THE WILSON JOURNAL OF ORNITHOLOGY • Vol. 124. No. 3. September 2012 
440 
Recapture distance categories (m) 
FIG. 2. Percentage of recaptures at different distances 
from the original capture site. Zero distance indicates 
capture in the same net. Nets were -50 m apart. 
per net had a clumped or clustered distribution on 
each plot (Indices of Dispersion - 2.19, 2.44, P < 
0.001 on Harpia and Puma, respectively; corre¬ 
sponding Morisita’s Index values were 1.13 and 
1.15, respectively). Number of captures per net 
had no correlation with captures at nearby nets 
despite the clumped distribution pattern. Correlo- 
grams had a significant positive correlation only 
at 525 m on Puma (Moran's I = 0.076. P = 0.04 
at 525 m; Bonferroni corrected significance for 
total correlation = 0.75) and no significant 
correlation at any distance on Harpia (Bonferroni 
significance for total correlation = 1.0). 
Number of captures per habitat category (4 
types were represented by nets on Harpia; swamp 
FIG. 3. Means (± SE) for recapture distances within 
and between years on each of two study plots at Tiputini 
Biodiversity Station. Ecuador. Betwcen-year capture dis¬ 
tances based on the distance between the last net the bird 
was captured in 1 year to the net it was first captured in 
during the second year. Values in parentheses are range of 
recapture distances. 
Number of captures per net 
FIG. 4. Number of captures per net on each of two 100- 
ha study plots (96 nets/plot) at Tiputini Biodiversity 
Station. Ecuador. 
habitats not represented) did not differ from 
expected with expected values based on the 
proportion of net sites within each habitat 
category (x; - 0.9, df =3 , P > 0.75). There 
were fewer captures than expected in swamp 
habitats on Puma and more than expected in 
mixed-upland forests (x z 17.7, df = 4, P = 
0.001; f = 11.0, df = 3, P = 0.029 if swamp 
habitats are excluded from comparison). 
Observation Pat a.—We recorded 490 Glypho- 
rynchus during February samples on Harpia (mean/ 
sample = 70 ± 4.7) and 384 on Puma (mean = 55 
± 5.7; Table 1). The number of observations was 
greater on Harpia during 6 of 7 years (paired /-test. 
t = 2.36. df =6. P = 0.057). Observations per 
sample were somewhat less variable on Harpia 
(CV = 17.8) than on Puma (CV = 27.7; variance- 
ratio lest Fbx, = 1.53, P = 0.31) and were not 
correlated across years between the two plots (r = 
0.08). Number of observations per sample was not 
correlated with number of captures during the same 
year on Harpia (r = 0.18) but was to some extent 
on Puma (r = 0.67). 
Observations were not evenly distributed 
throughout either plot (Fig. 5); Glyphnrynchus 
was not recorded in ~-28% of cells on Harpia and 
~40% on Puma. Combining cells with six or 
more Glyphorynchus observations (to avoid cells 
with <5 records), the distribution of cells with a 
given number of observations (0, 1. 2. 3. 4. 5. >5) 
differed between the two plots ij 2 = 19.8, df = 6. 
P = 0.003). mostly a consequence of the greater 
number of cells without any observations in Puma. 
The difference in distribution was not significant 
