We compare the chemical abundances at the sites of 12Â nearby ( z < 0.14 ) Type Ic supernovae ( SN Ic ) that showed broad lines , but had no observed Gamma-Ray Burst ( GRB ) , with the chemical abundances in 5 nearby ( z < 0.25 ) galaxies at the sites of GRB where broad-lined SN Ic were seen after the fireball had faded .
It has previously been noted that GRB hosts are low in luminosity and low in their metal abundances .
If low metallicity is sufficient to force the evolution of massive stars to end their lives as GRB with an accompanying broad-lined SN Ic , then we would expect higher metal abundances for the broad-lined SN Ic that have no detected GRB .
This is what we observe , and this trend is independent of the choice of metallicity calibration we adopt , and the mode of SN survey that found the broad-lined SN .
A unique feature of this analysis is that we present new spectra of the host galaxies and analyze all the measurements of both samples in the same set of ways , using the galaxy emission-line measurements corrected for extinction and stellar absorption , via independent metallicity diagnostics of Kewley & Dopita ( 2002 ) , of McGaugh ( 1991 ) and of Pettini & Pagel ( 2004 ) .
In our small sample , the boundary between galaxies that have GRB accompanying their broad-lined SN Ic and those that have broad-lined SN Ic without a GRB lies at an oxygen abundance of 12+log ( O/H ) _ { KD 02 } \sim 8.5 , which corresponds to 0.2 - 0.6 Z _ { \odot } Â depending on the adopted metallicity scale and solar abundance value .
Even when we limit the comparison to SN Ic that were found in untargeted supernova surveys , the environment of every broad-lined SN Ic that had no GRB is more metal rich than the site of any broad-lined SN Ic where a GRB was detected .