A Blog by

Study Of 1.5 Million Cows Shows Daughters Get More Milk Than Sons

For decades, the dairy industry has used data to supercharge the humble black-and-white Holstein cow into a milk-producing machine. Across the US, thousands of dairy farmers keep assiduous records about how much milk their cows produce, and the volume and composition of that milk. All of this information feeds into mathematical models that predict the total amount of milk a cow makes over its lifetime. Farmers use this information every day to decide how to care for and breed their animals. As a result, cows today make four times more milk than they did in the 1940s.

To Katie Hinde, the dairy records were a goldmine. Hinde studies the biology of milk; her Twitter handle, delightfully, is @mammals_suck. By analysing around 2,390,000 lactation records covering 1,490,000 cows, she found a clear pattern: Cows produce more milk for their daughters than their sons. The sex of the first calf is particularly important, and can influence how much milk their later siblings get.

These results suggest an easy way in which dairy farmers could squeeze more milk out of their herds. But they also contradict a longstanding hypothesis about how animals invest in their young, one which predicts that creatures like cows ought to favour sons over daughters.

Since the 1970s, biologists have shown that animal mothers don’t treat all of their offspring equally. Often, they invest more in one sex than the other. The most famous explanation for this pattern was proposed by Robert Trivers and Dan Willard. They argued that females in rude health should devote more resources to the sex that would benefit most from them.

That would usually be males, especially in species where males compete for females.  A really strong fit son could potentially impregnate dozens of females in a short space of time, while a weak one will impregnate none. A bit of extra parental effort could make a huge difference. By contrast, daughters spend a lot of time being pregnant and raising their existing young. This sets a ceiling on how many offspring they can raise, regardless of their physical condition.

So if you’re a baboon, a red deer or an elephant seal, and you want more grandkids, you should spend more effort on making your sons as competitive as possible. There are many ways of doing that. You could have more sons, or give birth to bigger sons.

Or, you could use milk. Milk is the main source of nutrients for mammal babies, but comes at great cost to mothers. As Hinde puts it, “Females dissolve parts of themselves to nourish their babies”. They could offer more milk, or more nutritious milk, to one sex over the other, and a few studies have looked for such biases in deer, seals, monkeys and even humans.

Hinde summarised this work in 2012, in a lucid and witty blog post that you should read. She concluded that “mothers make milk differently for males and females”, but noted that this was a niche field of research with many unanswered questions. For example, can male and female foetuses program their mother’s mammary glands in different ways, while they are still in the womb?

When dairy scientist Barry Bradford  read Hinde’s post, he realised that cows could help to address her question. The industry’s careful record-keeping would easily reveal how much milk cows offer to calves of either sex. And since calves are separated from their mothers shortly after birth, you could cleanly measure the influence of the foetus, without having to worry about whether the calves had an influence after birth.

Bradford contacted Hinde through Twitter and asked if she wanted to collaborate. She leapt at the chance. “He said, ‘I can get the records but I can’t interpret the data from a theoretical perspective’. I said, ‘Well, that’s great because that’s where my skillset is!’”

The team retrieved every lactation record from 1995 to 1999, and cleaned them up to get rid of incomplete or duplicated entries. They ended up with 2.39 million records, covering 1.49 million cows—a welcome change from the small samples that animal behaviour scientists normally settle for.

The results were very clear: Cows make more milk when they give birth to daughters than sons. The milk doesn’t contain any more fat or protein so its quality is the same—there’s just more of it. “Things turned out so elegantly,” says Hinde.

The first pregnancy is crucial. It kicks off the development of the cow’s mammary glands, and creates a baseline that affects all later pregnancies. And by focusing on records for 113,750 cows, Hinde’s team showed that the sex of the first calf has long-lasting effects.

 Image credit: Scott Bauer, USDA Creative Commons license

Image credit: Scott Bauer, USDA Creative Commons license

After giving birth, a female cow makes milk for 305 days and during that time, farmers often impregnate her again. This means that she’s pregnant with Calf 2 while she’s still breast-feeding Calf 1. If Calf 1 is a son, mothers are permanently handicapped in how much milk they can make, especially if they have another son the second-time round. If they have a daughter next, that partly reverses the deficit, but not completely.

By contrast, having a daughter the first time round ‘protects’ a cow from the negative effects of a second-round son. All in all, a cow that has a daughter first time round makes around 445 kilograms more milk across her first two lactations than a cow with back-to-back sons. That’s a sizeable amount, equivalent to a production boost of 2.7 percent.

The effect of sons and daughters on mum's milk production, across two lactations.
The effect of sons and daughters on mum’s milk production, across two lactations. S = son, D = daughter, numbers along x-axis indicate order of pregnancy. Credit: Hinde et al, 2014, PLOS ONE.

These results are a blow for the Trivers-Willard hypothesis, which cows really ought to uphold.  Dairy cows are domesticated versions of now-extinct cattle called aurochs. Bull aurochs were much bigger than females, and must have competed intensely for mates. They’re a classic example of the animal societies that Trivers and Willard were talking about. And yet, their modern descendants invest more in daughters than sons.

This doesn’t mean the hypothesis is “wrong”, just that it isn’t as universal as it’s sometimes made out to be. “The Trivers-Willard hypothesis has been such a dominating force in the literature, and the data for it is quite equivocal,” says Hinde. “Lots of people don’t want to believe that there could be a bias in favour of daughters, but this study was so clean.

So, why do cows invest more in daughters? It’s possible that the extra milk could speed up a daughter’s development, allowing her to reproduce at a younger age and amass more babies over her lifetime. Bulls aren’t in such a rush. They start reproducing later than females anyway, so they’ve got more time to make up for any shortfall in milk. This, says Hinde, means that “mothers may be able “under-invest” in a son with relatively less consequence for the number of offspring he will go on to sire.”

But Hinde cautions that people usually assume that these sorts of sex biases are adaptive for the mothers. But maybe it’s not about the mothers at all. Maybe the right question is: Why are foetal daughters better at manipulating their mothers into making more milk?

It could be that they make the right hormones. Female foetuses produce more oestrogens that could cross the placenta into the mother’s bloodstream, and influence the development of her breast glands. Sons can produce oestrogens too, but it might muck with their own genital development. Perhaps daughters can send more hormones into their mothers without any risk to themselves.

Either way, Hinde’s results have implications for the dairy industry. If they wanted to, dairy managers could ensure that most of the calves they breed are females, but they’d need to separate semen by sex to do so. In the past, some people have argued that this isn’t cost-effective, but it might be worth it if it leads to a 2.7 percent bump in milk production.

What about humans? A couple of small studies have found that women produce more nutritious milk for sons than daughters, but neither one measured milk volume. Indeed, it’s very hard to do that. One option would be to weight babies before and after feeding, but it would be hard to do that on a large enough scale to get reliable data.

Still, it’s a question that Hinde wants to address eventually. “Right now, formulas are the same for sons and daughters, and recommendations for neonatal intensive care units are the same,” she says. “We make different deodorants for men and women but we’re not thinking about whether the developmental needs of daughters and sons are identical.”

Reference: Hinde, Carpenter, Clay & Bradford. 2014. Holsteins Favor Heifers, Not Bulls: Biased Milk Production Programmed during Pregnancy as a Function of Fetal Sex. PLOS ONE. Citation tbc.

32 thoughts on “Study Of 1.5 Million Cows Shows Daughters Get More Milk Than Sons

  1. Here the female cows spend all this time gestating and lactating, year after year, and you think males need the extra?????

  2. Wouldn’t it be related to selection? I’m guessing more cows reproduce in the milk industry than bulls since the cows have to be pregnant it produce milk. A healthier cow would most likely reproduce more while a healthier bull might still not make the cut. What happens to the bulls? How many of them end up donating sperm to the cause?

  3. would be nice to do the same study in wild animals since cows are domesticated and the domestication could have affected the genes that make milk

  4. I bet that 3% more milk comes from the size of calf at 1st parturition. Bulls tend to be born larger which is more difficult for 2 yr old heifers to push out compared to the female counterparts. I would be curious to know what the difference in milk production was at 3 weeks after calving and also 75 days after calving. This would give some insight weather heifers are just transitioning into milk poorly/good due to dystocia

  5. Bull calves are larger. Difficult calvings are known to cause less milk production as the mother recovers. Was this taken into consideration?

  6. A most interesting piece of research, but as I see it there are several pitfalls in the hypothesis. The financial benefits are dubious unless you take into consideration calving index and not just 305 day lactation. The flaw that puts into the data used, straight away, is days open. This could have a definite reflection on the conclusions from the data, if all those producing heifer calves stayed open for more than twenty one days longer than the bull calves.
    The general hypothesis is however, not that surprising, or contrary to evolutionary theory. To maintain a species you need considerably more females to survive than males. As you quite rightly suggest every female is valuable, every male is not, how would we produce beef, lamb or pork without unwanted males? In a lifetime of observing cattle and their breeding nearly every female will reproduce females of a similar quality to themselves, but not every female will reproduce a male of breeding quality, in fact many outstanding females never produce a male of breeding quality, while other continually produce males of high quality even though they would not be considered as the best of the female population.
    Perhaps the data should have been assessed also for those females that not only produced above average yields from producing female progeny, but also produced as much or more from producing males; then it may be identifying those females that are the exception rather than the rule, and possibly those that are genuine bull mothers, rather than the also rans.

    1. From an earlier comment:
      “As you quite rightly suggest every female is valuable, every male is not, how would
      we produce beef, lamb or pork without unwanted males? ”

      Dairy cattle are not our main source of beef. Cattle raised for the beef industry are of different breeds and not expected to produce milk for consumers. I don’t know what happens to unwanted bulls in the dairy industry.

  7. The evolutionary pressures on a modern dairy cow are different from the evolutionary pressures on the aurochs. I’m not sure exactly how that would explain this, though. The daughters definitely will get bred, but the males (I assume) will only be bred if their mothers produce a lot of milk. So if anything you would expect opposite. It is sort of like the wild situation, but a little different. The amount of milk doesn’t directly help the calf, because calves are mostly fed formula. For dairy cows, it can only indirectly help the calves later on.

  8. Oh Ed you fell for it once again. It sounds great that the cow “invests” in the future of her offspring, but only because it is so anthropomorphized. Like others have pointed out and you wrote yourself “Bulls are bigger than females.” If “Females dissolve parts of themselves to nourish their babies.” then the cows already did so during bull gestation and further pay the price when they have to give birth to the larger fetus.

  9. It might not matter to where ever your view point is, but to the British and a considerable portion of Europe the “Dairy Industry” is exactly where over 50% of the beef industry comes from out of the dairy sector.

    1. “to the British and a considerable portion of Europe the “Dairy Industry” is exactly where over 50% of the beef industry comes from out of the dairy sector.”

      Ah! My experience with cattle has been that they’re different breeds. I do usually know better than to rely on my experience 😉 Thanks for the info!

      1. Dairy breeds are different. They are the bony ones like Holsteins (the black and white “cows” people often think of.) Their energy goes to making milk.

        By contrast, beef breeds are shorter and chunkier. Their energy all goes to making muscle, which is eaten as beef.

        Once in awhile a dairy-breed male will also go to beef, since they don’t need very many of them for breeding. That’s probably more than you’ll ever want to know about cattle breeds, but here’s hoping it’s helpful.

  10. These numbers in the histograms do not look different to me, and the SE’s are small because of the sample size. So the conclusions drawn are not strong.

  11. Were the cows in this study all Holsteins?Or were other breeds included?If a first pregnancy ends in a stillbirth,would that adversely affect future pregnancies?

  12. Artificial selection: most bulls will not breed. Most females will. If these animals developed parthenogenesis, we wouldn’t need bulls at all.

  13. Peter Talbot’s comments are interesting – as he says, you’ve got to control for effects of the birth.

    However, Peter, what you say about evolutionary theory is wrong: “to maintain a species you need considerably more females to survive than males”. That’s not how it works: natural selection operates on genes and individuals, not on species. This is why the sex balance is very near 50:50 in almost all animals – if natural selection operated at a species level, it would be heavily biased towards females.

    Curling River – you’re right of course about the different breeds for dairy and beef. What Peter didn’t explain is that only perhaps a third or a half of UK dairy cows are AI-d to a dairy bull, to produce replacement dairy heifers (most of the dairy-breed bull-calves are culled at birth). The other cows are put to a beef bull (nowadays often an extreme breed such as Belgian Blue), to produce cross-bred calves to be reared for beef – it is these which Peter was talking about.

    I wonder if the social structure of dairy herds is relevant here? I remember reading a study (I think in bighorn sheep) where females were somehow able to skew the sex ratio of their offspring according to their own social status: a dominant ewe produces more ram lambs. This fits evolutionary theory, because rams inherit their mother’s status and dominant rams have a good chance of producing many offspring. Subordinate rams produce no offspring, but all ewes produce some. Therefore if a subordinate ewe produces more ewe lambs she’ll get more grandchildren.

    I think the key here is that we’re talking about a particular breed (as several people have said), but also an extremely unnatural social structure, with huge herds composed entirely of adult females. Natural herds would be very much smaller, and would include lots of young stock and adult bulls.

    I wonder if cattle perhaps do something similar to the bighorn ewes, but using milk quantity rather than sex ratio? In a natural herd, this might mean that dominant cows favour bull calves by giving them more milk, and subordinate cows might favour their heifer calves. However, in a huge dairy herd, most cows must effectively be subordinates, as only a few cows in a herd will be dominant (domestic animals are also bred for docility, which may make most of them “naturally” subordinate). If a herd consists mainly of subordinate cows, then by this hypothesis you’d expect them to have a bias towards their daughters, which is what this study has found.

    So I’d like to see the study controlled by social position (easier said than done!).

    It’s an interesting illustration of the strengths and weaknesses of mining existing data: you get huge amounts of free information, but it may be biased or it may not include vital elements.

    1. Richard, I would disagree with your conclusion, because the ratio of births is 50:50, which it usually not, it is more like 52:48 male to female, that precludes the strategy that it requires more females to survive than males. In the first instance more males offspring die at birth than females. Less males make it to maturity in a physical shape to become dominant, and in a more natural environment most breeding males will have a limit to his harem size. Many of the herd species also breed at specific times of the year, to enhance their offspring survival rates, amongst other factors. I would also note that many of these herd species “hang together” as groups of single sexed individuals, many of the female groups being directly related. So I would suggest there is some correlation strategy for group survival.

      On the second point you make about the breeding of beef animals from the dairy herd, you would have been correct with your figures forty years ago, but today those figures have changed dramatically. Forty years ago the average dairy cow lasted 4.8 calvings, today it is 2.4 and decreasing. To replace itself a cow would need to be bred twice pure, and then it could be bred as you describe, and many where. That still leaves a considerable number of calves of pure dairy extraction reared in the UK for beef by various means. The issue of using males with muscle deformity’s to enhance the poorer quality carcass of a pure Holstein is a slightly different issue to the milk yield in relation to successful rearing of calves.

      The last question I would pose to you is your assumption that the Americas are full of very large herds. I know that the average size herd in the US in the 1970’s and 1980’s was the same as ours, about 30 to 40 cows. Like ours it did grow, but not as prolifically as ours did. Yes they did have their 3000 to 4000 individual cow units, but interestingly enough they were operating those as herds within herds, with no more than 80 cows in a group and no more than 800 in a section, over those numbers and the normal social structure broke down and delinquent behaviour became common.

  14. The Cows probably have realised by now that Males are taken from them wont see the light of day, to be kept in a crate for a matter of weeks and slaughtered for Veal ! The females will live for about three years the same as them giving milk day in day out until their bodies are so used the milk dries up and then they are taken to be slaughtered. So I guess the females need it more!

  15. If you were to dig a little deeper with all those record, you would find information on calving ease and mortality of calves born. First calf heifers that have bulls have a more difficult delivery than for heifers. First calf heifer that have difficult births produce less milk, because they take longer to recover from the difficult birth. It is almost unbelievable that these researchers did not contact a dairy farmer to review their conclusions. Every dairy farmer knows this already and is why breeding heifers to sexed semen to produce a female is now a common practice. It was the right conclusion that cows that have a female calf produce more milk, but your reasoning is flawed.

  16. Katie, I appreciate you taking time to respond. It does not take much to effect the lactation of a cow in a negative way. As a dairyman, and surrounded by dairymen, we know that cows that have a difficult birth in their first lactation produce less milk early on. But they also are more difficult to get bred back. A cow that is bred back later in lactation can produce more milk later in lactation, than the cow that gets bred back early in lactation. But the cow that gets bred back early, produces more calves and more milk in her lifetime. I respect your work and conclusions, but I have to agree with the many studies I have already seen on this work, done by universities and individual farms, that cows that have bulls (first lactation) produce less milk than their counterparts is due to calving ease and size of calf. Your comment seems to try to indicate just the opposite(that it does not matter), when NAAB has been telling us different for generations of data, not just a five year period.

  17. Dear Katie, I read the preprint and while you do consider birth difficulty based on calving scores these scores might not actually capture the impact of birth. If, as Mark writes, first calf heifers always have a more difficult delivery for bulls then the scores might be biased based downwards on the expectations of the farmer. It would be interesting to know what the distributions of calving scores etc. are in your data. I was quite surprised that the preprint did not contain any unbiased PCA or ANOVA type of analyses. Based on what Ed wrote regarding your prior research it seems a bit like you went in with one hypothesis in mind and were not really open to other explanations.

  18. The Trivers-Willard hypothesis, as you correctly state at first mention, is that healthier moms are more likely to produce sons. As a corollary, Trivers suggested that moms in better condition might invest more in sons. This study does not directly test this hypothesis. I agree that the evidence for T-W hypothesis is equivocal, but there is no evidence against it here (at least as it is presented). If one were to directly test it, a better way might be to look at offspring size as a function of maternal condition.
    Ultimately, what I’m saying is that the results are really cool, and I’d like it more if it were not pitched as a test of T-W.

  19. Hi Mark, more precisely to my statement above, in our study we went on to use a subsample in which we excluded cases of dystocia. As you state, dystocia does influence milk yield, well established in the literature and cited in our paper. The effect of more milk for daughters remained even after excluding such cases. Moreover the interaction between parities suggests something else is going on in terms of fetal influence on the mammary gland. Fetal daughters on pregnancy 2 seem to partially “recover” milk production on lactation 1, after gestating a son on pregnancy 1. And fetal sex on pregnancy 1 has continued effects on lactation 2. Our study isn’t intended to be a “counter” to established information about dystocia, rather it opens up new information about contributions to milk production, likely mediated directly through mammary gland architecture in as yet unstudied ways.

  20. Katie: I appreciate you following up on my comments. Your article could have more clearly indicated that calving ease was considered, It did indicate you used a “dairy scientist” but no indications that he wanted you to consider calving ease. It also indicates that dairy may want to use sex semen if but”people in the past argued this isn’t cost effective”. This statement also tends to lead me to conclude you don’t have a good grasp on the dairy industry. We have been using sexed semen for ten years. Some companies guarantee 90% heifers and some about 75% heifers. There is also a solution called Heifer plus which increases the percentage of heifers. It is very reasonably priced, and many dairymen use it on their virgin heifers. While you say your study is not to be “counter” to established information, the fact that the established information was not in the article, leads the reader to conclude that it is. You also state in the article that “This means that she’s pregnant with Calf 2 while she’s still breast-feeding Calf 1”. This is false. She is not breast feeding calf 1. Calf 1 is removed from a dairy cow the first day of birth. We milk the cow. They are tested each month. That is how you got the milk data from all those records. These are dairy cows, not beef cows. It is also interesting to know that if a cow calves early, she tends to have a heifer. If she calves late, she tends to have a bull. It is not a given every time, but a tendency. One of the problems we have with data for calving ease is defaults. Every program out there is set on the easiest calving setting for all calvings. So if the dairyman does not take time to record the calving ease, it just records the easiest setting. Some bull studs have taken to only using there own in house data with herds that they have contracted information to insure better information

  21. Mark,

    I am the ephemeral “dairy scientist” you speak of, so perhaps I can contribute to the discussion.

    This was a large-scale data mining study and no study of this sort can claim to have spotlessly clean data to use. In this type of work we really on the massive amounts of data to drown out whatever noise we are unable to weed out. Therefore, we cannot completely rule out the calving difficulty question you raise; however, it is highly unlikely in my mind that this is the primary explanation for the heifer benefit for 2 reasons. One, you are at least partly incorrect that the default calving ease entry in the database is a 1; hundreds of thousands of records in the database had a 0 entered, indicating that calving difficulty was not recorded. Secondly, if the mis-reported difficult calvings that remained in the 1 and 2 range were the cause of the response we observed, then we would have expected to at least see a decrease in the effect size when the reported dystocia cases were removed, and this was not the case. If one argues that every single heifer born is delivered more easily than any bull, then of course this effect is impossible to separate from any other sex-specific effect, but I have seen hundreds of calvings and I don’t buy this.

    Regarding the discussion in the paper about the economics of sexed semen, I am sure that you understand that the return on investment for nearly any management decision is dependent on the circumstances of the individual operation. According to the most recent data I can find, in the U.S., sexed semen is used on a little less than 1/4 of virgin dairy heifers. While I don’t question your decision making, the fact that one or even many producers chose to use a tool does not necessarily mean it is a profitable choice. To demonstrate the controversy on the economics of this tool (without considering a milk benefit), see these conflicting papers:

    Olynk, N. J. and C. A. Wolf. 2007. Expected net present value of pure and mixed sexed semen artificial insemination strategies in dairy heifers. J. Dairy Sci. 90(5):2569-2576.

    Chebel, R. C., F. S. Guagnini, J. E. P. Santos, J. P. Fetrow, and J. R. Lima. 2010. Sex-sorted semen for dairy heifers: Effects on reproductive and lactational performances. J. Dairy Sci. 93(6):2496-2507.

  22. Dear Katie,
    Would have been nice of you to address my comments. Congrats on getting your paper published in Plos One. I will see if I can comment on it there.

    Statements like “Moreover the interaction between parities suggests something else is going on in terms of fetal influence on the mammary gland. Fetal daughters on pregnancy 2 seem to partially “recover” milk production on lactation 1, after gestating a son on pregnancy 1.” are just absolutely ridiculous. You have zero evidence for an interaction of fetal sex with mammary glands! None. The simplest explanation, again, is that bulls tend to be bigger than heifers when born. Mark pointed this out multiple times along with the trend that bulls tend to be born later. I guess considering and publishing the simplest explanation doesn’t pay the bills at Harvard, but speculation like yours is hardly science.

  23. Dear ST,
    “You have zero evidence for an interaction of fetal sex with mammary glands! None.” Except, you know, what the mammary glands make- milk.

    If this were merely the extra cost of gestating a son or a few extra days of gestation of sons, then it seems implausible that the sex of the fetus on pregnancy 1 continues to be associated with milk production months/years later on lactation 2, especially considering how attentive dairy farmers are to energy balance? We are then left with considerations of how mammary gland architecture changes across parities (already well known in the literature) and infer/hypothesize that this additive change to mammary gland architecture may vary in part in relation to fetal sex. We do not suggest that *only* fetal sex matters, but the established explanations in the literature (dystocia, costs of son gestation) can not on their own explain the results presented here. Barry provides more info above.

    I will close in saying that I am disappointed that new data and a different perspective/alternative explanations generate repeated ad hominem attacks. That is counter-productive to any discussion of anything.

  24. Dear Katie,
    I tried to post this on your blog post about this research, but it doesn’t show up. The problem I have with your statement and the reason I say you have no evidence is that your figure regarding milk production during lactation 1 and 2 is incomplete (see the last figure above). In order to say “Fetal daughters on pregnancy 2 seem to partially “recover” milk production on lactation 1, after gestating a son on pregnancy 1.” you do need to have a comparison to a mother that has no second pregnancy. Given the data you show we simply do not know whether there is actually a “recovery” OR a reduction that is smaller than when having a son during pregnancy 2.

    I apologize for the ad hominem attacks. But like I said I read your blog post regarding this topic and if I find something to be counter-productive to scientific discussions, it is hyperbole.

Leave a Reply

Your email address will not be published. Required fields are marked *