Dead recovery model with small survival

[nturner]

Hi,

Just looking for some advice on running a dead recovery model...

Some background info, I have 7991 tagged and released individuals with 727 captures over three years.

I binned the data into even weekly time intervals (nocc=149) but am now interested in getting an annual estimate of survival.

Code: Select all

time.intervalsweek<-c(rep(0.019230769, 149))

wall17.processed=process.data(data,model="Brownie", time.intervals = time.intervalsweek)

wall17.ddl=make.design.data(wall17.processed)

wall17.ddl$S$Year=factor(c(rep("spring74Spring75", 52), rep ("spring75spring76", 52), rep("spring76march3177", 45)))

S.year=list(formula=~Year)
f.dot=list(formula=~1)
model.list1=create.model.list("Brownie")



I altered the time interval to be equal to one year (1/52) and binned the data into the yearly groupings and ran the brownie dead recovery model

Im getting annual survival of roughly 7% (one survival interval is at boundary) which seems unrealistically small.

Could this possibly due to the small initial tags released and <10% recovery, or am I interpreting something wrong with the time.intervals code set up??

Thanks in advance!

[darryl]

Presuming no coding error (haven't checked) then one option for a 'low' survival rate is permanent emigration, which depends on biology and study design. Are all tagged animals (that are still alive) within the area where recoveries are collected from? Or can animals move outside of the recoveries area?

Cheers
Darryl

[cooch]

Presuming no coding error (haven't checked) then one option for a 'low' survival rate is permanent emigration, which depends on biology and study design. Are all tagged animals (that are still alive) within the area where recoveries are collected from? Or can animals move outside of the recoveries area?

Cheers
Darryl

Darryl raises the key question -- if these are 'dead recoveries' (note: 'recovery' means 'dead thing', 'recapture' means 'live thing'. In the 'old days', our European colleagues would refer to live captures as 'recoveries', which caused no end of confusion. Applying 'reverse collonialism', ~10-15 years ago, at a meeting, we all agreed to adopt the convention of 'recapture = live', 'recovery = dead'). So, if these are 'dead recoveries', then the next question is...is this a game species, or not? If the former, then the usual assumption is that any dead individual is recovered as a result of harvest, and since harvest can (in theory) occur anywhere, then dead recovery survival estimates are generally unbiased by permanent emigration (again, under the assumption the you can potentially harvest anywhere). This is largely the case in North America, and it is this fact that underlies Ken Burnham's joint live-dead models, wherein you can parse out F - fidelity (i.e., 1 - permanent emigration rate), by making use of the fact that live recaptures occur at sampling site, whereas dead recoveries can occur outside live capture sampling sites. Chapter 9 in the MARK book lays out the basics.

If these are non-game species, and the dead recoveries are opportunistic, then the question is (as Darryl intimates) -- could an individual permanently leave the area in which a dead individual could be recovered? If so, then what Darryl suggests is more than likely a factor you'd need to consider.

[nturner]

So some background on the dataset im working with - this historical study tagged walleye across a large lake system and recoveries were made through the commercial fishery only (dead). Commercial activity could occur anywhere across the entire lake (in theory as was stated).

dead recovery survival estimates are generally unbiased by permanent emigration (again, under the assumption the you can potentially harvest anywhere)

If this is the case then the dead recovery model is still appropriate...?

Or, is it worth investigating a joint live dead model where fidelity can be set to 1 which will allow for true and apparent mortality to be separated (I have only briefly been through chapter 9... need to do a more in-depth read here)

Thanks for the advice!