Inferring phenotypic causal networks for tick infestation, Babesia bovis infection, and weight gain in Hereford and Braford cattle using structural equation models

Abstract

Tick infestation and associated diseases (i.e., babesiosis) constitute major drawbacks for improvement of beef cattle productivity in the tropics, mainly when purebred and crossbred taurine animals are used. Host-parasite-pathogen interactions form complex biological systems that are poorly understood and which significantly affect production and quality traits in ways yet to be dissected and described. This research was carried out to investigate potential causal relationships, through the use of structural equation modeling (SEM), among tick counts (TC), Babesia bovis infection level (IB) and the gains in weight: from birth to adjusted weaning age (WG), and from weaning to yearling (YG). Statistical analyses were conducted in three steps: 1) Partition of (co)variances into genetic and residual components using Bayesian multiple-trait modeling (MTM); of 2) Search for plausible causal structures by applying the inductive causation (IC) algorithm to the residual (co)variances obtained in the first step; and 3) Final analysis using SEM, which was based on the causal network learned from the IC algorithm. The most plausible results comprised three direct links between traits: WG→YG, TC→WG, and WG→IB with structural coefficients posterior means equal to -0.3026, 6.3620, and 0.0004, respectively. The final inferred directed acyclic graph (DAG) suggests that interventions on TC would directly affect WG, which would then affected YG; moreover, WG could also present a small positive effect on IB.