Disturbances are increasing globally due to anthropogenic changes in land use and climate. This study determines whether a disturbance that affects the physiology of individual trees can be used to predict the response of the ecosystem by weighing two competing hypothesis at annual time scales: (a) changes in ecosystem fluxes are proportional to observable patterns of mortality or (b) to explain ecosystem fluxes the physiology of dying trees must also be incorporated. We evaluate these hypotheses by analyzing 6 years of eddy covariance flux data collected throughout the progression of a spruce beetle (Dendroctonus rufipennis) epidemic in a Wyoming Engelmann spruce (Picea engelmannii)–subalpine fir (Abies lasiocarpa) forest and testing for changes in canopy conductance (gc), evapotranspiration (ET), and net ecosystem exchange (NEE) of CO2. We predict from these hypotheses that (a) gc, ET, and NEE all diminish (decrease in absolute magnitude) as trees die or (b) that (1) gc and ET decline as trees are attacked (hydraulic failure from beetle-associated blue-stain fungi) and (2) NEE diminishes both as trees are attacked (restricted gas exchange) and when they die. Ecosystem fluxes declined as the outbreak progressed and the epidemic was best described as two phases: (I) hydraulic failure caused restricted gc, ET (28 ± 4% decline, Bayesian posterior mean ± standard deviation), and gas exchange (NEE diminished 13 ± 6%) and (II) trees died (NEE diminished 51 ± 3% with minimal further change in ET to 36 ± 4%). These results support hypothesis b and suggest that model predictions of ecosystem fluxes following massive disturbances must be modified to account for changes in tree physiological controls and not simply observed mortality.