Increasing Millet Planting Density with Appropriate Fertilizer to Enhance Productivity and System Resilience in Senegal

Abstract

Climate forecasts show increased frequency and intensity of drought in the semi-arid regions of west Africa, which negatively impacts food and nutrition security. Developing and improving resilient cropping systems will require adequate varieties with improved agronomic practices. The purpose of the study was to evaluate grain and biomass production of newly released dual-purpose millet varieties under different fertilizer rates and planting densities across the millet-cropping regions of Senegal with different rainfall regimes (Bambey: 600 mm, Boulel: 700 mm, Nioro: 650 mm, and Sinthiou Malème: 800 mm). The experimental design was a split-split-plot with three replicates using one traditional variety (Souna 3) and four dual-purpose varieties (Thialack 2, SL28, SL 169, and SL423) as the main factor; two planting densities (E1: 12,500 seed hills ha⁻¹ and E2: 25,000 seed hills ha⁻¹) as second factor; and eight fertilizer combination as sub-sub-plots. Results showed that variety yield response differed with environments. Regardless of variety, increasing sowing density increased grain yields (1600 kg ha⁻¹ for E2 vs. 1000 kg ha⁻¹ for E1) and fodder yields (4200 kg ha⁻¹ for E2 vs. 3100 kg ha⁻¹ for E1). Fertilizer response differed between environments, but the application of 70N–10P–19K + 2.5 t ha⁻¹ cow manure produced appreciable yields in all the environments. The dual-purpose varieties (SL 28, SL 169, and Thialack 2) sown at E2 produced the best grain and fodder yields. Soil water content was greater for all varieties (Thialack 2: 12.32%; Souna 3: 5.32%; SL28: 6.32%; and SL423: 9.23%) at higher planting density compared with normal density (Thialack 2: 9.25%; Souna 3: 3.21%; SL28: 5.43%; and SL423: 7.47%). The highest agronomic-use efficiency (AEg) of inorganic fertilizer applied was observed at 25,000 seed hill ha⁻¹ and averaged 6.63 kg kg⁻¹ at Boulel, 9.20 kg kg⁻¹ at Sinthiou Malème, 4.67 kg kg⁻¹ at Bambey, and 8.32 at Nioro kg kg⁻¹. The AEg significantly varied among fertilizer combinations, with greatest AEg obtained with 70N–22.5P–22.5K (5.53 kg kg⁻¹) at Boulel and with 95N–17P–27K (4.66 kg kg⁻¹) at Bambey. This study provides crop-management options for millet-cropping systems in the semi-arid regions of west Africa for improving millet productivity while enhancing system resilience through better conservation and utilizing of soil water.

1. Introduction

The cropping systems in Senegal are mainly rain-fed and dominated by subsistence farming with limited access to inputs such as good-quality seeds, fertilizers, pesticides, and equipment as well as limited access to finance and credit, agricultural insurance, and improved post-harvest storage techniques. In general, farmers do not use sustainable land-management practices [1,2]. Other challenges include unproductive fallow periods, overgrazing, bush fires, and little to no fertilizer and limited nutrient input [2,3]. As a result, most of the agricultural lands in Senegal are degraded and agricultural production remains low while population growth continues to increase by 2.7% per year [4,5]. In addition, over 70% of the total crop production in Senegal is rain-fed. The agricultural sector is highly vulnerable to climate shocks, with low crop productivity, which will have negative impacts on crop and livestock productivity, food availability, and food prices [6]. Currently, food insecurity and malnutrition are 7.2% and 8.2%, respectively, with major regional disparities [7] that will further increase due to negative impacts of climate change, poor land-use planning, and management.

Pearl millet (Pennisetum glaucum (L.) R. Br.) is the main cereal crop in Senegal, is the staple food for more than 55% of the population, and is the most-produced rain-fed cereal [4]. Between 2010 and 2019, 837,078 ha of crop land was under pearl millet [8]. However, typical farmer millet yields of 500–900 kg ha⁻¹ remain below the attainable yield of 3500 kg ha⁻¹ [9,10]. Several studies showed the most limiting factors for cereal crops (i.e., maize, Zea mays L.; pearl millet; sorghum, Sorghum bicolor L. Moench) production in sub-Saharan Africa are the erratic rainfall pattern and low soil fertility [11,12]. The gap between actual and potential millet yield is nearly 1.7 t ha⁻¹ [13], which is largely a consequence of soil degradation, low organic matter soils, poor soil fertility management practices, and inappropriate and low use of fertilizer [14]. The current millet fertilizer recommendation of 150 kg ha⁻¹ NPK (15-15-15) with 100 kg of urea (46% N) was developed over 45 years ago for non-improved millet varieties and is prescribed as a single rate regardless of soil and climate variability [15]. Such a recommendation does not consider soil types, agro-ecological conditions, and the typology of farmers’ cropping systems, farm typology, and socio-economic conditions. Inorganic fertilizer applications alone could induce significant yield increases but is not efficient and would not be considered suitable for resource-constrained farmers. Consequently, the benefits of low-rate organic materials in combination with low or moderate levels of inorganic fertilizer can sustain production [16,17].

The annual precipitation gradient of Senegal ranges from <300 mm in the North to >1200 mm in the South. Soil pH ranges from >7.5 to <5.5, and soil organic matter (SOM) varies from 5% to <1% with over 50% of cropland close to 2% SOM. In the peanut basin that includes most part of the millet-cropping area in Senegal, prolonged dry periods near the beginning and end of the cropping cycles are frequent and account for the main causes of crop failure [18]. With increasing population density in the Sahel, livestock are often confined to smaller grazing areas during the rainy season to avoid crop damage, while in the dry season their feeding become more and more critical as the season progresses. This often results in animals’ undernourishment being shifted from the dry season to the wet season, and higher risks of pasture overgrazing [19]. More intensive grazing close to villages during the dry season can lead to the dominance of unpalatable or poorly productive, short-cycle species in rangelands. This can reduce livestock production and therefore reduce nutrient transfers to croplands, which in turn can diminish crop residue availability for livestock. As population density increases, particularly in areas with good access to urban markets, zero-grazing with improved dairy cattle and cultivated fodder becomes the predominant form of livestock management [20]. Despite limited availability of cropping land, millet as the main cereal of Senegal continues to be sown at a low planting density of 12,345 seed hills ha⁻¹, which results in lower yields. Therefore, opportunities exist to increase the planting density to increase yields. Millet has two main uses in Senegal farming systems: grain for human consumption and biomass as livestock feed. In response to the dominant uses of millet, dual-purpose pearl millet varieties (i.e., Thialack 2, SL28, SL423, SL169) [10] were developed. Bastos et al. [21] showed in controlled conditions in Bambey (450–500 mm yr⁻¹), Nioro (600–700 mm yr⁻¹), and Sinthiou Malème (700–800 mm yr⁻¹), and over three years (2019, 2020, and 2021), grain and biomass yields of both dual-purpose pearl millet lines (Thialack 2) and the local variety (Souna 3) can be increased up to 200% by simply increasing sowing density from the current rate of 12,345 to 24,691 seeds ha⁻¹ with fertilizer application. This needs to be further tested under different environmental conditions for the wide-scale adoption of dual-purpose millet varieties with appropriate agronomic practices including fertilizer recommendations. This is because there are wide precipitation gradients as well as soil fertility differences such as in pH and SOM content among agro-ecological zones in Sub-Saharan Africa [14].

About five years ago, the Feed the Future Sustainable Intensification Innovation Lab program supported Senegalese agricultural research programs in updating the fertilizer formula recommendation of the main cereal (millet) crop of the country and adapting site-specific fertilizer rates for each production environment (agro-ecological region). Bastos et al. (2022) screened 24 N, P, and K dose rates on grain and biomass yield performance of an old (Souna 3) and new (Thialack 2) variety following a rainfall and soil fertility gradient (Diourbel, Nioro, and Sinthiou Malème) in Senegal. Based on previous work, we hypothesize that developing site-specific fertilizer recommendations for other millet-production zones and dual-purpose millet (SL 28, SL 423 and SL169) varieties will contribute to identifying NPK rates that optimize millet grain and fodder yields across millet-production zones in Senegal. The objective of this study was to determine integrated agronomic management practices (planting density, variety, and fertilizer rate) for improved dual-purpose millet varieties, geared towards increased productivity and profitability in millet-production zones in Senegal.

2. Materials and Methods

2.1. Study Area

Field experiments were conducted in Bambey (14°43′12″ N, 16°36′41″ W; in Diourbel region), Nioro du Rip (13°45′0″ N, 15°48′0″ W; in Kaolack region), Boulel (14°18′18″ N, 15°31′48″ W; in Kaffrine region), and Sinthiou Malème (13°51′04″ N, 13°53′04″ W; in Tambacouda region) in 2021 (Figure 1). The climate in Diourbel is Sahelo-Sudanian and Coastal Sudanian, characterized by a mean annual temperature of 27–28 °C and mean annual rainfall values of about 520 mm from 1991 to 2016 [22]. The Kaolack region is characterized by a Coastal Sudanian climate with an average annual rainfall of 900 mm from 1991 to 2016 (Climate Research Unit of East Anglia, 2019). The soil types in Bambey (Diourbel) are Fe-rich tropical sandy soils, and slightly leached (Tappan et al., 2004). Nioro (Kaolack region) is located in the “Saloum Agricultural Region” (6413 km²) where the soils are Fe-rich tropical and ferralitic soils, and loamy sands over fine sandy loam at depth [23]. In Sinthiou Malème (Tambacouda region) and Boulel (Kaffrine region), the average annual rainfall ranged between 700 and 800 mm, and the experimental trials were on a tropical ferruginous soil, commonly called Dior soils, with a sandy texture.

2.2. Treatments and Management

This research was implemented in Senegalese Agricultural Research Institute’s (ISRA) research stations in different regions (Bambey, Nioro, Boulel, and Sinthiou Malème). The experimental design at each location was a split-split-plot design with three replications. The main treatments were four millet varieties. Three varieties (Souna 3, Thialack 2, and SL 28) were planted at all sites. In addition to these three varieties, one variety (i.e., SL 169) was recommended for Bambey and Nioro, and SL 423 was included for Boulel and Sinthiou Malème. In the sub-plots, the traditional planting density of 0.9 m × 0.9 m (12,500 seed hills ha⁻¹, E1) was compared with an increased planting density of 0.9 m × 0.45 m (25,000 seed hills ha⁻¹, E2). The sub-sub-plots consisted of eight fertilizer combinations with the composition described as D1 to D8 (

Table 1. Nutrient management details under different treatment and locations.

Boulel	Sinthiou Malème	Bambey	Nioro
D1 = 0N-0P-0K	D1 = 0N-0P-0K	D1 = 0N-0P-0K	D1 = 0N-0P-0K
D2 = 70N-10P-23K	D2 = 70N-10P-23K	D2 = 91N-13P-23P	D2 = 91N-13P-23K
D3 = 5 t/ha manure	D3 = 5 t/ha manure	D3 = 80N-10P-20K	D3 = 80N-12P-25K
D4 = 70N-13P-19K	D4 = 70N-13P-19K	D4 = 70N-15P-20P	D4 = 57N-10P-19K
D5 = 70N-22.5P-22.5K	D5 = 70N-22.5P-22.5K	D5 = 70N-22.5P-22.5K	D5 = 70N-22.5P-22.5K
D6 = 74N-14P-27K	D6 = 74N-14P-27K	D6 = 95N-17P-27P	D6 = 95N-17P-27P
D7 = 70N-10P-19K + 2.5 t/ha manure	D7 = 70N-10P-19K + 2.5 t/ha manure	D7 = 70N-10P-19K + 2.5 t/ha manure	D7 = 70N-10P-19K + 2.5 t/ha manure
D8 = 78N-18P-31K	D8 = 78N-18P-31K	D8 = 98N-20P-30P	D8 = 98N-20P-30P

). They were 4.5 × 7 m (31.5 m²) and comprised 5 rows for density E1 and 10 rows for density E2. To prevent potential fertilizer contamination between adjacent fertilizer treatments, buffer zones of 1.5 m were installed around each plot. The tested fertilizer combinations in this study are those identified by Batos et al. [21] in comparison to the control (D1) and the current country-recommended inorganic fertilizer rate (D5). The other combinations include D2 (higher grain yield); D3 (higher fodder yield); D4 (higher grain and fodder yield); D6 (current country organo-mineral fertilization recommendation); D7, which is D2 plus 25 kg ha⁻¹ of NPK (15-15-15); and D8, which includes D2 plus 50 kg ha⁻¹ of NPK (15-15-15). For these combinations, simple mineral fertilizers of urea (46% N), double super phosphate (25% P2O5), potassium chloride (KCl: 61.3 K2O), and complete fertilizer NPK (15-15-15) were used. Nutrient levels of cow manure at 12% moisture were 51 mg L⁻¹ potassium, 13% total carbon, 1.3% total nitrogen, and 0.3% total phosphorus on average.

All experimental sites were under two-to-three years fallow prior to the experimental setup. After mechanical ploughing, sowing was carried out manually under rainfed conditions without any fertilizer addition at planting. Nutrient treatments (full treatment amounts of P and K) were hand applied at the soil surface around seed hills and buried at 15 days after planting. The first split of N rate and the full rate of P and K were applied 15 days after planting (DAP) about 5 cm from the plant. The second split of N was applied at 45 DAP. Weeding was carried out frequently by hand to avoid contamination between treatments, and plots were maintained weed free. Disease and pest management consisted of the removal of infected plant materials.

2.3. Data Collection

Millet plant height, leaf chlorophyll index (SPAD value), total number of tillers, number of productive tillers, and fodder and grain yield were collected. Millet plant height was measured at 30, 45, and 60 DAP on eight adjacent plants in the middle row per plot. The chlorophyll index was taken at 45 DAP using a Soil Plant Analysis Development SPAD-502Plus^®meter. The measurements were made at middle of four leaves per plant, and the mean values were recorded [24]. The total tiller number and the number of tillers with panicles were collected at 45 DAP on the eight adjacent plants in the middle row per plot. Millet plants were harvested at 90–100 DAP at all the experimental sites. Millet panicles and fodder were hand harvested at physiological maturity from the center (16.2 m²) of each plot and oven dried at 60 °C to constant weight and yields reported based on 12.5% moisture content [4]. The panicles were threshed and weighed to determine grain yield. The grain and fodder yield from each plot was estimated in kg ha⁻¹.

In Bambey, the gravimetric soil moisture content was collected at harvesting time at different depths (0–20 cm, 20–40 cm, 40–60 cm, and 60–80 cm). The wet weight (P_h) of samples was determined in situ, whereas the dry weight (P_S) was determined in the laboratory after oven drying at 105 °C until a constant dry weight was attained. Soil moisture (SM) was determined by the following formula (Equation (1)).

$$SM (\%)=\frac{\left(P_h-P_s\right)}{P_s}*100$$

(1)

The agronomic-use efficiency of the inorganic N fertilizer for grain (AEg (kg kg⁻¹)) was calculated as the difference of grain yield (kg ha⁻¹) between D2 to D8 (D_x) and D1 divided by the amount of N applied (kg ha⁻¹) with the NPK fertilizer in D_x (Equation (2)).

$$AEg=\frac{Grain yield D_x-Grain yield D_1}{applied N dose}$$

(2)

Given the importance of the millet fodder for livestock and for soil fertility recycling in semi-arid regions, the agronomic-use efficiency of the NPK fertilizer for the fodder yield (AEf (kg kg⁻¹)) was also calculated (Equation (3)).

$$AEg=\frac{Fodder yield D_x-Fodder yield D_1}{applied N dose}$$

(3)

2.4. Statistical Analyses

Statistical analyses were performed with SAS 9.4 statistical software [25]. Prior to the analysis of variance (ANOVA), the normal distribution of the data was assessed using the Shapiro–Wilk normality test [26] and homogeneity of variance was evaluated using the Bartlett test [27]. The ANOVA was conducted for each site using PROC MIXED procedure. Millet varieties, planting density, and fertilizer treatments were used as a fixed effect, while block was considered as random effect. Significant fixed effects were further dissected by extracting means and performing Tukey’s Honestly Significant Difference pairwise comparisons with an alpha of 0.05. Relationships between grain and fodder yield and the millet growth variables were assessed using the Pearson correlation test. Regression analyses were performed to explore the relationship between soil moisture at harvest and the grain and fodder yields.

3. Results

3.1. Influence of Variety, Sowing Density, and Fertilization on Millet Growth

The ANOVA showed that the effects of variety, planting density, fertilization, and variety × density interaction were significant (p < 0.05) on most millet growth parameters in all sites (

Table 2. p-value of ANOVA test of varieties, planting density, and fertilization on millet growth in four regions in Senegal.

Source	Ddl	Height at 30 DAP (cm)	Height at 45 DAP (cm)	Height at 60 DAP (cm)	SPAD Unit	Total Tiller Number	Productive Tiller Number
Boulel
Varieties (V)	3	<0.0001	<0.0001	<0.0001	0.1455	<0.0001	0.012
Density (D)	1	0.3799	0.4978	0.0593	0.0435	<0.0001	0.009
Fertilization (F)	7	0.6871	0.0666	0.0004	0.0045	0.003	0.0278
V × D	3	0.0896	0.0271	0.0311	0.0325	0.2317	0.8776
V × F	21	0.8931	0.9046	0.3305	0.7215	0.645	0.3851
V × D × F	7	0.9996	0.9986	0.9896	0.6594	0.9036	0.7521
Sinthiou Malème
Varieties (V)	3	0.6831	0.0036	0.2444	0.6359	0.0089	0.0072
Density (D)	1	0.0215	0.1073	0.4369	0.0814	0.9876	0.9546
Fertilization (F)	7	0.815	0.0828	<0.0001	<0.0001	<0.0001	<0.0001
V × D	3	0.6173	0.9144	0.8621	0.5146	0.2367	0.0232
V × F	21	0.9965	0.7813	0.8881	0.9412	0.1882	0.2227
V × D × F	7	0.6573	0.6517	0.6909	0.1263	0.8184	0.4032
Bambey
Varieties (V)	3	0.0699	0.4506	0.4415	0.0038	0.4619	0.0042
Density (D)	1	0.5583	0.807	0.0337	0.4702	<0.0001	<0.0001
Fertilization (F)	7	0.9087	0.018	0.3365	0.013	0.0085	<0.0001
V × D	3	0.6044	0.7691	<0.0001	0.0104	0.0476	0.0882
V × F	21	0.9737	0.6948	0.7696	0.3399	0.8367	0.8014
V × D × F	7	0.9831	0.9915	0.8731	0.9632	0.8589	0.7899
Nioro
Varieties (V)	3	0.3765	<0.0001	0.1916	0.0901	0.235	0.0038
Density (D)	1	0.3199	0.2865	0.2127	0.0291	0.045	<0.0001
Fertilization (F)	7	0.8036	0.6691	0.0023	<0.0001	0.006	<0.0001
V × D	3	0.4371	0.0002	0.0052	0.5489	0.172	0.0005
V × F	21	0.9544	0.9963	0.7433	0.0002	0.557	0.5409
V × D × F	7	0.8800	0.9979	0.8147	0.0011	0.860	0.9406

). There were also main effects of density and fertilization. Due to interactions between variety and planting density, their main effects were not reported.

In all sites, the response of millet variety differed with planting density (

Table 3. Millet growth variable as affected by variety x planting density in four regions in Senegal.

Density	Varieties	Height at 30 DAP (cm)	Height at 45 DAP (cm)	Height at 60 DAP (cm)	SPAD Unit	Total Tiller Number	Productive Tiller Number
	Boulel
E1: 12,500 seed hills/ha	SL 28	19.84 c	57.16 b	93.21 cd	48.66 ab	2.67 e	0.93 ab
	SL 423	26.74 b	78.62 ab	107.40 bc	50.08 a	3.23 de	1.21 ab
	Souna 3	29.81 a	88.03 a	120.67 ab	49.03 ab	4.09 bcd	1.27 a
	Thialack 2	26.78 b	78.08 a	122.01 ab	47.18 b	3.78 cd	1.26 a
Mean		25.79 X	75.47 X	110.82 X	48.73 X	3.44 Y	1.16 X
E2: 25,000 seed hills/ha	SL 28	19.88 c	53.67 b	82.94 d	48.67 ab	3.52 de	0.80 b
	SL 423	28.80 ab	85.74 b	125.71 ab	49.78 ab	4.82 abc	0.99 ab
	Souna 3	30.82 a	88.56 a	113.62 abc	49.83 ab	5.15 ab	0.98 ab
	Thialack 2	30.79 a	97.05 a	131.82 a	50.30 a	5.51 a	1.11 ab
Mean		27.57 X	81.25 X	113.52 X	49.64 X	4.75 X	0.97 Y
	Sinthiou Malème
E1: 12,500 seed hills/ha	SL 28	66.26 a	141.32 ab	235.19	46.87	15.63 ab	5.46 b
	SL 423	63.89 c	142.37 ab	233.54	45.36	15.74 ab	6.63 a
	Souna 3	61.50 c	139.58 b	229.13	46.39	15.45 b	6.48 a
	Thialack 2	63.19 b	149.67 a	226.78	47.74	16.42 a	6.14 ab
Mean		63.71 Y	143.23 X	231.16 X	46.59 X	15.81 X	6.12 X
E2: 25,000 seed hills/ha	SL 28	67.04 a	140.07 ab	229.84	47.18	15.82 ab	6.03 ab
	SL 423	68 a	137.26 b	231.65	47.91	15.73 ab	6.22 ab
	Souna 3	67.68 a	132.91 b	230.88	47.93	15.71 ab	5.98 ab
	Thialack 2	65.86 b	143.28 a	222.32	47.88	15.98 ab	6.46 a
Mean		67.145 X	138.38 X	228.67 X	47.72 X	15.87 X	6.17 X
	Bambey
E1: 12,500 seed hills/ha	SL 169	22.35	52.30 b	173.60 b	56.91 ab	14.13 b	5.00 b
	SL 28	24.43	59.09 a	183.78 a	60.84 a	15.93 ab	4.41 c
	Souna 3	26.28	59.16 a	178.51 a	48.20 b	16.31 a	5.40 b
	Thialack 2	21.2	49.05 b	165.48 b	59.12 a	15.89 a	5.20 b
Mean		23.56 X	54.9 X	176.59 Y	56.26 X	15.56 X	5.00 X
E2: 25,000 seed hills/ha	SL 169	23.72	54.96 b	177.94 a	52.12 b	13.43 abc	5.76 b
	SL 28	23.45	51.58 b	180.04 a	51.90 b	11.51 c	6.50 a
	Souna 3	25.53	62.01 a	170.88 b	57.48 ab	12.76 c	7.28 a
	Thialack 2	23.59	56.64 a	183.08 b	50.50 b	13.15 bc	6.42 a
Mean		24.19 X	56.74 X	182.66 X	53.29 X	12.47 Y	6.73 Y
E1: 12,500 seed hills/ha	Nioro
	SL 169	36.15	136.00 ab	247.26 a	53.69	10.81	3.17 cd
	SL 28	38.35	121.97 c	242.16 ab	53.63	11.63	3.43 bcd
	Souna 3	39.20	138.16 ab	236.50 b	55.58	11.95	3.10 d
	Thialack 2	37.06	151.70 a	242.00 a	53.01	12.03	3.13 cd
Mean		37.69 X	136.95 X	241.98 Y	53.97	11.61 X	3.2 Y
E2: 25,000 seed hills/ha	SL 169	36.88	99.01 c	245.00 a	55.87	10.92	4.53 bc
	SL 28	48.40	120.81 c	244.00 ab	54.92	10.28	4.79 ab
	Souna 3	40.65	140.61 ab	239.11 b	55.55	10.79	4.19 a
	Thialack 2	40.73	152.03 a	245.07 a	54.34	11.42	6.69 b
Mean		41.66 X	128.11 X	243.29 X	55.17	10.85 X	4.8 X

For each variable, means followed by same letter are not significantly different, Tukey test, p < 0.05.

). At Boulel, millet varieties Thialack 2, Souna 3, and SL 423 sown at a density of 25,000 seed hills ha⁻¹ gave the greatest heights at 30 DAP, 45 DAP, and 60 DAP as well as the greatest values of total tiller number. At Sinthiou Malème, there was no significant difference between millet height at 30 DAP and 45 DAP for SL 28 and Thialack 2 under the two planting densities, but the response of SL 423 and Souna 3 was significantly different (Table 3). Thialack 2 at 12,500 seed hills ha⁻¹ and Thialack 2 at 25,000 seed hills ha⁻¹ were consistently among the greatest in total number of tillers and in the number of fruiting tillers, respectively, at Sinthiou Malème and all locations except Boulel. In Bambey, Thialack 2 and Souna 3 at 25,000 seed hills ha⁻¹ and SL 28 and Souna 3 sown at 12,500 seed hills ha⁻¹ presented the greatest heights at 45 DAP and 60 DAP, respectively. In addition, chlorophyll index and total tiller number were greater for varieties Thialack 2 and SL 28 or Thialack 2 and Souna 3, respectively, at the normal density, while the number of fruiting tillers were greater at an increased planting density for the same (Thialack 2, SL 28, and Souna 3) varieties at Bambey. At Nioro, average height was not significantly different among varieties at 30 DAP at any planting density and for varieties Souna 3 and Thialack 2 at 45 DAP. The number of fruiting tillers was generally greater at the higher density for all varieties at Nioro compared with normal density.

Fertilizer application had no significant effect on plant height at 30 and 45 DAP in all locations. At Boulel, D5 gave the greatest height at 60 DAP compared with D8, the greatest chlorophyll index and total tiller number compared with D1, and the greatest fruiting tiller number compared with D4 (

Table 4. Millet growth variable as affected by fertilization in four regions in Senegal.

Fertilizer Treatments	Height at 30 DAP (cm)	Height at 45 DAP (cm)	Height at 60 DAP (cm)	SPAD	Total Tiller Number	Productive Tiller Number
Boulel
D1 = 0N-0P-0K	26.16	73.26 ab	107.87 ab	46.88 b	3.12 b	0.92 ab
D2 = 70N-10P-23K	25.8	75.09 ab	107.17 ab	48.68 ab	4.26 a	1.18 ab
D3 = 5 t/ha manure	27.2	78.83 ab	110.17 ab	48.96 ab	4.24 a	1.04 ab
D4 = 70N-13P-19K	25.46	70.64 b	106.35 ab	49.18 ab	4.11 ab	0.86 b
D5 = 70N-22.5P-22.5K	28.65	91.91 a	133.83 a	49.77 a	4.6 a	1.36 a
D6 = 74N-14P-27K	27.09	81.55 ab	114.45 ab	50.3 a	4.24 a	1.06 ab
D7 = 70N-10P-19K + 2.5M	26.48	77.90 ab	115.23 ab	49.43 ab	4.33 a	1.16 ab
D8 = 78N-18P-31K	26.61	77.74 ab	102.33 b	50.34 a	3.88 ab	0.98 ab
Sinthiou Malème
D1 = 0N-0P-0K	65.54	131.84	204.19 b	42.51 b	10.88 c	4.63
D2 = 70N-10P-23K	66.5	149.44	206.89 b	45.82 ab	16.03 b	6.25
D3 = 5 t/ha manure	64.36	140.44	229.85 a	47.88 a	16.31 b	6.38
D4 = 70N-13P-19K	62.4	142.82	233.25 a	47.75 a	16.19 b	6.6
D5 = 70N-22.5P-22.5K	66.61	148.73	236.27 a	49.24 a	16.42 b	6.63
D6 = 74N-14P-27K	64.85	142.43	229.33 a	48.23 a	16.50 b	6.85
D7 = 70N-10P-19K + 2.5M	66.27	148.16	243.89 a	47.94 a	17.88 a	5.59
D8 = 78N-18P-31K	66.89	142.6	227.66 a	47.9 a	16.28 b	6.47
Bambey
D1 = 0N-0P-0K	25.67	51.2	158.02 b	56.09 b	11.67 b	4.46 c
D2 = 91N-13P-23P	23.48	58.11	173.71 ab	56.99 ab	13.99 ab	5.74 ab
D3 = 80N-10P-20K	23.91	57.1	183.01 ab	59.00 ab	14.90 a	5.92 ab
D4 = 70N-15P-20P	23.08	51.7	167.94 ab	59.15 ab	14.71 a	6.10 ab
D5 = 70N-22.5P-22.5K	23.51	61.2	186.80 a	60.49 ab	14.11 ab	5.70 abc
D6 = 95N-17P-27K	23.24	52.76	177.24 ab	60.61 ab	15.05 a	6.33 ab
D7 = 70N-10P-19K + 2.5M	23.07	55.86	188.38 a	60.94 a	13.51 ab	5.23 bc
D8 = 98N-20P-30K	24.59	56.87	173.22 ab	63.78 ab	15.17 a	6.49 a
Nioro
D1 = 0N-0P-0K	39.12	121.25	226.49 b	50.81 d	8.56 b	2.83 d
D2 = 91N-13P-23K	38.59	128.71	244.04 ab	54.78 bcd	11.42 a	3.58 abc
D3 = 80N-12P-25K	38.49	126.22	232.30 ab	51.55 bc	11.81 a	3.2 cd
D4 = 57N-10P-19K	36.98	129.82	248.56 a	57.02 a	11.67 a	3.73 ab
D5 = 70N-22.5P-22.5K	39.59	134.85	244.29 ab	57.90 a	11.71 a	3.94 a
D6 = 95N-17P-27K	38.39	134.95	245.61 a	56.07 ab	11.93 a	3.63 ab
D7 = 70N-10P-19K + 2.5M	38.61	136.42	243.59 ab	52.89 bcd	11.90 a	3.39 bc
D8 = 98N-20P-30K	39.36	131.07	236.21 ab	55.57 ab	11.77 a	3.75 ab

For each variable, means followed by same letter are not significantly different, Tukey test, p < 0.05.

). At Sinthiou Malème, D3, D4, D5, D6, and D8 gave the greatest heights and chlorophyll index at 60 DAP, and there was no significant difference in plant height and chlorophyll content at 60 DAP for D3, D4, D5, D6, D7, and D8. On the other hand, the greatest total tiller number was observed with D7. At Bambey and Nioro, D1 (the control treatment) resulted in consistently lower average heights, in addition to chlorophyll index, total tiller number, and fruiting tiller numbers, than top-performing treatment(s).

3.2. Influence of Variety, Sowing Density, and Fertilization on Millet Yield

The main effects of variety, planting density, and fertilization were significant on millet grain and fodder yields in most regions (

Table 5. p-value of ANOVA test of varieties, planting density, and fertilization on millet yield in four regions in Senegal.

Source	DDL	Grain Yield	Fodder Yield
Boulel
Varieties (V)	3	<0.0001	0.0005
Density (D)	1	<0.0001	<0.0001
Fertilization (F)	7	0.0001	0.0945
V × D	3	0.1879	0.0088
V × F	21	<0.0001	<0.0001
V × D × F	21	0.0506	<0.0001
Sinthiou Malème
Varieties (V)	3	0.0194	0.0355
Density (D)	1	<0.0001	<0.0001
Fertilization (F)	7	<0.0001	<0.0001
V × D	3	0.8068	0.5637
V × F	21	0.0212	0.0039
V × D × F	21	0.0767	0.0425
Bambey
Varieties (V)	3	0.1408	0.002
Density (D)	1	<0.0001	<0.0001
Fertilization (F)	7	<0.0001	0.0049
V × D	3	0.2412	0.0132
V × F	21	0.3463	0.2894
V × D × F	21	0.2888	0.3261
Nioro
Varieties (V)	3	0.1408	0.002
Density (D)	1	<0.0001	<0.0001
Fertilization (F)	7	<0.0001	0.0049
V × D	3	0.2412	0.0132
V × F	21	0.3463	0.2894
V × D × F	21	0.2888	0.3261

). Variety x density interaction was significant for fodder yield at Boulel, Bambey, and Nioro. The effects of variety x fertilization interaction were highly significant on millet grain and fodder yields in Boulel and Sinthiou (Table 5). Due to interactions between varieties and planting density, in most regions, their main effects were not reported.

Millet variety response varied among environments (Figure 2), and grain yields obtained with 25,000 seed hill ha⁻¹ was higher than typical farmers’ practice (12,500 seed hill ha⁻¹). The difference was greater at Boulel, where the average grain yield ranged from 470 to 681 kg ha⁻¹ under 12,500 seed hill ha⁻¹ and from 1500 to 2040 kg ha⁻¹ under 25,000 seed hill ha⁻¹ (Figure 2). At Sinthiou Malème, grain yield was not significantly different among millet varieties at 12,500 and 25,000 seed hill ha⁻¹. In Bambey, the highest grain yield was obtained with Thialack 2 (1495 kg ha⁻¹) at 25,000 seed hill ha⁻¹, whereas the least grain yield was obtained with SL 28, Souna 3, and Thialack 2 at 12,500 seed hill ha⁻¹ (1080, 1100, and 1037 kg ha⁻¹, respectively, for SL 28, Souna 3, and Thialack 2). The highest grain yields were obtained with Thialack 2 and SL 169 when planted at 25,000 seed hill ha⁻¹ in Nioro (averaged 1900 and 1950 kg ha⁻¹, respectively).

Fodder yields were higher at 25,000 seed hill ha⁻¹ compared with 12,500 seed hill ha⁻¹ regardless of millet variety (averaged 4200 kg ha−1 for 25,000 seed hill ha⁻¹ vs. 3100 kg ha⁻¹ for 12,500 seed hill ha⁻¹) for all environments (Figure 3). The difference between 25 planting densities was greatest at Sinthiou Malème and Nioro, where this difference ranged between 30 and 40% compared to 10–15% in Boulel and Bambey. Among varieties, Thialack 2, SL 28, SL 169, and SL 423 produced greater fodder yields at all sites.

Fertilizer response differed between environments (Figure 4). In Boulel and Sinthiou Malème, D7 yielded greater grain yield (1500 kg ha⁻¹) while D6 and D8 yielded the most grain yield in Bambey (1533 and 1525 kg ha⁻¹, respectively). In Nioro, grain yield was greatest under D6 (1726 kg ha⁻¹), D7 (1712 kg ha⁻¹), and D8 (1771 kg ha⁻¹). Higher fodder yields were obtained under D3 (3200 kgha⁻¹), D5 (3042 kg ha⁻¹), D7 (3050 kg ha⁻¹), D8 (2989 kg ha⁻¹) at Boulel; D7 (4870 kg ha⁻¹) at Sinthiou; D2 (4260 kg ha⁻¹), D5 (4500 kg ha⁻¹), D4 (4280 kg ha⁻¹), and D8 (4492 kg ha⁻¹) at Bambey; and D5 (5218 kg ha⁻¹) and D8 (4990 kg ha⁻¹) at Nioro (Figure 5).

Table 6. Summary of Person’s correlation coefficients between grain and fodder and growth parameters for millet.

Variables	H1	H2	H3	SPAD	TT	FT	GY
Boulel
GY	0.235 **	0.242 **	0.08 ns	0.182 *	0.248 **	0.76 ***
FY	0.051 ns	0.684 ***	0.86 ***	0.184 **	0.84 ***	0.033 ns	0.534 ***
Sinthiou Malème
GY	0.138 ns	0.192 ***	0.335 ***	0.249 ***	0.489 ***	0.737 ***
FY	0.021	0.156 *	0.198 **	0.216 **	0.664 ***	0.205 ***	0.7 ***
Bambey
GY	0.045 ns	0.581 ***	0.612 **	0.696 **	0.361	0.91 ***
FY	0.344 ***	0.445 ***	0.422 ***	−0.159 ns	0.006 ns	0.089 ns	0.718 ***
Nioro
GY	0.024 ns	0.47 **	0.113 ns	0.267 ***	0.036 ns	0.838 ***
FY	0.042 ns	−0.16 ns	0.109 ns	0.128 ns	−0.079	0.785 ***	0.613 ***

H1: height at 30 DAP in cm; H2: height at 45 DAP in cm; H3: height at 60 DAP in cm; SPAD: SPAD value at 45 DAP; TT: total tiller number; FT: fruiting tiller number; GY: grain yield in kg ha⁻¹; FY: fodder yield in kg ha⁻¹; ns: non-significant at the 0.05 probability level. * Significant at the 0.05 probability level. ** Significant at the 0.01 probability level. *** Significant at the 0.001 probability level.

summarizes the correlation coefficients between grain and fodder and growth parameters. The chlorophyll index, total number of tillers, and number of fruiting tillers were positively and significantly correlated with grain and fodder yield at most sites. This means that as the values of these parameters increase, grain yield and fodder yield also increase. The correlation is particularly high (˃70%) between the number of fruiting tillers and grain yield for all sites.

3.3. Influence of Variety, Sowing Density, and Fertilization on Soil Moisture

Figure 6 shows the variation in soil moisture under millet varieties and planting densities at Bambey. The soil moisture was very low under both densities, especially under the normal density (4–8%). The interaction between millet varieties and seeding rate was significant (p = 0.0078) on soil moisture and was greater for all varieties (Thialack 2: 12.32%; Souna 3: 5.32%; SL28: 6.32%; and SL423: 9.23%) at higher planting density compared with normal density (Thialack 2: 9.25%; Souna 3: 3.21%; SL28: 5.43%; and SL423: 7.47%).

3.4. Grain and Fodder Yield Versus Soil Moisture at Harvest

Regression lines for soil moisture at harvest versus fodder yield were similar for all varieties (Figure 7). The slope obtained from the regression analysis was around 0.0021 for all varieties. The soil moisture at harvest was highly correlated with fodder yield, and the effect was more pronounced for SL 28 (Figure 7A), SL 169 (Figure 7B), and Thialack 2 (Figure 7D) than Souna 3 (Figure 7C). The soil moisture at harvest was highly correlated with grain yield for SL 28 (R² = 0.70, slope = 0.0058), whereas a weak relationship was observed between soil moisture at harvest and grain yield for Souna 3 (R² = 0.30, slope = 0.0036) and Thialack (R² = 0.41, slope = 0.0033). For SL 169, the relationship between soil moisture at harvest and grain yield was not significant (Figure 7B).

3.5. Agronomic-Use Efficiency of the Applied Inorganic Fertilizer

In general, the planting density and millet varieties had a significant effect on the AEg of the N supplied by the inorganic NPK fertilizer (

Table 7. Effect of the planting densities on the agronomic-use efficiency for grain (AEg) and fodder (AEf) of N applied with fertilizer for each millet variety.

Planting Density	Varieties	Grain Yield Increase (kg ha−1)	AEg of N Fertilizer (kg/kg N Applied)	Fodder Yield Increase (kg ha−1)	AEf of N Fertilizer (kg/kg N Applied)	Grain Yield Increase (kg ha−1)	AEg of N Fertilizer (kg/kg N Applied)	Fodder Yield Increase (kg ha−1)	AEf of N Fertilizer (kg/kg N Applied)
		Boulel				Sinthiou Malème
E1: 12,500 seed hills/ha	SL 28	7.01 cde	0.34 cde	683.02 c	9.02 b	571.83 abc	7.69 abcd	867.38 b	11.774 b
	Souna 3	385.61 bc	2.05 cde	−1971.93 d	−24.43 c	492.56 bc	6.63 cd	1076.81 b	13.563 b
	Thialack 2	310.35 bcd	1.95 bcd	2072.69 a	20.02 a	697.56 ab	9.36 abc	1121.22 b	14.826 b
	SL 423	161.49 de	1.28 de	1038.98 bc	12.67 ab	375.71 c	5.070 d	1043.05 b	14.189 b
Mean		291.11 Y	1.81 Y	455.69 Y	4.32 Y	534.41 Y	7.19 Y	1027.12 Y	13.59 Y
E2: 25,000 seed hills/ha	SL 28	1106.22 a	14.54 a	862.29 c	10.97 b	484.71 bc	6.609 cd	2597.89 a	35.13 a
	Souna 3	336.90 bc	4.06 bc	494.28 c	6.19 b	561.81 abc	7.617 bcd	2802.21 a	37.94 a
	Thialack 2	206.25 bcd	1.16 bcd	1616.53 ab	20.59 a	827.44 a	11.228 ab	3208.25 a	43.68 a
	SL 423	504.33 b	6.739 b	1046.55 bc	13.07 ab	833.48 a	11.365 a	2810.67 a	37.93 a
Mean		538.43 X	6.63 X	1004.92 X	12.71 X	676.86 X	9.20 X	2854.76 X	38.67 X
	Bambey					Nioro
E1: 12,500 seed hills/ha	SL 169	27.05 ab	−0.04 ab	1969.85 a	23.61 a	531.74 bc	6.64 bc	729.57 bc	9.59 bc
	SL 28	285.51 b	3.365 b	2115.23 a	27.29 a	653.78 ab	8.48 ab	478.32 bc	5.58 bc
	Souna 3	473.04 a	5.724 a	1225.96 a	15.64 a	617.48 b	7.81 bc	1329.35 b	16.98 b
	Thialack 2	407.14 a	4.819 a	854.12 a	10.87 a	441.75 bc	5.71 bc	−405.56 c	−5.34 c
Mean		298.19 X	3.58 Y	1541.30 Y	19.35 X	561.19 X	7.16 X	532.92 Y	6.07 Y
E2: 25,000 seed hills/ha	SL 169	505.17 a	5.993 a	2258.41 a	27.34 a	578.39 bc	7.49 bc	3454.81 a	44.63 a
	SL 28	299.80 b	3.969 b	634.22 a	7.78 a	955.79 a	12.16 a	3056.56 a	39.32 a
	Souna 3	367.09 a	4.512 a	2348.44 a	29.20 a	286.87 c	3.68 c	3154.44 a	40.22 a
	Thialack 2	314.01 b	4.205 b	1791.90 a	21.31 a	556.34 bc	6.94 bc	3578.91 a	46.54 a
Mean		371.52 X	4.67 X	1758 X	21.41 X	594.35 X	7.57 X	3311.18 X	42.68 X

For each variable, means followed by same letter are not significantly different, Tukey test, p < 0.05.

). The highest N AEg was in the increased density (25,000 seed hills/ha), where it averaged 6.63 kg kg⁻¹ at Boulel, 9.20 kg kg⁻¹ at Sinthiou Malème, 4.67 kg kg⁻¹ at Bambey, and 8.32 kg kg⁻¹ at Nioro. N AEg was negative with SL 169 under the lower planting density in Bambey. Within each planting density, the AEg of N was significantly different among millet varieties and was greater with SL 28 and SL 423 at Boulel, Thialack and SL 423 in Sinthiou Malème, Thialack 2 and SL 169 at Bambey, and SL 28 and Thialack 2 in Nioro. AEg of N was also higher at 25,000 seed hills/ha compared to 12,500 seed hills/ha, but its values did not vary significantly among millet varieties within the planting densities except at Nioro, where Thialack 2 recorded a negative AEg of 5.34 kg kg⁻¹. AEg significantly varied among fertilizer combinations at Boulel and at Bambey where the highest AEg was obtained with D5 (5.53 kg kg⁻¹) and D6 (4.66 kg kg⁻¹), respectively; AEg did not vary significantly among D2–D8 in Sinthiou Malème and Nioro (

Table 8. Effect of the fertilizer combinations on the agronomic-use efficiency for grain (AEg) and fodder (AEf) of N applied.

Fertilizer Combinations	Grain Yield Increase (kg ha−1)	AEg of N Fertilizer (kg/kg N Applied)	Fodder Yield Increase (kg ha−1)	AEf of N Fertilizer (kg/kg N Applied)
	Boulel
D1 = 0N-0P-0K	0.00 e	0.00 d	0.00 c	0.00 c
D2 = 70N-10P-23K	190.37 ab	2.72 bc	776.39 b	11.09 a
D3 = 5 t/ha manure	463.64 a	0.229 d	1026.35 a	0.315 b
D4 = 70N-13P-19K	239.40 c	3.42 b	710.09 b	10.14 a
D5 = 70N-22.5P-22.5K	387.61 b	5.53 a	864.78 b	12.35 a
D6 = 74N-14P-27K	108.49 d	1.466 bc	762.57 b	10.30 a
D7 = 70N-10P-19K + 2.5M	161.19 d	2.32 bc	880.41 b	12.16 a
D8 = 78N-18P-31K	259.22 c	3.70 b	821.83 b	11.74 a
	Sinthiou Malème
D1 = 0N-0P-0K	0.00 c	0.00 b	0.00 c	0.00 b
D2 = 70N-10P-23K	711.74 ab	10.16 a	2006.96 b	28.67 a
D3 = 5 t/ha manure	635.08 b	9.77 a	1802.61 b	27.73 a
D4 = 70N-13P-19K	632.34 b	9.03 a	2114.48 ab	30.20 a
D5 = 70N-22.5P-22.5K	744.23 a	10.63 a	2208.26 ab	31.54 a
D6 = 74N-14P-27K	590.18 b	7.97 a	2207.97 ab	29.83 a
D7 = 70N-10P-19K + 2.5M	845.62 a	8.21 a	2850.21 a	27.67 a
D8 = 78N-18P-31K	685.90 a	9.79 a	2336.98 ab	33.38 a
	Bambey
D1 = 0N-0P-0K	0.00 e	0.00 e	0.00 c	0.00 c
D2 = 91N-13P-23P	149.00 c	1.63 c	2060.59 a	22.64 b
D3 = 80N-10P-20K	125.93 c	1.57 b	1350.36 b	16.88 b
D4 = 70N-15P-20P	41.05 a	0.58 d	2077.45 a	29.67 ab
D5 = 70N-22.5P-22.5K	16.83 d	0.24 d	2299.78 a	32.85 a
D6 = 95N-17P-27K	443.03 b	4.66 a	1814.11 b	19.09 b
D7 = 70N-10P-19K + 2.5M	151.02 c	2.15 b	1234.90 b	17.82 b
D8 = 98N-20P-30K	524.33 a	5.35 a	2360.96 a	24.09 b
	Nioro
D1 = 0N-0P-0K	0.00 b	0.00 b	0.00 b	0.00 b
D2 = 91N-13P-23K	616.88 a	6.77 a	2210.12 a	24.28 a
D3 = 80N-12P-25K	615.90 a	7.69 a	2086.37 a	26.08 a
D4 = 57N-10P-19K	529.73 a	9.29 a	1948.69 a	34.18 a
D5 = 70N-22.5P-22.5K	763.06 a	10.90 a	2633.14 a	37.61 a
D6 = 95N-17P-27K	691.91 a	7.28 a	2069.96 a	21.78 a
D7 = 70N-10P-19K + 2.5M	677.14 a	9.54 a	2086.97 a	29.69 a
D8 = 98N-20P-30K	727.53 a	7.42 a	2341.17 a	23.89 a

For each variable, means followed by same letter are not significantly different, Tukey test, p < 0.05.

).

4. Discussion

Although crop productivity in sub-Saharan Africa has increased over the last decade, per capita food and nutrition security have decreased. As such, resilient farming systems that aim not only to increase productivity but also to ensure its sustainability as well as adaptation to changing climatic conditions will improve the livelihoods of smallholders and provide a pathway out of poverty. We evaluated the impact of two plant densities and eight fertilizer combinations on growth and yield (fodder and grain) of four pearl millet varieties and soil moisture across the rainfall gradient in Senegal. In most of the studied sites, the greatest millet growth and yield were obtained with the dual-purpose varieties (Thialack 2, SL 169, SL 423, and SL 28) compared to Souna 3. These dual-purpose varieties stayed green at maturing and had a larger panicle length, which possibly had a positive impact on grain yield. The dual-purpose millet varieties have been selected to produce a high quantity of quality fodder without a trade-off in grain production. Our results agree with Gupta and Ndoye [28], who found Souna 3 to be lower yielding and that its yield was least stable across environments. Our results also showed that the dual-purpose varieties (SL 28, SL 169, and Thialack 2) are efficient in N use (Table 7) and conserve soil water compared to Souna 3 (Figure 6). This illustrates that fodder and grain are correlated with soil moisture at harvest, and higher fodder yield and grain yield are associated with higher soil moisture. This agreed with the results of previous study in maize crop productivity in Centre Benin [24]. These findings showed that dual-purpose millet can help smallholder farmers deal with trade-offs between the competing uses of scarce land for food and feed production [29]. Bell et al. [30] postulated that dual-purpose varieties can help to improve the profitability, environmental sustainability, and resilience of the whole farm system substantially compared to traditional varieties.

Planting millet at 25,000 seed hills ha⁻¹ (0.90 m × 0.45 m) was suitable for growth and yield (20–60% across the environments). Higher millet fodder and grain yields were obtained by planting at 25,000 seed hills ha⁻¹. This finding suggests that increasing the millet seeding rate will supply more grain for human consumption and fodder for livestock. In addition, more biomass production will allow at least a portion to be left in fields to provide soil cover and contribute more soil microbial activity, SOM, and improve soil fertility [31]. In the present study, the highest grain yields in each location were obtained with the highest sowing density (Figure 2), which can be explained by the fact that high densities increase light-use efficiency and possibly reduce water loss because of increased soil cover. Consequently, the plants under a high sowing density in this study produced more fodder (Figure 3), as it limits evaporation and increases soil moisture (Figure 6), and efficiently uses the N applied through fertilization (Table 7). Sivakumar and Salaam [32] reported that rapid early growth of millet leaves can contribute to a reduction in soil evaporative losses and increased water-use efficiency. Gregory et al. [33] and Cooper et al. [34] showed that early development of canopy cover helps crops to intercept more radiation, increase root development, and apportion more of the water extracted by the roots to transpiration. However, optimum plant density depends on growth habit and can differ among environments [35]. The optimum is defined as a point beyond which the competition between plants for light, water, and nutrients becomes important and can lead to decreased crop yields [36]. Buerkert et al. [37] reported that dense planting gives a statistically significant increase in grain yield under the marginal conditions of low fertility, low average rainfall, and frequent droughts. Nevertheless, grain yields typically increased with increased stand density in the years with regular rainfall [38].

Soil fertility is important for determining the productivity of all farming systems. Overall, lower millet height, chlorophyll index, total tiller number, number of fruiting tillers, and fodder and grain yields were observed with the control (D1: 0N-0P-0K), and the optimal fertilizer combination varied among regions. This variation across the experimental location can be explained by the difference in the local conditions (i.e., soil fertility, rainfall, and temperature). Maintaining the optimum combination of plant density and fertilizer amount, especially N, allows better growth and leads to higher yields [38]. Klaij et al. [39] recommended higher plant densities to achieving greater yields. However, significant increases in density should be made only in conjunction with appropriate fertilizer use to avoid nutrient limitations and competition among plants. We observed that fertilizer response differed between environments, but manure application (2500 kg ha⁻¹ manure) in combination with 70N-10P-19K provided higher grain and fodder yields in most of the sites. This result supports the importance of using the right nutrient sources. As reported by Tounkara et al. [3], long-term continuous cultivation without sufficient organic matter restoration may degrade soils to such an extent that they may become “non-responsive” to inorganic fertilizers with low use efficiency of the applied nutrients. This can be achieved through the utilization of assorted soil nutrients enhancing approaches such as animal manure, organic residues, and suitable inorganic fertilizer [40]. Fatondji et al. [41] argued that the application of good-quality organic amendments increases pearl millet yields. Brouwer and Powell [42] recommended not to exceed 2500 kg manure ha⁻¹ to minimize leaching of nutrients. However, the nutrient treatment rates identified in the current study were relatively high, especially for N, when compared to previous studies in west Africa (60–14–25 kg NPK ha⁻¹ [43]; 61–14–26–5000 kg NPKM ha⁻¹ [44]; 15–37–26–2000 kg NPKM ha⁻¹ [45]). The P rate that we identified is smaller compared to the current fertilizer recommended rate for millet (70N–22.5P–22.5K) in Senegal. Rockstrom and De Rouw [46] found that small quantities, 25 or 50 kg ha⁻¹ of N and 10 kg ha⁻¹ of P, buried next to individual pockets, were sufficient to double or triple the grain yields of millet in normal rainfall years. In a two-year experiment, Akponikpè et al. [47] observed that 2700 kg ha⁻¹ manure and 15 kg N + 4 kg P ha⁻¹ led to an average increase in grain yield of 132% compared to the unfertilized treatment.

Based on this, it is evident that the adoption of dual-purpose varieties with adequate management (plant density and fertilization) can improve grain and fodder yields of millet across Senegal. However, there is still limited information regarding the genetic resource, major production challenges, dissemination of production packages, and social factors related to dual-purpose variety production. A farmer’s decision to adopt a new variety is a complex process governed by productivity, social, economic, and technical factors [17]. Trait preferences of farmers further influence adoption, but these are not well documented for dual-purpose varieties [48]. Understanding farmers’ perspectives and preferences for specific traits of dual-purpose crops is critical to inform breeding and the targeting and development of improved dual-purpose varieties by researchers [49]. Therefore, further research strategies are required to alleviate problems in dual-purpose varieties production in sub-Saharan Africa.

5. Conclusions

The results showed that using dual-purpose millet varieties and adequate management (increased density and fertilizer management) improved grain and fodder yield and soil water content and agronomic-use efficiency for grain and fodder of the N applied. Grain and fodder yield responses varied by environment and increased with increasing sowing density (over 1700 vs. <1000 kg ha⁻¹ for grain and 4200 vs. 3100 kg ha⁻¹ for fodder). The responses of millet varieties to fertilizer differed by environment, but application of 70N–10P–19K + 2.5 t ha⁻¹ cow manure provided a greater grain and fodder yield in all locations. The soil water content measured at harvest was greater for all varieties (Thialack 2: 12.32%; Souna 3: 5.32%, SL28: 6.32%; and SL423: 9.23%) at higher planting density (25,000 seed hills ha⁻¹) compared to 12,500 seed hills ha⁻¹ (Thialack 2: 9.25%; Souna 3: 3.21%; SL28: 5.43%; and SL423: 7.47%). Thus, increasing millet sowing density could be a good management strategy to increase resilience of millet farming systems. For both grain and fodder qualities of these dual-purpose millet varieties, additional analyses are needed to evaluate the influence of increased sowing density and fertilizer on grain and forage quality and its impact on the productivity of livestock.