2.1 Study area

The Araripe-Apodi National Forest is located within the APA-Araripe in southern Ceará, between latitudes 07º 11’ 42” S and 07º 28’ 38” S and longitudes 39º 13’ 28” W and 39º 36’ 33” W. This conservation unit has a total area of approximately 38,626 ha, and includes parts of the municipalities of Barbalha, Crato, Nova Olinda, Jardim, Missão Velha, and Santana do Cariri, all located in the Brazilian state of Ceará (^{Brasil, 2004}). This conservation unit includes areas of cloud forest, savanna, savanna woodland, and carrasco scrub (^{Austregesilo Filho et al., 2001}). The climate of the Araripe National Forest is tropical rainy, with mean annual precipitation of 1368 mm, 80% of which is concentrated in the rainy season, between December and April, with a peak in March. The dry season lasts between five and seven months, peaking in September, which is usually the hottest and driest month (^{Brasil, 2004}).

The study area was delimited by the survey transects and adjacent areas. The area was estimated based on the geometric figures formed by the transects – an equilateral triangle and a right-angled triangle. The research reported here was authorized by the Brazilian government through permit number 29970-3 emitted by the Biodiversity Authorization and Information System (SISBIO).

2.2 Survey of guan populations

Quantitative data on the guan populations were collected in line transect surveys, using the standard procedures developed for the study of Neotropical vertebrate faunas (^{Buckland et al., 2001}; ^{Cullen Junior and Rudran, 2003}). The surveys were based on standardized walks of the pre-established transects at a constant speed of no more than 1 km/h (^{Buckland et al., 1993}). Distance sampling depends on five basic assumptions: (a) all animals on the trail are detected; (b) the perpendicular distances are measured with the maximum possible precision; (c) all animals are detected before they move; (d) all sightings are independent events, and (e) the same animal is not detected more than once during a given survey (^{Buckland et al., 2001}; ^{Cullen Junior and Rudran, 2003}).

For each sighting of a guan, the following data were recorded (see ^{Cullen Junior and Rudran, 2003}): (a) date and time; (b) species; (c) number of individuals sighted; (d) perpendicular distance between the animal or group and the transect (using a surveyor’s measuring tape).

The surveys were conducted along two existing transects of 4 km in length (Figure 1) that traverse the natural vegetation of the National Forest. The transects were walked between 05:00 h and 09:00 h during eight days per month (four days on each transect) between November, 2011, and October, 2012. During sightings, the animal detected during the surveys were observed using Nikon Monarch 8 × 42 DCF binoculars. The surveys were interrupted during days with intense precipitation.

Figure 1 Location of the Araripe Environmental Protection Area (APA-Araripe) and the Araripe-Apodi National Forest. The arrow indicates the study area surveyed in the present study of P. superciliaris and P. jacucaca. The red line represents transect 1 and the yellow line, transect 2. The white lines delimit the estimated area surveyed. 

2.3 Feeding records

The composition of the diet was defined based on two complementary methods: (i) the observation of feeding behavior along trails and feeding stations, and (ii) the analysis of fecal samples. Behavioral observations were conducted along pre-existing trails, which were walked randomly in order to guarantee the greatest possible number of encounters during fruit-feeding events. Feeding records were collected whenever one or more guans were observed consuming the fruits, flowers or leaves of a given plant species. Whenever the subjects moved to a new feeding resource during the same observation session, a new record was initiated, as proposed by ^{Altmann (1974)}. Sites at which a concentration of fruiting plants was found were used as observation points, referred to here as feeding stations.

2.4 Collection of fecal samples

The fecal samples were collected along the transects and at the feeding stations. The feces were easily identified due to their large size and presence of large seeds that are rarely ingested by smaller birds, as well as their uric acid content, which distinguishes bird feces from those of mammals (^{Mikich, 2002}). The identification of the samples was supported by the observations of the animals’ foraging behavior along the transects, where the animals would often defecate.

The fecal samples were stored in lidded plastic containers containing 70% alcohol for transportation to the Zoology Laboratory of the Cariri Regional University (URCA) in Crato, Ceará (Brazil). The samples were triaged using a stereoscopic microscope. The samples were removed, rinsed under running water using a 1 mm-mesh sieve for the separation of the items (seeds, leaves, and flowers). The seeds were sun-dried in Petri dishes, and identified through comparisons with material deposited in herbaria. The leaves and flowers were discarded once identified.

2.5 Collection of plant samples

Samples of the reproductive parts of all the plants exploited by the guans for their fruit or other structures were collected and prepared as herbarium specimens, for identification with the assistance of specialists, using the classification system of the Angiosperm Phylogeny Group (^{APG III, 2009}). The voucher specimens were deposited in the Dárdano de Andrade Lima herbarium of the Pernambuco Agronomics Institute (IPA) and the Caririense Dárdano de Andrade Lima herbarium (HCDAL) at URCA.

2.6 Analysis of the diaspores

Six fruits and seeds were collected randomly from each of five different plants (separated from one another by a distance of at least 10 m) representing each of the species identified in the diet of the guans. The length and breadth of each diaspore were measured using a digital caliper. The morphology and coloration of the fruits were classified following the scheme of ^{Barroso et al. (1999)}.

2.7 Data analysis

The density of the guan populations was estimated using the Distance 6.0 software (^{Thomas et al., 2010}). This program identifies a model or detection function that best represents the set of perpendicular distances recorded during the surveys. This function is then used to estimate the proportion of individuals that were not detected during the surveys, ultimately providing an estimate of population density (^{Buckland et al., 2001}; ^{Cullen Junior and Rudran, 2003}). The total size of the species’ population can be estimated by multiplying the density of the size of the study area. Sighting rates – number of individuals sighted per 10 km walked – were also calculated, to provide an index of abundance (^{Strahl and Silva, 1997}).

3.1 Abundance, density, and population size

A total of 384 km of transect were walked during the present study period, 192 km on each trail, during which Penelope superciliaris was encountered on 50 different occasions. Penelope jacucaca was not sighted during these surveys.

Based on the Akaike Information Criterion (AIC), the Distance program selected the half-normal model as the best fit for the data (Figure 2), which provided an estimate of 11.66 groups per square kilometer (km²), with a CI of 14.55-25.48 and CV of 13.98 (Table 1). Population density was estimated to be 19.17 individuals/km². The study area was estimated to cover 6.23 km² (623 ha).

Figure 2 Detection probability function for the population density estimate of Penelope superciliaris. The histogram represents the observed values and the curve, the expected values. 

3.2 Feeding ecology

A total of 52 feeding events were recorded during the present study, when the guans (P. superciliaris) were observed consuming the fruits and seeds of 14 different plant species. Most events occurred at heights of up to 5 m above the ground (Table 2). A total of 241 seeds were found in the 36 fecal samples collected during the study period. Fourteen (38.8%) of the fecal samples contained zoochoric seeds, while three (8.3%) had leaves and seeds, one (2.7%) contained only flowers, and one other (2.7%), fragments of insects.

Penelope superciliaris consumed the seeds of 14 plant species belonging to 12 families and 14 genera. Eleven feeding records were collected during the rainy season (December–April), while 15 were collected in the dry season (May–November). The consumption of six plant species – Miconia albicans, Tabernaemontana sp., Buchenavia capitata, Byrsonima sericea, Senna rugosa, and Eugenia puniciflora – was confirmed by 17 of the 36 fecal samples. The species found most frequently in the fecal samples was Buchenavia capitata, observed in nine samples, containing 30 seeds. The Myrtaceae was the most diverse family, being represented by three species, Myrcia multiflora, Psidium sp., and Eugenia punicifolia (Table 3).

3.3 Characteristics of the fruit consumed by the guans

Fruit samples of 12 of the 14 species consumed by Penelope superciliaris were collected, measured and identified. Mean diameters ranged from 6.3±1.35 mm in Miconia albicans to 29.9±1.7 mm in Pisidium sp. (Figure 3). Eight (66%) of the 12 species analyzed had fruits of between 15 mm and 30 mm in length (Table 2). Most fruits (n=5 or 41.6%) were yellow in color, with two species (16.6%) each having black, green, and red fruits (Table 2). Fleshy fruits of the baccate (50.0%, n=6) and drupe (33.3%, n=4) types were the most consumed by P. superciliaris (Table 2).

Figure 3 Mean size of the fruits consumed by Penelope superciliaris in the Araripe National Forest between November, 2011, and October, 2012. The vertical lines represent the standard deviation of each mean value. Plant species: 1. Anacardium microcarpum; 2. Senna rugosa; 3. Ocotea pallida; 4. Bryrsonima sericea; 5. Miconia albicans; 6. Tabernaemontana sp.; 7. Buchenavia capitata; 8. Eugenia punicifolia; 9. Myrcia multiflora; 10. Psidium sp.; 11. Matayba guianensis; 12. Protium heptaphyllum. 

4.1 Abundance, density, and population size

Reliable estimates of density are necessary for the systematic monitoring of populations over time and evaluate the impacts caused by anthropogenic disturbances (^{Hoyo, 1994}). Consistent and comparable population parameters are essential for the planning, implementation, and evaluation of management strategies, as well as conservation measures (^{Sutherland, 2000}).

The population density of guans (Penelope spp.) varies considerably (^{Setina, 2009}), although few data are available for Brazilian populations. In the forested areas of the Brazilian Pantanal wetlands, ^{Bernardo and Desbiez (2011)} estimated a density of 4.66 individuals/km² for the endangered Bare-faced curassow (Crax fasciolata), which is threatened by hunting and habitat destruction. ^{Bernardo et al. (2011)} provided density estimates for the Black-fronted piping Guan (Pipile jacutinga) from 11 protected areas in the Atlantic Forest biome of São Paulo, southeastern Brazil, that varied in area from 11,100 to 150,000 hectares (in comparison with 623 ha in the present study). Density estimates at these sites varied from 0.23 to 16.2 individuals/km², with sighting rates of 0.05 to 0.70/10 km, and group densities of 0.20 to 9.80/km², whereas in the present study, values of 19.17 individuals/km², 11.66 groups/ km², and 0.13 sightings/10 km, respectively, were recorded.

Studies of Penelope have been conducted in a number of other South American countries. Estimates of population density for Spix’s Guan (Penelope jacquacu) from Manu, Peru, varied between 2 and 19.8 individuals/ km² (^{Torres, 1997}). In the Andes of southern Ecuador, estimates of the population density of the Bearded Guan (Penelope barbata) in three different types of habitat within a study area of 400 ha varied from 2.3 to 17.1 individuals/km² (^{Jacobs and Walker, 1999}). In a study of the Andean Guan (Penelope montagnii) in the Guadera Biological Reserve in Ecuador, ^{Cresswell et al. (1999)} recorded densities of 20 individuals/km² on farmland, and 40 individuals/km² in a 650 hectare tract of primary forest. ^{Kattan et al. (2006)} surveyed populations of the Cauca Guan (Penelope perspicax) in two protected areas in Colombia, recording a density of 8.6 individuals/km² in one area of 559 ha, and 41.6-100 individuals/km² in the other area, of 489 ha. In a second study of P. perspicax in a 459 hectare protected area in Colombia, ^{Rios et al. (2008)} recorded a density of 31 individuals/km² in forested areas, and 88 individuals/km² in plantations of Andean Oak and Urupán.

The population density estimated for P. superciliaris in the present study was consistent with the values recorded in most of these other studies of cracids. Variations in population parameters may be related to the fragmentation of habitat, resource availability, seasonal migrations, and possibly also minor differences in study methods (^{Bernardo and Desbiez, 2011}). The Araripe National Forest is inserted within a much larger conservation unit (APA-Araripe), covering a total area of 10,000 km², and it is important to note that the estimates recorded in the present study provide researchers and reserve managers with only an approximate notion of the characteristics of the natural populations that persist in the region. As population parameters are related to the availability of resources, the considerable diversity of habitat types, and presumably also dietary and other resources within both protected areas may influence the movements of P. superciliaris within the region.

Groups of P. superciliaris may move freely to areas surrounding the National Forest in search of alimentary resources, nesting sites, and areas with reduced hunting pressure, which may all influence population density. Hunting practices in the conservation unit and surrounding areas (^{Brasil, 2004}) may also have a negative impact on population parameters. This is an important question, given that, while population parameters in the Araripe National Forest may be influenced by the movements of individuals in search of resources, they may also reflect the result of the history of hunting both within and around this protected area. While P. superciliaris may be considered to be a common bird (^{BirdLife, 2010}), its present-day populations are being impacted increasingly by habitat destruction and unsustainable hunting pressure, in particular within the Caatinga biome. The combination of hunting pressure and habitat destruction has contributed to a rapid decline in cracid populations in recent decades, in particular guans and curassows, which are primarily forest-dwellers (^{Brooks and Strahl, 2000}).

An important point raised by the present study was the absence of the White-browed Guan (Penelope jacucaca) in the surveys conducted in the Araripe National Forest. This species is endemic to the Brazilian Caatinga biome, but is also found in the transition zones with the Cerrado savanna and Atlantic rainforest and is known to occur on the Araripe Plateau (^{Brasil, 2008}). The biology of this species is still poorly-known (^{Brooks and Fuller, 2006}), even though it is classified as threatened with extinction (^{Machado et al., 2008}), given its vulnerability to hunting pressure and habitat disturbance (^{Silva et al., 2003}; ^{Machado et al., 2008}). ^{Brooks and Strahl (2000)} identified this species as one of the highest priority bird taxa for conservation. The most recent record of the species in the Araripe National Forest (^{Nascimento et al., 2000}; ^{Minns et al., 2010}) were obtained in areas of carrasco scrub, savanna, and dry forest, which may account for the lack of records in the present study, which focused primarily on more humid forest formations.

During a study of Lear’s Macaw (Anodorhynchus leari) in Serra Branca, in the Brazilian state of Bahia, ^{Lima et al. (2003)} recorded the presence of P. jacucaca frequently in the local Caatinga habitats. In Ceará, the species is known to occur in the Pentecostes Caatinga, and in the Mãe-da-Lua Private Natural Heritage Reserve in the municipality of Itapajé, where populations of more than 50 individuals were observed in 2012 (^{Redies, 2013}).

4.2 Feeding ecology

The Araripe National Forest is characterized by a mosaic of vegetation types (^{Austregesilo Filho et al., 2001}), which may favor the production of a diversity of feeding resources for P. superciliaris, as recorded in the present study, where feeding was recorded in trees of different heights – some over 10 m – along the trails surveyed. Forests with good cover are important for most cracids, especially guans and curassows, which are normally observed feeding in the canopy (^{Mikich, 1996}; ^{Guix and Ruiz, 1997}; ^{Brooks and Strahl, 2000}). In southeastern Brazil, ^{Mikich (2002)} observed a preference of P. superciliaris for fruit in the upper strata of the forest. At Araripe in the dry season months (August–November), however, the Rusty-margined Guans tended to forage in smaller groups and in the lower strata of the forest and on the ground (pers. obs.).

A number of previous studies have also confirmed a relatively frugivorous diet for P. superciliaris throughout the year (^{Mikich, 1996}, ²⁰⁰¹; ^{Zaca, 2003}), although the consumption of leaves, flowers, and insects was also confirmed in the present study. The available data indicate that these resources may be important alternatives during the dry season and the transition from the rainy to the dry season (^{Silva and Strahl, 1991}). ^{Sick (2001)} noted that the guan diet is composed of fruit, leaves, and shoots, and refers to the descent of the animals to the ground to retrieve fallen fruits or to drink water at river margins, where they may also ingest sand. Other cracids, such as the Red-billed Curassow curassow (Crax blumenbachii) may also feed on mollusks, grasshoppers, tree-frogs, and spiders.

In the Itapetinga Municipal Park in Atibaia, São Paulo (southeastern Brazil), ^{Zaca (2003)} found that the consumption of leaves and flowers varied significantly (χ²=61.42; p<0.001) over the 20 months of the study period, with two peaks in the dry season. A number of studies have also recorded animal prey, such as invertebrates in the stomach contents of guans (^{Théry et al., 1992}; ^{Merler et al., 2001}). ^{Sick (1970)} and ^{Teixeira and Snow (1982)} identified a number of the fruit species consumed by Crax blumenbachii. ^{González-García (1994)} recorded the consumption of 40 plant species by the Horned Guan (Oreophasis derbianus) in Mexico, ^{Caziani and Protomastro (1994)} analyzed the diet of the Chaco Chachalaca (Ortalis canicollis) in Argentina, while ^{Galetti et al. (1997)} reported the exploitation of 41 plant species by Pipile jacutinga in the Intervales State Park in São Paulo, Brazil.

Few data are available on the feeding ecology of Penelope. ^{Sick (1970)} reported the consumption of the fruit of a number of plant species, such as Virola bicuiba, Byrbicuiba sp., Lecythis pisonis, and Geonoma sp., by Crax blumenbachii and mentioned that these species were also exploited by P. superciliaris. ^{Mikich (1996)} recorded the consumption of Cabralea canjerana by P. superciliaris, while ^{Guix and Ruiz (1997)} reported the exploitation of Syagrus romanzoffiana by a number of Penelope species, based on observations of feeding behavior and the analysis of fecal samples.

In a study of the diet of P. superciliaris at three sites in southeastern Brazil, ^{Mikich (2002)} recorded the consumption of 55 species of fruit, of which 21 were recorded during observations and 34 were encountered in fecal samples. In the Itapetinga Municipal Park, also in southeastern Brazil, ^{Zaca (2003)} recorded 52 species, 12 through direct observation, and 40 in fecal samples. The number of species recorded in the present study was relatively low in comparison with these previous findings, especially the fecal sample data, although this may have been at least partly related to certain methodological differences among the studies, such as the collection of fecal samples at nighttime roosts during the two previous studies. Despite efforts to locate roosts during the present study, it was not possible to collect fecal samples from these locations. The chances of encountering feces containing seeds at these sites appear to be greater than in general, given that the birds may often use such roosts continually during fruiting peaks (^{Zaca, 2003}).

One other reason that may have contributed to the reduced number of fruit species recorded in the present study is the fact that the Chico Mendes Biodiversity Institute (ICMBio), which administrates the Araripe National Forest, has authorized the harvesting of pequi (Caryocar coriaceum) in this conservation unit by the local populations. This fruit is typical of the Brazilian Cerrado savanna, and is the principal source of income for many of the communities in the area surrounding the national forest. During pequi season, large numbers of harvesters use the trail system of the national forest to locate fruiting trees, and almost certainly interfere with the behavior of the local guans, which probably retire to more isolated and less accessible parts of the area, reducing the potential for observation and the collection of fecal samples.

Another important question here is that, during 2012, Ceará suffered the sixth worst drought recorded since 1950, according to data provided by the Ceará State Foundation for Meteorology and Hydrological Resources. Between January and May, only 352.1 mm of precipitation was recorded, 50.4% less than the average for the first five months of the year (^{FUNCEME, 2012}). Climatic variations and changes in environmental variables, such as rainfall and sunlight levels, temperatures, and winds, may affect reproductive phenology and alter fruiting patterns significantly (^{Laurance et al., 2003}). In the present case, the reduced precipitation of the study period may have provoked a marked decrease in the availability of fruit, in particular.

In southeastern Brazil, ^{Zaca (2003)} recorded 11 species of the family Myrtaceae in the diet of P. superciliaris, and as in the present study, this family contributed the largest number of species included in this guan’s diet. ^{Mikich (2002)} recorded two species – Campomanesia xanthocarpa and Eugenia florida – in the diet of P. superciliaris, while ^{Paccagnella et al. (1994)} also recorded this plant family in the diet of the cracid Pipile jacutinga. The importance of the Myrtaceae in the diet of cracids has been reported in other studies, such as those of ^{Théry et al. (1992)} and ^{Mikich (2002)}.

One notable exception recorded in the present study was the consumption of the fruit of the Fourleaf Buchenavia, Buchenavia capitata, which was recorded during observations and in nine fecal samples, with a total of 30 seeds. This species has not been recorded in previous studies of P. superciliaris, although it has been reported in the diets of other birds and primates (Weaver, 1991 apud^{Tabarelli et al., 2004}). In the Araripe National Forest, this tree attracts large groups of Rusty-margined Guans between June and August, leading to an increase in group size and observations of feeding behavior (pers. obs.). Local ICMBio staff and residents of the surrounding communities reported that the best period to hunt P. superciliaris coincided with the peak Buchenavia fruiting season, when large agglomerations of guans are observed visiting these trees. These observations indicate that Buchenavia capitata is a key species in the ecology of the P. superciliaris population in the study area.

4.3 Fruit characteristics

^{Mikich (2002)} recorded fruits of between 10 mm and 100 mm in size in the diet of P. superciliaris, but a predominance of those 20 mm in size. As in the present study, ^{Théry et al. (1992)} reported that most of the fruits consumed by the Marail Guan, Penelope marail were less than 30 mm in length, a pattern also observed in Pipile jacutinga by ^{Galetti et al. (1997)}. In southeastern Brazil, ^{Zaca (2003)} recorded P. superciliaris consuming fruits varying in size from 0.4 mm (Miconia cinnamomifolia) to 22.3 mm (Diospyros inconstans), although most species were between 4 mm and 16 mm.

In the present study, P. superciliaris consumed fruits of a variety of sizes, similar to that recorded by ^{Zaca (2003)}. The beak of this guan permits the ingestion of both small and principally large fruits. In some habitats, the seeds of large fruits may be less likely to be dispersed than those of smaller fruits, given their avoidance by birds with small beaks (^{Roda, 2003}). ^{Silva and Tabarelli (2000)} estimated that approximately one third of the tree species of the Atlantic Forest biome of northeastern Brazil, north of the São Francisco River, may be vulnerable to extinction due to the absence of large birds able to ingest and disperse their seeds. This reinforces the importance of the contribution of P. superciliaris and other cracids to the population dynamics of many plant species.

A number of studies have shown that red and black fruits are among the most common in southeastern Brazil, as well as other Neotropical regions and the Old World (^{Mikich, 2002}; ^{Zaca, 2003}). As in the present study, however, ^{Théry et al. (1992)} found that most of the fruits consumed by Penelope marail were either yellow or black. Fleshy dehiscent, indehiscent, and dry dehiscent fruits appear to be the most common zoochoric diaspores (^{Mikich and Silva, 2001}), although ^{Théry et al. (1992)} and ^{Zaca (2003)} also recorded a predominance of berry and drupe type fruits in their studies.