Insect. Soc. DOI 10.1007/s00040-010-0122-1
Insectes Sociaux
RESEARCH ARTICLE
Changes in the temporal pattern of antennal drumming behavior across the Polistes fuscatus colony cycle (Hymenoptera, Vespidae) S. Suryanarayanan • A. E. Hantschel C. G. Torres • R. L. Jeanne
•
Received: 29 April 2009 / Revised: 21 September 2010 / Accepted: 23 September 2010 Ó International Union for the Study of Social Insects (IUSSI) 2010
Abstract Social wasps in several genera exhibit a diverse array of conspicuous vibrational behavior patterns closely associated with larval feeding. Polistes, as the only genus in which these substrate-borne mechanical signals have been studied in some detail, is a useful system for understanding their functions. Most Polistes species examined perform antennal drumming (AD) in the context of feeding prey liquid to larvae. Two existing hypotheses on the function of AD propose that it is a behavioral releaser signal that regulates the release of larval saliva, but with opposite effects. One proposes that AD stimulates larvae to release their saliva for the drumming adult to imbibe, whereas the other proposes that AD inhibits saliva release. A recently proposed third hypothesis argues instead that AD has a modulatory effect on development: exposure to high levels of AD biases larvae toward a worker phenotype as adults. While the larval-saliva-release hypothesis for AD function has little support, predictions made by both the inhibition hypothesis and the mechanical switch hypothesis are yet to be tested within the broader ontogenetic framework of the Polistes colony cycle. We investigated the contexts, rates of performance, and actors associated with AD across 13 weeks of the P. fuscatus colony cycle. Mean colony-wide rates of AD
were high during pre-emergence and early post-emergence stages, but dropped dramatically following the third week after the first workers emerged. This variation in the temporal pattern of AD was correlated neither with the rate at which larvae were fed liquid, the number of larvae on the nest, nor with the adult-to-larva ratio, but was solely a function of colony stage. In contrast, rates of feeding liquid to larvae varied only as a function of the number of larvae in the nest. Queens drummed and fed liquid to larvae at much higher rates than did workers. Queen AD and feed-liquid rates decreased after the third week of worker emergence. During the same period, total feed-liquid rates of workers became as high as levels of queens during pre-emergence. Colony-wide AD rates dropped dramatically because workers seldom drummed while feeding liquid to larvae. The mean duration of AD bursts for queens also decreased after the second week of worker emergence. These results fail to support the salivary inhibition hypothesis, but provide indirect support for the mechanical switch hypothesis on AD function. Keywords
Polistes Caste Vibration Development
Introduction S. Suryanarayanan (&) A. E. Hantschel Department of Zoology, University of Wisconsin, Madison, WI, USA e-mail:
[email protected] C. G. Torres Department of Psychology, University of Wisconsin, Madison, WI, USA R. L. Jeanne Departments of Entomology and Zoology, University of Wisconsin, Madison, WI, USA
Social wasps in the independent-founding polistine genera Belonogaster, Mischocyttarus, Polistes, and Ropalidia perform a diverse array of conspicuous vibrational behavior patterns associated with brood feeding (Jeanne, 2009). While hypotheses on their function(s) have been posited, the role(s) of mechanical signals within these social wasp societies remain contested and unresolved (Jeanne, 2009). Polistes is the only genus in which substrate-borne mechanical signals associated with brood feeding have been studied in some
S. Suryanarayanan et al.
detail (Pratte and Jeanne, 1984; Brillet et al., 1999; Brennan, 2007; Suryanarayanan and Jeanne, 2008). Most Polistes species in which vibrational behavior has been studied perform antennal drumming (AD) (Pratte and Jeanne, 1984). The behavior is closely associated with the feeding to larvae of the liquid residue from prey malaxation: the female wasp strikes her antennae synchronously on the rim of a nest cell for about a second, then enters the cell and regurgitates crop liquid to the larva (Pratte and Jeanne, 1984; Suryanarayanan and Jeanne, 2008). Current understanding of the function(s) of AD and the contexts in which it occurs are based primarily on (1) studies that considered AD as part of investigations on other behavioral traits (Rau, 1928; Owen, 1962; Corn, 1972; Dew, 1983), (2) laboratory studies that specifically focused on AD in the North American species, P. fuscatus (Pratte and Jeanne, 1984; Suryanarayanan and Jeanne, 2008), or (3) laboratory studies of a related vibrational behavior, abdominal wagging (AW), which is also seen in the context of larval feeding in the European P. dominula (Brillet et al., 1999; Brennan, 2007). While prior studies converge strongly on AD’s being a signal directed toward larvae, they disagree on its postulated functions. Hypotheses on the function of AD in P. fuscatus have assumed the behavior to have releaser effects on the behavior of larvae (Rau, 1928; Owen, 1962; Pratte and Jeanne, 1984). Some authors have postulated that the signal stimulates larvae to release salivary fluid, which is rich in amino acids and sugars (Hunt et al., 1982) and is eagerly sought by adults (Rau, 1928; Morimoto, 1960; Owen, 1962). However, experimental manipulations using vegetable dye (Pratte and Jeanne, 1984) and radiotracers (Suryanarayanan and Jeanne, 2008) indicate clearly that adults pass liquid to the larvae following AD, not the other way around. These results do not support the stimulatory hypothesis in P. fuscatus. A second hypothesis proposes the opposite effect, that AD inhibits larvae from releasing salivary secretion. Working with lab-raised colonies of P. fuscatus, Pratte and Jeanne (1984) found some support, albeit weak, for this when they showed that larvae that had just been exposed to AD released 23% less salivary fluid upon artificial solicitation than larvae that had not been recently drummed. Pratte and Jeanne argued that the response is adaptive in that it reduces the likelihood that the larva will release saliva while the adult is attempting to feed it prey liquid. This posits that AD is an essential signal for feeding prey liquid to larvae. This ‘salivary inhibition hypothesis,’ so far based on findings from queens and brood on pre-emergence-stage colonies (up to the emergence of the first adult offspring), deserves to be tested further by situating it within the broader ontogenetic framework of the Polistes colony cycle.
In the typical colony cycle of a temperate Polistes such as P. fuscatus, the founding female(s) rear worker offspring during the pre-emergence phase. When the first adult workers emerge, initiating the post-emergence phase, they raise larvae destined to be overwintering, non-working and future reproductive females (gynes). The emergence of males and gynes characterizes the reproductive phase (Grechka and Kipyatkov, 1984; Reeve, 1991). Worker females have low cold tolerance and short life-spans. Gynes, which initiate colonies in the next nesting cycle, are less aggressive and have greater cold tolerance and longer lifespans (O’Donnell, 1998; Hunt and Amdam, 2005). The first few workers to emerge are smaller in body size compared to later-emerging workers and reproductives (West-Eberhard, 1969; Haggard and Gamboa, 1980; Miyano, 1983), although this size difference disappears under experimental surplusfood conditions (Karsai and Hunt, 2002). Jeanne (2009) recently proposed the alternative hypothesis that AD and other mechanical signals associated with brood care (such as AW in P. dominula) bias the development of recipient larvae toward worker-like traits. Arguing that different forms of mechanical signals that occur in the same context of larval feeding have similar functions, Jeanne (2009) drew on two key observations. First, females begin performing AW or AD on a nest only with the appearance of larvae at the third (of 5) instar (Brillet et al., 1999; Suryanarayanan and Jeanne, 2008). In Apis, Vespa, and Vespula the third instar is the stage at which the development of larvae begins to diverge between queen- or worker-destined trajectories (Ishay, 1975; Evans and Wheeler, 1999). This pattern is also consistent with the stimulatory and inhibitory hypotheses on larval saliva release, since the third instar is the earliest stage to produce measurable amounts of larval saliva (Suryanarayanan and Jeanne, 2008). Second, AW rates in P. dominula are high early in the colony cycle when worker larvae are being raised and low later in the colony cycle when gyne (potential queen) larvae are being raised (Brillet et al., 1999), a pattern that is not consistent with either the stimulatory or the inhibitory hypotheses on release of larval saliva. Whether this is also the pattern for AD in P. fuscatus has not yet been demonstrated. Here we document temporal changes in the AD behavior of P. fuscatus across its colony cycle, specifically in its rates of occurrence and in the mean duration of a burst of drumming. We also correlate changes in AD behavior to larval feeding, presence of specific larval instars, the social status of performing females, and environmental variables (temperature and humidity). In doing so, we test specific predictions made by the salivary inhibition and mechanical switch hypotheses. The salivary inhibition hypothesis proposes that AD is an essential signal for feeding liquid to larvae. It predicts, first, that the
Antennal drumming across the colony cycle in Polistes fuscatus
overall rates of AD performance should correlate closely with rates of feeding liquid to larvae. Second, since workers as well as queens feed liquid to larvae, it predicts that queens and workers would be expected to drum at rates that correlate with their rates of feeding liquid to larvae. On the other hand, the mechanical switch hypothesis predicts that the frequency of AD performance will be high during the development of worker-destined larvae (earlier in the colony cycle) and low during the development of gyne-destined larvae (later in the colony cycle). Since larvae that become workers are primarily foundress-fed (Hunt et al., 2007), foundresses are predicted to drum at high rates. Conversely, since larvae that become gynes are largely worker-fed (Hunt et al., 2007), workers are predicted to drum at low rates. If AD has little to do with the mechanics of adult-larva trophallaxis, then its rate of performance is expected to be independent of the rates at which larvae are fed liquid. Further, the greater the mean AD burst duration, the greater the probability of biasing an exposed larva toward worker traits. Therefore, AD burst durations are predicted to decrease across the colony cycle as well. We tested these predictions by recording the occurrences of AD behavior, its actors, and correlating them with developmental stages of colonies in the field.
Methods Colony observations A total of 31 colonies nesting under eaves of buildings at the University of Wisconsin Arboretum in Madison (43.1°N, 89.4°W) were observed in situ between May 4 and June 22, 2005, and between May 24 and August 25, 2006. Weather-related disruption of some observations and a high rate of colony failure due to queen disappearance and brood parasitism (Reeve, 1991) meant that not all colonies survived through all their developmental stages. Since our intent was to analyze brood-care behavior patterns as a function of stage of colony development, we used a crosssectional instead of a longitudinal analysis. That is, we determined the levels of AD, feed-liquid events, and burst duration at each of several developmental reference points in the colony cycle. Thus, the pre-emergence stage was divided into five sub-stages, as defined by the oldest immature stage present in the nest (egg, larval instars 1 or 2, instar 3, instars 4 or 5, pupa) (Suryanarayanan and Jeanne, 2008). Observations during the post-emergence stage of the colony cycle were binned into weeks, starting with the emergence of the first worker. Since not all colonies reached each sub-stage synchronously, these reference points do not align precisely by date.
Observations were randomized and uniformly distributed such that a colony was observed between 2 and 6 h each observation week between 0900 and 1600 h, for a total of 401 h over all colonies. Colony observations specifically noted the number of ‘feed-liquid’ events and ‘AD bursts’ performed by each individual wasp. A ‘feed-liquid’ event was a single contact between an adult wasp and a larva, during which the adult brushed the sides of the larva’s body with the tips of her antennae and regurgitated a droplet while slowly moving her head into the cell to bring the droplet into contact with the larva’s mouthparts. The mouth parts of the adult remained still during this behavior, and an inward telescoping of the gaster was sometimes seen. Only larval contacts with these distinguishing characteristics were recorded as feed-liquid events (Suryanarayanan and Jeanne, 2008). An ‘AD burst’ was recorded whenever a wasp positioned her head over the opening of a cell and hit the rim of that cell with a series of antennal strikes in a synchronous manner. In 2005, 20 colonies (19 single-foundress, 1 multiplefoundress) were followed during the period spanning nest initiation until the first week of worker emergence (week 0), for a total of 111 h. Queens on all colonies were marked on their thoraces with Testors paint (Rockford IL, USA). A brood census of each colony was carried out once per week. We recorded the number of cell visits, feed-liquid events, and AD bursts performed by each individual on each nest. Increasing colony size after the first week of worker emergence (June 22) meant that multiple individuals were involved in brood care, which made it unfeasible to continue this mode of manual data recording. In 2006, 11 colonies (8 single-foundress, 3 multiplefoundress) were followed during the period spanning the first appearance of fourth- and fifth-instar larvae to 8 weeks after worker emergence, for a total of 290 h. Only 3 (1 singlefoundress, 2 multiple-foundress) out of the 11 sampled colonies survived and completed the entire colony cycle. All observations were recorded on videotape (Sony TRV950 digital and Canon ES series Hi8 camcorders). Minute-tominute temperature and relative humidity measurements were taken daily with a remote temperature and humidity recording device (EasyLog EL-USB-2, Lascar Electronics, Erie PA, USA), which was placed under the eave of a building being used as a P. fuscatus nesting site. All adults, including newly emerging ones on all colonies, were marked on their thoraces and/or abdomens with Testors paint. Censuses of adults and brood were carried out at night twice per week uniformly on all colonies in both pre- and post-emergence stages. To facilitate the gathering of accurate adult and larval instar data from large postemergence stage colonies, we glued each nest during its preemergence phase onto a 6 9 6 9 1 cm3 plywood square that was hung by copper wires onto hooks at the original site
S. Suryanarayanan et al.
of the nest. To increase the probability that all adults (including foragers) were present on a nest, we performed inspections after sunset between 2100 and 0400 h as follows. (1) Each plywood piece with nest, brood, and adults was covered rapidly with a deep, wide-mouthed porous plastic cup that prevented any adults from escaping. (2) The plywood piece with cup in place was unhooked, wrapped in a plastic bag and placed on ice at 4°C for 20–30 min. (3) Chilled adults were removed, marked, and placed in glass vials. (4) The number of larvae of each instar stage and their nest cell locations were recorded on brood maps (Grechka and Kipyatkov, 1984). (5) The plywood piece with nest and brood was re-hooked to its original location, following which the marked adults were transferred from their glass vials back onto their natal nest. Video analysis We extracted the following data from the videotapes: (1)
(2) (3) (4)
Whenever a wasp left a nest, it was noted as a foraging trip. The times (h: min: sec) of departure and return were noted. This allowed us to calculate per-adult rates of occurrences of AD bursts and feed-liquid events per hour of time spent on the nest by each adult during an observation session. Number of feed-liquid events per adult. Number of AD bursts per adult. Using frame-by-frame analysis (30 frames per sec) in the digital 1 h videotapes, the duration of an AD burst was measured by recording the start and end time in video frames of the AD burst, only in cases where the head of a drumming wasp was clearly visible on the monitor.
Data analyses
(5)
following the emergence of workers, we did not attempt to analyze differences among foundresses of different dominance rank. In other words, subordinates were left out of this analysis. Only active workers that performed brood provisioning were included in our analysis of worker rates. Differences between queens and workers in their mean AD burst durations (only for 2006 data).
All rates of occurrences were calculated per hour per adult. Using this measure allowed us to independently assess the effects of changes in the numbers of particular larval instars on AD and feed-liquid rates. We used a mixed model analysis of variance to test the fixed effects of various factors and their interactions on rates of AD and feed-liquid events (Proc Mixed in the SAS statistical language). A mixed-model analysis of variance accounted for colonylevel variation by modeling colony as a random effect. This allowed the error degrees of freedom to be based more precisely on a ‘colony error’ rather than a ‘wasp error’ term. AD and feed-liquid rate data were square-root transformed and AD burst duration data were log10 transformed to meet normality assumptions. We fitted the response (AD rate or feed-liquid rate) to all the relevant sets of factors together with their interactions. Since this showed that interactions were non-significant, we employed backward elimination and included all the relevant factors without their interactions (Table 1). Multiple comparisons were controlled at an experiment-wise error rate of 5% using the Tukey–Kramer method. The effects of each factor and interaction were assessed at the 95% significance level. Table 1 A mixed model ANOVA of the effects of various factors on colony-wide DF
Type III F-ratio
p-value
\0.0001*
(A) Colony-wide AD rates 12
9.6
We analyzed the following for both single- and multiplefoundress colonies at each developmental stage:
Colony stage Feed-liquid rate
1
0.5
Adult-to-larva ratio
1
0.1
0.75
(1)
No. of larvae
1
0.9
0.35
12 1
1.8 2.1
0.08 0.15
1
0.01
0.92
1
6.0
0.02*
(2)
(3)
(4)
Colony-wide rates of AD performance as a function of colony stage, feed-liquid rates, adult-to-larva ratio, and the number of larvae. Colony-wide rates of feed-liquid events as a function of colony stage, AD rates, adult-to-larva ratio, and the number of larvae. Colony-wide rates of AD and feed-liquid events as a function of the ambient temperature and relative humidity (available only for 2006 data). Differences between queens (egg-layers) and workers in their mean rates of performance of AD and feedliquids to larvae. Since subordinate foundresses were on only 4 of the 31 colonies and also tended to disappear
Nest
30
Error
45
0.48
(B) Colony-wide feed-liquid rates Colony stage AD rate Adult-to-larva ratio No. of larvae Nest
30
Error
45
(A) AD rates and (B) Larval feed-liquid rates. DF denominator degrees of freedom *P \ 0.05
Antennal drumming across the colony cycle in Polistes fuscatus
Results Seasonal patterns of colony-wide AD and feed-liquid rates (mean ± S.E.) AD began on a colony with the appearance of third-instar larvae (Fig. 1). Its overall rate, in bursts per hour, subsequently increased, plateaued, and then decreased dramatically after the third week of worker emergence (31 colonies, 14 colony stages; Fig. 1). Larvae in pre-emergence stage colonies (from egg-hatch to first worker emergence) experienced much higher AD rates (17.1 ± 1.7 bursts per adult per hour; 26 colonies) than did those in post-emergence stages (8.9 ± 1.2 bursts per adult per hour; 12 colonies; P \ 0.001). AD rates dropped markedly after the third week following worker emergence (3.0 ± 0.7 bursts per adult per hour; 7 colonies) in comparison to all previous colony stages that showed AD behavior (from ‘larval instar 3’ through the third postemergence week) (16.1 ± 1.2 bursts per adult per hour; 27 colonies; P \ 0.0001). In contrast, the rates at which larvae were fed liquid did not differ between pre-emergence (29.8 ± 3.3 feeds per adult per hour; 26 colonies) and post-emergence stages (28.8 ± 2.8 feeds per adult per hour; 12 colonies; P = 0.47). But feed-liquid rates after the third week (30.8 ± 4.9 feeds per adult per hour; 7 colonies) remained at levels similar to those in prior colony stages (28.7 ± 2.3 feeds per adult per hour; 27 colonies; P = 0.25).
Fig. 1 Colony-wide AD and larval feed-liquid rates as functions of colony stage. Colonies are aligned on the x-axis at the point of the emergence of the first adult worker (‘zero’ on the x-axis). Prior to that event, data are binned by the oldest developmental stage of immatures present on the nest. Subsequent to that event, data are binned by weeks following worker emergence. Black bars are mean AD rates in bursts per hour. White bars are mean feed-liquid rates to larvae. Filled circles indicate the mean number of larval instars. Numbers above bars indicate sample sizes (colonies). Error bars are standard error
After eggs hatched into larvae, the variation in AD rate was best explained as a response to colony stage (13 stages, 31 colonies, F12,45 = 9.56, P \ 0.0001). In contrast, it was independent of the influence of feed-liquid rates, adult-tolarva ratios, the total number of larvae, number of thirdinstar larvae (P = 0.98), and the number of third to fifth instar larvae (P = 0.18) in the nest (Table 1A). On the other hand, the variation in feed-liquid rates was best explained as a response to changes in the total number of larvae (13 stages, 31 colonies, F1,45 = 5.9, P = 0.019) and was not significantly affected by colony stage, AD rates, or adult-tolarva ratio (Table 1B). Neither AD rates (52 observations, 11 colonies) nor feedliquid rates (52 observations, 11 colonies) were affected by seasonal changes in ambient temperature (AD: F1,40 = 0.56, P = 0.46; feed-liquids: F1,40 = 0.43, P = 0.52) or relative humidity (AD: F1,40 = 3.8, P = 0.06; feed-liquids: F1,40 = 3.6, P = 0.06) between late May and late August 2006. Differences between single- and multi-foundress colonies Colony-wide AD rates of single-foundress (SF) colonies (10.8 ± 1.2 bursts per adult per hour, N = 27 colonies) did not differ from those of multiple-foundress (MF) colonies (11.0 ± 2.1 bursts per adult per hour, N = 4 colonies) [F1,29 = 0.19, P = 0.67]. Both SF and MF colonies showed a similar pattern of AD rates across the colony cycle. In preemergence stages (from larval instar 3 through pupal substage), colony-wide AD rates were relatively high in both SF (16.6 ± 1.7 bursts per adult per hour) and MF (16.8 ± 5.5 bursts per adult per hour) colonies (P [ 0.05). In the post-emergence stages colony-wide AD rates were relatively low in both SF (9.0 ± 1.6 bursts per adult per hour) and MF (8.9 ± 1.9 bursts per adult per hour) colonies (P [ 0.05). However, colony-wide feed-liquid rates of SF colonies (23.1 ± 2 feeds per adult per hour, N = 27) were lower than those of MF colonies (35.1 ± 4.6 feeds per adult per hour, N = 4 colonies) [F1,29 = 6.57, P = 0.02]. This difference in feed-liquid rates was reflected in the lower total number of larvae that were reared in SF colonies (13.7 ± 0.9) compared to MF colonies (20.7 ± 1.8) (F1,28 = 13.4, P \ 0.001). Caste differences in rates of AD and feed-liquid events (mean ± S.E.) Workers performed AD at negligible rates (mean = 0.87 ± 0.21 bursts per adult per hour; 10 colonies, 41 workers) compared to queens (mean = 11 ± 1.2 bursts per adult per hour, 31 colonies, 34 queens; F1,9 = 33.6, P \ 0.0003;
S. Suryanarayanan et al.
Fig. 2). But this overall divergence in the performance of AD rates masked important variations based on colony stage. Queen AD rates varied significantly by colony stage (14 stages, F13,58 = 24.1, P \ 0.0001), while per-worker AD rates stayed relatively low throughout post-emergence (9 stages, F8,25 = 1.3, P = 0.28). Queens drummed at dramatically lower rates after the third week of post-emergence (5 stages, 6 queens, 2.1 ± 0.63 bursts per adult per hour) compared to previous colony stages that showed AD behavior (7 stages, 25 queens, 18.2 ± 1.4 bursts per adult per hour, P \ 0.0001). In fact, per-queen AD rates dropped to such an extent after the third week of post-emergence that they did not differ from either per worker AD rates (P = 0.07) or the combined AD rates of workers (3.3 ± 0.7 bursts per adult per hour, P = 0.32). Similarly, an average worker fed liquid to larvae at much lower overall rates than did an average queen (F1,9 = 23.6, P \ 0.0009; Fig. 3). But per-queen feed-liquid rates dropped following the third week of post-emergence to levels (5 stages, 6 queens, 10.5 ± 2.4 feed-liquids per adult per hour) that were not different from per-worker feed-liquid rates (9 stages, 41 workers, 5.6 ± 0.71 feeds per adult per hour, P = 0.06). Even though per-worker feed-liquid rates were low, the combined feed-liquid rates of workers during postemergence (41 workers, 26.1 ± 4.0 feeds per hour) were actually as high as the levels of feed-liquids performed by queens during pre-emergence (26 queens, 25.3 ± 2.7 feeds per adult per hour, P = 0.24). In summary, a queen’s AD rate was positively correlated with the rate at which she fed liquid to larvae (F1,57 = 4.6, P = 0.036). It was also a function of colony
Fig. 3 Variation among adults of queen and worker castes in mean feed-liquid rates. Rates were calculated as number of feed-liquid events per hour of presence on the nest. Closed circles are mean values per queen. Open circles are mean values per worker. Open inverted triangles are mean combined values of active workers from each sampled colony. Numbers near circles indicate the sample sizes (queens or workers). Error bars are standard error
stage (F13,57 = 13.1, P \ 0.0001). But worker AD rates, in contrast, were independent of both colony stage (F8,25 = 1.38, P = 0.28) and feed-liquid rates (F1,30 = 2.8, P = 0.1). Caste differences in AD burst durations (mean ± S.E.) Over the season as a whole, mean burst duration did not differ between queens (0.93 ± 0.5 s, 14 queens, 11 colonies) and workers (0.69 ± 0.03 s, 22 workers, 11 colonies) (F1,6 = 4.9, P = 0.07; Fig. 4). However, the two castes did differ significantly as a function of colony stage (11 colonies, 9 stages, F15,34 = 4.9, P \ 0.0001), influencing burst durations of queens (11 colonies, 9 stages, F8,20 = 5.6, P \ 0.0009), but not workers (7 colonies, 7 stages, F6,8 = 2.7, P = 0.09). In particular, mean burst durations for queens were significantly longer earlier in the colony cycle in the stages spanning the appearance of fourth- and fifthinstars in pre-emergence to the first 2 weeks of post-emergence (1.2 ± 0.04 s, 4 stages, 8 colonies, 8 queens) compared to subsequent stages (0.8 ± 0.03 s, 5 stages, 11 colonies, 14 queens, P \ 0.0001).
Discussion Fig. 2 Variation among adults of queen and worker castes in mean AD rates. Rates were calculated as number of AD bursts per hour of presence on the nest. Closed circles are mean values per queen. Open circles are mean values per worker. Open inverted triangles are mean combined values of active workers from each sampled colony. Numbers near circles indicate the sample sizes (queens or workers). Error bars are standard error
Our results do not support the salivary inhibition hypothesis for P. fuscatus. If AD were an essential signal to facilitate liquid feeding to larvae, we would expect colony-wide AD rates to be tightly correlated with liquid-feeding rates and the number of third to fifth instar larvae. However, AD rates
Antennal drumming across the colony cycle in Polistes fuscatus
Fig. 4 Variation among adults of queen and worker castes in mean AD burst durations. Closed circles are mean values per queen. Open circles are mean values per worker. Numbers near circles indicate the sample sizes (queens or workers). Error bars are standard error
in our study were not dependent on the feed-liquid rates and number of third to fifth larval instars, but rather solely on colony stage. Feed-liquid rates varied as a function of the number of larval instars present on a nest. Moreover, workers seldom performed AD even though they fed larvae at high rates comparable to those of queens during preemergence. In contrast, three findings from our study are predicted by the mechanical switch hypothesis. First, AD rates were highest early in the colony cycle, in between the stages that spanned pre-emergence and early post-emergence. Second, AD rates decreased sharply following the third week of worker emergence, whereas larval feeding rates remained high. Third, the duration of an average burst of AD, a measure of the intensity with which a wasp drums, became much shorter as the colony cycle progressed. In conjunction with a larger nest mass and size (Reeve, 1991) that could dampen AD vibrations, lowered AD rates and burst durations later in the colony cycle created a colony environment in which larval exposure to AD was greatly diminished. Our results showed that the marked decrease in the colony-wide AD rate was not a result of extrinsic factors such as seasonal changes in ambient temperature or humidity. Instead, it was likely due to a combination of changes that occurred in either the colony structure following worker emergence and/or the age of the queen. Feeding of larvae during the period after the third week of post-emergence shifted from being primarily a queen task to being carried out largely by workers, as evidenced by the lowered rates of feeding by queens and much higher combined rates of feeding by workers. Since workers seldom drummed while feeding larvae, feed-liquid rates stayed high but colonywide AD rates dropped. This shift could have been due to an
age-related transition in worker tasks, because older P. fuscatus workers are more likely to forage for prey and feed larvae than are younger workers (Jeanne, 1991), thus diminishing the likelihood that queens received prey loads and fed larvae. If this hypothesis were correct, it would predict that selective removal of older workers would cause an increase in the colony-wide AD rate due to an increased rate of larval feeding by the queen. However, if queen AD and feeding rates were to still remain low following this manipulation, an alternative explanation for the diminished rates of AD could be the queen’s increasing senescence (Jeanne, 1980; Dew, 1983). While it remains to be investigated why exactly colony AD rates diminished, the decrease in the queen’s AD burst duration could be explained, perhaps less ambiguously, as an effect of queen senescence. In support of this notion, we sometimes observed orphaned late-season nests where a worker that became dominant performed AD bursts at higher intensity than the missing queen. Colony-wide temporal patterns of AD behavior did not differ between single-foundress and multiple-foundress colonies. Neither did we observe differences in the form or rate of performance of AD produced by females on either kind of colony. These indicate that the presence of multiple adults does not have any measurable influence on changes in AD behavior. Our results thus support Brillet et al.’s (1999) interpretations regarding abdominal wagging associated with brood care in P. dominula. Furthermore, AD rates remained similar despite differences in the rates at which larvae were fed. Our findings suggest that the comparatively higher rate of feeding to larvae, which we saw in multiplefoundress colonies, is related to their raising a greater number of larvae. Our comparative finding on the number of larvae raised is further corroborated by the knowledge that multiple-foundress colonies have greater productivity compared to single-foundress colonies (Tibbetts and Reeve, 2003), although they have the same patterns of social organization and colony cycle (Reeve, 1991). AD and larval provisioning behavior were strongly associated in queens but not in workers. The correlation between dominance and vibratory behaviors in the context of brood feeding has been noted in other polistine species as well (Chandrashekara and Gadagkar, 1991; Brillet et al., 1999). Ovariectomized P. dominula foundresses continued to perform AW, suggesting that mechanical signals in the context of larval feeding are correlates of social dominance, but not reproductive dominance (Ro¨seler and Ro¨seler, 1989). High corpora allata (CA) activity and juvenile hormone (JH) levels can induce adults to become socially dominant (Ro¨seler et al., 1985). These cross-species comparisons suggest that high rates of AD behavior in Polistes queens are due to their high JH levels. Such a hypothesis predicts that allatectomized (CA removed) Polistes queens
S. Suryanarayanan et al.
on pre-emergence stage colonies with brood will perform AD at negligible rates compared to sham allatectomized controls (Sullivan et al., 2000). Based on the following lines of indirect evidence we argue that the AD of queens and workers are identical in their function: (1) Behavioral observations indicate that the context in which AD occurs is the same for both queens and workers in P. fuscatus, P. carnifex and P. instabilis (S. Suryanarayanan, unpublished). (2) Mean AD burst duration, a measure of the structure of AD, is not different between P. fuscatus queens and workers. The same is true for P. carnifex and P. instabilis as well (S. Suryanarayanan, unpublished). Another measure of the structure of AD, the number of antennal strokes per second, obtained from digitized sound recordings of AD in P. carnifex and P. instabilis, is also not different between queens and workers (S. Suryanarayanan, unpublished). Since AD in P. fuscatus has the same form and context as in P. carnifex and P. instabilis (Pratte and Jeanne, 1984), we believe that this is the case in the closely related P. fuscatus as well. Although it is possible that AD performed by workers has a different effect than AD performed by queens, this seems unlikely. While the findings of this study on AD are inconsistent with the salivary inhibition hypothesis, a previous study (Suryanarayanan and Jeanne, 2008) showed lack of support for the stimulatory hypothesis. Suryanarayanan and Jeanne (2008) used radiolabeled prey juice to show that queens did not receive saliva from larvae during or after a bout of AD bursts. Instead, our results support the hypothesis that AD plays a key role in biasing larval development toward a worker phenotype (Jeanne, 2009). The downward shift in AD rate and burst duration during the colony cycle coincided with a shift from queens to workers in the rearing of brood, providing indirect support for the ‘mechanical switch hypothesis’ on AD function. Queen-reared larvae, which received high AD vibrations, typically emerge as workers (Reeve, 1991). In contrast, worker-reared larvae, which received low AD vibrations, tend to emerge as gynes (and males) (Reeve, 1991). AD vibrations could bias the caste of developing larvae by modulating the developmental diapause pathways that are thought to underlie the gyne phenotype in Polistes wasps (Hunt et al., 2007). Artificial vibrational stress has been shown to induce changes in the growth and development of larvae in other insects (Hirashima et al., 1992, 1993). Thus, AD could be acting as a mechanical stressor that the egg-layer employs to manipulate the development of her offspring (Alexander, 1974). Alternatively, AD could be a fertility signal that informs the larvae of the presence of a viable egg-layer and cuing them to take the developmental trajectory that is likely to maximize their inclusive fitness in that context (Liebig et al., 2005). The rate at which larvae are fed is currently considered to be the means by which Polistes wasps bias caste in larvae,
with a higher rate leading to their development as gynes (Hunt and Amdam, 2005). Increased feeding has been linked to an increase in the number of adults per larva on a nest as the colony develops (Grechka and Kipyatkov, 1984). But in our study, after accounting for variation due to the number of larvae on a nest, adult-to-larva ratio did not significantly influence colony-wide rates of feeding to larvae. Instead, the adult-to-larva ratio could affect the total amount of protein arriving at a colony (Suzuki, 1981). Moreover, queen-raised larvae in pre-emergence nests were fed at rates comparable to worker-raised larvae in postemergence nests, a fact that is further supported by their similar development times (Klahn, 1981). This suggests that differential larval feeding rate alone cannot explain the biasing of castes (West-Eberhard, 1969). We postulate that AD is the required additional environmental input. Our findings are consistent with patterns in the contexts, actors, and temporal rates of performance of AW associated with larval feeding in P. dominula (Brillet et al., 1999). The close similarity between AD and AW strengthen the notion that they have the same function (Jeanne, 2009). Indeed, it was Brillet et al. (1999) who first speculated that AW vibrations in the context of larval feeding by P. dominula might influence the future caste of a developing larva. Clearly, this proposed role of AD and AW deserves to be more directly investigated. Our results do not support predictions made by the salivary inhibition hypothesis, but suggest indirect support for the mechanical switch hypothesis on AD function. This field-based study complements and builds upon insights into AD behavior gained from recent experimental studies (Suryanarayanan and Jeanne, 2008) and cross-species comparisons (Jeanne, 2009). Further, this study points to the need for similar detailed analyses in other paper wasp taxa such as Mischocyttarus, Ropalidia, and Belonogaster, which also exhibit bizarre and diverse mechanical signals in the context of brood feeding (Jeanne, 2009). Finding whether or not the colony-wide patterns that Brillet et al. (1999) and we report in Polistes sp. also hold for species in other polistine genera could shed light on the evolution of mechanical signals and their role in pre-adult caste determination (Jeanne, 2009). More broadly, our findings expand the scope of known roles played by mechanical signals. Most of the mechanical signals reported to date for eusocial insects in the context of caste control and reproductive conflict function as shortterm, behavioral modulators. For example, mechanical signals influence the success of reproductives (Tarpy et al., 2004; Schneider and DeGrandi-Hoffman, 2008; Boucher and Schneider, 2009) and of egg-laying workers in the honey bee (A. mellifera) (Schneider and McNally, 1991). They modulate queen behavior during colony reproduction and relocation, both in the honey bee (Pierce et al., 2007) and in
Antennal drumming across the colony cycle in Polistes fuscatus
termite-hunting ants (Ho¨lldobler et al., 1996). Repetitive mechanical signals may also modulate the levels of taskrelated activity among A. mellifera recipients commensurate with different stages of colony development (Cao et al. 2009). However, the longer-term developmental effects of mechanical signals on caste, such as we suggest here, are largely unknown (but see Brian, 1973). Mechanical signals transmitted via the substrate are believed to be an ancient channel of communication (Hill, 2008). We suggest that their role as modulators of caste may also have a long evolutionary history and may be more widespread than currently recognized. Acknowledgments The authors wish to gratefully acknowledge the University of Wisconsin Arboretum and its staff for providing key facilities for our field work. Undergraduate research assistants Jessica Bartlett, Theresa Cira, Gerald Elliott, Alex Heeren, Julia Hoeh, Victoria Konicek, Vanessa Machen, Jeff Rolling, Matthew Ruebl, Shail Shah, Adam Stone, Jeremy Tupper, Wa Vang and Meghan Wanzek performed valuable recordings and analyses of videotapes. Research supported by the College of Agricultural and Life Sciences, University of Wisconsin, and by a University of Wisconsin Department of Zoology research grant to SS.
References Alexander R.D. 1974. The evolution of social behavior. Annu. Rev. Ecol. Syst. 5: 325-383 Brennan B.J. 2007. Abdominal wagging in the social paper wasp Polistes dominulus: Behavior and substrate vibrations. Ethology 113: 692-702 Brian M.V. 1973. Caste control through worker attack in the ant Myrmica. Insect. Soc. 20: 87-102 Brillet C., Tian Chansky S.S. and Le Conte Y. 1999. Abdominal waggings and variation of their rate of occurrence in the social wasp, Polistes dominulus Christ. I. Quantitative analysis. J. Insect Behav. 12: 665-686 Boucher M. and Schneider S.S. 2009. Communication signals used in worker-drone interactions in the honey bee, Apis mellifera. Anim. Behav. 78: 247-254 Cao T.T., Hyland K.M., Malechuk A., Lewis L.A. and Schneider S.S. 2009. The effects of repeated vibration signals on worker behavior in established and newly founded colonies of the honey bee, Apis mellifera. Behav. Ecol. Sociobiol. 63: 521-529 Chandrashekara K. and Gadagkar R. 1991. Behavioral castes, dominance, and division of labor in a primitively eusocial wasp. Ethology 87: 269-283 Corn M.L. 1972. Notes on the biology of Polistes carnifex (Hymenoptera: Vespidae) in Costa Rica and Colombia. Psyche 79: 150157 Dew H.E. 1983. Division of labor and queen influence in laboratory colonies of Polistes metricus (Hymenoptera: Vespidae). Z. Tierpsychol. 61: 127-140 Evans J.D. and Wheeler D. 1999. Differential gene expression between developing queens and workers in the honey bee, Apis mellifera. Proc. Natl Acad. Sci. USA 96: 5575-5580 Grechka E.O. and Kipyatkov V.E. 1984. The seasonal cycle and caste determination in Polistes gallicus (Hymenoptera: Vespidae) 2. Dynamics of growth and productivity of colonies. Zool. Zh. 63: 81-94
Haggard C.M. and Gamboa G.J. 1980. Seasonal variation in body size and reproductive condition of a paper wasp Polistes metricus (Hymenoptera Vespidae). Can. Entomol. 112: 239-248 Hill P.S.M. 2008. Vibrational Communication in Animals. Harvard University Press. Cambridge, MA. ix ? 261 p Hirashima A., Uenoi R. and Eto M. 1992. Effects of various stressors of larval growth and whole-body octopamine levels of Tribolium castaneum. Pestic. Biochem. Physiol. 44: 217-225 Hirashima A., Nagano T. and Eto M. 1993. Stress-induced changes in the biogenic amine levels and larval growth of Tribolium castaneum Herbst. Biosci. Biotechnol. Biochem. 57: 2085-2089 Ho¨lldobler B., Janssen E., Bestmann H.J., Leal I.R., Oliveira P.S., Kern F. and Ko¨nig W.A. 1996. Communication in the migratory termite-hunting ant Pachycondyla (= Termitopone) marginata (Formicidae: Ponerinae). J. Comp. Physiol. A 178: 47-53 Hunt J.H., Baker I. and Baker H.G. 1982. Similarity of amino acids in nectar and larval saliva: The nutritional basis for trophallaxis in social wasps. Evolution 36: 1318-1322 Hunt J.H. and Amdam G.V. 2005. Bivoltinism as an antecedent to eusociality in the paper wasp genus Polistes. Science 308: 264267 Hunt J.H., Kensinger B.J., Kossuth J.A., Henshaw M.T., Norberg K., Wolschin F. and Amdam G.V. 2007. A diapause pathway underlies the gyne phenotype in Polistes wasps, revealing an evolutionary route to caste-containing insect societies. Proc. Natl Acad. Sci. USA 104: 14020-14025 Ishay J. 1975. Caste determination by social wasps cell size and building behavior. Anim. Behav. 23: 425-431 Jeanne R.L. 1980. Evolution of social behavior in the Vespidae. Annu. Rev. Entomol. 25: 371-96 Jeanne R.L. 1991. Polyethism. In: The Social Biology of Wasps (Ross K.G. and Matthews R.W., Eds), Comstock, Ithaca. pp 389–425 Jeanne R.L. 2009. Vibrational signals in social wasps: A role in caste determination? In: Organization of Insect Societies (Gadau J. and Fewell J., Eds), Harvard University Press, Cambridge. pp 241263 Karsai I. and Hunt J.H. 2002. Food quantity affects traits of offspring in the paper wasp Polistes metricus (Hymenoptera: Vespidae). Env. Entomol. 31: 99-106 Klahn J.E. 1981. Alternative reproductive tactics of single foundresses of a social wasp, Polistes fuscatus. PhD Dissertation, University of Iowa, Iowa City Liebig J., Monnin T. and Turillazzi S. 2005. Direct assessment of queen quality and lack of worker suppression in a paper wasp. Proc. R. Soc. B 272: 1339-1344 Miyano S. 1983. Number of offspring and seasonal changes of their body weight in a paperwasp, Polistes chinensis antennalis Pe´rez (Hymenoptera: Vespidae), with reference to male production by workers. Res. Popul. Ecol. 25: 198-209 Morimoto R. 1960. Experimental study on the trophallactic behavior in Polistes (Hymenoptera, Vespidae). Acta Hymenopt. 1: 99-103 O’Donnell S. 1998. Dominance and polyethism in the eusocial wasp Mischocyttarus mastigophorus (Hymenoptera: Vespidae). Behav. Ecol. Sociobiol. 48: 327-331 Owen J. 1962. The behavior of a social wasp Polistes fuscatus (Vespidae) at the nest, with special reference to differences between individuals. Ph.D. dissertation, University of Michigan, Ann Arbor, Michigan Pierce A.L., Lewis L.A. and Schneider S.S. 2007. The use of the vibration signal and worker piping to influence queen behavior during swarming in honey bees, Apis mellifera. Ethology 113: 267-275 Pratte M. and Jeanne R.L. 1984. Antennal drumming behavior in Polistes wasps. Z. Tierpsychol. 66: 177-188 Rau P. 1928. Trophallaxis in Polistes pallipes. Psyche 35: 153-156
S. Suryanarayanan et al. Reeve H.K. 1991. Polistes. In: The Social Biology of Wasps (Ross K.G. and Matthews R.W., Eds), Cornell University Press, Ithaca. pp 99-148 Ro¨seler P.F., Ro¨seler I. and Strambi A. 1985. Role of ovaries and ecdysteroids in dominance hierarchy establishment among foundresses of the primitively social wasp, Polistes gallicus. Behav. Ecol. Sociobiol. 18: 9-13 Ro¨seler P.F. and Ro¨seler I. 1989. Dominance of ovariectomized foundresses of the paper wasp, Polistes gallicus. Insect. Soc. 36: 219-234 Schneider S.S. and DeGrandi-Hoffman G. 2008. Queen replacement in African and European honey bee colonies with and without afterswarms. Insect. Soc. 55: 79-85 Schneider S.S. and McNally L.C. 1991. The vibration dance behavior of queenless workers of the honey bee, Apis mellifera (Hymenoptera: Apidae), J. Insect Behav. 4: 319-322 Sullivan J.P., Jassim O., Fahrbach S.E. and Robinson G.E. 2000. Juvenile hormone paces behavioral development in the adult worker honey bee. Hormones and Behavior 37: 1-14
Suryanarayanan S. and Jeanne R.L. 2008. Antennal drumming, trophallaxis, and colony development in the social wasp Polistes fuscatus (Hymenoptera: Vespidae). Ethology 114: 1201-1209 Suzuki T. 1981. Flesh intake and production of offspring in colonies of Polistes chinensis antennalis (Hymenoptera, Vespidae). Kontyuˆ 49: 283-301 Tarpy D.R., Gilley D.C. and Seeley T.D. 2004. Levels of selection in a social insect: A review of conflict and cooperation during honey bee (Apis mellifera) queen replacement. Behav. Ecol. Sociobiol. 55: 513-523 Tibbetts E.A. and Reeve H.K. 2003. Benefits of foundress associations in the paper wasp Polistes dominulus: Increased productivity and survival, but no assurance of fitness returns. Behav. Ecol. 14: 510514 West-Eberhard M.J. 1969. The social biology of polistine wasps. Misc. Pub. Mus. Zool. Univ. Mich. 140: 1-101
