The authors describe a volume-limited survey of all (64 in total) IR-
luminous (L_IR>3e9) galaxies within 15 Mpc with Spitzer-IRS. The
goal is to look for the [NeV] line at 14um, which is considered to be
an unambiguous tracer of AGN activity since the ionization potential
of this line is generally too high to produce in HII regions.
The BPT diagram above shows SDSS galaxies (faint grey points),
galaxies from their sample with no detected [NeV] (black squares),
and those with detected [NeV] (red squares). Dividing lines between
star-forming galaxies, LINERs, and Seyferts are also overplotted.
Although only 7 galaxies strictly meet the "Seyfert" classification
based on their optical emission lines, 17 show [NeV]; the authors
conclude that the BPT diagnostic misses more than half of AGN
activity in IR-luminous galaxies. However, most of the "optical non-
AGNs" classify as LINERs, so it seems a bit much to say they were
totally "missed" by the BPT diagnostic. Also, in a subsequent plot
it's apparent that most of these near-IR AGN are extremely weak -
between 1-5% of the total galaxy luminosity. However, there's
marginal evidence for a correlation between AGN activity and L_IR
(hence star formation) in this sample.