2017ApJ...841..109O

Galactic star formation scaling relations show increased scatter from kpc to sub-kpc scales. Investigating this scatter may hold important clues to how the star formation process evolves in time and space. Here, we combine different molecular gas tracers, different star formation indicators probing distinct populations of massive stars, and knowledge of the evolutionary state of each star-forming region to derive the star formation properties of ∼150 star-forming complexes over the face of the Large Magellanic Cloud (LMC). We find that the rate of massive star formation ramps up when stellar clusters emerge and boost the formation of subsequent generations of massive stars. In addition, we reveal that the star formation efficiency of individual giant molecular clouds (GMCs) declines with increasing cloud gas mass (M_cloud). This trend persists in Galactic star-forming regions and implies higher molecular gas depletion times for larger GMCs. We compare the star formation efficiency per freefall time (ε_ff) with predictions from various widely used analytical star formation models. While these models can produce large dispersions in ε_ff similar to those in observations, the origin of the model-predicted scatter is inconsistent with observations. Moreover, all models fail to reproduce the observed decline of ε_ff with increasing M_cloud in the LMC and the Milky Way. We conclude that analytical star formation models idealizing global turbulence levels and cloud densities and assuming a stationary star formation rate (SFR) are inconsistent with observations from modern data sets tracing massive star formation on individual cloud scales. Instead, we reiterate the importance of local stellar feedback in shaping the properties of GMCs and setting their massive SFR.