Influence of land use and land cover changes on ecosystem services in the Bilate Alaba Sub-watershed , Southern Ethiopia

Human well-being was obsessed with the natural scheme that provides various functions vital to support management at various levels. Land use/ land cover (LULC) dynamics over 45 years within four intervals (1972, 1986, 2008, and 2017) to evaluate its influence on ecosystem services. Geographic information system (GIS) and global value of coefficients’ database together with LULC dynamics were used to determine ecosystem service values (ESV). The results showed that cultivated land and settlement land expanded by 67.38 and 532% respectively whereas forest land, shrub land and grassland declined by 66.35 and 18.36% respectively over the analysis period. A decline of total ESVs from US$ 35.23 million in 1972, to 33.61, 27.91 and 25.87 million in 1986, 2008 and 2017, respectively. Approximately US dollar of 9.37 million ESVs were lost owed to LULC changes from 1972 to 2017 in the sub-watershed. In terms of ES functions, erosion control, nutrient cycling, climate regulation and raw material provisions were the key bringers to loss of ESV. Global ESV data sets together with LULC change information helps to make a possible judgment about past environmental changes and reliable results achieved to make sound decisions. The decline of ESV was an indication of environmental degradation in the sub-watershed and needs future appropriate intervention policies in land conservation.


INTRODUCTION
Ecosystem services (ES) defined as situations through which natural ecosystem support and sustain human life, maintain a healthy environment and support production of goods such as fuels or fibers (Daily et al., 1997), offer services varied both quality and quantity (MEA, 2005).Goods and services derived from Ecosystem functions *Corresponding author.E-mail: godebo09@gmail.com.
Author(s) agree that this article remain permanently open access under the terms of the Creative Commons Attribution License 4.0 International License benefit man both directly and indirectly in the territorial environment (Costanza et al., 1997).Ecosystem services categorized in different ways based on functional groupings, production, and information services (MEA, 2005;De Groot et al., 2002;Lobo, 2001).Agricultural activities including cultivation in various watersheds had modified the existing landscapes; the conversion of the natural ecosystem to agriculture would have a strong impact on the watershed capacity to produce important ecosystem functions (Power, 2010).Land use changes focusing on cultivation and settlements were the major human activities that influence the ES (Kindu et al., 2016;Tolessa et al., 2017a).
The changes in ecosystem services were varied on the spatial and temporal distribution of land use/land cover (Bryan, 2013;Hu et al., 2008).Ecosystem service is defined as the ecology provisions any kinds which make the sustainable life of human being in the biosphere (Li et al., 2017).In terms of functionalities, ES is categorized into four major components; provisioning, regulating, cultural, and supporting services (MEA, 2005).An ecosystem service is correlated to changes in LULC in certain areas in the global world (Yirsaw et al., 2017).
LULC dynamics has direct effects on ecosystem services (Hu et al., 2008;Polasky et al., 2011).LULC change influenced the variation of ES components (an increase of some services on the contrary decreasing others) that would affect human beings needs, indicating ecological disturbances (Polasky et al., 2011).Land use altered some ecosystem services, affected social and government practices (Garcia-LIorente et al., 2015).
Land use changes (cultivation and settlements) were dominant in rural landscapes influencing ecosystem services in most parts of Ethiopia.LULC and ecosystem services valuation information would facilitate to identify the mainly exposed to alter in ecosystem services at the watershed scale.Most studies conducted to monitor LULC change in Ethiopia given little attention to address the influence on ecosystem services (Tolessa et al., 2018).Inside current science, an ecosystem service global database is commonly used for assessment of ESs together with the investigation of LULC changes intended for various biomes (Costanza et al., 2014).
In Ethiopia rural landscapes, LULC changes were a very common occurrence in which agriculture and settlements had been affecting ecosystem services.Furthermore, most studies in the country focused on LULC detection and its causes Ethiopia (Tsegaye et al., 2010;Meshesha et al., 2014).The influence of LULC changes on rural ecosystem services which are important in watershed scales are not recognized (Kindu et al., 2016).The objective of this study is to evaluate the influence of LULC changes occurred over the past four decades  on ecosystem service values in the Bilate Alaba sub-watershed of the Southern Ethiopian and to investigate changes of individual ecosystem service function.

Description of the Study area
Bilate Alaba sub-watershed located in Alaba woreda, Southern Ethiopia about 310 km south of Addis Ababa and about 85 km southwest of the Southern regional state capital of Hawassa.The sub-watershed has lied UTM coordinates of 387500 to 413750 m north latitude and 797000 to 824500 m east longitude (Figure 1).The sub-watershed comprises 45 rural kebeles and one town of Alaba wereda (kulito town).
The elevation ranges from 1613 to 2201 m above seas level, but the majority of the sub-watershed is found at about 1880 m above sea level, the sub-watershed coverage estimated is about 403 km 2 , it is proper for crop production and animal husbandry because of its major portion is flat with regard to its landscape.
The major soils of the subwatershed are Andosol, Chromic Luvisols, Phaeozem and Nitisol (FAO, 1998).The soils in the study area are potentially fertile if properly managed through various soil management practices and smallholders can get reasonable yield without application of inorganic fertilizers.The nature of the soil in the study area have been detached by both water and wind resulted from the development of huge gullies in the northern and eastern parts; in some north-eastern part it was totally removed and degraded by the effects of cattle and human interference of the natural ecosystem (IPMS, 2005).
Agro-ecologically, the sub-watershed is characterized as Subtropical zone (IPMS, 2005) having the mean precipitation of 1093 mm per year and the average annual temperature value of 21°C.It has received bimodal rainfall where the main rainfalls (kiremt) are from July to October whereas small rains (belg) are between March and April.Rainfalls in both seasons were erratic unevenly distributed resulted in crop failures in most parts of the sub-watershed.Maize, sorghum, wheat, pepper and haricot beans are the common rain-fed crops grown in the area.
All these crops are mana ged using traditional agricultural techniques and equipment.Moreover, a few types of vegetables and livestock feed like rodus grass and cowpea are grown with the help of small irrigation scheme (IPMS, 2005).The population of livestock exceeded the available feed sources in the sub-watershed has affected animal production in existing crop-livestock farming systems.

DATA SOURCES AND MATERIALS
Time series data for LULC changes created from Landsat images of four periods (1972, 1986, 2008 and 2017) acquired from the United State Geological Survey (USGS) source (Table 1).Topographic maps were used for verification of 1972 Landsat image since Google earth was not functioning on this time series.ERDAS Imagine 14.0 was used for image process techniques, and ArcGIS 10.1 software was implemented for the production land use land cover maps.Dry season images were selected in order to get clear images, not having clouds to facilitate the image classification without difficulty, along years the same cropping season.

Classification and processing of images
Evaluation of LULC changes was carried out using supervised classification specifically maximum likelihood approach of the Landsat images (Jensen, 2007).Images of a similar season were used to reduce the misclassification.For this study, five LULC types were recognized (Table 2).Field visits, as well as discussion with key informants, were conducted to encompass a clear judgment of the major classes of LULC.
In addition to image classification, a field visit was conceded to  collect data for Ground Control Points (GCPs).Classified LULC using image classifications were cross-checked with ground truth data with the support of the global positioning system (GPS) which is generated during field trips.To monitor a correctness of the categorization method, 660 GCPs were collected using GPS from the field and Google Earth.Overall LULC classification was based on the general framework presented in Figure 2 (Alemu et al., 2015).
The overall producer's accuracy of LULC map of the subwatershed was 92.9%, overall user's accuracy was 93.3% and  overall kappa statistics was 91.1%, met the requirement outlined by (Anderson et al., 1976).Hence, the data considered for auxiliary evaluation of values of ecosystem services for five LULC classes.
ArcGIS was used to analyze LULC data and ecosystem services valuation (ESV) for different biomes was computed by following the methods of (De Groot et al., 2012;Li et al., 2007;Hu et al., 2008).
The global databases were used for five LULC classes to estimate the values of ecosystem services (Costanza et al., 1997).The identified major LULC was not matched with existing biomes (Costanza et al., 1997), a replacement for each LULC classes was used for forest, shrub and grassland, settlements and cultivated lands (Table 3).
Although the value coefficient proposed by Costanza et al. (1997) was criticized because of uncertainties (Nelson et al., 2009).The identified major LULC was not perfectly matched with the existing biomes, proxy for each LULC category were used for cultivated, shrub, settlements, forest and bare lands (Table 3).Standard methods deployed by Li et al. (2007) and Hu et al. (2008) was used to estimate the total ESV in the sub-watershed for 1972, 1986, 2008 and 2017, mathematically expressed in Equation 1: Where ESV is the estimated Ecosystem service value, AK is the area (ha) and VCk the value coefficient (US $ ha -1 yr −1 ) for LULC class k.
To estimate the impacts of LULC changes ecosystem services, it was attempted to make manipulation of individual ecosystem services (TEEB, 2010;Costanza et al., 1997).Individual ecosystem services values were calculated using Equation 2 developed so far Hu et al. (2008 Where ESVf is the estimated ecosystem service value of function f, Ak is the area (ha) and VCfk the value coefficient of function f (US $ha −1 yr −1 ) for LULC category k is taken from (Costanza et al., 1997).With the consideration of uncertainty (Costanza et al., 1997), sensitivity analyses were conducted due to the fact that biomes used as a proxy for LULC classes were not exactly contests.The coefficient of sensitivity (CS) was determined the robustness of calculated ESV, which was calculated based on Equation 3 outlined by Li et al., (2007 andHu et al. (2008).The value of the cultivated land, shrub and grassland and forest land coefficients were adjusted by 50% in sensitivity analysis.
Where CS is Coefficient of Sensitivity, ESVi and ESVj are initial and adjusted total estimated ecosystem service values respectively, and VCik and VCjk = initial and adjusted value coefficients (US $ ha -1 yr −1 ) for LULC type 'k'.

LULC dynamics analysis
Generally, there were five LULC classes identified; cultivated land, forest land shrub and grassland, bare land and settlement (Figures 3 and 4).The settlement was also overspread followed the similar trend as cultivated land, and its area became highest in 2017 compared with the land cover in 1972 (Table 4 and Figure 3).On the contrary, the forest land reduced from 9.28% in 1972 to 9.48% in 1986 to 4.16% in 2008, auxiliary to 3.15% into 2017.Similarly shrub and grassland declines from 55.39% in 1972 to 50.51% in 1986 to 47.70% in 2008, further to 45.19% in 2017.Nevertheless, settlement and bare land showed incoherent styles of changes (Table 4, Figures 4, 5 and 6).Similar result obtained by (Zeleke and Hurni, 2001;Tolessa et al., 2017b;Meshesha et al., 2016;Gashaw et al., 2017;Alemu et al., 2015) forest land was shrinking, while settlement and agricultural land increased significantly whereas Bewket (2002), Fentahun and Gashaw (2014) found the opposite, in terms of magnitude for changes.Zeleke and Hurni (2001) reported an increase in cultivated lands by 38% in 38 years .Similarly, Belay (2002) reported an increase in croplands only by 5.5% within the interval of 1957-2000; an increase in of forest land at the expense of cropland documented elsewhere in Ethiopia (Amare et al., 2011;Asmamaw et al. 2011).
Land use policy in Ethiopia has been changed remarkably since 1972 because there was a change of regime from feudal to the Derg regime (Reid et al., 2000).During Hail Selassie regime, he encouraged commercialization and mechanization of agriculture.Firms were easily accessed tractors and fertilizers on loan basis (Kibret et al., 2016).However, the state was owned land in the Military regime and it was communal property (land reform) with the promotion of cooperatives in the villagization programme across the country resulted in depletion of natural resources and cultivation of land became increased, forests were cleared, lands highly degraded.When Ethiopian People Revolutionary Democratic Front (EPRDF) took power after the downfall of military regime was kept the same land policy which encouraged smallholders to put extra forest area to cultivation to produce high value-crops for possible markets and agro-processing plants (Dejene et al., 2013) favored mixed economy.Rural land utilization proclamation was effective in SNNPR since 2007 focused on citizens had got land certificates and given the use right based on the federal level government land use policy; there were no significant changes in land management's by farmers after the effect of the proclamation (SNNPR, 2007) farmers had strong fear lead to conflicts due to land registration (Zerga, 2016).
The coefficients of sensitivity (CS) of these investigations were smaller than one in all land uses.The value of the ESV coefficients of selected land uses 50% adjustment results shown in (Table 8), CS varied from a small of 0.02-0.06for cultivated land to a larger of 0.73-0.84used for shrub and grassland.CS for shrub and grassland was the highest since relatively larger coverage in addition to 2 nd highest service value.The results of all analyses indicated that the ESVs calculated for the sub-watershed was fairly inelastic in relation to changes ESV coefficients, which also suggests the estimation of the ES value is reliable since all the CS values are less than one.
LULC changes were influenced ecosystems services,  the ecosystem service value of the sub-watershed was reduced over the analysis years due to the decline of important components of the sub-watershed especially forest, shrub and grassland, which is similar to the findings of Eshetu and Högberg (2000), Li et al. (2007) and Yirsaw et al. (2017).With regard to individual ecosystem functions, the raw material, nutrient cycling, and cultural services were reduced in the sub-watershed, which agreed with the results of Tolessa et al. (2017a), Hu et al. (2008) and Kindu et al. (2016).The robustness test carried out in sensitivity analysis confirmed that ESV calculated was reliable, agreed with the findings of Li et al. (2007), Hu et al. (2008) and Tolessa et al. (2017b).
However, the global database may underestimate the current potential land use practices by smallholders (Costanza et al., 1997).
Even though there were significant improvements of ecosystem services in developed nations, there was a huge loss of certain ecosystem services in developing countries, for example, in Ethiopia, due to land use changes derived from natural to agricultural (Haines-Young et al., 2012).Monitoring and quantification of each watershed functions in rural areas could help to understand the benefits and minimize associated losses to the natural ecosystem (Nelson et al., 2009).To diminish the cost of ground data collection which is expensive, estimation of ESV using LULC change and established global database is an alternative way, moreover, getting real data about land use in rural areas is challenging.

CONCLUSION AND RECOMMENDATIONS
There is a substantial influence of LULC data on ecosystem services at the local and global level to indicate how many services lost through human cultivation in both space and time.The ecosystem services reduced due to deforestation and overgrazing might have an effect on the livelihoods of local communities and hence the need for improving land for sustainable production is vital.Regards the major LULC change observed in the last 45 years , cultivated land and settlement land expanded by 67.38 and 532% respectively whereas forest land, shrub land and grassland declined by 66.35 and 18.36% respectively over the analysis period.
ESV decreased by 26.6%, among the ecosystem functions identified biological control (0.05%) was the highest positive value as compared to other, while other remaining services had negative value indicated decreasing trend; nutrient cycling, provision of raw materials, climate regulation and erosion control were major contributors for the loss of ESV indicated that the study area needs proper soil fertility management together with soil conservation measures.The CS values selected land use types (cultivated, forest and shrub and grassland) were less than one implied the estimation is robust.CS for shrub and grassland use has the highest among all land uses.It was estimated about US$ 9.37 million loss of service in 45 years has revealed of ecological degradation.The agricultural development under current pattern should take into financial related losses occurred at ordinary settings whether lost or transformed; crop production systems should have appropriate land use plan incorporating to protect forest, shrub and grassland having bigger ESV.
Many tropical countries land use policies encourage the alteration of woods to crop production outcome of the loss of important ESs (Lira et al., 2012), moreover, the current investment policy put most natural forest/shrub areas converted to agro-processing industries affecting the natural setup of ecology and this also practiced in the sub-watershed.The need for appropriate intervention of rural land policies and active participation of smallholders for the long-term management of land assets (forest, shrub and grassland) is crucial to prevent the degradation of the ecological resources.
and grassland ecosystem services decreased while cultivated land use ecosystem service increased.It was observed there was a rising trend in forest land use ESV from 1972 to 1986, then a decline for 1986-2017; general trend showed a decreasing tendency.Considering ESV values across land use classes, Shrub and grassland use system had the highest value in all study years (1972-2017).Ecosystem services value was reduced from US$ 35.23 million in 1972 to US$ 25.87 million in 2017, with the net loss of US dollar of 9.37 million (Tables
*United States Geological Survey.

Table 2 .
Land Use and Land Cover description of the study area.Land covered mainly eucalyptus trees, indigenous tree and not found near river courses Shrub &grassland Land covered mixed up both small shrubs and traditional grasses.

Table 3 .
Land use classes, proxy and equivalent VC.

Table 5 .
ESV for each LULC class during the study period from 1972 to 2017 in Bilate Alaba Subwatershed.

Table 6 .
ESV for each LULC category and changes of 1972and 2017 in Bilate Alaba Subwatershed, Southern Ethiopia.

Table 7 .
Estimated individual ecosystem functions (ESVf in US $ million per year).