Study area
The study was conducted at Teltele Woreda in the Borana zone of Southern Ethiopia (Fig. 8). The site was selected because it is one of the most arid parts of Borana zone and, therefore, the pastoral communities of this region are the most vulnerable to the rangeland degradation as a result of both human and climatic factors. It is located 666 km south of Addis Ababa, the capital city of Ethiopia. It lies approximately between 04° 56' 23''N latitude and 37° 41' 51''E longitude and the altitude are about 496-1500 m, the maximum altitude of 2059 m above sea level. The annual mean temperatures vary from 28-33oC with little seasonal variation. The rainfall in the region is characterized as bi-modal. That is to say that 60% of rainfall occurs from March to May and 27% of rainfall occurs from September to November with high temporal and spatial fluctuations [50] (Fig. 9). The potential evapotranspiration is 700-3000 mm [51]. The soil in the study area includes, 53% red sandy loam soil, 30% black clay, and volcanic light-colored silt clay and 17% silt and the vegetation mainly dominated by encroaching woody species, and those that frequently thinned out, include Senegalia mellifera, Vachellia reficiens and Vachellia oerfota [40,52]. According to the latest census conducted in 2015, the national census reported a total 70,501 of population for this woreda, of whom 36,246 men and 34,255 women; 4,874 or 6.91% of its population are urban dwellers. Cattle, goats, sheep, camel, mule, donkey and horse are the main livestock species reared.
Data sources and methods
This study combined multispectral satellite remote sensing data, in-depth fieldwork surveys and rangeland use policy analysis linked with rangeland vegetation change source. The Teltele rangeland shape file along with weather data (rainfall and temperature) from 1992-2019 were obtained from [53] to see the long-term trend in the study site. To monitor the spatial and temporal conditions of rangeland vegetation, we used the annual average of third Generation Standard Difference Vegetation Index (NDVI3g) data (1992-2019). The model we used to extract data from the study area and remove the biased from our data in order to adopt land use land cover (LULC) analysis is summarized in (Fig. 10). The data derived from the Global Inventory Modeling and Mapping Studies (GIMMS) with 8km grid resolution. Before extracting the data to our study area, we resampled them to 300 m resolution of digital elevation model of Ethiopia in order to increase the resolution of the data. For NDVI grid cell values we simply took the maximum, minimum and an average annual mean value in order to reduce disturbance in the trends, such as those attributable to bare soil and sparsely vegetated areas [54,55]. Vegetation maps of the Teltele district in 1992-2019 were obtained from the remote sensing data with spatial scale 1:100,000. The Landsat TM imageries acquired in 1992, 1995, 2000, 2005, 2010, 2015 and 2019 were used for range land vegetation cover classification and the characteristic of Landsat used for LULC change analysis was described at (Table 4). These years were chosen because of the availability of data, the quality of the images, and in order to compare the changes with in equal time intervals. Further, interviews and focal group discussions were conducted with the local pastoral community and stakeholders to verify the accuracy of the rangeland vegetation classified images analyzed by using ArcMap 10.3.1 software and furthermore, understand the possible major drivers and consequences of LULC changes in the rangeland. A total of 120 individuals (90 males and 30 females), 6 of them were stakeholders from different government sectors (4 males and 2 females) who have been lived 15 to 20 years in the study district, were selected, interviewed and discussed about the rangeland vegetation cover change and forage biomass production trend and as well as the major causes of change based on their observation and experience in the region. The priority driving factors for the changing of rangeland vegetation feature and biomass production were elaborated during the group discussions.
Table 4 Characteristic of Landsat used for LULC change analysis
Data
|
Year of acquisition
|
Bands/color
|
Resolution (m)
|
Spectral resolution/bands
|
Landsat Thematic Mapper (TM)
|
1992, 1995, 2000, 2005, 2010, 2015, 2019
|
Multi-
spectral
|
300
|
Band 1-5: 0.45-1.75
Band 6: 10.4-12.5
Band 7: 2.08 – 2.35
|
The data pre-processing, clipping the area of interest (AOI) and applying color composites with different reflectance grids, were used to improve visualization and interpretation [56]. The general techniques we used LULC analysis was described in the form of chart below at (Fig. 11).
Classification of vegetation cover change
In order to clearly understand the change of rangeland vegetation cover, a post classification comparison detection technique was used by classification and detection of each pixel using the remote sensing map and compute the coverage of the area change [57]. The classes were classified based on the Intergovernmental Panel on Climate Change (IPCC) Classes considered for the change detection and Land Cover Classification System (LCCS) Legend used in the Climate Change Initiative Land Cover (CCI-LC) maps for the Images obtained from different years (Table 5).
Table 5 Rangeland vegetation change classes and its definition in the study area
No.
|
Class
|
Definition
|
1.
|
Grassland
|
Land cover dominated by grass and herbs
|
2.
|
Agricultural land
|
Land area covered with crop fields with rural settlements
|
3.
|
Forestland
|
Land covered with higher indigenous plants
|
4.
|
Shrubland
|
Bush or shrub-dominated land with small range of grass
|
5.
|
Bare land
|
Area neither covered by vegetation nor crops
|
6.
|
Wetland
|
Areas seasonally or permanently waterlogged
|
The most widely used method for change detection is the comparative analysis of the spectral classification over a time series and filtered to reduce the poorly classified pixels [58,59]. Each classified image was compared for the detection of vegetation cover change and the summaries of the areas and percentages of change were calculated.
Forage biomass production dynamics
In order to quantify the forage biomass dynamics in different land cover classes, above ground biomass measurement was conducted. A 5 km transect was lied and systematically placed six 25X 25 m2 sampling plot at 500 m interval along a transect at each land class site. (in total 36 plots from the six land classes). In addition, within each plot three (3) 5x5m2 sub plots (in total 108) were placed. Finally, five (5) 1x1m2 quadrants was placed by randomly throwing them backwards in order to minimize any bias resulting from selective placement with in each sub plot for grass species samples collection. Then, all the above ground forage samples were cut by using cutter and collected in paper bag. The fresh weight of forage sample was measured in the field with a scale and taken to Yaballo Pastoral and Dryland Agriculture Research Center soil laboratory and oven dried for 24 hours at 105oC to determine the dry biomass. Then, the dry matter was measured after 24 h drying and converted into kilograms per hectare (kg/ ha). Data collection on grass species sampling was done twice per year (during dry and wet season).
Determining the linkage between forage biomass and NDVI value
In order to determine the linkage between the forage biomass and the NDVI value, the average NDVI values were derived from plot-specific extractions. The extracted plot-specific NDVI values were matched with the plot-specific forage biomass quantity for each land cover type monitored [60].
Socio-Demographic Profile of the Respondents
The Social-demographic status (age, sex, education level and income source) of the respondents was analyzed using Microsoft excel and descriptive statistics in the Statistical Package for Social Sciences (SPSS). The spatial and temporal trends in increasing number of agro-pastoralists, the drivers of rangeland vegetation cover and forage production change, the infestation rate of shrub plant species, the expansion agricultural lands, and rangeland indigenous management methods were analyzed using descriptive statistics.