[@Myers2019] - The Acoustic Diversity Index [[Villanueva-Rivera2011][]] is a spectral entropy measure and provides an estimation of bird [species richness][]. It is calculated by dividing the spectrogram into frequency bins and analysing the proportion of information in each bin above a pre-defined threshold [[Villanueva-Rivera and Pijanowski, 2016][]]. The Shannon index is then applied to the bins to give the final ADI value. ADI is calculated by the following formula:ADI=∑i=1Spilnpiwhere pi is the fraction calculated in each frequency bin and S is the number of frequency bins. ADI is shown to be the most appropriate proxy for bird species richness over multiple landscape scales [[Mammides2017][]], but it is important to remember that the index only takes into account the sounds that exist within the audio files, and are not directly related to the bird species per se [[Machado2017][]]. In a study from the tropics, ADI was shown to be one of the best performing indices, whilst ACI and BIO were found to be weakly correlated with bird diversity [[Mammides2017][]].
[@Jorge2018] - Surprisingly, the acoustic diversity index [ADI], which was found to correlate positively with the number of bird species in other complex habitats [e.g. [Machado2016][], [Mammides2017][]], showed the opposite correlation in our case. Usually, the ADI is higher as more frequency bands are filled due to the vocalization of more species [[Villanueva-Rivera2011][]]. In our records, however, the vocalizations of some bird species occupied a wide frequency range [[Table 1][]], resulting in high ADI when there were just few species vocalizing, explaining the negative correlation between the number of bird species and ADI reported here.