Socioeconomic Drivers of Land Use and Land Cover Change in Western Ethiopia

,


Introduction
Te relationship between humans and nature has been explained and conceptualized in many ways throughout human history, and signifcant heterogeneity still exists across cultures [1][2][3][4].Te interaction between humans and nature has recently led to LULC changes worldwide [5,6].Te impacts of land use/land cover change are determined by human activities [5].LULC change is a dynamic and complex process that can be caused by many interacting processes, ranging from various natural to socioeconomic factors [7][8][9].
Drivers that accelerate LULC changes can be social, economic, or environmental and can have positive or negative impacts on the planetary system [10][11][12].LUCC has been afected by a variety of important human endowments and biophysical phenomena [13,14].Te relationship between LULC change and its drivers is complex and dynamic, as some of the previous studies suggest that demographic changes contribute more than any other causal factor to land use/land cover change [15][16][17].LULC changes have led to global loss of native biodiversity and altered ecological processes and services in diferent ecosystems [18][19][20].
In developing countries, such as Africa, most of the populations are engaged in agriculture (both commercial and subsistence farming) and charcoal production as a source of income [21,22].In Ethiopia, the expansion of agriculture into forestland, logging, charcoal production, and fuel wood harvesting were the main drivers of LULC change [23,24].Drivers for land use/land cover dynamics are multiple and complex in space and time, requiring more investigation in Ethiopia [25,26,27].Agricultural land expansion, increased demand for fuel wood and building materials, illegal forest settlement, and illegal logging are said to be the main drivers of LULC change in Ethiopia [28,29].
In western Ethiopia, Wayu-Tuka district, where this study was conducted, human-induced changes in land use/ land cover are often observed [30,31].
A recent study by Negassa et al. [23] near the current study area showed that agricultural land expansion is one of the main drivers of changes in LULC changes in the district.However, the LULC change detection proposed by Negassa et al. [23] focuses on forest LULC change.Terefore, this study analyzes the socioeconomic drivers of land use and land cover dynamic changes in the study area.Understanding these issues requires a rigorous investigation of ongoing LULC changes in research.Research fndings contribute to the development of sound policy and management options for sustainable use and management of natural resources in the study area [30][31][32].Terefore, this study aimed to analyze the socioeconomic drivers of land use and land cover change in Wayu-Tuka district, East Wollega Zone, western Ethiopia.

Description of the Study Area.
Te study area, Wayu-Tuka district, is located in the administrative zone of Oromia National Regional State of East Wollega, approximately 322 km west of Addis Ababa.Te district is bordered by Sibu Sire district in the North and East, Leka Dulacha district in the South, and Guto Gida in the West.Specifcally, the feld of study is located between 8 °51′ 30″-9 °10′ 30″ north latitude and 36 °32′ 0″-36 °50′ 0″ east longitude (Figure 1).Te area has peaks and slopes such as Komto, Gara-achani, and Tuka with elevations of 3350, 3140, and 2350 m.a.s.l, respectively.Te area has a total area of 54,590.4hectares and includes 12 Kebeles (i.e., Kebeles-the smallest administrative unit in Ethiopia), 10 villages, and 2 urban centers.
Te district is divided into three agroecological divisions, of which 11% are upland, 49% are midland, and 40% are lowland [33].Te average annual rainfall in this area ranges from 1000 to 2551.4 mm, while the annual rainfall in the study area is about 2158 mm.Te recorded climate data show that the annual rainfall in the study area in 1993 and 1998 was about 2525.3 mm and 2551.4 mm, respectively (Figure 2).Recorded climate data show that the average annual minimum and maximum temperatures in this area are about 12.5 °c and 25.5 °c [34].
Te main vegetation types observed in the study area are Afromontane upland forest, interupland semideciduous forest, several forms of riparian forest, and plantation types.Southwestern Ethiopia hosts this moist evergreen montane forest, which is the native vegetation [35].Te soils of the study area are deep and belong to the orders Oxisols and Ultisols; of these, the main soil types and their spatial coverage in the district are 17,371.68ha (60%) of clay loom soil, 10,133.49ha (35%) of sandy soil, and 1,447.64ha (5%) of clay soil (Figure 3), which are suitable for agriculture such as cereal cultivation: maize, sorghum, and tef production in the district [36].
Subsistence agriculture is the principal economic activity in the study area, crop cultivation and livestock rearing.According to the 2015 demographic and housing census, the population of Wayu-Tuka district was estimated to be 66,194 [37].Of the 66,194 people, 32,391 were male and 33,803 were female (Figure 4).

Sources and Data
Processing.Te data used for Wayu-Tuka District LULC changes come from satellite images acquired in 1990 MSS, 2000 TM, 2010 ETM+, and 2020 OLI in four periods, downloaded free of charge from USGS: https://earthexplore.usgs.gov.LULC change classes such as bare land, settlements, agricultural land, forests, and information on changes in water bodies were obtained from Landsat images.
Landsat satellite images were obtained and processed of (path 170, raw 54) which contains an operational land imager (OLI), thematic mapper (TM), enhanced thematic mapper plus (ETM+), and multispectral scanner (MSS) used for LULC classifcation in the study area.Table 1 summarizes satellite data type, acquisition date, resolution, path and rows, and number of bands for the Landsat images used in this study (Table 1).
(1) Data Processing.In this study, the preprocessing of the images was georeferenced (UTM, WGS84) from the data downloaded.Tis image preprocessed corrects the image distortions and increases the quality of the image's data.Formerly, the Landsat images of each study year were independently classifed with a supervised classifcation technique and maximum likelihood classifer (MLC) applied to classify the LULC types in arc GIS used for supervised classifcation.Lastly, the postclassifcation was employed using separately classifed Landsat images made accuracy assessment pixel selection checked by using Google Earth for the LULC maps of 1990-2000, 2000-2010, and 2010-2020.[38].For these images, user accuracy, producer accuracy, overall accuracy, and the kappa coefcient were evaluated for each image in the LULC class in four periods 1990, 2000, 2010, and 2020 (Table 2).

Methods of Data
Producer accuracy and user accuracy were calculated to verify the accuracy of land use/land cover change classifcation types.
As shown in Table 3, the estimated accuracy values showed diferences between the fve land use and use classes within the same time period, and the accuracy values of each land use and use class also showed diferences from period to period in the three decades considered in the study (1990-2020).For example, for the Wayu-Tuka district in 1990 LULC classes, producer accuracy was 72.7%, user accuracy was 100% for farmland, user accuracy was 67.7%, and producer accuracy was 66.7% for forest, and UA (67.7%) of settlements and the AP of agricultural land (72.7%) were considered somewhat low.However, the levels of settlement of UA levels (66.6%) and (75%) of water bodies in 2000 were also relatively low.Te UA settlement level (62.5%) in 2010 was considered the lowest of all values for the fve LULC classifcations across all four study periods.Te maximum level of accuracy for all fve land use categories was 100% (Table 2).
Te overall accuracy levels for the LULC category for all four periods were 83.3% in 1990, 86.6% in 2000, 80% in 2010, and 83.3% in 2020 (Table 2) and above, which met the minimum requirements for Landsat imagery accuracy based on the classifed land use and land cover classes of the study area [39,40].

Sources and Data Collection Method.
Te data necessary for the study was generated through primary and secondary sources.Te primary sources of data include household surveys, feld observations, group discussions, and key informant interviews.Secondary data mainly come from relevant literature of published and unpublished materials, i.e., reviews using published and unpublished relevant literature.
Socioeconomic data were collected from communities through questionnaires, key informant interviews, focus group discussions, and feld observations.Survey tools were rational and used to collect information on household characteristics and drivers of LULC change.To meet the requirements of the main drivers of land use and land cover change, socioeconomic data should be integrated [41,42].Tis study also employs a combination of techniques to analyze socioeconomic variables that drive land use and land cover change.

Household Survey Data.
For the socioeconomic survey, sample households were drawn from the total population of villages (1340 HHs).Tree villages, Dalo Komto, Gara Hudha, and Kich, were selected based on their distance from the forest area (geographic location), and a questionnaire survey was conducted among the households.
Te formula used to determine sample size is [43] using the sample size formula.Finally, the sample size was calculated using the sample size determination correction formula, which is commonly used for sample size determination in most social science research when the target population is less than 10,000.Te sample size formula [44] is as follows: (1.96) 2 (0.5)(0.5) (0.5) 2 � 0.38416x100 ∼ 384. ( Terefore, for a population of 1,340, the required sample size is 384.However, since this sample size exceeds 5% of the population (520 * 0.05 � 26) [43], a formula for correction should be used to calculate the fnal sample size.Tese calculations are as follows: l � (no/(1 + (no/population))) n1 � 384/(1 + (384/ 520)) � 221.Terefore, the fnal sample size is 221.
Where no � sample size required when the population is greater than 10,000, n1 � limited population correction coefcient when the population is less than 10,000, t � normal standard deviation (95% confdence level is 1.96),Where t � the value of the chosen alpha level of 0.025 for each tail � 1.96 (an alpha level of 0.05 represents the level of risk the researcher is willing to take that the true margin of error may exceed the margin of acceptable error), n � sample size, where (p) (q) � variance estimate � 0.25.

International Journal of Forestry Research
According to the sample size determination formula above, 221 houses were randomly selected from 3 villages, 79 houses in Dalo Komto, 63 houses in Gara Hudha, and 79 houses in Kich.
Te district was chosen purposively because of the presence of forested areas in the area.Likewise, the sample/study villages were selected using a purposive sampling method based on the presence of nearby forests.Terefore, out of the 10 district villages, three villages were selected namely Dalo Komto, Gara Hudha, and Kich.Ten, in twelve questions, six were closedended that required a "YES" or "NO" answer (Tables 4-7), four were multiple-choice questions, and the remaining two were Sources: own summary of Landsat image results analysis via Arc GIS 10.3 (2023).Te diagonal bold values show that the areas correctly classifed for each year (1990-2020), the bold values of the last two rows for each year (1990-2020) represent total producer and producer accuracy (%), where as the last two columns for each year bold number represent total user and user's accuracy (%), and fnally, the last two rows bold values indicate that to focuses overall accuracy and overall kappa statistics.In addition, Pearson's chi-square test and logistic regression analysis were performed.Descriptive statistics using simple frequency analysis were used to describe the socioeconomic characteristics of households and to examine their responses and the classifcation of land use/ land cover change dynamics [45].Te drivers of LULC dynamics were analyzed using nonparametric tests, specifcally the Pearson chi-square test [46].In addition, quantitative binary logistic regression analysis was employed at the household level to identify the main socioeconomic determinants of observed land use changes and the drivers of land use dynamics [47].Logistic regression analysis is a statistical procedure suitable for examining the relationship between a variable (dependent) and several socioeconomic variables (independent) [48].
Tis is an efcient analysis technique when the dependent variable is binary [4] that estimates the LULC driving force of the independent (explanatory) variable on the dependent (response) variable: where Y � dependent variable representing the probability of Y � 1, α � intercept, β 1 . ..β n � coefcients of related independent variables, and X 1 . ..X n � independent variables.A logistic regression model was used to analyze the relationship between demographic data and socioeconomic conditions [49].In this study, the outcome variable (forest dependence) was taken from selected explanatory variables: age, sex, education, and property size were used as proxies for socioeconomic variables.
Figure 5 shows the framework and methodology fowchart used to create the LULC and socioeconomic drivers for LULC change (Figure 5).
Te year 1990 was the decade when the farmland in the study area accounted for approximately 36.6% (147.0 km 2 ) of the total area of the district, of which bare land, settlements, forest coverage, and water bodies each accounted for 28.1% (113.678km 2 ), 12.1% (49.0 km 2 ), 22.3% (90.0 km 2 ), and 0.6% (2.0 km 2 ) about more than two-third 66.1% of the area of Wayu-Tuka district in the same period in 1990.Te potion of water body (0.6%) is insignifcant in 1990 indicating that within the total area of the study area (Table 3).
In 2000, the district's farm land accounted for the largest proportion of the total area at approximately 53.4% (216.0 km2), while the rest bare land 10.7% (43.0 km2), forest cover 8.2% (33 .09km2), settlement for 25.7% (104.0 km2) and 2% of the water body (8.0 km2), respectively, accounted in 2000 and indicating a decline since the early 1990s.In fact, 2000 was also a period when bare land and forest coverage were 10.7% (43.0 km 2 ) and 8.2% (33.0 km 2 ), respectively, indicating a signifcant decrease compared with the area of 1990.On the other hand, the area percentage for farmland 53.4% (216.0 km 2 ) and water body 2% (8.0 km 2 ) of the study area increased in the year 2000 (Table 3).
Except the settlement of the district, the rest of LULC classes decreased compared with the previous period (2000).In 2010, the area of bare land 9.7% (39.0 km 2 ), farmland 51.7% (209.0 km 2 ), forest cover (16.0 km 2 ), and water body 0.3% (1.0 km 2 ) included in this research area.
Finally, in 2020, the Landsat images depending on the analysis showed that farmland with about 56.2% (227.0 km 2 ) was the LULC type which accounted the largest area share of the study site, and this accounted the largest from the periods of 1990, 2000, and 2010.Te settlement land was 36.0%(148.0 km 2 ) in the last period accounted for the district.Water body, forest cover, and bare land constituted about 3.7% (15.0 km 2 ), 2.6% (10.0 km 2 ), and 1% (3.0 km 2 ) in the study area, respectively (Table 3).
Figure 6 shows the LULC classifcation of four maps for fve (5) classifcation types in 1990-2020 from left to right.Te frst map shows the LULC classifcation of the TM in the 1990 image.Among them, most areas include 36.0%farmland, 28.1% bare land, 22.3% settlement areas, 12.1% forest coverage, and 0.6% water bodies.For the TM 2000 image, the LULC classifcation results show that most of the various types of land use/land cover fow to farmland and settlement areas, accounted for 53.4% and 25.7% of the  Water bodies cover the smallest area, accounted for 0.3% of the total area of the district.Finally, the LULC results for the 2020 OLI images were also similar, showing that the majority of LULC for all categories of drivers fowed toward farmland and settlement areas, covering 56.0% and 36.0% of the total area of the study area, respectively.Te water body and forest land coverage rates are 3.7% and 2.6%, respectively.During this period, the area of bare land was the smallest, accounted for 1% of the total area of all types in the district (Figure 6).From 1990 to 2000, the change results of LULC area of bare land and forest coverage showed positive changes, while the changes of cultivated land, settlement areas, and water body areas showed negative changes of km 2 .However, the LULC category type of area change subsidence showed negative changes from 2000 to 2010, indicating that settlement increased through area change.
Figure 7 also indicated that the area change results (2010-2020 km 2 ) of bare land, farm land, forest land, settlement areas and water bodies are 35.537km 2 , −17.9951, 5.457 km 2 , −9.2042 km 2 , and −13.788 km 2 , respectively.Finally, the area changes during 1990-2020 show that the largest area change during the same period was farmland

Driving Forces of Land Use/Land Cover Dynamics.
For the frst analysis of the relationship between dependent variables (land use and drivers of land use) and socioeconomic variables (age, sex, educational status, and size of land International Journal of Forestry Research ownership), correlation analysis was applied before modeling continued (Table 8).
Major socioeconomic variables (i.e., age, gender, educational status, and land-holding size) were analyzed to determine the relationship with LULC changes or not.According to Chen et al. [50], before conducting modeling analysis, Pearson correlation analysis can be used to explain the correlation between the dependent variable and the independent socioeconomic variables.Similarly, the Pearson correlation coefcient indicates positive and negative correlations.A positive correlation indicates that the dependent variable (in this case, the main socioeconomic variable that LULC dynamics) increases with the value of the independent socioeconomic variable, while a negative correlation indicates that the dependent variable increases with the value of the independent socioeconomic variable decreases as the value increases the values of socioeconomic independent variables that are the main drivers of LULC dynamics.Te independent variables (i.e., age and gender) have positive Pearson correlations with the drivers of LULC dynamics; this means that when these independent variables increase, the drivers of land use and land cover dynamics also increase.However, when the independent variable of education status is negatively correlated with the status of land holding and the number of educated population increases, the driving force of LULC cover dynamics, especially anthropogenic driving force, decreases, while the status of land holding increases, the driving forces of land use and land use dynamics decline.Te fgurative number one (1) correlation shows that the value of one variable can be accurately determined by knowing the value of the other variable, which is a perfect correlation (Table 8).

Analyzing Dependent Model.
Correlations with drivers of land use/land cover dynamics become important after identifying the main socioeconomic determinants using Pearson's chi-square.Te relationship between the (dependent) variable and various socioeconomic (independent) variables was examined according to [49] binary logistic regression model, which estimates the LULC of the independent (explanatory) variable on the dependent (response) variable driving force.

Independent Classifcation Encoding and Baseline
Used in the Model.Te independent categorical variable coded for education status in the model is divided into illiterate coded 0, high school coded 1, some college coded 0, and respondents who have completed a basic degree are recorded as 1, while respondent gender is coded for male as 0 and female as 1, and fnally, for land holding size, respondents no land are coded as 1, and respondents own land are coded as 0 (Table 9).
Te following part of the result, i.e., the baseline model of the drivers of LULC dynamics, is the result of the analysis without using any of the model's independent variables (i.e.,   Te baseline model would be used as a benchmark to compare the model to the results including the predictor variables.Te following classifcation tables a,b show that the overall accuracy is 59.3% (Table 10).

Model Goodness-of-Fit Statistics.
Goodness-of-ft statistics showed that the study was designed to determine whether the model is suitable to describe the data.Te model is signifcant, and the likelihood ratio chi-square test revealed that the full model represents a signifcant improvement in ft over a null model and represents a significant improvement in ft over a null model, x 2 (23.971),P < 0.001.Te Hosmer and Lemeshow test statistic indicates a poor ft if the signifcance value is less than 0.05 [51].In this test, the Hosmer and Lemeshow model has a good ft, as the chi-square value is 4.450, P � 0.814, which is greater than 0.05.
According to To further clarify, there is not much diference between the observed and expected drivers of LULC dynamics for natural forces and anthropogenic forces from the contingency test for Hosmer and Lemeshow, so the model adequately fts the data displayed clearly with the line graph (Figure 8)..As shown in Table 5, the classifcation table a provides an indication of the model's ability to predict the correct class when predictors are added to the study.Te percentages in the frst two lines, that is, whether socioeconomic variables are drivers of land use/ land cover change and whether natural and anthropogenic forces are selected, provide information about the specifcity and sensitivity of the model in predicting group membership of the dependent variable.Anyone can compare this classifcation table shown for Block 0 Table 10 to see how much improvement there is when the predictor variables are involved and these tables classifcations are used for comparing with the result of observed and predictor variables included in the model.In this result, there was variation between Block 0 Table 10  Specifcity which is called the true negative rate is the percentage of observed cases that fall into the nontarget (or reference) category [52].Te model correctly predicted 47.8% of respondents choosing natural forces as the driving force for LULC dynamics.In another word, the classifcation predicts that respondents would have chosen anthropogenic forces.

Variables in the Model Equation
Specifcally, it represents information about the degree to which observed results are predicted by the model used.
Sensitivity, also known as true positive rate, is the percentage of observed cases that belong to the target group, i.e., Y � 1, in which case the anthropogenic forces are chosen and the model correctly predicts that those who belong to this group (predict the chosen) anthropogenic forces model with a sensitivity of 78.6%.Generally, the accuracy is very high at 66.1%, as the overall percentage of correctly classifying the socioeconomic variables selected by respondents is the driver of the model-based LULC variation (Table 12).

Variables in Model Equations
. Table 13 provides the regression coefcients (B), Wald statistics (for testing statistical signifcance), and the most important odds ratios (Exp (B) for each variable category) and also shows the relationship between the predictor variables and the outcome.Te result for educational status is highly signifcant (Wald = 16.287,df = 3, p < 0.000), and B (Beta) is the predicted change in log odds for a unit change in the ratio between the odds (Table 13).
So, the Exp (b) of the variable "Gender" is 0.869, and the researcher can state that the probability of a respondent choosing the anthropic forces that provide International Journal of Forestry Research "Gender" is 0.869 times greater than that of the person who chooses "Gender," i.e., the person who chooses "anthropogenic forces" 86.9% higher than those who chose "natural force."Te variable educational status (1) has an Exp (b) of 2.596 and exp (b) of 2.596 times more likely than an uneducated person.People with a high school education have a 259.6% higher chance of choosing anthropogenic forces to participate in LULC dynamics than those without a high school education.For exp (b) educational status (2), the categorical variable is 2.610.
People with some college education have a 261% higher chance of choosing anthropogenic forces for LULC dynamics than people without some college education.In the case of exp (b), educational status (3) is 7.867.Tose with a college education had a 786.7% higher chance of choosing anthropogenic forces for LULC dynamics than those without a college education and not providing natural forces, with a 95% CI of 2.146 to 28.843.Te Exp (b) for the continuous variable age is 1.032.Te Exp (b) for the continuous variable age is 1.032.Te authors of [53] indicated results suggest that the aging of the study area is driving the observed changes (from the point of view of human activity), based solely on the basis of statistical models that provide adequate predictions of changes and comparisons of socioeconomic variables.Te poor predictive performance highlights the problem of using aggregated socioeconomic variables; i.e., more selective socioeconomic variables are available.In this study, the coefcient of the value of the continuous independent variable exp (b) is almost equal to 1, which indicates that subjects who present higher values of the variable have the same probability of success as subjects who present lower values of the variable (i.e., the independent variable has no signifcant efect on the response variable) (Table 11).

Socioeconomic Drivers of LULC Changes.
Socioeconomic data derived from distribution of survey questions to households, focus group discussions, key informant interviews, and observed feld data were analyzed.Te approach is comprehensive and strives to quantitatively and qualitatively characterize land use and the ongoing classifcation of the main drivers of land use change by combining many diferent data sources.
Te following survey questions were asked for the selected sample households in the study area.Distributed and collected from a sample of households selected in the study area, this question is "Is socioeconomic variables (age, sex, educational status, and size of land ownership) afect land use and the drivers (natural or anthropogenic) of land use dynamics?
In this study, researchers analyzed the socioeconomic variables that drivers of LULC change in the study area.Terefore, the socioeconomic variables for LULC change drivers in the study area are age, gender, educational status, and landholding size.Tese independent variables were analyzed one by one below.
Te question was asked to the selected respondents: "Is the socioeconomic variable of age infuence the driving forces of land use and land use dynamics in your village?"For this survey question, almost more than 60% of the respondents (i.e.62%) answered that age is the driving force of LULC dynamics in their village, of which 25.8% fall under 31-40 age group and the rest 0.3.2%,9%, 10%, and fnally, 13.6% were found 61-70, 51-60, 20-30, 41-50 under the age group, respectively (Table 4).
Regarding the gender of the respondents, the question asked to the selected respondents was "Is the socioeconomic variable gender a driver of LULC dynamics in your village?"For this survey question, more than 80% of the respondents (i.e., 80.5%) answered that gender, especially men, is the driving force of LULC dynamics in their villages, while 19.5% were women and were the driving force of LULC dynamics in their villages.Finally, 39.4% of the respondents responded that LULC change is driven by natural forces, not gender, they answered "no," while a majority of respondents around sixty percent answered that gender is a socioeconomic variable, with driving forces that LULC change (Table 5).
Relating to the educational status of the respondents, the question asked for selected respondents was "Is the socioeconomic variable educational status a driver of land use/ land dynamics in your village?"For this survey question, 19.5% of respondents answered that educational status is not a driver of LULC dynamics.Finally, 39.4% of respondents answered that LULC dynamics are driven by natural forces rather than educational levels, choosing "no," while the majority of respondents of similar gender, about 60% of respondents, answered that educational status is a socioeconomic factor (Table 6).
In the case of landholding size, as the number of landless respondents increases, the drivers of land use/land cover dynamics increase, while as the number of landless respondents decreases, the drivers of land use/land cover dynamics also decreases (Table 7).
3.6.Discussion.According to the respondent's explanations, household age groups play an important role in driving land use/land cover dynamics, especially for developing countries, because when their age groups reach 20 years and above; households want to increase their land holdings during illegal segregation, which leads to an alarming increase in land use rates or land use changes.Tis result is almost similar to the work of [10,54,55]  Some educated households responded separately that 1.8% undergraduate, 4.5% undergraduate, and 13.1% educational status were also not a driver of land use/land cover dynamics in their village.However, among the respondents, 7.7% of undergraduate students, 8.1% of partial college students, 22.6% of high school students, and 22.2% of illiterate students, respectively, responded that educational status is the driving force of land use and land use dynamics in their villages.Tis aspect is consistent with other works in Ethiopia [56] on the title "Drivers and Impacts of Land Use and Land Cover Change in the Central Highlands of Ethiopia," which mentioned that based on the socioeconomic characteristics of the households surveyed, 23% of the respondents were an illiterate person (see Table 6).

International Journal of Forestry Research
Land use/land cover changes are the result of human impacts, biophysical drivers, and natural processes [57][58][59].
Focus group discussions were conducted among selected household heads and local people regardless of their social status.
According to the ideas raised during the focus group discussion on "Is socioeconomic variable gender is a driving force for LULC dynamics?"Tere is a huge change and urgent decline of land use/land use dynamics in the study area as stated by household heads and local people.Te main reason for the decline is the independent variable of gender, especially women; men are responsible for the conversion of land cover from forests to agriculture and settlements.Te results from focus group discussions indicated that past LULCCs were prioritized based on ideas presented by men.
Te last question was posed for the informal discussion (interview): "Is the socioeconomic variable education level is a driver of LULC dynamics in your village?"In response to these joint analysis questions, key informant interviews were conducted with older adults who know and have lived in the study area for a long time to obtain more reliable information on the changes and drivers of LULC change in the study area.Te results of the informal discussions (interviews) were almost similar to the focused discussions, indicating that the key solution to changes in the past and current situation of LULCC is to increase the number of educated populations, especially in developing countries [60][61][62].According to these selected interviewers, the level of educational status is directly related to the drivers of LULC dynamics, suggesting that as educational status increases, the drivers behind forest decline (a specifc recent trend) are, increased agricultural production overwhelms forests through a variety of mechanisms decreased.Educational status, then, is the most important socioeconomic independent variable in LULC change drivers.

Conclusions
Tis study analyzes the main socioeconomic variables of LULC drivers of land use dynamics in the Wayu-Tuka district.Tis study analyzes the main socioeconomic variables of land use/land use drivers of the Wayu-Tuka district.Data were obtained from Landsat imagery classifed through Arc GIS used to develop the LULC map of the Wayu-Tuka district.Accuracy assessment and Kappa coefcient are used to confrm the accuracy of the LULC classifcation of farm land, settlement areas, bare land, forest land, and water bodies, which are the main land use/land cover changes in the district.Te forest coverage in the study area over the past thirty years (1990-2020) decreased from 12.1% in 1990 to 2.6% in 2020.For socioeconomic data, research questions were distributed to households and focus group discussions, and key informant interviews and feld data collection were analyzed.Pearson correlation was used to determine the relationship between the main socioeconomic determinants of the independent variables and the driving forces of land use/land cover dynamics in the selected district.Te result of the study showed that the independent variables (i.e., age and gender) have a positive Pearson correlation with the driving forces of LULC dynamics; this implied that as these independent variables increase, the driving forces of LULC change also increase, whereas education status has a negative correlation, showing that as the number of an educated population increases, the driving forces of anthropogenic forces for LULC dynamics decreased and vice versa.Te binary logistic regression model examined the relationship between the (dependent) and the main socioeconomic (independent) variables.Logistic regression was performed to determine how variables such as age, gender, educational status, and landholding size are the driving forces for LULC change (natural forces or anthropogenic forces).Te logistic regression model was statistically signifcant (x 2 � 23.971, df � 5, P < 0.001).Te model explained 13.9% (Nagelkerke R 2 ) of the variance in driving forces for LULC dynamics and correctly classifed 66.1% of cases.Age, gender, and educational status largely determine the driving forces for LULC dynamics.Te driving force for LULC dynamics has the greatest chance of selecting the anthropogenic forces.Te survey question was asked for selected sample HHs of the study area and almost more than sixty percent of the respondents responded to this survey question, i.e., H. 62% that age is the driving force for the LULC dynamics in their village: 25.8% are under 31-40 years age group, and the rest (0.3.2%, 9%, 10%, and 13.6%) were 61-70, 51-60, 20-30 and 41-50 among the age group, respectively.Finally, the result of this study also showed that 39.4% of the respondents had 16 International Journal of Forestry Research answered that natural forces are the driving forces of LULC dynamics not at the educational level and they chose "No", while the majority of the respondents were similar in gender, about sixty percent of respondents responded that educational status was a socioeconomic variable that facilitated the driving forces of LULC dynamics.Terefore, concerned bodies should minimize the driving forces of LULC dynamics, especially anthropogenic forces, to rehabilitate and restore the landscape in the Wayu-Tuka district, western Ethiopia.

Figure 3 :
Figure 3: Main soil types in the study area.

Figure 8 :
Figure 8: Comparison of observed and expected drivers of LULC dynamics (2023).

Table 1 :
Landsat images and their characteristics used in this study.

Table 4 :
Are socioeconomic variables age factors in LULC dynamics?
*Age of respondents cross-tabulation.sented in percentages, tables, and fgures.Te analytical data were useful in determining and quantifying the driving factors of the land use/land cover change in the study area.

Table 5 :
Are socioeconomic variables gender factors in LULC dynamics?
8 Gender of respondent cross-tabulation.8InternationalJournal of Forestry Research

Table 6 :
Is socioeconomic educational status is driving force for LULC dynamics?
*Educational status of respondents' cross-tabulation.

Table 7 :
Are socioeconomic variables land holding size driving forces for LULC dynamics?
* Size of respondent's land holdings cross-tabulation.

Table 8 :
Correlation coefcient values between SEV and LULC driving force.
Table 11 of the Hosmer and Lemeshow test contingency table, the test indicated that the model adequately fts the data.As anyone can see, there is no diference between what is observed and what is predicted models.Tis means that both values are almost the same.
classifcation Table a,b and Table 5 classifcation Table a .Te model correctly classifed a total of 66.1% of cases.Tis is the rate correct classifcation if the researcher always predicts that a respondent would choose anthropogenic forces.

Table 9 :
Coding of categorical independent variables.

Table 10 :
Classifcation Table a,b of dependent variables of LULC dynamic driving force.

Table 11 :
Contingency table for Hosmer and Lemeshow test.
titled "Socioeconomic Drivers of Land Use/Cover Dynamics and Teir Impacts in Walecha Watershed, Southern Ethiopia."Table 7 also improves the ones shown under each item percentage of reasons by age group.

Table 12 :
Classifcation Table a for step 1.

Table 13 :
Variable in the equation.Variables entered in step 1: age, gender, educational status, and landholding size. a