Introduction
The stem borer Busseola fusca (Fuller) (Lepidoptera: Noctuidae) is an important pest of maize [Zea mays L. (Poaceae)] and sorghum [Sorghum bicolor (L.) Moench (Poaceae)] in Sub-saharan Africa (Kfir et al., Reference Kfir, Overholt, Khan and Polaszek2002). Busseola fusca is oligophagous having a narrow host range consisting of a few grass species. Most field recoveries have been from cultivated cereals, specifically from maize and sorghum, but few from wild grasses (Le Ru et al., Reference Le Ru, Ong'amo, Moyal, Muchugu, Ngala, Musyoka, Abdullah, Matama-Kauma, Lada, Pallangyo, Omwega, Schulthess, Calatayud and Silvain2006a , Reference Le Ru, Ong'amo, Moyal, Ngala, Musyoka, Abdullah, Cugala, Defabachew, Haile, Matama-Kauma, Lada, Negassi, Pallangyo, Ravololonandrianina, Sidumo, Omwega, Schulthess, Calatayud and Silvain b ; Ong'amo et al., Reference Ong'amo, Le Ru, Dupas, Moyal, Muchugu, Calatayud and Silvain2006; Ndemah et al., Reference Ndemah, Schulthess, Le Ru and Bame2007). This might be due in part to differences in silicon (Si) content.
The grasses are generally richer in Si than many other plants and particularly than dicotyledonous plants (Epstein, Reference Epstein1999). Silicon is now well admitted to be a beneficial element for plant growth, characterized by helping the plants to overcome various stresses including biotic and abiotic stresses (Epstein, Reference Epstein1999; Richmond & Sussman, Reference Richmond and Sussman2003; Ma, Reference Ma2004; Ma & Yamaji, Reference Ma and Yamaji2006). For example, Si mediates plant resistance to insect herbivores by a Si barrier providing mechanical resistance to insect feeding (Djamin & Pathak, Reference Djamin and Pathak1967; Peterson et al., Reference Peterson, Scriber and Coors1988; Ma, Reference Ma2004; Kvedaras & Keeping, Reference Kvedaras and Keeping2007; Kvedaras et al., Reference Kvedaras, Byrne, Coombes and Keeping2009) and has been shown to disturb significantly the larval performance of Lepidoptera stem borers such as Sesamia calamistis (Lepidoptera, Noctuidae) and Eldana saccharina (Lepidoptera, Pyralidae), important pests of maize in West Africa and sugarcane in South Africa, respectively (Sétamou et al., Reference Sétamou, Schulthess, Bosque-Pérez and Thomas-Odjo1993; Keeping & Meyer, Reference Keeping and Meyer2002; Kvedaras & Keeping, Reference Kvedaras and Keeping2007; Kvedaras et al., Reference Kvedaras, Keeping, Goebel and Byrne2007) by an increase in leaf abrasion, which subsequently increases wear on insect mandibles, and may physically deter larval feeding (Raupp, Reference Raupp1985; Massey et al., Reference Massey, Ennos and Hartley2006; Keeping et al., Reference Keeping, Kveradas and Bruton2009; Kvedaras et al., Reference Kvedaras, Byrne, Coombes and Keeping2009). It is well known that Si level in plant is not only dependent to Si content in soil but also to plant genotype (Epstein, Reference Epstein1999; Keeping & Meyer, Reference Keeping and Meyer2006). In that context, cultivated grasses can be characterized by a lower Si level than cultivated ones such as maize? Such situation can explain in part why B. fusca has the tendency to avoid wild habitat with wild grasses, becoming an important pest of maize?
The purpose of this study was to test that hypothesis attempting to link the Si level of different grass species with B. fusca larval performance. Seven grass species, mostly present in the natural habitat where B. fusca occurred (Le Ru et al., Reference Le Ru, Ong'amo, Moyal, Muchugu, Ngala, Musyoka, Abdullah, Matama-Kauma, Lada, Pallangyo, Omwega, Schulthess, Calatayud and Silvain2006a ; Ong'amo et al., Reference Ong'amo, Le Ru, Dupas, Moyal, Muchugu, Calatayud and Silvain2006), were used. In complement, the direct influence of Si on larval performance was verified using artificial diets and potted maize plants amended with Si.
Materials and methods
Insects
The B. fusca larvae used in this study were from a colony maintained on a meridic diet according to Onyango & Ochieng’-Odero (Reference Onyango and Ochieng’-Odero1994) at the International Centre of Insect Physiology and Ecology (icipe), Nairobi, Kenya. To rejuvenate the colonies, feral individuals collected from maize fields in western Kenya were added thrice a year. For all experiments, 1-day-old neonates were used.
Plants
Maize (cultivar 511) and six wild Poaceae species were used in this study: wild sorghum, Sorghum arundinaceum (Desv.) Stapf., Napier grass, Pennisetum purpureum Schumach, Arundo donax L., Setaria megaphylla (Steud.) Th. Dur. & Schinz, Panicum maximum Jacq. and Panicum deustum Thunb. Among these wild grasses, B. fusca has been sporadically or once recovered on P. purpureum, A. donax, S. megaphylla and P. maximum but never on P. deustum (Le Ru et al., Reference Le Ru, Ong'amo, Moyal, Muchugu, Ngala, Musyoka, Abdullah, Matama-Kauma, Lada, Pallangyo, Omwega, Schulthess, Calatayud and Silvain2006a ; Ong'amo et al., Reference Ong'amo, Le Ru, Dupas, Moyal, Muchugu, Calatayud and Silvain2006). Maize was grown in 4-litre pots (one plant per pot) from seeds provided by Simlaw, Kenya Seed Company, Nairobi. The grasses were obtained from natural habitats in Kenya and grown from tufts or stem cuttings in 4-litre pots (one tiller per pot). All plants were grown in the same type of soil in a greenhouse at icipe. The environmental conditions were about 31/17°C (day/night) with a 12:12 h (L:D) photoperiod. The plants were watered three times weekly, once with a complete nutrient solution. For all experiments, 3-week-old plants were used, except for S. arundinaceaum where 5-week-old plants were chosen. These growth stages were found to be well accepted by B. fusca female for oviposition (Calatayud et al., Reference Calatayud, Juma, Njagi, Faure, Calatayud, Dupas, Le Ru, Magoma, Silvain and Frérot2008a , Reference Calatayud, Guénégo, Ahuya, Wanjoya, Le Ru, Silvain and Frérot b , Reference Calatayud, Ahuya, Wanjoya, Le Ru, Silvain and Frérot c ).
Larval development on different plant species
Potted plants were placed in a greenhouse avoiding any contact with neighbouring plants. A group of 30 B. fusca neonates was inserted by a camel brush in a 1 ml-Eppendorf tube, which was used for infestation by dropping directly the 30 neonates inside the whorl of each potted plant. The border of each pot was lined with a thin layer of petroleum jelly to prevent migration of older larval stages, though this did not prevent emi- or immigration via ‘ballooning-off’ of B. fusca neonates (Kaufmann, Reference Kaufmann1983). For each plant, assessment of percent live larvae remaining and relative growth rates of the larvae (RGR) were assessed at 7, 19 and 31 days after infestation (DAI). RGRs were calculated by subtracting the average weight per larva on each plant at infestation from average weight per larva on the same plant at 7, 19 and 31 DAI divided by the number of DAIs (Ojeda-Avila et al., Reference Ojeda-Avila, Woods and Raguso2003). For each plant species, there were 12 replications.
Analysis of Si contents in plant tissues
The young B. fusca larvae (from neonates to 3rd instars) feed on young and tender leaves. It is only from 3rd instars that larvae migrate to the lower parts of the plant where they penetrate into the stem (van Rensburg et al., Reference van Rensburg, Walters and Giliomee1987). However, larvae can remain in plant whorls, especially in older (6–8 weeks old) plants up to the 4th instar to feed on leaves (van Rensburg, Reference van Rensburg1997). In this context, the Si analysis of plant tissues was done from young fully expanded leaves.
The leaves were thoroughly cleaned with ultrapure water and subsequently harvested. The leaves were sliced into small pieces, put into plastic containers and freeze dried in a stoppering tray drier (Labconco-Germany). The dried leaves were then milled into a fine powder.
The dilute hydrofluoric acid (HF) extraction and spectrometric molybdenum-yellow method of Saito et al. (Reference Saito, Yamamoto, Sa and Saigusa2005) was used to determine Si content of leaf tissue. Briefly, 100 mg of dry weight of the powder of each plant species was digested in 2 ml of HF solution of 1.5 M HF – 0.6 M HCl. Digestion was carried out at room temperature for 1 h with occasional stirring at 10 min. intervals. After digestion, 8 ml of distilled water was added to each sample and the resulting mixture homogenized by vortexing before being allowed to settle down for further 20 min.
To obtain powder of Si for standard solutions, about 50 ml of pure sodium silicate (Sigma-Aldrich, 338443) was heated at 950°C for 2 h, then cooled and ground into a fine powder. In that temperature the Na2SiO3 was pulverized and the resulting powder contained only Si (Saito et al., Reference Saito, Yamamoto, Sa and Saigusa2005). The same powder was also used for artificial diet amendments. Thereafter, a Si stock solution was prepared by digesting 0.1 g of silica in 20 ml of a 0.3 M HF – 0.12 M HCl solution and then topped to 100 ml with distilled water to give a stock solution of a concentration of 1 mg ml−1. Standard solutions ranging between 0 and 1 mg l−1 were prepared by dilution from this stock solution.
For spectrophotometric determinations of Si, 100 μl of each sample was mixed with 2 ml of 0.1 M boric acid solution. Two millilitres of a molybdenum solution (0.025 M Mo–0.4 M H2SO4–0.25 M BH3O3) was then added to each sample, mixed and allowed to stand for 5 min to allow complete formation of molybdenum yellow complex. Four millilitres of 0.1 M citric acid was added and thereafter the mixture was vortexed. Optical densities were measured in a 5 ml cell on a Beckman DU 640 spectrophotometer at 400 nm after 4–10 min following citric acid addition. The amount of Si in the samples was calculated from a calibration curve generated using standard solutions. All reagents and solutions for this experiment were prepared in Si-free plastic wares soaked in 0.1% HF solution prior to the analyses.
Larval development on artificial diets amended with increasing Si levels
The aforementioned Si powder obtained for standard solutions was added to the artificial diet by Onyango & Ochieng-Odero (Reference Onyango and Ochieng’-Odero1994) at a rate of 0–20 mg ml−1 diet corresponding to a rate of 0–2% Si in the diet. These Si contents were similar to those found in the plant species used in this study (Table 1). Each diet type was dispensed in heat-sterilized glass vials (7.5 cm long × 2.5 cm diameter) with 12 vials (replications) per Si concentration. Each vial was inoculated with four B. fusca neonates tightly fitted with a cotton wool and kept at 80% relative humidity and a 12:12 h (L/D) photoperiod. The larvae were allowed to feed ad libitum and diets were replaced as necessary. The live larvae remaining on each diet were counted and weighed after 7, 19 and 31 days of rearing. Percentage of live larvae remaining per vial and the RGR were calculated.
Table 1. Percentage of live larvae remaining of B. fusca (mean 1 ± SE, n = 12) and RGR (mg d−1, mean 1 ± SE) after 7, 19 and 31 days of infestation on plants, and percentage of Si (mean 1 ± SE, n = 34) in leaves of the different Poaceae species used.

1 Means within a column followed by different letters are significantly different at 5% level (Tukey's contrasts test following ANOVA).
Larval development on potted plants amended with soluble Si
Similar to Sétamou et al. (Reference Sétamou, Schulthess, Bosque-Pérez and Thomas-Odjo1993) and Nabity et al. (Reference Nabity, Orpet, Miresmailli, Berenbaum and DeLucia2012), the plants were treated with soluble Si in the form of sodium metasilicate (Na2SiO3.5H2O) applied on soil for plant absorption by roots. The control plants were treated by sodium carbonate instead of sodium metasilicate since sodium was also present in Si-treated plants. Maize, which was the most suitable plant species according to the results in Table 1, was used in this experiment. Two-week-old plants were amended three times with 0.56 g sodium carbonate (control = absence of Si) or 0.56 g of sodium metasilicate (=Si-amended plants) per plant; this was the same rate used by Sétamou et al. (Reference Sétamou, Schulthess, Bosque-Pérez and Thomas-Odjo1993) who studied the influence of Si-treated maize plants on Sesamia calamistis Hampson (Lepidoptera: Noctuidae) survival. After 2 weeks, plants were individually infested with 30 neonates. For each plant, the percentage of live larvae remaining and RGR were assessed at 19 DAI (i.e., the minimum time required to get a significant influence of %Si on larval development under artificial diet conditions; see Table 2). Thereafter, each dissected plant was used for Si content analyses as mentioned before to determine if Si-amended plants exhibited higher Si contents than non-amended plants.
Table 2. Percentage of live larvae remaining of B. fusca (mean 1 ± SE, n = 12) and RGR (mg d−1, mean 1 ± SE) after 7, 19 and 31 days of rearing on artificial diets with increasing percentage of silica.

1 Means within a column followed by different letters are significantly different at 5% level (Tukey's contrasts test following ANOVA).
Statistical analysis
Data on plant Si content and RGR were log-transformed whereas data on % live larvae remaining were arcsine-transformed. Such arcsine transformation is appropriate for the data on proportions, i.e., data obtained from a count and the data expressed as percentages (McDonald, Reference McDonald2014). Untransformed results are presented in tables. All transformed data were subjected to Levene's test for homogeneity of variance with respect to treatments. Most of the data fulfilled Levene's data. Following this, repeated measures of multivariate analysis of variance (MANOVA) were conducted to compare the treatment effects on RGR or % of live larva remaining. In case of significant F-values of one-way ANOVA, the data were subjected to post hoc Tukey's contrast tests. The data obtained on responses of B. fusca to % Si in maize plants after 19 days of infestation on untreated (control) and Si-treated maize plants (Table 3) were analysed by a Student's t-test. Spearman rank-order correlations (ρ) were used to assess the relationships between Si contents in plants and % survival and RGR on different plant species. All the statistics were done using R (http://www.r-project.org/).
Table 3. Percentage of live larvae remaining of B. fusca (mean 1 ± SE, n = 12) and RGR (mg d−1, mean 1 ± SE) after 19 days of infestation on untreated (control) and Si-treated maize plants

1 Means within a column followed by different letters are significantly different at 5% level (Student's t-test).
Results
Larval development and Si contents of different plant species
The both factors studied, plant species and duration of infestation, induced a significant variation of the overall parameters evaluated on B. fusca larvae (AMOVA: F 6,140 = 18.253, P < 0.0001 for plant factor [A]; F 2,140 = 40.725, P < 0.0001 for duration of infestation factor [B]; F 8,140 = 15.718, P < 0.0001 for the intercept A × B). The percentage of live larvae remaining differed significantly between plant species within each duration of infestation (Table 1, ANOVA: F 6,77 = 3.585, P = 0.00347; F 6,77 = 10.68, P < 0.0001 and F 6,77 = 4.029, P = 0.00146 after 7, 19 and 31 days, respectively) with highest values at all sampling dates on maize and wild sorghum. On the other five grasses, this percentage was considerably lower and no larvae were found on P. deustum and S. megaphylla beyond 7 days. Similarly, RGR was significantly higher on maize and wild sorghum than the other grasses, after 7, 19 and 31 days of infestation (Table 1, ANOVA: F 6,57 = 25.72, P < 0.0001; F 4,49 = 6.086, P = 0.000467 and F 4,34 = 484.7, P < 0.0001 after 7, 19 and 31 days, respectively).
Silicon content of leaves differed significantly among plant species (Table 1, F 6, 231 = 109.2, P < 0.0001). Panicum maximum and P. deustum had double the levels of Z. mays and S. arundinaceum. Setaria megaphylla, P. purpureum and A. donax yielded similar and intermediate Si contents (Table 1). Plant Si was negatively related with % live larvae remaining and RGR of B. fusca larvae (for % live larvae remaining: ρ = −0.56; P = 0.0002; ρ = −0.55 P = 0.0003 and ρ = −0.43; P = 0.0066 after 7, 19 and 31 days, respectively; for RGR: ρ = −0.35; P = 0.0268; ρ = −0.19; P = 0.2486 and ρ = −0.66; P < 0.0001 after 7, 19 and 31 days, respectively).
Larval development on artificial diets amended with increasing levels of Si
The both factors studied, %Si in the diet and duration of feeding, induced a significant variation of the overall parameters evaluated on B. fusca larvae (AMOVA: F 4,164 = 10.451, P < 0.0001 for %Si in the diet factor [A]; F 2,164 = 74.817, P < 0.0001 for duration of feeding factor [B]; F 8,164 = 2.576, P < 0.0001 for the intercept A × B). However, although highly variable between diets the percentage of live larvae remaining did not differ significantly with silica level in the diet at 7, 19 and 31 DAI (Table 2, ANOVA: F 4,55 = 0.679, P = 0.609; F 4,55 = 1.726, P = 0.157 and F 4,55 = 1.475, P = 0.222 at 7, 19 and 31 DAI, respectively). RGR did not also differ significantly at 7 DAI, but it decreased significantly with increasing %Si in the diet at 19 and 31 DAI (Table 2, ANOVA: F 4,54 = 1.434, P = 0.235; F 4,55 = 5.502, P = 0.000847 and F 4,55 = 18.13, P < 0.0001 after 7, 19 and 31 days, respectively).
Larval development on Si-amended plants
Si-amendments significantly increased %Si in maize plants (Table 3; t = −2.4, df = 37.6, P = 0.0214). This induced a significant decrease of larval RGR (t = 3.5, df = 17.421, P = 0.00257) but not any change in larval survival (t = −1.8, df = 15.867, P = 0.0956).
Discussion
Among the seven Poaceae species used in this study, B. fusca larvae performed considerably better on Z. mays and S. arundinaceum than on the other wild grass species, confirming field observations that they are the most suitable hosts (Le Ru et al., Reference Le Ru, Ong'amo, Moyal, Muchugu, Ngala, Musyoka, Abdullah, Matama-Kauma, Lada, Pallangyo, Omwega, Schulthess, Calatayud and Silvain2006a , Reference Le Ru, Ong'amo, Moyal, Ngala, Musyoka, Abdullah, Cugala, Defabachew, Haile, Matama-Kauma, Lada, Negassi, Pallangyo, Ravololonandrianina, Sidumo, Omwega, Schulthess, Calatayud and Silvain b ; Ong'amo et al., Reference Ong'amo, Le Ru, Dupas, Moyal, Muchugu, Calatayud and Silvain2006; Ndemah et al., Reference Ndemah, Schulthess, Le Ru and Bame2007). Zea mays and S. arundinaceum are also hosts of other African stemboring crop pests such as the noctuid Sesamia calamistis Hampson, the pyralid Eldana saccharina (Walker) and the crambid Chilo partellus Swinhoe (Shanower et al., Reference Shanower, Schulthess and Bosque-Pérez1993; Schulthess et al., Reference Schulthess, Bosque-Pérez, Chabi-Olaye, Gounou, Ndemah and Goergen1997; Atachi et al., Reference Atachi, Sekloka and Schulthess2005; Sétamou et al., Reference Sétamou, Jiang and Schulthess2005; Mailafiya et al., Reference Mailafiya, Le Ru, Kairu, Calatayud and Dupas2010; Ong'amo et al., Reference Ong'amo, Le Ru, Calatayud and Silvain2013). Moreover, although the larval development on wild grasses tested in this study was inferior to that of maize or wild sorghum, marginal developments were recorded on P. purpureum, A. donax and P. maximum suggesting that the larvae can still utilize these wild plants as food source, probably when the preferred hosts are not available. Marginal utilizations of wild plants by B. fusca have been observed in the field particularly in seasons when maize or cultivated sorghum are not available (Ong'amo et al., Reference Ong'amo, Le Ru, Dupas, Moyal, Muchugu, Calatayud and Silvain2006). The minimal larval performance on most of the wild plants used in this study may suggest the presence of ‘antibiotic’ and/or ‘phagodeterrent’ properties in these plant species possibly as result of secondary metabolites or the plants were physically less suitable for feeding or lacked nutrients necessary for optimal larval growth.
The levels of Si found in this study were within the range of those reported for leaves of maize and other grass species (Massey et al., Reference Massey, Ennos and Hartley2006; Keeping et al., Reference Keeping, Kveradas and Bruton2009; Nabity et al., Reference Nabity, Orpet, Miresmailli, Berenbaum and DeLucia2012) and they negatively affected the performance of B. fusca larvae. This was confirmed by results from artificial diets and potted plants amended with increasing levels of Si. Similar results were obtained for other lepidopteran stemborer species (Peterson et al., Reference Peterson, Scriber and Coors1988; Kvedaras et al., Reference Kvedaras, Keeping, Goebel and Byrne2005; Keeping & Meyer, Reference Keeping and Meyer2006). However, the relationships between silica content and survival on artificial diets and potted maize plants amended with Si were not as strong as those found for wild grass species. This suggests that other factors besides Si such as phagostimulant/phagodeterrent factors influence larval survival. For example, several compounds such as sugars (including sucrose, the ‘universal’ phagostimulant), free amino acids, lipophilic constituents (fatty acids) and secondary metabolites are known to strongly influence insect feeding (see Schoonhoven et al. (Reference Schoonhoven, Jermy and van Loon1998) for review). Juma et al. (Reference Juma, Thiongo, Dutaur, Rharrabe, Marion-Poll, Le Ru, Magoma, Silvain and Calatayud2013) using the same wild grasses demonstrated that the balance of sucrose (a phagostimulant) and its isomer, turanose (phagodeterrent), in plant tissues influences the overall host plant choice and feeding by the B. fusca larvae. Silicon by itself appears not to be toxic to B. fusca larvae but most probably disturbs the food uptake process and feeding of larvae as indicated by the relationships between Si levels and RGR on artificial diets and on potted plants. This was also shown by Keeping & Meyer (Reference Keeping and Meyer2006), Kvedaras & Keeping (Reference Kvedaras and Keeping2007) and Kvedaras et al. (Reference Kvedaras, Keeping, Goebel and Byrne2007) for the stemborer E. saccharina. As reported by Schoonhoven et al. (Reference Schoonhoven, Jermy and van Loon1998), chewing Si containing plant tissues requires energy and causes severe wear of mouthparts as compared to feeding on softer tissues. The amorphous Si particles deposited in cell walls and cell lumen (Epstein, Reference Epstein1994) may serve as a harsh abrasive causing a complete loss of mandibular teeth during the feeding process, resulting in death by starvation (Schoonhoven et al., Reference Schoonhoven, Jermy and van Loon1998). Also, Djamin & Pathak (Reference Djamin and Pathak1967) showed that Si increases hardness of plant tissues and interferes with larval boring and feeding activity of Chilo suppressalis (Walker) (Lepidoptera: Pyralidae). In our study, it was observed during the experiments that B. fusca larvae feeding on plants high in Si or diets amended with Si fed on the surface rather than boring into the stem or diet.
From an applied point of view, the fact that increasing Si levels in maize plants disturbed the larval growth of B. fusca (as showed by the last experiment of the present study) suggests that Si amendments of maize plants may provide improved resistance to B. fusca as it has been reported for Sesamia calamistis (Hampson) (Lepidoptera: Noctuidae) following Z. mays tissue silification (Sétamou et al., Reference Sétamou, Schulthess, Bosque-Pérez and Thomas-Odjo1993). However, field trials are required to confirm this hypothesis.
In conclusion, the results provide insight into the possible mechanisms of oligophagy of B. fusca and provide a correlative support for a physical role of plant endogenous Si in impeding feeding of B. fusca larvae. However, this does not preclude the importance of phagostimulant or phagodeterrent compounds for the acceptance and feeding processes by B. fusca larvae.
Acknowledgements
The authors are grateful to Boaz Musyoka and Anthony Kibe for technical assistance and the staff of ICIPE's mass rearing unit for stemborer colony rearing. Thanks are also given to Fritz Schulthess for his review of the manuscript and to Florence Mougel for her statistical help. This work was funded by IRD, France and the French Ministry of Foreign Affairs for which the authors are very grateful.