Soybean (Glycine Max L.) Grain Yield Response to Inoculation with Novel Bradyrhizobia Strains Across Different Soil Fertility Conditions in Zimbabwe

Abstract

The agronomic effectiveness of biofertilizers is influenced by strain origin, genetic identity, crop genotype, soil type, and environmental conditions. For best results, both the plant and rhizobia strain must be adapted to the common harsh soil conditions in the tropics. While plant varieties have changed over the years, complementary research on new strains effectiveness under varying soil fertility conditions has lagged in southern Africa. Seven field experiments were established in the main soybean-producing areas of Zimbabwe in the north, central, and north–east regions to evaluate agronomic benefits of new rhizobia strains against the current exotic commercial strain (MAR1491). One site was irrigated (site 3), and the other six sites were rainfed (sites 1, 2, 4, 5, 6, and 7). While trends in inoculation response varied from site to site due to site conditions, inoculation with the strains NAZ15, NAZ25, and NAK128 consistently yielded high grain yields, which were similar to the current commercial strain MAR1491 and to application of mineral fertilizer (51.75 and 100 kg N ha⁻¹). Grain yield levels were generally below 2 t ha⁻¹ for sites 2, 3, and 5 and above 2 t ha⁻¹ for sites 1, 4, and 6, while for the irrigated site 3, they ranged upwards of 3 t ha⁻¹. When irrigated, all strains except NAK9 performed similarly in terms of grain yields and aboveground N uptake. Further testing on the inclusion of the indigenous strains NAZ15, NAZ25, and NAK128 in multi-strain commercial inoculant production targeting application in regions and soils where they excel beyond the current exotic strain MAR1491 is recommended.

1. Introduction

Production of soybean (Glycine max) in smallholder farming areas of Zimbabwe was popularized in the nineties due to the activity and support of the Soybean Promotion Taskforce [1]. From 2020 to 2023, the average national hectarage dedicated to soybean production was 41,234 ha, producing on average 63,588 t of grain, with average national yields of 1.6 t ha⁻¹ [2]. This production quantity represents only 29% of the annual national soybean requirement for food, animal feed, and other industries (220,000 t [3]). The average yield per hectare includes large-scale commercial production, but soybean yield in smallholder farming areas is usually much less than the 1.6 t ha⁻¹ national average due to multiple factors, which include rainfall, inherent soil fertility, low use of fertilizers, and lack of inoculation. An average yield of 1.6 t ha⁻¹ is, however, low compared to potential yields of above 5 t ha⁻¹ under irrigation in tropical environments [4]. The soybean crop is a strategic choice for national, regional, and global food and nutrition security because of its high protein (35–45%) and oil (15–25%) content. Soybean seeds contain 23–33% carbohydrates and 4% fiber and provide essential vitamins and minerals like potassium (K), magnesium (Mg), calcium (Ca), zinc (Zn), iron (Fe), and copper (Cu), along with antioxidants [5,6]. With rising population and the risks associated with climate change and variability, it is important to investigate strategies that improve soybean productivity as well as benefits realized from symbiotic nitrogen (N₂) fixation.

The use of N₂-fixing rhizobium inoculants is recommended to economically and sustainably supply the N needs of a soybean crop as opposed to using inorganic nitrogenous fertilizers. Benefits of inoculants include maintenance of high plant N supply due to symbiotic N₂ fixation and high grain yields. Nitrogen fixation is affected by several factors, including the presence, abundance, and efficiency of rhizobia; the amount of nitrogen in the soil; the plant’s genetic makeup and age; its interaction with rhizobia; and shifts in the soil’s physical and chemical properties [5,7]. Grain yield benefits could be constrained by harsh environmental conditions that affect the growth of the host legume, the microsymbiont (rhizobia), and the bacteria/plant symbiotic relationship [8]. Before rhizobium can form nodules with soybean, it must survive adverse soil conditions such as moisture stress, soil acidity, and high temperatures, known to reduce rhizobia survival in tropical soils [9,10]. Results from numerous field studies around the world show that increasing inoculant dosage could improve survival and therefore have greater yield impacts [11,12]. However, in terms of economics and agronomy, if available, providing a better-suited rhizobium strain is preferred because it is more economical as opposed to providing a less-suited strain in greater quantity to improve effectiveness.

In addition to the physical-environment limitations to rhizobium survival, poor soil fertility limits N₂ fixation by limiting nutrient availability for both rhizobia in soil and a growing legume [13,14]. Earlier studies in Zimbabwe reported low support for rhizobia populations in soils due to poor water-holding capacity and low fertility [15]. A more recent study by Chiduwa [16] reported a similar lack of persistence in soil of the introduced MAR1491 commercial rhizobia strain. To be agronomically effective, rhizobia that is introduced through commercial inoculants must be adapted to the common harsh soil conditions, which will improve survivability within the period from inoculation to successful nodule formation (nodulation). Rhizobia strain selection is found on environmental preference—a principle that states that certain strains are better suited for survival, nodulation, and symbiotic N₂ fixation in particular environments than others. A particular rhizobia’s ability to form nodules relates to plant yield response through nodulation intensity and expression of the Nod genes within the plant [17].

Despite the intensive soybean breeding for high yields and disease resistance that has taken place since the eighties in southern Africa, there is limited published research on isolation and selection of effective indigenous strains that could be more adapted to local conditions for high nodulation efficiency, effectiveness, and N₂ fixation. To fill this gap, recently in South Africa, rhizobia strains have been isolated and evaluated for their ability to nodulate with various soybean cultivars [18]. However, the main limitation of the research reported was the lack of yield data from field trials.

In Germany, three indigenous Bradyrhizobium isolates, GMF14, GMM36, and GEM96, were tested in comparison to an exotic Bradyrhizobium strain (USDA110), in combination with different soybean cultivars, under greenhouse and field conditions. Regardless of soybean cultivar, inoculation with GEM96 resulted in the highest shoot N content of 21.89 g kg^–1, which was similar to the USDA110 strain (21.91 g kg^–1). Additionally, the GEM96 strain promoted greater nodule formation, nodule mass, grain yield (1.38 t ha⁻¹), and protein content (412 g kg⁻¹), matching the performance of the USDA110 strain. These findings underscore the effectiveness of locally adapted Bradyrhizobium strains in enhancing soybean productivity [19]. A report on the higher competency of 4 native rhizobia isolates on soybean growth and nodulation, compared with commercial strains, under glasshouse conditions in Ethiopia, is also available [5]. The presence of effective indigenous rhizobia populations, which can relate to soybeans, was confirmed in Zimbabwe [20]. However, there is a lack of data on soybean yield response to inoculation with newly isolated rhizobia under different fertility conditions in Zimbabwe. This is despite the addition of newly isolated rhizobia strains to the Soil Productivity Research Laboratory (SPRL) Microbial Culture Bank through two parallel isolation studies under the N2 Africa project (https://n2Africa.org). Currently, soybean rhizobia inoculants in Zimbabwe are produced from a Bradyrhizobium diazoefficiens USDA110 strain (MAR1491) originally isolated over sixty-five years ago in Florida, USA.

The objectives of this study were to evaluate the effect of inoculation with recently isolated rhizobia strains on soybean grain yield and to assess how different soil fertility conditions influence the inoculation response. The main (alternate) hypothesis was that there were significant differences in the soybean yield response of recently isolated Bradyrhizobium strains compared to the exotic commercial strain.

2. Materials and Methods

2.1. Experimental Sites and Rhizobia Strain Details

Field experiments were established in the north, central, and north–east regions of Zimbabwe (Figure 1), which collectively account for approximately 94% of the total annual soybean production of Zimbabwe. Sites 1–3 were located at the Soil Productivity Research Laboratory (SPRL), Kushinga Phikelela Agricultural College (KPC), and the Horticultural Research Institute (HRI), all situated in Marondera district and classified as agro-ecological region IIa. This region is characterized by a mean annual rainfall of between 750 and 1000 mm with mean maximum annual temperatures that range from 23 to 27 °C This region constitutes the most agriculturally productive region in the country, second only to agroecological region I. Sites 4, 5, 6, and 7 were in the Wedza, Makoni, Mhondoro, and Chinhoyi districts, respectively, and represent communal farming areas. Sites 6 and 7 are classified as agro-ecological region IIb (750–1000 mm annual rainfall, 25–28 °C maximum temperatures), while sites 5 and 4 are classified as region III (650–800 mm annual rainfall; 25–28 °C maximum temperatures) and IV (450–650 mm annual rainfall; 27–29 °C maximum temperatures), respectively. All experimental sites were rainfed to mimic smallholder farmer conditions, except for site 3 (HRI), where drip irrigation was installed to supplement rain during the trial duration. Site 3 represented more commercial and resource-rich farmers.

The treatments consisted of 6 rhizobia strains, 5 of which had been isolated in an earlier phase of the N2 Africa project; a positive control with inorganic N fertilizer (at 51.75 or 100 kg N ha⁻¹ for rainfed and irrigated conditions, respectively); and a negative control without any N fertilizer or inoculant applied. Rhizobia strains NAZ15, NAZ21, and NAZ25 were isolated in 2011 from Goromonzi, Wedza, and Murehwa soils in Zimbabwe (and later deposited in the SPRL Microbial Culture Bank and renamed MAR1667, MAR1673, and MAR1677, respectively). Rhizobia strains NAK9 and NAK128 were isolated in Kenya within the same project. Rhizobium strain MAR1491 served as a reference strain because it is the commercial strain for soybean rhizobia inoculants in Zimbabwe.

Each strain was cultured in a standardized mother culture solution and mixed with a uniform carrier material, following the SPRL inoculation batch preparation protocol. This facilitated random quality control and quality assurance protocols for purity and rhizobial population analysis. Results from serial dilutions and plating on yeast extract mannitol (YEM) agar plates revealed that each inoculant sachet had at least 10⁹ viable rhizobial cells g⁻¹ of carrier media and no other identified species.

2.2. Trial Establishment and Setup

At all the farmer sites (sites 4–7), land was tilled using ox-drawn plows, while tractors were used for the on-station sites (1–3). A basal dressing of single superphosphate (SSP) fertilizer was applied to all plots at a rate of 200 kg ha⁻¹ (19% P₂O₅) to supply 16.6 kg P ha⁻¹. Trials were arranged in a randomized complete block design (RCBD) with eight treatments and four replicates with blocking against the slope at all sites. Each treatment plot measured 3 × 4 m². Inoculants with specific strains were applied as a slurry to soybean seeds using clean and sanitized implements treated with 70% ethanol. At each site, inoculated seeds were immediately planted, one treatment at a time to avoid rhizobia cross-contamination. Soybean (variety SC Safari from Seedco Ltd., Harare, Zimbabwe) was planted with an inter-row spacing of 0.3 m × 0.1 m within the row. Two seeds per station were planted and later thinned to one plant per station. The -N treatment control plots did not receive any N fertilizer from inoculant or a mineral form. At the rainfed sites (1, 2, and 4–7), the +N treatment received ammonium nitrate as a top-dressing fertilizer at a rate of 150 kg ha⁻¹ (51.75 kg N ha⁻¹), which was split into two equal applications at 7 and 30 days after planting. At the irrigated site 3, the rate of N was doubled to 100 kg N ha⁻¹ and similarly split as at other sites. All the plots at all the sites were routinely weeded as needed using hand hoes throughout the experiment period. Implements were sanitized between treatments. Application rates of N and P employed in this study were typical of what most smallholder farmers would apply as general recommendation rates for Zimbabwe based on soybean removal rates of 60 kg N ha⁻¹ and 15.3 kg P ha⁻¹ from soil [21].

2.3. Soil Sampling and Soil Chemical Analysis

At all sites, soil samples were collected for baseline soil fertility analyses. Composite samples were derived from ten subsamples (0–15 cm depth) before the establishment of the experiments. The composite samples were analyzed for pH, organic carbon (OC), mineral nitrogen (N), available phosphorus (P), and exchangeable potassium, calcium, and magnesium (K, Ca, and Mg) [22]. Mineral N was measured as ammonium and nitrate nitrogen after extraction in 2 M KCl (10 g soil/100 mL) for 1 h. Available P was determined colorimetrically [23] after extraction using the resin bag extraction method [24]. Exchangeable cations were determined after extraction in 1 M NH₄OAc for 30 min followed by filtering, and Mg content was read on an atomic absorption spectrophotometer, while K and Ca were read on a flame photometer [24].

2.4. Plant Material Sampling and Analysis

Due to unforeseen operational constraints that could not be resolved within the timeframe of this study, nodulation and biomass assessments were performed only for sites 1–3 at the flowering stage, and N uptake and estimation of N₂ fixation were performed only for the irrigated site 3. However, grain yield data and grain N content assessment were performed for all sites. For sites 1 and 2, an average of three plants were randomly sampled per plot. For site 3, ten plants per plot were randomly sampled. Whole plants were dug up carefully to preserve the roots, and the aboveground biomass was separated from the roots at the base of the stem. The roots were washed over a 1 mm sieve, blot dried using clean paper towels, and then root nodules were removed and counted. The fresh mass of the nodules was thereafter recorded. The aboveground plant biomass was dried to a constant mass in an oven at 70 °C and recorded.

At all sites, grain yields were determined after harvesting pods at physiological maturity from a net plot of 1 m², avoiding the plot borders by at least a meter. The pods were dried (to a constant mass), shelled, and the grain was weighed. Plant and grain samples were ground and analyzed for total N using the micro-Kjeldahl technique [22]. A 0.2 g plant sample was digested with concentrated H₂SO₄ (98%) in a Gerhardt Kjeldatherm^® digester, then distilled in 50% NaOH with N collection in a boric acid mixture, and then further titrated with 0.07 M H₂SO₄. Nitrogen uptake in grain was calculated by multiplying the %N content with the total grain mass.

For the irrigated site 3, nitrogen uptake was calculated by multiplying the average %N content in all aboveground plant parts with the total aboveground mass (including grain). Nitrogen derived from fixation (Dff) was estimated by subtracting the average N uptake of the uninoculated control from the N uptake in the inoculated treatments. The assumptions of this estimation were that the uninoculated control plants had poor nodulation and ineffective nodules. The N uptake from uninoculated plants was assumed to be mainly from available nitrogen in the soil and represented on average N uptake from soil by inoculated plants, assuming that all plants had similar root morphology and soil N uptake [25].

2.5. Statistical Analysis

All statistical analysis was performed in IBM SPSS software (Version 29.0.2.0 (20), IBM, Chicago, IL, USA). Data on biomass, nodule counts, nodule mass, total grain N uptake, and grain yields were assessed for normality using the Shapiro–Wilk test, and homogeneity of variance was assessed using the Levene statistic. For normally distributed data, a one-way ANOVA was run, and where there were significant differences, Tukey HSD was used to separate significantly different means at a 95% confidence interval. Generally, nodule counts and nodule mass data were not normally distributed, and the independent-sample Kruskal–Wallis test was employed to assess significance with pairwise separation of means thereafter. For normally distributed data with unequal variances (grain yield: site 4; grain P uptake: site 3), the Brown–Forsythe robust test of equality of means was used with mean separation using the Games–Howell statistic.

3. Results

3.1. Study Site Soil Characteristics

Soil fertility characteristics varied substantially across the experimental sites (

Table 1. Selected soil chemical characteristics at the seven experimental sites prior to trial establishment.

Site	Details/Names	pH 1	Soil OC%	Mineral N (ppm)	Avail. P (ppm)	Exchangeable Bases (me%)
						K	Ca	Mg
1	SPRL	4.5	0.35	22	5	0.09	0.64	0.38
2	Kushinga Phikelela College	5.3	1.08	11	37	0.42	2.4	1.51
3	Horticulture Research Institute	4.9	1.28	23	19.6	1.02	3.63	0.20
4	Wedza	5.0	0.23	5	4	0.05	0.73	0.35
5	Rusape/Makoni	5.1	1.35	44	2	0.25	9.41	4.1
6	Mhondoro	4.7	1.06	29	2	0.56	5.35	2.83
7	Chinhoyi	5.6	1.20	28	9	0.74	6.43	4.3

¹ CaCl₂ scale; Avail.—available; me%—milliequivalents per 100 g.

). Soils at the study sites ranged from slightly acidic to acidic, apart from site 7, which was neutral on the CaCl₂ scale (Table 1). Sites 1 and 4 had the most infertile soils, which had very low organic carbon content at 0.35 and 0.23% soil OC, respectively. Mineral N ranged from 5 to 44 ppm, with site 4 recording the lowest. Available P was also very low at the same two sites; however, sites 5 and 6 were acutely deficient in available P. Only site 2 had high available P at 37 ppm. Exchangeable K levels were low at site 1, site 4, and site 5.

3.2. Nodulation Counts and Nodule Mass of Soybeans During Flowering at the Researcher-Managed Sites

The +AN and -N treatments had negligible nodule counts and fresh nodule mass at the rainfed sites (1 and 2), while at the irrigated site (3), they had fresh nodule mass that was comparable to that of the NAZ21 and NAK9 treatments. The strains NAZ15, NAK128, and MAR1491 had significantly higher nodule counts and nodule mass than the uninoculated +AN and -N treatments at all three sites (

Table 2. Nodule counts and nodule fresh mass of soybean inoculated with various strains of rhizobia at the researcher-managed sites in Marondera, Zimbabwe.

	Site 1		Site 2		Site 3
N-Fixing Strain/Fertilizer	Nodules Plant−1	Fresh Nodule Mass Plant−1 (g)	Nodules Plant−1	Fresh Nodule Mass Plant−1 (g)	Nodules Plant−1	Fresh Nodule Mass Plant−1 (g)
NAZ15	36 d ± 1	0.98 c ± 0.07	25 d ± 1	0.38 bc ± 0.09	25 b ± 2	0.79 cd ± 0.11
NAZ21	24 bcd ± 2	0.57 ab ± 0.06	13 cd ± 1	0.38 bc ± 0.05	10 a ± 2	0.42 abc ± 0.05
NAZ25	7 ab ± 1	0.59 ab ± 0.03	11 abc ± 1	0.40 bc ± 0.04	17 ab ± 4	0.82 cd ± 0.16
NAK9	17 abc ± 2	0.60 ab ± 0.05	11 abc ± 1	0.22 b ± 0.02	10 a ± 2	0.43 abc ± 0.04
NAK128	27 cd ± 4	0.84 bc ± 0.07	12 bcd ± 1	0.43 c ± 0.04	22 ab ± 2	1.00 d ± 0.10
MAR1491	24 bcd ± 1	0.88 bc ± 0.04	16 cd ± 1	0.23 bc ± 0.04	21 ab ± 2	0.73 bcd ± 0.10
+AN	1 a ± 0	0.00 a ± 0	0.4 ab ± 0	0.04 a ± 0.02	12 a ± 4	0.37 a ± 0.05
-N	2 a ± 1	0.15 a ± 0.06	0.1 a ± 0	0.01 a ± 0.01	10 a ± 2	0.40 ab ± 0.08
Significance	**	**	**	***	**	***

Superscript letters represent statistical significance at α = 0.05, n = 4. Numbers after ± are standard error of means (SEM). +AN—ammonium nitrate added; -N—no nitrogen fertilization. ** p < 0.01, *** p < 0.001.

). The NAZ25 and NAK9 treatments had comparable nodule counts to the +AN and -N treatments across the same three sites. Although the strain NAZ25 had lower nodule counts than the NAZ15, NAK1, and MAR1491 treatments at the three sites, it had a considerable nodule mass, which was comparable to that of the former treatments at all sites except at site 1 and for NAZ15.

3.3. Biomass Production of Soybean During Flowering at the Researcher-Managed Sites

Biomass production during flowering was significantly influenced by treatment (p < 0.05) at the rain-fed sites 1 and 2 (Figure 2a,b) but not at the irrigated site 3 (p = 0.715) (Figure 2c). The rhizobia strains NAZ15, NAK9, and NAK128 produced soybean biomass yields that were similar to that of the commercial strain MAR1491 at site 1 (Figure 2a). At the same site, NAK128 and MAR1491 produced significantly higher biomass than NAZ21, NAZ25, +AN, and -N treatments. At site 2, while NAK128 produced the highest biomass, it was similar to the NAZ15, NAZ21, NAZ25, and NAK9 treatments (Figure 2b). At both sites, 1 and 2, the +N treatment was not significantly different from the -N treatment, and generally both treatments produced very low biomass compared to the rest of the inoculated treatments.

3.4. Soybean Grain Yields at All Sites

Grain yields across all the treatments were <2 t ha⁻¹ at sites 1, 4, and 6, while only site 3 and site 5 had the +N and all rhizobium inoculated treatments yielding >2 t ha⁻¹ of grain (

Table 3. Soybean grain yields and grain N uptake after inoculating with varying rhizobia strains.

N-Fixing Strain/Fertilizer	Site 1	Site 2	Site 3	Site 4	Site 5	Site 6	Site 7
Grain Yield (t ha−1)
NAZ15	1.84 b ± 0.14	1.47 b ± 0.08	3.31 b ± 0.21	0.81 b ± 0.09	2.15 ab ± 0.16	1.06 ± 0.04	1.12 a ± 0.12
NAZ21	1.31 ab ± 0.17	2.10 c ± 0.07	3.28 b ± 0.10	0.86 b ± 0.13	2.56 b ± 0.22	1.13 ± 0.14	1.63 ab ± 0.11
NAZ25	1.70 b ± 0.12	1.63 b ± 0.06	3.27 b ± 0.25	1.08 b ± 0.16	2.41 b ± 0.30	1.28 ± 0.21	1.73 ab ± 0.24
NAK9	1.08 a ± 0.12	1.45 ab ± 0.09	2.43 a ± 0.13	0.70 b ± 0.08	2.88 b ± 0.17	1.30 ± 0.23	1.60 ab ± 0.19
NAK128	1.69 b ± 0.16	1.59 b ± 0.12	2.90 ab ± 0.07	0.77 b ± 0.06	2.34 b ± 0.18	0.75 ± 0.08	0.83 a ± 0.16
MAR1491	1.63 ab ± 0.14	1.70 bc ± 0.13	3.26 b ± 0.17	0.63 ab ± 0.10	1.99 ab ± 0.28	1.59 ± 0.30	2.35 b ± 0.36
+AN	1.83 b ± 0.08	1.75 bc ± 0.08	3.47 b ± 0.18	0.30 ab ± 0.03	2.29 b ± 0.25	1.07 ± 0.16	1.61 ab ± 0.15
-N	1.39 ab ± 0.06	1.02 a ± 0.08	1.61 a ± 0.07	0.16 a ± 0.06	1.23 a ± 0.19	0.77 ± 0.15	1.49 ab ± 0.20
Significance	***	***	*	***	***	ns	*
Grain N Uptake (kg N / ha−1)
NAZ15	107.7 b ± 7.6	86.4 b ± 4.2	106.20 b ± 7.95	52.4 c ± 5.4	145.2 ab ± 13.6	62.1 ± 7.6	70.9 ab ± 11.6
NAZ21	77.1 ab ± 9.8	103.0 b ± 8.6	106.87 b ± 6.60	58.7 c ± 10.3	156.2 b ± 14.3	60.9 ± 6.3	93.0 ab ± 5.9
NAZ25	99.1 ab ± 7.2	95.5 b ± 7.2	104.58 b ± 9.72	64.4 c ± 7.3	148.3 ab ± 16.4	73.6 ± 12.8	103.8 ab ± 14.5
NAK9	61.6 a ± 6.4	68.2 ab ± 9.0	77.33 b ± 3.72	42.7 bc ± 5.1	182.4 b ± 12.6	71.6 ± 17.3	100.3 ab ± 13.8
NAK128	99.4 ab ± 9.3	92.9 b ± 12.9	66.35 ab ± 10.33	41.8 bc ± 4.8	141.4 ab ± 15.4	37.6 ± 4.8	52.9 a ± 11.5
MAR1491	95.0 ab ± 8.0	94.4 b ± 6.3	97.25 b ± 8.30	41.3 bc ± 7.8	121.0 ab ± 21.9	96.1 ± 21.2	126.5 b ± 20.9
+AN	113.8 b ± 2.7	86.8 b ± 6.5	106.91 b ± 7.58	17.3 ab ± 1.9	150.1 b ± 15.4	51.2 ± 5.7	100.2 ab ± 13.8
-N	67.8 a ± 12.9	43.4 a ± 3.4	50.80 a ± 1.45	10.0 a ± 3.6	76.9 a ± 11.5	42.7 ± 10.7	91.4 ab ± 15.4
Significance	*	***	***	***	***	ns	*

Numbers after ± are standard error of means (SEM), n = 4; superscript letters represent statistical significance at α = 0.05; ns—not significant, * p < 0.05, ** p < 0.01, *** p < 0.001.

). Inoculation or fertilization did not significantly affect (p = 0.57) soybean grain yields at site 6, while there were significant differences (p < 0.05) at the other sites. Inoculation with the strains NAZ15, NAZ25, and NAK128 consistently resulted in significantly (p < 0.05) higher grain yields than the -N treatment at sites 1–5, except for NAK128 at site 3. At the same time, the treatments were comparable to the commercial strain MAR1491 and the +AN treatments. Plants inoculated with NAK9 did not show a significant grain yield advantage compared to the -N treatment at sites 1, 2, 3, and 7. At site 7 and out of all the treatments, only the NAZ15 and NAK128 produced significantly less (p = 0.001) grain yields than the commercial strain MAR1491 (Table 3). Grain yields of soybean that received AN fertilizer were not significantly different from the -N control at sites 1, 4, 6, and 7.

Generally, a nearly similar trend to that of grain yields was observed for grain N uptake (Table 3). Inoculation or fertilization did not significantly (p = 0.068) affect soybean grain N uptake at site 6. At sites 1–3 and 5, the +AN treatment plots had significantly higher grain N uptake than the -N control. At the same sites, the +AN treatment plots were not significantly different in grain N uptake from the NAZ15, NAZ21, NAZ25, NAK128, and MAR1491 treatment plots. At site 4, the NAZ15, NAZ21, and NAZ25 treatment plots yielded significantly (p < 0.001) higher (at least 67% more) grain N uptake than the +AN treatment. Correlation analysis of pooled data from rainfed sites revealed that there was a significant (p < 0.01) positive correlation between grain yields and soil organic matter, initial available N, and exchangeable Ca and Mg (Figure 3). The same trend was true for grain N uptake and the soil properties.

3.5. Nitrogen Uptake and Estimated Nitrogen Derived from Fixation at Site 3

While biomass production was not significantly different across treatments during the flowering stage at site 3 (Figure 2c), there were significant differences between treatments at the physiological maturity stage. The range of aboveground biomass ranged from 4.1 to 5.5 t ha⁻¹ for the fertilized/inoculated treatments. The NAZ21, NAZ25, NAK128, MAR1491, and +N treatments produced significantly more (roughly double) aboveground biomass than the -N treatment (Figure 4a). Except for NAK128, all inoculated treatments and the +N treatment had comparable shoot N uptake (Figure 4b). The NAZ21 and +AN treatments had the highest N uptake, which was significantly different from that of the NAK128 and -N treatments. Estimated nitrogen derived from symbiotic N₂ fixation followed a similar trend to N uptake; the NAZ21 treatment had the highest amount of N₂ derived from fixation, which was comparable to the rest of the inoculated treatments except for the NAK128 treatment.

4. Discussion

While the lack of nodulation data at four out of the seven sites is a limitation of this study, the data for sites 1 to 3 sufficiently captured Bradyrhizobium strain-induced nodulation patterns in low- and medium-fertility soils, relevant to the objectives. The grain yield and grain N uptake data for all sites and across the strains tested reflect the N₂-fixing ability of the Bradyrhizobia strains in the varied soil fertility conditions. The near-zero nodule counts observed in the uninoculated treatments at sites 1 and 2 suggest that indigenous soybean-nodulating Bradyrhizobia populations were very low in these soils. This is because the soils were acidic, had low OC content, and underwent periods of low soil moisture, all of which limited microbial survival [10]. However, at site 3, the +N and -N treatments had nodulation counts and nodule mass that were comparable to those of NAZ21 and NAZ25. Partly a reason for this could be that the site is regularly irrigated, which reduces soil moisture stress on soil microbes. Moreover, the increased productivity of irrigated land compared to non-irrigated could lead to more microbial diversity and populations due to increased C input in the form of plant roots and residues. Land management significantly affects rhizobia survival and diversity, and activities that lead to increased soil OC, such as manure and biochar application, are reported to positively affect rhizobia abundance [15,26].

In soils such as sites 1 and 2, inoculation response can be very high at low soil N conditions, since competition with native strains will be minimal. Significantly higher nodule counts in inoculated treatments suggest that all the strains tested were very effective at forming nodules with soybean. Though the strain NAZ25 did not necessarily produce high nodule counts at site 1, it had considerable nodule mass at the three sites, indicating its potential as an efficient N₂-fixing strain. Due to the high energy requirements of N₂ fixation, the number of nodules is largely regulated by the host plant through a pathway known as autoregulation of nodulation (AON) [27]. Environmental conditions such as soil nutrient deficiency that negatively affect photosynthesis may therefore lead to greater nodulation restrictions by the plant. This happens by limiting the plant’s carbon resources allocated to the roots [28]. This AON is likely the reason that the same Bradyrhizobium strain differently nodulated the same host in different environments. At the same time, while nodule count is a good indicator of potential N₂ fixation, other studies have shown that taproot-based nodules instead of total nodule counts is a much better indicator of potential N₂ fixation [29].

The low biomass observed in the -N controls shows that a lack of a N source, whether through rhizobia symbiosis or mineral fertilizer, will result in limited vegetative growth of soybeans. Most soils in the smallholder areas are inherently deficient in nitrogen or depleted through extractive production methods [10]. At the national general recommendation rates for nitrogen employed in this study, increases in soybean grain after application of 57.5 kg N ha⁻¹ (rainfed sites) or 100 kg N ha⁻¹ (irrigated site) were comparable to most of the inoculated treatments. This confirms that inoculants can indeed replace mineral N fertilizer in these cropping systems. Despite the high need for N under such N-limiting systems, rhizobia supported by host plants are likely to adjust to meet high plant N requirements.

Soybean grain yields at the rain-fed sites were within the range of the reported Zimbabwe national average of 1.6 t ha⁻¹ [30]. A soybean yield gap decomposition highlighted multiple factors driving suboptimal yields in similar smallholder farm sites in Malawi, Mozambique, and Zambia, including crop management, and confirming genotype x environment and microbe (G × E × M) interaction effects [31]. The overall low grain yields are majorly attributable to poor soil fertility limiting crop productivity and, to some extent, the erratic and low rainfall that characterizes some of these areas. For example, site 4 had the poorest soil fertility, with very low OC, mineral N, K, and available P, which led to the lowest grain yields across all the treatments (<1.1 t ha⁻¹). In addition, reduced rainfall contributed to the low yields. At this site, soybeans exhibited yield non-responsiveness to the addition of ammonium nitrate fertilizer.

The low soil OC content at sites (0.23–1.35%) reported in this study is in line with those reported by Manzeke-Kangara et al. [32] of 0.4–1.5% in southern Africa and those reported by Nyawasha et al. [33] of <0.4% OC in north–east Zimbabwean smallholder farming areas. Low soil OC levels have been shown to result in less efficient N₂ fixation in inoculated Trifolium repens grown in South Africa [34]. Some studies have highlighted the synergic response on N, P, and OC, showing that balanced fertilization is critical in increasing N-use efficiency in such low-fertility soils [35,36]. Except for NAK9 inoculated plots, grain yields from inoculated or +N plots at the irrigated site 3 were within the range of the expected commercial yield level of 3.5 t ha⁻¹.

The trend in soybean grain yields with inoculation strain type was different at each site, suggesting that strain performance was influenced by site-specific factors, including rainfall and soil fertility. It is logistically impractical to tailor the strain used in commercial inoculant production for each smallholder setting, and selection of a strain that is consistent across the ecological zones in Zimbabwe is therefore essential. While inoculation with NAZ15, NAZ25, and NAK128 showed consistently good soybean grain yields on average, their performance was similar to the current commercial strain (MAR1491). This implies that there is limited justification to update the commercial inoculant strain solely on the basis of N₂ fixation. Strain selection should consider additional traits such as ecological fitness, which were not assessed in this work. Additionally, introducing a multi-strain inoculant with at least two strains (e.g., NAZ25 and MAR1491) may provide better agronomic efficiency of seed inoculation in different areas of Zimbabwe. The MAR1491 (USDA110) strain is used as a commercial strain in several African countries, for example, in Kenya [37] and Zimbabwe. The lack of differences between grain yields of inoculated soybean and the +N control implies that there could be no need to apply inorganic fertilizer when soybean has been inoculated with an efficient N₂-fixing rhizobium strain. At the same time, inoculation may yield better yield outcomes than inorganic N addition under severely soil fertility-limiting conditions, as observed for site 4, where inoculated treatments produced double the grain yield and grain N uptake (Table 3). The use of general N and P rates was a limitation in this study; however, the data generated are more realistic than the current farmer rates. While further studies should employ higher rates that are in line with soil nutrient deficiencies with proper correction of soil pH, a key limitation is water availability. Higher inorganic fertilizer rates increase chances of crop failure by osmotic stress when water is limited.

Soil pH and available P were surprisingly not correlated with grain yield and grain N uptake for the rain-fed sites, indicating that when P was supplied at 16.6 kg P ha⁻¹ in the tested soils, P supply was not a major bottleneck to crop productivity. However, this does not imply that soil P was not a limiting factor because total plant P uptake ranges from 15 to 20 kg P ha⁻¹ for a soybean grain yield of 2.5 t ha⁻¹ [38], and most of the sites were deficient in P (Table 1). A low rate of P supply (16.6 kg P ha⁻¹) was used because it represents the ‘general’ recommendation for soybeans and is realistic to what most smallholder farmers would apply without a soil test. This general recommendation is based on soybean removal rates of 60 kg N ha⁻¹, 15.3 kg P ha⁻¹, and 66 kg K ha⁻¹ from soil [21]. Soil OC, initial available N, and exchangeable Ca and Mg were all positively correlated with grain yields. Of all the tested soil attributes, soil OC had the highest moderately strong relationship with grain yield (Spearman coefficient: 0.645). Similarly, in an Africa-wide study by Kebonye et al. [39], plant biomass and soil OC were found to be highly correlated, explained by soil OC creating favorable conditions for nutrient uptake and releasing nutrients through mineralization. Our results highlight the importance of soil OC management and improvement in symbiotic N₂-fixing systems, which could result in increased inoculation response.

The high grain yields (2.9–3.3 t ha⁻¹) and similarity of grain yields in inoculated treatments at the irrigated site 3 (except for the NAK9 treatment) show that these strains have high inoculation response when water is not limiting. The reduction/elimination of soil moisture stress during the critical root nodulation stage enhances nodulation success by improving inoculant strain survival. The relatively high nodule numbers and fresh nodule mass per plant of inoculated plants corroborate this. All inoculated treatments at site 3 had a N uptake in aboveground biomass that was significantly higher than that of the -N treatment, indicating that there was indeed an increased N supply due to inoculation. Estimates of Ndff suggest that the NAZ21 and NAZ25 strains may have superior N₂-fixing ability to MAR1491 under irrigated conditions.

In terms of grain yield and inoculation response compared to other strains, the strain NAZ15 performed well in moderate soil mineral N conditions (sites 1 and 3), but less so when min N was high (site 5) or low (site 2). Considering NAZ25, the strain performed relatively well in low to acutely deficient available P, low OC, and exchangeable K and Ca soil conditions (sites 1, 3, 4, 5, and 6). The NAK9 strain could be better suited to soil conditions with high exchangeable Ca, where it performed well (sites 5 and 6), than low exchangeable Mg and K soil conditions, where grain yield inoculation response was low relative to other strains (sites 1, 2, 3, and 4). The NAK128 strain may be less effective in P-deficient soils, as observed for sites 4, 5, and 6. The commercial strain MAR1491 generally performed well across most sites except at site 4, which was one of the most infertile soils among the sites tested. While these results confirm that MAR1491 is a leading strain in terms of successfully fixing N₂ under many soil conditions in Zimbabwe, it may be less suited to multiple nutrient-deficient soils, such as at site 4. The strain NAZ25 could be better suited than MAR1491 for impoverished soils; however, further field testing is required.

5. Conclusions

Strains native to Zimbabwe (NAZ15 and NAZ25) can enhance nodulation capacity, biomass N uptake, and grain yield of soybean, comparable to the commercial exotic strain. Therefore, the inclusion of Bradyrhizobium strains NAZ15, NAZ25, and NAK128 in commercial multi-strain inoculants is recommended. Introducing a multi-strain inoculant may provide better agronomic efficiency of seed inoculation in different areas of Zimbabwe, and further testing is recommended. Results from this study also revealed that while type of strain is important, inoculation response is influenced by the soil fertility profiles in different parts of Zimbabwe. There is a need to couple soil organic and nutrient management with inoculants for improved N₂ fixation. Further studies of inoculation coupled with biochar addition to soils with very low levels of soil organic matter are recommended.