Motivation:

As seen in Dan Nemes's presentation  drupal.star.bnl.gov/STAR/system/files/AuAuMeeting-1.pdf

There are peaks in primary track η distribution depending on Vz.  (Dan used MB dataset)

Here is some effort to guess the possible cause

Dataset:

filetype=daq_reco_MuDst,events!=0,trgsetupname=pp200_production_2012,filename~st_physics,production=P13ib,storage!=hpss,sanity=1

370341 -- TOFMult4, 370361 -- TOFmult3*VPD

QA:

Event primary vertex Vz VS Primary track η  

There are some line structures. 

Event primary vertex Vz VS Global track η  

Independent of Vz.