While the conservation role of remaining natural habitats in anthropogenic landscapes is clear, the degree to which agricultural matrices impose limitations to animal use is not well understood, but vital to assess species' resilience to land use change. Using an occupancy framework, we evaluated how oil palm plantations affect the occurrence and habitat use of terrestrial mammals in the Colombian Llanos. Further, we evaluated the effect of undergrowth vegetation and proximity to forest on habitat use within plantations. Most species exhibited restricted distributions across the study area, especially in oil palm plantations. Habitat type strongly influenced habitat use of four of the 12 more widely distributed species with oil palm negatively affecting species such as capybara and naked-tailed armadillo. The remaining species showed no apparent effect of habitat type, but oil palm and forest use probabilities varied among species. Overall, generalist mesocarnivores, white-tailed deer, and giant anteater were more likely to use oil palm while the remaining species, including ocelot and lesser anteater, showed preferences for forest. Distance to nearest forest had mixed effects on species habitat use, while understory vegetation facilitated the presence of species using oil palm. Our findings suggest that allowing undergrowth vegetation inside plantations and maintaining nearby riparian corridors would increase the likelihood of terrestrial mammals' occurrence within oil palm landscapes.
Mitochondrial genomes provided the first widely used sequences that were sufficiently informative to resolve relationships among animals across a wide taxonomic domain, from within species to between phyla. However, mitogenome studies supported several anomalous relationships and fell partly out of favour as sequencing multiple, independent nuclear loci proved to be highly effective. A tendency to blame mitochondrial DNA (mtDNA) has overshadowed efforts to understand and ameliorate underlying model misspecification. Here we find that influential assessments of the infidelity of mitogenome phylogenies have often been overstated, but nevertheless, substitution saturation and compositional non-stationarity substantially mislead reconstruction. We show that RY coding the mtDNA, excluding protein-coding 3rd codon sites, partitioning models based on amino acid hydrophobicity and enhanced taxon sampling improve the accuracy of mitogenomic phylogeny reconstruction for placental mammals, almost to the level of multi-gene nuclear datasets. Indeed, combined analysis of mtDNA with 3-fold longer nuclear sequence data either maintained or improved upon the nuclear support for all generally accepted clades, even those that mtDNA alone did not favour, thus indicating "hidden support". Confident mtDNA phylogeny reconstruction is especially important for understanding the evolutionary dynamics of mitochondria themselves, and for merging extinct taxa into the tree of life, with ancient DNA often only accessible as mtDNA. Our ancient mtDNA analyses lend confidence to the relationships of three extinct megafaunal taxa: glyptodonts are nested within armadillos, the South American ungulate, Macrauchenia is sister to horses and rhinoceroses, and sabre-toothed and scimitar cats are the monophyletic sister-group of modern cats.