The Wilson Journal of Ornithology 124 ( 2):245 252 , 2012 
POPULATION SURVEY OF LEACH’S STORM-PETRELS BREEDING 
AT GRAND COLOMBIER ISLAND, SAINT-PIERRE AND 
MIQUELON ARCHIPELAGO 
HERVE LORMEE, 14 KARINE DELORD, 1 2 
BRUNO LETOURNEL,- AND CHRISTOPHE BARBRAUD 2 
ABSTRACT.—The Si Pierre and Miquelon Archipelago hosts the only French Leach’s Storm-Petrel ( Oceanodroma 
Irucorhoa ) colony. We conducted a survey during the 2008 breeding season to estimate the breeding population size on 
Grand Colombier Island. This survey included an estimation of burrow detection probability using a double-observer 
approach. We estimated that 3% of 1 .each's Storm-Petrels nests had failed before we stalled the survey. Nest occupancy 
probability was neither affected by slope nor vegetation type and was 0.546 ± 0.029. Burrow density was positively 
affected by slope and, consequently, was much lower on the plateau than on island slopes. Burrow detection probability was 
neither affected by observer nor by habitat anil was 0.89 ± 0.01. Wc estimated the population to be 363.787 [95% Cl = 
295.502—432.072] breeding pairs, which is among the largest Leach's Storm-Petrel colonies in the northwestern Atlantic 
Ocean. Received 10 May 2011. Accepted IV November 2011. 
Questions associated with the population dy¬ 
namics of colonial seabirds arc of intrinsic interest 
to biologists and managers of protected areas. 
Answers to basic questions about population 
estimates and trends are often needed, as seabirds 
are recognized as monitors of marine ecosystems 
and act as indicators of marine environmental 
changes (Cairns 19X4; Monteveechi 1993; Mon- 
tevecchi and Myers 1995a. b; Piatt el al. 2007; 
Einoder 2009). Breeding surveys arc particularly 
relevant for burrow-dwelling species such as 
storm-petrels as a decline could go unnoticed in 
these inconspicuous species, even for decades. 
These species may also be highly vulnerable to 
introduced predators or soil erosion in breeding 
colonics (Brooke 2004). Population estimates and 
trends must rely upon rigorous standardized 
protocols that should be developed throughout 
the range of species or population of concern 
(Walsh et al. 1995). 
Monitoring programs have two important 
sources of variation that must be considered in 
monitoring design: spatial variation and detection 
rate (Thompson 1992. Lancia et al. 1994. Nichols 
et al. 2000. Pollock et al. 2002). Spatial variation 
arises when the observer cannot monitor the entire 
1 ONCFS-CNERA Avifaunc Migralrice-Station dc Chize. 
Candour dc la Canauderic 79360 Villiers en Bois. France. 
CLBC-CNRS, l PR 1934. Carrcfour dc la Canaudcne 
79360 Villiers en Buis. France. 
ONCFS-Service dc la collcctiviid territorialc de Saint 
Pierre and Miquelon, B3 rue Boursaint BP 4244, 97500 
Saini Pierre, France. 
J Corresponding author; e-mail; herve.lormee@oncfs.gouv.fr 
area of interest. Monitoring effort thus has to 
concentrate on sample areas from which the results 
arc used to draw inference about the entire area of 
interest (Pollock et al. 2002). Detection probability 
is important as survey methods do not detect all 
individuals present in the sampled area. Monitoring 
has to incorporate methods for estimating effects of 
detection rate so any estimated temporal or spatial 
changes in the number of individuals counted 
reflect true changes and not differences in 
detection. Observer identity and experience are 
recognized as covariales likely to he relevant to 
variation in detection probability (Saueret al. 1994. 
Kendall et al. 1996); thus, estimation of detection 
probability is particularly required in long-term 
monitoring where inevitable changes in observers 
over time or between sample areas are likely to 
impact survey accuracy. 
Our objective was to estimate the population 
size of a Leach's Storm-Petrel (Oceanodroma 
leucorhoa) breeding colony in the French Saint- 
Pierre and Miquelon Archipelago in Ihe north¬ 
western Atlantic Ocean, relying on a land-based 
survey conducted in 2008. We applied similar 
survey methods, keeping with recent efforts to 
obtain up to date estimates for major Leach’s 
Storm-Petrel colonies in eastern North America 
(Robertson et al. 2006). We explicitly considered 
estimation of detection probability in the survey. 
These surveys are expected lo be regularly 
repeated in future years, and we emphasized the 
double-observer method (Nichols ct al. 2000). 
which appears less Jogistical/y expensive com¬ 
pared to other methods for estimating detection 
rates (i.e., capture-recapture methods). 
245 
