Introduction
Social insects are characterized by sophisticated organization and cooperation, such as division of labour in reproduction, as well as defence against predators and pathogens (Lach et al., Reference Lach, Parr and Abott2010). The evolutionary and ecological success of social insects has been attributed to this highly integrated organization and cooperation (Hölldobler & Wilson, Reference Hölldobler and Wilson1990). However, there are drawbacks associated with group living. For instance, social insects are sensitive to the spread of pathogens due to close relatives living in a humid and enclosed nest (Hughes et al., Reference Hughes, Eilenberg and Boomsma2002). As a result, social insects have evolved complex behaviours to defend against the threat of infection ranging from prophylactic to therapeutic strategies. In prophylactic strategies, healthy individuals take steps to prevent the intake and spread of pathogens in the nest. For example, wood ants incorporate antimicrobial tree resins into their nests to protect themselves and brood (Chapuisat et al., Reference Chapuisat, Oppliger, Magliano and Christe2007; Castella et al., Reference Castella, Chapuisat and Christe2008; Brütsch & Chapuisat, Reference Brütsch and Chapuisat2014). Similarly, honeybees produce propolis to defend against pathogens (Simone et al., Reference Simone, Evans and Spivak2009; Simone-Finstrom & Spivak, Reference Simone-Finstrom and Spivak2010). In therapeutic strategies, sick individuals and/or their nestmates usually alter their social behaviour in response to pathogens, including frequent hygienic care towards infected individuals (Ugelvig & Cremer, Reference Ugelvig and Cremer2007; Yanagawa & Shimizu, Reference Yanagawa and Shimizu2007; Baracchi et al., Reference Baracchi, Fadda and Turillazzi2012), social avoidance and even attack of infected nestmates (Heinze & Walter, Reference Heinze and Walter2010; Baracchi et al., Reference Baracchi, Fadda and Turillazzi2012; Bos et al., Reference Bos, Lefèvre, Jensen and D'ettorre2012).
Corpses of ants that have died due to pathogen infection pose a high risk to the colony due to the potential risk of the pathogen spreading (Diez et al., Reference Diez, Deneubourg and Detrain2012; Sun & Zhou, Reference Sun and Zhou2013). Various behaviours of social insects can reduce this risk. Moribund Temnothorax unifasciatus workers leave their nest and die in isolation (Heinze & Walter, Reference Heinze and Walter2010). In most ant species, dead nestmates are actively removed from the nest by workers (Diez et al., Reference Diez, Deneubourg, Hoebeke and Detrain2011, Reference Diez, Deneubourg and Detrain2012, Reference Diez, Moquet and Detrain2013), and in rare cases, corpses are buried inside and/or outside the nest (Renucci et al., Reference Renucci, Tirard and Provost2010) or even cannibalized (Howard & Tschinkel, Reference Howard and Tschinkel1976).
Owing to interesting characteristics and evolutionary importance, corpse removal in ants has attracted much attention. In social insects, most social behaviour is based on chemical communication among nestmates (Richard & Hunt, Reference Richard and Hunt2013). Previous studies have suggested that olfactory cues are likely involved in corpse recognition, and there are two non-exclusive main mechanisms suggested. The first is that the necrophoric behaviour of worker is triggered by the accumulation of decomposition products in or on the corpse, such as fatty acids and some esters (Howard & Tschinkel, Reference Howard and Tschinkel1976). In the archaic ant Myrmecia vindex, oleic acid, and to a lesser degree caproic acid, elicits workers to display burying reaction and sometimes the transport of contaminated objects to refuse middens (Haskins & Haskins, Reference Haskins and Haskins1974). The second is that the removal of dead nestmates is elicited by the rapid disappearance of chemical vitality signals. For example, in the Argentine ant (Linepithema humile), two compounds found in the pygidial gland secretion, dolichodial and iridomyrmecin, which are present on live ant cuticle, disappear within about 40 min after death (Choe et al., Reference Choe, Millar and Rust2009). However, there is evidence suggesting that synergism between tactile and chemical cues are involved in the corpse recognition and burial. In termites (Reticulitermes virginicus), both tactile and chemical stimuli are required to recognize corpses and elicit a necrophoric response; neither cue presented alone would elicit a necrophoric response (Ulyshen & Shelton, Reference Ulyshen and Shelton2012).
Necrophoresis (the removal of dead bodies from the colony) is critical for the survival of social insects (Diez et al., Reference Diez, Lejeune and Detrain2014). However, no study has compared necrophoric behaviour of workers towards corpses of nestmates in different developmental stages within social insects. In this paper, we used the red imported fire ant Solenopsis invicta as a model species because it is not only a serious pest in the warmer regions of the planet but also a well-established model system for studying ant behaviour. Like any other social insect, S. invicta colonies, with their rich store of food, abundant mass of immature larvae and pupae, as well as suitable temperature and humidity, suffer attack from numerous pathogens and parasites (Tschinkel, Reference Tschinkel2006). Given the fact that different chemical profiles are present on pupa and workers (Walsh & Tschinkel, Reference Walsh and Tschinkel1974), we hypothesized that S. invicta would respond differently towards corpses of different developmental stages. Similarly, adult workers are often well protected from diseases, but their brood is highly vulnerable and usually concentrated within the nest, making it a nutritious target for many diseases. We expected that corpses (especially the pupal corpses) infected by a fungus (Metarhizium anisopliae) would accelerate the necrophoric behaviour of workers. To test these hypotheses, we first investigated whether workers of S. invicta have different necrophoric responses to worker and pupal corpses as well as the effect of time since death on the ability of resident workers to discriminate a corpse from a live individual. We also investigated whether resident workers would react more quickly towards M. anisopliae infected corpses. In order to identify chemical cues that elicit necrophoric behaviour, the cuticular profile of the corpses of both workers and pupae at different time intervals since death were analysed with gas chromatography/mass spectrometer (GC–MS). Finally, because fatty acids accumulate with time on a pupal corpse, we examined the effects of these fatty acids on the necrophoric behaviour towards pupal corpses.
Methods
Ants
Four colonies of S. invicta were collected from four districts far from each other on the campus of the South China Agricultural University in Guangzhou, China in April 2013. Each colony was reared in plastic boxes (50 × 40 × 15 cm3) and fed ad libitum with Tenebrio molitor larvae and 25% sucrose water every other day. Laboratory conditions were kept at 25 ± 1°C and 85 ± 1% relative humidity, with a constant photoperiod of 12 h day−1. All behavioural experiments were performed at least 5 weeks after collection.
Preparation of conidial suspensions
M. anisopliae obtained from termite (Coptotermes formosanus) corpse was cultured in Petri dishes (9 cm diameter) containing potato dextrose agar and incubated in a constant temperature incubator at 25 ± 1°C, 85 ± 1% relative humidity (RH ) and 14-h daylight for 10 days. The conidia were brushed from cultures and suspended in 0.01% aqueous Tween-80, with concentrations measured using a haemocytometer.
Experimental set up
From the four colonies collected in the field, we established four standard laboratory nests containing about 200–300 workers and 50 broods. The ants were kept in plaster nests (10 × 10 × 8 cm3) connected to a foraging arena (10 × 8 × 8 cm3) with a corridor (20 × 2 × 4 cm3) where observations of necrophoric behaviour were made. The ants were acclimated under laboratory conditions for at least 2 weeks. The colonies were housed in environmental chambers under the above-mentioned rearing conditions. A Sony camcorder was placed directly above the corridor about 10 cm from the nest entrance (fig. 1). All ants behaved normally indicating that the camcorder did not influence the behaviour of ants.
A two-chamber connected arena was used to test necrophoric behaviour of S. invicta. The left chamber was the feeding area and the right chamber was the nest area. These two chambers were connected with a corridor. A Sony camcorder was placed about 10 cm directly above the corridor. The behaviour of workers in the corridor was recorded by the camcorder and viewed on the screen.
Necrophoresis towards either fungal-infected or non-infected worker corpses
We observed how necrophoric behaviour changed when faced with worker corpses that were either fungal infected or non-infected at different times since death. Corpses were all intra-colonial, and similar sized (media workers) haphazardly captured from each colony. Two days before the experiment, workers were exposed to the fungus by gently swirling and submerging them in a conidial solution at a 1 × 108 conidia ml−1 concentration for 10 s and then placing them on filter paper to dry. The controls were submerged and swirled in 0.02% Tween-80 solution and then let to dry on filter paper. Fungal-exposed ants and controls were kept in separate Petri dishes (9 cm diameter). We confirmed that the ants were infected after 2 days based on observations with scanning and transmission electron microscopes (online Figure S1 and Figure S2). The protocols for scanning electron microscope (SEM) and transmission electron microscope (TEM) sample preparation and observation follow Asensio et al. (Reference Asensio, Lopez-Llorca and Lopez-Jimenez2005). Two days later, live-treated workers were killed by freezing. Corpses were then maintained in a constant temperature incubator at 25 ± 1°C, 85 ± 1% RH for 0, 15, 30, 60 min or 1 day. Corpses were placed outside the nest about 3 cm from the nest entrance, using just one corpse per assay. There was at least 1 h interval between every two bioassays on the same colony to avoid the influence of short-term memory of ants, the same below. The experiment used a randomized block design with ten treatments (fungal-infected or normal worker corpses both with five time intervals post death), four experimental blocks (four ant colonies), with five replicates in each treatment (i.e., total of 200 experimental units). The necrophoric behaviour of workers towards the corpses was videotaped. We recorded the time between the introduction and removal of each corpse and the number of corpse discarded into refuse piles.
Necrophoric behaviour towards pupal corpses
Pupae were killed by freezing and treated with conidial solution at a 1 × 108 conidia ml−1 concentration or control Tween-80 solution as described above. We videotaped the necrophoric behaviour of workers towards ten pupal corpses in each of the four S. invicta colonies for corpses at each of the five time intervals since death, and for the two fungal treatments. This experiment followed a randomized block design with ten treatments (fungal-infected or normal pupal corpses both with five time intervals post death), four experimental blocks and ten replicates in each treatment (i.e., 400 experimental units). We recorded the removal time of each corpse and the number of corpses discarded into refuse piles.
Chemical analysis
We tested the cuticular chemicals presented on fungal-infected and control corpses of worker and pupae at different times since death. Four worker or pupal corpses were submerged in 200 μl of n-hexane (Sigma-Aldrich, ≥99% purity) for 15 min with four replicates. One-microlitre aliquot of each extract was injected into a gas chromatograph (Agilent Technologies 6890N) coupled with a MS (Agilent 5975), in a split-less mode with a solvent delay of 3 min. The GC system was equipped with a capillary column (30 m × 25 μm × 0.25 μm, HP-5MS). The oven temperature for corpses was initiated at 60°C and held for 1 min, raised to 200°C at a rate of 20°C min−1, 200–280°C at a rate of 4°C min−1, and maintained at 280°C for 2 min. Helium was used as a carrier gas at a flow rate of 1 ml min−1. The injector temperature was 260°C and the ion-source temperature was 200°C. Chemicals were identified according to standard compounds (Sigma-Aldrich, ≥99% purity) and compared with previous published data (Dall'Aglio-Holvorcem et al., Reference Dall'Aglio-Holvorcem, Benson, Gilbert, Trager and Trigo2009; Eliyahu et al., Reference Eliyahu, Ross, Haight, Keller and Liebig2011). Palmitic acid, linoleic acid and oleic acid each were diluted sequentially in n-hexane to concentrations of 1, 10, 100 and 1000 μg ml−1. One microlitre of each prepared solution was injected into the GC–MS. GC detection showed that the peak areas of fatty acids in pupal corpses extract were close to those of the standard samples at 10 μg ml−1. Therefore, 1 μl of authentic sample at 10, 20, 30, 40 and 50 μg ml−1 were injected in GC to obtain the calibration curves. The amount of each fatty acid on each pupal corpse was quantified by comparing the peak area with calibration curves.
Effect of fatty acids on necrophoric behaviour
Workers reacted significantly more quickly to worker corpses 1 h since death than freshly killed corpses. However, we did not detect any accumulation or dissipation of chemicals on the surface of worker corpses 1 h post death, but we did detect an accumulation of fatty acid on the surface of pupal corpses. We wondered whether and how these fatty acids influence necrophoric behaviour of resident workers towards pupal corpses. Freshly killed pupae were used as a base to test the level to which the additions of various fatty acids on a pupal corpse influence its removal, because freshly killed pupae rarely induce necrophoric behaviour (see the Results). Freshly killed pupae were placed in Petri dishes with filter paper and 0.5 μl of fatty acid (oleic acid, linoleic acid, palmitic acid or the combination of these three acids) solution were applied with a pipette. The amounts of these three fatty acids used were equivalent to fungal-infected pupal corpses that were dead for 2 days. By this time after death, most corpses were discarded to refuse piles. The solutions of fatty acids were prepared in hexane. Pure hexane was used as a control. Hexane was evaporated at room temperature for 10 min. For each fatty acid and for the combination of fatty acids, five pupal corpses were introduced into the arena simultaneously with three replicates for each colony. The per cent of corpses discarded onto refuse piles was calculated.
Statistical analysis
All data were analysed with statistical product and service solutions (SPSS) 13.0 Software. The normal distributions of data were tested using the Shapiro–Wilk test. Levene's test was used to test homogeneity of variance. Removal time and aggression towards worker corpses from different colonial origin was analysed using general linear model (GLM) with colonial origin, time since death and nest as three factors. Removal time and aggression towards fungal treated worker and pupal corpses was analysed using GLM with fungal treatment, time since death and nest as three factors. The quantities of fatty acids on pupal corpses as well as the impact of addition of fatty acids on corpses on necrophoresis were compared with GLM. The interactions between variables were also tested. We also perform analysis of variance (ANOVA) analysis after GLM. Multiple comparisons in GLM and in ANOVA were performed using a Tukey's test with bonferroni correction. At each time post death, t-tests were used to compare the percentage of discarded pupal corpses or fatty acid quantity between fungal-exposed and normal pupal corpses.
Results
Necrophoric behaviour towards worker corpses
There was no significant interaction between fungal treatment and time on removal of worker corpses (GLM: F 4,160 = 0.455, P = 0.769; fig. 2a). The interaction between fungal treatment and nest, as well as the interaction between time and nest, on removal time were not significant either (GLM: fungal treatment × nest, F 3,160 = 1.314, P = 0.272; time × nest, F 12,160 = 0.666, P = 0.782; fig. 2a). We tested if some S. invicta colonies inherently removed corpses more quickly than others, but we did not find any significant effect of colony on the removal time (GLM: F 3,160 = 0.174, P = 0.914; fig. 2a), and the fungal treatment also did not influence the removal time (GLM: F 1,160 = 0.701, P = 0.404; fig. 2a), but time since death significantly influenced corpse removal (F 4,160 = 34.85, P < 0.001; fig. 2a).
Removal time of fungal-infected or control worker corpses at different times post death.
Necrophoric behaviour towards pupal corpses
There was no interaction effect between three factors on removal time (GLM: fungal treatment × time, F 4,12 = 0.358, P = 0.833; nest × fungal treatment, F 3,12 = 0.277, P = 0.841; nest × time, F 12,12 = 0.77, P = 0.67; fig. 3). No significant effect of nest was found on pupal corpse removal (GLM: F 3,12 = 3.321, P = 0.057; fig. 3). However, both fungal treatment and time since death influenced the corpse removal significantly (GLM: fungal treatment, F 1,12 = 6.113, P = 0.029; time since death, F 4,12 = 84.04, P < 0.001; fig. 3). On average, there was a negative relationship between the number of pupal corpses that were retrieved into the nest with time, which means that an increasing number of corpses were carried to refuse piles with time (ANOVA: non-infected pupal corpses, F 4,15 = 23.244, P < 0.001; fungal-infected pupal corpses, F 4,15 = 20.703, P < 0.001; both with Bonferroni correction, fig. 3). The per cent of discarded pupae was significantly different at 1 day post death in fungal-infected and non-infected pupal corpses (t-test, P < 0.05), but it was no longer different at 2 days post death (t-test, P > 0.05).
Proportion (mean ± SE) of pupal corpses discarded to refuse piles by workers at different times since death (t-test; *, P < 0.05).
Chemical analysis
Even though the workers displayed different necrophoric behaviour towards freshly killed worker corpse and corpse 1 h post death, we did not detect any chemical alteration between freshly killed and 1 h post-death corpses (table 1 and fig. 4). However, we detected chemical differences in pupal corpses at different times post death, and fungal infection accelerated this chemical alteration (table 1 and fig. 5). We identified three compounds that increased with time since death: palmitic acid, linoleic acid and oleic acid. The amounts of these compounds and average per corpse increased were influenced by time post death and fungal infection (fig. 6). There was no significant interaction among time since death and fungal infection on the quantities of these fatty acids on the surface of pupal corpses (GLM: palmitic acid, F 3,24 = 1.647, P = 0.205; linoleic acid, F 3,24 = 0.441, P = 0.726; oleic acid, F 3,24 = 0.799, P = 0.507). However, we found a highly significant effect of time since death on palmitic acid quantities (F 3,24 = 77.22, P < 0.001). Fungal infection also significantly influenced the palmitic acid quantities (GLM, F 1,24 = 14.316, P = 0.001). Similarly, the quantities of linoleic acid and oleic acid were significantly influenced by the time since death (GLM: linoleic acid, F 3,24 = 136.71, P < 0.001; oleic acid, F 3,24 = 17.683, P < 0.001). Fungal infection significantly influenced the oleic acid quantity (GLM, F 1,24 = 5.692, P = 0.025), but not the linoleic acid (GLM, F 1,24 = 3.882, P = 0.06). One day after death, the average quantity of palmitic acid on each fungal-infected pupal corpse was 0.62 ± 0.076 μg, which was significantly higher than 0.38 ± 0.004 μg on non-infected corpse (t-test, P < 0.05). Fungus-infected pupal corpses also had a significantly higher accumulation of linoleic acid and oleic acid on the surface than non-infected corpses on 1 day post death (linoleic acid: P < 0.05, 1.02 ± 0.03 and 0.48 ± 0.04 μg on surface of infected and uninfected corpses, respectively; oleic acid: P < 0.05, 0.83 ± 0.22 and 0.14 ± 0.04 μg on surface of infected and uninfected corpses, respectively). The quantities of each fatty acid on corpses of 2 and 3 days since death were not different between fungal-infected and non-infected corpses (t-test, P > 0.05).
Example gas chromatogram of extract from the surface of worker corpses. Numbers correspond to compounds listed in table 1. In the upper right corner showed that gas chromatograms were not different between freshly killed worker corpses and worker corpses dead for 1 h.
Gas chromatograms of extracts from the surface of pupal corpses. (a) Freshly freeze-killed pupae; (b) control pupal corpses 1 day after death; (c) fungal-infected pupal corpses 1 day after death; (d) control pupal corpses 2 days after death; (e) fungal-infected pupal corpses 2 days after death. Compounds were listed in table 1, and compounds 23, 24 and 25 represent palmitic acid, oleic acid and linoleic acid, respectively.
Quantities (mean ± SE) of fatty acids on pupal corpses as a function of the fungal infection and time since death. (a) Palmitic acid, (b) linoleic acid and (c) oleic acid (t-test; *, P < 0.05).
Chemicals of whole body surface extracts of corpses of S. invicta workers and pupae.
Note: Chemicals of 1–22 present on the surface of corpses of workers and chemicals of 1–25 present on the surface of corpses of pupae.
1 Indicative mass spectrum fragments; M+ is the molecular ion.
2 Chemicals present only on the surface of pupae corpses.
Effect of fatty acids on necrophoric behaviour
The factor of colony did not influence necrophoric behaviour towards freshly killed corpses with 0.5 μl of fatty acid added (oleic acid, linoleic acid, palmitic acid or the combination of these three acids) (GLM: F 2,30 = 1.145, P = 0.332; fig. 7). However, the addition of different kinds of fatty acids significantly influenced necrophoric behaviour (GLM: F 4,30 = 33.687, P < 0.001; fig. 7). First, workers retrieved most of the control corpses treated with hexane into the nest; only 4% being discarded, indicating that the addition of hexane solvent on pupal corpses rarely triggered a necrophoric response of workers. The necrophoric response of workers to pupal corpses with added palmitic acid was similar to that of hexane; about 11.1% pupal corpses were discarded. However, the addition of linoleic acid or oleic acid significantly increased the necrophoric response of workers (F 4,40 = 32.798, P < 0.001, with Bonferroni correction): with added linoleic acid 50.5% of corpses were discarded, 68.9% with added oleic acid and 80% with the combination of three fatty acids (fig. 7).
Proportion (mean ± SE) of freshly killed pupal corpses discarded after addition of fatty acids on the surface of corpses (ANOVA, P = 0.05).
Discussion
Our work confirms that S. invicta workers display differential necrophoric behaviour towards corpses of different developmental stages in a colony. Workers removed worker corpses within 1 h post death, but they take a long time (24–48 h) to exhibit necrophoric responses towards pupal corpses. During their lives, workers use chemicals cues to recognize workers, larvae, pupae and the queen (Sturgis & Gordon, Reference Sturgis and Gordon2012). After death, chemical alterations on the cuticle of corpses are responsible for eliciting necrophoric behaviour (Diez et al., Reference Diez, Moquet and Detrain2013; Maák et al., Reference Maák, Markó, Erős, Babik, Ślipiński and Czechowski2014). S. invicta workers carried worker corpses to refuse piles within 1 h, suggesting that workers could discriminate corpses of workers quickly. But we did not find any alteration in the chemicals contributing to the different necrophoric response towards worker corpses within 1 h. Nevertheless, we found the accumulation of fatty acids on the cuticle of pupal corpses post death. Palmitic acid, linoleic acid and oleic acid were three major fatty acids that increased on the cuticle of pupal corpses over time post death. There were tiny amounts of these fatty acids on the cuticle of freshly killed corpses, and these corpses were constantly retrieved into the nest. As the amounts of fatty acids increased, the pupal corpses were gradually discarded to the refuse piles. In Argentine ant (L. humile), workers can recognize dead nestmates quickly; within 1 h iridomyrmecin and dolichodial that are present on live ants disappear and contribute to the rapid necrophoric response (Choe et al., Reference Choe, Millar and Rust2009). We suggest that there may be some polar chemicals or volatile or tiny amount of compounds we did not detect eliciting the quick necrophoric behaviour towards worker corpses of S. invicta. Moreover, tactile cues may play a partial role in recognition of dead nestmates (Ulyshen & Shelton, Reference Ulyshen and Shelton2012). Further chemical and behavioural analyses are needed to examine these hypotheses.
In S. invicta, workers killed by Beauveria bassiana are removed more quickly than freeze-killed ants (Fan et al., Reference Fan, Pereira, Kilic, Casella and Keyhani2012). However, we did not detect acceleration of necrophoric behaviour towards corpses infected with M. anisopliae, but we did find acceleration of necrophoresis toward fungus-infected pupal corpses. There are large amounts of hydrolytic and lipolytic enzymes in M. anisopliae (Schrank & Vainstein, Reference Schrank and Vainstein2010). Hydrolytic enzymes could turn triglycerides into oleic acid (Akino & Yamaoka, Reference Akino and Yamaoka1996) and lipolytic enzymes could accelerate the decomposition of lipid to linoleic acid (Gilby, Reference Gilby1965). Thus, fungal infection may result in the quick accumulation of fatty acids on the cuticle of corpses, which induce the accelerated necrophoric behaviour by workers. The accelerated removal of infected pupal corpses can reduce the risk of further fungal spread to the brood and queen. The fungal-infected pupal corpses were recognized and removed outside the nest more quickly than non-fungal-infected pupal corpses (47.5% normal and 73.8% infected pupal corpses were discarded on 1 day post treatment), which may result in protection against an epidemic within a colony.
Our work indicates that linoleic and oleic acids are critical chemical signals triggering necrophoric response towards pupal corpses; the application of these fatty acids to a freshly killed pupal corpse increased the percent of corpses discarded to the refuse piles. Similarly, in the red ant (Myrmica rubra), fewer freshly killed worker corpses were removed to refuse piles but significantly more corpses were removed between 1 and 6 days post death. Using GC–MS, Diez et al. (Reference Diez, Moquet and Detrain2013) found fatty acids were absent on freshly killed red ant worker corpses but appeared post death within several hours or days. The addition of pure fatty acids enhanced the level of necrophoric behaviour. Further, dummies treated with oleic acid were usually buried by workers; a behaviour also reported for the termite R. virginicus (Ulyshen & Shelton, Reference Ulyshen and Shelton2012). The role of oleic and linoleic acids as the chemical cues evoking necrophoresis in social insects may have evolved from repellence effects elicited by these fatty acids (Diez et al., Reference Diez, Moquet and Detrain2013). For example, oleic and linoleic acids induce repellent behaviour in gregarious cockroaches: cockroaches avoid shelters treated with these fatty acids (Rollo et al., Reference Rollo, Czvzewska and Borden1994). On the other hand, the same compounds can elicit opposite behaviour such as foraging, depending on the context. In the southern harvester ant (Pogonomyrmex badius), papers treated with oleic acid are carried to the refuse piles when a majority of workers are undertaking hygienic activities. However, when a majority of workers are foraging, oleic acid-treated papers are retrieved into the nest as though they were food items (Gordon, Reference Gordon1983).
Necrophoresis in eusocial insects is derived from non-eusocial ancestors avoiding dead conspecifics and recognizing fatty acids as death cue (Hölldobler & Wilson, Reference Hölldobler and Wilson1990). However, death recognition through the disappearance of a vital chemical sign may be a novel evolutionary trait in some eusocial insects (Sun & Zhou, Reference Sun and Zhou2013). Rapid recognition of a worker corpse by the disappearance of chemical cues may be more advantageous than pupal corpse recognition based on the accumulation of fatty acids associated with decomposition. This is because in the former case, ants can remove worker corpses quickly before corpse decomposition, regardless of its infection status, thus limiting the spread of pathogens in the colony. This is especially important in an enclosed nest with high humidity and relatively stable temperature.
Our work has shown that S. invicta workers displayed differential necrophoric behaviour towards corpses at different developmental stages. We confirmed the critical role of fatty acids in the removal of pupal corpses, and fungal-infected pupal corpses are removed earlier than non-infected ones. As worker corpses were removed quickly within 1 h, it will be worth investigating whether the chemical profile alteration occurs within several minutes after death eliciting necrophoric response of workers. Furthermore, because chemical profiles may differ among developmental stages and castes in a colony, one may wonder whether workers show different necrophoric responses toward larval, male, virgin female and queen corpses within a colony. It would also be interesting to investigate how the disappearance or accumulation of chemical signals on corpses of larvae, males and females would trigger the necrophoric behaviour of workers.
Supplementary Material
The supplementary material for this article can be found at http://www.journals.cambridge.org/BER
Acknowledgments
The authors thank M.P. Zalucki for significant scientific revisions and comments. This research was financially supported by the International Science and Technology Cooperate Programme (grant no. 2011DFB30040), Science and Technology Planning Programme of Guangdong Province (grant no. 2011B031500020) and Science and Technology Planning Programme of Guangzhou City (grant no. 2013J4500032).
