It seems unlikely at first sight to include prostrate slender-stemmed herbs in the same species as shrubs up to 2 m tall and extreme forms are vegetatively very dissimilar even as herbarium specimens. Nevertheless subsp. gazensis is decidedly variable in the Zimbabwe/Mozambique border region and the size and shape of the leaf parts sometimes approximate quite closely. For example F.R. Williams 81A, from Chimanimani (Melsetter) and without clear habit notes, is not certainly referable to either subspecies, though probably subsp. gazensis; conversely Goldsmith 38/68, from the same area, is clearly subsp. herbacea but with more rounded leaflets than usual.