2

I am struggling a bit with an analysis I need to do. I have collected data consisting of little owl calls that were recorded along transects. I want to analyse these recordings for similarity, in order to see which recorded calls are from the same owls and which are from different owls. In that way I can make an estimate of the size of the population at my study area.

I have done a bit of research and it seems that the package warbleR seems to be suitable for this. However, I am far from an R expert and am struggling a bit with how to go about this. Do any of you have experience with these types of analyses and maybe have example scripts? It seems to me that I could use the function cross_correlation and maybe make a pca, however in the warbleR vignette I looked at they only do this for different types of calls and not for the same type call from different individuals, so I am not sure if it would work.

J_rite
  • 572
  • 1
  • 12
  • 28
Roos
  • 31
  • 2
  • A usual approach would be to start with a controlled group, that means you must have sets of recordings where you know 100% sure that the samples within the set belong to the same individual. From there you can try and test the package you mention to see if with any of their methods you can seperate them apart such that you can prove the software you use is able to distinguish them on a individual level. Only then you can use it in the field to use it "reversed", i.e. assuming that those recording with the same similarities can reasonably be expected to belong to the same individual or not. – Merijn van Tilborg Apr 29 '22 at 10:58
  • Thank you! That is true indeed. However, I am mostly struggling with the R syntax and making the R program do what I want it to do. For example, I already don't really now how to load the data into R in such a way that they can be used in R. – Roos Apr 29 '22 at 11:55
  • Hi Roos, I removed the 'i hope you can help me' and 'thanks' part of the question. These are not used on this site. – J_rite Apr 29 '22 at 13:58

1 Answers1

0

to be able to run analyses with warbleR you need to input the data using the "selection_table" format. Take a look at the example data "lbh_selec_table" to get a sense of the format:

library(warbleR)

data(lbh_selec_table)

head(lbh_selec_table)

The whole point of these objects is to tell R the time location in your sound files (in seconds) of the signals you want to analyze. Take a look at this link for more details on this object structure and how to import it into R.

Marcelo Araya-Salas
  • 101
  • 2
  • 4
  • Thank you! I indeed now made a selection table. However, each of the sound files I have consists only of one individual call (and is thus very short), so the column 'selec' seems a bit redundant. – Roos May 06 '22 at 09:20
  • 1
    Yeah but the column 'selec' is still expected by the package. It doesn't hurt to keep it. – Marcelo Araya-Salas May 06 '22 at 16:41