Fall Flame Weeding: Targeting weed seeds before they enter the seedbank

Final Report for ONE11-141

Project Type: Partnership
Funds awarded in 2011: $12,238.00
Projected End Date: 12/31/2013
Region: Northeast
State: Maine
Project Leader:
Dr. Eric Gallandt
University of Maine
Expand All

Project Information


Flaming is commonly used on organic and low-external-input farms to control weed seedlings, especially annual dicots in slow-to-emerge crops such as carrot and beet. Flaming could, however, also be used to kill weed seeds before they enter the seedbank, a strategy used in recent years by Rob Johanson on the Goranson Farm in Dresden, Maine. In our field studies, pre-dispersal flaming did not affect viability of common lambsquarters or redroot pigweed seeds. Flaming, however, shows promise as a method to reduce density of weed seeds following dispersal. Greenhouse and field studies conducted on-farm demonstrated that flaming could kill weed seeds on the soil surface. Typical tractor speeds used for other flaming operations, e.g., 1.6 mph, killed about 50% of the most sensitive species (i.e., hairy galinsoga), however, with the flame dosage doubled, i.e., to 0.8 mph, flaming reliably killed 75% or more seeds of mustard, large crabgrass and hairy galinsoga. There was no advantage to further doubling the flaming dosage, as seed mortality was similar with both 0.8 and 0.36 mph treatments. With an estimated cost of $150 per acre for 0.8 mph treatment, fall flaming could prevent large weed seedbank credits, especially relatively sensitive species including hairy galinsoga.


Annual weeds are consistently ranked as a challenging production problem for organic and low-external-input farmers. While these growers often focus weed management efforts on controlling weed seedlings, often densities are very high requiring multiple cultivation events and, in high value crops, supplemental hand weeding. If, however, a broader perspective to weed management is used, and strategies are employed to reduce the weed seedbank, subsequent weed seedling densities are proportionally reduced and the outcome of physical weed control improves (Gallandt, 2006). Weed seedbanks vary widely across Northern New England organic farms, ranging from a low of approximately 2,500 germinable seeds m-2, to over 25,000 m-2 (Gallandt, unpublished, Figures 1 and 2). Practical strategies to reduce the germinable weed seedbank begin with effective cultivation. Targeted fallowing, timed to coincide with peak emergence potential of problem species is also foundational for managing the seedbank. Also, in diverse enterprises, intensive sequences of very short-season crops can incorporate frequent soil disturbance, which encourages weed germination, while completely preempting weed seed rain. The combined approach of encouraging germination losses or “debits” to the seedbank, while preventing seed rain “credits,” is described in a recent webinar by the PI: http://www.extension.org/pages/62445/cultivation-and-seedbank-management-for-improved-weed-control-webinar Despite a thoughtful and well-executed weed seedbank management plan, weeds escaping early season weeding may become large and produce abundant seed rain (e.g., Figure 3). In a previous, but related NE SARE research project (LNE06-237: “Managing weed seed rain: A new paradigm for organic and low-input farmers”) we examined contrasting fall weed management practices in an attempt to maximize post-dispersal seed mortality. In this study we demonstrated that complete preemption of weed seed rain resulted in rapid and large reductions in the weed seedbank. The central hypothesis motivating this work was related to fall cover cropping which we speculated would incorporate newly dispersed seed into the soil offering protection from seed predators (e.g., Figure 4). Using mesh exclosures post-dispersal through the following spring, we found that seed predation had a similar large effect on the seedbank, but this effect was inconsistent: in one year predation reduced the seedbank 40%, but in three other years predation losses were not detected. Rob Johanson, a diversified organic farmer in Maine, routinely flames potato to kill vines prior to harvest, typically travelling 1.5 mph, with 40 psi at the burners, but flaming speed depends on the condition of the vines and may range from 1.0 to 2.0 mph, occasionally two passes are made. Several years ago he started intentionally reducing his tractor speed to increase flaming dosage in fields with abundant fall seed rain. In short, he reasoned that post-dispersal seed flaming could be used to quickly and effectively kill weed seeds prior to their entering the seedbank. Standing weeds in potato fields are blackened and smoldering post-flaming, and it seemed logical that flaming would kill weed seeds on the soil surface. In 2010 we conducted several preliminary dose-response experiments examining effects of increasing flame duration on condiment mustard (Sinapas alba). Increasing exposure from 1 to 5 seconds in a propane flame of 900C reduced viability (Figures 5 and 6). The effective doses were, however, quite high indicating that practical tractor speeds and propane rates might be effective only on certain weed species. Our aim with this project was to evaluate flaming effects on weed seeds, both pre- and post-dispersal, and the effect on the subsequent weed seedbank, with a particular focus on species with average seed mass less than mustard that would likely be more susceptible to flaming. Flaming effects on weed seedlings (Ascard, Hatcher, Melander, & Upadhyaya, 2007; Ascard, 1994, 1998), and effects of field burning on seeds (Ensign, 1973; Young, Ogg, Dotray, & Press, 1990) have been studied previously. However, I found only a single report of flaming effects on weed seeds in the literature. Pali, Pomsar, & Reisinger (2007) examined flaming effects on seeds of several annual weeds and found mortality to ranged from a low of 80% for redroot pigweed (Amaranthus retroflexus) to 100% for maple-leaved goosefoot (Chenopodium hybridum) and barnyardgrass (Echinochloa crus-galli). These results, and the recent experience of Mr. Johanson at the Goranson Farm, indicates that seed flaming has promise as an additional tool for managing the weed seedbank.

Project Objectives:

1. Conduct quantitative evaluations of flaming effects on weed seeds; and 2. Present results of laboratory, greenhouse and field experiments to northern New England farmers and the scientific community.


Click linked name(s) to expand/collapse or show everyone's info
  • Dr. Randa Jabbour
  • Rob Johanson


Materials and methods:
Greenhouse / Controlled Environment Trials

Preliminary dose-response trials were conducted in a glasshouse using a typical hand-held propane wand, a single burner, attached to a 17 lb. liquid propane gas tank. This unit was attached to a set of wheels to improve consistency of both speed and burner height. Temperature 1 cm above ground were recorded by 4 thermocouples positioned 2 cm apart in a row perpendicular to the operating direction of the flamer. Thermocouples of type K (Mineral Insulated Thermocouple 304 St/St Sheath, 0.020” dia x 36” long with Miniature Plug) were used. The temperature data were collected with an OMEGA HH309A Data Logger Thermometer, at a sampling interval of 1 sec.

Pre-dispersal Seed Flaming

Pre-dispersal flaming effects on seed viability were tested at the Goranson Farm in Dresden, ME, using a liquid propane unit with 6 x 1 million BTU burners that Mr. Johanson constructed (Figures 7 and 8), using commercially available parts (Dennis Lutteke, Wells, MN). [we were unable to conduct the study on the Vermont farm due to complications in the measurement data collectors and other factors.] On October 10, 2011, mature, reproductive common lambsquarters and redroot pigweed plants were flamed at 1.64 mph, 40 psi, the typical flame dosage used by Mr. Goranson for killing potato vines. Six plants of each species were removed after flaming and three controls were randomly selected from an untreated area for both species (Figures 9 and 10). In the lab, the seed were threshed by hand and 20 seeds from each plant were selected randomly and treated with TTC. Each seed was bisected, placed on filter paper and treated with one drop of TTC. The treated seed was kept on a lab bench for at least 5 hours and then observed under a microscope. If there was easily observed color change anywhere on the seed (indicating respiration) the seed was deemed viable.

Post-dispersal Seed Flaming Trials

Post-dispersal flaming effects on seed viability were tested at the Goranson Farm in Dresden, ME using Mr. Johanson’s flaming unit described above. On November 15, 2011, seeds of condiment mustard, large crabgrass (Digitaria sanguinalis) and hairy galinsoga (Galinsoga cilata) were placed onto bare soil in steel mesh cages and flamed at 3 different tractor speeds (0.37 mph; 0.79 mph; and 1.64 mph) (see Video at YouTube, "zeroseedrain"; http://www.youtube.com/watch?v=gS0lXRvEEwU and Figure 11). Twenty-five seeds of each species were removed from the mesh cages, put in petri dishes with germination paper and placed in a germination chamber at 25 C and 12 h light-dark cycles and removed after four days to prevent fungal growth. Those seeds that did not germinate or did not show obvious signs of germination damage (i.e. integuments split yet no radicle emerged) were tested by various methods for viability. Mustard: The mustard seed showed 100% germination in the control so no further protocols, beyond using the germination chamber, were required to test the mustard’s viability. Crabgrass seed was tested conducting pinch tests: the seed coat was removed and the embryo observed/ pinched by tweezers under a microscope. If the seed had scorch marks or was too delicate to handle (e.g. it broke apart before the pinch test could be performed) it was considered non-viable. If, when pinched, the seed was chalky, white or dry, mealy after being pinched the seed coat is empty, aborted germination the seed was non-viable. Non-viable seeds were often expanded/swollen due to exposure to heat though this observation was not a factor in determining viability. Viable seed had no burn marks and when crushed, the seed would remain translucent and the pieces of seed that remain are large. Non-germinated GALCI seed was tested using a 1% Triphenyl Tetrazolium Chloride (TTC) solution. Each seed was bisected, placed on filter paper and treated with one drop of TTC. The treated seed was kept on a lab bench for at least 5 hours and then observed under a microscope. If there was easily observed color change anywhere on the seed (indicating respiration) the seed was deemed viable.

Flaming Effects on the Weed Seedbank

On October 10, 2011, we selected a relatively weedy field coming out of mid-season brassicas to establish a replicated set of flaming treatments including three flaming exposures based on decreasing forward tractor speed: 1.6 mph; 0.8 mph; and 0.36 mph. There were three replicates of each treatment, each 6 ft. wide (the width of the flaming unit), and 85 ft. long. The predominant weeds at the site were large crabgrass, common lambsquarters and redroot pigweed. On May 25, 2012, plots were located using a GPS and a bulb planter (33.2 cm2) was used to randomly remove five cores to a depth of 10 cm from each plot. All cores were placed in bucket, sifted and mixed before being placed in 1020 flats containing vermiculite and watered as needed. Seedlings were identified, counted and removed until no more germination occurred.

Research results and discussion:

Maximum temperatures (Tmax) in the field ranged from 314 F for the 6 x 1-million BTU liquid propane flaming unit at 40 psi travelling at 1.6 mph, to 1700 F at 0.36 mph (Figure 12). As noted previously, practical tractor speeds for flaming range from 1.0 to 2.0 mph; we included the slower speeds (higher flaming dosages) to increase likelihood of detecting seed mortality. Unexpectedly, these field measurements of Tmax, and the integral of total exposure, were challenging to measure. Heavy gauge thermocouples were durable but not sufficiently responsive to accurately measure Tmax given the relatively brief exposure to flames at higher speeds (lower flaming dosages). Fine-wire thermocouples, while responsive, were fragile and prone to failure. In conducting future flaming studies we will collaborate with engineering colleagues to design a precise and robust field unit for measuring exposure.

Pre-dispersal Seed Flaming

Despite smoldering and blackened foliage and seed heads, pre-dispersal flaming did not reduce the proportion of viable redroot pigweed or common lambsquarters seed compared to non-flamed control plants (Figure 13). This is, perhaps, not surprising for these two species in which seeds are retained in considerable maternal vegetation that likely offers protection from flames.

Post-dispersal Seed Flaming Trials

Post-dispersal flaming effects on redroot pigweed were particularly dramatic. Although the typical exposure of 1.6 mph did not affect redroot pigweed seed, increasing exposures with slower tractor speeds of 0.8 and 0.36 mph actually “popped” the seeds (Figures 14 and 15). Hairy galinsoga seeds were visibly scorched at even the lowest dose, 1.6 mph, and many combusted at the high dose of 0.36 mph (Figure 16). Condiment mustard, used here as a “surrogate weed” for wild mustard because of its high viability and rapid germination, was unaffected by flaming at 1.6 mph, but completely killed at 0.36 mph (Figure 17). Large crabgrass seed viability was similarly unaffected by flaming at 1.6 mph, but reduced by 80% or more at 0.8 and 0.36 mph (Figure 18). Overall, post-dispersal field experiments (Figure 18) demonstrated that: (i) flaming can kill weed seeds on the soil surface; (ii) typical tractor speeds used for other flaming operations, e.g., 1.6 mph used here, killed only about 50% of the most sensitive species (i.e., hairy galinsoga); (iii) flame dosage must be doubled, e.g., tractor speed reduced to 0.8 mph, to reliably kill 75% or more seeds of mustard, large crabgrass and hairy galinsoga; and (iv) there was no advantage to further doubling the flaming dosage as seed mortality was similar with both 0.8 and 0.36 mph treatments.

Flaming Effects on the Weed Seedbank

The germinable seedbank was similar across the three levels of flaming tested (P = 0.274, Figure 19). The rank order of means was consistent with the predicted dose-response, i.e., the seedbank was greatest at the 1.6 mph flame dose, intermediate at the 0.8 mph dose, and lowest at the 0.36 mph dose (Figure 19). However, due to the highly variable nature of weed seedbanks, the within replicates variability was sufficiently large that treatment effects were not detected. Future studies should include both weed seed additions in micro-plots, and increased sample sizes to overcome this problem.

Research conclusions:

Post-dispersal flaming is another tool that can be considered part of a multi-faceted strategy for managing the weed seedbank. While slow and costly, flaming at tractor speeds of 0.8 mph should offer a high level of control of recently dispersed annual weed seeds (see Figure 18). While preventing weed seed rain with short-season cash or cover crops is more effective and reliable, flaming may be an appropriate tool in higher value vegetable systems where management is focused on reducing the weed seedbank.

Participation Summary

Education & Outreach Activities and Participation Summary

Participation Summary:

Education/outreach description:

Predation, preemption, burial and flaming: Managing weed seed rain. Weed Science Society of America Annual Meeting, Baltimore, MD (February 5, 2013; 30 attending). Flame weeding and managing the weed seedbank. Maine Organic Farmers and Gardeners Association and University of Maine Cooperative Extension Farmer to Farmer Conference, Northport, ME (November 10, 2012; 28 attending). Johanson, R. and J. Goranson. Goranson Farm, Dresden, Maine. Flame Weeding. Maine Organic Farmers and Gardeners Association and University of Maine Cooperative Extension Farmer to Farmer Conference, Northport, ME (November 10, 2012; 35 attending). Germination, preemption, predation, and decay: Managing the weed seedbank. Organic Plant Production Graduate Course Lecture, Wageningen University, the Netherlands (March 28, 2012; 32 attending). Decay, predation, germination and preemption: Managing the weed seedbank. Northeast Organic Farming Association Vermont Winter Meeting, Intensive Seminar: Weed Management in a Wetter, Warmer Climate, Burlington, VT (February 10, 2012; 54 attending). Diversification, soil quality and integrated weed management. Integrated Weed Management Symposium, Northeast Weed Science Society Annual Meeting, Philadelphia, PA (January 5, 2012; 45 attending).

Project Outcomes

Project outcomes:

Our on-farm flaming trials included too small an area, using too little propane to measure actual lbs. of propane used per acre at a given flaming speed. However, in our controlled environment experiments, we were able to estimate propane consumption of 250 lbs. per acre at 0.8 mph. Considering the August 2013 price of propane in Maine at $0.61 per lb., the flame dosage required for effective control of the annual weed tested here would be $150 per acre. This is a reasonable expense for high value crops, or considering spot flaming only areas with abundant seed rain or fields that will be rotated into crops that require a high level of weed control. The results are sufficiently promising to warrant a more extended field experiment that includes flaming fields dominated by a range of weed species as compared to other fall seedbank management practices, and a partial budget analysis of weeding costs over the fall and subsequent cropping cycle.

Farmer Adoption

We did not conduct any post-project surveys to evaluate farmer adoption based on this project.

Assessment of Project Approach and Areas of Further Study:

Areas needing additional study

Here we have established a very solid set of dose-response datasets, documenting both dose and species variation in response to flaming, and furthermore, demonstrated that pre-dispersal flaming is ineffective. The next logical step is a longer-term field experiment comparing fall weed management techniques on the subsequent weed seedbank, and, importantly, through the cultivation and hand weeding in subsequent crop, including partial budget comparisons and multi-year simulation modeling.

Literature Cited

Ascard, J. (1994). Dose–response models for flame weeding in relation to plant size and density. Weed Research, 34(5), 377–385. Ascard, J. (1998). Comparison of flaming and infrared radiation techniques for thermal weed control. Weed Research, 38(1), 69–76. Ascard, J., Hatcher, P. E. E., Melander, B., & Upadhyaya, M. K. K. (2007). Thermal Weed Control. In M. K. Upadhyaya & R. E. Blackshaw (Eds.), Non-chemical weed management (pp. 155–175). CAB International. Ensign, R. (1973). Post-harvest field burning of seed crops in the Northwest. In Proceedings of the annual Symposium on Thermal Agriculture (pp. 37–52). St. Louis, Missouri: National LP Gas Association. Gallandt, E. R. (2006). How can we target the weed seedbank? Weed Science, 54(3), 588–596. Pali, O., Pomsar, P., & Reisinger, P. (2007). Thermal method to control dangerous weeds. In Cereal Research Communications (Vol. 35, pp. 885–888). Obervellach, Austria: VI. Alps-Adria Scientific Workshop. Young, F. L., Ogg, A. G., Dotray, P. A., &Press, A. (1990). Effect of postharvest field burning on jointed goatgrass (Aegilops cylindrica) germination. Weed Science, 4(1), 123–127.

Information Products

Any opinions, findings, conclusions, or recommendations expressed in this publication are those of the author(s) and do not necessarily reflect the view of the U.S. Department of Agriculture or SARE.