Passive acoustics monitoring (PAM) is a fast-growing ecological field. The data is very useful for monitoring animal presence/absence and studying their vocal behaviour. There are several limitations while using PAM for density estimation such as determining call rate. What are the other challenges to using PAM for animal density estimation?
$\begingroup$
$\endgroup$
1
-
6$\begingroup$ Ooooft, the answer to this could be a thesis....I would suggest perhaps trying to narrow it down a little if you can? - more specifics on species and hardware you are most interested in for example. $\endgroup$– user213Commented Jul 8, 2022 at 9:10
Add a comment
|
1 Answer
$\begingroup$
$\endgroup$
There are many challenges to PAM DE, and these are present both at the technical/hardware level, the logistical level and the analytical level. As Jamie notes, answering this question would be a thesis. In fact, probably several. At a hign level, and to begin with and a bit close to my heart, you need good ways to estimate the effective area of detection of your sensors. That's is fundamental to convert a number of sounds or animals into a density estimate. You can try to do that essentially via 3 paradigms, sound propagation modelling, distance sampling and spatially explicit capture recapture (personally, I much prefer an empirical answer based on plausible assumptions than one from a proagation model). Especially for the latter, you have new methods being proposed at a fast rate, so worth looking into that literature. On the other hand, you need reliable automatic detection and classification of sounds. So that is another area that I suspect will advance tremendosly in the next few years. Machine learning methods are flooding the field and CNNs in particular, where one transforms sound into images, say via spectrograms, and then uses the mnachinery that google et al have developed for image pattern recognition, sound very promissing. Without going into the weeds, I think those are, above and beyhond basic knowledge about the species acoustic behavious - in short, as you mentioned up front, cue rates - the key hurdles we need to overcome to do PAM DE. Mind you, for species for which sound production might be higly variable, ultimately dependent on various factors, and in particular density dependent (creating a circular problem) PANM DE might never work. It's not a panacea, it's just another tool in the applied ecologist toolbox.
