Comparative Analysis of Volatile Defensive Secretions of Three Species of Pyrrhocoridae (Insecta: Heteroptera) by Gas Chromatography-Mass Spectrometric Method

The true bugs (Hemiptera: Heteroptera) have evolved a system of well-developed scent glands that produce diverse and frequently strongly odorous compounds that act mainly as chemical protection against predators. A new method of non-lethal sampling with subsequent separation using gas chromatography with mass spectrometric detection was proposed for analysis of these volatile defensive secretions. Separation was performed on Rtx-200 column containing fluorinated polysiloxane stationary phase. Various mechanical irritation methods (ultrasonics, shaking, pressing bugs with plunger of syringe) were tested for secretion sampling with a special focus on non-lethal irritation. The preconcentration step was performed by sorption on solid phase microextraction (SPME) fibers with different polarity. For optimization of sampling procedure, Pyrrhocoris apterus was selected. The entire multi-parameter optimization procedure of secretion sampling was performed using response surface methodology. The irritation of bugs by pressing them with a plunger of syringe was shown to be the most suitable. The developed method was applied to analysis of secretions produced by adult males and females of Pyrrhocoris apterus, Pyrrhocoris tibialis and Scantius aegyptius (all Heteroptera: Pyrrhocoridae). The chemical composition of secretion, particularly that of alcohols, aldehydes and esters, is species-specific in all three pyrrhocorid species studied. The sexual dimorphism in occurrence of particular compounds is largely limited to alcohols and suggests their epigamic intraspecific function. The phenetic overall similarities in composition of secretion do not reflect either relationship of species or similarities in antipredatory color pattern. The similarities of secretions may be linked with antipredatory strategies. The proposed method requires only a few individuals which remain alive after the procedure. Thus secretions of a number of species including even the rare ones can be analyzed and broadly conceived comparative studies can be carried out.

The true bugs (Hemiptera: Heteroptera) have evolved a system of well-developed scent glands that produce diverse and frequently strongly odorous compounds that act mainly as chemical protection against predators. A new method of non-lethal sampling with subsequent separation using gas chromatography with mass spectrometric detection was proposed for analysis of these volatile defensive secretions. Separation was performed on Rtx-200 column containing fluorinated polysiloxane stationary phase. Various mechanical irritation methods (ultrasonics, shaking, pressing bugs with plunger of syringe) were tested for secretion sampling with a special focus on non-lethal irritation. The preconcentration step was performed by sorption on solid phase microextraction (SPME) fibers with different polarity. For optimization of sampling procedure, Pyrrhocoris apterus was selected. The entire multi-parameter optimization procedure of secretion sampling was performed using response surface methodology. The irritation of bugs by pressing them with a plunger of syringe was shown to be the most suitable. The developed method was applied to analysis of secretions produced by adult males and females of Pyrrhocoris apterus, Pyrrhocoris tibialis and Scantius aegyptius (all Heteroptera: Pyrrhocoridae). The chemical composition of secretion, particularly that of alcohols, aldehydes and esters, is species-specific in all three pyrrhocorid species studied. The sexual dimorphism in occurrence of particular compounds is largely limited to alcohols and suggests their epigamic intraspecific function. The phenetic overall similarities in composition of secretion do not reflect either relationship of species or similarities in antipredatory color pattern. The similarities of secretions may be linked with antipredatory strategies. The proposed method requires only a few individuals which remain alive after the procedure. Thus secretions of a number of species including even the rare ones can be analyzed and broadly conceived comparative studies can be carried out. PLOS  Introduction content with methyl-tert-butylether, hexane or dichlormethane [11,[21][22][23][24][25][26][27][28], or puncturing the gland with a glass capillary [29]. Such procedures are appropriate for studies of large species (e.g. shield bugs, Pentatomidae) in which only a few individuals are needed for dissection and extraction of sufficient quantities of secretion [14,25,30]. On the other hand, the method is overly time-consuming in smaller species in which it is necessary to dissect a large number of individuals (Farine et al. used up to 2000 specimens of Pyrrhocoris apterus [16]). Non-lethal sampling methods include blowing air across a vessel containing the bugs with subsequent collection of the substances on various types of adsorbents, such as Super Q, active carbon or Porapaq Q [31][32][33][34], sorption on a fiber (SPME) [35][36][37] or placing a paper napkin between the bugs and subsequent extraction of the napkin with dichloromethane [38]. The relatively newest procedure involves a temperature-desorption system that was used for analyzing pheromones [39,40]. However, these procedures do not include any irritation of the insects that would simulate a predator attack and assure the discharging of the defensive secretion. Gas chromatography (GC) combined with flame ionization, mass spectrometric or electroantennographic detection has been used for analysis of the secretions of true bugs [23,25,37,[41][42][43][44][45][46][47][48]. Mainly non-polar DB-or HB-5 capillary columns or parallel connected non-polar DB-1 and polar DB-WAX capillary columns have been used [20,28,36,39,41]. Because the secretions contain many different substances, preliminary separation can be carried out using high performance liquid chromatography prior to gas chromatography [49], but recently comprehensive two-dimensional gas chromatography (GCxGC) has begun to be used [24,50]. Substances are identified by comparison with a mass spectrometric database and by nuclear magnetic resonance [33].
Because of the great complexity of the secretion samples and the number of parameters that can affect the sampling process, it can be beneficious to employ the process optimization procedures that combine all the tested parameters together at the same time and substantially reduce the number of analyses required. Response surface methodology (RSM) is based on the fit of a polynomial equation to the multidimensional (multiparametric) experimental data which describes the behavior of the data set with the objective of making statistical predictions. RSM is a general approach for designing experiments; it efficiently reduces the development time and costs of the method. The main advantage of RSM as compared with the currently most widely used approach to optimization, namely one-factor-at-a-time OFAT (one parameter is changed while the others are fixed during optimization) is that only the most important experimental parameters are selected and optimized and that more than one parameter are changed simultaneously in one experiment according to the computer-designed plan of experiments [51][52][53].
In this study three species of the family Pyrrhocoridae, namely the Palaearctic Pyrrhocoris apterus (L.), Oriental P. tibialis (Stål), and Mediterranean Scantius aegyptius (L.) were chosen to analyze and compare their secretions since they are closely related but differ in many biologically important traits. Pyrrhocoris apterus and P. tibialis are congeneric, closely related but largely allopatric species [54]. They have similar host plant (various Malvaceae incl. Tilia spp.; in case of P. apterus also an introduced locust tree, Robinia pseudacacia) but a different life style, and both are reluctant to discharge their defensive secretion upon an attack. P. apterus is gregarious, mostly brachypterous, flightless species, and possesses a warning coloration of conspicuous red-and-black color pattern. P. tibialis is non-gregarious, readily flying, and rather cryptically colored in shades of gray and brown. Scantius aegyptius is a macropterous (but flightless), epigeic, Mediterranean species occupying the southwestern part of the range of P. apterus and sharing with the latter not only similar red-and-black warning coloration, but also the same habitat and major hostplant (Malva neglecta; Malvaceae). Like both Pyrrhocoris species, it is also reluctant to discharge its defensive secretion.
This study was carried out to i) develop a non-lethal method for sampling the defensive secretion of true bugs using various kinds of mechanical irritation that would simulate the attack of a predator, and also find a method suitable for the species that are reluctant to release their secretion; ii) select the most advantageous sampling of the secretion using SPME fibers with various polarities for subsequent GC-MS analysis; iii) optimize the sampling using the response surface methodology; iv) employ the developed method for analysis and comparison of infochemicals secreted by adults of both sexes in the three species of Pyrrhocoridae differing in their biology and antipredatory strategies, and to assess the similarities and differences in secretions.
MTG secretion of P. apterus has already been analyzed by Farine [20] using dissected MTGs and their reservoirs. Except for P. apterus, Dysdercus intermedius Distant, D. superstitiosus (F.), D. fasciatus Signoret, and D. cingulatus (F.) defensive secretion of no other pyrrhocorid species have been analyzed [20], and to our best knowledge such comparisons of secretion in closely related species of Heteroptera are surprisingly rare.

Instrumentation, separation conditions and identification of the volatile components
The analyses were performed using a GCMS-QP2010 Plus instrument (Shimadzu, Japan), equipped with 20 m × 0.15 mm i.d., 0.15 μm film thickness Rtx-200 column (trifluoropropylmethyl polysiloxane stationary phase, Restek, USA). This stationary phase was selected as a general purpose one with respect to its unique ability to separate substances in a relatively broad polarity range. Helium (99.999%, Linde, Czech Republic) was used as the carrier gas at a constant linear flow rate of 35 cm s -1 . Splitless-mode injection with SPME liner at 250˚C was employed (split valve closed for 1 min). The oven temperature was maintained at 35˚C for 3 min, ramped at 5˚C min -1 to 130˚C, then ramped at 20˚C min -1 to 300˚C and then maintained for 5 min (total run time, 35.50 min). The mass spectrometer was operated in the scan mode (m/z 35-500). The ion source and interface temperatures were 200 and 250˚C, respectively.
The data were collected and evaluated using the GCMS software (Shimadzu, Japan), Origin 8 (Origin Lab corporation, Northampton, MA, USA), Microsoft Excel 2003 (Microsoft Corporation, Redmond, DC, USA) and Minitab 16 (Minitab Inc., State College, PA, USA) programs. Identification of the secretion components was made by comparing the obtained spectra with those in the NIST 2008 Mass Spectra Library. Series of n-alkanes (C8-C20) and reference compounds (see Section Chemicals) were analyzed under the same experimental conditions as those used for the samples to either establish the retention indices or confirm the identity of the analytes. Retention times of reference compounds were measured by injection of their diluted solutions in pentane. The dilution was selected for each standard individually in such a way that its retention time was constant and reproducible and did not change with higher dilution. The identity confirmation of the separated compounds was performed by comparing the experimental retention times, retention indices and mass spectra of the compounds with those of concurrently analyzed reference standards.

Sample preparation and bug samples
For the study of sampling procedure Pyrrhocoris apterus was selected, because it is easily available in large numbers [19], it was frequently used as a model in experimental studies [55] and its secretion has already been studied [20]. Moreover, P. apterus is characterized by a moderate chemical defense with a delayed effect on predator and only releases its defense secretion when it is strongly irritated [15,19]. This species was used for optimization experiments focused on finding the most efficient approach for extracting and analyzing defensive secretion in Pyrrhocoridae. To optimize sampling conditions for all possible compounds four individuals of P. apterus were used together, always two males and two females.
Under the optimized sampling and separation conditions the defensive secretions of adults of three closely related true bug species (Hemiptera: Heteroptera: Pyrrhocoridae, the firebugs), were studied: (1) Pyrrhocoris apterus (L., 1758) from Prague, Czech Republic, a common Eurasian aposematically colored species introduced to other continents as well, (2) P. tibialis (Stål, 1874), from Tianjin, China, an East Palaearctic species, and (3) Scantius aegyptius (L., 1758) from Greece (Kos Island and Crete), a Mediterranean species. The adults were sexed, fed with seeds of their original host plants (Tilia cordata in P. apterus, Alcea rosea in S. aegyptius, and Hibiscus rosa-sinensis in P. tibialis), and maintained separately in plastic containers at 25 ± 2˚C and photoperiod L16:D8. Three males and three females of each species were used for the individual analysis, the sexes were sampled separately and the analysis was performed in triplicate (n = 3).
Three individuals were placed in 4 mL glass vial closed with a polypropylene stopper (Supelco, Bellefonte, PA, USA) with a teflon septum (Supelco, Bellefonte, PA, USA), and the vial was then placed in an ultrasonic bath (Elma, Singen, Germany) or shaker (Heidolph, Schwabach, Germany). In the experiment employing irritation with a plunger, three individuals were placed in a syringe with a barrel volume of 12.5 mL (Eppendorf, Hamburg, Germany), which was placed in an incubated shaker for tempering the specimens (bioSan, Riga, Lithuania). Following a certain tempering time at the given temperature, the bugs were carefully compressed with the plunger of a syringe so that they could not move and the syringe tip was then closed with a rubber stopper. The bugs were compressed with the plunger until a thin liquid film appeared on their body. Then the rubber stopper was removed from the syringe tip, 5 mL of air was drawn in and an SPME (solid phase microextraction) fiber was immediately inserted into the tip of the barrel.
The following experimental conditions were studied for each sampling method: i) irritation with ultrasonics: type of fiber, irritation temperature, intensity of the ultrasonics, irritation time, SPME sampling time and temperature of SPME sorption; ii) for irritation in a shaker: type of fiber, irritation temperature, rate of shaking, irritation time, SPME sampling time and temperature of SPME sorption and iii) for irritation using compression in the barrel of a syringe: type of fiber, temperature prior to compression, tempering time prior to compression, time and temperature of SPME sorption.
The volatile compounds were extracted from the syringe barrel using a manual SPME sampler with a selected fiber assembly (all Supelco, Bellefonte, PA, USA) coated with: i) 7 μm polydimethylsiloxane (PDMS), ii) 85 μm polyacrylate (PA) and iii) triple phase 50/ 30 μm divinylbenzene/carboxen/polydimethylsiloxane (DVB/CAR/PDMS). Each SPME fiber was thoroughly conditioned for 15 min before and after each analysis in an external syringe oven at 250˚C under vacuum. Before each analysis of defensive secretion, a control analysis of SPME sampler itself and SPME sampling of the syringe itself (blank) was performed. The insects were tempered during the sorption on SPME fiber to increase vapor pressure.

Comparison of secretions between true-bug species
In order to consider the intra-and inter-specific chemical diversity of defensive chemicals, we constructed a cluster tree based on qualitative similarities in chemical profiles among all analysed sampled. The clustering was based on distance matrix, in which the distance between each pair of samples was expressed as a number of compounds exclusively present in one of the samples and absent in the other. Single linkage clustering algorithm was used for tree construction. All 135 detected compounds were considered, including the fully identified compounds as well as partially identified chemicals, characterized by their retention behaviour and mass spectra with similarity (match factor in NIST 2008 data base) lower than 85%. Three independent biological replicates for each species and sex were used, each consisting of a pooled sample of three individuals, except for P. apterus females, in which only two replicates were available.

Selection of parameters and response definition
The studied parameters were selected with respect to several factors. One of the main requirements on sampling of volatile secretions was that the bugs would not be killed and destroyed, so the method would be suitable for a limited material of rare species. We found that the critical sampling temperature for P. apterus is 40˚C, because higher temperature is lethal. Therefore, this temperature has been arbitrarily set as the highest applied temperature also for other two tested species. For SPME, the choice of the polarity of the sampling fiber is the most critical condition for reliable and reproducible analysis of a sample. As the detailed composition of the secretions of the studied species of Pyrrhocoridae is unknown and can depend on many factors (described in the introduction), three different SPME fibers with different polarities were studied-polar PA, nonpolar PDMS fibers and composite bipolar DVB/CAR/PDMS fiber covering a wide range of volatilities and polarities. The type of fiber, irritation temperature and time, and the temperature of SPME sorption constituted the basic set of parameters for each of the proposed irritation methods. Ultrasonics was supplemented by the parameter of the ultrasonics intensity and the shaker by the shaking rate parameter. Selected parameters together with their tested levels (low, middle and high) are summarized in Table 1. The individual parameters for sampling using a syringe are listed on Table 2. For this sampling method the parameter of the type of fiber is omitted, because optimization was performed for each fiber separately. Sets of measurements, combining the factors and their levels are given in the Supporting Information (S1-S3 Tables).
In addition to developing a non-lethal method of sampling volatile secretions, we also attempted to obtain the largest possible amount of secretions. The following responses were chosen: the number of peaks with an area of at least 20,000 arbitrary units (a.u.) and the sum of their areas.

Optimization procedure
The secretion sampling method was optimized and Tables 3 and 4 provide the obtained optimum parameters.
It can be seen from Table 3 that, on the basis of the required responses (maximum number of peaks, maximum sum of the areas of all the peaks), the DVB/CAR/PDMS fiber seemed to be the best of the fibers studied. Irritation by ultrasonics was more effective than irritation in a shaker, because it generated a greater number of peaks and maximum sum of the areas of all the peaks. When the secretion was sampled using compression of the bugs by the plunger of a syringe (see Table 4), polar (PA) and nonpolar (PDMS) fibers exhibited much lower responses and again unambiguously the best results were obtained with DVB/CAR/PDMS fibers. The advantageousness of irritation by ultrasonics and compression by the plunger can be seen from comparison of the required responses obtained in all the tested systems. Irritation by ultrasonics generates a smaller number of peaks than compression with a plunger, but the overall response is somewhat higher than when using a plunger. The sampling method employing compression with the plunger of a syringe combined with DVB/CAR/PDMS was chosen as the most advantageous, generating the greatest number of peaks and probably simulating best the real danger to the bug when being seized by a large insectivorous predator, and when releasing its full set of antipredatory chemicals. For the responses "maximum number of peaks" and "maximum sum of their areas" the coefficients of determination (R 2 ) were 0.9710 and 0.8720, respectively. These results indicate very good agreement (97.1 and 87.2% of explained variability) between the experimental data and the built up model. Examples of two response surfaces of plots for this sampling method are given in Fig 1, depicting the interrelationships of selected parameters with the response. It can be seen from the two graphs that increasing time of SPME sorption are accompanied by an increase in the areas of all the peaks and also in their number. The increasing temperature of SPME sorption generates decreasing sum of peaks areas but increasing number of peaks which is more important for qualitative analysis.

Analysis of defensive secretions in adults of Pyrrhocoris apterus, Pyrrhocoris tibialis and Scantius aegyptius by GC-MS.
Under the optimized conditions defensive secretions of all the three studied species were analyzed. Sampling was performed under the conditions listed in Table 4 (bold column) by sorption on a DVB/CAR/PDMS fiber after compressing the bugs with the plunger of a syringe. The actual GC analysis was performed under the conditions described in Section Sample preparation and bug samples. The bugs were sexed and the identical experiments were carried out separately for males and females with three individuals in each analysis. Identifications of the individual peaks obtained for males and females of all three studied species, the corresponding retention times, their standard deviations and relative peak areas (n = 3) are summarized in Table 5, and only peaks with response (A) greater than 20,000 a.u. and similarity (match factor in NIST 08 database) higher than 85% were selected for the identification and following normalization.
We can scrutinize the similarities and differences among species from the viewpoints of two major hypotheses: (a) Pyrrhocoris apterus and P. tibialis should be more similar to each other (at least in critical traits) than any of them to Scantius aegyptius since they are more closely related as indicated by their generic classification (Pyrrhocoris x Scantius), (b) P. apterus and S. aegyptius should be more similar since they are Müllerian (or quasi-Batesian; the term referring to mimetic relations between unequally defended species [56][57][58]) mimics sharing the same conspicuous aposematic coloration and occurring together at the same localities, habitats and host plants.
Most of the compounds found in the secretions are highly volatile substances with low molecular weights and we may assume that most of them have either defensive function or act Table 5 Comparative Analysis of Three Species of Pyrrhocoridae by SPME-GC-MS as intraspecific chemical signals. Saturated short-chain aldehydes are very effective chemical irritants, hydrocarbons can facilitate penetration of reactive aldehydes through the cuticle of arthropod predators and some esters can also assist in wetting the cuticle of arthropod predators [20]. The secretions of all the studied species also contained a large number of minority branched hydrocarbons (especially methylated) which were not included in Table 5 because the determination of methyl group position is difficult without reference standards. Defensive secretions of S. aegyptius and P. tibialis match in 24 compounds, whereas defensive secretions of P. apterus and S. aegyptius match in 9, and P. apterus and P. tibialis in 14 compounds (all unisexual occurrences counted as presence in a species). A cluster tree based   It seems that interspecific differences and similarities reflect neither phylogenetic relationships between the species nor mimicry in colour pattern between P. apterus and S. aegyptius. The absence of several compounds in P. apterus may reflect its greagarious life style not requiring long-distance intraspecific communication. Some of the similarities and differences may also result from different natural host plants whose seeds were used for feeding the cultured populations.

Male S Female S Male S Female S Male S Female
In the present samples we find several patterns of shared similarity of chemicals. Those present in all the three species are probably conservative plesimorphic characters of the clade containing genera Pyrrhocoris (with P. apterus and P. tibialis) and Scantius (with S. aegyptius). Comparative Analysis of Three Species of Pyrrhocoridae by SPME-GC-MS Their number is high in hydrocarbons (6 compounds out of 12), but very low in alcohols (1 out of 17), aldehydes (2 out of 9) and nil in esters (0 out of 18), ketons (0 out of 4) and "others" (0 out of 6). This suggests that occurrence of the individual compound of alcohols, aldehydes and esters is mostly species-specific, while the few common ones of these groups (1-dodecanol, nonanal and decanal) are probably essential for functioning of the whole system.
The shared presence of a compound in two species is complementary to absence of this particular compound (or its replacement or transformation) in the third species. This happened in 3 hydrocarbons, 3 alcohols and 6 esters shared by P. tibialis and S. aegyptius, and 4 aldehydes shared by P. apterus and P. tibialis. Strangely, no compounds are shared exclusively by P. apterus and S. aegyptius. Of the complementary absences, there are 3 hydrocarbons, 3 alcohols and 6 esters missing in P. apterus, 4 aldehydes in S. aegyptius, but no unique absence was found in P. tibialis.
The secretion of the two Pyrrhocoris species contained greater number of aldehydes than the secretion of Scantius aegyptius ( Table 5, 6-8 versus 3). Since the aldehydes are known to have antipredatory defensive function [6,59,60], this difference corresponds with the difference in palatability of P. apterus and S. aegyptius for avian predators (A. Exnerova et al., unpublished), where S. aegyptius appears to be considerably less well defended of the two species and may be a quasi-Batesian mimic of P. apterus. Moreover, most of the aldehydes are shared by both Pyrrhocoris species, with (E)-2-hexenal being dominant. Thus the composition of aldehydes appears rather to follow the phylogenetic relationships between the species than be a result of mimetic convergence.
The composition of hydrocarbons was mostly similar in all the three species, with only some of them missing in P. apterus, which is likely to represent a derived situation, convergence between two non-congeneric species (P. tibialis and S. aegyptius) is less likely. In all the three species, one of the dominant hydrocarbons is tridecane, known for its antipredatory function against arthropod predators, where it facilitates penetration of aldehydes through the cuticle [61].
Interspecific differences in the composition of alcohols and esters are difficult to interpret, because of their less known and possibly multiple and context-dependent functions [6,60,62]. High proportion of both alcohols and esters is species-specific, which indicates their interspecific signaling function [60,[62][63][64]. The strikingly smaller number of different alcohols and esters in P. apterus may be connected with its gregarious lifestyle with smaller need for longdistance chemical communication.
Limonene, which is like other terpenoids obtained by the bugs from their host plants, was present only in S. aegyptius and P. tibialis. Its absence in P. apterus may indicate its unavailability in linden seeds in contrast to those of Alcea and Hibiscus. Along with other terpenoids, limonene plays role in a chemical defense of the bugs, as it is serves as an ant alarm pheromone [65].
Chromatograms obtained for the secretions produced by the males and females of all the studied true bugs are given in Fig 3. The chromatograms obtained for male and female of the individual species differ slightly. Whereas three dominant peaks corresponding to tridecane (9), (E)-2-hexenal (32) and methyl 2-hydroxy-3-methyl pentanoate (46) are present in the male, as well as in the female chromatograms of P. apterus, 2-propyl-1-pentanol (18) is only found in females. The chromatogram of P. tibialis female is enriched by five additional compounds, especially by (E)-2-hexen-1-ol (15) and (E)-2-octen-1-ol (19), whereas male and female chromatograms of S. aegyptus differ especially in the relative abundance of dodecane (7), 2-cyclohexen-1-ol (17) and cyclopentanone (39). Generally, the secretions of females of all the three species contained larger number of alcohols than the secretion of conspecific males (Table 5). This may suggest the epigamic pheromonal function of the particular alcohols [62]. On the other hand, there was minimum of sexual dimorphism in composition of aldehydes, supporting their interspecific function in an antipredatory defense [6,59,60], and in composition of esters, which were reported as non-epigamic intraspecific attractants and alarm pheromones [60,63,64].
In no true bug species the exact function of numerous infochemicals of the secretion is known-in antipredatory context it can range from signaling to irritation up to being lethally toxic [66]. In addition to metathoracic glands representing the major source of defense infochemicals in adults, also other specialized glands and sequestered noxious plant compounds may play role in defense [7,59]. It should be born in mind that also non-defensive infochemicals may be detected in the adults of true bugs (e.g. pheromones produced in dorsoabdominal glands persisting in adults from larval stage though with an altered function and no more secreting defense substances). We should emphasize that a similar extraction "in vivo" as  Table 4; numbering of the peaks corresponds to Table 5. having been used presently is also applicable to other arthropods (e.g., spiders, millipedes, a variety of insects in addition to true bugs), and may be used for all kinds of a rich array of exocrinous, supra-integumentally released, and often extremely complex defensive secretions.

Conclusions
A set of non-lethal methods has been developed for SPME sampling of the volatile secretions of three species of true bugs (Insecta: Hemiptera: Heteroptera: Pyrrhocoridae) as a part of their defensive mechanisms. Particular attention was paid to optimization of the input parameters, including the means of irritation, and the optimization was performed using multi-parameter response surface methodology. On the basis of the required information (maximum number of peaks, maximum sum of all the peaks), the best method seemed to be irritation of the bugs by compression with the plunger of a syringe and use of a composite DVB/CAR/PDMS fiber for SPME sampling. This method also simulates best the natural attack by a vertebrate predator. GC separation was performed using a column Rtx-200 with fluorinated stationary phase providing good selectivity even for substances with very different polarity scales.
The developed method was used for sampling of volatile defensive secretions of adult males and females of Pyrrhocoris apterus, P. tibialis and Scantius aegyptius (all Pyrrhocoridae). The method applied has the advantage that only several individuals of each sex are sufficient for the analysis and thus even rare and not easily available species can be analyzed. Moreover, the method is non-lethal for the insects, which may be later potentially used in further studies requiring live individuals. The outlined methods can be widely used in similar studies on other insects although the optimization procedure and fiber evaluation would be necessary to perform for any new species studied owing to large variation of chemicals potentially involved.
The chemical composition of secretion, particularly that of alcohols, aldehydes and esters, is species-specific in all the three pyrrhocorid species studied. The overall similarities in composition of secretion do not reflect the relationship of species or mimicry between Pyrrhocoris apterus and Scantius aegyptius. The similarity of secretions between the bug species is associated with their antipredatory strategies. For instance the small number of alcohols and esters in Pyrrhocoris apterus is probably associated with its high gregariousness while low numbers of aldehydes in Scantius aegyptius reflects its higher palatability. The sexual dimorphism in occurrence of particular compounds is largely limited to alcohols and suggests their epigamic intraspecific function.
Supporting Information S1 Table. Ultrasonics. Experimental parameters, their levels and modeling experimental plan of the face centered central composite design for sampling secretion using ultrasonics.