the expansion of ST8Sia gene family.

    Gene copies proliferation such as in the case AMY1 is a very crude way of upregulating gene expression. But, if it is effective, it is selected no matter how unelegant it looks.

  19. says

    @26: The same idea jumped out at me, as my brother has type I diabetes. However, I think NerdofRedhead @ 22 is correct. Since amylase is not involved in cellular absorption of glucose I don’t see an obvious relationship to diabetes, although it might make sense that a high incidence of type II diabetes and a tendency to have more copies of the amylase gene would be found in populations with starchier diets. (Just speculating on that, though.)

    Fascinating post, though, and now I’m going to have to go track down the paper to read in full.

  20. Nerd of Redhead says

    I forget to mention earlier #22, that the pancreatic beta cells produce insulin in response to blood glucose levels. So in type I diabetes, the body loses its ability to regulate glucose levels. In type II, the body is trying to regulate the glucose levels, but not succeeding like it should due to the lack of response to the insulin.
    That ends my knowledge of diabetes.

  21. Aaron says

    Since this is a science blog, I’m going to be skeptical of the adaptive explanations for the differences in amylase expression.

    One possibility, in line with the drift arguments suggested, is that the human population has exploded in the past few thousand years. There were likely to be many deleterious as well as inconsequential traits overrepresented in populations with high birth rates. Or to restate, if it is easy to have babies and many of them live, then selection is not as important of a factor.

    Would the high starch diet cultures not be more likely to have many more offspring? It’s a higher energy diet, I suspect, and allows for denser population growth. When the average individual is no longer just struggling for survival, then they can have more variant phenotypes and still reproduce.

  22. Sili says

    This is an important function: if you chew starchy foods made from rice, corn, or potatoes, saturating them with the enzyme, blood glucose levels afterwards are measurably higher than if you just swallow the food whole.

    So is generations of mothers telling their kids to chew their food at least thirty-two times the cause of the obesety epidemic?

    – Sili, 178cm, 67 kg, swallows.

  23. Jim says

    @ Helioprogenus
    Hang on just a bit longer, the $100 sequenced genome should be available in a year or two. I particularly like the gizmo Pacific Bioscience is putting togather (no i don’t on any stock).

  24. Eric says

    The comment PZ made about brain evolution is pretty interesting to me. Considering that the major energy source for our brain is glucose an increase in sugar in our diet should increase the amount of energy available for the brain and allowing it to grow without starving the organism.

    But, I’m just a college student speculating. A professional biologist would know better than me what’s going on.

  25. jfatz says

    You realize you just admitted an error, which means all scientists are wrong, and science is completely incorrect, right?


  26. says

    So is generations of mothers telling their kids to chew their food at least thirty-two times the cause of the obesety epidemic?
    – Sili, 178cm, 67 kg, swallows.

    more likely its the fact that the little darlins just sit infront of video monitors and exercise no other part of their bodies than their fingers…

    That and the craptastic loads of fat in most kids’ diets…

  27. jj says

    Another overlooked important factor about amylase, and I’d say even more important, is that this enzyme (well Alpha-amylase and beta-amylase actually) are the reason we have beer! Heat Barley Malt up to about 155F and it starts breaking itself down into simple, fermentation sugars! Oddly I do not think the enzyme is present in any other cereal grain. Beer is way more important than being able to digest complex starches.

  28. Nangleator says

    Heh. I figgered you atheists were all hot and bothered by AnalEze. Heh. Something about god and hell… um…

  29. Interrobang says

    I had a different diabetes-related question. If amylase-enhanced people have a higher blood sugar response to eating starchy foods, it would make sense, for purposes of blood sugar control, for diabetics to know whether or not they had more copies of the amylase gene or not, right? That’s not really to do with insulin, per se, but as a means of keeping the sugar from spiking in the first place.

    I’m thinking that would be useful in particular to people with Type II diabetes who often have to use diet as their primary or only means of sugar control, risk factors for that circumstance being poverty and Americanness.

  30. c-serpent says

    Do the same genes code for pancreatic amylase? While chewing food and mixing with salivary amylase gets the bolus rolling, salivary amylase doesn’t work so well once the bolus reaches the low pH of the stomach. That whole process takes about 30-60 sec if you chew your food well. Then it mixes with the low pH gastric secretions and pepsin and stays there for several minutes. This denatures salivary amylase. In the duodenum, the acidic chyme is neutralized by bicarbonate secretion but salivary amylase is already partially or wholly denatured. Pancreatic amylase takes over. By most estimates, salivary amylase only catalyzes about 5% of total starch digestion. The rest is handled by pancreatic amylase. I believe that the 5% estimate is a population average probably from the US or western Europe and thus representative of a high copy number group.

    Assuming that the amylase produced by both salivary glands and the pancreas are coded by the same genes (very likely) then this has no bearing on the adaptive argument and this is just a minor quibble with the physiology.

    But if they are encoded by different sets of genes then comparing copy numbers for both salivary and pancreatic amylase genes would be particularly useful. That is if drift explains the correlation of copy number with diet then one would not expect the copy number for pancreatic and salivary amylase both to correlate with diet. If it is adaptive, then one would expect the correlations to be the same for both organs.

  31. Matt, Sexual Jihadist says

    I don’t know too much about this fiber-FISH, though I have used FISH microscopy before. Does the red/green indicate the 5′–>3′ directionality, and if so, are those duplicates reversed over and over?

    Not sure if this observation is just plain ignorant, but I thought it looked interesting.

  32. Nerd of Redhead says

    GIL, if science needs god for an explanation, we’ll call. Tell god no need to wait by the phone though. Science is doing very well with him.

  33. Marc Abian says

    When the average individual is no longer just struggling for survival, then they can have more variant phenotypes and still reproduce.

    The obvious test is to look at other human genes to see how much their copy number has increased or decreased and that way you can get guess whether it was evolution or drift.

  34. BJ Marshall says

    See, this is why science is so cool:

    It’s totally all right to fess up when you find out you were mistaken. Science is self-correcting, and saying “whoops, my bad” actually shows WELL on the scientist for being intellectually honest.

    It’s not like when other people keep moving the goalposts to support the integrity of an arcane text that’s been potentially mistranslated your-choice-of-deity knows how many times.

  35. ennui says


    Thou tongueless open-arsed bugbear!
    Thou viperous clay-brained callet!
    Thou gorbellied plume-plucked miscreant!

    (thought it would be good to reprise an old favorite, the Shakespeare Insults Generator)

  36. 'Tis Himself says

    You’re obviously in the right profession, PZ. You’re an excellent teacher, explaining a complex process in a way understandable to amateurs yet not talking down to us.

  37. Aaron says

    @52: Yes, that is exactly the thought I had. Glancing over the original paper, it appears that Figure 4 shows these data. Specifically, they examine copy number differentiation at AMY1 versus other genome-wide loci in the Yakut (low starch) versus Japanese (high starch) populations. It appears that AMY1 is, indeed, an outlier, suggesting that other loci are not under the same influence, or, to rephrase, that drift is not sufficient.

    Having said that, I’m not an expert on these techniques, so perhaps I am misinterpreting the results.

  38. John C. Randolph says

    Mmmm.. Starchy carbohydrates.

    I suddenly have an urge to make some mashed potatoes.


  39. Malcolm says

    Randy Stimpson#57

    Is there any information on how recombination affects the number of copies of the gene that offspring have? Let’s say I have eight copies and my wife has five. What about our children?

    The answer would be somewhere between 2 and 11.
    It depends on how your copies are distributed between your to alleles. For example, individual A above would pass on either 10 or 4 copies, individual B would pass on 3, and Clint would always pass on one.
    This of course assumes no crossovers occur during meiosis.

  40. Malcolm says

    Now that I think about it, the answer to Randy’s question is between 0 and 13. The minimum number of copies per allele is zero.

  41. Fernando Magyar says

    I just downloaded my January edition of Sciam, the whole edition is devoted exclusively to all things evolution, including human evolution. Plus a bonus part stomping all over creationists. Yum! I’ll be up late reading.

  42. says

    I wonder if the chimps that have been observed digging for roots in Tanzania have started to show any signs of selection for increased amylase production…

    What chimps, you ask? I’ve been following the development of this starchy food digestion and human evolution hypothesis for a couple of years now: Here’s a post with links to three different stories that together suggest that digging up starchy roots may have been our own ancestors’ first use for tools. So much for Man, the Hunter. Hominins, the Potato Gatherers is looking more likely.

  43. RickrOll says

    G Felis, you mock the potato. It is as a God among the lowly tubers!

    There was also findings in South America that clearly suggest agrarian culture many thousands of years ago. All of the usual plus a small, unusual piece covered in white dried moisture, like a sickle!

    Fascinating stuff. Always. I really wish i could find this thread on the PhysOrg forums which there was a commenter who presented a great list of things that would indicate that human ancestors followed the shoreline in their expansion, and that much of the substance of their diet was fishy (syndactyly, which occurs all to frequently to not be a hint to this). It certainly provides the intelligence to Use tools as such, and bringing together the ideas here and the ideas of that commenter would provide a great insight into how humans developed anthropologically in the far-distant past.

  44. Thinker says

    Great stuff on the copy numbers of the amylase gene! Does anyone here know if there are studies on how this correlates with variations in the regulatory mechanism controlling amylase expression? An enhanced regulatory mechanism (for example, faster reacting) could be another adaption we might see in peoples with a higher starch diet.

  45. says

    Why is it that when natural selection possibly accounts for an evolutionary change, everybody must mention drift, whereas when genetic drift possibly shaped a biological feature there’s no obligation to mention selection?

    Okay I’m just trolling, but hey, that’s a sciencey trolley!

  46. David Marjanović, OM says

    much of the substance of their diet was fishy (syndactyly, which occurs all to frequently to not be a hint to this)

    I don’t understand. Please explain.

  47. Confused says

    Eric@38 and OP:

    Given proportionally more of the mass of the brain is lipid than the rest of the body (neurons have insane surface area to volume ratios, not to mention myelination), the impression I’d been given was that an increase in the number of fats in our diet was more likely to have increased brain size.

    If there is a connection, it strikes me as more likely that amylase copies would be a consequence of brain size – i.e., once we were dependent on a carb-rich diet to keep our brains running, then we start selecting for larger amylase CNVs.

  48. says

    This is awesome. In that “digest french fries better” quote, you effectively predicted the results of the study … although after the fact.

  49. SimonG says

    Sometimes the theists will try to claim that atheism is a dour belief, devoid of beauty and wonder. But who could fail to be filled with wonder at the discoveries being made through modern genetics? This sort of thing truly is marvellous.

  50. OrchidGrowinMan says

    Again, Late to the party, but I have a SPECULATION as to the purpose of salivary amylase. After all, as pointed-out above, only about 5% of carbohydrate digestion is accomplished on SWALLOWED carbs….

    First of all, do other creatures have salivary amylase? What about cud-chewers?

    As I understand it, the human digestive system is much attenuated in comparison to our relatives, who process massive quantities of fibrous materials.

    Could it be that our love of gum-chewing is related to salivary amylase and our maladaptation to very high-fiber diets?

    As a kid, I used to pick wads of grass “seeds” off the tops of stalks, stuff them in my mouth, chew them up, and suck the sweet juice until the wad wore out. I probably got some vitamins and stuff too, but it’s the sweetness developed by chewing I was after. MY kids chew-up, suck-on and spit-out the pulp/skin/seeds of certain fruits like oranges and tough-skinned (wild) grapes. I’m sure the amylase helps extract a bit of nutrition from this unappealing behaviour. What about chewing gum (tobacco)? Is that a vestige of an old instinctual casual way of getting nutrition from otherwise too-fibrous resources?

    I hope someone with more knowledge of such things sees this and comments, but in the meantime, I have a wad of “farmer’s gum” to enjoy. Why don’t we have salivary protease?'s+gum%22&source=bl&ots=le-u5vhrg_&sig=mIX7YNiZXpNvxHKVUcjasgig-_4&hl=en&sa=X&oi=book_result&resnum=2&ct=result#PPT35,M1

  51. RickrOll says

    Swiming is alot easier with webbed feet, David. Plus it occurs in around 1% of the population, which shows that it most certainly is a trait that is from our evolutionary past- quite recently too, it would seem.

    The Omega 3 fatty acids are very good brain food, as well. Fishing is also one of the first things man was ever good at. Man has never feared the seas. There is great evidence that some subset of polynesia showed up in the New World many thougsands of years ago.

    I’m sorry, the whole post is in PhysOrg forums. I ought to go over their and see if i can’t find it somehow. Sigh. It was really fascinating. I can’t properly discuss this without some source. Grrr.