PS 45-173 - Agricultural expansion and reserve connectivity in Tanzania

Wednesday, August 9, 2017
Exhibit Hall, Oregon Convention Center
Kate Tiedeman, Department of Environmental Science and Policy, University of California, Davis, Davis, CA and Robert J. Hijmans, Environmental Science and Policy, University of California, Davis, Davis, CA
Background/Question/Methods

Human land-use simultaneously shapes and is shaped by ecological patterns and processes (Burnsilver et al. 2003). Land use and land cover change in areas surrounding Protected Areas (PAs) may substantially influence park natural resources (Hansen and DeFries 2007). Human land use can interrupt landscape connectivity – the degree to which a landscape facilitates or impedes animal movement between resource patches (Taylor et al. 1993). The movement of organisms, both animals and plants, between habitats can influence community composition, population persistence, productivity, and reserve performance (Olds et al. 2012). In addition to connectivity, PA management type (national or local) impacts PA effectiveness in protecting species (Pailler et al. 2015), and may also influence land conversion. Community based management effectiveness is dependent on community membership and a mixture of authority and resources conferred to local governance (rather than central) (Hayes and Ostrom 2005, Sowman and Wynberg 2014).

It is therefore critical to know whether landscape permeability has changed over time around reserves, and whether community supported reserves experience different levels of agricultural encroachment/expansion outside their borders. To address this question, I computed a ‘landscape permeability index’ and applied this index to cropland surrounding PAs in Tanzania. I first computed the “naturalness” or the complement, the “human footprint” as a function of land cover types (agriculture specifically), population density, and presence of roads. I then used a Random Forest model (Rodriguez-Galiano et al. 2012, Ghosh et al. 2014) with the mean/median, maximum, minimum, and standard deviation of each band for each pixel as dependent variables, using 64,000 validation points (crop/non-crop) created through manual scoring of aerial photos.

Results/Conclusions

Preliminary results indicate that cropland has increased 10km around one particular reserve with a mixture of community and national governance, the Grumeti Reserve (IGGR). I used a moving window regression with time as the independent variable to assess the rate of change of agriculture over time, and found an overall trend of increased cropland, however the area exhibits considerable heterogeneity. From 2000 to 2015, cropland within a 10km buffer of the IGGR reserve increased by roughly 90%, where 3.3% of pixels in 2000 were classified as cropland, and in 2015 6.4% of pixels were classified as cropland. Future work will use classified cropland to create a permeability index from 1984-2015.

This research represents the first analysis of historical and current landscape permeability using satellite products for multiple years, to determine the impact of community supported reserve management.