Proximity and density of neighboring farms and water supply, as risk factors for bacteriosis: A case study of spatial risk analysis in tilapia and rainbow trout farms of Oaxaca, Mexico

In this work, we conducted a spatial analysis of three types of diseases: bacteriosis, trichodiniasis and girodactylosis in 336 tilapia and rainbow trout farms during a three-years-period. In order to identify and characterize spatial risk associated with exposure via water for each disease, we used the incidence of cases per type of water supply during the study period (R0), the reciprocal of distance to farms into the same water supply (R1), the inverse distance to farms with an epidemiological background into the same water supply (R2), the reciprocal of distance with an epidemiological background into the same water supply by their prevalence and incidence indexes (R3), and the reciprocal of distance to nearest wastewater domestic discharges that could affect the farm's water supply (R4). For this, through binomial linear multiple regression by generalized linear models (GLM), geographically weighted regression (GWR), logistic multiple regression by generalized additive models (GAM) and Partial Least Squares Path Modeling (PLS-PM); several models were tested to find the best fit to analyze the relationship between the prevalence and incidence index as the outcome variable and the calculated risk factors, as explicative variables. Consequently, the best models were utilized to construct risk maps with predicted relative risks values, employing interpolation methods such as ordinary Kriging and Inverse Distance Weighted Interpolation (IDW). As a result, an ANOVA of the residuals and fitted values from all models showed that the logistic binomial model GAM was the best to analyze all farms globally. The latter allowed predicting in a better way risk zones of bacteriosis in the cases presented during in the next year after the study period; that is to say that was more important to consider the nearness to an infected farm than its prevalence and incidence indexes in its epidemiological background of bacteriosis cases. From there, we found that bacteriosis cases were mainly affected by two of the five risks: the reciprocal of the distance to farms into the same water supply (R1) and the reciprocal of the distance to farms with an epidemiological background into the same water supply (R2). Also, these risk factors were used to estimate the relative risk and establish risk zones, using risk maps in tilapia and rainbow trout farms, whose water supply were lake and river. Finally, the cases of trichodiniasis and gyrodactylosis did not show a relationship to the five risks; accordingly, they could be associated with other not included risks in this work, such as the movement of organisms.