RR:C19 Evidence Scale rating by reviewer:
Potentially informative. The main claims made are not strongly justified by the methods and data, but may yield some insight. The results and conclusions of the study may resemble those from the hypothetical ideal study, but there is substantial room for doubt. Decision-makers should consider this evidence only with a thorough understanding of its weaknesses, alongside other evidence and theory. Decision-makers should not consider this actionable, unless the weaknesses are clearly understood and there is other theory and evidence to further support it.
The central claim of the article is that within-host recombination of SARS-CoV-2 has rapidly increased after the emergence of the Omicron clade, positing a connection with the selection of mutations associated with enhanced viral fitness. This is done by analyzing sequencing data belonging to the 31 Pango-designated recombinant lineages, considering their prevalence, geographical location, and dates of emergence and circulation. While analyzing the properties of these lineages is certainly worthwhile, my main concern is that this analysis does not adequately support the broad conclusions drawn about the role of recombination in SARS-CoV-2 evolution, mainly the suggestion that Omicron may have higher recombination rates than other lineages.
Viral recombination occurs during the co-infection of a host with multiple different strains. For the recombinant to become detectable, the parent strains must also be sufficiently genetically distinct, to create a discernible mosaic. Further, the recombinant strain must be sampled and sequenced, which generally requires onward transmission. The detection of the recombinant strain is then generally reliant on detecting patterns of mutation incompatible with a tree-like phylogeny (which is confounded by the presence of selection and recurrent mutation). Therefore, while it is well known that recombination is common among SARS-like viruses (and has been shown to be commonplace in-vitro for SARS-CoV-2 specifically), the detection of recombination was particularly challenging early on in the pandemic, due to insufficient levels of accumulated genetic diversity. In addition to the difficulties with detecting recombinants, the Pango designation rules  generally require that the recombinant lineage reaches a threshold level of frequency, or shows epidemiological significance, and contains recently sampled sequences. Thus, the set of recombinant lineages that receive a Pango designation will be a small and biased sample of all recombinant strains that have been in circulation. This can be expected to create bias in the analysis of the timing of the emergence of recombinants, and in the analysis of recombination breakpoints.
While the earliest Pango-designated recombinant lineage arose in July 2020, there is evidence that many recombinant lineages were in circulation during the first year of the pandemic , but at low frequencies (which would likely not have met the Pango designation criteria). I am thus not convinced that the number of Pango-designated recombinant lineages necessarily meaningfully represents the number of recombinant lineages circulating in general. Since the start of 2022, the dominant SARS-CoV-2 lineages worldwide have been different sub-clades of Omicron, co-circulating for reasonably long periods of time at middling frequencies and high prevalence . It thus seems entirely plausible that the rise in the number of detected recombinants between Omicron lineages around this time was due to these factors, combined with a general increase in genetic diversity and in detection efforts (rather than any inherent genetic properties of Omicron, as claimed in the article). The authors state that "increased recombination events observed during the Omicron wave were not due to higher genomic sampling or co-circulating genetic diversity", but this is not sufficiently substantiated by the evidence presented. The authors also find an enrichment of recombination breakpoints in ORF1ab and 3’UTR, which is somewhat at odds with other published work: for instance , who found a peak around the S gene (analyzing all recombination events detected in a sample of 1.6m genomes).
As more minor points: parent lineages for each designated recombinant strain are listed in Pango lineage reports  (where possible to identify), and the breakpoint positions in each corresponding GitHub issue. The authors use established tools to essentially replicate these findings but should compare the results to those already in existence. I am also struggling to see the reasoning for the phylogenetic analysis of recombinant strains presented in Figure 3. Due to the presence of recombination, the relationships between the different strains must be captured by a network rather than a tree (with different parts of the recombinant genomes being closely related to different other strains), so it is difficult to interpret the usefulness of this analysis.
 Pango lineages: Guidelines for suggesting novel and recombinant lineages, https://www.pango.network/pango-lineages-guidelines-for-suggesting-novel-and-recombinant-lineages
 VanInsberghe, D., Neish, A. S., Lowen, A. C., & Koelle, K. (2021). Recombinant SARS-CoV-2 genomes circulated at low levels over the first year of the pandemic. Virus Evolution: 7(2), veab059.
 Genomic epidemiology of SARS-CoV-2 with subsampling focused globally since pandemic start, https://nextstrain.org/ncov/gisaid/global/all-time
 Turakhia, Y., Thornlow, B., Hinrichs, A., McBroome, J., Ayala, N., Ye, C., ... & Corbett-Detig, R. (2022). Pandemic-scale phylogenomics reveals the SARS-CoV-2 recombination landscape. Nature: 609, 994–997.
 Pango Lineages: Latest epidemiological lineages of SARS-CoV-2, https://cov-lineages.org