Behavioural consequences of intraspecific variability in a mate recognition signal

Abstract

Mate recognition is paramount for sexually reproducing animals, and many insects rely on cuticular hydrocarbons (CHCs) for close-range sexual communication. To ensure reliable mate recognition, intraspecific sex pheromone variability should be low. However, CHCs can be influenced by several factors, with the resulting variability potentially impacting sexual communication. While intraspecific CHC variability is a common phenomenon, the consequences thereof for mate recognition remain largely unknown. We investigated the effect of CHC variability on male responses in a parasitoid wasp showing a clear-cut within-population CHC polymorphism (three distinct female chemotypes, one thereof similar to male profiles). Males clearly discriminated between female and male CHCs, but not between female chemotypes in no-choice assays. When given a choice, a preference hierarchy emerged. Interestingly, the most attractive chemotype was the one most similar to male profiles. Mixtures of female CHCs were as attractive as chemotype-pure ones, while a female–male mixture negatively impacted male responses, indicating assessment of the entire, complex CHC profile composition. Our study reveals that the evaluation of CHC profiles can be strict towards ‘undesirable' features, but simultaneously tolerant enough to cover a range of variants. This reconciles reliable mate recognition with naturally occurring variability.

1. Introduction

Finding a suitable mate is a crucial fitness prerequisite for sexually reproducing species. Attraction and courtship/mating behaviours towards a potential mate are based on the evaluation of signals and cues emitted by that individual. Thus, mate recognition is thought to be relatively strict towards deviations from the species-specific signals. Sexual communication signals can be visual, acoustical, tactile or chemical, and several sensory modalities may function in concert. However, many animals, and especially insects, rely heavily on chemical signals for sexual communication [1,2]. Generally, these sex pheromones are mixtures of several, specific compounds, with the respective composition (which compounds and their relative amounts in the mixture) considered to be vital for signal function [1]. They serve to attract conspecifics from a distance, govern the sequences of interaction between the sexes at close range, or elicit courtship and the initiation of mating [3–6]. Thus, sex pheromones can be instrumental in obtaining a high-quality mating partner [7]. Precopulatory isolation between closely related species can be effected by sex pheromones differing in the relative amounts of shared compounds, or the presence of one or more species-specific compounds [8,9]. Hence, the basic assumption is that the signals are subject to stabilizing selection [10,11]. However, intraspecific variation of the pheromone often occurs, such as between populations [12,13]. Diet can be a cause of pheromone variability [14,15], though more and more studies also find within-population variability correlated with factors such as age/maturity, body size or nutritional state [16–18]. Therefore, intraspecific sex pheromone variation, especially on a gradual scale, is likely much more common than previously assumed [19]. In some cases, even distinct pheromone phenotypes have been found within the same populations [11,20,21].

The existence of within-population pheromone variability leads to the question of which consequences this has for sexual communication. Do the receivers discriminate between the variants, and if they do, how and why? Signal variability may correlate with differences in the quality of potential mates, and females as well as males may benefit from choosing or preferring mates accordingly [7,22–24]. Investigating signal variability as well as its causes and effects is therefore essential to understand the intricacies of sexual communication. Further studies on pheromone production and perception may then, eventually, shed light on the long-standing questions of how chemical communication systems diverge and evolve, and the role they play in speciation [9].

When assessing a potential mating partner, the crucial steps take place in the immediate vicinity of that individual, and close-range pheromones may govern the final acceptance or rejection of mating. Among the compounds used at close range by insects, cuticular hydrocarbons (CHCs) are probably the most common. CHCs are constituents of the insects' cuticular lipid layer, and are present in a species-specific composition that can include n-alkanes, methylbranched alkanes and/or unsaturated hydrocarbons [6]. The primary function is to serve as barrier to desiccation [25,26], but CHCs are additionally involved in a plethora of close-range communication contexts [6,27]. Sexual communication at close range via CHCs has been reported for many species from several insect orders (e.g. Coleoptera [28,29], Hymenoptera [30,31], Diptera [4,32], Orthoptera [33], Hemiptera [34] and Thysanoptera [35]). The widespread use of CHCs for such a vital communication purpose is intriguing, since it is well established that CHC profiles can be influenced by several biotic and abiotic factors [4,15,36–41], which, in turn, could impact sexual communication. The resulting variability might be challenging for the receivers to recognize mates (communication requires reliable signals), but also has the potential to serve as basis for mate assessment. Consequently, CHCs may be under sexual selection [24]. However, few studies apart from ones on Drosophila (e.g. [37,42]) have hitherto investigated the effects of CHC variability for mate recognition [43,44]. To understand the consequences of this possibly common complication of CHC-based communication, more studies will be needed. Unfortunately, investigating the consequences of CHC variability can be very difficult, as changes are often complex, involving several compounds from different compound classes [4,43,45]. This makes it hard to evaluate which changes may be relevant and affect receiver behaviour. Thus, systems in which variability is easier to categorize (e.g. involving only one or a few compounds, or a suite of compounds for which relative abundances change in the same way) are the most promising to gain further insights.

Here, we studied a species that provides exactly such a prerequisite. We investigated male reactions to variations of the CHC profile in the parasitoid wasp Tachinaephagus zealandicus. In this species, both sexes produce the same set of CHCs. Female-derived (but not male-derived) CHCs elicit copulation attempts by males despite the fact that the tested CHC profiles of males and females are very similar in the relative amounts of CHCs. It was hypothesized that one way to distinguish between the sexes would be to detect and to react to small, but consistent differences in the relative amounts of the shared compounds [46]. Recently, females of this species were reported to show three distinct CHC chemotypes within the same population. The female profile (type M) that is very similar to male CHC profiles is characterized by large relative amounts of alkadienes, which are dominant in the male profiles. The other two female CHC profile types, while also consisting of the same CHCs, are characterized either by large relative amounts of alkenes (type L), or of methylbranched alkanes (type H; electronic supplementary material, figure S1) [47]. The same study indicated that the frequencies of the three chemotypes differed in the studied population, with type L being the most common (40–62% of analysed females), followed by type H (37–43%) and type M (1–18% [47]). The female chemotypes are probably genetically determined (intermediates seem to be rare; T.P. 2019, 2020, personal observation), and the mode of inheritance is under investigation. However, while CHCs are essential mate recognition signals in this species, the effects of the three female chemotypes for sexual communication are hitherto unknown. Males clearly distinguished between the similar CHC profiles of type M males and females, demonstrating their ability to detect seemingly minor differences in the relative compositions of CHC profiles [46]. Therefore, males should also be able to detect the more pronounced differences between the female CHC chemotypes. Here, we studied this remarkable system to gain insights into how CHC variability may impact mate recognition and preference. We investigated whether males differentiate between the three pheromone chemotypes, and in which way differences in the relative amounts of CHC compound classes influence the evaluation of the CHC profiles during mate assessment. We tested male responses towards natural CHC chemotypes using bioassays without and with an alternative choice (experiments providing more than one option can be more sensitive when investigating preferences [48]). As variation in mate recognition signals may correlate with the signaller's quality [16,49], female chemotypes in T. zealandicus might differ in fertility. Thus, we assessed the fitness returns for males paired with females from the three chemotypes. To investigate how males might evaluate the close-range sex pheromone, we manipulated variability by creating artificial ‘intermediates' of the naturally occurring CHC profiles. We tested whether the three female chemotypes are matched to distinct ‘recognition templates' by using mixtures of female-specific CHC profiles. And, as males reacted only to female CHCs even though one female CHC profile is very similar to males' CHC profiles [46], we combined male and female CHCs to elucidate whether male CHC profiles are simply ignored, or whether they may have unattractive properties.

2. Material and methods

(a) . Wasps and extracts

All wasps were from chemotype-pure, inbred matrilines that had been established from the same population (several lines per chemotype, each originating from trap catches in Gießen, Germany). Only females show pronounced CHC profile differences (electronic supplementary material, figure S1) [47]. Males originating from all three chemotypes (henceforth designated by the respective female chemotype label) were tested in the bioassays to take potential differences in male behaviour into account. Parasitoid wasps lay their eggs in or on an arthropod host, on which the developing offspring feeds. Female wasps were provided with third-instar larvae of Lucilia caesar or Calliphora vomitoria as hosts for oviposition, and kept at 23° ± 2°C in a thermal cabinet under a 16 : 8 h light : dark cycle. Wasps for experiments were dissected from the host pupae (females lay several eggs in each host [50]) approximately 2 days before eclosion from the pupal exuvia, and stored singly in 1.5 ml reaction vials in the thermal cabinet until wasp eclosion. While CHC profiles may differ to some extent between different insect body parts [51], previous work on T. zealandicus had shown that the mate recognition signal is present and bioactive in CHC fractions obtained from whole-body extraction [46]. Thus, female-derived CHC fractions were obtained by extracting batches of 100–150 females (originating from several different matrilines of each chemotype, freeze-killed at −20°C at an age of 24–48 h after eclosion, i.e. the age at which the wasps would emerge from the host and mate [47]) for 1 min in hexane (15 µl per individual). After transferral of the supernatant to clean vials, the wasps were rinsed with another 5 µl of hexane per individual, which was added to the extract. Solvent volume was reduced to about 200 µl under a gentle stream of nitrogen. The extracts were fractionated using silica gel columns as described by Jungwirth et al. [46], retaining only the non-polar fraction containing the CHCs. For bioassays testing male-derived CHCs, batches of 100 males were extracted and the extracts fractionated accordingly. The CHC fractions were carefully evaporated under nitrogen and redissolved in hexane (final concentration one animal equivalent (AE) per microlitre). They were either left unchanged, or combined to create mixtures (described below) before being aliquoted (20–25 AE per aliquot) and tested in bioassays. ‘Dummies' lacking chemical cues for use in bioassays were obtained by subjecting extracted females to 9 h of further extraction with a continuous supply of clean solvent using a Soxhlet set-up [46]. Naive male wasps were tested at an age of 24–72 h. Experiments were conducted in randomized block set-ups at climatized room temperature between 9.15 and 23.15. Each male was tested once only, with a maximum of two replicates per matriline origin for each bioassay. Sample size was always n = 30 per male origin and bioassay, and all bioassays scoring male behaviour live were performed by the same experimenter.

(b) . Male reactions to natural CHC chemotypes

Males originating from all three chemotypes (M, H, L) were tested in single stimulus set-ups (no-choice) towards dummies treated with two AEs of female-derived CHCs from each of the three chemotypes, or male-derived CHCs (chemotype M). Bioassays were conducted as described by Jungwirth et al. [46], using the same arenas with 20 mm diameter and 3 mm height set up on glass plates and covered with a glass slide. After treatment of a dummy with the respective CHC fraction, the solvent was allowed to evaporate for at least 10 min. Thereafter, it was placed in the arena near the side before introducing the test male into the arena at the opposite side using an aspirator. The glass slide was placed to cover the arena, and observation (300 s) began. Time of detection (antennae of the test male less than 1 mm from the stimulus), as well as start and end times of copulation attempts were documented using The Observer XT v. 11.5 (Noldus Information Technology, Wageningen, The Netherlands) in live observation mode. After each experiment, the test male was removed, the dummy was discarded, and the arena set-up was cleaned with ethanol. Observations of males that did not approach the dummy close enough to be defined as having detected it were discarded, and the same test was repeated with a fresh male and stimulus. Choice experiments were conducted similarly, though in a slightly smaller arena (15 mm diameter) to enable test males to encounter the two stimuli in quick succession. Two dummies treated with female chemotype-pure CHCs (2 AEs each) were presented together (M versus H, M versus L, L versus H) to males from all three chemotypes. Each experiment was videotaped to ensure being able to distinguish the dummies if male copulation attempts should move them away from the original position. The dummies were roughly size-matched before being treated with the respective stimuli. After evaporation of the solvent (at least 10 min), they were placed near the side and ca 2 mm apart on dots marked from the bottom on the glass base of the arena. The video was started, and the male was introduced at the opposite side before covering the arena with the glass slide. Video recording lasted for a minimum of 310 s after the male had been introduced. The position of the dummies was alternated between replicates to avoid side bias, and arena set-ups were cleaned with ethanol after each trial. Replicates in which the male detected only one or none of the dummies were discarded, and repeated with a new male and freshly prepared stimuli. Videos were analysed with The Observer XT v. 11.5 in offline observation mode. The start of the 300 s observation duration was set to the timepoint when the male started to move after being introduced in the arena. Time of detection, as well as start and end times of copulation attempts (for an example of male behaviour, i.e. copulation attempts versus no reaction, see electronic supplementary material, video S1) were documented separately for each dummy based on the original position of the dummies (left–right). Stimulus identity was matched with the respective position in later steps of the analyses. To normalize across individual male motivation, data visualization for copulation attempts shows relative male preference (i.e. the percentage of time spent attempting to copulate with one of the two dummies—M for choice tests of M versus H and M versus L; L for L versus H) in relation to the total duration of all copulation attempts.

(c) . Assessment of female chemotype-dependent fitness returns for males

Fifteen males per chemotype origin were mated to single females of all three chemotypes (non-sibling mates for combinations of same-chemotype males and females). Each individual mated once only. Females were then set singly into in small plastic dishes (10 mm height, 60 mm diameter) with 10 hosts (five third-instar larvae of each Lucilia caesar and Calliphora vomitoria) and a small drop of honey-water in the lid. After 24 ± 2 h, females were moved to a new dish with 10 fresh hosts and honey-water once per day until time of death, at which point the female was removed. Dishes were kept in a climate cabinet (25°C, 16 : 8 light : dark cycle). All hosts and emerging offspring were stored for 30 days (development time at 25°C is around 23 days [52]). Females that produced no offspring were excluded from further analyses. As, due to haplodiploidy, males gain direct fitness only through female offspring, the total number of female offspring was counted for each of the remaining females.

(d) . Male reactions to artificially created CHC profile mixtures—female-derived CHCs only

To create ‘intermediate' chemotypes several CHC fractions of each chemotype were pooled. One aliquot per chemotype remained unchanged to serve as control for tested males of the respective chemotype. The other aliquots were combined in binary mixtures (1 : 1 ratio; MH, ML, HL), and the ternary mixture (ratio 1 : 1 : 1; MHL) of the three chemotypes. In all experiments, each dummy was treated with two AEs. Assay set-ups were as described above. Males of all three chemotypes were tested using either natural female CHC profiles, a binary, or the ternary mixture in a no-choice assay, and in a choice assay towards chemotype-pure CHCs and the ternary mixture (M versus MHL, L versus MHL, H versus MHL). Analogous to the visualization of choice-assay results towards natural female chemotypes, relative male preference was calculated by dividing the time spent trying to copulate with the dummy treated with chemotype-pure CHCs (M, L or H) by the total duration of all copulation attempts.

(e) . Male reactions to artificially created CHC profile mixtures—male and female CHCs

Reactions of males from all three chemotypes towards either female CHCs or a mixture of male and female CHCs were tested in a no-choice set-up, using either one AE of pure female M-type CHCs, or two AEs of a binary mixture (1 : 1) of this and male M-type CHCs (i.e. the amount of female-derived CHCs was the same in both cases).

(f) . Analyses

Data obtained through The Observer XT v. 11.5 yielded the number of males responding with copulation attempts, as well as the total duration of copulation attempts for each tested male. Statistical analyses were conducted in R v. 4.2.1 [53]. All no-choice assays were analysed separately for males of each chemotype origin. Numbers of males responding to each of the tested stimuli were compared using χ²-tests with Bonferroni–Holm correction for pairwise comparisons. Total durations of copulation attempts shown in no-choice assays were analysed only for responding males. Shapiro–Wilk tests indicated non-normal distribution for reactions to one or more of the tested stimuli, thus analysis was conducted using Kruskal–Wallis (more than two tested stimuli) or Mann–Whitney U-tests (two tested stimuli). For choice assays, latency from detection to the first copulation attempt with the respective dummy was calculated by subtracting the corresponding time points. If one of the two dummies did not elicit copulation attempts, the ‘latency' was adjusted to span the time from detection of that dummy to the end of the experiment. This resulted in a latency value for each of the dummies. Replicates were excluded from further analyses if neither of the two detected dummies elicited copulation attempts. To evaluate whether the reactions of males originating from the three chemotypes differed, we calculated the difference between the respective values of latency/total duration of copulation attempts towards the two tested dummies. The resulting values were compared separately for latency and total duration of copulation attempts using Kruskal–Wallis tests (Shapiro–Wilk test results indicated non-normality). As male reactions did not differ between chemotype origins, the respective data were pooled, and the responses to the two tested stimuli were compared using Wilcoxon matched-pairs signed rank tests. Additionally, potential differences in the order of detection were assessed using χ²-tests. Female offspring numbers were compared between chemotypes using Kruskal–Wallis tests.

3. Results

(a) . Males react similarly to all three female chemotypes in no-choice assays

Female-derived CHCs of all three chemotypes elicited copulation attempts from the tested males of all chemotype origins while, in all but a single case, male M-type CHCs did not (χ²-tests: males M overall: χ² = 54.55, d.f. = 3, p < 0.001; males H overall: χ² = 43.77, d.f. = 3, p < 0.001; males L overall: χ² = 60.83, d.f. = 3, p < 0.001; for all males: pairwise comparisons female CHCs of each chemotype versus male CHCs, p < 0.001; figure 1a; electronic supplementary material, table S1). Overall, 85–90% of the tested males reacting with copulation attempts towards dummies with female-derived CHCs performed at least two copulation attempts. Males originating from chemotypes H and L did not react differently to female-derived CHCs irrespective of the chemotype profile tested (figure 1; electronic supplementary material, table S1). Males originating from chemotype M reacted less often to female M-type CHCs than to female L-type CHCs (Bonferroni–Holm corrected χ²-test: χ² = 6.65, p = 0.030; figure 1a). However, the males that did attempt to copulate with the dummy reacted equally long towards dummies with female M-type CHCs and those with CHCs from females of the other two chemotypes (Kruskal–Wallis test, χ² = 1.52, p = 0. 47; figure 1b; electronic supplementary material, table S1).

Figure 1.

Male reactions to natural cuticular hydrocarbon (CHC) profiles in no-choice tests. CHCs of females from chemotype M (first bar/box, dark grey), chemotype H (second bar/box, grey), chemotype L (third bar/box, light grey), or male-derived CHCs (fourth bar, black) were tested for bioactivity. (a) Number of males from each of the three chemotypes (M, H, L, n = 30 per tested stimulus) reacting with copulation attempts to dummies treated with two animal equivalents of CHCs. Different letters above the bars show significant differences. χ²-tests, pairwise comparisons Bonferroni–Holm corrected. (b) Total duration of copulation attempts shown by responding males (respective numbers listed below each box) towards only the female-derived CHC fractions (only a single copulation attempt was elicited by male-derived CHCs, not applicable for testing, n.a.). Boxes show the median (thick line), and range from the first to third quartile. Whiskers span the 25% lowest and highest values, respectively, excluding outliers (greater than 1.5 times box height, open circles). Comparisons within each male chemotype origin by Kruskal–Wallis tests.

(b) . Males show a preference hierarchy towards female chemotypes in choice assays

In choice tests, there was no difference between reactions of males originating from the three chemotypes (Kruskal–Wallis tests, all n.s.; electronic supplementary material, table S2), thus data were pooled for further analysis. Overall, males showed a preference hierarchy towards female CHCs, with M being more attractive than H (shorter latency between detection and the initiation of copulation attempts, as well as more time spent trying to copulate with the respective dummy; Wilcoxon matched-pairs signed rank tests: latency V = 1083, p = 0.049, duration V = 2199, p < 0.001) and a clear trend to be more attractive than L (latency V = 953, p = 0.091, duration V = 1575, p = 0.052), while L was more attractive than H (latency V = 1335, p = 0.049, duration V = 726, p = 0.036; figure 2a; electronic supplementary material, table S2). Males did not detect specific chemotypes faster than others (χ² tests, all n.s.; electronic supplementary material, table S2), and attempted to copulate first with the dummy that was encountered first more often than expected by chance (χ²-test, χ² = 8.4, p = 0.0038; electronic supplementary material, table S2). Most males (80%) attempted to copulate twice or more with at least one of the dummies.

Figure 2.

Relative preference (percentage of total duration of copulation attempts) by male T. zealandicus (n = 90 each) for one of two dummies treated with female-derived CHCs (two animal equivalents per dummy) in choice assays. Males reacting to neither of the two stimuli were excluded from the analyses, numbers of responding males listed below the boxes. Boxes show median (thick line), and range from first to third quartile; whiskers encompass the first and fourth quarter of the data. The dotted 50% line corresponds to no preference for either dummy. Data points next to the boxes depict the males' relative preference values, with values of 100 signifying that all copulation attempts were shown towards the natural CHC profile listed first in the x-axis label, while 0's indicate that the males in question only attempted to copulate with the respective other dummy. Wilcoxon matched-pairs signed rank tests. (a) Relative preference hierarchy towards the natural CHC chemotypes. Dark grey boxes show the relative preference for M-type CHCs (test combinations M versus H and M versus L), the light grey box depicts the relative preference for L-type CHCs (L versus H). (b) Relative preference for chemotype-pure CHC profiles when given the choice between either chemotype-pure CHCs (M-type: dark grey; L-type: light grey; H-type: grey) or a ternary mixture thereof (MHL, combinations: M versus MHL, L versus MHL or H versus MHL).

(c) . Chemotype-dependent mate preference does not result in direct fitness benefits

Overall, the total number of female offspring was highly variable between females, ranging from 6 to 222 (median 119; electronic supplementary material, table S3). Male chemotype origin did not have any effect on the total number of female offspring (Kruskal–Wallis tests: for M females χ² = 2.60, p = 0.27; H females χ² = 0.88, p = 0.64; L females χ² = 1.19, p = 0.55), and data were pooled for further analysis. Females of the three different chemotypes did not differ in the total number of female offspring (Kruskal–Wallis test, χ² = 1.71, p = 0.42).

(d) . Artificially created ‘intermediate' female chemotype mixtures are not discriminated

‘Intermediate' CHC profiles (binary or ternary mixtures of female-derived CHCs from the three chemotypes) were as attractive as chemotype-pure CHCs for males of all three types in no-choice tests, both in the number of responders as well as the time spent trying to copulate with the test dummy (χ²-tests, Kruskal–Wallis tests, all comparisons n.s.; electronic supplementary material, figure S2, and table S4). Males from all three chemotypes reacted similarly in choice tests (Kruskal–Wallis tests, all n.s.; electronic supplementary material, table S5), thus their reactions were pooled for further analysis. When given the choice between chemotype-pure CHCs and the ternary mixture MHL, the same preference hierarchy as for chemotype-pure CHCs was visible. MHL was at the same level of attractiveness as female-derived CHCs of chemotype L (Wilcoxon matched-pairs signed rank tests: latency V = 1948, p = 0.69, duration V = 2292, p = 0.33), less attractive than M-type CHCs (latency V = 1542, p = 0.042, duration V = 2916, p < 0.001), and more attractive than H-type CHCs (latency V = 2822, p < 0.001, duration V = 739, p < 0.001; figure 2b; electronic supplementary material, table S5). There was no difference in the number of first detections between the two stimuli for each tested combination (χ²-tests, all n.s.; electronic supplementary material, table S5). Males generally attempted to copulate first with the dummy that was encountered first (χ²-test, χ² = 223.14, p < 0.001; electronic supplementary material, table S5). Almost all males (265 of 270) attempted two or more copulations with at least one of the dummies.

(e) . Male responses are reduced if a CHC mixture contains male CHCs

The addition of male M-type CHCs to female M-type CHCs reduced male responses in comparison to pure female M-type CHCs of the same amount. Fewer males originating from type M reacted with copulation attempts (χ²-test: χ² = 6.67, p = 0.0098), and H-type males tended to respond less often (χ²-test: χ² = 3.30, p = 0.069; electronic supplementary material, figure S3A and table S6). While the number of responding L males did not differ between the two stimuli (χ²-test: χ² = 1.09, p = 0.30), they did, however, spend much less time attempting to copulate with dummies treated with mixed CHCs (Mann–Whitney U-test: W = 49, p < 0.001; electronic supplementary material, figure S3B and table S6).

4. Discussion

Close-range recognition in many insects is based on CHCs, even though CHC profiles can be influenced by several factors [4,15,36–40] that may entail signal variability. However, there is as yet a paucity of knowledge on how intraspecific variability in the CHC signal composition can influence the behavioural outcome.

This study investigated the consequences of pheromone variation in a species showing a clear-cut within-population CHC variation, a female polymorphism with three distinct CHC chemotypes. Surprisingly, males did not discriminate between the three female chemotypes in no-choice assays, despite the remarkable differences in the relative amounts of the CHC compound classes between them (see electronic supplementary material, figure S1) [47]. That female-derived M-type CHCs elicited responses from fewer M-type males than those of chemotype L females in this set-up is unlikely to be a representative result. First, the total duration of copulation attempts of M-type males that did respond was as long as towards the other two chemotypes. Second, M-type males reacted in similar numbers to female M-type CHCs as males originating from the other two chemotypes in all other bioassays. In only one of seven assays testing male M-type reactions to female M-type CHCs reported in Jungwirth et al. [46] was the number of responding males similarly low as observed in the first no-choice test of the present study. Thus, the result is likely not a general pattern of inbreeding avoidance or indicative of a lower attractiveness of female M-type CHCs for M-type males, but may rather be due to occasional males having a higher response threshold overall, or specifically to female-derived M-type CHCs.

The small differences between M-type female and male profiles had a strong impact on behavioural reactions, with males only showing copulation attempts towards the female-derived CHCs (as already shown for males of chemotype M by Jungwirth et al. [46]). Thus, the lack of discrimination between the clearly differing female chemotypes by males of all types was unexpected. In the leaf beetle Phaedon cochleariae, which also uses CHCs for mate recognition, host plant-induced CHC profile differences led to assortative mate choice for females that had developed on the same hosts [43]. And a study on two species of Heliothis moths revealed that males show preferences for female pheromone blends from their own region [54] (but see [55,56], where males did not distinguish between different female signals).

However, testing only single stimuli can be less sensitive for detecting preferences than bioassays using two or more [48]. Indeed, when given the choice between two female CHC profiles, male T. zealandicus showed a preference hierarchy. This demonstrates that the lack of discrimination in the no-choice tests was not due to an inability to detect the differences between the chemotypes. Males reacted similarly irrespective of their chemotype origin, indicating a generalized preference-pattern that was independent of which of the two stimuli had been encountered first. In all choice tests, neither of the two stimuli was detected first more often than the other. Although males usually directly attempted to mate with the first-encountered dummy, the total duration of copulation attempts shown towards each dummy differed corresponding to the observed preference hierarchy. Female-derived CHCs of chemotype M were more attractive than those of chemotypes L and H, with the latter being the least attractive pheromone phenotype. This is in contrast to the two pheromone races of the European corn borer moth Ostrinia nubilalis. Females differ in the E-/Z- isomeric ratio of their sex pheromone 11-tetradecenyl acetate, and this moth shows positive assortative mating (i.e. a preference for mates of the same pheromone race) [21]. In T. zealandicus, the most attractive chemotype was, interestingly, the rarest in the population of origin, constituting only 1–18% of analysed females [47]. In many cases of intraspecific sex pheromone variability, it is the common phenotype that is more attractive to responders. In studies on moths, females with pheromone component ratios that are rare/extreme or from allopatric populations are often less attractive than those with the ‘typical' or population-characteristic blend [11,54]. However, in a solitary bee, patrolling males preferred synthetic odour blends characteristic of females from allopatric populations, which was suggested to either promote outbreeding, or be caused by avoidance of previously encountered female odours to preclude futile mating attempts [57]. Such explanations seem unlikely for the ‘rare-type' preference observed here, since many species of gregarious parasitoid Hymenoptera regularly perform inbreeding [58], and males were naive at the beginning of the experiments. Additionally, they often returned to an already encountered stimulus and attempted to mate again. Chemical signals used by males for mate assessment may be informative of female quality, as they can indicate fecundity or fertility by gradual variation correlated with fitness-associated factors such as size, age or nutritional status [16,49]. Even in species with non-reversed or partially reversed sex roles (female/mutual mate choice), males would benefit by preferring high-quality females [22,23]. However, T. zealandicus of the preferred chemotype M did not produce more female offspring than L or H females, indicating no differences in potential fertility. Nonetheless, the female chemotypes might differ in realized fecundity, with M-type females potentially being superior in host finding ability or performance/survival. Alternatively, female quality-related variation of the CHC profiles may only be present within each of the three chemotypes. If differences in female quality would be better assessable for alkadiene-dominated profiles (and to a lesser extent for alkenes), such as through a sensory predisposition, the preference hierarchy could allow males to preferentially target (invest more time/effort in) females for which mate assessment is faster or more reliable. This could have resulted from an evolutionary scenario in which the female chemotype M was ancestral, when selection might have favoured males better at distinguishing between male and female-typical relative amounts of alkadienes. The chemosensory aspect of mate recognition in this system should be investigated in future studies.

While the preference hierarchy M > L > H seems to be innate, it was only discernible in choice trials. Encountering several females of different chemotypes in quick succession and in close proximity is likely the case under natural conditions. This parasitoid wasp is gregarious, laying several eggs into each host [50]. Many wasps emerge from each host in quick succession (T.P. 2019, 2020, personal observation). The hosts may additionally be aggregated in space and time (e.g. in close proximity to a decayed carcass), further raising the number of potential mates in the vicinity. Females of all three chemotypes emerged from hosts collected from the same carcass traps, and were shown to be able to produce offspring not only of their own, but also of the respective other chemotypes [47]. Superparasitism (parasitization of the same host by more than one female) may additionally occur. Thus, even females emerging from the same host can show different chemotypes. The persistence (though not the predominance) of the two less attractive female chemotypes in the population is probably due to the facts that all three female chemotypes are attractive when encountered singly, and that females usually only mate once (T.P. 2019, 2020, personal observation). Therefore, even in a ‘competitive' setting, females bearing a less attractive CHC profile would eventually be courted and mate [59]. Irrespective of which chemotype was ancestral in this species, these circumstances would correspondingly have allowed the establishment of the other chemotypes. The different CHC profiles could have emerged through, for instance, mutations affecting transcription factors, leading to a higher expression of genes involved in the biosynthesis of alkadienes/alkenes or methylbranched alkanes. Future behavioural, chemical and molecular examination of other populations and congeneric species might shed light on a possible evolutionary scenario for the three female CHC chemotypes.

All three chemotypes are found in virgin females of mating age [47] (i.e. they are not condition-dependent). CHCs may be under sexual selection [24], thus, the preference hierarchy shown by males of T. zealandicus could exert selection pressure on female CHC profiles. In Drosophila, laboratory studies demonstrated sexual selection on CHC profiles (e.g. [60,61]), and a study on crickets found evidence for sexual selection on CHCs in a wild population [62]. The low abundance of the most attractive CHC chemotype in the studied population of T. zealandicus (1–18%, [47]) is thus enigmatic. Perhaps mate assessment (and therefore also sexual selection) predominantly impacts within-chemotype CHC variability. However, it is also possible that the male preference hierarchy exerts only low to negligible selection pressure. As pointed out earlier, even less attractive females are likely to mate eventually. Furthermore, in the Hymenoptera unmated females do not experience a complete loss of fitness (unfertilized eggs develop into males), further attenuating the consequences of male preference. Future studies will be needed to investigate whether sexual selection affects the female CHC profiles (between or within chemotypes). However, the respective genetic variance is as yet not known, nor is the mode of chemotype inheritance.

It is generally assumed that a clear distinction between males and females is easier to detect or even required for mate recognition/preference. However, we showed that female-derived M-type CHCs were more attractive than those of the other two female chemotypes, even though the CHC profiles of M-type females are much more similar to male CHC profiles (both being dominated by large relative amounts of the same alkadienes [47]). The evaluation of the female CHC profiles could be based on matching them to different ‘templates', similar to the way nest-mate recognition is thought to be accomplished in social insects (reviewed in [63,64]). However, T. zealandicus responses do not seem to be tuned specifically to each of the three chemotypes, since artificial ‘intermediate' CHC profiles did not negatively impact male behaviour. Instead, male evaluation of the mate recognition signal seems to be sufficiently broad-tuned to cover the large differences between the female chemotypes [65]. This would also allow for mate recognition despite potential further CHC variability in response to external factors, such as temperature/humidity or the larval food source, all of which have been shown to influence insect CHC profiles [15,36,38,41]. Within such a broad response window, particular properties could nonetheless regulate male preference. Specifically, a preference for alkadienes > alkenes > methylbranched alkanes in female CHC profiles would result in the observed male preference hierarchy. Additionally, it would be an explanation for how the ternary mixture of all three chemotypes could be as attractive as the intermediately preferred L-type in the choice assays. The ternary mixture might balance out the attractive properties of the higher amounts of alkadienes with those of the less preferred methylbranched alkanes, and thus lead to an intermediate ranking. However, a preference for large amounts of alkadienes would have to be restricted to their occurrence in a female-specific composition. Results for T. zealandicus reactions towards male and female CHCs of type M in single stimulus assays showed that male CHCs were not bioactive, and the addition of male-derived CHCs to usually attractive female-derived CHCs made the mixture less so. Thus, male CHC profiles are not simply ignored, but seem to contain specific properties that distinguish them from female profiles. While both, male and female M-type profiles are characterized by large amounts of alkadienes, they differ in the relative amounts thereof, which are higher in males [47]. Large amounts of alkadienes could be attractive up to a threshold beyond which they might be perceived as repellent. In Nasonia vitripennis, both females and males produce (Z)-9-hentriacontene (9C31), though it is much more abundant in male CHC profiles [66]. Addition of male-typical amounts of 9C31 to female CHCs reduced the number of males responding, as well as the total duration of copulation attempts [67], a result that is very similar to T. zealandicus reactions to the male-female mixtures in the present study. Alternatively, methylbranched alkanes could differ between male and female CHC profiles in T. zealandicus if males and females produce different stereoisomers. These could, thus, constitute (the essential) part of the mate recognition signal. In another parasitoid wasp, Ooencyrtus kuvanae, male and female CHC profiles also lacked sex-specific compounds and were similar overall. However, one specific combination of stereoisomers of two methylbranched compounds was repellent, while another was attractive [68]. The authors concluded that males and females produce different stereoisomers of these two, and perhaps more, methylbranched CHCs, and that mate recognition by males is based on both, attraction to female and repellence of male-specific stereoisomers [68]. Nonetheless, the assumption that different stereoisomers of methylbranched compounds are responsible for mate recognition in T. zealandicus is unlikely to be the whole truth. While it would allow the distinction between males and females, the female chemotype H contains the largest relative amounts of methylbranched alkanes [47]. It should in that case have been the most attractive one, but was, in fact, the least preferred. Whether stereochemistry plays a role for mate recognition/preference in this species remains to be investigated. The results of the bioassays, the observed preference hierarchy, and the characteristics of the CHC profiles of males and females of the three chemotypes do not match the predictions of a single or a few compounds serving as the close-range sex pheromone in T. zealandicus. Overall, it seems most probable that the profile is perceived in its entirety, as has been suggested for some other parasitoid wasps [69,70], and may be the case in many other insect species, with the relative amounts of many different compounds being of importance for the chemical signal.

5. Conclusion

We show that for male T. zealandicus, male and female M-type CHC profiles, though similar, are clearly not the same [46], while the female CHC chemotypes are perceived as more alike than the remarkable chemical differences suggested. When given the choice, the discrete female sex pheromone chemotypes are nonetheless differentially preferred. Male responses do not seem to be based on matching the female CHCs to three distinct ‘recognition templates', since the comprehensive changes of compound abundances across the whole profile in the ‘intermediate' CHC profiles had no effect. We thus conclude that mate recognition is likely to be based on an integrative assessment of CHCs and their relative amounts that reconciles accurate mate recognition with intraspecific CHC variability.

Variability of CHC profiles is a common occurrence, and may therefore impact close-range CHC communication in a plethora of insect species. However, in many, if not most cases, the changes caused by factors influencing the CHC profiles are intricate and affect a number of different CHCs in concert [4,43,45]. This usually makes it hard to disentangle which changes potentially influence communication. Based on clearly delimited and pronounced natural variation, our study reveals an intriguing insight into how such multi-component, and potentially variable CHC profiles may be evaluated for mate recognition and mate assessment, which could apply similarly also to other species. Rather than individuals being more or less tolerant to divergence from their mate recognition signal, ‘positive' recognition may inherently cover a (sometimes large) degree of variability in the relative amounts of compounds most affected by natural variability (e.g. induced by temperature or humidity). This would minimize the risk of erroneously rejecting potential mates while allowing for the evaluation of specific, variable attributes such as the relative abundance of specific compounds or whole compound classes within the complex composition for mate assessment. At the same time, the receivers' acceptance limits could be clear-cut towards a few particular traits that are characteristic for heterospecifics or conspecifics of the same sex, such as if the relative amounts of key compounds surpass a threshold, or if ‘wrong' compounds are present (similar to the concept of ‘undesirable-present' distinction proposed for nest-mate recognition in ants [64]). Overall, an evaluation of CHC profiles in such a way would permit reliable mate recognition in spite of CHC variability.

Ethics

This work did not require ethical approval from a human subject or animal welfare committee.

Data accessibility

Supplementary material is available online [71].

Declaration of AI use

AI-assisted technologies were used in the last stage of manuscript finalization as spell-checkers.

Authors' contributions

M.S.: formal analysis, investigation; J.R.: resources, writing—review and editing; T.P.: conceptualization, data curation, formal analysis, funding acquisition, investigation, methodology, project administration, supervision, visualization, writing—original draft, writing—review and editing.

All authors gave final approval for publication and agreed to be held accountable for the work performed therein.

Conflict of interest declaration

We declare we have no competing interests.

Funding

This research was supported by a German Research Foundation grant (DFG grant no. PO 2550/3-1) to T.P.