Theory predicts that females should invest least in mate searching when young, but increase their effort with age if they remain unmated. Few studies have examined variation in female sexual signalling. Female Dawson's burrowing bees (Amegilla dawsoni) search for males by signalling their receptivity on emergence, but many leave the emergence site unmated and must attract males at feeding sites. Female bees prevented from mating on emergence had more extreme versions of cuticular hydrocarbon profiles that make them attractive to males, lending empirical evidence of adaptive shifts in female mating effort.
Theory predicts that males should expend the greater effort on mate searching when they benefit more from the acquisition of multiple mates, or when females suffer greater costs from searching . Searching for mates includes both physical movement, as well as various forms of signalling. In general, mate searching by males is the most commonly observed pattern; however, there are circumstances under which females are expected to search . Indeed, female mate searching is found in a variety of taxa; female moths emit pheromones , female fireflies emit flashes of light  and female bushcrickets produce acoustic signals . However, while mate searching by males has been the subject of considerable research effort, relatively little attention has been paid to elucidating the costs and benefits of mate searching by females.
Males frequently impose mating costs on females  so that females might be expected to exert only limited effort in signalling to attract them. However, females can experience reproductive failure if they fail to attract a mate . In a recent theoretical analysis, Umbers et al.  showed that even with very small costs associated with signalling, females should signal less intensely when young and increase their signalling effort the longer they remain unmated. In support of their prediction, Umbers et al.  presented evidence from the literature that a common pattern among moths is for unmated females to increase their signalling effort as they age. The cost of pheromone signalling for females has been difficult to document, but adaptive adjustments in signalling effort in response to elevating risk of reproductive failure provide good evidence of signalling costs . Umbers et al.'s  theoretical approach thus offers a useful framework with which to explore female mate search efforts more generally, which are often subtle in comparison to the overtly competitive nature of male mate searching. Here, I ask whether female Dawson's burrowing bee (Amegilla dawsoni) adjust their sexual signalling in response to the potential risk of remaining unmated.
Dawson's bee is a ground nesting solitary bee found in the arid regions of northwestern Australia . Males are protandrous, and search for newly emerging females that mate only once before the onset of nesting, which lasts for around six weeks. Emerging females signal their receptivity via cuticular hydrocarbons (CHCs) . Male mate searching is highly efficient, with 80–90% of females mating on emergence. Females are frequently discovered by multiple males and the struggles that ensue can be fatal for females. However, some 10–20% of females leave the emergence site unmated, either because they were undetected or because they escape groups of males during mating struggles. After leaving the emergence site, females feed at nearby stands of their host plants where unmated females can meet and mate with males that adopt an alternative mating tactic of searching for foraging females . When mated females return to nest at the emergence site, they are ignored by searching males as their CHC blends become unattractive .
Female insects often signal their receptivity to males using CHCs , which are costly to manufacture . Signalling sexual receptivity in a male-dominated mating system like that of A. dawsoni is also costly for females that attract too much male attention. Umbers et al.'s  theoretical model predicts that, given the costs of signalling receptivity, female A. dawsoni should signal less intensely at emergence when the probability of being discovered is high, but increase their signalling efforts when they remain unmated after leaving the emergence site.
2. Material and methods
(a) Sample collection
Fifty bees were collected at an emergence site 5 km south of Carnarvon, Western Australia in early August 2011. The emergence site was monitored from 10.00 to 16.00 each day over a period of one week. Emerging females were located by scanning the site for increased male activity around existing emergence holes or cracks in the surface of the ground that indicated a tunnel being excavated from below. Males were cleared from the area and the hole covered with a 30 ml vial. The emergence of a female was confirmed when her head capsule appeared at the opening. Forty emerging females were allocated haphazardly to one of four treatments: two mating treatments and two non-mating treatments.
Ten females were collected immediately on emergence and frozen. An additional 10 unmated females were placed into net enclosures containing their host plant, Trichodesma zeylanicum, and a pad of moistened cotton from which to source water. These females were left for 24 h before being frozen. Twenty females were allowed to copulate. Thus, the vial was removed from the emergence hole allowing a male to locate the emerging female. Immediately the male mounted, the pair were covered with the vial and copulation was observed. Copulation involves a period of genital contact lasting 2.5 min during which the female is inseminated, followed by 5.5 min of postcopulatory courtship . In order to assess the role of insemination and postcopulatory courtship in triggering non-receptivity, for 10 females the male was removed immediately after he disengaged his genitalia, before the onset of postcopulatory courtship. For 10 females, the male was allowed to perform postcopulatory courtship and the pair left to separate naturally. Males were released and females placed into net enclosures containing their host plant and a water source as described above. Mated females were frozen after 24 h. Finally, 10 females that were nesting in the emergence area were collected. When a female was observed entering a nest, it was covered immediately with a 30 ml vial. On leaving the nest, she was captured in the vial and frozen immediately.
(b) Cuticular hydrocarbon extraction and analysis
All bees were thawed within 24 h of freezing, and placed into glass vials containing 2 ml of AR grade hexane for 5 min. After the bee was removed from the hexane, the extract was reduced in volume by 50% via evaporation and then transferred to a 1 ml Chromacol vial and stored at 4°C. To control for any possible contamination that might arise by conducting this work in the field, hexane blanks were prepared alongside the samples by following the identical procedures but without placing a frozen bee into the glass vial.
Extracts were analysed by injecting 1 μl into a gas–liquid chromatography and mass spectrometer (GCMS, Agilent GC-6890N, MS-5975 with inert mass selective detector) fitted with a factor four VF-5MS column (30 m × 250 μm × 0.25 μm with 10 m guard). The GCMS was operated in the splitless mode, using helium as the carrier gas (1 ml min−1). The oven was programmed with an initial temperature of 100°C for 1 min and was ramped up by 7°C min−1 to 250°C for 15 min. Transfer from the GC to the MS was set at 250°C. MSD Chemstation (Agilent) was used for data quantification and spectra were identified with NIST MS Search 2.0 and an in-house library (see the electronic supplementary material). Peaks found in the hexane blanks were not considered in subsequent analyses.
Standard procedure for CHC data is the calculation of log-contrasts for use in statistical analyses . Thus, relative peak areas were calculated by dividing the area of each peak by the sum of all peak areas. Log-contrasts were calculated for each peak by diving relative peak area by the relative peak area of a randomly chosen peak and taking the log. Log-contrasts correct for non-independence introduced into the dataset by calculation of relative abundance. Pentacosene (C25H50) was used as the denominator. Log (1 + x) was used because some peaks were not present in all individuals. Log-contrasts were subjected to discriminant function analysis (DA). DA is a technique used when group membership is known a priori and scores on multiple predictor variables, such as the abundance of a number of CHC compounds, are available . It generates linear combinations of variables (canonical scores) that can then be used to determine how well the multivariate data predict group membership.
Twenty-one CHC compounds were identified from the hexane washes (table 1). Discriminant analysis returned four canonical axes (CAs) that collectively explained 100% of the variation in CHC profiles (table 2). The analysis correctly classified 92% of females into their groups, which differed significantly in the relative abundances of CHC compounds (Wilk's λ = 0.005, F80,105 = 3.80, p < 0.001). Separation of treatment groups on the first two CAs is shown in figure 1. The 95% confidence intervals on the mean canonical scores indicate that CA1 separated nesting females from newly emerging females, and among newly emerging females, unmated females 24 h after emergence from both unmated females on emergence and inseminated females (with or without postcopulatory courtship) 24 h after emergence. CA2 separated newly emerging females from all other groups.
Theory predicts that even with very small costs, females should signal less intensely for mates when young, but increase their signalling effort when they remain unmated . Newly emerging unmated female burrowing bees, A. dawsoni, differed in their CHC profiles from nesting females. Perfuming experiments have demonstrated that the CHCs of newly emerging females are highly attractive to mate searching males, whereas the CHCs of nesting females are unattractive . Among newly emerging females, those that remained unmated after 24 h had CHC profiles that scored higher on the major canonical axis of CHC variation that distinguishes attractive emerging females from unattractive nesting females. Thus, consistent with theoretical prediction , emerging females that remained unmated appeared to have increased their signalling effort by adjusting the intensity of the CHC blend on their cuticles that makes them attractive to searching males. Increased mate search effort makes adaptive sense because unmated female bees can only produce haploid male offspring. However, whether the observed change in CHC blend makes older unmated females more attractive requires behavioural demonstration. In general, there remains a lack of studies that have shown empirically that increased female signalling effort corresponds to increased attraction of males .
The compounds that weighted most heavily on the CAs tended to be those in low abundance. It has been suggested that minor compounds may have disproportionately large effects on discriminant analyses of CHC profiles . Analysis using only compounds with more than 5% abundance was successful in separating nesting females, but not the remaining female classes (electronic supplementary material). This suggests that subtle differences in minor compounds are necessary for statistical separation between classes of newly emerging females, calling for studies that isolate these compounds and confirm their impact on relative female attractiveness.
The females of many insects signal their receptivity to searching males via CHC profiles . However, as noted by Thomas , almost all previous studies have been unable to conclude whether females actively signal their receptivity to males or whether CHC profiles simply change with age. The second major axis of variation in CHC profiles separated newly emerging females from those aged 24 h (mated or unmated) or older (nesting females), indicating that CHCs do show ontogenetic changes. However, this axis of variation did not separate unmated from nesting females, indicating that ontogenetic changes in CHC profiles alone are unlikely to explain differences in attractiveness. However, it is clear that changes to the CHC profile that make nesting females unattractive take longer than the 24 h period allowed in these experiments.
Changes in female CHC profiles following mating are not owing to male-derived compounds administered during copulation . Insemination and/or postcopulatory male courtship may be responsible for changes in female signalling effort. However, females that had experienced postcopulatory courtship did not differ from those that had not, indicating that postcopulatory courtship is unlikely to be involved. Cessation of pheromone signalling by mated female moths has been linked to substances transferred in the ejaculate , which remains a likely mechanism triggering the changes in sexual signalling effort from receptive to nesting females of this species.
In conclusion, these data lend empirical support to the expectation that females should make adjustments in their mate searching efforts to balance the costs and benefits of attracting males. Whether the changes in CHC blends associated with lack of mating in these bees increases female attractiveness remains to be tested.
Data are available from Dryad: http://dx.doi.org/10.5061/dryad.2bm41.
The author declares no competing interests.
This research was supported by the ARC (DP110104594).
I thank Maxine Lovegrove and John Alcock for field assistance.
- Received April 13, 2015.
- Accepted June 2, 2015.
- © 2015 The Author(s) Published by the Royal Society. All rights reserved.