converting capture history from MARK to secr

[Bryan Hamilton]

I'm switching from MARK to secr. I worked very hard to get my capture histories coded for MARK with '0's and '1's

Code: Select all

            AOU_Code AnimalID StartYear CaptureHistory Gender Trap_Number
5203      PETR      004      2004           1011      F          D4
10931     PEMA      004      2004           1011      F          G3
11520     PETR      004      2004           1011      F          E5
3942      PEMA      005      2004           1000      F          D1
2861      PETR      006      2004           1111      M          A5
4353      PEMA      006      2004           1111      M          D6

Now I need to switch to a line based capture history where each individual capture occasion is represented by a line.

Code: Select all

   AOU_Code AnimalID StartYear CaptureHistory Gender Trap_Number occasion
5203      PETR      004      2004           1011      F          D4        1
10931     PEMA      004      2004           1011      F          G3        3
11520     PETR      004      2004           1011      F          E5        4
3942      PEMA      005      2004           1000      F          D1        1
2861      PETR      006      2004           1111      M          A5        1
4353      PEMA      006      2004           1111      M          D6        2

I'm hoping that I don't have to add the capture occasion line by line in Excel since I have nearly 6,000 captures. Are there some strategies or R code for adding the capture occasion to these data?

Thank you.

[murray.efford]

I'm hoping that I don't have to add the capture occasion line by line in Excel ...

That might be painful, and shouldn't be necessary if you have the data already in R.

I cannot offer a complete solution right off. The problem is that your data don't seem to include occasion-specific trap numbers, so how do we know where each capture happened?

secr has a function 'unRMarkInput' but it doesn't solve your problem.

Do you have multiple sessions (years)?

Murray

[Bryan Hamilton]

I do have multiple years. so there is alot of redundancy with animal id and trap id.

I can add occasion specific trap numbers pretty easily in R and probably get unique animal ids by combining year and animal id.

edit- I have the trap dates too so that may be the easiest way to code the capture occastion.

[murray.efford]

I don't follow. For example, the first row of your data represents 3 capture events, but there is only one trap number.

[Bryan Hamilton]

I think that is a typo. That animal (004) was caught in trap D4, G3 then E5.

edit- In the example data, each capture already has its own row associated with it. I just need to add the capture occasion.

[murray.efford]

I understand now. The 'PEMA' threw me.

So for any one animal the successive rows give the trap number corresponding to the successive detections in the (shared) capture history. I think it should be easy to attach occasion numbers - I'll look at it.

Murray

[Bryan Hamilton]

The PEMA threw me too... Its always worrisome to find errors in your carefully checked dataset. Especially on the first random sample.

Thanks for your help. I'm getting closer with the dates.

[murray.efford]

Here's my way of doing it - there may be better

Code: Select all

addOccasion <- function (df, IDfield = 'AnimalID', Yearfield = 'StartYear', CHfield = 'CaptureHistory') {
  df$ID <- paste(df[,IDfield], df[,Yearfield], sep='.') 
  dflist <- split(df, df$ID)
  dflist <- lapply(dflist, function (oneanimal) {
    occs <- strsplit(oneanimal[1,CHfield], '')[[1]]
    oneanimal$Occasion <- which(occs=='1')
    oneanimal
  })
  do.call(rbind, dflist)
}
newdf <- addOccasion(mydf)
head(newdf)
         AOU_Code AnimalID StartYear CaptureHistory Gender Trap_Number     ID Occasion
4.2004.1     PETR        4      2004           1011  FALSE          D4 4.2004        1
4.2004.2     PEMA        4      2004           1011  FALSE          G3 4.2004        3
4.2004.3     PETR        4      2004           1011  FALSE          E5 4.2004        4
5.2004       PEMA        5      2004           1000  FALSE          D1 5.2004        1

This assumes CaptureHistory has character values (to allow for 0 on first occasion).
Murray

[Bryan Hamilton]

Thanks for this. I will give it a try. I ended up just recoding each sampling date, which worked but was not nearly as elegant as your code.

[murray.efford]

So why did you ask if you weren't going to use the answer?