The Map shows the overlay of the number of land cover changes with NDVI decrease and increase between 2001 and 2011 referring to NDVI trends. Three classes among the trends are built. Besides a “tolerance class” meaning NDVI trends between -0.005 and +0.005 the dataset was classified into “decreasing” (NDVI Trend <-0.005) and “increasing” (NDVI trend >0.005) vegetation trends. The overlay highlights the southern part of Kenya, especially the counties Narok and Kajiado where a stable land cover and decreasing trends overlap. Within this overlap are also Kitui and Isiolo – both counties that were also highlighted in the OLS-regression output as underpredicting –, parts of Marsabit and some small areas along the coastline. Also again the northwestern area, mainly Turkana Region but also West Pokot and Baringo are expressing increasing trends and seem to be linked to a more stable land cover.

MODIS provides the Land Cover Type Product MCD12Q1 (Friedl et al., 2002) with 500m grid resolution which represents the same pixel size also used for the MODIS NDVI time-series analysis. Annually data provision and a matching pixel size with the MODIS NDVI data used earlier in this study were key elements for choosing this dataset. The Map shows the number of LULC changes as calculated based on the methods described in chapter II.3.3. Stable areas – where land cover changes are zero – can be identified in southern Kenya, Kajiado County in particular, but also in western Kenya north of Lake Victoria, around Lake Turkana, and in the northeastern part of Kenya bordering Ethiopia. Around 33.16% of the total land area experience zero changes from 2001-2011 while 16.11% changed once and 22.92% show two changes. Three (13.98%), four (9.53%) and five (3.42%) changes can still be observed in Map III.11 while areas experiencing more than five changes are occurring in less than 1% of the total land area. The different classes show the number of land cover changes within the observation period.

This map is included in a global study on mapping marginality focusing on Sub-Saharan Africa and South Asia. The Dimensions of Marginality are based on different data sources representing different spheres of life. The dataset used for this approach (Marginality Hotspots) can also be found here: (link to datasett???). Five different dimenstion of marginality were defined and based on their thresholds overlayed to identify those areas where more than only 1 or 2 dimensions occur but several once which make these areas more marginal. With regard to the project MARGIP especially those people are at risk who are marginalized and poor and are thereby lacking possibilities due to missing access to capital and resources but also by being remote. The number of poor are the once we want to make visible. Therefore data by HarvestChoice on Poverty Mass representing the number of people living in poverty were overlayed with dimensions of marginality to give an impression on how many people are living in these spots and are thereby being poor and marginal. See also: http://www.zef.de/fileadmin/webfiles/downloads/zef_wp/wp88.pdf .

Economies are ranked on their ease of doing business, from 1–190. A high ease of doing business ranking means the regulatory environment is more conducive to the starting and operation of a local firm. The rankings are determined by sorting the aggregate distance to frontier scores on 10 topics, each consisting of several indicators, giving equal weight to each topic. The rankings for all economies are benchmarked to June 2017.