Lishman and No! • MAGELLANIC PLOVER HABITAT SELECTION 
491 
10000 
1000 
Lake perimeter (m) 
FIG. 2. Lake perimeter (m) and area of aeolian lunette (ha) at endorheic lakes in southern Santa Cruz Province, 
Argentina. Solid symbols are lakes occupied by Magellanic Plovers and open symbols arc unused lakes. The size of solid 
symbols represents the number of Magellanic Plover pairs breeding on occupied lakes in 2006-2007: small is one pair, 
followed by one lake with two pairs, two lakes w ith three pairs, one lake with lour pairs, and the largest with 14 pairs. I here 
was no relationship hut ween length of lake perimeter or area of aeolian lunette and number of pairs on a lake. 
tic regression (Llosmer and Lemeshaw 2000). 
Sixteen candidate models were constructed based 
on all possible combinations; the constant model 
and AIC model selection were used to identify the 
most parsimonious models. Models with a AAIC, 
value <4 were considered the best subset. The 
most important parameters affecting microsite 
occupation were calculated by summing AkaikeLs 
weights from models including the parameter ol 
interest (Burnham and Anderson 2002). 
Nest Success and Habitat Characteristics.— We 
compared distance to lakeshore. vegetation, and 
freshwater channel, and percent cover of clay, 
vegetation, gravel, and cobble between successtul 
and unsuccessful nest sites using MANOVA. All 
statistical analyses were run using Program R (R 
Development Core Team 2010). 
RESULTS 
Hike Occupancy .—We searched 53 lakes (23 
occupied, 30 unoccupied) in 2006 (Fig. 1). We 
returned to 35 of these lakes in 2007 and 19 of 
those previously occupied were no longer occu¬ 
pied and had dried earlier than usual tor that 
season. The 22 territories visited regularly 
throughout the 2006 season were revisited in 
December 2007 and 16 were still occupied. The 
six territories that were not occupied were on 
lakes that had dried earlier in the season. We 
found only one pair of plovers on 4 km of the 
large Lago Argentino in 2006, while none was 
found on the 12.5 kin of rivers surveyed. We 
found 40 breeding pairs in both years. Most 
occupied lakes contained a single territory or pair 
while one lake had 14 territories (mean = 1.33 
territories/occupied lake. 95% Cl = 0.106-5.57). 
Eight of the 22 lakes searched on five occasions 
during 2006 were occupied on at least one 
occasion and 14 were not occupied. The monthly 
detection probability was 0.90 (95% Cl = 0.762- 
0.962). 
Aeolian lunette size and lake perimeter 
(Fig. 2) were moderately positively correlated 
(A\ = 0.67, P < 0.05). No significant relation¬ 
ships were found between number of territories 
per lake and lake perimeter. lunette size, or 
