The study was conducted at Bama agricultural research station ( Figure 1 ) during two consecutive agricultural wet cropping seasons in 2020 and 2021. The climate in the study area is South Sudanian, with an alternating rainy season from May to October and a dry season from November to April. Rainfall frequently exceeds 1000 mm [16] .

Field sampling was carried out on damaged tillers of cultivated rice (Oryza sativa) varieties Orylux6 and FKR64. The Orylux6 variety has a 100-day cycle and a potential yield of 6.5 t/ha; FKR64 has a 120-day cycle; its potential yield can reach 8 to 10 t/ha [17] .

The animal material consisted of the pre-immature stages of arthropods associated with rice.

Field equipment consisted of:

Data collection techniques depended on taxonomic group and arthropod development stage. Data were collected throughout the rice development cycle. Data collection sheets were prepared to make observations, take samples, make dissections, incubations/rearing in the field and in the laboratory.

Sampling consisted in observing the damage caused by rice stem borers. Observations were made from the 28th day after transplanting (DAT) to the 84th DAT with a frequency of 14 days.

Dead hearts were sampled during the vegetative phase of the rice plant, while white heads were sampled during the reproductive phase. These samples were taken randomly from 20 hills per sub-plot.

Pre-immature populations of stem borers were recorded after laboratory dissection of rice tillers showing symptoms of the damage mentioned above. The number of dissections was the same as for visual observations in the field.

The rearing of pre-immature stages of stem borers was carried out in the entomology laboratory of the Farako-Ba station, Bobo-Dioulasso. The biological material included chrysalids and larvae of various stages recorded during dissections of damaged tillers. Incubation of these specimens was carried out in flasks containing water-soaked cotton; one individual was placed per flask. Rearing took place under ambient laboratory temperature conditions. Daily observations were made until the complete emergence of either the adults of the rice stem borers, or the adults of their associated parasitoids. The identification of emerged individuals enabled us to monitor the level of parasitism of pre-immature populations of the insect pests.

At maturity each rice variety was harvested. The rice was harvested by hand, using a sickle and covering an area of 3.5 m². After harvesting, the paddy was sun-dried, threshed, winnowed and weighed, and the moisture content measured and corrected to 14% in order to estimate grain yield.

The family identification key by [18] and the insect identification manual, general entomology by [19] were used to identify the various species of stem borers emerged from rice damaged tillers; in addition to these documents, the reference collection of parasitoids associated with rice stem borers compiled by [14] was used for parasitoid identification.

The collected data were entered and grouped using Microsoft Excel 2010. The database was then imported into R software version 3.6.0 for various statistical analyses. The data were first tested for normality using the Shapiro-Wilk test. To determine the individual effect of the explanatory variables (variety, previous crops and mineral fertilization) and their interactions on the response variables (arthropods, dead hearts and white heads), the data were then subjected to the Kruskall-Wallis (K-W) test at the 5% significance level. Finally, to determine the combined effects of the factors studied, the data were subjected to a multi-factor analysis of variance at the 5% level. Analyses focused on descriptive statistics (percentages, means and graphical treatments).

Table 1 presents the results of the analysis of variance (ANOVA) of the factors studied on the formation of dead hearts and white heads during the consecutive wet seasons 2020 and 2021. These results revealed a very highly significant difference between subplots for the factors number of days after transplanting (DAT) and fertilization for the year 2020 with regard to dead hearts and white heads. For the variety factor, a highly significant difference in white heads was observed for the 2020 cropping season. A very highly significant difference for the interaction fertilization * DAT was recorded in the 2021 wet season for dead hearts and finally a very highly significant difference for the interaction fertilization * previous crops * variety * DAT in the 2021 wet season for dead hearts. Significant differences were also recorded between the interactions, previous crops * variety; fertilization * previous crops * DAT for dead hearts in wet seasons 2020 and 2021 and fertilization * previous crops * variety only for dead hearts in 2021. Finally, significant differences were observed for the interaction fertilization * previous crops for dead hearts in wet seasons 2020 and 2021 and the triple interaction fertilization * cultural previous * variety for dead hearts in 2020.

***: very highly significative; **: highly significative; *: significative; ns: non significative.

Figure 1 illustrates the impact of mineral fertilization level on average dead heart rates observed on rice over two consecutive cropping seasons. The results of the ANOVA revealed a significant difference between the four fertilization levels at the 5% probability threshold in relation to the average rates of dead hearts recorded, even though no significant difference between the two crops seasons was observed. There was a significant difference between the F0 and F1 fertilization levels whatever the growing season.

The combined effect of mineral fertilization and number of days after transplanting on dead heart formation over the two growing seasons is illustrated in Figure 2 . Average dead heart rates first increased from 28 to 42 days after transplanting, before decreasing from 42 to 84 days after transplanting for the two growing seasons. The results of the ANOVA on average dead heart rates revealed significant differences between the four levels of fertilization for the two growing seasons. The highest average dead heart rates were recorded at 42 DAT in all subplots. Plots receiving F1 fertilization had the highest average dead heart rates (1.50%) over the two growing seasons. This was followed by F3 (1.15%), F2 (0.60%) and F0 (0.40%).

The results of the ANOVA on the combined effect of mineral fertilization and previous crops on dead heart formation over the two consecutive growing seasons are shown in Figure 3 . The average rate of dead hearts varied from one combination to another, with the highest rates recorded in plots with cowpea and Mucuna as previous crops, while the lowest average rates were recorded in plots with Control and Organic matter as previous crops. The highest average rate of dead hearts was recorded in the cowpea * F3 plots (0.63% and 0.63%), followed by the Mucuna * F1 plots (0.59% and 0.57%) over the two cropping seasons. Damage was low in the Control * F2 (0.13% and 0.14%) and Mucuna * F0 (0.13% and 0.14%) plots.

Figure 4 shows the combined effect of previous crops and variety on dead heart formation. The ANOVA revealed a significant difference in the average rate of dead hearts for the previous crops * variety interactions. There were no significant differences in average dead heart rates for Mucuna, Cowpea, Control and the two varieties, except with organic matter, where average rates were 0.46% and 0.45% (variety FKR 64) and 0.12% and 0.11% (variety Orylux6) over the two consecutive cropping seasons.

The results of the combined effect of fertilization * previous crops * days after transplanting on dead heart formation are presented in Table 2 . ANOVA revealed a significant difference between treatments at the 5% threshold over the two growing seasons. Damage recorded was low at 28 DAT; some damage was observed with the F1 combination: NPK (200 kg/ha) + USG (72 kg/ha); it then increased at 42 DAT (0.86% over the two cropping seasons) and 56th DAT (0.82% over the two cropping seasons), then became rare at 70th and 84th DAT. No damage was recorded on the 28th DAT with the Mucuna * F0 * F1 * combination; the highest average rate of dead hearts (2.46%) was recorded on the 42nd DAT with the Mucuna * F1 combination, followed by the Cowpea * F3 combination (2.35%) over the two growing seasons. On the other hand, the lowest average rate of dead hearts was observed at 28 DAT (0.06% over the two growing seasons) with the Mucuna * F3 combination at 70 DAT (0.06% over the 2020 wet season and 0.07% over the 2021 wet season).

F0: Control without fertilization; F1: NPK (200 kg/ha) + USG (72 kg/ha); F2: NPK (100 kg/ha) + Simple urea (100 kg/ha); F3: NPK (200 kg/ha) + Simple urea (150 kg/ha). Numbers followed by the same letter are not significantly different from each other at the 5% probability level.

Figure 5 shows the results of the ANOVA on the combined effect of fertilization * previous crops * variety on dead heart formation over the two consecutive cropping seasons. A significant difference between treatments at the 5% threshold was found over the two growing seasons. The highest average rates of dead hearts (0.94% in 2020 and 0.87% in 2021) were recorded with the Orylux6 * organic matter * F1 combination, followed by the FKR64 * Mucuna * F3 combination with an average rate of 0.72% in 2020 and 0.74% in 2021. The lowest average dead heart rates were recorded with the Orylux6 * Cowpea * F1 and FKR64 * Control * F2 combinations, each recording 0.04% in 2020, and the Orylux6 * Organic manure * F0 0.04% in 2021.

1) Effect of mineral fertilization on white head formation

Figure 6 illustrates the effect of fertilization levels on the average white head rate over two consecutive cropping seasons. ANOVA revealed a significant difference between treatments at 5% level. The lowest average white head rates (0.26% in 2020 and 0.31% in 2021) were recorded in plots that received no fertilization (F0); plots that received F1, F2 and F3 fertilization recorded the highest average white head rates of 0.45%, 0.50% and 0.48% respectively for the 2020 wet season and 0.56%, 0.59% and 0.46% respectively for the 2021 wet season.

2) Effect of rice variety on white head production

Figure 7 shows the effect of rice variety on white head formation. The results of the ANOVA revealed a significant difference between varieties at 5% threshold. Average white head rates were higher (0.76% in 2020) and 0.71% in 2021 with the Orylux6 variety. The FKR64 variety recorded 0.10% in 2020 and 0.11% in 2021.

3) Combined effect of variety and number of days after transplanting on white head formation for both seasons

Figure 8 illustrates the effect of varieties on the evolution of the average rate of white heads. The results of the ANOVA showed a significant difference between varieties over the two growing seasons. The average white head rate increased from 42 to 70 DAT for the Orylux6 variety, before decreasing after 70 days. This curve followed the same trend for the two consecutive growing seasons. On the other hand, for the FKR64 variety, the first white heads were observed at 42 DAT; they increased slightly between 42 and 70 DAT, before rising again slightly at the end of the plant cycle (between 70 and 84 DAT). The highest average rate of white heads was recorded at 70 DAT for the variety Orylux6 (1.80% in 2020 and 1.75% in 2021) and at 84 DAT for the variety FKR64 (1.16% in 2020 and 1.13% in 2021).

1) Evolution of pre-immature populations of rice stem borers as a function of variety

Figure 9 illustrates the evolution of the number of individuals in the pre-immature populations of rice stem borers as a function of the varieties used. In the 2020 wet season, the FKR64 variety recorded 67.38 individuals of Chilo spp. and 130 in 2021, compared with 32.6 individuals in 2020 and 82 in 2021 for Orylux6. As for M. separatella, 96.15 individuals in 2020 and 04 in 2021 were hosted by Orylux6, compared with 3.85 individuals in 2020 and 34 in 2021 for FKR 64. Finally, 100 individuals of S. calamistis in 2020 and 07 in 2021 were found on Orylux6. During the 2021 wet cropping season, the Chilo spp. population was more abundant during the vegetative phase than during the reproductive one.

2) Relative importance of pre-immature populations of rice stem borers as a function of rice plant development phases

Figure 10 illustrates the results on the evolution of pre-immature populations of rice stem borers insects as a function of rice plant development phases.

Chilo spp. was the species recorded throughout the rice development cycle. An average rate of 24% of Chilo spp. populations was recorded during the vegetative phase, 59.29% during the reproductive phase and 17.25% during the maturation phase of the rice plant in 2020; 54% pre-immature population were recorded during the vegetative phase of the plant, 31% pre-immature population at rice reproduction and 15% pre-immature population at plant maturation in 2021. A percentage of 94% of M. separatella was observed during the vegetative phase and 5.55% during the reproductive phase in 2020. In the 2021 growing season, 92% of the pre-immature population of M. separatella was recorded during the vegetative phase, compared with 8% during the maturation phase of the rice plant. Sesamia calamistis showed the lowest average pre-immature population rate. This species was not recorded during the rice plant maturation phase in 2020 (50% of its population was recorded during the vegetative phase and 50% during the reproductive phase). In 2021, 14% of the pre-immature population was recorded during the vegetative phase and 86% during the reproductive phase of the plant. No individuals of this species were recorded during the rice maturation phase in any of the two consecutive growing seasons.

3) Evolution of parasitism affecting pre-immature populations of rice stem borers

Only pre-immature populations of Chilo spp. were parasitized ( Table 3 ). Larval parasitism of Chilo spp. was observed during the vegetative (14.28%) and reproductive (5.08%) phases of rice on the Orylux6 variety and during the vegetative phase on the FKR 64 variety (16.66%). The average rate of parasitism was higher during the vegetative phase for both varieties.

The results in Table 4 show the different parasitoids that emerged from Chilo spp. larvae. All parasitoids belong to the order Hymenoptera. The following families were represented: Ichneumonidae, Braconidae, Chalcidae and Bethylidae.

The results of the ANOVA for paddy rice yield are shown in Table 5 . Fertilization and variety had a very highly significant effect on rice yield in both cropping seasons, 2020 and 2021. The study of the fertilization * variety interaction revealed a significant effect on yield during the same growing seasons.

***: very highly significant; *: significant; ns: not significant. From Table 5 , the following variables were examined: the effect of fertilization on yield, the average yield per variety and the combined effect of fertilization * variety on yield for the year 2020.

1) Effect of mineral fertilization on rice paddy yield

Figure 11 illustrates the effect of mineral fertilization levels on yield. The ANOVA revealed a highly significant difference between treatments at the 5% threshold over the growing seasons. The highest average yields (4321 kg/ha in 2020 and 6912 kg/ha) in 2021 were recorded with F1, followed by F3 (4,182 kg/ha in 2020 and 5669 kg/ha) in 2021and F2 (3613 kg/ha in 2020 and 5,481 kg/ha in 2021). The lowest average yield was recorded for F0 (2683 kg/ha in 2020 and 4058 kg/ha in 2021).

2) Effect of varieties on paddy rice yields

Figure 12 shows the average paddy yield by variety. ANOVA revealed a highly significant difference between treatments at 5% threshold over the two consecutive growing seasons. The highest average yield was recorded with the FKR64 variety, i.e. an average of 4043.98 kg/ha in 2020 and 6049.51 kg/ha in 2021, compared with an average of 3356.01 kg/ha in 2020 and 5009.56 kg/ha in 2021 for the Orylux6 variety.

3) Combined effect of mineral fertilization and varieties on rice paddy yield

Rice yield was influenced by varieties and mineral fertilization levels. Figure 13 illustrates the effect of variety and fertilization level on average rice yield. The highest average yield was recorded with F1 fertilization for both FKR 64 (5018 kg/ha) and Orylux6 (3624 kg/ha) in wet season 2020 and (7275 kg/ha) for FKR 64 and 6549.83 kg/ha for Orylux6 in 2021 ; followed in decreasing order of average yield F3 (4796 kg/ha) with FKR 64 and 3568 kg/ha with Orylux6 in 2020, in the wet season 2021, 6464.86 kg/ha with FKR 64 and 4866.12 kg/ha with Orylux6; then F2 (3755 kg/ha) with FKR 64 and 3470 kg/ha with Orylux6 in 2020 and in the 2021 season with F2 (6051.52 kg/ha) for FKR 64 and 4911.09 kg/ha for Orylux6. The lowest average yields were recorded with F0, i.e. 2605 kg/ha with FKR 64 and 2761 kg/ha with Orylux6 in 2020 and in 2021 F0 (4006.66 kg/ha) with FKR 64 and (3711.19 kg/ha) with Orylux6.

Damage associated with Lepidoptera rice stem borers was relatively low (with levels often below 2%) over the two consecutive years covered by this study. Infestations increased at the beginning of the rice plant cycle. The highest levels of damage were recorded at 42 DAT for dead hearts during the vegetative phase, and at 70 DAT for white heads during the reproductive phase of the rice plant. The presence of M. separatella during the vegetative phase of the rice plant seems to be associated with the presence of young plants. Our results are in line with those of [20] , who showed that M. separatella females only lay eggs on plants at the tillering stage of the rice plant. Average dead heart rates were higher in plots that received higher NPK and legume as previous crops, and lower in control plots. In addition, white head rates were higher in plots that received mineral fertilizers and lower in control plots. Chilo species and S. calamistis species were more present during the vegetative and reproductive phases of the rice plant, while M. separatella species was more recorded during the reproductive phase of the rice plant. Rice Lepidoptera stem borers had a preference for plots that received mineral fertilization. Our results are similar to those reported by [21] , who showed that high doses of phosphorus favored the multiplication of Lepidoptera populations. A study of the combined effect of fertilization * previous crops * variety revealed that the highest rate of damage was recorded in the Orylux6 * Organic matter * F1 plots in 2020 and Orylux6 * Mucuna * F1 in 2021, and the lowest rate was recorded in the FKR64 * Control * F2 plots in 2020 and 2021 with the Orylux6 * Organic matter * F0. These results confirm that insect pests have a preference for nutrient-rich plots.

Our results revealed significant variations in abundance levels between species during the development cycle of the rice plant. Chilo species were more representative in terms of numbers of individuals during the vegetative and reproductive phases of the two rice varieties. The permanent presence of Chilo spp. may be associated with the fact that the species has two successive generations during the cropping season. Authors of [22] and [23] have shown that the first generation of Chilo spp. infests the tillering stage of the rice plant, while the second generation, with a higher number of individuals, attacks the heading and flowering stages. These results, reflecting the relative importance of this genus at Bama, are in line with those of [6] , who studied the seasonal abundance of Lepidoptera stem borers in western Burkina Faso, where they reported that Chilo spp. was the most abundant dry-season borer at Bama. Our findings also corroborate those of [1] [14] [15] and [24] , who reached the same conclusions. Maliarpha separatella species was present during the vegetative phase of the FKR64 variety; these results are similar to those of [25] . Although it has been established that M. separatella does not cause dead hearts, its presence in the stems reduces the plant’s vigor, disrupts its feeding and makes it vulnerable to fungal diseases. The presence of M. separatella during the vegetative phase of the plant could be due to the fact that this species attacks young organs, which could justify its abundance during this phase. Our results are in line with those of [18] , who showed that the female of M. separatella only lays eggs on plants in the tillering stage (15 days after transplanting at the earliest), and that this species becomes rare from heading onwards. The presence of S. calamistis was rare in the experiment. This result is similar to that reported by [14] , who noted a very low presence of this species in rice plots in Bama.

With regard to parasitoids, the four families listed all belong to the Hymenoptera order, all of which are associated with stem borers, particularly the Chilo genus. These findings could be explained by the fact that Chilo spp. is the most abundant and frequent stem borer in Bama. Our results are lower in terms of parasitoid species abundance than those reported by [14] , who listed 15 parasitoid species, and higher than those reported by [16] , who listed six parasitoid species, all of them belonging to the Hymenoptera order and associated with Chilo spp. This discrepancy could be associated, on the one hand, to the fact that parasitism is as dynamic as the evolution of host insect populations and, on the other, to differences in entomological sampling methods. In fact, our observations were limited to hills showing damage symptoms (“dead hearts” or “white heads”), while pevious authors sampled hole hills for dissection in the laboratory. The relative high rate of parasitism during the two wet seasons could be associated with the high rainfall and humidity that favoured egg hatching and the development of stem borers. Authors of [26] have shown that rainfall and altitude influence the absence or presence of parasitoids, and therefore parasitism rate.

Grain yield was influenced by variety, fertilization and fertilization * variety interaction. The results show that the lowest yields were recorded in the control plots, with an average of 2683 Kg/ha in 2020 and 4058.92 Kg/ha in 2021, while the highest yields were observed in the plots that benefited from mineral fertilization, with an average yield of 4321 Kg/ha in 2020 and 6912.41 Kg/ha. Increasing doses of NPK had an effect on yield. This could be explained by the multiple roles played by this nutrient, notably in stimulating root development, tillering, flowering and grain ripening. Our results corroborate those reported by [27] , who concluded that the use of high-performance varieties and appropriate doses of mineral nutrients in irrigated lowland rice maximized yield.