University of Edinburgh logo British Antarctic Survey logo National Oceanographic Centre logo University of Leeds logo

Projects

How fragmented are the world’s savannas?

Project background

Fragmentation is known to be widespread in the world’s tropical forests, with negative impacts on biodiversity and resilience. However, the situation in savannas is poorly known – there are no maps of how intact or fragmented savannas are and, as a result, it is impossible to target conservation work appropriately. It is thought that savanna biodiversity might be particularly sensitive to fragmentation, as many savanna animals need large home ranges1. It is also likely that fragmentation disrupts the provision of many ecosystem services that are critical to the livelihoods of 100s of millions of people who live in or near savannas2.

A fragmented savanna woodland in central Mozambique
Fig 1. The fragmented savanna woodlands of central Mozambique. Credit ref 1.

This project will address these major knowledge gap by quantifying the intactness, connectivity and fragmentation of the world’s savannas. This is both a conceptual and practical challenge: conceptually it is hard to define fragmentation in a naturally patchy and sparsely wooded landscape; practically it is difficult to distinguish open savannas from small scale agricultural landscapes using common remote sensing techniques.

This work builds on recent developments in mapping savanna structure3 made possible by a new generation of radar satellites and combines remote sensing with fieldwork at long term sites. You will be closely linked to a new NERC Large Grant in which a large team is addressing complementary research questions. The work will also be a collaboration with Yamaguchi University and the Japanese space agency (JAXA), to ensure access to the latest satellite radar data and products. You would be based at the University of Edinburgh as part of the LANDteam research group.

Research Questions

  1. How much of the savanna biome is fragmented, and to what extent?
  2. What opportunities are there for increasing connectivity in savannas, and where is most threatened by ongoing fragmentation?

Methodology

This project will work at a variety of scales, starting with a case study in Mozambique where the student can understand the reality of fragmentation in savanna landscapes and collect ground truth data for upscaling. This will make use of long term socio-ecological studies of three landscapes in Mozambique and will use a variety of methods including field ecology and drones to map fragmentation at landscape scale. At larger scales, you will use a suite of tools created by the University of Edinburgh LANDteam to map the structure of savannas using data from radar satellites (ALOS and Sentinel 1), with high resolution optical and field data as a reference. It is likely that this work will start with a regional focus (e.g. ref 3) and move to global mapping once the methods have been developed and evaluated.

Training

The student will gain state of the art skills much in demand across the environmental sector, including remote sensing, image analysis, spatial modelling, and ecological fieldwork. You will join a research group that uses R, Python and Google Earth Engine for most of its analysis, and will work closely with post docs and other PhD students conducing similar work. Full training will be provided in all the required methods.

Requirements

The successful student will need to have enthusiasm for learning new methods and be comfortable with quantitative analysis. As such this project would suit a student with a background in most natural sciences, and is also suitable for people with a physics, informatics or maths background, as long as you have an interest in ecosystem science. Fieldwork duration is flexible but is likely to involve 1-2 months away.

Further reading

[1] Tripathi (2017). PhD thesis. [2] Ryan et al (2016) Ecosystem services from southern African woodlands and their future under global change. Phil. Trans. Royal Society B, 371, 20150312; [3] McNicol, Ryan, Mitchard. (2018) Carbon losses from deforestation and widespread degradation offset by extensive growth in African woodlands. Nature Comm., 9, 3045.