Livestock trade network: potential for disease transmission and implications for risk-based surveillance on the island of Mayotte

Abstract

The island of Mayotte is a department of France, an outermost region of the European Union located in the Indian Ocean between Madagascar and the coast of Eastern Africa. Due to its close connection to the African mainland and neighbouring islands, the island is under constant threat of introduction of infectious diseases of both human and animal origin. Here, using social network analysis and mathematical modelling, we assessed potential implications of livestock movements between communes in Mayotte for risk-based surveillance. Our analyses showed that communes in the central region of Mayotte acted as a hub in the livestock movement network. The majority of livestock movements occurred between communes in the central region and from communes in the central region to those in the outer region. Also, communes in the central region were more likely to be infected earlier than those in the outer region when the spread of an exotic infectious disease was simulated on the livestock movement network. The findings of this study, therefore, suggest that communes in the central region would play a major role in the spread of infectious diseases via livestock movements, which needs to be considered in the design of risk-based surveillance systems in Mayotte.

Figure 1

In- and out-strength distributions of the livestock movement network. The in- and out-strength were aggregated over the study period, and their median values were presented as frequency distributions. The low and high ends of the vertical bars corresponded to the minimum and maximum value, respectively.

Figure 2

The livestock movement network in Mayotte from 2007 to 2014. Nodes and edges corresponded to communes and livestock movements between communes, respectively. The size of the nodes represented the number of livestock in the communes in 2010. The arrow and width of the edges represented the direction and the total number of livestock movements between communes over the study years, respectively. The nodes were expressed as pie charts, showing the proportion of communes’ structurally equivalent cluster over the study years.

Figure 3

In-strength, out-strength, betweenness (left y-axis) and closeness (right y-axis) by structurally equivalent cluster over the study years.

Figure 4

Time from disease incursion to commune infection (in days) aggregated by structurally equivalent cluster (Baseline model). For the three values of $p_{\inf }$ tested (0.1, 0.5, 0.9), the number of days from disease incursion to commune infection was aggregated by structurally equivalent cluster (central and outer communes) and plotted in the log scale (y-axis). Numbers in boxplots corresponded to the median values.

Figure 5

Relative time from disease incursion to infection of individual communes, using the commune of Sada as baseline (left y-axis), and percentage of simulations in which the commune was infected (right y-axis) (Baseline model). The time from disease incursion to commune infection was compared for each commune using the commune of Sada as a baseline. The median, and 1^st and 3^rd quartiles of individual communes were compared with the median of Sada and presented by filled points and vertical lines. For each commune, the percentage of simulations in which the commune was infected was presented by hollow points and horizontal lines. Different point shapes and colours represented different $p_{\inf }$ values and structurally equivalent clusters, respectively.

Figure 6

RVF IgG seroprevalence and normalised difference vegetation index (NDVI) by structurally equivalent cluster. Solid vertical lines represented the 95% confidence intervals of RVF IgG seroprevalence of each study year, with points representing the means. Dashed lines represented the medians of NDVI at the 16-day interval, with their background colour representing the range of the 1^st and 3^rd quantiles. NDVI data were extracted from the Moderate Resolution Imaging Spectroradiometer Terra satellite at 250 m spatial and 16-day temporal resolution.