Spatiotemporal trends of climate change and variability: impacts on coffee production in Abaya and Gelana Woredas, Southern Ethiopia

Abstract

The purpose of this study was to investigate the spatiotemporal trends and variability of climate impacts on coffee production in Abaya and Gelana Woredas. To clarify reliable data from the participants, the study utilized a mixed-research approach. Combining quantitative climate analysis (Mann–Kendall test, Sen’s slope, and rainfall indices) with qualitative data from surveys and interviews, this research assessed how climate variability, socioeconomic factors, and physical conditions affect coffee yield. Statistical analysis (regression and t-tests) reveals significant climate trends across the study area, including warming nighttime temperatures (T_min), cooling daytime temperatures (T_max), and seasonal rainfall fluctuations. Rainfall trends varied among kebeles: In Bunata, Belg (Z = 1.07) and Meher (Z = 1.03) conveyed moderate but non-significant increases, although annual rainfall showed a near-significant decline (Z =  − 1.84, Q =  − 0.076). In contrast, Guangawa Badiya, Giwe, and Jirme exhibited positive rainfall trends in both Belg (Z = 2.21) and Meher (Z = 2.67), while Odo Mike experienced negative rainfall trends, particularly in Meher (Q =  − 0.391) and annually (Q =  − 0.660). Temperature trends revealed a decrease in T_max across all sites (Bunata − 0.61, Guangawa Badiya − 0.66, Odo Mike − 0.45, Giwe − 0.43), while T_min increased entirely, with notable seasonal variability in T_max. Regression modeling showed a strong correlation (R = 0.871) between climate variability, soil erosion, land size, and coffee production, explaining 83.2% of the variation in yields. Key adaptation strategies reported by farmers included intercropping (8.7%), income diversification (8.7%), cultivar selection (8.6%), agroforestry (8.5%), and integrated pest management (IPM) (7.8%). While rising T_min, decreasing T_max, and rainfall variability contributed to variations in coffee production in Guangawa Badiya, Giwe, and Jirme, these changes led to a decline in Bunata and Odo Mike. Coffee production has been impacted by climate change due to reducing the diurnal temperature range, hindering blooming and bean development, and making pests more vulnerable. Intense rainfall causes soil erosion and nutrient loss, while irregular rainfall impacts important development phases, resulting in flower drop and low yields. This study underscores the importance of adaptive strategies such as intercropping, agroforestry, income diversification, enhanced water management, and government support in ensuring the sustainability of coffee farming amidst ongoing climate fluctuations.

Introduction

Climate change is an escalating global challenge that has far-reaching impacts on agricultural systems, with notable consequences such as rising temperatures, altered precipitation patterns, and the intensification of pest and disease pressures (IPCC, 2021; Raza et al., 2019). Among the crops most vulnerable to these changes is coffee, particularly Arabica coffee, which thrives within specific temperature and rainfall ranges (IPCC, 2022). Arabica coffee’s sensitivity to climate change makes it especially susceptible to reduced productivity and quality as climate variability intensifies. Rising temperatures beyond 30 °C and erratic rainfall patterns have already been linked to diminishing yields and deteriorating coffee quality (Gedefaw, 2023). Furthermore, the spread of pests like the coffee berry borer and leaf rust exacerbates these challenges, further compromising coffee production (Tavares et al., 2018).

In East Africa, particularly in Ethiopia, irregular rainfall patterns and escalating environmental stress increasingly disrupt agricultural practices, leading to reduced crop yields and increased risk of crop failure (Hobart et al., 2024; Suri, 2023). Ethiopia’s agricultural sector, which heavily relies on rain-fed systems, is particularly vulnerable to these climate changes (Awoke et al., 2022; Benti et al., 2021; Kumar et al., 2021). The country has already experienced an average temperature rise of 1.3 °C since 1960, with projections indicating a further increase of 1.1 to 5.1 °C by the century’s end (Girma, 2023; McSweeney et al., 2010; Worku, 2023). While some studies report a decline in rainfall during the Belg season, others show increased rainfall during the Kiremt season, pointing to regional inconsistencies in climate data and complicating the understanding of local climatic patterns (Asfaw et al., 2018; Gummadi et al., 2018).

While many researches have been conducted on the broader effects of climate change on coffee production in Ethiopia, significant gaps remain. Most studies focused on biophysical impacts, adaptation strategies, land use, and coffee quality (Ayalew et al., 2022; Hettig et al., 2016; Hussein et al., 2023), with little emphasis on the intricate interplay between local climate trends and coffee variability. Moreover, empirical analyses of spatiotemporal impacts on coffee production are notably scarce, particularly in key coffee growing areas like Abaya and Gelana Woredas (Lalicha, 2022; Malek & Verburg, 2020). The complex relationship between microclimatic fluctuations and the dynamics of coffee production, however, is still mainly unknown in both Abaya and Gelana Woredas. This disparity makes it more difficult to create climate-smart, locally relevant adaptation plans that are essential for maintaining coffee livelihoods in the face of increasing climatic unpredictability. The importance of addressing these gaps is underscored by the fact that climate impacts are often geographically variable, and understanding micro-scale climate trends is crucial for assessing their effects on local agriculture (Anteneh, 2022; Dibaba et al., 2020; Mekonnen et al., 2018).

This study aimed to fill these gaps by investigating the spatiotemporal trends of climate change and variability and their impacts on coffee production in Abaya and Gelana Woredas. Additionally, the study aimed to identify the adaptation strategies employed by farmers to cope with climate-induced challenges. Specifically, the objectives of this research were to (i) assess the spatiotemporal trends of climate change and variability from 1992 to 2022, (ii) analyze the impacts of climate change and variability on coffee production, and (iii) describe the adaptation strategies of farmers in response to the changing climate in the study area.

Materials and methods

Study area description

The study was conducted in Abaya and Gelana Woredas, located in the West Guji Zone of the Oromia Regional State in Southern Ethiopia. Abaya Woreda is situated approximately 365 km south of Addis Ababa, geographically positioned between latitudes 6°10′ N and 6°20′ N and longitudes 38°00′ E and 38°10′ E (Fig. 1). The Woreda is bordered by Lake Abaya to the west, Gelana Woreda to the south, and the Gedio Zone to the north and east (Gratzer et al., 2021). Gelana Woreda, located about 470 km south of Addis Ababa, is geographically positioned between latitudes 5°40′ N and 6°20′ N and longitudes 37°50′ E and 38°10′ E (Fig. 1). The Woreda is bordered by Gedio Zone to the north and northwest, Abaya Woreda to the northwest, Lake Abaya to the west, Gamogofa Zone to the southwest, Amaro Woreda to the south, and Bule Hora and Burji Woredas to the southeast (Tesfaye, 2018).

Fig. 1.

Map of the study area, showing Abaya and Gelana Woredas in the West Guji Zone, Oromia Region, Ethiopia. The map highlights the geographical boundaries of the two Woredas and their surrounding areas, illustrating the spatial context for analyzing spatiotemporal trends of climate change. The coordinates of each Woreda and key neighboring regions are indicated for reference

Agro-climatic conditions

Abaya Woreda is characterized by a yearly temperature between 25.1 and 30.6 °C, with annual rainfall between 470 and 1828.7 mm, through the central, southeastern, and southern areas receiving the most and the northeastern areas receiving the least. Agro-climatologically, the Woreda consists of 60% lowlands and 40% mid-highlands (Gratzer et al., 2021). There are approximately 103,348 people living in Abaya Woreda, with 3.42% of them residing in urban areas and the rest in rural areas (Csa, 2018). The land use in Abaya Woreda consists of 35% grassland, 15% forest land, and 41% arable land. Agricultural activities covered over 5000 hectares, where key crops like coffee, maize, and haricot beans are cultivated. The local economy is primarily based on subsistence supported by livestock and traditional farming (Abdisa et al.,2023).

Gelana Woreda, in contrast, is primarily lowland (70%) with temperatures ranging from 17 to 23 °C, and Gelana Woreda gets 1396 to 1710 mm of rain annually, with the southeast, southwest, and certain central areas getting somewhat more rainfall than the northwest, northeast, and portions of the central area. The economy of Gelana Woreda is also largely agricultural, relying on two erratic rainfall seasons that support crop and livestock production in a semi-arid landscape (Tesfaye, 2018). The population of Gelana Woreda grew from 71,369 to 102,141 in 2022, reflecting a 2.4% yearly growth rate (Csa, 2018). The agricultural land in Gelana spans 135,543 km², where both livestock and crops such as wheat, maize, and coffee are cultivated (Davis, 2018). The local farming system relies heavily on traditional irrigation to support staple crops such as Enset and pulses.

Both Abaya and Gelana Woredas provide favorable agro-ecological conditions for traditional agroforestry coffee farming. However, coffee production in this area faces significant challenges, including competition from khat cultivation, the impacts of climate change, poor agronomic practices, and declining coffee prices, all of which threaten the long-term sustainability of coffee farming (Lalicha, 2022).

Research method and design

This study employed a mixed-approach to assess the impacts of climate change on coffee production. A sequential explanatory design was used, which began with quantitative data collection and analysis, followed by qualitative insights to deepen the understanding of the findings (Creswell & Creswell, 2005). The quantitative analysis focused on climate data, socioeconomic factors, and adaptation strategies, while qualitative data were gathered through interviews to explore farmer experiences and perceptions of climate impacts.

Data sources and methods

The study employed a three-phase sampling technique. First, Abaya and Gelana Woredas were purposively selected due to their significant contributions to coffee production. Second, depending on their potential to produce coffee, five kebeles were selected at random: three from Abaya (Odo Mike, Guangawa Badiya, and Bunata) and two from Gelana (Giwe and Jirme). The sample size from the total household 7553 (4699 from Abaya and 2854 from Gelana Woredas) for the study was then calculated using Taro (1967) formula from both Woredas.

$$n=\frac{N}{1+N{(\mathrm{e})}^2}$$ 1

$$=\frac{7553}{1+7553(0.05)2}=\frac{7553}{1+7553(0.0025)}=\frac{7553}{1+18.883}=n=398$$

where n is the sample size, e is the precision level, and N is the entire population, the 95% confidence level. The error (0.05) represents the precision level.

The sampling frame for selected kebeles was as follows: Jirme had 63 (15.8%), Oddo Mike had 97 (24.4%), Guangawa Badiya had 84 (21.1%), Bunata had 68 (17.1%), and Giwe had 87 (21.9%) households. The study used primary data from 398 household heads (249 from Abaya and 150 from Gelana Woredas), observations, and interviews. Secondary data on climate trends (1992–2022) were obtained from the USGS and Ethiopian Meteorological Agency. The study also engaged stakeholders, including smallholder farmers, cooperative leaders, exporters, and experts, and analyzed secondary sources such as government and NGO reports to examine trends at the Woreda levels.

Data analysis

Survey data analysis

Data on climate change, variability, and household adaptation strategies were analyzed using SPSS (v26). Qualitative data were analyzed thematically to summarize farmers’ perceptions, while quantitative data were analyzed using multiple linear regression models to evaluate the relationship between climate variability and coffee production in the study areas. The regression model is specified as follows:

$$Y={\beta }_0+{\beta }_1{X}_1+{\beta }_2{X}_2+\cdots +{\beta }_n{X}_n$$ 2

where Y represents coffee production (in tons), and β are the regression coefficients for temperature, rainfall, and soil characteristics, respectively.

Spatial data analysis: spatiotemporal trends of climate change and variability

Spatiotemporal trends in temperature and rainfall were analyzed from 1992 to 2022 using grid data for minimum and maximum temperatures, rainfall, and farmer perceptions. The analysis followed the WMO guidelines for trend detection, using data from the USGS and the Ethiopian Meteorological Agency (EMA) (Chombo et al., 2020; Terefe et al., 2022). Numerous approaches for examining temperature and rainfall can be divided into two categories: trend analysis methods and variability analysis methods (Shitu et al., 2024). Various methods were used to assess climate trends and variability, including the Mann–Kendall test and Sen’s slope for trend analysis, and the coefficient of variation (CV), Precipitation Concentration Index (PCI), and Standardized Anomalies Index (SAI) for variability analysis.

1. Rainfall variability and trends

This study analyzed temporal variations in temperature and rainfall using the coefficient of variation (CV), Precipitation Concentration Index (PCI), and Standardized Anomalies Index (SAI) to understand their impact on coffee production.

I) Coefficient of variability (CV): Measures the distribution of temperature or precipitation data over time, indicating variability (Mangrio et al., 2013). Higher CV values indicate greater variability.

$$\text{CV}=\frac{\left(\mu \right)}{\sigma }\times 100\%$$ 3

where μ is the average temperature or precipitation, CV is the coefficient of variation, and σ is the standard deviation.

According to Ali et al. (2022), rainfall variability is classified as low (CV < 20), moderate (20 < CV < 30), or high (CV > 30). In this study, the coefficient of variation was calculated for each month from 1992 to 2022.

II) Precipitation Concentration Index (PCI): Assesses the concentration of rainfall throughout the year, with higher values indicating more concentrated rainfall. The annual PCI is calculated as follows:

$$\mathrm{PClannual}=\frac{{\displaystyle \frac{\sum }{\mathrm{i}=1}}{\mathrm{Pi}}^2}{{\left(\sum ,12,-,1,\mathrm{Pi}\right)}^{2-}}$$ 4

where Pi is the amount of rainfall during the ith month.

A PCI value under 10 indicates uniform rainfall, 11–15 suggests moderate concentration, 16–20 indicates high concentration, and values over 21 signify very high concentration (Lukić et al., 2018).

III) Standardized Rainfall Anomalies Index (SAI): Quantifies the deviation of annual rainfall from the long-term mean, identifying periods of above- or below-normal rainfall (Birara et al., 2018; Mamuye et al., 2024). Negative values indicate below-normal rainfall, while positive values signify above-normal rainfall (Taye et al., 2024) (Table 1), using Eq. (5).

$$\text{Z}=\frac{\left(X,i,-,X^{-}\right)}{S}$$ 5

where, Xi is the annual rainfall of a specific year, Z is the standardized rainfall anomaly, X̅ is the long-term mean annual rainfall over 30 years, and S is the standard deviation of annual rainfall over 30 years.

Table 1.

Standardized Rainfall Anomaly Index (SRAI) value clarification

SAI value	Category
2 +	Extremely wet
1.5 to 1.99	Very wet
1 to 1.49	Moderately wet
0.99 to 0.99	Near normal
1.49 to − 1	Moderately dry
1.99 to − 1.5	Severely dry

2. Statistical tests

According to Frimpong et al. (2022), the non-parametric Mann–Kendall statistic (MK) for trend analysis, as introduced by Mann (1945) and later expanded by Kendall 1975), was used to detect trends in temperature and rainfall, while Sen’s slope estimator provided the slope of the trend line, indicating the rate of change over time.

The analysis utilized the formula presented in Eq. (6).

$$s={\sum }_{i-1}N-i\sum \frac{Nj=i-1sgn\left(x,j,-,x,i\right)}{\mathrm{\sigma }}$$ 6

where S denotes the Mann–Kendall trend test statistic, i and j (j > I) denote the time series length, and x_i and x_j represent the sequential values in the time series, and σ is the standard deviation.

A higher Mann–Kendall statistic indicates a stronger trend, while a p-value above the significance level confirms statistical significance. Sen’s slope estimator calculates the slope of a linear trend in time series data (Jiqin et al., 2023), as described in Eq. (7).

$$Senslope=Median\frac{xi-xj,j<i}{i-j}$$ 7

where x_i and x_j are the variable’s change values at time steps i and j, respectively.

A slope close to zero indicates no change, while a positive slope reflects a strong positive trend, and a negative slope indicates a strong negative trend.

Results and discussions

Spatiotemporal trends of climate change and variability (1992–2022)

Spatiotemporal variability of rainfall

Coefficient of variability (CV)

The analysis of the CV from 1992 to 2022 reveals substantial differences in rainfall variability across the kebeles, with greater seasonal variability observed compared to annual variability (Fig. 2). Bunata kebele showed moderate annual variability (CV: 20–30%) but exhibited irregular high fluctuations in some years (CV > 30%). Meher rainfall in Bunata demonstrated a broad range of variability (CV < 20 to > 30), while Belg variability was relatively lower (CV < 20 to 20–30). Odo Mike and Guangawa Badiya exhibited high annual variability (CV > 30%) with moderate to low Belg variability (CV 20–30 to < 20) and moderate to high Meher variability (CV 20–30 to > 30). Giwe had mostly moderate annual variability (CV 20–30%), with low to high Meher variability and low to moderate Belg variability. Jirme experienced high annual variability (CV > 30%), moderate to low variability in Belg (CV < 20 to 20–30%), and moderate to high variability in Meher (CV 20–30 to > 30) (Fig. 2).

Fig. 2.

Coefficient of variations (CV) in Annual and Seasonal Rainfall Variability (%) across Kebeles a) Annual CV b) Belg season CV c) Meher season CV.

Seasonal rainfall variability (1992–2022) was higher than annual variability for the two study areas (Fig. 2). Belg rainfall exhibited relatively stable, low to moderate variability (CV < 20–30), while the Meher showed moderate to high variability (CV 20–30, > 30). Jirme exhibited the highest annual variability with the highest overall variability. Guangawa Badiya and Odo Mike showed high variability (CV > 30) and moderate variability in Bunata and Giwe (CV 20–30), with occasional extremes in Bunata.

Seasonal variability in rainfall poses challenges for coffee production, highlighting the critical need for consistent rainfall to support stable crop growth (Yona et al., 2024). The observed rainfall patterns align with findings from previous studies in the Ethiopian highlands (Haile, 2018), highlighting the detrimental effects of rainfall variability on agricultural stability.

Precipitation Concentration Index (PCI)

Figure 3 presents the PCI for each kebele. The PCI values reveal notable patterns of rainfall distribution throughout the year. Bunata displayed moderate annual PCI (ranging from 10 to 15), indicating a moderate concentration of rainfall, with some uniform conditions (PCI < 10) and periods of high concentration (16–20). Meher season PCI in Bunata was modest, while the Belg season PCI was mostly below 10 (uniform). Guangawa Badiya showed annual PCI values between 16 and 20, indicating a higher rainfall concentration, with modest Meher PCI and Belg PCI below 10. Odo Mike had annual PCI ranging from 10 to 15, and the Meher season PCI varied from low to moderate, with Belg PCI below 10 (Fig. 3). Giwe showed a mostly moderate annual PCI (10 to 15), but with irregular periods of high concentration (16–20). In the Meher season, PCI was mostly below 10, while the Belg season had occasional intermediate values (10–15) (Fig. 3). Jirme exhibited the highest annual PCI (> 20), with modest Belg PCI and Meher PCI consistently below 10. Overall, Odo Mike, Bunata, and Giwe had PCI between 10 and 15, while Guangawa Badiya had PCI between 16 and 20. These spatial differences in PCI highlight the need for localized climate adaptation strategies that take into account rainfall variability and seasonal distribution across kebeles.

Fig. 3.

Annual and Seasonal Rainfall Variability across Kebeles a) Annual PCI b) Belg season PCI c) Meher season PCI.

Overall, the results indicate that Jirme stands out for its intense rainfall concentration, which may present challenges for rain-fed agriculture and crop planning, while Odo Mike, Bunata, and Giwe may experience moderately concentrated rainfall with occasional uniformity, while Guangawa Badiya experiences higher concentration levels. These variations in PCI emphasize significant spatiotemporal shifts in rainfall distribution, which are critical for understanding agricultural impacts, particularly for crops like coffee that require specific precipitation patterns for optimal growth (Tessema & Simane, 2019).

These findings echo the results of earlier studies, which had identified rainfall unpredictability as a key determinant of agricultural productivity in Southern Ethiopia (Shigute et al., 2023). Furthermore, climate variability remains a key concern for food security and coffee cultivation (Olana et al.,2023); the studies confirmed that increased rainfall intensity is a growing challenge (Urgessa & Bewket, 2011).

Standardized Anomalies Index (SRAI)

Inter-annual rainfall anomalies from 1992 to 2022, measured using the SRAI, reveal variability across the five study sites: Bunata, Odo Mike, Guangawa Badiya, Giwe, and Jirme (Fig. 4). In Bunata, climatic conditions were generally near-normal, characterized by irregular dry spells, a moderately wet Meher season, and a predominantly dry Belg season. Guangawa Badiya exhibited relatively stable rainfall patterns, punctuated by occasional extreme events and moderate seasonal variability in both Belg and Meher periods. Odo Mike similarly demonstrated largely regular rainfall behavior, interspersed with modest extremes and alternating periods of moderate wetness and dryness across both seasons (Fig. 4). In contrast, Giwe experienced persistently dry conditions over much of the study period, with sporadic occurrences of near-normal, extremely wet, and moderately wet years. Seasonal analysis indicated a moderately wet Belg season with intermittent dry spells, whereas the Meher season was slightly drier than average (Fig. 4). Jirme presented the highest degree of rainfall variability among the five sites, encompassing years of extreme dryness, moderate drought, and episodes of very wet conditions. Seasonal anomalies at Jirme were particularly notable, with the Belg season marked by irregular, mild precipitation and dry spells, and the Meher season characterized by moderate, erratic dryness (Fig. 4).

Fig. 4.

Standardized Rainfall Anomaly Index (SRAI) for Bunata, Odo Mike, Guangawa Badiya, Giwe and Jirme a) Annual SRAI b) Belg season SRAI c) Meher season SRAI.

Overall, the research highlights the localized character of climatic variability by revealing significant differences in rainfall over time and space. Odo Mike and Guangawa Badiya had somewhat stable behavior, but Jirme and Giwe in particular displayed a great deal of fluctuation and extremes. These trends show how site-specific climate resilience plans are necessary, especially in regions that are susceptible to drought and threats to food security and agriculture. These observed patterns are consistent with recent findings by Geleta et al. (2024) and Sinore and Wang (2024), who report increasing rainfall intensity and variability across the Ethiopian highlands. Such shifts in precipitation regimes have significant implications for agricultural productivity and underscore the urgency of adaptive management strategies in rain-fed farming systems.

The analysis of annual rainfall anomalies from 1992 to 2022 reveals substantial inter-annual variability across the study sites. These findings underscore the significant role of rainfall variability in influencing agricultural productivity in the region. Our results are consistent with the conclusions of Habte et al. (2023), who highlighted the critical impact of rainfall fluctuations on agricultural outcomes. Additionally, the observed variability aligns with the findings of our findings is in line with the findings of Ademe et al. (2020), who emphasized the unpredictable nature of rainfall patterns and their direct implications for farming practices, particularly in areas dependent on rain-fed agriculture.

Trend of annual and seasonal rainfall distribution

Farmers’ perception of temperature and precipitation change

Farmers’ perceptions of climate change and variability were assessed across 398 participants (Table 2). A notable portion of respondents indicated awareness of shifts in climatic patterns. Specifically, 31.4% (125 farmers) reported an increase in temperature, while 36.2% (144 farmers) observed changes in precipitation patterns. Additionally, 16.6% (66 farmers) recognized an increased frequency of droughts and floods, highlighting the intensifying climate extremes in the region. A further 9.3% (37 farmers) noted an increase in precipitation volume, and 3.8% (15 farmers) expressed uncertainty about climate changes. Interestingly, only 2.8% (11 farmers) reported no perceived changes. These findings corroborate with the results of studies conducted in other regions, such as the Dabus watershed in Ethiopia, where farmers similarly recognized rising temperatures, alongside fluctuations in rainfall patterns (Awoke et al., 2022). Furthermore, literature highlights how farmers frequently perceive delayed rains, early cessation, and variations in rainfall intensity, all of which critically affect agricultural productivity (Benti et al., 2021).

Table 2.

Farmers’ perception of temperature and precipitation changes in study area kebeles

Farmers’ perception of climate change and variability	Frequency	Percent	Cumulative percent
No change	11	2.8	2.8
Uncertain	15	3.8	23.1
Increased frequency in drought and flood	66	16.6	19.3
Increased volume of precipitation	37	9.3	32.4
Increased temperature	125	31.4	63.8
Altered precipitations patterns	144	36.2	100.0
Total	398	100.0

Rainfall fluctuations in Odo Mike, Jirme, Guangawa Badiya, Bunata, and Giwe have significantly impacted coffee production, which relies on consistent rainfall (Fig. 5). Notably, in Bunata kebele, rainfall was characterized by significant fluctuations between 1992 and 2022, with peaks in 2020 (418.36 mm during Belg), 1996, 2005, and 2008. However, 2022 saw a decline in Meher rainfall despite high Belg rainfall. In Guangawa Badiya, extreme rainfall variation was observed, with a severe drought in 2000 and a peak in 2018, while Meher rainfall exhibited notable fluctuation. Odo Mike showed substantial fluctuations, including high rainfall in 1993, 2000, and 2021, followed by severe declines in 2011 and 2022. Giwe kebele experienced considerable variability, with Meher rainfall ranging from 65.31 mm in 2000 to 191.60 mm in 2006, to a peak of 286.52 mm in 2016, followed by a decrease to 102.63 mm in 2019. Similarly, Jirme kebele experienced fluctuations in Meher rainfall, with values increasing from 40.87 mm in 2000 to 214.45 mm in 2020, fluctuating between 29.88 mm in 2003 and 223.24 mm in 2020, while Belg rainfall varied from 24.61 mm in 2004 to 286.52 mm in 2013 (Fig. 5).

Fig. 5.

Trend of annual and seasonal rainfall in the kebeles (1992–2022)

These significant fluctuations in rainfall patterns highlight the vulnerability of rain-fed agricultural systems, particularly coffee production, which is highly sensitive to climatic shifts (Bedane et al., 2022). Without adaptive strategies such as altering planting dates, enhancing water management, and utilizing drought-tolerant cultivars, coffee farming remains at risk. In the absence of such adaptive measures, farmers’ livelihoods and food security are jeopardized (Tessema & Simane, 2019).

In Abaya and Gelana Woredas, analysis using the Mann–Kendall test and Sen’s slope method reveals mixed trends in rainfall trends. Most kebeles showed minimal change, although some, particularly during Belg and Meher, showed significant increases in seasonal rainfall (Table 3). The MK trend for Bunata conveys a modest increase during Belg (Z = 1.07) and Meher (Z = 1.03), but these trends are not statistically significant. However, the Sen’s slope for the Belg season (Q = 4.269) suggested a slight increase, whereas annual rainfall showed a significant decline (Z =  − 1.84, Sen’s slope =  − 0.076), indicating a potential long-term reduction that could adversely affect agriculture, especially coffee production. In Guangawa Badiya, the trend analysis shows a statistically not significant but positive trend (Z = 0.89, Sen’s slope = 1.652), with considerable increases in both the Belg (Z = 2.21) and Meher (Z = 2.67) seasons, suggesting improved crop productivity, including coffee. Conversely, Odo Mike displayed no significant rainfall trends, with Belg (Z = 0.36), Meher (Z =  − 0.41), and yearly rainfall (Z =  − 1.78) showing limited change. Negative Sen’s slopes for Meher (Q =  − 0.391) and annual rainfall (Q =  − 0.660) suggest considerable rainfall variation, which has implications for crop stability, particularly coffee (Table 3).

Table 3.

Mann–Kendall trend and Sen’s slope analysis of rainfall in study area selected kebeles (1992–2022)

Study sites	Mann–Kendall trend (Z test)			Sen’s slope (Q)
	Belg	Meher	Annual	Belg	Meher	Annual
Bunata	1.07	1.03	− 1.84	4.269	0.344	− 0.076
Guangawa B.	2.21	2.67	0.89	2.856	2.161	1.652
Odo Mike	0.36	− 0.41	− 1.78	0.462	− 0.391	− 0.660
Giwe	2.21	2.67	0.72	2.856	2.161	1.526
Jirme	2.21	2.67	0.85	2.856	2.161	1.667

In Giwe, the Meher (2.67) and Belg (2.21) seasons’ Z-scores are statistically significant, supporting trends in rising seasonal rainfall. Despite the lack of statistical significance in the yearly Z-score (0.72), the positive Sen’s slope (Q = 1.526) suggests a steady rise in annual rainfall. A beneficial tendency for producing coffee and other commodities was the increased availability of precipitation during the primary growing seasons (Table 3). Comparably, in Jirme, both the Belg (Z = 2.21) and Meher (Z = 2.67) seasons show statistically significant positive trends. Alongside the somewhat positive Sen’s slope (Q = 1.667) and the yearly Z-score (0.85), which is not statistically significant, there appears to be an upward trend in annual rainfall (Table 3). These results imply that Giwe and Jirme have advantageous seasonal agricultural situations, but the yearly patterns are not significant enough to support long-term shifts. While Bunata and Odo Mike did not exhibit any noteworthy patterns, Guangawa Badiya, Giwe, and Jirme were seen with statistically significant seasonal rainfall increases during Belg and Meher. However, Bunata’s negative yearly trend and Odo Mike’s diminishing trends might be early indicators of decreased rainfall. In Guangawa Badiya, Giwe, and Jirme, positive Sen’s slopes indicate a slow improvement in long-term rainfall conditions, despite largely insignificant yearly changes. According to Aboye et al. (2023), seasonal rainfall patterns significantly influence agricultural productivity, particularly for water-sensitive crops such as coffee. Previous research, Gezie (2019), has also highlighted the importance of flexible and adaptive agricultural practices to mitigate the adverse effects of such climatic changes on crop production.

Spatiotemporal variability of annual and seasonal temperature

The spatial patterns of annual, maximum, and minimum temperature

The spatial and temporal dynamics of annual, maximum (T_max), and minimum (T_min) temperatures across the study sites reveal notable heterogeneity with implications for agricultural productivity. In Bunata, average annual temperatures ranged from 27.79 °C in 2020 to a peak of 32.02 °C in 2003. This fluctuation may impact crop production, particularly for sensitive crops like beans, which can suffer under higher temperatures. Seasonal analysis indicates that Meher season T_max reached a maximum at 31.33 °C (2002), likely contributing to thermal crop stress, while Belg season T_max peaked at 34.34 °C in 2003, likely creating heat stress for crops, thus impeding early vegetative growth. T_min values in Bunata ranged from 12.40 °C in 2021 to 13.94 °C in 2002, with relatively warmer nighttime temperatures having been associated with favorable conditions for perennial crops such as coffee. During the Belg season, T_min varied between 13.24 °C (1993) and 16.44 °C (2019), and Meher T_min ranged from 11.68 °C (1992) to 13.90 °C (2014) (Fig. 6).

Fig. 6.

Mean annual maximum, minimum temperature, and seasonal temperature distribution maps for Bunata, Guangawa Badiya, Odo Mike, Giwe, and Jirme kebeles (1992–2022)

In Guangawa Badiya, T_max exhibited similar variability, ranging from 27.92 °C in 2013 to 32.02 °C in 2003. This range reflects a degree of thermal flexibility that may have supported crop growth in both anomalously warm years (e.g., 2003) and cooler periods (e.g., 2013 and 2018). The highest recorded Belg T_max was again 34.34 °C (2003), while Meher T_max reached 31.33 °C (2002). T_min values exhibited a discernible warming trend, ranging from 12.4 °C in 2021 to 13.7 °C in 1992. Seasonal T_min for the Belg period ranged from 13.2 °C in 2012 to 16.44 °C in 2019, whereas Meher T_min ranged between 11.75 and 13.9 °C (Fig. 6).

In Odo Mike, annual maximum temperatures ranged from 28 °C in 2013 to 32 °C in 2003, with pronounced seasonal increases during the Belg season. Notably, Belg T_max reached 34.34 °C in 2003 and Meher T_max was notably lower but still high, with a maximum of 26.39 °C in 2013. Minimum temperatures T_min showed a warming trend, with the highest Belg T_min at 16.44 °C in 2019, and Meher T_min reaching 13.77 °C in the same year. Annual T_min ranged from 12.99 °C in 2016 to 15.62 °C in 2011. These elevated nighttime temperatures may have intensified evapotranspiration rates and reduced soil moisture retention, thereby contributing to reduced yields of temperature-sensitive crops (Fig. 6).

In Giwe, annual T_max varied from 25.86 °C in 2013 to 29.37 °C in 2004. Warmer years, such as 2003 (29.23 °C) and cooler ones like 2020 (26.57 °C) affected coffee growth. Belg T_max was highest at 31.39 °C in 2011, likely challenging coffee cultivation due to thermal stress, while cooler years like 2020 (26.57 °C) may have supported more stable growth conditions. Meher T_max varied between 24.02 °C in 2013 and 31.25 °C in 2003, with cooler nights benefiting growth. Annual T_min ranged from 10.30 °C (1993) to 11.93 °C (2004), with Meher T_min between 9.85 °C (2016) and 11.72 °C (2022), and Belg T_min between 10.34 °C (1993) and 13.43 °C (2011). These temperature fluctuations have mixed effects on crop productivity, with recent warming trends likely increasing evapotranspiration and leading to higher levels of crop stress (Fig. 6).

In Jirme, annual T_max ranged from 25.86 °C in 2013 to 29.36 °C in 2004, mirroring trends observed in Giwe. Seasonal peaks were evident, with Belg T_max varying from 27.5 to 31.39 °C, raising evapotranspiration stress. Meher T_max ranged between 24.12 and 28.93 °C. These elevated temperatures during critical phonological stages may have increased the vulnerability of crops to heat stress and soil moisture deficits. Annual T_min values ranged from 10.79 °C (1996) to 11.94 °C (2014), with lower values increasing frost risk and higher values promoting early growth. Meher T_min fluctuated between 9.85 and 11.72 °C, while Belg T_min ranged from 11.2 to 13.8 °C. These trends suggest that Jirme has experienced significant inter-annual and seasonal thermal variability, which could influence planting dates, crop selection, and yield stability under future climate scenarios (Fig. 6).

All sites showed decreasing trends in T_max (e.g., Bunata − 0.61 °C, Guangawa Badiya − 0.66 °C), whereas T_min continuously increased, suggesting that the diurnal temperature ranges (DTR) were getting smaller. Due to its impact on the delicate balance between respiration and photosynthesis, this pattern may impede optimal coffee physiology. Long-term increases in T_min can improve insect reproduction cycles, interfere with blooming cues, and degrade bean quality because of incomplete development, even if they can help perennial crops survive. Seasonal variability, variable T_max, and rising T_min all work together to highlight the need for climate-resilient coffee types, modified planting schedules, and integrated pest control plans that are suited to local microclimates. These temperature fluctuations highlight the critical need for stable climatic conditions to optimize coffee production and reduce the risks associated with climate variability (Bracken et al., 2023; Cassamo et al., 2023).

Spatiotemporal variability in maximum (T_max) and minimum temperatures (T_min) across Bunata, Odo Mike, and Jirme has demonstrable effects on coffee productivity, underscoring the urgent need for targeted climate adaptation strategies. Inter-annual climatic fluctuations, often linked to large-scale phenomena such as El Niño and La Niña, significantly influence local temperature regimes (Bedane et al., 2022; Kassie, 2014). In contrast, the relatively stable temperature trends observed in Guangawa Badiya and Giwe emphasize the critical role of thermal stability in sustaining rain-fed coffee systems. Variations in both T_max and T_min directly affect crop phenology, yield potential, and water-use efficiency, with elevated temperatures exacerbating evapotranspiration and soil moisture depletion. These findings support the necessity of implementing adaptive agricultural interventions such as the selection and deployment of temperature-resilient coffee cultivars and the modification of agronomic practices to buffer the adverse effects of increasing temperature variability and to promote the long-term sustainability of coffee production under changing climatic conditions (Olana et al.,2023).

The Mann–Kendall (MK) trend test and Sen’s slope estimator were applied to assess long-term trends in T_max and T_min across kebeles in Abaya and Gelana Woredas (1992–2022) (Table 4). The results indicate spatially heterogeneous yet significant temporal patterns.

Table 4.

MK trend and Sen’s slope analysis of T_max and T_min in Abaya and Gelana Woreda selected kebeles (1992–2022)

Study sites	Mann–Kendall trend (Z test)						Sen’s slope (Q)
	Belg		Meher		Annual		Belg		Meher		Annual
	Tmax	Tmin	Tmax	Tmin	Tmax	Tmin	Tmax	Tmin	Tmax	Tmin	Tmax	Tmin
Bunata	–0.41	0.37	–0.64	1.09	–0.61	–0.22	–0.012	0.006	–0.044	0.015	–0.035	–0.003
Guangawa B	–0.53	0.41	–0.68	1.07	–0.66	–0.24	–0.022	0.005	–0.052	0.018	–0.038	–0.006
Odo Mike	–0.45	0.39	–0.67	1.05	–0.68	–0.25	–0.013	0.011	–0.045	0.016	–0.038	–0.004
Giwe	0.17	1.09	–0.82	0.54	–0.43	0.51	0.003	0.012	–0.042	0.016	–0.026	–0.002
Jirme	–0.21	0.27	–0.74	1.08	–0.51	–0.27	–0.032	0.007	–0.048	0.014	–0.025	–0.007

In Bunata, MK analysis identified a weak cooling trend in T_max (− 0.61) and a slight warming trend in T_min (0.006), with Sen’s slope indicating modest increases in T_min and slight decreases in T_max, particularly during Meher.

In Guangawa Badiya, both Belg (− 0.53) and Meher (− 0.68) seasons showed declining T_max trends, while T_min exhibited positive trends (0.41 for Belg, 1.07 for Meher). Annual trends similarly pointed to a mild decrease in both T_max (− 0.66) and T_min (− 0.24), suggesting a narrowing diurnal temperature range.

Odo Mike displayed comparable patterns, with negative trends in T_max during Belg (− 0.45) and annual (− 0.68), while T_min increased during both Belg (0.39) and Meher (1.05). Sen’s slope estimates further confirmed these shifts, pointing to a gradual warming in nighttime temperatures alongside declining daytime peaks (Table 4).

Temperature trends in Giwe were more variable. T_max displayed a marginally positive trend during Belg (Z = 0.17) but a negative trend during Meher (Z =  − 0.82), with minor declines in Sen’s slope for annual (− 0.042) and Meher (− 0.026) periods. Conversely, T_min displayed significant positive trends in Belg (Z = 1.09) and Meher (Z = 0.54), with gradual increases in Sen’s slope for Belg (0.012), Meher (0.016), and annual T_min (0.002), indicating a steady warming trend (Table 4).

In Jirme, the MK test reveals a rise in T_min during Belg (Z = 1.09) and Meher (Z = 0.54), while T_max decreased slightly in Belg (Z = 0.17) and Meher (Z =  − 0.82) seasons. Annually, T_min increased slightly (Z = 0.51), whereas T_max declined (Z =  − 0.43). Sen’s slope supported these findings, indicating a progress increase in T_min for Belg (0.007) and Meher (0.014), alongside minor decreases in annual T_max (− 0.025) and Belg (− 0.021), reflecting gradual warming for T_min but consistent T_max declines (Table 4). These temperature dynamics—marked by increasing T_min and declining or fluctuating T_max—suggest a narrowing diurnal temperature range, which holds significant implications for coffee physiology, water use efficiency, and crop productivity in the face of ongoing climate change (Bedane et al., 2022).

Overall, temperature trend analyses (1992–2022) in Abaya and Gelana Woredas demonstrate a consistent pattern of warming nights (T_min) and cooling days (T_max). All sites showed negative trends in T_max, with the steepest declines observed in Guangawa Badiya and Odo Mike across both Belg and Meher seasons. In contrast, T_min trends were positive across all kebeles, with the strongest warming during Meher, especially in Jirme and Giwe (Z = 1.08–1.09). Sen’s slope estimates support these findings, indicating modest decreases in T_max (ranging from − 0.012 to − 0.052 °C/year) and gradual increases in T_min (up to 0.018 °C/year). In Bunata and Guangawa Badiya, T_max decreased while T_min increased, especially during Meher. Odo Mike showed similar tendencies, with rising T_min and falling T_max. The narrowing diurnal temperature range is most evident in Giwe and Jirme, where T_min is increasing while T_max either declines or remains stagnant. These changes suggest warming nights and cooling days, affecting agricultural systems like coffee (Table 4). Global climatic patterns are consistent with rising T_min and cooling T_max throughout the Belg and Meher seasons (Korecha & Barnston, 2007). While the consistent T_min increases and T_max reductions in Odo Mike and Jirme signal subtle climatic shifts, the magnitude of change may not yet be sufficient to attribute directly to anthropogenic climate change (Mihiretu et al., 2021). Nevertheless, these emerging trends underscore the importance of ongoing, localized climate monitoring to better understand and anticipate the impacts of temperature variability on rain-fed agricultural systems.

Impacts of climate change and variability on coffee production socioeconomic determinants of coffee production

The study surveyed 398 coffee-producing households in Abaya and Gelana Woredas, revealing that 76.4% were male-headed and 23.6% female-headed, highlighting a gender imbalance that may influence access to climate adaptation resources (Fig. 7). Education levels among household heads show moderate literacy, with 46.7% having primary education, 32.4% secondary, and only 4.5% lacking formal education, suggesting a potential constraint on access to adaptation knowledge and extension services. Household size emerged as a determinant of adaptive capacity. Smaller households (0–5 members, 48.5%) were more agile in resource allocation, while medium-sized (6–10 members, 37.9%) households benefited from available labor. In contrast, larger households (11 + members, 13.6%) faced increased pressure on resources, though communal labor may mitigate these challenges.

Fig. 7.

Descriptive analysis of the household socioeconomic characteristics

Economic vulnerability is apparent in monthly income distribution: 16.8% (67) earned less than 500 birr, 63.1% (251) earned between 501 and 5000 birr, and 20.1% (80) exceeded 5000 birr.

Agriculture remains the primary livelihood, with 53.5% (213) relying on cattle, 31.9% (127) on crop farming, and only small proportions on fodder production, 3.8% (15), and forestry, 2.3% (9). However, 56% of respondents reported income from non-agricultural sources, underscoring the importance of livelihood diversification in building resilience to climate variability (Fig. 7). Managing climatic variability requires income diversification, which offers alternate sources of income to mitigate agricultural risks (Amare, 2013; Getahun et al., 2023). To enhance resilience, households diversify income sources, modify farming practices, and seek off-farm labor. Socioeconomic factors and climate unpredictability emphasize the need for targeted interventions in income, resources, and education (Chima et al.,2024). These socioeconomic characteristics critically shape household-level vulnerability and adaptive capacity in the face of changing climate conditions affecting coffee production.

Model performance and key drivers influencing coffee productivity

The multiple regression analysis (Table 5) reveals a robust model explaining coffee production variability in Abaya and Gelana Woredas. The adjusted R² of 0.832 indicates that approximately 83.2% of the variation in coffee productivity is explained by the independent variables, including climate variability, soil fertility, soil erosion, and socioeconomic factors. A strong positive correlation (R = 0.871) further supports the model’s predictive capacity. Multicollinearity diagnostics (VIF = 1.026–1.243) suggest minimal collinearity among predictors, enhancing the reliability of the regression coefficients. The model’s standard error of 0.55673 and the significant F-change (p = 0.029) indicate a good model fit. However, the Durbin-Watson statistic (1.119) suggests mild autocorrelation in residuals, which may warrant further investigation in future model refinement. Key factors significantly associated with coffee productivity include climate variability, soil fertility, livestock ownership, access to loans, and extension services. The amount of variance in the dependent variable that cannot be accounted for by the independent variables is represented by the residual sum of squares in the model, which is 117.781. The mean squared difference between the actual and anticipated values is rather small, as indicated by the related mean square error (MSE) of 0.310. This implies that although some variation is still unaccounted for, the model estimates the dependent variable with a respectable degree of accuracy. The ANOVA results (p = 0.005) confirm the overall significance of the model in explaining the observed variability in production across the 1993–2023 period.

Table 5.

Analysis of the regression model

Model statistics	Values
Correlation coefficient (R)	0.871
R-squared (R2)	0.416
Adjusted R-squared	0.832
Standard error of the estimate	0.55673
Durbin-Watson statistic	0.029
Significance of F-change	0.029
ANOVA results	Sum of squares	Mean square	Significance (p-value)
Regression	9.355	0.550	0.005
Residual	117.781	0.310
Total	127.136	398

a. Predictors: (Constant), climate variability, socioeconomic factors, soil fertility, and etc

b. Dependent Variable: Coffee production (qt/ha) in 1993 to 2023

Regression results demonstrate that climate variability has a significant negative impact on coffee productivity in Abaya and Gelana Woredas (B =  − 0.181, p = 0.0078), primarily due to temperature and rainfall fluctuations affecting fruit development. A strong baseline effect was observed (B = 1.754, p < 0.005), indicating underlying environmental or systemic influences on output.

Among socioeconomic variables, monthly income showed a modest positive relationship with productivity (B = 0.039, p = 0.0417), while income diversification and livestock ownership had negligible effects. Off-farm employment also had a non-significant influence (B = 0.077, p = 0.191). Key agricultural variables—including the number of coffee trees (B = 0.096, p = 0.002), years of cultivation (B = 0.142, p = 0.005), and land availability (B = 0.057, p = 0.008)—were positively and significantly associated with output, reinforcing findings from similar studies on the critical role of tree density and land access in enhancing yields (Duguma et al., 2024; Yirga et al., 2024). Experience growing coffee (B = 0.005) and land size (B =  − 0.051) have negligible and non-significant effects, respectively (Table A).

Contrary to expectations, access to markets and extension services showed negative associations, while access to credit had a weakly positive but negligible effect (B = 0.070). Surprisingly, longer residence in the area negatively impacted productivity (B =  − 0.119, p = 0.003), possibly due to entrenched farming practices or aging producer demographics. Other limiting factors, such as pests, diseases, and low soil fertility, also contributed to declining yields (B =  − 0.013, p = 0.008), in alignment with earlier work by Yona et al. (2024), which highlighted the negative consequences of erratic climatic trends on coffee cultivation (Table A). This highlights the urgent need for efficient soil management, adaptive strategies to counteract climate variability, and focused interventions to improve fertility and decrease soil erosion (Lal, 2014).

Adaptation strategies for minimizing the impacts of climate change and variability on coffee production

The findings from Table 6 illustrate a broad range of adaptation strategies that coffee farmers in Abaya and Gelana Woredas are employing to mitigate the effects of climate change and enhance coffee production resilience. Income diversification is the most widely used adaptation strategy overall, accounting for 45 responses (8.7%) of all strategies and being adopted by 96.5% of households. This high rate suggests that households are shifting their incomes away from climate-sensitive crops and toward alternative sources like small businesses, wage labor, and trade.

Table 6.

Potential adaptation strategies households use during climate change

Adaptation strategiesa	HH responses (N)	Percent of total responses	Percent of cases
Intercropping	35	8.7%	96.5%
Cultivar selection	34	8.6%	95.5%
Agroforestry	34	8.5%	94.7%
Income diversification	45	8.7%	96.5%
Shift to other crop	39	7.8%	87.4%
Integrated pest management (IPM)	31	7.8%	92.0%
Terracing	30	7.6%	84.4%
Stumping (number of coffee trees)	30	7.5%	83.4%
Soil bund (stone bund)	28	7.1%	78.6%
Mulching	27	6.8%	75.6%
Pruning (number of coffee trees)	26	6.6%	73.9%
Daily labor	16	4.0%	44.2%
Petty market	14	3.5%	39.4%
Irrigation	10	2.5%	32.7%
Total	389	100%	1114.3%

^aDichotomy group tabulated at value 1

Intercropping was equally prevalent, with 35 responses (8.7%) and 96.5% household adoption. It was essential for stabilizing yields, lowering climate risk, and increasing land-use efficiency. These strategies are consistent with the findings of Bracken et al. (2023) and Tilahun et al. (2023), who emphasize that intercropping fosters resilience through agricultural diversification and enhances ecosystem services. Similarly, agroforestry (34; 8.5%; 94.7%) and cultivar selection (34 responses; 8.6%; 95.5%) were common practices. In order to improve soil fertility, shade coffee plants, and boost biodiversity, all essential components of climate-smart agriculture farmers are depending more and more on drought-resistant cultivars and tree-crop integration. These tactics show a growing emphasis on ecological adaptation. This approach aligns with the recommendations of Obsi et al. (2023) who emphasize the importance of crop substitution in ensuring continued productivity under variable climatic conditions.

Additionally, households reported using integrated pest management (IPM) (31; 7.8%; 92.0%) and switching to other crops (39 responses; 7.8%; 87.4%) extensively, which reflects direct responses to the ecological pressures of a changing climate, such as temperature variability, pest outbreaks, and changes in rainfall. These actions demonstrate proactive risk management as well as agronomic flexibility.

Terracing (30 responses; 7.6%; 84.4%), coffee tree stumping (30; 7.5%; 83.4%), soil bunds/stone bunds (28; 7.1%; 78.6%), mulching (27; 6.8%; 75.6%), and pruning (26; 6.6%; 73.9%) were among the methods that were highlighted as being important for conserving soil and water. In order to maintain productivity in the face of increasingly unpredictable weather, these practices place a strong emphasis on domestic initiatives to retain moisture, stop soil erosion, and restore aging coffee trees. This highlights the importance of targeted initiatives to remove adoption barriers and ensure that farmers have equitable access to efficient and affordable solutions, as emphasized by Mairura et al. (2021). On the other hand, fewer people used tactics like irrigation (10; 2.5%; 32.7%), petty market activities (14; 3.5%; 39.4%), and daily labor (16; 4.0%; 44.2%). While reliance on labor migration and market trading demonstrates livelihood diversification born out of necessity rather than opportunity, the lower adoption of irrigation, in particular, may reflect financial and infrastructure constraints.

The table shows that most households use multiple strategies at the same time, with a total of 389 responses, or 1114.3% of cases. This diversity highlights a high degree of community awareness and agency in managing climate risk, in addition to the intricate and overlapping nature of local adaptation. An effective and sustainable model for household adaptation is presented by the combination of agronomic innovations, ecological practices, and economic alternatives; however, it still needs institutional support, better access to climate information, and resource-efficient technologies. Akinkuolie et al. (2024) and Gashure (2024) underscore the need for continuous support, including education, resource access, and income diversification, to enhance adaptive capacity and secure farmers’ livelihoods amid ongoing climatic uncertainties.

Conclusions and recommendations

This study provides a thorough analysis of the temperature and precipitation trends in Abaya and Gelana Woredas over a 30-year span (1992–2022), highlighting significant temporal and spatial climate variations. According to the data, there appears to be significant climatic variability across sites, with areas such as Jirme showing the most noticeable seasonal variations during the Meher and Belg seasons. Coefficients of variation (CV) above 30%, on the other hand, showed greater inter-annual rainfall variability in Guangawa Badiya and Odo Mike, which might be a reflection of erratic rainfall patterns. In contrast, Bunata displayed more moderate variability (CV between 20 and 30%), indicating comparatively more stable weather.

Records of precipitation show that the average annual rainfall varies significantly from year to year and that there is little long-term change. The study area’s erratic precipitation patterns are demonstrated by the years 1993, 1997, 2000, 2005, and 2022, which had higher rainfall totals. Guangawa Badiya had multiple wet Belg seasons, Giwe stayed comparatively dry, and Odo Mike saw fewer periods of intense precipitation. On the other hand, Jirme saw both major rainy seasons with mostly normal conditions interspersed with extreme occurrences. These results highlight how crucial it is to identify regional climatic trends in order to develop context-specific adaptation plans.

The local climatic heterogeneity was further highlighted by the spatial variation in maximum rainfall years across sites. For example, Guangawa Badiya reached its highest in 2018, while Bunata experienced its highest rainfall in 1996, 2005, and 2020. Giwe is classified as more arid due to its consistently low rainfall, whereas Odo Mike showed significant variations over time. This variance makes it more difficult to comprehend how various communities might be affected and emphasizes how important local climate data is for resource allocation and planning.

Sen’s slope estimator and the Mann–Kendall test were used to analyze trends and find new patterns in rainfall that could affect agricultural productivity, especially coffee. For instance, Bunata showed only slight seasonal increases in rainfall, but a downward yearly trend may make people wonder if coffee farming is still viable in the current climate. Positive seasonal and annual rainfall trends, on the other hand, were observed in Guangawa Badiya, which may indicate more conducive growing conditions. On the other hand, Odo Mike showed negative rainfall trends, particularly during the Meher season and on an annual basis, indicating a possible vulnerability to rainfall variability and the significance of focused adaptation measures. Variability in maximum temperatures (T_max) may lead to heat stress during sensitive growth stages, while overall increases in minimum temperatures (T_min) may promote plant growth and development. Although more research is needed to determine the precise physiological effects on coffee, these variations, especially during the busiest growing seasons, may have an impact on crop productivity and health.

Strong relationships between climatic factors and coffee production were highlighted by the climate impact model created for Abaya and Gelana Woredas, which explained 83.2% of the variation in coffee production. Coffee tree density and land availability were positively correlated with yields, but variables like rainfall variability and deteriorating soil fertility seemed to limit output. These results highlight the significance of adaptive agricultural strategies and integrated soil and water conservation practices as a safeguard against climate-related risks and to promote long-term productivity.

Based on these findings, several key adaptation strategies are recommended to improve resilience to climate change and ensure the long-term sustainability of coffee production in the region. These strategies include promoting crop diversification (8.7%), intercropping systems (8.7%), the selection of climate-resilient cultivars (8.6%), and income diversification (11.%). Income diversification is the most widely used adaptation strategy overall, accounting for 45 responses (11.6%) and enhancing water management practices such as irrigation and stone bunding. When effectively implemented, these strategies have the potential to mitigate the adverse effects of climate variability while optimizing resource use and maintaining coffee yields.

To enhance the adaptive capacity of coffee farmers, the study underscores the importance of implementing climate-resilient practices. Income diversification, alongside the introduction of climate-smart training programs and localized climate monitoring systems, will equip farmers with the tools and knowledge necessary to adapt to changing climatic conditions. Additionally, improving infrastructure for water management, such as efficient irrigation and rainwater harvesting systems, will be essential for ensuring stable coffee production in drought-prone areas. Government support is crucial in providing farmers with access to affordable credit, climate-tolerant coffee cultivars, and market access, alongside promoting capacity-building efforts through extension services.

In conclusion, the successful adaptation of coffee farming in Abaya and Gelana Woredas will depend on collaborative efforts among policymakers, researchers, and farming communities. By fostering an enabling environment that integrates climate resilience strategies with sustainable agricultural practices, coffee farmers can safeguard their livelihoods while ensuring the long-term viability of coffee production in the face of ongoing climatic challenges.

Appendix

Appendix A. Supplementary Data

Impact of Individual Components on Final Land Suitability Score for Coffee Production.

Coefficients^a

Table A: Results of Inferential Statistics

Model	Unstandardized Coefficients		Standard Sized Coefficients	t-test	Sig	95.0%Confidence Interval for B		Collinearity Statistics
	B	Std. Error	Beta			Lower Bound	Upper Bound	Tolerance	VIF
1 (Constant)	1.754	.348	.041	5.036	.000	1.069	2.438
Monthly income (in birr)	.039	.047	-.053	.812	.0417	.055	.132	.939	1.065
Income sources	-.027	.026	.004	−1.056	.007	.077	.023	.975	1.026
Livestock ownership	.003	.038	.067	.081	.936	-.072	.078	.938	1.066
Off-farming employment	.077	.059	.090	1.309	.191	-.039	.192	.920	1.087
Number of plot of land	.057	.033	.007	1.735	.008	.0018	.121	.910	1.099
Experience in coffee farming	.005	.038	.152	.140	.889	.069	.079	.950	1.052
Years of coffee trees	.142	.050	.119	2.815	.005	.043	.240	.839	1.192
Number of coffee trees	.096	.041	-.057	2.310	.002	.014	.177	.927	1.079
Land size for coffee farming(ha)	-.051	.046	-.016	−1.117	0.003	.142	.039	.932	1.073
Distance from coffee processing(km)	-.014	.047	.034	-.310	.756	-.106	.077	.960	1.042
Access to Credit Service	.070	.104	-.011	.671	.0503	-.135	.275	.941	1.063
Access to extension, medication & treatment services	-.027	.117	-.060	-.227	.0082	-.258	.204	.960	1.042
Access to markets	-.084	.074	-.021	−1.139	0.002	-.230	.061	.868	1.153
Perceived management a/t cooperatives	-.037	.088	-.160	-.419	.675	-.210	.136	.950	1.053
Years lived in the locality	-.119	.040	.012	−2.992	.003	-.197	-.041	.855	1.169
Climate variability impacts	-.181	.085	-.013	.311	.0078	-.045	.018	.824	1.243
Increase in temperature	-.073	.056	-.048	−2.371	0.004	.062	.045	.935	1.376
Decrease in temperature	-.152	.073	-.071	.432	0.001	.125	.304	.874	1.087
Fluctuation in rainy season	-.091	.081	-.019	−2.641	0.007	.1052	.176	.348	1.812
Fruit development and ripening	-.087	.076	-.052	.731	.0248	-.146	.075	.227	1.081
Occurrence of pest and disease	-.091	.019	-.057	.876	.0018	-.164	.063	.523	1.093
Access to climate variability & soil erosion info	-.023	.098	-.009	.231	.0018	-.170	.216	.914	1.094
Soil Fertility	-.013	.070	.041	-.183	.0085	-.150	.125	.916	1.092

a. Dependent Variable: Coffee production (qt/ha) in 1993 to 2023

Author contribution

The first author (TD) analyzed the data and drafted the manuscript, which is part of his PhD dissertation in Environmental Sciences at the College of Natural and Computational Science, Bule Hora University, Ethiopia. The second author (TT), the third author (SM), the fourth author (EB), and the fifth author (GE) contributed to the manuscript's improvement by providing edits and comments. All authors reviewed and approved the final version of the manuscript.

Data availability

No datasets were generated or analysed during the current study.

Declarations

Ethics approval and consent to participate

Every author has read and comprehended the Instructions for Authors’ statement on the “Ethical Responsibilities of Authors.” They attest to their compliance, if appropriate, and understand that, with very few exceptions, authorship cannot be changed after the work has been submitted.

In order to maintain a polite and transparent environment, participants were informed of the study’s objectives, confidentiality was guaranteed, and they were free to skip questions or discontinue participation at any moment.

Consent for publication

The authors have agreed and accepted the submission of the work.

Competing interests

The authors declare no competing interests.