Genomic analysis of the early COVID-19 pandemic in Haiti reveals Caribbean-specific variant dynamics

Abstract

Pathogen sequencing during the COVID-19 pandemic has generated more whole genome sequencing data than for any other epidemic, allowing epidemiologists to monitor the transmission and evolution of SARS-CoV-2. However, large parts of the world are heavily underrepresented in sequencing efforts, including the Caribbean islands. We performed genome sequencing of SARS-CoV-2 from upper respiratory tract samples collected in Haiti during the spring of 2020. We used phylogenetic analysis to assess the pandemic dynamics in the Caribbean region and observed that the epidemic in Haiti was seeded by multiple introductions, primarily from the United States. We identified the emergence of a SARS-CoV-2 lineage (B.1.478) from Haiti that spread into North America, as well as evidence of the undocumented spread of SARS-CoV-2 within the Caribbean. We demonstrate that the genomic analysis of a relatively modest number of samples from a severely under-sampled region can provide new insight on a previously unobserved spread of a specific lineage, demonstrating the importance of geographically widespread genomic epidemiology.

Fig 1. Fraction of SARS-CoV-2 sequences by nextclade lineage in global dataset versus Caribbean dataset.

(A) Count of samples sequenced from Haiti by their Pango classification. (B) The fraction of SARS-CoV-2 sequences of different lineages in the global dataset of SARS-CoV-2 sequences versus a Caribbean specific dataset (S2 Table). Color represents viral lineage, called by Nextclade [7].

Fig 2. Two most highly represented SARS-CoV-2 lineages from Haiti.

Maximum likelihood phylogenetic trees inferred from SARS-CoV-2 viruses from January 2020 through June 2020. Tips are shaded by Division in Haiti, and country of origin is indicated in the sample name. Triangles represent clades that were collapsed because they did not contain any Haiti data, circles represent samples downloaded from GISAID, and diamonds represent samples sequenced for this study. (A) Includes a cluster of Haiti samples from the pangolin B.1 lineage and (B) includes a cluster of Haiti samples from the B.1.478 pangolin lineage.

Fig 3. Four inter-Caribbean SARS-CoV-2 phylogenetic clusters.

Maximum likelihood phylogenetic trees inferred from SARS-CoV-2 viruses from January 2020 through June 2020. Tips are shaded by region of sample collection, country of origin is indicated in the sample name, and samples sequenced for this study are highlighted in yellow. (A-D) represent four cases of inter-Caribbean SARS-CoV-2 clusters.

Fig 4. International air travel loads before and after the pandemic.

Log of the average number of passengers traveling from an origin country to Haiti are plotted on a world map. Country color indicates the log of the number of passengers traveling to Haiti per month. (A) Plots travel patterns in January to February 2020 and (B) plots travel patterns in April to May 2020.

Fig 5. Ancestral State Reconstruction of B.1.478 lineage.

(A) Clade defining amino acid changes in the SARS-CoV-2 B.1.478 lineage. (B) Bayesian timescale MCMC tree inferred for the cluster of B.1.478 SARS-CoV-2 viruses. Tips are shaded by country of origin, branches are shaded by the state inferred from an ancestral state reconstruction, and samples sequenced for this study are highlighted in yellow. Posterior probabilities are provided for key nodes.