The advantage of converting to PDF is that the time to crank through the extra points happens on the local machine in creating the PDF, rather than tying up the network printer for 1/2 hour to print. I've never fiddled with the default PDF settings with Acrobat on my machine (it looks like choices include "high quality", "press quality", "standard", and "smallest file size"), I've been using "standard" (600 dpi, and several different resolutions for different image types - monochrome, color) and that has generally been given good print quality and usually a pretty significant file size reduction.
Decimation can be a good technique, but it is definitely data dependent. For instance you can decimate before the FFT, but only if the signal BW is within the Nyquist BW of the lower sample rate. To do this the data is filtered digitally and every nth sample is picked off to lower the sample rate. It is also possible to drop samples after the FFT, but this may (or may not) result in visible changes to the graph.
Given the availability of free PDF converters, I would give that a try first and see how that works before trying decimation.
Peter