Introduction

In the peach–potato aphid, Myzus persicae (Hemiptera: Aphididae), gene amplification leads to the overproduction of carboxylesterases that enhance the detoxification of insecticidal esters such as organophosphates (OPs) and carbamates (Devonshire and Moores, 1982; Field et al, 1988). In the same species, a point mutation of a voltage-gated sodium channel gene confers cross-resistance to pyrethroids and DDT (knockdown resistance or kdr) (Martinez-Torres et al, 1999). This latter mutation is common to many pyrethroid-resistant insect species, including Musca domestica (Diptera: Muscidae) (Vais et al, 2001).

It has long been proposed that the majority of insecticide-resistant organisms should show some differential survival in comparison with ‘wildtype’ organisms (eg Crow, 1957). However, despite the extensive literature on the subject, there are relatively few empirical studies that confirm that investment in such adaptations exacts a cost (Minkoff and Wilson, 1992; Carriere et al, 1994, 1995; Yamamoto et al, 1995; Chevillon et al, 1997; Alyokhin and Ferro, 1999; Boivin et al, 2001). Some authors have failed to reveal any detrimental effects of insecticide resistance (Follett et al, 1993; Tang et al, 1997, 1999; Baker et al, 1998) and some have shown such organisms to be even more successful than their susceptible counterparts in the absence of insecticides (Omer et al, 1992; Bloch and Wool, 1994; White and Bell, 1995; Mason, 1998; Haubruge and Arnaud, 2001). In other studies, some measures of fitness have been negatively affected, others positively (Brewer and Trumble, 1991) and different strains of insect, exhibiting resistance to the same compound, can show opposite associations (Chevillon et al, 1997; Hollingsworth et al, 1997; Oppert et al, 2000).

It is possible that the lack of any clear pattern in the literature results from the fact that evolutionary pathways can compensate for fitness disadvantages through the development of modifier genes (McKenzie and O'Farrell, 1993; Charpentier and Fournier, 2001) or through the replacement of mutant alleles by less costly ones (Guillemaud et al, 1998), but it may also reflect the fact that fitness costs are generally investigated by comparing biological parameters such as development times (Boots and Begon, 1993; Baker et al, 1998; Boivin et al, 2001), fecundity (Hollingsworth et al, 1997; Brewer et al, 1999; Boivin et al, 2001), and intrinsic rates of growth (Yamamoto et al, 1995) usually under optimal laboratory conditions. By concentrating purely on such obviously detrimental parameters and by failing to consider suboptimal conditions (stressful environments in which fitness differentials are at a premium), subtle pleiotropic effects on biology or behaviour are often overlooked (McKenzie, 1996). Examples of behavioural effects that have been shown to be correlated with resistance are therefore rare but include mating disruption and oviposition preference in dieldrin-resistant Anopheline mosquitoes (Rowland, 1991a,1991b), mating success in Bacillus thuringiensis (Bt)-resistant diamondback moths (Groeters et al, 1993), response to alarm pheromone in aphids (Foster et al, 1999), and overwintering success in Bt-resistant pink bollworms (Carriere et al, 2001), insecticide-resistant aphids (Foster et al, 1996, 1997) and blowflies (McKenzie, 1993). There is little understanding of the genotypic basis of these resistance-associated traits, but some may be examples of true pleiotropies, whereby one gene influences more than one aspect of an organism's phenotype. In pest-management terms, such effects clearly have the potential to exert a large influence on the selection and dynamics of genes conferring insecticide resistance.

The work reported here expands upon previous research on clones of M. persicae that showed behavioural effects to be correlated with elevated titres of carboxylesterases and with the presence or absence of kdr (Foster et al, 1996, 1997, 1999). It also reports on behavioural characteristics associated with kdr in the housefly (M. domestica). We present strong evidence that resistant genotypes may suffer pleiotropic fitness costs in the absence of insecticides, and that in the case of the kdr mutation, these are often likely to be identified in behavioural assays because of the effect of the resistant mutation on nerve function.

Methods and materials

M. persicae clones and diagnosis of resistance

In total, 22 M. persicae clones, collected from England and mainland Europe, showing different combinations of kdr and esterase resistance (Table 1) were used in the experiments. All aphids were maintained at 21±1°C on single excised leaves of Chinese cabbage (Brassica rapa L. var. pekinensis) in small perspex boxes under a 16:8 L:D regime. In M. persicae, kdr is caused by a single-point mutation (L1014P) in the para-type sodium channel gene. The genotypes of individual clones were determined by DNA sequencing of PCR-amplified para gene fragments as described previously (Martinez-Torres et al, 1999). Genotypes were scored as either susceptible homozygote (kdr-SS), heterozygote (kdr-SR), or resistant homozygote (kdr-RR) with the latter two being associated with resistance to pyrethroids and DDT (Martinez-Torres et al, 1999). Esterase levels, representing increasing levels of resistance (S, R1, R2, and R3) to OPs and carbamates (Devonshire et al, 1982, 1986) were based on the relative activities of immunocaptured E4 or FE4 carboxylesterases, the enzyme responsible for esterase-based insecticide resistance (Devonshire et al, 1986). None of these clones carried a third insecticide resistance mechanism known in M. persicae, termed modified acetylcholinesterase (MACE) (Moores et al, 1994).

Table 1 Myzus persicae clones assessed in leaf deterioration and alarm pheromone response bioassays
Full size table

Response of M. persicae clones to synthetic alarm pheromone

The aphid alarm pheromone (E)-β-farnesene is released from cornicle secretions exuded by aphids when they are physically disturbed, for example by foraging predators or parasitoids. Aphid response to synthetic alarm pheromone, applied at a wide range of concentrations, was assessed in the absence of insecticides in seven separate experiments. For each aphid clone tested, six first instar nymphs were obtained from laboratory stocks and grown to adulthood (in two small perspex boxes containing three aphids per clone). These first-generation (G1) adults were removed after they had produced approximately 30 G2 offspring per box (normally after about 10 days). The G2 aphids were grown to adulthood and transferred, using a fine paintbrush, to inverted 2 cm diameter Chinese cabbage discs (four G2 adult apterae per disc) held on 1.1% agar inside plastic tubs. The plastic tubs were left at 21±1°C overnight to allow the aphids to settle and produce nymphs. The adults were removed next morning leaving synchronised cohorts of settled offspring on each disc. Each replicate cohort was then assessed for response to alarm pheromone by applying a 1 μl droplet (ranging from 0.0001 to 10 mg (E)-β-farnesene ml−1 in hexane) to the central part of each leaf disc with a fine-needle syringe. Aphid behaviour was observed for 2 min using a binocular microscope (preliminary experiments showed that this period is sufficient for all responses to occur). Nymphs that withdrew their stylets and walked away were scored as responders. Control treatments with 1 μl droplets of hexane alone did not stimulate aphid movement. Each replicate batch of nymphs was tested once and then discarded.

Movement of M. persicae clones from deteriorating leaves at low temperature

Aphid movement from deteriorating excised Chinese cabbage leaves onto oilseed rape plants (Brassica napus L.) was assessed in seven separate experiments at low temperature (using one replicate per clone per experiment). Prior to each experiment, M. persicae clones were reared at 21±1°C under a 16:8 L:D regime in small perspex boxes. In each replicate, four young apterous adults were allowed to produce nymphs over 2 days on an excised Chinese cabbage leaf. All adults and surplus nymphs were then removed before the boxes, each containing 50 first/second instar nymphs, were transferred to 10±1°C (8:16 L:D regime) for 24 h temperature acclimation. Clones were then allocated (one replicate per plant) to 5-week-old plants (maintained under a 6±1°C, 8:16 L:D regime) arranged in a single row using an incomplete block design. Movement from each deteriorated excised leaf was assessed 5 days after it had been attached by the petiole and apex to the upper face of a healthy oilseed rape leaf using 4 cm clips. Numbers of aphids that moved to any part of the plant or remained on the deteriorated excised leaf were recorded at the end of each experiment.

M. domestica strains and maintenance

The origin of all housefly strains used is detailed in Table 2. They were reared at 28±1°C. Adults were kept in aluminium cages (0.3 m × 0.15 m × 0.15 m) with open ends; one covered with polythene netting, one with a terylene sleeve. Adults were fed on 20% sugar solution and encouraged to oviposit on cotton wool plugs soaked in milk. Eggs and young larvae were used to ‘seed’ a larval medium (milk, yeast, bran, and water in the ratio 2:2:3:7 by weight). Pupae were collected by flotation, gathered in a sieve, and transferred to the aluminium cages. The kdr (mutation L1014P) and super-kdr (mutation M918 T) genotypes were scored by DNA sequencing of PCR-amplified para sodium channel fragments from adult flies (Williamson et al, 1996).

Table 2 Origins of housefly strains
Full size table

Bioassays on M. domestica

All populations were tested for their responses to technical DDT and the pyrethroid deltamethrin. Serial dilutions were made up in acetone, and 1 μl of solution was topically applied to each fly. Control groups were treated with acetone alone and each group consisted of 10–15 female flies. Treated flies were kept in plastic containers with cotton wool soaked in sugar solution. Mortality was scored after 48 h. Each dose was tested against three groups and bioassays were conducted at least twice.

Temperature gradient studies on M. domestica

An aluminium sheet, 24 cm long and encased in a perspex cage (see Figure 1) was used to create a temperature gradient. One end of the sheet was kept in an ice bucket in which the ice was continuously replaced to create a temperature of 19±1.5°C at the 0 cm mark. The opposite end was immersed in a water bath to give a temperature of 29±1°C at the 24 cm mark. The quarter points of the aluminium sheet (6, 12, and 18 cm) were 21.5±1°C, 24±1°C, and 26.5±1°C, respectively.

Figure 1
figure 1

Apparatus for measuring temperature preference in houseflies.

Full size image

All experiments were carried out in a well-ventilated, temperature-controlled room (26±1°C) under infrared light to remove any visual cues. Individual female flies were kept in a refrigerator until they were sluggish and easy to manipulate. These were then dropped into the perspex case above the midpoint of the aluminium sheet. Each fly was given 30 s to recover before its position was noted. If a fly had not recovered in that time, it was removed and another released in its place. Between each observation, the aluminium sheet was swabbed with ethanol to remove any semio-chemical cues that could influence the movement of other flies. The orientation of the aluminium plate was also reversed between observations so that any bias resulting from the position of the apparatus was removed. All experiments were conducted between 12:00 and 18:00 h to standardise the part of the diel cycle that individuals were experiencing. Several strains were tested during each experimental period to randomise effects peculiar to that day (eg age of flies, barometric pressure, etc).

Once each fly had recovered, its position was noted 1, 5, 10, 15, and 20 min after introduction. After 20 min, observations were terminated. In total, 12–16 adult females were used for each strain, spread across at least three separate assay dates, and results were pooled for analysis.

Data analyses

Generalised linear models were fitted, using probit transformation, to the proportions of aphids in each replicate responding to alarm pheromone or moving from deteriorating leaves, including effects for clones, experiments, kdr genotype, and mean carboxylesterase activity (esterase level).

Housefly bioassays were combined for probit analysis (POLO-PC, 1984). Comparisons of the proportions moving to the warmest quarter of the temperature gradient were analysed using a generalised linear model with a binomial distribution.

Results

Response of M. persicae clones to synthetic alarm pheromone

Aphid response increased with higher alarm pheromone concentrations for all three kdr genotypes. Fitting a generalised linear model with a probit link and adjusting for esterase level and concentration effects, the tendency to respond was strongly associated with kdr genotype (Figure 2); the significance of differences between genotypes being as follows; SS vs SR, difference: 1.00, t715=8.6, P<0.001; SS vs RR, difference: 1.75, t715=14.7, P<0.001; SR vs RR, difference: 0.75, t715=6.6, P<0.001.

Figure 2
figure 2

Predicted mean response to alarm pheromone of Myzus persicae clones with kdr-SS, -SR, and -RR genotypes (clones within each genotype pooled).

Full size image

Furthermore, esterase resistance (adjusted for kdr effects and excluding the revertant clone) was inversely associated with pheromone response (Figure 3 and Figure 4) (slope: −1.31, SE 0.13, P<0.001). The revertant clone, 794Jrev, was not significantly different from the clone 794J that had retained expression of its highly amplified (esterase-R3 level) E4 genes (t715=1.7, P=0.09).

Figure 3
figure 3

Predicted mean response to alarm pheromone of esterase-R2 Myzus persicae clones with kdr-SS, -SR, and -RR genotypes (individual clones shown).

Full size image
Figure 4
figure 4

Predicted mean response to alarm pheromone of esterase-R3 Myzus persicae clones with kdr-SS, -SR, and -RR genotypes (individual clones shown).

Full size image

Movement of M. persicae clones from deteriorating leaves at low temperature

After adjusting for esterase level, the mean proportions (±SE) of aphids moving from deteriorating leaves in each genotype, for kdr-susceptible homozygotes (kdr-SS), kdr-heterozygotes (kdr-SR), and kdr-resistant homozygotes (kdr-RR) were, respectively, 0.48±0.03, 0.54±0.04, and 0.52±0.03. There were no significant differences between these three genotypes (kdr-SR vs kdr-SS: t76=1.25 P=0.22, kdr-SS vs kdr-RR: t76=0.87, P=0.39, kdr-SR vs kdr-RR: t76=0.38, P=0.71). Thus, movement from deteriorating leaves was not associated with kdr genotype (Figure 5).

Figure 5
figure 5

Predicted mean movement from deteriorating leaves of Myzus persicae clones with kdr-SS, -SR and -RR genotypes and esterase-S to -R3 levels.

Full size image

There was, however, a significant inverse relation between the tendency to move and esterase level (overall slope: −0.28, SE 0.12, P=0.027) (Figure 5). This relation was also apparent within the kdr-SS and kdr-SR categories (SS slope: −0.62, SE 0.20, P=0.003; SR slope: −1.18, SE 0.55, P=0.036). However, it was not apparent within the kdr-RR category, because all clones in this instance (with the exception of the two esterase-revertant clones) had very high esterase levels (RR slope: 0.08, SE 0.17, P=0.65). The two esterase-revertant clones (having the genotype for R3 amplified esterase, but the susceptible phenotype (Field et al, 1989) segregated with other high-esterase clones, in that they exhibited a limited propensity to move to fresh leaves. This suggests some association between behaviour and the esterase genotype rather than the phenotype.

Bioassays on M. domestica

The seven housefly strains could be divided into two groups on the basis of their collective responses to deltamethrin and DDT. Three strains: 213ab, 530, and 579 exhibited resistance to both DDT (resistance factors of 14, 43, and 7, respectively) and deltamethrin (resistance factors of 105, 497, and 14, respectively). Four strains: Cooper, 690ab, USA1, and USA2 were fully susceptible to one or both compounds (Table 3 and 4). Consistent with these results, the latter four strains were all confirmed as wild type for their domain II sodium channel sequences, whereas all three resistant strains contained the kdr (L1014P) mutation, with two strains, 213ab and 530, also carrying the super-kdr (M918 T) mutation. Previous studies have shown that the kdr mutation is associated with moderate (5 to 30-fold) levels of resistance to DDT and pyrethroids, whereas the additional presence of the super-kdr mutation increases resistance to over 100-fold for the more potent type II pyrethroids such as deltamethrin (Williamson et al, 1996). The USA2 and 690ab strains, showing an inconsistent response to DDT and deltamethrin, probably possessed a metabolic resistance mechanism specific to either DDT or pyrethroids.

Table 3 Response of housefly strains to μg/μl topical applications of DDT
Full size table
Table 4 Response of housefly strains to ng/μl topical applications of deltamethrin
Full size table

Temperature gradient studies on M. domestica

The majority of individuals (82–94%) of all four strains lacking kdr had, within 20 min, settled between temperatures of 26.5 and 29°C in the warmest quarter of the gradient (Figure 6a). Conversely, those containing sodium channel mutations (Figure 6b) exhibited no positional preference (20–25% in the warmest quarter–the fraction that would be expected if individuals had distributed themselves randomly). A generalised linear model showed that slopes of the lines fitted to the two groups of flies (kdr and non-kdr) were significantly different from each other (F1,80=17.93, P<0.001) as a result of an increasing number of non-kdr flies moving to the warmer part of the temperature gradient. The intercepts were similar (F1,80<3.84, P>0.05) because, at the beginning of each observation, flies were introduced at the centre of the temperature gradient.

Figure 6
figure 6

Responses of (a) the four strains without kdr/super-kdr mutations and (b) the three strains with kdr/super-kdr mutations on exposure to a 19–29°C temperature gradient (0–24 cm). Each point represents the mean response of a single housefly strain, the collective movement of which is charted over a 20 min period.

Full size image

Discussion

Intuitively, some types of resistance mechanism are more likely to confer fitness disadvantages than others. Attempts to review the relations between fitness and resistance mechanism (Roush and McKenzie, 1987) concluded that mechanisms based on the overproduction of detoxifying enzymes (eg carboxylesterases) often lead to serious disadvantages whereas kdr-type mechanisms of pyrethroid resistance may confer little or no effect on overall fitness. The results of our study concur with this conclusion in that genotypes conferring high carboxylesterase titres were shown, in aphids, to be associated with an inability to move from senescing leaves at low temperature. However, work on the behavioural correlates of resistance genotypes in both aphids and houseflies suggests that kdr mutations are also associated with detrimental behaviour.

Our study confirms earlier reports of biological and behavioural effects associated with high levels of carboxylesterase resistance in M. persicae (Foster et al, 1996, 1997), but the use of clones differing in both esterase titre and kdr genotype enabled a clearer resolution for the role of each mechanism in influencing responsiveness to environmental cues (either leaf deterioration or the presence of alarm pheromone). Increased esterase production was significantly associated with reduced movement from deteriorating leaves. A possible explanation of this and other side effects of esterase resistance in M. persicae is the large investment of resources in carboxylesterase production which in the most resistant (R3) clones account for ca. 3% of the total body protein (Devonshire and Moores, 1982). It has been shown that such high carboxylesterase titres reflect a proportionate increase in gene copy number, rising to approximately 80 copies of either E4 or FE4 in R3 M. persicae (Field et al, 1999).

Interestingly, esterase-revertant clones, carrying large numbers of unexpressed esterase genes, did not affiliate with true esterase-susceptible forms but instead showed similar responses, to alarm pheromone and leaf deterioration, to clones expressing high levels of esterase. This suggests that reduced response in this case was not directly related to phenotypic esterase overproduction, but rather to the presence of amplified carboxylesterase genes. It is worth noting that the kdr genotype was not associated with reduced movement from deteriorated leaves. This may reflect the complexity of the behavioural response in this instance. Mutations such as kdr, with direct effects upon nerve function, may be more likely to disrupt the perception of simple stimuli such as temperature or pheromones. Dispersal from deteriorating leaves, however, is probably a more intricate response dependent upon complex stimuli, and so may be less obviously associated with nerve function.

In a prior study (Foster et al, 1999), showing a correlated response to alarm pheromone with the presence or absence of kdr, only a single dose of alarm pheromone was used and those clones utilised did not display the same range of genotypes, nor were they as well characterised as the current subset. In the current study, the proportion of aphids disturbed by the presence of alarm pheromone increased with stimulus concentration for all three kdr genotypes, but there were clear differences between them. Heterozygotes (kdr-SR) were less responsive than homozygous susceptible forms (kdr-SS), and kdr-RR forms were least responsive of all. Once adjusted, statistically, for the presence of kdr, response to alarm pheromone was also inversely associated with carboxylesterase titres; with R3 aphids showing a lower response than R2 forms. An esterase-revertant clone (794J-rev) showed a response similar to the R3 clone from which it was derived. Although both possess the esterase-R3 genotype, only the nonrevertant clone (794J) expressed this enzyme, implying that it is the genotype to which the pheromone response is allied, and not the phenotype. This reduced response to alarm pheromone is likely to be a highly maladaptive behaviour, as the dispersal of aphids in response to this stimulus is considered to be an important behavioural adaptation for avoiding attack by natural enemies (Pickett et al, 1992)

Of the M. domestica strains used in this study, 213ab, 530, and 579 were confirmed as carrying kdr and/or super-kdr mutations and, as expected, this correlated with phenotypic resistance to both DDT and deltamethrin. The other four strains were susceptible to at least one of these compounds, and did not carry the kdr mutation. Our LD50 values for the strains Cooper 530 and 579 compare favourably with figures published previously. Farnham et al (1987) showed that kdr alone conferred 17- and 34-fold resistance to DDT and deltamethrin and 63- and 221-fold resistance, respectively, when super-kdr was also present. Our data showed 7- and 13-fold resistance conferred by kdr alone to DDT and deltamethrin, and 45- and 497-fold resistance, respectively with the addition of the super-kdr mutation. These mechanisms are therefore clearly stable in unselected laboratory populations.

The differential temperature preference shown by the kdr and non-kdr groups of flies strongly suggests an association between the presence of insecticide-resistant sodium channels and the perception of (or at least the response to) temperature gradients. It is worth noting that the genetic background of the strains 530 and 579 was extremely similar to that of the Cooper strain, having undergone a considerable programme of backcrossing in order to homogenise the genetic background in which the kdr and super-kdr genes existed (Farnham et al, 1987). The majority of individuals of all four kdr-susceptible strains had, within 20 min, moved to the warmest quarter of the gradient. Conversely, those carrying kdr and/or super-kdr sodium channel mutations exhibited no positional preference. It is difficult to assess whether this lack of preference is maladaptive, but if it represents a reduced ability to seek out optimal temperatures, then kdr strains will suffer decreased longevity, fecundity, and ovarian development rates. All these parameters are affected by temperature in houseflies (Elvin and Krafsur, 1984; Fletcher et al, 1990; Lysyk, 1991a, 1991b).

In pest-management terms, the pleiotropic effects described above clearly have the potential to exert a large influence on the selection and dynamics of genes conferring insecticide resistance. Such influences will be most obvious in the absence of selection pressure by insecticides and under stressful environments. It is therefore of note that the incidence of R2 and R3 M. persicae aphids in the UK decreases over the winter period when insecticidal selection pressure is relaxed, but that the proportion of kdr-resistant genotypes in the population has appeared to remain stable in recent years, perhaps reflecting the high amount of pyrethroid use (Foster et al, 2002). In northern Europe, kdr is the major mechanism of pyrethroid resistance in houseflies (Keiding, 1999). In Danish piggeries, resistance to pyrethroids is increasingly common, and this may reflect the near continuous selection pressure exerted in these environments (Kristensen et al, 2001). We are unaware of any attempts to monitor the frequency of kdr alleles in areas where pyrethroid use is decreasing.

The results presented above show that although the mechanisms that mediate the reduced responses to both leaf deterioration and alarm pheromone in insecticide-resistant M. persicae, and differential movement in response to a temperature gradient in M. domestica remain equivocal, there is increasingly sound evidence that they are pleiotropic effects of genes coding for insecticide resistance. In the case of the kdr sodium channel mutation specifically, this conclusion is strengthened by evidence that explicitly links qualitative and quantitative changes in the sodium channel structural gene para (paralytic) in Drosophila melanogaster to behavioural disturbances. For example, the temperature-sensitive paralysis exhibited by individuals carrying the point mutation napts (no action potential, temperature sensitive) results from a reduction in the expression of the para gene. It is thought that, when temperature rises, an increasing fraction of the available sodium channels is required to maintain propagation of action potentials. Thus, when sodium channel expression is decreased, the number of available channels cannot meet the demands of elevated temperature (Kernan et al, 1991). The mutation tipE (temperature-induced paralysis locus E) also disrupts para expression and confers temperature-induced paralysis as a result of a decrease in sodium channel numbers (Feng et al, 1995) The sbl (smellblind) mutation, which is not temperature sensitive, is also an allele of the para gene (Lilly et al, 1994). This associates sodium channel mutations with olfactory and chemotactic defects (Lilly and Carlson, 1990) and with changes in sexual behaviour (Tompkins et al, 1982, 1983). In another unrelated point mutation conferring resistance to the cyclodienes, a mutation in the structural gene encoding the γ-aminobutyric acid receptor (rdl or resistance to dieldrin) also appears to induce reversible temperature-sensitive paralysis (Ffrench-Constant et al, 2000).

Studies on kdr/super-kdr mutations incorporated into the insect para gene and expressed in vitro using Xenopus oocytes have revealed that these mutations reduce the sensitivity of the channel protein to pyrethroids and DDT and therefore confer resistance. Such mutations almost certainly have the same function in aphids, houseflies, and in all other insects in which they occur (Vais et al, 2001). However, in addition to making the channel protein less sensitive to some insecticides, these mutations also alter the normal gating properties of the sodium channel. This causes an elevation in their action potential threshold that, in vivo, would cause a general reduction in the excitability of the nervous system (Smith et al, 1997; Vais et al, 1997; Lee et al, 1999; Vais et al, 1997), and is likely to cause changes in the way that resistant individuals respond to external stimuli.

Such behavioural associations have wider implications beyond that of the sustainability of pest-management strategies alone. There is currently a debate over whether the pleiotropic effects of adaptive mutations can contribute to assortative mating and, therefore, speciation (Dieckmann and Doebeli, 1999; Emelianov et al, 2001). If the collective influence of genes conferring insecticide resistance on behavioural attributes, such as short- or long-range dispersal, can contribute to assortative mating between insects, it is possible that the pleiotropic effects of insecticide resistance might promote wider genetic differentiation than that which occurs at resistance loci alone.