Evaluation of organic strategies to control a new invasive pest, swede midge, Contarinia nasturtii (Diptera: Cecidomyiidae)

Final Report for ONE11-139

Project Type: Partnership
Funds awarded in 2011: $15,000.00
Projected End Date: 12/31/2013
Region: Northeast
State: Vermont
Project Leader:
Dr. Yolanda Chen
University of Vermont
Expand All

Project Information


Swede midge, Contarinia nasturtii, (Diptera: Cecidomyiidae) is an invasive insect pest that feeds on crops and weeds in the family Brassicaceae. In this study, we examined: (1) the use of row covers for early season midge control, (2) the efficacy and dosage of several OMRI-certified insecticides in controlling swede midge late in the season, and (3) examine the use of entomopathogenic nematodes and Bt var. israelensis as a soil drench to control the midge in the soil. Most of the larger trials were conducted at the Intervale Community Farm (ICF), in Burlington, VT. Early season row covers have the potential to be effective and are widely used elsewhere, but midge populations were too low in the field for the row covers so the data did not demonstrate that row covers significantly protected the broccoli plants. We tested several OMRI-listed insecticides to determine swede midge damage. Among them, Azadirachtin was the only insecticide that reduced the incidence of swede midge damage compared to the control plots (25% damage vs. 50%), but the results were not significant. More research is urgently needed to identify methods effective in controlling this pest prior to widespread economic losses for Brassica growers.


Swede midge, Contarinia nasturtii, (Diptera: Cecidomyiidae) is an invasive insect pest that feeds on cruciferous crops and weeds. The midge was first discovered in North America in 2000, in Ontario, Canada, and can now be found in New York, New Jersey, Massachusetts, Vermont, and Connecticut (Hallett and Heal 2001; CFIA 2006, 2008; Kikkert 2006). Swede midge larvae feed on the leaf folds, petioles, and in developing florets of Brassica crops, causing swelling or distortion of the tissues. Midge damage can prevent broccoli and cauliflower head formation, resulting in crops that are unmarketable (Hallett 2007). In 2000, Swede midge damage in several Brassica cultivars resulted in a yield loss of 85% (Hallett and Heal 2001). There have also been considerable losses within Vermont on one of our participating farms, with some Brassica fields experiencing over 95% damage levels. Within the United States, cultivars of Brassica oleraceae L. (kale, cabbage, broccoli, brussel sprouts) are grown on over 120,000 ha, yielding an economic value of $1.23 billion (USDA Economic Research Service 2007). Recent models predict that Swede midge can potentially colonize all of the Northeastern US, the Great Lakes States, south to Colorado and west to Washington State (Olfert et al. 2006, Mika et al. 2008). Therefore, it is widely expected that swede midge can cause serious losses.

The newness of the swede midge invasion has led to recommendations of prophylactic sprays on a calendar basis (Fraser et al. 2005). Given that preventative sprays will definitely lead to increased levels of non-target impacts and groundwater pollution, more sustainable pest management approaches need to be developed that may also be appropriate for organic agriculture. Research is urgently needed to identify management protocols appropriate for small-scale and organic growers to prevent serious economic losses. The current recommendations of systemic insecticidal treatments and long and widely-spaced crop rotation (Fraser et al. 2005), may be difficult to implement on their farms. Also, OMRI-certified pesticides have not yet been tested yet in the field (Wu et al. 2006).
Swede midge could have a major impact on organic vegetable production within Vermont and the Northeast, where vegetable production within Vermont is dominated by small diversified organic farms. Swede midge was first found in Vermont at ICF in 2006, and midge damage has steadily risen since then in the Intervale.

Project Objectives:

valuate the use of row covers for early season crop protection
This goal was accomplished.

2) Conduct an organic insecticide trial for swede midge control
This goal was accomplished.

3) Evaluate the field efficacy and application dosage of entomopathogenic nematodes for swede midge control
This goal was accomplished. Due to the flooding at the Intervale Community Farm (ICF) by Hurricane Irene, we were unable to perform a large replicated trial because not all Vermont farms have damaging populations of the midge. We found one farm had midge damage outside of the Intervale (Burlington, VT) in 2011. Wendy Ordway, who owns Farmstand at the Cobble had severe swede midge damage in 2011, but her Brassica acreage was very small. We conducted a field trial in the spring of 2012 to evaluate the effects of nematodes on midge emergence.


Click linked name(s) to expand


Materials and methods:

1) Evaluate the use of row covers for early season crop protection (season 1 and 2)
We evaluated the use of row covers for early season crop protection. In particular, we examined if midge damage differed in broccoli covered for 3 weeks, 4 weeks, and 5 weeks. Andy Jones, the farmer partner, was concerned about high losses, so we used the 3 week row cover as the control. The broccoli was planted on 4/16 and then on 4/27, in two rows within 6 ft beds, planted 18 inches apart.. The control was a row cover left on for 3 weeks from 4/23 to May 18. The 4 week and 5 week row cover treatment were then removed May 22, and May 29. Although this goal was accomplished, midge populations were not particularly high in the spring of 2012. Due to the general low damage levels of the midge early in the spring (April to June), we did not find that floating row covers significantly affected the marketability of the broccoli crop. We assessed the damage on June 11, using the broccoli damage rating system from Hallett et al.(2009). A total of 5 plants were sampled from each treatment in each block (total 90 plants).

Traps set on May 18 (1 at intervale community farm next to row cover treatments). We hung from stakes with the bottom of the trap set ca.30-40 cm above the soil. The trap placed within the field must be orientated in such a way so that the pheromone plume is dispersed down the row. This will maximize its effectiveness. Because the traps were effective for approximately 4 weeks, we replaced the sticky cards replaced every 2-3 weeks throughout the season. The sticky cards were replaced on 6/11/12, 6/26/12, 7/9/12, and 7/23/12.

2) Conduct an organic insecticide trial for swede midge control
We examined the efficacy of commercial formulations of Entrust, Aza-Direct, Pyganic, and Surround in reducing swede midge damage.

The study was conducted as random block design, replicated over 6 blocks. Each treatment within the block was 4.88 m x 4.88 m. A 3 m buffer plot separated all treatment plots, so the experimental plots were distributed across the field like a checkerboard. We used a 4 gallon RL Pro Diaphragm pump backpack sprayer will a hollow cone nozzle to apply commercial recommendations of the four insecticides. Pyganic – 15 mL, Entrust 3.75 mL, Aza-direct (71098-1-10163) 30 mL, Surround WP 1.5 cups to the plots (totaling 0.0059 acres). The pesticide applications occurred on August 20th and 29th 2012.

Because the severity of plant damage could change over time, we assessed the midge damage by marking 5 individual plants and sampling them repeatedly. Plant damage was scored using the damage index based upon Hallett et al. (2009), where 0= no damage, 1= mild twisting of stem or leaves and/or slight swelling of the petioles, slight corky scaring, 2= severe twisting of the stem and/or crumpling of leaves and/or swelling of florets, and 3=death of apical meristem and/or multiple compensatory stems.
5 specific plants within each treatment and marking the damage of those same five plants each week. We also randomly chose three plants per plot and evaluated damage on 9/11/12 and 9/18/12. We also evaluated damage on 5 plants randomly selected within the plot. These damage assessments were taken on 9/5/12, 9/11/12, and 9/18/12

3) Evaluate the field efficacy and application dosage of entomopathogenic nematodes for swede midge control
We applied the predatory nematode Heterorhabditis bacteriophora to a plot (~300 sq ft.) at the Farmstand at the Cobble (Hinesburg, VT) that had severe midge damage the previous season in 2011. Due to the small size of the field and lack of other fields that were known to be previously infested, we were unable to replicate the nematode application over multiple replicate blocks. The rows within the plot were downward sloping, which would minimize the amount of water moving between treatments. We applied the predatory nematodes on the downward side of half of the plot (~150 sq. ft.).

We purchased the predatory nematodes from Arbico Organics. We used 1/4 of the container mixed in 10 gallons of water, based upon the supplier’s recommendations. The mixture was placed into 1 gallon water cans with rose attachments (shower head-like) and spread 5 gallons over 3 short rows (50 ft long). The application occurred at dusk to avoid direct sunlight and hot temperatures on June 11, 2012.

In order to determine if predator nematodes reduced swede midge emergence, we placed 6 soil emergence traps in the field on June 12, 2012. Three traps were randomly placed in the control areas and three traps were placed in the treated area. We collected the traps approximately every other week: 6/18/12, 6/26/12, 7/9/12, and 7/23/12. The collected arthropods were placed into a glass bottle with 70% ethanol.

We identified the specimens to the family level, looking for the Cecidomyiid flies. We analyzed the whether the nematode treatment application influenced the number of cecidomyiid flies caught within the emergence traps using a Generalized Linear Model.

Research results and discussion:

The most important message is that swede midge is an exceedingly difficult pest to control, especially late in the season. We recommend that more basic research is needed to identify strategies that reduce midge oviposition on Brassica crops. Row covers will likely help in reducing damage early in the season, but we have not identified any management strategies for plant protection from July onwards, when damage can be most severe.

Due to the low midge populations early in the season, we could not determine the efficacy of row covers in protecting broccoli plants against the midge. While early season broccoli can currently escape midge damage because populations are low. Midge populations appear to peak in July, however the densities of trapped midges could have been dependent upon the phenology of the field plants. The midge damage late in the season is quite severe, and we found that 40% of the plants were damaged. None of the OMRI-listed pesticides proved to be very effective in that they were not significantly better than taking no action.

Among the pesticides, Azadirachtin was the most effective and appeared to provide the most protection among the OMRI-listed pesticides. Our applications were targeted at the plants, but we may have achieved greater control if the material had been applied as a soil drench. Aza-direct is derived from the neem tree, Azadirachtin indica, and it can repel insects and have systemic activity upon root application (Gowan Company, Yuma, AZ). Neem is known to be a feeding deterrent and growth regulator of insects (Saha et al. 2011). Soil drenches of neem seed extracts can be taken up systemically by the plant and confer resistance for at least 25 days (Gill and Lewis 1971). Furthermore, ground neem seeds applied as a soil amendment with transplants can act systemically to protect plants from insect herbivory for at least 15 days (Osman and Port 1990).

1) Evaluate the use of row covers for early season crop protection (season 1 and 2)
We did not find any damaged plants in this study evaluating the use of a floating row cover to protect broccoli from swede midge damage. Therefore, it was difficult to determine whether the lack of difference between our treatments was due to an absence of an active midge population early in the spring.

We found that midge populations appeared to be higher in the Intervale than the outlying farms (Fig.1). Although this difference was not significant using a Repeated Measures ANOVA, we still believe that the differences appear to be biologically significant. The figure below shows how low early season populations are. This data is from 2012, the year following Hurricane Irene.

ICF – Intervale Community Farm, Burlington, VT
FOTC – Farmstand at the Cobble, Hinesburg, VT
CGF – Common Ground Student Educational Farm, South Burlington, VT
Arethusa Farm, Burlington, VT

2) Conduct an organic insecticide trial for swede midge control
Swede midge caused significant damage within the experimental plots. A total of 40.2% of the assayed plants were significantly damaged and commercially unmarketable. We found that midge damage did not statistically differ among the treatments (Fig. 2). In other words, the insecticide sprays did not reduce the frequency of midge damage below the control. Among the pesticides tested, the only pesticide that appeared to show some protection against the midge was Azadirachtin.

3) Evaluate the field efficacy and application dosage of entomopathogenic nematodes for swede midge control
Nematodes did not significantly influence the number of emerging cecidomyiid flies (?2 = 0.47, P = NS).

Research conclusions:

Due to increase in midge damage in Northern Vermont and our activities, we have given outreach talks to over 200 growers, and raised the awareness around swede midge invasion within the state. From this project and our continued efforts to identify effective strategies for controlling the midge, we are identifying what works and what may hopefully be effective in the field. Given the lack of pest management tools, more research is urgently needed to identify strategies for controlling midge populations.

Participation Summary

Education & Outreach Activities and Participation Summary

Participation Summary

Education/outreach description:

One of the goals of this research was to identify new methods that are available to organic growers to help control swede midge. Other than row cover, most of the other strategies were not effective. Therefore, our outreach has mostly focused on raising awareness about the devastating damage that swede midge can cause. Midge damage seems to be increasing, and last year vegetable farmers in the Northern Finger Lakes region in NY saw 100% damage to Brassica crops. In Vermont, Dr. Chen has been in touch with more growers that are now seeing 50% damage on their Brassica crops.

NOFA-VT winter conference- February 2012
Twilight meeting at the Intervale Community Farm – August 2012

NOFA-VT winter conference – February 2013
Vermont Vegetable Berry Growers Association – Jan 28, 2013
NOFA-VT summer workshop – August 2013

Project Outcomes

Project outcomes:

Not Applicable

Farmer Adoption

Although we have not been successful in identifying methods that are effective in controlling swede midge, farmers have been appreciative of our efforts to both publicize the pending invasion of swede midge and our continuing efforts to identify methods to control the midge. Some growers are also starting to feel the economic impact of the midge.
Farmer feedback has use of row covers is limited mostly based on weather, and so there can be issues with Brassica quality if the covers are left on for too long.

Assessment of Project Approach and Areas of Further Study:

Areas needing additional study

There is an urgent need for more resources and personnel to develop alternative control options. The midge’s biology makes it particularly difficult to control. First, as a gall-forming pest, larvae are largely protected from foliar insecticides by plant tissue (Frey et al. 2004, Hallett et al. 2009). Second, the midge has short life cycle (15 days), resulting in multiple overlapping generations and the need for constant plant protection throughout the season (Hallett et al. 2007). On farms that practice sequential cropping, midge populations are able to rapidly build up, leading to devastating losses later in the season (Shelton and Hallett, pers. obs.). Third, the most effective strategy of long and widely-spaced (> 1 km) crop rotations is difficult to achieve on small vegetable farms within the Northeast. Even large growers will find such wide rotations to be challenging. Fourth, the midge is quite small and damage is difficult to identify until it is too late (Swede Midge Information Center for the US 2010). Biological control does not appear to be a viable option as the natural enemies in North America are generally ineffective (Corlay et al. 2007). In Europe where the midge is thought to originate, parasitism of swede midge is around 15%, too low to control the midge(Abram et al. 2012). Finally, all foliar insecticides, conventional and OMRI-listed become ineffective under high midge densities (Hallett et al. 2009)

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.