These histograms plot the fraction of reconstructed charged pions in each pion pT bin arising from gg, qg, and qq scattering.  I use the following cuts:

• fabs(mcVertexZ()) < 60
• fabs(vertexZ()) < 60
• geantId() == 8 or 9 (charged pions)
• fabs(etaPr()) < 1
• fabs(dcaGl()) < 1
• fitPts() > 25

I analyzed Pythia samples from the P05ih production in partonic pT bins 3_4 through 55_65 (excluded minbias and 2_3).  The samples were weighted according to partonic x-sections and numbers of events seen and then combined.  StEmcTriggerMaker provided simulations of the HT1 (96201), HT2 (96211), JP1 (96221), and JP2 (96233) triggers.  Here are the results.  The solid lines are MB and are identical in each plot, while the dashed lines are the yields from events passing a particular software trigger.  Each image is linked to a full-resolution copy:

Conclusions

1. Imposing an EMC trigger suppresses gg events and enhances qq, particularly for transverse momenta < 6 GeV/c.  The effect on qg events changes with pT.  The explanation is that the ratio (pion pT / partonic pT) is lower for EMC triggered events than for minimum bias.
2. High threshold triggers change the subprocess composition more than low-threshold triggers.
3. JP1 is the least-biased trigger according to this metric.  There aren't many JP1 triggers in the real data, though, as it was typically prescaled by ~30 during the 2005 pp run.  Most of the stats in the real data are in JP2.