I downloaded all recordings flagged in iNat for the bat genus Peropteryx. More than 90% of these recordings are horrible quality, as the users have not been trained in acoustics. Perhaps the equivalent of taking photos through frosted glass
There were 134 files to download and review. Most had incorrect IDs. I am slowly going through each observation to add the correct information. Eventually, I will examine other taxa, but as the GBIF includes iNat records and I am working on a paper on the complete acoustics of the genus, I needed to correct those that are incorrect. I am not at all sure that the original posting people will even note or correct the errors; many were included in projects such as the X-Prize and Map of Life projects.
it would be great if you can write a journal post on iNat explaining to users how to improve the quality of their recordings, and link it with your IDs; that way, future instances of similar low quality recordings will be reduced to at least some degree (for users who do respond to your IDs)
Citizen scientists need scientists to help them contribute to scienceâthey canât do it alone. If scientists want to use citizen science data, theyâve got to do exactly what youâre doingâcurate the observations. Even data in scientific collections contains numerous misidentifications (and must be curated by experts if those experts wish to publish papers that donât contain errors).
Isnât it great there thereâre data/observations there for you to curate and add to your research!
Hello bat_dude! Your efforts in this area are very welcome! iNaturalist seems to have very few people interested in bat acoustics, so best practices havenât really been established or promulgated. I tried adding some bat recordings many years ago,[1][2][3] but no one seemed interested in them so I didnât pursue it further. What are your suggestions for getting good quality bat recordings? Is it helpful to slow down the recordings (e.g. to 1/10th speed) so that they can be heard by humans or is that a bad idea? Should we upload spectrograms or just assume that IDers can generate their own? Any good tips for removing background noise or is that not important?
Hi Zygy, to display and measure important parameters for ID correctly, Bat calls need to be viewed in âbat callâ software. This is not at all compatible with human hearing. I will look at your calls you have posted and see what/how you were recording.
