1. Introduction

Wheat is a staple crop that provides both a significant source of income and food in Ethiopia [1]. Ethiopia’s efforts to increase its wheat self-sufficiency rely on tactics that include both vertical and horizontal agricultural growth [2]. A strategic shift towards intensified wheat cultivation, particularly through double-cropping in low and midlands where water resources abound, holds paramount significance at the national level [3]. Notably, the Ethiopian government’s recent adoption of a wheat irrigation policy underscores a concerted effort towards bolstering production through the adoption of cutting-edge irrigation technologies tailored to the diverse agro-ecological contexts of wheat cultivation [4]. When water resources are available to irrigate wheat, boosting wheat productivity in rain-fed and irrigable lowland and midland areas can pave the way for wheat self-sufficiency [5].

Nowadays, irrigated farming is classified as a significant catalyst for increased wheat crop production in Ethiopia [6]. Expanding the country’s irrigation potential can improve agricultural productivity and extend annual growing seasons reducing poverty, food insecurity, and import dependency with individual and collective action by governments, the private sector, and communities in rural and urban areas [7]. It has been demonstrated that both irrigated and rain-fed agriculture are important in the Ethiopian economy. However, the majority of wheat production goes to rain-fed farming, with irrigation accounting for only about 3% [8].

In Ethiopia, irrigation is regarded as a critical strategy for alleviating poverty and thus food security by converting the rain-fed agricultural system into a combined rain-fed and irrigation system, which is considered the country’s most prominent means of sustainable development [9]. Thus, irrigated agriculture is viewed as a necessity instead of an option, and irrigated wheat production areas must be expanded, particularly in Ethiopia’s arid and semiarid regions, to reduce commercial wheat imports. Therefore, Ethiopia has set goals to achieve wheat self-sufficiency by expanding irrigable lowland areas and increasing rain-fed wheat productivity [5]. In line with this, during the 2020/2021 season, irrigated wheat production was successfully carried out in the lowlands of the Arba Minch district, in South Ethiopia’s regional state [10]. However, very limited attempts were available to evaluate the performance of the lowlands wheat production and its future fate. In this regard, model-assisted studies are essential, which would enable one to predict future production based on historical/current achievements.

Models serve as invaluable tools for exploring and refining alternative management approaches aimed at maximizing yield while minimizing irrigation water usage in arid and semi-arid regions [11]. They also function as decision-support systems for system management [12], aiding in pre-season and in-season decision-making regarding various management practices such as irrigation, fertilizer application, cultivar selection, pesticide use, and cultural practices [1]. Moreover, these models provide critical support to policymakers by predicting climatic shifts, fertilizer and pesticide dynamics, crop yield forecasts, and soil erosion trends [13].

Numerous models are available for predicting wheat crop yield, including CERES-Wheat, CERES, DEMETER, SHOOTGRO, APES, CROPSYST, STICS, WOFOST, and DSSAT, among others, which have been extensively studied in the literature [14–16]. These models enable the evaluation of field conditions, forecasting of crop growth and yield based on environmental and meteorological factors, and streamlining of agricultural research processes [14, 17]. Additionally, they assist in optimizing farming practices and providing technical guidance to farmers and decision-makers [18, 19]. In recent years, there has been an increasing use of crop growth models at local and regional levels to benefit the agricultural community and improve productivity.

One notable recent development is the Agricultural Policy/Environmental eXtender (APEX) model, developed jointly by the United States Department of Agriculture (USDA) Agricultural Research Service (ARS) and Texas A&M University. APEX was designed to expand the capabilities of the Environmental Policy Integrated Climate (EPIC) model to encompass entire farms and watersheds. This enhancement was driven by the necessity to improve EPIC’s ability to simulate the effects of land management practices on small to medium-sized watersheds and diverse agricultural landscapes [20]. Recent advancements in crop growth and yield modeling have incorporated remote sensing applications [21] and machine learning techniques for prediction [22]. Nonetheless, validation using field data remains crucial to ensuring the reliability and applicability of these models across various conditions. In this study, the APEX model was used to evaluate the crop growth and yield conditions at Arba Minch. The model would shed light on better management of irrigated and/or rainfed farms or small watersheds to achieve sustainable production efficiency [20]. APEX can be utilized to assess how crop yield is affected by rainfed farming systems, irrigation water supply, and management [23, 24]. Therefore, the present study aims to model irrigated and rainfed wheat production in Arba Minch Zuria District, Southern Ethiopia, using an APEX model, which holds significant implications for advancing agricultural modeling techniques, understanding climate change impacts, informing policy decisions, guiding farming practices, and contributing to the scientific knowledge base.

2. Materials and Methods

2.1. Description of the Study Area

The particular study area is located in the Arba Minch Zuria district of the Gamo Zone in the South Ethiopia region, around 426 km south of Addis Ababa. Geographically, the study area is located between longitude: 37°20′0″–37°30′0″E, and latitude: 6°06′30″–6°10′30″N (Figure 1). The elevation of the study area ranges from 1150 to 3300 m above mean sea level. The study area is characterized by very gentle to steep slope terrain [25]. It has 29 “kebeles,” of which 13 kebeles are wheat producers (Figure 1). “Kebele” refers to a sub-district-level administrative area. The total population of the district is approximated to be 165,680, including 82,751 males and 82,929 females [26].

Based on the assessment of 35 years (1987 to 2022) of meteorological data from the neighboring stations in the study area, the average minimum and maximum temperatures are 17.40°C (December) and 30.63°C (March), respectively. Rainfall at Arba Minch station follows a bimodal pattern, with the first peak rainy season occurring from April to June and the second rainy period spanning from September to October, with average annual rainfall ranging from 800 to 1500 mm. Depending on the rainfall regime, irrigation is required throughout the dry season from January to March and from June to August. The agro-ecological zone of the study area is classified as 14% highlands, 53% midlands, and 33% lowlands [27]. The majority of farmers in the district grow crops such as corn, wheat, vegetables, and fruits [28].

The main soil texture types in the study area include loam, silt loam, and sandy loam soils, and the bulk density of the soil ranged from 1.25 to 1.55 Mg m⁻³. The soil pH value of the study area varied from 6.5 to 8.10, while the soil OM ranged from 2.07% to 2.5%. The total N content of the study area coverage is from 0.2% to 3.9%, Na ranged from 0.3 to 1 mol (+) kg⁻¹ and the electrical conductivity of soils has varied from 0.1 to 2.01 dS m⁻¹ in soils [29].

2.2. Methods

2.2.1. Data Collection and Sampling

The primary data including irrigated and rainfed wheat yield, all farm input materials, and farm activities to manage the crop and other relevant data were collected from the Arba Minch Zuria district. The main data used as input to the model were categorized as climate, crop, soil, field management, irrigation practices, and watershed characteristics from the Ethiopian Meteorological Institute, Ethiopian Water Ministry, region, and district agriculture offices. The primary data required concerning the farming and management of wheat in each kebele was collected using structured questionnaires, face-to-face interviews, direct surveys, and focus group discussion (FGD) techniques. The interviews and FGD were done with the selected leaders and experienced individuals in the district. Before using the above methods, we adopted a multi-stage sampling technique to arrive at the required sample size. A purposive sampling technique was applied for selecting the villages (kebeles) in the district. Each kebele’s wheat yields were gathered and averaged annually to represent the crop output for the entire study district.

2.2.2. Model Selection Criteria

Crop simulation models serve diverse purposes and are categorized into statistical, scientific, and engineering models [30]. Statistical models establish relationships between climatic parameters and crop yield components using statistical techniques like correlation and regression. Various complex scientific models, such as APSIM [31], CropSyst [32], EPIC [33], and the DSSAT cropping system [34] models predict crop growth and yield under different agro-ecological conditions. However, their complexity poses challenges for end-users and policymakers in farm and irrigation planning due to the extensive input and advanced skill requirements for calibration and operation [35].

Among these models, the APEX model is important for farm and watershed-level crop yield modeling [36]. Using the APEX model on a time scale, long sustained simulations can be carried out to simulate the effects of various nutrient management strategies, tillage operations, alternative cropping systems, conservation strategies, and other management strategies on crop yield, nutrient, sediment, and surface runoff [37]. In addition, the model can be set up for innovative land management techniques like the application of manure removed from waste storage ponds to the land, the effects of filter strips on pollutant losses from upslope crop fields, and intensive rotational grazing scenarios showing livestock movement between paddocks [20]. Therefore, the APEX model was used to analyze the performance of wheat yield after sampling from the farms in the Arba Minch Zuria District.

2.2.3. APEX Model Set-Up

The APEX model was created to expand the functionality of the EPIC model to encompass entire farms and watersheds. The model also simulates the effects of land management on small to medium-sized watersheds and diverse farms [20]. Various land management techniques, including fertilization, buffer strips, drainage, irrigation, terraces, manure management, lagoons, waterways, reservoirs, pesticide application, crop selection and rotation, grazing, and tillage, can be set up with it. The APEX model is a powerful and adaptable tool that can simulate land use and management effects for entire farms as well as small watersheds [37]. APEX can be applied to entire farms or watersheds that are segmented according to fields, landscape positions, soil types, or sub-watersheds. It is essentially a multi-field variant of the previous EPIC model. Crop yield in the APEX model is estimated by using the harvest index concept (Equation (1)) [20].

$$ ()

where YLD is the quantity of the crop yield harvested from the field in tons ha⁻¹, HI is the harvest index, and STL is the above-ground biomass in tons.ha⁻¹.

APEX functions on a time scale and can conduct long-lasting simulations and be used to model the effects of various management methods, including water management, alternative cropping systems, conservation techniques, tillage operations, and fertilizer management, on crop output [36]. Table 1 shows a list and description of the APEX input data used to simulate wheat crop yield.

Table 1. Input data for the APEX model [20].

APEX input parameters	Sub input variables
Climate conditions	Daily or monthly maximum and minimum temperature; humidity; solar radiation; wind speed; and rainfall
Crop parameters	Wheat crop components (yield); crop type; seed rate; plant population; Plant density
Soil physical parameters	Saturation soil water content; field capacity; PWP; Soil texture
Parameters of field management practices	Select tillage type; select equipment; fertilizer type and rate; Pesticide type and rate
Parameters of irrigation management practices	Irrigation method, type, application depth, and time of irrigation events
Watershed characteristics	Watershed name, location, and area of cultivated land

2.2.4. Sensitivity Analysis

The model sensitivity analysis is a technique for finding important parameters that influence model performance and are vital for model parametrization. The APEX model has large groups of parameters related to crops, sediment, hydrology, nutrients, and other environmental factors. Therefore, analysis of sensitivity is the initial step for crop models, which helps to analyze and narrow down the parameters for calibration.

The APEX model sensitivity analysis was carried out first by identifying the key input parameters that influence the model outputs. These parameters include soil properties, climate variables, crop management practices, and land use characteristics. Secondly, the variation of the identified key input parameters was done individually (one-at-a-time sensitivity analysis) to assess the combined effect of multiple factors. Then, for each set of parameter values, the APEX model was run to simulate the crop yield over a defined period. The model calculates the output crop yields under the given parameter settings. In the fourth stage, parameters that change the model outputs were determined, followed by the identification of indices or metrics (Sobol indices) that are used to quantify the comparative status of each input parameter on model outputs. Finally, the results of the sensitivity analysis were interpreted to understand the model’s behavior and the underlying factors driving variability in model outputs.

2.2.5. Model Calibration and Validation

Once the model-sensitive parameters were determined, model calibration and validation on wheat crop yields were carried out. There are several calibration and validation techniques exist [38]. In this research, the data set was classified into two parts: two-thirds of the data was used for model calibration and one-third of the data was used for model validation. One thousand simulation runs were made locally based on a combination of parameter values covering the full range of the selected sensitive parameters, while the rest of the nonsensitive parameters were all set to their default values.

The calibration and validation of the APEX model were completed using the data on agricultural practices, weather, management data, and soils collected from the farmers’ fields, and the method of model calibration and validation was done manually [37]. The model calibration for rainfed wheat yield was based on the data from 2009 to 2018, while the calibration for irrigated wheat crop yield was conducted using the data for the period from 2018 to 2021. Validation of rainfed wheat crop yield was evaluated based on the data from 2019 to 2023, and for irrigated wheat crop yield, the model run was based on data from 2022 to 2023. The study used the APEX crop parameters and their values for simulating wheat crops, which are detailed in Table 2.

Table 2. Parameters used to calibrate the APEX model.

Parameters	Description	Range	File location in APEX
TOP	The optimal temperature for plant growth	15–35	APEXCONT.DAT
TBS	Minimum temperature for plant growth	0–15	APEXCONT.DAT
ISW	Field Capacity/Wilting Point Estimation	0–5	APEXCONT.DAT
DMLA	Maximum area leaf area index	5–6	SUBAREA.SUB
BIR	Irrigation Auto Trigger	Infinitive	SUBAREA.SUB
ARMN	Minimum single irrigation application volume	100–200	SUBAREA.SUB
PST (Pest)	Insects and disease factor	0–10	PEST.DATA
PARM No. 3	Water stress-harvest index	0–1	PARM.DAT
PARM No. 5	Soil water lower limit	0.0–1.0	PARM.DAT
PARM No. 10	Pest damage cover threshold (t/ha)	1–10	PARM.DAT
PARM No. 12	Soil evaporation—plant cover factor	0–0.5	PARM.DAT
PARM No. 15	Runoff CN residue adjustment parameter	0–0.3	PARM.DAT
PARM No. 20	Runoff curve number initial abstraction	0.05–0.4	PARM.DAT
PARM No. 42	SCS curve number index coefficient	0.3 −2.5	PARM.DAT
PARM No. 90	Subsurface flow factor	1–100	PARM.DAT

2.2.6. Model Performance Evaluation

The APEX model performance was evaluated for the entire wheat producer kebeles. The statistical measures of the APEX model, including Nash-Sutcliffe efficiency (NSE), percentage bias (PBIAS), observation standard deviation ratio (RSR), and coefficient of determination (R²) [23] were used to evaluate the model. The coefficient of determination (R²) describes the degree of collinearity between simulated and observed data. R² ranges from 0 to 1, with higher values indicating less error variance, and typically values greater than 0.5 are considered acceptable, which is calculated as shown in Equation (2) [39].

The NSE is a normalized statistic that measures the relative size of residual variance (“noise”) vs. observed data variance (“information”) [40]. The NSE (Equation (3)) runs from −∞ to 1.0 (1 inclusive), with NSE = 1 being the ideal value. Values between 0.0 and 1.0 are considered acceptable levels of performance, but values <0.0 imply that the mean observed value is a stronger indicator than the simulated value, indicating inadequate model performance. NSE values of 0.75–1.00, 0.65−0.75, 0.5–0.65, and NSE < 0.5 indicate very good, acceptable, satisfactory, and unsatisfactory performance, respectively.

Percent bias (PBIAS) quantifies the simulated data’s average tendency to be smaller or larger than their measured equivalents. PBIAS’s ideal value is 0.0, with small magnitude values identifying accurate model simulation. Positive values imply model underestimation bias; negative values suggest model overestimation bias. Equation (4) illustrates how PBIAS is determined. PBIAS scores <10, 10–15, 15−25, and >25 are classified as very good, good, satisfactory, and unsatisfactory, respectively [41]. To ensure that the reported values and consequent statistics apply to a wide range of constituents, the Root Mean Standard Deviation Ratio (RSR) statistic was used. RSR integrates the advantages of error index statistics with a scaling/normalization factor. RSR ranges from a big positive value to the ideal value of 0, which denotes excellent model simulation with 0% RMSE or residual variation [42]. RSR values 0–0.5, 0.5–0.6, 0.6–0.7, and >0.7 are classified as excellent, good, satisfactory, and unsatisfactory, respectively [42].

$$ ()

$$ ()

$$ ()

$$ ()

where Yoi and Ysi represent the ith observation and simulated value for the component being evaluated, respectively; Ym and Ys represent the average data of observed and simulated values, respectively; and n represents the total number of measurements. The flowchart for the overall study is summarized in Figure 2.

3. Results

3.3. Yield Simulation of Irrigated Wheat

The potential wheat crop yields in response to climate, soil properties, irrigation water application, and management factors during the dry season were simulated using the APEX model. The model captured the yield of the irrigated wheat since the observed and simulated wheat yields on an annual basis are in fair agreement (Figure 3). It has been observed that the yield of irrigated wheat crops was slightly above the model-computed yields over 6 years (2018–2023) (Figure 3). The recorded crop yields through the district wheat producer kebeles in the research area varied from 10 to 11.5 tons on 20 ha area coverage, while the estimated yields ranged from 9.57 to 11.31 tons on 20 ha area during the crop growth years from 2018 to 2023. In the initial crop growth year (2018), the simulated yields were slightly higher than the actual yields, but for the subsequent years (2019–2023), the simulated yields were slightly lower. This indicates a favorable relationship between the simulated and measured yields during the dry season when an irrigation system was utilized.

The APEX crop model was manually calibrated by modifying the parameter values of the all-input technologies database to calibrate it for the dry season, particularly when irrigation was used in the Arba Minch Zuriya district in South Ethiopia. This calibration process involved utilizing 4 years of measured irrigated wheat crop yield data from 2018 to 2021, followed by validation from 2022 to 2023 (Figure 3). The statistical performance measures indicated that the APEX model’s performance in simulating irrigated wheat crop yield was very good, with NSE > 0.7 (very good), RSR ≤ 0.5 (very good), and PBIAS ≤ ±10% (very good). Additionally, the determination coefficient for both calibration and validation demonstrated very good model performance (R² > 0.80) (Table 5). Consequently, the APEX model’s performance in simulating irrigated wheat crop yield exhibited a very good agreement with the annually measured irrigated wheat crop yield for both calibration and validation in the Arba Minch Zuria district.

Table 5. APEX model performance at Arba Minch Zuria district irrigated wheat production.

Period	Observed		Simulated		Performance measures
	Mean	CV	Mean	CV	R2	NSE	RSR	PBIAS (%)
Calibration (2018–2021)	10.88	6.3	10.77	5.7	0.79	0.70	0.50	10.00
Validation (2022–2023)	10.88	9.9	10.55	11.8	0.93	0.73	0.50	10.00

3.4. Yield Simulation of Rainfed Wheat

The APEX model was used to forecast the yield of rainfed wheat in the research area. As a result, the model effectively predicted rainfed wheat yields under different weather conditions, soil characteristics, and agricultural practices during the rainy season. The predicted rainfed wheat yield was compared with the actual yields measured over 15 consecutive years (2009–2023) in the Arba Minch Zuria district (Figure 4). Across the study area, the simulated and observed yields ranged from 8.82 to 11.55 tons and from 9 to 11.6 tons on a 20 ha average area, respectively, during the crop growth years from 2018 to 2023. The comparison revealed that in some years (2009, 2011, 2014, 2016, and 2023), the model slightly overestimated the yields, while in the remaining 10 years, it slightly underestimated them (Figure 4). Consequently, the relationship between the simulated and observed rainfed yields during the rainy season was deemed to be favorable.

The APEX crop model underwent calibration and validation using 10 years of rainfed wheat crop yield data (2009 to 2018) and 5 years of data (2019 to 2023), respectively (Figure 4). Assessment based on statistical performance measures indicated that the APEX model’s performance in simulating irrigated wheat crop yield was very good, with NSE > 0.7 (very good), RSR ≤ 0.5 (very good), and PBIAS ≤ ±10% (very good). Additionally, the determination coefficient for both the calibration and validation periods demonstrated a very good model performance (with R² > 0.80) (Table 6). Consequently, the APEX model’s simulation of rainfed wheat crop yield exhibited very good agreement with the annually measured rainfed wheat crop yield for both calibration and validation in the Arba Minch Zuria district.

Table 6. APEX model performance at Arba Minch Zuria district rainfed wheat production.

Period	Observed		Simulated		Performance measures
	Mean	CV	Mean	CV	R2	NSE	RSR	PBIAS (%)
Calibration (2009–2018)	10.66	7.2	10.69	6.5	0.91	0.91	0.29	−3.4
Validation (2019–2023)	11.02	5.8	10.89	4.4	0.87	0.80	0.43	5.50

4. Discussion

This study employed the APEX model to simulate both irrigated and rainfed wheat production in the Arba Minch Zuria district of southern Ethiopia. The results showed that the application of the APEX model facilitated the evaluation of various factors influencing wheat production in both irrigated and rainfed conditions. The simulation of the APEX crop model was done by integrating soil characteristics, climate data, and crop management practices influences factors. This comprehensive analysis provided insights into the productivity potential of wheat cultivation [20] to offer valuable implications for sustainable agricultural development.

The sensitivity analysis was intended to identify the key parameters affecting wheat production under irrigated and rainfed conditions in the Arba Minch Zuria district. All relevant parameters for APEX crop yield simulation were considered in the sensitivity analysis. Nine parameters were found to be sensitive for irrigated wheat production, while 10 parameters were sensitive for rainfed wheat production. These parameters were categorized into climate, soil properties, and management activities, reflecting the diverse range of factors influencing wheat yield [20].

Among the identified parameters, certain factors emerged as particularly influential in both irrigated and rainfed wheat production. TBS (minimum temperature for plant growth) was identified as a critical parameter affecting wheat yield under both rainfed and irrigation regimes. Other parameters, such as ISW (field capacity/wilting point), PST (pest factor), and various soil-related parameters, also demonstrated significant sensitivity. These results are matched with the findings of Assefa et al. [23], who reported that TBS and ISW are the most sensitive parameters on crop yield especially for dry land or irrigated wheat season. Tadesse et al. [24] also mentioned that the most sensitive parameters on crop yield for the APEX model are related to soil and water. The calibration and validation findings revealed that the APEX model demonstrated excellent performance for both irrigated and rain-fed conditions in the Arba Minch Zuria district. Similar results were reported by earlier studies [24, 36, 37, 43] confirmed the robustness of the APEX crop model in simulating wheat grain yield.

Wheat production has shown significant variation through space and time. The global yield average has increased from 1.09 tha⁻¹ in 1961 to 3.45 tha⁻¹ in 2020 [3]. The grain yield increased from 1.3 tha⁻¹ in 1960 to more than 5 tha⁻¹ in 2000 in China, while it remained low in Africa due to less consumption of mineral fertilizer in the region. Wheat productivity has only slightly increased, from 1.3 tha⁻¹ in 1970 to 2.1 tha⁻¹ in 2014, in Sub-Saharan African countries [44]. However, there are also reports that global wheat production is over 15 tha⁻¹ [45] and yields as low as below 1 tha⁻¹ [4] in certain cases and nations. Global wheat production per hectare varies widely depending on factors such as agricultural practices, climate conditions, water availability, management, and technological advancements. Decreases in area coverage and production in Ethiopia have been observed since the 1960s against the increasing demands [3].

Ethiopia is the 2^nd largest producer of wheat crop in Africa next to Egypt [3, 4, 44]. However, the country imports a substantial amount of wheat to fill the supply-demand gap. Therefore, the government of Ethiopia is looking to have irrigated lowland wheat cultivation to boost wheat production. Ethiopia’s annual wheat production is about 5.8 million tons with a mean productivity of 3 tons per hectare (tha⁻¹) [46], which is relatively lower than the attainable yield of the crop, reaching up to 5 tha⁻¹ [47]. The wheat yields in Ethiopia have been increased from 0.7to 2.97 tha⁻¹ from 1967 to 2020 [4]. This shows that wheat yield increased from the 1970s to 2020, and it was mainly attributed to the introduction and use of organic fertilizers. It can be noted that there are also areas and times where and when wheat yield can go beyond 1 tha⁻¹ [4]. The present study gathered data on the annual yields from each wheat-producing kebeles and in some of the kebeles found that the crop yields were well below 1 tha⁻¹. The significant reduction in yield may be attributed to abiotic stresses, farming practices, climate conditions, availability and accessibility of farm inputs, and crop diseases and pests.

Wheat yield is very sensitive to abiotic stresses such as drought, extreme temperatures (both high and low), salinity, and nutrient deficiency, which can lead to significant yield reduction [48]. The present study area is prone to frequent drought and higher temperature stresses (especially in the lowlands), which might have affected the wheat crop yield. Soil salinity and nutrient deficiency are becoming an alarming problem in the study area as reported by earlier researchers which also contributed to the yield losses [49, 50]. Moreover, it is also anticipated that the traditional farming practices, climate conditions, availability and accessibility of farm inputs, and biotic stresses (crop diseases and pests) might also have affected the crop yield. Farmers in the study are more inclined to other crops than wheat probably due to its lesser yield. The sensitivity of the APEX model also suggested the effect of temperature on the wheat crop. Therefore, it is essential to select the appropriate growing period for the wheat crop in the district, which reduces the effect of abiotic stresses.

The wheat harvest in Arba Minch District, Ethiopia, sheds light on regional agriculture and provides valuable insights into the factors influencing wheat productivity. Yields ranged from 0.3 to 0.9 tons per hectare for rainfed and 0.6–1.1 tons per hectare for irrigated systems, revealing a significant gap compared to other wheat-producing areas. Previous studies also underscored disparities in wheat yields across different regions and periods. For instance, South Wollo and North Gonder reported minimum yields of 1.2 tha⁻¹, while West Arsi reached a maximum of 2.9 tons per hectare during wheat growing seasons from 2009 [4, 51]. Moreover, leading wheat producers in Africa, such as Egypt and South Africa, achieved higher yields ranging from 3 to 4 tha⁻¹ and 4.8 to 6.5 tha⁻¹, respectively [24, 52, 53]. Globally, the average potential wheat yield stands at 8.94 tons per hectare [54].

The primary challenges contributing to lower wheat yields in the study area compared to other wheat-producing regions in Ethiopia, Africa, and globally include variations in soil fertility, water availability, and climatic conditions, encompassing both abiotic and biotic stresses. Additionally, inadequate adoption of fertilizers, improved wheat varieties, agrochemicals, efficient irrigation water management practices, and proper land preparation techniques significantly affect wheat yields in the study district. For instance, ineffective irrigation water management practices in local communities lead to over- or under-irrigation, resulting in decreased yields. Previous studies by Shikur [4], Silva et al. [51], and Netsanet et al. [55] have highlighted wheat production policies and poor coordination in input markets in Ethiopia as key factors influencing productivity differences. Moreover, factors such as water management practices, diseases, pests, climate variability, limited technical knowledge utilization, restricted access to information and agricultural technologies, and the absence of cooperatives contribute to variations in wheat yields in Ethiopia, Africa, and globally [24, 54, 56, 57].

In conclusion, the application of the APEX model simulates both irrigated and rainfed wheat yield in the Arba Minch Zuria district, southern Ethiopia demonstrates successfully its efficacy in various wheat production scenarios. Sensitivity to climate, soil properties, and management practices underscores its versatility. Evaluation through statistical measures shows strong agreement between measured and simulated yields, suggesting the model’s value in assessing and enhancing wheat production. These insights are pivotal for decision-making in agriculture and resource allocation, with potential implications for food security. Future research could explore the model’s utility in assessing climate change impacts, irrigation strategies, and site-specific recommendations for enhancing wheat production practices.

Conflicts of Interest

The authors declare no conflicts of interest.

Funding

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.