Dispersal of first "workers" in social wasps: Causes and implications of an alternative reproductive strategy

Abstract

Many "workers" in north temperate colonies of the eusocial paper wasp Polistes fuscatus disappear within a few days of eclosion. We provide evidence that these females are pursuing an alternative reproductive strategy, i.e., dispersing to overwinter and become nest foundresses the following spring, instead of helping to rear brood on their natal nests. A female is most likely to stay and help at the natal nest (i.e., least likely to disperse) when it is among the first workers to emerge and when it emerges on a nest with more pupae (even though worker-brood relatedness tends to be lower in such colonies). The latter cause may result from the fact that pupae-laden nests are especially likely to survive, and thus any direct or indirect reproductive payoffs for staying and working are less likely to be lost. Disappearing females are significantly smaller than predicted if dispersal tendency was independent of body size (emergence order-controlled), suggesting that the females likely to be most effective at challenging for reproductive rights within the natal colony (i.e., the largest females) are also most likely to stay. Thus, early dispersal is conditional on a female's emergence order, the maturity of its natal nest, and its body size. Finally, we present evidence that foundresses may actively limit the sizes of first-emerging females, perhaps to decrease the probability that the latter can effectively challenge foundresses for reproductive rights. The degree to which foundresses limit the size of first-emerging females accords well with the predictions of the theory of staying incentives.

Full text

Proc. Natl. Acad. Sci. USA
Vol. 95, pp. 13737–13742, November 1998
Evolution
Dispersal of first ‘‘workers’’ in social wasps: Causes and
implications of an alternative reproductive strategy
H. KERN REEVE
†‡
,JOHN M. PETERS
†
,PETER NONACS
§
, AND PHILIP T. STARKS
†
†
Section of Neurobiology and Behavior, Cornell University, Ithaca, NY 14853-2702; and
§
Department of Biology, University of California, Los Angeles, CA 90095
Communicated by Thomas Eisner, Cornell University, Ithaca, NY, September 8, 1998 (received for review June 20, 1998)
ABSTRACT Many ‘‘workers’’ in north temperate colonies
of the eusocial paper wasp Polistes fuscatus disappear within a
few days of eclosion. We provide evidence that these females
are pursuing an alternative reproductive strategy, i.e., dis-
persing to overwinter and become nest foundresses the fol-
lowing spring, instead of helping to rear brood on their natal
nests. A female is most likely to stay and help at the natal nest
(i.e., least likely to disperse) when it is among the first workers
to emerge and when it emerges on a nest with more pupae
(even though worker-brood relatedness tends to be lower in
such colonies). The latter cause may result from the fact that
pupae-laden nests are especially likely to survive, and thus any
direct or indirect reproductive payoffs for staying and working
are less likely to be lost. Disappearing females are signifi-
cantly smaller than predicted if dispersal tendency was inde-
pendent of body size (emergence order-controlled), suggesting
that the females likely to be most effective at challenging for
reproductive rights within the natal colony (i.e., the largest
females) are also most likely to stay. Thus, early dispersal is
conditional on a female’s emergence order, the maturity of its
natal nest, and its body size. Finally, we present evidence that
foundresses may actively limit the sizes of first-emerging
females, perhaps to decrease the probability that the latter can
effectively challenge foundresses for reproductive rights. The
degree to which foundresses limit the size of first-emerging
females accords well with the predictions of the theory of
staying incentives.
Recent studies of social insects have revealed surprisingly high
levels of plasticity in the expression of altruism, i.e., the amount
of reproductive self-sacrifice exhibited by one member of the
society on behalf of another. In some termites, for example,
workers can facultatively switch from worker morphology and
behavior to winged, dispersing reproductive forms (alates)
when the resources available to the colony become sufficiently
scarce. In addition, even soldiers can become fertile under
certain conditions (1). In the eusocial Hymenoptera, mem-
bership in either the worker or the reproductive caste is
strongly dependent on larval nutrition, which presumably
affects the benefits and costs of challenging for reproductive
dominance or for breeding solitarily (2). The relative invest-
ment by females in worker or queen-associated activities also
appears to be affected by their histories of success in compet-
itive interactions with nestmates, with losers increasing their
levels of altruistic investment in nestmate relatives (3, 4).
In temperate social wasps of the genus Polistes, females can
flexibly assume worker or reproductive roles depending on
nutrition, time of year, and social conditions such as the
presence of brood (5, 6). In addition, workers produced earlier
in the colony cycle sometimes leave their natal colonies and
found their own nests solitarily in the same season (7–9),
replace the foundress queen (10), or even lay eggs in the
presence of the queen (11). In an extreme example of plasticity
in altruism, a fraction of first-brood females in the temperate
halictid bee, Halictus rubicundus, do not become workers but
instead disperse, enter early diapause, and reappear as nest
foundresses near their natal sites the following year (12, 13).
The prevalence of facultative, early female dispersal in social
insects is unknown, because relatively few field studies couple
the systematic marking of potential workers (henceforth de-
scribed simply as first-brood or ‘‘early’’ adult females) with
follow-up censuses of adults founding nests at the females’
natal sites the following year. Moreover, early emerging fe-
males that disperse away from their natal sites will be unde-
tected in censuses of the following year’s colonies at these sites.
Finally, very little is known about the ecological, social, and
genetic factors influencing the decision to become an early
disperser, and whether early females respond to these factors
in ways that maximize their inclusive fitnesses.
We provide demographic and genetic evidence that a sub-
stantial fraction of early females in the temperate, eusocial
paper wasp Polistes fuscatus disperse as future nest foundresses
instead of staying and helping at their natal nests. Thus,
facultative early female dispersal must have evolved indepen-
dently at least twice in the eusocial Hymenoptera (polistine
wasps and halictid bees). In addition, we show how the
probability of early dispersal by a female is affected by its
emergence order, its body size, the number of foundresses on
its natal nest, and its natal nest’s size and maturity. Finally, we
provide evidence that foundresses may nutritionally manipu-
late the adult sizes of the earliest first-emerging females, i.e.,
females that are likely to stay at the natal nest and become
dominant workers.
MATERIALS AND METHODS
Study Populations. We studied field-nesting Polistes fuscatus
colonies at five sites around Ithaca, New York in 1993 and
1994. In our study population, foundresses initiate nests near
mid-May (31y107 5 29% of sampled colonies were founded by
multiple foundresses in both years), and the first female
progeny (potential dispersers or workers) begin to emerge in
the first week of July. This worker-production phase proceeds
until the first males and gynes (late-season females destined to
become foundresses the following spring) begin emerging
around 1 August.
In 1993, we studied 15 colonies (9 single-foundress and 6
multiple-foundress colonies) from 23 June to 2 August at two
sites separated by approximately 1 km. A total of 21 found-
resses and 105 early females (potential workers) were given
individually specific paint-marks with Testor’s enamel, and the
winglengths of foundresses and their adult offspring were
measured (from tegulum to wingtip) as soon as they were first
censused. Each colony was censused six times during the study
period. On 2 August, the remaining females of surviving
The publication costs of this article were defrayed in part by page charge
payment. This article must therefore be hereby marked ‘‘advertisement’’ in
accordance with 18 U.S.C. §1734 solely to indicate this fact.
© 1998 by The National Academy of Sciences 0027-8424y98y9513737-6$2.00y0
PNAS is available online at www.pnas.org.
‡
To whom reprint requests should be addressed. e-mail: hkr1@
cornell.edu.
13737

colonies (n 5 11) were collected and frozen for microsatellite
genetic analyses.
On 29 July, 1993, five females were found (accidentally)
between the stacked combs of a partly exposed, abandoned
Vespula nest. Four wasps had marks of early females that
emerged prior to 12 July on colonies within 100 m and were in
apparent early diapause, i.e., the wasps were extremely slug-
gish, as if cooled. These wasps were collected and frozen for
genetic analyses.
In 1994, we focused on 47 colonies (31 single-foundress and
16 multiple-foundress colonies) from 2 June to 1 August at
four different sites ranging from 1 to 4 km apart. A total of 69
foundresses and 173 early females were paint-marked for
individual identification; all foundresses and a subsample of
workers from 13 colonies were measured for winglength. The
colonies at one site, the Liddell field laboratory (19 colonies),
were frequently censused between 2 June and 1 August (36
censuses, with at least 1 day between censuses). Between 6
June and 2 August there were 8 censuses at the other three sites
(28 colonies total). On 8 August, the adults present on 12
surviving colonies (4 single-foundress and 8 multiple-
foundress) were collected and frozen for microsatellite genetic
analyses.
Census Data. For each colony, we recorded the identifica-
tions of all resident adults and marked new adults, noting
whether the latter were newly eclosed wasps, as indicated by
black eyes. We define a cohort as a group of newly eclosed,
early females appearing together for the first time at a census.
Cohorts were categorized by their order of emergence, with the
first cohort of females being the first censused group of newly
eclosed (black-eyed) unmarked females, the second cohort
being the next censused group of newly eclosed females, etc.
Females of all of the cohorts in our study had been reared by
foundresses exclusively (not by workers). Different cohorts
emerged within only a few days of each other. For example,
second-cohort females emerged on average only 3.4 days later
than first-cohort females. Thus, first-cohort workers played no
role in the second cohort’s rearing because the latter were
already at the pupal stage (when no feeding occurs).
The nest tenure for each early female, i.e., the number of
days each female spent on nest, was estimated for the inten-
sively censused 1994 Liddell colonies as follows. The female’s
first day was calculated as the midpoint between the census day
before the female was first seen and the census day that the
female was first seen and marked; the female’s last day was
calculated as the midpoint between the census day it was last
seen and the first census day after its (permanent) disappear-
ance. Nest tenure was then estimated as the difference between
the female’s last and first days. Nest size was measured as the
total number of brood cells on 30 June, which is just prior to
initial worker emergence. Overall nest maturity was measured
as the number of pupae divided by the total number of brood
cells on the same date.
As a measure of the propensity of early females to disappear
from a colony, we calculated the proportion of females eclosing
on or prior to 12 July (for 1993 colonies) or 8 July (for 1994
colonies) that remained with the colony until the end of the
worker phase on 1 August.
Genetic Analyses. After screening microsatellite primers
derived from Polistes and Parachartergus wasps sent to us by
Joan Strassmann and David Queller (Rice University), we
found eight primers detecting polymorphic loci having from 4-
to 21-length alleles and with heterozygosities ranging from 0.60
to 0.91. The primer loci with corresponding mean heterozy-
gosity and number of alleles are, respectively: Pbe128TAG,
0.61, 6; Pbe269bAAG, 0.82, 12; Pbe411AAT, 0.84, 9;
Pbe424AAT, 0.86, 12; Pbe440AAT, 0.91, 21; Pbe442AAT,
0.70, 8; Paco3155TAG, 0.65, 6; Paco3219AAG, 0.60, 4.
We used the methods of Choudhary et al. (14) to obtain
microsatellite genotypes at all eight loci for a total of 98 early
females from 11 1993 colonies (7 single-foundress and 4
multiple-foundress) and 12 1994 colonies (4 single-foundress
and 8 multiple-foundress). We used the
RELATEDNESS program
(version 4.2) of Goodknight and Queller (15), based on the
logic of Queller and Goodnight (16), to estimate the mean
relatedness among wasps.
Statistical Methods. Parametric tests were used only when
assumptions of normality appeared satisfied. Proportions of
workers staying on the nest were arcsine square root-
transformed for all statistical tests. A three-way analysis of
variance of the effect of site, foundress number, and foundress
presence (i.e., whether at least one foundress remained on the
nest until 1 August) was performed on the transformed
proportion of staying workers. For this analysis, the five sites
were combined into two sites (the five sites fell into two spatial
clusters .1 km apart) to obtain a more balanced design. Early
female tenures measured in the 1994 Liddell colonies were
square root-transformed to yield approximate normality for
parametric tests. All tests are two-tailed.
RESULTS
Genetic Data. The mean relatedness among single-
foundress early females (sampled 2–8 August) was 0.74 6 SE
0.04, which is close to the theoretical value of 0.75 for singly
mated, outbreeding queens (n 5 11 colonies). In multiple-
foundress colonies, the mean relatedness among early females
(sampled 2–8 August) was only 0.38 6 0.04 (n 5 11 colonies
with multiple early females), which is significantly less than
that for single-foundress colonies (P 5 0.0002; two-tailed
Mann–Whitney U test). The latter result suggests that cofound-
resses share in reproduction.
Evidence of Early Female Dispersal. In 1993 and 1994, four
unambiguously identifiable females that emerged between 3
and 12 July permanently disappeared from their natal nests
within 3 days of eclosion. Three of the females were later found
sitting on inactive combs (two on 22 July, 1993, one on 7
September, 1994, on nests .10 m from their natal nests), and
one female was on a distant active nest late in the season (7
September 1994 on a nest .100 m from the natal colony). Late
in the season, the mixing of gynes (future foundresses) from
different natal colonies on the few remaining active large
colonies is common, as these females attempt to parasitize
dwindling resources from the few colonies still possessing some
foraging workers (Reeve, unpublished data).
In 1994, a female that emerged on 8 July disappeared by 13
July and returned on 16 May, 1995 to singly found a new nest
near its natal nest. [A second probable case of spring nest-
founding by an early female marked the previous summer was
documented in 1997, and a third case has been observed in a
Michigan population of P. fuscatus (G. Gamboa, personal
communication)].
On 29 July, 1993, five females were found between the
stacked combs of a partly exposed, abandoned Vespula nest.
The wasps, four of which had the same marks as females
disappearing from nests within 100 m prior to 12 July 1993,
were lethargic and apparently in diapause. The females had a
low mean relatedness to each other (r 5 0.138, SE 5 0.13), but
two of them (likely full sisters) had identical genotypes (r 5
1.00) that closely matched the genotype of a foundress in a
multiple-foundress colony about 30 m away (mean relatedness
to foundresses 5 0.54). Indeed, females with marks exactly
matching those of the diapausing wasps had disappeared from
the latter colony by 12 July. The other two marked females
could not be matched genetically to their natal colonies
because there were no adults in several (vacated) colonies at
the time of collection, but their marks corresponded to the
marks of missing early females (and not foundresses or gynes)
from colonies in the area. (In July, 1998, two more disappear-
ing, first-cohort, marked females were recovered in a diapause
13738 Evolution: Reeve et al. Proc. Natl. Acad. Sci. USA 95 (1998)

aggregation with unmarked females approximately2mfrom
the natal colony.)
Overall, 10 of 278 marked early females in 1993–1994 (3.6%)
disappeared from their natal nests shortly after eclosion and
were later recovered in contexts associated with overwintering
and nest founding in the following season. Nearly 75% of early
females disappeared during the worker phase (Fig. 1). The low
percentage of disappearing early females that were later
recovered (5%) may greatly underestimate the actual percent-
age of early dispersers (as opposed to workers dying while
working for their colonies) if such females typically disperse
out of their natal sites (before or after overwintering) or have
high mortality rates. Both population-genetic and mark-and-
recapture evidence for our population indicate that dispersal
out of the immediate vicinity of the natal nest before nest
founding andyor a high mortality rate also is typical of
late-season females (gynes) known to become foundresses the
following spring. For example, only 11 of 130 or 9% of fall
gynes (females emerging after 1 August) marked in 1994 at
Liddell colonies were recovered as foundresses at the same site
the following spring, in contrast to many Polistes populations
in which marked foundresses typically return with high prob-
ability to found nests near their natal nests (17).
Determinants of Early Female Nest Tenure. Early females
frequently disappeared from nests shortly after eclosion: 47%
of first- and second-cohort wasps and 31.5% of last-cohort
wasps had mean nest tenures of less than 10 days post-eclosion
(mean tenures calculated by cohort across 1994 Liddell colo-
nies). Nearly one-third (29%) of all early females had tenures
of less than 5 days post-eclosion (1994 Liddell colonies). The
overall frequency distribution of early female tenures (1994
Liddell colonies pooled) is strongly bimodal, with most females
either disappearing within 5 days of emergence or staying at
their natal colonies for the entire worker phase and beyond
($30 days; Fig. 1). These frequency distributions are markedly
different from the monotonically decreasing, negative expo-
nential distributions expected if there was a constant proba-
bility of disappearance per unit time.
A three-way ANOVA controlling for site effects shows that
early females were (i) more likely to disappear from single-
foundress colonies than from multiple-foundress colonies and
(ii) more likely to disappear from colonies without foundresses
than from colonies with at least one foundress present (Table
1; 1993 and 1994 data on proportion of staying workers
combined).
For the 1994 Liddell data only, for which we had detailed
data on tenures of marked early females (see Materials and
Methods), we performed a two-factor ANOVA on early female
tenure versus foundress number (multiple or single) and
presence or absence of the foundress(es). For the first female
cohort, the results were similar to the above: mean tenures
were significantly longer in multiple-foundress than in single-
foundress colonies (P 5 0.02) and also marginally significantly
longer in colonies with foundresses present versus those
without foundresses present throughout the entire worker
phase (P 5 0.05; mean-square-root tenures of first cohort 5
2.96 for single-foundress colonies without a foundress, 4.07 for
single-foundress colonies with a foundress, 4.41 for multiple-
foundress colonies with no foundresses, and 6.40 for multiple-
foundress colonies when at least one foundress remained).
However, there were no such relationships for the second-
female cohort (P 5 0.70 for queen-number effect and P 5 0.33
for foundress-presence effect; mean-square-root tenures of
first cohort 5 3.79 for single-foundress colonies without a
foundress, 3.90 for single-foundress colonies with a foundress,
2.41 for multiple-foundress colonies with no foundresses, and
4.45 for multiple-foundress colonies when at least one found-
ress remained).
The overall mean early female tenure in single-foundress
colonies was 13.41 days (SD 5 9.33) for the first cohort and
17.27 (SD 5 12.87) days for the second cohort. The overall
mean early female tenure in multiple-foundress colonies was
30.18 days (SD 5 17.09) for the first cohort and 15.75 (SD 5
20.36) days for the second cohort. The difference in square
root-transformed tenures between first and second cohorts in
single-foundress colonies was significantly less than that in
FIG. 1. Frequency distribution of tenures for first-, second-, and
later-cohort early females (data from pooled 1994 Liddell colonies).
Table 1. Proportions of early females staying on the nest
throughout the worker phase as a function of foundress presence
and initial foundress number (n 5 45; 1993 and 1994
colonies combined)
Foundress status
Proportion of early females staying on nest
Single-foundress Multiple-foundress
Absent 0.098 0.109
Present 0.112 0.259*
*Three-factor ANOVA (site, foundress presence, foundress number)
revealed a significant positive effect of foundress presence (F test 5
7.70; P 5 0.0086) and a significant positive effect of foundress
number (F test 5 7.66; P 5 0.0088).
Evolution: Reeve et al. Proc. Natl. Acad. Sci. USA 95 (1998) 13739

multiple-foundress colonies (P 5 0.02, Student’s t test). Thus,
the first emerging workers in multiple-foundress colonies were
especially likely to stay; all others tended to disappear at
similarly high rates.
To determine the most important proximate factors in a
worker’s decision to stay rather than to disperse, we performed
a multiple regression of the square root of the tenure of the
first cohort (1994 Liddell colonies) as a function of four
(intercorrelated) variables that are likely assessable by work-
ers: queen number, foundress presence (1 5 present; 0 5ab-
sent), comb size on 30 June (just prior to emergence of first
females), and nest maturity on 30 June. Only the nest maturity
was a significant predictor of worker staying (partial regression
coefficient 5 6.07; P 5 0.01); queen number, foundress
presence, and comb size (P 5 0.45, 0.13, and 0.64, respectively)
exerted effects only indirectly through their correlations with
mean brood maturity. Although mean brood maturity was the
single most important correlate of worker tenure, more mature
combs did tend be larger. Indeed, the number of pupae (i.e.,
the product of mean brood maturity and cell number) was
strongly positively correlated with the mean tenure of the first
worker cohort (Fig. 2). Thus, it appears that first-emerging
workers stay only if the proportion or number of pupae in their
colonies is sufficiently large. Because there are more pupae in
multiple than in single-foundress colonies, and because found-
resses are more likely to disappear from nests with few pupae,
such a staying rule explains why workers are most likely to stay
and help in multiple-foundress colonies with at least one
foundress present.
Worker Size and Dispersal. The mean sizes (winglengths) of
staying and leaving early females in single-foundress colonies
(1.31 cm 6 0.02 SE and 1.25 cm 6 0.02 SE, respectively) were
significantly less than the corresponding mean sizes of early
females in multiple-foundress colonies (1.39 cm 6 0.020 SE;
P 5 0.01 and 1.34 cm 6 0.02 SE; P 5 0.028, respectively;
two-tailed t tests; 1993 and 1994 colonies).
First-cohort females had significantly smaller winglengths
(mean 5 1.29 cm 6 0.02 SE) than second- or later-cohort
females (mean 5 1.34 cm 6 0.02 and 1.36 cm 6 0.02,
respectively; P , 0.0001, repeated-measures ANOVA, n 5 18
1993 and 1994 colonies with measured workers from at least
three cohorts). The latter two cohorts did not differ signifi-
cantly in mean size.
First-cohort mean female size was significantly positively
correlated with the mean size of their foundress(es) in all 1993
and 1994 colonies for which both sizes were available (Fig. 3).
However, despite the fact that later-female cohorts also had
been exclusively reared by foundresses, there was no significant
correlation between the mean size of second- or later-cohort
females and the mean size of their foundress(es) (Fig. 3). In
single-foundress colonies, the association between first-cohort
female mean size and foundress size was positive but not
significant (n 5 10 colonies; regression slope 510.37; P 5
0.28); however, this association was both positive and signifi-
cant in multiple-foundress colonies (n 5 10 colonies; regres-
sion slope 510.73; P 5 0.04).
Early females that permanently left the nest were, on
average, significantly smaller than the size expected if leaving
was independent of size (Fig. 4; expected size 5 mean size of
all second-and later-cohort females, because second- and
FIG. 2. Correlation between early-female tenure in days (square
root-transformed) and number of pupae at initial worker eclosion
(1994 Liddell colonies).
FIG.3. (Top) Correlation between mean size of first cohort of early
females and mean size of foundresses. (Middle) Correlation between
mean size of second cohort of early females and mean size of
foundresses. (Bottom) Correlation between mean size of last cohort of
early females and mean size of foundresses. n 5 18–20 colonies (1993
and 1994 Liddell colonies for which both foundress and first- and
later-cohort winglength data were available).
13740 Evolution: Reeve et al. Proc. Natl. Acad. Sci. USA 95 (1998)

later-cohort females were most likely to leave). Conversely,
staying females were significantly larger than expected if
leaving was independent of size (Fig. 4; expected size 5 mean
size of all first-cohort females, because first-cohort females
were most likely to stay). In addition, the correlation between
the square root-transformed mean tenure for first-cohort
females and mean foundress size was significantly negative (r 5
20.52; P 5 0.026; n 5 19 1994 Liddell colonies).
DISCUSSION
Our evidence indicates that many early disappearing ‘‘work-
ers’’ are pursuing selfish reproductive options, e.g., overwin-
tering to become foundresses the following year, instead of
helping to rear brood on their natal nests. The probability of
early female dispersal is affected by the state of its colony. A
first-eclosing female is most likely to stay and help at the natal
nest when the female (i) is large and (ii) is among the first to
emerge on a colony founded by multiple foundresses (despite
the lower mean relatedness to brood in single- versus multiple-
foundress colonies). A partial regression analysis indicates that
the proportion or total number of pupae on the natal nest may
be the single most important proximate determinant of the
decision to stay and help.
Why are early females more likely to help on nests with more
pupae despite their lower relatedness to nestmates on such
nests? We propose three hypotheses. (i) Workers are more
likely to escape queen reproductive policing on larger nests.
However, this hypothesis by itself does not explain why females
work, and, moreover, existing data suggest that worker pro-
duction of reproductives in P. fuscatus colonies with found-
resses present is rare (18). (ii) Early females prefer to become
workers on larger, more mature nests because more resources
(per adult) that are critical for overwintering can be obtained
on such nests. However, the latter benefit is devalued to the
extent that risky or energetically costly helping itself reduces
the probability of successful overwintering (5, 6, 8). (iii) More
mature nests will produce more nest-tending adults sooner and
thus are less likely to fail from chance loss of all nest-tending
adults. Indeed, this mortality factor generates the principal
advantage for nest cofounding by multiple foundresses in our
study population (19). Thus, in accordance with Queller’s (20,
21) ‘‘head-start’’ or survival-insurance hypothesis, workers
prefer to stay and help on larger, more mature natal nests
(which will soon have additional helping adults) because the
worker’s direct or indirect reproductive payoffs are less likely
to be lost through colony failure. This hypothesis also explains
the otherwise puzzling observation that the proportion or
number of pupae may be the most potent single proximate
determinant of worker staying.
Why are earlier emerging females more likely to become
workers? Because older workers generally are dominant to
younger workers in Polistes (reviewed in ref. 22), it follows that
first-emerging workers would have a relatively high direct
reproductive stake in the natal colony [either via reproduction
in the presence of the queen or through queen supersedure, the
latter being well documented (reviewed in ref. 22)]. This
greater opportunity for personal reproduction in the natal
colony may make staying favorable for earlier emerging but not
later emerging females. Alternatively, slightly later emerging
females may be more likely to disperse because the prior
presence of workers diminishes the degree to which these
females could further improve colony success—as when colony
success is a concave function of the number of helping workers
(23).
Departing early females are significantly smaller than ex-
pected if dispersal tendency was independent of body size.
Conversely, staying workers are larger than expected (emer-
gence-order controlled; Fig. 4). These results make sense if a
larger worker is more effective at challenging foundresses for
reproductive rights within its natal nest and thus benefits more
from staying. In support of this hypothesis, first-cohort early
females had shorter mean tenures when their foundresses were
larger (see Results).
Intriguingly, our evidence suggests that foundresses nutri-
tionally manipulate (reduce) the sizes of early females that are
most likely to become reproductive threats. First-emerging
early females, which are likely to be dominant over younger
workers if they remain at the natal nest (reviewed in ref. 22),
are significantly smaller than later-emerging early females. The
latter result may reflect foundress regulation of brood nutri-
tion to ensure that first-emerging workers are not too large and
thus not too effective at challenging for reproductive rights.
It may be argued that the smaller size of earlier emerging
workers is simply a reflection of a passive tendency for smaller
workers to develop faster or a lower food availability for the
first brood. However, the involvement of the foundress in
precisely regulating the sizes of the early females is suggested
by the significant, positive correlation between the mean size
of the foundress(es) and their first-emerging workers, with
mean foundress size consistently exceeding worker size (Fig.
4). Importantly, there is no such significant correlation be-
tween sizes of foundress(es) and sizes of slightly later emerging
females, the latter often being larger than the foundress(es)
(Fig. 4); these later emerging females, like wasps of all cohorts
in our study, had been reared exclusively by foundresses.
Together, these data suggest that foundresses nutritionally
ensure that first-emerging females are no larger than some
threshold size that is positively related to the sizes of the
foundresses themselves. By limiting a first-emerging female’s
size, the foundress can lower the relative fighting ability of a
female likely to become a dominant worker, thereby reducing
the effectiveness of reproductive challenges from that female
or the size of any peace incentives conceded to the female.
Because first-emerging females are most likely to become
highly ranked workers (reviewed in ref. 22), foundresses may
be favored to manipulate only the earliest-developing brood.
The degree to which foundresses reduce the size of first-
emerging females should be constrained by the greater pro-
pensity of smaller females to leave the colony altogether (see
above). In particular, foundresses should reduce the first-
emerging female’s size just to the point at which leaving the
colony would become favored for the latter. This theoretical
minimal worker size can be derived from the theory of staying
incentives (25, 26), if we assume that the probability of direct
reproduction by a worker is an increasing function p(s
w
ys
f
) of
the ratio of the worker’s size s
w
to the controlling foundress’s
size s
f
. Let r be the potential worker’s mean relatedness to the
foundress’s offspring, divided by its relatedness to its own
offspring (27). Let x be the expected success of a dispersing
FIG. 4. Mean observed minus expected winglengths for early
females that disappeared from or stayed at the colony.
Evolution: Reeve et al. Proc. Natl. Acad. Sci. USA 95 (1998) 13741

early female and k be the colony success if the potential worker
stays, both standardized relative to a success of 1.0 for the
foundress if the early female disperses. The foundress should
then reduce the early female’s size just to the point at which
p~s
w
ys
f
! 5
x 2 r~k 2 1!
k~1 2 r!
~r , 1! [1]
(see refs. 26 and 27). (Note: The foundress benefits from
retaining an early female as a worker if x , k 2 1.)Ifthe
foundress’s optimal ratio s
w
ys
f
equals a*, then it follows that
the size of the first-cohort female should be linearly related to
the size of the foundress, i.e., s
w
5 a*s
f
. Moreover, according
to Eq. 1, the slope a* should be higher in multiple-foundress
colonies (lower r) than in single-foundress colonies (higher r),
exactly as observed (see Results). The model further predicts
that early females in multiple-foundress colonies should be
larger than those in single-foundress colonies, which also was
observed (see Results). Thus, our data are consistent with the
staying-incentive model of worker-size manipulation by found-
resses.
The remarkable behavioral plasticity exhibited by early P.
fuscatus females reinforces growing evidence that social be-
havior is strongly context-dependent, even in insects (24). In
addition, the ability of early females to disperse as future
foundresses has crucial implications for tests of sex ratio and
‘‘optimal skew’’ theory. For tests of sex ratio theory, early
female dispersal means that previous empirical studies have
probably underestimated the degree to which social wasp
sex-investment ratios are female-biased and thus are an ex-
pression of worker versus queen genetic interests (28). For
tests of optimal skew theory (25, 26), it now becomes necessary
to consider the possibility that dominant foundresses allow
subordinates to produce reproductively valuable ‘‘workers’’ as
a direct incentive to cooperate in a multiple-foundress asso-
ciation.
We thank Lee Dugatkin and George Gamboa for valuable com-
ments. The 1998 early-female diapause aggregation was discovered by
Sandra Nasrallah. H.K.R. and P.N. were supported by a National
Science Foundation grant; H.K.R. was also partly supported by a New
York State Hatch grant.
1. Shellman-Reeve, J. (1997) in Social Behavior in Insects and
Arachnids, eds. Choe, J. & Crespi, B. (Cambridge Univ. Press,
Cambridge, U.K.), pp. 52–93.
2. Wheeler, D. E. (1986) Am. Nat. 128, 13–34.
3. West-Eberhard, M. J. (1967) Science 157, 1584–1585.
4. West Eberhard, M. J. (1981) in Natural Selection and Social
Behavior, eds. Alexander, R. D. & Tinkle, D. W. (Chiron, New
York), pp. 3–17.
5. Solis, C. R. & Strassmann, J. E. (1990) Funct. Ecol. 4, 531–542.
6. Mead, F., Pratte, M. & Gabouriaut, D. (1993) Insectes Soc. 37,
236–250.
7. Strassmann, J. E. (1981) Behav. Ecol. Sociobiol. 8, 55–64.
8. Mead, F. & Gabouriaut, D. (1995) Insectes Soc. 42, 385–396.
9. Page, R. E., Post, D. C. & Metcalf, R. A. (1989) Am. Nat. 134,
731–748.
10. Queller, D. C., Peters, J. M., Solis, C. R. & Strassmann, J. E.
(1997) Behav. Ecol. Sociobiol. 40, 3–16.
11. Miyano, S. (1986) Res. Pop. Ecol. 28, 347–361.
12. Yanega, D. (1988) Proc. Natl. Acad. Sci. USA 85, 4374–4377.
13. Yanega, D. (1989) Behav. Ecol. Sociobiol. 24, 97–107.
14. Choudhary, M., Strassmann, J. E., Solis, C. R. & Queller, D. C.
(1993) Biochem. Genet. 31, 87–96.
15. Goodknight, K. F. & Queller, D. C. (1994) Goodknight Software.
16. Queller, D. C. & Goodknight, K. F. (1989) Evolution 43, 258–275.
17. Klahn, J. E. (1979) Behav. Ecol. Sociobiol. 5, 417–424.
18. Metcalf, R. A. (1980) Am. Nat. 116, 642–654.
19. Reeve, H. K. & Nonacs P. (1997) Behav. Ecol. 8, 75–82.
20. Queller, D. C. (1989) Proc. Natl. Acad. Sci. USA 86, 3224–3226.
21. Queller, D. C. (1994) Proc. R. Soc. London Ser. B 256, 105–111.
22. Reeve, H. K. (1991) in The Social Biology of Wasps, eds. Ross, K.
& Matthews, R. (Cornell Univ. Press, Ithaca, NY), pp. 99–148.
23. Nonacs, P. (1991) Oikos 61, 122–125.
24. Nonacs, P. & Reeve, H. K. (1995) Ecology 76, 953–967.
25. Vehrencamp, S. L. (1983) Anim. Behav. 31, 667–682.
26. Reeve, H. K. & Ratnieks, F. A. (1993) in Queen Number and
Sociality in Insects, ed. Keller, L. (Oxford Univ. Press, Oxford,
U.K.), pp. 45–85.
27. Reeve, H. K. & Keller, L. (1995) Am. Nat. 145, 119–132.
28. Noonan, K. M. (1978) Science 199, 1354–1356.
13742 Evolution: Reeve et al. Proc. Natl. Acad. Sci. USA 95 (1998)