In [19] it was explained how one can naturally associate a Deitmar scheme (which is a scheme defined over the field with one element, $mathbb{F}_1$) to a so-called loose graph (which is a generalization of a graph). Several properties of the Deitmar
scheme can be proven easily from the combinatorics of the (loose) graph, and known realizations of objects over $mathbb{F}_1$ such as combinatorial $mathbb{F}_1$-projective and $mathbb{F}_1$-affine spaces exactly depict the loose graph which corresponds to the associated Deitmar scheme. In this paper, we first modify the construction of loc. cit., and show that Deitmar schemes which are defined by finite trees (with possible end points) are defined over $mathbb{F}_1$ in Kurokawas sense; we then derive a precise formula for the Kurokawa zeta function for such schemes (and so also for the counting polynomial of all associated $mathbb{F}_q$-schemes). As a corollary, we find a zeta function for all such trees which contains information such as the number of inner points and the spectrum of degrees, and which is thus very different than Iharas zeta function (which is trivial in this case). Using a process called surgery, we show that one can determine the zeta function of a general loose graph and its associated { Deitmar, Grothendieck }-schemes in a number of steps, eventually reducing the calculation essentially to trees. We study a number of classes of examples of loose graphs, and introduce the Grothendieck ring of $mathbb{F}_1$-schemes along the way in order to perform the calculations. Finally, we compare the new zeta function to Iharas zeta function for graphs in a number of examples, and include a computer program for performing more tedious calculations.
