Evaluating connectivity for two mid-sized mammals across modified riparian corridors with wildlife crossing monitoring and species distribution modeling

Abstract

The movement of wildlife can be constrained by river development projects owing to the presence of artificial structures. This study analyzes the spatial riparian crossing patterns of two mammal species, the leopard cat (Felis bengalensis euptilura) and water deer (Hydropotes inermis). We examine their ability to traverse from habitat on one side of the river to the other, through riparian vegetation, embankments, and the river,. We used 34 months of monitoring on 104 stream segments with varying levels of human modification on the Seom river, South Korea, to analyze their riparian crossing preferences across the river between suitable habitats on either side of the steam. Environmental variables were used for two scales of analysis: a watershed scale and a stream segment scale that examined both riparian habitat suitability and riparian crossings. At the watershed scale, habitat suitability was determined using species distribution modeling. The resulting maps were inputted with other predictor variables to logistic regression models to predict areas of high probability for the species' occurences. For the river segment analyses, we determined species’ presences and potential crossings from snow tracking, heat sensor cameras, and scat. We compared the predicted crossing locations from the watershed scale model to observed crossings. The watershed models indicate that leopard cats prefer upland forest, while water deer prefer the forest edge and riparian corridor. For both target species, the best riparian habitats were characterized by the presence of vegetation cover on the embankment and by at least one side of an embankment being adjacent to farmland or forest cover. The two species showed different riparian crossing preferences. Water deer crossed through large culverts or under bridges, whereas leopard cats utilized drainage pipes and smaller culvert boxes. Stream reaches located close to a river tributary had the highest connectivity values, and areas modeled as good habitats were linked near riparian and watershed areas with high density for both species as well as culvert boxes with a high openness ratio. Artifacts such as steep banks, concrete embankments, and adjacent roads were found to degrade the riparian connectivity of wildlife. These outcomes can be used to identify suitable riparian wildlife habitats of mammals, to evaluate lateral connectivity of riparian corridors for mammals, and to develop criteria for river conservation and restoration in urbanizing areas and developing countries. Keywords: Modified landscapes, Species distribution models, Maximum entropy, Logistic regression