RESULTS

We detected 38 species of mammals in the camera trap surveys, 27 occurring in the Middle Magdalena, 23 in the Orinoquia landscape, and 12 in both (Table S1). The probability of detection ranged from 0.09 to 0.90 in the MML and from 0.01 to 0.68 in the OL. In general, for shared or similar species across landscapes, detection was higher in the MML, with the exception of the tapir and ocelot. Occupancy rates varied from 0.09 to 0.99 across landscapes, with larger species like the giant ant-eater, the puma and the tapir (> 30 Kg) having higher occupancy rates in the OL, while smaller species like the agouti, paca and armadillo had higher occupancy in the MML (Figure 1, F1,16= 9.5, P=0.007).
The variables influencing occupancy differed from one landscape to another. For species occurring only in the MML, the area of secondary vegetation in the cell, distance to human settlements, and area of forest were most determinant, while for those species occurring only in the OL, distance to water, area of natural savannah and distance to human settlements were the key predictors of occupancy (Figure 2). For species occurring in both landscapes, the most important variable influencing occupancy always changed from one landscape to another (Figure 2). For medium-sized herbivore mammals and birds, occupancy was influenced mainly by distance to human settlements and area of artificial pastures in the MML (anthropic factors), while in the OL the area of natural grassland, distance to water bodies, and area of riparian forest were the most determinant factors (Figure 2). Carnivore species were also influenced by distinct variables in each landscape. For this group in the MML, occupancy rates were high in areas with more forest, less artificial pasture, and away from human settlements. These factors switched to area of natural grassland, distance to water and distance to human settlements, respectively, in the OL (Figure 2).
Co-occurrence with another species was also important to determine occupancy in some cases, reflecting inter-specific interactions (Table 1). Such interactions changed between landscapes, with the MML presenting fewer interactions (zero values) and all of them positive (occupancy of species B increases in the presence of species A). In the MML the agouti increased its occupancy in the presence of the puma and jaguar; the puma had a higher occupancy in the presence of paca and jaguar, and the jaguar had a higher occupancy in the presence of paca and puma. The paca itself was not affected significantly by the presence of other species. In the OL, more combinations of species had significant interactions, and some of them were negative. In this landscape, the paca had higher occupancy rates in the presence of agouti and puma, the agouti increased its occupancy in the presence of paca, but decreased it in the presence of puma (negative interaction), and the puma increased significantly in the presence of the paca but decreased in the presence of agouti (Table 1). In most cases, species interactions were mediated by another variable, which in the Middle Magdalena was distance to human settlements (except for Jaguar-paca interaction) (Figure 3). In the OL all interactions involving the agouti were mediated by the area of riparian forests and the puma-paca interactions by distance to water. Some of the relationships with the environmental variable in turn, changed in the presence and absence of the co-occurring species (Figure 3). The clearest example is forDasyprocta in the MML, where results indicate that its occupancy is higher in the presence of Puma concolor and Panthera onca . The relationship to the other important factor, distance to human settlements, is positive in those cells with presence of these predators, but it becomes negative (higher occupancy at smaller distances to settlements) in their absence (Figure 3).