Introduction

Pheromones are involved in intraspecific chemical communication (Wyatt 2013), and the honey bee (Apis mellifera) has long been the subject of chemical ecology studies (Breed et al. 2015; Free 1987). However, the glands associated with compounds used in nestmate recognition in honey bees remain elusive. This search is difficult since nestmate cues can arise from both within the colony, and from the environment (Kalmus and Ribbands 1952). For example, Downs and Ratnieks (1999) found no evidence that honey bee guards used heritable cues; instead, guards appear to rely exclusively on environmental cues to distinguish nestmates from non-nestmates. However, nestmate cues can be produced by the individual, and thus must be under genetic control (Breed 1983; Page Jr et al. 1991). A further factor is that the wax used to build the colony is both produced and manipulated by the bees, which means it may be a medium into which recognition cues are transferred (Breed et al. 1998). Therefore, Breed et al. (1998) stated that no single factor is responsible for nestmate recognition in honey bees; rather, all three factors (genetically determined cuticular signatures, exposure to comb wax, and environmental cues e.g. floral cues) seem to work together.

In ants, the post-pharyngeal gland (PPG) found in the head can absorb, store, and metabolize its own lipids (Decio et al. 2016), and the hydrocarbon profile (ratio of compounds) of the PPG is very similar to that of the cuticle (Bagnères and Morgan 1991; Kaib et al. 2000). The PPG may also act as a mixing ‘gestalt’ organ (Soroker et al. 1994), since cuticular hydrocarbons (CHCs) can be transferred to an individual’s PPG by nestmates via the sharing of secretions (trophallaxis) or grooming (Meskali et al. 1995; Morgan 2010). Thus, any genetic (CHCs) or environmental differences between individuals can be smoothed out, producing a single unified nestmate odour.

Work, predominantly on social insects, has shown that each species of insect has a unique CHC profile (Kather and Martin 2015), which is remarkably stable over large geographical areas (Guillem et al. 2016). In ants, qualitative differences in these unique species profiles indicates that the CHCs can also contain a nestmate signal (Martin et al. 2008; van Zweden and d’Ettorre 2010). Several studies have indicated the importance of CHCs in nestmate recognition in honey bees. For example, newly emerged bees, which have fewer hydrocarbons in their cuticle, are accepted more readily into an unrelated colony, while the removal of hydrocarbons from older bees improves their acceptance (Breed et al. 2004). Further studies indicated that a sub-set of hydrocarbons, the long-chained alkenes, maybe involved in honey bee nestmate recognition (Châline et al. 2005; Dani et al. 2005; Pradella et al. 2015). However, the location where hydrocarbons, especially alkenes, are stored prior to release onto the cuticle remains unclear.

The five major glands studied (mandibular [MG], hypopharyngeal [HG], cephalic salivary [CSG], thoracic salivary [TSG], and Dufour’s gland [DG]) (Fig. 1) are all vital to an adult worker performing its various tasks both inside and outside the hive successfully (Duffield et al. 1984; Katzav-Gozansky et al. 2001; Poiani and Cruz-Landim 2017). The MG, HG, CSG, and TSG are all paired glands whose ducts open into the buccal cavity (Snodgrass 1956). The primary role of the HG is providing protein rich secretions to feed the larvae during the nursing phase (Crailsheim and Stolberg 1989; Huang et al. 1989), and in foragers these glands secrete enzymes that aid the processing of honey (Simpson 1960). The MG initially secretes fatty acids, again used for larval nutrition (Plettner et al. 1997). They then switch to the secretion of ‘forage-marking’ (Vallet et al. 1991) and alarm pheromones such as 2-heptanone in foragers. In worker honey bees, the salivary gland develops into two glands with the CSG found in the head and the TSG in the thorax. Simpson (1960) noted the secretion of the TSG was aqueous, containing water soluble digestive enzymes, whereas the CSG produced an oily secretion that helps with wax manipulation. However, Katzav-Gozansky et al. (2001) detected a series of n-alkanes in both the TSG and CSG, which partly conflicting with what Simpson (1960) found. Recently, and based on a single colony, Poiani and Cruz-Landim (2017) suggested that the CSG secretion may be used to replenish the CHCs. In contrast the TSG secretions are important for honey maturation (Maurizio 1975), possibly contain pheromones (Maurizio 1975) and help moisten pollen and wax (Simpson 1963). The DG is associated with the sting apparatus and although well studied in queens (Niño et al. 2013) and workers in queen-less colonies, its role in workers from queenright colonies is not well defined (Mitra 2013). It is known to produce hydrocarbons (Katzav-Gozansky et al. 2001) that are used for waterproofing the cuticle and possibly involved in chemical communication (Abdalla and Cruz-Landim 2001).

Fig. 1
figure 1

The position of the five glands, three in the head, one in the thorax, and one in the gaster, evaluated in this study (adapted from Tofilski 2012)

Full size image

The aim of this study was to investigate the presence or absence of hydrocarbons contained within five honey bee exocrine glands, at three different key points in the adult honey bee worker’s life (newly emerged, nurse, and foraging phases). This data was then compared with the CHC profiles from workers doing the same tasks (Kather et al. 2011). This will determine which gland is acting as a reservoir for hydrocarbons that are detected on the surface of the honey bee i.e. CHCs, and will also help determine whether the role of the glands changed as the tasks of the bee changed.

Methods and Materials

Samples Collection and Preparation

All studies were conducted during the summer of 2014 using four honey bee colonies maintained at the University of Salford apiary Manchester, UK. Worker bees of known age were obtained by removing a frame containing mature pupae and placing it into a temperature cabinet controlled at 30 °C ± 5 °C and RH 60% ± 2%. The thorax of approximately 500 newly born workers per colony were marked with a non-toxic Uniposca® pen before being returning to their natal colony. Thereafter, from each of the four colonies, 50 workers were sampled when they emerged (newly emerged bees [day 0]). After 10 days bees were collected from the surface of a frame whilst feeding larva (nurse bees), and after 30 days bees were collected returning to the hive bearing pollen (forager bees). Since hydrocarbons are known to be stable under a wide range of storage conditions (Martin et al. 2009), all collected bees were stored at −20 °C until their glands were removed. Only workers were sampled since CSG glands are poorly developed in male honey bees (Poiani and Cruz-Landim 2010).

Gland Extraction

The five glands (Fig. 1) were dissected out in water using fine tweezers under a Leica binocular microscope at ×25 magnification. The head was separated from the thorax using micro scissors and the TSG extracted from the thorax using fine tweezers. From the head, the buccal apparatus was removed by creating a hole near the mandibles, which was then extended into the ventral region to allow the removal of the MG. The exoskeleton in the dorsal region of the head was removed allowing the HG and CSG to be removed. By removing the stinger, the DG could easily be extracted.

Chemical Analysis

Each sample was a pool of 10 glands to ensure a clear chemical signal. For each colony, five replicates of each age group (0, 10, and 30 days) were prepared. This was replicated across the four colonies giving a total of 300 samples to be analysed (four colonies x five replicates x three age groups x five glands), from 600 bee dissections. Each sample was immersed in 30 μl of high-performance liquid chromatography (HPLC) grade n-hexane and evaporated to dryness at room temperature before being stored at −20 °C. Immediately before analysis samples were re-suspended in 30 μl HPLC grade n-hexane and analysed on an Agilent 7890-GC (equipped with a Vf-5ht UltiMetal column; length: 30 m; ID: 0.25 mm; film thickness; 0.1 μm) connected to an Agilent 5975-MSD (quadrupole mass spectrometer with 70-eV electron impact ionization). Samples were injected in splitless mode with the injection port at 325 °C and the MS in scan mode. The oven temperature programme was 70 °C (held for 1 min), 40 °C min−1 to 200 °C, 4 °C min−1 to 250 °C and finally 25°Cmin−1 to 350 °C with a final five-minute hold. The carrier gas of helium was used at a constant flow rate of 1.0 ml min−1. Compounds were identified using standard MS databases, diagnostic ions, and Kovats indices. This method allowed us to detect hydrocarbons and esters up to chain lengths of forty-two carbons (n-C42). The presence of oleic acid in samples was variable, potentially due to contamination from other body tissues, so was excluded from the analysis. Each peak was transformed into a relative proportion based on the total amount of hydrocarbons (C19 to C33) plus wax esters, if present. Any compound which was consistently less than 1% across all three age groups was excluded from further analysis and recorded as trace (Table 1). An internal standard was not used so all comparisons are relative.

Table 1 The percentage of hydrocarbons and oleic acid wax esters detected in four major glands in three different age groups of honey bees
Full size table

Statistical Analysis

The raw data was arcsine square root transformed to reduce the range of data values, and non-metric Multidimensional Scaling (NMDS) performed using Euclidean distances in SPSS v23. Goodness of fit was measured using the Normalized raw stress. The lower the stress value, the better the data are represented, with values of <0.05 considered excellent, and where values of >0.2 are considered a poor representation of the data and should be interpreted with caution. NMDS does not rely on the assumptions of normality of multidimensional data, thus having the advantage of omitting any distributional assumptions required for other methods (McCune and Grace 2002). Moreover, the pre-assumed hypothesis-driven groupings such as those used in discriminant analysis, which can bias the final result, as can minor compounds (Martin and Drijfhout 2009). Also, if the number of variables is sufficiently large, which is often the case, the canonical variates separate the groups regardless of the actual group distributions (Mitteroecker and Bookstein 2011). The CHC profiles from workers (Kather et al. 2011) and CSG (this study) came from the same honeybee colonies, but the CHC profiles were analysed using a standard GC column, so before the datasets were compared the high temperature oleic acid waxy esters were removed from the data-set (see Fig. 3b).

Results

In the TSG trace amounts of n-pentacosane (n-C25) were detected, irrespective of age group and no other non-polar compounds were seen. As expected no hydrocarbons were detected in the HG, while the MG of new-born bees contained only small amounts of the n-alkanes and alkenes typically found in the CSG and DG, which became trace amounts of n-C25 and n-heptacosane (n-C27) in older age groups (Table 1). The DG contained predominantly (70–88%) odd chained n-alkanes (n-heneicosane [n-C21] to n-hentriacontane [n-C31]), with small proportions of alkenes (3–8%), whereas the methyl branched-alkanes occur predominantly in newly emerged bees, and the oleic acid wax esters were found mainly in the foragers (Table 1).

The CSG contained, relative to the other glands, a large quantity of n-alkanes, alkenes, methyl branched-alkanes, and oleic acid wax esters (Table 1; Fig. 2). The proportions of these compounds changed dramatically as the bees aged (Table 1; Fig. 3a) and mirrored the changes seen in the adult CHC profiles (Table 1: Fig. 3b). In newly emerged bees (day 0) 10% of the CSG consisted of methyl branched-alkanes, which dropped to just 2% after 10 days and to trace amounts at 30 days. The dominant hydrocarbons (47%) in newly emerged workers were the three n-alkanes (n-C19, n-C21, and n-C23), these were replaced by C31:1 and C33:1 alkene isomers (30%) after 10 days, before the appearance of the oleic acid wax esters (48%) in the 30-day old bees (foragers) (Fig. 4; Table 1). This changing pattern was mirrored in the DG (Table 1; Fig. 3a). In the CSG gland the C31:1 and C33:1 alkene isomers were present in all age groups but were a major part of the profile in 10-day old bees (nurses). At this time colony-specific patterns in the C31:1 and C33:1 isomer ratios were seen in all colonies. In other words, there was little within colony variation of the isomer pattern for all four colonies, whilst between colony difference could be detected (Fig. 5). These patterns cannot be detected in day 0 and day 30 bees due to the small amounts of alkenes detected. However, such colony-specific patterns were not seen in any other group of compounds.

Fig. 2
figure 2

The relative amount, based on total ion counts of CHCs and oleic acid wax esters, in the five exocrine glands (HG = hypopharyngeal, MG = mandibular, CSG = cephalic salivary, TSG = thoracic salivary, and DG = Dufour’s gland), at the three different honey bee ages; day 0 = newly emerged, day 10 = nurses, and day 30 = foragers. The error bars represent +1 SD. The pie charts indicate the proportions of the CHC and oleic acid wax esters at each age (CSG = clear, MG = grey, DG = dotted and MG = black)

Full size image
Fig. 3
figure 3

a NMDS analysis of the four glands contain different hydrocarbons that change as the bees aged (see Table 1). Each dot represents the mean profile for each colony. b NMDS analysis indicates the similarity between the CSG and CHC profile for each age group (dimension 1). Each dot represents the mean chemical profile (excluding esters) for each group across all samples in each of the four colonies. The Normalized raw stress values were a) 0.05184 and b) 0.01813, indicating good representation of the data

Full size image
Fig. 4
figure 4

Typical Total Ion Chromatograms of the cephalic salivary gland from (a) newly emerged bees, (b) nurse bees, and (c) foragers. The numbers indicate the compounds presented in Table 1. The pie diagrams indicate the proportion of n-alkanes (clear), alkenes (lines), methyl branched-alkanes (black) and oleic acid wax esters (dots). * = Oleic acid (C18)

Full size image
Fig. 5
figure 5

a The colony specific differences in the C31:1 and C33:1 isomers ratios between the four colonies detected in the cephalic salivary glands at day 10. Colonies are sometimes distinct (1, 4 and 2) or less distinct (2 and 3). b These observations were supported by an NMDS analysis with a normalized raw stress value of 0.00779. Each point represents a pooled 10-day CSG sample from each of the four colonies

Full size image

Discussion

This study found that, among the five major exocrine glands studied, the CSG was the only one to contain large amounts of CHCs. The production, or storage, of lipids in the CSG, and not the TSG, is supported by evidence from both physiological (Poiani and Cruz-Landim 2010) and proteomic (Fujita et al. 2010) studies. The CHCs detected in the CSG are typical of those reported on the surface of adult honey bees (Table 1, Kather et al. 2011; Pradella et al. 2015), thus supporting the findings of Poiani and Cruz-Landim (2017). Across a wide range of Hymenoptera the DG also contains hydrocarbons found on their cuticle. These include bumblebees (Oldham et al. 1994), social Polistes (Dani et al. 1996) and Ropalidia wasps (Mitra and Gadagkar 2014), as well as solitary wasps (Howard and Baker 2003) and solitary bees (Pitts-Singer et al. 2017). Interestingly, in the stingless bee Melipona bicolor the DG has been lost in the workers (Abdalla et al. 2004) but remains in the queen and contains hydrocarbons and esters, as found in many social Hymenoptera queens (Le Conte and Hefetz 2008).

In this study the MG contained a range of hydrocarbons in newly emerged bees that became reduced to only n-pentacosane and n-heptacosane as the bees aged, with tritriacontene (C33:1) appearing in the nurses (Table 1). However, in the stingless bee Melipona quadrifasciata, n-alkanes and alkenes found in the MG persist throughout the entire adult life and are similar to the adults CHCs (Cruz-Landim et al. 2012). Therefore, the functions of the CSG and MG may be interchangeable in different genera of bees.

The predominance of alkene isomers C31:1 and C33:1 found in the CSG, especially in nurse bees, suggests a role in sequestering recognition compounds. It has been suggested that these alkenes serve as nestmate recognition compounds in honey bees (Châline et al. 2005; Dani et al. 2005; Pradella et al. 2015). Furthermore, a large diversification of alkene isomers are uniquely found among bees (Kather and Martin 2015a) indicating that these play an important role in chemical communication in this group. Interestingly, a similar pattern of n-alkanes, alkenes, and oleic acid wax esters found in the CSG are also present in comb wax (Aichholz and Lorbeer 2000; Fröhlich et al. 2000; Waś et al. 2014). Of the three age groups studied, only the nurse bees have active wax glands, which reach their maximum size in 5 to 15-day old workers (Blomquist et al. 1980). The wax scale produced by the nurse bee is manipulated into brood cells by adding a frothy saliva substance that increases lipolytic activity, helping to change the waxes physical structure (Kurstjens et al. 1985). During this time compounds from the CSG could be added to the comb wax. This may help explain why cues gained from comb wax can affect a honey bees’ level of aggression towards its nestmates (Breed et al. 1998; Downs and Ratnieks 1999). In male honey bees the CSG is poorly developed and regress with age (Poiani and Cruz-Landim 2010). However, in male Bumblebees (Bombus spp) the CSG is well developed and contains species-specific sexual marking pheromones (Šobotnik et al. 2008).

This study and a previous one (Poiani and Cruz-Landim 2017) both found similar age-related changes in the hydrocarbons detected in the CSG despite being conducted in Brazil and UK and on different sub-species of bee (Africanised and Buckfast). This suggests these patterns of change are widespread across A. mellifera. That is, the methyl branched-alkanes associated with newly emerged bees disappear from the CSG and CHCs after 10 days. In contrast, alkene production peaks during the nurse phase, potentially allowing them to be incorporated into the wax-comb. While the CSG of foragers is dominated by several long (>C32) waxy esters. Similar heavy (>C35) waxy esters have been detected on the surface of 11 species of insects from nine genera of Hymenoptera using high temperature GC-MS (Sutton et al. 2013) and could be ubiquitous in social insects, acting as waterproofing compounds (Nelson et al. 2001), although they could also play a role in communication. For example, Amsalem et al. (2009) found a series of C24-C28 wax esters in the DG of Bombus terrestris workers that did not lay eggs and suggested they signal a worker’s functional sterility to other members of the colony. Furthermore, the honey bee brood pheromone is composed of a series of C17-C20 fatty acid esters that act as a primer pheromone (Le Conte et al. 2001). These esters are produced in the salivary gland of the larvae prior to it bifurcating into the CSG and TSG during the pupal stage.

This study indicated that the CSG in honey bees has a parallel function to the PPG in ants and wasps (Herznera et al. 2011), functioning as a reservoir for hydrocarbons that are also present on the surface of honey bees. In addition, the CSG contains a group of oleic acid wax esters whose function needs to be elucidated in the future.