Dependence of optical Active Galactic Nuclei identification on stellar population models

We have conducted a study to quantify the systematic differences resulting from using different stellar population models in optical spectroscopic identification of type II AGN. We examined the different AGN detection fractions of 7069 nearby galaxies (z <= 0.09) with SDSS DR8 spectra when using the Bruzual & Charlot (2003, BC03), Vazdekis et al. (2010, MILES), and solar metallicity Maraston and Stromback (2011) (MS11solar) stellar population models. The line fluxes obtained using BC03 and MS11solar are publicly available from SDSS data releases. We find that the BC03 templates result in systematically higher BPT line ratios and consequently higher AGN fractions and the MS11solar templates result in systematically lower line ratios and AGN fractions compared with the MILES templates. Using MILES as the standard, BC03 results in 25% "false positives" and MS11solar results in 22% "false negatives" when using the Kewley et al. (2001a) boundary for AGN identification. The fraction of galaxies whose AGN identification changes for different templates is luminosity dependent, ranging from a few percent for L[OIII]5007 >= 10^40 erg s-1 and increasing to ~ 50% for L[OIII]5007 <= 10^38 erg s-1. These results suggest that template choice should be accounted for when using and comparing the AGN and emission line fluxes from different catalogs.
Comment: 21 pages, 20 figures, 1 table; accepted for publication in The Astrophysical Journal