In contrast to predictions from nitrogen limitation theory, recent studies have shown that herbivorous migratory insects tend to be carbohydrate (not protein) limited, likely due to increased energy demands, leading them to preferentially feed on high carbohydrate plants. However, additional factors such as mechanical and chemical defenses can also influence host plant choice and nutrient accessibility. In this study, we investigated the effects of plant protein and carbohydrate availability on plant selection and performance for a migratory generalist herbivore, the Australian plague locust, Chortoicetes terminifera. We manipulated the protein and carbohydrate content of seedling wheat (Triticum aestivum L.) by two means: 1) we increased the protein:carbohydrate ratio using nitrogen (N) fertilizer, 2) we sought to increase carbohydrate accessibility by grinding cell walls after drying the plants. Using a full factorial design, we ran both choice and no-choice experiments to measure preference and performance. We confirmed locust preference for plants with a lower protein-carbohydrate ratio (unfertilized plants). Unlike previous studies with mature wild grass species, we found that intact plants supported better performance than dried and ground plants, suggesting that cell wall removal may only improve performance for tougher or more carbohydrate-rich plants. These results add to the growing body of evidence suggesting that several migratory herbivorous species perform better on plants with a lower protein:carbohydrate ratio. 

Plant treatment and nutrient analysis

Wheat treatments

We purchased seeds of red hard winter wheat (Triticum aestivum L.) from Sustainable Seed Company (South Salt Lake, Utah) and stored them in a freezer at -20°C until the beginning of the experiment. We chose this variety because of its hardiness and popular use as both a crop and as a dietary staple for lab-reared locust colonies. Wheat was grown hydroponically in a greenhouse at temperatures ranging from 20–22°C from November to January (light cycle 10.5 light hr:13.5 dark hr).

Seeds were first soaked for 18–24 hours in a cool dark area to initiate germination. We then placed 700–730 seeds in perforated containers (food-safe plastic, 16 x 13 x 4 cm) and covered them for two days. Once germinated, we placed those perforated containers in flood trays (Active Aqua AALR24B Low Rise Black Flood Table, ABS plastic, 121 x 61 x 13 cm). Every 8 hours, each tray was flooded for 15 minutes. Three days after being placed into the flood trays, using the same watering regimen, the wheat for the fertilized treatment received 4.792 g.L^-1 of urea (Greenway Biotech Inc. 46-0-0), an optimal amount for field-crop wheat (Kaiser, 2018). We dissolved the granulated urea in water and added to sump of each hydroponic system. This fertilization period lasted 3 days to support optimal nitrogen uptake (Franzen, 2015, 2018). The control treatment received water for three weeks. For both control and fertilized treatments, we replaced the water used to flood the wheat every 4–5 days.

Once plants had reached the desired age of 21 days old, we set half the wheat aside for the live-grass experiment, cut the remaining half of both treatments' wheat at the base and dried the leaves at 60°C for 48 hours. Afterwards, we ground the dried wheat to particles of < 10 m diameter (following Clissold et al., 2009) using a Retsch MM 400 ball mill at 30 Hz for 30 s. Ground leaves were then frozen at -20°C in airtight containers until use. Wheat and other cereal grains are most vulnerable to locust damage at the seedling stage, while more mature plants may be less affected by leaf herbivory (APLC, 2019); therefore, we used three-week old seedlings.

Protein and carbohydrate analyses

Plant protein content was determined with a Bradford assay and the non‐structural carbohydrate content using the phenol‐sulphuric acid method on the dried and ground plant material (e.g., Clissold et al., 2006; Deans et al., 2018).

Australian plague locust and experimental design

Locusts

Our C. terminifera lab colony is hosted at Arizona State University (Arizona, USA) and was established in 2015 from a colony originating from The University of Sydney (New South Wales, Australia). The Australian lab colony was started with wild locusts collected in 2005 and 2006 from Eastern and Western Australia (Clissold et al., 2014). The ASU colony is reared on a 14 h light:10 h dark cycle with RH = 20–50%, and a 30 ± 2ºC (light) 25 ± 2 (dark) ºC daily temperature cycle. Locust colonies are fed hydroponically grown wheat seedlings, supplemented with wheat bran treated with tri-sulfa for colony health.

Fresh wheat: choice experiment

All experiments started when locusts molted into their last nymphal instar (5th instar) at which point they were weighed, sexed, and placed in an experimental enclosure. In total, we used 12 wire mesh cages (45 cm long x 45 cm wide x 45 cm tall), each containing ten individuals (five males and five females). We did this to more accurately measure the amount of consumption as the individual consumption measurements would be more prone to inaccuracy due to small differences in amount consumed. The food source needed to be cut into “patties” containing wheat sprouts still connected to bare roots to remain turgid throughout the duration. We fed each cage of locusts two clipped and pre-weighed wheat patties presented in 8 x 6 cm food containers (SI, Figure 1). One food container was filled with nitrogen-fertilized wheat and the other with unfertilized (control) wheat. After 24 hours, locusts were removed and weighed. The remaining wheat was dried for 48 h and weighed to measure consumption. We estimated dry consumption using a regression equation linking the mass of fresh wheat to the mass of dried wheat. For this, we recorded the mass of 15 patties of control wheat and 15 patties of fertilized wheat, dried them for 48 h at 60ºC, and recorded their dried mass. The regression equation is presented in the Supporting Information (SI, Table 1).

Fresh wheat: no-choice experiment

For the no-choice experiment, we used ten cages per treatment (fertilized and control). Each cage contained six individuals (three females and three males) that were individually marked on the pronotum with Sharpie brand paint markers. We replaced the wheat patty every day until the locusts molted or died. We recorded locust mass and frass production (mg) every three days, as well as development time. We recorded consumption for days 0–3, days 0–6, and total consumption; locust body mass change for days 0­–3, days 0–6, and total locust body mass change; development time; survival; total frass production; and assimilation (total amount of food consumed−frass produced)/total amount of food consumed)).

Ground wheat: choice experiment

We placed 26 freshly molted 5th instar locusts (roughly half males and half females) into individual 17.5 x 11.5 x 4.5 cm perforated polystyrene cages with a perch for roosting and a water tube. Each cage contained two pre‐weighed dishes: one filled with fertilized and the other control (unfertilized) ground wheat. After three days, we removed any frass present in the dish and dried the diets for 24–36 hours at 60° C and then weighed the diets to measure the amount of ground wheat consumed.

Ground wheat: no-choice experiment

The no-choice setup was identical to the choice experiment, except that a locust received only one food dish per cage (fertilized ground wheat or control ground wheat). We used 26 individually housed locusts per treatment. We then removed the diet dishes after three days and dried, weighed and replaced the dishes with new pre-weighed dishes. The no-choice experiment ended when the locusts molted or died. We recorded consumption for day 0–3, day 0–6, total consumption, locust mass change for day 0–3, day 0–6, day 0–9, total locust mass change, development time and survival, frass production, and food assimilation.

Artificial diet vs ground wheat: no-choice experiment

To compare the effects of dried, ground wheat to a dried and powdered artificial diet containing all nutrients needed for locust growth and development, we ran a final experiment. This experimental setup was identical to the ground no-choice experiment but instead of comparing the performance of wheat treatments, we compared the performance of control wheat to an artificial diet with a ratio selected by field populations of C. terminifera (p14:c28) (Lawton et al. 2021). This experiment was run to test if dry foods, those with a complete lack of edible water, was to blame for high mortality and poor molting success found in other experiments. We used 16 locusts (8 for each treatment; 1:1 sex ratio) and followed the same protocol as the previous section dictates.

Statistical analysis

Prior to any statistical analysis, we assessed all data collected for normality and homoskedasticity. To compare protein and carbohydrate contents between fertilized and control wheat, we performed a MANOVA. For all experiments, we analyzed consumption and locust mass change using ANCOVAs with locust initial mass as a covariate to account for size differences and sex as a cofactor. For development time and survival, Kaplan-Meier survival analyses were used. For both the ground and the fresh grass experiments, we calculated frass and consumption rates (e.g., consumption/days in experiment) and analyzed both using ANCOVA’s and locust initial mass as a covariate. For all analyses besides the survival analysis, locusts that were not alive for the duration of the interval recorded (e.g. day 0–3 or day 0–6) were removed from the analysis. We presented the cumulative results for standardized time periods (days 0–3 and days 0–6) as well as for the whole experiment (day 0 to time of molt). We conducted the statistical analysis using R studio version 1.3.1073. as well as JMP Pro 15.2.0.

Microsoft Excel