The objective of this study was to analyze the combined effects of climate and land use/cover changes on livestock feed resources and livestock species composition. The land use/cover data were generated from Landsat images of 1986, 1995, 2010 and 2018. The images were classified using a maximum likelihood classifier algorithm. The result from meteorological data and the land use/cover change were compared with household perception on livestock feed availability and livestock species composition. The result showed that the dominant land use/cover in the eco-region throughout the study period was bush/shrub cover followed by woody vegetation. This study revealed transition of land use/cover from grassland, woody and forest vegetation cover to bush/shrub and cropland in the study areas. The analysis of time series meteorological data showed significantly increasing patterns of temperature, and the highly variable nature of rainfall during 1986-2018. The pattern of livestock population throughout the analysis period exhibited a significantly increasing trend. The land use/cover, temperature and rainfall significantly affected livestock feed availability and quality. Over the last 32 years, potential grazing resources had been declined with a resultant increase in the proportion of bush/shrub feed resources available for livestock. The inter-annual variation of rainfall during the analysis period was 40.45%, which implies that the rangeland is in non-equilibrium dynamics. The rangelands carrying capacity was significantly decreased from 3.76 TLU/ha/year in 1986 to 1.74 TLU/ha/year in 2018. However, the stocking rate was increased from 1.8 TLU/ha/year in 1986 into 7.15 TLU/ha/year during 1986-2018. It is recommendable to choose camel and goat more likely than cattle and sheep raising with increasing temperature and decreasing pattern of rainfall that favour bush/shrub feed resources. Hence, available feed resources and the probability of choosing livestock species vary with eco-region and land use/cover that indicates the need for site-specific feed and rangeland management scheme.
Keyword: Land use/cover; Feed resource; Livestock; Climate; Rangeland
INTRODUCTION
Climate and land use/cover change have strong interconnection and the concurrent mutual influence on each other (Dale, 1997; Fan, Ma, Yang, Han, & Mahmood, 2015). According to Fahey, Doherty, Hibbard, Romanou, and Taylor (2017), land-use changes such as deforestation and soil cultivation changes the surface brightness. The positive radiative forcing (lower land surface albedo) can produce through the abandonment of pasture and forest cover. Research findings indicate that conversion of land use/cover to agriculture is a crucial driver of climate change. For instance, agricultural activities (crop with livestock) carried out in the field directly accounted for 13.5% of GHG emission. Whereas clearing forest for agriculture roughly accounted for an additional 17% of global GHG emission (Faurès et al., 2013). Furthermore, conversions of pastureland/ native vegetation to cropland in semi-arid environments decreased soil organic carbon stock by 30% within five years of cultivations (Lipper, McCarthy, Zilberman, Asfaw, & Branca, 2017). Conversion of tropical forest to agriculture accounts for more than 60% of soil carbon loss (Eshetu & Hailu, 2020).
Land use/cover changes are mainly caused by human activities including overgrazing, expansion of built-up, forest clearing and cultivation (Feddema et al., 2005; Garedew, Sandewall, Söderberg, & Campbell, 2009; Homewood et al., 2001; IPCC, 2019; Reid et al., 2000). According to Benin, Ehui, and Pender (2003) and IPCC (2019), changes in land use/cover patterns and climate variabilities are primarily changing livestock feed resources in terms of composition, availability and quality. Under the dynamics of land use/cover changes and climate variability, the livelihood of pastoral communities is under threat (Elias, 2008; Gebru, Desta, & Coppock, 2003; Müller, Linstädter, Frank, Bollig, & Wissel, 2007). Furthermore, Benin et al. (2003) and Aklilu, Gerard, Kindie, Lisanework, and Duncan (2014) revealed declining patterns of grazing feed resources due to the combined effects of land use/cover and climate change. Moreover, Wolde-Georgis, Aweke, and Hagos (2000), Hidosa and Guyo (2017) and Husein (2018) reported that the quality and availability of natural pastures are considerably affected by continuous land use/cover and climate change. The supply of quality feed with the required quantity determines livestock productivity (Tolera, Yami, & Alemu, 2012). In Ethiopia, livestock feed resources are mainly derived from foraging over large areas of grazing and browsing lands (Mengistu & Salami, 2007; Tolera et al., 2012). However, the contribution of free-ranging feed resources is subjected to the types and species of livestock reared (Rahman et al., 2008).
The current growing concern of scientists is mainly focused on investigating how and to what extent climate and land use/cover changes are affecting livestock feed resources bases in pastoral regions (Nelson, 2012; Oba, 2012). Forage availability, pastoral herd mobility and herd composition are primarily affected by the current dynamics of climate and land use/cover changes. Land use/cover and climate change significantly vary across the agro-climatic condition, topography/altitude, and types of main livelihood and production system (Deressa, 2007; Yesuph & Dagnew, 2019; Žurovec & Vedeld, 2019). According to Faurès et al. (2013), climate impact mitigation and adaptation potentials depend on the agro-climatic condition and land use/cover types. Therefore, possible interventions could be provided depending on intensity and types of land use/cover and specific to particular environments (Dale, 1997).
According to FAO (2015), livestock diversity, both species and genetic, is a source of resilience and facilitates adaptations in the face of environmental challenges. Species and genetic diversity are potentially required to cope up with the continual changes in climate, feed resources and newly emerging diseases. The adaptive capacity of livestock species to climate extremes and harsh environmental condition is likely to sustain the livelihoods of the producers. The livelihoods of the poor pastoral community depend on diverse livestock species, and hence the value of species diversity remains essential in the face of environmental challenges.
Land use/cover change studies have extensively conducted concerning drivers of the change and drought vulnerability (Aklilu et al., 2014; Meshesha, Tsunekawa, & Tsubo, 2012; Reid et al., 2000). However, few studies dealt with livestock feed availability with land use/cover change (Aklilu et al., 2014) and none have dealt with the availability of livestock feed resources with climate and land use/cover changes in the southeastern pastoral region of Ethiopia.
The interconnection of climate and land use/cover change with household perception on trends of livestock feed resources is less understood. The current trends of climate and land use/cover changes and its potential impacts on the availability of livestock feed resources, pastoral herd composition, grazing land, the viability of livestock production and livelihoods of pastoral communities have not yet investigated in a southeastern pastoral region of Ethiopia. Furthermore, integrating land use/cover and meteorological information with ground-based community perception could provide a complete picture of the combined effects of climate and land use/cover change on livestock feed availability and its driving factors on herd composition. Therefore, this study presents the combined effects of climate and land use/cover change on livestock feed resource availabilities and livestock species composition in east Guji zone of southeastern pastoral region of Ethiopia.
  1. MATERIAL AND METHODS
  2. Description of the study area
East Guji zone is one of the zones in Oromia regional state and is the richest in livestock production in southeastern Ethiopia. NegelleBorana is the capital town of the Zone, which is located 610 km to the southeast of Addis Ababa. Three pastoral and agro-pastoral districts with the similar agro-ecological condition, namely Goro-Dola, Liban and Gumi-Eldallo, were selected for this study. The selected districts lie between 40 38 55” and 50 33 7” N latitude and 390 9 25” and 39058 37” E longitude. The study locations cover about 742,644.14 ha. The districts are located at a lower altitude (1370 and 1560 m.a.s.l). Thus, the climatic condition of the study area is a mostly semi-arid condition with the bimodal rainy season. The average annual rainfall is about 526.75 mm, and the temperature ranges from 24 to 30 ℃ (Adi & Swoboda-Reinhold, 2003; Aklilu et al., 2014). The soil type of the rangelands in the study areas are mainly vertisols (Abate, 2016; Dalle, Maass, & Isselstein, 2006).
  1. Data collection
  2. Land use/cover change analysis methods
  3. Data sources and methods of data collection
Land use/cover change analysis was conducted using time-series satellite imageries downloaded from USGS web site (https:/glovis.usgs.gov/). Landsat images of 1986, 1995, 2010 and 2018 with 167 path and 056/057 rows were used to generate land use/cover. As the study is large in size, two adjacent Landsat images were downloaded for each period (Table 1).
Image analysis methods
The analysis was conducted in three different stages such as pre-processing, image classification and post-processing. In the pre-processing stage, the images were checked for their geometric and radiometric errors. As the images downloaded from USGS are already geo-referenced, there is no need to conduct geometric correction. With regard to radiometric errors, all the images were subjected to haze reduction and atmospheric correction. Once the images were checked for their geometric and radiometric errors, the adjacent images were mosaicked using ERDAS IMAGINE 14 software.
In the image classification stage, 24 sample-training areas were collected from each land use/cover type and used for training as well as accuracy assessment. Maximum likelihood classifier algorithm was used to classify the images. Accordingly, six different land use/cover classes such as agricultural land, bushland, forestland, woodland, grassland and settlement were generated.
In the third stage, the classified images were checked for accuracy using accuracy assessment technique. The training areas used for image classification were used to assess the accuracy of the classified images as well. Kappa coefficient was computed, and report about producer and user accuracy was generated. In addition, a change matrix was computed in ERDAS IMAGINE 14 software to determine the contribution of each land use/cover class.
Estimation of potential dry-matter production of different land-use type
Rangeland condition assessment was conducted at different years in the same study area, and the condition varies over time. Potential dry-matter supply of the rangeland during 1986, 1987-95 and 1996-2010 was estimated using Wroe (1988) based model. The model considers the range of precipitation and rangeland condition of the study area. Average annual precipitation of the study area falls within the range of 550-650 mm at all study period (1986, 1987-1995 and 1996-2010). Accordingly, excellent and fair rangeland condition was considered to determine the potential dry matter supply of grassland in 1986 and 1987-1995, respectively. At the same time, the average of good and excellent rangeland condition was used to estimate total pasture supply of grassland during 1986. The grassland forage supply of the study area for 2018 was estimated from the sample taken from two different enclosed Kalo (reserve grazing). About 37 sample plots were selected from two different sites the grass was cut at the ground during the beginning of May (5-10) by throwing 1 m2 quadrant to measure the total dry matter supply in the field. The clipped grasses were dried at room temperature over several days, and then dried sample weight was weighed using digital sensitive balance and recorded. Thus, the average value of dry-matter productivity from the whole plots was used to estimate the total dry matter supply of the grassland in the study area.
The total dry matter supply of bush or shrubland, woodland and dry forest were determined using a simplified simulation model of Timberlake and Reddy (1986). The model considers potential evapotranspiration (PET), annual rainfall and soil water retention capacity of the study area. Annual rainfall and temperature data were obtained from the national meteorological agency, Ethiopia. Potential evapotranspiration data was determined from the monthly average maximum and minimum temperature using DrinC software version 1.7 (91). About 233 mm of water retention capacity of the soil was used, since vertisols at a depth of 0-1.5 m can have 233 mm of water retention capacity in the lowland of Ethiopia (Virmani, Sahrawat, & Burford, 1982). About 75% of utilization potential was considered since the woodland, shrub and dry forest flora of tropical Africa contain about 75% of browse species that can be potentially utilized by goat and camel (Wickens, 1980).
\(Rp=\ \overset{\overline{}}{R}-t_{50+p}*\left(\mu\right)\)………………………………………… Equation 1
\({R^{\prime}}_{p}=\ \left(\frac{R_{p}}{\text{PET}}\right)*1800\)……………………………………………Equation 2
\(\text{AE}_{p}=\ {R^{\prime}}_{p}-\text{RO}_{p}\)………………………………………………Equation 3
\(\text{RO}_{p}=\ \left(\frac{15.2482}{K^{0.8}}\right)*\left(\frac{{R^{\prime}}_{p}}{100}\right)^{3}\)………………………………………….Equation 4