Organic management of the invasive swede midge (Diptera: Cecidomyiidae) on Brassica vegetables: multiple dead ends necessitate novel approaches

Abstract

The distinctive biology and ecology of the invasive swede midge, Contarinia nasturtii Kieffer (Diptera: Cecidomyiidae), render organic management in North America particularly challenging, necessitating the search for novel out-of-the box approaches to control it. Native to Eurasia, this pest was first confirmed in North America in 2000 and has since spread to several states and provinces in the United States and Canada. As a galling midge, swede midge feeding causes distorted growth in Brassica (Brassicales: Brassicaceae) vegetables, resulting in major losses of marketable produce. While conventional growers typically use systemic insecticides for plant protection early in the season, equivalent approaches for organic production do not exist. The lack of effective organic management approaches for swede midge has resulted in devastating losses for heading Brassica vegetables, which are the most sensitive to midge feeding. Here, we review over 2 decades of research focused on developing organic approaches to manage swede midge. To encourage more rapid progress on swede midge management, we believe that it is important to review the widest array of work, including recent publications as well as our unpublished research. We conclude by highlighting the most promising strategies that should be utilized on farms and explored further for organic management of swede midge in vegetable crops.

Swede midge, Contarinia nasturtii Kieffer (Diptera: Cecidomyiidae), is an invasive fly native to Eurasia that attacks Brassica oleracea L. (Brassicales: Brassicaceae; broccoli, cabbage, cauliflower, collards, kale, etc.) and B. napus L. (canola) in the northeastern and midwestern United States and adjacent eastern Canada in its invasive range (Hallett 2007, Williams and Hallett 2018). Heading Brassica spp. are particularly sensitive to larval swede midge feeding (Hallett 2007). In the regions where it has become established (Michigan, New York, and Vermont in the United States and Ontario and Québec, Canada), swede midge has caused sporadic, but severe economic losses of organic broccoli (Hallett and Heal 2001, Chen et al. 2011; Hodgdon, Hoepting, and Chen, personal observation). Vegetable growers report that swede midge has caused average economic losses of $5,894 USD per acre per year (Hodgdon et al. 2022a), which exceeds the expected profitability margins for broccoli and cabbage (Wiswall 2012). For conventional Brassica production, management recommendations rely heavily on systemically translocated insecticides from seedling to heading crop stages (Chen et al. 2011, Hallett and Sears 2013). Despite over 10 years of focused research since the last review on swede midge management (Chen et al. 2011), organic vegetable growers still lack economically sustainable and practical tools for managing the midge. Given that the midge is now found in 15 US states and 5 Canadian provinces, further research and development is needed to advance alternatives to synthetic insecticides for swede midge management. We recently found that both organic and conventional growers (n = 112) would be interested in spending more on alternative management tactics to manage swede midge without insecticides (Hodgdon et al. 2022a).

The midge was first discovered in North America in 2000 in southern Ontario, Canada, and has since spread into New York, Vermont, New Jersey, Massachusetts, Connecticut, and Ohio (Hallett and Heal 2001, Kikkert et al. 2006, Chen et al. 2011). Although more widespread major losses are currently limited to New York, Vermont, and Michigan in the United States, the midge has recently invaded New Hampshire, Maine, Minnesota, Illinois, and Wisconsin (Phillips 2015, Hoepting and Kikkert 2017, Estes 2018). Swede midge is difficult to detect until it has become established and has caused serious economic losses (Chen et al. 2011, Hodgdon et al. 2017). Climate-based models predict that swede midge could colonize many Brassica-producing regions in the United States (Olfert et al. 2006, Mika et al. 2008), and there is concern for the swede midge’s potential to invade irrigated agroecosystems in California (Chen, personal observation). In Canada, the midge has the potential to spread west as far as British Columbia (Olfert et al. 2006, Mika et al. 2008). Although the swede midge is an economic pest in provinces in eastern Canada where its presence has been confirmed, swede midge has not been detected in Canada west of Ontario in 15 years (Vankosky et al. 2023). It remains a threat, however, for canola production in western Canada.

Due to the small size of the larvae and their feeding location deep in the apical bud, swede midge cannot be included in traditional scouting programs for other Brassica pests. Cecidomyiid larvae secrete saliva for extraoral digestion of plant material, which distorts plant growth, scars tissues, and prevents proper head formation (Fig. 1; Hallett 2007, Stuart et al. 2012). Plants need to be dissected to find the tiny larvae (~1 mm; Fig. 2), and plant damage only becomes apparent a week after the larvae have already left the plant (Stratton et al. 2018). We found that a single swede midge larva within the growing point can lead to an unmarketable cauliflower head (Stratton et al. 2018). In comparison to foliar-feeding caterpillars and other Brassica pests, swede midge feeding uniquely impacts meristematic growth in a way that prevents head formation.

Furthermore, swede midge adults are very small and difficult to detect. Rather than visual scouting of plants, swede midge must be monitored using pheromone traps. Monitoring using pheromone traps can be time consuming, requires specialized knowledge for accurate identification, and purchase of costly lures. The midge is active from May through October or November and has multiple overlapping generations, so there is need for constant monitoring and plant protection throughout the growing season. Swede midge nearly continuously emerge over 2–3 prolonged overlapping generations (Liu 2019). When growers practice sequential cropping on a continuous land base, swede midge populations rapidly build up, leading to devastating losses later in the season (Hallett et al. 2009b). Because all swede midge life stages are challenging to detect visually, swede midge damage is often misdiagnosed (Hallett and Heal 2001). Therefore, the typical scout-and-spray integrated pest management (IPM) strategies commonly used for vegetable crops are not well suited to the swede midge life cycle and biology.

For conventional growers, swede midge is most effectively managed with multiple insecticide applications (Wu et al. 2006, Chen et al. 2011, Seaman et al. 2014, Hoepting and Kikkert 2017). The recommendations for conventional growers are to use a systemic neonicotinoid treatment as a tray drench to protect seedlings over the first 3–5 weeks, followed by applications of products containing active ingredients with systemic or contact activity. Labeled products in the United States include imidacloprid and acetamiprid (neonicotinoid chemical class), cyantraniliprole (diamide chemical class), spirotetramat (tetramic acid chemical class), and pyrethroids such as active ingredient lambda-cyhalothrin (Hallett et al. 2009a, Chen et al. 2011). Swede midge outbreaks have increased insecticide use, as some growers have resorted to weekly sprays to manage them (Hallett and Sears 2013). However, even these intensive spray regimes can be ineffective with large midge populations (Hallett et al. 2009a). Additionally, intensive insecticide use may increase the risk of swede midge developing resistance to insecticides, indicating a need for diversified IPM approaches (Chen et al. 2011).

Organic growers lack comparable insecticide options, since materials listed by the Organic Materials Review Institute (OMRI-listed) do not have comparable systemic or translaminar activity. The OMRI-listed insecticides are largely ineffective for managing swede midge (Seaman et al. 2013). Consequently, extension professionals and crop advisors typically do not recommend them for swede midge management (Hodgdon et al. 2017, Martinez et al. 2020). As a result, organic vegetable growers search for noninsecticidal management strategies.

The challenge of finding effective organic management strategies for swede midge has motivated us to explore a range of novel approaches for this invasive pest, including exogenous plant hormones, plant essential oils, biological control, pheromone mating disruption, exclusion fencing and netting, ground barriers, intercropping, and more. To guide future swede midge research and IPM program development, we review the results of those studies here, including promising and unpromising tactics. Since novel approaches are needed to manage swede midge effectively, our review of nonsignificant results can help guide research forward on promising leads. We summarize over 2 decades of research on swede midge management in an open-access format readily accessible by agricultural service providers and vegetable producers, along with suggestions for further research and development for this challenging invasive pest.

Cultural Control

Crop Susceptibility

Although B. oleracea L. is a single species, the cultivated forms include varieties that are highly diverse in physical and chemical characteristics. Crop characteristics, such as growth habit, biochemical profiles, and marketable plant parts, play important roles in susceptibility to swede midge and potential for economic losses. The species includes varieties that produce a single marketable crown (e.g., broccoli [var. italica] and cauliflower [var. botrytis]), leafy vegetation that either forms a single head (e.g., cabbage [ssp. capitata]) or multiple heads (e.g., Brussels sprouts [ssp. Gemmifera]), or individual leaves that can be harvested more than once in a given season (e.g., kale [var. acephala]) (Mohd Saad et al. 2021). As a specialist of B. oleracea L., swede midge can consume all varieties of its host but may prefer some subspecies or varieties over others (Thompson 1988, Wang et al. 2017).

Broccoli cultivars vary in a broad range of biologically active phytochemicals, including glucosinolates, tocopherols, and carotenoids (Renaud et al. 2014), so variations in their concentrations could influence the resistance of a cultivar to disruptions caused by swede midge feeding. Even minor variation in phytochemical or physical characteristics of a crop can influence its susceptibility to herbivores (Snyder et al. 2020). Gall midges are intricately linked with their host plant physiology and have direct effects on plant growth and development (Carneiro et al. 2017, Miller and Raman 2019). In a test of crop susceptibility among 8 broccoli cultivars in a Vermont field trial, all cultivars experienced >80% loss of marketable heads (Jones et al. 2018). In an Ontario field experiment, we assessed swede midge damage to broccoli cultivars using a rating scale that ranks the increasing severity of swede midge damage symptoms to the plants, including scarring, swollen petioles, and lack of head formation. The varieties ‘Triathlon’ and ‘Regal’ experienced reduced and delayed swede midge damage symptoms and severity compared to other broccoli varieties we evaluated (Hallett 2007). These cultivars, however, are no longer available commercially (E. A. Hodgdon, personal observation). Of note, we identified ‘Paragon’ as one of the most susceptible broccoli cultivars in the experiment, a hybrid that has Chinese broccoli (B. oleracea var. alboglabra L.) as a parent. We hypothesized that higher sucrose levels in the stems of Chinese broccoli may have contributed to the increased swede midge preference, which should be explored further when testing the susceptibility of newer commercially available broccoli cultivars to swede midge.

Along with broccoli, ‘Red Russian’ kale (B. napus L.) incurs high levels of swede midge damage. In a monitoring project on small-scale organic farms in New York, we compared swede midge damage incidence and severity (using a similar damage rating system to Hallett (2007)) within 28 Brassica plantings that included several different plant types and cultivars (Table 1; Hoepting 2018a, 2018b, 2020). Repeatedly, broccoli and ‘Red Russian’ kale received the most damage of all other Brassica plant types (Hoepting 2018a, 2018b, 2020). Often, swede midge would infest these 2 crops when other crop types were available, only infesting others when broccoli and ‘Red Russian’ kale were not present. In these studies, broccoli and ‘Red Russian’ kale sustained swede midge damage in the field over a prolonged period, as the plants compensated with new shoots and growing points where the midges could lay eggs. Swede midge’s preference for ‘Red Russian’ kale was also observed in a Vermont field experiment, where ‘Red Russian’ experienced significantly more damage than ‘Lacinato’, ‘Red Curly’, and ‘Vates’, with the 3 kale cultivars also experiencing little damage when ‘Red Russian’ was not present in the field (Izzo and Lewins 2022a, 2022b).

Compared to other crops, Brussels sprouts and cabbage experience lower swede midge damage ratings and economic loss (Hallett 2007, Hoepting 2018a, 2018b, 2020). Once the cabbage is heading and the growing points are covered by older leaves, it becomes unsuitable for swede midge (Barnes 1946, Hallett 2007). Savoyed cabbage (B. oleracea L. var. subauda), Napa, or Chinese cabbage (B. rapa L. var. pekinensis), and bok choy (B. rapa L. ssp. chinensis) are either less preferred by swede midge or more tolerant of midge feeding. These cabbages were consistently undamaged in fields with high swede midge pressure, even when other Brassica spp. were not available (Hoepting 2018a, 2018b, 2020), and they may be produced in fields with large (>100 swede midge/trap/week) swede midge populations without withstanding economic damage.

In addition to Brassica vegetable crops, noncrop plants and weeds in the Brassicaceae family have differing susceptibility to swede midge. Several common brassicaceaeous weeds can foster low levels of swede midge in agroecosystems, presenting a challenge in utilizing crop rotation as a method for controlling swede midge. The common weeds field mustard (B. rapa L.), wild mustard (Sinapis arvensis L.), shepherd’s-purse (Capsella bursa-pastoris L.), and dog mustard (Erucastrum gallicum (Willd.) O.E. Schulz) are swede midge hosts (Hallett 2007, Chen et al. 2009). While native and weedy plants in the Brassicaceae family serve as hosts for swede midge, they may only play a small role in fostering swede midge populations in agroecosystems. Ultimately, wild Brassicaceae were found to host fewer swede midge larvae than domesticated Brassica crops, suggesting that swede midge prefers cultivated Brassica crops or that the crop plants can structurally support more larvae (Readshaw 1961, Chen et al. 2009).

Noncrop brassicaceous plant species that are either resistant, tolerant, or not preferred by swede midge may serve as sources of genetic material in crop breeding for swede midge resistance. We monitored 10 Brassica weed species and did not find evidence of swede midge damage on small-seeded false flax (Camelina macrocarpa Andrz. ex DC), wormseed mustard (Erysimum cheiranthoides L.), field pepper grass (Lepidium campestre (L.) R.Br.), wild mustard, or flixweed (Descurainia sophia (L.) Webb ex Prantl) (Hallett 2007). These plants, and other wild Brassicaceae, could be studied further for traits to be used in Brassica vegetable crop breeding projects for swede midge tolerance.

Crop susceptibility to swede midge should be considered with crop rotation as a management strategy (Hoepting and Vande Brake 2020). Since new Brassica crop cultivars are released every year, continued trialing will assist growers in selecting less preferred cultivars for swede midge management. Currently, the physical and chemical traits of crop and wild plants correlated with reduced swede midge infestation are untested and could be identified in future experiments. Additionally, the ability of Brassica cover crops such as radish (Raphinus sativus L. Domin) and field mustard to contribute to swede midge populations has not been studied and is an area for future research.

Crop Rotation

The life history of swede midge can undermine cultural management efforts. Swede midge has a broad host range, an extended facultative diapause, and a short generation time, which allows swede midge populations to rapidly increase (Hallett and Heal 2001, Hallett et al. 2009b, Des Marteaux et al. 2015). Since heading Brassica spp. are most susceptible to swede midge infestations prior to heading or flower development, cultural management efforts should focus on reducing the exposure of susceptible developmental stages of host plants to peaks in adult swede midge populations. To disrupt the swede midge life cycle, preliminary crop rotation recommendations advised growers to rotate away from Brassica crops by at least 914 m (3,000 ft) for a minimum of 3 years (Hodgdon et. al. 2017, Hoepting and Vande Brake 2020). This conservative recommendation assumes that swede midge are weak fliers and that pupae can persist in soil for at least 2 years (Des Marteaux et al. 2015). However, implementing spatiotemporal rotations of that magnitude can be impractical for small farms if a large proportion of the Brassica production occurs over the entire season on a small land base (Hoepting 2018a, 2020, Hoepting and Vande Brake 2020).

Researchers have examined how manipulating spatiotemporal crop rotations could avoid damage from the spring emergence of swede midge. In New York, growers can expect fewer swede midge and less severe swede midge damage when they return to a field previously cropped in Brassica spp. if they wait until the end of July (i.e., after the spring and early summer emergence of swede midge have subsided) to plant fall Brassica spp. (Hoepting 2018a, 2020, Hoepting and Vande Brake 2020). In Ontario, broccoli transplanted late in the growing season (late July/early August) showed significant reductions in swede midge feeding damage compared to broccoli planted in mid- to late-May and early-June, the standard times for planting (Evans 2017). Conversely, cabbage transplants planted early showed lower damage levels than later plantings (Corlay and Boivin 2008, Evans 2017). These results may be explained by differences in the developmental characteristics of these 2 crops and the presence and size of swede midge populations at the different planting dates. Whereas broccoli is susceptible to midge damage until early crown formation, cabbage is susceptible for a shorter period after transplanting.

Our recent data indicates that a ~152 m (500 ft) distance between a swede midge-infested site and a Brassica crop was far enough to prevent swede midge from dispersing only when the fields were secluded and separated by a significant barrier such as a wooded area, hedgerows, or fences (Hoepting 2018a, 2020). Cultural practices such as temporal avoidance through planting date manipulations and strategically selecting isolated fields should be considered as part of an IPM program. By understanding crop phenology, susceptibility, and landscape and population dynamics, growers can manipulate planting dates and crop rotation to avoid swede midge damage.

Interplanting

In general, increasing plant diversity in the field through interplanting (an intentional mixture of crops or noncrop plants within the same field) reduces crop damage by generalist insect herbivores more than specialist herbivores (Koricheva and Hayes 2018). Diversifying crop fields can provide a physical and/or chemical barrier that makes it difficult for pest insects to find and damage their crop hosts (Finch and Collier 2000). Nonhost plants introduce physical diversity into cropping systems, but each plant species also produces a unique blend of volatile organic compounds (odors) that vary in biological activity (Vivaldo et al. 2017). Insect herbivores respond to the volatile organic compounds emitted by plants, which may be attractive or repellent to ovipositing insects (Conchou et al. 2019). There is considerable potential for integrating plant diversity into crop fields, especially if we improve our understanding of the impacts of nonhost plants on insect behavior (Poveda and Kessler 2012).

Interplanting can contribute to pest management in many ways, through deterring pests, preventing pests from finding host plants, or directing pests toward noncrop plants. As Brassica specialists, swede midge is responsive to Brassica signals and may be repelled by nonhost visual and scent cues. Interplanting broccoli within surrounding rows of sweet alyssum (Lobularia maritima (L.) Desv.) or love-in-a-mist (Nigella damascena L.) did not reduce midge damage in a field study (C.A. Stratton and E.A. Hodgdon, personal observation), although they showed promise in a preliminary greenhouse screening (Brion 2015). In Ontario, we tested whether nonhost crops could serve as a barrier to swede midge when planted surrounding a Brassica crop (Evans and Hallett, unpublished). The nonhost barriers comprised of onions (Allium cepa L.), corn (Zea mays L.), or peas (Pisum sativum L.). None of the nonhost plants tested significantly decreased swede midge damage. One possible explanation for the failure of interplanting to influence insect behavior is that habituation may occur, where the continued presence of odor causes insects to become less sensitive to repellent odors (Glanzman 2011, Twick et al. 2014). Because interplanting has shown little promise for swede midge management in the field and would add complexity to cropping system management on farms, further research in this area is not warranted until new candidate interplants and planting schemes are identified.

Trap Cropping

Trap crops have been used to manage pests in agricultural systems for decades. When effective, trap crops can attract pest insects away from economic crops and/or concentrate them in an area where they can be managed more easily (Sharma et al. 2019). However, by definition, trap crops are only effective in systems where pests exhibit variable preference within their range of host plants (Shelton and Badenes-Perez 2006). To test the efficacy of trap cropping for swede midge management in broccoli, large cage and open-field trials were conducted in New York over 2 yr (Hoepting 2018b, 2020). Treatments included monoculture broccoli and ‘Red Russian’ kale (more susceptible), Chinese cabbage and bok choy (less susceptible), and mixed 2 species plantings of broccoli with either ‘Red Russian’ kale, Chinese cabbage, or bok choy. In each of the 2 large cage trials conducted with high swede midge pressure, compared to monoculture broccoli, broccoli grown beside the kale (trap crop) had a 43–47% lower incidence of swede midge infestation, 30–75% more marketable heads, and fewer heads with severe damage (Hoepting 2018b, 2020). In these studies, there was no difference in incidence and damage severity between kale monoculture and kale grown beside broccoli. Although ‘Red Russian’ kale monoculture had significantly higher (56%) swede midge damage incidence than broccoli monoculture (Hoepting 2018a), midge damage did not differ between monoculture broccoli field plots and broccoli grown beside ‘Red Russian’ kale (Hoepting 2018b). While using a trap crop to manage swede midge has potential, more research is needed with different crop/trap crop combinations, planting configurations, and trap crop management to prevent swede midge population increases.

Physical Control

Exclusion Netting and Fencing

Exclusion barriers, such as screens, netting, and fencing, can effectively suppress pest infestation severity by reducing the number of insects entering planted fields and delaying colonization. Since swede midge host plants are more susceptible during early stages of development, delayed colonization may allow host plants to mature before being exposed, when they are less prone to damage (Vernon and Mackenzie 1998, Hallett 2007). Our work has tested both insect exclusion materials over hoops and as vertical fencing to manage swede midge.

We evaluated polyamide insect exclusion netting (25 g/m², ProtekNet with 0.35 mm holes) for its potential to exclude swede midge from Brassica plantings in 9 small plot on-farm field trials and demonstrations in New York (Hoepting 2018a, 2018b, 2020). Insect netting over hoops resulted in <1% unmarketable plants, whereas the unprotected treatments had 50–89% unmarketable broccoli or kohlrabi (B. oleracea L. var. gongylodes) heads (Hoepting 2018a, 2018b, 2020). The netting was most effective in areas where the Brassica plants were planted into ground that had been rotated out of Brassica spp. for at least 2 years. When spring broccoli was planted in a field that had been infested the previous fall, netting was ineffective and crop losses were severe (Hoepting 2018a). In this case, the exclusion netting effectively trapped emerging midges. Thus, for netting to be effective, it should not be used on ground that has had midge-infested Brassica spp. for at least 2 years prior to planting. It is also important to note that netting alters the growing environment for the crop, which can delay or advance maturity. Additionally, netting must be secured well and repaired when tears occur to effectively exclude midges.

The cost of materials and labor can be a significant barrier to using insect exclusion netting. We have found that growers are dissatisfied with using netting because of the cost and logistical challenges (Hodgdon et al. 2022a). When trialed in 2015, the netting system with black plastic mulch cost around $17,705 USD per acre (Table 2; Hoepting 2018a). Netting to control swede midge on broccoli is only feasible when farmers expect > 50% crop loss and broccoli prices exceeded $6.60 USD/kg or $3 USD/lb (Hoepting 2018b). However, it can be economically beneficial for reducing overall pest damage; netting also excludes flea beetle (Phyllotreta spp.) and imported cabbage worm (Pieris rapae L.). Careful management to avoid tearing and allow reuse of the netting improves the economic feasibility of its use.

We evaluated 2 heights of vertical “no-see-um” mesh fencing (85 cm [2.7 ft] and 150 cm tall [4.9 ft]) as barriers to swede midge in Ontario field trials. The fencing was installed around plots containing broccoli, which were evaluated for swede midge damage. Shorter fencing (85 cm) did not reduce swede midge damage in the broccoli. In plots surrounded by the taller exclusion fencing (150 cm), the onset of damage was delayed, reducing damage severity and doubling the number of marketable plants in 2 of the 3 study years (Evans 2017). Vertical sampling captured midge adults at heights above the 150 cm upper boundary of the insect exclusion fencing. It was not clear whether captures at these heights were the result of swede midge flying under their own power or being carried by turbulent wind gusts (Evans 2017). The high labor and material costs of insect exclusion fencing (Meadow and Johansen 2005) and swede midge captures at heights up to 240 cm (8 ft) above the soil surface (Evans 2017) suggest that this physical exclusion barrier is not a viable management tactic. However, reduced damage in fenced plots suggests that if combined with a temporal avoidance planting date strategy, insect exclusion fencing could offer promising results.

Ground Barriers

Prior to spring emergence, all swede midge are in the pupal stage, presenting an opportunity to focus management efforts on a sessile, uniform population (Chen and Shelton 2007). Despite laboratory results showing that swede midge emergence can be reduced by burying pupae in soil, deep tillage is not recommended as it promotes emergence (Chen and Shelton 2007, Allen et al. 2022). Ground barriers, however, may be applied over the soil to kill larvae and pupae by solarization, prevent larval entry into soil, and inhibit adult emergence.

In an Ontario study, we laid a 6-mm-thick clear polyethylene sheet over infested soil following an early-season broccoli harvest and left it in place until the following spring (Evans 2017). Given that the majority of swede midge pupate in the top 1 cm of the soil (Chen and Shelton 2007), covering infested soil with clear plastic sheeting during July and August could maximize the solarization effect of summer peak temperatures in southern Ontario (Government of Canada 2016). Although solarization reduced the number of spring-emerging adults (~50% of control), the reduction was not statistically significant (Evans 2017). The lack of significance for this study may have been due to a low midge emergence at this time, which minimized the differences between treatments.

In addition to solarization, we evaluated ground barriers for their ability to prevent swede midge larvae from entering the soil for pupation. In a laboratory study, most larvae (70–90%) did not travel more than 1 cm on top of different ground covers, suggesting that growing Brassica crops in ground barriers could inhibit larvae from reaching soil (Lemay et al. unpublished data). In an artificially infested on-farm cage study, planting in black plastic significantly reduced total swede midge trap catches (37%), suggesting that the ground barrier interfered with larval ability to reach the soil, pupation, and/or emergence (Hoepting 2020).

In another field trial in Ontario, we evaluated a white spunbond polyethylene floating vegetable row cover as a physical barrier aimed at preventing swede midge larvae from entering the soil to pupate. Broccoli seedlings were planted through 5-cm slits in the material in late summer and the row cover was left in place through the winter. Though swede midge emergence the following spring was reduced by ~35% compared to uncovered bare soil treatments, consistent adult captures confirmed that the soil beneath the row cover did become infested (Evans 2017). Larvae may have entered through the planting holes or potentially through the pores in the fabric.

In New York, we used field cage studies to test whether different ground barriers could reduce swede midge emergence from artificially infested soil (Hoepting 2018b, 2020; unpublished data). Barriers including tarp, black biodegradable fabric, and landscaping fabric reduced swede midge emergence by 98–100% following a 30- to 46-day coverage period after soil infestation when compared to bare ground infestation. Landscape fabric with planting holes and plastic mulch did not effectively reduce swede midge emergence. Additionally, straw mulch was completely ineffective as a barrier to swede midge emergence (Hoepting 2018a). Our laboratory experiments confirm that midges emerged similarly among clear and black plastic mulch-covered and uncovered soil indicating that the mulches may have trapped adults, preventing them from entering the environment (E. Schoeppner, unpublished data).

Tarps or landscape fabric without planting holes appear most promising for reducing swede midge populations; however, more research is needed to determine the mechanisms behind the reduction in swede midge emergence. Current research is determining the length of time needed for ground barriers to effectively reduce swede midge emergence in naturally infested soil. The major downside to ground covers is that this practice requires producers to take land out of production during part of the growing season.

Biological Control

Parasitoid Natural Enemies

Surveys were conducted in Europe from 2008 to 2011 to identify parasitoids that could be released in North America as part of a classical biological control program for swede midge (Abram et al. 2012, 2013), and 5 species of swede midge larval parasitoids were identified during the surveys: Macroglenes chalybeus Haliday (Abram et al. 2012, 2013), Synopeas myles Walker (Buhl and Notton 2009, Abram et al. 2012), Sy. ventrale Westwood (Buhl and Notton 2009), Sy. osaces Walker (Abram et al. 2012), and Inostemma opacum Thomson (Abram et al. 2012b). Macroglenes chalybeus and Sy. myles were the most common and widely distributed species observed (Abram et al. 2012). However, neither species was recommended for importation and release in North America because both had relatively low parasitism rates of swede midge (<5% on average) and relatively broad host ranges of multiple cecidomyiid species (Abram et al. 2012).

Despite the decision not to introduce this species to North America, we discovered Sy. myles parasitizing swede midge in canola in Ontario and Québec in 2016. Since then, annual surveys of spring canola have revealed that Sy. myles is widely distributed throughout both provinces, and the mean parasitism rate of swede midge by Sy. myles ranged from 1% to 60% in Ontario and 0.2% to 71% in Québec (Gradish et al. in press), higher than what was observed in Europe (Abram et al. 2012). Because Sy. myles is established and actively parasitizing swede midge in canola, a conservation biological control approach aimed at maintaining and/or increasing Sy. myles populations and parasitism rates may improve the suppression of swede midge populations. To develop conservation biological control tactics for swede midge, we are conducting research on the phenology and behavior of Sy. myles, and its compatibility with current IPM tactics for swede midge in canola. To date, however, no studies have examined whether Sy. myles parasitizes swede midge in vegetable crops. Future research quantifying parasitism levels in vegetable crops is a logical next step.

Studies on the potential of sweet alyssum as a nectar-bearing supplemental plant for supporting Sy. myles populations in canola have also been conducted. Consuming floral nectar generally, and the nectar of sweet alyssum specifically, can increase the fecundity or longevity of parasitoid adults (Landis et al. 2000, Berndt and Wratten 2005, Lahiri et al. 2017, Aparicio et al. 2018, Jado et al. 2018, Chen et al. 2020), which can increase parasitism. Access to flowering sweet alyssum increased the longevity and parasitism rates of Sy. myles in lab experiments (Ferland 2020, McLennan 2021). Furthermore, swede midge females do not preferentially oviposit on sweet alyssum, and swede midge larvae do not develop on sweet alyssum (Brion 2015, McLennan 2021). Taken together, these results suggest that adding sweet alyssum plants to canola agroecosystems may help to maintain or increase Sy. myles populations. However, it is unclear if supplemental plantings lead to increased parasitism of swede midge by Sy. myles in the field. Information gained from canola studies may be used to inform research and development of management practices to increase Sy. myles populations in vegetable fields.

Entomopathogenic Nematodes

Entomopathogenic nematodes are parasitic roundworms that utilize soil-dwelling life stages of insects as hosts, and pest managers use inoculative and inundative releases of nematodes (Steinernema carpocapsae Weiser, St. feltiae Filipjev, and Heterorhabditis bacteriophora Poinar) on farms as part of biological control programs for many pests with soil-dwelling life stages. Laboratory experiments demonstrated that the nematode strains of H. bacteriophora, St. carpocapsae, and St. feltiae isolated from Ontario could readily infect all soil-dwelling life stages of swede midge (Corlay et al. 2007, Evans et al. 2015). We observed significant mortality of larvae, pupae, and cocoons resulting from infection with these entomopathogens. These strains were also able to successfully reproduce within swede midge larvae, suggesting there is potential for longer-term persistence of nematodes in the soil with swede midge hosts (Evans et al. 2015). While laboratory tests of nematodes have been promising, field trial results with soil-applied nematodes have been mixed. Nematodes did not significantly lower emergence rates of swede midge on a commercial vegetable farm in Ontario but did reduce midge emergence in 1 year at a research farm site (Evans et al. 2015). The trials, however, were conducted in very small-scale plots for 1 growing season and did not consider long-term effects of nematode populations. Multi-year studies using nematode strains adapted to persist in northern climates may be a logical next step for swede midge research. Specifically, New York strains selected for persistence have demonstrated prolonged pest management over several years, requiring longer-term studies (Shields 2015, Shields and Testa 2020).

Entomopathogenic Fungi and Bacteria

Applications of the entomopathogenic fungi and bacteria, such as Beauveria bassiana (Bals.-Criv.) Vuill., Burkholderia spp., and Chromobacterium subtsugae Martin et al., have been largely ineffective for swede midge management. For entomopathogenic fungi, such as Beauveria bassiana (Bals.-Criv.) Vuill., sufficient moisture is required for fungal spore germination and insect infection; hot and dry conditions may have limited the efficacy of Botanigard 22WP (Beauveria bassiana (Bals.-Criv.) Vuill.; Laverlam Corp., Butte, MT) in our Ontario field trials (Evans and Hallett 2016). Venerate XC (Burkholderia spp.; Marrone Bio Innovations, Davis, CA) and Grandevo WDG (Chromobacterium subtsugae Martin et al.; Marrone Bio Innovations, Davis, CA) did not increase marketable yields when trialed in New York by Seaman et al. (2013, 2014). In a separate field study, however, we observed reduced swede midge emergence following a soil drench treatment with Metarhizium brunneum Petch to a previously infested site in 1 yr of the study (Evans et al. 2015). Further work may be warranted for M. brunneum in larger scale experiments examining effects on crop damage in addition to midge emergence. Since there are many different strains of entomopathogenic fungi and bacteria held in repositories used for pest control research, further exploration of these materials for swede midge infectivity and practical management may be warranted. In the studies reviewed here, only a small number of species and strains have been tested, limited to commercially available strains.

Chemical Control

Organic Materials Review Institute-listed Insecticides

Swede midge management in organic agriculture is challenging due to the lack of OMRI-listed insecticide options. Contact with larvae in the growing point is difficult to achieve with foliar-applied insecticides that are not translocated. A limited number of botanical pesticides can act systemically, translocating active chemicals throughout the entire plant, which could help protect Brassica crops against swede midge. While some laboratory and greenhouse studies have shown that application of OMRI-listed products can result in reduced swede midge infestation, results from field studies have been unsuccessful and inconsistent across field trials in the United States and Canada (Table 3).

Spinosad, pyrethrin, azadirachtin, and kaolin clay products have demonstrated intermittent efficacy in reducing damage in field trials, particularly when overall swede midge pressure was low (Table 3; Seaman et al. 2014, 2015, Evans and Hallett 2016). Organic Materials Review Institute-listed insecticides have failed when swede midge populations are large (Seaman et al. 2013). It should be noted that not all OMRI-listed products tested include swede midge on their labels, which may restrict their use for this pest depending on local regulations.

Azadirachtin is a botanical, systemic pesticide derived from the neem tree (Azadirachta indica A. Juss.), and formulations derived from the neem seed have the highest levels of the active ingredient (Thoeming et al. 2006). In laboratory trials using potted plants, we found that azadirachtin soil drenches using 1.5, 3, or 4.5 ml/liter AZA-Direct (Gowan, Yuma AZ) and 1, 2, or 3 ml/liter AzaMax (Perry America Inc., Sacramento, CA) significantly reduced the density of swede midge larvae on cauliflower plants (K. Jacobs, unpublished data). The application of OMRI-listed formulations of azadirachtin to the growing point of broccoli plants resulted in significant reductions in the number of swede midge larvae in separate greenhouse studies. These reductions were attributed to both oviposition reduction and larvicidal properties (Wu et al. 2006, Evans and Hallett 2016). Drenches of azadirachtin have not yet been tested in the field but are worth exploring due to its systemic activity.

Due to the inconsistencies observed in field studies of OMRI-listed insecticides and the cost of these products, extension professionals and crop advisors typically do not recommend their use for swede midge management (Hodgdon et al. 2017, Martinez et al. 2020). While future research investigating the repellent effect of foliar-applied products may be warranted to understand how they reduce swede midge oviposition, the most pressing need is for new OMRI-listed products with translocating and larvicidal properties. Additional research directions could include testing adjuvants and application methods to more effectively apply insecticides deeper within the meristem in order to reach the larvae.

Plant Defense Elicitors

When plants are attacked, they increase the production of chemical defenses that cause them to be less preferred by subsequent insect herbivores (Karban and Baldwin 1997). Generally, chewing insect herbivores induce the jasmonic acid pathway, whereas fungi and bacteria induce the salicylic acid pathway (Walling 2000). Galling flies appear to escape detection by plants or prevent the induction of both the jasmonic and salicylic acid pathways during natural interactions (Tooker et al. 2008).

We tested foliar- and soil-applied elicitors to observe whether plants with induced defenses impacted swede midge oviposition or larval survival. Using whole broccoli plants in cage assays, we tested how soil applications of defense elicitors affected larval abundance using benzothiadiazole, chitosan, salicylic acid, or methyl jasmonate. We observed the lowest average larval density on broccoli treated with benzothiadiazole and the highest on broccoli treated with salicylic acid (R. Pilischer and R. Roman, unpublished data). Although broccoli plants drenched with benzothiadiazole had the lowest larval density, the average density was 24.45 ± 3.77 larvae/plant, which exceeds the single midge threshold for causing marketable damage (Stratton et al. 2018). Therefore, none of the treatments were effective as control measures.

Using laboratory behavioral assays, we tested if foliar-applied elicitors of salicylic and jasmonic acid pathways influenced female behavioral choices in a y-tube olfactometer. Plants were sprayed with benzothiadiazole, jasmonic acid, methyl jasmonate, or a water control solution just prior to use in the experiment. Mated females were significantly more likely to fly towards untreated plants or did not fly towards either of the plants when exposed to untreated versus benzothiadiazole- or methyl jasmonate-treated plants (P. Filho, unpublished data). Although results suggest that Brassica plants treated with plant defense elicitors are less attractive for oviposition to mated female swede midge, they have fallen short in the field. Swede midge damage did not differ between plants that were pretreated with foliar applications of jasmonic acid, salicylic acid, or water plus adjuvant 1 week prior to placement in an organic field setting (H. Eiseman, unpublished data). While repeat applications may be required to deter swede midge oviposition, plant defense elicitors do not appear to be promising for swede midge management overall.

Plant Essential Oils

As naturally occurring compounds, plant essential oils could be low-risk alternatives to insecticides for pest management as repellents, yet it remains unclear which plant odors are most effective at deterring specific pests (Regnault-Roger et al. 2012). Among the 20 plant odors tested in our laboratory and greenhouse assays, garlic (Allium sativum L.), eucalyptus lemon (Corymbia citriodora (Hook.) K.D. Hill & L.A.S. Johnson), spearmint (Metha spicata L.), and lemongrass (Cymbopogon citratus (DC.) Stapf) consistently disrupted swede midge host selection behavior and reduced larval density on host plants. The most effective essential oils were derived from species less related to Brassicaceae, with odors that were more chemically similar to Brassica compounds (Stratton et al. 2019).

Since some insect herbivores may mate only in the presence of their host plant, nonhost odors may interfere with female and male reproductive behavior (Conchou et al. 2019). To test how female mating behavior varies in the presence of nonhost essential oils, the frequency of calling (extension of pheromone gland and release of pheromones) was tested in the presence of host plant volatiles alone or with garlic or eucalyptus lemon essential oils in laboratory trials. While eucalyptus lemon reduced the frequency of calling by approximately 55%, garlic reduced calling by approximately 92% compared to the host plant alone (J. Samuel, unpublished data). Essential oils that influenced female behavior were also tested for their impacts on male behavior. In a laboratory study, males were given a choice between 1% dilutions of wintergreen essential oil (Gaultheria procumbens L.), lemongrass, wormwood (Artemisia absinthium L.), lily (Lilium auratum Lindl), garlic, thyme (Thymus vulgaris L.), eucalyptus lemon, cinnamon (Cinnamomum verum J. Presl), or oregano (Origanum vulgare L.) compared to water (control). The distribution of males flying toward non-Brassica essential oils did not differ from water, suggesting that male swede midge do not respond to nonhost odors (P. Judge, unpublished). Plant essential oils released at a high dosage may reduce swede midge mating and could be tested in a mating disruption system.

While essential oils are a lower cost alternative to insecticides (Nerio et al. 2010), further field testing is needed to develop practices that improve efficacy due to limitations including dissipation and degradation (Regnault-Roger et al. 2012). In field trials, we tested Garlic Barrier (Garlic Research Labs, Inc; Glendale, CA), a commercially available OMRI-listed product derived from garlic essential oil, along with lemongrass essential oil in kale in 2015 and with eucalyptus lemon essential oil in broccoli in 2017. In both years, Garlic Barrier reduced swede midge damage and increased marketable yield by 20% compared to untreated plants (C. A. Stratton, unpublished data). Eucalyptus lemon slightly reduced swede midge damage but was mildly phytotoxic and caused some scarring on the broccoli leaves. Although leaf scarring would lessen the marketability of leafy Brassica spp., the broccoli crowns were not scarred and had higher harvestable weight compared to untreated plants (C. A. Stratton, unpublished data).

Field trials of garlic essential oil in New York, however, did not reduce swede midge damage. Here, we applied food-grade garlic essential oil (1%, Bulk Apothecary, Aurora, OH) once per week from 1 week after transplanting until the start of head formation in broccoli (6–8 applications in total) in 2 field trials (Hoepting 2018a). Garlic oil failed to control swede midge in both of these high-pressure trials. The following year, an adjuvant with spreader-sticker properties, Nu-Film-P 0.25% (Miller Chemical, Hanover, PA), was added to garlic essential oil at 1% to improve the residual and efficacy of this product in the field. The garlic essential oil and adjuvant mixture was applied to kohlrabi weekly for 4 weeks. This treatment did not reduce swede midge incidence or damage severity in the crop under moderate-high swede midge pressure.

Pheromone Mating Disruption

Pheromone mating disruption (PMD) has shown some promise in reducing swede midge populations on farms but also has several logistical and economic challenges. By continuously treating crop fields with high doses of the synthetic female sex pheromones, PMD disrupts insect pest mating by preventing males from locating females, altering female pheromone signals, and disabling male sensory systems (Welter et al. 2008, Miller and Gut 2015). Ultimately, pest populations may be reduced on farms using large-scale setups of pheromone dispensers for mating disruption. Although research studies have shown that pheromone dispensers can disorient male swede midges (Samietz et al. 2012, Hodgdon et al. 2022b), the ability of PMD to provide crop protection is still inconclusive. In a Swiss study, PMD resulted in a 90% reduction in Brussels sprout damage when comparing plots treated with the 3-component stereospecific swede midge pheromone blend (blend naturally produced by females) to untreated plots (Samietz et al. 2012). In our field studies in Ontario, however, PMD did not prevent economic damage in broccoli test plots despite using a similar plot setup. Even with evidence of disorientation through trap shut down (i.e., the lack of pheromone trap captures in treated plots), broccoli sustained high rates of plant damage (Hodgdon et al. 2022b).

Several aspects of swede midge ecology, behavior, and population dynamics in its invasive range present challenges for commercial use of mating disruption. Our research suggests that the midge mates upon emergence, i.e., midges mate in fields where Brassica crops were previously grown (Hodgdon et al. 2019). If dispensers are placed within the current crop, they may be ineffective in preventing mating of midges emerging elsewhere. Due to diverse cropping rotations on vegetable farms, pest managers may need to treat multiple fields to manage emerging midges during the growing season for this tactic to successfully reduce populations.

The female-produced, stereospecific swede midge pheromone is complex to synthesize and thus expensive, further increasing the cost of treating multiple fields or whole farm landscapes. We found that lower-cost racemic (all possible stereoisomers) and single-component pheromone blends were less effective than the natural blend in disorienting males and providing crop protection (Hodgdon et al 2022b). Recently, the availability of less expensive precursor compounds has helped to make the stereospecific swede midge pheromone affordable. Other opportunities to reduce PMD cost for swede midge include reducing emitter pheromone output through timed devices, reducing pheromone load within dispensers, and altering crop rotation strategies to minimize emergence sites requiring treatment. Research and development of swede midge reproductive biology and PMD are ongoing, and large-scale field trials at the farm level are needed as the next step to demonstrate efficacy for commercial application.

Mass Trapping

There is potential for using the highly attractive sex pheromone used in PMD and swede midge monitoring for a mass trapping system. Mass trapping systems have successfully managed a wide range of insect pests by using odor or pheromone lures to bait insect traps and reduce pest pressure and economic damage (El Sayed et al. 2006). A key component of mass trapping is targeting the pest population before mating occurs, which would prevent oviposition and larval feeding. Mass trapping systems could maximize efficacy by focusing treatments in field sites with previous infestation and the earliest swede midge emergence of the growing season before treating subsequent Brassica crop plantings. While the swede midge pheromone lures successfully trap male swede midge, an additional attractant for female swede midge could result in greater population suppression. There is no attractant currently available for female swede midge. Additional research on the reproductive behavior of swede midge as well as female-based attractants is ongoing.

Discussion

Over 2 decades have passed since swede midge damage was first detected in North America (Chen et al. 2011). There has been considerable investment in time, personnel, and funding dedicated to understanding swede midge in its invasive range while testing the efficacy of many different management strategies to disrupt its life cycle (Fig. 3). Most strategies that are effective in other IPM programs have fallen short in reducing swede midge damage below economic threshold levels. Conventional and organic vegetable growers still report large economic losses of Brassica crops and dissatisfaction with available management strategies for swede midge (Hodgdon et al. 2022a). Continued investment in both basic and applied understanding of swede midge is imperative as invasive species have only become more prevalent in an increasingly globalized society (Bright 1999, Hulme 2009, Bertelsmeier 2021). Maintaining effort in swede midge research will not only help vegetable growers in the short term but will also provide foundational knowledge to jumpstart research if another vegetable gall midge invades North America. We conclude our review with summaries of the most effective organic management strategies developed for swede midge thus far and most promising research areas warranting further attention.

Because OMRI-listed insecticides are largely ineffective for swede midge, an integrated approach using multiple control techniques that reduce swede midge populations could prove effective on farms. Spatiotemporal crop rotation and insect exclusion netting are tactics that separate or block swede midge from their hosts and have the most support for efficacy in the literature. However, limited land for production prevents some growers from using recommended rotations. Netting allows growers to produce Brassica crops adjacent to fields with known infestations and avoid loss but cannot protect crops transplanted into infested soil. Therefore, growers need to be aware of when Brassica plants were last cropped in each field for the last 3+ years while monitoring active populations with commercially available traps and lures (Evans 2017). While growers can maintain historical records of where Brassica crops were last grown, the high cost of labor and supplies associated with managing exclusion netting and limited land can inhibit growers from using these strategies (Hodgdon et al. 2022a).

Tarping, or using plastic sheets/ground barriers over soil, has recently become a popular practice for small-scale organic vegetable growers in the northeastern United States (Lounsbury et al. 2022). The impact of ground barriers on soilborne stages of insects, however, has not been extensively studied. Further studies of effects on soilborne arthropods will inform ground barrier use so that multiple benefits can be obtained (e.g., a ground barrier that is suppressing weeds and emerging swede midge adults). For swede midge, ground barriers effectively block adult emergence from the soil; however, ground covers also take land out of production. Identifying the maximum length of time needed for covers to stay in place over infested soil could reduce the amount of time that land is left fallow.

In a 2018 survey of vegetable growers, biological control was identified as a major priority for future swede midge research programs (Hodgdon et al. 2022a). The parasitoid wasp Sy. myles is now present in Canadian canola fields, but to date, it is unclear if it has established in vegetable fields in Canada or the United States. Because Sy. myles is not a swede midge specialist, researchers are hesitant to recommend the releases of this insect (Abram et al. 2012, Ferland 2020). Lack of a specialist natural enemy, despite international efforts to identify one, poses the most limiting factor for classical biological control of swede midge. Conservation biological control practices to promote adventive Sy. myles populations through landscape alteration and incorporation of nectar-providing plants, like sweet alyssum, as interplantings are the most promising for biological control of swede midge (McLennan 2021). Long-term use of native strains of entomopathogenic nematodes may also be a promising research area for biological control of swede midge. In general, the diversity of entomopathogen strains—bacteria, fungi, and nematodes—for swede midge control has yet to be fully explored.

Pheromone mating disruption and other chemical ecology-based approaches are promising but will require large-scale field trials before major recommendations can be made. The effectiveness of mating disruption is difficult to evaluate within small plots as females can mate outside of the treated area and then immigrate into the Brassica field. Replicating a mating disruption field study for swede midge management will require a collaborative effort from professionals throughout the Great Lakes and northeastern region of North America, in both Canada and the United States, where swede midge populations are present.

Finally, the development of new OMRI-listed pesticides would offer an important tool for a rounded IPM program to control swede midge. Bacillus thuringiensis ssp israelensis de Barjac targets Diptera (flies; Pérez et al. 2005, Brühl et al. 2020) and could be formulated into a product labeled for swede midge on vegetable crops. Despite our efforts to broaden B. thuringiensis insecticide options for field production, currently available products are costly and marketed for greenhouse pest management. Few new OMRI-listed insecticide products are released each year with new active ingredients, and more chemical management tools are required for swede midge and other challenging pests.

Our review should serve as a call to action for researchers, vegetable producers, and funding agencies alike. Research and development for swede midge management has encountered many dead-ends, highlighting the need for farms to use several strategies simultaneously to disrupt different aspects of the swede midge life cycle. For organic growers, combining crop rotation, exclusion netting, and growing fewer susceptible crops has been most effective in preventing swede midge populations from building on farms. Our decision tree (Fig. 4) of these most widely used and successful management strategies may assist commercial vegetable growers in planning a swede midge IPM program for their farm. Of the management strategies tested, small reductions in swede midge demonstrated in the research could contribute to a balanced IPM program and provide an alternative to insecticidal management for growers who are looking for additional options. We hope that the research reviewed here can help catalyze continued innovation toward integrated programs that will rein in this elusive and devastating pest.

Author Contributions

Elisabeth Hodgdon (Conceptualization [Equal], Investigation [Equal], Project administration [Lead], Writing—original draft [Equal], Writing—review & editing [Equal]), Chase Stratton (Investigation [Equal], Writing—original draft [Equal], Writing—review & editing [Supporting]), Christine Hoepting (Investigation [Equal], Writing—original draft [Equal], Writing—review & editing [Supporting]), Andrea Campbell (Conceptualization [Equal], Investigation [Equal], Writing—original draft [Equal], Writing—review & editing [Equal]), Angela Gradish (Investigation [Equal], Writing—original draft [Equal], Writing—review & editing [Equal]), Braden Evans (Investigation [Equal], Writing—original draft [Equal]), Rebecca Hallett (Investigation [Equal], Writing—original draft [Equal], Writing—review & editing [Supporting]), and Yolanda Chen (Conceptualization [Lead], Investigation [Equal], Methodology [Equal], Project administration [Supporting], Writing—original draft [Equal], Writing—review & editing [Equal])

References