Antifeedant and insecticidal effects of alfalfa saponins in the management of the Japanese beetle Popillia japonica

Popillia japonica is a quarantine pest of priority interest for the EU, given its potentially important economic, social and environmental impacts. Alternative strategies to chemical methods are essential to limit its spread in newly infested areas with favourable climatic and environmental conditions. Saponins are biologically active molecules widely distributed in plants, displaying a well‐known repellent activity combined with a mortality effect against insects. In this context, saponins were extracted from alfalfa Medicago sativa, where medicagenic and zanhic acid glycosides and Soyasaponin I were the most abundant compounds and used in the laboratory and semi‐field experiments for treating leaves of susceptible host plants for P. japonica. Under laboratory conditions, a food deterrence effect and a significant mortality rate were observed using Corylus avellana leaves treated at increasing saponin concentrations, ranging from 1% to 5% w/v. Semi‐field condition experiment supported the food deterrence effect, as a significant food preference was observed for untreated plants of Vitis vinifera compared to treated plants. The promising results obtained suggest that alfalfa saponins could represent a potential eco‐friendly approach for Japanese beetle control.

Popillia japonica is usually a univoltine species, and generally, adults emerge in summer with a lifespan of about 40 days (Campbell et al., 1989;Fleming, 1972;Potter & Held, 2002). P. japonica is a polyphagous species that feeds on over 300 species with a wide variety of trees, shrubs, wild plants and crops (Campbell et al., 1989;Fleming, 1972;Ladd Jr, 1987Potter & Held, 2002). Plants are damaged by both adults and larvae: adults feed on leaves and fruits, whereas larvae feed just below the soil surface damaging the roots of host plants (Potter & Held, 2002).
The high reproductive rate, the pathway of introduction and the dispersal ability, as well as the polyphagous diet, have contributed to the classification of P. japonica as a quarantine pest of priority interest for the EU, according to Commission Delegated Regulation (EU) 2019/1702, given the potential economic, environmental and social impact for the European territory.
In the US, chemical methods are mainly used for the P. japonica control but with adverse effects on ecosystems, with the possibility of non-target effects (Althoff & Rice, 2022). For this reason, an ecofriendly approach is thus appreciated and encouraged, especially by the European Commission, which is pursuing a pesticide-reducing policy ('Farm to Fork' European Commission strategy -https:// ec.europa.eu/food/syste m/files/ 2020-05/f2f_actio n-plan_2020_ strat egy-info_en.pdf).
Given the current limited distribution of P. japonica in mainland Europe, sustainable and effective containment strategies are essential to limit its spread in newly infested areas (EFSA, 2018). So far, environmentally friendly approaches have been proposed using attract and kill strategies (Marianelli et al., 2018;Paoli et al., 2023), biological control agents such as entomopathogenic nematodes (Marianelli et al., 2018;Mazza et al., 2017;Simões et al., 1993;Torrini et al., 2020) and entomopathogenic fungi Benvenuti et al., 2019;Giroux et al., 2015). Nevertheless, the integrated pest management approach emphasizes a reduction in the use of synthetic pesticides and is crucial to increase the effectiveness of P. japonica containment with environmentally acceptable alternatives.
Saponins are biologically active glycosides widely distributed in plants, consisting of a sugar moiety linked to a hydrophobic aglycone (sapogenin) with a triterpenoid or a steroid structure (Hostettmann & Marston, 1995). These specialized metabolites play a central role in plant defence strategy preventing and deterring phytopathogenic agents, like fungi, bacteria and insects (Abdelrahman et al., 2017;Zaynab et al., 2021). It has been also reported that saponins have a significant effect on insects' reproduction, development, activity and mortality rate. Indeed, their repellent activity, combined with their ability to impair the digestion process, leads to a drastic reduction in food intake, which in turn leads to starvation and related health issues (Singh & Kaur, 2018). Moreover, saponins bind with a complex of cholesterol, causing the insects' ecdysial failure (De Geyter et al., 2012;Qasim et al., 2020;Taylor et al., 2004). is the most cultivated forage crop in the world and represents an important source for the extraction of these bioactive molecules (Tava et al., 2022). Saponins from this genus display fungicidal, molluscicidal, nematicidal, antibacterial, antiviral and antitumoral properties (Abbruscato et al., 2014;Avato et al., 2017;D'Addabbo et al., 2011D'Addabbo et al., , 2020Paparella et al., 2015;Tava & Avato, 2006).
Despite the above-mentioned significant effects on insect survival and feeding strategies, the effect of saponins against the major pest P. japonica has been poorly investigated. Kreuger and Potter (1994) highlighted that unripe holly fruits had a higher saponin content than ripe fruits, resulting in less desirable, probably also due to the low sugar level. By contrast, in another paper, Keathley and Potter (2008) reported no relationship between saponins and resistance and found that some of the more susceptible host plants, such as the sassafras Sassafras albidum (Nutt.) Nees, have a higher leaf content of saponins, sugar and tannins, comparing to resistant plants. Apart from these suggestions and correlational observations, to date, no direct experimental tests have investigated the effects of saponins on P. japonica.
In this study, we aimed to investigate alfalfa saponins regarding their possible effects as a repellent, feeding deterrent and/or biocidal capability at different concentrations against P. japonica adults under laboratory conditions and semi-field assay when applied to the leaves of two of the most susceptible host plants, hazelnut (Corylus avellana L.) and grapevine (Vitis vinifera L.).

| Extraction and purification of M. sativa leaf saponins
Alfalfa (Medicago sativa cv. Equipe) plants were grown in an open field at CREA-ZA (Lodi, Italy) and collected in July 2019. Leaf samples were oven-dried at 50°C for 2 days, ground and used for saponin extraction, following general procedures (Abbruscato et al., 2014;D'Addabbo et al., 2020;Tava et al., 2020). Briefly, about 800 g of powdered alfalfa leaves were first defatted with CHCl 3 in a Soxhlet apparatus and then saponins were extracted with 80% MeOH under reflux (48 h). After removing the solvent with a rotary evaporator, the residue was resuspended in 30% MeOH and loaded onto a 200 × 60 mm C18 column (Lichroprep RP-18, 40-63 μm; Merck, Darmstadt, Germany), preconditioned with 30% MeOH. Elution was first performed with 40% MeOH to remove some polar compounds, and then the saponins were eluted with 90% MeOH and dried under reduced pressure. The saponin mixture obtained (2.5% yield) was kept in airtight vials until used. The saponin mixture was dissolved in distilled water and used at the indicated concentration for different experiments conducted in laboratory and semi-field conditions.
Purified saponin mixtures were also characterized for their quantitative aglycone composition as previously reported by Tava et al. (2017). In brief, 5 mg of saponin mixtures were treated with 15 mL of 2 N HCl in 50% aqueous MeOH under reflux for 8 h. After cooling, MeOH was eliminated with a stream of N 2 and aglycones extracted with AcOEt (3 × 10 mL) and dried under N 2 .

Finally, gas chromatography-flame ionization detection (GC-FID)
and gas chromatography-mass spectrometry (GC-MS) (Clarus 500; Perkin-Elmer) analyses were performed on sapogenins as their methyl-trimethylsilyl derivatives. Gas-chromatographic analyses were performed with a 30 m × 0.32 mm, 0.25 μm i.d., DB-5 capillary column. Retention times and mass spectra data were compared with those of previously purified and identified sapogenins ).
Traps were emptied every 30 min and only active, apparently healthy and free of morphological defects adults were used. After collection, P. japonica adults were transferred to the quarantine room at CREA-DC in Florence (26°C, 51.6% HR, 8:16 L:D) for laboratory experiments. Males and females were separated by the observation of the tibial shape and tarsal length, according to EPPO (2006) and kept in plastic boxes with perforated lids. P. japonica adults were fed with fresh leaves of the host plant Corylus avellana and used for the experiments after 5 days. A water-soaked cotton swab was placed in each box to hydrate the adults during rearing. For the semi-field experiment, the adults were collected as above, transferred to plastic boxes to the nursery 'Vivai Scarlatta' (Biella, Piedmont region, Italy) and used within 1 h.

| Food preference in laboratory conditions
The food preference between treated and untreated leaves of C. avellana was evaluated under laboratory conditions through a choice test. Leaves were collected from untreated hazelnut plants and randomly selected for control and treatment. Two leaves with a comparable surface, one soaked in water and the other one in a 1% saponin-water solution (10 mg/mL of saponin mixture in water) for 3 s, were placed at the maximum distance from each other in a plastic box (27 × 18 × 10 cm) together with one single specimen (n = 10 males and n = 9 females). We were not able to use the same number of specimens for both sexes because one female did not look in good health. Before the experiment, each leaf was scanned to record its original surface. Water was supplied using a cotton swab soaked in water and placed on the leaf petiole to wet both the P. japonica adults and the leaf itself. After 24 h, leaves were collected and scanned to quantify leaf consumption. Images were analysed using ImageJ 1.52a software (Schneider et al., 2012), and for each replicate, the eaten surface was calculated for both leaves by estimating the ratio between the final surface and the original one. As the differences between the pairs were not normally distributed (Shapiro Wilk test: W = 0.519, p value = 7.451 × 10 −7 ), the non-parametric paired-samples Wilcoxon signed-rank test (statistic: V) was used to compare the ratio between the final surface and the original one in the control and the treated leaves. Data were analysed using R statistical software (version 4.1.3) (R Core Team, 2020).

| Deterrent and biocidal effects of saponins
Different concentrations of the saponin mixture were tested against P. japonica adults under controlled laboratory conditions in order to evaluate (1) the deterrent effect rating the percentage of leaf consumption and (2) the biocidal effect recording the dead individuals daily. The experimental unit consisted of a fresh leaf of C. avellana, soaked in water for 3 s for the control or in the saponin-water solution (concentrations 1%, 3% and 5%, corresponding to 10 mg/ mL, 30 mg/mL and 50 mg/mL, respectively) for the treatments and placed inside a plastic box (14 × 14 × 8 cm) containing 5 P. japonica adults. For each treatment, 10 replicates were carried out (a total of 50 adults for each of the four treatments). Before each treatment, each leaf was scanned to record its original surface. Water was supplied as described above. After 48 h, the leaves were collected and scanned to quantify leaf consumption and processed as previously described in paragraph 2.4, using ImageJ 1.52a software (Schneider et al., 2012). To assess potential long-term mortality, after the experiments, P. japonica adults were reared for 5 days and fed with untreated leaves moistened with cotton. Each box was checked every 24 h, and mortality was recorded daily.
As the assumption of normal distribution was not met by some of the data distributions (Shapiro Wilk test, p < 0.05), Kruskal-Wallis test (statistic: H) followed by post-hoc Dunn test (statistic: Z) with Benjamini-Hochberg correction for multiple comparisons were used to compare leaf consumption (i.e., the ratio between the eaten surface and the original one) and adult among mortality treatments (i.e., the control leaves and the treated leaves with 1%, 3% and 5% concentration of saponins). Data were analysed using R statistical software (version 4.1.3) (R Core Team, 2020).

| Food preference in semi-field conditions
To evaluate the food preference between saponin-treated and untreated plants, a semi-field experiment was conducted in the Vivai Scarlatta nursery using nine bug-dorm cages (75 × 75 × 130 cm). Florence to be scanned. The captured images were analysed using ImageJ 1.52a software. Contrary to the previous experiments, the original leaf surface was calculated after the experiment through the remaining leaf margin, as the leaves that were eaten by P. japonica adults retain the vascular system, and it was thus possible to determine the original leaf perimeter and the initial surface. A generalized mixed model (tweedie distribution) was used to assess the effect of treatment on the leaf area eaten, using this parameter as a dependent variable, cage ID and plant ID as random factors, and treatment and total leaf area as predictors. The interaction term between treatment and total leaf area was not significant (p = 0.996) and it was thus removed from the analyses. This choice was confirmed by the lower AIC/BIC values of the model without the interaction.

| Purification and structural elucidation of M. sativa leaf saponins
Purified saponins were obtained as a whitish powder in a pure grade (85%-90%) and their purity was first inspected by HPLC (Figure 1).

Medicagenic acid and zanhic acid were the most abundant detected
sapogenins within the saponin mixture extracted from M. sativa leaves, representing 52.2% and 31.5% of the total, respectively, followed by soyasapogenol B (12.1%), bayogenin (2.0%) and hederagenin (1.5%) (Figure 2). Alfalfa leaf saponin mixture was predominantly composed of bidesmosidic compounds, with two sugar chains attached to aglycons (Figure 2), accounting for more than 85% of the total saponins.

| Food preference in laboratory conditions
The food preference for all the tested individuals (n = 19) was evalu-

| Deterrent and biocidal effects of saponins
To evaluate the deterrent effect of the saponin mixture, the eaten surface of the treated leaves was compared with the untreated ones.

| Food preference in semi-field conditions
The laboratory choice test has shown that 3% saponin solution works as a deterrent as well as 5% one and we did not observe any difference between the leaves eaten surfaces treated with those concentrations, thus we decided to use the 3% solution in the semi-field experiments. We found an effect of the treatment on the amount of leaf area eaten, with greater leaf consumption on control plants than F I G U R E 3 (a) Boxplots (red for control and blue for treatments) represent the median and the interquartile range for the ratio between the eaten leaf surface and the total one. Different letters among treatments indicate significant differences to Dunn's Multiple Comparison  Figure 5a). We also found a significant effect on the total area, that is the larger the leaf, the greater the area eaten (Table 1, Figure 5b).

| DISCUSS ION
In this work, we have investigated the possible effect of alfalfa saponins as a repellent, feeding deterrent, and insecticide, under laboratory and semi-field conditions, against P. japonica adults. Alfalfa leaf saponin mixture was characterized through LC-MS/MS analysis and single compounds quantitated by HPLC analysis. The most abundant compounds were identified as bidesmosides of medicagenic acid (compounds 13 and 14) and Soyasaponin I (compound 18, a monodesmoside saponin of soyasapogenol B), representing 9.9%, 25.2% and 9.9% of the total saponins, respectively (Figures 1 and 2).
The activities of alfalfa saponins have been reported against some insect pests belonging to Hemiptera, Coleoptera and Lepidoptera but data are scanty and mainly concerning stored products (Singh & Kaur, 2018;Zaynab et al., 2021). The most observed effects are increased mortality, lowered food intake, weight reduction, delayed development and decreased reproduction (Qasim et al., 2020;Tava & Avato, 2006;Zaynab et al., 2021). Among various tested saponins, 3-GlcA-28-AraRhaXyl-medicagenic acid (compound 14, Figure 2), the most abundant compound in our saponin mixture is reported to be potentially responsible for insecticidal activity (Singh & Kaur, 2018). However, as suggested by Li et al. (2022), Tian et al. (2021 and Ye et al. (2023), also some other minor constituents with particular functional groups could act as pesticides, but further research is needed to solve this topic. To the best of our knowledge, the employment of M. sativa saponins in the control of the P. japonica pest has not been investigated so far, while few works against other Coleoptera have been reported. Saponin extract from alfalfa roots has been used as a growth regulator in Tropinota squalida, disturbing the development and reproduction (Hussein et al., 2005). Moreover, saponins extracted from three Medicago species were included in the diet of Colorado potato beetle larvae showing a food deterrent and insecticidal effect (Szczepaniak et al., 2001;Szczepanik et al., 2004).
Several repellent and/or antifeedant deterrents have been tested in laboratory choice tests and field experiments against P. japonica.
Among the plant-derived molecules, the cardenolide glycoside neriifolin showed a feeding deterrent effect for 5 days on soybeans in laboratory experiments (Reed et al., 1982) as well as the cucurbitacin B used in choice test with primrose leaves (Tallamy et al., 1997).
Neem-based products have been also used both in laboratory tests and greenhouse experiments, proving to be effective against P. japonica using different plant hosts, such as soybean, linden, rose and sassafras (Held et al., 2001;Ladd Jr et al., 1978;Mmbaga & Oliver, 2007). leaves with respect to the treated ones. Moreover, the deterrent effect of the saponin mixture was assessed using different concentrations (1%, 3% and 5%) based on previously reported data on alfalfa saponins tested against leafhoppers and Colorado potato beetle at concentrations of 5% and 0.5% respectively (Qasim et al., 2020;Szczepaniak et al., 2001;Szczepanik et al., 2004).

F I G U R E 4
Therefore, we evaluated the eaten leaf surface by observing that the consumed surface was lower as the concentration of the treatment increased, even if no statistical differences were highlighted between 1% and 3% of treated leaves, showing that the adults dislike leaves treated with 5% saponins over the other concentrations. The survival of adults of P. japonica was monitored for 5 days, observing that it was significantly higher in the control than in the treatments, without significant differences among saponin concentrations.
Finally, a semi-field experiment was conducted on Vitis vinifera, which is one of the most favourite host plants of P. japonica, to evaluate the food preference between treated and untreated leaves. We decided to use the saponin mixture at 3% concentration since in the laboratory choice tests no significant differences have been highlighted with respect to the highest concentration used. In addition, it may last longer on leaves in semi-field conditions and is still in the range of potential feasibility in terms of costs. With this experiment, we confirmed that saponins from M. sativa act as a feeding deterrent against P. japonica also in semifield conditions.
Our study aimed to increase the knowledge about saponins as an ecologically sound and safe method for controlling the Japanese beetle. Our results, using saponins extracted from alfalfa, showed a food deterrence effect and a significant mortality rate. These promising results suggest that saponins could represent a potential eco-friendly approach for the control of this pest and require further research to deeply investigate this topic. Writing -review and editing. Giulia Torrini: Conceptualization; investigation; writing -review and editing.

ACK N O WLE D G E M ENTS
The authors thank Enrico Furno, the owner of the nursery Vivai

Scarlatta in Biella and Giovanni Bosio, Agostino Strangi and Claudia
Benvenuti for the field support. The authors thank B. Pintus of CREA, Lodi, Italy for her technical support for saponin analyses.

CO N FLI C T O F I NTE R E S T S TATE M E NT
All authors declare that they have no conflicts of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are openly available