I assume this pertains to Wolfram|Alpha.
Try replacing the 1/3 power with CubeRoot[ ((x+2)(x^2+4x +1))].
It's not a bug but rather a judgement call as to whether that power should be interpreted as the real-valued CubeRoot function, or the complex-valued principal third root (as written).
By the way, the plot is not identical to that of the square root, though they are similar.
