Introduction

Bemisia tabaci Gennadius (Hemiptera: Aleyrodidae) is one of the most important agricultural pests worldwide (Oliveira et al. 2001; Baldin and Beneduzzi 2010), including China where, since the 1990s, it largely affects many major crops resulting in huge yield losses every year (Luo and Zhang 2000; Ren et al. 2001). Multiple studies have reported that natural enemies could act along with other mortality agents to inflict high levels of mortality in populations of B. tabaci in various crops (Naranjo and Ellsworth 2005; Stansly and Naranjo 2010). Biological control of whiteflies has proved to be successful, notably using generalist predators, parasitoids or fungal pathogens (Faria and Wraight 2001; Gerling et al. 2001; Ren et al. 2004). Seventeen predator species has been already identified as preying largely on B. tabaci in Chinese agricultural systems, and the native coccinellid Serangium japonicum Chapin (Coleoptera: Coccinellidae) has shown promising potential for effective biological control of B. tabaci in various crops, notably on cotton (Ren et al. 2001, 2004). Larvae and adults of S. japonicum prey on eggs and all juvenile instars of B. tabaci, and its high daily consumption (730 prey per adult per day) could slow down or suppress B. tabaci population growth (Yao et al. 2004).

However, management of B. tabaci in China relies primarily on broad spectrum chemical insecticides in most cropping systems, which are noxious to beneficial arthropods (Desneux et al. 2007). Extensive use of insecticides has lead to the development of resistant strains of B. tabaci to most conventional insecticidal compounds (Palumbo et al. 2001; Ahmad et al. 2010), including relatively new classes of insecticides (Cahill et al. 1996; Wang et al. 2010; Zhang et al. 2011). Therefore, it is necessary to develop integrated pest management (IPM) programs that do not rely only on insecticides and that could provide sustainable control of B. tabaci in China. As IPM aims at reducing pest status to tolerable levels by using methods that are effective, economically sound and that mitigate the harmful effects of pesticides to natural enemies (Croft 1990), understanding risks posed by pesticides to natural enemies is of absolute importance (Desneux et al. 2006a, 2007). Natural enemies can be exposed to pesticides through direct exposure to spray droplets or residues on the crop foliage when foraging for hosts or prey (Jepson 1989), or through dietary exposure when feeding on pesticide-contaminated food (e.g. prey) (Ahmad et al. 2003; Huo et al. 2004; Torres and Ruberson 2004).

Acute toxicity (lethal effect) and selectivity of pesticides has been examined in predators (Zhu et al. 1998; Qiu et al. 2000), including in predatory coccinellid beetles (Hao et al. 1990; Zhu et al. 1998; Wu et al. 2007; Urbaneja et al. 2008; Xi 2008; Cabral et al. 2011). However, in addition to their lethal effect, pesticides also can induce various sublethal effects. Such effects (either physiological or behavioral) are defined as effects on individuals that survive pesticide exposure and they can occur after an exposure to dose/concentration/rate that can be either sublethal or lethal (see Desneux et al. 2007 for a thorough review). Occurrence of such effects is likely given the multiple ways natural enemies could be exposed to pesticides. Sublethal effects can impair various key processes of natural enemies’ efficacy against pests (Desneux et al. 2006c, 2007; Thornham et al. 2008; Suma et al. 2009; Evans Samuel et al. 2010; Arnó and Gabarra 2011; Saber 2011; Stara et al. 2011; Castro et al. 2012). Given the multiple possible behavioral effects, functional response of predators to prey could be impaired by pesticides (Wang and Shen 2002; Poletti et al. 2007). However, sublethal effects of insecticides on coccinellids, notably neonicotinoid insecticides (e.g. imidacloprid which is used on a large scale in China), are scarcely documented (Smith and Krischik 1999; Wang and Shen 2002; Papachristos and Milonas 2008; Eisenback et al. 2010) and no information is currently available on sublethal effects of insecticides on S. japonicum.

In this context, the aims of the present work were to provide a lethal and sublethal toxicity assessment of the neonicotinoid imidacloprid on S. japonicum, and to assess risks posed by imidacloprid use in crops. We studied the lethal toxicity of imidacloprid to S. japonicum through various exposure routes that aimed at representing various exposure scenarios (and because exposure methods could have an impact on results, Stark et al. 1995; Desneux et al. 2006a). We also studied the effects of imidacloprid on functional response of S. japonicum to B. tabaci eggs and on predator voracity as sublethal effects on these key predator traits (O’Neil 1997; Cabral et al. 2011; Madadi et al. 2011) could greatly reduce predator impact on B. tabaci. This study constitutes the basis for future guidelines for imidacloprid rational use, and it can help optimizing IPM programs involving use of both natural enemies and imidacloprid against B. tabaci in China.

Materials and methods

Insects

Adults of B. tabaci (biotype B) and adults of S. japonicum were collected from vegetable fields in Fuzhou city (Fujian Province, China) in 2008 and 2009, respectively. These insects (100 S. japonicum adults and 1,000 B. tabaci adults) were used to establish colonies in the laboratory. B. tabaci was reared on cotton plants (Gossypium hirsutum) and S. japonicum was reared on cotton plants infested with B. tabaci (all plants were free of any insecticides). All colonies were reared in growth chambers at 25 ± 1 °C, 65 ± 5 % RH, l4:10 L.D. and the coccinellids used in all experiments were 24–48 h old adult females (mated prior experiments).

Insecticide solutions

Imidacloprid-based commercial product, BIDAN 10WP (Shenyang Chemical Ltd., Shenyang, China), was diluted with deionized water. In toxicity bioassays, final concentrations tested were as follows: the recommended field rate on cotton against B. tabaci (4 g active ingredient per 100 l, i.e. 40 ppm [part per million]) and reduced rates (25, 20, 15, 10 and 5 ppm) to mimic lower rates that usually occur in fields during the days/weeks/months following initial insecticide application (owing to degradation of pesticides) (Desneux et al. 2005; Zhang et al. 2011).

Toxicity of imidacloprid residues on glass

Coccinellid adults were exposed to dry residues of imidacloprid on glass in tubes, using a worst-case exposure scenario (accordingly to Candolfi et al. 2000), using methods developed in previous studies (Zhu et al. 1998; Desneux et al. 2004a). Imidacloprid solutions were applied to the inner surface of glass tubes (length: 18 cm; diameter: 1.8 cm). To obtain a homogeneous deposit and total coverage of the internal surface of the tube, they were first fully filled (i.e. 45.8 ml) with the various solutions (or water as control), then the solutions were poured off and tubes were rotated until no more droplets were seen on the glass. The tubes were left on the bench to allow complete evaporation of solutions before introducing S. japonicum (ten adult females per tube). Tubes were then sealed with gauze and placed in growth chambers (25 ± 1 °C, 65 ± 5 % RH, l4:10 L.D.). Mortality was recorded after 24 h of exposure. Each treatment was replicated six times, with at least 60 coccinellids exposed per concentration tested. Individuals that did not react when pushed with a brush were considered dead.

Toxicity of imidacloprid through systemic treatment

We assessed acute toxicity of imidacloprid to S. japonicum adults when feeding on B. tabaci eggs deposed on cotton leaf treated with imidacloprid through systemic application (as it would occur in case of imidacloprid-dressed cotton seeds or direct soil applications). The petioles of cut cotton leaves infested with B. tabaci eggs were immersed in an imidacloprid solution at field rate (40 ppm) or in water as control. After 24 h at 25 ± 1 °C in a ventilated hood, leaves were placed with their adaxial surface downwards onto agar in a Petri dish (diameter: 11 cm). Ten S. japonicum adults were transferred onto the leaf-discs. Each Petri dish was then covered with a perforated lid and placed in growth chambers (25 ± 1 °C, 65 ± 5 % RH, l4:10 L.D.). Each treatment was replicated six times (total of 120 coccinellids tested). Mortality was recorded after 24 h of exposure using the mortality criterion described above.

Impact of imidacloprid on functional response of S. japonicum to B. tabaci eggs

The assessment of effects of imidacloprid on the functional response of S. japonicum to B. tabaci eggs was done by testing different exposure scenarios: (i) B. tabaci eggs were provided on cotton leaves treated systematically with imidacloprid at field rate (40 ppm) (described in “Toxicity of imidacloprid through systemic treatment” section). In this case S. japonicum is not exposed to the insecticide directly. (ii) B. tabaci eggs were treated with imidacloprid at 5 ppm (reduced rate and actually a sublethal rate for S. japonicum adults, see “Results” section) by egg-dip method, based on the method developed by Horowitz et al. (2002). Briefly, cotton leaves infested with 1 day-old eggs, were dipped for 5 s in the 5 ppm imidacloprid solution and then leaves were allowed to dry. In this situation S. japonicum was exposed to both imidacloprid residues on cotton leaves and B. tabaci eggs that had been treated with imidacloprid. Because imidacloprid is mainly used as foliar spray application in China, assessing side effects induced by exposure to cotton leaves and prey treated through the egg-dip method is highly relevant (realistic). (iii) S. japonicum adults were first exposed to imidacloprid residues (5 ppm) on glass for 24 h (method described in “Toxicity of imidacloprid residues on glass” section) and B. tabaci eggs were sprayed with water. This situation mimics what may happen when a predator forages in a treated field in which it experiences a discontinuous exposure to the insecticide (alternation between pesticide-treated and refuge areas, Desneux et al. 2005). (iv) S. japonicum and B. tabaci eggs were not treated with imidacloprid (control group). These scenarios are referred hereafter as “systemic”, “egg-dip”, “discontinuous” and “control” respectively.

We followed the method developed by Wang and Shen (2002) to estimate the functional response of the predator to B. tabaci eggs under these different exposure scenarios. We placed the cotton leaves infested with different B. tabaci egg densities (100, 200, 300, 400, 500, 600, 800, 1000 and 1200 eggs per leaf) on agar within Petri dishes (diameter: 11 cm). Within each Petri dish, we introduced one S. japonicum adult female that was previously starved for 24 h. Each Petri dish was then covered with a perforated lid and placed in growth chambers (25 ± 1 °C, 65 ± 5 % RH, l4:10 L.D.). The number of B. tabaci eggs consumed was counted 24 h after predators had been introduced into Petri dishes. Each prey density treatment was replicated ten times.

Impact of imidacloprid on predator voracity

Cotton leaves infested with 1000 B. tabaci eggs were exposed to the 5 ppm imidacloprid solution (i.e. sublethal rate, see “Results” section) using the egg-dip method (see above) or in water as control and leaves were placed onto agar within a 11 cm diameter Petri dish after they dried. The density of B. tabaci (1000 eggs) was chosen because it was higher than maximal daily consumption of eggs by S. japonicum (see results on functional response). One S. japonicum adult (previously starved for 24 h and not previously exposed to imidacloprid) was introduced per Petri dish and dishes were placed in growth chambers (25 ± 1 °C, 65 ± 5 % RH, l4:10 L.D.). After 24 h, we counted the number of B. tabaci eggs consumed and the surviving coccinellids were transferred onto untreated cotton leaves bearing 1000 untreated B. tabaci eggs. The number of eggs consumed was then counted after 8, 16 and 24 h to assess recovery of potential effects of imidacloprid on S. japonicum voracity. Each treatment was replicated ten times.

Statistical analysis

Datasets were first tested for normality and homogeneity of variance using Kolmogorov–Smirnov D test and Cochran’s test respectively, and transformed if necessary. We compared mean mortality among treatments using ANOVA followed by a Tukey’s post hoc test for multiple comparisons, and the lethal rate 50 (LR50) was determined using a log-probit model (Finney 1971). In addition, we estimated the hazard quotient (HQ) value which is calculated by dividing crop-specific application rates by the LR50 derived from worst-case-scenario laboratory study (exposure to residues on glass) (Candolfi et al. 2001) and which gives an indication of the risk when testing commercial pesticide products.

The functional response model proposed by Holling (1959) was used to describe the variation in the number of prey consumed by the predator, estimating Nα by:

$$ N\alpha = \frac{\alpha TN}{1 + \alpha ThN} $$

where α is the attack coefficient, T is the duration of the experiment (T = 1 day in the present study), Th is the handling time by the predator during the processes of prey identification, capture, attack and consumption, and N is prey (egg) density. The parameters of the equations were estimated by nonlinear regression. Nonlinear regression and ANOVA were conducted using SAS Version 8.0 (SAS Institute Inc., Cary, USA). The peak consumption in each treatment was estimated using Th reciprocal.

Results

Toxicity on glass and through systemic exposure, LR50 and HQ

Dried residues of imidacloprid induced significantly mortality to S. japonicum when exposed by contact on glass (F 6,41 = 147.59, P < 0.001; Fig. 1). Linear regression of the dose–mortality relationship (Y = −3.85 + 3.62 X) was fitted to the observed data (i.e. no significant deviation between the observed and the expected data; χ² = 2.93, df = 4, P = 0.569) and the LR50 was considered as valid (inf lim < LR50 < sup lim: 10.08 < 11.54 ppm < 12.91). Mortality in the control group was lower than 2 %. The recommended field rate of imidacloprid on cotton in China (40 ppm) was higher than the LR50 and HQ was equal to 3.47. At the recommended field rate (40 ppm) all individuals died within the 24 h exposure period. Mortality rates of more than 40 % in exposed individuals were recorded for all reduced rates except at 5 ppm. Mortality induced by 5 ppm imidacloprid was not significantly different than control mortality and therefore this rate could be designated as a sublethal rate of imidacloprid for S. japonicum (Desneux et al. 2007) and we refer to it as such hereafter.

Fig. 1
figure 1

Mean percentage (±SEM) of mortality after exposure to dried residues of imidacloprid on glass. Histograms bearing different letters are significantly different (P < 0.05, ANOVA followed by Tukey’s post hoc test)

Full size image

Imidacloprid applied systemically to cotton leaves at the field rate (40 ppm) did not induce significant mortality when S. japonicum adults fed on B. tabaci eggs on these leaves (P > 0.05).

Functional response

In all imidacloprid treatments, S. japonicum adults showed similar patterns of functional response when compared with the control, as the number of prey consumed per predator increased quickly as the prey density offered increased, leveling off with additional prey (above 800 eggs or 450 eggs depending of treatments) (Fig. 2). We fitted the type II functional response equation to the dataset and estimated the parameters α and Th for the various treatments (Table 1). S. japonicum showed significantly longer prey Th when it was exposed to both imidacloprid residues on leaves and imidacloprid-contaminated prey (egg-dip method). In contrast a similar Th was observed among systemic application treatment, when S. japonicum was exposed to dried residues of imidacloprid on glass at the sublethal rate (5 ppm) before the trial and the control group. S. japonicum exhibited similar attack coefficient (α) for all treatments (Table 1).

Fig. 2
figure 2

Effect of imidacloprid on functional response of S. japonicum adults on B. tabaci eggs. Exposure scenarios: systemic B. tabaci eggs treated with imidacloprid by systemic application (40 ppm), egg-dip B. tabaci eggs treated with imidacloprid by egg-dip method (5 ppm), discontinuous S. japonicum adults were pretreated with imidacloprid by residual application for 24 h (5 ppm), and control S. japonicum and B. tabaci eggs were not treated with imidacloprid. Dots represent means and vertical lines their respective standard errors, continuous lines represent nonlinear models fitted for each treatment

Full size image
Table 1 Estimates of functional response parameters of S. japonicum on B. tabaci eggs for the various exposure scenarios tested
Full size table

The peak consumption estimated for S. japonicum was reduced by approximately 60 % in the “egg-dip” treatment when compared to control group. In contrast, it was not affected in case of “systemic” and “discontinuous” treatments (Fig. 3).

Fig. 3
figure 3

Effect of imidacloprid on peak consumption (estimated using Th reciprocal) of S. japonicum female adults on B. tabaci eggs (no. eggs/24 h). Exposure scenarios: systemic B. tabaci eggs treated with imidacloprid by systemic application (40 ppm), egg-dip B. tabaci eggs treated with imidacloprid by egg-dip method (5 ppm), discontinuous S. japonicum female adults were pretreated with imidacloprid by residual application for 24 h (5 ppm), and control S. japonicum and B. tabaci eggs were not treated with imidacloprid

Full size image

Predator voracity

When applied through the egg-dip method for 24 h, the sublethal rate (5 ppm) of imidacloprid significantly affected S. japonicum voracity on B. tabaci eggs (Table 2). The number of B. tabaci eggs consumed on treated leaves was significantly lower than that on untreated ones (F 1,18 = 171.48, P < 0.001). However, after transfer to untreated leaves, S. japonicum voracity (consumption of eggs) rapidly returned to control levels (Table 2, all P > 0.05).

Table 2 Mean consumption (SEM) of B. tabaci eggs by S. japonicum on imidacloprid-treated and untreated leaves
Full size table

Discussion

Insecticides are mainly used as foliar spray applications in Chinese fields. Therefore natural enemies likely experience exposure to insecticide residues when foraging on treated plants/leaves after treatments and it may cause various direct and indirect effects. We demonstrated that residues of imidacloprid on leaves, when applied at the sublethal rate of 5 ppm (according to the definition of sublethal dose by Desneux et al. 2007) using the egg-dip method, affected the functional response of S. japonicum adults, markedly we observed an increase in Th, and a reduced peak in B. tabaci eggs potentially preyed (Table 1; Fig. 3). The sublethal rate of imidacloprid also impaired predator voracity with 41 % reduction in predation on B. tabaci eggs (Table 2). In all cases tested, predators recovered rapidly after the end of the exposure period and sublethal effects disappeared within a few hours following end of exposure. Consistently, exposure for 24 h via contact with dried residues of imidacloprid on glass (at sublethal rate), prior to assessment of S. japonicum functional response on imidacloprid-free cotton leaves, did not show any effects (“discontinuous” treatment, Table 1). Recovery after exposure to sublethal or low lethal rates of insecticides has previously been reported for other natural enemies exposed to pesticides (Desneux et al. 2004a, c, 2007), although it is not always found (Desneux et al. 2004b; Moser and Obrycki 2009).

Lethal effect

There was a good fit between the observed dose–mortality relationship and the linear regression, so our experimental conditions gave reliable estimates of LR50. The ratio between the field application rate and the LR50, i.e. HQ, gives an indication of the risk. For use of imidacloprid in cotton, the HQ value is higher than the trigger value of 2 (Candolfi et al. 2001) which indicates a risk for S. japonicum. This result is in concordance with previous laboratory studies showing that spray formulations of imidacloprid were highly toxic to various predatory mite species (Galendromus occidentalis, Neoseiulus fallacis and Amblyseius andersoni) (James 2003), and studies reporting high toxicity of imidacloprid to the parasitoid Anagyrus pseudococci (Krischik et al. 2007) and to the predatory coccinellid Hippodamia undecimnotata (Papachristos and Milonas 2008). However, previous studies on toxicity of imidacloprid also showed some selectivity of imidacloprid toward predators (i.e. more toxic to pests than to predators, Qiu et al. 2000; Tang et al. 2007). In addition, in field conditions, planting imidacloprid-dressed soybean seeds did not affect coccinellid and predatory bug populations (Seagraves and Lundgren 2012). However, because HQ was higher than 2 in our study, additional assessments of potential sublethal effects of imidacloprid on predator voracity and functional response to B. tabaci were required (Candolfi et al. 2001). For such assays, we used almost exclusively the sublethal rate of 5 ppm. In our study, systemic application imidacloprid at the field rate (40 ppm) was safe for S. japonicum adults and it may mean that imidacloprid-dressed seeds are safer for coccinellid predators than foliar applications (consistent with Ohnesorg et al. 2010; Seagraves and Lundgren 2012).

Sublethal effects on S. japonicum predation activity

Imidacloprid specifically targets the nicotinic acetylcholine receptors of insects (Palumbo et al. 2001) and therefore behavioral effects in predators are likely because, for example, sophisticated nervous activity is required in the process of prey detection by predators (Desneux et al. 2007). Previous studies already reported side effects of imidacloprid on predatory traits. Sublethal doses of imidacloprid significantly increased Th, decreased searching efficiency and consumption rate, and reduced the peak consumption of the coccinellid Harmonia axyridis on its aphid prey (Wang and Shen 2002) and of the predatory mite Neoseiulus californicus on spider mite eggs (Poletti et al. 2007). Adverse effects of imidacloprid on motor activity in insects have been also reported (Lambin et al. 2001), notably in predatory beetles (Smith and Krischik 1999; Eisenback et al. 2010). In our study, imidacloprid residues were harmful to S. japonicum adults, and the adverse effects on predator voracity and its functional response to B. tabaci eggs may be linked to multiple nervous disruptions following contact exposure with the residues. Other insecticides have been shown to adversely affect functional response of natural enemies because of sublethal effects on searching efficiency and attacking coefficient (α in our equation) (Gu et al. 1991; Claver et al. 2003). In our experiments, reduction in food intake may have been linked to haphazard movement and increased restlessness in predators (Claver et al. 2003; Tran et al. 2004) and disturbances in coccinellids feeding behaviors that consume imidacloprid-contaminated prey may also indicate antifeedant effects (Desneux et al. 2007). Such effects have previously been reported for imidacloprid at sublethal doses on Hemiptera (Nauen et al. 1998) and honey bees (Ramirez-Romero et al. 2008; Han et al. 2010). Specific behavioral and/or physiological assays are required to identify more precisely which traits are disturbed but our results clearly demonstrated that sublethal effects occur in individuals foraging on cotton plants previously treated by foliar applications of imidacloprid.

Exposure routes and implications for risk assessment

Imidacloprid has known good root-systemic characteristics and is quite toxic through both contact and ingestion by insects. These properties allow imidacloprid to be applied in various ways such as foliar sprays, seed treatment and soil application (Palumbo et al. 2001). As pointed out previously, insecticides are mainly used as foliar spray applications in Chinese crops. Therefore natural enemies likely experience high exposure through direct contact with insecticide residues when foraging on treated plants. However, this exposure may be discontinuous because natural enemies likely alternate foraging in treated areas and untreated refuge areas as previously reported in other agro-ecosystems (Cilgi and Jepson 1992; Koch and Weisser 2001; Desneux et al. 2005).

In the specific case of predators, a major route of exposure to pesticides is through ingestion of contaminated prey. Studies already reported adverse effects on coccinellid predators when fed on pesticide-contaminated prey (i.e. prey were previously treated) (Wang and Shen 2002; Ahmad et al. 2003; Huo et al. 2004; Urbaneja et al. 2008). However, in our present study, we did not find direct mortality nor sublethal effect on S. japonicum when imidacloprid was applied systemically at recommended field rate. This is unexpected because imidacloprid is toxic to B. tabaci eggs when placed on systemically treated plants (He et al. 2011; He and Wu unpublished data), i.e. the eggs did receive imidacloprid (through water uptake by the pedicel of B. tabaci eggs from the treated plants, Buckner et al. 2002; Li et al. 2009) and they contained imidacloprid in such exposure situation. It may mean that the imidacloprid dose present in B. tabaci eggs through this type of exposure is not toxic for S. japonicum when the predator consumes such contaminated prey.

More broadly, imidacloprid may be more toxic to S. japonicum through contact with residues than when ingesting imidacloprid-contaminated prey as reported in previous studies involving coccinellids and imidacloprid (Ohnesorg et al. 2010; Seagraves and Lundgren 2012). However, we think that this conclusion should be tempered though because (i) B. tabaci nymphs may contain higher concentrations of imidacloprid when starting feeding (after eggs hatch) on imidacloprid-treated plants, and (ii) coccinellids did not feed on plant materials during our assays (but it could occur in the field as they can fed on imidacloprid-contaminated nectar or pollen). Taken as a whole, our experiments can be seen as realistic scenarios as they mimic what would happen at the initial phase of infestation of the plants by the pest. However, it is not totally realistic if we consider that eggs may hatch after few days and then young nymphs do fed on imidacloprid-contaminated sap. In addition, exposure to imidacloprid may be problematic in case of other major predators like hemipterans (Desneux et al. 2006b; Ragsdale et al. 2011), which are known to consume plant sap. Further studies should be conducted on such predators if they prove to be of interest against B. tabaci on cotton in China.

Our study shows that aside from direct high residual-contact lethal toxicity, imidacloprid residues on cotton leaves occurring after foliar application can also cause adverse effects on functional response of S. japonicum on B. tabaci eggs. As imidacloprid applied systemically did not induce adverse effects on S. japonicum, there might be an opportunity to optimize imidacloprid application methods in cotton fields in China (to minimize potential negative effects on natural enemies). In addition, further longer-term studies are necessary to investigate the possible effects on development of larvae and fecundity of S. japonicum adults (e.g. see Biondi et al. 2012) when exposed to imidacloprid and/when consuming imidacloprid-contaminated prey. A reduction in the consumption capacity of S. japonicum may affect its reproductive capacity, inhibiting its population growth and consequently affect its performance in pest management programs (Stark and Banks 2003; Desneux et al. 2007).