Think about it this way. Assume a cambered airfoil, not a symmetrical airfoil. Assume the wing has zero twist and zero incidence. To place the wing at the zero-lift angle-of-attack (zero CL), the fuselage and wing will have to fly at a somewhat nose-down pitch attitude relative to the airflow. This does not yield the lowest possible drag coefficient, regardless of whether we are looking at the wing, the fuselage, or the whole aircraft.
Bear in mind that in this part of the flight envelope (near the zero-lift angle-of-attack), drag is dominated by profile drag, not induced drag.
Related: Is drag coefficient lowest at zero angle of attack?
Now imagine using the same data to generate a graph of CD versus airspeed, assuming Lift=Weight. What would that look like? Also what would a graph of Drag versus airspeed look like?