Reiter and Andersen • ARCTIC FOXES, LEMMINGS, AND CANADA GEESE 
267 
nesting waterfowl species (but see Wilson and 
Bromley 2001). 
Eastern Prairie Population (EPP) Canada Geese 
(Branta canadensis interior) at Cape Churchill, 
Manitoba have exhibited substantial inter-annual 
variation in nest success over nearly 30 years 
(1976-2004). Arctic fox depredation accounts for 
>80% of nest failures in some years (Walter 
1996) and local arctic fox dens have been 
monitored regularly for occupancy. Multi-annual 
cycles have been documented in collared lemming 
I Dicrostonyx richardsoni) populations, a primary 
prey of arctic foxes in this area (Bahr 1989), near 
the town of Churchill (Shelford 1943. Scott 1993) 
and more recently at Cape Churchill ( Roth 2003, 
Reiter and Andersen 2008b). However, the extent 
to which annual fluctuations in nest success of 
dispersed-nesting EPP Canada Geese coincide 
with variation in lemming abundance and arctic 
fox den occupancy is unknown. We used existing 
data from Cape Churchill, Manitoba between 
1993 and 2004 to quantify factors influencing 
Canada Goose nest success in the context of the 
predictions of the bird-lemming hypothesis. We (1) 
present time-series of Canada Goose nest success, 
nest density, hatch date, relative arctic fox abun¬ 
dance, and relative abundance of lemmings be¬ 
tween 1993 and 2004; and (2) model nest success as 
n function of years with peak lemming abundance, 
years with a lemming population trough I year 
following a peak, arctic fox den occupancy, nest 
density, and median hatch date. We also examined 
the impact of fox removal, which occurred on our 
study area from 1994 to 1997, on nest success. 
METHODS 
Study Area. —The EPP Canada Goose breeding 
range includes —101,500 km 2 in northern Mani¬ 
toba. Canada (Malecki et al. 1980). The highest 
density of breeding Canada Geese occurs along a 
s tnp of coastal tundra on western Hudson Bay 
w >thin the broader ecosystem of the Hudson Bay 
Lowlands (Reiter 2009). The Nestor One study 
(-48 km 2 ; 58 : 34' N, 93 11' W) was south 
of Cape Churchill and -60 km east-southeast of 
the town of Churchill. Manitoba (Fig. I). Nestor 
One is characterized by low relief, continuous 
Permafrost, and coastal tundra vegetation (Didiuk 
a nd Rusch 1979, Brook 2001). Coastal salt 
tnarshes, beach ridges, sedge (Carex spp., Erio- 
Phoruni spp.) meadows, and interior sedge 
nieadow complexes comprise the major habitat 
types. The northern boreal forest ecosystem is 
~10 km inland from Nestor One and the Hudson 
Bay coastline. 
Arctic fox breed in relatively high densities on 
the coastal tundra in this region, using established 
dens along large, elevated beach ridges (Bahr 
1989, Roth 2003). Collared lemmings, the only 
lemming or vole species that occur on the Nestor 
One study area (Roth 2002), are the primary food 
for arctic foxes. Nesting waterfowl (Anseri- 
formes) and shorebirds (Charadriiformes) are 
abundant in this region and are consumed by 
foxes (Bahr 1989; M. E. Reiter, unpubl. data) 
during the summer. A breeding colony of 
>20,000 nesting Lesser Snow Geese ( Chen 
caerulescens caerulescens) and Ross's Geese (C. 
rossii) occurs <20 km from Nestor One at La 
Perouse Bay (Fig. 1; Cooke et al. 1995). 
Nest Success ami Hatch Date —Ground surveys 
of Canada Goose nests and estimates of nest 
survival were completed annually at Nestor One 
since 1976 following standardized protocols 
(Didiuk and Rusch 1979, Walter 1999). Nest- 
search crews (4-6 people) spaced at regular 
intervals systematically searched Nestor One. 
Females were flushed at all nests encountered 
and the incubation day of the nest was estimated 
using candling and flotation methods (Westerkov 
1950. Weller 1956, Reiter and Andersen 2008a). 
Nest locations were marked with a 7.5 X 12.5-cm 
plastic orange flag placed 10 m north of the nest 
bowl. We revisited nests at or subsequent to 
predicted hatch date, based on a 28-day incuba¬ 
tion period, to ascertain fate (Didiuk and Rusch 
1979. Walter 1999). We categorized nests as (1) 
successful if >1 egg hatched, indicated by the 
presence of goslings in the nest or eggshells and 
intact membranes consistent with hatching, or (2) 
failed if few or no eggshell fragments were in the 
nest. We calculated daily survival rate (DSR) and 
annual nest success using exposure days and the 
number of nests that failed after initial discovery 
(Mayfield 1975). We followed methods described 
by Johnson (1979) to construct 95% confidence 
intervals (95% Cl) for DSR and nest success. We 
also estimated the median hatch date for all 
Canada Goose nests in each year as a measure of 
spring phenology, as hatch date is correlated with 
timing of snow melt (Walter 1999). 
Nest Density.—We calculated nest density for 
each year from 1993 to 2004 as the number of 
Canada Goose nests/100 ha of wetland nest 
habitat. We used estimates from Walter and 
Rusch (1997) to correct observed nest counts for 
