When we extract species-level information from complex multi-species soundscapes, how does the acoustic complexity index (ACI) of the recording environment affect the reproducibility of our extraction results? Are simpler soundscapes more reproducible, or does complexity provide redundant information that actually improves reliability?