When I started check_zpools.sh, I had manually set the warning and critical thresholds which are used to compare against the current pool usage. This is basically how I start a new plugin. If everything works, I move on to the next step and add options (e.g. -w).
Well, it seems that I forgot to remove the manually defined warning and critical thresholds after having added the options ;-).
The new released version now supports to run without any thresholds (so there won't be any usage comparison on the zfs pool).
Thanks to user "xrated" on Nagios Exchange making me aware of this bug.