The trial was conducted in an industrial banana plantation in Southeastern Côte d’Ivoire (5˚41'04.09'' N, 3˚3'53.95'' W, 100 m altitude). The climate, which is tropical and humid, is characterized by four seasons, two dry and two rainy [13]. The vegetation is diverse and includes coastal savannas, mangroves, swamp forests, riparian forests, and evergreen forests [14]. During the study (April 2019 to July 2020), the average temperature fluctuated around 27˚C and rainfall was recorded at approximately 2915.11 mm (Figure 1).

Plants of Arachis repens, and Desmodium adscendens (non-food legumes), as well as banana trees (Musa sp., AAA, Cavendish, Grande Naine cultivar) were used as plant material. The individuals of the first two species mentioned, aged 84 days, were grown from cuttings taken in a nursery using coconut fiber. The nurseries were set up in the greenhouse at the experimental station. For the third species, the plants were obtained after 77 days of acclimatization in the nursery from imported vitroplants.

Figure 1. Rainfall and temperatures of the study site during the test period.

For the trial, a banana plantation (3.6 ha), which had been left fallow for one year, was selected. At the time of plot preparation, the plantation contained banana trees aged 6 weeks. Within the plantation, three successive blocks, each with an area of 1080 m² (108 × 10 m), were selected. Fourteen days before the cover plants were installed, the weed flora was treated with glufosinate (SL 200 g/l) at a dose of 2 l/ha, using a backpack sprayer. Subsequently, weeding with a hoe was carried out to remove any weeds that were still alive. In addition, each block was subdivided into three elementary plots with an individual area of 360 m² (36 × 10 m) and identified with signs. The cover plants from the nursery were transplanted during the rainy season over the entire surface of the elementary plots. The plants, removed from the seedling trays, were transplanted at a spacing of 30 × 30 cm, a density of 111,111 individuals/ha. Before and after transplanting, the plot and the plants were watered.

Trial maintenance consisted of manually pulling up resistant weeds and removing them from the banana plantation. For the control plots, weeding was done by applying glufosinate every 4 weeks during the first two months of the trials, then glyphosate (SL 360 g/l; 3 l/ha) every 8 weeks until the end of the experiment according to the standard local practice in banana cultivation. Glufosinate, a contact herbicide, acts rapidly on young broadleaved weeds and has low phytotoxicity for banana young plants, whereas glyphosate, a systemic herbicide, provides broader-spectrum control later in the cycle when weed communities are more established and has low phytotoxicity on banana old plants. These practices were carried out when weeds began to dominate the cover crops or invade the control plots. Other maintenance activities were those commonly used in banana plantations. These included fertilization (NPK 12.5-04-28, urea 46, DAP 18-46, KCl 50, etc.), irrigation (2 hours per intervention every 2 days during the dry season), chemical and physical control of major pests and parasites (sigatoka, nematodes, black weevils), and care of banana trees and fruit.

The trial was arranged in a Fisher block design with three replicates of three treatments. The factor studied was the method of weed management, with the treatments A.repens, D.adscendens, and herbicides (control). Each elementary plot or experimental unit was equivalent to one replicate of a treatment (factor treatment). Each replicate consisted of 60 banana plants planted at a density of 1820 plants/ha (2 × 2 staggered rows, with a spacing of 2.2 m between banana plants in the row and 1.7 m between double rows). For the elementary plots treated with cover crops, transplanting was carried out at a density of 111,111 plants/ha (spacing of 30 cm × 30 cm). For each cover crop, the number of plants transplanted per experimental unit was 4000. Within the same block, the experimental units were spaced 3 m apart. Two successive blocks were separated by a drain 1 m wide and 1 m deep.

The weediness of the experimental plots was assessed during the two trial phases in the banana plantation, starting on the 28^th day after the cover plants were transplanted. Floristic surveys were conducted every two months using the systematic sampling method [15]. This method involves the use of observation squares measuring 1 m² each, randomly arranged at each assessment on the diagonals of the experimental units, at a rate of five per elementary plot [5]. The species observed were identified on site or collected and placed in a herbarium after being dried for three days at a temperature of 55˚C, then sent to the National Floristic Center (CNF) for identification (Figure 2). The data collection focused on qualitative variables, namely the specific diversity of weed species and their biological spectrum. The weeds inventoried were classified according to species, genera, families, classes, and biological types based on the model developed by Ake-Assi [16][17], adapted from that of Raunkiaer [18]. Verification and correction of species names were carried out using the CNF herbarium and the works of Lebrun and Stork [19]-[22] and Ake-Assi [16][17]. The abbreviations of the biological types used are those defined by Ake-Assi [16][17]. The percentage of a biological type was calculated from the ratio between the sum of the abundance-dominance indices of its weeds and that of all species, multiplied by 100, using the formula used by Tialou etal. [23]. The collected data were entered and processed using Excel 2021 spreadsheet software and its XLSTAT add-in, version 2016. Shapiro-Wilk’s test for normality and Levene’s test for homogeneity were performed before analysis of variance (ANOVA). These analyses aimed to compare the numbers of weed species, genera, families, and biological types among treatments. In case of rejection of the hypothesis of equality between means, Duncan’s test at the 5% significance level was performed to identify homogeneous groups (p < 0.05).

Figure 2. Weed species included in the herbarium.

A total of 315 floristic surveys were carried out across the entire experimental plot, at a rate of 105 per modality. Analysis of variance (ANOVA) showed statistically significant differences (p < 0.05) in the number of species and genera among treatments, but not for families and classes of weeds. The herbicide-treated plots and those associated with Desmodiumadscendens exhibited higher richness of genera and families than the Arachisrepens plots. For the entire study plot, three classes comprising 25 families, 50 genera, and 71 species were recorded (Table 1 and Table 2). Dicotyledons were most dominant, accounting for at least 55% of the total, regardless of treatment. Monocotyledons contributed between 41% and 34% to the diversity of the flora, while pteridophytes were represented by only two species, accounting for less than 5% (Table 3). The families most represented in terms of species are grouped in Table 4. Overall, these were Poaceae (11.27% to 18.31%), Cyperaceae (11.27% to 15.49%), Euphorbiaceae (4.22% to 9.85%), Asteraceae (5.63% to 8.45%), Passifloraceae and Rubiaceae (2.82% to 4.22%), Amaranthaceae and Fabaceae (1.41% to 4.22%). In the Poaceae, Passifloraceae, Fabaceae and Rubiaceae families, the differences observed between A.repens, D.adscendens, and the control are not statistically significant (p > 0.05). For Cyperaceae, the highest values were recorded under D.adscendens and in the herbicide-treated plots, unlike A.repens, where the proportion is significantly lower. In the case of the Asteraceae and Euphorbiaceae, the control plots show higher proportions. Among Amaranthaceae, the proportion is significantly lower in plots under A.repens (p < 0.05). Overall, A.repens significantly reduced the species richness of several major weed families (Cyperaceae, Amaranthaceae, Asteraceae and Euphorbiaceae), while D.adscendens is similar to the herbicide-treated plots for these families.

Table 1. Class, family, genus, and codes of all weed species inventoried during the study.

Table 2. Number of classes, families, genera, species, and proportions of weed flora according to banana plantation weed management treatment.

P: proportions; In the same column, values followed by the same letter are not significantly different at the 5% level according to the Duncan test.

Table 3. Number and proportion of weed species by class according to weed management treatments in banana plantations.

For a treatment within the same column, values followed by the same letter are not significantly different at the 5% level according to the Duncan test.

All biological types were present in the adventitious vegetation inventoried on the experimental site. These include megaphanerophytes (MP), mesophanerophytes (mP), microphanerophytes (mp), nanophanerophytes (np), geophytes (G), hydrophytes (Hyd), chamaephytes (Ch), hemicryptophytes (H), and therophytes (Th). However, mesophanerophytes were absent from plots treated with herbicides and megaphanerophytes from those associated with D. adscendens. Chamaephytes contributed the least, proportionally, to the weed flora inventoried in plots associated with cover crops. Hydrophytes were predominant in the A. repens plots and geophytes in the D. adscendens plots. In the control experimental units, hemicryptophytes were dominant (Figure 3).

Table4. Numbers and proportions of species from the weed families most commonly found in banana plantations, according to weed management treatments.

A.r: Banana plantations associated with A.repens; D.a: Banana plantations associated with D.adscendens; Hb: Control banana plantations treated with herbicides; E. P: Entire experimental plot; For a family within the same variable, values followed by the same letter are not significantly different at the 5% level according to the Duncan test.

A: entire experimental site; B: control; C and D: plots associated with A.repens and D.adscendens, respectively. MP = Megaphanerophyte (tree over 30 m tall), mP = Mesophanerophyte (tree between 8 m and 30 m tall), mp = Microphanerophyte (shrub between 2 m and 8 m tall), np = Nanophanerophyte (shrub between 25 cm and 2 m tall), Ch = Chamaephyte (species whose perennial buds are located less than 25 cm above the ground), H = Hemikryptophyte (species whose buds are located on the ground or very close above it), G = Geophyte (species whose buds are hidden in the ground), Th = Therophyte (annual species that spends the unfavorable season in the form of dormant embryos, protected inside seeds); Hyd = Hydrophyte (species whose buds are located in water).

Figure 3.Biological spectra of weeds identified in banana plots.