BIAS IN CANADA GOOSE POPULATION-SIZE ESTIMATES FROM SIGHTING DATA

Application of capture-recapture methods to sightings of marked indivi duals is a common technique for estimating Canada goose (Branta canade nsis) population size. Hestbeck and Malecki (1989) described a method using the Jolly-Seber estimator (M) over cap(i) to estimate the number of marked geese in the population at time i. Total population size wa s estimated as (N) over cap(i) = (M) over cap(i) (R) over cap(i) where (R) over cap(i) was the estimated ratio of total geese (marked and un marked) per marked goose in the population. The ratio (R) over cap(i), was estimated from counts of marked and unmarked geese in observed ne cks. The (M) over cap i is defined as the number of individuals, captu red and marked in samples 1,..., i - 1, that are alive at time i. When applied to neck band sighting data, (M) over cap(i) estimates the num ber of marked geese initially sighted in samples 1,..., 1 - 1 and aliv e at the ith sample. This differs from the total number of marked gees e at time i because frequently marked geese in the population at i wer e never sighted in samples 1,..., i - 1. Bias can occur when estimatin g (R) over cap(i) if there are sightings of marked geese when the indi viduals are not identified (unverified sightings). We used Monte Carlo simulations to demonstrate a positive bias in population size estimat es from (M) over cap(i) when the proportion of sightings that were unv erified was greater than or equal to 5%. An alternative method is to u se the Jolly-Seber estimator (N) over cap(i) to estimate the number of marked geese at time i. The appropriate ratio of total geese to marke d geese corresponding to (N) over cap i is unbiased by unverified sigh tings. We used sightings of dusky Canada geese (B. c. occidentalis) ma rked with plastic neck and tarsal bands to demonstrate bias in populat ion size estimates due to unverified sightings.