Genomic epidemiology of SARS-CoV-2 variants during the first two years of the pandemic in Colombia

Colombia reported its first confirmed SARS-CoV-2 infection on 06 March 2020. By April 2022, more than 6 million infections and over 135,000 deaths had occurred. The epidemic featured four distinct waves and multiple non-pharmaceutical interventions (NPIs), including mobility restrictions, closures, and vaccination campaigns. Global genomic surveillance has identified variants of concern (VOC) and variants of interest (VOI) with enhanced transmissibility, virulence, or immune escape. However, the dynamics and impacts of specific variants can vary by country. Colombia is notable for the emergence and extensive local spread of the Mu variant (B.1.621). This study investigates, at country scale, how SARS-CoV-2 variants were introduced and spread across Colombia, how their transmissibility (effective reproductive number, Re) and effective population size (Ne) changed over time, and how these dynamics related to public health measures and mobility.

The paper situates its work within global efforts to classify and monitor SARS-CoV-2 genetic diversity using the Pango lineage system and WHO labels for VOCs (Alpha, Beta, Gamma, Delta, Omicron). Prior large-scale sequencing enabled identification of mutations associated with transmissibility and immune escape (e.g., D614G, E484K, L452R, P681R) and supported phylogenomic studies tracking origins, introductions, and spread at global and local scales. While Alpha rapidly became dominant globally after September 2020, Colombia exhibited different dynamics with substantial circulation of other lineages, notably the Mu variant, highlighting the need for fine-scale genomic epidemiology. The study builds on phylodynamic frameworks (e.g., Bayesian Skyline, Birth-Death Skyline) and prior regional analyses (e.g., Brazil’s Gamma emergence) to analyze Colombian variant dynamics.

Sampling and sequencing: Nasopharyngeal swabs were collected from 10,674 residents in Bogotá (Capital District), Cali (Valle del Cauca), and Córdoba (capital and small towns). Samples were tested by RT-qPCR using an in-house protocol targeting the E gene per WHO guidelines. Positive samples with available metadata (travel history, patient status, collection date, vaccination status) were eligible; 610 were selected (86 Córdoba, 122 Cali, 402 Bogotá). RNA was extracted with the GeneJet RNA Extraction Kit (Thermo Fisher K0732). Libraries were prepared following the ARTIC Network protocol and sequenced. Public data: 13,444 Colombian genomes were downloaded from GISAID, yielding 14,049 complete genomes for analysis across all 32 states. Variant assignment used Pango lineage nomenclature; additional quality control and clade assignment followed standard tools (e.g., Nextclade/Nextstrain referenced). Analytical framework: Bayesian phylogenetic and phylodynamic analyses were conducted to infer times of most recent common ancestor (tMRCA) for major variants, effective reproductive numbers (Re), and effective population sizes (Ne). Models included Birth-Death Skyline (BDSKY), Bayesian Skyline, and Coalescent Epoch Plots (BICEPS), enabling time-varying estimates of Re and Ne and comparison across methods. Temporal dynamics were mapped to Colombia’s four epidemic waves and key policy milestones (lockdowns, reopening, vaccination phases). Statistical associations: Linear regressions assessed relationships between Re or Ne (by variant over time) and predictors including mobility, government stringency index, vaccination coverage, mask use, and reported cases (where available). Geographic distribution analyses summarized variant prevalence across Colombian states and the capital district.

- Scope of sequencing and diversity: 14,049 complete genomes analyzed from all 32 states; 188 Pango lineages detected since pandemic start. Colombia contributed the fifth largest volume of SARS-CoV-2 genomic data among South American countries over the period considered. - Temporal waves and dominant variants: Four waves were observed. Early waves were dominated by ancestral lineages (e.g., B.1, B.1.1, B.1.420, B.1.348). In 2021, Gamma was dominant early but was later replaced by Delta. The Mu variant (B.1.621), first detected in Colombia, became dominant during wave 3 (March–July 2021), representing 48.42% of sequenced cases in that period, with Gamma, Lambda, and Alpha at 21.41%, 4.59%, and 0.47%, respectively. Omicron emerged in late 2021 and rapidly displaced prior variants during wave 4. - Transmissibility (Re) ranges and averages (selected periods): During late 2020–early 2021, estimated Re ranges were Delta 0.45–2.68, Gamma 0.40–2.78, Mu 0.73–4.0, Alpha 0.28–2.11, with average Re approximately Delta 1.29, Gamma 1.24, Mu 1.80, Alpha 1.20. During wave 3, average Re were roughly Gamma 1.52, Lambda 1.23, Alpha 0.82, Mu 1.17, Delta 1.37, with peaks observed for Mu and Delta. - Effective population size (Ne) dynamics: After emergence, Ne rapidly increased for several variants and often plateaued; Mu and Gamma showed the highest Ne peaks during wave 3, consistent with their large transmission clusters. Delta’s Ne increased rapidly approaching waves 3–4; Omicron’s emergence led to rapid displacement and dominance in late 2021–early 2022. - Introductions and timing: Multiple introductions of distinct variants were inferred. tMRCA estimates for early lineages indicated circulation near February–April 2020 (e.g., B.1 around 23 Feb 2020; B.1.1 around 27 Feb 2020; B.1.420 around 10 Apr 2020; B.1.348 around 28 Mar 2020). Lambda’s tMRCA was estimated around 15 Dec 2020 despite later detection. - Geographic distribution: The most populated regions (Antioquia, Cundinamarca, Valle del Cauca) contributed the most sequences. Mu showed highest prevalence in Bogotá (≈19.43%) and remained high in Antioquia (≈19%). - Associations with interventions and mobility: Significant positive correlations (p < 0.05) were observed between Re changes and mobility, vaccination, and stringency for multiple VOCs during late 2020–early 2021. Tables of adjusted R-squared and p-values indicated, for example, Delta’s Re correlated with mobility (0.54), stringency (0.40), vaccination (0.36), and mask use (0.47); Gamma’s Re correlated with mobility (0.51), stringency (0.50), vaccination (0.57), mask use (0.50). For Mu during wave 3, Ne positively correlated with case counts, while Re showed no significant association with mobility. - Growth advantage: Mu exhibited a growth advantage over Lambda, Gamma, and Delta during wave 3, consistent with peaks in Re and Ne and partial immune escape discussed by the authors.

The study demonstrates that variant dynamics in Colombia were shaped by multiple introductions, changing public health measures, behavioral factors, and variant-specific properties such as transmissibility and immune escape. The emergence and rapid expansion of Mu in Colombia—coinciding with reduced NPIs and gradual vaccine rollout—help explain why Alpha and Delta did not dominate as strongly during certain periods as seen elsewhere. Significant correlations between Re/Ne and mobility, stringency, vaccination, and mask use show that both policy and behavior influenced transmission dynamics over time. The timing of introductions and inferred tMRCAs, combined with geographic prevalence patterns, illuminate how variants established regionally and then spread nationally. The findings underscore the value of continuous genomic surveillance and phylodynamic modeling to detect and respond to shifts in variant fitness and dominance, informing targeted interventions at national and subnational scales.

This work provides a comprehensive, country-scale phylodynamic reconstruction of SARS-CoV-2 variant introductions, transmissibility, and population dynamics in Colombia across the first two pandemic years. By integrating 14,049 genomes with Bayesian phylodynamic models and epidemiological covariates, the authors show how dominant variants oscillated across waves, highlight Mu’s emergence and growth advantage, and quantify associations between transmission and public health measures. The study supports the implementation of sustained, periodic genomic surveillance to guide timely, evidence-based responses to emerging variants. Future research should expand geographically representative sampling, integrate immunological and vaccine effectiveness data, refine subnational mobility and intervention metrics, and further develop real-time phylodynamic pipelines for rapid situational awareness.

Sequencing coverage varied over time and across waves and regions (e.g., only 0.16% of cases sequenced during wave 3; differing proportions in other periods), which may introduce sampling bias in phylodynamic estimates. Some predictor variables (e.g., mask use, vaccination) were not consistently recorded across the full circulation periods of all variants (NA entries in tables), limiting certain regression analyses. The reliance on publicly available genomes and metadata may entail reporting delays and detection lags. As with observational analyses, correlations with policy and mobility data may be confounded by contemporaneous changes and unmeasured factors.