DISCUSSION
Although we found a high activity overlap between jaguars and pumas, such as other studies (Scognamillo et al . 2003; Harmsen et al . 2009; Harmsen et al . 2011; Foster et al . 2013), their activity pattern was significantly dissimilar. Corroborating the findings of Hernández-SaintMartín et al. (2013), we also observed a negative correlation between the peak of activity of jaguars and the peak of activity of pumas. Pumas used much more of the daylight compared to jaguars, especially in EP. This can be characterized as a time partitioning between these predators, where pumas might be negatively influenced by having smaller body sizes (Iriarte et al. 1990). Therefore, as pumas are subordinate to jaguars, they tend to avoid the peak of the activity of jaguars to reduce the probabilities of IGP and IK.
However, different temporal activity patterns, such as the case of jaguars and pumas, may allow each predator to make use of different preys more efficiently and thus, facilitate coexistence between competing species. Although we did not sample important prey species for jaguars, such as capybaras and caimans, this predator overlapped temporally with eight prey species in our studied area. Contrary, pumas are temporally overlapped with pacas in RD, with peccaries in EP, and in both areas with deers. The temporal overlap between the puma activity and the deer activity is not a surprise as deers are commonly found in puma’s diet (Scognamillo et al. 2003; Novack et al. 2005; Moreno et al. 2006), and the shift in the activity pattern found for pumas (i.e., nocturnal in RD, but cathemeral in EP) might be related to the deer activity patterns. Red brocket deers prefer forested areas (i.e., RD) and are nocturnal, whereas gray brocket deers are more generalist and diurnal (Ferreguetti et al. 2015). Therefore, it is possible that pumas increase the likelihood of preying on red brockets in RD by using more of the nighttime and gray brockets in EP by using more of the daytime. This shift in the activity pattern found for pumas reinforce the high plasticity of the species to adapt to different environmental conditions (De Angelo et al. 2011; Moss et al. 2016).
Ocelots with jaguars showed a high and significant temporal overlap in all areas and in RD with pumas, suggesting that coexistence might be facilitated by differences in other niche dimensions (Davies et al. 2010) or even facilitated by the low density of jaguars in our study area (Viana 2006), which might result in few encounters with this species (Davies et al. 2010). Also, ocelots prey mainly on rodents (Mezaet al . 2002; Moreno et al . 2006; Booth-Binczik et al . 2014) but can prey on primates (Bianchi and Mendes 2007; Abreuet al . 2008; Bianchi et al . 2010). Importantly, smaller preys are selected by ocelots when in sympatry with jaguars (Morenoet al . 2006). Therefore, ocelots are probably limited to smaller preys in our studied areas, which may reduce competition with the larger felids in this potentially competitive scenario, where ocelots are clearly a victim of IGP and IK due to its smaller body mass (Oliveira and Pereira 2014).
Crab-eating-foxes with jaguars, pumas, and ocelots showed a significant temporal activity overlap, but the species prefers areas with intermediate forest cover and broader trails (Goulart et al . 2009), which are characteristics found in EP. The crab-eating-fox activity pattern was dissimilar from pumas in EP, which combined to the low density of jaguars, may result in low risks of IGP and IK. Also, the great body mass differences with jaguars and pumas and the low density of jaguars may result in low risks of IGP and IK. Similarly, the small body mass difference between crab-eating-foxes and ocelots may reduce IGP and IK risks. Therefore, because the IGP and IK risk for the crab-eating-fox are unlikely, the species may benefit greatly for hunting preys (i.e., tapetis) in EP.
As expected, tayras showed a diurnal activity pattern and, therefore, no significant overlap was found with the temporal activity pattern of jaguars, pumas, and ocelots. Although mustelids have one of the highest competitive pressure among the American carnivore families, tayras are omnivorous and can use arboreal strata and aquatic environment, which may facilitate the coexistence with dominant predators (Hunter and Caro 2008).
Coatis with pumas in EP showed a significant temporal activity overlap, but as coatis are mainly diurnal, have an omnivorous feeding habits, and frequently uses the arboreal strata, it present low risks of IGP and IK by pumas. This is somehow expected, given the lowest competition pressure among the American carnivores of the Procyonidae family, possibly due to their ability to change the spatial and temporal use of resources (Hunter and Caro 2008).
Contrary to our initial hypothesis, most predators did not show less diurnal activity where human activity was higher (i.e., EP). Jaguars and coatis showed similar temporal activity in RD and EP. The few records of crab-eating-foxes and tayras did not allow us this comparison, but crab-eating-foxes only occasionally use RD. Only ocelots were more nocturnal in EP, but this might be a response to the tapetis activity pattern, an important and apparently abundant prey that also was more nocturnal in EP. Of course, this higher nocturnal activity of ocelots in EP also might result in fewer encounters with both pumas and humans, which was suggested to be a mechanism (or strategy) adopted by ocelots to allow its coexistence with pumas and humans in Atlantic Forest remnants (Massara et al., 2018). Surprisingly, pumas were more diurnal where the human activity is higher (i.e., EP), contradicting other studies (Paviolo et al. 2009; Carter et al. 2012; Schuette et al. 2013). We have two hypotheses to explain this result. Firstly, both study areas are so isolated and saturated with predators that competition is high and the benefits of hunting during the daytime outweigh the risks of encountering humans. Secondly, human activities in EP may not be as intense as we expected and because the eucalyptus management is limited mainly to smaller areas at each given time thus, leaving the rest of the area untouched, the encounters with humans are unlikely.
We observed that temporal partitioning contributes to the coexistence among predators in our studied areas. Pumas avoided conflicts with jaguars by using more the daytime, which increased also the likelihood of encountering diurnal preys, especially in EP. Ocelots avoided conflicts with pumas in EP by being more nocturnal, but the coexistence in RD with jaguars must be facilitated by the different use of preys. Tayras and coatis were diurnal, which might result in a low probability of agonistic encounters with nocturnal felids, besides the different niches among them. The temporal partitioning seemed unimportant only for crab-eating-foxes, but they probably coexist with felids by using the habitat and preys differently. In other words, our findings suggest that temporal partitioning contributed to the coexistence of predators by shifting the temporal activity pattern of the subordinate predator to hours that the dominant predator is less active, thus reducing the chances of direct conflicts. The shift in the activity pattern by the subordinate predator may also contribute to reducing competition by increasing its chances of encountering different preys.
However, in some cases the temporal partitioning did not occur or it was very subtle, since it depends on some other factors (or variables), such as the density of the dominant predators (Durant 1998; Davies et al. 2010), the other niche dimensions (e.g., diet) (Davies et al. 2010), the prey availability (Carrillo et al. 2009), and the human activities (Paviolo et al. 2009; Carter et al. 2012; Schuette et al. 2013), which needs to be investigated by further studies. We suggest that future studies use a combination of the spatial (e.g., telemetry data), temporal and trophic (e.g., diet data) dimensions of the niche to evaluate the drivers that may facilitate species coexistence in the Atlantic Forest fragments. Also, it is reasonable to expect that species will respond differently to human activities and matrix type and thus, these studies should also consider natural patches in different disturbance scenarios, which may also favor (or not) species dispersal. These studies are even more urgent for those more specialist species, such as jaguars, which are more susceptible to extinction in the current scenario of the Atlantic Forest.