Rainfed-based production of Megathyrsus maximus in sub-Saharan Africa: the case of the semi-arid environment of Sudan

The performance of rainfed-based Megathyrsus maximus (syn. Urochloa maxima and Panicum maximum) was investigated in the semi-arid pastures of Sudan. Split-plot complete design experiments with three replications were applied for two consecutive seasons (2020–2021). The treatments were two in situ rainwater harvesting systems [i.e. ridges plus terraces (RD) and terraces (TR)], three seeding rates (i.e. 1.5, 2.5 and 3.5 kg ha−1) and two urea fertilisation rates [0 kg ha−1 (Z) and 95 kg ha−1 (F)]. These treatments were compared to a control (a flat, unfertilised and zero-tillage plot). Hydrological, biological and chemical indicators were used in the assessment. The results showed that the adopted RWH improved semi-arid pastures, with the RD treatment providing the best results for sustaining biomass production, water use efficiency, nutritional quality and soil quality. The highest plant density (112 000 plants ha−1) was associated with the RD3.5 treatment. The FSR1.5-RD treatment resulted in the greatest plant length (63.5 cm) and number of leaves per plant (34.59), whilst the FSR3.5-RD resulted in both the highest fresh biomass (25.9 t ha−1) and dry biomass (6.3 t ha−1). The chemical compositions of M. maximus (i.e. crude protein, organic matter and nitrogen contents) were also substantially improved by fertilisation. The water use efficiency of M. maximus was plant-, management- and climate-dependent.

The 2019 report of the Intergovernmental Panel on Climate Change (IPCC) has unfolded the gloomy picture of land uses worldwide, especially for the arid and semi-arid environments of sub-Saharan Africa (SSA), which provide indispensable natural pasture resources despite the detrimental impacts of the severe drought on pastures (de Araujo et al. 2018;Mendoza-Labrador et al. 2019;Li et al. 2020;Lutta et al. 2020).
Sudan is a large country in SSA, with a large area used for crop production (84 million ha, Mha, of fertile lands).The world's largest irrigation scheme (i.e.Gezira scheme, 0.88 Mha) is also located in Sudan.Approximately 50% of the agricultural contribution to the gross national product comes directly from the livestock sector, especially the traditional ones.The country is endowed with a vast area of natural pasture (i.e.96 Mha), of which the majority (80%) is located in arid and semi-arid environments (Zaroug 2006;Sudan's country report 2015;Mohamed et al. 2021).However, approximately 26% of the country's rangeland has suffered degradation problems (Ayoub 1998).Drought, overgrazing, fire practices, expansion of rainfed agriculture and erosion have reduced the availability of the country's most nutritious rangeland species and trees (e.g.Andropogon gayanus, Blepharis linariifolia, Vachellia tortilis), leaving only inherently low-quality native species with a crude protein content of <8% (e.g.Cenchrus biflorus, Eragrostis tremula, Cajanus cajan) (Ayoub 1998 and1999;Zaroug 2006;Sudan's country report 2015;Ezzat et al. 2016;Mohamed et al. 2021).
The cultivation of exotic nutritious pasture species is popular worldwide (Sudan's country report 2015; Ward et al. 2017).However, introductions of this nature need to be carefully considered, particularly prior to implementation in fragile arid and semi-arid lands (Ainsworth and Long 2021).Among the exotic pasture species, Guinea grass (Megathyrsus maximus Jacq.) has received much attention (Habermann et al. 2019;Ouachinou et al. 2018;Benabderrahim et al. 2021;Mohamed et al. 2021;Rocha et al. 2022).It is a native animal fodder and is largely introduced (in authorised and unauthorised manners) in all tropical regions in Africa (Benabderrahim et al. 2021).Through the potential alteration of ecological functions, the introduction of exotic pasture species in general and (syn.management practices are central to sustainable rangeland biomass production (da Silva et al. 2015;Benabderrahim et al. 2021;González Marcillo et al. 2021).The persistence and dry matter production of M. maximus under pastures conditions is reputed to be soil nitrogen-supply dependent, especially under intensive pasture systems (Santos et al. 2012;Euclides et al. 2022), however the effects of soil fertilisation on M. maximus in particular, as well as plants in general, is less well understood (Paciullo et al. 2017).

Rainfed-based production of
A recent report by the IPCC (IPCC 2019) has claimed that 'land degradation can be avoided, reduced, or reversed by implementing sustainable land management, restoration, and rehabilitation practices'.This entails integrated water-land management practices, of which rainwater harvesting (RWH) is the most affordable (Falkenmark et al. 2014;Lebel et al. 2015;Adimassu et al. 2017;Velasco-Muñoz et al. 2019;Tolossa et al. 2020;Mganga et al. 2022).However, to maximise the benefits of RWH field-based research is required, especially in arid and semi-arid lands (Castelli et al. 2017;Ghimire and Johnston 2019).The adoption rate of RWH remains very low, especially in rangelands (Ghimire and Johnston 2019;Mohamed et al. 2021).The intensive vegetation cover and fire resistant characteristics of M. maximus emphasise its potential for controlling soil erosion in arid and semi-arid areas (Shashikanth et al. 2013;Lai et al. 2018).
The objectives of this study were to assess the impacts of M. maximus on rainfed pastures in a semi-arid area of Sudan, enhanced with two in situ RWH practices, three different seeding rates and two urea fertilisation doses.
The study hypothesises that in situ RWH practices will significantly improve the rainfed-based production of M. maximus in a semi-arid environment by increasing soil moisture content and by enhancing the adoption of rainfedsoil fertilisation practices and the optimum seeding rate
In both the experimental seasons, the experimental plots were prepared early (15 th of June).The process of preparing the RD plots was as follows: primary ploughing (20 cm in depth using a wide disc plough), levelling, furrowing (0.8 m apart using a mouldboard) and final terracing (30 cm in height using the Bander disk) of plots.The same preparations were carried out for the TR plots, except for the furrowing process.The seeds of M. maximus 'Mombasa'were spread manually on 28th July 2020 and 1st August 2021, for the first and the second seasons, respectively.

Measurements
A rain gauge was installed at the experimental site to monitor the daily amount and intensity of the rainfall events.Daily data for the air temperature, relative humidity, wind speed and sunshine duration were taken from the nearby Wad Medani Meteorological Station (14°22′50″ N, 33°28′44″ E).
Field-based phenological measurements were directly carried out, mainly by averaging the different percentiles (Wang et al. 2018).Plant heights were measured using a tape measure.The leaf area index (total leaves area/total land area, m 2 m −2 ) was measured directly, by multiplying the estimated area of a single leaf x number of leaves x number of plants per 1 × 1 m 2 .This process was repeated three times per plot, with care taken to select representative plant samples (Yang et al. 2022).For the fresh and dry biomass, plant samples were taken randomly from a 1 × 1 m 2 quadrat per plot.Collected samples were immediately weighed (fresh biomass, g plant −1 ), then oven dried (70 °C for 24 h) and the dried samples subsequently weighed for estimating the dry biomass (g plant −1 ).The estimated biomass was then converted to kg ha −1 , based on the plant density of the given treatment (Alebele et al. 2020).The soil moisture contents (0-100 cm profile depth) and the soil bulk density were monitored over the seasons.Three undisturbed soil samples (0-100 cm depth) per plot were collected at 10-day intervals using an auger.The collected soil samples were immediately weighed (the wet weight, g) and then oven dried (105 °C for 24 h) for obtaining the dry weight (g).To determine the bulk density, the volume of the collected soil sample was also determined (cm 3 ).

Soil water balance
The volumetric soil moisture contents and the soil bulk density were determined using the oven-dry technique at the soil laboratory of the Faculty of Agricultural Sciences, the University of Gezira, as follows (Pereira and Alves 2005;Rowlandson et al. 2013): w w w (Equation 1) 4) where θ g and θ ν stand for the gravimetric (g g −1 ) and the volumetric moisture contents (cm 3 cm −3 ), w w and w d are the wet and dry weights respectively, ρ b is the soil bulk density (g cm −3 ), m s is the soil dry weight (g), V t is the soil sample total volume (cm 3 ), z is the soil depth (cm) and WC is the moisture content at z (cm).
The reference evapotranspiration (ET o ) was estimated based on the Penman-Monteith model (Allen et al. 1998).The actual crop evapotranspiration (ET c ) at day i was estimated based on a simple soil water balance as shown below (Pereira and  .(Equation 7) in which R 0 is the runoff in (10 -6 m 3 ) and R is the rainfall volume in m 3 .Two fertilisation doses of 0 kg N ha −1 and 90 kg N ha −1 were applied.The effects of nitrogen application on M. maximus were assessed at the plant and canopy levels.The former was indicated by the impact of fertilisation on leaf characteristics (length and number), whereas biomass production was used at the canopy level (affected by the plant density).

Water use efficiency
The water use efficiency (WUE) at the canopy level was assessed using green water footprints and water productivity indicators (Hoekstra et al. 2011;Vadez et al. 2014;Hatfield and Dold 2019;Mohamed et al. 2021).8) 9) where, WFP is the water footprint (m 3 kg −1 ), GWU is the green water consumption (m 3 ha −1 ), BM is the produced biomass (kg ha −1 ), WP is the water productivity (kg m −3 ) and ET c is the actual evapotranspiration (m 3 ).

Nutrition quality
The nutritional quality of M. maximus was based on three chemical components of the produced biomass (i.e.organic carbon, OC; crude protein, CP; and nitrogen, N contents), which were all analysed in the soil laboratory of the Faculty of Agricultural Sciences, University of Gezira.

Statistical analysis
The analysis of variance was performed using SPSS software version 15.0 (SPSS Inc. 2006) and spreadsheets.
Throughout the study, the statistical significance level is 0.05.The null hypothesis was that 'none of the treatment groups will have different impacts'.The split-plot ANOVA (RCB model, complementary to the Bonferroni post hoc test) was used to analyse the datasets.The main plot consisted of three groups, namely RD, TR and CT.The subplots were the combinations of SR and fertiliser treatments, with the main plots and subplots sometimes rearranged, e.g. to analyse the seasonal effects (a hard-to-change factor).The experimental seasons were assigned to the main plots.The dependent variables were a set of measured hydrological, biological and chemical indicators.If there was a significant difference, a subsequent simple one-factor ANOVA analysis was performed to separate means.

Effects of applied in situ rainwater harvesting techniques
Table 1 compared the impacts of in situ RWH on M. maximus, based on selected indicators, i.e. hydrological, biological, water use and chemical indicators.All indicators under the adopted RWH systems presented a significant improvement compared to the control.The amount of rainfall in the first and the second seasons were almost the same, namely 262 mm and 252 mm (only a difference of 10 mm), respectively.The amount of rainfall in both seasons were slightly lower (8-10%) than the normal rainfall .The applied RWH systems significantly increased the soil moisture contents compared to the control by 46-49% and 35-42% for the first and the second seasons, respectively.The soil bulk densities were reduced by 11-19%, and 9-13% relative to the control, respectively.
As RWH systems were used, the rainfed-based production of fresh biomass by M. maximus had increased by 122-640% in the first season, and by 122-560% in the second season relative to the control.This was associated with a tremendous increase in the leaf area index (LAI) of 83-12628% and 50-10809%, compared to the control, respectively.The plant leaf length under RWH systems (125-175 cm for the RD and 121-167 cm for the TR) was significantly higher than that of the control (24-60 cm).During the first season, M. maximus grown under the control condition wilted at soil moisture contents of 23.4% (by weight).
The water use of M. maximus was significantly improved by the RWH systems, as the water consumption in the dry biomass production was substantially reduced by 54-85% for the first season, and 52-83% for the second season, relative to the control.This was coupled with the improved nutritional quality of M. maximus, attested by the significant increase in OC, N and CP contents of 63%, 58% and 62% over the control, respectively.
The highest CP% and N% contents of 13.5% and 2.22% respectively, were associated with the lowest rainfall amounts received during the second season (252 mm) compared to the first season (262 mm).The first and the second experimental seasons also presented significant differences in rainfall distributions, indicated by dry spells (i.e.days with rainfall amount <1.0 mm).This infers a potential relationship between chemical components of M. maximus, amount of rainfall and distribution.
Megathyrsus maximus under RWH systems is significantly better than that of the control.Among the RWH systems treatments (RD and TR), the RD resulted in the best performance, which is attributed to its increased soil moisture contents (Figure 1).

Seeding rates
The impacts of M. maximus plant density (seeding rates) on selected phenological and chemical components were summarised as supplementary materials (1).At the plant level, increases in the seeding rate had negative impacts on the phenological (plant leaf length, number of leaves, fresh biomass and dry biomass) and chemical components of M. maximus.At the plant level, the plant density presented contradicted impacts on applied RWH plots and the control ones.In the former the increased seeding rate resulted in decreased phenological performance (plant leaf length, number of leaves, fresh biomass and dry biomass), whilst the opposite holds for the latter (Supplementary Table 1a).In addition, the higher the seeding rate, the lower the nutritional quality of M. maximus (Supplementary Table 1b).At the canopy level, the highest plant density (112 000 plant ha −1 ) was associated with the RD35, compared to 60 000 plants ha −1 for the control.At both the plant and the canopy levels the RD performed better than the TR and control, respectively.

Fertilisation
Supplementary Table 2 summarises the effect of fertilisation.The applied fertilisation rate of 95 kg ha −1 significantly improved the phenological characteristics of M. maximus in the semi-arid conditions of Sudan; on average, fertilisation increased the plant length significantly (p = 0.024) from 21.8 cm (the control) to 66.3 cm, coupled with a significant increase (p = 0.03) in the number of leaves from 16 leaves (the control) to 31.6 leaves; the production of fresh biomass increased significantly (p = 0.05) from 35.4 g plant −1 (the control) to 121.6 g plant −1 , but an insignificant increase (p = 0.49) was attached to the dry biomass production, from 17.7 g plant −1 (the control) to 28.2 g plant −1 .Fertilisation increases tremendously (almost 2 folds) the chemical percentages (OC, N and CP) of M. maximus compared to the control (Supplementary Table 2); among all treatments, the fertilised SR15 plots under RD conditions showed the highest nutritional quality of 15.3% for OC, 2.3% for N and 14.5% for CP (Supplementary Table 2).

Seasonal effects
The experimental seasons, S (main plots) interacted insignificant effects (p = 0.45) with the seasonal fresh biomass of 15.4 t ha −1 and 13.5 t ha −1 for the first and the second seasons, respectively; however, there are significant seasonal differences (p = 0.0156) among the subplots (RD, TR, and CT) in the fresh biomass of 21 t ha −1 , 19 t ha −1 and 5 t ha −1 , respectively.The season insignificantly interacted (p = 0.64) with RD, TR, and CT treatments (season x treatment) with a different magnitude, e.g.[|S 1 CT -S 2 CT|] in produced fresh biomass of 1.97 t ha −1 for the S-CT, 3.94 t ha −1 for the S-TR, and 5.39 t ha −1 for the S-RD.Thus, rather than the season, the adopted RWH systems managed to significantly explain the variations in the seasonal produced fresh biomass.The soil moisture profile for ridges with terracing (RD) and terraces (TR), as in situ rainwater harvesting systems, compared to the control (CT), during the second experimental season (June-December 2021), in a semi-arid environment, Sudan.The control is a flat, unfertilised and zero-tillaged plot

Seeding rates
Table 2 summarises the responses of M. maximus to the three seeding rates (SR15, SR25 and SR35), regardless of the fertilisation effects.The highest plant leaf length (62.8 cm), number of leaves per plant (34.4 leaves), fresh biomass (25.1 t ha −1 ) and dry biomass (6.2 t ha −1 ) were observed at SR15-RD, SR15-RD, SR35-RD and SR35-RD, with significant differences among the main plots (RD, TR and CT), as well as at the split plots (seeding rates).However, the significance of the interaction between RD, TR and CT and seeding rates depended on the plant index concerned.The seeding rate interaction was significant for both plant leaf length and number of leaves, but not with the production of fresh biomass (p = 0.07) and dry biomass (p = 0.52).The RD treatment provided the best conditions, although its magnitude of difference with the TR treatment was not significant.

Fertilisation
Table 3 compares the responses to applied fertilisation treatments (95 kg ha −1 and 0 kg ha −1 ).At the main plot (RD, TR and CT), only plant leaf length increased significantly (p = 0.04) in response to the fertilisation treatments.

Combined effects of seeding rate and fertilisation
Table 4 shows the responses to the combined seeding rates (1.5, 2.5 and 3.5 kg ha −1 ) and soil fertilisation (95 kg ha −1 ) with RWH treatments, as well as the control.The FSR15-RD treatment had the greatest plant leaf length (63.5 cm) and number of leaves per plant (n = 34.59),whilst the FSR35-RD treatment resulted in the highest fresh biomass (25.9 t ha −1 ) and dry biomass (6.3 t ha −1 ).The whole plot showed significant differences.The interaction was significant for length and number of leaves, but not for the yielded fresh and dry biomass.The subsequent one-factor ANOVA analysis showed that there were significant differences between the RWH treatment and the control, but not among the in situ RWH treatments (RD and TR).

Water-use efficiency
The actual ET of rainfed-based M. maximus was estimated at 1 893 m 3 ha −1 and 1 648 m 3 ha −1 during the first and the second seasons, respectively.The results for canopylevel water productivity (WP) were shown in Figure 2. The average WP of the RD treatments (14.2 kg m −3 ) was slightly higher than that of the TR treatments (13.8 kg m −3 ).There were insignificant differences among the applied seeding rates, as well as the fertilisation rates.In general, the higher the seeding rate and the fertilisation dose, the higher the fresh biomass-based WP f (12.6-15.9kg m −3 for the fertilised plots compared to 12.7-15.3kg m −3 for the unfertilised ones).The WP f of fertilised and unfertilised conditions under the RD treatment was 2-5% higher than that of the TR.In contrast, the dry biomass-based WP d of unfertilised conditions for the TR treatment was 2% higher than the RD.Among all treatments, the fertilised treatment with a seeding rate of 3.5 kg ha −1 for the RD (RDFSR35) showed the outstanding WPs of 15.9 kg m −3 for the WP f and 3.98 kg m −3 for the WP d .In contrast, the lowest WPs (10.8 kg m −3 and 2.7 kg m −3 , respectively) were associated with the fertilised seeding rate of 1.5 kg ha −1 for the RD treatment (RDFSR1.5).
The WP f showed a relatively better correlation with the LAI than the soil moisture content (SM) based on Kendall's test, with Kendall rank order correlation coefficients of 0.60 (p = 4.05e -07 ) for the LAI and -0.38 for the SM (p = 0.001).The multiple regression showed there was a significant correlation between the WP f and LAI (plant and management dependent, m 2 m −2 ), and SM (m 3 ha −1 ), with a root mean square error (RMSE) of 2.27 kg m −3 , as expressed in Equation 10: . .R 2 = 0.54 Equation 10

Seasonal effects
Rainfall (amount and distribution) is highly variable in the semi-arid climate of Sudan.Among treatments, only the control presented seasonal variations.In addition, the season failed to explain the magnitude of differences among adopted RWH systems and the control.This supports the finding of Pezzopane et al. (2017), who claimed the insignificant impacts of the season on the production of 1.5, 2.5 and 3.5 kg ha −1 with two in situ rainwater harvesting systems [i.e.ridges with terraces (RD), and terraces (TR)] and the control (CT) (flat zero-tillage, unfertilised plots), in a semi-arid environment in Sudan.LN = number of leaves per plant, FBM = fresh biomass and DBM = dry biomass of M. maximus under the wet conditions in Brazil.By increasing the soil moisture contents, the RWH systems led to a substantial reduction in the seasonal variability in terms of rainfall amount and distribution.The semi-arid environment is famous for its high rainfall variability (amounts and distribution).The amounts of rainfall received during the experimental seasons were insignificantly different by 10 mm, however this no longer holds for the distribution of rainfall considering the significant differences in the seasonal length of dry spells of 3.4 and 7.2 days, on average, respectively (Figure 3).The RWH systems significantly reduced the detrimental impacts of dry spells on M. maximus, compared to the control.

Rainwater harvesting effects
Megathyrsus maximus is significantly sensitive to water stress conditions (Rocha et al. 2022).The adopted in situ RWH systems significantly increased the soil moisture contents compared to the control during both experimental seasons, supporting the finding that the dry biomass production (BM) of M. maximus depends mainly on conserved soil moisture contents (de Araujo et al. 2018;Nawaz et al. 2014).The magnitude of differences between the RWH systems and the control is very large, namely 44 mm (440 m 3 ha −1 ).The explanation for the magnitude of the differences among plant indices (i.e.leaf length, number of leaves and biomass production) hinges significantly on the type of RWH system.This states also the varied phenological response of M. maximus to variability in agro-climatic conditions (Nawaz et al. 2014;Mendoza-Labrador et al. 2019).Hence, the sensitivity of M. maximus to water stress in the semi-arid environment is cultivar, soil, and soil moisture management (RWH) dependent, as M. maximus grown under the control condition showed higher wilting points (23.4% by weight) than those of 5.8-7.3%claimed by Durr and Rangel (2003).

Concerns about soil quality
The improved soil moisture contents of the control during the second season, compared to the first season, could be attributed to the improved soil bulk density, as the second season experienced a reduction of 7% in its soil bulk density compared to the first season.Reduced soil bulk density increases rainfall infiltration (leading to increased soil moisture contents), hence the higher the soil bulk density is the lower the soil quality (Arriaga et al. 2017).The soil bulk density of the RD treatments (i.e.conventional ridging practices) of the second season was 2% greater than that of the first season, in contrast to the 2% reduction of the soil bulk density under TR conditions (i.e.zero-ridging practices).Thus, the long-term application of ridging practices might degrade the soil hydrological quality (i.e.lead to the development of hardpans).However, the best performance of M. maximus is associated with the RD treatment under rainfed semi-arid conditions in Sudan.This raises a compromise between sustaining the soil quality (adopting the zero-tillage practice, TR) or the higher production of M. maximus via RD-RWH practices.
The minimum use of tillage remains a sustainable practice compared to conventional tillage (Aziz et al. 2013).Therefore, differences (p > 0.05) between TR and RD treatments were considered, as well as the additional cost of tillage compared to only terracing.

Seeding rate effects
The biomass production of M. maximus possesses a linear relationship with the applied seeding rates.This indicates the current maximum applied seeding rates (3.5 kg ha −1 ) could be increased  bearing in mind that the current fresh biomass under RWH systems (25.5 t ha −1 ) was obtained using a seeding rate of 3.5 kg ha −1 .This is almost twice that claimed for a wetter condition (González Marcillo et al. 2021).However, the interaction of applied seeding rates could not be used to explain the magnitude of differences in the fresh and dry biomass production under the main plots (RD, TR and CT), as they were statistically insignificant.

Fertilisation effects
While the effect of fertilisation on rainfed-based M. maximus production could reduce pressure on natural pasture (de Oliveira et al. 2020), the significance of the effect remains uncertain (Wright et al. 2018;Xu et al. 2020).Xu et al. (2020) stressed that current studies of the impact of fertilisation on M. maximus remained enormously empirical and no one has determined directly the optimum fertilisation formula for M. maximus.The accrued benefits from fertilisation are significantly influenced by inter alia the prevailing climatic conditions, soil water conditions, varieties and management.The responses to fertiliser rate (95 kg N ha −1 ) varied, depending on the plant indices (dependent variables) and the plot level (main or sub).The fertilisation treatments had a significant impact only within the main plots (RD, TR and CT) using plant leaf length as a dependent variable, whilst its interactions with the other plant indices (e.g.fresh and dry biomass) were insignificant.
The current linear relationship of M. maximus production with fertilisation also offered the potential to increase the currently applied fertilisation dose (95 kg N ha −1 ) as well as other fertilisation sources (e.g.phosphate) for rainfed conditions in SSA.

Nutritional quality
The quality of M. maximus depends generally on cultivar and rangeland management practices (Hare et al. 2015;Benabderrahim et al. 2021).The nutritional quality of M. maximus 'Mombasa' was significantly improved by the RWH systems in the semi-arid environment of Sudan.
Interestingly, the N and CP contents of M. maximus in the second season (the lowest rainfall and poorest distribution compared to the first season) were better than that of the first season.This is attributed to the accumulated nitrogen effects from the first season.However, the opposite held for OC contents, which increased with increased amounts of rainfall (12.2% for the first season compared to 11.2% for the second season).Mureva et al. (2018) reported negative correlations between soil nitrogen, soil carbon and annual rainfall under semi-arid conditions.They claimed that gains in soil carbon were permitted for rainfall amounts of 750-900 mm.Consequently, building up soil OC requires a longer time compared to N in semi-arid conditions.The chemical components of fertilised M. maximus plots were better than that of the unfertilised ones.In summary, the higher the seeding rate, the poorer the contents of OC, N and CP for M. maximus (Euclides et al. 2022).

Water use
The rainfed-based M. maximus showed a varied water use efficiency under semi-arid environments in Sudan, depending mainly on the treatment.This is attributed to the wide-ranging physiological response of M. maximus to water deficits (i.e.dry spells) and field management practices like RWH systems (Izaurralde et al. 2011;Hatfield and Dold 2019).In addition, plants in arid environments have well-established mechanisms for drought tolerance, emphasising the substantial role of field practices and improved varieties for reducing water use (Morison et al. 2008).The obtained rainfed-based WFP of M. maximus under semi-arid conditions in Sudan compares favourably with the North African ones of 0.08-0.31kg m −3 , claimed by Yerou et al. (2021).It is worth noting that the WP f showed a better correlation with the LAI than the SM.However, 44% of the variability in the WP f could not be explained by the LAI and SM factors, and hence may be a response to climate variability.The negative sign of the SM in the developed model (Equation 10) stresses the detrimental impacts of evaporation on water use efficiency.This indicates that the bulk of the SM is lost through evaporation rather than the beneficial transpiration process.It is evident that the water use efficiency of M. maximus hinges on plant, management and climatic factors in the semi-arid climate of Sudan.

Conclusion
Applied in situ rainwater harvesting techniques (ridges and terraces) provided the best conditions for M. maximus compared to the control in the semi-arid environment of Sudan.While the performance of the ridges plus terraces technique outweighed that of terraces alone, its long-term impacts on soil quality remain questionable, implying that the less tillage there is, the better the soil quality.Further studies on the long-term impacts of tillage in SSA are urgently needed.
Empirically, the response of M. maximus to fertiliser treatments has been poorly described.This should provide strong motivation to identify the optimum fertilisation formula for accelerating the growth of M. maximus under rainfed conditions, enhanced with rainwater harvesting techniques.The applied fertilisation dose (95 kg N ha −1 ) improved the nutritional quality of M. maximus leading to a build up of soil nitrogen but the effects on soil organic carbon were highly variable in the semi-arid environments in SSA.
Megathyrsus maximus genetic characteristics and management of soil moisture contents are central to improving the water use efficiency.There is a potential yet to improve the water use efficiency of M. maximus by reducing evaporation losses, which are easier to achieve compared to improvements in M. maximus through breeding.There is a compelling need for more applied studies.
In situ rainwater harvesting is a promising technique to mitigate the current accelerated rate of pasture degradation in Sudan, especially by cultivating M. maximus, but care should be taken regarding the impacts of M. maximus on native pasture and rangeland species.

Figure 2 :
Figure2: Average water productivity (WP) of Megathyrsus maximus based on different applied seeding rates (1.5, 2.5, and 3.5 kg ha −1 ), soil fertilisation rate of 95 kg ha −1 (F), zero-fertilisation rate (ZF) and two in situ rainwater harvesting techniques (i.e.ridges with terraces and terraces) in a semi-arid environment, Sudan.F1.5 stands for a fertilised treatment with a seeding rate of 1.5 kg ha −1 whereas ZF2.5 stands for a zero-fertilisation rate with a seeding rate of 2.5 kg ha −1 .(a) For the fresh biomass-based WP, and (b) for the dry matter-based WP

Figure 3 :
Figure 3:The distribution of dry spells (days with rainfall amounts < 1.0 mm) during the first and the second experimental seasons (2020-2021).The dry spells showed significant seasonal differences (p = 0.05), indicating the importance of rainfall distribution for sustaining Megathyrsus maximus in a semi-arid environment of Sudan

Table 3 :
Interactions of soil fertilisation with selected indices of rainfed-based Megathyrsus maximus in a semi-arid environment of Sudan, enhanced by two in situ rainwater harvesting systems [i.e.ridges with terraces (RD) and terraces (TR)] compared to the control (CT) (flat zero-tillage, unfertilised plots).F = fertilisation (95 kg ha −1 ), Z = unfertilised, LN = number of leaves, FBM = fresh biomass and DBM = dry biomass