Tag type changing mid-project

[rirvine]

Hi All,

I am working on a multi-year mark recapture program on three fish species in a large river. When the program started (6 years ago), there was not enough money in the budget for PIT tags, so T-bar (external) tags were used. However, for the past two years, we have been alotted enough money to use PIT tags. A group variable was added to the encouter history file to test for survival differences between the two types of tags with 1 0 for T-Bar and 0 1 for PIT. However, I am now encountering fish that the field crews have had to replace a rotting T-Bar tag with a PIT tag. This means that the fish has encountered 1-5 years with a T-Bar tag and is now back in the river without the T-Bar and with a PIT tag. I am unsure how to code the tag group for these fish and I don't want to throw the data out if at all possible given that these fish are multi, inter-year recaptures and are quite important to the estimation process.

Any assistance would be greatly appreciated.
Thanks,
Robyn

[Bill Kendall]

There is no need to throw out data. Simply censor the animal after the point at which the tag is switched, and enter it in as a new animal in the other group at that point. For example suppose in a 6-year study an individual is tagged in period 1 with tag type 1 and then captured again in periods 3, 4, and 6. Its history would then be

101101 1 0;

Suppose further that in period 3 tag type 1 is replaced with tag type 2. To correct for this properly the history above would be replaced with two histories:
101000 -1 0;
001101 0 1;

If it had an individual covariate that is applicable throughout, it could be carried along for both parts of the history.

[rirvine]

Thanks for the response. Now that I am actually starting the analysis, I have a subsequent question. I am using the Robust- closed captures design and am predominantly dealing with live recoveries. However, on occasion, there is a dead fish or two. The dead recoveries are minimal (9 out of 5000 for one species), but I am struggling with how to code the data. I have read the new chapter 16 on robust designs and have re-read the intro chapter on coding data, but I have not found the information (apologies if it is in there and I didn't see it). There are individual covariates for each fish, so the possibility of just putting a negative value in the frequency column is not an option. I initially put a separate group in 'alive dead' that had either a 1 or -1 value, but there are not enough dead fish to really estimate most of the parameters for this as a separate group and I find it confusing as to how MARK is seeing this data. Can I simply put a -1 in the encounter history when a fish is known to be dead or does this not function in the Robust design?
Thanks for any assistance that can be provided.
A sample of the data is below. The three groups are currently: alivedead, TBAR tagged, PIT tagged and the two standardized individual covariates are fork length and weight.

/*13941*/ 000000000000000000000000000100000000 1 1 0 0.5022 0.3961;
/*13942*/ 000000000000000000000000000100000000 1 0 1 0.6758 0.2557;
/*13943*/ 000000000000000000000000000100000000 1 1 0 0.6179 0.6013;
/*13944*/ 000000000000000000000000000101000000 -1 0 1 -0.8714 -0.9037;
/*13945*/ 000000000000000000000000000100000000 1 1 0 0.3721 0.0576;
/*13947*/ 000000000000000000000000000100000000 1 0 1 1.8035 2.5673;