Serial polygyny in the primitively eusocial wasp Ropalidia marginata: implications for the evolution of sociality

Queen Number and Sociality in Insects

EDITED BY

LAURENT KELLER

Museum of Comparative Zoology, Harvard University, Cambridge, USA

and Museum of Zoology, Palais de Rumine, Lausanne, Switzerland

Oxford New York Tokyo

OXFORD UNIVERSITY PRESS

1993

9

Serial polygyny in the primitively eusocial wasp Ropalidia marginata: implications for the evolution of sociality

Raghavendra Gadagkar, Krishnappa Chandrashekara, Swarnalatha Chandran, and Seetha Bhagavan

Social insects usually live in colonies comprising one or a small number of reproductive individuals and a few or large number of sterile individuals.

In termites only, both sexes are represented among the reproductives as well as among the sterile workers. In other social insects, namely ants, bees, and wasps, males do not participate significantly in the social life of colonies, which involves primarily the fertile queens and sterile female workers (Wilson 1971). The haplodiploid genetic system found universally in the Hymenoptera creates an asymmetry in genetic relatedness such that a female is more closely related to her full sister (coefficient of genetic relatedness, r = 0.75) than to her offspring (r = 0.5). This makes inclusive fitness theory (Hamilton, 19640, b) particularly applicable to the evolution of sterile worker castes in the social Hymenoptera (Wilson 1971;

Hamilton 1972).

However, two phenomena tend to break down this genetic asymmetry and decrease the inclusive fitness that workers might have otherwise gained by rearing siblings instead of their own offspring (Hamilton 1964b).

One is polyandry, or multiple mating by queens. Hymenopteran queens can mate with several males, store sperm from all of them in their spermathecae, and use their sperm simultaneously, thus producing several patrilines of daughters who are half-sisters (r = 0.25). This can lead to considerable reduction in the relatedness between workers and their female siblings. Th~ implication of polyandry for the evolution of sociality has received considerable attention (see reviews in Starr 1984; Gadagkar 1985b;

Page 1986). The second is the presence of multiple queens or egg-layers within a colony (polygyny). Polygyny leads to different matrilines within a colony but its consequences for worker-brood genetic relatedness depend on the number of queens, relatedness among them, and relatedness between workers and queens (see Queller this volume).

Polygyny can be of two kinds. One involves the simultaneous presence of multiple egg layers in the colony while the second involves frequent

190 Raghavendra Gadagkar et a/.

analysis to study the effects of polyandry and pedigree analysis to study the effects of serial polygyny.

The primitively eusocial wasp Ropa/idia marginata

RopaJidia is a large Old World polistine genus with about 136 known i species occurring in Africa, Southeast Asia, Australia, and the Okinawa islands of Japan (Vecht 1967). This genus contains both relatively highly eusocial, swarm-founding, enveloped-nest building species with large colony :. sizes as well as primitively eusocial, independent-founding, open-nest

building species with relatively small colony sizes (Jeanne 1980; Gadagkar 1991a). RopaJidia marginata, a species that belongs to the latter group, is abundantly distributed in peninsular India. Colonies of R. marginata may be initiated throughout the year by one or a small group of. females (Gadagkar ec aJ. 1982). In multiple foundress nests, only one female becomes the egg layer while others assume the role of workers. Female offspring may either leave their natal nests to found new single- or multiple-foundress nests or they may stay and assume the role of workers (Gadagkar 1991a). R. marginata exhibits a typical polistine-like colony cycle with rapid growth in the pre-emergence and early post-emergence phases, followed by a declining phase after which, a small number of females may stay to start another colony cycle on the same nest. In this manner a series of colony cycles may. be encompassed within one nesting cycle (Gadagkar 1991a; Jeanne 1991; Chandrashekara et aJ. 1990). An important consequence of such an indeterminate colony cycle is that colonies often outlive their queens (Gadagkar et aJ. 1990). This provides opportunities for other adults of the colony to take over the role of queens. Because such change-over of queens may occur at various times in the colony cycle, workers who are offspring of one queen often rear brood that are offspring of another. Apart from such serial polygyny, R. -marginata exhibits strict monogyny at any given time. This is of course difficult to prove because it is based on negative evidence. But in well over 1000 hours of observation of dozens of colonies during the last 10 years, we have never documented .the simultaneous presence of two or more egg layers. Some but not all workers in R. marginata are inseminated. Even the queen may sometimes be uninseminated at the time of taking over the role of the queen although she may mate subsequently

(Chandrashekar a and Gadagkar 1991).

Data collection

Our studies reported in this chapter are based on four colonies of R.

marginata that were established by transplanting four post-emergence

~ colonies (including the nests and adults) into cages located at the Centre

I

L -

Serial polygyny in Ropalidia marginata 191

for Ecological Sciences, Indian Institute of Science, Bangalore. These cages were made of wire mesh which allowed wasps to forage freely outside but prevented the hornet Vespa tropica from attacking the brood.

Long-term studies of R. marginata are almost impossible without such protection from high rates of V. tropica predation. Colony TOI was collected from a site 7 km away on 14 September 1986 and was monitored for 606 days until it was abandoned on 17 May 1988. This colony had 33 Cggs, 25 larvae, 8 pupae, and 4 empty cells when it was collected and transplanted. Colony T02 was collected from a site 2 km away on 9 February 1987 and was monitored for 474 days until it was abandoned on 24 May 1988. This colony had 14 eggs, 15 larvae, 7 pupae, and 5 empty cells when it was collected and transplanted. Colony T08 was collected from a site 15 km away on 6 June 1987 and was monitored for 315 days till it was abandoned on 15 April 1988. It contained 43 eggs,

17 larvae, and 10 pupae when it was collected and transplanted. Colony Tll was initiated naturally inside one of our protected cages, by a group of females who had abandoned their transplanted nest on 5 August 1987, and was monitored for 272 days till it was abandoned on 2 May 1988.

Starting from the day of transplantation, we maintained a map of each nest, so that every cell was numbered and its contents noted at intervals of approximately one to two days. All adults present at the time of collection and transplantation, as well as those eclosing subsequently, were marked with unique spots of enamel paint. A census of all the animals present was made at night at intervals of approximately one to two days. The identity of the cell from which each animal eclosed was also noted. These colonies were under behavioural observation of differing intensities for other experiments. Consequently, the identity of the egg layer was known for all nests at all times. A change in the queen of any colony became known within a day or two of the event.

The system of serial polygyny in Ropalidia marginata

Figure 9.1 illustrates the indeterminate colony cycle of Ropalidia mar- ginata and shows that the timing of queen replacements bears no obvious relation to the phase of the colony cycle. Data in Table 9.1 give a glimpse of the system of serial polygyny in R. marginata. While the durations of our studies ranged from 280 to 606 days, the number of queens seen per colony ranged from 2 to 10. The mean tenure of queens was about 80 days but it varied enormously, ranging from 7 to 236 days.

The ages of queens also varied considerably, with values of 4 to 78 days at the beginning of their tenure and 37 to 262 days at the end of their tenure. The productivities of different queens were highly variable, as was the proportion of eggs laid by queens that successfully eclosed as adults. The maximum value for the latter was only 0.68 indicating that

192 Raghavendra Gadagkar et al.

T01 T02

JAN JUL JAN APR OCT APR

1987 1987 1988 1987 1987 1988

1 50 0

100 0

50

0 0

20 10

1 0 5

0 0

40

1 00 20

0 ~o

100 40

20

0 0

300 100

In 200 50 In

~ 100 ~

..0 0 0 ..0

E 30 20 E

:J :J

Z 20 10 Z

1 0

0 0

1 5 1 0 5

0 0

30

20 1 0

10

0 0

40 40

20 20

0 0

60 6040

40 20

20

0 0

JUL OCT JAN APR OCT JAN APR'

1987 1987 1988 1988 1987 1988 1988

TaB T11

Fig. 9.1 Growth and development of the four colonies used in this study. The survival of these colonies for long periods was associated with possible serial polygyny. Events of queen turnover are shown by vertical arrows.

~ZZZZZZZ»::iZO ;.":3 ° I: I: I: I: I: I: 1:0000 ~ I: I: ~:. 'O3333333nn-3"" ~=-, <:r <:r <:r <:r <:r <:r <:r ° ° § <:r~, It onnnnnnn""""'"nO o """""""""""""",O,Oo""=, ~ nOOOOOOOI:I:""oo

~

-

'"

~ ~

p~~~~~~~~~~~~ ~ O.

Serial polygyny in Ropalidia marginata 195

any social insect species. Note that for colonies TO8 and for Tll, the . relationship between queen 1 and queen 2 was unknown. This is because both queen 1 and queen 2 were present on the nest at the time of collection and transplantation. In the analysis, we first make the assump- tion that queen 2 was the daughter of queen 1 and then repeat the analysis assuming that queen 2 was the sister of queen 1. Worker-brood genetic relatedness falls to rather low values (see below) even with these assump- tions. Therefore, we did not repeat our analysis with assumptions of even more distant relatedness between queen 2 and queen I. The central message that emerges from Fig. 9.2 is, once again, that extensive variation exists among colonies. The simplest pedigree is that of colony TO1, where each queen is the daughter of the previous one, and at the other extreme is the rather complex pedigree in colony TO8. When the pedigrees of all four colonies are considered simultaneously, new queens are seen to be daughters, sisters, nieces, or cousins of their immediate predecessor queens (Fig. 9.2).

Genetic relatedness between workers and brood

The pedigrees in Fig. 9.2 lead to temporal and inter-colony variations in genetic relatedness between workers and brood. The relationships actually observed depend on the extent of overlap between adult offspring and brood of various queens. The extent of such overlap is shown in Fig. 9.3. When all colonies are considered, we find brood to be sisters, brothers, nieces and nephews, cousins, cousins' offspring, mother's cousins, Table 9.2 Genetic relationships between successive queens and between workers and brood observed in the four colonies.

Relationship between Relationship between workers and brooda queens and their

immediate predecessors

(a) Daughters (I) Sisters (0.75 or 0.53) (b) Sisters (2) Brothers (0.25)

(c) Nieces (3) Nieces and nephews (0.375 or 0.265) (d) Cousins (4) Cousins (0.1875 or 0.1325)

(5) Cousins' offspring (0.0938 or 0.0663) (6) Mother's cousins (0.0938 or 0.0663) .

(7) Mother's cousins' offspring (0.0469 or 0.0331) (8) Mother's cousins' grand-offspring (0.0234 or 0.0165)

a Brood were sisters, brothers, etc., of the workers. Values of relatedness given in parenthesis are for single

mating (r for sisters = 0.75 or multiple mating, average r for sisters = 0.53, based on electrophoretic data from Muralidharan et al. (1986) and Gadagkar (l990a); see the section on genetic relatedness between workers and brood).

196 Raghavendra Gadagkar et al.

T01 T02

JAN JUl JAN JUl OCT JAN APR

1987 1987 1988 1987 1987 1988 1988

2 00

f;:;:: ':::: :;;1 r

I/) 1 00 W, W. W3 W. W3 w~. 1 5

ti o

w, w. ~

~ 0 CI>

'- ~

0

~ ^~-~--!1 ~ ^{~'~ '-}

~ b. 0

-300 b ~ b "- 1 00 ~

"8 1 ..b ' -'.oJ "tJ

0 0 "" bs 0

b 0 e

30 .D

'0

~ ^{.,J:~ IW9 ~} W;<~/\J 2 0 -

.w6 0

15 "' w7

~ w3 W "' 10 In

..."' .6 '-

.D 0 ' W, w. CI>

E 0 .D

~ 7 5

Jt

t:

^{=;~=:;1 ~}

0 1 0

OCT JAN APR OCT JAN APR

1987 1988 1988 1987 1988 1988

T08 T11

Fig. 9.3 The extent of overlap between workers belonging to different matrilines (WI' W 2, etc.) and brood belonging to different matrilines (b I , b 2, etc.) in the four colonies under study.

mother's cousins' offspring, or mother's cousins' grand-offspring of the workers. These relationships, along with the appropriate coefficients of genetic relatedness, are given in Table 9.2. In computing the coefficients of genetic relatedness, we considered single mating by queens as well as the possibility of multiple mating. In the case of multiple mating, we used the average relatedness between sisters in a colony of 0.53, obtained from electrophoretic analysis of colonies with single matrilines (Muralidharan et at. 1986; Gadagkar 1990a). Thus, the coefficients of genetic relatedness betweerl workers and brood ranged from O. ~5 to 0.0165.

Variation in worker-brood genetic relatedness

For convenience, all quantitative data analysis was done at weekly inter- vals. From the nest map and census data, the proportion of workers belonging to each matriline amongst the total pool of workers present on any given day was computed. Similarly, the proportion of brood belonging to each matriline out of the total pool of brood present on any 'given day was calculated. The mean genetic relatedness between workers and brood for any given day was calculated as

n n

r = ~ ~Wibj rij (9.1)

i=lj=1

Serial polygyny in Ropaliala marginatE. 197 where, Wi is the proportion of workers that belong to the; th matriline,

hj is the proportion of brood that belongs to the j th matriline, r ij is the coefficient of genetic relatedness between the workers belonging to the

; th matriline and brood belonging to the j th matriline, and n is the number of matrilines. Since we had no way of sexing the brood, we computed two sets of worker-brood relatedness values assuming in the first case that all brood were females, and in the second case that all brood were males. Although Pamilo and Crozier (1982) have correctly argued that the appropriate directionality of relatedness is from bene- ficiary to altruist (in our case, from brood to workers), we shall, in keeping with common usage in the literature, refer to the conditional probability that an allele present in the workers is also present in the brood as the genetic relatedness of workers to brood or simply, worker-brood genetic relatedness.

Worker-brood genetic relatedness varies considerably over time in all four colonies (Figs. 9.4 and 9.5). As expected, th~ relatedness to female brood drops after queen replacement because workers who are rearing their sisters will now rear nieces, cousins, or other distantly related female brood. However, when queens are mated singly, a change in the queen may lead to an increase in relatedness between workers and male brood

TO1 TO2

JAN JUL JAN JUL OCT JAN APR

1987 1987 1988 1987 1987 1988 1988

VI 08 08 In

VI In

c,I c,I

c 0.6 0.6 c

u u

c,I c,I

iU 0.1, 0.1, §

-c,I~ '-

0.2 02

u .~

~

--

c,I c,I

01 08 0.8 01

u u

8 0.1, 0.1, 8

'- '-

.Do 0.D

I I

'- '-

c,I c,I

~ 0.1, 01, ~

~ 0 0 ~

OCT JAN APR OCT JAN APR

1987 1988 1988 1987 1988 1988

TOB T11 .

Fig. 9.4 Variation in mean worker-brood genetic relatedness computed at weekly intervals in the four colonies under study. Vertical arrows indicate events of queen turnover. Queens are assumed to be mated singly. For colonies TO8 and TII, the upper panel assumes the unknown relationship between queens (see Fig. 9.2) to be that of daughters while the lower panel assumes the relationship to be that of sisters.

198 Raghavendra Gadagkar et a/.

TO1 TO2

JAN JUL JAN JUL OCT JAN APR

1987 1987 1988 1987 1987 1988 1988

I/) 08 -' 08 I/)

I/) oj, J, I/)

~ ~

c: 0 06 c:

u u

~ ~

~ 0 04.:!9

~ ~

0 0 2

u u-

~

~ ~

0108 08 01

oj, oj, ,j, ,j, u

801. -=::::-- ~ -~ ~ 01. e

...~---~--~~~::>---~-C;--d~- .D

.D 0 0

I I

...

k ~::::=:::=:;:=~1 9 ~

~ 0 I. 9 .01. ~

~ ~~_. 0" 0 :3:0

0

OCT JAN APR OCT JAN R

1987 1988 1988 1987 1988 1988

TO8 T11

Fig. 9.5 As in Fig. 9.4 but queens are assumed to be mated multiply such that the average relatedness between sisters in a colony is 0.53 as determined in a previous electrophoretic study (Muralidharan et at. 1986; Gadagkar 1990a).

because nephews (r = 0.375) are more closely related to workers than brothers (r = 0.25). Such cyclical changes in worker-brood genetic re- latedness in response to successive replacements of queens are seen most clearly in colony TOI where there was a considerable time-lag between queen replacements.

Among the events occurring after a queen replacement, the rate of egg-laying by the new queen contributes to the initial change in worker- brood relatedness while the eclosion of offspring of the previous queen, the rate of death of workers who are offspring of the previous queen, and the rate of eclosion of female offspring of the new queen all contribute to a return to the original level of worker-brood relatedness. In colonies such as T02 and especially T08, queen replacements are so frequent that the relatedness values are uniformly low for long periods of time.

With the assumption of multiple mating, the pattern of variation in worker-brood genetic relatedness (Fig. 9.5) is qualitatively similar to that obtained under the assumption of single mating (Fig. 9.4) but it is quite different quantitatively; the values are uniformly lower in the case of multiple mating. A striking feature in Fig. 9.5, especially in TOI and the initial part of T02, is fhe near constancy in worker-male brood genetic relatedness. This is because, in the multiple mating case, worker relatedness to nephews (0.53 x 0.5 = 0.265) is nearly the same as that to brothers (0.25).

Serial polygyny in Ropalidia marginata 199

Evaluating the observed levels of relatedness

Clearly, serial polygyny leads to a reduction in genetic relatedness. But how significant is this reduction? What are the consequences of such reduction for the evolution of worker behaviour? Reduced worker-brood genetic relatedness values may last for only a very short time or may

occur only when colony sizes are very small, in which case the overall effect of serial polygyny may be only marginal. We have therefore com- puted for each colony, a grand mean worker-brood relatedness from the mean relatedness calculated at weekly intervals by using the equation:

s

L; (Wk + Bk)r k

=;:= = k=l (9.2)

s

L; (Wk + Bk) k = I

where, W k is the total number of workers present on the kth day, B k is the total number of brood present on the kth day, r k is the mean worker-brood relatedness on the kth day computed from equation (9.1), and s is the total number of days for which mean worker-brood genetic relatedness was computed using equation (9.1). Values of grand mean worker-brood relatedness for male and female brood, under single mating and multiple mating assumptions are given in Table 9.3. For each colony, the values for female brood range from 0.65 to 0.22 and the values for male brood range from 0.29 to 0.18.

How low are these values? Are they low enough to permit the con- clusion that the advantage of genetic asymmetry created by haplodiploidy for social evolution is entirely lost on account of serial polygyny and/or polyandry? A useful way of stating Hamilton's rule (Hamilton 19640) for social evolution is that a worker allele will be favoured if:

njrj > noro (9.3)

where n j is the number of individuals reared by a worker in the colony, r j is the average genetic relatedness between workers and the brood they rear, no is the number of offspring that a solitary individual can rear ar~d r 0 is her genetic relatedness to her offspring (Craig 1979; Gadagkar 19850). Inequality (9.3) can be achieved in two contrasting ways. The first is if n j > no, even if r i :S r o. The second is if r i > r 0' even if n j :S no. In other words, when rj > r 0 this may create conditions which might be sufficient for the evolution of worker behaviour without the need for any environmentally caused limitations on the productivities of solitary nest foundresses (Gadagkar 19850, 1 990b, 1991b). As men- tioned before (Gadagkar 1991 b), this is not to deny the importance of~

oC .00 ~= 0 ~",,-2 8:G,)o,c -E '" 0 .co"oo~... ~" G,) C '" a----"'001.. -~ .s:.G,)G,) ~..~~~ ~ N ..., 0 ~ ~ ~0 G,)-o ...,..., N N...,..., 00II ~~~ '=' '=' '=' '=' '=' '=' .g

'" 00 0I.. 0 -G,) 0 0 e

I.. U '".-""c .c'" I.. -

C .~ ~ ~ '" ~ 0 ~ \0 -BG,) ...I.. 'v N N N -N N '"~ ~~ 2: '=' '=' '=' '=' '=' '=' ~00... G,)... ..,CG,) 00,- 00 u'::',c

coo ...

~0-E'" ~ '" ~'" G,) '" I...G,)

EG,) ,c ~

'" G,) CC

.-

-..

U 00 -5'::, 0 cG,) = 0 .-c ~'-I.. '"G,)E°,c ~00 "'0.c

'" -

:: 5 ~ ~ 0 0 0 0 0 0 .S

.~ 00,- 00 ~~ = 0 0 ""~ 0 .'"e G,) ~ I.. ~ ~EG,) ,c 0 "".= a- '"~ 0000 ~ ~ 0= G,) ~ -~.5 ~ ~ E "'..., '" N..., r- 10 ~I.. --, \0 '" ..., ..., \0 '" "">- '" G,) ,-;, 00'"= VI.. -0 0 0 0 0 0 ..g>- 00 ~ .;00 = Co ..,., ->- G,) G,) Co G,) G,) Co '-' .c

-~ .s:. ==.- ==.- ..00 '" G,) oo=.s:. '" oo=.s:. -"'- ::c:-CoE= > "'001.. "'00"'1 ~ ~-o=.;n", c = "".~=G,)~"".~=G,) G,)~ G,) a-8~ G,) '::'~~= ,.. ,.. ~EooE.s:. ~EooE'" E.s:. E", ::..c'".--,..- """ ='::' 00"""'=.::' I..:soo:sl..I.. 00 ,g...u~ 0 0 _0~~:S:s_0~~:s~ "':s "'~ ..; 0~ .5 G,) ~ U:s = = -= = -~ ~ -= = -~ '" '" ~ '" '" ~-'- C/) ~ ,c ~ C' ~ ~ <I: ~ :s ~ ~ 00 <I: ~ :s ~ ~';n <I: 00 <I: ';n ...~

0 , c c,'" 0 10 u

c ...U I.. .., -,~ G,) '" .-u-,c='"

'-

Serial polygyny in Ropalidia marginata 201

environmental factors in social evolution but to test the role of what can be measured today with some degree of precision (genetic relatedness, r i and r 0) by not mixing it up with what cannot be measured precisely today (differential productivities in the social and solitary modes, n i and

no)'

One way to examine whether haplodiploidy leads to genetic predis- position for the evolution of worker behaviour is to assume that workers can skew investment in female and male brood in proportion to their genetic relatedness to each (Trivers and Hare 1976) and ask if the resultant J weighted mean relatedness between workers and the brood they rear is greater than 0.5, the value expected for a solitary foundress (Gadagkar 1990b, 1991b). A second way is to argue that, during the origin of social life, workers may even be selected to rear an all-female brood and thus any value of relatedness to female brood greater than 0.5 will be sufficient to cause genetic predisposition for the evolution of worker behaviour

(Gadagkar 1991 b). These arguments yield respectively two possible thres- holds in worker-brood relatedness that can be compared with the values obtained for the four colonies. One threshold is a weighted mean related- ness to brood of 0.5, taking both female and male brood into consi- deration and assuming that workers bias investment in female and male brood in proportion to their relatedness to them. The second, more con- servative threshold is simply a value of 0.5 for relatedness to female brood. If the values obtained are greater than these thresholds, this would suggest that haplodiploidy results in a genetic predisposition for the evolution of worker behaviour in R. marginata. If the values obtained are lower than these thresholds, thenhaplodiploidy does not cause genetic predisposition for the evolution of worker behaviour, and we must invoke other factors to explain the evolution of the worker caste in R. marginata.

The weighted mean relatedness to brood after sex ratio bias is never greater than 0.5 under multiple mating, and is greater than 0.5 in only two out of six cases under single mating (Table 9.3). Similarly, relatedness to female brood is never greater than 0.5 under multiple mating and is less than 0.5 for colony T08 even under the assumption of single mating.

Whichever threshold is used, the values obtained are always lower than the thresholds, under the assumption of multiple mating. Thus, we con- clude that the genetic asymmetry, potentially created by haplodiploidy, is broken down by polyandry and serial polygyny to the extent that the presence of a worker caste in this species cannot be attribut~d solely to haplodiploidy.

Simulation models

In an attempt (i) to extend these conclusions from the four colonies studied to R. marginata in general and to other species of social insects

202 Raghavendra Gadagkar et a/.

exhibiting serial polygyny and (ii) to gain some insight into the mechanism leading to the observed values of relatedness, we constructed simple simulation models of serial polygyny. Queens are assumed to lay one egg every day whose development into adults and subsequent death are monitored in the model. The three parameters used in the model and their numerical values derived empirically from studies of R. marg;nata are shown in Table 9.4. In the first model, we have merely considered the mean values of these parameters, a queen tenure of 80 days, a brood developmental period of 62 days, and a worker life span of 31 days. New queens are assumed to be either daughters or sisters of their predecessor queens. As in the case of analysis of empirical data, the model is repeated for single mating and multiple mating. As before, a mean relatedness between sisters of 0.53 derived from electrophor{tic data (Muralidharan et a/. 1986; Gadagkar 1990a) is used in the case of multiple mating.

Variation with time in the mean worker-brood relatedness in the model (Fig. 9.6) mimics clearly that in the empirical data (Figs. 9.4 and 9.5).

An examination of the grand mean worker-brood relatedness and the weighted mean relatedness of workers to brood after sex ratio bias yields results which are similar to those of the empirical analysis. Our conclusion from the simulation model is therefore identical to that from the empirical .analysis. By the criteria of either threshold, the evolution of worker behaviour cannot be attributeq only to genetic predisposition on account of haplodiploidy, whether new queens are sisters or daughters of their predecessors (Table 9.5). Serial polygyny in R: marg;nata thus reduces significantly worker-brood genetic relatedness. Even when poly- andry is ignored, haplodiploidy results in a genetic predisposition for the evolution of worker behaviour only if new queens are daughters of their predecessors and only if we use the more conservative threshold, namely a relatedness to female brood of 0.5. The relatedness value corresponding to this restricted scenario is the only one in Table 9.5 that is greater than 0.5.

Table 9.4 Parameters used in simulation models of serial polygyny in R. marginata.

Parameter Mean :t SE

(Values derived from natural colonies) Queen tenure 79.7 :t 15.8 days a

.

Brood development period 61.5 :t 1.4 days b Worker life span 31.2 :t 3.4 daysb

a Data in this chapter.

bGadagkar 1990c.

Serial polygyny in Ropalidia marginata 203 Ti me (days)

VI 0 100 200 300 0 100 200 300 VI

gJ 08 0 8 VI

c: (a) ~

"206 06"0

- ^t1J

^-

-04 C\I ---' .-, --, , 04!J. C\I

I- ',,~... ', '0 l-

v 0.2 --0 2 v ---

+::

C\I -

c: 0 C\I

~, 0 c: -:

~ C\I

0108 0801 )i

"8 (c) (d) "0 I

0 0-6 0-6 g r

I- "- r

.Q .Q

I 0-4 Q 04 I

~ 02 02 ~

~ ~

I- I-

0 0 0 0

:3: 0 100 200 300 0 100 200 300 3:

Ti me (days)

Fig. 9.6 Variation in worker-brood genetic relatedness derived from a simulation model using empirically derived mean value of 80 days as the queen tenure, 62 days as the brood developmental period and 31 days as the worker life span. Notice that worker- brood relatedness is zero for the first 62 days when no wQrkers have yet eclosed. Queens are assumed to mate singly in (a) and (b). In (c) and (d), queens are assumed to mate multiply, as in Fig. 9.5. New queens are assumed to be daughters of previous queens in (a) and (c) while they are assumed to be sisters of previous queens in (b) and (d).

The foregoing analysis ignored the variation seen in the empirically derived values of the parameters used in the model. Therefore, in the next model, we attempted to incorporate this variation with the use of Monte Carlo techniques. To do this, we reconstructed a normal distri- bution for each of the parameters of the model using the mean and standard error values in Table 9.4. We then repeated each simulation 1000 times, using random values from the normal distribution for each of

the parameters. This generated a distribution of worker-brood relatedness , values. The results of these simulations indicate that in spite of taking I into consideration the natural variation in queen tenure, brood develop- ; mental period, and worker life spans, the entire distribution of the one f thousand 'vorker-brood relatedness values (under multiple mating) is

below the threshold required for haplodiploidy-induced genetic predis- position for the evolution of worker behaviour (Fig. 9.7). This is true whether we use a weighted mean relatedness to brood (after sex-ratio bias) of 0.5 as the threshold or the more conservative threshold of relatedness to female brood of 0.5. In the case of single mating, some portion of the distribution may lie above the threshold, but even that is small except when new queens are daughters of their predecessors and

'"

a '" -

.cII '0 ~'" 0 ~...0 0 ~J!. "'... ~'" ",.0 c

"

C .~ ~ -; N -..~ ~6 ~ 00 "E

oo~ 5~ .=C~ 00 '0 ~.-.0 0 0 ~...C ... ~

a

a

-u ~

.c~ '0.-~e.o .0~ '" .":! ...I:" ~ ..~ ~~~ 0 B

~

I:0 .- >-

a

.-~.o -.to ..., .--~- ...! ::> ~ .9 0 0 ~...;> ...~ ~'" .~~ .....'0 " u,£ 00 ~-g...0 -~"~ ~ .0 ~ ...'"' 0 u~ ~ N- "'."

a.-~ -; ..., N ..u

I:~o ~ 00 .".-

S 5~ OZ.-00 '0 ..0~ ~ CO 0 (Q'."

-.-~ '" 0 ~ 0= ~ ~""" -e

a

.-C ~ ..; 0'" ~ '0] -; ,,~

~ ~ a~ a ""N OJ ~

= ~ II"!-.t. I: I:0 -rn o~ ~ 00 ~-g'" .c ..

" -. -.-u= ...~'" ..I:~ ~ ..~ 0. ~ u.'" '" =II) ~ ~ ...=.".C > ~ 0 u~ 0 5.1;; '" :c. '" .: ~

.-~ ~ C ~... "..~~ ~~u~ =J!. ..u.c ~ ~ u =Q~.~ 0 ~~ ~.o~O' rJ)...o

206 Raghavendra Gadagkar et a/.

Threshold values of queen tenure, brood developmental period, and worker life span

Next, we explore the effects of varying one of the three parameters at a time. This allows us to determine the threshold value of each parameter above which haplodiploidy can cause genetic predisposition for the evo- lution of worker behaviour. With increase in queen tenure, the related- ness to female brood, as well as the weighted mean relatedness to brood, increase (Fig. 9.8). In the case of multiple mating, given a brood develop- mental period of 62 days and a worker life span of 31 days, there is no value of queen tenure which will yield a weighted mean relatedness greater than 0.5 (Fig. 9.8(g),(h». There are values of queen tenure which will. yield a relatedness to female brood of 0.5 or more, but these are very high -465 days if new queens are daughters of their predecessors (Fig. 9.8(c» and 715 days if new queens are sisters of their predecessors (Fig. 9.8(d».

Keeping the queen tenure fixed at 80 days and the worker life span at 31 days, we then explore the consequences of varying the brood developmental period. As expected, worker-brood relatedness decreases when the brood developmental period increases (Fig. 9.9). Under multi- ple mating, there is no value of brood developmental period for which the worker -brood developmental period is higher than either threshold (Fig. 9.9(c),(d),(g),(h». Similarly, when queen tenure and brood develop- mental period are held constant at 80 and 62 days respectively, worker- brood relatedness decreases with increased worker life span (Fig. 9.10).

Once again under multiple mating, there is no value of worker life span for which worker-brood relatedness is higher than either threshold (Fig.

9.10(c),(d),(g),(h». These results add to the robustness of our conclusion that haplodiploidy does not cause the expected genetic predisposition for the evolution of worker behaviour in R. marginata. The values of the relevant parameters derived emprically for R. marginata are not merely marginally different from those required for genetic predisposition-they are very different indeed.

Mapping the parameter space where haplodiploidy can lead to genetic predisposition for the evolution of worker behaviour In an attempt to derive conclusions that may have wider applicability to other species exhibiting serial polygyny, we have delineated regions in the parameter space where haplodiploidy leads to genetic predisposition for the evolution of worker behaviour. First, we conducted an extensive search in the entire parameter space of queen tenure, brood developmental period, and worker life span, and found that the simulation model can be described adequately by two parameters, nRmely (i) queen tenure and (ii) the sum

Serial polygyny in Ropalidia marginata 209

A B

0.8

Ea B

0.6 (~! ~~!- 06

0.4 0.4

0.-2 2 0.2 "8

44 0

-00 015

0

6

.t o. 6 ~~!- ~~ 0.6 :

~ 04 041/1

-~IU c:

E 0.2 02 -0

~ 10 ~

0 0 "'iiJ

o

B ^~

~ o. 6 ~~~ ~~~ 0 6 ~

c: 0.4 04 ~

-0 E

~ 0.2 0.2

IU 0

~ 0 0 ~

a: .c

(d) .Q1

0.6 0 6 S:~

0.4 04

02 02

0 0

1 30 60 90 I 30 60 90

Worker life span (days)

Fig. 9.10 Variation in worker-brood genetic relatedness as a function of worker life span in simulation models. See legend to Fig. 9.7 for general explanations.

may be drawn from Fig. 9.11, when queens mate singly, new queens are daughters of their predecessors, and when a relatedness to female brood of 0.5 is sufficient to drive social evolution, the sum of brood develop- mental period and worker life span should be no more than 1.33 times the queen tenure. Similarly, if all other conditions in the above example remain unchanged but if a weighted mean relatedness to brood (after sex-ratio bias) of 0.5 or more is required for worker behaviour to be selected, then the sum of brood developmental period and worker life span should be no more than 0.8 times the queen tenure.

Conclusions

Serial polygyny in R. marginata leads to a significant reduction in worker -brood genetic relatedness. This implies that the evolution of worker behaviour cannot be attributed solely to haplodiploidy unless, as

Serial polygyny in Ropalidia marginata 211

argued earlier, two conditions are satisfied: (i) queens mate singly and (ii) a relatedness value to female brood greater than 0.5 is sufficient for such evolution. An important assumption made in this analysis is that workers act altruistically towards all brood present in a colony. If workers care preferentially for brood that belong to their own matriline or patriline, or in general, for brood that are relatively more closely related to them, the conclusion drawn here may not be valid. However, our limited present knowledge suggests that, at least in primitively eusocial species, workers do not show such a care bias (Gadagkar 1985b; Gamboa et of. 1986, Venkataraman et of. 1988). Another assumption made in this analysis is that R. morginoto is outbred. This is very likely as mating never takes place on the nest. Males leave their natal nests within a few days of eclosion (Gadagkar et of. 1982) and lead a nomadic life, hovering around vegetation in an apparent attempt to mate with foraging females.

The assumption of outbreeding also appears to be reasonable for most social insects studied so far (for a compilation of data, see Gadagkar 1991b).

Since we have concluded that the presence of a worker caste in R.

morginoto cannot be attributed to the genetic asymmetries created by haplodiploidy, it seems reasonable to end with a brief consideration of alternative causal factors. Curiously, the very data presented in this chapter provide a clue. Recall that the average number of adults success- fully produced by queens is 76 while most solitary nests fail before pro- ducing any adult offspring. Clearly, there is a substantial difference in the productivities of single-foundress and multiple-foundress nests. This, coupled with the fact that queen replacements can be quite frequent, supports strongly the hypothesis that female wasps stay on their natal nests in the hope of becoming queens in future (Gadagkar 1990c, 19910).

Even if the probability of becoming a queen is quite small for any given individual, the fitness gained by those who succeed can be so great that it offsets the cost incurred by the other, unsuccessful bearers of the hypothetical 'gambling' allele, which makes its bearers stay in a social group and await their chances of becoming queens (Gadagkar 1990c, 19910). But why should the productivity of a queen in a multi-female nest be so much greater than that of a solitary foundress. Ecological factors such. as differential sensitivity to predators (such as ants) and parasites (such as ichneumonid wasps and tachinid flies) are surely important but quantitative data on these are hard to come by. It ~ becoming increasingly clear too that a variety of physiological and demographic factors may interact in complex ways and lead to other kinds of predis- positions to the evolution of worker behaviour by boosting the inclusive fitness of a completely or partially sterile worker relative to that of a solitary nest foundress (Gadagkar et of. 1988, 1991; Gadagkar 1990d, 1991c).

212 Raghavendra Gadagkar et a/.

Summary

Ropa/idia marginata is a tropical Old World primitively eusocial polistine wasp with a perennial indeterminate colony cycle. Although colonies have only one egg layer at any given time, they exhibit serial polygyny due to frequent queen replacements. In a long term study, the tenure of queens varied from 7 to 236 days. New queens were daughters, sisters, nieces, or cousins of their immediate predecessors. Brood were sisters, brothers, nieces and nephews, cousins, cousins' offspring, mother's cousins, mother's cousins' offspring, or mother's cousins' grand-offspring of the workers that reared them. Electrophoretic analysis showed that queens mate multiply and use sperm simultaneously from different males, resulting in an average relatedness between sisters of 0.53. When polyandry and serial polygyny were considered simultaneously, the relatedness of workers to female brood ranged from 0.22 to 0.46 and that to male brood ranged from 0.18 to 0.26. Even if workers invest in female and male brood in proportion to their relatedness to them, the weighted mean relatedness to brood that would result ranged from 0.20 to 0.38, values which are much less than the 0.5 that a solitary foundress would obtain by rearing her own off- spring. Simulation models show that the combination of queen tenures, brood developmental periods, and worker life spans seen in R. marginata result in the complete breakdown of the genetic asymmetry created by haplodiploidy. It seems reasonable to conclude that worker behaviour in R. marginata must be attributed at least in part to factors other jhan genetic asymmetry created by haplodiploidy.

Acknowledgements

It is a pleasure to thank Niranjan Joshi, Anindya Sinha, Madhav Gadgil, Vidyanand Nanjundiah, Kenneth Ross, Robert Jeanne, Mary Jane West- Eberhard and Laurent Keller for comments on an earlier version of this chapter. Supported in part by grants from The Department of Science and Technology, Government of India and the Ministry of Environment and Forests, Government of India.

References

Bourke, A. F. G. 1988. Worker reproduction in the higher eusocialRymenoptera.

Quart. Rev. Bioi., 63, 291-311.

Chandrashekara, K. and Gadagkar, R. 1991. Unmated queens in the primitively eusocial wasp Ropalidia marginata (Lep.) (Hymenoptera: Vespidae). Ins. Soc., 38,213-216.

Chandrashekara, K., Bhagavan, S., Chandran, S., Nair, P., and Gadagkar, R.

1990. Perennial indeterminate colony cycle in a primitively eusocial wasp. In Social insects and the environment. Proc. 11th Int. Congo IUSSI (ed. G. K.