There have been a number of Florida cases where the appellate court reversed because scientific evidence didn't meet the Frye test -- Ramirez v. State, 651 So. 2d 1164 (Fla. 1995), for one. If the jury were entrusted to weigh that evidence, there would be no Frye test and no reversals based on it.
I don't know if there are any cases where that error is held to be harmless.
But my point was a different one, and it went to how I speculate Nelson will rationalize allowing the testimony. I don't think the testimony is admissible as expert testimony, just on its face and without going through the sophisticated wordplay that lawyers and judges love so much. The experts can't reach a reliable conclusion, therefore their opinion isn;t helpful to the layman. That the experts disagree, using the same methods! is rock solid evidence of inability to reach a reliable conclusion on this evidence.
As I mentioned above, I don't know, at the conclusion of the case, that the error matters. It certainly does not matter if Zimmerman is acquitted. Which introduces my other thought, that Nelson expects an acquittal, and so is unconcerned about making errors that cut against the defense. The defense wins the trial, there is no appeal on having admitted the state's unqualified expert opinions.