Effect of chemical and alternative weed management on weed composition in banana greenhouses with full and deficit irrigation levels

Abstract

Background:   In banana greenhouses in Türkiye, extensive herbicide use and high irrigation demand create environmental and management challenges. However, the effects of herbicide application frequency and irrigation on weed control strategies and their effectiveness remain unclear.

Objective:   The research, conducted in Mersin in 2022–2024, evaluated the effects of alternative weed control methods (mowing and geotextile mulching) and sequential herbicide treatment programs (oxyfluorfen, pendimethalin, indaziflam, glyphosate, diquat) under full (100%) and deficit (50%) irrigation.

Methods:   The experiment was established using split-plot design, with irrigation levels assigned to the main plots and randomized subplots arranged in three replications. Control for coverage (% population thresholds), weed density (plant m^-2), and end-of-experiment dry biomass (g m^-2) were measured.

Results:   Weed community changes were evaluated using Shannon-Wiener and Simpson indices, and treatment–species relationships were analyzed by canonical discriminant analysis (CDA). Herbicide treatments (especially indaziflam+diquat; 1.00 plant m^-2; 7.92 g m^-2) reduced weed populations compared to the weedy control. Mulching (0.00 plant m^-2; 0.00 g m^-2) was the most effective. Mowing demonstrated partial effectiveness, achieving only limited weed suppression (41.09 plant m^-2; 125.00 g m^-2). Irrigation levels did not significantly affect weed density or biomass, whereas control methods significantly altered weed composition, species abundance, and diversity. Setaria viridis decreased depending on treatments, while Cardamine occulta, Oxalis spp., Urtica urens, and Amaranthus viridis increased. CDA revealed that weed treatments significantly influenced species composition.

Conclusions:   Integrating alternative and chemical control programs, and minimizing chemical use, are vital for sustainable banana production and for maintaining ecological balance.

Keywords:
Community; Modified Control Programs; Mulching; Suitable Herbicides; Sustainability

1. Introduction

Bananas (Musa spp.) are a major agricultural commodity grown primarily in the Americas, the Caribbean, and selected Asian countries. They play a critical role in global food security (Daniells, 2009). Bananas are widely consumed in raw, dried, and juice forms, and they number among the most consumed agricultural products worldwide after rice, wheat, and maize (Singh et al., 2016). Banana cultivation in Türkiye, concentrated along the Mediterranean coast, has expanded significantly to 921,925 tonnes produced on 13,506 hectares in 2023; Türkiye is the leader in productivity, at 70 tonnes ha^-1 (Food and Agriculture Organization, 2022; Türkiye İstatistik Kurumu, 2022). Bananas are also an important source of income worldwide (Arumugam et al., 2023; Castillo et al., 2023).

In Musa spp. cropping systems, inadequate weed control can lead to significant yield losses. When not managed effectively, weed competition can cause reductions in banana yields of approximately 35–47% (Prameela et al., 2009; Fongod et al., 2010). In addition to direct competition for resources, weeds pose further challenges— particularly during the early growth stages—by harboring pests and diseases (Fongod et al., 2010). Therefore, effective weed management supports banana growth and yields by minimizing weed competition and associated risks (Costa, 2007; Prameela et al., 2009; Lanza et al., 2017).

Chemical control remains the preferred method of weed management, due to its cost-effectiveness and ease of use. In Türkiye, the use of herbicides in banana production has increased with the expansion of the banana cultivation area. Recommendations for banana fields include contact and systemic herbicides in the growing season (Achard, 1993). Although the practice is widespread, there is a growing need to minimize herbicide use and to use alternative methods in order to reduce environmental impact.

Alternative control strategies are essential to mitigating damage to the ecosystem and protecting nontarget organisms from exposure to chemicals. Less environmentally harmful methods—such as (biodegradable) mulching, mechanical control, irrigation practices, and mixed cropping—also contribute to weed control and help maintain weed biodiversity while protecting the agricultural ecosystem (Bauri et al., 2010; Concenço et al., 2014). The goal of weed control is not to eliminate all weed species but to keep them below economic thresholds. Understanding weed distribution and the relevant banana-specific conditions is essential for preventing competitive pressure and for optimal crop growth (Singh et al., 2006). Crop management practices have a significant impact on banana–weed interactions, and banana growers often use more ecological practices than herbicides (Dassou et al., 2023). Research on alternative pest control methods has increased due to the reduction of chemicals under the Green Deal (Hall et al., 2020).

At the same time, Türkiye faces a potential water scarcity, and efficient irrigation practices are needed to maintain agricultural productivity and quality (Çeliktopuz, 2019). Accordingly, a focus on deficit irrigation studies is crucial for optimal crop water use. Bananas, with their shallow roots and permanent canopies, requires continuous irrigation for optimal growth (Robinson, 1996). Understanding the effects of water stress and developing crop management strategies for bananas are critical for sustainable production-especially in the Mediterranean region, where growers rely on empirical knowledge for irrigation and protection decisions.

The primary objective of this study was to evaluate the effectiveness of different weed management strategies for greenhouse production of bananas under Mediterranean climatic conditions. The effectiveness of these strategies was assessed based on weed density, biomass, and species composition over time. The secondary objective was to determine the influence of two irrigation regimes (Full-100% and Deficit-50%) on weed dynamics under these management strategies. The study aimed to identify practical and sustainable weed control practices that reduce dependence on herbicides while supporting efficient banana cultivation.

2. Materials and Methods

This study was carried out over a two-year period between 2022 and 2024 in a banana greenhouse located at Alata Horticultural Research Institute (36.626239°N, 34.339746°E) in Mersin Province, a Mediterranean region of Türkiye. The experimental material consisted of banana plants (Musa acuminata cv. Alata-Azman). The 5-m-tall greenhouse was designed to maintain optimal microclimatic parameters for banana growth: the average temperature was kept around 15°C and relative humidity at approximately 75%, providing favorable conditions for banana development and minimizing frost risk during winter months. When greenhouse temperatures exceeded 35°C, ventilation and shading systems were used to reduce heat stress and to prevent humidity-related diseases. Frost protection was ensured through misting systems.

The experiment was conducted using a split-plot design, with irrigation levels as the main plots and weed control treatments as the subplots. Two irrigation treatments were applied: full irrigation (100%) and deficit irrigation (50%). Seven weed control treatments—including four herbicide programs, two alternative control methods, and one untreated control—were randomly assigned to subplots, with each replicated three times (Table 2). Banana plants were spaced 1.5 m apart within rows and 2.3 m apart between rows. Each subplot contained three plants covering approximately 6 m² (4.5×1.5 m), with 0.8-m buffer zones between subplots to minimize interference; the total experimental area amounted to 475 m². The first experimental year (2022-2023) examined initial banana plants, established in 2022. In the second experimental year (2023-2024), newly emerged suckers (ratoon plants) from the same mother plants were used as main pseudostems. To avoid carryover effects, weed control treatments were randomly reassigned to the new plants in different subplot locations each year, which allowed the experiment to be repeated under comparable environmental conditions while evaluating different plants.

Irrigation was provided through drip and mini-sprinkler systems, to maintain uniform soil moisture levels. From April 14, 2022, to March 12, 2024, total irrigation amounts were 3301 mm for full and 1,727 mm for deficit, with 178 irrigation events. Class A pan evaporimeters, calibrated by the US Weather Service, measured daily evaporation (Epan) to assess water loss and crop demand. Crop coefficients (Kc) were set at 1.0 for full and 0.5 for deficit. Irrigation water applied (IR) was adjusted using the following formula, where Sp is distance between plants (m), Sl is distance between rows (m), P is plant cover, and Ea is irrigation efficiency (%) (Equation 1).

(1), $\text{IR}=\frac{\text{Sp}\times \text{Sl}\times \text{Epan}\times \text{kc}\times \text{P}}{\text{Ea}}$

Weed assessments were performed by counting and recording plants within a defined quadrat frame, with wooden stakes used to delineate the experimental plots. Geotextile mulch was applied in relevant treatments, and mechanical weed control was performed using a motorized trimmer. Herbicides were applied using a rechargeable backpack sprayer equipped with double TeeJet nozzles operating at a pressure of 304 kPa. Applications were conducted when weeds were at the 2-4 leaf growth stage.

The herbicides contained both pre- and post-emergence active ingredients. Oxyfluorfen, a pre-emergence herbicide, causes lipid peroxidation in cell membranes, leading to rapid necrosis of photosynthetic tissues. Pendimethalin, also applied pre-emergence, inhibits microtubule formation and prevents cell division, thereby suppressing root and shoot development during weed seedling emergence. Indaziflam is a soil-applied herbicide that disrupts cellulose biosynthesis, impairing cell wall formation and resulting in weed mortality at early growth stages. As a post-emergence herbicide, glyphosate potassium salt inhibits the enzyme 5-enolpyruvylshikimate-3-phosphate synthase (EPSPS), blocking the synthesis of aromatic amino acids and causing systemic death of the entire weed. Diquat dibromide interferes with electron transport in photosystem I, leading to the formation of reactive oxygen species and rapid cell death in green plant tissues (Table 1).

Table 1
Experimental characters in the banana greenhouse

Trade names	Active ingredient (g L-1)	Doses (g ha-1)	HRAC/WSSA Grouping
Goal 4F	480 Oxyfluorfen	192 (pre)	Protoporphyrinogen oxidase (14)
Stomp Aqua	455 Pendimethalin	1365 (pre)	Microtubule assembly (3)
Allion	500 Indaziflam	50 (pre)	Cellulose synthesis (29)
Round Up Star	441 Glyphosate potassium salt	1323 (post)	Enolpyruvylshikimate phosphate synthase (9)
Bustop 200 SL	200 Diquat dibromide	1600 (post)	PS I electron diversion (22)
Alternative
Geotextile mulch	Black perforated polyethylene, mulch weight 40 g m-2, thickness 0.33 mm.
Mowing	Repeated mowing with string trimmer.

The pre-emergence herbicides (oxyfluorfen, pendimethalin, and indaziflam) were applied in early spring (April), 1 day after planting (DAP), to suppress early-season weed emergence. Post-emergence herbicide applications were performed based on weed coverage thresholds. Glyphosate was first applied at 42 DAP and reapplied at 110 DAP when weed coverage approached 10%, in accordance with established guidelines (TAGEM, 2022). Mechanical weed control through mowing was carried out three times during the growing season at 56, 110, and 215 DAP, corresponding to early-, mid-, and late-season weed growth stages. Geotextile mulch was applied once, while untreated subplots were maintained as weedy controls for comparison. Herbicide application programs were scheduled according to the weed population dynamics and coverage thresholds observed in the experimental plots (Table 2).

Table 2
Banana greenhouse treatment programs from start to pre-harvest in one year in 2022–2024

Day After Planting (DAP)	1st	42th	56th	110th	215th	Total treatments
Treatments	1st	1st	1st	2nd	3rd
Herbicide treatments
-Oxyfluorfen+Glyphosate+Diquat	Oxyfluorfen			Glyphosate	Diquat	3 applications
-Pendimethalin+Glyphosate+Diquat	Pendimethalin			Glyphosate	Diquat	3 applications
-Indaziflam+Diquat	Indaziflam				Diquat	2 applications
-Glyphosate+Glyphosate+Diquat		Glyphosate		Glyphosate	Diquat	3 applications
Alternative treatments
-Mowing (3 times)			Mowing	Mowing	Mowing	3 applications
-Geotextile mulching	Mulching					1 application
-Weedy control						No application

• ^*
  Mulching was applied once. After indaziflam treatment, a second application of diquat was performed at 215 DAP for coverage thresholds. Observations for weed density were performed on the 75th, 145th, and 245th DAP and at the pre-harvest stage.

Weed observations were performed at 75, 110, 145, 215, and 245 DAP as well as at the banana pre-harvest stage, resulting in a total of six observation dates. Weed observations were conducted in each subplot using a fixed 1-m² quadrat to determine weed coverage (%), density (plants m^-2), and dry biomass (g m^-2). After the pre-harvest assessment, all weed biomass within the quadrat area was cut at the soil surface. Collected samples were oven-dried at 105°C for 24 h to determine dry weight. Weed density for DAP and dry biomass were analyzed using one-way ANOVA after verification of normality and homogeneity of variances. Data from the two years (which involved different plant materials) were pooled, and subplots were randomized within each year. Significant differences among treatments (irrigation, treatments, and irrigation × treatments) were further examined using Tukey’s HSD at 95% confidence level (p<0.05).

Weed species composition and diversity under full and deficit irrigation were evaluated using the Shannon-Wiener (H’), Simpson diversity (1-D), and Simpson dominance (Sd) indices (Gorelick, 2006). In the indices, p_i represents the proportion of the ith species, ln is the natural logarithm, n_i is the number of individuals of a given species, and N is the total number of individuals. The Simpson index ranges from 0 to 1 (Magurran, 2004) (Equations 2,3,4). Species dynamics and indices were determined based on species densities recorded in subplots at different observation dates (DAP) in relation to the applied treatments.

Shannon-Wiener diversity

(2), ${\text{H}}^{\prime }=-{\displaystyle \sum {\text{p}}_{\text{i}}\ln \left({\text{p}}_{\text{i}}\right)}$

Simpson diversity

(3), $1-\text{D}=1-\frac{{\displaystyle \sum {\text{n}}_{\text{i}}\left({\text{n}}_{\text{i}}-1\right)}}{\text{N}\left(\text{N}-1\right)}$

Simpson dominance

(4), $\text{Sd}=\frac{{\displaystyle \sum {\text{n}}_{\text{i}}\left({\text{n}}_{\text{i}}-1\right)}}{\text{N}\left(\text{N}-1\right)}$

Canonical discriminant analysis (CDA) was conducted using JMP PRO 16 to examine relationships between irrigation, weed control treatments, and species composition at 110 DAP, 215 DAP, and pre-harvest. For CDA, species-level density data for each treatment were used after averaging across the two years. Individual species within each treatment were evaluated separately, allowing for the inclusion of zero values, and rare species were retained to reflect the complete community structure.

CDA was used to assess how sequential herbicide, mulching, and mowing treatments influenced weed species under different irrigation regimes. This multivariate approach clarified the degree of separation among treatments, revealing the impacts of irrigation level and management strategy on weed community composition and on suppression efficiency.

3. Results

3.1 Impact of weed control on weed populations

Over the two-year experiment period (2022–2024) in the banana greenhouse, 27 different weed species were identified using the Flora of Turkey and East Aegean Islands book series (Davis, 1965–1988). These species belong to 16 weed families, with the highest number of species found in the families Euphorbiaceae (5 species), Poaceae (4 species), and Urticaceae (3 species). Among the identified species, Setaria spp. had the highest density during the experimental period, with 5.33 plant m^-2 (41.13%), followed by Oxalis spp. with 2.43 plant m^-2 (18.75%), Amaranthus spp. with 2.26 plant m^-2 (17.40%), Cardamine occulta Hornem. with 0.77 plant m^-2 (5.94%), and Mercurialis annua L. and Portulaca oleracea L. with 0.72 plant m^-2 (5.52%) each (Table 3).

Weed observations were performed at 75, 110, 145, 215, and 245 DAP as well as at the banana pre-harvest stage, resulting in a total of six observation dates. Weed observations were conducted in each subplot using a fixed 1-m² quadrat to determine weed coverage (%), density (plants m^-2), and dry biomass (g m^-2). After the pre-harvest assessment, all weed biomass within the quadrat area was cut at the soil surface. Collected samples were oven-dried at 105°C for 24 h to determine dry weight. Weed density for DAP and dry biomass were analyzed using one-way ANOVA after verification of normality and homogeneity of variances. Data from the two years (which involved different plant materials) were pooled, and subplots were randomized within each year. Significant differences among treatments (irrigation, treatments, and irrigation × treatments) were further examined using Tukey’s HSD at 95% confidence level (p<0.05).

Weed species composition and diversity under full and deficit irrigation were evaluated using the Shannon-Wiener (H’), Simpson diversity (1-D), and Simpson dominance (Sd) indices (Gorelick, 2006). In the indices, p_i represents the proportion of the ith species, ln is the natural logarithm, n_i is the number of individuals of a given species, and N is the total number of individuals. The Simpson index ranges from 0 to 1 (Magurran, 2004) (Equations 2,3,4). Species dynamics and indices were determined based on species densities recorded in subplots at different observation dates (DAP) in relation to the applied treatments.

Table 3
Weed species and average density in the experimental treatments (2022–2024)

Weeds	Family	EPPO codes	2022-2024
			Average density(plant m-2)	Proportion(%)
Setaria viridis (L.)P.Beauv.	Poaceae	SETVI	5.33	41.13
Oxalis spp.	Oxalidaceae	OXASS	2.43	18.75
Amaranthus viridis L.	Amaranthaceae	AMAVI	2.26	17.40
Cardamine occulta Hornem.	Brassicaceae	CAROC	0.77	5.94
Mercurialis annua L.	Euphorbiaceae	MERAN	0.72	5.52
Portulaca oleracea L.	Portulacaceae	POROL	0.72	5.52
Cyperus rotundus L.	Cyperaceae	CYPRO	0.15	1.12
Heliotropium europaeum L.	Boraginaceae	HEOEU	0.10	0.73
Euphorbia prostrata Aiton	Euphorbiaceae	EPHPT	0.09	0.66
Urtica urens L.	Urticaceae	URTUR	0.07	0.54
Stellaria media (L.)Vill.	Caryophyllaceae	STEME	0.07	0.50
Sorghum halepense (L.)Pers.	Poaceae	SORHA	0.06	0.46
Euphorbia nutans Lag.	Euphorbiaceae	EPHNU	0.04	0.31
Eleusine indica (L.)Gaertn.	Poaceae	ELEIN	0.04	0.27
Convolvulus arvensis L.	Convolvulaceae	CONAR	0.03	0.23
Malva spp.	Malvaceae	MALSS	0.03	0.19
Euphorbia helioscopia L.	Euphorbiaceae	EPHHE	0.02	0.15
Lamium amplexicaule L.	Lamiaceae	LAMAM	0.02	0.15
Veronica arvensis L.	Plantaginaceae	VERAR	0.02	0.12
Capsella bursa-pastoris (L.)Medik.	Brassicaceae	EPHHE	0.01	0.08
Chrozophora tinctoria (L.)A.Juss.	Euphorbiaceae	CRZTI	0.01	0.08
Pilea microphylla (L.)Liebm.	Urticaceae	PILMI	0.01	0.08
Sonchus oleraceus L.	Asteraceae	SONOL	0.01	0.08
Digitaria sanguinalis (L.)Scop.	Poaceae	DIGSA	<0.01	<0.01
Parietaria judaica L.	Urticaceae	PAIDI	<0.01	<0.01
Physalis spp.	Solanaceae	PHYSS	<0.01	<0.01
Solanum nigrum L.	Solanaceae	SOLNI	<0.01	<0.01

As shown in Figure 1, the weedy control plots consistently exhibited the highest weed coverage compared to all other treatments. For all treatments except indaziflam, when weed coverage reached the 10% threshold at 110 DAP, a second application was implemented for all treatments; in the indaziflam plots, weed cover did not reach the 10% threshold, and so no additional application was required at that time. After the second assessment at 145 DAP, control efficacy was maintained in most treatments. By 215 DAP, weed cover had once again surpassed the 10% threshold, prompting a third application. With regard to the alternative controls, mulching did not allow any weed growth, but mowing did not appear to be sufficient, particularly for the control of short-stem and broad-leaved weed species. A noticeable decline in weed density and biomass was observed following these applications. Nevertheless, given that weed cover remained below the 10% threshold for longer periods with the mulching and sequential indaziflam+diquat treatments, these proved to be the most successful, with the lowest number of applications in both irrigation regimes (Figure 1).

Figure 1
Average weed coverage (%) for application thresholds by banana planting dates and weed treatments (2022–2024)

The analysis showed that there was no significant relationship between the two-year average density and dry biomass in the irrigation factor and the irrigation × treatment factor; only the treatment factor was significant for both weed observation criteria (p<0.001).

Parallel increases in weed density within plots were observed, with weed cover reaching or exceeding the 10% threshold. The greatest densities occurred in the weedy control and the mowing treatment, with peaks recorded during the summer period of the experimental season. In weedy control plots, the weed population declined after the end of the summer, followed by the emergence of winter weed species. In the mowing plots, an increase in weed density was observed before application, and then a decrease following treatment.

The most effective treatments for reducing weed density were mulching and indaziflam+diquat. A pronounced reduction in density was evident after the second application (145 DAP). Mulching proved effective, with no significant differences observed compared to herbicide treatments in the same group. Nevertheless, before the third application (215 DAP), weed density increased again, indicating a temporary loss of control efficacy. Following the third application (245 DAP), interventions effectively reduced weed density in all plots. The suppressive effects of the sequential application of herbicides and mulching persisted until the pre-harvest observation period. Although indaziflam+diquat and mulching resulted in lower weed densities with fewer applications, these densities were not statistically different from the other herbicide treatments. In contrast, mowing was ineffective in reducing weed density throughout the experimental period. Finally, no significant interactions were detected between irrigation regimes and weed control treatments (Table 4).

Table 4
Average weed density (plant m^-2) before and after application observations by planting date (2022-2024)

Number of applications	After-1st (July)	Before-2nd (August)	After-2nd (September)	Before-3rd (November)	After-3rd (December)	Pre-harvest
Observation Days(DAP)	75th	110th	145th	215th	245th
Variation Source
Irrigation (F)	0.363NS	0.588NS	0.002NS	2.381NS	0.909NS	0.280NS
Irrigation (p)	0.552	0.450	0.969	0.134	0.348	0.601
Treatments (F)	18.879*	20.096*	39.031*	10.178*	33.272*	52.271*
Treatments (p)	0.000	0.000	0.000	0.000	0.000	0.000
IrrigationxTreatments (F)	0.278NS	0.319NS	0.480NS	1.682NS	0.234NS	1.012NS
IrrigationxTreatments (p)	0.943	0.922	0.817	0.162	0.962	0.438
Irrigation
Full-(100%)	21.23±6.40	2718±8.18	16.39±3.87	33.53±4.94	12.93±4.66	16.17±3.99
Deficit-(50%)	18.56±5.41	23.02±6.67	16.30±3.94	26.48±5.24	10.53±3.84	15.17±3.89
Treatments
Oxyfluorfen+Glyphosate+Diquat	7.50±1,15b	9.47±1,33c	7.71±1.77b	37.67±8.16a	0.34±0.25b	8.00±0.76b
Pendimethalin+Glyphosate+Diquat	7.67±2.69b	9.69±3,62c	7.96±2.19b	29.17±5.96ab	1.42±0.72b	8.92±1.24b
Indaziflam+Diquat	3.30±1.13b	4.05±1.22c	5.46±0.80b	8.50±2.20bc	0.26±0.18b	1.00±0.50b
Glyphosate+Glyphosate+Diquat	9.50±2.14b	12.34±2.41c	9.63±2.09b	36.09±10.06ab	0.84±0.65b	8.59±0.57b
Mowing(3-times)	44.50±3.44a	54.69±4.59b	39.38±1.90a	52.92±5.36a	35.09±2.36a	41.09±3.68a
Mulching	0.00±0.00b	0.00±0.00c	0.00±0.00b	0.00±0.00c	0.00±0.00b	0.00±0.00b
Weedy-Control	66.80±13.42a	85.47±16.51a	44.25±5.77a	45.67±767a	44.17±7.81a	42.09±5.18a
IrrigationxTreatments
Full-(100%)
Oxyfluorfen+Glyphosate+Diquat	7.17±1.98	9.34±2.62	7.09±2.09	54.00±3.41	0.50±0.50	9.00±0.29
Pendimethalin+Glyphosate+Diquat	11.50±4,12	14.88±5.79	10.42±3.50	38.34±9.34	2.51±1.16	10.84±1.17
Indaziflam+Diquat	1.75±0.39	2.26±0.58	5.34±0.73	9.34±2.81	0.51±0.29	1.51±0.87
Glyphosate+Glyphosate+Diquat	7.75±2.26	11.09±3.46	7.76±3.01	29.84±2.93	1.67±1.17	8.50±1.05
Mowing(3-times)	47.09±2.61	57.92±4.20	42.17±2.18	50.00±9.02	36.50±4.50	37.00±5.30
Mulching	0.00±0.00	0.00±0.00	0.00±0.00	0.00±0.00	0.00±0.00	0.00±0.00
Weedy-Control	73.34±22.06	94.80±28.29	41.92±8.00	53.17±9.71	48.84±14.75	46.34±10.14
Deficit-(50%)
Oxyfluorfen+Glyphosate+Diquat	7.84±1.60	9.59±1.39	8.34±3.30	21.34±7.38	0.17±0.17	7.01±1.33
Pendimethalin+Glyphosate+Diquat	3.84±2.13	4.50±2.19	5.50±2.39	20.00±2.47	0.34±0.17	7.01±1.61
Indaziflam+Diquat	4.84±1.95	5.84±1.97	5.59±1.63	7.67±3.95	0.00±0.00	0.51±0.50
Glyphosate+Glyphosate+Diquat	11.25±3.87	13.59±3.94	11.50±3.04	42.34±21.41	0.00±0.00	8.67±0.73
Mowing(3-times)	41.92±6.75	51.46±8.78	36.59±2.34	55.84±7.34	33.67±2.32	45.17±4.77
Mulching	0.00±0.00	0.00±0.00	0.00±0.00	0.00±0.00	0.00±0.00	0.00±0.00
Weedy-Control	60.25±19.26	76.13±21.81	46.59±9.84	38.17±11.99	39.50±8.10	37.84±3.61

• ^*
  =Statistically significant (Tukey, p<0.001). NS=Not significant.

However, dry biomass average was identified as a significant factor related to weed control treatments (p<0.001): apart from sequential mowing and weedy control plots, it was observed that all sequential herbicide applications and mulching resulted in similar groupings for dry weed biomass. The lowest biomass was observed in indaziflam+diquat plots, followed by oxyfluorfen+glyphosate+diquat and        glyphosate+glyphosate+diquat,        and        then pendimethalin+glyphosate+diquat; the highest was recorded in weedy control plots. A statistically significant difference was observed among the treatments: when only the treatment effect on biomass averages is considered, a pattern similar to that observed for weed density was evident, in that sequential mowing was similarly insufficient to the weedy control, whereas mulching proved to be the most effective in reducing dry biomass (Table 5).

Table 5
Effect of weed control methods and irrigation on dry biomass (g m^-2) at pre-harvest observation (2022–2024)

Dry Biomass(g m-2)	Average
Variation Source
Irrigation (F)	1.641NS
Irrigation (p)	0.211
Treatments (F)	13.564*
Treatments (p)	0.000
IrrigationxTreatments (F)	0.237NS
IrrigationxTreatments (p)	0.961
Irrigation
Full-(100%)	49.28±12.66
Deficit-(50%)	36.43±10.91
Treatments
Oxyfluorfen+Glyphosate+Diquat	18.34±1.91b
Pendimethalin+Glyphosate+Diquat	30.00±5.33b
Indaziflam+Diquat	7.92±3.62b
Glyphosate+Glyphosate+Diquat	19.17±2.87b
Mowing(3-times)	99.59±22.39a
Mulching	0.00±0.00b
Weedy-Control	125.00±23.27a
IrrigationxTreatments
Full-(100%)
Oxyfluorfen+Glyphosate+Diquat	18.34±3.01
Pendimethalin+Glyphosate+Diquat	35.00±7.64
Indaziflam+Diquat	11.67±5.84
Glyphosate+Glyphosate+Diquat	25.00±2.50
Mowing(3-times)	113.34±7.93
Mulching	0.00±0.00
Weedy-Control	141.67±23.47
Deficit-(50%)
Oxyfluorfen+Glyphosate+Diquat	18.34±3.01
Pendimethalin+Glyphosate+Diquat	25.00±7.64
Indaziflam+Diquat	4.17±4.17
Glyphosate+Glyphosate+Diquat	13.34±0.84
Mowing(3-times)	85.84±29.63
Mulching	0.00±0.00
Weedy-Control	108.34±43.34

• ^*
  =Statistically significant (Tukey, p<0.001). NS=Not significant.

3.2 Management effect on weed diversity

Experimental plots under full and deficit were compared across the different weed control treatments. Weed community changes were assessed using the ShannonWiener, Simpson diversity, and Simpson dominance indices for the experimental years (Tables 6,7,8). Te final pre-harvest observations in the mulching treatment were excluded from the analysis, due to the absence of weeds.

Weed observations were performed at 75, 110, 145, 215, and 245 DAP as well as at the banana pre-harvest stage, resulting in a total of six observation dates. Weed observations were conducted in each subplot using a fixed 1-m² quadrat to determine weed coverage (%), density (plants m^-2), and dry biomass (g m^-2). After the pre-harvest assessment, all weed biomass within the quadrat area was cut at the soil surface. Collected samples were oven-dried at 105°C for 24 h to determine dry weight. Weed density for DAP and dry biomass were analyzed using one-way ANOVA after verification of normality and homogeneity of variances. Data from the two years (which involved different plant materials) were pooled, and subplots were randomized within each year. Significant differences among treatments (irrigation, treatments, and irrigation × treatments) were further examined using Tukey’s HSD at 95% confidence level (p<0.05).

Weed species composition and diversity under full and deficit irrigation were evaluated using the Shannon-Wiener (H’), Simpson diversity (1-D), and Simpson dominance (Sd) indices (Gorelick, 2006). In the indices, p_i represents the proportion of the ith species, ln is the natural logarithm, n_i is the number of individuals of a given species, and N is the total number of individuals. The Simpson index ranges from 0 to 1 (Magurran, 2004) (Equations 2,3,4). Species dynamics and indices were determined based on species densities recorded in subplots at different observation dates (DAP) in relation to the applied treatments.

Table 6
Shannon-Wiener diversity, Simpson diversity, and Simpson dominance indices of weed populations at different irrigation levels of weed management (2022–2024)

Number of Treatments*	Treatments	2022-2023						2023-2024
		Full Irrigation(100%)			Deficit Irrigation(50%)			Full Irrigation(100%)			Deficit Irrigation(50%)
		H’	1-D	Sd	H’	1-D	Sd	H’	1-D	Sd	H’	1-D	Sd
Oxyfluorfen+Glyphosate+Diquat
1st(110-DAP)	Oxyfluorfen	1.55	0.82	0.1 8	1.12	0.70	0.30	1.20	0.66	0.34	1.22	0.69	0.31
2nd(215-DAP)	Oxyfluorfen+Glyphosate	0.53	0.23	0.77	0.77	0.42	0.58	1.49	0.76	0.24	1.22	0.69	0.31
3rd(Pre-harvest)	Oxyfluorfen+Glyphosate+Diquat	1.76	0.80	0.20	1.62	0.79	0.21	0.56	0.50	0.50	0.00	0.00	1.00
Pendimethalin+Glyphosate+Diquat
1st(110-DAP)	Pendimethalin	1.50	0.71	0.29	1.49	0.82	0.18	1.34	0.68	0.32	0.93	0.57	0.43
2nd(215-DAP)	Pendimethalin+Glyphosate	0.64	0.37	0.63	1.23	0.65	0.35	1.14	0.60	0.40	1.59	0.76	0.24
3rd(Pre-harvest)	Pendimethalin+Glyphosate+ Diquat	1.86	0.85	0.15	1.81	0.82	0.18	1.24	0.64	0.36	0.00	0.00	1.00
Indaziflam+Diquat
1st(110-DAP)	Indaziflam	1.12	0.63	0.37	1.13	0.54	0.46	1.00	0.67	0.33	1.09	0.61	0.39
-	-	1.44	0.72	0.28	0.99	0.62	0.38	1.24	0.74	0.26	1.44	0.74	0.26
2nd(Pre-harvest)	Indaziflam+Diquat	0.60	0.48	0.52	1.10	1.00	0.00	0.00	0.00	1.00	-	-	-
Glyphosate+Glyphosate+Diquat
1st(110-DAP)	Glyphosate	1.47	0.69	0.31	1.56	0.76	0.24	1.50	0.82	0.18	0.80	0.38	0.62
2nd(215-DAP)	Glyphosate+Glyphosate	0.55	0.24	0.76	0.51	0.24	0.76	1.56	0.78	0.22	1.42	0.70	0.30
3rd(Pre-harvest)	Glyphosate+Glyphosate+Diquat	1.96	0.86	0.1 4	1.80	0.83	0.17	1.23	0.66	0.34	-	-	-
Mowing+Mowing+Mowing
1st(110-DAP)	Mowing	1.71	0.78	0.22	1.08	0.68	0.32	0.95	0.50	0.50	1.11	0.56	0.44
2nd(215-DAP)	Mowing+Mowing	1.13	0.63	0.37	0.95	0.65	0.35	1.29	0.66	0.34	1.40	0.69	0.31
3rd(Pre-harvest)	Mowing+Mowing+Mowing	1.21	0.56	0.44	1.05	0.50	0.50	1.38	0.65	0.35	1.55	0.66	0.34
Weedy-Control
110-DAP	-	1.69	0.77	0.23	0.86	0.84	0.16	1.23	0.64	0.36	1.06	0.50	0.50
215-DAP	-	0.94	0.54	0.46	1.29	0.67	0.33	1.33	0.68	0.32	1.32	0.62	0.38
Pre-harvest	-	1.05	0.50	0.50	1.35	0.61	0.39	1.56	0.78	0.22	1.86	0.80	0.20

• ^*
  For weed control, the first-line treatments refer to the evaluation at 110 DAP of the 1^st application, the second-line treatments refer to the evaluation at 215 DAP of the 2^nd application, and the last-line treatments refer to the evaluation before harvest of the 3^rd application. (H’):Shannon-Wiener, (1-D):Simpson diversity, (Sd):Simpson dominance

Table 7
Simpson dominance indices of weed species numbers with weed control methods with full and deficit irrigation in 2022–2023

Treatments 2022-2023 Sd	Oxyfluorfen+ Glyphosate+Diquat			Pendimethalin+ Glyphosate+Diquat			Indaziflam+Diquat			Glyphosate+ Glyphosate+Diquat			Mowing(3-times)			Weedy-Control
	1st(110-DAP)*	2nd(215-DAP)	3rd(Pre-harvest)	1st(110-DAP)	2nd(215-DAP)	3rd(Pre-harvest)	1st(110-DAP)	-	3rd(Pre-harvest)	1st(110-DAP)	2nd(215-DAP)	3rd(Pre-harvest)	1st(110-DAP)	2nd(215-DAP)	3rd(Pre-harvest)	1st(110-DAP)	2nd(215-DAP)	3rd(Pre-harvest)
Weed species	Full Irrigation(100%)
AMAVI	0.00	0.00		0.00	0.04	0.00	0.00	0.20		0.01	0.00	0.00	0.00	0.01	0.01	0.01	0.00	0.00
CAROC	0.02	0.00	0.14	0.01		0.07	0.02	0.05		0.00	0.00	0.04	0.01		0.00	0.03	0.00	0.00
CONAR
CYPRO				0.08			0.33	0.01	0.48
MERAN	0.02	0.00	0.00	0.00	0.00	0.02				0.00	0.00	0.00	0.00	0.00	0.00	0.01	0.00	0.03
OXASS	0.03	0.00	0.01	0.00		0.04		0.01		0.01			0.05	0.13	0.41	0.02	0.12	0.47
POROL	0.00	0.00		0.01			0.02			0.04	0.00		0.05	0.00		0.00	0.00
SETVI	0.10	0.77	0.03	0.20	0.59	0.02		0.00		0.25	0.76	0.07	0.10	0.23	0.00	0.16	0.34	0.00
SORHA
STEME	0.00		0.00					0.00	0.05			0.01			0.01
URTUR			0.01			0.00						0.02						0.00
VERAR			0.00			0.00						0.01
Weed species	Deficit Irrigation(50%)
AMAVI	0.01	0.03		0.01	0.04	0.00	0.01	0.10			0.00	0.02	0.01	0.01	0.02	0.03	0.02	0.01
CAROC			0.00	0.01	0.04	0.00				0.06	0.01		0.00	0.00		0.00	0.00
CONAR	0.01			0.01									0.00	0.00
CYPRO				0.13		0.00										0.00
MERAN	0.03	0.00	0.07		0.00	0.06	0.00			0.01		0.00		0.00	0.00	0.01	0.00	0.00
OXASS		0.00					0.00			0.01			0.02	0.08	0.47	0.03	0.21	0.36
POROL					0.00		0.00	0.02		0.15	0.00		0.03	0.00		0.02
SETVI	0.25	0.54	0.10	0.01	0.27	0.01		0.26		0.01	0.76	0.04	0.26	0.25	0.00	0.06	0.10	0.01
SORHA							0.44
STEME			0.01			0.08			0.00			0.02			0.00
URTUR			0.00			0.03			0.00			0.08			0.00			0.01
VERAR			0.02									0.01						0.00

• For weed control, the first-line treatments refer to the evaluation at 110 DAP of the 1^st application, the second-line treatments refer to the evaluation at 215 DAP of the 2^nd application, and the last-line treatments refer to the pre-harvest evaluation of the 3^rd application.
• A. viridis: AMAVI; C. arvensis: CONAR; C. occulta: CAROC; C. rotundus: CYPRO; M. annua: MERAN; Oxalis spp.: OXASS; P. oleracea: POROL; S. halepense: SORHA; S. media: STEME; S. viridis: SETVI; U. urens: URTUR; V. arvensis: VERAR.

Table 8
Simpson dominance indices of weed species numbers with weed control methods with full and deficit irrigation in 2023–2024

Treatments 2023-2024 Sd	Oxyfluorfen+ Glyphosate+Diquat			Pendimethalin+ Glyphosate+Diquat			Indaziflam+Diquat			Glyphosate+ Glyphosate+Diquat			Mowing(3-times)			Weedy-Control
	1st(110-DAP)*	2nd(215-DAP)	3rd(Pre-harvest)	1st(110-DAP)	2nd(215-DAP)	3rd(Pre-harvest)	1st(110-DAP)	-	3rd(Pre-harvest)	1st(110-DAP)	2nd(215-DAP)	3rd(Pre-harvest)	1st(110-DAP)	2nd(215-DAP)	3rd(Pre-harvest)	1st(110-DAP)	2nd(215-DAP)	3rd(Pre-harvest)
Weed species	Full Irrigation(100%)
AMAVI	0.08	0.00		0.00	0.00		0.28			0.01	0.00		0.01	0.00		0.06	0.04	0.04
CAROC		0.03		0.24	0.28	0.32		0.1 6	1.00		0.11	0.29	0.00	0.00	0.01	0.00	0.02	0.04
CYPRO				0.00						0.08	0.05	0.01
EPHHE						0.01
LAMAM
MERAN		0.03			0.00	0.00		0.03			0.05			0.01	0.01		0.00	0.00
OXASS	0.02	0.09		0.04	0.00			0.00		0.01	0.02	0.00	0.02	0.22	0.31	0.03	0.01	0.05
PILMI							0.03
POROL
SETVI	0.24	0.09		0.03	0.12		0.03	0.07		0.08	0.00		0.46	0.10	0.01	0.26	0.24	0.09
STEME			0.50			0.00						0.04		0.00	0.00
URTUR	0.00	0.00	0.00			0.03						0.00			0.00			0.00
Weed species	Deficit Irrigation(50%)
AMAVI	0.07			0.01			0.34	0.21		0.02	0.00		0.06	0.02	0.00	0.02	0.01	0.00
CAROC		0.00			0.00			0.00		0.00				0.00	0.00	0.00	0.00	0.01
CYPRO
EPHHE
LAMAM								0.02			0.00				0.00	0.00
MERAN		0.17			0.12			0.00			0.04			0.00	0.01		0.00	0.02
OXASS	0.19	0.08			0.04		0.03	0.00			0.21		0.00	0.09	0.32	0.01	0.33	0.14
PILMI
POROL				0.06						0.00						0.00
SETVI	0.05	0.05		0.35	0.07		0.03	0.04		0.59	0.05		0.38	0.20	0.01	0.46	0.04	0.02
STEME
URTUR			1.00		0.00	1.00									0.00			0.00

• ^*
  For weed control, the first-line treatments refer to the evaluation at 110 DAP of the 1^st application, the second-line treatments refer to the evaluation at 215 DAP of the 2^nd application, and the last-line treatments refer to the pre-harvest evaluation of the 3^rd application. A. viridis: AMAVI; C. occulta: CAROC; C. rotundus: CYPRO; E. helioscopia: EPHHE; L. amplexicaule: LAMAM; M. annua: MERAN; Oxalis spp.: OXASS; P. microphylla: PILMI; P. oleracea: POROL; S. viridis: SETVI; S. media: STEME; U. urens: URTUR.

In the first year (2022–2023), the highest weed diversity was observed particularly in herbicide-treated plots and in plots with both full and deficit irrigation. In the second year (2023–2024), species diversity generally decreased in plots receiving herbicide treatments, while it slightly increased in the mowing and weedy control plots. The Simpson diversity index showed a similar trend, with relatively high values in herbicide-treated plots during the first year and a decline in the second year, while mowing maintained more consistent diversity across the years. Thus, herbicide applications tended to reduce weed diversity over time (Table 6). The Simpson dominance index indicated that while the herbicide-treated plots had high species richness in the first year, there was a decline in species diversity in treatments in the following year; in indaziflam+diquat and glyphosate+glyphosate+diquat plots, there were species that disappeared completely in the second year. In contrast, the Simpson dominance index remained consistently high in the sequential mowing and weedy control treatments in both years, indicating the dominance of a few weed species rather than high overall diversity.

Changes in diversity and dominance were observed following the first and second treatment applications; however, the most relevant criterion for evaluation was final weed control, at the end of the growing season (Table 6). After the first and second weed control applications under both irrigation regimes, the dominant weed species varied according to the Simpson dominance index. For example, after the third application under full irrigation, Cardamine occulta (CAROC) became dominant in oxyfluorfen+glyphosate+diquat and pendimethalin+glyphosate+diquat plots, while Cyperus rotundus (CYPRO) dominated indaziflam+diquat plots after diquat. Moreover, Oxalis spp. (OXASS) emerged as the dominant species after the third application in mowing and in weedy control plots. Notably, there was a shift in the dominant species under deficit irrigation in various weed treatment plots, with Mercurialis annua (MERAN) and Setaria viridis (SETVI) dominating in oxyfluorfen+glyphosate+diquat plots and MERAN and Stellaria media (STEME) dominating in pendimethalin+glyphosate+diquat plots (Table 7). However, in the second year, all species were eliminated from the indaziflam+diquat and glyphosate+glyphosate+diquat plots. In sequential herbicide treatment plots, changes in dominant species were also observed, with Amaranthus viridis (AMAVI) emerging as the dominant species. In summary, while certain species were dominant in the plots in the first year, different species came to dominate in the second year (Table 8).

In the following year, after the first and second weed control applications, a shift in dominant species was observed: the dominance of SETVI decreased, while that of CAROC and OXASS increased. In particular, under full irrigation, STEME dominated in oxyfluorfen+glyphosate+diquat plots, while CAROC, OXASS, and SETVI dominated in pendimethalin+glyphosate+diquat,           indaziflam+diquat, and glyphosate+glyphosate+diquat plots, respectively. Under deficit irrigation, Urtica urens (URTUR) dominated in oxyfluorfen+glyphosate+diquat and glyphosate+diquat plots, while OXASS dominated in the mowing and weedy control plots. Interestingly, shifts in dominant species were observed after sequential herbicide treatments. Dominant species occurred over the course of the experiment, with different species observed between the first and second years (Table 8).

The diversity indices were used to quantify shifts in weed community structure among the treatments. The dominant species in the trial were SETVI, OXASS, and AMAVI, as confirmed by discriminant analysis (Tables 6,7,8). In the first season, weed control treatments exhibited distinct effects on weed populations under both the deficit and full irrigation regimes. After the initial herbicide applications at 110 DAP, treatments (except mowing and weedy control) showed similar weed community compositions across the irrigation regimes. At 110 DAP, CDA indicated that under deficit irrigation, C1 and C2 explained 99% and 92% of the total variation, respectively, and the discriminant function was statistically significant (p<0.05); under full irrigation, C1 and C2 accounted for 96% and 89% of the variation, respectively, and the model was likewise significant (p<0.05). After the second application at 215 DAP, divergence among treatments became more evident—particularly under deficit irrigation, where the weedy control separated from the clustered herbicide treatments. At this stage, although C1 and C2 explained 97% and 86% of the variation under deficit irrigation and 93% and 78% of the variation under full irrigation, the discriminant functions were not statistically significant (p>0.05) for either irrigation regime. By the pre-harvest period, weed communities were clearly separate according to treatment, indicating substantial temporal shifts in species composition. Before harvest, C1 and C2 each explained 99% of the total variation under both irrigation regimes. The discriminant function was statistically significant under deficit irrigation (p<0.05), but no statistically significant separation was detected under full irrigation (p>0.05). Under deficit irrigation, herbicide treatments and the weedy control were primarily associated with SETVI, AMAVI, and CYPRO; under full irrigation, mowing favored POROL, EPHPT, CONAR, and Digitaria sanguinalis (DIGSA), while SORHA and SONOL clustered under high-moisture conditions, suggesting enhanced competitiveness (Figure 2).

Figure 2
Discriminant analysis graphs of weed control treatments under deficit and full irrigation regimes for 2022–2023

With the onset of changes in weed populations during the second year, treatments showed similar weed community compositions under both irrigation regimes at 110 DAP, with the weedy control and mowing treatment somewhat distinct from the others. At this stage, under deficit irrigation, C1 and C2 explained 95% and 69% of the total variation, respectively, and the discriminant function was statistically significant (p<0.05); similarly, under full irrigation, C1 and C2 accounted for 96% and 76% of the variation, respectively, with a significant discriminant function (p<0.05). However, unlike the first year, after the second herbicide application at 215 DAP, treatments diverged completely across subplots. At 215 DAP, C1 and C2 explained 99% and 96% of the variation under deficit irrigation, whereas under full irrigation they accounted for 99% and 95%, respectively. Significant differences among treatments were observed under full irrigation (p<0.05), whereas no differences were detected under deficit irrigation. CDA for the 2023–2024 season again demonstrated clear separation among treatments under both irrigation regimes, and by the pre-harvest period, weed communities observed under both irrigation regimes were completely distinct. Before harvest, under deficit irrigation, C1 explained 98% and C2 explained 59% of the variation, although the discriminant function was not statistically significant (p>0.05). In contrast, under full irrigation, C1 and C2 explained 99% and 96% of the variation, respectively, and the discriminant function was statistically significant (p<0.05). Under deficit irrigation, plots treated with oxyfluorfen and the weedy control plots were dominated by HEOEU, MERAN, and EPHNU, indicating their tolerance to limited moisture and reduced canopy closure; under full irrigation, mowing and glyphosate-based herbicide treatments were more closely associated with ELEIN, SOLNI, and MALSS. Notably, CONAR remained distinct across all treatments and years, highlighting its persistence and adaptability under varying irrigation conditions. Compared to the previous season, greater species segregation and reduced overlap in canonical space in this season indicated increased ecological specialization and lower overall diversity. These results reflect the cumulative long-term effects of irrigation regimes and weed control strategies on the composition and dynamics of the weed community (Figure 3)

Figure 3
Discriminant analysis graphs of weed control treatments under deficit and full irrigation regimes for 2023–2024

4. Discussion

This two-year experiment sheds light on the dynamics of weed management in greenhouse banana production in Mediterranean conditions, particularly with regard to striking a balance between herbicide efficiency and sustainable agricultural practices. The results clearly show that reducing the frequency of herbicide use and integrating alternative, non-chemical strategies can achieve comparable levels of weed suppression to those of conventional herbicide-intensive systems, under both full and deficit irrigation regimes. The diversity indices in this study were not used to directly assess beneficial or harmful species but to quantify community shifts among the treatments, thereby providing insight into how different management strategies influence weed assemblages.

Of the herbicide treatments, a sequential application program consisting of indaziflam followed by diquat (indaziflam+diquat) resulted in numerically lower weed densities compared to oxyfluorfen+glyphosate+diquat, pendimethalin+glyphosate+diquat,                             and glyphosate+glyphosate+diquat programs. It should be noted that these results reflect the overall performance of sequential herbicide programs, rather than the effects of individual active ingredients. Additionally, statistically significant differences were detected only at specific observation dates; at all other sampling times, no differences were observed among the herbicide treatments. Overall, the results suggest that sequential programs starting with a pre-emergence herbicide with longer residual activity (indaziflam) followed by a post-emergence contact herbicide (diquat) can provide effective control of both early-emerging and persistent weed species. Similar findings were reported in Lugo-Torres et al. (2015), who demonstrated that sequential use of clomazone before planting and glyphosate after planting improved weed suppression efficiency in banana production in Puerto Rico.

Notably, in this study, the geotextile mulch treatment completely suppressed weed growth throughout the experiment, resulting in zero weed density and biomass. This finding corroborates the observations of Fongod et al. (2010) and De La Cruz et al. (2001), who reported that mulching effectively suppresses weed emergence, improves soil moisture retention, and enhances banana plant development. In contrast, manual mowing with three applications and the weedy control resulted in the highest weed densities and biomass, confirming that mechanical control alone (when not combined with other management strategies) is insufficient for high-humidity greenhouse conditions. Comparable results were reported by Hauser et al. (2006), who found that herbicide programs relying on a single mode of action (such as repeated glyphosate applications) failed to maintain long-term weed suppression, due to regrowth and species adaptation.

Interestingly, irrigation regimes (100% and 50%) had no significant effect on weed density or biomass. This suggests that, under controlled greenhouse conditions, irrigation level has a minimal influence on weed proliferation compared to the overall weed management program. This finding contradicts the commonly held assumption that higher soil moisture enhances weed proliferation. This lack of a significant effect may be due to the relatively uniform microclimatic conditions in the greenhouse, to the limited variability in soil moisture between treatments, or to the overriding influence of the weed management programs, which could mask subtle irrigation effects. On this point, Cortazar et al. (2017) found that planting density and canopy shading exert a greater influence on weed suppression than irrigation in banana systems.

Sequential herbicide programs (indaziflam+diquat and oxyfluorfen+glyphosate+diquat) significantly reduced both weed species richness and abundance, whereas untreated (weedy control) or inadequately managed (sequential mowing) plots exhibited higher species diversity and biomass accumulation. This response can be attributed to the initial application of soil-applied herbicides (oxyfluorfen, pendimethalin, and indaziflam), which effectively prevented weed emergence. Moreover, greenhouse irrigation conditions likely enhanced herbicide persistence in the soil, further limiting weed species establishment and resulting in the observation of only a few dominant species. Over the two-year experimental period, weed community composition shifted notably: SETVI dominated in the first year, while CAROC and OXASS became dominant in the second year. These changes highlight the adaptive and successional nature of weed communities in response to weed management programs and microclimatic conditions, rather than to individual herbicide active ingredients. Comparable findings were reported by Lanza et al. (2017) and Concenço et al. (2014), who observed that continuous shading and monoculture banana production systems favor the proliferation of shadetolerant and herbicide-resistant weed species.

CDA confirmed that weed management practices, rather than irrigation levels, were the dominant factors influencing weed species diversity and biomass. Both the diversity indices and CDA revealed that the distinctions among treatments were driven by specific dominant species and their associations with particular management practices. For instance, under deficit irrigation, herbicide applications and weedy control were primarily associated with SETVI, AMAVI, and CYPRO, whereas under full irrigation, mowing facilitated the prevalence of POROL, EPHPT, CONAR, and DIGSA. Additionally, species such as SORHA and SONOL clustered distinctly under highmoisture conditions, indicating greater competitiveness. The strongest correlations were observed at 110 and 215 DAP in the first year and at 110 DAP in the second year.

By integrating these CDA results with density, biomass, and diversity indices, it becomes clear that treatments such as sequential herbicide programs and geotextile mulching effectively reduced species richness and the dominance of persistent weeds, while mowing and weedy control allowed the proliferation of tolerant or opportunistic species. These findings underline the importance of comprehensive weed management approaches that integrate chemical, cultural, and physical methods to ensure long-term control and ecological balance. Grundy et al. (2010) and Tuck et al. (2014) also pointed out that field configuration, crop rotation, and neighboring crops substantially influence weed community dynamics and must be considered during the design of sustainable weed management strategies.

Furthermore, the present study demonstrates that strategically-timed sequential herbicide programs (2–3 per season) can effectively control weeds while minimizing chemical input. This supports Vezina (2016), who noted the diminishing effectiveness of continuous herbicide use due to the adaptation of persistent perennial species. The integration of mulching and optimized planting density offers a practical pathway for reducing herbicide dependency while maintaining effective control, in line with the observations of Bhattacharyya and Madhava (1987) and Costa (2007).

5. Conclusions

This study demonstrated that the sequential herbicide program involving indaziflam+diquat and geotextile mulching were the most effective strategies for controlling weeds in the greenhouse production of bananas under both full and deficit irrigation. Irrigation level had no significant impact on weed density or biomass, emphasizing the greater influence of weed management programs. Reducing herbicide applications to two or three times per year, using products with different modes of action, was sufficient for effective control. Geotextile mulching provided a sustainable, non-chemical alternative with comparable efficacy. The composition of weed species shifted over time, highlighting the need for adaptable, integrated weed management programs. Management strategies that combine reduced chemical use with ecological approaches can lower input requirements, preserve biodiversity, and support the long-term sustainability of banana agroecosystems.

Acknowledgements

This project is funded by the Republic of Türkiye, Ministry of Agriculture and Forestry, General Directorate of Agricultural Researches and Policies (TAGEM/BSAD/B/20/A2/P1/1553).

Data Availability

Data will be made available on request.

References

• Achard R. Weed control in banana plantations in Cameroon. Fruits. 1993;48(2):101-5.
• Arumugam S, Pugazhenthi G, Selvaraj S. Investigations on mechanical, thermal, wettability properties of processed banana leaves for sustainable food packaging applications. Biomass Convers Biorefin. 2023;14:22527–37. Available from: https://doi.org/10.1007/s13399-023-04655-2
  » https://doi.org/10.1007/s13399-023-04655-2
• Bauri FK, Sarkar SK, Bandyopadhyay B, Misra DK, Debnath S, Chakraborti K. Banana-cowpea association in the perspective of weed management in banana plantation under new alluvial zone of West Bengal. J Crop Weed. 2010;6(2):72-5.
• Bhattacharyya RK, Madhava RVN. Yield of ‘Robusta’ banana as influenced by soilcovers and soil moisture regimes. Banana Newsl. 1987;10:9-10.
• Castillo M, Guzman MJK, Aberilla JM. Environmental sustainability assessment of banana waste utilization into food packaging and liquid fertilizer. Sustain Prod Consump. 2023;37:356-68. Available from: https://doi.org/10.1016/j.spc.2023.03.012
  » https://doi.org/10.1016/j.spc.2023.03.012
• Çeliktopuz E. [Effects of different irrigation levels and bioactivator application on yield, fruit quality, and nutrient element content in two strawberry cultivars] [dissertation]. Adana: Çukurova Üniversitesi Fen Bilimleri Enstitüsü; 2019. Turkish.
• Concenço G, Motta IS, Correia IVT, Santos SA, Mariani A, Marques RF et al. Infestation of weed species in monocrop coffee or intercropped with banana, under agroecological system. Planta Daninha. 2014;32(4):665-74. Available from: https://doi.org/10.1590/S0100-83582014000400001
  » https://doi.org/10.1590/S0100-83582014000400001
• Cortazar SMU, Wolf ED, González IA. Effect of plant density on growth and yield in Barraganete plantain (Musa paradisiaca (L.) AAB cv. Curare enano) for a single harvest cutting in Provincia de Los Ríos, Ecuador. Acta Agron. 2017;66(3):367-72. Available from: https://doi.org/10.15446/acag.v66n3.52198
  » https://doi.org/10.15446/acag.v66n3.52198
• Costa JNM. [Production system for banana crops in the State of Rondônia]. Porto Velho: Embrapa Rondônia; 2007[access May 27, 2024]. Portuguese. Available from: https://ainfo.cnptia.embrapa.br/digital/bitstream/item/24653/1/sp-29-banana.pdf
  » https://ainfo.cnptia.embrapa.br/digital/bitstream/item/24653/1/sp-29-banana.pdf
• Daniells JW. Global banana disease management-getting serious with sustainability and food security. Acta Hortic. 2009;828(1):411-6. Available from: https://doi.org/10.17660/ActaHortic.2009.828.43
  » https://doi.org/10.17660/ActaHortic.2009.828.43
• Dassou AG, Tovignan S, Vodouhè F, Vodouhè SD. Meta-analysis of agroecological technologies and practices in the sustainable management of banana pests and diseases. Environ Dev Sustain. 2023;26:21937–54. Available from: https://doi.org/10.1007/s10668-023-03570-w
  » https://doi.org/10.1007/s10668-023-03570-w
• Davis PH. Flora of Turkey and East Aegean Islands. Edinburgh: University Press; 1965-1988.
• De La Cruz R, Rojas CE, Lobon HL, Burgos C. [The role of weeds in reducing nutrient leaching in banana crops in the humid tropics]. Man Integr Plagas. 2001;62:29–37. Spanish.
• Food and Agriculture Organization – FAO. Crops and livestock products. Rome: Food and Agriculture Organization; 2022[access Apr 3, 2024]. Available from: https://www.fao.org/faostat/en/#data/QCL
  » https://www.fao.org/faostat/en/#data/QCL
• Fongod AGN, Focho DA, Mih AM, Fonge BA, Lang PS. Weed management in banana production: The use of Nelsonia canescens (Lam.) Spreng as a nonleguminous cover crop. Afr J Environ Sci Technol. 2010;4(3):167-73. Available from: https://doi.org/10.5897/AJEST09.154
  » https://doi.org/10.5897/AJEST09.154
• Gorelick R. Combining richness and abundance into a single diversity index using matrix analogues of Shannon’s and Simpson’s indices. Ecography. 2006;29:525-30. Available from: https://doi.org/10.1111/j.0906-7590.2006.04601.x
  » https://doi.org/10.1111/j.0906-7590.2006.04601.x
• Grundy AC, Mead A, Bond W, Clark G, Burston S. The impact of herbicide management on long-term changes in the diversity and species composition of weed populations. Weed Res. 2010;51:187-200. Available from: https://doi.org/10.1111/j.1365-3180.2010.00831.x
  » https://doi.org/10.1111/j.1365-3180.2010.00831.x
• Hall RM, Penke N, Kriechbaum M, Kratschmer S, Jung V, Chollet S et al. Vegetation management intensity and landscape diversity alter plant species richness, functional traits, and community composition across European vineyards. Agric Syst. 2020;177:1-14. Available from: https://doi.org/10.1016/j.agsy.2019.102706
  » https://doi.org/10.1016/j.agsy.2019.102706
• Hauser S, Ngoumbe S, Nkongmeneck BA. Effects on plant species composition of glyphosate application in a plantain system after secondary forest clearing. In: Asch F, Becker M, editors. Prosperity and poverty in a globalized world: challenges for agricultural research. Conference on International Agricultural Research for Development; 2006 Oct 11-13; Bonn, Germany. p. 321.
• Lanza TR, Machado AFL, Martelleto LAP. Effect of planting densities of “Brs Princess” banana tree in the suppression of weeds. Planta Daninha. 2017;35:1-11. Available from: https://doi.org/10.1590/s0100-83582017350100054
  » https://doi.org/10.1590/s0100-83582017350100054
• Lugo-Torres ML, Díaz M, Acín N. Use of clomazone in plantain (Musa acuminata). J Agric Univ P R. 2015;99(1):59–61. Available from: https://doi.org/10.46429/jaupr.v99i1.2527
  » https://doi.org/10.46429/jaupr.v99i1.2527
• Magurran AE. Measuring biological diversity. Oxford: Blackwell; 2004.
• Prameela P, Menon R, Suma A. Identification of critical stages of weed competition and its effect on banana. Indian J Weed Sci. 2009;41(3-4):167-9. Available from: https://isws.org.in/IJWSn/File/2009_41_Issue-3&4%20Supplymentary_167-169.pdf
  » https://isws.org.in/IJWSn/File/2009_41_Issue-3&4%20Supplymentary_167-169.pdf
• Robinson JC. Bananas and plantains. Wallingford: CAB International; 1996.
• Singh B, Singh JP, Kaur A, Singh N. Bioactive compounds in banana and their associated health benefits – A review. Food Chem. 2016;206:1-11. Available from: https://doi.org/10.1016/j.foodchem.2016.03.033
  » https://doi.org/10.1016/j.foodchem.2016.03.033
• Singh HP, Batish DR, Kohli RK. Handbook of sustainable weed management. Boca Raton: CRC; 2006.
• Tarımsal Araştırmalar ve Politikalar Genel Müdürlüğü - TAGEM. [Standard methods for weed control herbicide trials]. Ankara: Tarımsal Araştırmalar ve Politikalar Genel Müdürlüğü; 2022. p. 1-152. Turkish.
• Tuck SL, Winqvist C, Mota F, Ahnstrom J, Turnbull LA, Bengtsson J. Land-use intensity and the effects of organic farming on biodiversity: a hierarchical meta-analysis. J Appl Ecol. 2014;51:746-55. Available from: https://doi.org/10.1111/1365-2664.12219
  » https://doi.org/10.1111/1365-2664.12219
• Türkiye İstatistik Kurumu (TR). [Central distribution system]. Ankara: Türkiye İstatistik Kurumu; 2022[access Apr 5, 2024]. Turkish. Available from: https://biruni.tuik.gov.tr/medas/?kn=92&locale=tr
  » https://biruni.tuik.gov.tr/medas/?kn=92&locale=tr
• Vezina A. Weed management. Promusa. Sept 13, 2016[access Mar 12, 2024]. Available from: https://www.promusa.org/Weed+management
  » https://www.promusa.org/Weed+management

Edited by

• Editor in Chief:
  Carol Ann Mallory-Smith

• Associate Editor:
  Silvia Fogliatto

Publication Dates

• Publication in this collection
  17 July 2026
• Date of issue
  2026

History

• Received
  24 Nov 2025
• Accepted
  26 Mar 2026