Insects

Two populations of R. dominica were collected from storage units in Campo Grande, Mato Grosso do Sul (CG lineage) and Santa Rosa, Rio Grande do Sul (SR lineage) in 2010. At the time of collection, the population from Campo Grande was effectively controlled by protective insecticides of stored grains, while in Santa Rosa the control of these pests was a failure. The studied species were reared on insecticide-free wheat grains until the installation of the experiments.

Insecticides

In order to tests the effectiveness of the insecticcedes, we used the following formulations: Sumigranplus^® (500 g fenitrothion + 25 g esfenvalerate per liter of the commercial product), Sumigran^® (500 g fenitrothion per liter of the commercial product) and Sumidan^® (25 g esfenvalerate per liter of the commercial product) were used. To estimate LC₅₀ analytical standards of fenitrothion (99.1 % a.i.) and esfenvalerate (100 % a.i.) were utilized, (Iharabras S.A., Sorocaba, SP, Brazil).

Grain treatment and bio-assays

Grains were treated in order to obtain the theoretical concentration of 10 and 0.5 mg.kg^-1 fenitrothion and/or esfenvalerate, respectively, as recommended by the manufacturer on the labels. The method of application in use was proposed by (^{Vásquez-Castro et al. 2007a}, ^2007b). Three replications were performed for each insecticide. Temperature values and relative humidity during spraying were 18 °C and 72 %, respectively. After grain-spraying, they were placed into kept-open plastic bags and stored in the laboratory under uncontrolled conditions of temperature and relative humidity.

For the effectiveness tests, 40 g of grains were placed into plastic pots and infected with 40 adult insects aged 10 to 20 days. After this, the pots were closed with voile fabric and elastic and maintained in the laboratory under uncontrolled environmental conditions. Mortality evaluation was accomplished 15 days after infestation, considering dead the immobile individuals, that is, unable to move even when stimulated by an incandescent light source placed a few centimeters from them (^{Vásquez-Castro et al. 2012}). The effectiveness experiments were installed as of 15 days, 30 days and subsequently at monthly intervals until mortality ceased. After mortality evaluation occurred, all individuals of R. dominica were eliminated, either dead or alive, and grains were maintained for 90 days to assess the number of emerged progeny (F1) during this period.

Bio-assays to estimate LC₅₀ were adapted from studies of resistance of R. dominica to insecticides (^{Guedes et al. 1996}; ^{Guedes et al. 1997}; ^{Guedes and Zhu 1999}; ^{Fragoso et al. 2003}). Adult insects aged 10 to 20 days were evaluated by exposure to dry insecticide-residuals on the inner surface of glass vials (2.3 cm diameter x 4 cm height). Five milliliters of the insecticide solutions were transferred to the tubes and then evaporated by air-moving assistance previously dried on blue silica-gel desiccant filter. Afterwards, 25 individuals were placed inside the tubes and maintained under controlled conditions (25 ( 1 °C temperature and 74 % relative humidity). In order to prevent insects from leaving the treated area, a thin layer of liquid vaselineage was spread over the border of treated and untreated areas of the tube. Mortality evaluation was carried out 24 h after the installation of the experiments, using the same criterion of the effectiveness tests. Four replications for each concentration and five or six concentrations were used in each bioassay. Control treatment was also included but received only the solvent (acetone). Concentration-mortality data were submitted to Probit for analysis (^{Leora Software 2016}).

Statistical analysis

As the response variable of the experiment (number of dead insects, d = 40 insects) has binomial distribution, effectiveness tests data were analyzed by using logistic regression, that is, through the logistic procedure of SAS (^{SAS 1999}). The model ƞ=α_ij+β_ijx was practiced; whose lineagear predictor includes different intercepts and inclination coefficients (related to time after insecticide application) for the combinations of insecticides and lineages. Based on this model, estimates were calculated for (_s, (_s and lethal time 50, when mortality of 50 % insects occurred. For progeny data analyses (F1), it was first considered that this variable had a Poisson distribution since it is a count, but due to data over-dispersion, the response variable was evaluated using the negative binomial distribution, through GENMOD procedure of ^{SAS (1999)}. Lineage R. dominica lineage and insecticide (categorical variables) and time after insecticide application (quantitative variable) were considered explanatory variables.

Effectiveness experiments

The model in use indicates the existence of significant differences between ( and ( as to insecticides and pest lineages. Table 1 shows the estimated values of these two parameters for each combination of insecticide and pest lineage, as well as lethal time to kill 50 % of insects, based upon estimates of these two parameters. Deviances and χ² statistics indicate that the lineagear logistic model fit well in the 6 models of the combination of insecticide and pest lineages, and the lower significance level was 0.3171 for the mixture of insecticides fenitrothion + esfenvalerate used in the CG lineage of R. dominica. Moreover, a significant lineagear regression effect (P > 0.05) occurred in all cases; that is, a time-effect happened after insecticide application on insect mortality.

In some cases, it was not possible to estimate the wanted-lethal time so as to reach 50 % mortality (LT₅₀) because values lower than 50 % were obtained at the beginning of the experiments, as per SR lineage exposed to insecticides fenitrothion, esfenvalerate and fenitrothion + esfenvalerate mixture.

Comparing the adjusted curves of lineages within insecticides, as well as insecticides within lineages, significant differences (P < 0.0001) were observed in all cases, the same happens when comparing curves of insecticides within lineages. Figures 1 to 3 depict mortality observed and adjusted curves, on each, and for each insecticide and each lineage. Comparison amid more similar mortality curves as to those within each insecticide by way of difference of deviances displayed significant dissimilarities (P < 0.05) in curves. The control showed 0.13 % average mortality.

Figure 1 Mortality of two lineages of Rhyzopertha dominica and adjusted curves using the lineagear logistic model over time after treatment of grains with fenitrothion.  

Figure 2 Mortality of two lineages of Rhyzopertha dominica and adjusted curves using the lineagear logistic model over time after treatment of grains with esfenvalerate. 

Figure 3 Mortality of two lineages of Rhyzopertha dominica and adjusted curves using the lineagear logistic model over time after treatment of grains with the fenitrothion + esfenvalerate mixture. 

Table 2 presents the characterization of susceptibility of the studied lineages to fenitrothion and esfenvalerate. One can note that the results obtained through dry-film bio-assays meet results of effectiveness tests. Regarding fenitrothion, the SR lineage of R. dominica was 1.6 times more resistant than the CG lineage. As for esfenvalerate, the SR lineage of R. dominica was 8.9 more resistant than the CG lineage.

In all cases the tested populations presented LC₅₀ heterogeneity lower than 4, except for the SR lineage of R. dominica treated with esfenvalerate. Notwithstanding, for the same combination of lineage and insecticide, the lowest angular coefficient and the higher confidence interval amplitude were obtained. The SR lineage showed slight response to the increase of esfenvalerate concentration, up to the state in which the increase in insecticide concentration did not increase mortality. This might explain the output result.

In general, R. dominica was more susceptible to esfenvalerate than to fenitrothion. Considering only the results of the dry film test, one can affirm that esfenvalerate is effective in controlling R. dominica, but the grain protection experiments demonstrated the contrary. In this sense, performing both types of bio-assays (grains treatment and dry film) is recommended in order to characterize the susceptibility of pests to the insecticides used to protect stored grains.

Potentiation effect occurred for the fenitrothion + esfenvalerate mixture in controlling the two lineages of R. dominica. In the CG lineage, the toxic effect of the additive type between insecticides was initially observed; as of the second month after treatment, the toxic effect was potentiation. In the SR lineage, control levels were very low; and so, this potentiate-effect was observed from the beginning of the experiments.

F1 progeny of Rhyzopertha dominica

Considering as a random component the Poisson distribution, and all the explanatory variables and interactions, the starting analysis presented residual deviance (7513) and χ² (7243) as very high compared to the 260 residual degrees of freedom, indicating over-dispersion of data. Thus, the negative binomial distribution was considered as a random component of the model, that is, one of the ways to analyze over-dispersed tallied data. The adjustment of the negative binomial model presented residual deviance (317.9) and χ² (249.7) values closer to the 260 residual degrees of freedom.

Table 3 shows the existence of significant effect (P < 0.01) of all the evaluated causes of variation, including triple interaction. This indicates that the number of individuals of F1 progeny depends on the lineage of R. dominica, insecticide and time after insecticide application.

Table 4 presents the estimates of ( and ( parameters of the negative binomial lineagear model for the combination insecticide-lineage.

The estimated value of number of insects of F1 progeny is determined by e ^α+βx , for each combination of insecticide and pest lineage. Time after treatment is indicated by an x in the expression, and e (= 2.7183) is a natural number.

Significant effect (P > 0.05) of lineagear regression was verified in all cases, except for fenitrothion, utilized to control the SR lineage of R. dominica (P = 0.1301). This result indicates that time-effect after insecticide application occurred on a number of insects of F1 progeny in seven out of eight combinations of insecticide and pest lineage.

Figures 4 and 5 show tallies observed and adjusted curves of F1 progeny insects, respectively, for each R. dominica lineage and each insecticide.

Comparing the four curves of insecticides for each lineage show significant differences (P < 0.0001), for both lineages, indicating the existence of significant differences amid them for at least two insecticides in these populations.

Figure 4 Average observed number of adults (F1) of SR lineage of Rhyzopertha dominica and curves adjusted by the negative binomial model over time after treatment of grains with fenitrothion, esfenvalerate and the fenitrothion + esfenvalerate mixture. 

Figure 5 Average observed number of adults (F1) of CG lineage of Rhyzopertha dominica and curves adjusted by the negative binomial model over time after treatment of grains with fenitrothion, esfenvalerate and the fenitrothion + esfenvalerate mixture. 

SR lineageage (Fig. 4) report presented an estimated curve over the number of insects of F1 progeny holding higher values than those other treatments; however, this curve was not significantly different from that of the insecticide fenitrothion (P = 0.1725). For the same lineage, esfenvalerate presented the lowest estimated values of response as of the end of the experiment; nevertheless, the estimated curve was not significantly different from that of the fenitrothion + esfenvalerate mixture (P = 0.3081). Curves of both insecticides were significantly different from that of control (P < 0.0001).

As for the CG lineage (Fig. 5), significant differences were observed between all insecticide curves, and the fenitrothion + esfenvalerate mixture was the most effective in reducing F1 progeny. Although fenitrothion was ineffective in controlling this lineage, it was able to reduce significantly the progeny. Conversely, there was a potentiation effect for the fenitrothion + esfenvalerate mixture in reducing offspring, thus agreeing with the results obtained in the control of parents.

Effectiveness experiments

When insecticides fenitrothion and esfenvalerate were singly utilized the CG strain of R. dominica presented the highest mortality % (Figs. 1 and 2), although the insecticide was able to kill only 60 % of tested individuals as of 15 days after grain treatment. These insecticides caused less than 40 % mortality at any time of the evaluation, that is, SR lineage of R. dominica.

The lower locomotion ability of R. dominica on grain mass is a characteristic-behavior that favors tolerance of this species because the insect is exposed to a lower amount of insecticide (^{Surtees 1964}). Although R. dominica presents great flight ability, it rarely infects grains in the field. The migration of this species occurs mainly from one silo to another (^{Athié and Paula 2002}) and consequently the evolution of resistance might also be favored by this characteristic.

Higher resistance to the tested insecticides was observed in the SR lineage of R. dominica compared to the CG lineage. At collection time, the SR lineage of R. dominica was more difficult to control than the CG lineage, as after rearing both lineages during 6 years in the absence of selection pressure by insecticides, theoretically a resistance regression was expected in the SR lineage. Nevertheless, the SR lineage was always more resistant than the CG lineage regarding tested insecticides. Hence, results suggest that the SR lineage presents stability of resistance, which would have practical complications to management of this biological phenomenon, because the principle of insecticide-rotation or suspension of the chemical treatment as a management approach is based on the reversion of resistance in the absence of selective pressure. ^{Mason (1998)} , studying the evolution of resistance of Oryzaephilus surinamensis (L.) (Coleoptera: Silvanidae) to malathion in absence of insecticide, observed different responses in evaluated lineages, existing selection for and against alleles that confer resistance, and possibly in favor of modified alleles that reduce the fitness cost of resistance in an insecticide-free environment.

Differences of susceptibility between the evaluated lineages agree with results obtained by (^{Pacheco et al. 1993}) which did not detect resistance to malathion, nor to pirimiphos-methyl or fenitrothion in populations of R. dominica coming from the state of Mato Grosso do Sul, and by (^{Lorini and Galley 1999}) who registered moderate to high grounds for resistance to deltamethrinin in populations gathered in Santa Rosa, Rio Grande do Sul. The southern region of the country is a pioneer in storing food grains in bulk, mainly wheat (^{Puzzi 2000}) and the number of storage units in Rio Grande Do Sul is greater than in Mato Grosso do Sul (^{Beskow and Deckers 2002b}). These characteristics would explain why most cases of insecticide resistance in populations of R. dominica originate in the state of Rio Grande do Sul, as an elevated selection pressure for an extended time must have favored the development of this phenomenon.

Some studies have shown a toxic-effect of the additive type (^{Desmarchelier 1977}; ^{Daglish 1998}) and potentiate-type (^{Daglish et al. 1996}) when mixing an organophosphate and a pyrethroid insecticide to control coleopteran pests of stored grains. The results obtained in this report indicate that the toxic effect of the mixture of insecticides depends on the molecules involved and, on the species, to be controlled; thus, the mixture of these insecticides for simultaneous control of R. dominica and Sitophilus spp. is not always an appropriate strategy.

Reduced movement on treated surface was found in Brazilian lineages of R. dominica resistant to deltamethrin (^{Lorini and Galley 1998}; ^{Beckel et al. 2004}). This behavioral characteristic would be one responsible for resistance because the insect is exposed to a lower amount of insecticide.

F1 progeny of Rhyzopertha dominica

As expected, fenitrothion failed to reduce the offspring of the SR lineage as it had previously failed in the control of parents. Although fenitrothion and esfenvalerate have caused the same level of control in parents, esfenvalerate significantly reduced progeny, possibly due to the negative effect of this molecule in some aspect of the biology of that lineage. Sub-lethal concentrations of insecticides can cause adverse effects on pest development (^{Loschiavo 1960}; ^{Fila et al. 2003}). Instead, no potentiation effect occurred for the fenitrothion + esfenvalerate mixture on progeny reduction, as observed in the control of parents. Thus, offspring reduction for this lineage was exclusively of esfenvalerate effect.

Two important factors to be considered in the control of R. dominica are oviposition and feeding habits. This species is characterized by laying eggs on the outside of grains, and after hatching the larvae it perforates the grains and feeds on all its content, reducing them all to empty shells (^{Athié and Paula 2002}). So, most insecticide-susceptible development stages like first-larval and egg being exposed to these chemical products, will be relatively-easily controlled. Instead, surviving larvae feed on the entire content of the grain including the germ tissue where insecticides are concentrated, especially the more lipophilic; consequently, the larvae are killed by exposure to high concentrations of these products. Apparently, this is a disadvantage of the species, as observed in the CG lineage, but the strong selection pressure favors the evolution of resistance and, therefore, the loss of insecticide-effectiveness, as it is observed in the SR lineageage.

When evaluating control treatment, it was verified that the CG lineage presented greater F1 progenies than SR (P = 0.0001), and the inverse occurred when insecticides were applied. The result suggests that when both lineages are reared in the absence of fenitrothion and esfenvalerate, the SR resistant lineage presents adaptive disadvantage compared to that of susceptible CG lineage. Differently, when reared on treated grains, the resistance lineage presented higher adaptive value than the susceptible lineage. The adaptive disadvantage of populations resistant to insecticides in the absence of these chemicals has been demonstrated in some pest species of stored grains (^{Muggleton 1983}; ^{Fragoso et al. 2005}). Thus, the effect of insecticide on pest population growth rate depends on the chemical molecule, species, and pest lineage, among other factors.

In the 90's, fenitrothion and esfenvalerate insecticides were effective in the control of R. dominica in Brazil (^{Furiatti et al. 1999}), the continuous use of these insecticides without an adequate rotation program has led to the failure of these tools to control the lesser grain borer. On the other hand, cross-resistance between pyrethroids has been demonstrated in populations of R. dominica from Brazil (^{Lorini and Galley 2001}). It is likely that the intensive use of other pyrethroids has favored resistance to esfenvalerate.

Currently, phosphorus and pyrethroid insecticides have lost efficacy in the control of the lesser grain borer in many countries (^{Edde 2012}). The genetic and bioecological characteristics of this pest added to the inadequate use of insecticides has generated this problem.

Special attention should be given to the fact that both R. dominica populations were reared for more than six years in the absence of any insecticide. In the case of the SR lineage, the resistance remained stable despite the years that passed. This shows that both insecticides are not viable for the control of R. dominica and must be temporarily removed from the control programs of this pest in Brazil.