DISCUSSION

Occupancy values varied across species and landscapes, however a clear pattern emerged when comparing mammal species shared between the two landscapes, that larger species have a higher occupancy in the OL while smaller species have a higher one in the MML. The few previous studies in these regions agree with these results. Gómez et al., (2016) found high levels (>0.55) of occupancy for ungulates (tapirs and both species of peccaries) in the eastern highland savannas of El Tuparro National Park; while Boron et al., (2019) reported occupancy levels even lower than the ones reported here (<0.30) for smaller mammals (Cuniculus, Dasypus, Eira) in a human-modified landscape in the Middle Magdalena Valley. No studies however, had compared the same species across landscapes.
While differences in occupancy values between landscapes for a same species could be expected naturally (different ecosystems and context), the pattern of size-related occupancy suggests that human modification of the landscape is a determining factor. Populations of larger species like tapirs, pumas and giant ant-eaters have been reduced in the Middle Magdalena modified ecosystems. Larger mammals tend to be more affected by deforestation and human pressures like hunting, as they require larger areas of habitat and are more susceptible to hunting (Chichorro et al., 2019; Forero-Medina et al., 2009; Ripple et al., 2017). On the other hand, smaller, common species (not endemic, small-ranged), and especially rodents, habitat generalists with high reproductive rate, may become more abundant at sites with substantial human use (Dias et al., 2019; Gibb et al., 2020). Such is the case for Cuniculus ,Dasyprocta and Dasypus , in the MML.
Both natural and human-mediated factors influenced occupancy of species in the two landscapes. However, in the less modified areas of the OL, mostly natural factors, such as area of flooded grassland, area of different types of forests and distance to water bodies were determinant, both for unique species and for species shared with the MML. Only two species in the OL were influenced by distance to human settlements, the puma and the tapir. On the other hand, in the more disturbed MML, ten out of the thirteen species were influenced by anthropic factors determining occupancy, such as distance to human settlements and the area of artificial land cover types. The area of forest was also an important factor in this landscape. Most notably, all shared species except the puma and tapir (same variable opposite effect), were influenced by a different factor in each landscape, usually moving from an anthropic factor in the MML to a natural factor in the OL.
These differences could highlight how in the more modified landscape, where forests are now represented by small, fragmented patches surrounded by crops and other agricultural land cover types, human-mediated factors have a stronger influence on the presence or absence of some terrestrial birds and mammals. This effect can be negative or positive, with some species apparently thriving in more disturbed conditions, while others avoiding them. Previous studies in tropical ecosystems have shown how many species of mammals and other game vertebrates are affected by human activities, like hunting, and therefore their abundance or occupancy increases away from human settlements or in no-hunting areas (Cavada et al., 2019; Nuñez-Iturri et al., 2008). On the other hand, some species of rodents and small mammals can benefit from such changes in the landscape, being favored by human settlements, through the increase in open areas, water reservoirs and food availability from garbage or agriculture (Fedriani et al., 2001; Mendelssohn & Yom-Tov, 1999; Oro et al., 2013). Species of Dasyprocta, Cuniculus and Pecari have been reported as some of the main consumers of human-cultivated plants (crop raiders) in the Amazon (Abrahams et al., 2018), and armadillos have been reported to consume arthropods associated with cultivated lands (Gallo et al., 2019). In the MML, these rodent species and armadillos are increasing their occupancy closer to human settlements or in open, disturbed areas, perhaps as a consequence of the presence of crops; however, these same species in the OL, are more influenced by natural variables like distance to water, and area of riverine forests. In a less disturbed site (like the OL) natural factors become more important.
Understanding the effects of natural and anthropic variables in vertebrate occupancy in these two sites is further complicated by inter-specific interactions (Table 1). Species can interact directly or indirectly, through competition, mutualism and trophic relations, all of which can be mediated by other variables. In the MML, all interactions found were positive, and were influenced by the distance to human settlements, while in OL species interactions were mediated by natural factors.
We did not find evidence of exclusion between Cuniculus andDasyprocta, two rodents of similar size. In the MML there were no interactions between them and in OL the occupancy of each species was higher in the presence of the other. However, in the MMLDasyprocta occupancy was positively correlated with the presence of Puma and Jaguar, which in turn were positively correlated with the presence of Cuniculus , one of their preferred prey (Ávila-Nájera et al., 2018; Gutiérrez-González & López-González, 2017; Novack et al., 2005), raising the question of a potential competition release between the rodents by these predators, who could be reducing the abundance of Cuniculus (not eliminating it), favoring the presence ofDasyprocta and explaining its higher occupancy in the presence of the predators. Curiously, the relationship between Puma andDasyproct a is inverted in the OL, where the interaction becomes negative in both directions. This could be related to the fact that the occupancy of both Dasyprocta and Cuniculus is much lower in the OL, therefore there may not be competition release by the predators but instead Dasyprocta avoids sites occupied by Puma because without Cuniculus it also becomes a common prey. While specific experimental designs would be necessary to test these hypothesis, results clearly indicate how species interactions may vary from landscape to landscape as a result of interactions with other species and the influence of environmental or human-mediated variables.
Both Puma and jaguar occupancy was influenced by distance to human settlements, with both preferring higher distances away from humans in the MML, which has less remnant forest and more human-predator conflict, while the Puma was closer to humans in the OL, where there is less conflict with predators and less human density. There was no exclusive interaction between these two predators in the MML, on the contrary, the occupancy of each species was higher in the presence of the other. This is not rare and has been documented before (Scognamillo et al., 2003), and could be related to the fact that although their diets can be somehow different in a well-preserved ecosystem, in a disturbed site like the MML predators will necessarily have to occupy the few forest remnants, where habitat and preys are available.
In the MML occupancy of predators and prey presented opposing relationships with distance to human settlements. Apparently prey species are favored by the heterogeneity created by human activities, while the felids like jaguar and puma tend to distance from these areas, perhaps as a consequence of conflict killings. This indicates a spatial segregation, mediated by human presence, between predators and preys in this study, a pattern that has been reported before in temperate systems (Dorresteijn et al., 2015; Muhly et al., 2011). In natural systems, prey species like Cuniculus and Dasyprocta prefer dense forests and are influenced by water bodies, to escape from predators (Goulart et al., 2009; Péres, 1992), just as it was observed for the OL, where these variables affected their occupancy and mediated their co-occurrence patterns with Puma.
Despite the limitations of extrapolating a local study to broader taxonomic and geographic categories and regions, our work presents a contribution to the understanding of the way in which human activities impact species and the way they interact with each other in these tropical landscapes. Considering the characteristics of the MML, it is hard to isolate the effects of the different human activities, habitat loss, fragmentation, hunting, subsidies for certain species (food, garbage, etc). However, this work clearly shows that although the effects are not simple and in the same direction across all species, humans play a key role in the structuring of vertebrate communities. Although species richness is similar between the two studied areas, the probability of occupancy of species is very different, generalist species present higher occupancy values in the MML, but lower interactions with predators. These changes, coupled to the low occupancy of large seed disperses like the tapir and active seed predators like the peccaries, can result in significant changes in the structure and composition of the vegetation and the ecosystems (Terborgh et al., 2008). In this sense, understanding such changes in vertebrate occupancy, both positive and negative, and the factors that cause them, should be a key question to inform conservation and restoration initiatives for vertebrate species and their habitat.