The site for the experiments was located at Kamboinse in the central plateau of Burkina Faso (12˚26'48"N and 1˚33'45"W). Experiments were performed on the experimental site of the International Institute for Water and Environmental Engineering (2iE) from August to December 2011. The physical and chemical properties of the soil used in the experiments were described in our previous report [25] .

The site for experiments is located in the central plateau of Burkina Faso which supports five months of rainy season (Juin-October) with peaks from July to August, and seven months of dry season (November-Mai). During the period of study (August-December) thirty rainy days with an average rainfall of 284.8 mm were recorded (Figure 1). The average of daily temperature was 28˚C and ranged from 24.6˚C (minimum) to 30.9˚C (maximum) (Figure 2). The daily relative hygrometry (RH), varied from 39% to 78% (Figure 2), with a sunshine average of 7.26 hours/day (Figure 2).

For the experiment, FKR 19 rice variety (Farako-Bâ rice) was used. This variety originates from the Asian species Oryza sativa L (Japonica varietal group), and is adapted to the rainfed rice. The variety results from a cross- breeding performed by the Institute for Environment and Agricultural Research (INERA Farako-Bâ, Burkina Faso).With a cycle of seedling-maturity of 120 days, this rice variety has an average yield estimated to 5 - 6 tons/ha.

Experiments were carried out as previously described by Otoidobiga et al. [25] and Keïta [26] . Twelve microplots

of 1.50 m × 1.20 m × 1.20 m in size (total internal volume: 1.31 m³) were constructed in reinforced concrete to prevent cracks due to temperature changes following the alternation of the rainy and the dry seasons. At the bottom of each microplot (1.45 m depth), perforated PVC pipes were installed and connected to an external tap through the concrete wall for sub-drain the soil (Figure 3). The design was intended to improve circulation of water and oxygen within the rice roots zone. Microplots were filled with the clay-loam soil described above [25] , and then flooded for 1 week. After the flooding period of soil, two 15-day-old plants of FKR 19 rice were transplanted at 20 cm hill intervals. Then, the plants were continuously flooded from rice transplanting until rice maturity and harvest. Three replications and four modes of drainage were performed throughout the study: without drainage (control: T) and drainage for 7 days (P1), 14 days (P2) and 21 days (P3), respectively by regulating the drained water flow from the bottom tap (Figure 3). Chemical fertilizers (NPK and urea) were applied in the microplots according to the recommended doses of 300 kg/ha for N-P-K (14 - 23 - 14) at transplanting and 100 - 150 kg/ha for urea in two further dressings (at rice transplanting, and 60 days after transplanting, respectively) [27] [28] .

Populations of Sulfate-Reducing Bacteria were enumerated in microplots during the rice growth stages by the Most Probable Numbers methods using lactate and sulfate as substrates. The basal medium for enumeration was the medium described by Dianou and Traoré [19] .The medium composition was as follows (per liter of distilled water): [A] sodium lactate 60% (12 ml), K₂HPO₄ (0.35 g), KH₂P0₄ (0.25 g), Na₂S0₄ (4.00 g), MgS0₄, 7H₂O (2.00 g), NaCl (0.5 g), NH₄Cl (2.00 g), yeast extract (1.00 g), resazurin solution 0.1% (1 ml), 1 ml of trace elements solution [29] ; [B] Na₂S, 9H₂O (2.5 g), Chlorhydrate of cysteine (1.25 g), NaHCO₃ (4.00 g). The heat-stable salts [A] solution was autoclaved and cooled under an atmosphere of N₂. Then, the medium was supplemented with components [B] solution, to final concentration of 2%. The pH was adjusted to 7.2 by addition of NaOH 1N solution. Tubes were incubated at 37˚C for two weeks. Formation of colloidal CuS from total sulfides (H₂S, S²⁻ and HS⁻) according to Cor-Ruchwich [30] was used for detection of positive tubes after incubation period. The most probable numbers of SRB were calculated from a table of MPN for three tubes.

From the soil sampled for bacterial enumeration and at the same periods during the rice cultural cycle, the sulfides were extracted using extraction medium, according to the method adapted from Chaudhry and Cornfield [31] . The extraction medium composition was as follows (per litter of distilled water): CH₃ (COO)₂ Zn, 2H₂O (50 g), CH₃COONa, 3H₂0 (12.5 g). Cor-Ruschwich [30] method was then used to measure the content of sulfides in the soil solution. The extracted sulfides solution (2 ml) was removed by syringe and rapidly injected

into tubes containing 2 ml of a copper reagent. Immediately after mixing (for 5s), the absorbance is measured at 480 nm using a Spectronic 61 photometer.

Data obtained were analyzed for SRB populations’ development and activity, drainage mode, rice growth stage and rice yield variations using the Student’s t-test and XLSTAT 7.5.2 statistical software. Mean parameters were compared according to the Newman Keuls’ test at 5% probability level.

In the present study, densities of SRB enumerated on soil before flooding, when the soil was dried and at transplanting day after soil flooding, are shown in Figure 4 and expressed as log 10 (most probable population number g⁻¹ dry soil).

Sulfides content in soil before flooding when the soil was dried and at transplanting day after one week of soil flooding, are shown in Figure 5 and expressed as log 10 (µg/g dry soil). The concentration of sulfides increased after one week of flooding in the soil for all the microplots.

Rice paddy fields represent a freshwater environment from which sulfate reducers have been isolated and in which sulfate reduction occurs [1] [6] [16] [19] [31] . Many studies showed that sulfate-reducers are common in flooded soils and they are also found near rice roots [8] [32] .

Thus ours results are in agreement with those obtained by Dianou et al. [33] , who reported a population increase of 200%, 139%, 119%, 66%, 40%, 59%, 42% and 22% in the soils of Burkina Faso, after flooding. Garcia et al. [33] reported that SRB populations quantified in the paddy soils of Senegal were influenced by the redox potential related to flooding [34] . Indeed, soon after the flooding of the rice fields, O₂ is removed from the bulk soil via respiratory processes, and large amounts of minerals and nutrients available for bacterial growth were released in the soil solution [16] [35] . In these anaerobiosis conditions the strictly anaerobic bacteria, including sulfate reducers, become active. Thus, significant reduction of sulfate occurs in a wide range of rice fields [16] .

The number of SRB in the soil near rice roots increased after one week of flooding for all microplots (Figure 4). An increase in SRB population number of 95.32%, 93.03%, 86.62% and 88.7% was recorded in the soil of the control (T), P1, P2 and P3 microplots, respectively. Our study also revealed that the number of SRB in soil near rice roots increased gradually with fluctuations from the transplanting day to rice flowering and maturity stages in all the paddy microplots (Figure 4). In most microplots, the highest densities of SRB were recorded from rice tillering and flowering to maturity stages, at which bacterial population number could reach 10⁷ to 10⁸/g dry soil (Figure 4).

The sulfides content in soil near rice roots also increased gradually with fluctuations during the rice cultural cycle and the highest levels were recorded from rice tillering and flowering to maturity stages at which it could reach 10² to 10³ µg/g dry soil (Figure 5).

Thus, SRB population density and sulfides concentration in soil near rice roots appeared particularly high at the tillering flowering and maturity stages of rice growth (Figure 4 & Figure 5).

Wind et al. [36] who reported that the number of Sulfate-Reducing Bacteria were higher in planted than in unplanted rice microcosms support our findings. Many studies reported also that sulfate reduction rates and sulfate concentration were high at the root surface and near the rice roots [36] [37] .

Our results are also in agreement with those of Ouattara [18] and Dianou and Traoré [19] who reported that after rice transplanting the numbers of SRB increased significantly in the plant rhizosphere. Dianou and Traoré [19] reported also that populations increased gradually with fluctuations from transplanting day to rice flowering and maturity stages in the soil near rice roots during paddy plots experiments. Dianou and Traoré [19] and Jacq et al. [38] showed that bacterial population number could reach values of 10⁷ to 10⁹/g dry soil in most of paddy plot soils of Senegal and Burkina Faso, from rice tilling and flowering to maturity stages, respectively. The rice heading and repining periods which correspond to the highest level of reduced soil conditions in paddy flooded soil, may be conducive to the growth of SRB as underlined Dianou and Traoré [19] . Thus, the high physiological activity of rice plant at these stages may result in the production of more substrates available for bacterial growth [19] [20] [38] .

The variance of the SRB density in soil near rice roots, in relation to drainage and sampling period during rice growth is presented in Table 1. The Newman Keuls’ test revealed that the density of SRB in the paddy soil near rice roots was not significantly related to drainage (P = 0.187) (Table 1). Moreover, no significant difference was observed between the average numbers of SRB population in the soil of all the microplots (drained or not) (Table 2). However, the number of bacteria in the paddy soil near rice roots was significantly related to sampling period (P < 0.0001) and to combined both factors (P = 0.000, Table 1). These contrasted results could be partially explained by the MPN method used. Indeed, the analysis of bacterial communities by culturing methods, although valuable to characterize metabolic activities of their members, allows the characterization of only 1% to 3% of the microscopically-detectable cells in soils [39] [40] . Thus, molecular methods could be an indispensable tool to provide a more comprehensive description of the SRB community evolution in relation to drainage [3] [5] [40] [41] .

The variance of the soil sulfides content in relation to drainage and sampling period during rice growth is presented in Table 2. It appeared that sulfides content in the soil near rice roots was significantly related to drainage (P = 0.032) and to the sampling period (P < 0.0001). SRB have been traditionally considered as strict anaerobes [16] [42] . However, in the present study, the SRB population number in the soil of P1, P2 and P3 drained microplots increased slightly as compared to the control (18.45%, 1.06%, 8.76%, respectively) indicating that some among the SRB can survive and grow in aerobic compartments in the presence of low pressure of oxygen where a surplus of oxygen is released by drainage and by healthy roots [38] [43] .

These results are in agreement with the ones of Johnson et al. [44] and Dolla et al. [45] who found that some species of SRB were capable of slow linear aerobic growth in sulfate-containing medium in the presence of very low concentrations of oxygen. Thus, the abundance and metabolic activity of SRB in oxic zones of numerous biotopes are frequently evaluated as higher than those in neighboring anoxic zones [46] [47] . The high number of SRB found in these oxic environments indicates that these organisms are able to deal with temporarily exposures to elevated oxygen concentration [16] [48] . Wind et al. [36] show also that sulfate reduction also takes place on the roots of intact rice plants when O₂ is allowed to diffuse to the roots through the aerenchyma system of the plants.

The density of SRB populations and the sulfides concentration in drained microplots could also be explained by heterogeneous distribution of oxygen through the drainage system leading to the formation of anoxic compartments [25] . A lack of oxygenation of such microsites in drained microplots may promote the SRB population survival [49] and the sulfides production [50] . Therefore, an efficient oxygenation of soil by water drainage would significantly reduce anaerobic SRB population’s development and activity.

DF = degree of freedom; F = Fisher F; ^*significant P < 0.05; ^**significant P < 0.01; ^ns not significant P < 0.05.

Means with a same letter within a column are not significantly different according to Newman Keuls’ test P > 0.05.

In our study, symptoms of sulfides toxicity for rice based on IRRI standard evaluation system [51] weren’t evidenced during the microplot experiments. Although no significant difference (P = 0.209) was found for rice yield among treatments, drained microplots, in particular P2 ones showed the highest yield. Dobermann and Fairhurst [21] reported that an excessive concentration of hydrogen sulfide in the soil results in reduced nutrient uptake due to a decrease of root respiration. The typical symptoms linked to sulfide toxicity involves: rotting of roots, bronzing of leaves, poor growth at the reproductive phase, reduced ability to oxidize iron in the rhizosphere and poor yield [52] - [60] . The symptoms of sulfide toxicity can occur throughout the growth cycle of the rice [21] ; however no critical levels have been established to test sulfide toxicity. Our result can be explained by the chemical fertilizers (NPK and urea) applied in all the microplots. Indeed, according to Dobermann and Fairhurst [21] and Fageria et al. [61] , sulfides toxicity is generally associated with iron toxicity, which is defined as a multiple nutritive disorder further through excesses of H₂S [61] . Deficiencies in minerals may enlarge the root permeability that enhances exudation, oxygen consumption and sulfate reduction, finally leading to severe sulfide toxicity [21] [61] - [63] . Many studies reported that sulfides toxicity for rice depends on nutrient supply; thus, amendment of paddy soil in mineral elements may reduce symptoms of sulfide toxicity for rice [64] [65] .

One raison of our results may be that the FKR 19 rice variety is not susceptible to sulfides toxicity as reported some studies [26] .

Rice plants can develop also physiological avoidance mechanisms to survive under toxic-sulfide condition [36] [49] . Indeed, sulfide toxicity depends on the strength of rice root oxidizing power, H₂S concentration in the soil solution and root health [21] [38] .

Soil oxygenation by water drainage can affect sulfides profiles [49] [66] . The release of oxygen from the roots of rice causes radial redox gradients around the roots, leading to various chemical and microbial oxidation processes in the rhizosphere along with the oxidation of sulfide to sulfate [67] .