The frequency of the ECG fibrillatory f-waves (Ff) in atrial fibrillation (AF) shows significant variations over time. Cardiorespiratory interactions through the autonomic nervous system have been suggested to play a role in such variations. Here, we tested whether the spatial distribution associated with the release of the parasympathetic neurotransmitter acetylcholine (ACh) could affect the frequency of atrial reentrant circuits. Computational simulations in a human persistent-AF 3D atrial model were performed. We evaluated two different patterns of atrial innervation: ACh release restricted to the area of the ganglionated plexi (GP) and the nerves departing from them, following the so-called octopus hypothesis, and ACh release distributed uniformly randomly throughout the atria. In both cases, ACh release sites occupied 8% of the atria. The temporal pattern of ACh release was simulated following a sinusoidal waveform of frequency 0.125 Hz (respiratory frequency). Different mean levels and peak-to-peak variation ranges of ACh were tested. We found that variations in the dominant frequency Ff followed the simulated temporal ACh pattern in all cases, with Ff modulation being more pronounced for increasingly larger ACh variation ranges. For the tested percentage of ACh release sites (8%), the spatial distribution of ACh did not have an impact on Ff modulation.