The study was conducted in a 7000 km² agricultural area in northwestern La Pampa province, Argentina (Fig. 1). This area lies within the Pampas phytogeographical province and its transitional zone (ecotone) with the Espinal phytogeographical province (Soriano et al., 1991). The climate in this region is classified as dry sub-humid, with average annual rainfall ranging from approximately 350–850 mm, predominantly occurring in spring and summer (Lorda et al., 2008). In the late 18th century, vegetation consisted of temperate grasslands dominated by various tussock grasses, with an absence of trees (Cabrera, 1976). However, this landscape has been significantly transformed by intensive mixed agricultural and livestock activities (Caviglia et al., 2010). Currently, the vegetation comprises a mosaic of crops (including cultivated pastures, alfalfa, corn, sunflower, soybean, wheat, and barley), low grasslands, and exotic tree stands primarily used for shade for cattle and as windbreaks. Some patches of native forest, mainly composed of Caldén (Neltuma caldenia), remain in the area, originating from the surrounding Espinal phytogeographical province (Williamson, 1967; Cano, 1980; Soriano et al., 1991; Merenson et al., 2004).

Fig. 1.

Study area and geographic location of the sample sites corresponding to Aplomado Falcon breeding territories in La Pampa, Argentina, showing political and biogeographic boundaries.

Between September and January of 2010–2014, intensive surveys were conducted throughout the study area to locate Aplomado Falcon (Falco femoralis) individuals and breeding pairs. Field searches covered approximately 10,000 km annually. During this period, falcons were more conspicuous and easily detected through reproductive and territorial behaviors such as courtship displays, copulations, and agonistic interactions. Once nests were located, they were georeferenced using GPS and monitored every 10–15 days to minimize disturbance, from discovery until fledging or breeding failure. Territories were revisited annually to confirm occupancy, and some were re-sampled in multiple years to collect pellets and prey remains. Field procedures followed standard methodologies for raptor breeding surveys and diet studies (Liébana, 2015; Marti et al., 2007; Macías-Duarte et al., 2004).

To assess the influence of landscape cover on the trophic habits of the Aplomado Falcon and its role as an ecosystem service provider in pest control, we established a buffer zone with a radius of 1000 m around each nest (approximately 314 ha). This radius corresponds to the mean distance between the nearest nests in our study area (Liébana, 2015). For each buffer, we generated a shapefile by manually digitizing all land covers using a georeferenced Google Earth® image (≈1 m per pixel). The land covers within each polygon were verified in the field to create a Geographic Information System (GIS) for each Aplomado Falcon territory, which were classified into four types: exotic tree stands (primarily composed of Eucalyptus spp. and Ulmus pumila), native forest (dominated by Caldén), agricultural lands (including crops, pastures, grasslands and stubble), and peri-domestic areas (comprising residences, surrounding structures and associated parks or gardens; roads were not included, as they were unpaved, narrow, and represented a negligible proportion of the landscape). This analysis allows us to better understand how landscape composition affects the availability of prey species and, consequently, the falcon's foraging behavior.

During the reproductive seasons (September to January) from 2010 to 2013, we collected pellets and prey remains from nests and roost sites belonging to 60 reproductive territories, some of which were sampled repeatedly in different years (Fig. 1). Samples were collected during each visit to the nesting territories throughout the breeding season, following the same general schedule as nest monitoring. However, collection frequency varied among territories and years depending on accessibility and breeding activity. During each visit, potential perching and feeding sites near the nests were also inspected for additional remains. All samples were collected following standardized protocols: they were carefully removed from under nests and roosts, placed in labeled containers, and preserved by air-drying to prevent decomposition and fungal growth. In the laboratory, pellets were manually disaggregated, and prey items were sorted for identification (Marti et al., 2007). From all the samples (pellets and preys remains) birds were identified to the species level using feathers, beaks, and other remains, supported by reference collections and, when necessary, through the examination of feather microstructures (nodes and barbs) (Reyes, 1992). Mammals were identified using skulls, teeth, and hairs, employing keys (Chehébar and Martín, 1989; Pearson, 1995) and reference collections. Insects were identified to family level based on mandibles, heads, elytra, and other parts using reference materials. To determine the number of individual prey items in each sample, we counted mammal and bird skulls. For arthropods, we considered whole heads, legs, elytra, and mandibles. When only hairs, bones, or feathers were found, they were treated as a single individual. Unidentified prey items were recorded and included in total prey counts; their proportion relative to the total diet was noted but excluded from species- or taxon-specific analyses.

For each identified taxon, we calculated its relative frequency (% Ni), defined as the number of individuals of prey i over the total number of prey expressed as a percentage, and the biomass (% Bi), calculated as the proportion of the total biomass consumed represented by prey i. The mean mass of vertebrate prey was sourced from literature (Camperi and Darrieu, 2005) and our own data. For arthropods, we employed a mean weight of 1 g, following Vargas et al. (2007). We used the Shannon diversity index (H') to assess dietary diversity and Levins' niche breadth index (B), as these metrics provide critical insights into resource use patterns, feeding flexibility, and potential responses to habitat changes in managed landscapes (Colwell and Futuyma, 1971).

To estimate prey availability and abundance within the reproductive territories, we conducted bird surveys in each territory during the reproductive seasons (September–January) from 2010 to 2013. Surveys were carried out using fixed-point counts with a 200 m radius and a 10-minute duration per point, conducted in the morning (7:00–10:30) under favorable weather conditions (Bibby et al., 1992). All birds seen or heard were included in the counts, including flying individuals, as falcons are capable of capturing prey in flight, making these individuals part of the available prey pool. Due to logistical constraints, only 1–2 counts were conducted per territory; when more than one count was performed, these were temporal replicates at the same point. Each count point was located as close as possible to the nest while ensuring adequate visibility of the surrounding area. This approach together with the analysis of diet composition allows us to link prey availability to the falcon's functional role in agricultural landscapes.

Because each sampling point was visited only once or two times per breeding season, detection probabilities were not estimated, and abundance values should therefore be interpreted as indices of relative abundance rather than absolute population estimates. This limitation was considered when comparing prey availability with dietary data.

For statistical analyses, prey items were grouped into three main categories representing the dominant trophic components of the Aplomado Falcon diet: Columbiformes, Other Birds, and Arthropods. The Other Birds category included primarily Passeriformes, but also a few non-passerine species of similar size and ecological role (e.g., Piciformes and Cuculiformes). To determine whether the analyzed land covers explained variation in the consumption of the main prey types, we performed Generalized Linear Models (GLM; McCullagh and Nelder, 1989) with a quasibinomial error distribution to address the observed overdispersion. The dependent variable was the relative frequency of prey (e.g., the number of arthropods out of total identified prey for a territory sample), while the explanatory variables were the land cover types of each territory. Year was included as a fixed factor in the models to account for potential interannual variation in prey availability and consumption. For modeling purposes, we only included territories where a minimum of ten prey items were identified (n = 29). Taxa such as Mammalia, Reptilia, and Amphibia were excluded from the analysis due to their representation being below 5% of the total biomass consumed, and we included only one nesting pair from different years to avoid pseudo-replication.

Additionally, we constructed linear models to test the effects of land cover and year on Levins’ standardized food-niche breadth and Shannon diversity index values for each nest. A stepwise removal procedure was applied, excluding all variables with p > 0.05, resulting in a final model that included only significant effects (Hosmer and Lemeshow, 1989). We employed likelihood ratio tests (LRT) to compare models differing by the inclusion of a focal variable, with associated chi-squared values taken as a measure of support for retaining the focal variable in the model (Crawley, 2007). Residuals of each model were examined post-fitting by examining residual plots to verify assumptions of normality, homoscedasticity, and absence of influential outliers (Crawley, 2007). For binomial and quasibinomial GLMs, we calculated dispersion parameters and adjusted for mild overdispersion where needed. Finally, to verify the association between bird prey consumption and their abundances in the territories, we performed a Spearman rank correlation between the proportion of each bird prey species in the diet and their corresponding abundance in the field (n = 47 nesting territories). Due to limited prey sample sizes per year and territory, we pooled diet and survey data across years for each territory to obtain more stable estimates. While this approach reduced temporal resolution, it allowed us to retain sufficient data for a conservative, non-parametric correlation analysis. All statistical analyses were conducted using R version 3.1.2 (R Development Core Team, 2011).

The percentages of mean land uses (±SE, range in brackets) around Aplomado falcon nesting sites in the study area during the 2010–2013 breeding seasons confirm that agricultural lands are the predominant surface, accounting for 90.8% (±20.7; 0–102.9) of the landscape surface. Among the other three cover types, native forest was the largest, accounting for 5.3% (±14.8; 0–56.1), followed by exotic tree stands with 2.3% (±3.5; 0–15.3) and peri-domestic areas with 1.6% (±2.5; 0–13.1). Within the category previously described as agricultural lands, the pastures and grasslands were the predominant cover type, representing 55.9% (±34.3; 0–40.5) of the surface. This was followed by stubbles with 23.6% (±32.3; 0–24.8) and crops with 20.5% (±33.4; 0–34.6).

During the study period, we collected and analyzed 589 regurgitated pellets and 278 prey remains, identifying a total of 2119 prey items belonging to 83 species across four classes of vertebrates (Aves, Mammalia, Reptilia, and Amphibia) and two classes of arthropods (Insecta and Arachnida) (Table 1). In terms of relative frequency, Aplomado Falcons primarily consumed arthropods (52%), followed by birds (44%). The other groups did not exceed 5% occurrence: reptiles (%Ni = 2.5), mammals (%Ni = 1.9), and amphibians (%Ni < 1) (Table 1). Birds contributed the most biomass to the diet of Aplomado Falcons (83.1%), followed by reptiles (11.9%), mammals (3.9%), arthropods (1.1%), and finally amphibians (<1%) (Table 1). Among the birds, columbiformes were the main prey species, representing 94% and 91% of the relative frequency and biomass, respectively, and constituting 66.1% of the total biomass. Eared Doves were the most common item among the total prey items (Table 1). The standardized food-niche breadth observed for the Aplomado Falcon in the study area was relatively high, with a value of 0.7. Totals for the main prey groups (Columbiformes, Other Birds, and Arthropods), used in subsequent analyses, are shown in Table 1.

Regarding the influence of land cover on diet composition, tree stand cover was the only variable retained in the model for the occurrence of Columbiformes in the Aplomado Falcon's diet (χ² = 13.45, p = 0.027). This effect was positive and significant (β = 0.02 ± 0.01), indicating that falcons nesting in areas with larger tree stand surfaces consumed higher levels of this prey type. Model diagnostics for this GLM indicated mild overdispersion (ĉ = 2.93); thus, a quasibinomial model was applied.

For Other Birds (mostly Passeriformes), tree stand cover also had a positive effect (β = 0.04 ± 0.01, χ² = 23.28, p = 0.01), and prey consumption differed among years (χ² = 38.59, df = 3, p = 0.01). In contrast, arthropod consumption was negatively associated with tree stand area (β = –0.04 ± 0.008, χ² = 67.1, p = 0.008) and with the proportion of natural forest (β = –0.007 ± 0.0029, χ² = 18.05, p = 0.0029). No strong deviations or influential outliers were observed upon inspecting residuals, and model fit was considered adequate across all GLMs.

Although the effect sizes were moderate, their direction and significance consistently support the interpretation that wooded habitats (especially tree stands) increase avian prey availability, while more open or forested areas support a higher proportion of arthropod prey. This pattern suggests a landscape-mediated shift in trophic composition.

During prey sampling (2010–2013), a total of 34,630 individuals were recorded in the study area, representing 63 species belonging to 23 avian families (Table 2). The Eared Dove was the most abundant species (79% of the total bird species recorded). A positive association was confirmed between the abundance of birds in the field and their appearance in the diet of the Aplomado Falcon (Fig. 2; n = 47 nesting territories; ρ = 0.73; p < 0.001).

Fig. 2.

Relationship between the relative abundance of bird species in the study area (based on point count surveys) and their frequency in the Aplomado Falcon’s diet in La Pampa, Argentina. Each point represents a bird species detected both in diet samples and in surveys. The green line represents the Spearman rank correlation between prey abundance and dietary frequency across pooled data from all territories (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article).