Soil sampling was performed on the rice-growing sites of Karfiguela, Banzon, and Vallée du Kou based on five plots per site. Five samples of three (03) kg of soil were taken in the first 10-20 cm along the two diagonals of each plot. The samples were then pooled to form a composite sample of 15 kg per plot. The samples are bagged and then labeled by assigning them a code, the name of the site, the date of collection, and the GPS coordinates of the plot. In the laboratory, we sifted the soil to collect the seeds of the weeds which were put in pots in the greenhouse for germination.To determine their host capacity, all weed species were inventoried after emergence and then inoculated.

In addition, during the period of a high prevalence of BLB and BLS in the field (September- October), we performed observations on weeds in the plots on the main irrigated sites: Vallée du Kou, Bazon, Karfiguela, Douna, Di, Niassan, Mogtédo, and Bagre. Symptomatic and asymptomatic samples were collected to determine the presence/absence of Xoo and Xoc strains.

Two improved rice varieties, TS2 and FKR62, from National Research Institute were used to test the pathogenicity of the strains on weeds. These varieties were previously identified by Wonni et al [8]. are susceptible to Xoo and Xoc strains representing Xo diversity in Burkina Faso. The Xoo strains BAI3 (Race A1) and Xoc Strain BAI10 [4, 14] were used to inoculate weed and rice varieties, respectively.

To obtain the weed seeds contained in each sample, we sieved them using two (02) sieves of different mesh (5 and 0.2 mm). The grains of sand passed through the sieves and the particles retained by the mesh, called refusals, usually contain weed seeds. The refusals obtained were dried in the shade and then weighed.

The seeds were grown in soil (2/3 soil and 1/3 sand) previously sterilized at 100 °C to avoid any development of unsampled weeds. The refusals of each sample were divided into four batches (or four replicates), and were mixed in the soil of each pot. Fifteen days after weed germination, to avoid competition between species for mineral elements, we applied 0.4 g of NPK in each pot in accordance with the recommended dose per hectare (200 kg/Ha) in rice fertilizing. Before the inoculation tests, we applied the same dose again, to eliminate the doubt that the symptoms would be due to a deficiency in mineral elements.

We carried out a visual observation and a count of each species. The inventory is updated every 15 days after sowing for three (03) months. We used the Adventrop guide (Weeds of African rice fields and Weeds of subtropics) to identify the species even in the juvenile state. abundance index [15] was used to assess the frequency of individuals of a species within the same family. The frequency of occurrence (number of pots where the species is present) and the maximum abundance per station (maximum score obtained by the species) were then recorded.

Weed diversity is assessed using three diversity indices namely: the Shannon-Weave Index (Ish) which is an indicator of diversity taking into account not only the species richness but also the proportion of each species represented within the community, it varies between 1 and 5,

Ish = -Σ (ni/N) log_2 (ni/N), where ni is the number of species i and N the total number of species;

Simpson's index (SI), expresses the probability that two (02) individuals chosen at random in an infinite population, belong to the same species, it varies between 0 and 1,

IS = Σ (ni/N)2, where ni is the number of species i and N the total number of species;

Piélou Equitability Index (Eq), measures the degree of diversity achieved by a stand in relation to its maximum value and allows comparison between two groups that do not have the same number of species. It is between 0 and 1,

Eq =ISh/(log_2 N), where N is the total number of species.

The asymptomatic leaves of the weeds after growing, were collected for Xo strain isolation on Peptone-Sucrose Agar medium. Then, Multiplex PCR specific to both pathogens, described by Lang et al. [16] was achieved from the leaf’s suspension.

Multiplex PCR combines four pairs of primers. It can be performed directly on bacterial colonies or from genomic DNA. The reaction volume of the PCR is 25µl composed of 5µl buffer 5X containing MgCl2, 0.5µl dNTPs, 2.5µl primers, 0.05 µl Taq polymerase, 5 µl of leaf’s suspension. Samples were initially denatured for 5 min at 95 °C, then 35 reaction cycles including (i) denaturation for 30s at 94 °C, (ii) hybridization for 30 s at 60 °C, (iii) elongation for 2mn at 68 °C, followed by a final elongation phase for 10 mn at 68 °C. The amplicons were separated by electrophoresis on 0.8%.

To determine the capacity of the weed species to harbor Xoo and Xoc strains, we inoculated the weed leaves 30 days after sowing with 108 bacteria/ml suspension. The leaf reactions after inoculation were assessed as follows: (i) presence of typical symptoms after inoculation; (ii) no typical symptoms, but the presence of necrosis at the site of inoculation; (iii) presence of yellowish or whitish lesions at the point of inoculation.

Two weeks after inoculation, the inoculated leaves are collected for the detection of Xoo and Xoc by isolation or multiplex PCR.

The data were entered into Excel 2010 software. To compare the level of weed and bacterial populations, histograms were constructed and variance analyzes were performed with SPSS20 software. The phytosociological and ordination analyses were performed with PC-ORD version 5.0 ordination software [17].

The inventories performed in the three prospected sites identified 24 species including the three (03) major families of weeds, namely, Poaceae, Cyperaceae, and dicotyledons. In fact, we have counted 12 species belonging to the broad leaf weed family with nine (09) species of the Poaceae family and three (03) species of the Cyperaceae family. The Poaceae were represented by Pennisetum pedicellatum Trin, Setaria pallide-fusca (Schumach.) Stapf & C.E. Hubbard, Paspalum scrobilatum L., Echinochloa colonum (L.) Link, Echinochloa stagnina (Retz.) P. Beauv., Rottboellia cochinchinensis (Lour.) W.D. Clayton, Oryza longistaminata A. Chev. & Roehr, Eleusine indica (L.) Gaertn., Eragrostis tenela (L.) Roem. & Schult., and Digitaria horizontalis Willd. As for Cyperaceae, we identified Cyperus esculentus L., Cyperus difformis L., and Fimbristylis littoralis Gand.

The broad-leaved species inventoried were Phyllanthus amarus Schum. & Thonn., Corchorus tridens, L., Portulaca oleracea L., Melochia corchorifolia, L., Physalis angulate L., Cleome viscosa L., Euphorbia hirta L., Stachytarpheta angustifolia (Mill.) Vahl, Triumfetta rhomboidea Jacq., Ammania prieureana Guill. & Perr., Ludwigia hyssopifolia (G. Don) Exell, Mollugo nudicaulus Lam., Sphenoclea zeylanica Gaertn.and Hyptis spicigera Lam..

Three major families of weeds, namely, Poaceae, Cyperaceae, and Dicotyledonous, were inventoried at the sites of Bama, Karfiguèla, and Banzon (Fig. 1). However, broadleaf (dicotyledonous) weeds were most abundant at all three sites with 51% at Banzon, followed by Cyperaceae (41.02%) at the Karfiguèla site and Poaceae Bama (33.51%).

Fig. (1). Weed populations inventoried in soil form Banzon, Bama, and Karfiguela.

Fig. (2). Dynamic of weeds populations inventoried in soil per site. GV4: Group of weeds where the effect of the treatments is the most significant; 1; 13; 14; 18: Sub-group of species according to the effect of the most significant treatments.

Concerning the dynamics of weeds, the ordination diagram (Fig. 2) shows that there is a slight link (13%) between the classes, the capacity to be a reservoir of Xoo and Xoc of weeds, and the distribution of collection sites. Indeed, regarding the capacity to be a Xo reservoir, it appears that dicotylodonous (Dicot) species do not host Xo strains at any site. However, Cyperaceae (Cyp), reservoirs of Xo are mainly present in Karfiguèla. As for the Poaceae (Gra) reservoirs of Xo, they are more present in Banzon and Karfiguèla.

The Cyperaceae genus are dominated by two species, Cyperus difformis L. with a sum of 99.4 individuals per sample, and Fimbristylis littoralis Gand. (Sum = 24.3) (Table 1). The two most important species of Poaceae are Echinochloa colona (L.) Link (sum=16.3) and Eleusine indica (L.) (sum= 4.9). Regarding the broad-leaved species, the most inventoried species are Stachytarpheta angustifolia (Mill.) Vahl (sum=28.2), and Ludwigia hyssopifolia (G. Don) Exell (sum=22.8).

The Cyperaceae genus obtained the highest specific diversity with F. littoralis (Shannon`s diversity index: H = 1.03) but also the lowest for C. difformis (H =0.41).

In order to determine the possible transmission of BLB and BLS by grasses and Cyperaceae seeds, we evaluated by PCR and biological tests, the presence/absence of Xoo and Xoc strains in the weed’s leaves. Of all Poaceae and Cyperaceae leaves collected, no weed species exhibited symptoms of BLB and BLS after seed emergence and plant development. PCR analysis of the leaf suspension collected from each species of Poaceae and Cyperaceae revealed the absence of Xoo and Xoc strains.

Surprisingly, foliar infections performed with foliar suspensions from Poaceae and Cyperaceae species collected in experimental conditions, on the two susceptible rice varieties, induced apparent BLS symptoms. These are samples of Echinochloa colona and Rottboellia cochinchinensis from Bazon; Eleusine indica from Bama and Digitaria horizontalis from Karfiguela. PCR analysis of infected rice leaf samples after inoculation of foliar suspensions of weeds revealed the presence of Xoc.

During our surveys, we collected weeds showing typical symptoms of BLS on Oryza longistaminata, Sacciolepis Africana, Paspalum vaginatum, Paspalum polystachyum, and Echinochloa colona. However, some weeds exhibited non-specific symptoms of BLB and BLS, which are leaf blight-like. Diagnosis by isolation and multiplex PCR tests has been made to identify mainly Xoc strains from these samples which are Paspalum vaginatum, Brachiaria lata, and Pennisetum pedicellatum sp.

Under the infiltration tests in the greenhouse, all species of Poaceae and Cyperaceae have the capacity to harbor Xo strains, without exhibiting typical symptoms of BLB or BLS. These are identified as hosts following reactions:

(i) typical symptoms of BLS (Fig. 3A-C) on Oryza longistaminata, Sacciolepis africana, Paspalum vaginatum, and Echinochloa colona;

Fig. (3). Foliar reaction of weed at 7 days after inoculation with Xoc BAI10 strain. (A): Echinochloa colona; (B): Paspalum vaginatum; (C): Paspalum vaginatum; (D): Pennisetum alopecuroides; (E): Brachiaria lata.

Fig. (4). Dynamic of Xanthomonas oryzae strains in weed leaves. (A): Xathomonas oryzae pv. oryzae; (B): Xathomonas oryzae pv. oryzicola.

(ii) No specific symptoms such as necrosis, whitish and yellowish lesion, or total absence of reaction at the inoculation point were induced by the two pathovars. However, the presence of Xoo and Xoc has been detected by multiplex PCR or after infiltration of left suspension from infected weeds with FKR 62N and TS2.

Furthermore, to assess the bacteria's ability to multiply in weeds and leaves, we quantified the bacterial populations at 7, 14, and 21 days after inoculation. These quantifications made it possible to confirm the host nature of some weed species.

The results showed that Xoo and Xoc strains survive in the asymptomatic leaves beyond one (01) month after inoculation. Indeed, high bacterial growth is observed during the first week of inoculation and decreases gradually from 14 days after inoculation (Fig. 3A and B). In general, bacterial growth is greater in grasses than in Cyperaceae. On the other hand, Xoo grows more abundantly in the leaves of O. sativa than those of weeds. Dicotyledons inoculated with both strains show necrosis and bleaching on the inoculation.

Three major families of weeds, namely, Poaceae, Cyperaceae, and Dicotyoledons, have been inventoried on the sites of Bama, Karfiguela, and Banzon. However, broadleaf (dicotyledon) weeds were the most abundant, followed by Cyperaceae (28%) and Poaceae (23%) at all three sites. These results agree with those [18] who showed that broadleaf weeds constitute the most important group of weeds in irrigated rice and lowland areas followed by Cyperaceae and Poaceae.

Similar results are obtained in irrigated rice with the same weed species by [19], where the species such as Cyperus difformis, L., Fimbristylis littoralis Gand., Echinochloa colona (L.) Link, Oryza longistiminata A. Chev. & Roehr. Ludwigia abyssinica A. Rich., Spilanthes uliginosa Sw. and are some major species found in the rice plains of Bama and Karfiguela. Fifty weed species was inventoried in the rice fields in Ivory Coast, that half of them belong to dicotyledons, the rest are distributed equally between Cyperaceae and Poaceae [20]. Studies achieved possible after it [20-22], in the valley of the Senegal river and in Wianga made it to inventory 90 species of weeds, distributed in 27 families where Poaceae and Cyperaceae are the most important with respectively 29% and 16%. Given the importance of Poaceae and Cyperaceae, which represent 50% of the weeds present in the rice fields surveyed in our study, they could constitute potential reservoirs for the two pathovars of X. oryzae. These results agree with studies by Sanou et al. [23]. which showed that farmers consider Poaceae and Cyperaceae to be the most harmful weeds in the irrigated rice system.

The ability of weeds to be a reservoir of X. oryzae pv. oryzae and X. oryzae pv. oryzicola would depend on the shape of the leaves, stems, and root system. Indeed, from the distribution of species according to these parameters, it appears that broad-leaved weeds (dicotylodones) are not the host of Xoo and Xoc. This classification of weeds by Defoer et al. [24], allowed rice farmers to identify Cyperaceae and Poaceae as the most damaging weed weeds [23].

Cyperaceae and Poaceae have been identified as the host group of Xoo and Xoc. In addition, they contained the greatest specific diversity, in particular with the species Fimbristylis littoralis. Their ability to host Xoo and Xoc would be explained by their morphology; but also their proliferation capacity.

We identified latent infections due to Xoc strains from the leaves of Oryza sativa, Echinochloa colona, Eleusine Indica, Digitaria horizontalis, and Rottboellia cochinchinensis following inoculation of leaf suspensions of these species on two susceptible rice varieties including FKR62N and TS2.

However, PCR analysis of the leaf’s suspension from asymptomatic weeds did not reveal the presence of Xoo and Xoc. This result could be explained by a low concentration of inoculum in the leaves, therefore, a small amount of amplifiable DNA. Indeed, rice seeds are the main source of Xoc inoculum [25-28], which can be transmitted to the plant. Xoc may also infect Leersia sp (p)., Leptochloa filiformis, wild and cultivated species of Oryza, Paspalum orbiculare, Zizania palustris, and Zoysia japonica [29]. Bracharia lata, Oryza longistaminata, Pennissetum sp, and Paspalum vaginatum have been reported as possible Xoc hosts [4, 15]. Consequently, the offending species weeds are Poaceae such as rice, which could support the hypothesis of Xoc transmission by the seeds of these weeds.

All species of Poaceae and Cyperaceae retain Xoc after inoculation except for Fimbristylis littoralis. Indeed, this species is characterized by hard and thin leaves, difficult to infiltrate. Our results confirm those [30], who indicated that Xoo and Xoc strains survive on wild or low-cultivated Poaceae. Analysis of the bacterial population dynamics in the leaves reveals Xoc strain's presence of 21 JAI with or without induced necrosis at the inoculation site [31]. Reported epiphytic survival of Xoo on wild grass leaves for 140 days. In our study, we noticed a larger Xoc population in the leaves of Poaceae than that of Cyperaceae. Indeed, Cyperaceae by their leaf structure can limit the penetration of bacteria and consequently their growth. They have a cylindrical shape at the level of their rachis, which gives them resistance to attacks from both external and internal factors.