60 
THE WILSON JOURNAL OF ORNITHOLOGY • Vol. 124. No. 1. March 2012 
TABLE 1. Data for nine best sampled pairs of the Serra 
Finch in Serra do Cipo National Park. Brazil. Values of 
100% minimum convex polygon (MCP); and by 95% and 
50% kernel in ha. 
Bird pair 
Number of 
locations 
Sampling 
days 
Sampling 
period" 
100% 
MCP 
95% 
Kernel 
50% 
Kernel 
A 
43 
6 
435 
2.11 
3.29 
0.85 
B 
57 
14 
412 
3.30 
3.62 
1.02 
C 
52 
9 
273 
2.68 
2.84 
0.63 
D 
44 
6 
168 
3.09 
4.69 
1.16 
E 
14 
3 
32 
0.78 
2.66 
0.75 
F 
13 
3 
113 
1.46 
4.87 
1.23 
G 
30 
4 
169 
1.42 
2.32 
0.55 
H 
12 
3 
112 
1.24 
4.78 
1.15 
K 
16 
4 
169 
0.88 
1.85 
0.40 
Number of days elapsed between the first and the last locations. 
territory) and B (3.00%), and 19.34 nr for pairs C 
(0.07%) and D (0.06%) (Fig. I). The estimated 
overlap in area between territories (95% fixed 
kernel) was 7,204.59 nr for pairs A (20.72%) and 
B (19.35%), and 3,673.51 nr for pairs C (13.09%) 
and D (7.94%) (Fig. 2). The eore areas over¬ 
lapped for 1,155.62 m 2 for pairs A (12.03%) and 
B (10.99%), and did not overlap between pairs (’ 
and D. 
approaches. Dry and wet grassland were essen¬ 
tially avoided by all individuals (Table 3), and we 
excluded them from the compositional analysis. 
The proportion of habitat use based on 95% 
kernel territories in the second-order approach 
was non random relative to that available in the 
study area (A = 0.13, df = 4. X 2 = 18.30, P < 
0.05). A ranking matrix ordered the habitat types 
in the sequence: wet scrubland > riparian 
woodland > rocky outcrop > candeial > dr. 
scrubland (Table 4). Serra Finch territories had 
significantly more wet scrubland and nparian 
woodland than dry scrubland. 
Habitat use of territory locations versus 95% 
kernels in the third-order approach significantly 
differed from random (X = 0.20, df - 4. X : = 
14.41, 1 1 < 0.05). The matrix ranked the habitat in 
the order: candeial > riparian woodland > rocky 
outcrop > dry scrubland > wet scrubland. 
Candeial was significantly more used than rocky 
outcrop, dry scrubland, and wet scrubland, the last 
was significantly less used than riparian woodland 
(Table 4). 
DISCUSSION 
We did not observe any changes in the 
composition of mated pairs or in their territories, 
indicating these birds have annual site fidelity and 
a socially monogamous mating system. All pairs 
were observed singing duets on perches, appar¬ 
ently defending their territories throughout the 
year, even outside the breeding season. We 
observed agonistic events („ = 4) on territory 
boundaries, when two pairs stayed in close 
proximity performing a unison duet, and ending 
in a physical confrontation on two occasions. One 
of those, between pairs A and B. occurred within 
the MCP overlap area. 
Habitat Selection .-The 318 locations recorded 
in the habitat types were not distributed as 
expected from availability in the study area (I 1 
< 0.001, x 2 = 686.82, df = 6; Table 2). The 
candeial. rocky oulcrops, wet and dry scrubland, 
and riparian woodland habitats were used more 
t an expected by chance as the proportions of 
expected use were below the Bonferroni confi¬ 
dence intervals for observed use. Use of dry and 
we grassland habitats was lower than what could 
be expected by chance (Table ~>) 
We used the location data of nine Serra Finches 
which had >10 locations (Tables 1-3) to examine 
hab, lat Wfecllon by (he .second and third orZ 
territory Distribution .—The number ot — 
Serra finches that we found is high compared 
with other species that inhabit open areas in 
central Brazil. Our results revealed higher densi¬ 
ties than 11 species studied by Braz (2008). 
including the closely related Black-masked Finch 
(Cotyphuspiza melanntis) (0.23 individuals/hai 
and Wedge-tailed Grass Finch [Emberizoides 
Iwrbicola) (0.15 individuals/ha). Silva (2008) 
reported 11 individuals/ha for Serra Finch. Hoff¬ 
man (2011 ) estimated a density of 0.44 ± 0.08 
individuals/ha for Gray-backed Taehuri ( Polystic■ 
tus supereiliaris). a tyrant-flycatcher with a range 
distribution similar to the Serra Finch (Vasconce- 
los 2008). Population density is highly dependent 
on the quality of a particular ecosystem (Makar- 
ieva et al. 2005, Johnson 2007). Our density data 
are for a population of the Serra Finch occurnng 
in an undisturbed area, which may serve as a 
baseline lor comparison with future studies o! this 
species. 
Mattos and Sick (1985) reported 400 in as the 
distance between pairs of Serra Finch, which is 
higher than the average distance estimated for the 
birds at our study site. Our results were derived 
Irom small sample sizes, but it is the most 
accurate estimate available, particularly given 
