Understanding the drivers of habitat distribution patterns and assessing habitat connectivity are crucial for conservation. Habitat fragmentation is considered a major cause of biodiversity loss. It results in interruption of network between habitat patches and subsequent difficulty for gene flow between populations. Maintenance of landscape connectivity can promote populations move between habitat patches, which is essential to biological dispersal and gene flow in mosaic landscapes. In Yunnan, agriculture encroachment upon high altitude forest is one of the major threats to black and white snub-nosed monkey (Rhinopithecus bieti) habitats. In this study, we used maximum entropy distribution modeling (MaxEnt) to identify the climatic and topographic factors driving the distribution of the habitat type of the snub-nosed monkey. We also analyzed potential habitat connectivity between the groups reported in Yunnan, using genetic, least-cost path and Euclidean distance. We show that genetic distance is better explained by human disturbance and land-cover least-cost paths than by Euclidian distance. Furthermore, we used graph theory and least-cost distance to model the ecological network of the Yunnan snub-nosed monkey and to evaluate the connectivity improvement provided by potential measures of reforestation of agricultural lands. This approach was also used to assess the impact of highway development (including tunnels deviating traffic from the mountain roads) on the network connectivity. We show that potential habitat area will be reduced by 15% in total but increases mainly in the north of the distribution range at high altitude (due to tunnels) and decreases at lower altitude (due to habitat fragmentation). Other scenarios were also explored: on the landscape scale, the scenario consisting to reforest with optimal habitat for monkeys provides the biggest increase in connectivity, of course. On patch scale, a patch addition algorithm was applied on 1482 cropland patches to identify the 10 that most increase connectivity at different threshold of (hypothetical) dispersal distances. Graph methods can help modeling connectivity on different scales for species whose habitats are fragmented and thus help planning habitat restoration and conservation measures.