Between 2012 and 2015, a survey was performed in the south of CAR in several localities bordering Bangui city and other locations further in the south at 75 km from the capital (Figure 1). Leaves samples were harvested from cultivated rice and wild grass including Oryza species in lowland, upland and irrigated conditions (Table 1) and processed as indicated by Konate et al. [8] . Wild grasses were identified using a document entitled “weeds of rice in West Africa” [9] . The visual diagnostic of RYMV are yellowing and mottling of leaves as well as severe stunting in some rice genotypes [10] . Thus, leaves were collected based on these typical symptoms (Figure 2) of RYMV using plastic bags. Samples were labeled with name location and ecological condition and they were stored in the ice box. Thereafter, samples were transferred to the laboratory and stored in the freezer at −20˚C for further processing.

Typical symptoms of rice yellow mottle disease could be wrongly attributed to iron or nitrogen deficiency. Therefore, the mechanical transmission test was done using the susceptible rice variety IR64 to confirm RYMV transmission. Leaf samples from fields were ground in the mortar at the ratio of 1:10 (w/v) with 0.01 M phosphate buffer and pH 7.0. The powder of carborundum (600 mesh) was added to the extracts, which were rubbed onto leaves of 2 week-old seedlings [11] [12] [13] . All experiment was conducted in a greenhouse under insect proof conditions.

The presence of RYMV in the collected samples was confirmed using DAS-ELISA (double antibody sandwich enzyme_linked immunosorbent assay) method [14] . A polyclonal antibody raised against isolates from Madagascar known to react strongly with isolates from west and Central Africa was used as coating antibody [6] . It was kindly offered by the Doctor Denis Fargette (IRD, Montpellier, France). All procedures of the test were as previously described [8] .

Leaf samples (1 g) were ground in 10 ml PBS buffer (Phosphate buffered saline pH

7.4) containing 0.05% of tween 20 (PBS-T) and 2% polyvinylpyrolidone (PVP) or extraction buffer. The resulting extracts were centrifuged for 10 min at 8000 rpm at 4˚C. The procedure was the following: immunoplates were coated with 5 µg/ml IgGs anti-RYMV in 5 mM carbonate buffer, pH 9.6 for 2 hours at 37˚C. After washing three times with PBS-T, the plates were incubated with 3% skimmed milk in PBS-T for 30 minutes. Thereafter, extracts were deposited to each well and incubated for 2 hours at 37˚C. IgGs anti-RYMV conjugated with alkaline phosphatase diluted in extraction buffer containing 1% skimmed milk were incubated for 2 hours. The substrate (p-nitrophenylphosphate) was diluted (1g/ml) in 10% diethanolamine, pH 9.8. Absorbance was recorded at 405 nm after 1 h substrate incubation using a multiscan

^aYM, yellowing mottle; GM, green mottle; M, mosaic; ^bNumber of positive samples out of total number.

microplate reader Metertech model. For the whole procedure, each step was followed from washing the plate three times with PBS-T excepted blocking step (skimmed milk incubation step).

For immunological typing, triple antibody sandwich-ELISA (TAS-ALISA) procedure was performed to determine the serological profiles of RYMV isolates which were detected positive with both biological and serological detection. So, 45 isolates chosen according to sites of provenance were used. Two monoclonal antibodies were used (Mab A and Mab D) to distinguish Ser1 from Ser2 [6] [7] . Thus, the procedure was the same as described above with two differences: after leave extracts incubation, monoclonal antibodies were incubated followed by the incubation of Goat anti mouse IgGs conjugated with phosphatase alkaline (GAM-PAL).

In order to test a possible relationship between cultural practices and RYMV occurrence in rice fields, the Chi-square independence test with Yates continuity’s correction was used using R statistical software https://www.r-project.org/.

For biological detection, typical symptoms of RYMV were induced on susceptible rice variety IR64 only by samples belonging from cultivated rice (Table 1). Up to 96% (201 out of 208) of leaves samples induced the characteristic yellow discoloration and mottling of leaves (as shown in Figure 2) after mechanical inoculation between 7 and 10 day after inoculation (DAI). By contrast, all samples collected from wild grasses did not infected IR64 for 30 DAI. In addition, the immunological detection using polyclonal antibodies showed positive reactions with all leave samples which were positive at biological test. The positive reaction including positive control ranged from 0.851 and 2.35. Absorbance between 0.028 and 0.2 were attributed to negative samples as they corresponded to the negative control. Negative samples were those from wild grasses. Failure to detect RYMV in these samples suggested that symptoms observed on collected wild grasses were not due to infections by RYMV.

Taking account of the ecology from which samples were collected and cultural practices, we found that infection rates depended on rice growing practices i.e. whether rice was directly seeded or seedlings were transplanted from pre-established seedbeds. Infection rates were significantly higher with transplantation (94.2%) than with direct seeding (22.8%) according to Chi-square test (c² = 108.83, df = 1, P < 0.001). This indicates that cultural practices, such as transplantation, likely play a role in spread of the disease in the field. However, as shown in Table 2, in both upland and lowland rice fields where direct seedlings were performed, few positive samples were detected.

The serological profile of RYMV isolates tested with monoclonal antibodies was homogenous. Indeed, all tested isolates were reacted strongly with Mab A and failed to react with Mab D. The virus titers estimated through absorbance values ranged from 1.063 to 2.628 with an average of 1.909 ± 0.360 (average ± standard deviation, n = 45) for Mab A and from 0.018 to 0.296 ± 0.069 with an average of 0.115 (average ± standard deviation, n = 45) for Mab D (Figure 3). The positive controls of both Mabs had similar absorbance with average of 1.774 ± 0.525 and 1.675 ± 0.346 for Mabs A and D respectively. This suggested that all isolates belonged to Ser1 serotype, thus RYMV is not diversified at serological level into this region.

^a + and ? for positive and negative reactivity respectively.