NAs in mrClosed

[droglenc]

NAs cause problems with mrClosed(). Need to make a catch and send the user a warning.

Here is the issue. In a Schnabel method the first sample will have no marked fish (and thus m[1]=0) and it is irrelevant how many marked fish are returned to the population after the last sample (thus, either R[length(R)]=0 or R[length(R)]=n[length(n)]. However, some users may enter NA values for these positions, which will cause an error as shown below.

> WAL <- mrClosed(n =c(321,412,178,415,367),
+ m = c(NA,45,55,93,113),
+ R = c(321,412,178,415,NA),
+ method ="Schnabel")
> cbind(summary(WAL),confint(WAL,verbose=TRUE))
 Error in if (object$sum.m < 50) type <- "Poisson" else type <- "normal" : 
  missing value where TRUE/FALSE needed  

A better result would be to catch these and replace them with zeroes with a warning sent to the user.

[droglenc]

This has been addressed in the latest development version of FSA.

The version with NAs ...

> WAL <- mrClosed(n =c(321,412,178,415,367),
+                 m = c(NA,45,55,93,113),
+                 R = c(321,412,178,415,NA),
+                 method ="Schnabel")
Warning messages:
1: NA for last sample of 'R' was ignored. 
2: NA for first sample of 'm' was ignored. 
> cbind(summary(WAL),confint(WAL,verbose=TRUE))
The normal distribution was used.
        N 95% LCI 95% UCI
[1,] 3280    2832    3898

... gives the same results but with warnings as the version without NAs ...

> WAL <- mrClosed(n =c(321,412,178,415,367),
+                 m = c(0,45,55,93,113),
+                 R = c(321,412,178,415,367),
+                 method ="Schnabel")
> cbind(summary(WAL),confint(WAL,verbose=TRUE))
The normal distribution was used.
        N 95% LCI 95% UCI
[1,] 3280    2832    3898