I am developing a framework for the use of conspecific playbacks to elicit a response in frogs for passive acoustic monitoring. If playbacks were successful in eliciting a response, could this increased detection in acoustic recordings serve to improve population estimates made from acoustic monitoring?
Add a comment
|
3 Answers
$\begingroup$
$\endgroup$
1
If your objective is indeed population/density estimation, then, IMO, eliciting reponses violates the basic assumption that detections are not conditioned by the observation.
Edit: To address a comment to this answer:
Observing frog calls is equivalent to cue counting, and you must then compensate the cue count with the cue rate. Usually the cue rate is estimated independently and typical for the species.
If you elicit frog calls with playbacks then you need an independently obtained relation between playback and call-rate, which may depend on the distance and type of playback and therefore not a simple number.
If this functionality is not known, the population estimated is unreliable. You will get an estimate, but there may be a bias and the correct variance of the population is difficult to obtain.
-
$\begingroup$ Does violating this assumption, if it has been stated, cause the data to become void? The aim here is just to improve detection in passive monitoring, would population estimations not be more accurate with improved detection? $\endgroup$– DajabaCommented Jul 12, 2023 at 10:00
$\begingroup$
$\endgroup$
In general, I would advise against using playbacks to improve detection rate, though most of my research is in the marine realm, and focused on endangered species, so that perhaps colours my view on this.
In my experience, when using acoustic detections for population abundance or density estimation, the detection rate is rarely the limiting factor. Often the variability in acoustic abundance estimates is driven by uncertainty in the cue rate. Cue rate variability cuts both ways, often with errors in understanding the relationship between detection rate and call production rate (perception bias), as well as variability and error in the relationship between cue rate and animal abundance. Using playbacks to improve the number of detections doesn't address either of these sources of error/bias.
Furthermore, using playbacks to trick animals into producing calls may have negative consequences for those animals. This could be in the form of extra energetic costs required to produce calls, or increased risk of predation by not remaining cryptic and hidden. So, unless you're studying an invasive species that you're aiming to eradicate anyway, these extra risks might not be worth the increased number of detections. Especially if the number of detections isn't the largest source of bias/error/variability.
If anyone does decide to follow the suggestion of SM1 that this is a "great question for a research project," Then I hope they'll consider this and perhaps start with a thought experiment/simulation before moving onto to actual playbacks.
$\begingroup$
$\endgroup$
This is probably a great question for a research project in itself.
