Plant chemicals affect trade-offs between adult preference and larval performance of the rice water weevil, Lissorhoptrus oryzophilus

Herbivores use plant chemicals for host-plant selection to maximize their own and/or offspring performance. Since host plants that are optimal for mother and offspring are often different and spatially/temporally separated, how plant chemicals affect trade-offs between adult preference and larval performance remains unclear. We found that adults of the rice water weevil (Lissorhoptrus oryzophilus), one of the most important pests on rice in the world, preferred volatiles from barnyard grass over rice, tended to feed and oviposit on barnyard grass compared with rice. In contrast, larvae performed better on rice roots than on barnyard grass roots. Chemical analysis further show that rice roots had higher nitrogen and soluble sugar but lower lignin and cellhouse contents than barnyard grass. Together, these results suggest that violate, nutritive and defensive chemicals could jointly determine trade-offs of the adult preference and larval performance on these two hosts. As developing chemical-based technology is one of the main approaches for control of pest insects, our ndings may also contribute to the future efforts for management of the rice water weevil. the leaf area removed (cm 2 ) and the number of eggs between rice and barnyard grass were separately compared using the paired samples t-test. The volatile compounds identied (peak areas corrected by dry plant weight in grams) from rice and barnyard grass were analysed by a principal component analysis (PCA) to detect the differences between rice and barnyard grass volatiles. The PCA was conducted using the function prcomp in R, and further screened by partial least squares discriminant analysis (PLS-DA). We compared the 12-day fecundity of RWW female adults feeding on rice leaves or barnyard grass leaves using the independent samples t-test. We also compared the larval weights feeding on rice roots or barnyard grass roots using the Mann-Whitney U test, as the data were not normally distributed even after being transformed. The differences in nitrogen, C: N ratio, soluble sugar, soluble protein, cellulose, lignin, avonoid and oxalate contents in roots and leaves of rice and barnyard grass were separately compared by using the independent samples t-test. Data analyses were performed with R (v3.6.3) using the mixOmics and RVAideMemoire packages.


Introduction
Insect host selection behaviour and performance are affected by plant primary and secondary metabolites (Awmack and Leather 2002;Jakobs and Müller 2019;Wang et al. 2020). When insects locate and select potential host plants, volatile secondary metabolites emitted by plants can function as longdistance cues (Allmann et al. 2013;Dahlin et al. 2014;Liu et al. 2020). For example, plum psyllid (Cacopsylla pruni) can use plant volatiles to locate their favorable food resources from a distance (Gallinger et al. 2019). When insects come into contact with plants, nutritive and defensive metabolites can determine host plant suitability and subsequently affect their survival, growth, and reproduction (Gonçalves-Alvim et al. 2004;Michael 2018;Brzozowski et al. 2020). Therefore, simultaneously assessing the roles of plant volatiles, nutritive and defensive chemicals in determining the relationship between insect host selection and their performance is of the utmost importance to better understanding the underlying mechanism of host plant selection. This knowledge may also contribute to the development of environmentally sustainable integrated pest management strategies, such as breeding high resistant cultivars, developing push-pull technologies and volatile-based traps, given plan primary and secondary metabolites can signi cantly affect pest insect preference and performance.
Many previous studies have reported the relationships between insect adult preference (e.g., feeding and oviposition) and larval performance (e.g., growth development) (Wiklund 1975;Jaenike 1978;Fox 1993;Gripenberg et al. 2010). According to the preference-performance hypothesis (or the mother knows best hypothesis) (Jaenike 1978), female adults are expected to optimize their tness by laying eggs on the host plants on which their offspring will perform best. Findings from many previous studies supported this theory (e.g., Heisswolf et al. 2005;Zhang et al. 2012), however, some other studies also found that host plants preferred by parental adults for oviposition might not be the best hosts for development of their offspring larvae (Scheirs et al. 2000;Scheirs et al. 2004;Jiao et al. 2012;Smith et al. 2018). These adult-larva con icts, or trade-offs between adult preference and larval performance may be attributed to a lot of factors (Wiklund 1975;Fox 1993;Van Nouhuys et al. 2003;Merwin et al. 2020), such as variation in host suitability. For example, insect adults and larvae often feed on different plant tissues (leaves vs. roots) and these tissues are often spatially and temporally separated (Krebs and Davies 1997;Clark et al. 2011;Lee et al. 2016), different chemicals in leaves and roots may then differently affect adult preference and larval performance (Huang et al. 2013). Therefore, understanding chemical-mediated host plant selection in the context of trade-offs between adult preference and larval performance may help us to better understand the ultimate and proximate reasons for insect host plant selection and development.
Plant species differ in volatile, nutritive, and defensive chemicals, thus differently affecting insects and their interactions with host plants (Mattson 1980;Bezemer and van Dam 2005;Aartsma et al. 2019).

Different plants emit variable chemicals that attract different insects or different numbers of individuals.
It is also well-known that high contents of nitrogen, or low carbon/nitrogen in plants may favour insect development as of more proteins, while high content of lignin and celluloses can inhibit insect digestion (Awmack and Leather 2002;Kitajima et al. 2012;Armani et al. 2020). Moreover, avonoids and oxalate are often considered defensive chemicals in some crops, such as rice (Yoshihara et al. 1980;Nenaah 2013;Dai et al. 2019), and contents of these chemicals may differ between host species. Therefore, variations of these volatile, nutritive, and defensive chemicals in host plant species may determine the difference of adult host location behaviours, oviposition, and larval development, leading to potential trade-offs between adult preference and larval performance. Tests on this prediction is critical in understanding of the chemical-mediated insect host plant selection and performance.
Here, we report how host plant chemicals affect adult location behaviours and larval performance of the rice water weevil (abbreviated RWW), Lissorhoptrus oryzophilus (Coleoptera: Curculionidae). RWW is one of the most destructive pests on rice (Oryza sativa L.) in the world (Saito et al. 2005;Aghaee and Godfrey 2014). In many introduced ranges such as China, RWW undergoes parthenogenetic reproduction without mating and there has been no males invaded. Adults primarily feed on leaves and lay eggs on submerged areas of the leaf sheaths, and their larvae develop in the plant roots (Stout et al. 2002;Aghaee and Godfrey 2014). Larvae pupate on roots. L. oryzophilus has a wide range of host plants, including rice and barnyard grass (Echinochloa crusgalli L.). The insect occurs one generation per year in central China where we conducted our study. Previous studies have found that females performed better on barnyard grass, while their larvae survived better on rice (Tindall and Stout 2003). However, it remains unknown that how host plant volatiles, nutrients and defensive compounds affect host selection and performance when the optimal adult and offspring plants are different.
In this study, we focus on the roles of chemicals in host selection and performance in L. oryzophilus with rice (O. sativa) and barnyard grass (E. crusgalli). Speci cally, we rst compared the responses of L. oryzophilus adult females to volatiles from rice and barnyard grass in a Y-tube olfactometer. We also measured volatiles from rice and barnyard grass to compare the volatile composition difference. We then determined the feeding and oviposition preferences of L. oryzophilus adult females between rice and barnyard grass, as well as larval performance (i.e., weight) on these two hosts. Finally, we analysed the nutritional and defensive compounds in the roots of rice and barnyard grass to examine their effects on larval performance.

Plants and insects
We used the rice variety Shenliangyou 3117 for this experiment. This variety is widely grown in the Yangtze River basin. Rice seeds were purchased from Hubei Seed (Wuhan, Hubei Province, China). The barnyard grass seeds were collected from a rice eld (Shenliangyou 3117) near Wuhan city, Hubei province, China (31°54′ N, 115°24′ E). The seeds were sown in seedling trays with 72 plugs lled with commercial nutrition soil for germination. After 10 days, similar-sized seedlings were individually transplanted into plastic pots (25 cm height and 20 cm diameter) with a substrate of standard paddy soil and arranged in an outdoor common garden. Potted plants at the early tillering stage (32-36 cm height) were used for following experiments. To prevent herbivore damage, each pot was enclosed by a small nylon screen cage (60 × 60 × 60 cm). The RWWs for this study were collected from the same rice eld mentioned above and were reared on potted rice seedlings (Shenliangyou 3117) in an insectary at 32 ± 1°C , 70% relative humidity and 16:8 h photoperiod.

Adult feeding and oviposition preferences
To determine RWW adult preference in feeding and oviposition between rice and barnyard grass, we conducted two choice experiments in the summer of 2019 at Wuhan Botanical Garden, Chinese Academy of Science, Wuhan city, Hubei Province, China (30°32′ N, 114°25′ E). The choice arena was set up as follows. Two potted plants, one of each species (i.e., rice and barnyard grass), were carefully placed into basins (28 cm in height and 48 cm in diameter). Each basin was then lled with water until the water depth inside the basin was approximately 8 cm above the soil surface. To prevent the weevils from escaping, each basin was covered with a large screen cage (100 × 100 × 100 cm).
In the feeding preference experiment, we released ve RWW adults into each cage and allowed them to feed for 10 days. Then, we collected all the leaves of both plant species to measure total area and damaged areas of each leaf using a Vernier calliper (0.02 mm accuracy). This experiment was repeated 24 times.
In the oviposition preference experiment, we released six RWW adults into each cage for oviposition. After 5 days, we removed all adults and counted the number of eggs on each plant according to the method described in Stout et al. (2002). After cleaning the plants with running water, whole plants were placed in 95% alcohol for bleaching. The number of eggs on each plant was counted under a dissecting microscope. This experiment was repeated 15 times.

Y-tube olfactometer assays
To test the effect of volatile on adult preference, we carried out a choice experiment using a Y-tube olfactometer. The experimental setup and parameter settings are based on those in our previous work . Brie y, the Y-shaped glass tube consisted of a base tube (12 cm length and 2 cm diameter) and two 12-cm branching arms. The angle between the two arms was 90°. Each arm of the Ytube was connected through Te on tubing to a glass cylindrical container (90 cm height and 22 cm diameter) with the odour source. The Y-tube was placed horizontally during the experiments. The experiments were conducted under dark conditions in a climate-controlled room (28 °C and 70% relative humidity) to ensure air purity and exclude light effects.
RWW adults were starved for 24 h before the choice experiment, and only healthy and active adults were used. For each test, an adult was released into the downwind arm of the Y-tube and given 2 min to make a choice. A positive choice was recorded when the adult entered the right or left arm of the olfactometer and remained there for at least 10 s. If the adult failed to make a choice within 2 min, it was removed from the Y-tube and excluded from the statistical analyses. To avoid positional effects, we alternated between stimuli in each Y-tube arm for every test. The Y-tube olfactometer was also cleaned with alcohol between tests. Each adult was tested only once. In total, 60 female adults were recorded in this experiment.

Volatile collection and identi cation
To examine whether there is a difference in volatile composition and determine which active compounds RWW adults might use to locate their preferred host plant, we collected and then identi ed volatiles from the two host plants. Dynamic headspace methods were used to collect the volatiles as described by Turlings et al. (1998). Individual plants were bagged with polyethylene oven bags (406 × 444 mm; Reynolds, Richmond, VA, USA). Volatile collection lasted for 24 h and was replicated 10 times for each plant species. We also carried out the same procedure with empty oven bags (N = 10) to obtain negative controls. After collection, the volatiles were eluted from the adsorbent (80/100 mesh Porapak Q adsorbent, Sigma, USA) with 1.5 mL of dichloromethane (Sigma-Aldrich) for gas chromatography-mass spectrometry (GC-MS) analyses.
We conducted volatile analyses using a GC-MS system (GC-2010 Plus; Shimadzu Inc., Japan) equipped with a fused silica capillary column (Rxi-5 MS; Shimadzu Inc., Japan) according to Sun et al. (2019). We used the NIST08 MS spectral library database to identify the plant volatiles. We obtained the percentage of each compound by integrating their peak areas. We con rmed the identities of the compounds using chromatographic comparisons with commercial standards (Sigma, USA). We used principal component analysis (PCA) to analyse these data.

Adult oviposition in no-choice test
To examine how a single host plant affected RWW ovipositon, we conducted a test on the insect fecundity in a no-choice test. We reared and observed the weevils from individual ooded pots (25 cm height and 20 cm diameter) that each contained a plant. A single female adult from the colony was introduced to each pot. To con ne the adult to the plant, the plant in the pot was sealed with a nylon bag. The cylindrical nylon bag (30 cm diameter, 100 cm height) was made of white netting (0.02 mm mesh size) and secured with a string. Every 4 days, we collected the plant from each pot, counted the number of eggs on the plant, and transferred the adult to a new potted plant of the same species. The experiment was terminated after 12 days. During the experiments, if an adult weevil could not be found on a test plant, that replicate was removed, and a new replicate was established to ensure that 20 similar-sized adults would be tested for each plant species. The experiment was conducted in an insectary (32 ± 1 °C, 70% relative humidity and 16:8 h photoperiod).

Larval performance tests
To examine the effects of host plants on the growth of RWW larvae, we conducted a larval performance test with rice and barnyard grass in an insectary (32 ± 1 °C, 70% relative humidity and 16:8 h photoperiod). A female adult was transferred to each pot (n=15 for each species of rice and barnyard grass) and allowed to lay eggs for 1 day. The pot was covered with a cylinder (15 cm diameter and 50 cm height) with one end inserted into the soil to con ne the weevil. One day later, the adult weevil was removed. Twenty-ve days later, we carefully uprooted the plants from the pots and searched for larvae on the roots. Each larva was weighed using an electronic balance (Sartorius BS 110 S, Germany). In total, 23 and 21 larvae were examined on rice and barnyard grass, respectively.

Leaf and root chemicals
To determine primary and secondary chemicals in leaves and roots, we harvested plant materials to make measurement of total nitrogen, carbon, soluble sugars and avonoids, oxalates, lignin and cellhouse. The leaves and roots of the potted plants at the tillering stage were separately collected. Each pot contained 3-5 healthy individual plants, and 8 pots were used for each plant species, resulting in 8 leaf samples and 8 root samples for each plant species. The samples were rst divided into two parts before being weighed. One part of each sample was immediately frozen in liquid nitrogen and stored at −80 °C until the oxalate content analysis. The other part of each sample was oven-dried at 105 °C for 30 min and then oven-dried at 65 °C for 48 h for analysis of N, soluble sugar, soluble protein, cellulose, lignin and avonoid contents.
The oxalate concentration of each sample was analysed by high-performance liquid chromatography (HPLC) according to Libert (1981) and Xu et al. (2006). The concentrations of avonoid, soluble sugar, soluble protein, cellulose and lignin were analysed with an ultraviolet and visible spectrophotometer (UVS) (Thermo Scienti c GENESYS 10S, Waltham, MA, USA). The avonoid concentrations were analysed as described by published methods (Sun et al. 2016). The soluble protein and soluble sugar were determined as described by Bradford (1976) and Elleuch (2007), respectively. The plant cellulose and lignin contents in dry tissue were determined by the methods of Updegraff (1969) and Morrison (1972), respectively. The total nitrogen content of each sample were measured using an elemental autoanalyser (Vario MAX CN, Germany).

Statistical analyses
Data analyses were performed with R (v3.6.3). All data were tested for normality using the Shapiro-Wilk test before being compared. For the Y-tube olfactory assay, the proportions of RWW female adults on rice and barnyard grass were compared using Wilcoxon's signed rank test, as the data were not normally distributed even after being transformed. To test for the feeding and oviposition preferences of RWW female adults, the leaf area removed (cm 2 ) and the number of eggs between rice and barnyard grass were separately compared using the paired samples t-test. The volatile compounds identi ed (peak areas corrected by dry plant weight in grams) from rice and barnyard grass were analysed by a principal component analysis (PCA) to detect the differences between rice and barnyard grass volatiles. The PCA was conducted using the function prcomp in R, and further screened by partial least squares discriminant analysis (PLS-DA). We compared the 12-day fecundity of RWW female adults feeding on rice leaves or barnyard grass leaves using the independent samples t-test. We also compared the larval weights feeding on rice roots or barnyard grass roots using the Mann-Whitney U test, as the data were not normally distributed even after being transformed. The differences in nitrogen, C: N ratio, soluble sugar, soluble protein, cellulose, lignin, avonoid and oxalate contents in roots and leaves of rice and barnyard grass were separately compared by using the independent samples t-test. Data analyses were performed with R (v3.6.3) using the mixOmics and RVAideMemoire packages.
Y-tube olfactometer assays and volatile identi cation L. oryzophilus females showed a stronger preference for the volatiles from barnyard grass (E. crusgalli) than for those from rice (O. sativa) (V = 4.99, p < 0.001; Fig. 2a). In total, 105 volatile compounds were identi ed from barnyard grass and rice via GC-MS (electronic supplementary material, Table S1). The PCA plot indicated that the volatile compositions of rice and barnyard grass were signi cantly different (Fig. 2b). Twelve compounds were primarily responsible for the difference in volatile composition between rice and barnyard grass (electronic supplementary material, Table S1).

Leaf and root chemicals
There were signi cant differences in the nutrient and defensive chemicals in the leaves of rice and barnyard grass (Fig. 4a-d). Nitrogen contents were signi cantly higher in the rice leaves than in the barnyard grass leaves (t 13.77 = 9.08, d.f. = 1, p < 0.001; Fig. 4a Fig. 4h) contents between rice and barnyard grass leaves.

Discussion
In this study, we found RWW adults preferred barnyard grass over rice for feeding and oviposition while their larvae developed better on rice than on barnyard grass which are in line with previous studies (Tindall and Stout 2003). Our Y-tube tests on adult behaviours and analysis of volatile chemicals indicate that chemicals emitted by barnyard grass might attract more adults than those of rice. Moreover, our analyses of nutritive and defensive chemicals show rice roots had higher nitrogen and soluble sugar but lower lignin and cellhouse contents than barnyard grass, being more suitable host for larvae. Therefore, our study provides evidence showing chemicals affect trade-offs of RWW adult preference and larval performance on these two hosts, which is critical in understanding of the ecological interactions of this global pest insect and its important crop host.
In general, herbivorous insect adults use chemical cues to choose suitable host plants for feeding and oviposition (Trona et al. 2013;Knolhoff and Heckel 2014;Webster and Cardé 2016;Bertea et al. 2020).
Our olfactometer assays showed that RWW adults had a stronger preference for the volatiles from barnyard grass than for those from rice. Through volatile analysis, we further found that the types and contents of volatile compounds differed considerably between rice and barnyard grass. This suggests that RWW adults can distinguish differences in the quality and/or quantity of volatile compounds between barnyard grass and rice and that they rely on volatile cues to locate host plants. In the choice bioassay, we also found that RWW adults preferred to feed and lay their eggs on barnyard grass over rice.
Our adult performance tests on oviposition showed the same results (Fig. 3a), further con rming that barnyard grass is better host for adults than rice. Together, in this study volatile cues may provide RWW adults with reliable and easily assessed information to use to select high-quality host plants.
Similarly, we also found that the nitrogen and soluble sugar content were lower in barnyard grass leaves than in rice leaves. We are unclear how such differences are related to the high RWW fecundity on barnyard grass leaves. However, we found that the avonoid and oxalate contents were lower in barnyard grass leaves, and the lignin and cellulose contents were similar between two host plants. Previous studies reported that avonoids could negatively affect food consumption and utilization by adult insects (Nenaah 2013), thus high avonoid in barnyard grass leaves might have negative impact on RWW fecundity in our study. Similarly, oxalate and its calcium salts are widely existing in plants and often deter insect herbivory (Franceschi and Nakata 2005), thus rice varieties with high rich oxalate could have high resistance to insect herbivory (Yoshihara et al. 1980). Overall, in this study, the low defensive compounds such as avonoid and oxalate in barnyard grass leaves might explain the better adult performance on this host.
In contrast to adults, larval performance was better on rice than on barnyard grass in this study. Abundant evidence has shown that high nitrogen contents contribute to the enhancement of herbivore performance (Atijegbe et al. 2020;Eberl et al. 2020). Consistently, our results showed that the nitrogen and soluble sugar content were higher in rice roots than in barnyard grass roots, suggesting that rice roots would provide more nutrients to larvae than barnyard grass. Moreover, the low lignin and cellulose in rice roots might facilitate food digestion in the larvae, thus also improving their performance. In addition, in this study, avonoid and oxalate in roots appear to have less effects on RWW larvae than the effect of those chemicals in leaves on adults. Due to their limited mobility and food exploration ability, larvae likely have a stronger ability to detoxify plant defences than adults (Mason et al. 2019). We also acknowledge that some other chemicals such as lipids which were not measured in this study might affect RWW growth in this study. Finally, as RWW laid more eggs on barnyard grass, competition between larvae for food and space could potentially have affected their development. Future work is needed to explicitly reveal how these factors differently affect larval developments on the two host plants.
Adults often choose high-quality host plants for maximizing their offspring performance (Minkenberg and Ottenheim 1990;Thompson and Pellmyr 1991;Gripenberg et al. 2010;Heisswolf et al. 2005). Our ndings of the trade-offs between adult preference and larval performance on barnyard grass and rice are inconsistent with the "mother knows best" theory (Scheirs et al. 2000;Mayhew 2001). Previous studies show that genetic correlations, natal-habitat experience, natural host plant range and maternal effects can affect adult preference and offspring performance, leading to their trade-offs (Wiklund 1975;Fox 1993;Merwin et al. 2020). For RWW, such trade-offs on barnyard grass and rice were reported in several previous studies and the maternal effects were likely small (Tindall and Stout 2003;Tindall et al. 2004).
In our study, the differences in volatile, nutritive and defensive compounds between the two hosts could largely explain differences in the adult preference and larval performance. However, some other ecological and evolutionary factors may also affect the adult host selection and larval performance. RWW is native to North America where barnyard grass is its one of the ancestral host (Webb 1914;Lange and Grigarick 1959), however, rice, which is native to Asia, is a relatively novel host of this insect.
Therefore, lack of long-term co-evolution with rice may affect the adult host selection, even if rice is more nutritive and less defended than its ancestral host barnyard grass. In addition, since adult host selection is often affected by natural enemies, different predators or parastoids associated with the two hosts and their eld habitats may also determine adult behaviours. In this regard, future work is needed to reveal how evolution history and natural enemies affect the adult host selection and larval performance.
Moreover, in this study, our data could not show the linkages between the adult feeding and larval development, which may be mediated by the volatile cues and/or the herbivory-induced resistance that may vary between adults and larvae, future experiments focusing on these issues may also assist to explicitly unveil mechanisms behind these complex interactions.
In conclusion, this study shows chemicals determine the trade-offs of RWW adult preference and larval performance on barnyard grass and rice. Our ndings highlight the importance of considering plant nutrients, defences and volatiles for better understanding the relationship between adult selection and offspring performance. Since RWW has become one of the most important invasive pests worldwide affecting rice production, our ndings may also be useful in future efforts around integrated pest management. Our results on rice root chemicals will facilitate future selection and breeding for rice cultivars with high chemical resistance to belowground pests such as RWW larvae. Moreover, because RWW adults showed a strong tendency for volatiles of barnyard grass, it may be possible to use barnyard grass as a bait plant growing nearby rice to attract adults. We identi ed 12 volatile compounds that were responsible for the differences between the volatile compositions of rice and barnyard grass, further research is also needed to identify the speci c volatile compounds used for host selection by RWW adults. This will be a basis for further development of volatile-based trap to manipulate the pest populations in the eld.

Declarations
Author's contributions JD and WY conceived the idea and designed the experiments, QW conducted the experiments and analysed the data. JD and WY, WH drafted the manuscript. All authors revised and approved the manuscript.   Olfactory preferences (a) of L. oryzophilus female adults for host plant volatiles between rice (O. sativa) and barnyard grass (E. crusgalli) in the pairwise Y-tube olfactometer assays (*** p < 0.001, Wilcoxon's signed rank test). Principal component analysis (b) of volatiles emitted from rice and barnyard grass. The length and direction of the arrows indicate the relative contribution and correlation of each volatile compound, respectively. The ellipses represent 95% con dence intervals, in which one point denotes one sample (n = 10 for each plant species). Principal components 1 and 2 explained 50.1% and 13.3% of the variance, respectively.

Figure 3
Adult and offspring performance. (a) Twelve-day fecundity of L. oryzophilus female adults on rice (O. sativa) and barnyard grass (E. crusgalli) (*** p < 0.001, independent samples t-test). (b) L. oryzophilus larval weight (mg) on rice (O. sativa) and barnyard grass (E. crusgalli) (** p < 0.01, Mann-Whitney U test). Boxes represent inter-quartile ranges, the centerline inside the boxes represents the median value, the yellow diamond in the middle of box plot represents the mean value, and the tails represent the nonoutlier range and the empty dots represent outlier values. Each solid dot represents one replicate.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download. ElectronicsupplementarymaterialtableS1.docx