From time to time one sees ‘traditional wisdom’ that inclining the radials of a VHF ground plane to raise its feed point resistance degrades it performance significantly.
I have constructed NEC-4.2 models of a 52MHz ground plane with four 45° inclined radials at 10m height above ‘average ground’ (σ=0.005, εr=13) on and connected to a conductive support pole which is bonded to ground at the lower end, and one with horizontal radials.
Comparing the patterns at low angles shows there is not much in it, but below 32° elevation which tends to be of greater interest at VHF, the winner is actually the inclined radials though the difference is less than 1 dB.
Now the patterns are dependent on height, soil type, common mode current on the masting, and radial inclination… but trials at a range of practical heights did not reveal a significant difference between the configurations.
In this case, the radials needed to be inclined down 58° for a 50Ω match, and the patterns are very slightly different.
Above, a trivial difference between the 45° and 58° patterns at low elevation.
It is one of the many cases of accumulated ham wisdom that does not stand scrutiny with modern analytical tools. This stuff is stated as fact with evidence, and part of the ham tradition is to learn it without question and repeat it like the pros.
Effect of feed point shunt capacitance
Small amounts of lumped shunt capacitance at the feed point will ‘detune’ the antenna and require lengthening of the radiator to achieve minimum VSWR at the desired frequency, and in the process it will raise the feed point R component a little (by virtue of the L network created).
Choosing sufficient C could be used to match a GP with horizontal radials, but even inadvertent small amount of C in an antenna mount might account for a GP matching with lesser inclination of radials that the models above indicate.
Common mode current
The model included a conductive mast from radials to ground. Most analyses of GP antennas ignore common mode current, indeed it seems hams by and large insist that there is no common mode current by virtue of the radials and it is unusual to see counter measured deployed. The common mode current affects pattern to some extent, may contributed to EMC problems, and in the case of HF RX may degrade noise floor.