Introduction

Maize is the major food, feed, and industrial crop globally, and the leading staple food crop in many developing countries. With 1147 million metric tons (MMT) produced on 193.7 million hectares globally1, maize is the leading cereal in terms of production. However, the average maize yields have declined globally by about 3.8% over the last decades due to climate related shocks2. Maize potential and attainable yields were estimated to reduce by 2.1% and 8.0%, respectively from 1961 to 2009 due to climate change3. Maize yield may see a decline of 24% by late century, with declines becoming apparent by 2030, when the global cereal demand for food and animal feed alone is expected to increase by 50% higher than in 20004 and there is another projection of 60% increase in food demand by 20505. This is mainly due to shifts in rainfall patterns and temperatures regimes arising from increases of CO2, which would probably make growing food crops more difficult to in the tropics.

Climate change increases uncertainty in food production by undermining efforts of mitigating threats to global food security in the 21st century. Previous studies have showed that the negative impacts of climate change on maize yield were mainly associated with the warming and increased drought frequency during the growth periods and reduction of farm lands eventually reducing production and creating yield gaps6,7,8. In Sub-Saharan Africa, gaps between the potential and the actual maize yields are caused by biotic (diseases, weeds, insects and other arthropod pests), abiotic (drought and nutrient deficiencies), or socioeconomic (adoption, input and market) constraints9,10,11 leading to recurring food shortages12. This wider yield gaps should be filled by matching agronomic adaptation options and seed genetic improvement that counterfeit the ongoing climate change and combat food insecurity challenges3,13,14,15.

The likely response of cultivar or management response to the impacts of climate change on crop yield can be determined through the field experimentation and simulation modeling research. A number of simulation models are being used for fast-tracking crop–environment processes, in filling gaps of field experiments, by minimizing the expensive and time-consuming agronomic trials, and for analyzing the yield gaps between the potential and actual productivity of various crops under the changing climate and in exploring both indigenous and modern adaptation responses16,17,18,19,20. Among the models, Decision Support System for Agricultural Technology Transfer (DSSAT) presents an important opportunity for scoping the short-duration field experimental results into the long-term, by using the current, as well as the projected climate and soil information21,22. DSSAT model has been tested and used in the environments of Sub-Saharan Africa including Ethiopia23, calibrated and tested for different maize cultivars9 and recommended for further use by these authors. The cumulative thermal time from seedling emergence to the end of the juvenile phase (P1), and from silking to physiological maturity (P5) are among the most important determinants of maize yield which could be used in yield prediction21,22. Thus, the germination process of maize benefits from abundant solar radiation and an increased temperature within the optimum range of crop’s heat requirement24,25,26. Hence, the DSSAT simulation assumes that reasonable increase of the values of the P1 and P5 would mitigate the negative impacts of future climate change on maize yields, implying that cultivars with longer growing season should be developed in order to cope with heat load and benefit from warmer future climate13.

Planting date adjustments and fertilizer applications conducted following forecasted rainfall are among the most widely studied responses of adapting to climate risks9,13,26,27,28,29. This is because optimal planting dates and fertilizer regimes could offset adverse impacts of climate change by matching crop growth with changing temperature and rainfall distribution. Modifications of mulching or tillage rates and methods have also been used as adaptation options30,31, but their impact on crop production depends on climate32, soil texture, and slope33,34. Mulching would reduce soil water evaporation and run-off35, increase topsoil organic matter and improved near-surface soil aggregate properties36, potentially bearing positive effects on crop productivity37. Khurshid et al.31 also attributed a yield contribution of about 20% to tillage related changes. However, adaptation options like mulching had the opportunity cost of nitrogen immobilization in the short run in that, for every 10 units of carbon sequestered in the soil; there is a need to immobilize one unit of nitrogen38. Similarly both N fertilization and residue retention significantly enhance ammonia (NH3) and nitrous oxide (N2O) emissions39. This calls for splitting application of N fertilizer to improve the temporal synchronicity between crop N demand and soil N availability, thereby increasing crop yield and reducing residual soil nitrogen and its environment risk26,40. Hence, there is a need to consider fertilizer rating and timing to meet the nutrient demands of crop growth and avoid nutrient loss when crop biomass decreases due to water and temperature stress over time under climate risks13,15. Thus, adaptation options have their own management requirements that vary with local natural resource endowment rendering them location specific, which in turn calls for optimization of adaptation options across different agro-ecologies.

Southern central Rift Valley of Ethiopia is one of the most vulnerable regions to climate risks, where increased temperatures and changes in rainfall patterns have been altering the evapotranspiration rate and reducing water availability for maize crop41,42. Moreover, soil conditions have been changing due to the associated land degradation processes over the years and hence the old management recommendations may not be efficient, thus making generation of new adaptation options a necessity. Hence, quantifying climate related risks and adapting agriculture to the changing climate is inevitability daunting challenge in the study area. Thus, exploring effective adaptation strategies and measuring their effect on the performance of maize crop is crucial to improving maize production under future climate scenarios. However, past climate change studies carried out in Ethiopia have been limited to assessment of impacts of crop production without accounting for potential adaptation measures43,44. Moreover, information on climate related yield variability, climatic yield potentials and the magnitude of yield gaps due to nutrient and water limitation, and their associated management was scanty9. In addition, the difference between maize yield potential and the actual yield varies from place to place. Therefore, the objective of this study was (1) to analyze existing yield gaps using DSSAT model in the southern central Rift Valley of Ethiopia, and (2) to explore potential adaptation strategies and develop agronomic recommendations that reduce the negative impacts of future climate risks on maize yield in the study areas.

Materials and methods

Description of the study area

This study was carried out in southern plains of central Rift Valley of Ethiopia located between 6.38 latitudes in Dilla area (South) to 7.72 latitudes in Bulbula area (North). The Western margin corresponds with 37.75 longitudes (Wolita sodo) to 38.68 latitudes in Woteraresa (East). This covers an area of 1,021,332 ha (Fig. 1). The study area has been grouped using prnicipal component and cluster analysis of physiographic, climatic and topographic attributes into four distinct clusters each having a number of districts represnting the area as shown in Markos et al.45.

Figure 1
figure 1

The study area map with the four distinct clusters (Map of the study area with four clusters was created by using free version of QGIS version 3.32, https://www.qgis.org/en/site/forusers/download.html/).

Full size image

Historic climate

The long term average annual rainfall for Shamana, Bilate, Hawassa and Dilla clusters is 972.0 ± 97.8 mm, 758.7 ± 125.8 mm, 953.9 ± 57.5 mm and 1298.2 ± 70.4 mm42, respectively. About 33.0 and 42.9% of the average annual rainfall in Shamana cluster is received during ONDJ and JJAS period, respectively. About 36.0 and 37.2% of the average annual rainfall in Bilate cluster is received during FMAM and JJAS period, respectively. About 33.0 and 42.9% of the average annual rainfall in Hawassa cluster is received during ONDJ and JJAS period, respectively. About 35.3 and 33.3% of the average annual rainfall in Dilla cluster is received during FMAM and JJAS period, respectively45. Shamana cluster have a mean minimum and maximum temperature of 15.5 and 26.4 °C, respectively. Similarly Bilate, Hawassa and Dilla clusters have mean minimum and maximum temperature of 19.3 and 28.4 °C, 14.2 and 27.1 °C and 15.8 and 28.0 °C, in respective areas42.

Soil

In Shamana cluster, soils are developed from volcanic sedimentary to lacustrine deposits and have relatively higher clay and dominantly belong to Vertic Luvisols, Eutric Cambisols, and Chromic Vertisols based on FAO soil classification45. In Bilate cluster, the most dominant soil is riverine or lacustrine alluvium derived from basalt, ignimbrite, lava, or ash with dark brown loamy coarse sands and sandy loams with often calcareous subsoil, and belong to Calcaric Fluvisol, Orthic Andosol, Ortic Phaeozems, and Chromic Luivisols. In Hawassa cluster, the dominant soils are Vitric Andosols, Eutric Cambisols, and Leptosols. The soils had lacustrine and pyroclastic deposits of sands and silts, interbedded with pumice. In Dilla cluster the soils are originated from basalt rocks, and had deep red-dish to brown clayey to clay loam texture classified as Haplic Luvisols and Chromic Vertisols. Low bulk density and weak structure of soils in Bilate and Hawassa clusters render them vulnerable to erosion even on gentle slopes compared with those in Shamana and Dilla clusters (Markos et al.57). The soils of study area at Shamana, Bilate, Hawassa and Dilla had organic carbon content of 3.99, 3.61, 4.33 and 4.96%, respectively.

Farming system

The Shamana cluster stands for high-land transitional moist areas with upper mid-altitude category46 or highland sub-humid areas47. The Shamana cluster had enset/maize based potato farming system in Shamana, maize based tef farming system in Shone and maize based faba bean farming system in Bitena areas. The cluster is characterized by intermediate to high altitudes and cooler temperatures allowing December–January planting of maize, and double cropping of the second crop usually starting in July. The crop residues have been used for cattle feed. The Bilate cluster is located in the valley bottom of the rift valley attaining the lowest altitude and hottest temperature among the clusters; sorghum/finger millet/maize production begins in April–May period in this cluster. The Bilate cluster represented maize production agro-ecology with large number of livestock termed as semiarid lowlands of central Rift Valley or dry mid-altitudes46 and low moisture areas47. Currently, the Hawassa cluster is known for rain-fed maize-based production systems and modest livestock rearing9. It possesses intermediate altitude and dominant maize cropping allowing common beans for intercropping in most cases. The Hawassa cluster was one of the sub-moist mid altitude areas in central Rift Valley which is also called mid-altitude sub-humid area47. Finally low to mid- altitude areas adjacent to mountain chains of Sidama and Gedeo, also known for indigenous coffee, enset and livestock based agroforestry practices, were grouped into Dilla cluster. The Dilla cluster is represented as humid tropics48, moist low to mid-altitude areas46, and as lowland sub-humid areas47.

Data sources

Average farmers’ yield representing farmer managed maize plots was obtained by using questioniare interview from 60 farmers per district representing a given cluster. This has been added and then divided to obtain the average average farmers yield per cluster. Furthermore, the obtained yield has been trianglated with reports of district offices of agriculture for each cluster. Next on farm trial yields, maize yield from researcher designed researcher managed trials, were collected from reports of researchers49,50,51,52,53,54,55,56. The model was calibrated using historic weather data of thirty years (1991–2020), field measured values of crop parameters and soil properties during 2011–2017 cropping season. Thus, calibration was done using experimental data for the period between 2011 and 2017 whereas validation was done using experimental data for the period between 2016 and 2018 period. Using genetic coefficients already calibrated and evaluated for hybrid maize variety Shone57, and current and future climate data downscaled from contrasting multimodel ensembles of 20 Global Climate Models (GCMs) over Representative Concentration Pathways (RCP) 4.5 and 8.558, maize yields were anticipated. Thus, the water-limited potential yield for each cluster was obtained through simulation using DSSAT 4.8, where water was switched off under baseline (current) scenareo using the hybrid maize variety (Shone) and commonly practiced management practices. Under future climate scenarios, the maangment practices were alttered as designed in the adaptation strategies section of this manuscript and the water remained switched off for the same hybrid maize variety.

Yield gap (Yg) analysis

Yield gap analysis was performed at three levels following procedures established by Kassie et al.59 as shown below in Eqs. (1, 2, 3):

$${\text{Yield }}\;{\text{gap }}\,{\text{I}}\; = \;Yb - {\text{Ya}}$$
(1)
$${\text{Yield }}\;{\text{gap}}\;{\text{ II}}\; = \;Yc - {\text{Yb}}$$
(2)
$${\text{Yield }}\;{\text{gap}}\;{\text{ III}}\; = \;Yc - {\text{Ya}}$$
(3)

where Yield Gap I—Gap between simulated water-limited potential yield and average farmers’ yield.

Yield Gap II—Gap between water-limited potential yield and on-farm tal yield.

Yld Gap III—Gap between on-farm trial yield and average farmers yield.

Yb = Simulated water limited potential yield.

Ya = Average farmers’ yield.

Yc = On-farm trial yield.

Adaptation strategies

Four adaptation options each having different levels of strategies were explored as in Table 1.

Table 1 Adaptation strategies used in the simulated experiment.
Full size table

Analysis of variance (ANOVA)

The output of DSSAT model was statistically analyzed using the analysis of variance (ANOVA) technique to evaluate the impact of climate change on maize production. Accordingly the four management and their levels collectively 54 management strategies (three nitrogen rates × three planting dates × three mulch rates × two tillages) were compared using Randomized Complete Block Design in factorial combination of the treatments. The year effect, which has 30 levels, was used as replications (blocks) as in the DSSAT software because the maize yield in 1 year under a given treatment was not affected by another year. Since each simulation year had unpredictable weather conditions, the soil organic carbon and moisture were reset each year, formal randomization of simulation years was not needed16,60. The ANOVA was calculated using the SAS software package and treatments averages were separated using least significance difference (LSD) at 5% level of probability wherever difference between averages exist.

Estimation of yield changes under different scenarios

The biophysical impact of climate change across southern central rift valley of Ethiopia was obtained by calculating changes in crop yield between the current and the proposed adaptation options for hybrid maize variety called Shone. Then the changes were calculated for each of early, mid and late century period relative to the current practice for the respective GHGs emission scenarios. Accordingly impacts of adaptation options were computed based on61 given in Eq. (4) as shown below.

$${\text{Yield }}\;{\text{change }}\left( {\text{\% }} \right) = \frac{{\left( {{\text{Yanticipated}} - {\text{Ycontrol}}} \right){*}100}}{{{\text{Ycontrol}}}}$$
(4)

where Y anticipated—simulated yield under new adaptation option.

Y control—simulated yield under control.

Ethics approval and consent

Hawassa University hosts graduate students from higher educational institutions and agricultural research centers, who are publicly allowed to gather, analyze and report local data to global scientific communities. All methods of experimental research and field studies used in this study comply with relevant institutional, national, and international guidelines and legislations. The study has been exempted from the requirements of the human subject protections regulations.

Results

Average farmers’ yield and yield gaps

The result showed that the average of thirteen years farmers’ average yield was high at Hawassa (5 t ha−1) and Shamana (4.2 t ha−1), but low at Bilate (3.4 t ha−1) and Dilla (3.2 t ha−1) (Fig. 2). The standard deviations of maize yield were highest at Bilate depicting higher variability across years, which is manifestation of environmental vulnerability of the cluster. This showed that the locations Hawassa and Shamana had high potential for maize production compared with those of Dilla and Bilate, which could be due to climatic and agro-ecological suitability of the clusters.

Figure 2
figure 2

Box and Whisker plot of average farmers’ yield (t ha−1) across the four clusters.

Full size image

The average actual maize yield (2.9–5 t ha−1) of districts located within a given cluster formed yearly average farmers’ yield, which is the lowest compared with experimental (5.1–7.3 t ha−1) and water limited potential yield (6.3–10.6 t ha−1) (Table 2). Farmers' yield is 46–54% of on-farm experimental yield and in turn on farm experimental yield was 23.5–45.2% of the water limited potential yield in southern central rift valley of Ethiopia, which was twice the gap between research and farmers’ average yield. Farmers’ yield was 46–54% of on-farm experimental yield mainly due to lack of improved seeds/fertilizers, lack of credit, poor extension services, and poor field management for weeds, insect pests and diseases. This calls for integrated crop management, credit to farmers, timely supply of inputs, research and extension collaboration to transfer the new technologies trough training, demonstrations, field visits and advice by extension agents etc. On-farm trial yield (5.1–7.3 t ha−1) lied in between farmers’ average yield (2.9–5 t ha−1) and water limited potential yield (6.3–10.6 t ha−1), which is attributed to higher use of fertilizers, intensive weed, insect pest and disease control. This agrees with previous findings of62 that states average yields in rain-fed systems are commonly 50% or less of yield potential, suggesting ample room for improvement, However, the potential yields were lower than those reported by some authors63,64, who stated 12.5 t ha−1 simulated potential yield for maize yield in Ethiopia. The yield obtained from experimental field is constrained by rainfall and soil water characteristics, and can be enhanced through use of increased investment in agricultural research infrastructure including molecular laboratories, tissue culture facilities, irrigation structures, nut tunnels, screen houses, etc), policy alignment with research and development to enhance technology generation and transfer and skilled research personnel. The water limited potential yield reported in this study was lower than previous report by other scholars due to increased uncertainty in growing season weather in the study clusters, and should be enhanced by increasing genetic yield potential, using heat tolerance and high harvest index cultivars, which requires breakthroughs in genetic improvement research. The result of this study is in agreement with those of Kassie et al.59 who elaborated actual yields to be 28–30% of the water-limited potential yield and 44–65% of on-farm trial yields in relative terms while the on-farm trial yields were 50–73% of the water-limited potential yield.

Table 2 Comparison of farmers yield, experimental and potential yield in the study area.
Full size table

Analysis of variance for projected maize yields against adaptation options

The mean square of error (MSE) values depicted that use of mulch (P < 0.001), nitrogen (P < 0.001), and tillage × planting date (P < 0.001) affected maize yield significantly during baseline scenario at Shamana cluster (Table 3). The same trend was projected across medium and high emission scenarios during early, medium and late century for these practices at Shamana. The interactions of tillage × planting date × nitrogen or mulching would affect maize yield significantly (P < 0.05) across high emission scenario during early, mid and late century at Shamana. At Bilate cluster, use of mulch (P < 0.001), tillage × nitrogen × planting date (P < 0.001), and tillage × mulching × planting date (P < 0.001) affected maize yield significantly during baseline scenario (Table 3). This trend was maintained across medium and high emission scenarios during early, medium and late century at Bilate. Moreover, use of nitrogen affected maize yield significantly (P < 0.05) across RCP 4.5 and RCP 8.5 during mid and late century at Bilate. Lack of response of maize yield to application of nitrogen at Bilate to during baseline and early century could be attributed to sub-optimal moisture in the cluster.

Table 3 MSE for grain yield across medium and high GHGs emission scenarios during early, medium and late-century period.
Full size table

At Hawassa cluster, use of mulch (P < 0.001) and nitrogen (P < 0.001) affected maize yield significantly during baseline scenario (Table 3). The same trend was maintained across medium and high emission scenarios during early, medium and late century. During mid and late century, the interaction of tillage × nitrogen × planting date or tillage × mulching × planting date were projected to affect maize yield significantly (P < 0.05) across RCP 4.5 and RCP 8.5 at Hawassa. At Dilla cluster, use of mulch (P < 0.001) and nitrogen (P < 0.001) affected maize yield significantly during baseline scenario (Table 3). This trend was projected to continue across medium and high emission scenarios during early, medium and late century. Across RCP 4.5 and 8.5, interaction of tillage × nitrogen × planting date or tillage × mulching × planting date was projected to affect maize yield significantly (P < 0.05) during early, mid and late century at Dilla.

Trends of yield change in response to adaptation strategies

Tillage as adaptation options

The projection of maize yield (8.5 t ha−1) is highest at Hawassa due to conservation tillage across medium GHGs emission scenario during early century. The lowest grain yield of 3.1 t ha−1 was projected in Dilla due to 30 cm till or conventional tillage across high GHGs emission scenario during late century. Considering baseline scenario, conservation tillage is projected to increase maize yield by 1.4% in Shamana and Dilla clusters. No yield change is projected at Bilate and Hawassa under respective baseline scenarios. Maize grain yield increased by 5.7 and 2% during early and mid-century, respectively due to conservation tillage under medium GHGs emission scenarios at Shamana compared with conventional tillage (Figs. 3 and 4). However, across high GHGs emission scenarios during early, medium and late-century, and medium GHGs emission scenario during late-century changing tillage changes to conservation tillage will not be able to counteract the negative effects of climate change on maize crop at Shamana.

Figure 3
figure 3

Change in grain yield (%) of shone variety in response to conservation tillage relative to traditional tillage during early, mid and late-century across baseline, RCP 4.5 and 8.5 emission scenario at Shamana, Bilate, Hawassa and Dilla clusters, southern central Ethiopia.

Full size image
Figure 4
figure 4

Maize yields (t ha−1) and error bars with standard deviation due to change in tillage practices under medium and high GHGs during early, mid and late century period across Shamana, Bilate, Hawassa and Dilla clusters.

Full size image

At Bilate, conservation agriculture produced 1.5 and 3.1% higher yields compared with conventional tillage under medium GHGs emission scenario during early and medium-century. Moreover, there will be yield advantage of 1.4 and 1.8% under high GHGs emission scenario during early and mid-century, respectively. During late-century, there was not any advantage due to alterations in tillage at Bilate. At Hawassa, only 1.2 and 1.5% yield increments were anticipated due to conservation tillage under medium and high GHGs emission during early and mid-century, respectively. There was 2% yield decline due to conservation tillage compared with conventional tillage at Hawassa under high GHGs emission scenario during late-century. At Dilla, 1.6% grain yield advantage was foreseen due to conservation tillage compared with conventional tillage under medium GHGs emission scenario during mid-century period. However, changing tillage practices from conventional to conservation resulted in 1.5 and 2.3 yield decrease under high GHGs emission scenarios during early and mid-century, respectively. The yield decline due to conservation tillage was also 1.6% under medium GHGs emission scenario during late-century at Dilla.

Planting dates as adaptation response

The projection of maize yield (8.5 t ha−1) is highest at Hawassa due to April planting across medium GHGs emission scenario during early century. The lowest grain yield of 3.1 t ha−1 is projected in Dilla due to February planting across high GHGs emission scenario during late century. Shifting planting window backwards (eg. from January to December in Shamana, from May to April in Bilate, from April to March in Hawassa or from March to February in Dilla) did not show any yield advantage under baseline conditions. Thus, an early sowing date wouldn’t be conducive for maize production under baseline scenario. However, shifting planting window from January to February would present 1.4% yield advantage under baseline conditions in Shamana cluster (Figs. 5 and 6). Shifting planting dates from January to December resulted in 1.8 and 2% grain yield decline under medium GHGs emission scenarios during early and mid-century period, respectively in this location. Conversely, shifting planting period to February would result in 2% yield advantage under medium GHGs emission scenario during mid-century period in Shamana.

Figure 5
figure 5

Change in grain yield (%) of shone variety in response to planting date relative to control during early, mid and late-century across baseline, RCP 4.5 and 8.5 emission scenario at Shamana, Bilate, Hawassa and Dilla clusters, southern central Ethiopia.

Full size image
Figure 6
figure 6

Maize yields (t ha−1) and error bars with standard deviation due to change in planting date under medium and high GHGs during early, mid and late century period across Shamana, Bilate, Hawassa and Dilla clusters.

Full size image

At Bilate, shifting planting window from May to June would result in 1.5% grain yield penalty under medium GHGs emission scenario during early century period. However, there would be 1.4% yield increase under GHGs high emission scenarios due to the same forward shifting of planting window for the same duration. Shifting planting window forward or backward did not show any difference under medium and high GHGs emission scenarios during mid-century period. However April planting resulted with 2.2% yield decline under high GHGs emission scenario during late century period. Thus June planting would help to adapt climate change under high GHGs emission scenario during early century at Bilate with 1.4% yield advantage over the commonly practiced May planting. At Hawassa and Dilla, the DSSAT model did not simulate any gains or losses in grain yield due to changes in planting window either under medium or high GHGs emission scenarios during early, mid and late century periods. Delaying planting was useful under baseline and medium GHGs emission scenarios during early century at Shamana, and under high climate emission scenario during early century at Bilate. This is mainly because delayed planting avoids the high-risk periods of heat and drought stress in the maize growth period. This agrees with some authors who attributed the advantage of delayed planting to increased precipitation during late July to early August, which resulted in higher pollination rates and kernel numbers without water stress65,66. However, delayed planting previously recommended by Disasa and Yan67 for Hawassa cluster was not validated with multiple climate ensembles considered in this study.

Nitrogen as adaptation response

The projection of maize yield (8.5  t ha−1) is highest at Hawassa due to 64 kg ha−1 nitrogen across medium and high GHGs emission scenario during early century. The lowest grain yield of 2.9 t ha−1 was projected in Dilla due to 128 kg ha−1 nitrogen across high GHGs emission scenario during late century. Considering yield changes, increasing nitrogen rate from 0 to 64 kg ha−1 resulted in 7.1, 7.8 and 6.1% increase of maize grain yield under medium GHGs emission scenario during early, medium and late century period, respectively compared with the control (0 kg ha−1 N) at Shamana. Increasing nitrogen rate further to 128 kg ha−1 resulted with 5.5, 7.8 and 6.4% increases in maize yield for the same period and GHGs emission scenario (Figs. 7 and 8). Thus, increasing nitrogen rate increased maize grain yield under baseline and medium GHGs emission scenario conditions during early, mid-term and late century period. However, under high GHGs emission scenarios, increasing nitrogen rate resulted in yield loss compared with control in early and late century period in Shamana.

Figure 7
figure 7

Change in grain yield (%) of shone variety in response to nitrogen application relative to control during early, mid and late-century across baseline, RCP 4.5 and 8.5 emission scenario at Shamana, Bilate, Hawassa and Dilla clusters, southern central Ethiopia.

Full size image
Figure 8
figure 8

Maize yields (t ha−1) and error bars with standard deviation due to change in nitrogen rate under medium and high GHGs during early, mid and late century period across Shamana, Bilate, Hawassa and Dilla clusters.

Full size image

At Hawassa, maize yield increased by 3.4% each when application rates are increased from 0 to 64 and 128 kg ha−1 during the baseline period. However, maize yields showed a decreasing trend during early, medium and late century period under medium and high GHGs emission scenarios. This depicts the vulnerability of the Hawassa cluster for small changes in emission of greenhouse gases. At Dilla cluster, increasing nitrogen rate from 0 to 64 and 0 to 128 increases maize yield by 1.4 and 5.3%, respectively during the baseline period. During early century period, increasing nitrogen application from 0 to 64 and 0 to 128 kg ha−1 increases maize yield by 4.5% each during medium GHGs emission scenario. At Dilla, increasing nitrogen application from 0 to 64 and 0 to 128 increases maize yield by 3.1% and 8.7%, respectively under medium GHGs emission scenario during early century period. The simulation result showed that increasing nitrogen application from 0 to 64 and 0 to 128 kg ha−1 results in yield disadvantage under high and medium GHGs emission scenarios during both mid and late century periods at Dilla cluster.

Mulching rate as adaptation response

The projection of maize yield (12.1 t ha−1) is highest at Dilla due to mulching maize field with 5 t ha−1 across baseline scenario (Fig. 9). Considering yield changes, increasing mulching rate from 0 to 10 t ha−1 reduced maize yield under baseline, medium and high GHGs emission scenarios during early and mid-century at Shamana (Fig. 10). Moreover, maize yields increased by 2 and 4.3% under medium and high GHGs emission scenario, respectively due to application of 5 t ha−1 during late century period at Shamana (Figs. 9 and 10). Similarly, maize yields increased by 0.9 and 4.3% under medium and high GHGs emission scenario due to application of 10 t ha−1 during late century period at Shamana. At Bilate, increasing mulch rate to 5 t ha−1 increased maize yield by 4.2 and 3.7% during baseline and late century period under baseline and high GHGs emission scenarios, respectively. However, increasing mulch rate to 5 t ha−1 would result in negative yield change under high GHGs emission scenario during mid and late century period at Bilate. Further increase of mulching rate to 10 t ha−1 would inflict reduction in maize gain yield by 5.6, 7.8 and 6.3% in Bilate cluster during early, mid and late century periods, respectively under medium GHGs emission scenario. When mulching rate was increased from 0 to 10 t ha−1, maize yield would increase by 7.1 and 7.3% compared with baseline during early and late century period, respectively under high emission scenario in Bilate, but maize yield would decrease by 8% during mid-century period under medium GHGs emission scenario.

Figure 9
figure 9

Maize yields (t ha−1) and error bars with standard deviation due to change in mulching rate under medium and high GHGs during early, mid and late century period across Shamana, Bilate, Hawassa and Dilla clusters.

Full size image
Figure 10
figure 10

Change in grain yield (%) of shone variety in response to mulching relative to control during early, mid and late-century across baseline, RCP 4.5 and 8.5 emission scenario at Shamana, Bilate, Hawassa and Dilla clusters, southern central Ethiopia.

Full size image

At Hawassa, increasing mulching rate from 0 to 5 t ha−1 increased maize yield by 5.5 and 4.3% under medium and high GHGs emission scenarios, respectively during early century period. Further increase of mulch to 10 t ha−1 resulted in 1.4 and 0% yield increase under medium and high GHGs emission scenarios in the same period. At Dilla, simulation of mulch application resulted in yield loss during baseline and early century period. For instance, application of 5 t ha−1 mulch resulted in decrease of grain yield of maize by 3.6 and 5.8% under medium and high GHGs emission scenarios, respectively during early century. However, application of 5 t ha−1 mulch resulted in 5 and 4.8% yield increases under high GHGs emission scenario during mid and late century, respectively. Similarly, application of 5 t ha−1 mulch resulted in 1.1 and 1.2% yield increases under medium GHGs emission scenario during mid and late century.

Simulation of integrated adaptation response

Results of combination of adaptation options including planting early in December with conservation tillage and use of 64 kg ha−1 N provided 3.5% yield advantage over the control under medium GHGs emission scenario during mid-century period at Shamana (Fig. 11). Similarly, delaying planting to February, use of 64 kg ha−1 N and employing conservation tillage provides yield advantage of 5.8% over the control under medium GHGs emission scenario during mid-century period at Shamana. Conversely, delaying planting to February, use of high nitrogen rate (128 kg ha−1 N) and use of conventional tillage results in yield decline of 4.8% compared with the control. In Bilate, use of 128 kg ha−1 N resulted in yield change of − 1.1 to 8.4% compared with the control regardless of tillage and planting date adjustments. A combination of conservation tillage, April planting and 128 kg ha−1 N produced yield advantage of 1.9% compared with the control at Hawassa. If tillage is altered to conventional, then May planting suits as it renders 2.2% yield advantage in Hawassa under medium GHGs emission scenario during mid-century (Fig. 11). At Dilla, use of conservation tillage and 128 kg ha−1 N results in 6.5% yield advantage if planting date would be shifted from March to April. The advantage of applied adaptation options would be low in Bilate and Hawassa clusters during mid-century period which could be mainly due to recurrent moisture stress projected to occur in the clusters. Mulching 5 t ha−1 was projected to produce 4–5% yield advantage in Hawassa cluster during mid-century period regardless of changes in tillage or planting window (Fig. 12). Under high GHGs emission scenario during mid-century period, over 13.4% yield advantage was projected in Bilate cluster due to conservation tillage and late planting, which is June (Fig. 13).

Figure 11
figure 11

Percentage yield change from baseline in response to nitrogen, tillage and planting dates during mid-century across medium GHGs emission scenario for Shamana, Bilate, Hawassa and Dilla clusters.

Full size image
Figure 12
figure 12

Percentage yield change from baseline in response to mulching, tillage and planting dates during mid-century across medium GHGs emission scenario for Shamana, Bilate, Hawassa and Dilla clusters.

Full size image
Figure 13
figure 13

Percentage yield change from baseline in response to nitrogen, tillage and planting dates during mid-century across medium GHGs emission scenario for Shamana, Bilate, Hawassa and Dilla clusters.

Full size image

Intigrated use of conservation tillage, 5–10 t ha−1 mulch and April planting produced over 11% yield advantage in Bilate across high GHGs emission scenarios during mid-century period. Similar yield advantage could be obtained if tillage remained conventional and planting maintained in May so long as 5–10 t ha−1 mulch retained in the soil at Bilate (Fig. 14). In Dilla cluster, use of 10 t ha−1 mulch and conservation tillage and early planting (February) would result about 1.8% yield advantage compared with the control either in medium or high GHGs emission scenarios.

Figure 14
figure 14

Percentage yield change of maize from baseline in response to mulching, tilage and planting dates during mid-century across medium GHGs emission scenario for Shamana, Bilate, Hawassa and Dilla clusters.

Full size image

Discussion

Considering a time period most relevant to large agricultural investments, which typically take 15–30 years to realize full returns68, impacts of adaptation strategies were closely looked at mid-century period in this study. During mid-century period, delaying planting to February, use of 64 kg ha−1 N and employing conservation tillage would provide yield advantage of 5.8% over the control under medium GHGs emission scenario at Shamana. At Bilate, conservation tillage and June planting produced over 13.4% yield advantage under high GHGs emission scenario during the same period. At Hawassa, mulching 5 t ha−1 is projected to produce 4–5% yield advantage during mid-century period regardless of changes in tillage or planting window. At Dilla, use of conservation tillage and 128 kg ha−1 N would result in 6.5% yield advantage if planting date would be shifted from March to April. The lowly performance of conventional tillage was attributed to moisture loss and erosion hazards particularly in Bilate and Hawassa clusters69 whereas better performance from conservation agriculture was due to conserving soil structure, storing soil carbon and nutrient which would ultimate;y enhance water retention70. Overall, future climate is expected to have a negative impact on maize yield as yields still go down with adaptations but the decrease is about less compared with conditions without adaptation strategies. This manifests the urgent need to adopt moisture conserving managment practices in all study areas.

Use of mulch and nitrogen has been proven responsive in highland Shamana cluster, mid-land Hawassa cluster and also in high rainfall receiving Dilla cluster across RCP 4.5 and RCP 8.5 during baseline and early century. The nutritional role of nitrogen on yield enhancement under future climate couldn’t be over emphasized, but the beneficial role of mulches could be due to enhancement of soil physico-chemical properties, suppression of weeds and reduction of evapo-transpiration that would ultimately lead to better maize productivity in dry land agriculture71. Conversely, lack of response to mulch application could be observed occasionally due to extensively low soil moisture and low nitrogen fertility as in Bilate cluster, and the humid climate of Dilla cluster where slow decomposition of already available mulch layers occur72. The variability in yield improvements due to nitrogen applications observed in the current study across clusters agreed with findings of Liu and Basso73 that attributed spatial variability in response of nitrogen strategies to differences in weather, soil biophysical and chemical properties of soils, and their interactions with management.

The climate in baseline and early century is projected to reward a better grain yield due to soil mulching. Similarly, increasing nitrogen rate increased maize yield under medium and high GHGs emission scenario across early and mid-century period. During late century period, positive increases from applied nitrogen were simulated only for moderate GHGs emission scenario and not for high emission scenario. The yield loss anticipated across most clusters under high GHGs emissions were attributed to likely enhanced rate of dvelopment and pollination due to high temperature stress under future climate74. According to this authors, high temperatures lessens the pollen viability, fertilization, and grain creation in maize74. This has been excerbated by frequent low moisture regimes which subsequently reduces nutrient uptake and response to nitrogen applications. Thus, despite benefiting from elevated (CO2) with increased use of N fertilizers, maize plants will be increasingly negatively impacted as temperature increases and/or if rainfall decreases under high GHGs emission scenarios. Previous research also showed both retention of harvest residue and N fertilization under RCP 8.5 would enhance maize yield39; however, also reported that increased CO2 concentration would enhance the labile C input to the soil, which may increase N2O and NH3 emissions through encouraged microbial activity75, which would probably hamper productivity. Hence the yield penalty remain high because the benefits of elevated CO2 were unlikely to offset the negative impacts of increasing temperature and variable rainfall.This calls for controlling GHGs emissions to enhance maize production and obtain maximum rewards from investment in adaptation strategies.

The current study calls for changing conventional tillage to conservation tillage in Shamana and Dilla clusters during baseline scenarios, and also starting early century at Bilate and Hawassa clusters. Changing tillage from conventional to conservation was found beneficial across most clusters under present and future climate scenario which agreed with Hemmat and Eskandari76 and Kephe et al.16 who attributed improvement in crop establishment, growth and yield due to enhanced soil water content under conservation tillage. Dong et al.77 associated higher yields under conservation tillage due to improvement of soil organic carbon and crop water use efficiency. Although the DSSAT model did not simulate any gains or losses in grain yield due to delayed planting at Hawassa and Dilla, there were positive yield gains in Bilate and Shamana. This is mainly because delayed planting avoids the high-risk periods of heat and drought stress in the growth period of the crop. This agreed with some authors who attributed the advantage of delayed planting to increased precipitation during late July to early August, which resulted in higher pollination rates and kernel numbers65,66. However, delayed planting previously recommended by Disasa and Yan67 for Hawassa cluster was not validated with multiple climate ensembles considered in this study.

The positive yield gains observed with 128 kg ha−1 nitrogen in Shamana, Bilate and Dilla under baseline and medium GHGs emission scenarios were in agreement with findings of He who recommended 150 kg N ha−1 in Canada based on the DSSAT model13 and Kassie et al.59 that indicated increasing nitrogen fertilizer rate of 60 kg ha−1 to increase maize yield by 78–89% in Ethiopia. The yield loss observed at Hawassa due to increased nitrogen use under future climate scenarios could be due to high nitrogen application despite low nutrient use efficiency leading to resources wastage, greenhouse gas emissions and nitrate leaching13 and has been attributed to the lower soil water regime after fertilization75. The current study also proves findings of Smith et al.15 who disclosed yield loss across high emission scenarios during late century despite use of alternative adaptation options.

The current study showed that most promising and least risky practices among simulated strategies would be enhancing integrated use of nitrogen and mulching across the clusters in combination with either tillage or planting date options. However, planting date and tillage interventions remained least promising and highly risky practices if not integrated with residue or nitrogen in order to adopt climate change in the studied clusters. Hence, nitrogen fertilization and residue mulching require suitable moisture regimes prior application. This necessitates integration of nitrogen or mulches with planting date or tillage in maize fields, which is vital to reduce higher emission of green house gases. Moreover, this study is limited to biophysical impact of only four adaptation options, and didn’t include bio-economic analysis. Similarly adaptation responses related to rotation, water harvesting, irrigation, phosphorus application or modification of cropping systems were not addressed. To generate other effective adaptation options that reflect local conditions, there is a need to understand the characteristics of the maize crop under simultaneous and relay intercropping with N-fixation legume cover crops and split applications of mineral fertilizer, which may be important in high rainfall cluster of Dilla as nitrate leaching could lead to low response of applied nitrogen. In low rainfall and warm clusters of Bilate and Hawassa, drought tolerant maize varieties, water-harvesting technologies (stone lines, tied ridges, zaï pits and contour ridges), and in low laying Bilate cluster, cultivars adapting heat stress may help reduce production losses. For all the clusters, accounting for P stress may help to reduce model uncertainty as maize crop responds to applied P in all clusters except Hawassa. There is a need also to explore supplemental irrigation as one of adaptation strategies where water can be availed through alternative sources particularly during critical periods of moisture stress to the crop. If the growth time of maize from silking to maturity could be prolonged and the accumulated temperature could be raised, which requires sufficient moisture during his stage, the dry matter accumulation of maize would effectively increase, which would have an obvious effect on maize yield.

Summary and conclusion

The present investigation is undertaken to explore effects of adaptation responses including tillage, mulching, planting dates and nitrogen on productivity of maize variety using DSSAT model v 4.8. Impacts and adaptation options were evaluated using projections by 20 coupled ensemble climate models under two representative concentration pathways (RCPs) 4.5 and 8.5. It was found that farmers' yield is 46–54% of on-farm experimental yield, and in turn on farm experimental yield was 23.5–45.2% of the water limited potential yield in southern central rift valley of Ethiopia leaving 1.9 to 3.1 t ha−1 gap between research and farmers’ yield and 3.4–5.6 t ha−1 gap between water limited potential yield and farmers’ yield. Under RCP 4.5, maize grain yields changed by 1.4–5.7%, 1.2–3.1%, 0–1.2% and 1.4–1.6 due to conservation tillage at Shamana, Bilate, Hawassa and Dilla clusters, respectively. However under RCP 8.5, maize grain yields changed by 0%, 1.4 to 1.5%, −2 to 1.5% and − 2.3 to 1.6 due to conservation tillage in the same locations. Thus, there is reduced response by maize yield for tillage change at high emission scenario across most clusters. Changing planting dates forward (from January to February) changes maize yield by − 2 to 2% and 0 to 1.4% at Shamana during RCP 4.5 and baseline scenario, respectively. Delaying planting dates by a month (changing from January to December) increased maize yield by 0 to 1.8% across RCP4.5 at Shamana. Thus under high emission scenario, delayed planting would act as adaptation option but early planting did not at least in the study clusters. Planting date changes did not affect maize yield significantly (P < 0.05) at Hawassa and Dilla clusters under future climate.

Maize grain yields changed by − 8.5 to 17.5%, − 4.5 to 5.1%, − 2.3 to 7.1% and − 9.8 to 7.1% due to increased use of nitrogen compared with control (0 t ha−1) at Shamana, Bilate, Hawassa and Dilla clusters, respectively. Thus, better response is projected across highland and mid-altitude clusters of Shamana and Hawassa compared to low land clusters of Bilate and Dilla thereby reducing vulnerability of maize production to risks related to climate change in former agro-ecologies. Yield loss from application of nitrogen were evident across RCP 8.5 scenario which was mainly due to the lower soil water regime after fertilizer application. Thus, under high GHGs emission scenarios, increasing nitrogen rate increased risk of yield decline compared with control in early and late century, which necessitates controlling emissions to enhance maize production and maximize benefit from applied nitrogen fertilizers. At Hawassa, increasing mulching rate from 0 to 5 t ha−1 increased maize yield by 5.5 and 4.3% under medium and high emission GHGs scenarios, respectively during early century period. Moreover, application of 5 t ha−1 mulch resulted in 5 and 4.8% yield increases under high GHGs emission scenario during mid and late century, respectively at Dilla. Generally, maize grain yields changed by − 6.3 to 4.3%, − 15 to 7.7%, − 1.6 to 8.5% and − 5.8 to 4.8 due to the use of mulches compared with control at Shamana, Bilate, Hawassa and Dilla clusters, respectively. Yield loss from use of mulches were evident across RCP 8.5 scenario. Higher response due to applied mulches is projected at Bilate and Hawassa compared with the reponse in Shamana and Dilla, which shows that mulches would become important component of climate change response strategy.

The combined result of adaptation responses including February planting, use of 64 kg ha−1 N and conservation tillage provided yield advantage of 5.8% over the control under medium GHGs emission scenario at Shamana during mid-century period. At Bilate, conservation tillage and June planting produced over 13.4% yield advantage under high GHGs emission scenario during the same period. Mulching 5 t ha−1 was projected to produce 4–5% yield advantage in Hawassa cluster during mid-century period regardless of changes in tillage or planting window. Under high GHGs emission scenarios, over 13.4% yield advantage was projected in Bilate cluster during mid-century period due to conservation tillage and June planting, which is late planting for the cluster. A combination of conservation tillage, April planting and 128 kg ha−1 N produced yield advantage of 1.9% compared with the control at Hawassa during mid-term period under medium GHGs emission scenarios. At Dilla, use of conservation tillage and 128 kg ha−1 N would result in 6.5% yield advantage if planting date is shifted from March to April. In all the clusters, conservation tillage allowed better yield advantage if planting is delayed. Mulching was found to benefit maize production more in Hawassa and Bilate clusters than Shamana and Dilla, the former being high potential but moisture stressed compared with the later. Hence, altering planting date alone, integrating planting date with residue, tillage with nitrogen or planting date with tillage remained least promising and highly risky practices if not integrated with mulch application or nitrogen use in order to adopt climate change in the study clusters. Thus, most promising and least risky practice among simulated strategies would be amalgamation of nitrogen and mulching with either tillage or planting date options, and hence optimized agronomic crop management practices can be considered as effective adaptation strategies to climate change for maize production in southern central rift valley of Ethiopia.