Balancing crop water requirements through supplemental irrigation under rainfed agriculture in a semi-arid environment

Abstract

Rainfed farming is a dominant agricultural system in Tigray, Ethiopia. However, rainfall is characterized by short duration, intense and erratic subjected to late-onset and early cessation, suggesting a pressing need for Supplemental Irrigation (SI) to fill the crop water demand. Understanding the effects of SI during rainfall late-onset, early cessation, or both, along with their underlying causes, is a critical knowledge gap globally. Although wheat is one of the principal food crops in Tigray, it is subjected to moisture stress during critical growth stages, limiting its potential productivity. Studies specifically related to impacts due to the late-onset of rainfall on wheat are non-existent. Here, we investigated the agrometeorological characteristics of rainfall variability, onset, cessation, and length of the growing season to evaluate the use of SI for balancing the moisture stress in rainfed farming. Meanwhile, using an on-farm experiment, we also evaluated double-season (2017 and 2018) SI application during late-onset (Pre), early cessation (Post), and its combined effects (Pre + Post) on yield and water productivity (WP) of wheat. Yield and WP were significantly (P < 0.05) affected by SI application with higher grain yield (3298 kg/ha) and WP (0.538 kg/m³) obtained from applying Pre + Post. Applying Pre + Post has increased grain yield, biomass, and WP of wheat by 45.6, 27.7, and 21.5% over Rain-fed farming, respectively. Thus, balancing crop water requirements using SI during inadequate rainfall distribution is key for improving WP and wheat production in semi-arid environments. Particularly, the application of SI both during the late-onset and early cessation of rainfall is suggested for greater wheat productivity in semi-arid regions.

1. Introduction

Crop production in the semi-arid area is highly influenced not only by the amount of rainfall but also by its extreme variability, with high intensities, few events, and poor spatial and temporal distribution [1,2]. In semi-arid areas, meteorological droughts occur on average once or twice every decade, while dry spells occur more frequently [1]. Likewise, Tigray is located in northern Ethiopia which is characterized by a semi-arid climate where crop production is highly influenced by rainfall distribution. In Tigray, rainfall is a key source of agricultural water for crop production. On the contrary, it is erratic, torrential, highly variable, and poorly distributed over the growing season. More than 70% of annual rainfall in the region is found only in July and August [[3], [4], [5], [6]]. In addition, the rainfall is either too much at one time (short and intensive) or no rain at another time [4,7]. Thus, frequent dry spells and shorter growing periods due to the late-onset and early cessation of rainfall and dry spell in between is the major causes of crop failures in Tigray [[7], [8], [9]]. Accordingly, the rainy season is characterized by erratic rainfall with inadequate spatial distribution for rainfed agriculture. These imply that there is a pressing need for supplying water to the crops as rainfall is inadequate to fulfil the crop water requirement.

Bread wheat (Triticum aestivum L.) is one of the key cereal crops being cultivated in Tigray both in the middle and highland areas. Wheat cultivation covers about 107,929.86 ha and produces nearly 214,003.14 tons annually [10]. Nevertheless, the average productivity of wheat in Tigray (19.8 kg/ha) is lower than the national average (27.4 kg/ha) [10]. This is mainly attributed to the moisture stresses resulting from rainfall fluctuations as compared to the other parts of Ethiopia [11]. However, the government of Tigray and its partners have been engaged in an immense effort in providing improved agronomic practices to increase wheat production. These include providing improved wheat varieties, inorganic fertilizers, pesticides, and herbicides to wheat-growing farmers. Despite these efforts, Tigray is still expected to face a growing wheat supply deficit due to the ever-increasing agriculture-dependent population. In addition, wheat, which is predominantly sown in the mid of June strongly affects by the occurrence of water stresses. Because the rainfall often starts late and ends before the wheat reaches its grain-filling stage [9].

To cope with low and uneven distribution of rainfall in rainfed farming, supplemental irrigation (SI), defined as an artificial application of a limited amount of water to rainfed grown crops when rainfall fails to provide optimum moisture for plant growth [12,13], is used for increasing crop production and productivity. This practice is used to alleviate the adverse effects of unfavorable rain patterns through applying a limited amount of water for rainfed crops, especially in dryland areas [2,14,15]. However, a clear understanding of the agrometeorological characteristics such as late-onset, early cessation, dry spell, and growing length at the regional level is very important to schedule supplemental irrigation. Unlike full irrigation, SI scheduling for wheat can't be determined in advance due to the uncertainty of rainfall [16]. However, the critical periods of water shortage during the wheat growing season can be identified either through analyzing long-term daily, decadal, monthly, or yearly records of rainfall patterns.

In Tigray, SI has been practiced by smallholder farmers, mainly during the early cessation of rainfall. Few studies (e.g., Refs. [9,17,18]) have been conducted to understand the effects of SI, during only early cessation of rainfall. On the other hand, the application of SI due to late-onset and early cessation of rainfall or their combined effects is lacking in wheat production. Studies especially related to impacts due to the late-onset of rainfall on wheat, are non-existent globally. Hence, studying the effects of SI on balancing the wheat water requirements via fulfilling the crop water demand is important for increased wheat production and water productivity. The main objectives of this study are (1) to examine agrometeorological risks such as late-onset, early cessation, and short growing length and (2) to evaluate the effect of SI (Pre, Post, and combined) on wheat yield and water productivity using field experiments.

2. Materials and methods

2.1. Study area

The study was conducted at the Tigray Agricultural Research Institute experimental site, located in the Kilte-Awlaelo district (14.73°N and 39.60°E, at an altitude of 2020 m. a.s.l) in the eastern zone of Tigray, Northern Ethiopia (Fig. 1a). The study area is characterized by unimodal rainfall ranging from June to September with erratic spatial and temporal distributions (Hagos et al., 2016). The long-term (1992–2016) mean annual rainfall is about 576 mm. More than 75% of the rainfall occurs in July and August, with the maximum effective rainy season ranging from 50 to 60 days.

Fig. 1.

(a) Location map of the study area and (b) Long-term (1992–2016) monthly rainfall, and maximum and minimum temperatures.

The dry season over the study area extends up to ten months, from October to May. The average potential evapotranspiration (ETo) estimated based on FAO Penman-Monteith (Allen et al., 1998) is about 1495 mm. The long-term mean maximum and minimum temperatures are 27.9 and 11.3 °C, respectively (Fig. 1b). The soil type in the experimental site is dominated by sandy clay loam with field capacity and permanent wilting points of 25.9% and 12.6%, respectively (Table 2).

Table 2.

Soil physio-chemical properties of Korir experimental site at a depth of 0–20 cm.

Particle size distribution (%)			Soil texture	pH	TN (%)	Av. P (ppm)	OM (%)	EC (ds/m)	BD (g/cm3)	FC (%)	PWP (%)	TAW (mm/m)
Sand	Silt	Clay
53	26	21	Sandy clay loam	7.45	0.11	2.86	0.93	0.79	1.36	25.9	12.6	133.0

Note: TN = Total nitrogen, Av. P = Available phosphorus, OM = Organic matter, EC = Electrical conductivity, BD = Bulk density, FC = Field capacity, PWP = Permanent wilting point, and TAW = Total available water.

2.2. Agrometeorological characteristics

2.2.1. Estimation of onset, cessation, and length of growing days

Long-term (1992–2016) daily climatic data for the study area was obtained from the Ethiopian Meteorological Service Agency. This long-term daily rainfall data were used to analyse seasonal and annual rainfall variability and its characteristics including onset and cessation dates. The amount, temporal distribution, dates of onset, cessation of rainfall, and length of the growing seasons were analyzed by processing daily rainfall data using the Instat climatic model [19].

Previous studies applied empirical modeling approaches and threshold values to determine the onset and cessation of rain [7,8,[20], [21], [22]]. In this study, the onset and cessation of the rainy period were determined from the rainfall-reference evapotranspiration (ETo) relationship followed by Refs. [7,8]. Accordingly, onset was assumed to occur after 15 June when long-term cumulative pentad rainfall is greater than or equal to the cumulative half of the pentad ETo for at least two consecutive pentads in which the rainfall sum exceeds 25 mm and within 30 days there is no dry spell of more than 7 consecutive days. Similarly, cessation was supposed to happen after 20 August which was a week preceded by a pentad with a ratio of cumulative effective rainfall to cumulative ETo lower than 0.5 and more than 10 days of dry period [7]. The period between onset and cessation was considered as the length of the growing period (LGP).

2.2.2. Seasonal deficit in crop water requirement

Wheat crop water requirement (ETc) of the study area was determined from the reference evapotranspiration (ETo) and respective crop coefficients (Kc) of each growth stage [23]. The Kc of wheat in the different growth stages was adopted from Ref. [23]. Hence, the seasonal water requirement of wheat was computed using the FAO CROPWAT program [23] given in Equation (1).

$$ETc=Kc\times ETo$$ (1)

The supplemental irrigation water requirement (SIWR) was computed from crop water requirement (ETc) and effective rainfall (Pe) as provided in Equation (2).

$$SIWR=ETc-Pe$$ (2)

In addition, Assefa et al. [24] computed the relative water satisfaction (RWS) as the ratio of effective rainfall (Pe) to crop water requirement (ETc) as given in Equation (3).

$$RWS=\left(\frac{Pe}{ETc}\right)\times 100$$ (3)

2.3. Experimental design and treatments setup

The experiment was conducted in a randomized complete block design with three replications in two consecutive production years (2017 and 2018). Similar treatment setups were followed in both years and the experimental plots. Treatment details are presented in Table 1 and the randomization and layout of the treatments can be referred to in supplementary information (Fig. 2). The plot size was 4 × 5 m with 2 and 1 m spacing between blocks and plots, respectively.

Table 1.

Treatments setup and description.

Treatments	Symbol	Descriptions
Rain-fed (Control)	Rain-fed	No supplemental irrigation
Pre rainfall SI	Pre	Application of measured amount of water under conditions of late onset of rainfall (Two weeks before normal onset)
Post rainfall SI	Post	Application of measured amount of water under conditions of early cessation of rainfall
Pre + Post rainfall SI	Pre + post	Application of measured amount of water under conditions of both late onset and early cessation of rainfall

Fig. 2.

Layout of the experimental plots (RI, RII, and RIII represent 1st, 2nd, and 3rd replications, respectively).

2.4. Irrigation application method and water productivity

The amount of irrigation water applied to each experimental plot was measured using a container (Jerry can) of known volume (20 L) and hence the application efficiency was 95% as the method of application is hand watering. The supplemental irrigation was applied in three ways as specified in the treatment setups, i.e., only before the onset of rain, only after the cessation of rain, and at both before onset and after cessation of the rain. Irrigation amounts of 15 mm were applied to the plots (20 m²) every 5 days. This amount was sufficient to maintain the topsoil at field capacity during the growing season. The number of irrigations depends on the length of the dry periods. Accordingly, 3–5 irrigations were applied based on the water deficit of the seasons.

The irrigation method was evaluated based on the use of the irrigation water by crop, which was estimated as crop water productivity (CWP). CWP (Kg/m³) can be described in different forms based on the type of water applied. For instance, rainfall water productivity (RWP), irrigation water productivity (IWP) and total water productivity (TWP) can be mentioned. The described CWP indicators were calculated following [25] using Equations (4), (5), (6)).

$$RWP=\frac{Yt}{P}$$ (4)

$$IWP=\frac{Yt}{I}$$ (5)

$$TWP=\frac{\Delta Y}{P+I}=\frac{Yt-Yr}{P+I}$$ (6)

where Yr, Yt, and ΔY are the rainfed yields, irrigated yield, and increase in yield due to SI, respectively. P is the total rainfall from sowing to harvest, and I is the total SI amount in each treatment measured in m³/ha.

2.5. Soil sampling and analysis

Composite soil samples from plants' root depth i.e., surface soil depth of 0–20 cm were collected using a soil auger from three random spots of the experimental plot before planting. These soil samples were analyzed at Mekelle Soil Research Centre's soil laboratory following standard laboratory procedures. Accordingly, the soil texture of the experimental site was determined to be sandy clay loam. The physio-chemical characteristics analyzed from the composite samples associated with top 0–20 cm soil depth are given in Table 2.

2.6. Agronomic management and statistical analysis

The experimental field was well prepared, and the wheat crop was grown under optimal planting practices. The full dose of DAP (Di Ammonium phosphate) fertilizer was applied as a triple super phosphate (46% P₂O₅) at sowing, while Urea (46% N) was applied by splitting, half dose at sowing and half at 30 days after emergence for all treatments. Following the standard recommendations for the study area, 50 kg N/ha and 100 kg P₂O₅/ha were applied at the time of sowing, the remaining 50 kg N/ha was top-dressed at the vegetative stage of wheat. Each plot received the same fertilizer dose and field management in both years [26]. The improved wheat variety (locally called Kekeba) was used as a testing crop and was hand drilled at 150 kg/ha with a row spacing of 10 cm. Wheat was sown on 24 June for the Pre and Pre + Post and 8 July for the Rain-fed and Post for both cropping years. Data were statistically analyzed using GenStat 16th edition [27]. For yield and water productivity, a combined analysis of the experiments conducted during 2017 and 2018 has been provided helpful for a better decision. Treatment means were also compared using the least significant difference (P = 0.05) procedure [28].

3. Results

3.1. Annual and seasonal rainfall variability and rainy days

The long-term mean annual rainfall (1992–2016) of the study area was 575.6 mm (Fig. 1, Table 3), out of which 486.4 mm (85%) was received during the rainy season (June–September). The amount and distribution of seasonal rainfall is a critical rainfall feature that indicates useful information on temporal rainfall variability. The long-term mean monthly rainfall of the rainy season was found to be 38.8 mm (7.8%), 198.7 mm (40.9%), 219.5 mm (45.1%), and 29.4 mm (6.0%) for June, July, August, and September, respectively (Table 3). Furthermore, the two months (July–August) received 86.0% (418.2 mm) of the rainy season totals (486.4 mm). However, June and September are exposed to water deficit, which only receives 20%, which could lead to yield reduction. This indicates high variability in the temporal distribution of rainfall during the length of the wheat growing period (rainy season: June–September). Similarly, the average annual and seasonal rainy days were found to be 52.7 and 41.2 days, respectively. Out of the rainy season rainy days, 31.9 days (77.3%) occurred during July and August. The long-term mean monthly rainfall for the growing season (June to September) has been highly variable (CV > 79%) indicating large rainfall variability. In addition, the CV for rainy days was also large (CV > 30) indicating higher rainy days variability within the growing season except for July. A study by Tesfaye & Walker [29] and Araya et al. [8] verified that crop water stress and crop failure were caused by the mismatches of crop growing period with the seasonal rainy period. Similar results were also reported by Hagos et al. [17] and Gebremicael et al. [30] that 80% of the rainfall occurs during the rainy season from July to August.

Table 3.

Long-term (1992–2016) mean annual and seasonal rainfall and rainy days variability.

Parameters	Measurements	Kiremti (rainy season months) (Rainy season) months						Annual
		Jun	Jul	Aug	Sep	Jul–Aug	JJAS
Rainfall	Mean	38.8	198.7	219.5	29.4	418.2	486.4	575.6
	St. Dev.	36.6	93.8	119.3	35.4		71.3	170.6
	CV	94.2	47.2	54.4	120.2		79.0	29.7
Rainy days	Mean	5.3	16.8	15.1	4.1	31.9	41.2	52.7
	St. Dev.	2.9	3.4	6.3	4.2		4.2	14.9
	CV	55.4	20.0	41.8	103.0		55.0	28.3

Note: St. Dev. = Standard Deviation and CV = Coefficient of Variation.

3.2. Rainfall onset, cessation, and length of growing period

Analysis of long-term daily rainfall revealed that the start of rain in the study area ranges from 183 days of the year (DOY) that is 01 July to 199 DOY (17 July) and the end of rain ranges from 235 DOY (22 August) to 254 DOY (10 September). The average Onset of Rain (OR), Cessation of Rain (CR), and length of the growing season (LGS) were found to be 186 DOY (4 July), 246 DOY (2 September), and 60 days, respectively (Fig. 3). The LGS analyzed based on the difference between onset and cessation was too short as compared to the actual duration of the wheat-growing period in the study area. Wheat requires more than 100 days (from sowing to maturity) but most of the analyzed length from onset to cessation falls below 70 days. In this analysis, the observed onset, cessation, and LGS for the wheat growing season revealed that the rain season was challenging for optimum wheat crop production due to the late date of rainfall onset and early cessation date that leads to a decrease in LGS. Thus, it is crucial to support the growing season with SI at both pre-onset and post-offset of rainfall with an optimum amount of SI based on the length of the dry period. In line with this, Araya & Stroosnijder [7] reported that the most frequent onset and cessation of crop growing period around the study area was July 1–10 and September 1–10, respectively. Similarly, Hadgu et al. [11] reported that the onset and cessation of the growing period were July 3–5 and September 8–16, respectively. According to Yibrah et al. [31], the onset date, cessation date and LGP for JJAS (rainy months) season around the study area were 177 DOY (25 Jun), 184 DOY (02 Jul), 258 DOY (14 Sep), 261DOY (17 Sep) and 87, 79 days, respectively.

Fig. 3.

Dates of Onset of Rainfall (OR), Cessation of Rainfall (CR), and Length of Growing Season (LGS) for wheat growing season.

3.3. Seasonal deficit in crop water requirement

The seasonal deficit of wheat water requirement estimated using long-term mean climate inputs, used for planning purposes was found to be 111.7 mm (Table 4) for both seasons. Likewise, about 107.8 mm was also found using daily climate data of only the growing seasons for the study years (Table 4). The long-term agrometeorological analyses indicated that the mean effective rainfall (Pe) of the wet seasons of 2017 and 2018 was 301.3 and 360.7 mm, respectively. Accordingly, the relative water satisfaction was 97.8% and 100% for 2017 and 2018, respectively. These results revealed that in terms of total seasonal amount the crop water requirement of wheat was relatively stable in this area. However, its high temporal variations could cause an indeterminate drought that deters wheat production.

Table 4.

Decadal supplemental irrigation water needs of wheat for 2017 and 2018.

Year		Jun		Jul			Aug			Sep			Total
		II	III	I	II	III	I	II	III	I	II	III
2017	ETo, mm	49.4	49.4	36.5	36.5	36.5	34.2	34.2	34.2	43.7	43.7	43.7	442.0
	Kc	0.3	0.3	0.3	0.4	0.7	1.1	1.2	1.2	1.1	0.8	0.5
	ETc	14.8	14.8	11.0	13.1	25.6	36.6	39.3	39.3	48.9	36.3	20.5	300.3
	Pe, mm	0.0	16.9	30.5	74.4	40.0	156.0	79.2	89.7	9.6	0.0	0.0	496.2
	SIR, mm	14.8	0.0	0.0	0.0	0.0	0.0	0.0	0.0	39.3	36.3	20.5	111.0
2018	ETo, mm	49.4	49.4	36.5	36.5	36.5	34.2	34.2	34.2	43.7	43.7	43.7	442.0
	Kc	0.3	0.3	0.3	0.36	0.7	1.07	1.15	1.15	1.12	0.83	0.5
	ETc	14.8	14.8	11.0	13.1	25.6	36.6	39.3	39.3	48.9	36.3	20.5	300.3
	Pe, mm	13.6	25.6	144.0	80.8	72.8	60.0	26.4	91.2	8.9	4.0	2.4	529.7
	SIR, mm	1.2	0.0	0.0	0.0	0.0	0.0	12.9	0.0	40.1	32.3	18.1	104.6
Mean	ETc	14.8	14.8	11.0	13.1	25.6	36.6	39.3	39.3	48.9	36.3	20.5	300.3
	Pe, mm	6.8	21.2	87.2	77.6	56.4	108.0	52.8	90.4	9.2	2.0	1.2	513.0
	SIR, mm	8.0	0.0	0.0	0.0	0.0	0.0	6.5	0.0	39.7	34.3	19.3	107.8

Note: I, II, and III are 1st, 2nd^, and 3rd decades of a month. Pe is effective rainfall and SIR is supplemental irrigation requirement, decades = 10 days.

The mean monthly rainfall distribution of June, July, August, and September was found to be 28 mm (5.5%), 221.2 (43.1%), 251.2 (49.0%), and 12.4 (2.4%), respectively (Table 4). Moreover, the two months (July–August) received 92.1% (472.5 mm) of the mean seasonal rainfall indicating the growing season results in low (60%) rainwater use efficiency (the ratio of water usage by the crop to the amount of rain received by the crop). Conversely, June and September are the most essential months, when germination and yield formation takes place, respectively, and during these months the crops are more sensitive to water stress. Accordingly, the amount of SI needed during June and September to compensate for the deficit in wheat water requirement was 108 mm but 80% of it was applied in September as the rainfall is low in September compared to June. Fig. 4 also shows crop water requirements, effective rainfall, and irrigation water requirements as per the wheat crop growing stages separately indicating the existence of a clear variability of water demand among the crop growth stages. Similar results were obtained from different studies [13,19,32] who reported that the SI amount required by wheat crops to produce higher yield was increased as rainfall amount decreases. Based on the long-term cropping season rainfall analysis, the total amount of SI applied was set to be 50, 75, and 100 mm for the Pre, Post, and Pre + Post respectively. To ensure optimal conditions for crop growth, 20 mm of water was applied to both the Pre and Pre + Post areas three days prior to sowing (on June 21st). This pre-irrigation brought the soil moisture in the root zone to field capacity, creating favorable conditions for wheat germination. Then a fixed depth of irrigation (15 mm) was applied at every 5-day interval for each of the wheat cropping seasons of 2017 (Fig. 5a) and 2018 (Fig. 5b).

Fig. 4.

Wheat crop growing stages water demand levels using long-term climate inputs.

Fig. 5.

Daily rainfall and irrigation water applied (a) for the 2017 and (b) for the 2018 cropping season at the experimental site; DOY = Days of the year.

3.4. Yield and water productivity

3.4.1. Effect of supplemental irrigation on yield components

The combined analysis results (Table 5) showed that the yield components of wheat (days to 50% heading, panicle length, grains per panicle, and thousands of grain weight) were significantly (P < 0.05) influenced by the application of SI. However, there were no significant differences among the treatments in days to 50% maturity, plant height, and number of tillers per plant. The Rain-fed registered the shortest number of days to flowering (52 days), while the Pre + Post took the longest number of days (58 days) from sowing to flowering. The panicle length in the Pre + Post was significantly (P = 0.05) higher than the remaining treatments. The grains per panicle and thousand grain weight were also significantly lower than those of other plant parameters in the Pre and Rain-fed. Generally, the yield components tend to increase with the increasing application of SI strategies.

Table 5.

Yield components of wheat (2017 and 2018 combined analysis).

Treatments	DTH (days)	DTM (days)	PH (cm)	PL (cm)	Tillers /Plant	Grain/Panicle	TSW (g)
Pre + Post	56.0ab	90.8	76.6	8.08a	5.6	38.1a	32.7a
Pre	57.7a	90.7	74.3	7.3ab	4.5	32.5b	26.6b
Post	53.7bc	91.0	74.3	7.7ab	4.7	38.2a	29.5ab
Rain-fed	52.0c	91.0	74.4	7.2b	4.0	34.4ab	28.1ab
Mean	54.8	90.9	74.9	7.6	4.7	35.8	29.2
LSD (0.05)	3.32	Ns	Ns	0.83	Ns	5.20	4.57
CV (%)	3.5	0.7	3.5	5.5	21.2	7.3	7.8

Note: DTH, DTM, PH, PL, and TSW are days to 50% heading, days to 50% maturity, plant height, panicle length, and thousand seed weight, respectively. Means within columns followed by different letters differ significantly at P = 0.05.

3.4.2. Effect of supplemental irrigation on grain and biomass yields

The SI treatments have shown significant (P < 0.05) differences in the average biomass and grain yields along with the corresponding harvest index (Table 6). The maximum biomass yield (8718 kg/ha) and grain yield (3298 kg/ha) were obtained under Pre + Post while the minimum yield was obtained under the Rain-fed treatment. Application of SI before on-time onset and after early cessation of rainfall has increased the grain and biomass yield of wheat by about 45.6% and 27.7% over the Rain-fed, respectively. As compared to the Rain-fed, applying SI has shown the minimum and maximum yield increments ranged from 137 to 1033 kg/ha for the Pre and Pre + Post, respectively. The harvest index also ranged from 33.43 to 37.69% for the Rain-fed and Pre + Post, respectively.

Table 6.

Yield productivity of wheat (2017 and 2018 combined analysis).

Treatments	Biomass yield (kg/ha)	Grain yield (kg/ha)	Harvest Index (%)	Yi (kg/ha)
Pre + Post	8718a	3298a	37.69a	1012.58
Pre	7351bc	2402b	32.83b	137.50
Post	8279ab	3035a	36.78ab	815.33
Rain-fed	6828c	2265b	33.43b	0.00
Mean	7793.8	2750.1	35.2	655.14
LSD (0.05)	1020.25	365.07	4.08	–
CV (%)	6.6	6.6	5.8	–

Note: Yi = Yield increments due to SI application. Means within columns followed by different letters differ significantly at P = 0.05.

3.4.3. Effect of supplemental irrigation on water productivity

As indicated in Table 7 the SI treatments have significantly (P < 0.05) affected the average water productivity. The maximum IWP and TWP were estimated to be 1.087 kg/m³ and 0.538 kg/m³ under the Post and Pre + Post, respectively. Rainwater productivity for the Rain-fed was very low (0.442 kg/m³) as compared to IWP and TWP. This is due to the uneven temporal distribution of rainfall that resulted in a low yield (2265.0 kg/ha) with high rainfall amount (5130.0 m³/ha). Thus, under the same agronomic practice application of SI on Pre + Post has increased the TWP of wheat by 4.3% and 21.5% over Pre- and Rain-fed conditions, respectively. In contrast, the application of SI on Pre condition has increased the IWP of wheat by 6.8% over the Pre + Post.

Table 7.

Water productivity of wheat (2017 and 2018 combined analysis).

Treatments	Rainfall (P) (mm)	Irrigation (SI) (mm)	Total (P + SI) (mm)	IWP (kg/m3)	TWP (kg/m3)
Pre + Post	513.0	100	613.0	1.013a	0.538a
Pre	513.0	50	563.0	0.275b	0.427b
Post	513.0	75	588.0	1.087a	0.516a
Rain-fed	513.0	0.0	513.0	0.000	0.442b
Mean	513.0	75.0	569.3	0.59	0.48
LSD (0.05)	–	–	–	0.335	0.06
CV (%)	–	–	–	28.2	6.3

Note: IWP and TWP represent supplemental irrigation (SI) and Total water (P + I) productivity, respectively. Means within columns followed by different letters differ significantly at P = 0.05.

4. Discussion

In Tigray, the current wheat production has been insufficient to meet the increasing food demand due to its various forms of rainfall interruptions. Moreover, the average productivity of wheat is low (19.8 kg/ha) as compared to the Ethiopian national average (27.4 kg/ha) (CSA, 2018). The major cause of yield reduction and crop failures include shorter growing periods due to the late-onset of rain, early cessation, and frequent dry spells [7,8]. The two critical factors for improving water and wheat productivity are the timing and amount of rainfall [13]. Moreover, there are only four months of rainfall in Tigray with only July–August as the active rainy months. The average seasonal rainfall and rainy days of the study area were 486.4 mm and 41.2 days, respectively. Out of these, July and August received 86.0% and 77.3% of the mean seasonal rainfall and rainy days, respectively leaving June and September with a water deficit. Considering wheat growing periods application of the measured amount of water during the late-onset of rainfall in June and early cessation of rainfall in September or at both conditions is therefore critically important.

For this purpose, field experiments were conducted during the 2017 and 2018 rain seasons in the Korir irrigation scheme under conditions of erratic rainfall distributions. This study was aimed at evaluating wheat crop productivity by applying different forms of SI application periods, during the late-onset and early cessation of rainfall useful for increasing the use of the SI. Therefore, application of 50 mm of SI before onset (late June) and 75 mm after cessation (early September), and 100 mm under both Pre + Post based on the increase or decrease of rainfall amount were applied to balance the wheat crop water deficit. The two years combined results indicated that the application of SI at pre-onset and post-cessation of rain had a significant effect on the yield, yield component, and water productivity of wheat. From this experiment, the application of SI has increased the grain and biomass yields of wheat by about 45.6% and 27.7% over the Rain-fed (Control), respectively. Several studies have found the effects of SI on yield and water use efficiency are higher as compared to rainfed farming [2,12,14,15,32,33]. Previous studies [13,34,35] also reported that the SI amount depends on the amount of rainfall that the SI for wheat increases as rainfall amount decreases since SI is only applied when the rainfall fails to provide the moisture needed by crops. Early studies have shown that applying two or three SIs (80–200 mm) to wheat could increase crop grain yields by 36 to 45% [33]. Oweis & Hachum [36] also reported that rainfed wheat yield has increased from 21.6 kg/ha to 46.1 kg/ha by applying only 68 mm of supplemental irrigation water. Similarly, Fox & Rockström [37] showed that the application of 60–90 mm SI amount per season alone has resulted in an average wheat grain yield of 712 kg/ha. Balancing crop water requirements through SI under rainfed conditions is key for higher wheat production in Tigray. In this experiment, a fixed depth (15 mm) of SI to each experimental plot was also applied using a user-defined interval (5 days) based on field investigation to refill the field capacity of the soil and reduce water stress which is in line with the studies conducted by Li et al. [38].

Rockström et al. [39] and Allen et al. [23] reported that water stresses during flowering or yield formation reduce wheat yields as compared to water deficits during the vegetative period and ripening stages. Late-onset and early cessation of rainfall implying water stress at later development stages (during the reproductive stage) can also negatively affect crop productivity. On the other hand, the late-onset of the rainy season together with long dry spells leads to crop/seedling failure because most farmers, traditionally, sow their seed in dry soil [20]. The application of SI can also produce similar or even higher grain yields than in rain-independent fully irrigated conditions. Thus, rainfed farming being the dominant agricultural production system, supplementing water to crops when either during the late-onset or after early cessation of rainfall is important. On the other hand, the use of optimum water use technologies such as drip irrigation and efficient irrigation scheduling and water application methods could bring valuable benefits to the water-limited schemes [40,41]. Additionally, as salinity hazard is evolving in the irrigation schemes introducing salinization-reducing techniques helps to minimize soil degradation hazards in Tigray [42].

Finally, it is essential to point out some of the limitations during field experimentations. It was challenging to fully control the effects of birds and rodents, especially on the Pre and Pre + Post as the sowing dates were two weeks before the normal sowing (onset of rain) and were early matured as compared to the other treatments. At this time no crop was sown, and no wheat or other crops matured except for the Pre and Pre + Post. As a result, there were some effects by birds and rodents after early sowing and early maturity. This is because farmers were not familiar with using SI at early sowing or pre-onset of rainfall. However, farmers had some practices on dry sowing and waiting for the onset of rain and they noticed that damage by rodents and birds was the main challenge in their practice of early sowing. Such problems can be solved if all or most farmers in an area used similar practices of early sowing using pre-onset SI application.

5. Conclusion

Analysis of seasonal rainfall showed a high coefficient of variation (79%), which is an indicator of large rainfall variability showing mismatches among distributions of rainfall, the length growing period, and crop water requirement of crops. To balance the crop water deficit in late June (due to late-onset) and early September (due to early cessation), when germination and yield formation takes place, respectively, the application of measured SI based on the deficit level is important to minimize yield reduction and even to increase yield. Grain yield, yield components, and water productivity of wheat were strongly influenced by the application of SI. Maximum grain yield of 3298 kg/ha and the highest water productivity of 0.538 kg/m³ were recorded in response to the application of SI on the late-onset and after cessation of rainfall combined. The lowest grain yield of 2265 kg/ha and the lowest water productivity of 0.442 kg/m³ were recorded in response to the rainfed production without application SI. The seasonal crop water requirement deficit obtained from analysis of long-term climate data is similar to that of the growing season. Therefore, farmers who have the option of water harvesting (available water) can apply 100 mm of SI on average in late June and early September, based on the rainfall condition. SI increases the yield and water productivity of wheat in semi-arid areas and can be applied in other similar agroecologies. These indicate that the SI technique is of typical use in areas where rainfall is a limiting factor of production hindering crop and water productivity. Thus, SI during insufficient rainfall is imperative to increase wheat production in Tigray.

Author contribution statement

Teferi Gebremedhin: Performed the experiments; Analyzed and interpreted the data; Wrote the paper.

Gebremedhin Gebremeskel Haile: Conceived and designed the experiments; Analyzed and interpreted the data; Wrote the paper.

T.G. Gebremicael: Analyzed and interpreted the data.

Hintsa Libsekal: Kidane Welde Reda: Contributed reagents, materials, analysis tools or data; Wrote the paper.

Data availability statement

Data included in article/supp. material/referenced in article.

Additional information

No additional information is available for this paper.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.