The Genetic Legacy Of Our Ancestors' Worst Traumas

A "hard selective sweeps" is a population-wide change in the frequency of genetic traits due to some natural selection inducing cause (called a "selection pressure"), often caused by something like the Black Plague or a massive famine, which kills almost all people without the right genetic traits, or at least prevent them from reproducing. Specifically, as the introduction explains:

In the case of humans, a consistent finding has been the apparent paucity of classical ‘hard sweep’ signals in modern genomic datasets (that is, where a new beneficial mutation increases to 100% frequency in a population). 

This has resulted in suggestions that humans adapted to new environmental and sociocultural pressures through alternate modes of selection—for example, ‘soft sweeps’ (where the beneficial allele sits on multiple genetic backgrounds) or polygenic selection (subtle frequency shifts across numerous loci with small fitness effects)—though the evidence for these processes has also been challenged. The apparent ubiquity of admixture in human population history suggests instead that the absence of hard sweep signals in modern human genomes may be a consequence of the masking effects of historical admixture events. Moreover, growing evidence suggests that admixture events pervade the history of most natural populations, which may have led to a potentially biased view of the modes of selection operating in nature.

But, suppose one population experiences a hard selective sweep, and another does not, and then, after the event causing the selective sweep passes, the populations merge, at least in part, a process which is called admixture. In those cases, it is hard to tell that one of the parent populations of the admixed population underwent a hard selective sweep. So, a naive examination of the genomes of human populations that ignores admixture effects greatly underestimates how many hard selective sweeps the parent populations experienced. 

But, a new study whose abstract appears below, informed by 1162 samples of Eurasian ancient DNA and new methods, reveals that hard selective sweeps have been much more common in the last 10,000 years than previously known (although the authors conservatively estimate that as many of 10% of the 57 hard sweeps they identify could be false positives).

For example, there was a hard sweep favoring certain immune system related genes in Anatolian first farmers that was not present in Western European hunter-gatherers or steppe peoples, the two other main sources of European ancestry, that was diluted when these sources of ancestry admixed because whatever selection pressure had caused the hard sweep in Anatolian first farmers had abated by the time that the admixtures occurred.

They find that 41 out of 57 of the hard sweeps, however, probably started long before the Holocene era and acted primarily on hunter-gatherer populations ancestral to modern Europeans. It makes sense that genetic populations of that era which were small and vulnerable would be more prone to hard genetic sweeps.

The role of natural selection in shaping biological diversity is an area of intense interest in modern biology. To date, studies of positive selection have primarily relied on genomic datasets from contemporary populations, which are susceptible to confounding factors associated with complex and often unknown aspects of population history. In particular, admixture between diverged populations can distort or hide prior selection events in modern genomes, though this process is not explicitly accounted for in most selection studies despite its apparent ubiquity in humans and other species. 

Through analyses of ancient and modern human genomes, we show that previously reported Holocene-era admixture has masked more than 50 historic hard sweeps in modern European genomes. Our results imply that this canonical mode of selection has probably been underappreciated in the evolutionary history of humans and suggest that our current understanding of the tempo and mode of selection in natural populations may be inaccurate.

Yassine Souilmi, et al., "Admixture has obscured signals of historical hard sweeps in humans" Nature Ecology & Evolution (October 31, 2022).

The discussion section of the paper explains that:

Our analyses of >1,000 ancient West Eurasian genomes has uncovered strong evidence for 57 hard sweeps in Early Holocene to Middle Holocene populations that have been almost entirely erased from descendent populations in modern Eurasia. Most of these selected loci had probably swept to high frequencies well before the Holocene era; this is supported by the moderate-to-strong selection strengths that were inferred for the sweeps. These selection coefficients are comparable with the strongest currently known for human populations (that is, the LCT locus having s between ~2% and 6% in the recent history of European populations) and suggest that such strong positive selection events have been much more common in recent human history than previously recognized.

Our empirical and simulation results implicate Holocene-era admixture as the primary factor attenuating these historical sweep signals, which has led to them being missed in previous studies or detected instead as empirical genome-wide outliers of haplotype-based selection statistics. Such haplotype-based outliers are typically interpreted as ongoing partial sweeps resulting from recent selection, as the underlying haplotype patterns decay quickly over time and become largely undetectable for selection starting more than 30,000 years ago in humans. However, our analyses suggest that most (85%) of the 41 sweeps that overlap with a haplotype-based outlier were already under selection by 30 ka, implying that these signals are more likely to result from admixture-driven dilution of hard sweeps that mostly began before 30,000 years ago.

An intriguing implication arising from the simulations is that the selection pressure(s) underlying the sweeps may have eased during the Holocene period in some cases. This period marked the introduction of new technologies and diets, a stable warm climate and living conditions that introduced new selection pressures (for example, selection on the LCT gene to reduce the costs associated with milk consumption in adulthood) and may have also reduced the intensity of historical selection pressures underlying the 57 hard sweeps. 

Alternatively, these patterns could have resulted from the introduction of new ancestry sources into Europe following the Bronze Age period, which would have further diluted sweep signals and left insufficient time for their reappearance even if the selection pressure was still present. Indeed, this may explain why modern Tuscans and Finns had the fewest sweeps of any Eurasian population examined in this study, with both populations descending from post-Bronze Age mixing events that introduced distinct ancestry components (from Northern African and Siberian groups, respectively) that were not apparent in the other surveyed Eurasian populations.

Although our analyses point to well-known admixture events during the Holocene as the prime driver of the diluted sweep signals observed in modern European genomes, it is possible that the three populations directly ancestral to present-day Europeans (that is, Mesolithic hunter-gatherers, Anatolian Neolithic farmers and pastoralists from the Pontic–Caspian steppe) were also admixed to some degree. However, the much stronger genetic differentiation observed between the three ancestral populations relative to the later Holocene European groups suggests that potential admixture events involving the three ancestral lineages were probably less influential or frequent than subsequent admixture phases in the Holocene. This implies less perturbation of any underlying sweep signals, although we note that the occurrence of such mixing events would mean that historical hard sweeps were even more frequent than identified in our study.

In addition to masking historical sweeps in human populations, the obscuring effect of admixture might explain why species-wide selective sweep signals are rare in many species while being abundant in others. Crucially, recently admixed populations often lack detectable signs of structure, in which case admixture will not be correctly accounted for in any subsequent selection tests. For example, modern European populations have been considered to be sufficiently genetically homogeneous for the purpose of selection scans, despite ancient DNA studies revealing that Europeans have multiple diverged ancestry components. Indeed, modern genomic data is often insufficient to establish past admixture events, with widely used principal component analysis (PCA) and ancestry decomposition methods being unable to detect historical admixture signals when suitable proxy populations for the admixing groups are lacking. Similarly, although admixture creates temporal variation in genomic coalescence rates, these patterns can be equally well explained by historical population size changes in a single continuous population. Accordingly, species with little apparent population structure may be more susceptible to the confounding effects of admixture on selection scans than those in which structure is evident and can be directly accounted for, and this may partly explain why species with distinctive population structuring often show strong local selection signals (for example, Swedish Arabidopsis thaliana, African Drosophila melanogaster and Microbotryum lychnidis-dioicae), whereas genetically homogeneous taxa or populations tend to lack fixed hard sweeps but harbour abundant partial or soft sweep signals (for example, European humans, North American D. melanogaster and M. silenes-dioicae).

In accordance with previous work, our results emphasize the importance of incorporating historical population structure and admixture events into the null models of selection tests. If this information is not available—for example, because DNA from ancestral source populations is lacking, the default scenario for most species—then the interpretation of historical selection signatures could be highly misleading and heavily weighted against the detection of historical hard sweep events. Although these factors imply that the extent of past hard sweep events has probably been underestimated in natural populations in general, we caution that they do not directly challenge results from previous studies proposing a substantial role for soft sweeps in the adaptive history of human and other species. Rather, we reiterate the conclusions of other recent work that advocate for improved modelling of complex but widespread evolutionary and demographic processes to achieve an unbiased understanding of the mode and tempo of adaptation in natural populations.