The Relationship between CO Emission and Visual Extinction Traced by Dust Emission in the Magellanic Clouds

To test the theoretical understanding that finding bright CO emission depends primarily on dust shielding, we investigate the relationship between CO emission ($I_{\rm CO}$) and the amount of dust (estimated from IR emission and expressed as "$A_V$") across the Large Magellanic Cloud, the Small Magellanic Cloud, and the Milky Way. We show that at our common resolution of 10 pc scales, $I_{\rm CO}$ given a fixed line-of-sight $A_V$ is similar across all three systems despite the difference in metallicity. We find some evidence for a secondary dependence of $I_{\rm CO}$ on radiation field; in the LMC, $I_{\rm CO}$ at a given $A_V$ is smaller in regions of high $T_{\rm dust}$, perhaps because of an increased photodissociating radiation field. We suggest a simple but useful picture in which the CO-to-H$_2$ conversion factor (\xco) depends on two separable factors: (1) the distribution of gas column densities, which maps to an extinction distribution via a dust-to-gas ratio; and (2) the dependence of $I_{\rm CO}$ on $A_V$. Assuming that the probability distribution function (PDF) of local Milky Way clouds is universal, this approach predicts a dependence of \xco\ on $Z$ between $Z^{-1}$ and $Z^{-2}$ above about a third solar metallicity. Below this metallicity, CO emerges from only the high column density parts of the cloud and so depends very sensitively on the adopted PDF and the H$_2$/{\sc Hi} prescription. The PDF of low metallicity clouds is thus of considerable interest and the uncertainty associated with even an ideal prescription for \xco\ at very low metallicity will be large.