The experimental field sites were located at the University of Sarh in southern Chad (09˚04.51'N, 18˚25.22'E, at 375 m elevation), while in Cameroon, it was established within the University campus (07˚24.61'N, 13˚34.24'E, at 1155.8 m elevation). Trials were carried out during the cropping season extending from October 2015 to Febuary 2016 at Sarh-Chad (temperature: 25˚C - 39˚C, rainfall: 800 - 1200 mm/year), and from March to July 2015 at Ngaoundere-Cameroon (temperature: 22˚C - 28˚C, rainfall: 1500 - 2500 mm/year).

Groundnut (Arachis hypogaea L.), a flower variety 11 provided by the Chadian Institute for Agricultural Research and Development (ITRAD) of Bebedja in Chad, has a life cycle of about 90 days. It is a plant with yellow flowers, pinnate leaves that give at maturity two or very rarely three seeds per pod. Some pods may have only a single seed. Bambara groundnut (Vigna subterranea (L.) Verd.), commonly known as voandzou was a local Djar variety, also supplied by ITRAD. It has 90 to 100 days at maturity, with seeds having beige black hile outline. Flowers are yellow in colour, leaves are trifoliolate, while pods give at maturity one, but very rarely two seeds. Rhizobium strains were produced as liquid inoculum from groundnut, Bambara groundnut cowpea, and soybean root nodules at the microbiology laboratory of the Institute for Agricultural Research and Development (IRAD) at Wakwa, Ngaoundere in Cameroon.

Soils were sampled before sowing and after harvest. Physico-chemical analysis of soils was carried out following AFNOR method [15] at the Chadian Institute for Agronomic Research and Development (ITRAD) laboratory of N'Djamena.

The laboratory-adapted experimental set-up consisted of 2 randomized complete block fields, one field per legume crop species. Each block has 6 plots randomly arranged according to different treatments: control (Ctrl), Groundnut rhizobia (GR), Cowpea (CR), soybean rhizobia (SR), Bambara groundnut rhizobia (BR), and the Mixture of rhizobia (MR) from the 4 legume crop species. These six treatments were replicated uniformly in the three blocks numbered I, II, III, for a total of 18 plots per field trial. The experimental unit was a rectangular plan of (3 × 2) m². Plots were separated 0.50 m apart, while blocks were separated by a distance of 1 m.

Control plots were sown first. Each plot had 7 lines with 9 sowing holes each, giving a total of 63 plants per plot. The sowing interval between plants was 25 cm, and 30 cm between lines (Ngakou et al. [14] ). The distance between the outer lines and the plot edges was about 25 cm. For each cross-inoculation formula- tion, seeds were coated with 100 ml of liquid inoculum solution, and 25 g of Nido milk serving as adhesive carrier. Different treatments were applied to seeds at sowing time. Weeding was done on the 12, 26 and 45 DAP with hoes and hands, while adding soften soil around each plant to avoid exposure of pods out of the soil surface.

The plant size was measured from the crown to the last leaf at 30 and 60 DAP on 30 randomly selected plants per treatment (Zakaria [16] ). These measurements were performed early in the morning before 8 a.m. to avoid wind disturbances. Plants were dried at 65˚C in a P-Selecta oven (0˚C to 250˚C) for 48 hours and then, weighed using a Gilbertini brand scale at 0.001 g sensitivity as previously reported (Ngakou et al. [12] ).

All data were subjected to the Analysis of Variance (ANOVA) using the Stat Graphics plus software program. Treatment means were compared using the Duncan multiple Range Test at 5% significance. Correlations between variables were determined using the statistical package for social science (SPSS).

Table 1 sumarizes the physico-chemical composition of soils prior to sowing of crops and after harvest. The soil of Sarh was sandy clay in texture, and was slightly rich in silt, nitrogen and organic matter compared to that of Ngaoundere, but the latter was rich in phosphorus with the C/N ratio 3 folds higher than that of Sarh soil before sowing. After harvest, the soil organic matter content was reduced by about 50% in Ngaoundere-Cameroon, while at Sarh-Chad, there was a slow increase in soil organic matter content. This result suggests that the cultivation of legumes on Cameroonian soil contributed to a loss of carbon stocks following the mineralization of organic matter. This mineralization would

For each granulometric or organic matter parameter in a column and within a study site, values affected by the same letter are not significantly different at the indicated level of probability.

have been enhanced by favorable growth conditions to mineralizing bacteria present in the soil. This indicates that plants have used organic matter from the soil (Ngakou et al. [14] ). The decrease in the soil organic matter content is in agreement with other results obtained in the range of between 20% and 30% (Ngakou et al. [10] ). An increase of 20.66% in soil nitrogen content was observed after harvest at Ngaoundere-Cameroon compared to 3.27% at Sarh- Chad.

The enhanced soil C/N ratio by 40.34% in Chad after harvest justifies the increment of carbon in the soil. The phosphorus contents were 25.88% and 38.09%, respectively in Sarh and Ngaoundere soils, and this could be attributed to the presence of phosphorus solubilizing bacteria in the soil. At Sarh during harvest, soil pH increases towards neutrality, suggesting that free hydronium ions were fixed during the transformation of phosphorus complexes into assimilable phosphorus (Ndeye [17] ; Nwaga et al. [18] ).

At 30 DAP, groundnut plant size was improved by CR, MR and BR respectively by 5.93%, 17.79% and 19.49%, compared to the control at Sarh (Table 2). Plants

For each parameter in a column, values affected by the same letter are not significantly different at the indicated level of probability. GR: Groundnut Rhizobia, CR: Cowpea Rhizobia, SR: Soybean Rhizobia, BR: Bambara groundnut Rhizobia, MR: mélange des quatre Rhizobia.

inoculated with BR, MR were significantly taller (p = 0.002) than those inocu- lated with GR, SR and uninoculated plants. The size of plant treated with cowpea rhizobia was significantly higher (p < 0.0001) than that of plant treated MR, as well as the control plants at 30 DAP in Ngaoundere.

Inoculation of Bambara groundnut plants with rhizobia accelerated their growth height (Table 2) at 30 DAP. This increment in plant size varied from 69.16% to 97.15% compared to that of untreated plants at Sarh. The size of plants treated with SR was more important than that of other treatments. This suggests that these bacteria may have contributed to the production of growth hormones that have accelerated the plant growth in size. Inoculation of cowpea plants with two strains of rhizobia was revealed to greatly increase the stem size (Abed Nour el Houda [19] ). Results obtained from this study line with other findings which reported plant size ranging from between 20 and 38 cm de- pending on the seed varieties used (Ndiang et al. [20] ).

At Ngaoundere, inoculation of Bambara groundnut with nitrogen-fixing bacteria resulted in a considerable improvement in plant size (Table 2) at 30 DAP, from between 25.28% and 88.57% compared to untreated plants. Plants treated with BR significantly have greater size (p = 0.0001) than plants of other treatments. This result is above those pointed out from a research performed in Adamawa (Gomoung [21] ). Inoculation of Bambara groundnut with rhizobia and mycorrhiza was claimed to positively improve the plant biomass (Ngakou et al. [14] ). The positive effect of inoculum on plant size can be explained by the production of plant growth hormones such as auxin and gibberellins (Vargas et al. [22] ). In a related research, soybean inoculated with Bradyrhizobium strain was revealed to exhibit greater biomass than non-inoculated plants through AIA production by bacteria symbiont (Afzal et al. [23] ).

The use of the inoculum significantly improved (p = 0.006) the plant biomass of groundnut plant at Sarh (Table 2). The biomass of inoculated plants was more elevated and varied from 16.62% in CR to 45.45% in MR treated compared to that of uninoculated plants. The enhanced plant biomass was reported to be attributed to soils nutrient improvement by symbiotic bacteria (Ama-Abina et al. [24] ). At Ngaoundere, the plant biomass of untreated groundnut plants was significantly (p = 0.0001) higher than that of inoculated plants, indicating that indigenous rhizobia were more efficients on groundnut than in introduced strains. However, plants inoculated with CR and SR had higher dry biomass than those inoculated with GR and MR.

At Sarh, the most important Bambara groundnut plant biomass at 30 DAP was obtained from SR-inoculated plants. There was a highly significant difference between the biomass of CR, BR, MR, SR treated plants (Table 2). The inoculum improved the Bambara groundnut biomass from 39.33% to 87.86% as compared to that of the control. GR did not affect in anyway the Bambara groundnut biomass. This result lines with another finding which yielded an average biomass of 5 g per plant (Ngakou et al. [14] ).

At Ngaoundere rhizobia inoculum moderately improved the groundnut plants biomass (Table 2). BR treatment contributed to 73.70% improvement of bio- mass as compared to the control. There was a positive and significant correlation (r = 0.63, p = 0.0001) between the size and Bambara groundnut biomass.

Seed yield (kg∙ha⁻¹) of rhizobia inoculated groundnut was improved from 9.83% to 63.73% compared to the control at Sarh (Figure 1(a)). The highest yield was obtained from CR treated plants. This result was favored by the improvement of soil fertility through atmospheric nitrogen fixation (Dey et al. [31] ), and the solubilization of soil phosphorus by specific rhizobia, making it available for the host plant (Andriamananjara [32] ). In pure culture, inoculation of plant with Rhizobium was claimed to be an alternative to improve plant growth and productivity (Diop et al. [33] ).

At Ngaoundere, the enhanced seed yield in groundnut inoculated treatments was attributed either to improvement of soil fertility through atmospheric

nitrogen fixation (Badawi et al. [34] ), or the soil phosphorus solubilizing rhizobia (Mekhemar et al. [35] ). Other reported results have indicated seed yield of between 975 kg∙ha⁻¹ and 1.432 t∙ha⁻¹ (Ndiaye [26] ), bellow our findings. Seed yield per hectare of untreated plants (1.57 t∙ha⁻¹) was found to be statistically equal to the average pod yield (1.63 t∙ha⁻¹) of five cultivars of groundnuts introduced into northern Cameroon (Betdogo et al. [25] ).

Results on Figure 1 indicate that Bambara groundnut seed yield was in the range between 491.25 and 910.375 kg∙ha⁻¹ depending on the treatments at Sarh. The performance of the BR treatment was greater (p < 0.0001) than that of the control. BR improved the seed yield of Bambara groundnut by 72.71%. These results are in agreement with recent findings which reported yield values varying between 300 to 800 kg∙ha⁻¹ (Brink [36] ), or between 524.25 kg and 912.1 kg∙ha⁻¹ in Bambara groundnut dually inoculated in the field with rhizobia and mycorhiza (Ngakou et al. [14] ), but are twice those from which seed yield of between 130 to 470 kg∙ha⁻¹ in was obtained (Salon et al. [30] ).

The highest seed yield was that of CR treatment (699.77 kg∙ha⁻¹), whereas the lowest was observed (Figure 1(b)) in SR treatment (327.26 kg∙ha⁻¹) at Ngaoun- dere. Rhizobial inoculation at sowing was reported to increases seed dry weight and yield (Mnasri et al. [37] ). This yield increment was attributed to the improvement in mineral nutrition of plants, as well as seeds content in N and P, following inoculation (Ngakou et al. [12] ). Similar to our findings, grain yield varying from 485 to 1322 kg∙ha⁻¹ depending on Bambara groundnut varieties has been reported (Pungulani et al. [38] ).

The affinity between yield parameters of each studied crop was express through the Pearson correlation (Table 4). A positive and significant correlation (r = 0.5, p < 0.0001) was detected between the number of seeds and the dry weight of the pods at Ngaoundere. At Sarh, the correlation between the number and dry weight of pods was positive and highly significant (r = 0.55, p < 0.0001). Similarly, the the pod weight highly and significant correlated (r = 0.59, p < 0.0001) with the number seeds per pod on the same site. On the overall, all the yield parameters within each of the studied site positively and significantly correlated each other.