Perception of Smallholder Farmers about Climate Change and Its Impacts on Crop Production across Agroecological Zones of the Gassera District, Southeastern Ethiopia.

doi:10.21203/rs.3.rs-5289926/v1

Climate change and variability have threatened rainfedagriculture by affecting the livelihoodsof rural communities in Ethiopia. The study area, Gassera District, is among the high-potential crop production areas of the Bale Zone and is severely impacted by recurrent droughts resulting from climate change. This study evaluated smallholder farmers' perceptions of significant climate change and its effects on food crop production across the agroecological zones of the Gassera District. A cross-sectional survey design was employed to collect data from 444 farm households via multistage random sampling techniques. Multiple linear regression (MLR) models were used for the data analysis. The results revealedthat 98.5% of the interviewed farmers were aware of climate variability and that 51.6% understood its impact to a reasonable extent. However, over half of the farmers did not perceive climate change as the greatest threat to their livelihood. Most farmers experienced rain becoming more erratic, starting late, and ending early as medium climatic factors (60.4%, 68.1%, and 66.2%, respectively), affecting their crop production. The results revealed that rainfall had a negative and insignificant decreasing trend (2.92 mm/year). The annual mean temperature exhibited a positive and statistically significant increasing trend (Ρ < 0.01). Crop production is positively and linearly correlated with the amount of annual rainfall at the Ρ ≤ 0.05 level of significance. The findings revealed that the greatestcrop yield loss was associated with lowland agroecology. The MLR results revealed that farmer agroecology, age, sex, and chemical use had substantial impacts on crop yield loss. We urge farmers to understand the long-term effects of climate change on their livelihoods.

Climatology

Climate change

climate variability

impact on crops

farmers' perceptions

vulnerability.

Over the past 50 years, we have observed climate change and its effects on agricultural production worldwide[1]. As their economies are built on agricultural products, residents in underdeveloped nations are vulnerable to the effects of climate change[2],[3]. Countries in Sub-Saharan Africa (SSA) are characterized by a low capacity to adapt to climate change, with approximately 62% of the population currently employed in the climate-sensitive agriculture sector[4],[5]. The productivity of all agricultural sectors has been affected by climate-related extremes, with detrimental effects on food security and livelihoods being severe for those living in SSA[6]. Some studies have investigated the effects of smallholder rain-fed crop production on different crops to various extents, and the timings have been used to predict these impacts[7],[8]. The IPCC report additionally revealed that maize and wheat yields in SSA have decreased by an average of 5.8% and 2.3%, respectively, owing to climate change[6].

Ethiopia has also been negatively affected by the impacts of climate change due to changing rainfall patterns and warming trends, which could increase the risk of rain-fed agriculture and increase food insecurity [9]–[11]. Similarly, climatic data gathered in the study area in the past two decades from 1983–2012 also revealed a significantly increasing trend in the mean surface temperature and decreasing rainfall with recurrent droughts [12]–[15]. The Gassera district used to be one of the high-potential crop production areas of the Bale zone, which is severely affected by a recurrent drought resulting from climate change [13],[16].

Some studies have examined trends, farmers' perceptions, and the impacts of climate change on farm households' food crop production in different parts of the country [17]–[21]. Some scholars have used correlation and regression models [3],[15],[19],[21]–[23] and the Ricardian approach [17],[20] to investigate the impacts of climate change on farm households' food crop production in Ethiopia. For example, a study conducted in the Konso district, in Ethiopia, revealed that both crop and livestock production were positively correlated with the amount of rainfall [21]. Another study in southern Ethiopia indicated that over the last three decades, climate change has been found to negatively affect household food security, and crop production has been constrained by poor rainfall, severe erosion, and temperature increases [3]. Moreover, in a study of farm households in 50 districts from different AEZs in Ethiopia, the results from crop revenue per hectare regressed on climate, household, and soil variables were found to have a significant effect on the net crop revenue per hectare of farmers [23].

Some studies in and around the Gassera district of the Bale Zone have also assessed farmers' awareness, trends, and impacts of major climatic change variability[14]–[16],[24]. Earlier studies have identified a gap in farmers' awareness levels and trends of climate change in the Agarfa and Sinana districts of the Bale Zone[15]. Another study examined livelihood vulnerability and coping mechanisms in the Rayitu, Guradamole, and Dawe Qachen districts of the Bale zone, focusing on pastoral communities with limited coping mechanism variables[24]. In addition, a study conducted in the Sinana, Ginir, Goro, and Gololcha districts of the Bale Zone also examined the economic impact of climate change on crop production and farmers' adaptive capacity [14]. This study revealed that an increase of 1°C in the mean summer temperature decreased farmers' net revenue per hectare by approximately ETB 1185.14.

A study by Legese et al. (2018)[16] examined recent trends in climatic variability, farmer perceptions, and adaptation strategies in response to climate change in the Sinana, Ginir, Goba, Gassera, and Goro districts of the Bale highlands. For the last 20 years, the average annual temperature trend in the Gassera district increased by approximately 0.08°C, and the mean annual rainfall decreased at a rate of 6.82 mm/year from 1995–2014. However, measuring the trend of climate change variability in a given area usually requires 30 years of climate data as more scientific recommendations [6],[25]. In addition, this study used only one climate change variable, i.e., temperature, to measure farmers' perceptions and did not capture the impacts of climate change on farm household crop production in the Gassera district at the agroecological level. In addition, most of these studies have focused on the Bale highlands[14]–[16],[24] and identified uniform intervention options as major adaptation strategies for a changing climate in different agroecological zones (AEZs).

In addition, farmers' perceptions and expectations of climate variability play a major role in understanding how well they adjust and cope [26],[27] and are considered crucial prerequisites for improved farming practices and adoption [28],[29]. Previous studies have shown that households in different agroecological zones (AEZs) differ in their perceptions of climate change and hence use different adaptation methods [30]–[33]. These studies also revealed that variations in climatic factors such as soil conditions and other factors across different agroecology can influence farmers' perceptions of climate change and their decisions to adapt. The views and perceptions of climate change and its impact on various agroecology may influence how the local people react and adjust to various climatic effects accordingly. Despite assessing trends in major climate variables, there is no adequate evidence from studies conducted on farmers' perception levels and impacts of climate change. Its impacts on farm households' food crop production in the Gassera district, as well as at the agroecological level, are also important.

Hence, this study examined the effects of climate change on the production of cereal crops as well as other variables, such as rainfall, important sociodemographic parameters, and agroecological aspects, that are thought to influence farmers' perceptions of climate change variability.

Therefore, the purpose of the current study was to evaluate how farmers perceive climate variability and investigate how climate change may affect major cereal crop production among smallholder farmers in the Gassera area. As a result, the study outputs will be crucial for developing local and regional plans for adapting to climate change and for helping policymakers create interventions that would help farmers manage risks associated with climate change and thus enhance rural livelihoods.

Socioeconomic characteristics of the sample households.

The study revealed that 81% of households (HHs) were male-headed, with 72.4% from lowlands, 92.2% from midlands, and 80.0% from highland regions. The remaining 19% of households were female-headed. This result is similar to that of the national census, which showed that women account for approximately 16% of HHs in rural areas of the country[34],[35].

Furthermore, details regarding the age group of the respondents revealed that the majority (37.1%) of the HH heads were 50 years and older, whereas a few (7.8%) HH heads were between 20 and 29 years of age. From the agroecological perspective, 37.1% (29.4% from highlands, 49.6% from midlands, and 36.2% from lowlands) of the surveyed headed households were 50 years and older, followed by 35.6% (48.2% from highlands, 29.2% from midlands, and 23.3% from lowlands), who were between 40 and 49 years of age. The mean age of the respondents was 45.64 years, with a standard deviation of 11.45 years.

In terms of marital status, approximately 82.5% of the surveyed-headed households were married, followed by 9.0% who were divorced and 6.0% who were single-headed. Additionally, 57.1% (71.8% from the highlands, 36.3% from the midlands, and 56.0% from the lowlands) of HHs were from families of 4–6 persons, 18.5% from 7–8 members, and 17.5% from 1–3 families. The average family size was five persons per HH, which is similar to a previous report[35] that revealed an average rural area of the country had approximately five individuals (5.2 in rural areas and 3.6 in towns).

Concerning educational status, approximately 38.6% (49.4% from highlands, 26.5% from midlands, and 34.5% from lowlands) of the HHs replied that they attended basic education levels (grades 5–8). Approximately 30.8%, 22.1%, and 8.5% of the HHs sampled in all the AEZs replied that they had attended primary (grades 1–4), secondary general, and illiterate schools, respectively. With respect to religion, the largest percentage of HHs were Orthodox Christians (54.6%), Muslims (44.4%), and Protestants (1.0%) (Table 1).

Table 1

Respondents' socioeconomic characteristics across the AEZs. Note: The numbers in columns (1)–(4) are percentages. M (s.d.) in column (5) is the mean value for age, family size, and education status of all respondents, and its standard deviation is given in parentheses.

Source: Author's survey (2022).
Variable	Response	Farmer Agro-ecological Zone			Total	M (s.d.)
Variable	Response	Highland (1)	Midland (2)	Lowland (3)	(4)	(5)
Sex	Male	80.0	91.2	72.4	81.0
	Female	20.0	8.8	27.6	19.0
	Total	100.0	100.0	100.0	100.0
Age	20–29	5.9	7.1	11.2	7.8	45.64 (11.45)
	30–39	16.5	14.2	29.3	19.5
	40–49	48.2	29.2	23.3	35.6
	50 and more	29.4	49.6	36.2	37.1
	Total	100.0	100.0	100.0	100.0
Marital status	Single	6.5	2.7	8.6	6.0
	Married	79.4	93.8	75.9	82.5
	Divorced	12.9	1.8	10.3	9.0
	Widowed	1.2	1.8	5.2	2.5
	Total	100.0	100.0	100.0	100.0
Family size	1–3	13.5	17.7	23.3	17.5	5.37 (2.38)
	4–6	71.8	36.3	56.0	57.1
	7–8	13.5	27.4	17.2	18.5
	9 and above	1.2	18.6	3.4	6.8
	Total	100.0	100.0	100.0	100.0
Education status	Illiterate	7.1	13.3	6.0	8.5	5.84 (3.18)
	Primary (Grade 1–4)	28.2	51.3	14.7	30.8
	Basic (Grades 5–8)	49.4	26.5	34.5	38.6
	Secondary (9–12)	15.3	8.8	44.8	22.1
	Total	100.0	100.0	100.0	100.0
Religion	Muslim	36.5	61.1	39.7	44.4
	Orthodox	61.2	38.9	60.3	54.6
	Protestant	2.4	0.	0.	1.0
	Total	100.0	100.0	100.0	100.0
Main cereal crop	Barley and wheat	87.6	75.2	0.0	58.6
	Maize	12.4	21.2	12.1	14.8
	Sorghum and teff	0.0	3.5	87.9	26.6
	Total	100.0	100.0	100.0	100.0

Crop production of sample households.

Barley and wheat (58.6%), sorghum and teff (26.6%), and maize (14.8%) were the major cereal crops produced in the study area. Barley and wheat were the major cereal crops, accounting for 87.6% of the highland (Balo Amina) and 75.2% of the midland (Homa Abu), whereas sorghum and teff, accounting for 87.9%, were the dominant cereal crops in the lowland (Melka Olifigi) AEZ.

Perception of change in climate variability by smallholder farmers.

Farmers across the AEZs were asked to rate their knowledge of climate change and perceptions of any changes in temperature and rainfall (late, early, and erratic rainfall) in the study area. Moreover, the Pearson correlation test was used to determine the important socioeconomic factors that influence farmers' perceptions of climate change (temperature and rainfall).

The overall survey results in the three AEZs revealed that approximately 98.0% of the HHs had replied that they were aware of climate change before, and the remaining 2.0% did not have any clue at all. The study results also revealed variations among the AEZs. A number (98.3%) of HH heads in the lowlands were aware of climate change before the HHs in the midland (98.2%) and highland (97.6%) AEZs. Regardless of agroecology, the survey results also revealed that 51.6%, 41.9%, 3.3%, and 3.3% of the HHs knew about climate change to a reasonable extent, little, to a great extent, and did not have any clue at all, respectively. Similarly, the study results revealed differences among the AEZs in terms of farmers' extent of knowledge about climate change. Most of the smallholder farmers in the midlands (65.5%) replied that they knew little about climate change, whereas most HHs in the highlands (65.9%) and lowlands (49.1%) replied that they knew to a reasonable extent about climate change. This implies that most of the sampled respondents in the study areas were aware of climate change.

Farmers were asked again to respond to changes in climate variability (temperature and rainfall) in their locality. Accordingly, the overall survey results in the three agroecological areas indicated that the majority (98.5%) of the HHs had an equal understanding of the climate variability of temperature and rainfall.

Moreover, farmers' perceptions of changes in the climate variability of temperature and rainfall varied across the three AEZs. More HHs in the highland (98.8%) than in the midland (98.2%) and lowland (98.3%) AEZs experienced significant temperature changes. Moreover, the statistical analysis of the perception of rainfall change also revealed significant variation among the different AEZs. A significant rainfall change was perceived more in the lowland (100%) than by farmers in the midland (98.2%) and highland (97.6%) AEZs. Regardless of agroecology, the survey results also revealed that 51.6%, 41.9%, 3.3%, and 3.3% of the HHs knew about climate change to a reasonable extent, little, to a great extent, and did not have any clue at all, respectively. Similarly, the study results revealed differences among the AEZs in terms of farmers' extent of knowledge about climate change.

The overall survey results indicated that 87.5%, 10.8%, and 1.8% of the HHs perceived an increase, a decrease, and did not know the variability in rainfall starts late in their local area, respectively. Similarly, the majority (86.5%) of the respondents replied with an increase in erratic rainfall, followed by 10.8% with decreases, 1.5% with no change, and 1.3% without a clue. In the rainfall that ends early, 81.5%, 17.3%, and 1.3% of the household heads experienced an increase, a decrease, and no clue at all, respectively (Table 2).

Table 2

Farmers' perceptions (%) of changes in climate and its correlation with socioeconomic factors. Note: The numbers in columns (1)–(4) are percentages. Positive (+) and negative (-) signs in the correlation indicate direct and indirect relationships, respectively. Asterisks (**) indicate the correlation is significant at the < 0.01 level (two-tailed); asterisks (*) indicates the correlation is significant at the < 0.05 level (two-tailed).

Source: Author's survey (2022).
Variables		Responses		Farmer Agroecological Zone			Total
Variables		Responses		Highland (1)	Midland (2)	Lowland (3)	(4)
Have you heard of climate change before now?		Yes		97.6	98.2	98.3	98.0
		No		2.4	1.8	1.7	2.0
		Total		100.0	100.0	100.0	100.0
Have you noticed a significant temperature change?		Yes		98.8	98.2	98.3	98.5
		No		1.2	1.8	1.7	1.5
		Total		100.0	100.0	100.0	100.0
To what extent do you know about climate change?		Do not know		2.9	1.8	5.2	3.3
		Know little		25.9	65.5	42.2	41.9
		To a reasonable extent		65.9	32.7	49.1	51.6
		To a great extent		5.3	0.0	3.4	3.3
		Total		100.0	100.0	100.0	100.0
Have you noticed a significant rainfall change?		Yes		97.6	98.2	100.0	98.5
		No		2.4	1.8	0.0	1.5
		Total		100.0	100.0	100.0	100.0
Have you observed/noticed any changes in rainfall starting late?)		Decreased		18.2	3.5	6.9	10.8
		Increased		78.8	94.7	93.1	87.5
		No change		0.0	0.0	0.0	0.0
		I do not know		2.9	1.8	0.0	1.8
		Total		100.0	100.0	100.0	100.0
Have you observed/noticed any changes in rainfall that end early?		Decreased		18.2	3.5	29.3	17.3
		Increased		80.0	94.7	70.7	81.5
		No change		0.0	0.0	0.0	0.0
		I do not know		1.8	1.8	0.0	1.3
		Total		100.0	100.0	100.0	100.0
Have you observed any changes in erratic rainfall?		Decreased		20.6	1.8	5.2	10.8
		Increased		75.3	96.5	93.1	86.5
		No change		2.4	0.0	1.7	1.5
		I do not know		1.8	1.8	0.0	1.3
		Total		100.0	100.0	100.0	100.0
Pearson correlation between socioeconomic factors, rainfall, and temperature.
Variables		Socioeconomic factors
Variables		Farmer AEZs	Gender	Age	Marital status	Family size	Education level
Rainfall	Pearson Correlation	.127*	.267**	.088	.225**	− .016	− .177**
	Sig. (two-tailed)	.011	.000	.078	.000	.747	.000
Temperature	Pearson Correlation	− .070	.154**	.101*	.179**	− .061	− .123*
	Sig. (two-tailed)	.164	.002	.044	.000	.226	.014

Moreover, 78.8% of respondents from the highlands and the same percentages from the midlands and lowlands 94.7% and 93.1%, respectively replied an increase in the rainfall starts late in the study area. Nonetheless, an increase in the incidence of irregular rainfall was felt by 96.5% of respondents from the midlands, (93.1%) of respondents from the lowlands, and (75.3%) from the highlands.

The results of the correlation analysis between socioeconomic variables and temperature changes revealed a significant (positive) relationship between gender and marital status at the < 0.01 level and a (negative) correlation with educational status at the < 0.05 level. Gender, age, marital status, and education level are important socioeconomic factors that influence farmers' perceptions of temperature variability (Table 2). Moreover, farmers' AEZs, gender, and marital status were found to have a significant (positive) correlation with factors that influence farmers' perceptions of rainfall variability, and it had a (negative) correlation with educational attainment at the < 0.01 level.

Perception of the impacts of climate phenomena on crop production by farmers.

To capture perceptions of climate change impacts, farmers were asked about the impacts of principal climatic parameters (rainfall and temperature) on crop production over the last ten years. The impacts of each climate parameter were categorized as high, medium, low, and no (Fig. 1). Accordingly, regardless of agroecology, the survey results indicated that 48.6%, 36.1%, 12.8%, and 2.5% of the HHs perceived the variability in temperature to result in high, medium, low, and no impact on crop production in their local area, respectively. Similarly, the overall results indicated that the majority (42.9%) answered that rainfall variability has had a medium influence on crop production, followed by 41.1% with high impact, 14.3% with low impact, and 1.8% with no impact.

Moreover, the results revealed variations among the AEZs. A large number (58.6%) of HH heads in lowland agroecology responded that rainfall variability resulted in high impacts on crop production, whereas most of the farmers in highlands (42.9%) and midlands (61.1%) replied that rainfall variability resulted in a medium impact on crop production.

Additionally, the largest percentage of respondents perceived that rains occurred more erratically, started late, and ended early as a medium factor affecting their crop production. For instance, 68.1% and 16.3% of the farmers perceived that rains started late had medium and high impacts on crop production, respectively, over the last 10 years, whereas 15.1% and 0.5% of the farmers replied that rains had low and no impacts on their crop production, respectively.

However, the medium impacts of late-started rainfall on crop production also revealed some percentage differences among the AEZs in terms of farmers' perceptions. More HHs in the lowland (72.4%) than in the highland (70.3%) and midland (60.4%) AEZs perceived the medium impacts of late-started rainfall on food crop production.

Overall, 66.2% of the HHs perceived a medium impact on crop production from early ending rains in their local area. Moreover, the medium impact on crop production because of rains ended early for the highland, midland, and lowland AEZs at 71.9%, 62.2%, and 62.1%, respectively. The results also indicated that, regardless of agroecology, rains became more erratic, and the largest percentage of the respondents, 238 (60.4%), replied as having a medium impact on crop productivity over the last 10 years, followed by 98 (24.9%) with a high impact, 52 (13.2%) with a low impact, and 6 (1.5%) with no impact.

Despite this awareness of the climate, most HHs do not perceive climate change to be the greatest risk to their livelihoods and success. Since the general interviewed farmers, only approximately half of the HHs replied that rainfall (41.1%) and temperature (48.6%) variability have had a high impact on their crop production over the last 10 years.

Moreover, among all the interviewed HHs, the largest percentage (60.4%, 68.1%, and 66.2%) of the respondents perceived that rains became more erratic, started late, and ended early, respectively, as a medium climatic factor that affected their crop production.

The trend of climate variability and crop yield.

Linear regression and correlation tests were used to examine the trends and changes over time in the climate variables from 1990–2019 and to examine the relationships between the annual rainfall and crop yield in the study district. Accordingly, the results from the meteorological analysis revealed that the annual temperatures in the district showed statistically significant increasing trends at the < 0.01 level, whereas the annual rainfall showed insignificant (Ρ > 0.05) decreasing trends over the past 30 years.

Moreover, for the period from 1990–2019, the mean annual temperature of the Gassera district increased by 0.037°C per year (Fig. 2). As shown in the time series plot with the estimated linear trend, the mean annual rainfall decreased by 2.924 mm/year during the same period. The lowest annual mean temperatures and rainfall were recorded in 1993 and 2015, respectively, whereas the highest annual mean temperatures and rainfall were observed in 2015 and 1992, respectively.

With respect to the relationship between annual rainfall and crop production, the value of crop yield is presented in Fig. 2. The results revealed that as the amount of annual rainfall increased, the crop yield also increased. The values of barley, wheat, sorghum, maize, and teff yields (Qt/ha) were high in the years 2010 (with a rainfall of 1368 mm) and 2014 (with a rainfall of 1010 mm). The lowest yields of barley, wheat, sorghum, and teff yield crops were obtained in 2012 (with a rainfall amount of 826 mm), except for the yield of maize.

Effect of rainfall on yield and loss of crop yield by smallholder farmers.

Moreover, correlation and regression were used to investigate the effects of or relationship between our response variable (crop production) and the explanatory variables of annual rainfall in the study area.

Effect of annual rainfall on crop production.

Below are the results of the ANOVA, the linear regression analysis, and the model summary. The model coefficient, as shown in Table 3, was determined to be .989, which indicates that 98.9% of the variation in how rainfall changes affect crop output in Quintal (Qt) of the study region can be accounted for by the variables in the model. The model explained approximately 96.3% of the variability in crop yield, according to the adjusted R square. The linear correlation coefficient was displayed as R.995a, and the combination of these variables strongly predicted the dependent variables, as indicated by Sig.026.

Table 3

Model summary, ANOVA, and results of the linear regression analysis. *Note*: Barley production in Quintal (continuous); Wheat production in Quintal (continuous); Sorghum production in Quintal (continuous); Maize production in Quintal (continuous); and Teff production in Quintal (continuous). Positive (+) and negative (-) signs indicate direct and indirect relationships, respectively. Asterisks (*) are statistically significant at the$\:\:<0.05$ level (two-tailed); ^ns, nonsignificant.

Source: GDARDO, 2022.
Model summary ^b
Model	R	R Square	Adjusted R Square	Std. error of the estimate	Sig. F Change
1	.995^a	.989	.963	45.93546	.026
^b indicates all predictors/independent variables: (constant).
ANOVA ^a
Model	Source	Sum of Squares	df	Mean Square	F	Sig.
1	Regression	391389.174	5	78277.835	37.097	.026^b
1	Residual	4220.132	2	2110.066
	Total	395609.306	7
^a indicates the dependent variable: Annual rainfall (mm),$\:\:\text{a}\text{n}\text{d}\:$^b indicates the independent variables: the value of crop production in Quintal (Qt).
Model	Variables	Coefficient	Std. Deviation	t-value	P-value	Correlations
1	(Constant)	-312.894		-1.646	.026*
	Barley production (Qt)	.080	269382.617	.564	.629 ^ns	.673
	Wheat production (Qt)	.896	1044607.093	4.714	.042*	.141
	Sorghum production (Qt)	-1.274	71529.463	-6.165	.025*	− .165
	Maize production (Qt)	− .645	179087.131	-5.621	.030*	− .540
	Teff production (Qt)	1.428	140918.535	6.462	.023*	.176

The least-squares estimator of the intercept, β0, was found to have a value of -312.894 and was labeled a constant. The least-squares estimators of the five partial slopes (080, .896, -1.274, − .645, and 1.428) represented barley, wheat, sorghum, maize, and teff production in Quintal, respectively. Accordingly, wheat, sorghum, maize, and teff production in Quintal (Qt) were found to be significant at the < 0.05 level variables that predicted the response variable effects of annual rainfall change on crop production in the study area at a 5% level of significance. The $\:\:\text{P}$-values for wheat (0.042) and teff (0.023) crop production in Quintal were found to be positively significant variables affecting crop production. Thus, the study results revealed that the amount of wheat production increased by 0.89 Qt as the amount of rainfall increased by one millimeter. Similarly, the amount of teff production increased by 1.43 Qt as the amount of rainfall increased by one millimeter.

Moreover, the $\:\:\text{P}$-values for sorghum (0.025) and maize crops (0.030) were much lower than 0.05 and were found to be significantly negatively related to the effects of annual rainfall change on crop production at the 5% level of significance. For instance, as the amount of annual rainfall increased by one millimeter, the amount of sorghum production decreased by 0.645 Quintal (Qt).

This might be because of climate variability and extreme events of a high level of seasonal rainfall variability, unexpected decreases in rainfall, and the frequent occurrence of drought episodes in the study area. For instance, a study conducted between 1983 and 2012 on climate variability across possible crop-growing regions in Ethiopia revealed a considerable shift in the rainfall coefficient of variation, with greater interannual variability than annual variability in the Bale Zone [12].

Climate-induced loss of crop yield by smallholder farmers.

Moreover, farm householders were asked about the most vulnerable human elements, the outcome of climate change with multiple optional responses, and the estimated amount of crop yield loss (in kg) over the last ten years due to climate change. Accordingly, regardless of agroecology, the survey results revealed that the most vulnerable family members (human elements) in the study area to the impacts of climate change were 84.5% (337) women, 77.4% (309) children, 44.1% (176) elders, and 2.5% (10) men. In addition, regardless of agroecology, all respondents (100.0%) reported that the decline in crop yield was the outcome of climate change, followed by food shortages (97.2%) and increased food prices (92.2%) (Table 4).

However, only 40.6%, 24.6%, and 23.6% of the smallholder farmers in the three AEZs agreed that soil nutrient depletion, erosion hazards, and early maturity of crops were the outcomes of climate change, respectively. Thus, approximately two-fifths of the farmers did not consider soil nutrient depletion, erosion hazards, early cropping, or flooding as potential outcomes of climate change and variability. This implies that all the respondents in the three AEZs confirmed that the decline in crop productivity over the last few years in the study area was the principal outcome of climate variability.

In addition, the HH heads in the three AEZs that responded to the decrease in crop yield were again asked to mention the total estimated amount of loss of crop yield (in kg) due to climate change in the last ten years. Accordingly, survey results from 399 smallholder farmers revealed that, on average, 2,700 kg of crop yield was lost because of climate change (Table 4).

Table 4

Farmers' perceptions of the outcome and amount of crop yield loss (kg). s.d. is the standard deviation for the farmers' estimated amount of crop yield loss (kg).

Source: Author's survey (2022).
Variables			Responses		Frequency		Percentage
Outcome of climate change (Multiple answer option)			Decrease in crop yield		399		100.00
			Increase in crop yield		22		5.5
			Food shortage		388		97.2
			Food price increase		368		92.2
			Early maturity of the crops		94		23.6
			Flooding		74		18.5
			Erosion hazard		98		24.6
			Soil nutrient depletion		162		40.6
Who has been most affected by climate change?			Children		309		77.4
			Women		337		84.5
			Men		10		2.5
			Elders (50 and more)		176		44.1
Farmers estimated the amount of loss of crop yield due to climate change (in kg) for the last ten years.
Variables	Responses
Estimated crop yield loss across AEZs and crop type.	AEZs	Crop type	Mean	Range	s.d.	Variance	Sum
	Highland	Barley and wheat	1,409.53	5,800	1,074.03	1,153,531.14	239,620
	Midland	Maize	2,902.65	11,000	2,189.93	4,795,796.46	328,000
	Lowland	Sorghum and teff	4,396.55	9,600	2,651.63	7,031,118.44	510,000
	Total		2,700.80	11,800	2,330.85	5,432,864.18	1,077,620

Statistical analysis also clearly revealed that the range, variance, and total amount of crop yield lost because of climate change were 11,800 kg, 5,432,864 kg, and 1,077,620 kg, respectively. From an agroecological perspective, the study results from the sample respondents revealed that the highest (510,000 kg) total amount of crop yield loss was found in lowland (Melka Olifigi) agroecology, followed by midland/Homa Abu (328,000 kg) and highland/Balo Amina (239,620 kg) in the last ten years, resulting from climate change.

This implies that lowland agroecology has the greatest crop yield loss, which is most affected by climate change variability, whereas highland agroecology is the least affected by climate change variability. The results from MLR also revealed that, compared with farmers living in midland and lowland agroecology, farmers living in highland agroecology experienced a decrease in crop yield of 143.44 kilograms per hectare (kg ha^− 1). The results of the FGDs and interviews verified that the climate was changing and confirmed that the decline in crop productivity was the principal outcome of climate variability over the years in the study area.

Impacts of climate variability on crop production.

This section addresses the findings of the MLR model, that revealed which parameter was more responsible for productivity fluctuations and to what extent the independent variables explained the effects of climate variability on the crop production loss of smallholder farmers.

Goodness of fit and estimation of multiple regression coefficients.

The SPSS software was used to find the $\:{\text{R}}^{2}$ and least squares estimates for $\:{\beta\:}_{0},\:{\beta\:}_{1},\:{\beta\:}_{2},\:{\beta\:}_{3},\:{\beta\:}_{4},\:{\beta\:}_{5},\:{\beta\:}_{6},\:{\beta\:}_{7},\:{\beta\:}_{8},\:{\beta\:}_{9},\:{\beta\:}_{10},\:\text{a}\text{n}\text{d}\:{\beta\:}_{11},\:$ in the multiple regression models. The results are displayed in the table below. Table 5 shows that the model $\:{\:\text{R}}^{2}\:$ value was .956, which indicates that the variables in the model accounted for 95.6% of the variability in the consequences of climate change on the loss of agricultural output experienced by smallholder farmers (kg ha^− 1). Moreover, the adjusted R square in this instance showed that the model explained almost 95.5% of the variability in farmers' crop production loss.

The multiple correlation coefficient is displayed as R.978^a, and Sig.000 suggested that the variables together significantly (P < 0.05) predicted the dependent variables. Furthermore, the coefficients for each variable in the model related Y to X₁, X₂... X₁₁ shows how much one could anticipate that crop yield loss would change as a result of climate variability given a one-unit change in that variable's value and the assumption that all other variables in the model would remain constant.

The least-squares estimator of the intercept, β0, was found to have a value of 871.309 and was labeled constant. The least-squares estimators of the eleven partial slopes − 143.440, -134.712, 221.591, -25.802, 221.591, -25.802, 215.850, -93.624, 326.631, -28.970, -68.180, 63.814, and 49.779 represented farmers' agroecology type (highland, midland, and lowland), sex, religion, to rate their knowledge of climate change, whether they have access/primary sources of information regarding climate change or not, use of chemicals (such as herbicides or insecticides), use improved seed varieties of crops, marital status, education status, age of household heads, and family size, respectively. Table 5 presents the point estimation of the regression coefficients and the P-values of the variables.

Table 5

Model summary, ANOVA, and results of multiple linear regression analysis. Note: HH head agroecology type (highland = 0, midland = 1, and lowland = 2), headship type (male = 0, female = 1); religion, HH head religion affiliation (Muslim = 0, orthodox = 1, protestant = 2, catholic = 3, others = 4); HH head rating their knowledge status about climate change, (do not know = 0, knew little = 1, to a reasonable extent = 2, to a great extent = 3); do farmers have access/ primary source of information regarding climate change (did not have = 0, have = 1); HH head use of chemicals such as herbicide, insecticide (not use = 0, use = 1); HH head use of improved seed varieties of crops (not use = 0, use = 1); marital status (single = 0, married = 1, divorced = 2, widowed = 3); HH head education status, maximum class attended by the household head (continuous); age of household head in years (continuous); family size, total family size in HH (continuous). Positive (+) and negative (-) signs indicate direct and indirect relationships, respectively. Asterisks (*) are statistically significant at the < 0.05 level; ^ns, nonsignificant.

Source: Author's survey (2022).
Model summary ^b
Model	R	R Square	Adjusted R Square	Std. error of the estimate	Sig. F Change
1	.978^a	.956	.955	72.90632020	.000
^b indicates all predictors/independent variables: (constant).
ANOVA ^a
Model	Source	Sum of Squares	df	Mean Square	F	Sig.
1	Regression	45123208.672	11	4102109.879	771.751	.000^b
1	Residual	2057033.300	387	5315.332
	Total	47180241.972	398
^a indicates the dependent variable: the value of farmer's yields loss of crop (kg ha^− 1) and ^b indicates predictors: (constant), family size, and use of any insecticide/pesticide on a farm for cultivation within the last two years.
Model	Variables		Coefficient	Std. Err	t–value	Ρ–value
1	(Constant)		871.309	35.305	24.679	.000*
	Farmer's agroecological zone		-143.440	4.622	-31.035	.000*
	HH heads Sex		-134.712	12.771	-10.548	.000*
	The religion of HH heads		221.591	7.412	29.898	.000*
	Knowledge extent of climate change		-25.802	6.605	-3.906	.000*
	Access to climate change information		215.850	26.044	8.288	.398 ^ns
	Use of any insecticide or pesticide		-93.624	21.986	-4.258	.036*
	Use of improved seed varieties of crops		326.631	15.090	21.646	.000*
	Marital status of the HH heads		-28.970	10.377	-2.792	.435 ^ns
	The highest-grade level of HHLs completed		-68.180	1.345	-50.688	.000*
	Age of the HH heads		63.814	5.189	12.299	.000*
	Family size		49.779	5.691	8.746	.000*

Hypothesis test of the multiple linear regression model.

Analysis via the multiple linear regression model revealed that farmer agroecology type, sex, religion, knowledge of climate variability, use of chemicals, use of special seeds, highest grade level HH heads completed, age of HH heads, and family size were the significant (P<0.05) independent variables predicting the response variable impacts of climate change on crop production loss of smallholder farmers at the 5% level of significance. This implies that the nine explanatory variables have additional predictive power for predicting the response variable impacts of climate change on the crop production loss of smallholder farmers. Accordingly, farmer's AEZs, use of chemicals, farmer's knowledge of climate variability, education, and sex of the HH heads were found to be negatively significant (P<0.05) variables that predicted the response variable effects of climate change on crop production loss of smallholder farmers at the 5% level of significance.

For instance, the Ρ–values for the farmers' agroecological type (highland, midland, and lowland) and use of chemicals (insecticide/pesticide) were much lower than 0.05, which implied that there was significant evidence that the null hypothesis was accepted at the 5% level of significance. Thus, for a unit increase in the level of the farmer's agroecological zone, the amount of food crop yield loss (kg ha^− 1) due to climate change decreased by 143.44 when the remaining variables were kept constant. These results implied that, compared with farmers who lived in the midland and lowland AEZs, smallholder farmers in the highland AEZs experienced a decrease in crop yield loss of 143.44 kg ha^− 1. Similarly, farmer holders who used a litter of chemicals (insecticide/pesticide) on their farm for cultivation decreased the amount of crop yield loss due to climate change by 93.62 kg compared with household head farmers who did not use it at all.

Moreover, the P–values (0.00) for the farmer's rating of their knowledge extent about climate variability, education status, and sex of the HH heads were much lower than 0.05 and were found to be significant negative variables that influence the amount of crop production loss due to climate change. For example, HHs who were aware of climate change decreased the amount of crop production lost by 25.80 kg to a greater extent than did farmers who were familiar with climate change or who were not aware of climate change. The analysis also revealed that as the education status of the HH heads increased by one grade, the amount of crop yield loss decreased by 68.18 kg. Compared with female-headed HHs, male-headed HHs likely decreased the amount of crop production loss by 134.71 kg.

However, the age of the HHs, religion, family size of the HHs, and use of improved seed varieties were found to be positive significant factors affecting the crop yield loss of the smallholder farmer. Thus, the study results revealed that the amount of crop production loss of the smallholder farmers increased by 63.81 kg as the age of the HHs increased by one year. The amount of crop production loss increased by 49.78 kg as the household size increased by one person. Similarly, the religious affiliation of HH heads increases the impact of climate variability on crop damage to smallholder farmer communities by 221.59 kg units when the remaining variables are kept constant. Farmholders who used improved seeds increased the impact of climate variability on the livelihoods of smallholder farmer communities by 326.63 kg compared with household head farmers who did not use special seed varieties.

The results from the meteorological analysis revealed that the annual temperatures in the district have significantly increased over the past 30 years. This increase pertained to the current national temperature increase (0.03°C per year) in Ethiopia [36]. This result is in line with previous findings [27],[32],[37]–[39], which revealed a statistically significant increasing trend of temperature in different parts of Ethiopia. Similarly, the outcomes of our study are consistent with those of previous studies [27],[32],[37],[40] that reported a decreasing rainfall trend in different parts of the country. The survey results from all the interviewed farmers indicated that most HHs perceived climate variability in terms of temperature and rainfall. According to research conducted in Ethiopia's Woleka subbasin, 96% of the farmers perceived that the climate was variable, and a study conducted in the Central Rift Valley found that 90% of the farmers surveyed believed that the climate was changing[41][40].

Our findings are consistent with those of Xie et al. (2022)[42], who revealed that the majority, approximately 83%/77% of the respondents, felt a change in temperature or rainfall in the district. Moreover, this study revealed that a greater number of HHs perceived significant temperature changes in the highlands, whereas significant rainfall changes were perceived more in the lowlands.

Similarly, a study conducted in northwestern Ethiopia revealed that farmers in the lowlands (84.8%) perceived rainfall changes more than did respondents in the highlands (60.9%), which also supported this finding[37]. This might be because of their educational status; as more educated HH heads were found in the agroecology of the lowlands, they became keener in knowing about climate change [37],[43].

The correlation test revealed significant socioeconomic factors related to perceptions of climate variability among the different farmers' agroecology types, gender, marital status, and educational attainment, which is in line with the findings of different studies [21],[27],[30],[32],[36]–[38].

Despite the awareness of climate in this study, approximately half of the interviewed HHs replied that rainfall variability has had a medium influence on their crop production. Some site-specific farmers' perceptions of climate variability and impacts on crop production studies in the Bale Zone, such as research conducted in the Sinnana and Agarfa districts by Henok et al. (2016)[15]; in the Sinana, Ginir, Goro, and Gololcha districts by Duressa et al. (2020)[14]; and in the Rayitu, Guradamole, and Dawe Qachen districts by Henok and Dejene (2017)[24], were also identified. This implies that approximately half of the HH heads included in this study replied that rainfall and temperature variability were the key climatic parameter influencing the decline in crop production over the last 10 years (similar to the findings of Henok et al. (2016) [15]. A study conducted by Duressa et al. (2020)[14] in the Bale zone revealed that rainfall variability has a significant negative effect on annual net revenue from cereal crop production. Another study noted that climate variability was the principal factor for the decline in crop production in smallholder agro-pastoral communities[24].

Moreover, the survey results revealed that women and children were the most vulnerable family members to the impacts of climate change. Similarly, a World Bank (WB) report revealed that female-headed households, children, and older HHs were particularly vulnerable [44]. Despite the decline in crop yield as a climate variability outcome, approximately two-fifths of the HH heads did not consider soil nutrient depletion, erosion hazards, or early maturity of crops as potential outcomes of climate change in their locality. In line with this finding, a study conducted in southern Ethiopia revealed that climate change was found to harm 60.5% of households' food security status and crop production[3]. Our findings are consistent with those of Deressa and Hassan (2009)[17], Degefu (2013)[20], and Duressa et al. (2020)[14], who revealed that climate change and variability have a significant negative impact on annual net revenue from cereal crop production.

MLR analysis revealed that farmers' AEZs, chemical use, knowledge of climate variability, education, and sex were found to be negatively significant factors of climate change's impact on crop yield loss in smallholder farmers. In line with these findings, studies have revealed that the effects of climate change on crop production loss are relatively greater at lower altitudes than at higher altitudes[14],[20]. Degefu (2013)[20] also revealed that the probability of household crop production loss decreases as the educational status of farmholders increases by one grade, which supports this finding.

The variation in this study might be due to the differences in the various socioeconomic aspects of the surveyed smallholder farmers, as shown in Table 1. This impact has been widely studied and reported by Degefu (2013)[20], WB (2015)[44], Wossen (2016)[45], Duressa et al. (2020)[14], and Ayal et al. (2021)[32] in Ethiopia, who captured the altitude, gender-specific, and other socioeconomic effects of climate change on agricultural production. Previous studies revealed that farmers in lowland agroecology, female-headed households, and households with low levels of education were the groups most affected by climate change impacts on crop production[14],[32],[45]. For instance, a study showed that educated farmers can perceive local climate conditions and have a greater probability of knowing the impacts on crop yield because of the changing climate than illiterate farmers do[20],[32].

Similarly, a finding revealed that female farmers have heavier workloads and fewer hours to allocate to agricultural production than male-headed households do in most parts of rural Ethiopia, all of which contribute significantly to lower levels of productivity[44]. Another study also revealed that female-headed households participate in various community decision-making activities, own fewer assets, are less educated, have less access to extension services than male-headed households do, and are more affected by the impacts of climate variability due to the endowment effect[45]. We argued that household farmers with male heads, more education, and high access to climate information were less likely to be affected by the impacts of climate change on crop production loss.

Moreover, the study argued that the age of the HHs, family size, religion, and use of improved seeds were found to be positive significant factors affecting the crop yield loss of smallholder farmers. The magnitude of the effects of climate change on crop production loss might be due to the existing differences in the surveyed age groups of the smallholder farmers. As described in the HHs' socioeconomic characteristics, from the agroecological perspective, the majority 49.6% (in the midland) and 36.2% (in the lowland) of the surveyed HHs were 50 years and older, whereas the largest percentage (48.2% in the highland) were 40–49 years of age. This might be due to the failure of farmers to use improved seed varieties as they grow older. This result is supported by Ayal et al. (2021)[32], who found that as the age of farmers increases, they become more reluctant. A previous report (2020)[35] revealed that the time spent on agricultural activities decreases as the age of farmers increases, which affects agricultural production. In addition, Mekonnen et al. (2021)[3], in which the probability of a household being food secure decreases by a factor of 0.33 as the household size increases by one person, support this finding. Agricultural surveys and studies, especially those focusing on smallholder farming, often found that improved seed usage was very rare. Similarly, a study conducted by Wossen (2016)[45] on smallholder farming in Ethiopia also indicated that the use of improved seed varieties decreased in female-headed households, particularly because of climate variability or erratic rainfall. This finding contradicts a study conducted by Duressa et al. (2020)[14], who noted that farmers choosing not to use improved seeds decreased the net revenue per hectare.

The Gassera district's trend study findings verified rising temperatures and falling yearly precipitation. The research area's temperature and rainfall variability were perceived by most farmers across the three agroecology groups, according to the overall results. AEZs, gender, marital status, and educational attainment were found to be significant socioeconomic factors influencing farmers' views of climate variability according to the correlation analysis. The study findings also demonstrated that climate change and variability were the major causes of the crop output drop AEZs. The medium impacts of climatic factors were associated with the effects of rainfall variability by most smallholder farmers, which affected food crop production.

This suggested that farmers had a limited perception of the long-term effects of climatic variability. In terms of activities and livelihoods, approximately half of the communities were still unable to recognize or comprehend the long-term effects of climate change and variability. Increasing awareness of the possible dangers and effects of climate change on smallholder farmers in the Gassera district is one of the challenges associated with putting adaptation methods into practice.

Moreover, wheat, sorghum, maize, and teff crop production showed a linear relationship and a significant correlation with the amount of annual rainfall in the study area. This implies that as the amount of annual rainfall increases, the crop yield also increases. The additional findings on the agricultural impact of climate change that caused the reduction in crop yield were uneven agroecologically, possibly having a more significant effect on the lowlands. The lowland AEZ has experienced the most substantial negative impacts of climate change on crop yield loss and will become increasingly vulnerable over time. Therefore, we conclude that climate change and variability (e.g., reduced rainfall and higher temperatures) negatively impact agricultural output and livelihoods. Therefore, increasing climatic trends and variations have led to crop production in the study area being negatively affected by rising temperatures and precipitation fluctuations. These results hold significance for additional investigations that concentrate on the agroecological surroundings of the adjustment interventions required to ensure the future of smallholder farmers in general and those in lowland AEZs in particular.

The research results also offer valuable perspectives on policy tools to address differences in climate effects among different AEZs and design successful approaches accordingly. MLR models were developed and able to account for much of the variation in explanatory variables, with selected variables of cereal crop production loss among smallholder farmers. The findings showed that the use of improved seeds increases the impact of climate variability on the crop yield loss of smallholder farmers. Because of a lack of weather information, most farmers have failed to use improved seed varieties due to climate variability or erratic rainfall. On the other hand, the use of chemicals, farmers' education level, and extent of knowledge about climate variability were found to decrease the impact of climate change on crop production losses among smallholder farmers. It is essential to increase farmers' knowledge about climate variability and provide weather information to reduce the impact of climate variability on crop production loss. This will also enhance farmers' use of improved seed crop varieties. The agricultural sector in the study area was impacted by climate variability, particularly temperature and rainfall variability, resulting in a negative impact on the livelihood of the community. The primary sources of income for the community under study are maize and sorghum. The fluctuation in productivity has a significant effect on the well-being of the community under study.

Description of the study area.

Location. The study was conducted in the Gassera District of the Bale Zone of the Oromia National Regional States, Ethiopia. According to the socioeconomic profile of the Bale Zone Finance and Economic Development Department (BZFEDD), the zone comprises 18 districts, and Gassera is an administrative district located in the northwestern part of the zone[13]. Geographically, Gassera is situated between 7° 10'–7° 30' N latitude and 39° 50'–40° 25' E longitude and covers approximately 1,114.09 km². As shown in Fig. 3, the district shares a common boundary with Agarfa in the east, Sinana in the north, Ginir in the southwest, Gololecha district in the east, and Arsi district in the south[46]. The elevation of the district ranges from 1050–2573 m above sea level (m.a.s.l.). The district consists of mountains (Gango, Mudo, Abro), flat plains (Asandabo, Shanaka, and Dinkiti), river valleys and courses (Wabe Shabelle River), and deep gorges (Wabe Gorges)[47]. The district is classified into three AEZs: lowland (Kolla, 1050–1500 m.a.s.l.), midland (Woina–Dega, 1500–2300 m.a.s.l.), and highlands (Dega, 2300–2573 m.a.s.l.). Kolla (where Melka Olifigi kebele was found) covers 134.70 Km² (12.09%), Woina Dega (where Homa Abu kebele represented the agroecology) covers 570.30 km² (51.19%), and Balo Amina represents the Dega agroecology, which covers 409.09 Km² (36.72%) of the district [13],[47].

Temperature. The temperatures of the highland, midland, and lowland areas of the district are summarized as warm temperate, temperate, and cool temperate zones, respectively. The mean annual temperature of the district is approximately 20°C, with minimum and maximum temperatures of 10°C and 25°C, respectively. The average annual temperatures in the highlands, midlands, and lowlands are 10–15°C, 15–20°C, and 20–25°C, respectively.

Rainfall. The mean annual rainfall in the Gassera district is approximately 819.25 mm. The district experiences a bimodal rainfall regime known as Belg for the short and Meher-long rainy seasons. Between March and June, the Belg season occurs and spans nearly all areas of the district, except for certain sections along the southeastern margin. The Meher season is between July and November and covers the margin of the southeastern part of the district[13],[36].

Agricultural activities. Agriculture provides a means of occupation and an important economic sector in the district, on which the lives of the majority depend. A farming system is a decision-making unit comprising the farm household, cropping, and livestock systems that transform land, capital, and labor that can be consumed or sold [48]. The district has highly and moderately cultivated land for mixed farming. Sedentary agriculture is predominantly practiced for barley in the highlands and for wheat and maize in the midland areas, whereas animal rearing (a nomadic way of life) with sorghum is practiced in the lowlands and border areas of the district. According to the Gassera District Agriculture and Rural Development Office (GDARDO), crop production is rain-fed and accounts for the largest percentage of its total area for grain production. Cereals and pulses, followed by oilseeds, vegetables, and fruits, are the dominant crops grown in the district. Wheat and barley, followed by oats, maize, and teff, are the dominant cereal crops in the district. The district ranks crop production first in terms of total agricultural production [13],[49].

Research design.

A descriptive cross-sectional survey design was employed in this study, given the purposive selection of the district and Kebeles based on existing variations, which enabled the simultaneous collection of both quantitative and qualitative data at a single point in time. This study used a mixed research approach that involved both quantitative and qualitative methods to gain a comprehensive understanding of the research problem rather than relying on either approach alone [50]. Thus, the study used two major methodologies: the quantitative approach, as a survey, and the qualitative approach, which included focus group discussions and key informant interviews.

Sample size and sampling techniques.

A multistage random sampling technique was used to select an appropriate sample size and study units. First, the Gassera district was purposefully selected because of its high vulnerability, heterogeneous agroecology, and susceptibility to the adverse effects of climate change events. The district has 20,758 households, and the total population was 103,792[49], according to a recent national report, with an average household size of five [34],[51]. In the second stage, a stratified sampling technique was used to divide the 21 Kebeles in the district into three agroclimatic zones: three lowlands (Kolla, 1050–1500), ten midlands (Woina–Dega, 1500–2300), and eight highlands (Dega, 2300–2573) Kebeles, based on altitude and topographic effects[47]. One kebele was randomly selected from each zone: Balo Amina from the highlands, Homa Abu from the midlands, and Melka Olifigi from the lowlands.

In the three selected Kebeles, which are the lowest administrative units, there are approximately 2,596 households (smallholder farmers). Smallholder farmers are small-scale farmers who manage areas varying from less than one hectare to ten hectares that are characterized by a family-focused farm household system[52]. Finally, a simple random sampling technique was used to select the sample households from the three Kebeles, considering the probability proportional to their size. Eq. (1) was used to determine the sample size[50]:

$$\:n=\frac{N}{1+N{\left(e\right)}^{2}}$$

1

In this equation, $\:n$= the required sample size, $\:N\:$ represents the total number of households in the three selected Kebeles (2,596), and $\:e$ represents the degree of accuracy needed or margin of error, which is normally set at the .05 alpha level (.05). According to a document obtained from the Gassera district administration, the total number of households in the respective Kebeles were Balo Amina (1095), Homa Abu (757), and Melka Olifigi (744), with 2,596 households [49]. Thus:

Using Eq. (1) above, by adding 15% (52) for pilot testing and nonreturned questionnaires, 399 sample households were determined. The number of sample HHs from each agroecology/kebele was determined by the proportional population size (PPS) method in Eq. (2) to maintain an equal representation of HHs in each Kebeles. On this basis, the number of HHs from each kebele was calculated as follows:

$$\:\frac{{n}_{1}}{{N}_{1}}=\frac{{n}_{2}}{{N}_{2}}=\frac{{n}_{3}}{{N}_{3}}=\frac{{n}_{4}}{{N}_{4}}=\dots\:=\:\frac{n}{N}\:\:\:$$

2

where N = Total HHs size, n = Total sample size, N₁ = HHs size 1, n₁ = Sample size 1 ·····

Thus, 399 household heads (highland = 169, midland = 116, and lowland = 114) were selected for the household surveys (Table 6).

Table 6

Distribution of sample respondents across the AEZ and *Kebeles*.

*Source*: Author's compilation based on GDARDO (2020).
District	Kebeles in the AEZ	Total HHs	Sample HHs	FGD	KII	Total sample
	Balo Amina within the Highland	1,095	169	10	5	184
Gassera	Homa Abu within the Midland	757	116	10	5	131
	Melka Olifigi within the Lowland	744	114	10	5	129
Total		2,596	399	30	15	444

The study collected 444 samples via three data collection tools: 399, 30, and 15 samples through household surveys, focus group discussions (FGDs), and key informant interviews (KIIs), respectively. Moreover, a purposive sampling technique was used to select KIIs of village elders (two from each Kebeles), development agents (two from each Kebeles), two experts, and the head of the agricultural office participants and to select FGD participants (local elders, females, and men farmers) in a village, considering the length of their residence and lifelong experience.

Data collection.

Household surveys, KIIs, and FGDs were used to collect primary data from a sample of households of smallholder farmers and agricultural office workers for this study using a structured questionnaire. Before data collection, the study used pilot testing to enhance consistency, validity, and reliability. A pilot survey of 52 selected sample households was conducted, and the enumerators checked the questionnaire prepared for collecting data before conducting the final household survey. The questionnaire was validated for consistency, accuracy, and adequacy. Adequate enumerators (three development agents in sample Kebeles) were selected and trained on important procedures and precision in data collection.

The household survey questionnaire generated household-level data on sociodemographic characteristics. The second section captures the farmer's awareness level of any changes in temperature and rainfall (early rain, late rain, and erratic rainfall) in their locality. This section also provides the potential threats of the principal climatic parameters, i.e., temperature and rainfall, to the crop production of smallholder farmers. The third section investigated the most vulnerable human elements and perceptions of the outcome of climate change in the last ten years. The last section barley, wheat, maize, teff, and sorghum predicted responses to the effects of climate change on food crop production and the estimated amount of yield loss (kg ha^− 1) in response to the impact of climate variability.

Moreover, the study used information on the area harvested, annual food crop production (2009–2016), and time series climate data (1990–2019) from secondary data sources. The time series rainfall and temperature data of the study district over the past 30 years (1990–2019) were obtained from the Gassera Meteorological Station (GMS). Eight years of major food crop (barley, wheat, maize, sorghum, and teff) production/yield data from 2009–2016 were obtained from the Gassera District Agriculture and Rural Development Office (GDARDO) and Bale Zone Finance and Economic Development Department (BZFEDD) and were crosschecked with regional and national data.

Methods and models for data analysis.

Descriptive statistics, correlations, and MLR were used to analyze the perceptions of smallholder farmers toward the impacts of climate change and to examine the impacts of climate change on barley, wheat, maize, teff, and sorghum crop production in the three AEZs. Descriptive statistics and key variables are presented in tables, graphs, and percentages. The Statistical Package for Social Sciences (SPSS version 27.0) was used to analyze the collected data. Pearson correlation analyses between socioeconomic and climate change variables were used to examine the important socioeconomic factors that influence farmers' perceptions of climate variability (temperature and rainfall). The time series rainfall and temperature data of the study district over the past 30 years (1990–2019) were obtained from the Gassera Meteorological Station (GMS) to show the trends and changes over time. Annual temperature and rainfall trends were analyzed using simple linear regression. In the simple linear regression model, the dependent variable (the change in temporal trends of climate variability) was considered in the regression equation at a 5% level of significance to test the independent climate (annual rainfall and temperature) variables of trends in the data.

Based on a review of previous scholars that address the impacts of climate change on agricultural crop production [3],[4],[20]–[22],[53],[54], this study used a regression model. Hence, correlation and linear regression models were used to analyze the relationships and effects of climate change on the production of major food crops (barley, wheat, maize, teff, and sorghum) in the district, among others.

Moreover, the MLR was used to analyze the impact of climate change on farmers' yield loss of food crops in kg per hectare (kg ha^− 1) and was regressed against the selected eleven nonclimate predictors (socioeconomic control variables), assuming other socioeconomic predictors were constant. The R-squared value of the model has been used as an indicator of how much of the major food crop (barley, wheat, maize, teff, and sorghum) yield loss (kg ha^− 1) is explained by the socioeconomic control variables. Using the MLR, 11 independent variables were considered in Eq. (3) to be included in the regression equation at a 5% level of significance for predicting the response variable of the impact of climate variability on the food crop yield loss (kg ha^− 1) of the smallholder farmers. The equation of the regression model was calculated as follows[55]:

$$\:{y}_{i}=\beta\:ₒ+{\beta\:}_{1}{X}_{1}+{\beta\:}_{2}{X}_{2}+{\beta\:}_{3}{X}_{3}+{\beta\:}_{4}{X}_{4}+{\beta\:}_{5}{X}_{5}+{\beta\:}_{6}{X}_{6}+{\beta\:}_{7}{X}_{7}+{\beta\:}_{8}{X}_{8}+{\beta\:}_{9}{X}_{9}+{\beta\:}_{10}{X}_{10}+{\beta\:}_{11}{X}_{11}+\epsilon\:$$

3

where $\:{y}_{i}$ is the response/dependent variable and represents the value of the number of farmers per hectare per year yield loss of the food crop when each explanatory/independent variable$\:\:\left(\mathcal{X}\right)$ was zero. The parameters $\:{\mathcal{X}}_{1,\:\:}\:{\mathcal{X}}_{2,\:\:}{\mathcal{X}}_{3,\:\:}{\mathcal{X}}_{4,\:\:}{\mathcal{X}}_{5,\:\:}{\mathcal{X}}_{6,\:\:}{\mathcal{X}}_{7,\:\:}{\mathcal{X}}_{8,\:\:}{\mathcal{X}}_{9,\:\:}{\mathcal{X}}_{10,\:\:}\:and\:{\mathcal{X}}_{11}\:$represented the independent variables. These include farmers' agroecological type (highland, midland, and lowland), sex of the HH heads, religious affiliation, knowledge of HH heads about climate change, access to climate change information, use of chemicals (such as herbicides and insecticides), use of improved seed varieties of crops, marital status of HH heads, the highest grade level HH heads completed, age of HH heads, and family size. 𝛃₁…𝛃₁₁ were their respective partial slopes. Moreover, the scale of each independent variable that was considered continuous and dummy for the MLR model analysis is described in Table 5. The other parameter $\:\varvec{\beta\:}\varvec{ₒ}\:\:$in the multiple regression equation refers to the intercept of the model. Each $\:{\beta\:}_{j}\ne\:0$ represents the expected change in response $\:\left(Y\right)$ for a unit increase in $\:{{\rm\:X}}_{j}$ an explanatory variable, while all other explanatory variables are held constant. The study showed that the model was valid based on the Ρ-value and coefficient of determination $\:\left({R}^{2}\right)$, and explanatory variables with Ρ < 0.05 were deemed statistically significant.

The model explanatory power, measured by $\:{R}^{2}$, validated the MLR model. The R-squared value of the model was used to measure the extent to which the explanatory variables explained the farmer’s yield loss of food crops (kg ha^− 1)[56].

AEZs: agroecological zones; BZFEDD: Bale zone finance and economic development department; EGSA: Ethiopian Geospatial Agency; FGD: Focus group discussions; GMS: Gassera meteorological station; GDARDO: Gassera district agriculture and rural development office; HH: household; KIIs: key informant interviews; kg ha^-1: kilogram per hectare; m.a.s.l.: meters above sea level; MLR: Multiple linear regression; Qt ha^-1: Quintal per hectare.

The authors declare that no funds, grants, or other support were received during the preparation of this manuscript. The authors have no relevant financial or nonfinancial interests to disclose that are directly or indirectly related to the work submitted for publication.

Funding

This Ph.D. thesis work was supported by Hawassa University.

Competing interests

The authors declare that they have no conflicts of interest.

Author contributions

All the authors contributed to the conception and design of the Ph.D. study. Research design, material preparation, data collection, and database construction were performed by the principal author H.W. H.W. who conducted the analysis and wrote the first drafts of the paper. T.B. is a major advisor who supervised, revised, and edited the entire work; W.W. is a coadvisor who revised, commented, and edited the work. All authors revised and edited previous versions of the manuscript. The authors read and approved the final manuscript.

Ethics approval

The purpose of the study was clearly explained to household respondents and agricultural experts. The author declared that all the principles of anonymity and confidentiality, informed consent and voluntary participation, no harm to participants, and no invasion of privacy were implemented. Ethical approval for this study complies with national and Hawassa University postgraduate research guidelines and legislation. The ethical approval was also granted by the administration and advancement of the college research ethics sector and ethical committee.

Consent to participants

The data were collected from household respondents, local administrators, and district agricultural experts with their consent. Data collected from the BZFEDD and the GDARDO were requested formally through the official letter written by Hawassa University, College of Forestry and Natural Resources. The author declared that part of the Ph.D. thesis manuscript was original research work and acknowledged all sources of materials dearly.

Data availability

"The datasets for this Ph.D. thesis part generated during and/or analyzed during the current manuscript preparation are not publicly available owing to [the information used for this manuscript was collected from expertise/household farmers' after a cleared explanation of the study purpose and based on their consent. The author also assured the participants that the information they provided would be confidential and used purely for academic purposes but is available from the corresponding author upon reasonable request.].”

Acknowledgments

We would like to sincerely thank Hawassa University for its financial support in carrying out this Ph.D. study. We are also thankful to the BZFEDD and the GDARDO for facilitating and providing support in the data collection. We thank Mr. Behailu Legesse for his support in developing maps of the study area and honor the contribution of people at the research site for their responses and support during the data collection.

Castro, P., Azul, M., Leal, W., Ulisses, F. & Azeiteiro Editors, M. Climate Change Management Climate Change-Resilient Agriculture and Agroforestry Ecosystem Services and Sustainability. Ecosystem Services and Sustainability (Springer Nature Switzerland, 2019).
Alvi, S., Jamil, F., Roson, R. & Sartori, M. Do farmers adapt to climate change? A macro perspective. Agric. 10, 1–12 (2020).
Mekonnen, A., Tessema, A., Ganewo, Z. & Haile, A. Climate change impacts on household food security and adaptation strategies in southern Ethiopia. Food Energy Secur. 10, 1–14 (2021).
Chukwuone, N. Analysis of the Impact of Climate Change on Growth and Yield of Yam and Cassava and Adaptation Strategies by Farmers in Southern Nigeria. (2015).
USAID. Vulnerability, Impacts, and Adaptation Assessment in the East Africa Region. in Vulnerability, Impacts and Adaptation Assessment in the East Africa Region. (eds. Galu, G., Omondi, P., Okumu, M. & Odingo, P.) 1–92 (Camco Advisory Services (K) Ltd, 2017).
IPCC. Climate Change: Impacts, Adaptation, and Vulnerability. Contribution of Working Group II to the Sixth Assessment Report of the Intergovernmental Panel on Climate Change. (2022).
Pereira, L. Climate Change Impacts on Agriculture across Africa - Oxford Research Encyclopedia of Environmental Science. 10.1093/acrefore/9780199389414.013.292 (Oxford University Press, USA., 2017).
Bryan, E. et al. Adapting agriculture to climate change in Kenya: Household strategies and determinants. J. Environ. Manage. 114, 26–35 (2013).
Zegeye, H. Climate Change Mitigation and Adaptation. J. Biodivers. Endanger. Species s2, 18–35 (2018).
Irish Aid. Ethiopia Country Climate Risk Assessment Report. Irish Aid (2018).
FDRE. Federal Democratic Republic of Ethiopia (FDRE). Ethiopia Climate Resilient Green Economy: National Adaptation Plan (NAP), Implementation Roadmap. (2020).
Suryabhagavan, K. V. GIS-based climate variability and drought characterization in Ethiopia over three decades. Weather Clim. Extrem. 15, 11–23 (2017).
BZFEDD. Bale Zone Finance and Economic Development Office (BZFEDD). Physical and Socio-Economic Profile of Bale Zone. (English Version) Unpublished report. vol. 2 (2014).
Duressa, B., Datiko, D. & Kanaa, Y. The Economic Impact of Climate Change on Agricultural Crops in Bale Zone, South Eastern Ethiopia. International Journal of Scientific & Engineering Research vol. 11 (Ph.D. Thesis. IJSER, 2020).
Henok, W., Derso, D., Bedeso, G. & Erasu, D. Analyzing farmer’s awareness of climate change and trends of climate change in Bale Zone of Oromia National Regional State, Ethiopia. African J. Agric. Sci. Technol. 4, 818–823 (2016).
Legese, W., Koricha, D. & Ture, K. Perception of Farmers on Climate Change and their Adaptive Strategies over Bale Highlands, Southeastern Ethiopia. J. Earth Sci. Clim. Change 09, 1–6 (2018).
Deressa, T. T. & Hassan, R. M. Economic impact of climate change on crop production in Ethiopia: Evidence from cross-section measures. J. Afr. Econ. 18, 529–554 (2009).
Kassie, B. T. Climate variability and change in Ethiopia: Exploring impacts and adaptation options for cereal production. (Ph.D. Thesis. Wageningen University, 2014).
Tessema, I. & Simane, B. Vulnerability analysis of smallholder farmers to climate variability and change: an agro-ecological system-based approach in the Fincha’a sub-basin of the upper Blue Nile Basin of Ethiopia. Ecol. Process. 8, 1–18 (2019).
Degefu, B. Economic Impacts of Climate Change on Teff Production in Lume and Gimbichu Districts of Central Ethiopia. (PhD Thesis. Sokoine University. pp. 1-126., 2013).
Araro, K., Legesse, S. A. & Meshesha, D. T. Climate Change and Variability Impacts on Rural Livelihoods and Adaptation Strategies in Southern Ethiopia. Earth Syst. Environ. 4, 15–26 (2020).
Bewket, W. Rainfall Variability and Agricultural Vulnerability in the Amhara Region, Ethiopia. in Ethiopian Journal of Development Research vol. 29 823–836 (2008).
Deressa, T. T. Measuring The Economic Impact of Climate Change on Ethiopian Agriculture : Ricardian Approach. (2007).
Henok, W. & Dejene, T. Livelihoods vulnerability of climate variability and coping mechanism in South Eastern Ethiopia. Mondopoli 5–8 https://www.mondopoli.it/wp-content/uploads/2019/01/CLIMATE-VARIABILITY-HENOK-YOHANNES.pdf (2017).
WMO. World Meteorological Organization. State of climate service. Agriculture and Food Security. WMO-No. 1242 https://library.wmo.int/idurl/4/56884 (2019).
ATPS. Farmers’ perception and adaptive capacity to climate change and variability in the Upper Catchment of Blue Nile, Ethiopia. [Bewket Amdu, Azemeraw Ayehu, Andent Deressa], ATPS (African Technology Policy Studies Network). Working Paper No. 77. https://www.africabib.org/htp.php?RID=383156459 (2013).
Ayal, D. Y. & Filho, W. L. Farmers’ perceptions of climate variability and its adverse impacts on crop and livestock production in Ethiopia. J. Arid Environ. 140, 20–28 (2017).
Belay, A. et al. Knowledge of climate change and adaptation by smallholder farmers : evidence from southern Ethiopia. Heliyon 8, 1–15 (2022).
Jha, K. & Gupta, V. Farmer’s perception and factors determining the adaptation decisions to cope with climate change : An evidence from rural India. Environ. Sustain. Indic. 10, 1–15 (2021).
Jaana, S., Kaseva, J. & Peltonensainio, P. Farmer views on climate change a longitudinal study of threats, opportunities, and action. Clim. Change 164, 1–19 (2021).
Deressa, T., Hassan, R. & Ringler, C. Perception of and adaptation to climate change by farmers in the Nile Basin of Ethiopia. J. Agric. Sci. 149, 23–31 (2011).
Ayal, D. Y., Tilahun, K., Ture, K. & Terefe Zeleke, T. Psychological dimensions of climate change: perceptions, collective efficacy, and responses in Berehet District, North Shoa, Ethiopia. Clim. Chang. 165:32, 1–18 (2021).
Gezie, M. Farmer’s response to climate change and variability in Ethiopia: A review. Cogent Food Agric. 5, 1–13 (2019).
CSA. Central Statistics Agency. Ethiopia - Population and housing census of 2007. CSA https://www.statsethiopia.gov.et/census-2007-2/ (2012).
Caroni, P. Actin cytoskeleton regulation through modulation of PI(4,5)P2 rafts. EMBO Journal vol. 20 https://doi.org/10.48529/k739-c548 (2001).
MFAN. Climate Change Profile: Ethiopia. MFAN (Ministry of Foreign Affairs of the Netherlands). www.government.nl/foreign-policy-evaluations (2018).
Marelign, A., Addisu, S., Mekuriaw, A., Dar, B. & Dar, B. Observed and Perceived Climate Change and Variability and Small Holder Farmers’ Vulnerability: The Case of Janamora District, Northwestern Ethiopia. J. Environ. Earth Sci. 9, 33–44 (2019).
Addis, Y. & Abirdew, S. Smallholder farmers’ perception of climate change and adaptation strategy choices in Central Ethiopia. Int. J. Clim. Chang. Strateg. Manag. 1–20 (2020).
Getachew, B. Trend analysis of temperature and rainfall in south Gonder zone, Ethiopia. J. Degrad. Min. Lands Manag. 5, 1111–1125 (2018).
Belay, A., Recha, J. W., Woldeamanuel, T. & Morton, J. F. Smallholder farmers’ adaptation to climate change and determinants of their adaptation decisions in the Central Rift Valley of Ethiopia. Agric. Food Secur. 6, 1–13 (2017).
Asfaw, A., Simane, B., Bantider, A. & Hassen, A. Determinants in the adoption of climate change adaptation strategies: evidence from rainfed-dependent smallholder farmers in north-central Ethiopia (Woleka sub-basin). Environ. Dev. Sustain. 21, 2535–2565 (2019).
Xie, S., Ding, W., Ye, W. & Deng, Z. Agro-pastoralists’ perception of climate change and adaptation in the Qilian Mountains of northwest China. Sci. Rep. 12, 1–15 (2022).
Saguye, T. S. Assessment of Farmers’ Perception of Climate Change and Variability and its Implication for Implementation of Climate-Smart Agricultural Practices: The Case of Geze Gofa District, Southern Ethiopia Geography & Natural Disasters. J. Geogr. Nat. Disaster 7, 1–9 (2017).
WB. World Bank. Ethiopia Poverty Assessment. Ethiopia Poverty Assessment. Report No. AUS6744. www.worldbank.org/ethiopia (2015).
Wossen, T. Environment for Development Gender-Differentiated Impacts of Climate Variability in Ethiopia. (2016).
EGSA. Ethiopian Geospatial Agency. Location of Gassera district, Bale Zone, Ethiopia. Extracted from EGSA, 2019. at (2019).
EGSA. Ethiopian Geospatial Agency. Kebeles of the Gassera district across the agroecological zone, Bale Zone, Ethiopia. Extracted from EGSA, 2019. at (2019).
FAO. Food and Agriculture Organization. Eastern Africa Climate-Smart Agriculture Scoping Study: Ethiopia, Kenya, and Uganda. vol. 5485 www.fao.org/publications (2016).
GDARDO. Gassera District Agriculture and Rural Development Office. Total population of Gassera district Peasant Association (PA) across sex. Bale Zone, Gassera, Ethiopia. Unpublished report. (2020).
Adam, A. Sample Size Determination in Survey Research. J. Sci. Res. Reports 26, 90–97 (2020).
ESS. Ethiopia Socioeconomic Survey ( ESS ). CSA and WB vol. 4 (2018).
FAO. Food and Agriculture Organization. Smallholders and Family Farmers. Smallholders and Family Farmers Sustainability Pathways. FAO 1–4 www.fao.org/nr/water/docs/Enduring_Farms.pdf (2012).
Byishimo, P. Assessment of Climate Change Impacts on Crop Yields and Farmers’ Adaptation Measures: A Case of Rwanda. (Ph.D Thesis. University of Zimbabwe., 2017).
Weldearegay, K. & Tedla, G. Impact of climate variability on household food availability in Tigray, Ethiopia. Agric. Food Secur. 7, 9 (2018).
Bhattacherjee, A. Social Science Research: Principles, Methods, and Practices. Textbooks Collection.3. http://scholarcommons.usf.edu/oa_textbooks/3 (USF Tampa Library Open, 2012).
Leng, G. & Huang, M. Crop yield response to climate change varies with crop spatial distribution patterns. Sci. Rep. 7, 1–10 (2017).

The authors declare no competing interests.

Perception of Smallholder Farmers about Climate Change and Its Impacts on Crop Production across Agroecological Zones of the Gassera District, Southeastern Ethiopia.

Status:

Version 1

Abstract

Figures

Introduction

Results

Discussion

Conclusion

Materials and methods

Abbreviations

Statement and declaration

References

Additional Declarations

Status:

Version 1