590 
THE WILSON JOURNAL OF ORNITHOLOGY • Vol. 124. No. 3. September 2012 
FIG. I. Study area, northeastern Buenos Aires Province, central-eastern Argentina; shaded areas represent the surface 
on which buildings, paved roads, or other construction replaced the natural substrate (= ‘built habitat’). Sites are arranged 
from north to south. Numbers correspond to those in Table 1. 
of Buenos Aires is formed by buildings and paved 
streets with patches of parks and open green areas 
(Morello et al. 2000, Cavia et al. 2009). Towards 
the north, west, and south, urbanization is 
gradually replaced by pastures and cultivated 
fields, where grasses occupy small relicts along 
field borders and roads (Soriano et al. 1991). The 
landscape in some littoral areas is still composed 
of small, highly fragmented, fringes of humid 
grasslands intermixed with reduced patches of 
riparian thickets and xerophylous forests (Mat- 
teucci et al. 1999). The climate is characterized by 
mean annual precipitation of 1.014 mm and mean 
temperatures of 23 C in January (summer) and 
10 C in July (winter) (Murphy 2008). 
Data Collection .—Fresh pellets were collected 
mostly between 2005 and 2006 at nest and roosting 
sites from 24 localities (Fig. 1). One to six collections 
were made at each site. All results from each roost 
were combined, and each site provided only one 
sample. Pooling data was necessary to minimize the 
effects of possible seasonal and annual biases, given 
there were insufficient data to use in the analysis 
(Clark and Bunck 1991, Love et al. 2000). Analyses 
were performed only with samples from sites with 
>1(X) prey items. Studied sites were ordered along a 
urban to rural gradient (Table I) considering build¬ 
ings. paved roads, or other construction that replaced 
the natural substrate or the percentage of tree and 
herbaceous cover, among others (Table 1; Hercolini 
2007). We follow Morello et al. (2000) in delining 
‘urban’, ‘periurban’, and ‘rural’. We used the area 
covered by human constructions (‘built habitat’. 
sensu Whitney 1985) as an approximate measure of 
urbanization for statistical procedures (Hercolini 
2007. Cavia et al. 2009; Table I). Landscape 
variables, including ‘built habitat', were recorded 
for each site in an area of 2.5-km radius using die 
continuous land classification of Hansen et al. (2002) 
and data from MODIS (Moderate Resolution 
Imaging Specirorudiometer). We used a 2 5-ktn 
radius coinciding with the mean home range of Bam 
Ow ls (Hercolini 2007). 
Vertebrate prey items were identified to species 
level by comparisons with reference collections 
housed at die Museo Argentino de Ciencias Natural^ 
‘Bernardino Rivadavia’ (Buenos Aires. Argentina). 
We computed die food niche breadth for each sample 
(FNB = 1/(Y pi : ) where pi is the proportion ot each 
class i in the diet (Orders = class) and the 
standardized food niche breadth iFNBst = FNB 
I)/(;? — 1) where n is the total number of prey classes 
(Marti 1987). A Principal Component Analyst 
(PCA) was performed using software Infostat Di 
Rienzo et al. 2010) to detect and describe changes in 
the composition and abundance of prey categories; 
i.c.. Orders Anura, Columbiformes. Passerifom*”'- 
Didelphimorphia. Chiroptera. Rodcntia (native anJ 
exotic species were considered separately following 
Clark and Bunck 1991), and Lagomorpha. 
