Spillover rate is not predictive of host jump risk

Brandon J Simony,David A Kennedy

DOI: https://doi.org/10.1101/2024.10.14.618291

2024-10-17

Abstract:Spillover of infectious diseases is a significant issue for human, animal, and plant health, in part due to their potential to host jump. Current approaches to predict and prevent host jumps are structured around discovering new viruses, drawing associations between pathogen characteristics and host jump risk, and managing spillover, but prediction and prevention are still elusive. Spillover is a critical step in the process of completing a host jump, so it is often assumed that a pathogen's spillover rate is predictive of host jump risk. However, this may not necessarily be the case. While a pathogen that spills over into a novel host frequently will indeed have many opportunities to establish itself, the lack of success establishing itself in previous spillovers demonstrates a poor ability to do so. In comparison, a pathogen that rarely spills over will have few opportunities to establish itself, but may be well suited to do so, and could, in theory, be more likely to host jump despite these limited opportunities. Here we address the question of which of these poses the greater risk. Using a Bayesian framework, we create a model to quantify the probability that a host jump occurs in some future time interval. This model only depends on the number of spillover events in the past and future, and our uncertainty regarding the probability that a single spillover will result in a host jump. We derive an analytical solution to this model, which shows that when past and future spillover rates are linearly correlated, a pathogen's inherent rate of spillover is a poor predictor of host jump risk owing to two complementary factors. First, the pathogens that pose the greatest host jump risk may be either those that spillover extremely rarely or extremely commonly, with the result depending on our prior uncertainly in the probability that a spillover will result in a host jump. Second, as spillover rate becomes large, the increase in the number of opportunities for a host jump to occur conferred by the increase in the number of spillover events is exactly balanced by a decrease in the probability of success for each individual spillover event. Therefore, we show that spillover rate is not a useful metric for evaluating host jump risk.

Systems Biology