Why have rh negative frequencies gone down since Bronze age?

The Bronze age in Europe lasted from around 3,200–600 BCE.

According to Blood groups in ancient Europe, it tuns out that the O allele is at high frequency in hunter-gatherers, but relatively rare on the Steppe. The B allele seems to be absent in both hunter-gatherers and early farmers, and seems to be introduced from the steppe in the Bronze Age. The Rh- allele seems to be relatively common in hunter-gatherers and, particularly, in steppe populations, and relatively rare in early farmers, partly confirming Haldane and Cavalli-Sforza’s hypotheses. Allele frequency estimates are in the figures below (bars show 95% binomial confidence intervals).

If we compute expected phenotypic frequencies, this suggests that around around 65% of Mesolithic hunter-gatherers would have been type O, compared to around 40% in present-day Europeans, and around 40% of Steppe-ancestry individuals would have been Rh-, compared to around 24% of hunter-gatherers, 4% of early farmers, and about 16% of present-day Europeans.

Traditionally the Proto-Indo-European society is considered to have been primarily patrilineal and patriarchal. However, excavations at Yamna kurgans reveal that approximately 20% of the remains are female, which appear to have lived and died similarly to their male counterparts. This has been compared to later Scythian warrior woman burials.
According to Unterländer et al. (2017), Iron Age Scythians from the southern Ural region, East Kazakhstan and Tuva can best be described as a mixture of Yamnaya-related ancestry and an East Asian component, the latter occurring only at trace levels – if at all – among earlier steppe inhabitants.

Since around 40% of Yamnas have been rh- and Scotland and Ireland showing highest patriarchal ancestry among them, it allow us to now look closer at the history of the world through the eyes of rh negative blood:

If Basques are primarily descendants of Hunter-Gatherers having been 24% rh negative, then why are their frequencies higher today? Be aware, that the author of the page referenced is a geneticist, but he wouldn’t be the first to be caught in a major mistake should his estimates be proven wrong by someone who has more time than me at the moment.

It would mean there are potentially additional factors deciding a rise in rh negatives unless you want to assume that the increase of rh-% came from the Celtic invasion. Proto-Celts were basically descendants of Yamna men mixing with women along the way from the Black Sea region to western Europe.

The Scots and Irish also happen to have the highest percentage of combined Celto-Germanic R1a (L664 and Z283 subclades) and R1b (P312 and U106), and therefore the highest percentage of patrilineal Yamna ancestry.
Yamnayan DNA tested by Haak (2015), Wilde (2014), Mathieson (2015) showed that Yamna people (or at least the few elite samples concerned) had predominantly brown eyes, dark hair, and had a skin colour that was moderately light, lighter than Mesolithic Europeans, but somewhat darker than that of the modern North Europeans. This is not unexpected considering that these samples had about 25% of recent admixture from the Iranian Plateau (before the Indo-European migrations brought Northeast European genes to the region), which would have darkened their pigmentation. Other tests have confirmed that the vast majority of Mesolithic Europeans had blue eyes, and the high incidence of red hair among Northwest Europeans (who have the highest percentage of Yamna ancestry) as well as in the Volga-Ural region and in ancient Chinese depictions of the Tocharians from the Tarim Basin strongly suggest that red hair was found among Yamnayans, and that the genes for red hair (which also include some mutations for fair hair) were spread by R1b Indo-Europeans. (=> see The Origins of Red Hair)

By the time that the Proto-Celts arrived in the Pyrenees, the Proto-Basques had been living in the region for thousands of year.
So we might now have a situation where the only factor needed is early farmers bringing down the frequencies of rh negative in Europe.

Broadly speaking, the present-day frequencies are consistent with genome-wide ancestry, in the sense that they are just a mixture of the ancient populations (roughly equal to the Bronze Age). The Rh- frequency in present-day Europe is a bit lower than in the Bronze Age, which might be evidence for selection against the allele, but might also just be because there’s a bit less Steppe ancestry in 1000 Genomes than in the samples Bronze Age populations. This doesn’t really explain the Basque frequency though. The Basque population doesn’t particularly have a lot of hunter-gatherer or Steppe ancestry. But perhaps there is substructure in Rh- frequencies within the Basque population, or within the hunter-gatherer populations. Finally, we’ve explained the present-day frequency in terms of mixtures of ancestral populations, but really we have just pushed the question back ten thousand years. Why was the Rh- frequency so high in hunter-gatherer and Steppe populations? Clearly there has to be some selective advantage to this allele at some point, otherwise it would just have been removed everywhere. Perhaps in these populations that selective advantage was stronger. Or perhaps they were just sufficiently small that it was able to drift to high frequency.

More here: Blood groups in ancient Europe