Modeling with limited data

Forum for discussion of general questions related to study design and/or analysis of existing data - software neutral.

Modeling with limited data

Postby molly_takacs » Wed Mar 10, 2021 8:37 pm

Hello,

I have modeled mark-recapture data using POPAN in RMark to estimate juvenile fish populations. A little background on the data and why we chose POPAN:

The data set consists of 25 years of mark recapture data on juvenile sturgeon. The project did not begin as a sturgeon project so the first few years had little to no sampling design. fishing occurred whenever they could wherever they could. the last 15 years were standardized to 5 months of the year with reduced sampling effort. The sampling area consisted of a 1 km drift. That being said some years of data are very sparse with little to no recaptures. For analysis I had to standardize the data which left even less data to work with (15 captures with 6 recaps for one year).

I could not consider the population to be closed for several reasons. Sampling occurred over several months and only in a 1 km stretch. From what we know about juvenile sturgeon they move all over the place. juveniles would have been moving in and out of the sampling area and new individuals would be moving in as well. movement is beneficial for the population but I cannot consider it closed. Model POPAN was chosen for this and the lack of data.

I initially ran the data with individual day capture histories but got questionable pop estimates with ridiculous confidence intervals for years with little data. I then re-ran all years with capture histories pooled into months for a total of 5. Pop estimates and CI's improved for some years but got worse for others. Some years were almost identical between daily and pooled ch while others were drastically different.

My main question is what should I consider to be more reliable (pooled or daily) when there are "good" and "bad" population estimates in both versions. Intuitively we think it should be pooled because there are far less parameters in the model than dailies with over 34 encounter histories and little to no recaptures. I have been told to consider a more robust design and to look at recaptures between years but I am looking for annual population estimates specifically.

Thank you in advance for the help.
molly_takacs
 
Posts: 4
Joined: Fri Apr 17, 2020 6:00 pm

Re: Modeling with limited data

Postby jlaake » Wed Mar 10, 2021 8:49 pm

I suggested Molly post this message here which was originally in the RMark forum but had little to do with RMark. I'm finding that folks are posting there message to RMark because they used RMark when in fact the question is of a more general nature about how to best analyze their data.

I believe she misunderstood my message to her when she said
I have been told to consider a more robust design and to look at recaptures between years but I am looking for annual population estimates specifically.


I told her she may want to use the robust design models in which she could use short periods within the year as closed and open between years. Note that I believe she has been analyzing each of the 25 years in a separate analysis rather than with a 25 year time period. I now know these are sturgeon which live awhile so it would seem worthwhile to combine the 25 year data set or at least a decade or more worth of data. If anyone else wants to jump in here please do. I keep busy enough just handling questions specific to RMark.
jlaake
 
Posts: 1417
Joined: Fri May 12, 2006 12:50 pm
Location: Escondido, CA


Return to analysis & design questions

Who is online

Users browsing this forum: No registered users and 13 guests