Behavioral Ecology and Sociobiology

, Volume 59, Issue 4, pp 490–499

Queen fertility, egg marking and colony size in the ant Camponotus floridanus

Authors

    • Department of Behavioral Physiology and Sociobiology BiocenterUniversity of Würzburg
  • Jürgen Liebig
    • Department of Behavioral Physiology and Sociobiology BiocenterUniversity of Würzburg
  • Bert Hölldobler
    • Department of Behavioral Physiology and Sociobiology BiocenterUniversity of Würzburg
    • School of Life SciencesArizona State University
Original Article

DOI: 10.1007/s00265-005-0073-0

Cite this article as:
Endler, A., Liebig, J. & Hölldobler, B. Behav Ecol Sociobiol (2006) 59: 490. doi:10.1007/s00265-005-0073-0

Abstract

In ant societies, workers do not usually reproduce but gain indirect fitness benefits from raising related offspring produced by the queen. One of the preconditions of this worker self-restraint is sufficient fertility of the queen. The queen is, therefore, expected to signal her fertility. In Camponotus floridanus, workers can recognize the presence of a highly fertile queen via her eggs, which are marked with the queen's specific hydrocarbon profile. If information on fertility is encoded in the hydrocarbon profile of eggs, we expect workers to be able to differentiate between eggs from highly and weakly fertile queens. We found that workers discriminate between these eggs solely on the basis of their hydrocarbon profiles which differ both qualitatively and quantitatively. This pattern is further supported by the similarity of the egg profiles of workers and weakly fertile queens and the similar treatment of both kinds of eggs. Profiles of queen eggs correspond to the cuticular hydrocarbon profiles of the respective queens. Changes in the cuticular profiles are associated with the size of the colony the queen originates from and her current egg-laying rate. However, partial correlation analysis indicates that only colony size predicts the cuticular profile. Colony size is a buffered indicator of queen fertility as it is a consequence of queen productivity within a certain period of time, whereas daily egg-laying rate varies due to cyclical oviposition. We conclude that surface hydrocarbons of eggs and the cuticular profiles of queens both signal queen fertility, suggesting a major role of fertility signals in the regulation of reproduction in social insects.

Keywords

Queen signalHonest signalingPheromoneCuticular hydrocarbonsWorker policingConflictFormicidae

Introduction

The evolution of social insects is considered one of the major transitions in evolution (Maynard Smith and Szathmary 1995). In insect societies, only a minority of individuals reproduce directly, whereas all others usually forego personal reproduction and instead help in rearing the offspring of relatives. Understanding the ultimate and especially the proximate mechanisms of this kind of reproductive division of labor is still a challenge to evolutionary and behavioral biologists.

Kin selection theory of Hamilton (1964) provides the framework for understanding division of reproductive labor in most insect societies (Bourke and Franks 1995; Pamilo and Crozier 1996). According to this theory, a helper benefits when the condition r>c/b is true where r is the relatedness between the altruist and the beneficiary of the help, c is the loss of direct fitness, and b is the gain of indirect fitness benefits of the altruist. It is immediately obvious that we have two parts in the term, relatedness and the cost–benefit ratio, that determine whether helping is advantageous.

Relatedness is especially important in social Hymenoptera, since offspring is asymmetrically related to each other due to their haplo-diploid sex determination system. If females have the same mother and father, they are related to their sisters by 0.75 but only by 0.25 to their brothers and by 0.375 to their nephews. On the other hand, they are related by 0.5 to their male and female offspring. Under these conditions, females should prefer their own sons or nephews over their brothers (Ratnieks 1988).

Since workers are restricted to male production in most hymenopteran societies, we expect them to produce sons in colonies with a single queen (monogynous) that is singly mated (monandrous) purely on relatedness grounds (Ratnieks 1988). However, a comparative analysis shows that relatedness is not sufficient to explain the pattern of male production in the social Hymenoptera (Hammond and Keller 2004). The common absence of worker-produced males in monogynous and monandrous ant species may also be explained by colony efficiency. In the presence of a fertile queen, worker reproduction may divert resources from brood rearing, which imposes costs on colony productivity and reduces the indirect fitness gains of workers (Cole 1986; Pamilo 1991). Workers are expected to control each other's reproduction (worker policing) (Ratnieks 1988; Monnin and Ratnieks 2001), which may finally induce them to refrain from reproduction (self-policing) if the reduction of their indirect fitness is sufficiently high (Ratnieks and Reeve 1992; Frank 1995; Wenseleers et al. 2004).

Egg destruction and aggression towards fertile individuals are two efficient mechanisms which either prevent worker-laid eggs from reaching adulthood or inhibit worker reproduction. In bees (Apis sp.) and the common wasp Vespula vulgaris, workers eliminate worker-laid eggs via oophagy (egg-eating) in queenright colonies (Ratnieks and Visscher 1989; Ratnieks 1993; Visscher 1996; Foster and Ratnieks 2001; Halling et al. 2001; Oldroyd et al. 2001). In ants, worker policing may either be employed by attacking workers with developed ovaries (Hölldobler and Carlin 1989; Gobin et al. 1999; Kikuta and Tsuji 1999; Liebig et al. 1999; Monnin and Peeters 1999; Hartmann et al. 2003; Iwanishi et al. 2003) or by selective destruction of worker-laid eggs as well, as in the ant Diacamma sp. (Kikuta and Tsuji 1999), Camponotus floridanus (Endler et al. 2004), Pachycondyla inversa (D'Ettorre et al. 2004a), and Formica fusca (Helanterä and Sundström 2005). Nevertheless, in several species, individuals sometimes escape worker policing (Visscher 1989; Walin et al. 1998; Barron et al. 2001; Dietemann et al. 2003).

Worker policing, however, is only advantageous in the presence of a related fertile queen. If the colony is orphaned, workers generally start laying eggs (Bourke 1988a; Choe 1988). We also expect that workers lay eggs if the productivity of the queen decreases and rearing capacities are unused (Seeley 1985; Keller and Nonacs 1993).

In several species, pheromones are involved in the regulation of reproduction between queen and workers (Passera 1980; Hölldobler and Wilson 1983; Hoover et al. 2003; Endler et al. 2004) and between the queen and her gyne offspring (Vargo 1988; Vargo and Hulsey 2000). These pheromones inform both about the presence and the fertility of a queen (Ortius and Heinze 1999; Hannonen et al. 2002). In many species, this kind of queen pheromone seems to be represented by cuticular hydrocarbons (CHCs) that change their profile with fertility (Bonavita-Cougourdan et al. 1991; Ayasse et al. 1995; Monnin et al. 1998; Peeters et al. 1999; Liebig et al. 2000; Sledge et al. 2001; Cuvillier-Hot et al. 2002, 2004a; Hannonen et al. 2002; Heinze et al. 2002; Tentschert et al. 2002; De Biseau et al. 2004).

In large colonies, not every colony member can monitor the queen's condition directly, suggesting indirect communication of her presence or fertility. In the honeybee (Apis mellifera), for example, messenger bees distribute the queen pheromone within the colony (Seeley 1979; Naumann et al. 1991). Alternatively, queen-laid eggs may be used as vehicle for the distribution of a queen signal among nestmates.

Such a mechanism is present in the ant C. floridanus, where the queen indirectly signals her presence to workers via her eggs (Endler et al. 2004). The eggs are marked with a queen-specific hydrocarbon profile that also occurs on the queen's cuticle. Queen and worker profiles show qualitative differences in the composition of the compounds; approximately 46% of the total amount of CHCs is found only in highly fertile queens (10 of the major 35 compounds, see Endler et al. 2004).

In C. floridanus, workers refrain from own reproduction in the presence of queen eggs even though they are not in direct contact with the queen (self-policing). Worker eggs do not elicit such an effect, but instead, they will be destroyed when presented to workers from a queenright colony (worker policing). Worker eggs lack the respective queen-specific hydrocarbon components, but, when they are coated with hydrocarbon extracts of the queen's cuticle, their rate of acceptance by workers increased by 34% (Endler et al. 2004). The results on C. floridanus indicate that workers differentiate between hydrocarbon profiles in the context of reproduction, as suggested by behavioral and electrophysiological experiments in the ant species Myrmecia gulosa and P. inversa (Dietemann et al. 2003; D'Ettorre et al. 2004b). However, it is unknown whether the eggs also transmit information about the fertility of the queen.

In the study of Endler et al. (2004), only highly fertile Camponotus queens from large colonies of more than 1,000 workers were investigated. Nothing is known about queen signaling in smaller incipient colonies in which the egg-laying rate of the queen is relatively low. We hypothesize that in small colonies, queens lack either single or all compounds of the CHC profile, which characterizes queens in large colonies. As cuticular and egg profiles are similar in highly fertile queens, eggs from weakly fertile queens originating from incipient colonies should lack similar parts of the profile. If this is the case, we expect workers from large colonies not to be able to distinguish the eggs from weakly fertile queens from worker eggs. They should accordingly destroy them like worker eggs due to lack of the appropriate queen signal (Endler et al. 2004).

In this study, we investigated whether workers from C. floridanus differentiate between eggs originating from highly fertile queens from large colonies and from weakly fertile queens from small colonies. We offered workers from large queenright colonies eggs from (a) queens from large colonies, (b) queens from incipient nests, and (c) workers as a control. We furthermore investigated the profiles of the surface hydrocarbons (SHCs) whether differences exist among egg layers as this would support the idea that hydrocarbons are responsible for egg recognition. Since we argue that the hydrocarbon profiles represent a fertility signal, we investigated the egg-laying rates of the queens to see whether egg-laying rate is correlated with changes in the hydrocarbon profile. However, as egg-laying rates tend to fluctuate very much, a better indicator of fertility may be colony size as large colonies can only be maintained by highly fertile queens. Thus, colony size is included as another factor to explain variation in CHC profiles of queens as an indicator of overall ovarian activity.

Materials and methods

Animals and egg discrimination experiment

Queens of Camponotus floridanus (N=90) were collected at the Florida Keys, USA, after the mating flight in August 2001, July 2002, as well as June 2003. The very common carpenter ant C. floridanus reaches colony sizes of over 10,000 individuals. The colonies have only one single-mated queen (Gadau et al. 1996) but often several subnests. If the queen dies, the colony will die as well, as re-queening is unknown in strictly monogynous ant species (Hölldobler and Wilson 1990; Heinze and Keller 2000) except Nothomyrmecia macrops (Sanetra and Crozier 2002). Founding queens were transferred to the laboratory and were cultured in plaster nests at 25°C and 50% humidity (12-h day and 12-h night). Twice a week, they were fed with honeywater and cockroaches (Nauphoeta cinerea). Subsequently, approximately 80% raised colonies with 1,000–2,000 individuals within 1 to 2 years from which the experimental colonies originated.

In our egg discrimination experiment, we measured the differential survival of three groups of eggs originating either from queens from incipient colonies, from queens from large colonies, or from workers to establish whether queen eggs from incipient colonies can be identified by workers. Freshly orphaned worker groups (<2 h isolated) each containing 130 individuals (30 foragers and 100 individuals from inside the nest) were provided with eggs of different origin: they received either 30–35 eggs from unrelated workers, 30–35 eggs from a foreign queen originating from a colony with more than 1,000 workers, or 30–35 eggs from a foreign queen originating from an incipient colony with less than 10 workers. Egg number varied, since we wanted to minimize the risk of unknowingly damaging an egg by singling them out from egg clumps. Only eggs without any conspicuously embryonic development were used (identifiable by the formation of the chorionic cavity). This indicates that the eggs have been laid relatively recently. Eggs from workers were laid in colonies that were orphaned for more than 60 days.

For the discrimination experiment, we used ten queenright colonies. For one experimental trial, we isolated three worker groups from one queen colony, resulting in a paired experimental design. The survival of the eggs was calculated as the percentage of the eggs remaining from the original number added to the respective worker group. Survival was measured by counting the eggs 1, 2, and 24 h after the transfer. During the experiment, each worker group was provided with honeywater and one cockroach (N. cinerea).

Determination of fertility

To determine the relationship between fertility and queen signals, we measured oviposition rates of differently fertile queens. We isolated queens of four categories from their colonies in small plastic boxes for 24 h. Group A contained founding queens with less than 10 workers (N=2), group B queens with 50 to 80 workers (N=5), group C queens with 200 to 300 workers (N=10), and group D queens from colonies over 1,000 workers (N=12). Each queen was provided with ten workers and fed with honeywater and a half cockroach (N. cinerea). After 24 h, the eggs were counted, and the queens were returned to their colonies. Each queen and, thus, each colony was measured once; thus, each data point reflects an independent measurement. Sample sizes referring to eggs always represent different individuals that laid the eggs but never different eggs laid by a single individual.

Chemical analysis

As the surface hydrocarbons (SHCs) of eggs have been shown to be a major determining factor of egg discrimination in workers (Endler et al. 2004), we investigated the profiles of eggs originating from workers and queens. We only used eggs less than 24 h old, since eggs change their profile with time (Endler and Liebig, unpublished data). As we focus on differences in fertility among queens and fertility of queens is presumably associated with colony size, we analyzed the egg profiles of queens from differently sized colonies (A, N=20; B, N=16; C, N=14; D, N=19). To establish a connection between egg and cuticular profiles, we also investigated the CHC profiles of queens. A high similarity between the two profiles supports the idea that CHC profiles act directly as a signal and, furthermore, allows using CHC profiles of the egg layers instead of those of the eggs for further investigations. This avoids the time-consuming and disturbing isolation of the egg layers which is necessary to obtain fresh eggs.

The CHCs were extracted with solid-phase microextraction (see e.g., Monnin et al. 1998; Liebig et al. 2000; Endler et al. 2004). The fiber was swiftly rubbed on the tergites of queens for 3 min and on eggs for 2 min. Afterwards, the fiber was directly injected into the injection port of a ThermoQuest Trace gas chromatograph (GC) with a split/splitless injector. Further details are described in Endler et al. (2004). Peak areas were computed with Chrom-Card 1.19 (CE Instruments). The compounds were identified by using Kovats indices on the basis of a GC/mass spectrometry (GC/MS) analysis (see Endler et al. 2004).

Since the differences in the CHC profiles are mostly qualitative and thus represent absence–presence data, principal component or discriminant analyses are not applicable. As the transition from A to D queens is also accompanied by major changes in the presence of some hydrocarbon compounds, we use the most parsimonious way to statistically describe the transition. We divided the cuticular CHC profiles of queens and the SHC profiles of eggs into two parts: (1) the shorter-chained queen compounds (retention time 7.0 to 14.0 min, n-pentacosane to 10-methyl-, 12-methyl-, and 14-methyloctacosane) that are specific to queens of group D and (2) the longer-chained compounds present in both queens and workers (retention time >14.0 min, 12,16-dimethyloctacosane to 5,9,13,17-tetramethyltritriacontane) (see Endler et al. 2004). Then, we calculated the sum of the peak areas of the compounds of the respective parts. This allows describing changes with simple statistical correlations.

In addition to the qualitative differences, we also compared quantitative differences in egg profiles based on compounds present in eggs from A queens, D queens, and workers from orphaned worker groups. These profiles are represented by the compounds or compound mixtures indicated by the numbers 11 (12,16-dimethyloctacosane) through 34 (5,9,13,17-tetramethyltritriacontane) described in Endler et al. (2004). In addition, we included the alkanes n-pentacosane, n-heptacosane, and n-octacosane, since they occur regularly on all eggs. The areas of the peaks were standardized to 100%. The resulting values were transformed according to Aitchison (1986) (see Dietemann et al. 2003). Six variables were excluded from the analysis as they either deviated significantly from the normal distribution (Kolmogorov–Smirnov test with Bonferroni correction, p<0.01, p<0.005, and three times p<0.0001) or did not show homogenous variances (Levene's test after Bonferroni correction, p<0.05 for one variable) (Table 1). The remaining 20 variables were subject to a stepwise backward discriminant analysis. We identified five variables that significantly contributed to the separation. These are represented by three alkanes (n-pentacosane, n-heptacosane, and n-hentriacontane) and a mixture of 4-methyltriacontane and 12,16-dimethyltriacontane and the compound 4,8,12,16-tetramethyldotriacontane. The compounds were identified in comparison to the GC/MS analysis from Endler et al. (2004) using Kovac indices. Statistical tests were performed with STATISTICA 6.1 (Statsoft) and SPSS 12.02 (SPSS Inc.).

Results

A greater fraction of eggs laid by D queens was alive after 24 h compared to eggs laid by A queens and workers, whereas there was no difference between the latter two kinds of eggs (overall difference, Friedman's analysis of variance, p<0.0001; Wilcoxon–Wilcox test for multiple comparisons, eggs from D queens vs eggs from A queens p<0.01, eggs from D queens vs worker eggs p<0.01, eggs from A queens vs worker eggs p>0.1; Fig. 1).
https://static-content.springer.com/image/art%3A10.1007%2Fs00265-005-0073-0/MediaObjects/265_2005_73_Fig1_HTML.gif
Fig. 1

Discrimination of queen- and worker-laid eggs expressed as survival rate after transfer in worker groups. Presented are medians, quartiles, and range. Stars denote extreme values. N (egg samples)=10 for each group

The differences in egg survival correspond to differences in the composition of their SHCs. The median percentage of shorter-chained compounds in the total profile is 47.4 in eggs from D queens (range 31.7–60.5%), 2.2% in eggs from A queens (range 0.6–5.4%), and 1.9% in eggs from workers (range 0.3–10.7%). High proportions of shorter-chained compounds in A queens and workers are due to large amounts of linear alkanes. Differences are significant between D queens and A queens or workers, but not between A queens and workers [Kruskal–Wallis test: egg samples from D queens N=19, from A queens N=19, and from workers N=13, H=35.15, p<0.0001; post hoc comparison (Siegel and Castellan 1988), D queens vs A queens or workers p<0.0001, A queens vs workers p>0.3].

In the longer-chained part of the profile, all focal compounds were present in all individuals allowing for a discriminant analysis (Fig. 2). This part of the profile shows significant differences as well (N as before, Wilks–Lambda N=0.009, p<0.0001). The profiles of eggs from D queens are well separated from those of the other eggs by the first discriminant function (100% correct classification). However, the profiles of eggs laid by workers and by A queens overlap completely on the first axis. The separation of the two groups is largely due to the second axis (91% correct classification), which, however, explains only 1.8% of the variance. The profiles of eggs laid by D queens largely overlap with the profiles of the eggs from the other two groups on the second axis, which indicates that fertility, colony size, sex of the egg, or caste of the egg layer are not associated with the separation on the second axis. The compounds that were excluded from the analysis for statistical reasons show a pattern that supports the results of the discriminant analysis (Table 1).
https://static-content.springer.com/image/art%3A10.1007%2Fs00265-005-0073-0/MediaObjects/265_2005_73_Fig2_HTML.gif
Fig. 2

Comparison of the profiles of eggs originating from workers, A queens, and D queens. The percentage denotes variance explained by the discriminant function. Egg samples originate from 13 workers, from 19 A queens, and from 19 D queens

Table 1

Compounds excluded from the discriminant analysis of egg profiles with their respective transformed values

Compound

Worker

A queen

D queen

Median

Range

Median

Range

Median

Range

9-Methylnonacosane

−1.14

−2.09 to −0.90

−1.44

−2.50 to −1.00

2.02

0.97 to 2.52

13,17-Dimethyl-, 11,15-dimethyl-, and 9,13-dimethylnonacosane

−2.27

−2.77 to −2.09

−2.76

−3.56 to −1.95

1.72

0.89 to 2.06

7-Methyl-, 9-methylhentriacontane, and 13,17-dimethylhentriacontane

0.36

0.08 to 0.73

0.72

0.35 to 0.92

−1.3

−2.11 to 0.30

11,15-Dimethylhentriacontane

−0.04

−0.64 to 0.42

−0.22

−1.17 to 0.77

−0.63

−1.04 to −0.04

N-dotriacontane

1.64

1.05 to 2.06

1.58

1.28 to 1.78

0.45

-0.07 to 1.22

N-tritriacontane

−1.03

−1.59 to −0.66

−0.65

−1.18 to −0.23

−2.36

−2.78 to −1.75

The percentage of the shorter-chained part in the egg profile is positively correlated with the respective part of the CHC profiles of the egg-laying queens from groups A to D (Spearman rank correlation NA queens=8, NB queens=7, NC queens=6, ND queens=19, rs=0.93, p<0.0001; Fig. 3). Overall, the CHC profiles of egg layers and the profiles of the SHCs of their eggs are similar in the proportions of shorter-chained hydrocarbons and in the pattern of the profile (Figs. 3 and 4).
https://static-content.springer.com/image/art%3A10.1007%2Fs00265-005-0073-0/MediaObjects/265_2005_73_Fig3_HTML.gif
Fig. 3

Comparison of the proportion of queen-specific compounds on the cuticle (CHC) and the egg surface (SHC). Queens and their eggs originate from group A (N=8); B (N=7); C (N=6), and D (N=19)

https://static-content.springer.com/image/art%3A10.1007%2Fs00265-005-0073-0/MediaObjects/265_2005_73_Fig4_HTML.gif
Fig. 4

Chromatograms of the CHCs of typical representatives of their categories together with the SHCs of their respective eggs. Only compounds between 6 and 25 min are shown, which represent the hydrocarbon profile. Minor, not reproducible peaks between 0 and 6 min are not shown. For better comparison, the elution times for n-alkanes with chain length from 25 to 33 are indicated. Colony sizes are indicated by letters (see “Materials and methods”)

As we were interested in the information content of the hydrocarbon signal, we investigated the egg-laying rate of the queens of different groups and its relation to changes in the hydrocarbon profile. A queens laid about 1 egg/day on average (median N=7, range 0–9), B queens 9.5 eggs/day (median N=8, range 3–12), and C queens 14.5 eggs/day (median N=14, range 4–41). The highest egg-laying rate with 28 eggs/day on average was observed in D queens (median N=17). Nevertheless, the daily egg-laying rate of queens from these large colonies varied strongly, covering the whole range from 0 to 83. Although some queens produced none or only a few eggs, their colonies contained much brood that developed into females. Furthermore, these queens lived for at least another year after counting egg-laying rate, which suggests that these queens were healthy and produced female brood.

The higher the average daily egg-laying rate, the higher was the relative amount of shorter-chained hydrocarbons in the cuticular profile of the queens. A queens with an egg-laying rate of 0 or 9 eggs/day had less than 0.6% shorter-chained hydrocarbons in their profile, whereas D queens with egg-laying rates between 20 and 83 eggs/day (median 37.5) had between 28.7 and 59.4% shorter-chained hydrocarbons in their profile (N=12, median 50.3%; Fig. 5).
https://static-content.springer.com/image/art%3A10.1007%2Fs00265-005-0073-0/MediaObjects/265_2005_73_Fig5_HTML.gif
Fig. 5

The amount of shorter-chained hydrocarbons on the queen cuticle (CHCs) vs the oviposition rate of the queens in 24 h and colony size. The size of the colonies the queens originated from are indicated by A to D. See text for details and statistics

Besides egg-laying rate, the amount of shorter-chained hydrocarbons in the CHC profile of queens is also associated with the size category of the colony. The larger the colony is from which the egg-laying queen originates, the more shorter-chained hydrocarbons are present in her CHC profile, whereas the strongest variation is present in C queens (Fig. 5). The amount of shorter-chained hydrocarbons increases with both egg-laying rate and colony size (Spearman rank correlation N=29, rs=0.79, p<0.0001 and rs=0.88, p<0.0001, respectively). As egg-laying rate and colony size are positively correlated (Spearman rank correlation rs=0.89, p<0.0001), we used partial correlation analysis to factor out first the effect of colony size and then that of egg-laying rate. When controlling for colony size, egg-laying rate had no effect on the amount of short-chained hydrocarbons (rs=0.04, N=29, p>0.2), but when controlling for egg-laying rate, the amount of short-chained hydrocarbons increased significantly with colony size (rs=0.63, N=29, p<0.001). This indicates that colony size is a better predictor of the CHC profile of egg-laying queens than egg-laying rate.

Discussion

Our results clearly show that workers of C. floridanus extract information about the fertility of queens from their eggs. The treatment of eggs corresponds to the hydrocarbon profiles of the egg surface and the cuticular profiles of the respective queens. A good predictor of the hydrocarbon profiles is the size of the colony the respective queen originates from. Since a large colony size can only be maintained by a certain productivity of a queen, it may represent the moving average of the queen's egg-laying rate, whereas egg-laying rate itself is subject to daily or weakly variation. A strong correlation between colony size and hydrocarbon profiles thus supports the idea that hydrocarbon profiles represent a fertility signal that can be perceived via eggs and the cuticle of the respective egg-laying queen.

The fertility signal on the eggs also seems to regulate the policing of worker-laid eggs by workers in C. floridanus (Endler et al. 2004) as well as in other ant species; for example, differences in the hydrocarbon profiles of eggs are associated with worker policing in the ant P. inversa (D'Ettorre et al. 2004a) and “queen” policing in Dinoponera quadriceps (Monnin and Peeters 1997), although the differences between the eggs are not as pronounced as in C. floridanus. In our experiment, the destruction of eggs from A queens by workers may be a consequence of the inability of the workers to differentiate between eggs from A queens and those laid by workers, since the eggs of A queens lack the fertility signal similar to worker eggs. Hence, workers may falsely perceive eggs laid by A queens as worker-laid ones and therefore destroy them.

A further reason for the destruction of eggs laid by A queens could be their origin. Potentially, workers may recognize that neither these nor worker-laid eggs originated from their colony. However, D-queen eggs also originated from foreign colonies in this study as well as in the study of Endler et al. (2004), and they were accepted. In contrast, eggs laid by sister workers in the study of Endler et al. (2004) were destroyed in worker groups. Thus, foreign colony membership cannot generally be a cause of destruction. If information about colony membership is present, the fertility signal on D-queen eggs presumably prevents destruction. This would, however, not work for foreign eggs laid by A queens and workers. Thus, destruction of eggs laid by workers or A queens in this experiment could potentially be the consequence of the detection of foreign colony membership or of worker policing.

Instead of worker policing, worker-laid eggs have a high mortality per se, as in the honeybee, A. mellifera (Pirk et al. 2004). New data from honeybees, however, strongly oppose these results (Beekman and Oldroyd 2005). Our results do not support differential egg mortality as a factor either, as there was no difference in survival between eggs laid by A queens and workers. In addition, the transfer of queen hydrocarbons increased the survival of worker-laid eggs in C. floridanus in another study (Endler et al. 2004). Thus, we exclude increased mortality of worker-laid eggs as a factor explaining our results. Furthermore, the indiscriminate destruction of eggs laid by A queens or workers excludes other potential causes for differential egg treatment, e.g., differences in morphology, physiology, or other potentially sex-specific properties.

Although workers use the differences in the hydrocarbon profiles to discriminate among eggs, the question is what the differences indicate. If the hydrocarbons represent a fertility signal, the most obvious measure is egg-laying rate. Although egg-laying rate differed among queens, this difference was also associated with differences in colony size. When we corrected for colony size, the partial correlation analysis shows no correlation between CHC patterns of the queens and their current egg-laying rate. On the other hand, colony size strongly correlates with the changes in the hydrocarbon profiles. Thus, colony size is a better predictor for the changes in the CHC profiles of the queens than current egg-laying rate. This may be the consequence of cyclical oviposition of the queens (Endler, personal observation), which is known from other ants (Hölldobler and Wilson 1990; Dietemann et al. 2002, Cuvillier-Hot et al. 2004b). On the other hand, colony size does not vary on a day-to-day basis. Since egg-laying rate of the queen determines colony size in combination with the mortality rate of workers, colony size is a buffered and delayed indicator of the fertility of the respective queen. Potentially, colony size could determine the fertility of the queen as a kind of feedback mechanism. This, however, seems only partly to be the case in C. floridanus, as this species is monogynous and founds colonies solitarily. A queen has to produce a surplus amount of brood which determines colony growth. On the other hand, successful raising of brood depends on a sufficient number of workers, which leads to the close correlation of colony size and egg-laying rate in our analysis.

The change in the hydrocarbon profiles of the queens is not only associated with an increase in colony size as a secondary effect but primarily with a change in physiology. Many studies showed a close link between CHC patterns and fertility (Bonavita-Cougourdan et al. 1991; Ayasse et al. 1995; Monnin et al. 1998; Peeters et al. 1999; Liebig et al. 2000; Sledge et al. 2001; Cuvillier-Hot et al. 2002, 2004a; Hannonen et al. 2002; Heinze et al. 2002; Tentschert et al. 2002; De Biseau et al. 2004). These CHC patterns revert to those of infertile individuals if ovarian activity stops in a ponerine ant (Liebig et al. 2000). In another ponerine ant, Streblognathus peetersi, the experimental manipulation of juvenile hormone (JH) patterns in fertile ants inhibited ovarian activity and this was preceded by changes in the CHC patterns of the respective individuals (Cuvillier-Hot et al. 2004a).

The underlying physiological mechanism also explains the close connection between the surface profiles of eggs and the CHC profiles of the respective queens. This could be a consequence of the same physiological pathway responsible for the transportation of hydrocarbons to target tissues. In the German cockroach Blattella germanica, hydrocarbons are distributed by a hemolymph high-density lipophorin to different tissues in the insect body, including the cuticle and the ovaries where they are incorporated in developing oocytes (Schal et al. 1998; Fan et al. 2003). A similar lipophorin transport pathway is known from the ant Pachycondyla villosa (Lucas et al. 2004). In C. floridanus, the SHCs may thus be already present in the ovaries in similar proportions as on the cuticle and are presumably not transferred from the cuticle on the eggs during egg laying. Eggs thus carry the same information as the cuticle of the egg layer. Since the proportions of shorter-chained hydrocarbons in the surface profile of eggs and the cuticular profile of the respective queen increase with fertility, they both provide information about queen fertility.

Although it is clear that workers use this fertility signal (Endler et al. 2004) for their reproductive decisions, the finer mechanisms of their decision making is generally unclear. Is there any fertility threshold of the queen below which workers should start laying eggs? Should they start laying eggs at all if there is still an egg-laying queen in the colony? This probably depends on the potential benefits they receive from egg laying, and these vary depending on the situation (e.g., Foster 2004). One indication for a high threshold for egg laying in this species is the relatively long delay between orphanage and first worker oviposition. In an isolation experiment of C. floridanus, worker eggs appeared not before 60 days after orphanage, and some worker groups did not yet oviposit after 158 days of isolation (Endler et al. 2004). In several ant species, this delay is much shorter [earliest egg laying after orphanage: 5 (mean 29.5) days (S. peetersi; Cuvillier-Hot et al. 2004b), 6 days (Harpagoxenus sublaevis; Bourke 1988b), less than 13 days (Diacamma ceylonense; Cuvillier-Hot et al. 2002), 17 days (Aphaenogaster cockerelli; Hölldobler and Carlin 1989), less than 21 days (Harpegnathos saltator; Liebig et al. 1999), and 30 days (Gnamptogenys menadensis; Gobin et al. 1999). This large variation in oviposition delay indicates that workers probably respond differently to variations in the fertility signal due to different potential benefits of own reproduction. One would expect therefore that workers invest more in monitoring the fertility of the queen if the potential benefits of own reproduction are higher. A better understanding of fertility signaling will therefore remain at the core for understanding of the mechanisms and evolution of reproductive regulation in social insects.

Acknowledgements

We thank Thibaud Monnin, Christian Peeters, Kazuki Tsuji, a referee, and especially Lotta Sundström for providing helpful comments on the manuscript; we also thank the German Science Foundation (SFB 554 TP C3), the European Union (INSECTS, Integrated Studies of the Economy of Insect Societies, HPRN-CT-2000-00052), and Aventis (travel grant to AE) for funding. The Social Insect Working Group of the Santa Fe Institute provided insightful discussions.

Copyright information

© Springer-Verlag 2005