Herbivores are often predicted to be limited by protein because plants typically have low nitrogen and, by extension, protein content relative to animals (1–4). However, herbivores with high energetic demands, such as migrating locusts, prefer and perform better when fed diets that are carbohydrate-biased (5–10). Most of these recent studies employed the Geometric Framework for Nutrition (GFN), which demonstrates consumers’ ability to balance dietary macronutrients like protein and carbohydrates. This regulation is key for maximizing performance (11). Thus, herbivores with high energetic demands are predicted to select plants with a low protein:carbohydrate ratio so they can meet their metabolic needs without overconsuming protein, as protein has been shown to have deleterious effects when consumed in high quantities (12–15). However, other factors such as mechanical and chemical defenses can influence host plant choice and herbivore performance. Here, we tested the interactive effects of host plant fertilization and wheat seedling state (intact or ground) on the preference and performance of the Australian plague locust, Chortoicetes terminifera (Walker) (Orthoptera: Acrididae), the most economically important and widespread locust species in Australia.

Locusts are grasshoppers that aggregate at high density and migrate long distances; they are also generalist herbivores (16). Long distance flights are fueled primarily by fat stores that are typically built via carbohydrate consumption (7, 17, 18). Indeed, locust outbreaks are often found in areas containing low protein, high carbohydrate plants, as has been shown in China (19), Australia (10, 20), and West Africa (8, 9, 21). When given the choice to balance dietary macronutrients with two complementary artificial diets using GFN methods, field populations of locusts select carbohydrate biased diets on which they have the highest growth and survival: Oedaleus asiaticus in China (5, 19), Oedaleus senegalensis in Senegal (8, 9), Schistocerca cancellata in Paraguay (7), and Chortoicetes terminifera in Australia (10, 22). Furthermore, Oedaleus asiaticus locusts fed their preferred protein:carbohydrate ratio fly for longer periods of time (5, 23). This pattern holds when eating plants. S. cancellata nymphs collected from marching bands preferred and gained more weight when fed plants with high carbohydrate contents (7). While O. senegalensis is found in environments where rapidly growing plants are often protein-rich, locusts are more numerous in fallow fields where soil fertility is lower (21) and plants contain more carbohydrates (9). This species preferred unfertilized over fertilized millet leaves, and had higher survival and laid heavier eggs when kept in field cages over unfertilized vs. fertilized millet (8). Collectively, these studies indicate that balancing protein and carbohydrate, and especially ensuring adequate carbohydrates, is an important factor influencing host plant choice for locusts. However, limited studies have investigated the interactive effects of plant nutrients and their mechanical properties on herbivory and on herbivores themselves.

In addition to raw nutrients, plant physical attributes, cell walls in particular, may restrict nutrient access and limit performance of herbivores (24, 25, 2004; 26). The plant cell wall is an extracellular matrix made of two main layers, the middle lamella and the primary cell wall, that encapsulate much of the plant cell’s nutritional content (27). The primary cell wall is made of cellulose which is undigestible, except by specialized consumers such as ruminants (28). Plant cell walls can hinder nutrient assimilation for herbivorous insects. For example, the Australian plague locust can only assimilate 40% of plant carbohydrate content when the cell wall is present compared to 90% when this barrier is mechanically overcome by grinding the plants (25). Surprisingly, insects were able to assimilate 80% of plant protein content when cell walls were intact (29). This suggests that carbohydrates are less accessible in some plant species than protein, although the mechanisms behind this are poorly understood. These studies were conducted using non-agricultural (undomesticated) grasses common to Australia and it is unknown if this pattern of inaccessibility holds for domesticated grass varieties that may be less defended mechanically and typically more protein-rich (25) or across agricultural regimes that may affect nutrient availability, such as soil amendments.

During outbreak years, the Australian plague locust, C. terminifera, invades rangeland and agricultural fields (30, 31), and wheat is grown on 42% of the 19.7 million ha of crop-growing land vulnerable to these outbreaks (32). We tested how locusts responded to wheat that was either fresh or ground and with different levels of nitrogen fertilizer. We predicted that fertilization would decrease, and grinding would increase, preference and performance by mechanically breaking down the cell wall and making carbohydrates more accessible. Understanding how both plant nutrient content and accessibility affect locust choice and performance when eating key crops is strategic for improving management programs for this serious pest.

Growing evidence indicates that final instar locusts prefer and perform best on diets with a lower protein to carbohydrate ratio, whether feeding on artificial diets or plants (reviewed in 6), and this study provides some support for that pattern. Locusts increased consumption on control wheat treatment, which had both a higher caloric and carbohydrate density. This increase in consumption corresponded with an increase in mass, likely as lipid (7). However, besides growth, this increase in food consumption did not improve nutrient assimilation, molting success, or survival. In contrast to previous studies (25, 29), we did not find that breaking down plant cell walls increased locust performance. In fact, locusts performed better on intact wheat than they did on dried ground wheat, which may have been due to mechanical or nutrient differences (38, 39) in the grass species used and/or how plants were dried. This research question could benefit from further investigation.

When locusts are given the choice between foods differing in macronutrient content, they will self-regulate by compensatory feeding of each plant to strategically meet their ideal p:c ratio (6). Locusts and swarming grasshoppers tend to select a carbohydrate-biased p:c intake target (6). For example, final (5^th) instar C. terminifera selected 1p:1.13c in a lab population (37) and 1p:1.8c to 1p:1.37c in field populations (10). However, protein-biased nutritional landscapes are common in agricultural settings and for young rapidly growing plant tissues (8, 40). Indeed, both control and fertilized wheat in our study contained more protein than carbohydrates: control was p1.94:c1 and fertilized was p3.81:c1 ( Figure 2 ). Thus, locusts in our study were constrained to a protein-biased diet for both plant choice and plant no-choice experiments. Accordingly, the self-selected p:c ratios in the choice experiment were both protein-biased, albeit slightly less protein-biased for the dried ground wheat (2.42p:1c vs. p2.26:c1).

Results from all experiments indicated that locusts preferred the control wheat over the fertilized wheat. In choice experiments, locusts ate about 2–3 times the amount of control relative to fertilized wheat; in no choice experiments, locusts ate about 30–70% more over 6 days when confined to control wheat relative to fertilized. Locusts conferred some benefits in eating the lower p:c wheat. In fresh wheat experiments, locusts eating control wheat had a faster weight gain in the first 6 days, though there was not a significant effect of fertilization on food assimilated or on molting or death rate. Our study was conducted on a single nymphal stadium and it is likely that longer-term experiments would show stronger effects. For example, a long term study on caterpillars (Heliothis virescens) showed that there is only a narrow range of p:c that maximizes performance over the course of their lifespans (41). Because balancing p:c intake is a primary driver of foraging behavior and growth for insect herbivores (42), it is likely the lower p:c ratio of control plants increased preference and growth rate. Given that protein amounts were similar in control and fertilized wheat plants ( Table 1 ), our results suggest that carbohydrates may be a key factor in this choice. Locusts may have chosen plants based on total energy content as unfertilized plants were more macronutrient dense (36.15% ± 1.84 SE and 41.86% ± 8.25 SE, respectively). It is also possible that using a simple N fertilizer like urea decreased micronutrients; other studies have shown (43) usage of urea can cause a negative effect that producers may avoid by using fertilizers with more micronutrients like aminochelates. Nevertheless, our results corroborate other regional studies (China (5, 2012), Australia (10, 2020), and West Africa (8, 9, 21, 44)) indicating that late-instar locust and swarming grasshopper species prefer and perform best on low p:c diets.

Our expectation was that grinding wheat to particles smaller than 10μm would increase nutrient accessibility, particularly soluble carbohydrates, and therefore increase performance. Another study on 5^th instar C. terminifera nymphs found that grinding freeze dried Mitchell grass (Astrebla lappacea) improves nutrient accessibility and assimilation (25). Furthermore, this same study (25) showed that locusts were able to extract 50% more carbohydrates when plant cell walls were removed by grinding than when they were consuming fresh plants (25). Therefore, we expected preference for unfertilized wheat (less protein-biased plant) to be more pronounced when using intact plants since carbohydrates should be harder to access. However, we found similar preference for unfertilized plants in both fresh and dried ground studies. Moreover, for locusts fed dried ground plants in our study, we recorded lower food consumption, less successful molts to adult, and lower survival.

We have identified potential factors that may explain the differing effects of dried ground plants on locusts between Clissold et al. (25) and our study. In the Clissold et al. (25) study, lyophilizing (freeze drying) the plants instead of desiccating them in a drying oven may have helped maintain nutrient and vitamin levels that may have been degraded through the drying process in our samples. Another potential explanation may lie in plant structure as leaf toughness can affect nutrient accessibility and plant choice (25, 29, 45). Here we used a seedling (3-4 weeks old) cereal crop, wheat, which is not as thick as the wild Mitchell grass (A. lappacea) used in the Australian study (12-14 weeks old). Thus, grinding the tough Mitchell grass may have released more nutrients than grinding soft wheat sprouts. Grasses (Poaceae) can contain high concentrations of silica (46) which can wear out herbivore mandibles and decrease consumption (47, 48). Silica has been shown to deter locusts which prefer to eat plants with lower silica concentration (49–52). It is possible that grinding the plants released silica structures inside the gut of the insects, causing internal damage (51, 53), though that would not explain higher performance in locusts eating ground grass in the Clissold et al. (25) study. A final explanation may be major differences in nutritional content between these two plant species. Adult Mitchell grass contains 9.7% protein and 24.1% carbohydrates (25) while our seedling unfertilized wheat plants contained 27.6% protein and 14.2% carbohydrates. Thus, grinding wheat seedlings may not have released the same amount of soluble carbohydrates. More studies using additional combinations of plant and grasshopper species are needed to disentangle the relative importance of these factors.

We contrasted experiments using control dried ground plants with dry powder artificial diets to ensure low consumption and growth rates were not due to diets being a dry powder. Locusts eating the artificial diet had substantially higher consumption and growth rates than locusts eating dried ground plants. This result may be partially due to the p:c ratio of the artificial diet (14p:28c) being a better match to the preferred p:c of C. terminifera populations (10, 37), which likely further supported improved growth. However, rates for the artificial diet overlapped with consumption and growth rates for locusts eating control fresh wheat, so there was likely a combination of nutritional and structural factors at play.

In conclusion, we showed that fertilizing young wheat makes it less preferred by Australian plague locusts than control wheat, and that locusts will decrease consumption if confined to fertilized wheat seedlings. It was important to test young plants because seedling stages are typically more vulnerable to locust attacks on foliage since older plants with more leaves can better tolerate herbivory (32, 54). While seedling wheat crops regularly sustain considerable damage from C. terminifera in Australia (32), we showed that their nutritional profile is far from the optimal p:c ratio for this locust species. Our results suggest that carbohydrate is potentially a limiting nutrient for C. terminifera, particularly in agricultural settings where nymphal bands attack seedling plants that may be richer in protein than what is usually assumed for grasses. This study represents an important step in bridging the gap between theoretical knowledge developed using artificial diets and practical advances that can form the basis of a nutritionally-based management program for herbivorous pests.

The data presented in the study are deposited in the Dryad repository, https://datadryad.org/stash/dataset/doi:10.5061/dryad.5hqbzkh9d. Further inquiries can be directed to the corresponding author.

JB and MG contributed to the conceptualization and design of the study. JB and SM collected the data, and JB organized the database. JB and MG performed the statistical analysis. JB made figures and tables. JB and MG wrote the original manuscript. MG administrated the project and supervised it. AC and RO acquired funding and provided the resources necessary to complete the project. All authors contributed to the revisions of the manuscript. All authors read and approved the submitted version.