The quality and availability of resources influence the geographical distribution of species. Social species need safe places to rest, meet, exchange information and obtain thermoregulatory benefits, but those places may also serve other important functions that have been overlooked in research. We use a large soaring bird that roosts communally in cliffs, the Andean condor (Vultur gryphus), as a model species to elucidate whether roost locations serve as a refuge from adverse weather conditions (climatic refuge hypothesis, CRH), and/or from predators or anthropogenic disturbances (threats refuge hypothesis, TRH). The CRH predicts that communal roosts will face in the opposite direction from where storms originate, and will be located in climatically stable, low precipitation areas. The TRH predicts that communal roosts will be large, poorly accessible cliffs, located far from human-made constructions. We surveyed cliffs used as communal roosts by condors in northwestern Patagonia, and compared them with alternative non-roosting cliffs to test these predictions at local and regional scales. We conclude that communal roosting places provide refuge against climate and disturbances such as, for instance, the threats of predators (including humans). Thus, it is not only the benefits gained from being aggregated per se, but the characteristics of the place selected for roosting that may both be essential for the survival of the species. This should be considered in management and conservation plans given the current scenario of global climate change and the increase in environmental disturbances.
Citation: Lambertucci SA, Ruggiero A (2013) Cliffs Used as Communal Roosts by Andean Condors Protect the Birds from Weather and Predators. PLoS ONE 8(6): e67304. https://doi.org/10.1371/journal.pone.0067304
Editor: Fabio S. Nascimento, Universidade de São Paulo, Faculdade de Filosofia Ciências e Letras de Ribeirão Preto, Brazil
Received: March 28, 2013; Accepted: May 16, 2013; Published: June 24, 2013
Copyright: © 2013 Lambertucci, Ruggiero. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: Funding provided by Agencia Nacional de Promoción Científca y Técnica de Argentina (ANPCyT) PICT 38148. The funder had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
The geographical distribution of a species is strongly influenced by resource availability. The quality of available resources can vary due to intrinsic characteristics as well as the ecological requirements of the species that use them . For birds, location of food, nesting and roosting sites can be essential . Particularly, for social species it is important to access safe places to rest, meet, exchange information or obtain thermoregulatory benefits . Knowledge about how a species selects this resource at different geographical scales can provide clues to the ecological aspects associated with habitat use and behaviour, and can be useful to establish management strategies –. Moreover, this information can be used to analyse the potential habitat for a species, and to understand behaviours from an ecological and evolutionary point of view , –.
Communal roosting is widely distributed among animals, and some of the proposed benefits of aggregation include the exchange of information for finding food, mate acquisition, and thermoregulatory purposes , –. There are a great number of studies devoted to the analysis of communal roosting behaviour in association with feeding habits , –. The potential advantages of communal roosting behaviour, unassociated with feeding behaviour, have been less studied (but see , , ). Moreover, the physical and environmental characteristics that may influence the selection of communal roosting places have also been rather overlooked .
Many species, including several raptors, use rocky cliffs as roosts, to rest or hunt , . In particular, large soaring birds depend on places that provide refuge from predators (i.e., safe places), and allow them to take off easily, among other requirements , . Hence, cliffs can be useful to analyse other potential advantages of communal roosting behaviour beyond its association with feeding habits. In particular, Andean condors (Vultur gryphus) roost communally in cliffs different from those in which they choose to nest (more detailed information is given in Methods, Study species) , .
The distribution and the aggregation patterns of condors in roosting places might be limited by the occurrence of roosts offering sun or protection, among other characteristics , . Here, we proposed that communal roosts may serve as refuges from adverse weather conditions (“Climate Refuge Hypothesis”, CRH). In northwestern Patagonia (Argentina), the climate is cold-temperate, seasonally variable in terms of temperature, wind and precipitation, and adverse weather (strong winds, snow, etc.) comes mainly from the west-northwest . Large flying birds are strongly limited by weather conditions, and avoid flying on rainy days . The CRH predicts that: (1) On a regional scale, Andean condor communal roosts will be located in places with favourable weather conditions. Thus, in northwestern Patagonia (Argentina), communal roosts will be associated with warmer temperatures, lower variability in temperature and lower precipitation compared with cliffs not used for roosting. (2) At a local scale, where regional weather conditions are similar, the cliffs used for communal roosting will face in the direction opposite to that of predominant yearly winds, snow and rain.
Rocky cliffs can also provide a safe environment from potential predators , , and this could favour the evolution of the formation of communal roosts in birds , . Condors face risks mainly when they are on the ground (e.g., eating), where they are very cautious –, and also when they are roosting. At present, Pumas (Puma concolor), foxes (Dusicyon sp.), dogs (Canis familiaris), and humans could be perceived as a threat to condors during roosting , , . Thus, if communal roosts offer refuge from threats posed by predation or anthropogenic disturbances (“Threats Refuge Hypothesis”, TRH), we predict that they will be in cliffs inaccessible to terrestrial predators and humans. To test this hypothesis at a local scale, we compared the geomorphologic characteristics of communal roosts with closed alternative non-roosting cliffs. On a regional scale, we predicted that if humans are perceived as a threat, cliffs used as communal roosts will be farther from towns, rural houses or roads, than alternative cliffs.
The Andean condor is a Near Threatened large scavenger inhabiting the Andes mountain ranges in South America, and has suffered important retractions in several areas of its original geographical distribution . Current populations are rare and reach maximum numbers in the southern area of the species’ distribution range . The Andean condor does not breed communally but uses communal roosts to overnight . Their nests may be located some hundreds of meters away in the areas surrounding their communal roosts, but can commonly be found as far as several kilometres away –, . Breeding adults frequently roost in the vicinities of the nest, but those roost-sites are used only by one individual rather than communal. Aggregations of individuals in the communal roosts may be more or less dispersed depending on the size of cliffs. Some of those shelves allow for the aggregations of dozens of individuals and others are small caves that can be used by one individual each. All sexes and age classes are represented in the communal roosts , , .
The study was carried out in a cold-temperate region in north-western Patagonia (Argentina) (ca. 40°–42°S and 70°–72°W). The Andes to the west act as a barrier to westerly winds at these latitudes, which produce variable weather conditions , and the predominance of winds blow from the west-northwest. The air masses from the Pacific Ocean are driven up and over the Andean mountain ranges causing the air to lose much of its moisture as precipitation on the Chilean side. Upon reaching the leeward side of the Andes the dry air descends and picks up any available moisture from the landscape below. This produces a high west-east gradient of precipitation in Argentina, where precipitation declines exponentially with distance from the Andes .
Towards the east, the Patagonian landscape is predominated by plains and erosive forms that have resulted in a number of outcrops serving as distinguishing characteristics of the local landscape. The number of roads and cars, and human density in the study area are low: 0.06 km of road/km2, from about 100 vehicles/day on gravel roads to 1550 vehicles/day on the national roads, and <0.6 inhabitants/km2 .
Choice of communal roosts and alternative cliffs.
We surveyed the south of Neuquén and the northwest of the Río Negro provinces in Patagonia looking for Andean condor communal roosts based on information from local people and researchers, and previous data on roost location , . Permission to study cliffs used by condors as communal roosts were provided by Dirección de Fauna Silvestre de Río Negro, the Argentine National Park Administration, and the owners and managers of local farms. We did not collect or manipulate birds in this study.
Proximity to food sources is a potentially important factor that may influence the presence of communal roosts in several species , , . However, condors have large home ranges and can fly more than two hundred kilometres per day in order to find food sources  (Lambertucci et al. unpubl. data). Moreover, the same bird can move between communal roosts over a period of days , . Nonetheless, to minimise the potential effect of food availability on our results we selected a geographically restricted study area that condors are able to cross in a few hours , and where the spatial variation of food resources is low (see below).
Data on a regional scale.
We mapped a total of 29 communal roosts, within a rectangular study area of 150 km north-south×90 km west-east, to analyse the ecological and geomorphological determinants of the occurrence of communal roosts at a regional scale. We generated 29 random geographic coordinates within the same area using the “animal movement” extension of ArcView v3.3 . These points were imported into Google Earth (http://earth.google.es/) to select the closest cliff to each random location. Then, we verified in the field that the location corresponded to a cliff not used by condors (i.e., no birds or faeces on the shelves). We estimated the same geomorphological climatic and anthropogenic variables for cliffs used and not used as communal roosts (see Table 1).
Data on a local scale.
We selected 24 communal roosts at the centre of our study area out of the total 29 communal roosts mapped within the region. We walked along the four cardinal directions looking for the closest cliffs with apparently similar topographical and environmental characteristics to the central cliff used as a communal roost, but with no indication of use. We selected 3 or 4 of those alternative cliffs, depending on their availability and accessibility to the research team. These locations were also characterized in situ by the same set of variables estimated for communal roosts (see variables in Table 1). In total, we surveyed 109 cliffs: 24 communal roosts, and 85 alternative cliffs.
We used the extension Grid Analyst v1.1 in ArcView v. 3.3. (ESRI, CA, USA) to assign values of five climatic variables to the geographic coordinates of each cliff based on the WorldClim digital database at a spatial resolution of 1 km2 (Hijmans et al. , available http://www.worldclim.org/current): 1) mean annual temperature, 2) mean annual precipitation, 3) diurnal mean range in temperature, 4) isothermality (diurnal mean range of temperature/annual range of temperature), and 5) seasonality of precipitation (Table 1).
We estimated the angular direction of the wind within our study region from records of wind direction registered by a weather station located in the town of Bariloche (41°07′04′′S–71°24′39′′W). This station registered 60961 records in 2007.
Geomorphologic variables and altitude.
Each cliff was characterized in the field by aspect, height (m), width (m), the top-floor altitude of the cliff (m above sea level), distances from borders to shelves and accessibility of the cliff (Table 1).
Distances to the closest human buildings were included as indicators of anthropogenic disturbance. We distinguished between living places (houses, farms, towns) and roads (Table 1). At a local scale, we recorded the distance from each communal roost to the first building without any discrimination. Because the distances between communal roosts and alternative cliffs were short (mean distance between communal roosts and alternative cliffs was ca. 1.3 km) it can be assumed that they were at the same distance from distant built-up areas (towns). At a regional scale, we separately estimated the distances to the closest human building (house or farm) and to the closest built-up area (village or town; Table 1).
First, we compared the environmental attributes between communal roosts and alternative cliffs using non-parametric statistics . We then applied logistic regressions  to determine the variables that better explained the use of a cliff for communal roosting, based on environmental predictors estimated at both local and regional scales; we included as a positive response (1) the communal roosts, and as a negative (0), the alternative roosts (Table 1). Variables were combined in the same statistical model provided they had a correlation of r <0.6 to reduce the problem of multicolinearity (see Table S1 in File S1).
We employed a multi-model selection approach  that involved an exhaustive search of all possible (single and multiple) logistic models to account for the presence of communal roosts. We included every possible combination of (low correlated, r<0.60) variables for each subset of predictors representing geomorphology, weather conditions, and anthropogenic disturbance. We used the Akaike’s Information Criterion (AIC, ), to find the best ecological model supported by our data among all possible models. For each group of predictors, the model with the lower AIC was selected as that model best supported by our data. Models with ΔAIC<2 from the best supported model were considered equivalents . When several models had a ΔAIC<2, from the best models we selected the most parsimonious (i.e., with the lower number of variables). We combined the effects of variables from each of the best (geomorphologic, anthropogenic and climatic) models, and performed an exhaustive search of the “best final model” based on AIC. We used the McFadden’s rho test as a measure of the variation explained by the best final model. The McFadden`s rho is similar to the coefficient of determination (r2) of a linear regression, although it shows lower values. Values of McFadden’s rho between 0.2 and 0.4 are satisfactory . We considered a coefficient estimate to be statistically significant when it was more than two standard errors away from zero .
The autocorrelation of variables across the geographic space is an inherent property of most ecological data, which often complicates the statistical testing of hypotheses by standard methods of analysis because it can inflate type I error rates, and may result in model instability . We used SAM v. 4.0  for previous analyses and to evaluate the effects of spatial structuring of variables on the performance of our environmental models. SAM v.4.0 allows the elaboration of a spatial correlogram using Moran’s (I) coefficient to describe the magnitude of spatial autocorrelation of variables for different distance classes. This coefficient range from −1 (maximal negative spatial autocorrelation) to +1 (maximal positive spatial autocorrelation), and values close to zero indicate no spatial correlation. We checked the adequacy of each environmental model through the examination of patterns of spatial autocorrelation in the residuals. Independent of the pattern of spatial autocorrelation in the original (predictors and response) variables, if no spatial autocorrelation is found in the residuals after including environmental predictors in the statistical model, then it can be concluded that the model has taken into account all spatial structure in the original data, and there is no statistical bias in the overall statistical analysis .
Finally, we applied circular statistics  to compare the aspect of roosts and alternative cliffs. We used the Rayleigh test to determine if the mean aspect angle, i.e. averaged over all roosts, differed from a random distribution around the 360° . We applied the Watson U2 test  to look for significant differences between the aspect of communal roosts and alternative cliffs, and to compare the aspect of communal roosts with respect to angular wind direction.
Test of the effect of site and food availability as potential confounding variables.
Since at a local scale communal roosts and alternative cliffs are located close to each other, we applied a logistic regression with “site” as the only explicative variable; given that there was no significant effect of site (Wald = 0.499, P = 0.99), we did not control for site effects in any subsequent analyses. We also evaluated the possible effect of food availability on the use of communal roosts. We must emphasize here that condors in our study area were able to come and go from any of the roosts studied  (Lambertucci et al. unpubl. data), hence it was reasonable to assume that food resources were not a limiting factor influencing the use of any cliff for roosting. Nonetheless, to confirm this assumption we selected a subset of 18 communal roosts and 17 alternative cliffs for which data on the density of livestock in the surroundings were available . Given that we did not find evidence that the presence of communal roosts was associated with the density of livestock (Logistic Regression Model including only livestock abundance as an explanatory variable, Wald = 0.097; P = 0.755), food availability was not included in any subsequent statistical analyses.
Overall Differences between Communal Roosts and Alternative Cliffs
Communal roosts were larger and less accessible, with shelves more distant from the borders of the cliff, than alternative cliffs. However, both were at similar altitude above sea level and at similar distances to the closest building or road (Table 2).
Factors Associated with the Presence of Communal Roosts at a Local Scale
The model including geomorphological variables only was the best supported by our data (Table 3; Table S2 in File S1). The presence of communal roosts was associated with wide cliffs of low accessibility, with a long distance (a mean of 23 m) from the floor to the lowest shelf, and eastern aspect. Cliff height, distance from the top to the highest shelf, and distances to roads and buildings had no significant association with the presence of communal roosts at a local spatial scale (Table 3). There was a very low (I = −0.12), negative spatial autocorrelation in the original response variable at the lowest distance class (<5 km); the lack of spatial autocorrelation in the residuals from the best geomorphological model (i.e., for all distant classes I<0.1; Fig. 1a) suggested that our statistical models were robust and adequate to account for the spatial variation in the use of roosts at a local scale.
Moran’s index to detect the presence of spatial autocorrelation (arrows = I<0.05) in the response variable “presence of cliffs” (black dots) and in the residuals (open dots) from the best logistic model fitted at a local (a), and at a regional scale (b) (see Methods).
Factors Associated with the Presence of Communal Roosts at a Regional Scale
A mixed model that combined climatic and geomorphological variables was the best supported by our data to account for the presence of communal roosts at a regional scale (Table 4; Table S3 in File S1). Communal roosts were located in places with high isothermality and low precipitation; they faced toward the east and had a high floor-shelf distance. Although the distance to town was an important predictor of the presence of communal roosts in the best anthropogenic model, it was not in the final mixed model (Table 4). The spatial autocorrelation present in the response variable near 100 km (I = −0.24) approached zero in the residuals from the best ecological model (Fig. 1b). This suggests that the environmental predictors adequately accounted for the use of communal roosts at large geographic scales within the spatial extent of our study. At the shortest distance classes (<30 km), spatial autocorrelation remained in the residuals after model fit, although it was low (I<0.23) and hence, this suggested that the effect of spatial autocorrelation did not severely influence the performance of our statistical model (Fig. 1b).
Cliff Aspect and Direction of Winds
The aspect of non-roosting cliffs in northwestern Patagonia was randomly distributed (Rayleigh, N = 82; Z = 1.10; P = 0.33; Fig. 2a). In contrast, cliffs aspect of condor communal roosts was not randomly distributed (Rayleigh, N = 24; Z = 7.11; P<0.001), and the mean angle was orientated toward the east-southeast at 108.9°N (r = 0.544; CI (95%) = 81.6°–136.1°; median = 107.5°; Fig. 2b). Consequently, the aspect of cliffs used as communal roosts and alternative cliffs differed (Watson, U2 = 0.42; P<0.001).
(a) Circular plots showing the distribution of aspects of cliffs (black points represent cliffs) not used as communal roosts in the northwest of Patagonia; (b) Aspect of 24 Andean condor (Vultur gryphus) communal roosts. Points in the radius of the circle correspond to the number of roosts. (c) Wind direction throughout 2007 in the NW of Patagonia. Each point corresponds to ca. 870 data points registered by the weather station. In every graph we indicate the values of the angles (numbers outside the circle), and for (b) and (c), the mean aspect and its confidence interval (CI 95%) with a thin black line.
The wind direction was not randomly distributed in north-western Patagonia (Rayleigh, N = 60961; Z = 12027.2; P<0.0001), and its mean angle of direction at 326°N (r = 0.444; CI (95%) = 325.6°–327.0°; median = 337.5°; Fig. 2c) is opposite to that of the aspect of communal roosts (Fig. 2b).
We found that the use of a cliff for roosting may involve a twofold selection process for sites that offer shelter against natural or anthropogenic threats while also serving as a refuge from unfavourable weather conditions at both local and regional scales. Thus, the proximity to food sources or other benefits of aggregation are not unique in their role in determining the use of a place as a communal roost as is often suggested , , , . At a local scale, we showed that the geomorphology of the cliff is important in distinguishing between roosting cliffs and those not used for roosting, which agrees with studies in other species (e.g., for a mammal: , for a bird: ). At a regional scale, cliffs used for communal roosting are located in places with low precipitation and more stable temperatures, which suggest that climatic variables may play an important role in the use of sites to overnight. The lack or low levels of spatial autocorrelation that remained in the residuals after the fit of our statistical models suggests that our results were robust and not severely influenced by the presence of spatial autocorrelation in our original data . As far as the local spatial scale of our analysis was concerned, the density of livestock did not account for the presence of communal roosts, suggesting that it is unlikely that proximity to food resources had been an important factor underlying the use of particular cliffs for communal roosting by the Andean condor.
We showed that climate was an important factor to account for the presence of communal roosts at a regional scale, thus supporting the climatic refuge hypothesis (CRH). Weather conditions are known to be very important in modulating the behaviour and habitat selection of many species –. The location of communal roosts coincides with sites of lower precipitation and lower variation in temperature rather than alternative cliffs. Moreover, condor communal roosts were typically located in large cliffs that faced in the direction opposite to the predominant winds. Cliff aspect may also affect the microclimatic conditions at the roosting place, as it influences the quantity of radiation (daily, seasonal and latitudinal), temperature, wind and precipitation received at a local scale . Cliff facing in the direction opposite to the wind can accumulate lower amounts of snow , and are frequently selected by different raptor species , .
It has been proposed that aggregated roosting imparts thermoregulatory benefits , . Although condors may benefit from aggregation in close proximity to conspecifics in the communal roosts (not evaluated in this work), our results suggest that the roosts themselves may provide thermoregulatory benefits due to physical and geographic characteristics associated with morphology, aspect and geographical location. Specifically, large cliffs, facing opposite to the predominant winds, and located in places with low temperature variability and low precipitation, may be suitable under adverse weather conditions , when communally roosting birds could be injured . Therefore, cliffs may be important climate refuges for condors, and thus it seems reasonable to predict that if future global warming alters the predominant weather conditions within the region , this might modify their quality as refuges.
On the other hand, cliffs selected as communal roosts were large, poorly accessible outcrops, which supports the idea that they may be good refuges from possible threats (predation or anthropogenic disturbances, TRH). The extent to which cliffs are accessible to predators seems to be relevant for communal roosting. We found that distance from the floor to the lowest shelf was associated with the presence of a communal roost. Cliffs with lower shelves, at short distances from the floor, could be more accessible to species that can be considered as potential predators of condors (e.g., pumas or humans) and, thus, dangerous to roosting birds. In contrast, the distance from the highest shelf to the top of the cliff was not important. This is not surprising given that the top of the cliffs are inaccessible for terrestrial species that may be considered as a potential predator. The dilution effect hypothesis proposes that the gathering of many individuals in communal roosts may reduce the individual depredation risk  with the centre of the roost being the safest . Our study favors the idea that condors use large and inaccessible cliffs to diminish predation and disturbance, but other explanations could also affect roost selection. For instance, large-heavy birds such as condors are limited in their capacity to fly , , , which presents the possibility that these birds use large cliffs, with high shelves, to help them take off. Moreover, communal roosts oriented toward the sunrise (east) favor thermal lift and benefit soaring flight. In the southern hemisphere, east and north-east facing cliffs are warmer, mainly during the morning, which could be important to increase thermal activity more so than would occur in west-facing cliffs and to warm the birds for an easier transition in leaving the location. On the other hand, the amount of overhang at the top of the cliff and above the shelves could also affect the terrestrial radiation useful during nighttime thermal conditions. None of these hypotheses are mutually exclusive, and deserve detailed consideration in future studies.
Although cliffs used as communal roosts were in general far from buildings, distance to town was not important in accounting for the presence of a communal roost in the presence of climatic and geomorphological variables. At present, the number of roads, car traffic and extent of human population is still low in our study area , which might explain why variables representing human disturbance were less explicative in our study. Other large avian scavengers are known to avoid human constructions selectively (California condor ,  Old world vultures , ); sometimes they can be tolerant to human constructions when they find a location appropriate, for instance, to rest or breed . However, this can pose other problems such as the introduction of competitor species, furtive hunting, and nesting failure , , , . Thus, as a precautionary principle, changes in land use should be considered and anthropogenic disturbances included in future evaluations of hypotheses on the use of communal roosts.
Previous studies that test hypotheses on communal roosting behaviour were mainly focused on the species’ decisions once the birds selected a particular roost (eg., , , ), but not on the identification of environmental characteristics that might be involved in the decision of which places were used to roost. Individual birds that visit a communal roost may be independently searching for places with particular characteristics , and we found that condor communal roosts typically provide protection against unfavourable weather conditions, and refuge from possible threats, supporting both the climatic and the threats refuge hypotheses. Warm, climatically protected, inaccessible roosting places can reduce the possible survival costs of overnighting in very cold places, or the mortality risk of being in places highly exposed to predators or human disturbances , , . Therefore, communal roosts can be a valuable resource and deserve special attention in the development of long-term conservation practices for condors, and likely for other species as well , , , under the current scenario of climatic change and increases in anthropogenic disturbances.
Table S1, Correlations between anthropogenic, geomorphological and climatic variables included in the statistical models. Table S2, Exhaustive search of best models to account for the presence of communal roosts at a local scale. Table S3, Exhaustive search of best models to account for the presence of communal roosts at a regional scale.
We thank all volunteers from Global Vision International, O. Mastrantuoni, F. Barbar, C. Lambertucci, H. Lambertucci and K. Speziale who assisted in the field during the collection of data. We also thank K. Bildstein, A. Margalida and an anonymous reviewer for useful comments on the manuscript. Managers and owners of La Buitrera del Ñirihuau, San Ramón, Siete Cóndores, Pilcañeu, and Cerro Negro farms, and the Argentinean National Park Administration gave permission to visit their lands to study the roosts. We thank the INTA and Centro Atómico Bariloche, especially S. García and M. Caputo, for providing the institutional weather stations reports.
Conceived and designed the experiments: SAL. Performed the experiments: SAL. Analyzed the data: SAL AR. Contributed reagents/materials/analysis tools: SAL AR. Wrote the paper: SAL AR.
- 1. Tilman D (1982) Resource competition and community structure. New Jersey, USA.: Princeton University Press.
- 2. Johnson DH (1980) The comparison of usage and availability measurements for evaluating resource preference. Ecology 61: 65–71.
- 3. Beauchamp G (1999) The evolution of communal roosting in birds: origin and secondary losses. Behav Ecol 10: 675–687.
- 4. Sutherland WJ, Newton I, Green R (2004) Bird ecology and conservation: a handbook of techniques. Oxford University Press, USA.
- 5. Battin J, Lawler JJ (2006) Cross-scale correlations and the design and analysis of avian habitat selection studies. Condor 108: 59–70.
- 6. Sergio F (2003) From individual behaviour to population pattern: weather-dependent foraging and breeding performance in black kites. Anim Behav 66: 1109–1117.
- 7. Ward P, Zahavi A (1973) The importance of certain assemblages of birds as “information-centres” for food-finding. Ibis 115: 517–534.
- 8. Blanco G, Tella JL (1999) Temporal, spatial and social segregation of red-billed choughs between two types of communal roost: a role for mating and territory acquisition. Anim Behav 57: 1219–1227.
- 9. Eiserer LA (1984) Communal roosting in birds. Bird Behavior 5: 61–80.
- 10. Weatherhead PJ (1983) Two principal strategies in avian communal roosts. Am Nat 121: 237–243.
- 11. Wright J, Stone RE, Brown N (2003) Communal roosts as structured information centres in the raven, Corvus corax. J Anim Ecol 72: 1003–1014.
- 12. Margalida A, Boudet J (2003) Dynamics and temporal variation in age structure at a communal roost of Egyptian vultures (Neophron percnopterus) in northeastern Spain. J Raptor Res 37: 252–255.
- 13. Hatchwell BJ, Sharp SP, Simeoni M, McGowan A (2009) Factors influencing overnight loss of body mass in the communal roosts of a social bird. Funct Ecol 23: 367–372.
- 14. Newton I (1979) Population ecology of raptors. Berkhamsted, U.K: Poyser.
- 15. Pennycuick CJ, Scholey KD (1984) Flight behavior of andean condors Vultur gryphys and turkey vultures Cathartes aura around the Paracas Peninsula, Peru. Ibis 126: 253–256.
- 16. Lambertucci SA, Luis Jácome N, Trejo A (2008) Use of communal roosts by Andean Condors in northwest Patagonia, Argentina. J Field Ornithol 79: 138–146.
- 17. Lambertucci SA, Mastrantuoni OA (2008) Breeding behavior of a pair of free-living Andean Condors. J Field Ornithol 79: 147–151.
- 18. Donázar JA, Feijóo JE (2002) Social structure of Andean Condor roosts: influence of sex, age, and season. Condor 104: 832–837.
- 19. Paruelo JM, Beltran A, Jobbagy E, Sala OE, Golluscio RA (1998) The climate of Patagonia: general patterns and controls on biotic. Ecología Austral 8: 85–101.
- 20. Hiraldo F, Donázar JA (1990) Foraging time in the Cinereous Vulture Aegypius monachus: seasonal and local variations and influence of weather. Bird Study 37: 128–132.
- 21. Larson DW, Matthes U, Kelly PE (2000) Cliff ecology: pattern and process in cliff ecosystems. Cambridge, U.K.: Cambridge University Press. 364 p.
- 22. Donázar JA, Travaini A, Ceballos O, Rodríguez A, Delibes M, et al. (1999) Effects of sex-associated competitive asymmetries on foraging group structure and despotic distribution in Andean condors. Behav Ecol Sociobiol 45: 55–65.
- 23. Speziale KL, Lambertucci SA, Olsson O (2008) Disturbance from roads negatively affects Andean condor habitat use. Biol Conserv 141: 1765–1772.
- 24. Lambertucci SA, Speziale KL, Rogers TE, Morales JM (2009) How do roads affect the habitat use of an assemblage of scavenging raptors? Biodivers Conserv 18: 2063–2074.
- 25. Koford CB (1966) The California Condor. Reprint Edition. Dover Publications. 154 p.
- 26. Lambertucci SA, Speziale KL (2009) Some possible anthropogenic threats to breeding Andean condors (Vultur gryphus). J Raptor Res 43: 245–249.
- 27. BirdLife International (2012) Species factsheet: Vultur gryphus. http://www.birdlife.org. Available: http://www.birdlife.org. Accessed 10 March 2013.
- 28. Lambertucci SA (2010) Size and spatio-temporal variations of the Andean condor Vultur gryphus population in north-west Patagonia, Argentina: communal roosts and conservation. Oryx 44: 441–447.
- 29. Lambertucci SA (2013) Variability in size of groups in communal roosts: influence of age-class, abundance of individuals and roosting site. Emu 113: 122–127.
- 30. De Martino E (2009) Estudio de home range y estacionalidad en el comportamiento de vuelo de ejemplares de cóndor andino (Vultur gryphus) liberados en la Argentina y Chile. Buenos Aires, Argentina: Universidad CAECE.
- 31. Alcaide M, Cadahía L, Lambertucci SA, Negro JJ (2010) Noninvasive estimation of minimum population sizes and variability of the major histocompatibility complex in the Andean condor. Condor 112: 470–478.
- 32. Hooge PN, Eichenlaub B (1997) Animal movement extension to ArcView ver. 1.1. Anchorage, AK, USA.: Alaska Science Center, Biological Science Office, U.S. Geological Survey.
- 33. Hijmans RJ, Cameron SE, Parra JL, Jones PG, Jarvis A (2005) Very high resolution interpolated climate surfaces for global land areas. Internat J Climatol 25: 1965–1978.
- 34. Conover WJ (1980) Practical nonparametric statistics. Wiley New York.
- 35. Allison PD (1999) Logistic regression using the SAS system: theory and application. North Carolina, USA: SAS Publishing.
- 36. Burnham KP, Anderson DR (2002) Model selection and multimodel inference: a practical information-theoretic approach. New York: Springer. 528 p.
- 37. Gelman A, Hill J (2007) Data analysis using regression and multilevel/hierarchical models. Cambridge University Press New York. 648 p.
- 38. Diniz-Filho JAF, Bini LM, Hawkins BA (2003) Spatial autocorrelation and red herrings in geographical ecology. Global Ecol Biogeogr 12: 53–64.
- 39. Rangel TF, Diniz-Filho JAF, Bini LM (2010) SAM: a comprehensive application for Spatial Analysis in Macroecology. Ecography 33: 46–50.
- 40. Batschelet E (1981) Circular statistics in biology. New York, USA: Academic Press.
- 41. Zar JH (1998) Biostatistical Analysis. 4th ed. Prentice Hall. 929 p.
- 42. INDEC (2002) Censo Nacional Agropecuario. Instituto Nacional de Estadísticas y Censos. Available: http://www.indec.mecon.gov.ar. Accessed 15 January 2013.
- 43. Walker RS, Novaro AJ, Branch LC (2003) Effects of patch attributes, barriers, and distance between patches on the distribution of a rock-dwelling rodent (Lagidium viscacia). Landscape Ecol 18: 185–192.
- 44. Peters KA, Otis DL (2007) Shorebird roost-site selection at two temporal scales: is human disturbance a factor? J Appl Ecol 44: 196–209.
- 45. Martin TE (2001) Abiotic vs. biotic influences on habitat selection of coexisting species: climate change impacts? Ecology 82: 175–188.
- 46. Sæther BE, Grøtan V, Tryjanowski P, Barbraud C, Engen S, et al. (2006) Climate and spatio-temporal variation in the population dynamics of a long distance migrant, the white stork. J Anim Ecol 75: 80–90.
- 47. Bauer S, Van Dinther M, Høgda KA, Klaassen M, Madsen J (2008) The consequences of climate-driven stop-over sites changes on migration schedules and fitness of Arctic geese. J Anim Ecol 77: 654–660.
- 48. Bartlett RM, Matthes-Sears U, Larson DW (1990) Organization of the Niagara Escarpment cliff community. II. Characterization of the physical environment. Can J Bot 68: 1931–1941.
- 49. Williams RN (1984) Eyrie aspect as a compensator for ambient temperature fluctuations: A preliminary investigation. J Raptor Res 18: 153–155.
- 50. Sarasola JH, Negro JJ, Salvador V, Maceda JJ (2005) Hailstorms as a cause of mass mortality of swainson’s hawks in their wintering grounds. J Wildlife Dis 41: 643–646.
- 51. Sala OE, Chapin FS, Armesto JJ, Berlow E, Bloomfield J, et al. (2000) Global biodiversity scenarios for the year 2100. Science 287: 1770–1774.
- 52. Elgar MA (1989) Predator vigilance and group size in mammals and birds: a critical review of the empirical evidence. Biol Rev Camb Philos Soc 64: 13–33.
- 53. Shepard ELC, Lambertucci SA, Vallmitjana D, Wilson RP (2011) Energy beyond food: foraging theory informs time spent in thermals by a large soaring bird. PloS ONE 6: e27375.
- 54. Snyder NFR, Ramey RR, Sibley FC (1986) Nest-site biology of the California Condor. Condor: 228–241.
- 55. Donázar JA (1993) Los buitres ibéricos: biología y conservación. J.M. Reyero, Madrid, Spain. 260 p.
- 56. Morán-López R, Sánchez JM, Costillo E, Corbacho C, Villegas A (2006) Spatial variation in anthropic and natural factors regulating the breeding success of the cinereous vulture (Aegypius monachus) in the SW Iberian Peninsula. Biol Conserv 130: 169–182.
- 57. Ellis DH, Craig T, Craig E, Postupalsky S, LaRue CT, et al. (2009) Unusual raptor nests around the world. J Raptor Res 43: 175–198.
- 58. Sánchez-Zapata JA, Calvo JF (1999) Raptor distribution in relation to landscape composition in semi-arid Mediterranean habitats. J Appl Ecol 36: 254–262.
- 59. Carrete M, Lambertucci SA, Speziale K, Ceballos O, Travaini A, et al. (2010) Winners and losers in human-made habitats: interspecific competition outcomes in two Neotropical vultures. Anim Conserv 13: 390–398.