TLDR: nobody knows. Blood type arises from variations on a single gene called ABO. But it's always hard to infer "why" from genes, genetics is not a source of just-so stories.
Hypothesis given: "So it’s possible that each strain of norovirus has proteins that are adapted to attach tightly to certain blood type antigens, but not others. That would explain why our blood type can influence which norovirus strains can make us sick."
There is no purpose or planning behind evolution, so blood types could just be a random genetic mutation that does not reduce the chances for individuals of having offspring that survives long enough to have children. Just like curly vs flat hair or blue vs green eyes.
I'm not a biologist so I might very well be wrong about that, but it seems that one possible explanation is that the question has no meaningful answer.
I don't think blood types are a neutral mutation. For starters proteins are not free to make and blood cells make up 5% of our total body weight. But also incompatibility between blood types between a mother and child can lead to hemolytic disease[0] (which is mostly a problem with the rhesus factor). Being rhesus positive in a mostly rhesus negative population is a very strong disadvantage (and vice versa) so I'm not sure the initial mutation would proliferate.
Secondly, one would expect a neutral mutation to eventually be present in 50% of the population and homozygous in 25%. Or for three variations, we'd expect 1/9th of people to be O, 1/3rd A, 1/3rd B en 2/9ths AB. We've had blood types for millions of years, and no population has anything close to the distribution, which makes me believe there is some kind of evolutionary pressure.
Is this a three-way choice? I though it is two two-way choices, and I would expect 50% to have A, 50% not-A, similarly 50% to have B, 50% not-B, and hence, assuming zero correlation between them, 25% not-A-and-not-B (= O), 25% A-and-B (= AB), 25% A-but-not-B (= A) and 25% B-but-not-A (= B)?
Also, if there is evolutionary pressure, given that we know blood groups of millions, it shouldn't be that hard to find some correlation between blood group and life expectancy, number of kids, etc. Are you aware of any?
Blood type is determined by just one gene, not two. The gene, which has the unoriginal name ABO, has three alleles A, B and O, with O being a dysfunctional variant. This means there are nine possibilities: OO, OA, OB, AO, AA, AB, BO, BA, BB. A and B are both dominant over O, and code for blood type AB when they're both present. So there is 1 in 9 possibilities for O, 3 in 9 for both A and B and 2 in 9 for blood type AB.
It would be extremely hard to find a statistically significant indicator of increased fitness. Evolution is really good at amplifying a 0.1% advantage over countless generations to dominate a population, but to distill that signal from the noise you'd need a study with hundreds of thousands of participants.
Another problem is that the advantage from certain blood types might have only been relevant in prehistoric times, i.e. resistance to diseases that we now can easily treat and avoid. The norovirus mentioned in the article was quite dangerous back then, and still is in developing countries, but is little more than an annoyance here.
Lastly, blood type is correlated with ethnicity, which is correlated itself with life expectancy and resistance to various ailments, making it even harder to unentangle statistically.
"Blood Group AB Is Associated with Increased Risk for Severe Dengue Disease in Secondary Infections"
Not statistically significant? The malaria protection, in particular, seems well known (that doesn't make it true, but I find several papers in what look like respectable sources making such claims) or is severe malaria so rare that being protected against it isn't worth much?
In general, gene variants start out with > 99% the original variant and < 1% the new variant. So at some point, most people were probably OO, and a baby was born with mutant AO genotype. Then A spreads from the mutant.
For modeling purposes, I'll start with 50000 people, set the generation span at 20 years, and on average, each female will have 2.004 surviving children. Well, then the mutant population grows at the same rate as the general population, so remains at a tiny fraction.
So I update the model to provide a tiny advantage to the mutant by giving the mutant phenotype more surviving children per generation.
If I set up the model to start with 49998 OO, 1 AO, and 1 BO, set type O reproductive success at 2.004, type A at 2.013, type B at 2.012, and type AB at 2.013, and let it go for 2500 generations (50k years), the ending population is about 20M total, 41% O, 42% A, 11% B, and 4% AB.
That amounts to a relative advantage of only 0.45%. Obviously, the A and B mutations are very unlikely to have occurred at the same time, but it looks like both may have occurred after humans had already migrated from Asia to North and South America, as natives of South America are almost entirely type-O, and the presence of type A in North America may be explainable by Clovis migrants from Europe. So a 50ky timespan is actually a bit too long.
It seems as though there must be a selective advantage to A and B antigens, or that they be linked with an advantage, otherwise, they would not be observable as a significant population now.
[Edit:] Note that this only applies to pre-medicine. Blood type compatibility for the purposes of transfusions is definitely a survival advantage now. If current technology and politics were fixed for the next 50ky, it is likely that the "O" gene variant would become rare, and AA, BB, and AB genotypes would dominate.
I always thought that everything happens/happened , all the possibilities were tried by nature and 99% of those possibilities couldnt produce furthur (got killed diseases/stability) .
I would definitely consider it a meaningful answer to say, "There is no selection pressure in favor of that one gene vs its subsitute(s), so both/all kinds of antigens [expressed by those genes] persist with equal regularity in successive generations [modulo the effects of recessiveness/dominance]."
I thought I read a while ago that there's a theory that different blood types started out with O-, and evolved as a way to be more resistant to certain diseases, but that there were downsides to the other types (certain health problems or less efficient use of energy, I forget now).
Some micropopulations have developed their own, unique, blood signatures (Bombay group is a good example here, they lack the H antigen that makes them universal donors but Bombay only acceptors). Blood types in general are super complex combinatorially.
The concept of hybrid vigor (Expressing two different alleles expressed may yield a competitive advantage over expressing a single allele twice) has always interested me in regards to blood, though I haven't gone digging for a basis for that being a driving force.
The main 8 ABO+RhD types are encoded by just 2 genes. But there are about 200 rare blood types that arise from interactions with another 49 genes. Those 51 genes can express 379 blood antigens.
Hypothesis given: "So it’s possible that each strain of norovirus has proteins that are adapted to attach tightly to certain blood type antigens, but not others. That would explain why our blood type can influence which norovirus strains can make us sick."