Abstract
We investigated whether field assessments of honey bee (Apis mellifera L.) colony health explain subsequent colony size, honey production, and survival. Field detections of visible diseases, Varroa destructor (Anderson and Trueman) and queen replacement events were recorded during a multisite cohort study, which also incorporated fumagillin and protein supplementation as colony-level treatments. Together, treatment groups and field observations explained between 5% of the variability in adult bee counts and 28% of the variability in honey production among colonies, after accounting for the effects of region and date. In particular, detections of minor disease symptoms, mainly chalkbrood, were associated with large reductions in honey production and approximately doubled the short-term probability of colony death. Although the effects of treatments and field-observed events were significant, unexplained variability among similarly managed colonies was much greater. Consequently, beekeepers may be unable to detect the effects of these field-observable factors, or distinguish effective treatments from ineffective ones. Despite this, interventions to reduce the prevalence of varroa and visible diseases, and to prevent queen loss, are likely to improve honey bee health and productivity.