The two endoparasitoids were reared in our laboratory at the University of Neuchatel. The C. glomerata rearing was started with individuals that had emerged from Pieris brassicae caterpillars collected from cabbage plants grown in gardens around Neuchatel, Switzerland. The offspring was reared on P. brassicae (first–second instar) that was fed on cabbage. C. marginiventris were initially obtained from the United States Department of Agriculture-Agricultural Research Service (USDA‑ARS), Biological Control and Mass Rearing Research Unit (Stoneville, MI, USA), occasionally replenished with individuals from field collections in Mexico and reared as described in Tamò et al. [23]. The hosts were about three-day old Spodoptera littoralis caterpillars (first–second instars), which were fed with a wheat germ-based artificial diet. The eggs of S. littoralis were provided by Syngenta (Stein, Switzerland). To obtain virgin adult wasps, each parasitoid cocoon was placed in a 1.5-mL centrifuge tube until a wasp emerged. Then, virgin females and males were kept separately in two Bugdorm-1 cages (30 × 30 × 30 cm, Mega View Science Education Services Co. Ltd, Taiwan), provided with honey and moist cotton wool in a 25 °C incubator (LD 16:8 h) for three days before each test. To obtain mated individuals, one female and one male were placed in a Petri dish (90 × 15 mm) and then tested immediately or stored in Bugdorm-1 cages with honey and moist cotton wool in a 25 °C incubator for about 18 or 40 hours before tests.

The bioassays were carried out in a six-arm olfactometer, as described by Turlings et al. [24]. Each arm had an air flow of 0.6 L per minute that entered the central release chamber (Figure 1). Six virgin wasps were released at the same time in this central chamber, where they were allowed to choose among the arms within 30 minutes. From previous studies (i.e., [23,24]), we know that the wasps are initially attracted to the light above the olfactometer. When attracted by an odor, they will enter the respective arm and walk until they encounter metal screens that prevent them from walking further into the arm. At this point, they move upwards into a glass bulb that is located just before the metal screen (Figure 1). They will readily stay in these bulbs for the duration of the experiment. Thirty minutes after their release, the wasps in the bulbs were counted and removed, before a new group of six wasps was released.

Test for the presence of female pheromones: Six virgin females, six newly mated females and six unmated males were placed into three non-adjacent arms, respectively, and the three control arms in between were left empty. Several overlapping layers of metal meshes kept the “treatment” wasps from entering the central chamber, prevented the choosing wasps from approaching the pheromone source and reduced any potential visible cues (Figure 1A). Unmated males were released in the central chamber. The test was repeated four times, with six releases of six unmated males (4 × 6 × 6) on each experimental day, for a total of 144 wasps per experiment. Between replicates, the positions of the different treatments in the setup were changed.

Test for the presence of male pheromones: Six unmated males, six newly mated males and six virgin females were placed as odor sources in the olfactometer with the same design as above. There were four releases of six virgin females per replicate. This assay was replicated four times, with a total of 96 wasps.

Test for repellency of mated females: To test if the observed poor responses to already mated females in the previous experiments were due to repellency, we combined odor sources as follows. Six virgin females were combined with six mated females; six virgin females were combined with six unmated males; and six virgin females were kept alone in each of the three non-adjacent arms separated by empty arms (control). The combined groups of six wasps were placed in two chambers in the olfactometer, as shown in Figure 1B: Chamber 1 contained the virgin females, and Chamber 2 included the six mated females (0, 18 or 40 hours after mating), or six unmated males, respectively, or was left empty. Each test was replicated four times with new wasps as odor sources and 6 × 6 unmated males each time (144 wasps).

Test for species specificity of Cotesia pheromones: Six virgin females of C. glomerata and C. marginiventris were placed into two opposite arms, and the remaining four arms were left empty. Unmated males (6 × 6) of either species were released into the central chamber to test their responses to these odor sources. An additional test was done to assess the responses to the interspecific pheromone in the absence of the conspecific pheromone. For this, only one arm of the olfactometer contained six virgin C. glomerata or C. marginiventris females, while the other five arms were left empty. Unmated males of the other species (6 × 6) were released into the central chamber. Each test was repeated four times (144 wasps).

Statistics: To test whether the differences among the responses of the parasitoids to the treatments were significant, we used generalized linear models (GLMs) with the assumption that the arms are equally chosen by parasitoids without stimuli [24]. The models take into account the possible effects of over-dispersion caused, for instance, by positional biases or wasps affecting each other’s responses [25]. Each model was fitted by maximum quasi-likelihood estimation in the software package, R. In the figures, the number of wasps choosing empty arms was divided by the number of empty arms present in the setup to make it comparable to the other treatments.

Our study reveals that virgin Cotesia females attract conspecific males by volatizing pheromones. In the two studied species, virgin males do not attract females from a distance (about 15 cm in our test, Figure 1A). As a gregarious parasitoid, it should be easier for males of C. glomerata to find females to mate as compared to the solitary sister species. We frequently observed that males of C. glomerata emerged earlier and stayed on or near the cocoon cluster, waiting for the females to emerge. The attractant has been shown to be emitted already at the late developmental “black-eye” stage of the cocoons [21]. Mating normally takes place soon after female emergence within 10 cm from the natal patch, which is within the pheromonal functional range [22]. In our tests, the odor cues were attractive at a distance of at least 18 cm (Figure 1B). Not all individuals mate on the natal patch: the percentage of C. glomerata wasps mating on the natal patch has been estimated at about 60% [22]. Since there are very few females that parasitize their hosts without mating in the field [1], outbreeding must be considered an important part of the mating strategy of C. glomerata. However, the pheromones are apparently active only over relatively short distances. We found that the attraction over a distance of 18 cm (Figure 4A,C) was lower compared to a 13 cm distance (Figure 2A), as indicated by the response rate. Response rates are further reduced at a greater distance [26]. Over longer distances, the mating strategies of these and other parasitoids may also involve host‑induced plant volatiles (HIPVs). Cotesia plutellae, for instance, probably uses both sex pheromones and plant volatiles for mate finding at a long distance [27], and a few studies suggest that HIPVs aid in the mate finding in other genera and families of hymenopteran parasitoids [28,29], but direct evidence is still lacking. C. glomerata is also able to use visual cues to locate mates [30].

Compared with the gregarious parasitoids, for solitary wasps, the challenge to find a mate would logically require detection over longer distances. Yet, certain solitary wasps may mate in their natal patch, because the hosts are aggregated, as is the case for aphids and scale insects [31]. This is not the case for caterpillars that serve as hosts for C. marginiventris, which generally disperse after hatching [32,33,34]. Therefore the offspring hatching from these hosts has to engage in a longer-range mate search. Our results imply that males actively search for females and that they use a pheromone released by females.

Cotesia females mate only once [16,19]. After mating, the female parasitoids switch their foraging behavior to host finding. Indeed, we found that mated females of the two Cotesia species lost their attractiveness rapidly after mating (Figure 2), and mated females of C. glomerata even became repellent to males (Figure 4). Something similar has been reported for the braconid aphid parasitoids, Aphidius nigripes and Aphidius ervi: in a field study, mated females attracted significantly fewer males than did virgin females [5,7]. Secretions transferred during male ejaculation may inhibit pheromone production in females. This phenomenon is known as pheromonostasis [35,36] and is known to occur in dipterans and lepidopterans [11,13,37]. For certain Lepidoptera, including Pieris species that serve as hosts of C. glomerata, it has even been found that males transfer so-called “anti-aphrodisiacs” to females, resulting in reduced attraction and even repellency [38,39]. This in turn may attract egg parasitoids in search of a ride to oviposition sites [40,41]. An analogous phenomenon has been reported for Drosophila melanogaster, where cuticular hydrocarbons that are characteristic for males have anti-aphrodisiac properties and are transferred to females during mating [42,43]. Several hours later, mated females even appear to mimic males by synthesizing this same compound to deter courting males [42]. It is possible that the transfer of “anti-aphrodisiacs” also occurs in the studied Cotesia wasps. In C. glomerata, it could explain why mated females, as well as males are repellent to conspecific males, but the fact that the females are not yet repellent right after mating (Figure 4) suggests a change that occurs in the females. Alternatively, the males might deposit a repellent pheromone during copulation that needs some time to be released or activated.

Due to their distinct distribution within a colony, gregarious and solitary parasitoids can be expected to differ in their pheromone communication system. For C. glomerata, mated females were found repellent to males. This allows males to optimize their efforts to find unmated females and minimizes disturbance of females by males while they forage for hosts. In contrast, mated females of C. marginiventris exert no repellent effect upon males (Figure 4).

Parasitoids are expected to choose the lower-cost option between actively searching for mates and the biosynthesis of pheromones in sexual communication [10]. It is not likely for both sexes to release the pheromones at the same range. When the individuals hatch close to each other, as is the case with gregarious parasitoids, it could be expected that the wasps do not invest in costly pheromone release. This is clearly not the case for C. glomerata, since the females emit pheromones in spite of the large proportion of females that mate in the natal patch [1,22]. The attractiveness of the pheromones was at least as strong in C. glomerata as in C. marginiventris, suggesting that at this level, there is no difference in the investment in pheromones between the solitary and gregarious parasitoid.

Cotesia females mate only once, but the males are able to mate several times during their lifetime [16,19]. Interestingly, females of C. glomerata avoided mated males in the six-arm olfactometer (Figure 3A). Apparently, females give a priority to unmated males for mating, possibly because they would transfer more sperm. In addition, males of C. glomerata are repelled by conspecific males. Males emerge earlier than females, and then, most of them stay around the clustered cocoons, fanning their wings or waiting motionless on an unemerged cocoon [44]. Therefore, the observed mutual repellency may promote male dispersal and reduce competition. Considering that most males in a patch will be 100% related, this competition avoidance will be highly adaptive. The male-male repellency was not observed in C. marginiventris, where competition among brothers is much less evident, as their cocoons are much more widely distributed.

In several parasitoids, some cross-attraction between related species has been observed [15,16,17]. In contrast, we found the specificity of pheromonal responses in the two Cotesia species to be very high. C. glomerata males even avoided the odor of C. marginiventris females (Figure 5B). To our knowledge, no evidence of repellency between different parasitoid species has previously been reported. The apparent recognition between the two Cotesia species indicates a close evolutionary relationship and enforces prezygotic reproductive isolation. The genus, Cotesia, is exceedingly diverse [18], and pheromone-mediated avoidance among different genotypes may have contributed to this diversity. The identification of the actual pheromones that are involved in the interactions is necessary to further analyze the relationships among Cotesia species.