Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) variants are clusters of mutations that create distinct genetic lineages of SARS-CoV-2. Several such lineages have spread widely and developed new phenotypic properties, altering transmissibility, pathogenesis, diagnosis, treatment, and prevention in ways that have repeatedly reshaped the coronavirus disease 2019 (COVID-19) pandemic. In this chapter, we describe the emergence of viral variants and highlight their origin, emergence, impact on diagnostics, therapeutics, and vaccination. We argue that the lens of evolutionary biology and population genetics provides the clearest perspective through which to understand viral variants. Variants have impacted the pandemic through repeated selected sweeps of combinations of mutations that provide a fitness advantage. In large populations, where selection dominates over genetic drift, these fitness advantages behave as deterministic effects on the change in allele frequency over time in populations, resulting in competitive displacement of preexisting lineages. The mechanisms underlying fitness advantages are various, incompletely understood, and include enhanced transmissibility, increased angiotensin-converting enzyme 2 (ACE2) binding, increased furin cleavage, and immune escape. Fitness is time-dependent, with an increasing emergence of antibody escape mutations arising because convalescent and vaccine-induced immunity have become widespread. Understanding the theoretical principles and empirical patterns of variant evolution through this perspective provides insight into the mutational drivers of fitness, where variants may have come from, and their role in current and future pandemics.

The emergence of SARS-CoV-2 variants has arguably been the most unexpected and consequential development of the COVID-19 pandemic. When SARS-CoV-2 first emerged in December 2019, viral evolution was not expected to play an important role in the outbreak, largely for two reasons. The first reason was that coronaviruses encode proofreading exoribonucleases; thus, it was expected that the rate of SARS-CoV-2 evolution would be slow relative to other highly transmissible RNA viruses like influenza A virus. The second reason is that viral genetic variation was not known to have an important role in prior outbreaks. Both justifications turned out to be erroneous.

The substitution rate of SARS-CoV-2, once measured, was not too different from other positive-sense RNA viruses,¹ and mutations soon began to emerge in consequential areas of the virus. During March 2020, several groups noted the emergence of a Spike glycoprotein mutation at position D614G.^{2, 3 and 4} The rate of growth of variants with this mutation appeared to be higher than the rate of growth of strains bearing the ancestral residue.² During the first year of the pandemic, there was ongoing skepticism about the relevance of viral variation in the SARS-CoV-2 outbreak (perhaps best exemplified by reference⁵), but extensive scientific evidence emerged through 2020 to document the important consequences of genetic variation in SARS-CoV-2. Structural and functional studies revealed that S:D614G alters the conformation and dynamics of the Spike protein in important ways that increase infectivity.^{2,3,6, 7, 8, 9 and 10} Rates of evolution differ across the proteins in SARS-CoV-2, with the highest rates of evolution occurring in Spike.¹¹

By late 2020, the importance of viral variation was clearly established.^12,13 Early reports suggested both increased transmissibility/growth rate¹⁴ and escape from plasma neutralization responses.^15,16 This change in perspective was catalyzed in part by the emergence of a stereotyped triad of nonsynonymous mutations at Spike positions 501, 417, and 484 that occurred independently in a variety of lines.^{14,17, 18 and 19} These mutations occurred in the Pango B.1.1.7/WHO Alpha variant, first identified in England; the B.1.351/Beta variant,²⁰ first identified in South Africa; and the P.1/Gamma variant,¹⁹ first identified in Brazil. A tree of major variants is shown in Figure 5.1. The earliest signal of increased fitness came from the United Kingdom, where high quality of genomic surveillance enabled the first region-by-region analysis of growth.²¹ The Alpha variant was found to grow quickly in all seven UK subregions, suggesting that it harbored intrinsic fitness increases relative to the variants that it replaced. This increase in fitness was later estimated at approximately 70% compared to the wild-type SARS-CoV-2.^14,22

Viral variants reflect the real-time evolution of SARS-CoV-2; thus, in order to understand how variants spread, it is important to consider evolutionary theory. Selection and drift are the two most fundamental evolutionary forces at play in natural populations. Genetic drift is a stochastic force that results in changes in allele frequency between generations and derives from the sampling of alleles between generations. Because genetic drift is a result of sampling, its effects are magnified in small populations. In contrast, selection is a deterministic force that increases the probability of survival of an allele from one generation to the next. As a consequence of this dichotomy, selection and drift act differently based on population size. Selection dominates in large populations (reviewed, for example, in reference²³). Because SARS-CoV-2 populations are large, major changes in allele frequency in the population have been driven by selection. One may observe this by first noting that population genetic theory predicts that the relative frequency of a selectively advantageous variant relative to others will follow a logistic growth curve.²⁴ Such trajectories have repeatedly been observed in practice (eg, Figure 5.1 and references^14,25,26). Thus, the way in which SARS-CoV-2 variants repeatedly outcompete one another is that populations are large and new variants contain relative fitness advantages over previous variants. This has been exploited in the widespread use of logistic regression to estimate fitness advantages.^14,25,26

Fitness advantages may occur for a variety of reasons. Selective forces act on individuals to increase their reproductive fitness in the population. This general concept will be true at all times; however, the mechanisms by which evolutionary change enhances fitness are multiple and appear to have changed during the pandemic. During the initial period of the pandemic, when the majority of the population had not been infected or vaccinated, variants evolved mutations to increase their ability to transmit person-to-person or to infect cells, for example, by increasing the affinity to the ACE2 receptor,²⁷ enhancing the efficiency of S1/S2 cleavage,²⁸ or promoting the efficiency of viral genomic RNA packaging.²⁹ Spike mutations such as N501Y and N501T, which are major contact residues with ACE2, increase the affinity to the ACE2 receptor.²⁷ Spike mutations such as R681R, at the S1/S2 cleavage site, increase the efficiency of cleavage by host furin-like proteases.²⁸ Mutations in the Nucleocapsid (N) protein, such as R203K and nearby substitutions, enhance the efficiency of viral genomic RNA packaging.²⁹ The Spike D614G mutation enhances the infectivity of SARS-CoV-2 by altering the structure of Spike and the dynamics of its conformational changes, as well as increasing steady-state levels of S1 on virions. These changes were all seen early in the pandemic when evolutionary pressure from immunity—both convalescent immunity and vaccine-induced immunity—was limited.²⁶

By the end of 2020, there started to emerge a second class of mutations that promoted immune evasion.¹⁶ Mutations in this class included the Spike E484K mutation,^30,31 the delta69-70 mutation,³² and other mutational changes in the Spike N-terminal domain. Many of these mutations were first detected in immunocompromised hosts in response to weak antibody pressure.³³ There is a theoretical basis for accelerated rates of evolution in response to weak evolutionary pressure³⁴ (see Figure 5.2 and note 29 in reference³⁴): When selective pressure against a virus is strong, the virus

is likely to die out quickly and limited change will take place. When selective pressure is weak, selective forces will take a very long time to introduce changes in the population and therefore play a limited role. At intermediate values of selection, the rate of evolution is maximal. Thus, early studies of chronically infected hosts have previewed important mutational changes that were subsequently observed in variants of concern (VoCs) and served as a breeding/training ground for future variants with functionally important changes, especially with regard to immune escape.^{32,33,35, 36 and 37} These mutations preferentially affected the binding of neutralizing antibodies that appear to be critical for blocking transmission; in contrast, T-cell responses have, to date, been relatively preserved.^38,39

During 2021, as immunity in the population began to increase because of a greater number of convalescent cases and the rollout of effective vaccines, the rates of antibody escape mutations increased.^26,40 From a broader perspective, these observations indicate that the pandemic underwent a “phase change” in late 2021 when immune escape became the dominant factor driving variant evolution. The apotheosis of this new era of SARS-CoV-2 immune escape was the BA.1 lineage, with its extensively substituted Spike, containing over 30 mutations (Figure 5.3) that produced almost complete escape from neutralizing antibodies. This antibody escape effectively increased the population of susceptible hosts by enabling reinfections and greatly increasing the probability of infection among vaccinated individuals. As a consequence, the Omicron BA.1 lineage spread widely across the globe. The ability of SARS-CoV-2 to accomplish such a feat was presaged by the creation of a highly substituted Spike, termed PMS20, designed to avoid all antibody responses.⁴¹ The resemblance of the BA.1 Omicron lineage to PMS20 with many of the very same mutations, many of which had previously been discovered through long-term passage of Spike glycoprotein in a replication-competent vesicular stomatitis virus (VSV) backbone,⁴² highlighted the role of convergent evolution and the importance of key mutations for escaping monoclonal antibody (mAb) responses.

The relative fitness advantage conferred by particular constellations of mutations produces rapid competitive displacement of other circulation lineages. This pattern is known as a selective sweep and has been repeatedly observed during the pandemic (eg, Figure 5.2 and references^19,21,25). These selective sweeps have several notable consequences. First, they can be exploited to estimate relative fitness advantages of emerging variants.^25,26,43 Second, the rate of change of the logistic growth curve for allele frequencies is maximal at 0.5. This explains the phenomenon that even highly fit variants such as Alpha, Delta, or Omicron tend to circulate at low levels for a long period of time after their first detection before seeming to “take over” in a matter of days or weeks. This behavior is simply a consequence of the dynamics of allele frequency change when there is a relative selective advantage of one genotype over another during exponential growth. Third, not all mutations that compromise the constellation of genotypes will have a selective advantage. Mutations with neutral and negative selective advantage can “hitchhike” to increased allele frequency when linked with other, advantageous mutations. Because recombination is rare and requires coinfecting genotypes, all mutations in a variant are tightly linked. In fact, some of the mutations in variants are likely disadvantageous. The accumulation of mutational burden of such disadvantageous mutations (termed Muller ratchet⁴⁴) may be one of the reasons for the recent proliferation of recombinant genotypes of SARS-CoV-2.^45,46

The origin of SARS-CoV-2 is not clearly known, but certain characteristics shared across variants have provided clues to how and where they might have evolved. A hallmark of SARS-CoV-2 emerging variants has been the large jump in the number of mutations that they possess. This is sometimes shown by plotting their evolutionary clock against the overall epidemic’s clock. The rate of accumulation of genetic diversity of novel variants occurs at the same rate (same slope) but the intercept is markedly different.¹⁷ These variants are said to be “off the clock.” Equivalently, variants evolve along a “long branch” for which there are often no intermediate, precursor sequences. The lack of such intermediates has made it difficult to ascertain the origin of variants. A number of competing hypotheses have been proposed.