The evolution of adaptive behavior often requires changes in sensory systems. However, rapid adaptive changes in sensory traits can adversely affect other fitness-related behaviors. In the German cockroach, a gustatory polymorphism, ‘glucose-aversion (GA)’, supports greater survivorship under selection with glucose-containing insecticide baits and promotes the evolution of behavioral resistance. Yet, sugars are prominent components of the male’s nuptial gift and play an essential role in courtship. Behavioral and chemical analyses revealed that the saliva of GA females rapidly degrades nuptial gift sugars into glucose, and the inversion of a tasty nuptial gift to an aversive stimulus often causes GA females to reject courting males. Thus, the rapid emergence of an adaptive change in the gustatory system supports foraging, but it interferes with courtship. The trade-off between natural and sexual selection under human-imposed selection can lead to directional selection on courtship behavior that favors the GA genotype.

Cockroach strains. All cockroaches were maintained on rodent diet (Purina 5001, PMI Nutrition International, St. Louis, MO) and distilled water at 27°C, ~40% RH and a 12:12 h L:D cycle. The WT colony (Orlando Normal) was collected in Florida in 1947 and has served as a standard insecticide-susceptible strain. The GA colony (T-164) was collected in 1989, also in Florida, and shown to be aversive to glucose; continued artificial selection with glucose-containing toxic bait fixed the homozygous GA trait in this population (approximately 150 generations as of 2020).

Generating recombinant lines and life history data. To homogenize the genetic backgrounds of the WT and GA strains, two recombinant colonies were initiated in 2013 by crossing 10 pairs of WT♂ × GA♀ and 10 pairs of GA♂ × WT♀ (Fig. 3a). At the F8 generation (free bulk mating without selection), 400 cockroaches were tested in two-choice feeding assays (see below) that assessed their initial response to tastants, as described in previous studies^{11, 26}. The cockroaches were separated into glucose-accepting and glucose-rejecting groups by the rapid Acceptance-Rejection assay (described in Feeding Bioassays). These colonies were bred for three more generations, and 200 cockroaches from each group were assayed in the F11 generation and backcrossed to obtain homozygous glucose-accepting (aa) and glucose-averse (AA) lines. Similar results were obtained in both directions of the cross, confirming previous findings of no sex linkage of the GA trait²⁷. These two lines were defined as WT_aa (homozygotes, glucose-accepting) and GA_AA (homozygotes, glucose-averse). To obtain heterozygous GA cockroaches, GA_Aa, a single intercross group was generated from crosses of 10 pairs of WT_aa♂ × GA_AA♀ and 10 pairs of GA_AA♂ × WT_aa♀.

The GA trait follows Mendelian inheritance. Therefore, we used backcrosses, guided by two-choice feeding assays and feeding responses in Acceptance-rejection assays, to determine the homozygosity of WT and GA cockroaches. The cross of WT♂ × WT♀ produced homozygous F1 cockroaches showing maximal glucose-acceptance. The cross of GA♂ × GA♀ produced homozygous F1 cockroaches showing maximal glucose-aversion. The cross of WT × GA produced F1 heterozygotes with intermediate glucose-aversion. When the F1 heterozygotes were backcrossed with WT cockroaches, they produced F2 cockroaches with a 1:1 ratio of WT and GA phenotypes.

The two-choice feeding assay assessed whether cockroaches accepted or rejected glucose (binary: yes-no). Insects were held for 24 hrs without water, or starved without food and water. Either 10 adults or 2 day-old first instar siblings (30–40) were placed in a Petri dish (either 90 mm or 60 mm diameter × 15 mm height). Each Petri dish contained two agar discs: one disc contained 1% agar and 1 mmol l^-1 red food dye (Allura Red AC), and the second disc contained 1% agar, 0.5 mmol l^-1 blue food dye (Erioglaucine disodium salt) and either 1000 mmol l-1 or 3000 mmol l^-1 glucose. The assay duration was 2 hrs during the dark phase of the insects’ L:D cycle. After each assay, the color of the abdomen of each cockroach was visually inspected under a microscope to infer the genotype.

We assessed whether the recombinant colonies had different traits from the parental WT and GA lines. We paired single newly eclosed females (day 0) with single 10–12 days-old males of the same line in a Petri dish (90 mm diameter, 15 mm height) with fresh distilled water in a 1.5 ml microcentrifuge tube and a pellet of rodent food, and monitored when they mated. When females formed egg cases, each gravid female was placed individually in a container (95 × 95 × 80 mm) with food and water until the eggs hatched. After removing the female, her offspring were monitored until adult emergence. We recorded the time to egg hatch, first appearance of each nymphal stage, first appearance of adults and the end of adult emergence. The first instar nymphs and adults in each cohort were counted to obtain measures of survivorship. Although there were significant differences in some of these parameters across all four strains, we found no significant differences between the two recombinant lines, except mating success, which was significantly lower in GA_AA♀ than WT_aa♀ (Supplementary Table 11).

Mating bioassays. All mating sequences were recorded using an infra-red-sensitive camera (Polestar II EQ610, Everfocus Electronics, New Taipei City, Taiwan) coupled to a data acquisition board and analyzed by searchable and frame-by-frame capable software (NV3000, AverMedia Information) at 27°C, ~40% RH and a 12:12 h L:D cycle. For behavioral analysis, tested pairs were classified into two groups: mated (successful courtship) and not-mated (failed courtship). Four distinct behavioral events (Fig. 1c, Contact, Wing raising, Nuptial feeding, and Copulation) were analyzed using seven behavioral parameters as shown in Supplementary Table 2.

We extracted behavioral data from successful courtship sequences, defined as courtship that led to Copulation. For failed courtship sequences, we extracted the behavioral data from the first courtship of both mated and not-mated groups, because most pairs in both groups failed to copulate in their first encounter, and there were no significant differences in behavioral parameters between the two groups.

To assay female choice, we conducted two-choice mating assays (Fig. 1a). A single focal WT♀ or GA♀ and two males, one WT and one GA, were placed in a Petri dish (90 mm diameter, 15 mm height) with fresh distilled water in a 1.5 ml microcentrifuge tube and a pellet of rodent food (n = 25 WT♀ and 27 GA♀). To assay male choice, a single focal WT♂ or GA♂ was given a choice of two females, one WT♀ and one GA♀ (n = 27 WT♂ and 18 GA♂). Experiments were started using 0 day-old sexually unreceptive females and 10–12 days-old sexually mature males. Newly emerged (0 day-old) females were used to avoid the disruption of introducing a sexually mature female into the bioassay. B. germanica females become sexually receptive at 5–7 days of age, so the mating behavior of the focal insect was video-recorded for several days until they mated. Fertility of mated females was evaluated by the number of offspring produced. We assessed the gustatory phenotype of nymphs (either WT-type or GA-type) to determine which of the two adult cockroaches mated with the focal insect. Each gravid female was maintained individually in a container (95 × 95 × 80 mm) with food and water until the eggs hatched. Two day-old first instar nymphs were starved for one day without water and food, and then they were tested in Two-choice feeding assays using 1000 mmol l^-1 glucose-containing agar with 0.5 mmol l^-1 blue food dye vs. plain sugar-free agar with 1 mmol l-1 red food dye. If all the nymphs chose the glucose-containing agar, their parents were considered WT♂ and WT♀. When all the nymphs showed glucose-aversion, they were raised to the adult stage. Newly emerged adults were backcrossed with WT cockroaches, and their offspring were tested in the Two-choice assay. When the parents were both GA, 100% of the offspring exhibited glucose-aversion. When the parents were WT and GA, the offspring showed a 1:1 ratio of glucose-accepting and glucose-aversive behavior. Mate choice, mating success ratio and the number of offspring were analyzed statistically.

We conducted no-choice mating assay using the WT and GA strains (Figs. 1b and 1d). A female and a male were placed in a Petri dish with fresh water and a piece of rodent food and video-recorded for 24 hrs. The females were 5–7 days-old and males were 10–12 days-old. Four treatment pairs were tested: WT♂ × WT♀ (n = 20, 18 and 14 pairs for 5, 6 and 7 day-old females, respectively); GA♂ × GA♀ (n = 23, 22 and 35 pairs); GA♂ × WT♀ (n = 21, 14 and 17 pairs); and WT♂ × GA♀ (n = 33, 19 and 15 pairs).

To confirm that gustatory stimuli guide nuptial feeding, we artificially augmented the male nuptial secretion and assessed whether the duration of nuptial feeding and mating success of GA♀ were affected (Fig. 2c). Before starting the mating assay with 5 day-old GA♀, 10–12 days-old WT♂ were separated into three groups: A control group did not receive any augmentation; A water control group received distilled water with 1 mmol l^-1 blue dye (+Blue); A fructose group received 3000 mmol l^-1 fructose solution with blue dye (+Blue+Fru). Approximately 50 nl of the test solution was placed into the tergal gland reservoirs using a glass microcapillary. No-choice mating assays were carried out for 24 hrs. n = 20–25 pairs for each treatment.

We evaluated the association of short nuptial feeding (Fig. 1C) and the GA trait we conducted no-choice mating assays using females from the recombinant lines (Fig. 3c). Before starting each mating assay with 4 day-old females from the WT, GA and recombinant lines (WT_aa, GA_AA and GA_Aa), the EC50 for glucose was obtained by the instantaneous Acceptance-Rejection assay using 0, 10, 30, 100, 300, 1000 and 3000 mmol l^-1 glucose (WT♀ and WT_aa♀, non-starved; GA♀, GA_AA♀ and GA_Aa♀, 1-day starved). After the Acceptance-Rejection assay, GA_Aa♀ were separated into two groups according to their sensitivity for rejecting glucose; the GA_Aa_high sensitivity group rejected glucose at 100 and 300 mmol l^-1, whereas the GA_Aa_low sensitivity group rejected glucose at 1000 and 3000 mmol l^-1. We paired these females with 10–12 days-old WT♂ (n = 15 WT_aa♀, n = 20 GA_AA♀, n = 20 GA_Aa_high♀ and n = 17 GA_Aa_low♀).

Feeding bioassay. We conducted two feeding assays: Acceptance-Rejection assay and Consumption assay. The Acceptance-Rejection assay assessed the instantaneous initial responses (binary: yes-no) of cockroaches to tastants, as previously described7, 22, 27. Briefly, acceptance means that the cockroach started drinking. Rejection means that the cockroach never initiated drinking. The percentage of positive responders was defined as the Number of insects accepting tastants/Total number of insects tested. The effective concentration (EC50) for each tastant was obtained from dose-response curves using this assay. The Consumption assay was previously described²⁷. Briefly, we quantified the amount of test solution females ingested after they started drinking. Females were observed until they stopped drinking, and we considered this a single feeding bout.

We used the Acceptance-Rejection assay and Consumption assay, respectively, to assess the sensitivity of 5 day-old WT♀ and GA♀ for accepting and consuming the WT♂ nuptial secretion (Figs. 2a and 2b). The secretion was diluted with HPLC-grade water to 0.001, 0.01, 0.03, 0.1, 0.3 and 1 male-equivalents/µl (n = 20 non-starved females each). The amount of nuptial secretion consumed was tested at 0.1 male-equivalents/µl in the Consumption assay (n = 10 each).

The Acceptance-Rejection assay was used to calculate the effective concentration (EC50) of glucose for females in the WT, GA and recombinant lines (Figs. 3a and 3b). A glucose concentration series of 0.1, 1, 10, 100 and 1000 mmol l^-1 was tested with one-day starved 4-day old females (n = 65 GA_Aa♀, n = 50 GA_AA♀ and n = 50 GA♀) and non-starved females (n = 50 WT_aa♀ and n = 16 WT♀).

The effects of female saliva on feeding responses of 5 day-old WT♀ and GA♀ were tested using the Acceptance-Rejection assay (Fig. 4a). Freshly collected saliva of WT♀ and GA♀ was immediately used in experiments. Assays were prepared as follows: 3 µl of 200 mmol l^-1 maltose or maltotriose were mixed with 3 µl of either HPLC-grade water or saliva of WT♀ or GA♀. The final concentration of each sugar was 100 mmol l^-1 in a total volume of 6 µl. This concentration represented approximately the acceptance EC70 for WT♀ and GA♀27. Nuptial secretion (1 µl representing 10 male-equivalents) was mixed with 1 µl of either HPLC-grade water or saliva from WT♀ or GA♀, and 8 µl of HPLC-grade water was added to the mix. The final concentration of the nuptial secretion was 1 male-equivalent/µl in a total volume of 10 µl. This concentration also represented approximately the acceptance EC70 for WT♀ and GA♀ (Fig. 2A). The mix of saliva and either sugar or nuptial secretion was incubated for 300 sec at 25°C. Additionally, we tested the effect of only saliva in the Acceptance-Rejection assay. Either 1-day starved or non-starved females were tested with water only and then a 1:1 mixture of saliva and water. Saliva alone did not affect acceptance or rejection of stimuli. n = 20–33 females from each strain.

To evaluate whether salivary enzymes are involved in the hydrolysis of oligosaccharides, the contribution of salivary glucosidases was tested using the glucosidase inhibitor acarbose in the Acceptance-Rejection assay (Fig. 4b), as previously described27. We first confirmed that the range of 0–125 mmol l-1 acarbose in HPLC-grade water did not disrupt the acceptance and rejection of tastants. Test solutions were prepared as follows: 2 µl of either HPLC-grade water or saliva of GA♀ was mixed with 1 µl of either 250 µmol l^-1 of acarbose or HPLC-grade water, then the mixture was added to 1 µl of 400 mmol l^-1 of either maltose or maltotriose solution. The total volume was 4 µl, with the final concentration of sugar being 100 mmol l^-1. For assays with nuptial secretion, 1 µl of either HPLC-grade water or saliva from 5 day-old GA♀ was mixed with 0.5 µl of either 250 µmol l^-1 of acarbose or HPLC-grade water. This mixture was added to 0.5 µl of 10 male-equivalents of nuptial secretion (i.e., 20 male-equivalents/µl). HPLC-grade water was added for a total volume of 10 µl and a final concentration of 1 male-equivalent/µl. The mix of saliva and either sugars or nuptial secretion was incubated for 5 min at 25°C. All test solutions contained blue food dye. Test subjects were 5 day-old GA♀ and 20–25 females were tested in each assay.

Nuptial secretion and saliva collections. The nuptial secretion of WT♂ was collected by the following method: Five 10–12 days-old males were placed in a container (95 × 95 × 80 mm) with 5 day-old GA♀. After the males displayed wing-raising courtship behavior toward the females, individual males were immediately decapitated and the nuptial secretion in their tergal gland reservoirs was drawn into a calibrated borosilicate glass capillary (76 × 1.5 mm) under the microscope. The nuptial secretions from 30 males were pooled in a capillary and stored at -20°C until use. Saliva from 5 day-old WT♀ and GA♀ was collected by the following method: individual females were briefly anesthetized with carbon dioxide under the microscope and the side of the thorax was gently squeezed. A droplet of saliva that accumulated on the mouthparts was then collected into a microcapillary (10 µl, Kimble Glass). Fresh saliva was immediately used in experiments.

GC-MS procedures for analysis of sugars. Standards of D-(+)-glucose (Sigma-Aldrich), D-(+)-maltose (Fisher Scientific) and maltotriose (Sigma-Aldrich) were diluted in HPLC-grade water (Fisher Scientific) at 10, 50, 100, 500 and 1000 ng/µl to generate calibration curves. Samples were vortexed for 20 sec and a 10 μl aliquot of each sample was transferred to a Pyrex reaction vial containing a 10 μl solution of 5 ng/μl sorbitol (≥98%) in HPLC-grade water as internal standard and dried under a gentle flow of N2 for 20 min.

Samples containing degradation products from nuptial secretions were prepared by adding 15 μl of HPLC-water to each sample in a 1.5 ml Eppendorf tube, vortexed for 30 sec and centrifuged at 8000 rpm (5223 RCF) for 5 min to separate lipids from the water layer. The water phase was transferred to a reaction vial using a glass capillary. This procedure was repeated with the remaining lipid layer and the water layers were combined in the same reaction vial containing 10 μl of a solution of 5 ng/μl sorbitol and dried under N₂ for 20 min.

For derivatization of sugars and samples, each reaction vial received 12 μl of anhydrous pyridine under a constant N2 flow, then vortexed and incubated at 90°C for 5 min. Three μl of N-methyl-N-(trimethylsilyl)trifluoroacetamide (MSTFA; Sigma-Aldrich) was added to each reaction vial and centrifuged at 1000 rpm (118 RCF) for 2 min. Vials were incubated in a heat block at 90°C for 1.5 hr and vortexed every 10 min for the first 30 min of incubation.

The total volume of sample was ~10 μl, and 1 μl was injected into the GC-MS (6890 GC coupled to a 5975 MS, Agilent Technologies, Palo Alto, CA). The inlet was operated in splitless mode (17.5 psi) at 290°C. The GC was equipped with a DB-5 column (30 m, 0.25 mm, 0.25 μm, Agilent), and helium was used as the carrier gas at an average velocity of 50 cm/s. The oven temperature program started at 80°C for 1 min, increased at 10°C/min to 180°C, then increased at 5°C/min to 300°C, and held for 10 min. The transfer line was set at 250°C for 24 min, ramped at 5°C/min to 300°C and held until the end of program. The ion source operated at 70eV and 230°C, while the MS quadrupole was maintained at 200°C. The MSD was operated in scan mode, starting after 9 min (solvent delay time) with a mass range of 33–650 AMU.

For GC-MS data analysis, the sorbitol peak area was obtained from the extracted ion chromatograms with m/z = 205, the sorbitol base peak. The area of peaks of glucose, maltose and maltotriose were obtained from the extracted ion chromatograms using m/z = 204, the base peak of the three sugars. The most abundant peaks of each sugar were selected for quantification36, and these peaks did not coelute with other peaks. Then, the peak areas of the three sugars were divided by the area of the respective sorbitol peak in each sample to normalize the data and to correct technical variability during sample processing. This procedure was performed to obtain the calibration curves and quantification of sugars in our experiments.

The results of sugar analysis using GC-MS are reported in Supplementary Figures 1-4.

Analysis of nuptial secretions. We focused the GC-MS analysis on glucose, maltose and maltotriose in WT♂ nuptial secretion (Fig. 4c). To quantify the time-course of saliva-catalyzed hydrolysis of WT♂ nuptial secretion to glucose, 1 µl of GA♀ saliva was mixed with 1 µl of 10 male-equivalents/µl. We incubated the mixtures for 0, 5, 10 and 300 sec at 25°C, and added 4 µl of methanol to stop the enzyme activity (n = 5 each treatment). Each sample contained the nuptial secretions of 5 males to obtain enough detectable amount of sugars. For the statistical analysis, the amounts of sugars were divided by 5 to obtain the amount of sugars in 1 male (1 male-equivalent). These amounts were also used for generating Figure 4c and Supplementary Table 9. In calculations of the concentration of the three sugars (mmol l^-1), the mass and volume of the nuptial secretion were measured using 70–130 male-equivalents of undiluted secretion of each strain (n = 3). The mass and volume of the nuptial secretion/male, including both lipid and aqueous layers, were approximately 30–50 µg and 40–50 nl. Because it was difficult to separate the lipid layer from the water layer at this small scale, we roughly estimated that the tergal reservoirs of the four cockroach lines had 30 nl of aqueous layer that contained sugars.

To quantify the time-course of saliva-catalyzed hydrolysis of maltose and maltotriose to glucose, 1 µl of GA♀ saliva was mixed with 1 µl of 200 mmol l^-1 of either maltose or maltotriose (Figs. 4d and 4e). Incubation time points were 0, 5, 10 and 300 sec at 25°C and methanol was used to stop the enzyme activity. Controls without saliva were also prepared using HPLC-grade water instead of saliva and 300 sec incubations. n = 5 for each treatment. 

Photomicroscopy. The photographs of the tergal glands and mouthparts (Fig. 5) were obtained using an Olympus Digital camera attached to an Olympus CX41 microscope (Olympus America, Center Valley, PA). 

Statistics and Reproducibility. The sample size and number of replicates for each experiment are noted in the respective section describing the experimental details. In summary, the samples sizes were: Mating bioassays, n = 18–80; Feeding assays, n = 16–65; Sugar analysis, n = 5; Life history parameters, n > 14. All statistical analyses were conducted in R Statistical Software (v4.1.0; R Core Team 2021) and JMP Pro 15.2 software (SAS Institute Inc., Carey, NC). For bioassay data and sugar analysis data, we calculated the means and standard errors, and we used the Chi-square test with Holm’s method for post hoc comparisons, t-test, and ANOVA followed by Tukey’s HSD test (all α = 0.05), as noted in each section describing the experimental details, results, and in Supplementary Tables 1–11.

The file "Rapid_evolution_of_an_adaptive_taste_polymorphism_disrupts_courtship_behavior_2022_Data" contains the raw data for Sapplementary Tables.

List of Data sheet

Data sheet TableS1: Supplementary Table 1. Comparison of successful and failed courtship sequences in two-choice mating assays

Data sheet TableS2: Supplementary Table 2. Comparison of successful and failed courtship sequences in no-choice mating assays

Data sheet TableS3: Supplementary Table 3. Feeding responses (acceptance, rejection) of females to nuptial secretion

Data sheet TableS4: Supplementary Table 4. Effects of augmenting the nuptial secretion with fructose in no-choice mating assays

Data sheet TableS5: Supplementary Table 5. Effective concentration of glucose that elicits feeding acceptance in 50% of 4 day-old WT♀ and WT_aa♀ and feeding rejection in 50% of 4 day-old GA♀, GA_AA♀ and GA_Aa♀

Data sheet TableS6: Supplementary Table 6. No-choice mating assays using 5 day-old females from the recombinant lines

Data sheet TableS7: Supplementary Table 7. Effect of saliva on feeding responses (acceptance, rejection) of WT♀ and GA♀

Data sheet TableS8: Supplementary Table 8. Involvement of salivary glucosidases in sugar degradation and sugar acceptance by GA♀ 

Data sheet TableS9: Supplementary Table 9. Time-course of saliva degradation of nuptial secretion

Data sheet TableS10: Supplementary Table 10. Time-course of saliva degradation of maltose and maltotriose

Data sheet TableS11: Supplementary Table 11. Life history parameters of WT (Orlando Normal), GA (T-164), WT_aa and GA_aa cockroaches