246 
THE WILSON JOURNAL OF ORNITHOLOGY • Vol. 124. No. 2. June 2012 
METHODS 
Study Site. —Field work was conducted on 
Grand Colombier Island (46 49' N, 56 10' W), 
Saint-Pierre and Miquelon Archipelago, in the 
northwestern Atlantic Ocean from 18 June to 18 
July 2008 (Fig. I). Grand Colombier Island has an 
estimated surface area of 480,000 nr and hosts 
large breeding populations of seabirds (Table 1), 
particularly Atlantic Puffin ( Fratercula arctica) 
and Leach's Storm-Petrel (Desbrosses and Etch- 
eberry 1989). The island is free of all mammal 
species but one, the meadow vole (Microtus 
pensylvaniais). The main topographical features 
of the island are a central plateau surrounded by 
vegetated or rocky slopes and cliffs. Grand 
Colombier has dry soils and is densely vegetated 
(mainly ferns, Dryopteri, s spinulosa, and grami- 
noids, Deschampsia Jlexuosa), providing diverse 
and highly suitable breeding sites for burrow¬ 
dwelling petrels and Atlantic Puffins. 
Leach’s Storm-Petrels breed almost exclusively 
on Grand Colombier Island within the Saint- 
Pierre ami Miquelon Archipelago where they nes 
in more or less aggregated burrows, forming 
relatively dense colonies (Desbrosses and Etch 
eberry 1989). Petrels return to colonies in May 
lay their single egg in June, and start visiting 
colonies at dusk and during the first part of the 
night. Young hatch in July and fledge ir 
September. Males and females alternate foraging 
trips at sea during breeding and feed on fish, krill, 
and squid (Montevecehi et al. 1992). 
Sampling Design.- We used a systematic 
sampling approach to estimate Leach’s Storm- 
Petrel density (Harris and Murray 1981) following 
Catry et al. (2003). We conducted line transects (n 
- 19) from 8 to 18 July 2008. crossing the entire 
island from north to south, during the second half 
ot the incubation and early brooding periods. The 
irst transect location was chosen randomly and 
die following transects paralleled the first' one. 
The distance between successive transects was 
;, m - Each dansect starting point was located 
with a Global Positioning System (GPS) and 
plotted on a map (Fig. 2). 
We stopped at counting points every 50 m 
along each transecl (measured using a 10-m rope). 
e appl,cation ot these procedures resulted in all 
plots („ - |62) being pseudo-randomly located in 
region to habitat features and burrow deresHy 
One field worker stood a. the center of the plot a, 
ach location, holding the tip of a 3-m rope, while 
a second observer holding the other tip walked in 
circle (total surface of the plot = 28.27 nri and 
counted all burrow entrances that were within the 
plot. The slope angle of each plot was estimated 
using a clinometer. Petrel burrows were identified 
by entrance diameter (4-5 cm). Burrows consist 
of a tunnel of 23 cm (range = 12-39 cm I depth on 
average which may be straight or with several 
turns (Huntington ct al. 1996). The nest chamber 
is at the end of the tunnel. We only counted 
burrows used by Leach's Storm-Petrels (Hunting- 
ton et al. 1996) excluding double-entrance 
burrows and vole tunnel entrances. Leach's 
Storm-Petrel and meadow voles, respectively, 
use burrows and tunnels with similar size 
entrances, hut which generally have different 
tunnel shape (vole tunnels stay just below the 
surface) and habitat requirements in temis of 
vegetation cover and soil substrate (Cramp and 
Simons 1977. Huntington et al. 1996). We 
counted burrows with a clear entrance. Burrow 
entrances overgrown by vegetation were consid¬ 
ered inactive and were not counted. Non-surveyed 
areas (lakes, sleep cliffs) represented 2.2 % of 
Grand Colombier Island. 
Burrow density may be affected by habitat 
characteristics including slope angle and/or veg¬ 
etation type. Fern patches on Grand Colombier 
Island w'erc highly associated with steep slopes nf 
the island while herbaceous vegetation was nearly 
exclusively on the plateau and low slopes. We 
investigated the relationship between burrow 
density and slope angle using a generalized linear 
model dog link, negative binomial distribution). 
Estimating Harrow Occupant's Probability — 
We estimated burrow occupancy probability by 
acoustic playback, to minimize disturbance of 
breeding birds, on a sample of 301 burrows at 19 
stations spread over the island. Sample plots were 
selected to be representative of the habitat 
diversity and slope range; 11 and eight plots were 
sampled, respectively, in fern and graminaceous 
habitats, covering a large range in slope (from 5 *•> 
38.1%). I.each's Storm-Petrel calls are sex-specific 
(Huntington et al. 1996); calls of both males and 
females were played into each burrow with a digital 
voice recorder during 1 min and we recorded 
whether or not a bird responded tRateliffe et al. 
1998, Amhagis 2004). It is known that a proportion 
of Leach s Storm-Pfetrels present in burrows may 
not respond to playback (Amhagis 2004), and we 
systematically inspected burrows with no response 
to playback using a burrow-scope. We checked nest 
