Clustered star formation as a natural explanation of the Halpha cutoff in disc galaxies

From Pflamm-Altenburg and Kroupa (arXiv:0905.0898v1)
Caption reads:
"The Hα-luminosity surface density versus the total gas surface density observed for seven disc galaxies15 averaged over annuli at different galactocentric radii is plotted (black squares) after correcting for photon leakage from H ii regions (see Supplementary Discussion). These galaxies have a mean star formation rate of SFR=6.9 M⊙ yr−1 (3.2 – 16.4 M⊙ yr−1 )2, 15 , a mean total gas mass of Mgas = 2.1 · 1010 M⊙ (0.6 – 3.6 · 1010 M⊙ )2, 15 and a mean scale length of rd = 4.4 kpc (3.9 – 5.2 kpc)25–28 . These mean values define our model standard disc galaxy. For a choice of γ = 2 the LIGIMF-theory predicts an ΣHα -Σgas relation which matches the observations excellently (solid line). Note that the underlying true star-formation density as derived from UV observations1 is directly proportional to the gas surface density (N = 1) and is shown after converting it into an Hα surface luminosity using the wrong linear Kennicutt Hα-SFR relation2, 29 (dashed line) and shows the expected ΣHα -Σgas relation based on the classical picture which is in disagreement with the observations."
Basically, since stars form in clusters, and because you have lower mass star clusters at lower densities, you expect relatively less massive stars and therefore a Hα cutoff