Phylogeographic and Host Interface Analyses Reveal the Evolutionary Dynamics of SAT3 Foot-And-Mouth Disease Virus

Abstract

Foot-and-mouth disease virus (FMDV) serotype SAT3 is a rarely studied serotype primarily circulating in southern Africa, with African buffalo (Syncerus caffer) serving as its key reservoir. In this study, we performed a comprehensive phylogenetic and phylodynamic analysis of SAT3 based on 81 full-length VP1 gene sequences collected between 1934 and 2018. Maximum likelihood and Bayesian analyses revealed five distinct topotypes, each with clear geographic and host associations. Notably, topotypes I, II and III were observed in both African buffalo and cattle (Bos taurus), while topotype IV appeared restricted to African buffalo. Likelihood mapping indicated moderate to strong phylogenetic signal, and the mean substitution rate was estimated at 3.709 × 10^-3 substitutions/site/year under a relaxed molecular clock. The time to the most recent common ancestor (TMRCA) was traced back to 1875. Discrete phylogeographic reconstruction identified Zimbabwe as a major center, with multiple supported cross-border transmission routes. Host transition analysis further confirmed strong directional flow from buffalo to cattle (BF = 1631.09, pp = 1.0), highlighting the wildlife-livestock interface as a key driver of SAT3 persistence. Together, these results underscore the evolutionary complexity of SAT3 and the importance of integrating molecular epidemiology, spatial modeling, and host ecology to inform FMD control strategies in endemic regions.

Keywords: SAT3; cross-species transmission; evolutionary; foot-and-mouth disease; phylogeographic.

Figure 1

Likelihood mapping analysis of the VP1 gene sequences from SAT3 FMDV revealed sufficient phylogenetic signal for tree reconstruction. Analysis was based on 10,000 randomly selected sequence quartets using TREE-PUZZLE. Only 3.6% of quartets fell into the central unresolved region, indicating moderate-to-strong phylogenetic signal supporting robust downstream evolutionary inference.

Figure 2

Maximum likelihood phylogeny of 81 non-recombinant SAT3 VP1 sequences revealed seven distinct topotypes with host and geographic associations. The circular tree showed seven well-supported topotypes (topotypes I–VII) with country and host annotations. Bootstrap values ≥ 95% are shown. Sequences from South Africa, Zimbabwe, and Zambia are widely distributed, while Botswana and Mozambique are more topotype-restricted.

Figure 3

Root-to-tip regression plot showing the temporal signal of the VP1 dataset. (a) Full dataset (n = 81); (b) filtered dataset (n = 75). Analyses were performed in TempEst with best-fit root placement. The fitted linear trend (dashed) showed a positive correlation between sampling year and root-to-tip distance, indicating temporal signal in both datasets and supporting the use of relaxed molecular-clock models for evolutionary rate estimation and TMRCA inference.

Figure 4

Time-scaled MCC phylogeographic tree of FMDV SAT3 (n = 75) in southern Africa. Branches are colored by posterior probabilities values (red = high support), and 95% highest posterior density (HPD) intervals for node dates are indicated by horizontal bars. Five topotypes (I–V) are highlighted by shaded backgrounds. Sampling countries are indicated by side color strips, and host species are marked by symbols: Bos taurus (blue squares) and Syncerus caffer (orange circles).

Figure 5

Bayesian Skyline Plot of SAT3 VP1 gene showing effective population size over time. Estimates are based on 75 VP1 sequences under a relaxed lognormal molecular clock with Bayesian Skyline coalescent prior.

Figure 6

Network diagram of SAT3 geographic migration routes. Phylogeographic reconstruction of the spatiotemporal spread of SAT3 foot-and-mouth disease virus (FMDV) in southern Africa. Arrows represent the most supported viral migration pathways between countries, with arrow thickness indicating the strength of support based on Bayes factor (BF > 3, BF > 10, and BF > 100). Colors indicate different countries: Uganda (yellow), Malawi (green), Zambia (blue), Mozambique (cyan), Zimbabwe (light violet), Botswana (pink), and South Africa (purple), with the background in gray representing Africa.

Figure 7

Results of codon-level selection and co-evolutionary analysis of the VP1 coding region. (a) Codon sites under episodic diversifying selection identified by MEME (red) and FEL (blue). The y-axis shows the p-values, with the dashed line indicating the significance threshold (p = 0.05). MEME detected 13 codons (22, 24, 25, 98, 110, 111, 142, 143, 153, 160, 161, 174 and 202) under significant selection, whereas FEL detected no sites that passed the strict threshold but showed borderline signals at codons 44 and 161 (p slightly > 0.05). (b) Co-evolving codon pairs inferred using the Bayesian Graphical Model (BGM) approach. Each node represents a codon site, and edges denote co-evolutionary relationships with associated posterior probabilities. Edge labels indicate the estimated directional and bidirectional probabilities (P[Site1→Site2]/P[Site2→Site1]/P[Site1↔Site2]), highlighting pairs of codons that may evolve in a correlated manner.