Arabidopsis thaliana ecotype Columbia (Col-0, ABRC Stock no. 6000) was grown under controlled conditions at 22°C to 25°C, 150 μmol m^-2 sec^-1 photosynthetically active radiation and a 10 h light - 14 h dark photoperiod. Seeds were stratified for 24 h at 4°C prior to planting in potting mix [90% Sunshine Mix No. 1 (Sun Gro Horticulture, Bellevue, WA) and 10% sand].

Seramis^® clay granules http://www.seramis.de are readily available in Europe with currently limited distribution in North America (Ace Gardening Products, Kitchener, Ontario, Canada; http://www.seramis.de/weltweit.html). Fifty ml polypropylene conical tubes (Fisher Scientific, Suwanee, GA) were prepared by drilling five 3.5 mm diameter holes around the bottom and a 25 mm hole in the cap (Figure 1).

Tubes were then filled with Seramis^® clay granules and wrapped with aluminum foil to reduce light penetration and algal growth. The top of each tube was covered with plastic wrap and the caps were reattached. A small hole large enough for the root mass to pass through was punctured in the plastic wrap using the tip of a scalpel. The tubes were then placed in a test tube rack and submerged in Hoagland's solution 22 to moisten the substrate. Four-week-old Arabidopsis plants grown in potting mix were carefully removed from their pots (Figure 1). Roots were repeatedly submerged in water to remove as much of the potting substrate as possible. A 1.25 cm cube of Rockwool was divided and placed beneath the rosette, around the top of the roots. Plants were then transferred to Seramis^®-containing tubes and allowed to grow. The Rockwool and plastic wrap stabilized the plant until its roots had grown into the clay granulate. The clay granulate was kept moist by submerging the culture tubes in Hoagland's solution for 10 to 15 min every other day. After four weeks, when the plants started to bolt, roots had grown to the bottom of the tube. During the feeding experiments, plants were watered every other day by applying 10 ml of Hoagland's solution to the surface and allowing excess media to drain freely from the tube.

Fungus gnat larvae were collected from greenhouse soil by the "potato disk" method 23 and subsequently used to establish a laboratory colony. The colony was maintained in 8 L plastic containers with screened openings for ventilation. The culture medium consisted of 4 L of moist Sunshine Mix No. 1 enriched with 1.5 kg of shredded potato tubers 24 . The cultures were kept at ambient greenhouse conditions (21°C to 23°C; indirect sunlight). To maintain the colony, 0.5 L of medium (containing Bradysia larvae, pupae, adults, and eggs) was transferred to a fresh container of soil and potato mix every three to four weeks. Specimens were identified by Dr. Raymond J. Gragné (USDA Systematic Entomology Laboratory, Washington, DC) as a mixed colony of Bradysia coprophila (Lintner) and Bradysia impatiens (Johannsen).

Larvae were collected from the culture using a modified flotation/extraction method previously described by Cloyd and Zaborski 24 . Approximately 1 L of culture medium containing Bradysia larvae was transferred to a 2 L wide-mouth Erlenmeyer flask. The flask was filled with room temperature tap water and agitated to break up the culture substrate and release the larvae. The flask was then inverted and placed on top of a 2 L graduated cylinder filled with cold water, so that the mouth of the flask opened just below the water surface in the cylinder (Figure 2A). In this position the flask remained filled with water with the graduated cylinder serving as both a container and a rack to hold the flask. Denser material including fungus gnat larvae sank from the flask into the cylinder while most of the potting mix remained floating in the flask. After approximately 5 min, the flask was removed, and its contents were discarded. Water, larvae, and the remaining organic matter were poured from the graduated cylinder into a plastic container. After the larvae had settled to the bottom of the container, the water was decanted and a 1.5 M MgSO₄ solution was added. Larvae floated to the top because of their lower density compared to the MgSO₄ solution, while most of the remaining organic matter sank to the bottom. Larvae were then collected in a sieve (1 mm mesh size), rinsed in water, and placed in a Petri dish with moist filter paper without food for 20 to 24 h prior to their use in feeding assays. The described isolation method collected approximately 1000 Bradysia larvae of all four instars from one liter of culture medium (Figure 2B).

Four weeks after their transfer to aeroponic culture tubes, Arabidopsis plants were infested with fungus gnat larvae. To this end, the plastic tube caps were carefully removed to avoid leaf damage. Approximately 200 to 300 second- and third-instar larvae, collected as described above, were transferred to a single aeroponic culture tube. First instar (~1 mm) and fourth instar larvae were excluded from the feeding experiments since they were either too difficult to apply or too close to pupation. Larvae were submerged in 1 mL of Hoagland's solution (Hoagland's solution was used instead of water to avoid washing nutrients from the clay pellets) and then released with a pipette into the Seramis^® substrate. After two and four days of larval feeding, roots were removed from the tubes by submerging the plant and the tube in water. Roots were then separated from the granulate by simply holding the plant rosette and moving it back and forth in the water. The heavier clay granules fell to the bottom of the container, while the roots stayed attached to the rosette. The shoots and roots were then blotted dry with Kimwipes (Fisher Scientific, Suwanee, GA) and their fresh weights were recorded. Tissue dry weight was measured after two to three days of air drying at room temperature. Under lab conditions, this time was sufficient to completely dehydrate the root mass and additional drying did not significantly alter weight measurements. The shoot and root biomass and percentage of water weight for each tissue were calculated. For these measurements a T-test was performed (n = 12, α = 0.05) to test the null hypothesis of no biomass change for the variables. To determine if there was a significant difference in root mass after larval feeding, a one-way analysis of variance (ANOVA) followed by Tukey-Kramer HSD (n = 8) was performed.

To validate the feeding damage by Bradysia larvae on roots of aeroponically-grown mature Arabidopsis plants, plants were challenged with larval feeding at the seedling stage in potting mix. Individual seedlings were grown in pots (8 cm × 8 cm × 10 cm) under the described conditions. Bradysia larvae (second and third instar, Figure 2B), collected as described above, were dislodged from the bottom of the Petri dish with 1 mL of water, and ten larvae were transferred with a Pasteur pipette into each pot. Larvae that did not move into the potting substrate within 5 min were removed and replaced. Larvae were transferred to pots at seed germination, ten and 14 days after germination. Infested and control plants were kept in net enclosures. The percentage of surviving seedlings from three independent experiments was determined seven days after inoculation. Plants were removed from the pots, roots were rinsed in water, and fresh and dry weights of the entire seedling were recorded. For data analysis, the weights were log transformed before performing a one-way ANOVA and Tukey-Kramer HSD (n = 8-10).

Statistical data analysis was performed as described for the individual experimental procedures. Analysis of variance was accomplished by the JMP program (Version 8, SAS Institute Inc., Cary, NC, USA) statistical software.

To cultivate Arabidopsis under aeroponic conditions, plants were first grown on potting mix for approximately four weeks prior to their transfer to aeroponic culture tubes. By four weeks after transition to the clay granulate, roots had grown from the residual potting substrate particles to the bottom of the plastic tube. In comparison to Arabidopsis plants grown in potting substrate, aeroponically grown plants appeared morphologically normal in primary and secondary root growth as well as in the formation of root hairs 25 (Figure 3 and see below). The Rockwool was removed without difficulty since roots had not grown into the Rockwool substrate. Approximately 1 g of healthy root tissue was obtained from each individual plant (Table 1). The root-to-shoot ratio for fresh and dry weight was 0.40 and 0.45, respectively (Figure 4C).

Approximately 200 to 300 second and third instar larvae were released into each aeroponic culture tube. Within 24 h, larvae, observed through the clear plastic tube, were actively feeding on the root tissue. Two days after the release of the larvae, severe feeding damage was visible on most roots (Figure 5A and 5B). The majority of root consumption was observed within the upper 5 cm of the tube (Figure 5A) indicating that most larvae fed from the top and gradually moved toward the bottom of the tube as the food source became depleted. Larvae stripped away the root epidermis and cortex, but generally avoided consuming the vascular tissue (Figure 6E-H). Some fine roots were completely severed. Feeding damage was most severe on root tips, young roots, and root hairs suggesting a preference by larvae for these cells and tissues. Thick tap roots displayed little or no feeding damage.

Larval root consumption significantly reduced Arabidopsis root biomass. On day two and four of larval feeding, the root fresh weight was reduced by 58% and 55%, respectively (Figure 5B). These values reflect the highest reduction of root mass observed in one out of ten independent experiments; the average loss of root mass for all experiments was 37%. In addition to root consumption, stems were poorly supported and rosette leaves were beginning to wilt (Figure 5A). To determine whether leaf wilting was caused by a decreased uptake of water, shoot fresh and dry weight were measured and the weight attributed to water was determined (Figure 7). After four days of larval feeding, shoots showed significantly reduced water content, while shoot dry weight remained unaffected. By contrast, both root fresh and dry weight were significantly reduced with no significant change in the water content of roots due to feeding damage (Figure 7).

Bradysia larvae reduced Arabidopsis seedling establishment. Only 24 to 33% of seedlings survived when exposed to ten Bradysia larvae at germination (Figure 8). When 10-day-old seedlings were treated with larvae, survival rates increased to 45 and 51%. However, more than 95% of seedlings survived when larval treatment started after the first pair of true leaves had developed ten to 14 days after germination. Nevertheless, all of the surviving plants were considerably affected by herbivore damage since they had significantly reduced biomass in comparison to unchallenged seedlings and were noticeably stunted in growth (Figure 9).

Investigating root herbivory has remained a challenge in the study of belowground plant-organism interactions because of limited access to root tissue for subsequent analyses. Several culture methods have been developed previously to make Arabidopsis roots more accessible for molecular and biochemical studies. These methods are primarily based on growing plants in liquid culture for optimal root biomass production, but they are not suitable for investigating root herbivory 22 26 . We tested whether hydroponically grown roots removed from liquid culture could be used for Bradysia feeding experiments; however, larvae did not actively feed on these roots. Similarly, larvae did not feed on roots of plants grown in solid sterile media, but instead burrowed into the agar and suffocated. Arabidopsis hairy root cultures have been previously used to investigate root-aphid interactions and the emission of volatile organic compounds in response to aphid attack 27 . However, in this study aphid feeding was confined to floating root tissue, which simulates the natural habitat of the herbivore only marginally and prevents analyzing responses of the whole plant.

To study herbivory on Arabidopsis roots, it was necessary to establish a culture system that could accommodate belowground herbivores while still providing easy access to the root tissue for further molecular or biochemical analysis. Initial attempts to grow Arabidopsis in sand and vermiculite resulted in poor plant growth in comparison to soilless potting mix. By contrast, optimal growth conditions were achieved by growing plants in nonsterile perlite or Seramis^® clay granules. Seramis^® substrate was preferred over perlite since roots could easily be removed from the clay granules by submersion in water without causing any substantial tissue damage.

Optimal survival of plants was achieved when plants were first grown on potting mix for approximately 4 weeks prior to their transfer to aeroponic culture tubes. Transfer of younger plants was avoided due to inconsistent growth performance under aeroponic conditions. Growing seedlings directly on Rockwool placed on top of the clay substrate is less suitable since plants are easily stressed by over- or under-watering. A similar problem has been encountered in establishing hydroponic cultures and is considered a major limitation in the success of germination and seedling survival in this system 22 26 . Within two to three days after transplanting, plants adapt to the aeroponic conditions and grow as phenotypically healthy mature plants with no obvious signs of impaired root growth or damage after 4 weeks (Figure 3 and 6A-D). The culture requires eight weeks to progress from seed to mature plants, but beyond transferring the plants from soil to aeroponic substrate, only watering with Hoagland's solution is necessary. The harvest of roots from the aeroponic system is rapid, complete, and causes negligible damage to roots (Figure 5A and 6A-D) in contrast to removal of potting mix (Figure 1B-D and 9).

Aeroponically-grown plants produce a root mass comparable to that obtained in other previously reported Arabidopsis culture systems (Table 1). For example, plants grown for 24 days in axenic liquid culture supplemented with 1 to 3% sucrose yield approximately 1 g of root fresh weight 26 , which is similar to that of roots grown in the aeroponic system (Table 1). When compared to Arabidopsis grown under hydroponic conditions, the average root-to-shoot ratio in aeroponic culture (0.45) is approximately twice the ratio observed for hydroponically-grown plants 22 . Thirty percent of the total aeroponic plant biomass (dry weight) is comprised of root tissue while only 15 to 25% of total hydroponic plant biomass has been attributed to root tissue 22 (Table 1). The higher percentage of root mass under aeroponic conditions may be the result of increased aeration in the root environment 28 29 . Moreover, enhanced root growth might be caused by the intermittent nutrient supply in aeroponic culture as opposed to continuous nutrient availability in hydroponic systems.

Taken together, the technique shown here represents a simple culture method, which can be established under ordinary growth conditions without the use of sophisticated hydroponic equipment such as bubble stones, air pumps or sprayers. In comparison to hydroponic cultivation, the culture system reflects more closely the conditions of the soil environment. Since clay granules such as Seramis^® are used as a common plant medium in home gardens and interior landscapes http://www.seramis.de, aeroponic cultures similar to the one presented here can be established for a variety of different plants including those with a longer life span than Arabidopsis. Because of the loosely packed consistency of the clay substrate, it seems possible that root herbivores of larger size than Bradysia larvae can be accommodated in this substrate, which will allow testing this method for other root-herbivore systems.

Fungus gnats are common greenhouse pests and frequently infest Arabidopsis growth rooms. While feeding of Bradysia larvae on Arabidopsis seedlings has been previously reported 19 , the extent of feeding damage on mature wild-type plants has not been investigated. When placed in aeroponic clay substrate deprived of other food sources, larvae quickly began to feed on root tissue. Within two to four days of feeding, larvae significantly reduced root mass by an average of 37%. Despite the considerable damage of tissue belowground, plants appeared to be less affected aboveground. However, Bradysia-damaged plants showed reduced shoot mass, which was due to the loss of water, suggesting that water absorption and most likely nutrient absorption were compromised by root herbivore damage.

In the presented experiments, approximately 200 to 300 larvae at the second and third instar were applied per plant. Considering that only 20 to 25 gnats released into an area with 15 pots containing Arabidopsis plants can yield 300 to 500 larvae per pot within 14 days 19 , it was within reason that this density can occur under greenhouse conditions. The observed root damage in aeroponic culture may be magnified due to the absence of other food sources such as organic matter. Bradysia larvae primarily consumed root tips, root hairs, and young lateral roots (Figure 6E-H). Less feeding was observed on older roots, where larvae preferably removed the epidermis and cortex but none of the vascular tissue and only parts of the endodermis. This feeding preference may be caused in part by increased root secondary growth. An increase in cell wall lignification can affect root herbivore activity as shown for wireworm feeding on tobacco roots 30 . Moreover, differences in formation and concentration of constitutive and induced secondary defense metabolites such as glucosinolates and terpenoids can contribute to the observed differential feeding behavior and overall root consumption (Vaughan, Tokuhisa, and Tholl, unpublished results). The role of direct defense metabolites was also demonstrated in spinach roots, where Bradysia larvae led to a dramatic up-regulation of defense compounds 15 . A recent study investigating glucosinolate distributions in canola (Brassica napus) roots showed that the highest concentration of glucosinolates occurs in the outer layer of roots with secondary growth 31 32 . To what extent Bradysia feeding is affected by the distribution of glucosinolates in Arabidopsis roots requires further investigation.

Depending on the developmental stage, Arabidopsis seedlings grown in potting mix were severely affected by Bradysia larval feeding. A moderate infestation 18 of ten larvae per plant reduced seedling survival by 50% or less depending on the plant age at the time the biotic stress was introduced. Surviving seedlings were much smaller and appeared to have reduced growth. This reduction in plant growth can be attributed to larval root consumption and limited uptake of nutrients but may also reflect fundamental changes in the physiological state of the plant due to trade-offs between resource allocation to defense and growth 33 34 or changes in the potting mix due to frass deposition.

To our knowledge, a bioassay for studying belowground herbivory on intact Arabidopsis plants has not previously been reported. We propose that Arabidopsis grown in aeroponic culture and the generalist herbivore Bradysia (B. coprophila and B. impatiens) can be used as a novel system to investigate belowground plant defense responses to herbivore attack such as the release of secondary metabolites. The chemistry of plant defense against root herbivores is one of the most neglected aspects of root biology 12 . Using this system, Arabidopsis mutants with altered profiles of secondary metabolites, such as glucosinolates 35 or terpenoids, can be applied to study the biological effect of individual defense compounds belowground. Finally, knowledge gained from these and various other mutants may provide closer insight into the still poorly understood molecular and physiological mechanisms of belowground and aboveground interactions 36 .