Circadian activity patterns and temporal overlap among cracids (Aves: Cracidae) within a vegetation mosaic in the Pantanal of Rio Negro, Brazil

Vertebrates, overall, present a daily activity pattern when managing their needs, such as foraging, resting or searching for sexual partners. Most of the available information regarding the circadian rhythm in birds comes from controlled laboratory conditions, and little is known about these patterns in the wild. In this study we used camera traps to describe the daily activity patterns of three cracid species in the Pantanal of Rio Negro, Brazil. We had a sampling effort of 9,617 camera trap-days along 231 days (5,544 hours) from September 2013 to May 2014. This resulted in 4,833 independent records of cracids from a total of 7,713 individuals. Crax fasciolata was the species with the most records (nrec = 3,792) and individuals (nind = 5,781), followed by Ortalis canicollis (nrec = 934; nind = 1,758) and Aburria grayi (nrec = 107; nind = 174). None of the species was uniformly distributed throughout the day, thus evidencing a periodization of their activities. The mean vectors of the activity patterns of C. fasciolata, O. canicollis and A. grayi were, respectively, mμ = 10:36 ± 04:26 (SD), mμ = 11:42 ± 03:57 and mμ = 11:44 ± 03:47. We observed a temporal overlap between A. grayi and O. canicollis, whereas C. fasciolata significantly differed from them. Because of their large home ranges, cracids are important indicators of environmental quality, and, as frugivores, they play key roles in the ecological dynamics of forests. In this sense, and given that cracids are notably more susceptible to extinction, the knowledge on their circadian activity patterns may be useful when establishing effective management and conservation strategies.


INTRODUCTION
The daily activity period, or circadian cycle, corresponds to the 24-hour span which establishes the biological cycles of nearly all living species. Such patterns have existed throughout the history of life on Earth, and virtually every life form has had its rhythm adapted to it (Beale et al., 2016). Activity patterns may be influenced by abiotic factors, such as photoperiod, lunar cycle, temperature, precipitation, tides, and latitude, as well as by biotic factors, like food availability, reproduction, and interactions with predators and humans (Kronfeld-Schor & Dayan, 2003;Pita et al., 2011;Ross et al., 2013;Bennie et al., 2014;Sassi et al., 2015;Díaz-Ruiz et al., 2016;Gaynor et al., 2018). Knowing how species explore their habitats and spotting the basic resources and conditions for their occupation, survival and reproduction is essential for stakeholders when determining potential threats and planning efficient management programs (Willems & Hill, 2009;Morrison et al., 2006).
Animals perform a number of activities within their circadian cycle (e.g., foraging, resting or searching for reproductive partners) that may concentrate in specific times of the day or night. The activity pattern refers to the continuous period in which an individual concentrates its daily actions (Maffei et al., 2002;Monroy-Vilchis et al., 2011). The circadian cycle is involved in the regulation of several daily functions, including the patterns of greater activity (Pittendrigh, 1981;Laposky et al., 2008;Cassone, 2014). Therefore, as ecosystems grow in complexity, organisms go on to explore not only their physical dimensions but also their temporal niches (Hut et al., 2012). As a consequence, all life forms share physiological and behavioral processes for specific times of the day (West & Bechtold, 2015). Behavioral plasticity, although limited in some species, becomes evident in others from observations of intraspecific variation in foraging patterns (Ashby, 1972;Hertel et al., 2016) that may lead to changes in the temporal niche (Fenn & MacDonald, 1995;Ensing et al., 2014).
Monitoring rare cryptic species is not a feasible task for researchers. Camera traps have been used to overcome this and successfully record, monitor and estimate the density of enigmatic taxa (Heilbrun et al., 2006;Kelly & Holub, 2008). This tool is highly flexible and has been used for fauna inventories (e.g., Jiménez et al., 2010), relative abundance indexes (Monroy-Vilchis et al., 2011), and to describe activity patterns (Hernández-SaintMartín et al., 2013;Lafleur et al., 2014) and land and habitat use (Michalski et al., 2015;Blake et al., 2017;Pardo et al., 2017;Pérez-Irineo & Santos-Moreno, 2018). In birds, most of the available information regarding the circadian rhythm comes from controlled laboratory conditions, while data from the wild remains poorly reported. Studies from the last decade have shown the efficacy of camera traps for studying large birds in the wild in Brazil, especially in the Atlantic Forest (Srbek-Araújo et al., 2009Kuhnen et al., 2013), and in other countries (Negret et al., 2015).
In this study we deployed camera traps in the Pantanal of Rio Negro to describe the daily activity patterns of three cracid species (A. grayi, O. canicollis and C. fasciolata) and to check whether or not they overlap in time.

Study area
The study was performed in a 26,500 ha area in the Pantanal wetlands of Rio Negro, in Aquidauana, Mato Grosso do Sul state, Brazil (Silva & Abdon, 1998). Most of the surveyed area (22,300 ha) was within the Barranco Alto ranch (19°33′35″S, 56°09′22″W, 100-120 m a.s.l.). The remaining sampling sites were located in a contiguous section of the Diacuí and Vera Lúcia ranches. The landscape is characterized by several bays and salt lakes, intertwined with open savannas, dense shrublands, fields and gallery forests. Non-flooded forests compose the most elevated areas. Secondary savannas and planted pastures occur in areas of greater anthropic influence (Silva et al., 2011). The climate in the region is classified as tropical wet (Aw) in the Köppen-Geiger's system (Kottek et al., 2006), with rainy summers (June to September) and dry winters (December to March). The mean minimum and maximum annual temperatures range from 21° to 33℃, respectively (Soriano & Alves, 2005), while the mean annual precipitation is approximately 1,350 mm (Bergier et al., 2018).

Sampling design and data collection
The birds were recorded by camera traps (Bushnell Trophy Cam HD E3). This method is recognized for its versatility and efficacy in obtaining ecological data, mainly when detecting rare species and determining their activity periods, as it provides a big volumes of information simultaneously (Hernández-SaintMartín et al., 2013;Lafleur et al., 2014;Pardo et al., 2017). Data was collected over nine months, from September 2013 to May 2014. The traps were spread over a grid of 90 sampling stations (Fig. 1). All sites were at least 500 m apart from each other to avoid sampling gaps. The cameras possess infra-red sensors activated by heat and movement and were placed approximately 40 cm above the ground so that the targeted animal could be recorded 2-3 m away (Tobler et al., 2008). The traps were set to work 24 h/ day and make 60-second videos for every record, with 30-second intervals between successive records. The sites were checked every 30 days for battery and memory card replacement.

Data analyses
All registered animals were identified to species level. When the species could not be determined, the records were classified at the smallest scale within the following: 1) genus; 2) family; 3) order; 4) class; 5) not identified. Every video made for the sampled species was considered a record. They were subdivided in: 1) total, which are all records obtained, and; 2) independent, excluding repeated records of the same species at the same sampling site in intervals ≤ 20 min (Tobler et al., 2008). Every record informed the recorded species, the number of individuals, date, and time of day. Videos depicting more than one species were considered independent records for each of them.
The sampling effort was calculated by multiplying the number of stations by the number of sampled days (1 day = 24 hours), and the catch rate by dividing the number of records by the sampling effort (Srbek-Araújo & Chiarello, 2005). The area covered by the camera traps was measured by the minimum convex polygon method (MCP) using all of the sampling stations. We performed a Shapiro-Wilk test for inspecting the normality of abundance distributions throughout the day.
We used circular statistics through the Rayleigh uniformity test (Z-test) to determine the circadian preferences of each species (Jammalamadaka & SenGupta, 2001). This test is employed to observe whether the records are uniformly distributed along a day. Our null hypothesis was a uniform circular distribution of the data, while our alternative hypothesis was a non-uniform distribution, with peaks of activity along the day. For this, the 24 hours of the day were distributed along 360°, with 1 h = 15° or 1 min = 0.25°. The mean vector (mμ), generated in degrees, was then converted into hours and corresponded to the mean hour of greater activity. We performed a nonparametric Mardia's two-sample test (Watson's U²) to check for overlap between the mean active hours of the species (Mardia, 1967;Batschelet, 1981;Zar, 2010). The null hypothesis acknowledges that all samples come from a single population, whereas the alternative hypothesis considers that populations differ in their mean active hours and present time partitioning. All these analyses were performed on R 3.5.1 (R Development Core Team, 2011) with the packages vegan 2.5-5 (Oksanen et al., 2020), circular 0.4-93 (Agostinelli & Lund, 2017) and plotly 4.9.2.1 (Sievert, 2020). We adopted significance at p < 0.05.

RESULTS
We had a sampling effort of 9,617 camera trap-days along 231 days (5,544 hours). The cameras covered an area of 23,391 ha in a 60.3 km perimeter, according to the MCP. We found a total of 26,559 records of vertebrates, 4,833 (18.2%) of which were independent records of the three studied cracids, C. fasciolata, O. canicollis and A. grayi, with 7,713 individuals detected. Crax fasciolata was the most registered (N rec = 3,792) and abundant species (N ind = 5,781), followed by O. canicollis (N rec = 934; N ind = 1,758) and A. grayi (N rec = 107; N ind = 174). The varia- tion in their abundances throughout the day had a nonnormal distribution (Shapiro-Wilk normality test for all species, p-values < 0.05). The distribution of the active time of the three species was non-uniform along the 24 hours, which evidences a periodization of their activities (Table 1). The mean vectors of the activity patterns of C. fasciolata, O. canicollis and A. grayi were, respectively, mμ = 10:36 ± 04:26 (SD), mμ = 11:42 ± 03:57 and mμ = 11:44 ± 03:47.
Watson's U² test showed a temporal overlap between A. grayi and O. canicollis, while their peak of activity differed significantly from C. fasciolata (Table 2).

DISCUSSION
Our results reveal that the three studied species (C. fasciolata, A. grayi and O. canicollis) share the activi-ty patterns of other cracids. For instance, the red-billed curassow (Crax blumenbachii) has a peak of activity in the early morning that gradually decreases until midday, restarts in the afternoon and peaks again before dusk (Srbek-Araújo et al., 2012;Hernández-SaintMartín et al., 2013;Fernández-Duque et al., 2013;Schaaf et al., 2014;Pérez-Irineo & Santos-Moreno, 2018). Some other species, however, show distinct temporal patterns. The Sira Curassow (Pauxi koepckeae), in the Peruvian Andes, stays active between 10:00 a.m. and 06:00 p.m. (Beirne et al., 2017). The razor-billed curassow (Pauxi tuberosa) is active mostly in the morning, whereas the Spix's guan (Penelope jacquacu), in the Xeruã river, Amazonas, peaks at midday (Saebø, 2016). The incompatibility with the bimodal activity pattern in such species may be a result of the environments they inhabit, i.e., mountain and tropical forests, with scarce solar radiation on the ground vegetation that allows animals to forage longer along the day (Saebø, 2016;Beirne et al., 2017).
Even though the activity periods of some cracids have been previously documented (e.g., Crax and Penelope), studies focused on A. grayi are deficient. The semi-arboreous habits of this bird may be responsible for the low number of records in our study. This species feeds primarily on leaves, seeds and fruits, which are collected in the canopy, and eventually forages on the ground (Schubart et al., 1965;Sick, 1997). In Trinidad and Tobago, the congeneric Trinidad piping-guan A. pipile occupies canopies above 5 m (Hayes et al., 2009). In that same study, A. pipile was registered more frequently in the morning and late afternoon, similarly to A. grayi in the Pantanal.
Contrary to Aburria, species in the Crax genus primarily inhabit the forest floor (Delacour & Amadon, 1973) in areas that either are or are not prone to seasonal floodings (Michalski et al., 2015;Costa et al., 2018). Seasonally floodable areas like the Pantanal present remarking differences in the density and richness of animal species between distinct seasons, which is driven by the effects of the flooding cycles on fruit production and habitat use by the animals (Haugaasen & Peres, 2007Alvarenga et al., 2018). Behavioral adaptations arise under environ-   mental pressure, while changes in temporal distribution may also occur according to the heterogeneity of the site (Presley et al., 2009;Reyes-Arriagada et al., 2014). In our study, C. fasciolata was more active in the morning, with a gradual decrease in activity along the day. This pattern had already been observed for an Argentinean population in another camera-trap study (Fernández-Duque et al., 2013). In the Vale Natural Reserve in the Espírito Santo state, Brazil, activities of C. blumenbachii peaked between 06:00-07:00 a.m. (Srbek-Araújo et al.,  2012), a similar pattern to what we observed for C. fasciolata in the Pantanal. Camera-trap assessments in Central America and other parts of South America revealed bimodal diurnal activity patterns for plain chachalaca O. vetula, great curassow C. rubra, and C. fasciolata, with activities beginning between 06:00-07:00 a.m., reaching a first peak still in the morning and another less evident one in the afternoon (Hernández-SaintMartín et al., 2013;Fernández-Duque et al., 2013). The peak in activity for O. canicollis also happened in the morning, between 07:30-09:30. Nevertheless, this species presented a more homogeneous pattern throughout the day compared to the two other cracids.
Camera-trap samplings have shown that temporal segregation is an important strategy for the coexistence of ecologically similar species (e.g., Di Bitetti et al., 2010;Monterroso et al., 2014;Sunarto et al., 2015). In the present study, the overlap in activity was prominent between A. grayi and O. canicollis but differed significantly for C. fasciolata. When resources are abundant, species tend to vastly overlap their use (Pianka, 1981). Estevo et al. (2017) observed a great overlap in time of activity between two sympatric land birds, brown tinamou Crypturellus obsoletus and tataupa tinamou C. tataupa. A study on the coast of the Argentinean Patagonia showed a high isotopic niche overlap among 14 seabird species during mating season that was maintained by a superabundance of food (Forero et al., 2004). Moreover, similar patterns of co-occurrence between species sharing the same food resources at the same times of the day have also been widely observed for mammalian carnivores (Davis et al., 2018). Therefore, resource availability could be acting as a limiting agent to the co-occurrence of cracids in the Pantanal.
Cracids are important indicators of environmental quality, as they occupy large territories. They are frugivorous animals with a fundamental role in forest dynamics (Sedaghatkish, 1996;Jordano et al., 2006;Muñoz & Kattan, 2007;Bueno et al., 2013;Galetti et al., 2013Galetti et al., , 2016. Studies on activity patterns and interspecific co-occurrence present a straightforward application in the conservation strategies for protected areas, as they incorporate a zonal approach aiming at habitat protection for the cracids and also for other codependent species (Leite et al., 2018). Continuous multiannual camera-trap studies would allow a better comprehension of how local alterations and anthropic disturbance affect the distribution and activity of species in this taxon.
This study presents the first data on the activity patterns of cracid birds in the Pantanal of Rio Negro obtained through camera-trapping, a reliable tool for the evaluation of activity rhythm in vertebrates (Ridout & Linkie, 2009;Monterroso et al., 2013;Torretta et al., 2017;Mori et al., 2020). The presented data contribute substantially to fill the information gaps concerning our target species and their activity patterns, particularly in the Pantanal of Rio Negro.
The activity patterns described here suggest a clear niche partitioning within their circadian cycles. Furthermore, these patterns are similar to results previously found for other cracids. Given the environmental heterogeneity of the Pantanal, differences in behavior along the day may as well be guided by the capacities of each species, as the high habitat and resource availability allows these animals to occupy a wide variety of niches. Therefore, investigating how organisms occupy a landscape along time may provide support for researchers to better understand the patterns in community structure and interspecific coexistence.
From a conservationist viewpoint, cracids are particularly relevant as they have highly specific ecological requirements and significantly different daily activity patterns allowing species to coexist. Because they are seed predators and dispersers and perform essential services for the dynamics of vegetal communities, understanding their life history is of great conservation interest.

FUNDING INFORMATION:
We are grateful to Jaguar Conservation Institute (JCF) for support this research project. Field research in Pantanal was also supported by Barranco Alto ranch.

ACKNOWLEDGMENTS:
We thank the team of the Ornithology Lab in CESC/UEMA, especially Surama Pereira, Hilda Raianne and Anderson Teixeira, for the help to survey and identify the collected material. We are grateful to Natália M. Tôrres, Ananda Barban and all collaborators of the Jaguar Conservation Institute (JCF). Ana Luiza Catalano kindly revised the English version of the manuscript. We thank Fábio Maffei and Lucas Lima for reviewing the preliminary version of this text.