RMark (version 3.0.0)

wwdo.popan: White-winged dove Jolly-Seber POPAN Analysis Details

Description

This dataset represents 2 years of capture-mark-recapture data collected on uniquely identifiable leg-banded (size 4) white-winged dove captured in Alice, Texas, USA (Latitude 27.25, Longitude -98.07) between mid-February and mid-September during 2009 and 2010. The package was developed such that others could recreate the analysis developed by Collier, B. A., S. R. Kremer, C. D. Mason, J. Stone, K. W. Calhoun, and M. J. Peterson. 2012. Immigration and recruitment in an urban white-winged dove breeding colony. Journal of Fish and Wildlife Management, In review., and see how the data and results were used to estimate population level recruitment (number juveniles in population over number adults in population).

Format

The format is 2 data frames, one for 2009 and one for 2010. 2009 has 5101 unique captures, 2010 has 3502 unique captures.

Prefix

Unique band prefix identifier (usually 0914 or 0984)

Suffix

Unique band suffix (numeric value)

E0-E13

0/1 representing whether a dove was captured (1) or not captured (0) during that (E) sampling occasion

Age

Age class with AHY=after-hatch year and HY=hatch year

Details

White-winged doves were captured (aged: AHY=after-hatch year, HY=hatch year) continuously in baited walk-in dove traps beginning in Februrary and ending in September in each year (2009 and 2010). During 2009 5,101 white-winged doves were captured (2,894 AHY, 2,207 HY) while in 2010 3,502 white-winged doves were captured (3,106 AHY, 486 HY). We used approximately 2 week date windows to categorized our encounter histories for analysis in MARK http://www.phidot.org/software/mark/ via RMark using these dates: 27 Feb; 13 March; 27 March; 10 April; 24 April; 8 May; 22 May; 5 June; 19 June; 3 July; 17 July; 31 July; 14 August-End; giving us 13 encounter occasions.

I wanted to force b0=0 for the first time frame, as none could be there when we started as they had not arrived yet in any real number. If you take the first column out, the numbers get ridiculously screwy for the super population size and the entry parameters, because the initial population has individuals in it, thus the JSPOPAN estimates something like 40 the population pre-trapping, which is biologically impossible.

Initially, because of the parameter structure in a JS-POPAN model and the fact that the initial entry probability is 1 minus the sum of the resultant entry probabilities for subsequent sampling occasions, and because occasionally a couple of doves were captured during the initial time frame, we were getting entry values for the initial time period representing >40 To obtain reasonable results a 'fake' encounter occasion (time -1) with no captures ("0") was appended to beginning of each capture history to force b0=0. As such, the data for wwdo.09 and wwdo.10 will have 14, not 13 encounter histories.

Important to note, in case you don't read Collier et al. (2012), is the fact that the 2010 dataset is kind of screwy relative to the estimation of the entry parameters for HY wwdo. Basically, what happened was we caught a bunch of AHY birds, but when we captured HY birds, we only caught a few (~400 in 2010) and of those we captured, we rarely, if ever had any recaptures. Without getting into a bunch of speculation on what happened, as we don't really know, we suspect it had something to do with the fact that 2009 was a extreme drought in Texas, as to where 2010 was extremely wet, so mast based food sources (mulberry tree's are everywhere) and such were readily available in the urban environment, as such, lower trapping success. So, when you fit the 2010 dataset using the same candidate model set at the 2009, you get pretty nonsensical answers for the entry b parameters. As such, we did not use a group specific entry model for 2010. If you want more detail, see Collier et al. (2012).

Within the model code below, you can see that we fixed the both the Phi and p parameters for those periods in the analysis when HY birds could not be in the population (e.g., when all birds migrated into the breeding colony and no HY birds had been produced yet). There is probably a little slack in the range, as it is possible that there were some HY white-winged doves in the population in the last period we fixed, but we did not catch any there so we opted to fix it as well.

Note that the R function wwdo.popan is set up in RMark speak, and will run either the wwdo.09 or wwdo.10 datasets as long as you specify one in the lines where I have listed data(wwdo.09) and wwdo=wwdo.09. If you want to see the 2010 analysis, just change those lines to wwdo.10.

References

Collier, B. A., S. R. Kremer, C. D. Mason, J. Stone, K. W. Calhoun, and M. J. Peterson. 2012. Immigration and recruitment in an urban white-winged dove breeding colony. Journal of Fish and Wildlife Management, In review.

Examples

# \donttest{
# This example is excluded from testing to reduce package check time
# Starting with R version 4, examples wrapped in donttest are still run. Therefore,
# I have commented out the model runs below because this example takes a lot of computing
# time. If you want to run, you can copy to an editor and remove the # to run this example
data(wwdo.09)
wwdo=wwdo.09
wwdo.popan=function(){
wwdo.proc=process.data(wwdo, model="POPAN", groups="Age")
wwdo.ddl=make.design.data(wwdo.proc)

#Fixing Phi Parameters for sampling periods where HY WWDO were not available in population
	hy.phi1=as.numeric(row.names(wwdo.ddl$Phi[wwdo.ddl$Phi$group=="HY" & 
                      wwdo.ddl$Phi$time==1,]))
	hy.phi2=as.numeric(row.names(wwdo.ddl$Phi[wwdo.ddl$Phi$group=="HY" &
                      wwdo.ddl$Phi$time==2,]))
	hy.phi3=as.numeric(row.names(wwdo.ddl$Phi[wwdo.ddl$Phi$group=="HY" & 
                      wwdo.ddl$Phi$time==3,]))
	hy.phi4=as.numeric(row.names(wwdo.ddl$Phi[wwdo.ddl$Phi$group=="HY" & 
                      wwdo.ddl$Phi$time==4,]))
	hy.phi5=as.numeric(row.names(wwdo.ddl$Phi[wwdo.ddl$Phi$group=="HY" & 
                      wwdo.ddl$Phi$time==5,]))
	hy.phi6=as.numeric(row.names(wwdo.ddl$Phi[wwdo.ddl$Phi$group=="HY" & 
                      wwdo.ddl$Phi$time==6,]))
	hy.phi7=as.numeric(row.names(wwdo.ddl$Phi[wwdo.ddl$Phi$group=="HY" & 
                      wwdo.ddl$Phi$time==7,]))
	hy.phi.fix=c(hy.phi1, hy.phi2, hy.phi3, hy.phi4, hy.phi5, hy.phi6, hy.phi7)
	
#Fixing PENT Parameters for sampling period where HY WWDO were not available in population
	hy.pent2=as.numeric(row.names(wwdo.ddl$pent[wwdo.ddl$pent$group=="HY" & 
                       wwdo.ddl$pent$time==2,]))
	hy.pent3=as.numeric(row.names(wwdo.ddl$pent[wwdo.ddl$pent$group=="HY" & 
                       wwdo.ddl$pent$time==3,]))
	hy.pent4=as.numeric(row.names(wwdo.ddl$pent[wwdo.ddl$pent$group=="HY" & 
                       wwdo.ddl$pent$time==4,]))
	hy.pent5=as.numeric(row.names(wwdo.ddl$pent[wwdo.ddl$pent$group=="HY" & 
                       wwdo.ddl$pent$time==5,]))
	hy.pent6=as.numeric(row.names(wwdo.ddl$pent[wwdo.ddl$pent$group=="HY" & 
                       wwdo.ddl$pent$time==6,]))
	hy.pent7=as.numeric(row.names(wwdo.ddl$pent[wwdo.ddl$pent$group=="HY" & 
                       wwdo.ddl$pent$time==7,]))
	hy.pent.fix=c(hy.pent2, hy.pent3, hy.pent4, hy.pent5, hy.pent6, hy.pent7)
	
#####
#Real Parameter Definitions
#####
#Detection process
	p.dot=list(formula=~1)
	p.time=list(formula=~time)
	p.group=list(formula=~group)
	p.g.time=list(formula=~group:time)
	
#Survival process
	Phi.dot.fix=list(formula=~1, fixed=list(index=hy.phi.fix, value=c(0,0,0,0,0,0,0)))
	Phi.time.fix=list(formula=~time, fixed=list(index=hy.phi.fix, value=c(0,0,0,0,0,0,0)))
	Phi.age.fix=list(formula=~group, fixed=list(index=hy.phi.fix, value=c(0,0,0,0,0,0,0)))
	Phi.timeage.fix=list(formula=~time:group, 
     fixed=list(index=hy.phi.fix, value=c(0,0,0,0,0,0,0)))
	
#Entry Process-always time dependent, otherwise makes no sense in my situation
	pent.time.fix=list(formula=~time, fixed=list(index=hy.pent.fix, value=c(0,0,0,0,0,0)))
	
	#Model.1=mark(wwdo.proc, wwdo.ddl, 
 #  model.parameters=list(Phi=Phi.dot.fix, p=p.dot, pent=pent.time.fix, N=list(formula=~group)),
 #  invisible=FALSE,threads=1,options="SIMANNEAL",delete=TRUE)
	#Model.2=mark(wwdo.proc, wwdo.ddl, 
 # model.parameters=list(Phi=Phi.time.fix, p=p.dot, pent=pent.time.fix, N=list(formula=~group)),
 #  invisible=FALSE,threads=1,options="SIMANNEAL",delete=TRUE)
	#Model.3=mark(wwdo.proc, wwdo.ddl, 
 # model.parameters=list(Phi=Phi.age.fix, p=p.dot, pent=pent.time.fix, N=list(formula=~group)), 
 #   invisible=FALSE,threads=1,options="SIMANNEAL",delete=TRUE)
	#Model.4=mark(wwdo.proc, wwdo.ddl, 
 # model.parameters=list(Phi=Phi.timeage.fix, p=p.dot, pent=pent.time.fix, N=list(formula=~group)), 
 #   invisible=FALSE,threads=1,options="SIMANNEAL",delete=TRUE)
	#Model.5=mark(wwdo.proc, wwdo.ddl,  
 # model.parameters=list(Phi=Phi.timeage.fix, p=p.time, pent=pent.time.fix, N=list(formula=~group)), 
 #  invisible=FALSE,threads=1,options="SIMANNEAL",delete=TRUE)
	#Model.6=mark(wwdo.proc, wwdo.ddl, 
 #  model.parameters=list(Phi=Phi.timeage.fix,p=p.g.time, pent=pent.time.fix,
 #             N=list(formula=~group)), 
 #  invisible=FALSE,threads=1,options="SIMANNEAL",delete=TRUE)
	#collect.models()
}
wwdo.out=wwdo.popan()
wwdo.out
# }