Remote sensing of wildlife connectivity networks and priority locations for conservation in the Southern Agricultural Growth Corridor (SAGCOT) in Tanzania

Abstract

Land conversion is causing habitat loss and fragmentation worldwide, particularly in Africa, where the proliferation of agricultural development corridors may threaten vital areas for ecological connectivity and wildlife survival. To conserve connectivity, careful landscape planning is necessary, which strongly relies on remotely sensed land cover maps. Here, we present a remote sensing-based framework that efficiently identifies priority locations for connectivity conservation. We applied the framework in the Kilombero catchment and development cluster of the Southern Agricultural Growth Corridor (SAGCOT) in Tanzania, where new agricultural development projects could act as barriers for connectivity. Using satellite imagery from Sentinel-1 and 2, we mapped the mixture of mountain and lowland land covers and habitats with an overall accuracy of 75%. Then, we assessed ecological connectivity to predict African elephant corridors and prioritize them in two ways. First, we identified elephant corridors that contribute the most to current landscape connectivity, and second, we identified those whose restoration would significantly enhance landscape functionality and improve the current connectivity level. We mapped 214 potential elephant corridors, identified 43 of them as priority for conservation, and 43 as target for restoration. Our model predicted four already known corridor areas in and around the Kilombero valley floodplain, and other important corridors not yet identified by previous studies in the south of the basin. Priority elephant corridors inside the floodplain showed narrow widths and low permeability, indicating low functionality in the connectivity network. Nevertheless, the abundance of priority corridors for restoration suggests that connectivity could be recovered if the recommended measures are applied during SAGCOT planning and implementation process. Our findings demonstrate the possibilities of combining multispectral and radar data for guiding biodiversity management in development corridors and for assessing ecological connectivity worldwide.