Insecticide resistance in the bed bug comes with a cost

Adaptation to new environmental stress is often associated with an alteration of one or more life history parameters. Insecticide resistant populations of insects often have reduced fitness relative to susceptible populations in insecticide free environments. Our previous work showed that three populations of bed bugs, Cimex lectularius L., evolved significantly increased levels of resistance to one product containing both β-cyfluthrin and imidacloprid insecticides with only one generation of selection, which gave us an opportunity to explore potential tradeoffs between life history parameters and resistance using susceptible and resistant strains of the same populations. Life history tables were compiled by collecting weekly data on mortality and fecundity of bugs from each strain and treatment throughout their lives. Selection led to a male-biased sex ratio, shortened oviposition period, and decreased life-time reproductive rate. Generation time was shortened by selection, a change that represents a benefit rather than a cost. Using these life history characteristics we calculated that there would be a 90% return to pre-selection levels of susceptibility within 2- 6.5 generations depending on strain. The significant fitness costs associated with resistance suggest that insecticide rotation or utilization of non-insecticidal control tactics could be part of an effective resistance management strategy.

began to evolve in one generation in the laboratory 22 . This rapid evolution under laboratory conditions gave us an opportunity to explore the hypothesis that life history costs would be associated with decreased susceptibility to the combination insecticide in an environment no longer containing the insecticide.

Results
Mortality levels caused by the selected exposure times to label-rate Temprid SC in the second filial generation of the unselected line were very similar to the parental lines as would be expected in the absence of genetic drift (Table 1). Mortality in selected lines of LA1, CIN1 and NY1 were reduced by the selection that had occurred two generations earlier. This historical selection affected some, but not  Table S1 for analysis of variance results). Longevity, per cent hatch and per cent reaching the adult stage were not affected by the history of selection (Figs. 1, 2A,B). Unexpectedly, the sex ratio at maturity was biased towards females in unselected lines and biased towards males in selected lines (Fig. 2C). The generation time (G) was significantly less in all populations with a history of insecticide exposure compared to those not selected (Fig. 2D). Generation times were significantly reduced by selection by 5.1, 10.3 and 20.2% in LA1, CIN1, and NY1, respectively. Reproductive rates (R o ) were significantly lower in selected lines compared to their unselected counterparts (Fig. 2E). Reproductive rates were significantly reduced by 15.5, 42.5 and 35.3% in LA1, CIN1 and NY1, respectively. Additionally, the oviposition duration from the first egg laid to the last egg laid was significantly shorter for selected compared to unselected groups (Fig. 2F). Oviposition durations were reduced by 13.8, 29.4, and 46.9% in LA1, CIN1, and NY1, respectively.

Discussion
Three different strains of bed bugs with different initial levels of susceptibility to Temprid SC incurred significant life history costs after selection with this pyrethroid/neonicotinoid combination product 22 .
The unselected strains varied in their levels of resistance to Temprid SC as is reflected in the exposure times that were required to lead to ca. 80% mortality (i.e., LA1< CIN1< NY1). These initial differences in resistance were inversely related to the initial R o s of these strains (i.e., ET 80 s of 0.1 h, 1 h, and 19 h and R o s of 59.2 (± 5.2), 52.6 (± 6.4), and 38.7 (± 5.6) for LA1, CIN1, and NY1, respectively). Similarly, an earlier study investigating population growth potential of different strains of bed bugs with different insecticide susceptibility profiles found that insecticide resistant strains had fitness costs compared to more susceptible strains 25 . Further, and more convincing, support for the hypothesis of a cost associated with resistance comes from the observation that parental selection for resistance resulted in decreased R o s in the F 2 of all three populations. Similar tradeoffs between insecticide resistance and life history parameters have been recorded in other insect pests [3][4][5][6][7][8]26 .
In this system, one possible mechanism of the observed costs could be a tradeoff between production of detoxifying enzymes and allocation of resources for fecundity. A previously published study investigating the molecular mechanisms of resistance in the CIN1 selected strain found that four cytochrome P450s and one carboxylesterase was significantly over expressed compared to the CIN1 unselected line 27 . Increased detoxification is likely at least one mechanism of resistance in the LA1 and NY1 selected strains as well. Research investigating mechanisms of resistance in many populations of C. lectularius found that the P450 class of enzymes frequently confers a level of resistance to pyrethroid insecticides 16,[27][28][29][30][31] . Given this information, resources may be shunted from fecundity to the production of detoxifying enzymes. This implies a significant metabolic cost associated with the production of these enzymes. Further experiments involving the use of RNA interference could be used to elucidate the molecular mechanism of the observed costs and their correlation with detoxification-mediated insecticide resistance.
The R o ratios of selected and unselected lines are 0.70, 0.42 and 0.57 for LA1, CIN1 and NY1, respectively. Numbers greater than 1 suggest an advantage, whereas numbers less than one suggest a disadvantage. Thus, a reversion to susceptibility should occur rapidly in insecticide-free environments. We calculated the time course to 50 and 90% recovery of pre-insecticide selection levels of susceptibility using a model proposed earlier 32 . T r denotes the time required to reach a significant degree of susceptibility, G denotes generation time, p f denotes the proportion of the population required to be susceptible, p 0 denotes the initial proportion susceptible, w s denotes the fitness of the susceptible and w r denotes the fitness of the resistant. The R o s for each treatment and strain were used for the fitness variables w s and w r . A return to 50% of pre-selection levels of susceptibility would take 3.03, 4.88 and 0.71 generations for LA1, CIN1 and NY1, respectively. Additionally, the same model predicted that the number of generations required for a strain to return to 90% level of pre-selection susceptibility was 6.51, 5.95 and 2.06 for LA1, CIN1 and NY1, respectively. Given that Gs were less than 20 weeks, reversion to 90% pre-selection levels of resistance should occur within ca. 1 to 2.5 years in the absence of selection. Currently, the combination pyrethroid/neonicotinoid products are some of the most effective choices for control in the field 21,22 . In theory, rotation to products utilizing alternative modes of action could reverse resistance 3,10,11 assuming that alleles for susceptibility still exist in the population. However, such reversion is complicated by public intolerance for resident populations (which contrasts with agricultural resistance management 11 ), limited gene flow that would help re-establish susceptibility 33 , limited choices of alternative chemical classes 21 and independence of actions of pest managers when a more coordinated approach might be required. Despite the complications, integration of, or rotation among, both chemical and non-chemical approaches (heat, vacuuming, encasement of beds, etc.) will be necessary for sustained management of populations of bed bugs.

Material and Methods
Insects. Three strains of bed bugs were used for this study. The LA1 strain was collected from Los Angeles in 2007 and was susceptible to pyrethroids 13 . The strain CIN1 was originally collected from Cincinnati, OH in 2005 and was resistant to pyrethroids 13 . Subsequently, its originally high level of pyrethroid resistance has declined but not to the level of the susceptible colony LA1 27 . The NY1 strain was collected from New York City, NY in 2007 and was resistant to pyrethroids 17 . However, a reversion toward susceptibility has also been recorded for this strain, though not to the degree of CIN1 27 . For each strain, two samples of bugs were selected overtime, and two separate lineages of selected and unselected strains were initiated for each strain (LA1, CIN1 and NY1; Fig. 3) by exposing these strains to residual  Table S1, and p-values depicted in the figure represents the effect of selection on the specific parameter investigated.
Scientific RepoRts | 5:10807 | DOi: 10.1038/srep10807 deposits of the pyrethroid/neonicotinoid combination product Temprid SC for a time calculated to kill 80% of the population (ET 80 ) at the label rate 22 . All subsequent evaluations of susceptibility were performed using the same exposure times and bioassay. Bugs from an untreated F 1 generation were used to establish an F 2 generation (Fig. 3). Insects were housed in incubators away from any insecticide exposure at 26.7° C, 65 ± 5% RH, and a photoperiod of 14:10 (L:D) h. All bed bugs were fed weekly on defibrinated rabbit blood (Quad Five, Ryegate, MT) that was warmed to 39 °C with a circulating water bath using an artificial membrane feeder 34 . Life history variables. Life history data were collected for each strain and treatment. A group of 20 eggs was gathered within a 24 hour period from three to 12 females (Supplemental Table S2) and maintained within a small petri dish (5.1 cm diameter) lined with black filter paper. These eggs were allowed to hatch and the resulting individuals were fed weekly for the duration of their lives. Adult offspring were allowed to mate ad libitum. Longevity and fecundity were recorded weekly in order to calculate per cent hatching, per cent reaching adulthood, proportion of females, G, R o and oviposition duration. In addition, this information was used to generate weekly survival (l x ; female specific survival could not be determined initially due to a lack of information about sex until eclosion to the adult stage) and fecundity (m x ; in this case oviposition rates). These recorded values were used to calculate net R o as the number of female offspring per individual per generation. Because sex determination was only made during the adult stage, estimates of l x , m x , and R o were based on an assumption of a 1:1 sex ratio of eggs laid by the F 2 generation. The F 2 generation for each replicate was followed for up to 70 weeks when the last individual died. Within strains, samples of replicates 1 and 2 were separated in time by 2 to 4 weeks.

Data analysis.
A nested analysis of variance was used to investigate the effects of strain, treatment and replicate(strain) on the hatch rate of eggs laid by F 1 mothers, the per cent of those eggs that reach adulthood, the observed sex ratios after reaching maturity, G, R o and oviposition duration [Systat software. SYSTAT 13. San Jose, CA (2008)]. Net R o was calculated as the sum of the weekly l x *m x for each replicate with the assumption of a 1:1 sex ratio of eggs laid by F 2 females. Generation time was calculated by taking the summation of l x *m x *x divided by the summation of l x *m x 35 . A Z-test was performed to investigate the contrast in susceptibility to Temprid SC at the ET 80 of unselected and selected bugs in the  Experimental design for selection experiment. Bugs were exposed to label rate Temprid SC ® for a time calculated to kill 80% of the respective strain of bed bugs 22 . Two parental lines of selected and unselected bugs were created for each strain. Using the subsequent F 2 generation from these lines that receive no further exposure to insecticide, two samples were taken from each replicate to generate data. Replicates were not synchronous and separated by weeks.