Temporal and spatial analysis of Plasmodium falciparum genomics reveals patterns of parasite connectivity in a low-transmission district in Southern Province, Zambia

Abstract

Background: Understanding temporal and spatial dynamics of malaria transmission will help to inform effective interventions and strategies in regions approaching elimination. Parasite genomics are increasingly used to monitor epidemiologic trends, including assessing residual transmission across seasons and importation of malaria into these regions.

Methods: In a low and seasonal transmission setting of southern Zambia, a total of 441 Plasmodium falciparum samples collected from 8 neighbouring health centres between 2012 and 2018 were genotyped using molecular inversion probes (MIPs n = 1793) targeting a total of 1832 neutral and geographically informative SNPs distributed across the parasite genome. After filtering for quality and missingness, 302 samples and 1410 SNPs were retained and used for downstream population genomic analyses.

Results: The analyses revealed most (67%, n = 202) infections harboured one clone (monogenomic) with some variation at local level suggesting low, but heterogenous malaria transmission. Relatedness identity-by-descent (IBD) analysis revealed variable distribution of IBD segments across the genome and 6% of pairs were highly-related (IBD ≥ 0.25). Some of the highly-related parasite populations persisted across multiple seasons, suggesting that persistence of malaria in this low-transmission region is fueled by parasites "seeding" across the dry season. For recent years, clusters of clonal parasites were identified that were dissimilar to the general parasite population, suggesting parasite populations were increasingly fragmented at small spatial scales due to intensified control efforts. Clustering analysis using PCA and t-SNE showed a lack of substantial parasite population structure.

Conclusion: Leveraging both genomic and epidemiological data provided comprehensive picture of fluctuations in parasite populations in this pre-elimination setting of southern Zambia over 7 years.

Keywords: Genomics; Plasmodium falciparum; Transmission; Zambia.

Fig. 1

Theoretical construct for continued transmission in low transmission settings. Malaria transmission in a low endemicity setting is likely the result of a combination of persistence through the dry, low transmission season and importation from other regions. However, the relative contribution of these remains unknown (red question marks). Parasite genomics can help understand these relationships. Different parasite lineages (colours of circles) may or may not survive through a dry season, shown by the reduction in diversity during the dry season. However, genetic diversity may be enhanced through importation through an external source

Fig. 2

Malaria trends and complexity of infections. A Trends of RDT positive cases at the 8 health centres over the course of the study (blue dotted line) and the number of dried blood spot (DBS) samples collected by month over the course of the study (gray bars). Dark green bars show dry seasons (June to August). B Spatial heterogeneity of complexity of infections. C Seasonal variation of multiplicity of infections. The box- and whisker-plots were generated from the median number clones determined per sample. Boxes indicate the interquartile range, the line indicates the median, and the whiskers show the 95% confidence intervals. Dots indicate any outlier values and colours indicate seasons

Fig. 3

Temporal and spatial patterns in pairwise genetic relatedness. A Shows the genetic relatedness between all samples using the F statistic. The insert shows the number of samples with high relatedness (IBD ≥ 0.25). B Shows that isolates collected on the same day and within the same health centre were more likely to be highly related. C Shows a pattern of isolation by increasing geographic distance. D Shows the proportion of highly related parasite pairs shared between health centres. The width of the yellow connecting line corresponds to the proportion of shared highly related parasites. Major roads connecting health centres are shown in white

Fig. 4

IBD network of monoclonal samples with IBD ≥ 0.25 shows transmission across multiple seasons and through the dry season. Networks spanning multiple seasons are circled in blue. The samples are coloured by the season of collection and only monogenomic samples (COI = 1) are included. Thirty-eight samples were contained in networks spanning multiple seasons, with five representing samples collected in the low transmission dry season. Values indicate IBD cutoff and number of samples (n) above the IBD cutoff

Fig. 5

IBD networks of clonal samples (IBD ≥ 0.95) show transmission across seasons and across the study site. A Shows networks across seasons using only monoclonal samples. B Shows the geographic locations of samples from one of these clusters. These locations are connected by lines coloured to correspond to the season of collection. Arrows on these lines correspond to the order in which the samples were collected. The insert represents the cluster from A that is circled in blue and displayed on the map

Fig. 6

Temporal population structure P. falciparum parasite population Choma District, Southern Zambia using PCA (A) and t-SNE (B). Only single representative samples from each highly related cluster were included in both analyses (see Additional file 1: Fig. S9A for analysis with all samples)