Introduction
Diabrotica speciosa (Germar) (Coleoptera: Chrysomelidae) is an insect native from South America. During adulthood, this species feeds on many crops, such as soybean (Glycine max L.; Fabales: Fabaceae), common bean (Phaseolus vulgaris L.; Fabales: Fabaceae), and squash (Cucurbita pepo L.; Cucurbitales: Cucurbitaceae). Larvae feed on roots of several crop plants such as maize (Zea mays L.; Cyperales: Poaceae), peanut (Arachis hypogaea L.; Fabales: Fabaceae), common bean, soybean, wheat (Triticum aestivum L.; Cyperales: Poaceae), forage turnip (Brassica rapa L.; Brassicales: Brassicaceae), and potato tubers (Solanum tuberosum L.; Solanales: Solanaceae) (Gassen, Reference Gassen1984; Cabrera Walsh, Reference Cabrera Walsh2003; Cabrera Walsh and Cabrera, Reference Cabrera Walsh, Cabrera, Jolivet, Santiago-Blay and Schmitt2004; Ávila et al., Reference Ávila, Bitencourt and Silva2019).
Adults of D. speciosa are popularly known as bean leaf beetle, and these insects may cause defoliation in soybean and bean (Medina et al., Reference Medina, Trecha and Rosa2013), besides several other crops (Link and Costa, Reference Link and Costa1978). Significant damage can occur if cotyledons of soybean plants are injured by D. speciosa, resulting in abnormal growth of plants or even leading to death (Martins et al., Reference Martins, Baldin and Marques2004). Notably, damages are greater in soybean preceded by second season maize crops (Sosa-Gómez et al., Reference Sosa-Gómez, Moscardi, Corrêa-Ferreira, Oliveira, Hoffmann-Campo, Panizzi, Corso, Bueno, Hirose, Gazzoni and Oliveira2010). Losses caused by larval feeding on maize roots have been significant throughout Brazil, especially in the continuous maize growing areas in southern Brazil where the soils are generally rich in organic matter and retain moisture more efficiently. Losses have been even greater in irrigated areas of southeastern and midwestern Brazil (Viana, Reference Viana2010).
Pest management strategies to control D. speciosa are restricted to application of chemicals. Adults are controlled by foliar sprays with synthetic insecticides, such as acephate, thiamethoxam, and lambda-cyhalothrin (Agrofit, 2022), which tend to lose efficacy after ~21 days, and therefore posterior reinfestation often occurs. To avoid this, farmers routinely perform several insecticide applications during the crop season (Arruda-Gatti and Ventura, Reference Arruda-Gatti and Ventura2003).
One control method within the precepts of Integrated Pest Management is the use of resistant plants. According to Rausher (Reference Rausher, Roitberg and Isman1992), plant resistance is any characteristic that has a negative effect on the development, survival, preference, and/or reproduction of a herbivore. Conversely, a susceptible plant may be preferentially consumed by insects, generally for being a food of superior quality. Overall, it is expected that in normal conditions resistant plants are less injured than susceptible plants (Smith, Reference Smith2005). An example of a resistant plant used in agriculture is the soybean commercial varieties containing Rag genes, such as Viking 2188AT12N (Rag1 and Rag2 genes), which confers resistance to aphids (Hanson et al., Reference Hanson, Bhusal, Lorenz and Koch2019).
Feeding non-preference was originally defined by Painter (Reference Painter1951) as a set of plant traits and insect responses that make a plant less suitable for insect feeding than another plant lacking these traits and the insect responses to them. A range of mechanisms may be associated with expression of non-preference resistance in a plant, including trichomes (Agrawal and Karban, Reference Agrawal, Karban, Tollrian and Harvandll1999; Dalin et al., Reference Dalin, Ågren, Björkman, Huttunen, Kärkkäinen and Schaller2008), flavonoids (Piubelli et al., Reference Piubelli, Hoffmann-Campo, Arruda, Franchini and Lara2003, Reference Piubelli, Hoffmann-Campo, Moscardi, Miyakubo and Oliveira2005; Treutter, Reference Treutter2006), and plant nutrition (Elden and Kenworthy, Reference Elden and Kenworthy1994; Poschenrieder et al., Reference Poschenrieder, Tolrà and Barceló2006). Therefore, the current study aimed to evaluate the feeding non-preference to D. speciosa in soybean genotypes, also investigating the role of trichome density and chemical compounds as mechanism of resistance.
Materials and methods
The experiments were performed at São Paulo State University (Universidade Estadual Paulista - UNESP), Jaboticabal, SP, Brazil, in laboratory under controlled climatic conditions (temperature: 25 ± 2 °C, relative humidity: 70 ± 10%, and photoperiod: 12L:12D h). Insects of D. speciosa used in the experiments were reared according to the methodology of Ávila et al. (Reference Ávila, Tabai and Parra2000).
Ten soybean genotypes were used, of which IAC 100, PI 227687, PI 274454, and DM 339 were assigned as resistance standards (Miranda et al., Reference Miranda, Braga, Lourenção, Miranda, Unêda and Ito2003; Lima and Lara, Reference Lima and Lara2004; Piubelli et al., Reference Piubelli, Hoffmann-Campo, Moscardi, Miyakubo and Oliveira2005; Costa et al., Reference Costa, Souza, Barbosa and Boiça Júnior2014a, Reference Costa, Ribeiro, Souza and Boiça Júnior2014b), and BR 16 as susceptible standard (Piubelli et al., Reference Piubelli, Hoffmann-Campo, Moscardi, Miyakubo and Oliveira2005). These genotypes were compared to five other soybean genotypes, namely BRSGO 8360, IGRA RA 626 RR, BRS Valiosa RR, Dowling, and IGRA RA 516 RR. Soybean seeds were provided by Embrapa Soja, Londrina, PR, Brazil. The experiments were carried out in free- and no-choice conditions.
Whole-plant test
Plants of the 10 soybean genotypes were grown in a greenhouse, in 300 mL plastic containers (Copaza, Içara, SC) filled with substrate composed by soil, sand, and organic fertilizer (manure) at a ratio of 3:1:1. The soil type used in the experiments was eutrophic dusky red latosol (Centurion et al., Reference Centurion, Andrioli, Marques Júnior and Marcgiori1995). After development of the first two expanded leaves (~15 days after plant emergence), plants of the 10 genotypes were brought to the laboratory where assays were carried out. In free-choice test, plants were placed equidistantly inside glass cages (40 cm length × 30 cm width × 30 cm height) where 10 seven-day-old D. speciosa adults were released per cage. In no-choice test, the plastic containers in which the plants were grown were covered with plastic containers of the same type that were sealed together with adhesive tape to avoid insect escape, thus forming a symmetrical cage. One insect was released per plant (replicate).
In free-choice test, the attractiveness was evaluated at intervals of 5, 10, 15, and 30 min and 1, 2, 6, 12, 24, 30, and 48 h after release of insects, and in no-choice test at intervals of 5, 10, 15, and 30 min and 1, 2, 6, 12, 24, 30, 48, and 72 h. Attractiveness was attributed to insects that were feeding on the leaflets. Leaf area consumed (LAC) was assessed on the two expanded leaves at the end of the experiments, assigning LAC percentages through visual evaluation accomplished by two researchers. Eight replicates were used per treatment for both free- and no-choice tests.
Leaf disk test
Plants of soybean genotypes used in this experiment were grown in 7-L plastic pots filled with substrate composed by soil, sand, and organic fertilizer (tanned bovine manure) at a ratio of 3:1:1. Aiming to evaluate attractiveness of soybean genotypes and leaf consumption by D. speciosa, leaflets of the upper part of 60-day-old plants were collected, since D. speciosa adults prefer to feed on young leaves (Medina et al., Reference Medina, Trecha and Rosa2013). Next, leaf disks (2.54 cm diameter) were obtained with the aid of a metal punch.
For the free-choice test, glass arenas 26.2 cm diameter × 5.0 cm height lined with moistened filter paper were used, where leaf disks of the 10 soybean genotypes were placed, and subsequently 10 seven-day-old adults obtained from a lab rearing colony were released per arena. Each glass arena consisted of a replicate in the free-choice test. In the no-choice test, Petri dishes 9.0 cm diameter × 1.2 cm height lined with moistened filter paper were used, and a single leaf disk of one of the 10 genotypes was used per replicate. In the no-choice test, a Petri dish was considered a replicate. One adult D. speciosa was released per Petri dish (replicate), and 10 replicates were used per treatment for both free- and no-choice tests.
In free-choice assay, leaf disk attractiveness was assessed at intervals of 1, 3, 5, 10, 15, and 30 min and 1, 2, 6, 12, and 24 h after release of insects. In no-choice test, attractiveness assessment was performed at intervals of 1, 3, 5, 10, 15, and 30 min and 1, 2, 6, 12, 24, 20, and 36 h after release of insects. Attractiveness was attributed to insects that were feeding on the leaflets. Upon finishing the experiments after 36 h, the LAC was assessed in each genotype using an electronic leaf area meter (Model 3000A, LI-COR®, Lincoln, NE).
Leaf trichome density of soybean genotypes
Counting of simple trichomes (Smith, Reference Smith2005) was carried out using a stereoscopic microscope (40 × magnification). Evaluations were done by examining two circular areas of 3.14 mm2 (radius = 1 mm), one to the left and another to the right of the leaf midrib, equidistantly from each other and from leaf margin, on the abaxial and adaxial surfaces of a fully expanded leaf of the upper part of plants. Ten replicates were tested for each genotype using both 15- and 60-day-old plants.
Foliar flavonoids content of soybean genotypes
Quantification of flavonoids content described below was performed at the Department of Chemistry of the Federal University of São Carlos. Plants of the 10 soybean genotypes were grown in a greenhouse, and their leaves were used for analysis of constitutive levels of flavonoids. Fully expanded leaves of the upper part of 30 plants (15- and 60-day-old plants) of each genotype were collected with the aid of scissors, labeled and brought to the laboratory in paper bags. Leaves were washed using demineralized water and Extran MA02 (Merck KGaA, Darmstadt, Germany) at 0.5%. The leaves were rinsed four times with demineralized water, and then dried in an oven at 45 °C for 48 h. Dried leaves were ground to a fine powder using an analytical knife mill (IKA, A10 Basic, Wilmington, NC, US). Then, 500 mg of each sample was weighted and transferred to a 15-ml falcon tube. The flavonoids were extracted with 10 ml of MeOH:HCl 0.5 M 80:20 (v/v) solution placed in ultrasound bath (USC1400 Unique, Indaiatuba, Brazil) for two hours, with temperature increasing from 25 to 40 °C from the beginning to the end of the extraction process. Samples were centrifuged at 12,000 × g (Eppendorf, 5810R, Eppendorf, Hamburg, Germany) for 15 min, at 25 °C. The supernatant was filtered in 0.2 μm mesh. The precipitate was once again extracted, using the same protocol. At the end, both supernatants were grouped, evaporated to dryness, resuspended in 5 ml of HPLC-grade methanol, filtered through a 0.22-μm microporous regenerated cellulose membrane filter (Macherey-Nagel, Düren, Germany), diluted, and conditioned in HPLC vials for analysis.
The flavonoids rutin, isoquercitrin, quercetin (flavonols), daidzin, daidzein (isoflavones), hesperidin, naringin, and naringenin (flavanones) were quantified using high-performance liquid chromatography (HPLC). The analyses were adapted from the methodology proposed by Klejdus et al. (Reference Klejdus, Vacek, Benesová, Kopecký, Lapcík and Kubán2007). Calibration curves were built through external standardization, in the concentration between 0.5 and 50 μg ml−1 showing a good linearity throughout all analyzed concentration range, and analytes. All determination coefficient values were higher than 0.994. The linearity and concentration were obtained by calculating the regression equation (y = ax ± b) by the least squares method, where: umbelliferone y = 422.3x + 322.2; daidzein y = 8131.0x + 7449.6; daidzin y = 2313.3x + 880.1; hesperidin y = 6410.3x + 1803.4; isoquercetrin y = 11,766.0x + 3171.8; naringenin y = 7926.0x + 6721.4; naringin y = 14,172.6x + 7501.7; quercetin y = 57,537.6x + 1528.7; and rutin y = 16,245.9x + 6073.2.
Quantitative analyses were performed using an Agilent 1200 series (Agilent Technologies, Santa Clara, USA), configured with a degasser G1322A, quaternary pump G1311A, autoinjector G1367B, compartment of thermostatic column G1316A and a detector of diode array (DAD) G1316A.
The chromatography method was optimized by gradient elution reducing the time of experiment and overlap of possible compounds, using a column Alltech Prevail® C18 100 × 2.1 mm i.d., 3 μm. The mobile phase consisted of 0.5% (v/v) acetic acid aqueous solution (solvent A) (deionized water, Direct-Q 8 UV Milli-Q, Millipore, Molsheim, France), and 0.5% (v/v) acetic acid in acetonitrile (J.T. Baker, Mexico) as organic solvent (solvent B). Gradient started with 75:25 (solvent A:B % ratio) for 6.0 min followed by sloping to 90% of solvent B in 0.5 min, which was maintained for another 13.5 min; total run time was 20 min. The temperature of column oven was kept at 35 °C, flow of 250 μl min−1 and injection volume of 5 μl. The output of the liquid chromatograph was connected to MS inlet after 1:10 splitting, where 25 μl min−1 were directed to a Mass Spectrometer with triple quadrupole analyzer API® 2000 (AB/MDS Sciex, Framingham, MA, USA).
Mass Spectrometer was operated with a Turbo Ion Spray ionization source by electrospray (ESI-MS/MS) and ionization in negative mode, and the remainder directed from the DAD. Nitrogen was used as the carrier, heater, and collision gas. Selected Reaction Monitoring (SRM) method was used in tandem mass spectrometry for identification and quantification of flavonoids after choosing appropriate precursor and product ions through Q1 full scan and product ion (Q3) analyses. The mass spectrometric parameters were optimized for each analyte using continuously infusing standard solution at the rate of 1 μg ml−1. The ion transition chosen for SRM were: umbelliferone (Q1 m/z 161→Q3 m/z 133); daidzein (Q1 m/z 253→Q3 m/z 132); daidzin (Q1 m/z 415→Q3 m/z 252); hesperidin (Q1 m/z 609→Q3 m/z 301); isoquercetrin (Q1 m/z 463→Q3 m/z 300); naringenin (Q1 m/z 271→Q3 m/z 151); naringin (Q1 m/z 579→Q3 m/z 151); quercetin (Q1 m/z 301→Q3 m/z 151); and rutin (Q1 m/z 609→Q3 m/z 300). Selection and tuning of transitions, as well as analyte-dependent parameters were performed using direct infusion of each flavonoid solutions in methanol at a concentration of 10 μg-ml−1. The optimized parameters were as follows: turbo ion spray temperature, 350 °C; ion spray voltage, – 4500 V; declustering potential (DP) – 75 V; enhanced potential (EP) – 10 V; collision energy (CE) – 50 V; collision cell entrance potential (CEP) – 20 V; collision cell exit potential (CXP) – 2.4 V; curtain gas (CUR) 25 psi; nebulizer gas (GS1) 40 psi; and heater gas (GS2) 50 psi. The software Applied Biosystems Analyst V.1.4.2 was used to control the LC–ESI-SRM system, for collection and analysis of the data.
Foliar nutrients content of soybean genotypes
Analysis of nutrients was carried out at the Laboratory of Soil Fertility of UNESP. Soybean plants were grown in a greenhouse, and their leaflets were used to quantify the inorganic nutrients. Leaflets of 15- and 60-day-old plants (30 plants per genotype) were cut with scissors, stored in paper bags, labeled, and brought to the laboratory where they were subjected to the same procedures described in the above section, including washing, rinsing, drying, and milling. The macronutrients nitrogen (N), phosphorus (P), potassium (K), calcium (Ca), magnesium (Mg), and sulfur (S), and the micronutrients boron (B), copper (Cu), iron (Fe), manganese (Mn), and zinc (Zn) were quantified according to methodology of Miyazawa et al. (Reference Miyazawa, Pavan, Muraoka, Carmo, Mello and Silva1999).
Experimental design and statistical analysis
A completely randomized design was used in all experiments. Data on attractiveness, LAC, and trichome density were checked for normality of residuals (Kolmogorov-Smirnov's test) and homogeneity of variances (Bartlett's test). Owing to the occurrence of non-normal and/or non-homoscedastic data, these were log(x + 5)-transformed and then subjected to one-way analysis of variance. When significant differences were found, means were separated by Tukey's test (α = 0.05). Statistical analyses were performed using the software Assistat 7.7 (Silva and Azevedo, Reference Silva and Azevedo2006).
Principal component analysis (PCA) was performed using the mean values of LAC, total trichomes, flavonoids (raw data at Online Resource 1), and nutrients (raw data at Online Resource 2) obtained from soybean genotypes at different stages with the objective of grouping the genotypes that displayed the highest similarity. The PCA was performed using the software Statistica 7.0 (Statsoft Incorporation, 2004).
Results
Whole-plant test
In the free-choice test, the least attractive genotypes (F = 4.13; df = 9, 90; P = 0.0003) were IAC 100, IGRA RA 626 RR, PI 227687, PI 274454, and BR 16, whilst the highest mean number of D. speciosa adults feeding on the plants was observed on BRSGO 8360. Regarding leaf consumption of D. speciosa, IAC 100 and PI 227687 were less consumed (F = 3.56; df = 9, 90; P = 0.0013) than IGRA RA 516 RR (Table 1).
Table 1. Number (mean ± SE) of Diabrotica speciosa adults attracted (average of attractiveness) and leaf area consumed (LAC, %) on leaflets of soybean genotypes (15-day-old plants), in free- and no-choice tests
ns, non significant.
Means followed by the same letter within columns are not significantly different by Tukey's test at 5% probability. For statistical analysis, data were transformed in log (x + 5).
In the no-choice test, no significant difference was recorded for soybean genotypes attractiveness to adult of D. speciosa (F = 1.16; df = 9, 90; P > 0.05). However, lower LAC was noted for the genotypes IGRA RA 626 RR and PI 227687 (F = 2.25; df = 9, 90; P = 0.0286) compared to BRS Valiosa RR, Dowling, and IGRA RA 516 RR (Table 1).
Leaf-disk test
No significant differences were found for attractiveness of soybean leaf disks to adult of D. speciosa in free-choice test (F = 1.69; df = 9, 90; P > 0.05). The genotypes IAC 100, DM 339, BR 16, PI 227687, and PI 274454 showed lower LAC (F = 5.66; df = 9, 90; P < 0.0001) by adult D. speciosa than the cultivars BRSGO 8360 and IGRA RA 626 RR (Table 2).
Table 2. Number (mean ± SE) of Diabrotica speciosa adults attracted (average of attractiveness) and leaf area consumed (LAC) on leaf disks of soybean genotypes (60-day-old plants), in free- and no-choice tests
ns, non significant.
Means followed by the same letter within columns are not significantly different by Tukey's test at 5% probability. For statistical analysis, data were transformed in log (x + 5).
In no-choice test, soybean leaf-disk attractiveness to D. speciosa was similar among genotypes (F = 1.07; df = 9, 90; P > 0.05). On the other hand, significant differences were observed for LAC (F = 2.40; df = 9, 90; P = 0.0174), with the genotypes IGRA RA 626 RR, Dowling, PI 227687, and PI 274454 exhibiting lower values, whereas the most preferred genotypes for adult D. speciosa feeding were BRSGO 8360 and IGRA RA 516 RR (Table 2).
Leaf trichome density of soybean genotypes
Trichome density differed among soybean genotypes, regardless of plant age. For leaflets of 15-day-old plants, the lowest trichome density was found on genotype IGRA RA 626 RR on both adaxial (F = 43.99; df = 9, 90; P < 0.0001) and abaxial (F = 33.00; df = 9, 90; P < 0.0001) leaf surfaces, as well as in total trichome density (F = 40.26; df = 9; P < 0.0001), not differing from genotype IGRA RA 516 RR (Table 3). On the other hand, the highest trichome density was shown by genotype BR 16, which exhibited similar numbers of trichomes compared to genotype Dowling.
Table 3. Number (mean ± SE) of trichomes/3.14 mm2 on leaves of soybean genotypes at different plant ages
Means followed by the same letter within columns are not significantly different by Tukey's test at 5% probability. For statistical analysis, data were transformed in log (x + 5).
On leaflets of 60-day-old soybean plants, significant differences were found for trichome density in the adaxial and abaxial surfaces and for total trichome density (Table 3). The lowest trichome densities on the adaxial region (F = 25.07; df = 9, 90; P < 0.0001) were observed for the genotypes Dowling and IGRA RA 516 RR, not differing from the genotypes BRSGO 8360, IGRA RA 626 RR, BRS Valiosa RR, and BR 16. Conversely, the highest trichome density on the adaxial surface was observed in PI 227687, which did not differ from genotype PI 274454. With respect to the abaxial surface, the lowest trichome density (F = 21.03; df = 9, 90; P < 0.0001) was found in Dowling, while the highest density was estimated on leaflets of PI 227687. For total trichome density (F = 24.38; df = 9, 90; P < 0.0001), the lowest mean was observed in genotype Dowling, whereas the highest mean was observed in PI 227687.
Multivariate analysis
Considering the PCA applied to variables collected in the whole-plant test (15-day-old plants), the first principal component (PC1) held 30.19% variability of the original variables, and the most influential parameters in group separation were LAC in free-choice test (−0.84), concentrations of the macronutrients Ca (−0.80) and S (−0.85), of the micronutrients Mn (−0.88) and Zn (−0.74), and of the flavonoid rutin (0.61) (fig. 1). The second principal component (PC2) concentrated 18.06% variability existent in the original variables, and the parameters with greater influence in group separation were the flavonoid daidzein (0.80) and trichome density (−0.65).
Fig. 1. Principal component analysis graph based on data of feeding non-preference assay to Diabrotica speciosa in soybean genotypes using whole 15-day-old plants. LAC15N-C: leaf area consumed in no-choice test; LAC15F-C: leaf area consumed in free-choice test; N: nitrogen; P: phosphorus; K: potassium; Ca: calcium; Mg: magnesium; S: sulfur; B: boron; Cu: copper; Fe: iron; Mn: manganese; Zn: zinc.
According to the results obtained from the whole-plant test (15-day-old plants), by considering the most influential factors (leaf consumption, nutrients etc.), the soybean genotypes were separated into four groups (fig. 1).
The genotypes PI 227687 and DM 339 were positioned in the first quadrant by displaying one of the lowest LAC in free-choice test; lowest contents of Ca; intermediate amounts of S, Mn, and Zn; and one of the highest rutin contents. IGRA RA 626 RR, IGRA RA 516 RR, and BRSGO 8360 were grouped together in the second quadrant, due to a higher LAC in free-choice test (mainly the latter two genotypes); intermediate amounts of Ca, Mn, and Zn; and the highest S content. BRS Valiosa RR, Dowling, and BR 16 were found in the third quadrant, displaying intermediate values of LAC in free-choice test; intermediate amounts of S; one of the highest means of Zn concentration and total trichome density. Lastly, the genotypes IAC 100 and PI 274454 were isolated in the fourth quadrant for possessing one of the lowest LAC in free-choice test; lowest S contents; and intermediate values of Ca content and total trichome density (fig. 1).
The PCA applied to variables evaluated in the leaf disk assay (60-day-old leaves) showed that PC1 concentrated 26.47% data variability existent in the original variables, and the parameters that most influenced group formation according to eigenvalues obtained from the variables (values in parentheses) were LAC in no-choice test (0.64), and concentrations of the macronutrient S (0.84), and concentrations of the micronutrients B (−0.84) and Fe (−0.73) (fig. 2). PC2 accounted for 23.63% data variability in the original variables, and the parameters with greater influence in group separation were total trichome density (0.77), and concentrations of the micronutrient Mn (−0.77), and of the flavonoid hesperidin (0.74). Thus, according to the PCA results for data obtained in the leaf-disk assay (60-day-old plants), when considering the most influential factors (leaf consumption, nutrients etc.), the soybean genotypes were separated into four groups according to similarity and dissimilarity among them (fig. 2).
Fig. 2. Principal component analysis graph based on data of feeding non-preference assay to Diabrotica speciosa in soybean genotypes using leaf disks from 60-day-old plants. LAC60N-C: leaf area consumed in no-choice test; LAC60F-C: leaf area consumed in free-choice test; N: nitrogen; P: phosphorus; K: potassium; Ca: calcium; Mg: magnesium; S: sulfur; B: boron; Cu: copper; Fe: iron; Mn: manganese; Zn: zinc.
The genotype DM 339 was isolated in the first quadrant and had intermediate values of S, B, Fe, Mn, total trichome density, and low hesperidin content. The genotypes IAC 100, PI 274454, and PI 227687 were positioned in the second quadrant, and displayed low LAC in no-choice test and the highest total trichome density. BRS Valiosa, Dowling, and BR 16 were grouped together in the third quadrant, and possessed high contents of B, Fe, and Mn. Lastly, IGRA RA 626 RR, IGRA RA 516 RR, and BRSGO 8360 occupied the fourth quadrant, showing high S content, low Fe content, and the lowest total trichome density (fig. 2).
Discussion
Feeding non-preference to adult D. speciosa was observed in both free- and no-choice assays by PI 227687 when testing both whole plants and leaf disks, and by PI 274454 when testing leaf disks. No other soybean genotype expressed non-preference in both free- and no-choice tests. For example, the genotype IGRA RA 626 RR was one of the least consumed by D. speciosa in the leaf disk no-choice test; however, in free-choice test, IGRA RA 626 RR showed one of the highest means of LAC.
In the whole-plant assay (15-day-old plants; fig. 1), genotypes grouping was very similar compared to the leaf disk experiment (60-day-old plants; fig. 2), with exception of line PI 227687, which was found isolated from the genotypes PI 274454 and IAC 100. As plotted in the PCA analysis (fig. 1), this probably happened due to a lower LAC in PI 227687 in the whole-plant test, and also for its high rutin content. Another difference between assays is that, in the whole-plant assay the genotype DM 339 was found in the same quadrant of PI 227687, most likely due to the low LAC in free-choice test, low Ca content, and high rutin content. Similarly, Veiga et al. (Reference Veiga, Rossetto, Razera, Gallo, Bertoletto, Medina, Tisselli Filho and Cione1999) reported higher rutin concentration in PI 227687 than in other genotypes, with similar content in genotype IAC 100, which was one of the least consumed genotypes in our study. In this context, Hoffmann-Campo et al. (Reference Hoffmann-Campo, Harborne and McCaffery2001) reported that rutin is likely associated with resistance of PI 227687 to Trichoplusia ni (Hübner) (Lepidoptera: Noctuidae). Rutin is a flavonoid that may cause feeding deterrence to insects, in addition to impairing their growth, delaying period of development, and decreasing survivorship because of its anti-nutritive effects (Stamp, Reference Stamp1994; Hoffmann-Campo et al., Reference Hoffmann-Campo, Ramos Neto, Oliveira and Oliveira2006; Silva et al., Reference Silva, Almeida, Moura, Silva, Freitas and Jesus2016).
In the PCA applied to variables evaluated in the whole-plant assay, the genotypes PI 227687 and DM 339 were clustered in the same group, and one of the major reasons for this is the low content of the macronutrient Ca. Calcium is an essential component of all living organisms and is present in most living cells (Clark, Reference Clark1958). This macronutrient is also an important secondary messenger, in the form of the ion Ca2+ in the hemolymph and tissues of insects, playing an important role in cell signal transduction, and is involved in developmental regulation and stress responses by combining with calcium binding protein (Wasserman and Fullmer, Reference Wasserman and Fullmer1989; Capozzi et al., Reference Capozzi, Casadei and Luchinat2006). Therefore, lack of Ca in resistant soybean genotypes may impair insect development, clarifying (at least partly) why D. speciosa adults rejected those genotypes for feeding.
In this same PCA analysis, the genotypes IAC 100 and PI 274454 formed another group. This occurred primarily due to their low content of the macronutrient S. Sulfur is a component of the vitamins biotin and thiamine, and is also a component of the enzyme acetil-coA; therefore, S is very essential to the Krebs cycle, influencing metabolism of lipids, carbohydrates, and proteins (Taiz et al., Reference Taiz, Zeiger, Møller and Murphy2017). These characteristics possibly justify the high intensity of pest attack on leaves with high S contents, since its higher availability can increase the concentration of sulfur amino acids in plants sap (Bastos, Reference Bastos1998). This was observed by Katzel and Moller (Reference Katzel and Moller1993), whom reported increased sulfur content after application of sulfur dioxide in Pinus sylvestris L. (Pinales: Pinaceae), with a subsequent elevation in concentrations of free amino acids in the sap, which favored the attack of Dendrolimus pini (L.) (Lepidoptera: Lasiocampidae). Similarly, Laine et al. (Reference Laine, Itamies, Orell and Kvist1994) reported higher abundance of 50 taxa of arthropods on Picia abies (L.) Karst. (Pinales: Pinaceae) when its leaves exhibited higher S and N concentrations. Thus, perhaps D. speciosa avoided feeding on the resistant genotypes also because of their poor nutritional value.
The PCA analysis detected other leaf nutrients that exerted influence in the genotypes grouping. A low Mn concentration was also found in leaves of genotype PI 274454, whereas an intermediate content of this micronutrient was detected in leaves of genotype IAC 100 (15-d old plants). This micronutrient is considered an important element for proper insect growth and development (Ito, Reference Ito1967) that is associated with chitin in insect mandibles and maxillae, increasing their hardness (Schofield et al., Reference Schofield, Nesson and Richardson2002; Cribb et al., Reference Cribb, Stewart, Huang, Truss, Noller, Rasch and Zalucki2008). Therefore, lack of Mn in leaves of genotype PI 274454 might impair insect survival and performance, and this nutritional absence is possibly associated with less D. speciosa defoliation, once a plant can express resistance due to the lack of essential elements (Smith, Reference Smith2005).
Zinc was detected at low concentrations in 15-day-old plants of the resistant lines PI 227687 and PI 274454. According to U.S. National Library of Medicine (2022), Zn is found in cells along the body of animals, and it is necessary for proper functioning of the immune system, playing an important role in cell division and growth, wound healing, and carbohydrate breakdown. This corroborates Diener et al. (Reference Diener, Zurbrügg and Tockner2015), whom speculated that zinc is an essential element for insect development. Furthermore, similarly to Mn, Zn is associated to chitin in the mandibles and maxillae of insects, contributing for their hardness (Schofield et al., Reference Schofield, Nesson and Richardson2002; Cribb et al., Reference Cribb, Stewart, Huang, Truss, Noller, Rasch and Zalucki2008). Therefore, low Zn content in the PI's lines is possibly associated with soybean resistance, negatively affecting insect development by malnutrition.
The results obtained by testing leaf disks of 60-day-old plants are in accordance with results reported by Lara et al. (Reference Lara, Elias, Baldin and Barbosa1999), which indicated that genotype PI 274454 displays feeding non-preference to D. speciosa, as manifested in free- and no-choice tests. These authors also mentioned that PI 274454 expressed feeding non-preference when its aqueous extract was applied in filter paper disks, which indicates that resistance in PI 274454 can be closely associated with allelochemicals. However, this genotype did not display resistance when tested as 15-day-old plants, what may suggest that qualitative and/or quantitative variation in allelochemical composition occurs in function of plant age in this and probably in other soybean genotypes. Interestingly, in most evaluations, no soybean genotype stood out concerning attractiveness to D. speciosa. Perhaps this is associated to the greater mobility that adult D. speciosa presents compared with less mobile insects, e.g., caterpillars, enabling D. speciosa to perform several feeding attempts.
The PCA showed that genotypes grouping was very similar considering the two assays performed. In the leaf disk assay (60-day-old plants) the PI lines and genotype IAC 100 were grouped in the same quadrant, representing the group of the resistant genotypes. These genotypes were less consumed by D. speciosa adults and showed a high trichome density. Piubelli et al. (Reference Piubelli, Hoffmann-Campo, Moscardi, Miyakubo and Oliveira2005) reported that the flavonoid genistin was identified in the genotypes PI 274454, PI 227687, and IAC 100, and its concentration in PI 274454 was ~ two-fold higher than in the other genotypes, which are also considered resistant to multiple insect pests. Thus, resistance of PI 274454 may also be associated with the presence and high concentration of genistin. In the same graph (fig. 2), the genotype DM 339 was isolated due to its moderate level of resistance. The other genotypes were classified as susceptible, and were separated in distinct groups according to similarities and dissimilarities between genotypes.
Considering the density of simple trichomes observed on leaves of soybean genotypes, the trend is that genotypes less consumed by D. speciosa display one of the highest total trichome density, as PI 274454 and PI 227687, while, overall, genotypes that had greater LAC displayed lowest densities of trichomes, as IGRA RA 516 RR. However, exceptions can be noted; for example, 15-day-old plants of the genotype Dowling had a greater trichome density, but were one of the most consumed by D. speciosa. Ootani et al. (Reference Ootani, Souza, Rodrigues, Silva, Melo and Aguiar2014), after investigating resistance of soybean genotypes to Cerotoma arcuata (Olivier) (Coleoptera: Chrysomelidae), reported that trichome density on genotypes MSOY-9988, ENGOPA-314, and A-7002 played an important role against the beetle feeding, with consequences in soybean grain yield. However, it is important to emphasize that, according to results obtained in the current research, simple trichomes are probably auxiliary structures in soybean resistance to the feeding of D. speciosa adults, potentially acting together with flavonoids, enzymes (e.g., proteinase inhibitors), and nutrients.
The genotype PI 227687 is the most used source of resistance to soybean defoliating insects worldwide, while genotype PI 274454 was similarly used in Brazil (Piubelli et al., Reference Piubelli, Hoffmann-Campo, Moscardi, Miyakubo and Oliveira2005). It is worth mentioning that the resistant lines PI 227687 and PI 274454, and the moderately resistant cultivar DM 339 were previously found to show antibiosis (Costa et al., Reference Costa, Souza, Barbosa and Boiça Júnior2014a) and oviposition non-preference (antixenosis) to D. speciosa (Costa et al., Reference Costa, Ribeiro, Souza and Boiça Júnior2014b), highlighting their potential for plant breeding. Also, it seems that resistance in PI 274454 is expressed only in older plants, since 15-day-old plants of this line did not display any resistance to D. speciosa in our study. Conversely, resistance in PI 227687 was observed in plants at both plant ages (15- and 60 day old).
Our study assessed different soybean genotypes, including commercial cultivars and breeding lines, and also investigated possible morphological and chemical traits associated with the resistance to D. speciosa. The PI lines expressed resistance to the pest, and therefore they may be valuable sources of resistance genes that can be introgressed in plant breeding programs aiming to develop commercial cultivars resistant to D. speciosa. This study added new and important information to the literature on soybean insect-resistance, providing more clues on the resistance mechanisms that could be focused in plant breeding.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/S0007485322000220
Data
Raw data are available upon request.
Acknowledgements
We thank Dr Clara Beatriz Hoffmann-Campo from Brazilian Agricultural Research Corporation for the donation of soybean seeds, and Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq) for granting a master's degree scholarship to the second author (grant number 132376/2011-3).
Author contributions
E. N. C., A. L. B. J. and M. R. F. conceived and designed research. E. N. C., B. H. S. S., M. R. F., B. P., and M. C. P. C. conducted experiments. E. N. C., A. L. B. J., and M. R. F. analyzed data. E. N. C. wrote the manuscript. All authors read and approved the manuscript.
Conflict of interest
The author(s) declare none.
