As a cornerstone of disease ecology and evolution, the virulence--transmission trade--off hypothesis guides numerous research agendas. The hypothesis predicts that pathogens with intermediate virulence levels will have the highest transmission rate, with empirical and theoretical work often focusing on the relationship between transmission and pathogen replication. In order for virulence--transmission trade--off to hold true, two other relationships must also be true. First, virulence increases linearly as within-host replication increases. Second, pathogens with intermediate levels of replication will have the highest transmission rate. Recent narrative reviews suggest that an increasing number of empirical studies support these predictions and , thus, bring us closer to making generalizations regarding this trade--off. Yet, others have found contradicting evidence. To quantitatively assess the generality of the trade--off predictions and to identify gaps for future research, we conducted a Bayesian meta-analysis of the literature. Unlike most meta-analyses that rely on summary statistics derived from the data, we chose to directly test the predictions of the virulence--transmission trade--off using the raw data, which allows a more robust assessment of the type of functional relationship that best describes the data. We collected data from a total of 22 studies on 22 host pathogen-systems that met our criteria of inclusion. We compared three model types: an intercept-only, a linear and a polynomial model in their ability to parsimoniously explain the relationships between replication and virulence, replication and transmission, and virulence and transmission.
We found strong support for a linear model explaining the relationship between within-host replication and virulence, and moderate support for the decelerating polynomial relationship between within-host replication and transmission rate. Even though the polynomial model was the most parsimonious explaining the relationship between virulence and transmission rate, the average trend did not decelerate and the slopes were highly uncertain. Our results support the linear relationship between pathogen replication and virulence but suggest, in spite of optimism from previous narrative reviews, more studies are required to generalize the predictions of the virulence--transmission and the replication--transmission relationships. Overall, we provide the first quantitative synthesis that shows support for two out of the three predictions of the trade-off hypothesis. While we show that more data are needed, the cornerstone that the virulence--transmission trade--off represents appears strong and still forms a solid and common framework to understand the links between disease virulence and transmission.