Disentangling contributions of different hosts to disease emergence and spread is highly complex, but critical for improving predictions, surveillance, and response. This is particularly challenging in wildlife, with pathogens often infecting multiple species and data collection being difficult. Using the emergence of Usutu virus (USUV) in the Netherlands as a case study, we demonstrate the use of an Approximate Bayesian Computation framework on diverse data sources to uncover hidden drivers of spatio-temporal wildlife disease emergence. We calibrated single- and multi-host mechanistic transmission models to five types of wildlife surveillance and research data, describing molecular and serological evidence of USUV in wild birds. Although Eurasian blackbirds, the primary sentinel species, were the most severely affected species, our models indicated that USUV could not persist in blackbirds alone. Our data-model framework provided statistical support for additional bird species to have contributed to transmission. This population of bird species is characterised by limited infection mortality, a longer lifespan, and likely further dispersal than blackbirds. Immunity in this reservoir population appears to have protected blackbirds from further USUV-related population decline. Continued surveillance could help identify these reservoir species. Our results underscore the importance of considering multiple host populations to understand outbreak dynamics. Neglecting the multi-host context of disease transmission can impact the reliability of future predictions and the projected impact of intervention strategies.Significance statementUnderstanding which species contribute to disease transmission is critical for improving predictions, surveillance, and response. However, disentangling these contributions is difficult, especially for wildlife diseases where pathogens often infect multiple species and data collection is notoriously difficult. Using Usutu virus (USUV) in the Netherlands as an example, we show how a computational framework can integrate diverse wildlife data sources to uncover hidden contributors to transmission. Our findings show that blackbirds, the most severely affected species, cannot sustain USUV transmission alone. Additional bird species, characterised by lower infection mortality and longer lifespans, are important for transmission and may protect more vulnerable species. This work emphasizes that studying transmission in a multi-host context is crucial for accurate outbreak predictions and effective interventions.