A knot/link diagram can be checkerboard colored (as in, the regions between the strands can be two-colored), and a checkerboard surface is what you get by taking all the regions of a particular color and attaching twisted bands at the crossings; the boundary of a checkerboard surface is the link. Every link diagram has two checkerboard surfaces $\Sigma_1$ and $\Sigma_2$. (Mind that the diagram is in $S^2$, so one of them contains $\infty$ as it were.)
In a paper by Joshua Howie ("A characterization of alternating knot exeriors"), he says that by "a simple Euler characteristic argument" $$\chi(\Sigma_1)+\chi(\Sigma_2)+n=2$$ where $n$ is the number of crossings in the diagram.
I am not sure what argument he had in mind. I came up with one which I will post as an answer, but I am afraid I am missing something about the Euler characteristic, and I would appreciate another argument.
I ran into Josh Howie a few months ago and he told me the argument he had in mind. At this point, I don't exactly remember what he said, but the following shouldn't be too far off:
Consider triangulations of $\Sigma_1$ and $\Sigma_2$ which are compatible at the crossings in that they both have an edge spanning the crossing.
The union of $\Sigma_1$ and $\Sigma_2$ is homotopy equivalent to a sphere, so $\chi(\Sigma_1\cup\Sigma_2)=2$.
$\chi(\Sigma_1)+\chi(\Sigma_2)$ double counts those $n$ spanning edges, and since the number of edges is subtracted in the formula for Euler characteristic, $\chi(\Sigma_1)+\chi(\Sigma_2)+n=2$. (It also double counts the link, but Euler characteristic of a link is $0$.)