LPG Flaming—A Safe Post-Emergence Weed Control Tool for Direct Seeded and Bulb Onion

: The demand for pesticide-free food has increased the need for sustainable organic farming. Onion ( Allium cepa L.) is an important vegetable crop cultivated worldwide. The available weed control tools for intra-row weeds following onion emergence are limited. This study aimed to evaluate the potential use of liquefied petroleum gas (LPG) flaming as a pre- and post-emergence weed control method for both direct-seeded onion seedlings and transplanted onion bulbs. The safety of cross-row, where the flames are targeted to the intra-row area from both sides of the row, and broadcast flaming for bulb onion was compared. Cross-row flaming at twelve days after planting had no effect on onion dry weight, while broadcast flaming-treated plants’ dry weight was reduced by 36% as compared to controls. For the cross-row technique, the tested burners’ angle (45° and 30°) and inter-burner distances (30 and 40 cm) had no impact on weed control efficacy, and similar control levels, between 55% (15 cm) and 45% (10 cm), were observed 15 cm from both sides of the row-center. Direct-seeded onion cultivars were treated at various growth stages. The pre-crop-emergence stage was completely safe for the crop, and the second leaf stage exhibited a wide range of tolerance levels to flaming treatment across the different onion cultivars, with dry weights ranging between 39 and 117% compared to non-treated control in the flaming sensitive and tolerant cultivars, respectively. These results were validated under field conditions using the two most tolerant cultivars (Orlando and Browny); no yield reductions were observed for either cultivar when treated from the third leaf stage. In bulb onion, flaming had no impact on dry weight of shoots or roots when applied from four weeks after planting. This study demonstrates, for the first time, the potential of using flaming as a post-emergence weed control tool for direct-seeded and bulb onion, and at earlier time points than previously shown. Cross-row flaming proved effective for controlling intra-row weeds and can lower weeding costs. Future research should evaluate the safety of sequential applications and test complementary control methods for the initial growth stages.


Introduction
Onion (Allium cepa L.) is an important vegetable crop, grown in most parts of the world via direct seeded, bulb or transplanted cultivation [1][2][3]. Onion is a weak competitor with weeds [4] due to a slow seedling emergence and establishment rate [1], and its shoot architecture is of thin vertical leaves that do not produce much shade throughout the growing season [5]. It has been shown that season-long infestation of annual weeds, such as lambsquarters (Chenopodium album L.), can result in >90% yield reduction, with 43% reduction observed when weeds remain untreated during the seedling establishment stage [6]. Thus, developing efficient and sustainable weed management for onion, mainly during early growth stages, is essential to ensure high yields [7].
In recent years, increased consumer awareness for the negative effect of pesticide application on both the environment and human health [8,9], drove a shift toward organic crop production [10]. Yet weeds remain the central challenge hindering widespread adoption of organic agro-systems [11], owing to the limited availability of non-chemical weed control options [2,3]. While cultivation is a common mechanical control practice, the shallow root system of onion can be easily damaged by the cultivator blades resulting in crop injury and yield loss [1]. Hand weeding is a tedious and time-consuming method, which in many parts of the world is also expensive as human labor becomes limited [12]. Previous studies reported on the potential use of the other mechanical method, vertical brushes, which provided 74% weed control when combined with pre-emergence harrowing or flaming in bulb onion [2]. However, the limited control level achieved by these methods, coupled with the lack of control options for intra-row weeds (weeds growing within the crop row) emphasize the need to adopt new non-chemical control tactics that will be safe and effective for organic onion, particularly for intra-row weeds.
Liquefied petroleum gas (LPG) weed flaming (hereafter, flaming) is an alternative non-chemical weed control practice that can be implemented in organic farming [11]. This method has become the prime thermal weed control method [13] due to its high control potential, low environmental impact and low soil erosion [14,15]. The burners create high temperatures (up to 1900 °C), which instantly raise the temperature of the treated plant tissues to lethal levels [16], leading to plant cell membrane rupture and eventual tissue desiccation [17,18]. There are two main techniques for weed flaming: Broadcast and cross-flaming [19,20]. For broadcast flaming, the burners are mounted parallel to the crop row, with overlapping flames that cover the entire area of the field [21,22]. For cross-row flaming, the burners are mounted in a staggered pattern and angled perpendicular to the crop row; the flames are targeted to the intra-row area from both sides of the row [3]. Previous study showed that broadcast flaming can be safely used in bulb onion as a pre-emergence (PRE) treatment [2]. More recently, [3] delivered six sequential cross-row flaming after-emergence (POST) treatments to transplanted onion, and reported on the safety and efficacy of the technique. However, in these experiments onion plants were transplanted when 60 days old and treatments initiated 10 days after transplanting, i.e., on 70-day-old plants. Furthermore, only one onion cultivar was used in the study and there is no knowledge about potential differences in flaming-tolerance between onion cultivars. In Israel, direct seeding is the prime cultivation method (~90%), while bulb onion is the minor one (~10%) that is used between the direct seeded onion seasons due to its fast establishment and reduced time window of competition with weeds. Nonetheless, there are no data about the potential use of PRE or POST flaming treatment for these onion types, its impact on crop safety, or the safe time window for application. The current study aimed to identify the optimal flaming technique and the time window for safe POST flaming application in planted bulb and direct-seeded onion, and to evaluate flaming tolerance differences between direct-seeded onion cultivars.

Plant Material and Growth Conditions
Seven direct-seeded onion cultivars and one bulb onion cultivar were used in this study for the controlled and field trials (Table 1). Vaquero and Cimmaron are long-and intermediate-day cultivars, respectively, while the other cultivars are short-day cultivars, which are commonly used in the Israeli growth systems. For the effective flaming width experiments (Section 2.3), bread wheat (Triticum aestivum L.) plants (cultivar Galil, Hazera Genetics LTD., Kiryat Gat, Israel) were used as a model.  (1 L) were filled with clay soil (57% clay, 23% silt and 20% sand, on a dry-weight basis and 2% organic matter). Five uniform seeds were sown in each pot, and 7 days after emergence, thinned to one plant per pot. For the bulb experiments, one bulb per pot was planted so that the upper 2 cm of the bulb was above the soil surface. Pots were placed in a greenhouse and watered manually. For the flaming treatments, pots were taken from the greenhouse, placed 50 cm apart on a straight 10-m-long line and flaming was performed in a straight line ( Figure S1). Following treatment, pots were returned to the greenhouse and after 14 days, shoots were harvested separately and dry weight (DW) were measured and recorded after 72 h of drying in a 70 °C oven. For the bulb onion safe window trial (Section 2.4), dry roots, bulb weight and bulb diameter were also documented.

Optimal Burner′s Setup
Experiments were conducted on bulb onion, as preliminary experiments showed higher bulb tolerance to flaming at early growth stages as compared to direct-seeded plants. Plants were treated 12 and 19 days after planting (DAP), using an unshielded Red Dragon two-burner system equipped with two liquid-phase torches (LT 1 1/2 × 6; Flame Engineering Inc., LaCrosse, KS, USA). The burners were connected to a 12-kg LPG tank mounted on a cart to simulate application from moving platform, i.e., commercial tractor application ( Figure S1). The cart was manually pushed over the pots at a constant speed of 3 km h −1 , with LPG pressures of 25 psi and 50 psi that were converted to per area dose: 48 kg ha −1 and 83 kg ha −1 , respectively. For the broadcast technique, burners were mounted 30 cm apart, positioned 20 cm above the soil surface, parallel to the crop row and angled 30° toward the soil. For the cross-row technique, burners were mounted at a 45° angle with respect to zenith, 20 cm from the crop line and 20 cm above the soil surface. The shoot DWs were determined as described in Section 2.1. The experiment was performed at complete randomize design with 10 replicates.

Effective Flaming Width
Wooden containers (55 × 20 cm) filled with soil, were used to evaluate the effective control width in different cross-row setup combinations ( Figure S2). Two rows of wheat were sown, with 4 cm between rows and 1 cm between plants within rows ( Figure 1). At 20 DAP, the plants were treated with 48 kg ha −1 LPG. Two burner mounting angles, 30° and 45°, and two distances between burners, 30 cm and 40 cm, were tested ( Figure S2). Two days after treatment, the shoot DW of the wheat plants was determined from five sections (5 and 10 cm) from the center of the wheat seed line ( Figure 1). The experiment was performed at complete randomize design with four replicates.

Determination of the Safe Flaming Time Window
To determine the safe time window for flaming treatments for bulb and direct-seeded onions, seven direct-seeded cultivars and one bulb onion cultivar were planted and grown as described above. Plants were treated with 48 kg ha −1 LPG, using a single-burner setup, positioned at a 30 cm distance between the burners and 45° inclination angle. The direct-seeded onions were treated pre-emergence (PRE), at the cotyledon, first leaf and second true leaf stages. The bulb onions were treated every week at seven timings: From one week to seven weeks after planting, starting four days after planting. Each plant was treated once. Dry weight evaluations were performed as described above. For the bulb onions, bulb diameter and root DW were also documented. The experiment was performed at complete randomize design with five replicates.

Field Validation
To validate the safe time window determined in Section 2.4, the impact of flaming on crop safety was assessed under field conditions. Weed control efficacy was not tested, thus, control treatments that were used for the commercial plot were also applied at our experimental area. Experiments were held in the 2017/8 growing season (winter-spring) in two commercial fields: Gazit (Lat 35°27.1′, Long 32° 37.96′ N) and Kefar Baruch (Lat 35°13.5′, Long 32°39.2′). Irrigation and fertilization treatments were similar to those delivered to the adjacent commercial field. Table 2 summarizes the cultivar, seeding dates, seeding stands, bed sizes and application timings of the two experimental sites. Plots consisted of one raised bed wide by 10 m long. The left seeding line of each plot was flamed with the burner setup described in Section 2.4 ( Figure S3). The right seeding line of each plot served as a control. One month after treatment, bulb diameter and shoot and root DW were documented from 1 m sections collected from the center of each plot. At the end of the growing season and before harvesting, the counts and weights of bulbs collected from a 1 m section of each plot were documented. Experiments were performed at complete randomize designed with five replicates.

Statistical Analysis
All statistical analyses were conducted using the JMP ® ver. 14 statistical package (SAS Institute, Cary, NC, USA). The data-sets were tested and the assumptions of homoscedasticity and normality were met using Levene′s tests. For the optimal setup experiment (Section 2.2), parameters were presented as percent of non-treated control. Data were subjected to ANOVA and two-way analyses were conducted to determine the interaction between flaming technique and LPG dose at the two application timing (12 and 19 DAP) on the crop DW. For the effective width experiment (Section 2.3), data from the 5 (L1), 10 (L2), 15 (L3) and 25 (L4) cm sections ( Figure 1) from both sides of the center section were averaged in each container. Data were analyzed as percent of non-treated control. Data were subjected to ANOVA and three way analyses were performed to determine the interaction between burner angle, distance between burners and distance from the row center on wheat DW. As the distance between burners and distance from the row center were not a significant factors (Section 2.2) data were pooled, and means were separated using Tukey HSD test. For the bulb onion safe timing experiment (Section 2.4), shoot and root DW and bulb diameters of the different cultivars were analyzed by comparing means of control and treated plants using a t-test, at the 5% level. For the direct seeded onion, one way ANOVA was preformed to determine the impact of cultivar on the shoot DW (parameter was presented as percent of non-treated control) and means were separated using Tukey HSD test. For the field validation experiment, parameters (except final yield) are presented as percent of non-treated control. Means of control and treated plants at the different application timings were compared using a t-test, at the 5% level.

Flaming Technique Affects Onion Safety
Comparison of the cross-row versus broadcast flaming techniques demonstrated that cross-row flaming 12 DAP had no significant impact on crop safety, while broadcast treatment led to a significant (p = 0.001) 36% reduction in shoot DW in treated versus control plants. Furthermore, no LPG dose by flaming technique interaction was observed (p = 0.592), suggesting that the cross-row treatment is safer regardless of the applied LPG dose. In contrast, when treated 19 DAP, none of the main factors (flaming technique and LPG dose) nor the interaction between them had a significant impact on the crop safety (p > 0.414), suggesting that the differential impact of the tested flaming techniques on crop safety is relevant at early growth stages only and that developed plants are more tolerant.

Burner Angle and Distance Do Not Impact Effective Control Width
Comparison of the effective control width of various cross-row flaming setups demonstrated that neither burner angle nor distance had a significant impact on the control width (p = 0.706 and p = 0.374, respectively (Table 3). Some interaction was observed between the angle and distance (p = 0.046) suggesting a minor effect of the combination on cross-row flaming efficacy. Burner distance from the center of the row was the only significant factor (p < 0.0001). Dry wheat weight was significantly higher at 25 cm from the row center as compared to shorter distances from the row center, 0, 5, 10 and 15 cm (Figure 2). The DW median value at these distances was between 60 and 65% of non-treated control, compared to 82% of control at a distance of 25 cm from the center. The limited dry weight reductions observed in this trial may be attributed to the short time span between the flaming treatment and the final evaluations, two days. Longer duration would probably resulted in higher impact on wheat dry weigh, however, the main objective here was to compare different burner setups rather than control efficacy.

Bulb Onions Can Be Safely Treated by POST Cross-Row Flaming Starting 4 WAP
A significant reduction in shoot and root DWs and bulb diameter was recorded following the first flaming treatment (one week after planting [WAP]) (p = 0.009, p = 0.009 and p = 0.007, respectively; Figure 3). More specifically, median shoot DW values dropped from 0.58 g in control samples to 0.51 g in treated plants, while the median bulb diameter declined from 5.6 cm to 4.8 cm.
In terms of shoot DW, the subsequent applications were completely safe and no reductions were observed. However, the roots and bulb DW were more affected by the flaming treatments and treated plants exhibited significant reduction in the first three application timings (p = 0.009, p = 0.018 and p = 0.027, respectively). Starting four WAP, flaming treatments were safe for the onion plants and no shoot and root DW or bulb diameter reductions were observed (Figure 3).

Flaming Tolerance Varies across Seeded Onion Cultivars
This experiment aimed to set the safe flaming time window for direct seeded onion, and to evaluate the heat tolerance level of different cultivars. Flaming at the first two phenology stages, PRE emergence and cotyledon, resulted in similar trends across all tested cultivars. The PRE application was completely safe, with no significant differences (p = 0.293) between the control and treated plant shoot DWs. In contrast, flaming at the cotyledon stage led to a significant reduction (>95%) in DW in all tested cultivars (p = 0.0002), with tolerance variations across cultivars manifesting from the first leaf time point, ranging from 43% to 10% compared to control. At the second leaf stage, the flaming-tolerance spectrum became wider, with DWs of the highly tolerant cultivars, Orlando and Browny, being 117% and 85% of control, respectively (Figure 4). The rest of the cultivars had low-tolerance for the flaming treatment and their DWs were significantly lower, between 39% and 54% of control, respectively (Figure 4).

LPG POST Flaming Is Effective under Field Conditions
For field validation, the two flaming-tolerant cultivars were used: Orlando and Browny. The PRE treatment was applied only for Browny, and was safe with no effects on growth or yield as compared to the non-treated plants (Figures 5 and 6). In the case of flaming treatment at the second leaf stage, Orlando showed a significant reduction in shoot and root DWs and bulb diameter, in the treated compared to the non-treated plants (p < 0.009) ( Figure 5). For example, the median value of shoot DW of treated plants was ~70% of the non-treated control plants (Figures 5 and S4). For Browny, only the shoot DW underwent a significant (p = 0.003) reduction, with the median DW value of treated plants was ~50% of the non-treated control plants. However, by the end of growing season, the treated Orlando plants fully recovered from the flaming treatment and no significant differences in marketable yield were observed between the treated and non-treated plots ( Figure 6). In contrast, final Browny yield was significantly (p = 0.0023) lower in the treated compare to the non-treated plots ( Figure 6).  Later flaming applications (third and fourth leaf) had no negative effects on either cultivar ( Figure S4). Despite the significant (p = 0.022) reduction in Browny shoot DW following the third leaf application, the plants recovered from the treatment; no yield reductions were observed at the end of the season (Figure 6). At the fourth leaf application, no significant differences were noted in DWs and final yields of treated versus non-treated plots. The Orlando showed a minor (p = 0.045) yield reduction following the eighth leaf application, with median value dropping from 0.16 to 0.12 kg plant −1 between the non-treated and the treated plots ( Figure 6).

Discussion
Weed flaming following crop emergence can be accomplished either by broadcast flaming [8,21,23,24] or by cross-row flaming [15,25,26]. However, optimal flaming techniques for weed control must avoid damaging the crop plant growth rate and reduce the growth of competing weeds [27]. This study demonstrated the potential safe usage of cross-row weed flaming for direct-seeded and bulb onion. Our results showed that the cross-row technique was superior to the broadcast technique, primarily at the first application timing at 12 DAP (Table 4). Due to the location of the onion apical meristem between the leaves at the soil surface [28], the cross-flaming approach, in which the flames are targeted at the crop center from both sides, seemingly causes less damage to this growth tissue and has less of an impact than broadcast flaming on plant development. Cross-row flaming may also have economic benefits, e.g., lower LPG consumption as application is restricted to the seed line; however, this aspect is of secondary importance [26]. Optimal LPG doses must effectively control weeds while eliciting minor damage to the crop [27,29]. In our study, the lowest tested LPG dose (48 kg ha −1 ) proved safest, when applied at early growth stages, and was not associated with any biomass reduction. This dose effectively controlled weeds in transplanted onion in previous study [3], and was also reported to effectively control local Mediterranean annual species, e.g., Sinapis arvensis, Lavatera trimestris and Avena sativa [30]. In both studies, this dose was applied via cross-row flaming. It can be assumed that early growth stages of direct-seeded onion are more sensitive than bulbs to flaming treatment, thus, the lower dose was used in subsequent experiments.
While cross-row flaming aims to control the intra-row weeds [31], the width of the effective control strip achieved by various burner setups has not been previously evaluated. Our results showed that the burner angle and the inter-burner distance had no impact on the width of the effective control strip, suggesting flexibility in cross-row setups. These results are in agreement with previous study that evaluated similar setting (angle and the inter-burner distance) effects on broadcast flaming effectiveness [32]. The observed 30 cm-wide effective flaming strip supports the conclusion made by others [13,15,25] regarding the need to complete weed control on the rest of the bed using another control method. However, in high-density crops (e.g., direct-seeded onion in Israel) with <30 cm distance between seeding lines, the cross-row technique may result in crop injury by adjacent burners targeting nearby seed lines. In such cases, reducing the number of seed lines per bed and increasing the line spacing must be considered to ensure the crop safety.
The tested onion cultivars showed a wide range of tolerance to the flaming treatment, mainly at the second leaf stage, with DWs of the tolerant cultivars Orlando and Browny approximately two-times higher than the low-tolerant cultivars (Figure 4). These findings can be harnessed to improve LPG-based management in onion by integrating and genetically breeding flaming-tolerant cultivars, as is done with herbicide-tolerant cultivars [33]. Direct-seeded onion has limited solutions for the control of intra-row weeds, thus identification of cultivars with high flaming tolerance may broaden the available toolbox for onion weed management [34]. Additionally, information regarding flaming tolerance level diversity across crop cultivars can be useful for plant breeders [35,36]. Taking into consideration tolerance for flaming at early stages of the selection process can result in cultivars with high tolerance levels at the end of the process, which in turn may allow flaming applications at higher doses and better weed control.
Determining the safe time window for flaming application is a key to commercial application of the method [26]. The window will vary across crops and cultivars [25,37,38]; for direct-seeded onion, the second leaf was found to be the earliest growth stage at which the treatment was safe, with no effects on shoot and root DW or yield. However, in some scenarios (e.g., Browny field trial), further development is needed and the third leaf stage is recommended. Relying on the phenological features rather than on chronological age to define the safe time window enables assessment and application across varied climates and growing systems [39]. For bulb onion, the four WAP treatment was safe and had no impact on DW. In terms of chronological age, this time point is earlier than the second and third leaf stages, yet, comparison between these onion types is misleading as the bulb onion is a more vigorous and tolerant crop. The 4 WAP (<30 days) period is shorter than the 70-day period set by [3], thus, our findings allow for earlier and safe LPG application in bulb onion compared to transplanted.
As with other crops whose earliest flaming timing requires additional control methods to prevent weed establishment, both onion types evaluated here will require integration of another control method, e.g., stale seedbed or cultivation [31]. It is likely that more than one flaming treatment will be required to ensure commercial control level throughout the growing season. However, a previous study demonstrated that six sequential LPG applications are safe for transplanted onion and is likely to be suitable for direct-seeded and bulb onion as well [3].

Conclusions
Flaming is a non-chemical weed control method widely used worldwide. To the best of our knowledge, this is the first attempt to assess its post-emergence application in direct-seeded and bulb onion. Our findings showed that flaming can be safely used in both direct-seeded and bulb onion before (PRE) and after (POST) crop emergence. Interestingly, the burner angle and inter-burner distances had no impact on control effectiveness. The effective control width was 30 cm and must be taken into consideration when <30 row spacing is used. Onion sensitivity to the flaming treatments varied across phenology stages and genotypes. The PRE emergence treatment was completely safe, whereas treatments at the cotyledon and first leaf stage resulted in plant death. Some of the high-resistance cultivars can be safely flamed starting from the second leaf stage, but waiting until the third leaf stage will ensure complete safety. These findings were validated under field conditions, under which neither growth inhibition nor yield reduction were observed following flaming at the third leaf stage. Future research should focus on determining the safe number of sequential flaming applications in these onion types. Our findings demonstrate the potential of integrating POST flaming treatment into onion crop weed management protocols, either in organic agro-systems, or as part of integrated weed management programs for conventional systems.  Acknowledgments: The authors are grateful to Nachshon from Gazit and Shay from Gevat for facilitating and supporting the field trials.

Conflicts of Interest:
The authors declare no conflict of interest. The funders had no role in the design of the study, in the collection, analyses, or interpretation of data, in the writing of the manuscript, or in the decision to publish the results.